Infant Formula - NBK 2004

http://www.nap.edu/catalog/10935.
html
We ship printed books within 1 business day; personal PDFs are available immediately.
Infant Formula: Evaluating the Safety of New

Ingredients
Committee on the Evaluation of the Addition of
Ingredients New to Infant Formula
ISBN: 0-309-53022-9, 224 pages, 6 x 9, (2004)
This PDF is available from the National Academies Press at:
http://www.nap.edu/catalog/10935.html
Visit the National Academies Press online, the authoritative source for all books
from the National Academy of Sciences, the National Academy of Engineering,
the Institute of Medicine, and the National Research Council:
• Download hundreds of free books in PDF
• Read thousands of books online for free
• Explore our innovative research tools – try the “Research Dashboard” now!
• Sign up to be notified when new books are published
• Purchase printed books and selected PDF files
Thank you for downloading this PDF. If you have comments, questions or
just want more information about the books published by the National
Academies Press, you may contact our customer service department toll-
free at 888-624-8373, visit us online, or send an email to
[email protected].
This book plus thousands more are available at http://www.nap.edu.

Copyright © National Academy of Sciences. All rights reserved.
Unless otherwise indicated, all materials in this PDF File are copyrighted by the National
Academy of Sciences. Distribution, posting, or copying is strictly prohibited without
written permission of the National Academies Press. Request reprint permission for this book.
Infant Formula: Evaluating the Safety of New Ingredients
INFANT
FORMULA
EVALUATING THE SAFETY OF NEW INGREDIENTS
Committee on the Evaluation of the Addition of Ingredients

New to Infant Formula
Food and Nutrition Board
INSTITUTE OF MEDICINE
OF THE NATIONAL ACADEMIES
THE NATIONAL ACADEMIES PRESS

Washington, D.C.
www.nap.edu

THE NATIONAL ACADEMIES PRESS • 500 Fifth Street, N.W. • Washington, DC 20001
NOTICE: The project that is the subject of this report was approved by the Governing Board of the
National Research Council, whose members are drawn from the councils of the National Academy of
Sciences, the National Academy of Engineering, and the Institute of Medicine. The members of the
committee responsible for the report were chosen for their special competences and with regard for
appropriate balance.
This study was supported by Contract No. 223-01-2460, TO4, and Contract No. 4500033271 be-
tween the National Academy of Sciences and the Food and Drug Administration, U.S. Department of
Health and Human Services, and Health Canada. Any opinions, findings, conclusions, or recommen-
dations expressed in this publication are those of the author(s) and do not necessarily reflect the view
of the organizations or agencies that provided support for this project.
Library of Congress Cataloging-in-Publication Data
Institute of Medicine (U.S.). Committee on the Evaluation of the Addition of Ingredients New to Infant
Formula.
Infant formula : evaluating the safety of new ingredients / Committee on the Evaluation of the
Addition of Ingredients New to Infant Formula, Food and Nutrition Board.
p. ; cm.
Includes bibliographical references.
ISBN 0-309-09150-0 (pbk.)
1. Infant formulas—Standards—United States. 2. Infant formulas—Law and legislation—United
States.
[DNLM: 1. Infant Formula—standards. 2. Evaluation Studies. 3. Food Analysis—methods. 4.
Food Contamination—prevention & control. 5. Infant Formula—legislation & jurisprudence. 6.
Infant Nutrition. WS 120 I59 2004] I. Title.
RJ216.I538 2004
613.2′083′2—dc22
2004008412
Additional copies of this report are available from the National Academies Press, 500 Fifth Street,
N.W., Lockbox 285, Washington, DC 20055; (800) 624-6242 or (202) 334-3313 (in the Washington
metropolitan area); Internet, http://www.nap.edu.
For more information about the Institute of Medicine and the Food and Nutrition Board, visit the
IOM and FNB home pages at www.iom.edu and www.iom/fnb.edu.
Copyright 2004 by the National Academy of Sciences. All rights reserved.

Printed in the United States of America.
The serpent has been a symbol of long life, healing, and knowledge among almost all cultures and
religions since the beginning of recorded history. The serpent adopted as a logotype by the Institute of
Medicine is a relief carving from ancient Greece, now held by the Staatliche Museen in Berlin.


The National Academy of Sciences is a private, nonprofit, self-perpetuating society of distinguished

scholars engaged in scientific and engineering research, dedicated to the furtherance of science and
technology and to their use for the general welfare. Upon the authority of the charter granted to it by
the Congress in 1863, the Academy has a mandate that requires it to advise the federal government on
scientific and technical matters. Dr. Bruce M. Alberts is president of the National Academy of Sciences.
The National Academy of Engineering was established in 1964, under the charter of the National
Academy of Sciences, as a parallel organization of outstanding engineers. It is autonomous in its
administration and in the selection of its members, sharing with the National Academy of Sciences the
responsibility for advising the federal government. The National Academy of Engineering also spon-
sors engineering programs aimed at meeting national needs, encourages education and research, and
recognizes the superior achievements of engineers. Dr. Wm. A. Wulf is president of the National
Academy of Engineering.
The Institute of Medicine was established in 1970 by the National Academy of Sciences to secure the
services of eminent members of appropriate professions in the examination of policy matters pertain-
ing to the health of the public. The Institute acts under the responsibility given to the National
Academy of Sciences by its congressional charter to be an adviser to the federal government and, upon
its own initiative, to identify issues of medical care, research, and education. Dr. Harvey V. Fineberg is
president of the Institute of Medicine.
The National Research Council was organized by the National Academy of Sciences in 1916 to associ-
ate the broad community of science and technology with the Academy’s purposes of furthering knowl-
edge and advising the federal government. Functioning in accordance with general policies determined
by the Academy, the Council has become the principal operating agency of both the National Academy
of Sciences and the National Academy of Engineering in providing services to the government, the
public, and the scientific and engineering communities. The Council is administered jointly by both
Academies and the Institute of Medicine. Dr. Bruce M. Alberts and Dr. Wm. A. Wulf are chair and vice
chair, respectively, of the National Research Council.
www.national-academies.org

COMMITTEE ON THE EVALUATION OF THE ADDITION OF INGREDIENTS

NEW TO INFANT FORMULA
RICHARD J. DECKELBAUM (Chair), Institute of Human Nutrition and Pediatric

Gastroenterology and Nutrition, Columbia University, New York, New York
LINDA ADAIR, Department of Nutrition, Schools of Public Health and Medicine,
University of North Carolina at Chapel Hill
MARK APPELBAUM, Department of Psychology, University of California, San Diego
GEORGE L. BAKER, Snowmass Village, Colorado
SUSAN S. BAKER, Department of Pediatrics, State University of New York at Buffalo
CHESTON M. BERLIN, Department of Pediatrics, Pennsylvania State University,
Hershey
WILLIAM C. FRANKE, Center for Advanced Food Technology, Rutgers, State
University of New Jersey, New Brunswick
MICHAEL K. GEORGIEFF, Department of Pediatrics, University of Minnesota,
Minneapolis
JAMES M. NTAMBI, Department of Biochemistry, University of Wisconsin, Madison
THEODORE D. WACHS, Department of Psychological Science, Purdue University,
West Lafayette, Indiana
Consultants
CARMI ZVI MARGOLIS, University Center for Health Sciences, Ben-Gurion
University of the Negev, Israel
PEARAY OGRA, State University of New York at Buffalo
ODED SUSSKIND, Brookline, Massachusetts
Editor
SARAH CARROLL, Minnetonka, Minnesota
Staff
MARIA ORIA, Study Director (from June 2003)
PAULA TRUMBO, Study Director (September 2002–June 2003)
SANDRA SCHLICKER, Study Director (until September 2002)
LESLIE J. VOGELSANG, Research Assistant
SANDRA AMAMOO-KAKRA, Senior Project Assistant (from September 2002)
HARLEEN K. SETHI, Senior Project Assistant (until September 2002)

FOOD AND NUTRITION BOARD
CATHERINE E. WOTEKI (Chair), College of Agriculture, Iowa State University, Ames

ROBERT M. RUSSELL (Vice Chair), Jean Mayer U.S. Department of Agriculture Human
Nutrition Research Center on Aging, Tufts University, Boston, Massachusetts
LARRY R. BEUCHAT, Center for Food Safety, University of Georgia, Griffin
SUSAN A. FERENC, SAF*RISK, LC, Madison, Wisconsin
NANCY F. KREBS, Department of Pediatrics, University of Colorado Health Sciences
Center, Denver
SHIRIKI KUMANYIKA, Center for Clinical Epidemiology and Biostatistics, University of
Pennsylvania School of Medicine, Philadelphia
REYNALDO MARTORELL, Department of International Health, Emory University,
Atlanta, Georgia
LYNN PARKER, Child Nutrition Programs and Nutrition Policy, Food Research and
Action Center, Washington, D.C.
BARBARA O. SCHNEEMAN, Department of Nutrition, University of California, Davis
NICHOLAS J. SCHORK, Polymorphism Research Laboratory, University of California,
San Diego
JOHN W. SUTTIE, Department of Biochemistry, University of Wisconsin, Madison
STEVE L. TAYLOR, Department of Food Science and Technology and Food Processing
Center, University of Nebraska, Lincoln
BARRY L. ZOUMAS, Department of Agricultural Economics and Rural Sociology,
Pennsylvania State University, University Park
Staff
LINDA D. MEYERS, Director
GAIL E. SPEARS, Staff Editor
GERALDINE KENNEDO, Administrative Assistant
ELISABETH RIMAUD, Financial Associate
vi

Reviewers
This report has been reviewed in draft form by individuals chosen for their diverse
perspectives and technical expertise, in accordance with procedures approved by the NRC’s
Report Review Committee. The purpose of this independent review is to provide candid and
critical comments that will assist the institution in making its published report as sound as
possible and to ensure that the report meets institutional standards for objectivity, evidence,
and responsiveness to the study charge. The review comments and draft manuscript remain
confidential to protect the integrity of the deliberative process. We wish to thank the follow-
ing individuals for their review of this report:
M. Tom Clandinin, MTI MetaTech, Inc.; Susan Ferenc, SAF*RISK, LC; Kenneth Fisher,
KD Consultants; Cutberto Garza, Cornell University; William C. Heird, Baylor College of
Medicine; Charlie Homer, National Initiative for Children’s Healthcare Quality; Nancy F.
Krebs, University of Colorado Health Sciences Center; Mary F. Locniskar, Hill’s Pet Nutri-
tion, Inc.; Michael Posner, University of Oregon; Catharine Ross, Pennsylvania State Uni-
versity; Steve L. Taylor, University of Nebraska, Lincoln; John E. Vanderveen, San Antonio,
Texas.
Although the reviewers listed above have provided many constructive comments and
suggestions, they were not asked to endorse the conclusions or recommendations nor did
they see the final draft of the report before its release. The review of this report was
overseen by Dennis M. Bier, USDA/ARS Children’s Nutrition Research Center at Baylor
College of Medicine. Appointed by the Institute of Medicine, he was responsible for
making certain that an independent examination of this report was carried out in accor-
dance with institutional procedures and that all review comments were carefully consid-
ered. Responsibility for the final content of this report rests entirely with the authoring
committees and the institution.
vii


Preface
Infant formulas are unique because they are the only source of nutrition for many
infants during the first 4 to 6 months of life. They are critical to infant health because they
must safely support growth and development during a period when the consequences of
inadequate nutrition are most severe. The safety standard for an ingredient added to infant
formula is defined as a “reasonable certainty of no harm.” In recent years infant formula
manufacturers have proposed ingredients (e.g., long-chain polyunsaturated fatty acids [LC-
PUFAs], probiotics) that have created new scientific challenges that existing regulations may
not address. Meanwhile the Food and Drug Administration (FDA) has been working to
revise several aspects of its infant formula regulations, including requirements for quality
factors and current good manufacturing practices.
In 2002 FDA and Health Canada asked the Institute of Medicine’s Food and Nutrition
Board (FNB) to formulate an expert committee to review and identify gaps in methods
currently used to assess the safety of ingredients new to infant formulas. The committee
was asked to identify tools to evaluate the safety of these ingredients under intended
conditions of use in term infants. This charge included determining the data needed to
demonstrate the safety of a component already present in human milk, that is, the effect of
the matrix, and the utilization of preclinical and clinical studies and postmarket monitoring
in the safety assessment process. FDA also asked the committee to apply its recommenda-
tions to the recent addition of LC-PUFAs as new ingredients in infant formulas, and others
as appropriate.
A 10-member committee was appointed with expertise in the areas of pediatric nutri-
tion, pediatric gastroenterology, epidemiology and public health, statistics, food technology,
food regulatory processes, pediatric neurology, biochemistry, and infant formula manufac-
turing. Members brought a diversity of experience from research laboratories, industry, and
hospital and clinic settings. Many of the committee members had never met, yet the group
developed a cohesion that allowed them to work through and agree upon several difficult
issues over the 18-month project.
The committee sought additional expertise and background from several consultants.
ix

x PREFACE
Two medical education consultants worked with the committee to develop several decision-
making tools (algorithms) that emerged as important components of the committee’s recom-
mendations. An editorial consultant helped the committee bridge the gap from rough draft
to final report to ensure uniformity of style and maximum readability for all possible users.
Another consultant provided targeted guidance in the areas of immunology and endocrinol-
ogy. In addition, the committee held two open sessions to learn more about updated systems
and information that was relevant to its charge. In one open session, the committee reviewed
differences that might apply to safety of foods as compared with drugs. In the other open
session, one specialist discussed the use of growth charts and another from FDA’s Center for
Food Safety and Applied Nutrition discussed the Adverse Event Reporting System.
The committee held six meetings and utilized frequent conference calls to develop this
report. The committee decided at an early stage that as a general approach, it would first
review current (and past) approaches to establishing the safety of ingredients new to infant
formulas. Next, gaps or limits of the current systems were to be identified, and then recom-
mendations were to be proposed to improve current approaches. Realizing that detailed
recommendations for every possible new ingredient would be impossible to achieve within
the framework of the committee’s work, generic templates were designed to be utilized (and
modified) to fit a large range of potentially new ingredients with varying characteristics and
different levels of safety concern.
The committee proposes the use of a hierarchical approach to assess the safety of
ingredients new to infant formulas. These hierarchies, which often utilize algorithms, assist
decision-making through a step-by-step process that places priority on commonly used
assessments (e.g., screening) and progresses to more specific assessments (e.g., neuroimaging)
when early indicators suggest safety concerns. The primary benefit of using algorithms is
that they provide an appropriate balance of information and flexibility to navigate through
a decision-making process without being prescriptive. This is especially useful given the wide
diversity in possible new ingredients.
On behalf of the committee, I am grateful to FNB’s staff for their support and contribu-
tions. This report was guided by three study directors: Sandra Schlicker laid the initial
foundation for the committee’s work, Paula Trumbo provided input through the middle
stages of deliberation, and Maria Oria assisted the committee in bringing the report to its
conclusion. The committee is grateful to Leslie Vogelsang, research assistant, and Sandra
Amamoo-Kakra, senior project assistant, for their support and dedication. The committee
would also like to thank Linda Meyers, director of FNB, for her objective insights and
invaluable expertise that informed committee deliberations and conclusions and Allison
Yates for her leadership during the early stages of the project. This report would not have
been possible without the leadership and dedication of the FNB staff.
Finally, I wish to thank each member of the committee for his or her unique contribu-
tions and unfailing dedication to a report that has the potential to improve the safety of
infant formulas for a new generation.
Richard J. Deckelbaum, Chair

Committee on the Evaluation of the Addition of
Ingredients New to Infant Formula

Contents
EXECUTIVE SUMMARY 1
1 INTRODUCTION AND BACKGROUND 17
Infant Formula Regulations and Guidelines, 18
Charge to the Committee, 20
The Committee’s Approach, 22
Organization of the Report, 27
Summary, 27
References, 27
2 DEFINING SAFETY FOR INFANTS 29
Abstract, 29
Introduction, 30
U.S. Regulatory Agencies, 30
Fundamental Concepts of Safety Regulations, 30
Statistical Considerations in Food Safety Determination, 31
Models of Safety Assessment, 33
Special Considerations for Infant Formulas, 36
Summary, 39
References, 40
3 COMPARING INFANT FORMULAS WITH HUMAN MILK 41
Abstract, 41
Background, 42
History of the Development of Infant Formulas, 42
Challenges of Matching Human-Milk Composition and
Breastfeeding Performance, 44
Performance Advantages of Breastfeeding, 46
Risks of Breastfeeding, 50
xi

xii CONTENTS
Summary, 51
References, 52
4 STRENGTHENING THE CURRENT PROCESS TO EVALUATE NEW
INGREDIENTS FOR INFANT FORMULAS 55
Abstract, 55
Introduction, 56
The Current U.S. Regulatory System, 56
Proposed Safety Assessment Process, 61
Summary, 69
References, 69
5 TESTING INGREDIENTS WITH PRECLINICAL STUDIES 70
Abstract, 70
Introduction, 71
The Importance of Appropriate Preclinical Studies, 71
Current Regulatory Guidelines for Preclinical Studies, 72
Overview of Recommended Levels of Assessment, 73
Conducting Preanimal Tests, 73
Conducting Animal Toxicity Studies, 77
Conducting Neurological Tests, 87
Summary, 94
References, 94
6 GOING BEYOND CURRENT CLINICAL STUDIES 98
Abstract, 98
Introduction, 99
The Importance of Clinical Studies, 99
Current Regulatory Guidelines for Clinical Studies, 100
General Approach to Conducting Clinical Studies, 101
Growth, 105
Specific Organ Systems, 113
Developmental-Behavioral Outcomes, 124
Summary, 144
References, 150
7 SELECTING AN IN-MARKET SURVEILLANCE PLAN 160
Abstract, 160
Background, 161
Criteria and Methods for In-Market Surveillance, 166
Summary, 172
References, 172

CONTENTS xiii
APPENDIXES
A Acronyms and Glossary 175
B Composition of Infant Formulas and Human Milk for Feeding Term
Infants in the United States 179
C Redbook Table of Contents 183
D Applying the Recommended Approaches 186
E Biographical Sketches of Committee Members 204


Executive Summary
OVERVIEW
Existing guidelines and regulations for evaluating the safety of conventional food ingre-
dients (e.g., vitamins and minerals) added to infant formulas have worked well in the past;
however they are not sufficient to address the diversity of potential new ingredients pro-
posed by manufacturers to develop formulas that mimic the perceived and potential benefits
of human milk. Proper nutrition, while important throughout life, is particularly critical
during infancy when growth and development are most rapid and when the consequences of
inadequate nutrition are most severe. Not all organ systems are fully mature at birth, and
many are highly susceptible to environmental inputs as they undergo further development.
Thus optimization of nutrition and minimization of exposure to potentially harmful sub-
stances in the food supply is of heightened importance during infancy.
Multiple health organizations endorse breastfeeding as the optimal form of nutrition for
human infants because of its potential advantages to the infant, including prevention of
infectious diseases and its role in neurodevelopment. Despite these recommendations, the
vast majority of infants worldwide are fed infant formulas (e.g., liquid or reconstituted
powders) at some point in their first year of life, whether as their sole source of nutrition or
in combination with human milk, supplemental foods, or both. Infant formulas have been
modified over the years to improve flavor, increase shelf life and, recently, to mirror the
composition of human milk and the performance of breastfeeding.
Existing guidelines and regulations to assess the safety of ingredients new to infant
formulas do not provide a clear and complete set of tools to address the new scientific
challenges created by the addition of new ingredients (e.g., probiotics and other complex
ingredients). For example, in the United States the Generally Recognized as Safe (GRAS)
Notification is the most common approach that manufacturers use when they seek to add a
new ingredient to infant formula. This process, while scientifically rigorous and transparent,
was developed to regulate food ingredients for the general population, not for infants who
are a more vulnerable population.

2 INFANT FORMULA: EVALUATING THE SAFETY OF NEW INGREDIENTS
This report, prepared at the request of the Food and Drug Administration (FDA) and
Health Canada (with potential international utility), addresses the regulatory and research
issues that are critical in assessing the safety of the addition of new ingredients to infant
formulas.
COMMITTEE CHARGE AND APPROACH

The Committee on the Evaluation of the Addition of Ingredients New to Infant For-
mula, convened by the Institute of Medicine, was asked to:
• review methods currently used to assess the safety of ingredients new to infant
formulas,
• identify gaps in current safety assessment guidelines, and
• identify tools to evaluate the safety of ingredients new to infant formulas under
intended conditions of use in term infants.
The committee was asked to focus on ingredients that are regulated under the food
provisions of the law and to consider the health and well-being of term infants from birth to
12 months of age. This charge included determining which new ingredients or classes of
ingredients are of lesser or greater concern, which additional data would be needed to
demonstrate the safety of a component already present in human milk when it is added to
the matrix of infant formula, the usefulness of certain safety tools and approaches, and the
utilization of preclinical and clinical studies and in-market monitoring. Finally, the commit-
tee was asked to apply its recommendations to long-chain polyunsaturated fatty acids (LC-
PUFAs), recently determined GRAS, as a new ingredient in infant formulas and to other
ingredients as appropriate.
The committee reviewed U.S., Canadian, and European laws and regulations to examine
current processes for manufacturers who wish to add new ingredients to infant formulas: a
GRAS Notification and a Food Additive Petition. The committee drew on this review,
especially the GRAS Notification process, as it developed its recommendations. The com-
mittee also reviewed the special needs of infants and their implications for evaluating the
safety of infant formulas.
The committee developed and used algorithms throughout the report to graphically
depict the overall process and recommends the use of stepwise decision-tree approaches for
the process and for preclinical studies, clinical studies, and in-market surveillance. Each
algorithm is a step-by-step decision tree that depicts the logic of a process but does not
denote a particular chronology. Algorithms provide a useful tool and a visual way to explain
the process of planning the type and depth of safety assessments; to improve data collection,
problem solving, and decision making; to incorporate multiple levels of information into
a single document; and to utilize a linear approach to identify critical information needed
at major decision points. The committee also applied its recommended approaches to
LC-PUFAs and to probiotics, recently determined GRAS, to examine the utility of the
approaches.
The committee recognizes that some of its recommendations may require statutory
changes. Even with this limitation, the committee encourages dialogue among members of
government agencies, the public, industry, and academia to act on the recommendations set
forth by this report in the best interest of our most vulnerable members of society—our
infants.

EXECUTIVE SUMMARY 3
FINDINGS AND RECOMMENDATIONS
Approach for Evaluating Safety

Safety evaluation of food ingredients in the United States is based on “reasonable
certainty of no harm.” Sections 201(s), 201(z), 409, and 412 of the Federal Food, Drug and
Cosmetic (FD&C) Act, associated regulations, and FDA’s Toxicological Principles for the
Safety Assessment of Direct Food Additives and Color Additives Used in Food, also known
as the Redbook,1 provide guidelines that are used to assess the safety of food ingredients and
infant formulas. The Canadian Food and Drugs Regulations, administered by Health
Canada, include specific requirements for infant food, novel food, and food ingredients.
In Canada and the United States, the food additive petition processes are similar and
require premarket review and approval. In the United States, under the proposed GRAS
Notification rule, a manufacturer can declare that a substance is GRAS if there is scientific
consensus among qualified experts about its safety under the conditions of intended use. The
manufacturer then notifies FDA, and if the agency has no questions, a letter of no objection
is issued.
A primary difference between the two routes is that the Food Additive Petition places
the responsibility of declaring that a substance is safe and approved under the conditions of
use with the regulatory agency, whereas the GRAS Notification process places the responsi-
bility of demonstrating that a substance is GRAS, and therefore safe under the conditions of
use, with the manufacturer. The GRAS Notification and Food Additive Petition procedures
are intended to ensure the safety (“a reasonable certainty of no harm”; FD&C Act Section
409), not the efficacy, of the proposed ingredient. The vast majority of new ingredients will
likely follow the GRAS process to establish safety.
In addition to the regulations, FDA provides preclinical guidelines in its Redbook. This
document was prepared to assist in the design of protocols for animal studies conducted to
test the safety of food ingredients and includes detailed guidelines for testing the effects of
food ingredients on mothers and their developing fetuses. However the special conditions
surrounding infancy require distinct procedures to ensure the safety of infant formulas.
The GRAS process is rigorous, flexible, credible, and transparent. However its applica-
tion does not clearly address possible concerns for the multitude of potential new ingredients
in infant formulas. New ingredients may possess a variety of chemical characteristics, nutri-
tional contributions, and pharmacological and physiological activities. Ingredients may be
derived from novel sources or processes (e.g., products of fermentation or biotechnology).
Such diversity requires safety guidelines that are clear but not overly prescriptive because of
the disparity in the issues that each class of ingredient to be added to infant formulas may
present. Because of its wide use, the committee used the GRAS Notification process as a
starting model in developing its proposed approach for assessing the safety of new ingredi-
ents added to infant formulas, without being prescriptive. Some elements of the system
proposed by the committee are currently in place.
Infant formula is the only source of nutrition for many infants and, therefore, additional
safeguards have been established for infant formulas to which new ingredients are added
that are not required for other foods. Regulations under Sections 201(s), 201 (z), and 412 of
the FD&C Act have been implemented to ensure the safety of an infant formula to be
marketed with a new ingredient. Under these sections of the Act, a proposed rule would
1This FDA document should not be confused with the American Academy of Pediatrics’ Red Book of childhood
infectious diseases.

mandate that manufacturers must demonstrate that the formula containing the new ingredi-
ent is capable of sustaining physical growth and development over 120 days when formula
is likely to be the sole source of infant nutrition. The committee believes that data from
growth and development studies should be submitted as part of the material demonstrating
safety.
Important Safety Considerations When Regulating Infant Formulas

Most of the principles that govern the safety of ingredients new to infant formulas derive
from the same principles that govern food safety for older children and adults. The commit-
tee concluded that the six issues listed below must be considered as important safety issues
when regulating infant formulas:
• Infant formulas are the sole or predominant source of nutrition for many infants.
• Formulas are fed during a sensitive period of development and may therefore have
short- and long-term consequences for infant health.
• Animals may not be the most appropriate model on which to base decisions of safety.
• “One size fits all” food safety models may not work for all new additions to formulas.
• Infant formulas could be considered as more than just food (i.e., as a delivery system
for non-nutritional agents).
• Potential benefits, along with safety, should be considered when adding a new ingre-
dient to formulas.
Use of a Hierarchical Approach

Since infants have distinct needs and vulnerabilities, a set of guidelines should be devel-
oped to provide a hierarchy of decision-making steps for manufacturers seeking to add new
ingredients to infant formulas. Because specific safety assessments will need to be targeted
according to the nature of the ingredient, the set of guidelines should allow for flexibility in
the approach, while being rigorous and scientifically based.
A hierarchical approach assists in determining the appropriate level of assessment by
considering: (1) the harm (e.g., toxicity), and (2) the potential adverse effects of a new
ingredient. This hierarchical approach will guide the level of assessments to be applied to the
new ingredient by considering the following factors:
• the reversibility of potential harmful effects,

• the severity and consequences of adverse effects,
• the time of onset of manifestations of the adverse effects,
• the likelihood that a new ingredient could adversely affect a specific system, and
• whether the effect would be common or rare.
This approach to evaluating the safety of new ingredients to be added to infant formulas
was based on the uniqueness and vulnerability of the infant population. Therefore each step
in the process requires empirical evidence from many disciplines and the application of the
highest standards, whether using methods of bioassay, nutritional analysis, or basic chemis-
try. This approach is valuable in determining the relative importance of potential adverse
effects for each specific new ingredient by providing generic templates for different steps in
the safety assessment process rather than specific recommendations for each compound. It is
neither realistic nor desirable to design individual templates for each new ingredient; rather,

EXECUTIVE SUMMARY 5
expert panels can refine the generic templates as needed. This approach is designed for a
broad spectrum of ingredients and could be applied to new ingredients to be added to infant
formulas regardless of the regulatory process used.
The hierarchical approach is graphically presented by algorithms throughout the report
and is applied in Appendix D to LC-PUFAs and probiotics. Each algorithm (see Figures ES-
1 through ES-7) is a step-by-step decision tree that depicts the logic of the process but does
not denote a particular chronology. For example, a manufacturer may initiate several differ-
ent studies and procedures at the onset of the process, the results of which could be assessed
at different steps in the algorithm. Any new ingredient considered for use in infant formulas
must be considered in the context of its form, the matrix, and other ingredients with which
it may interact.
Expert Panels
The committee recommends that manufacturers establish balanced, qualified expert
panels in consultation with the regulatory agency. The existing GRAS process requires
consensus by qualified experts to evaluate the safety of the ingredient under consideration;
this consensus is often reached through a panel of scientific experts. The current system does
not specify the composition of the panel, and manufacturers may be uncertain about the
selection of appropriate panel members.
The panel should have experts that will ask the right questions and form an opinion that
is robust and of the highest scientific integrity. The committee strongly recommends that the
expert panel include a physician experienced in clinical study assessments, preferably a
pediatrician. Guidelines for selecting a panel early in the process could improve the effi-
ciency and objectivity of the process. Each expert panel should be responsible for determin-
ing the requisite levels of preclinical and clinical studies and in-market surveillance needed to
ensure the safety of new ingredients by utilizing evidence-based approaches and high-quality
scientific data.
Additional Elements of a Safety Assessment

Elements of the safety assessments of infant formulas need to be standardized (e.g.,
toxicity studies, risk assessment). A scientifically rigorous set of guidelines must also allow
for flexibility in their approach since specific safety assessments must be targeted according
to the nature of the ingredient.
The committee recommends that bioavailability be specifically addressed in any evalua-
tion of the safety of infant formulas. Other factors that should be considered for safety are:
tolerance, allergenicity, impact of gastrointestinal flora, and possible nutrient imbalances (if
ratios or cofactors are important). While bioavailability is a factor considered under the
proposed infant formula regulations, other factors are also of special importance to infants
and should be specifically addressed in any evaluation of the safety of infant formulas.
The committee recommends that the manufacturer implement an appropriate in-market
surveillance strategy that is based on findings from preclinical and clinical studies and the
potential for harm to infants. Infant formula manufacturers routinely conduct passive sur-
veillance (e.g., contact with health care professionals), but guidelines do not exist to assist
manufacturers in determining the appropriate type and length of surveillance, from simple
methods (e.g., incoming toll-free calls) to rigorous methods (e.g., clinical follow-up of the
original study population).

PROPOSED PROCESSES
1
New ingredient proposed for infant
formula
2
Manufacturer establishes assessment process
to determine the safety of ingredients new to
infant formula
Sidebar A: Preclinical Studies

3 Characterize chemical and physical properties of the new
Review of literature ingredient. Conduct preclinical studies to evaluate toxicity
- history of safe use and neurological safety.
(See Chapter 5)
4
Preclinical Studies Sidebar B: Clinical Studies
(See Sidebar A)
Conduct clinical studies to assess symptoms and laboratory
indicators for specific organ systems, absorption and
metabolism, and developmental and behavioral outcomes.
5 (See Chapter 6)
Clinical Studies
(See Sidebar B )
Sidebar C: In-Market Surveillance
Establish ex pert panel to evaluate in-market monitoring,
6 Manufacturer selects an expert panel in review submitted evidence,surveillance data, and ongoing
consultation with regulatory agency to review literature reviews. Determine necessary follow-up studies.
results and determine safety of new
(See Chapter 7)
ingredient
7 Manufacturer submits to
regulatory agency its
demonstration of safety of new
ingredient 9
Manufacturer provides
answers to questions
8 Regulatory agency raises

questions concerning the
Y es
scientific results, process, or
other components?
No
10 Regulatory agency has 11

no objection Ensure safety of formula
Yes
concerning the safety containing new ingredient
of new ingredient
No
No
15 14 REGULATORY AGENCY
12 REGULATORY AGENCY
DISCONTINUE DOES NOT APPROVE
APPROVES INFANT
PROCESS INFANT FORMULA WITH
NEW INGREDIENT FORMULA WITH NEW
INGREDIENT
13
In-Market Surv eillance
(See Sidebar C)
FIGURE ES-1 Proposed process for evaluating the safety of ingredients new to infant formulas
algorithm. In-market assessment should be planned in conjunction with preclinical and clinical test-
ing. This algorithm is modeled after the U.S. Generally Recognized as Safe Notification process;
similar schemes can be adapted to other regulatory processes. = a state or condition,
= a decision point, = an action, sidebar = an elaboration of recommendation or
statement.

EXECUTIVE SUMMARY 7
Preclinical Studies for Evaluating Safety

Preclinical studies are a vital first step to assess the safety and quality of ingredients new to
infant formulas. Guidelines and regulations should consider the special needs and vulnerabilities
of infants. Also, the diversity in the ingredients and issues requires clear but flexible preclinical
guidelines. For example, a complex mixture or ingredients derived from novel sources or pro-
cesses present unique safety issues. Also, guidelines should provide the most appropriate animal
models at relevant developmental stages for testing infant formulas. For example, the most
commonly used animal models for general toxicological studies are the rat and mouse, which are
of limited use for developmental studies because of the difficulty of feeding formula to a
preweanling rodent. The nonhuman primate and the piglet are more amenable for these types of
studies because they readily accept infant formula as a nutrient source.
Current guidelines, provided in the FDA Redbook, and regulations do not address the
special needs and vulnerabilities of infants or the diversity of potential ingredients.
The committee recommends the use of a two-level assessment approach to determine
the potential toxicity of the ingredient, its metabolites, and their effects in the matrix on
developing organ systems (see Figure ES-3, Sidebars A and B). Level 1 assessments include
standard measures for each organ system (gastrointestinal, blood, kidney, immune, endo-
crine, and brain). Level 2 assessments include in-depth measures of organ systems that
would be used to explicate equivocal level 1 findings or specific theoretical concerns not
typically addressed by level 1 tests. Preclinical assessments range from cellular-molecular
through whole animal studies (see Figure ES-2, Sidebars A and B).
The committee recommends that a distinct set of procedures that use an appropriate
number and type of animal model at relevant developmental stages be included in preclinical
studies. At least two animal models should be selected, and justification for the studies, as
well as limitations in extrapolating to humans, should be clear, especially regarding the
comparative development between the animal model and the human. The appropriate spe-
cies, age, and safety factors, as well as other factors, such as the timing, bioavailability of the
ingredient, and differences in pharmacokinetics and dietary imbalances, should also be
considered.
Clinical Tests for Evaluating Safety

Human clinical studies in infants are a vital second step in the safety assessment process
after preclinical studies have been satisfactorily completed. Clinical studies can predict how
a new infant formula may affect growth and development, organ systems, and developmental-
behavioral outcomes. Clinical studies in human infants are needed for several reasons. First,
extrapolation from animal studies may be limited by differences between animal and human
structure, physiology, and development. Second, extrapolation from isolated tissue studies is
limited by the inability of such models to assess functions in the context of whole organ
systems where coordination and integration are the rule. Third, there may be no available
animal or tissue model to test specific conditions or functions (e.g., tests for allergenicity or
higher cognitive functions found only in humans).
There are no explicit requirements for clinical testing of ingredients new to infant
formula specified under Canadian or U.S. regulations. However FDA recommends that the
studies conform to guidelines presented in section VI.A. of the Redbook. In addition, in the
proposed rule under Section 412 of the FD&C Act, FDA proposes that researchers use the
120-day growth study as the main method to assess the ability of an infant formula to
sustain normal infant growth.

PROPOSED PRECLINICAL ASSESSM ENT

1
New ingredient proposed for
infant formula
2
Are the active component or the
impurities well known ?
AND Yes
Is adequate literature available to
determine their safety?
Sidebar A: Chemical and Physical
Characterization
3 Structure, Stability, and Solubility

Initiate preclinical studies to evaluate toxicity - High performance liquid chromatography,
No liquid chromatography-mass spectrometry, and
and neurological safety (See Figure ES-3)
thin layer chromatography.
- The stability to temperature, ultraviolet light as
well as the solubility properties of the ingredient.
8 - The percentage of the unidentifiable materials
in ingredient.
Chemical and Physical Characterization
- The kind of so lvents,suspending agents,
(See Sidebar A)
emulsifiers, or other materials that will be used
in administering the ingredient whether in
in vitro or animal studies.
- Ingredient should be stored under condit ions
9 4 that maintain its stability, quality, and purity
Are chemical and Toxicity Assessment until the subsequent studies are complete.
Yes - These studies should be repeated with the
physical purity assured? (See Sidebar B)
ingredient in the solution or matrix that would
be fed to the human infants.
Sidebar B: Toxicity Assessment
1. Genetic tests
2. Cellular studies
3. Animal toxicity studies
- Acute,subchr onic, and chro nic
toxicity
- Developmental toxicity
- Absorption, distribution,
metabolism,,and excretion
4. Organ level studies
- Gastrointestinal tract
No
- Hepatic
- Renal
- Hematology
- Immune
- Endocr ine
- Neurolo gic
5
Any positive test for
toxicity or concern for Yes
safety?
No
6
10 7
DISCONTINUE Neurological Safety Assessment DISCONTINUE
PROCESS (See Figure ES-3) PROCESS
FIGURE ES-2 Proposed preclinical studies algorithm. = a state or condition, =

a decision point, = an action, sidebar = an elaboration of recommendation or statement.

EXECUTIVE SUMMARY 9
P
PROPOSED LEVELS OF PRECLINICAL ASSESSMENT
1
Initiate preclinical studies to evaluate

toxicity and neurological safety
2
Any evidence of abnormalities
based on previous human data,
other preclinical studies, or on Yes
theoretical plausible perturbation
of a metabolic pathway?
No
Sidebar A: Level 1 Assessment
7
Standard measures of genetic tests,
Lev el 1 Assessment
cellular studies, animal toxicity studies,
(See Sidebar A)
major organ systems, and neurological
preclinical screening measures.
Sidebar B: Level 2 Assessment

8 3 Detailed measures of genetic tests,
Any evidence of
Lev el 2 Assessment cellular studies, animal toxicity studies,
adverse Yes
(See Sidebar B) major organ systems, and neurological
effect/event?
preclinical measures.
No
4
Any evidence of adverse
effect/event or concern Yes
for safety?
No
9 6 5
Continue to clinical Re-evaluate results before DISCONTINUE
studies considering clinical trials PROCESS
FIGURE ES-3 Proposed levels of assessment for preclinical studies algorithm. = a state or
condition, = a decision point, = an action, sidebar = an elaboration of recom-
mendation or statement.

Growth
The committee recommends that growth studies should continue to be a centerpiece of
clinical evaluation of infant formulas and should include precise and reliable measurements
of weight and length velocity, and head circumference. Appropriate measures of body
composition should also be assessed (see Figure ES-4, Sidebar A). These measures help
researchers understand the impact of an ingredient new to infant formulas. For example,
weight is responsive to acute insults, such as infectious morbidity or changes in nutrient
intakes, and recumbent length is an overall indicator of linear or bone growth.
The committee recommends that clinical growth studies follow the study participants
for the entire period when infant formula remains a substantial source of nutrients in the diet
of the infant. The committee believes that a 120-day growth study, in the 1996 FDA
proposed rule, may be insufficient for several reasons. Currently human milk is recom-
mended as the sole nutrient source for infants ages 4 to 6 months; infant formula, intended
as a human milk substitute, is recommended for the same period of time. Ideally formula
should be tested for the entire period for which it is intended to be fed as the sole source of
infant nutrition (up to 6 months, or roughly 180 days, consistent with breastfeeding guide-
lines) rather than the currently proposed 120-day period. An additional and more serious
limitation of the 120-day growth study is that it does not allow for the determination of
delayed effects or for understanding longer-term effects of early perturbations in growth.
The committee recommends the development of specific guidelines that define normal
growth and establish a level of difference that represents a safety concern. Specifically, the
committee recommends that any addition of an ingredient new to infant formulas should be
judged against two controls: the previous iteration of the formula without the added ingre-
dient and human milk. The proposed rule does not define “normal” growth, nor does it
identify what represents a biologically meaningful difference among groups of infants con-
suming different formulas. The committee recognizes that there is very little scientific evi-
dence to establish a level of difference associated with long- or short-term health conse-
quences. However the committee concluded that any systematic and statistically significant
difference in size or growth rate among infants fed a formula with the new ingredient versus
human milk or an already approved formula should be a safety concern.
Organ Systems, Neurobiological Development, and Behavior

In addition to growth studies, the committee recommends a two-level assessment ap-
proach to assess organ, immunological, and endocrinological systems (see Figure ES-5,
Sidebars A and B). Level 1 assessments include standard measures for each organ system
(gastrointestinal, blood, kidney, immune, endocrine). Level 2 assessments include in-depth
measures of organ systems that would be used to explicate equivocal level 1 findings or
specific theoretical concerns not typically addressed by level 1 tests.
These major organ systems must be studied because growth deficits are likely to appear
only secondary to effects on specific organs or tissues and may not appear for some time
after nutritional insult. For example, slow or inadequate growth is a common denominator
of impaired gastrointestinal function, but it does not identify the function that is impaired.
In addition, some organ systems (e.g., immune and endocrine) are immature at birth; every
effort must be made to ensure that ingredients new to infant formulas will not affect the
development of these systems or the expression of their function.
The committee recommends a three-level assessment approach to assess developmental-
behavioral outcomes, including sensory-motor, cognitive development, temperament, and

EXECUTIVE SUMMARY 11
CPROPOSED CLINICAL ASSESSMENT

1
formula
Initiate clinical studies to evaluate the

impact of new ingredient on growth and
development
Sidebar A: Growth Studies
Assess:
- Weight velocity
3 - Length velocity
Grow th Studies - Head circumference
(See Sidebar A) - Body composition
and
Clinical Endpoints
(See Sidebar B and Figure ES-5)
Sidebar B: Clinical Endpoints
Assess symptoms and adverse
laboratory indicators in the following:
- Kidney
- Blood
- Immunological system
4
- Endocrinological system
Abnormal growth or
Assess absorption, distribution,
adverse effect/event on metabolism, and excretion of
Yes
specific organ, immune, or
ingredient where appropriate
endocrine systems
Sidebar C:
No Developmental-Behavioral
Assessment
6
Assess:
- Sensory and motor function
Dev elopmental-Behav ioral Assessment
- Cognitive development
(See Sidebar C and Figure ES-6)
- T emperament
- Neurological function
7 5
Abnormal function in major DISCONTINUE

Yes
developmental areas PROCESS
No
8
MANUFACT URER/REGULAT ORY
AGENCY DET ERMINES INGREDIENT
IS SAFE
FIGURE ES-4 Proposed clinical studies algorithm. = a state or condition, =a

decision point, = an action, sidebar = an elaboration of recommendation or statement.

PROPOSED LEVELS OF CLINICAL ASSESSM ENT OF MAJOR SYSTEMS

1
form ul a
Known or theoretical i ndi rect

Y es
li nk to major organ system s?
No Major organ systems screening

6 measures in the following:
Adverse effect/event documented - Liver
in precli ni cal trials? - Kidney
OR - Blood
Y es
Evi dence of signifi cant individual - Immune
difference i n susceptibil ity to the - Endocrine
ingredient?
No
7 Major organ systems detail ed m easures

Lev el 1 Assessment in the foll owi ng:
(See Sidebar A) - Gastrointestinal tract
- Liver
- Kidney
- Blood
8
- Immune
Evi dence of - Endocrine
adverse Yes
effect/event?
3
Lev el 2 Assessment
(See Sidebar B)
No 4
Evi dence of
adverse Yes
effect/event?
No
9 5
Continue to
DISCONTINUE
neurobehavioral clincal
PROCESS
studi es
FIGURE ES-5 Proposed levels of assessment for clinical studies of major organ, immune, and endo-
crine systems algorithm. = a state or condition, = a decision point, =
an action, sidebar = an elaboration of recommendation or statement.

neural functions, with appropriate measurement instruments and study design features (see
Figure ES-6, Sidebars B, C, and D). Level 1 assessments include developmental screening
measures. Level 2 assessments include in-depth measures of child functions in major devel-
opmental areas (single assessment with one instrument). Level 3 assessments include in-
depth measures using repeated assessments with multiple instruments.
These developmental-behavioral outcomes are important for the following reasons:
• Behavioral outcome measures are sensitive to exposure to toxic substances.

• Developmental-behavioral measures can have long-term predictive value.
• Bidirectional brain-behavior links exist (e.g., brain development mediates changes in
behavioral competence, but the child’s interactions with his or her environment also can
influence brain development).
The proper application of developmental-behavioral studies requires the use of the most
appropriate measurement instruments and study design features to assess sensory and motor
functions, cognitive development, infant temperament, and neurological function.
In-Market Surveillance to Detect Adverse Effects

Although satisfactory completion of the appropriate preclinical and clinical studies
diminishes the likelihood of systematic adverse reactions, the risks for adverse reactions
cannot be ignored. Adverse effects may not be detected in preclinical studies if the wrong
animal model was chosen, if the assessment instrument chosen measured a function other
than the one adversely affected by the new ingredient, or if a subpopulation of individuals
who are highly sensitive to the new ingredient added to infant formulas was not sufficiently
represented in clinical studies. Also, brain areas that are adversely affected by a new ingredi-
ent may not become functionally apparent until later in development. Therefore in-market
surveillance is needed to ensure safety and normal development of the infant population.
The committee recommends that all submissions seeking to add a new ingredient to
infant formula include a systematic plan for continued in-market monitoring and long-term
surveillance (see Figure ES-7). Formal regulatory guidelines for in-market surveillance do
not exist for infant formulas. Surveillance is generally limited to consumer reporting of
adverse events through toll-free numbers or Internet sites established by the manufacturer or
the regulatory agency. There are a number of reasons why this approach is inadequate. One
reason is the risk of underestimating actual negative occurrences given that not all caregivers
will report a problem. Furthermore, caretakers will be less likely to link a child’s problems to
earlier intake of infant formula.
The committee, however, recognizes that there are methodological factors (e.g., length
of follow-up, confounding factors, lack of statistical power), practical concerns (e.g., conti-
nuity of the research team, tracking of subjects, record retention), and cost considerations
that limit the implementation of certain in-market surveillance programs.
The committee recommends a three-level assessment approach to determine appropriate
in-market surveillance strategies. Level 1 assessments include monitoring the toll-free line or
Internet website (passive surveillance). Level 2 assessments include in-market panels to
review existing data (both published and proprietary). The same selection and composition
recommendations presented earlier also hold with regard to in-market panels. Level 3 assess-
ments include conducting retrospective and/or follow-up studies (active surveillance).
The committee recommends that an expert panel determine the level and strategies to be
utilized based on conditions under which potential adverse effects of a new ingredient added

PROPOSED LEVELS OF CLINICAL ASSESSM ENT OF DEVELOPM ENT AND BEHAVIOR

1

formula Sidebar A: Developmental-Behavioral
Assessment
2 Cri teria for choosing neural -behavioral

assessment measures:
Known or theoreti cal li nk to - Age appropri ateness
Yes
neurobehavior? - Predictive value
- Sensi tivi ty
- Brai n-behavior l inks
No - Cross-species general izabil ity
- Function speci ficity
7 - Ease of ad ministration
Adverse effect/event
Study design requirements:
documented in preclinical
- Adequate stati stical power
trial s?
- Avoi d over-control of mediator variabl es
OR Yes - Use measurement aggregation
Evidence of significant
- Use repeated measures
indi vidual difference in
susceptibility to the
ingredient?
No
11
Neural and behavioral screening
Known or theoreti cal measures admi nistered duri ng a routine
indi rect l ink to other Yes well-baby physical exam or through
organ systems? parent reports.
No
12
Sidebar C: Level 2 Assessment
Lev el 1 Assessment Detai led measures of function in maj or

(See Sidebars A and B) child developmental areas. Si ngle
assessment for each area using one
instrument.
13 8 Sidebar D: Level 3 Assessment

Evidence of
Lev el 2 Assessment
adverse Yes Detai led measures of function in maj or
(See Sidebars A and C)
effect/event? child developmental areas on at least two
separate occasions using two
recommended instruments for each area.
9 3
Evidence of
Lev el 3 Assessment
adverse Yes
(See Sidebars A and D)
effect/event?
No
4
Evidence of
adverse Yes
effect/event?
No
No
14 10 6 5
MANUFACTURER/REGU
MANUFACTURER/REGULATORY
DISCONTINUE LATORY AGENCY DISCONTINUE
AGENCY DETERMINES
PROCESS DETERMINES PROCESS
INGREDIENT IS SAFE
INGREDIENT IS SAFE
FIGURE ES-6 Proposed levels of assessment for clinical studies of development and behavior algo-
rithm. = a state or condition, = a decision point, = an action, sidebar
= an elaboration of recommendations or statement.

PROPOSED IN-MARKET SURVEILLANCE

1
Marketed formula with new Sidebar A: Indication for Level 3
ingredient Assessment
Lev el 3 assessment should be considered if there

is potential f or harm. For example:
2 1. When action of new ingredient could af f ect:
Did any of the follow ing exist prior to -Slower developing brain regions,
marketing: -Endocrine or neurotransmitter action, and
- Evidence for signi ficant individual -Ea rly behavior t hat can impact the q ual i ty of
or pop ulatio n differences in ongoing parent-child relations.
2. Primary care phy sicians identif y changes in
susceptib ility to the new ingredient?
the in d ividual's growth or other outcomes,
- Adverse effect/event reported in Yes such as changes in weight, height, and head
prec li n i c al or cli n i c al tri als? circumference percentiles; skin or hair
- Scienti fic evi dence linki ng the new changes; muscle atr ophy; mood changes;
ingre dient or new ingredient source to anorexia; vomiting; and diarrhea.
development of any of the areas of
infant func tion (See Chapter 6) ?
The f ollowing issues should be considered in lev el 3
No assessment:
- The domains to be inv estigated and the
9 instrume nts to be used wil l vary depending on
Lev el 1 Assessment:
PASSIVE SURVEILLANCE the organ or functional systems that are most
like ly to be affecte d by the ingredient and
1-800 li ne or Internet web si te result s from in-market, preclinical, or clinical
studies.
- Assessment should include times when children
10 make major life transitions, such as ent ry into
Surveill ance data indicates problems school, the onset of pube rty, high-school
in a function area beyond graduation, and vocational cho ice. Ev en further
follow-up in adult life m ay be desirable with
expectati on based upon previous Yes
tr ansitions points, such as post-high-school
survey, cl ini cal studi es, or population
education and vocation status.
base rates ? - The location of those inf ants participating in the
clinical trials should be tracke d after the trials
are over.
No
3
11 Lev el 2 Assessment:
Convene an expert panel (in consultation with
Any adverse effect/event
the regulatory agency) to review existing
reported in l iterature Yes
published or proprietary data, submitted
published after marketing ?
evidence, survei llance data, and ongoing
li terature revi ews
Is there harm linked to marketed

YYes
formula with new ingredient?
Formula is pulled from

No No market
Is there potential for harm?

No
(See Sidebar A)
Yes
8
Expert panel makes
7
recommendations to
Lev el 3 Assessment: regulatory agency about
12 Initi ate studies at level and type needed long-term safety status of
Conti nue surveil lance to establi sh safety of formula formula with new ingredient
(See Sidebar B) and whether further long-term
follow-up is needed
FIGURE ES-7 Proposed in-market surveillance algorithm. = a state or condition,

statement.

to infant formulas might have been missed in preclinical or clinical trials involving the
ingredient.
In-market monitoring information must be assessed for each area of function reviewed,
as appropriate. Level 1 assessments are recommended only when all the following condi-
tions occur:
• There is no evidence of adverse effects in preclinical or clinical studies, including

adverse effects with potentially plausible alternative explanations (e.g., the effects are viewed
as the result of random chance or the reviewers believe that there may be methodological or
statistical problems in the studies).
• A review of the relevant scientific literature indicates there is no link between the new
ingredient, metabolites, secondary effectors, or source and development of any of the areas
of infant function previously described.
Level 2 assessments are required when any one of the above conditions does not occur
or when level 1 assessments indicate unexpected problems in a function area, based on
previous surveys, clinical information, or population-based rates.
In-market follow-up information must be assessed for each area of function reviewed, as
appropriate. Level 2 assessments are recommended when any one of the following condi-
tions occurs:
• A review of the relevant scientific literature indicates that there is existing evidence
linking the new ingredient, metabolites, secondary effectors, or source to the growth and
development of organ systems that could result in cumulative adverse effects over time.
• There is evidence of adverse effects in preclinical or clinical studies, including adverse
effects with potentially plausible alternative explanations.
• In-market monitoring reveals any adverse effects reported for the new ingredient,
metabolites, secondary effectors, or source.
Level 3 assessments for in-market follow-up are required when any one of the above
conditions occurs and level 2 assessments of in-market follow-up (review by the expert
panel) indicate potential for harm in a function area.
CONCLUDING REMARKS
This report describes the critical need to ensure the safety of infant formulas resulting
from a number of converging issues:
• Infancy is a uniquely vulnerable period of life.

• Infant formulas are consumed by the vast majority of infants and are the sole source
of nutrition for a large segment of infants up to the first 6 months of life.
• Manufacturers are increasingly interested in adding new ingredients to formulas in an
attempt to mimic the perceived and potential benefits of human milk.
• Existing guidelines and safety regulations lack clarity and completeness in adequately
addressing the unique growth and development requirements of infants and the vast diver-
sity of potential new ingredients.
The committee is confident that this report will provide regulatory agencies—FDA,
Health Canada, and others—with the recommendations, tools, and resources required to
improve guidelines to ensure the safety of infant formulas for generations to come.

Introduction and Background
Infant formulas are liquids or reconstituted powders fed to infants and young children
to serve as substitutes for human milk. Infant formulas have a special role in the diet because
they are the only source of nutrients for some infants. In the United States and other
industrialized countries, the vast majority of infants receive infant formula at some time
during their first year of life (Hediger et al., 2000; Ryan et al., 2002) as the number of infants
breastfed after birth rapidly decreases (Figure 1-1). Many infants receive formula in combi-
nation with breastfeeding. During these mixed feeding routines there are potential interac-
tions between the components of human milk and those contained in formulas.
Over recent decades ingredients have been added to infant formulas not only to better
simulate the composition of human milk, but also to impart health benefits. Examples
include fortifying formulas with iron, adding nucleotides, and changing the composition of
fat blends. Recently infant formulas containing added sources of arachidonic acid and
docosahexenoic acid have been made available in the United States, Europe, and elsewhere.
In the United States, new ingredients, such as probiotics and compounds produced by
genetic engineering, are currently being considered for addition to formulas.
Infancy is a uniquely vulnerable period of rapid growth and development and as such,
feeding changes have the potential to impart benefit or harm in the short term, into early
childhood, and even later into adulthood. Thus measurements of safety parameters during
infancy need to be equally or even more stringent than at other periods during the life cycle.
The introduction of new ingredients to formulas must pose no or minimal risk to infants. In
the United States and worldwide there is a paucity of guidelines or recommendations from
national and international organizations regarding approaches to assess the safety of ingre-
dients added to infant formulas.
The Committee on the Evaluation of the Addition of Ingredients New to Infant Formula
was convened by the Institute of Medicine at the request of the Food and Drug Administra-
tion (FDA) and Health Canada to review methods currently used to assess new ingredients
to be added to infant formulas, including preclinical and clinical studies and in-market
monitoring, and to identify gaps in current safety regulations and guidelines. The committee
17

50
45
40
Infants (%)
35
30
25
20
15
10
5
0
Birth 4 5 6 7 8 9 10 11 12
Age (months)
FIGURE 1-1 Percent of infants who were exclusively breastfed from birth to 12 months of age in the
United States.
SOURCE: Adapted from Hediger et al. (2000) and Ryan et al. (2002).
met six times over 18 months to fulfill its charge. This report is intended to provide recom-
mendations for regulatory bodies, industry, and basic and clinical investigators involved in
determining the safety of new ingredients added to infant formulas.
INFANT FORMULA REGULATIONS AND GUIDELINES
United States
Food products that are designed and marketed for infants are regulated under the
Federal Food, Drug and Cosmetic (FD&C) Act of 1938 (21 U.S.C. §301) (Vanderveen,
1991). FDA, an agency in the U.S. Department of Health and Human Services (HHS),
regulates infant formulas and evaluates the safety of food and color additives.
Two sections of the FD&C Act, 409 and 412, are the primary laws that relate to infant
formulas. Section 409 gives authority to HHS to ensure the safety of new food ingredients
(e.g., food additives and Generally Recognized as Safe [GRAS] substances) “under the
conditions of its intended use” (e.g., in infant formulas). Manufacturers may propose the
addition of new ingredients to infant formulas in the United States by either filing a Food
Additive Petition with FDA to request a formal premarket review, or making a GRAS
determination. It is important to point out that it is the use of the substance, rather than the
substance itself, that is eligible for the GRAS determination (see Chapter 4).
Under Section 412 of the FD&C Act, regulations have been promulgated and ultimately
implemented that are intended to ensure proper formulation for infants to thrive. They
include:
• Infant Formula Quality Control Procedures (21 C.F.R. §106),

• Records and Reports Regulations (21 C.F.R. §106.100),
• Infant Formula Labeling Requirements (21 C.F.R. §107.10–107.30),

INTRODUCTION AND BACKGROUND 19
• Exempt Infant Formulas (21 C.F.R. §107.50),

• Nutrient Requirements for Infant Formulas (21 C.F.R. §107.100), and
• Infant Formula Recall Requirements (21 C.F.R. §107.200–107.280).
FDA (1996) has proposed to revise these regulations to establish quality factors, current
good manufacturing practices, and revised quality control procedures. Table 1-1 lists some
of the U.S., Canadian, and European Union laws and regulations related to adding new
ingredients to infant formulas.
Other Countries
Canada
The Canadian Food and Drug Regulations (Canada, 2001) include specific require-
ments for infant formulas, novel foods,1 and other ingredients. Division 25 of the Regula-
tions provides for the addition to infant formulas of nutritive substances, in addition to
specified vitamins and mineral nutrients, found in human milk, provided the nutritive sub-
stance is added to the formulas to the level found in human milk (section B.25.056). The
Regulations include an inclusive list of those food additives that may be added to infant
formulas (section B.25.062). Similar to the process in the United States, a new food additive
must undergo premarket approval. The manufacturer must submit a request to the Minister
of Health that includes all the details laid out in Division 16 of the Regulations. The request
must include “detailed reports of tests made to establish the safety of the food additive under
the conditions of use recommended” and “data establishing that the food additive will have
the intended physical or other technical effect” (section B.16.002). Health Canada reviews
the request and, if accepted, the Minister of Health recommends to the Governor-in-Council
that the ingredient be added to the list of food additives in the Food and Drug Regulations.
The Food and Drug Regulations also require a premarket notification for novel foods. The
manufacturer of a novel food must submit a premarket notification for the food as required
under Division 28 of the Regulations. Health Canada issues a written notice to the manufac-
turer if it is satisfied that information submitted establishes that the novel food is safe for
consumption.
European Union
The European Union also has regulations in place for food additives, novel foods, and
genetically modified organisms. However the European regulations are not specific for
adding new ingredients to infant formulas.
1As defined by the Food and Drug Regulations, a novel food is “a substance, including a microorganism, that
does not have a history of safe use as a food; a food that has been manufactured, prepared, preserved or packaged
by a process that (i) has not been previously applied to that food, and (ii) causes the food to undergo a major
change; and a food that is derived from a plant, animal or microorganism that has been genetically modified (i) the
plant, animal or microorganism exhibits characteristics that were not previously observed in that plant, animal or
microorganism, (ii) the plant, animal or microorganism no longer exhibits characteristics that were previously
observed in that plant, animal or microorganism, or (iii) one or more characteristics of the plant, animal or
microorganism no longer fall within the anticipated range for that plant, animal or microorganism” (Canada,
2001, B.28.001).

TABLE 1-1 Selected Laws and Regulations Related to New Ingredients Added to Infant
Formulas
Country or
Organization Regulations (Reference) Identifying Terms
United States Federal Food, Drug, and Cosmetic Act Food additive, Substances Generally
(21 U.S.C. §301) Recognized as Safe (GRAS)
Food Additive Food Additive
(21 C.F.R. §170, 171, and 172)
GRAS (21 C.F.R. §182, 184, and 186) GRAS
Canada Food and Drugs Regulations Food additive, novel food
(Canada, 2001)
European Union European Communities (Infant Formulae and Established by generally accepted
Follow-on Formulae) Regulations, 1998 scientific data
(FSAI, 1999)
Council Directive 89/107/EEC (EEC, 1989) Food additive
European Parliament and Council Food additive
Directive 94/34/EC (EEC, 1994)
Regulation EC No. 258/97 (EC, 1997) Novel foods, novel food ingredients
Council Directive 90/220/EEC (EEC, 1990) Genetically modified organisms
In June 2002 representatives of academia, the infant food industry, the European Com-
mission, food regulatory bodies of some European Union member states, and consumer
organizations met to discuss opportunities to evaluate the safety of ingredients new to infant
formulas (Koletzko et al., 2002). Topics included consumer expectations, preclinical and
clinical evaluations, investigations of infant growth and nutrient bioavailability, and in-
market surveillance. The participants also discussed a position paper written by members of
the European Society for Paediatric Gastroenterology, Hepatology and Nutrition’s Commit-
tee on Nutrition (Aggett et al., 2001). The paper provided 13 points for consideration for the
evaluation of infant formulas; these points are presented in Box 1-1.
CHARGE TO THE COMMITTEE

FDA and Health Canada asked the committee to review methods currently used to
assess new ingredients to be added to infant formulas, including preclinical and clinical
studies and in-market monitoring, and to identify gaps in current safety regulations and
guidelines. As part of its task, the committee was requested to provide recommendations
to strengthen the scientific approaches used in assessing the safety of ingredients added to
infant formulas. The committee was asked to focus on:
• ingredients new to infant formulas that are regulated under the food provisions of the
law, not as drugs or therapeutic agents;
• the health and well-being of healthy term infants (delivered between gestational age
of 37 to 42 weeks with birth weights of 2.5 kg or more) from birth to 12 months of age; and
• the effects that ingredients new to infant formulas could have on metabolism, physi-
ology, neurological function, and normal growth and development, as well as on specific
systems, including the immunological, renal, hepatic, hematological, and gastrointestinal
systems.
The committee was asked to address the composition of human milk and to evaluate
how the presence or absence of a substance in human milk may be factored into the safety

BOX 1-1 Key Points of the European Society for Paediatric

Gastroenterology, Hepatology, and Nutrition’s Committee on
Nutrition Commentary on the Nutritional and Safety Assessment
of Breast Milk Substitutes and Other Dairy Products for Infants
1. Although human milk composition can be a guide to that of breast milk substitutes, the com-
parison of outcomes with those seen in healthy infants who have been exclusively breast fed for 4–6
months is considered a better approach.
2. Appropriate clinical studies of nutritional and safety assessment should be performed par-
ticularly for components, and combinations of components, which have not been previously included in
infant formulas and other dietary products for infants. Technological as well as compositional modifica-
tions to infant formulas should be assessed nutritionally.
3. The introduction of any modification to a formula or other dietary product for infants should be
based on a systematic review of the relevant existing information to develop a clear hypothesis of the
expected functional and clinical benefits. These reviews should be published or be made publicly avail-
able in other forms. Studies should be designed primarily to test these hypotheses, as well as making
general nutritional assessments.
4. Infant formulas or other products modified for reasons other than to provide a novel functional
or clinical benefit, or which are based on products already on the market, should, at least, be subjected
to studies of acceptability, and of nutritional equivalence to the existing products.
5. All infants in clinical trials should be characterized with regard to factors which might affect the
planned outcomes. Blind randomization with respect to the allocation of test and reference formulas is
important, and all studies should comply with Good Clinical and Good Laboratory Practices.
6. For all clinical trials on nutritional products, ethical approval should be acquired, informed pa-
rental consent obtained and this should be declared in the publication of results.
7. Modifications of infant formulas and other dietetic products for infants need to be evaluated for
their safety. It is important that the possibility of unexpected adverse outcomes be addressed by ad-
equate clinical monitoring of participants, and by incorporating, into the study design, arrangements for
the independent scrutiny of the accumulating data.
8. The general principles, design, execution and the data analysis of evaluative studies of infant
formulas and other substitutes for breast milk need to be determined to detect relevant short- and long-
term (i.e., in later childhood and adult life) outcomes. The design should consider from the outset the
statistical power of the study, and the confidence limits of any differences found should be included in
the published reports.
9. Preliminary pilot studies of the proposed study design are often useful to identify and anticipate
outcomes and issues which would inform definitive studies and enable protocols to be adapted and
would enable the views of the infants’ carers to be taken into account. This approach would be ex-
pected to enhance the co-operation of carers and the quality of the methodology of the subsequent
definitive assessment.
10. Manufactures and scientific, academic, and professional groups should collaborate to the
extent of agreeing on an essential portfolio of data and outcomes, which should be recorded in all
nutritional studies performed during early life. This would enable the later consolidation of information
from individual studies into larger databases which would be appropriate for the assessment of long-
term nutritional efficacy and safety, as well as being able to detect unanticipated outcomes of early
feeding practices and dietary exposure.
11. A register of current trials of infant formulas should be established, and wherever possible,
this information should be accessible to manufacturers and to clinical researchers. It should be used
to reduce overlap between investigations, should avoid unnecessary replication of studies, and encour-
age collaborative projects particularly in the evaluation of pre-competitive modifications. Similar col-
laborations would facilitate the creation of cohorts, which should be large enough to enable follow-up of
the studied infants through their childhood and into adulthood. It was considered possible to achieve
this without compromising intellectual property rights, commercial confidentiality and competition be-
tween manufacturers.
(continued on next page)

BOX 1-1 (continued)

12. Original study records, with protection of the participants’ confidentiality, should be preserved
wherever possible; from these an anonymous data archive could be made publicly available for retro-
spective epidemiological assessment of associations with adverse and beneficial outcomes.
13. The results of studies and nutritional trials in infants that have not been completed or have not
been published for other reasons should at least be made available publicly, and to the consolidated
database, together with the specific reasons for the cessation of the study. Similarly, the specific rea-
sons why children withdrew from completed studies should be included in the published reports.
SOURCE: Reprinted, with permission, from Aggett et al. (2001).
assessment of that substance for use in infant formulas. The committee was also asked to
consider how the process of estimating intakes and safety of substances intended for infant
formulas has evolved over time and to discuss whether and how this process is changing in
light of the current state of clinical science to safeguard the health and well-being of infants
enrolled in clinical studies. In addition, the committee was requested to apply its recom-
mended approaches to the specific situation of adding long-chain polyunsaturated fatty
acids (LC-PUFAs) to infant formulas for use in term infants. The approaches were to also be
applied to other ingredients, if appropriate.
The committee’s primary charge was to provide guidelines for assessing the safety of
new ingredients added to infant formulas under U.S. regulations, with possible international
applications. Under U.S. regulations, the process that evaluates the safety of ingredients is
stated in the FD&C Act, but it does not address issues of efficacy (i.e., health benefits).
Although the committee recognizes that efficacy should be considered when assessing new
ingredients to be added to infant formulas, efficacy is not a consideration under current or
proposed regulations for infant formulas or infant formula ingredients. The committee
therefore focused its attention on matters related to safety as delineated in its charge.
Similarly, although the committee recognizes that cost can be a factor for expert panels
making decisions about the appropriate studies to be used to assess safety, cost issues were
not included in the charge and, consequently, were not part of its deliberations.
The committee recognizes that some of its recommendations could not be implemented
under the current U.S. laws and may require statutory changes. Even with this limitation, the
committee encourages dialogue among members of government agencies, the public, indus-
try, and academia to act on the recommendations set forth by this report in the best interest
of our most vulnerable members of society—our infants.
THE COMMITTEE’S APPROACH
Types of Ingredients
The committee focused on new potential ingredients and existing ingredient ratios that
could be added to infant formulas to:
• impart potential health benefits within the first year of life, later in childhood, and
perhaps even during adult life; and
• change color, extend shelf life, or modify marketing or manufacturing processes
because they could potentially be associated with harmful short- or long-term effects.

Several compounds have been seriously considered for addition to formulas, but no final
decision has been made about their addition for a variety of reasons. For example, choles-
terol is present in human milk at levels higher than is present in cow’s milk. A recent article
reported that breastfed infants had lower total cholesterol and low-density lipoprotein
cholesterol in adulthood and suggested that infant formulas should have added cholesterol
to more closely match that of human milk (Owen et al., 2002). On the other hand, animal
studies of formulas with added cholesterol found no evidence of any beneficial short- or
long-term effects (LSRO, 1998). In addition, lysozyme and lactoferrin are also present in
human milk at levels higher than in cow’s milk. These factors may be important in growth
or host defense and have been considered for addition to formulas (Lo and Kleinman, 1996).
Many substances found in human milk have not been added to infant formulas, perhaps due
to the lack of information on their function in human milk or their effect on the infant.
The committee proposes a functional classification of potential target areas of ingredi-
ents new to infant formulas, as summarized in Table 1-2, according to potential targeted
endpoints. The committee also considered the types of molecules or ingredients that could be
considered as new additives, including:
• new sources of existing ingredients;

• prebiotics and probiotics;
• lipids;
• nitrogen-containing ingredients (e.g., recombinant proteins; single amino acids, such
as glutamine, and immunoglobulins; and bioactive peptides/polyamines/proteins, such as
lactoferrin, enzymes, and hormones);
• oligosaccharides;
• vitamins, minerals, and flavonoids;
• flavoring and coloring agents; and
• nonprotein recombinant-derived products.
However the committee did not simply consider molecules or new ingredients as iso-
lated substances, it also considered three important characteristics: (1) the compound itself
TABLE 1-2 Functional Classification of Potential Target Areas of New Ingredients

Examples of Possible
Class Examples Adverse Consequences
Anti-allergic Prebiotics, probiotics, cytokines Gastrointestinal side effects
Anti-infective Antiviral/bacterial agents, probiotics, Adverse immunological changes,
prebiotics, lactoferrin allergic effects
Flavor/color/texture/ Organoleptics, flavors, aeromatics Toxicity, carcinogenicity
stabilizing agents
Immunologic Long-chain polyunsaturated fatty acids, Immunological, inflammatory
enzymes, cytokines, nucleotides effects
Metabolic Enzymes, hormones Allergic immunological changes,
allergic effects
Neurodevelopmental Long-chain polyunsaturated fatty acids, Decreased growth, decreased
and behavioral growth factors, choline, oligosaccharides, cognitive ability
precursors of neurotransmitters,
neurotransmitters, amino acids
Trophic Growth factors, hormones Abnormal increase or decrease in
growth, endocrine effects

(the molecule), (2) the matrix in which it is delivered, and (3) its amount and ratio relative to
other constituents in the infant formula. The committee also realized that certain potential
new additions to infant formulas might be in the form of complex ingredients or biologicals
(e.g., probiotics). Some of the recommended chemical, physical, and in vitro characteriza-
tion steps might not apply to complex ingredients (see Chapter 5).
The committee also reviewed whether ingredients derived from genetically engineered
techniques should have the same safety recommendations for use in infant formulas as
compared with other classes of new ingredients. Even though compounds derived from
genetic engineering techniques would have properties similar to those of natural compounds,
they would still have to be tested. Similar to other ingredients, the short- and long-term
unintended compositional changes that result directly or indirectly as a result of genetic
engineering would need to be considered. In addition, the unintended functional conse-
quences of a successful, specific intentional alteration would need to be considered and
monitored over long time periods. Thus the committee recommends that standards for
safety should be the same for ingredients derived from genetically engineered techniques as
they are for other classes of ingredients. Guidelines should maintain emphasis on a “reason-
able certainty of no harm” for all ingredients.
Safety Definitions
The committee reviewed other existing reports and guidelines, including those available
from FDA and the Life Sciences Research office (LSRO), and it considered current recom-
mendations relating to safety. While the committee recognized that its recommendations
had to be targeted with safety as the ultimate objective, additions of ingredients new to
infant formulas need to consider potential efficacy (i.e., health benefits) considerations.
The concept of “safety” refers to a reasonable certainty of no harm per the FD&C Act;
to safe and adequate levels of nutrients, including essentiality, stability, history of use, and
toxicity per LSRO; or, in some circumstances, a reasonable balance between costs (e.g.,
risks, harm) and benefits. Safety, therefore, is not an inherent biological property, but rather
a point on a continuum that is influenced by intellectual concepts and judgment. Generally,
a “hazard” refers to a substance or combination of substances that produce undesired
outcomes; in the case of infant formulas, the undesired outcome is health related. “Risk”
implies that an adverse event will be expressed under specified conditions, and “harm”
refers to the nature of an undesired outcome associated with a hazard. Details about the
concept of safety and surrounding issues are provided in Chapter 2.
Use of a Hierarchical Approach to Safety Assessment

The committee recognized that its charge was to provide comprehensive guidelines for
evaluating the safety of the addition of ingredients new to infant formulas—not to produce
a “how to” document. With this in mind, the committee discussed an initial outline and
agreed that for each area it would review the existing systems, identify gaps or limitations of
the systems, and make recommendations for revised guidelines. The committee also agreed
on the concept of a hierarchical approach to determine levels of safety assessment needed for
each ingredient.
To evaluate the safety of the addition of new ingredients to infant formulas, the commit-
tee proposes the use of a hierarchical approach, which involves matching the level of safety
assessment with the degree of concern about the potential harm. Such an approach needs to

be considered in terms of a new ingredient’s potential: (1) harm (e.g., toxicity), and (2)
adverse effects. This hierarchical approach will guide the levels of assessment applied de-
pending on the nature and purpose of the new ingredient.
Considerations about the degree of concern and, therefore, about the level of safety
assessment required would include: whether potential harmful effects of a new ingredient
would be reversible or irreversible, the severity of the effect, and the consequences. For
example, levels of safety assessment might not be as intense for an ingredient with a proven
lack of adverse effects (e.g., an ingredient added for formula stability) as compared with a
new ingredient added to induce a positive biological effect (e.g., better visual acuity). In its
hierarchical approach, the committee includes assessments that consider whether an adverse
effect manifests immediately (while the infant is ingesting the formula) or later (even into
adulthood and future generations). The committee also recommends assessing the likelihood
that a new ingredient could adversely affect a specific system and whether the effect would
be common or rare.
To help facilitate the hierarchical approach, the committee recommends the use of
algorithms. An algorithm diagrams the process into a step-by-step decision tree. The steps in
the algorithm include:
• recommended observations and assessments,

• decisions to be considered, and
• actions to be taken.
Standardized symbols are used in a flowchart to display each step in the algorithm
(SMDMC, 1992) (Figure 1-2). Arrows connect the numbered boxes, indicating the order in
which the steps should be followed. A letter within a box of an algorithm refers the reader
to corresponding text (or sidebar). The sidebar elaborates on the recommendations and
statements that are found within each box of the algorithm. Readers should keep in mind
that the algorithm depicts the logic of the process, but does not denote a chronology (e.g., a
manufacturer may initiate several different studies and procedures at the onset of the pro-
cess, the results of which are assessed at different steps of the algorithm).
The algorithms presented in Chapters 4 through 7 are designed to encompass a
broad spectrum of components for evaluating the safety of new components added to
infant formulas. The use of algorithms in applying the hierarchical approach provides
several advantages. First, the utilization of algorithms should simplify the process of
planning the type and depth of safety assessments for each new ingredient. Second,
evidence suggests that an algorithm approach improves data collection, problem solv-
ing, and decision making. Third, multiple levels of information are incorporated into a
single, unified document. Finally, the algorithmic format allows the regulatory agency
and the manufacturer to follow a linear approach to critical information needed at the
major decision points.
The algorithms in this report are provided as generic guides and as tools for stepwise
approaches to be used in assessing the safety of ingredients new to infant formulas. The
committee realizes that it cannot provide specific recommendations for each compound and
that some variation in these approaches may be needed for specific ingredients. Thus the
committee recommends that the manufacturer and an expert review panel establish the
relative importance of potential adverse effects for each specific new ingredient and deter-
mine the depth of preclinical and clinical studies and in-market surveillance needed to assess
safety.

Yes
No
4
3
No
ACTION and ENDPOINT
FIGURE 1-2 An example of an algorithm.
Using this flexible, stepwise approach, each potential new ingredient is considered using
evidence-based approaches and high-quality scientific data to assess potential adverse effects
on:
• growth and development (including temperament),

• the immune system (including allergic and infectious risks),
• metabolism (e.g., acid-base balance, effects related to lipid and carbohydrate process-
ing), and
• other organ systems (e.g., hematological, gastrointestinal).

For each distinct area of safety assessment (i.e., preclinical, clinical, and in-market
monitoring) the committee designed algorithms or stepwise decision trees to be applied to
new ingredients for infant formulas. The committee’s approach was based on the uniqueness
of the infant population. Therefore each step in the process requires empirical evidence from
many disciplines and the application of the highest standards, whether using methods of
bioassay, nutritional analysis, or basic chemistry. This approach could be applied to new
ingredients to be added to infant formula regardless of the regulatory process used.
ORGANIZATION OF THE REPORT

The committee structured this report so that the first three chapters provide the back-
ground information and rationale for the report. Chapter 2 reviews the parameters consid-
ered by the committee when defining “safety” and how to approach it from a practical,
theoretical, and statistical point of view. Chapter 3 compares the biological and behavioral
advantages of human milk with infant formulas and reviews how infant formulas were
developed to meet the biological advantages of human milk.
The remainder of the report reviews the current regulatory processes involved in evalu-
ating the safety of ingredients new to infant formulas and provides recommendations for the
overall process (Chapter 4), preclinical studies (Chapter 5), clinical studies (Chapter 6), and
finally, in-market surveillance (Chapter 7). In Appendix D the committee provides two case
studies that have recently utilized the current regulatory process by submitting a GRAS
Notification. The addition of LC-PUFAs and probiotics to infant formulas are used as
examples of how the recommendations of the committee could be utilized for new ingredi-
ents in infant formula. LC-PUFAs may affect diverse functions within the brain and there-
fore demonstration of safety is critical; probiotics is presented because it is a complex
ingredient that could potentially impact diverse organ systems and could potentially raise
allergenicity concerns. The purpose of this exercise was not to make conclusions about the
completeness of the information or the safety of the new ingredients, but to point out
important assessments or steps that could be considered. In doing so, the committee empha-
sizes the importance of including some steps, such as follow-up surveillance and level 2
assessments, in future assessments of the safety of the addition of new ingredients to infant
formulas.
SUMMARY
The Committee on the Evaluation of the Addition of Ingredients New to Infant Formula
provides FDA, Health Canada, and other regulatory bodies worldwide with a thorough
discussion of the challenging issues surrounding the determination of the safety of new
ingredients to be added to infant formulas. The committee encourages dialogue among
members of government agencies, the public, industry, and academia to act on the recom-
mendations set forth by this report in the best interest of our most vulnerable members of
society—our infants.
REFERENCES
Aggett PJ, Agostini C, Goulet O, Hernell O, Koletzko B, Lafeber HL, Michaelsen KF, Rigo J, Weaver LT. 2001.
The nutritional and safety assessment of breast milk substitutes and other dietary products for infants: A
commentary by the ESPGHAN Committee on Nutrition. J Pediatr Gastroenterol Nutr 32:256–258.

Canada. 2001. Departmental Consolidation of the Food and Drugs Act and the Food and Drug Regulations.
Ottawa: Minister of Public Works and Government Services Canada.
EC (European Communities). 1997. Regulation (EC) No 258/97 of the European Parliament and of the Council of
27 January 1997 concerning novel foods and novel food ingredients. Off J Eur Communities L043:1–7.
EEC (European Economic Communities). 1989. Council Directive of 21 December 1988 on the approximation of
the laws of the member States concerning food additives authorized for use in food stuffs intended for human
consumption (89j/107/EEC). Off J Eur Communities L40:27.
EEC. 1990. Council Directive of 23 April 1990 on the deliberate release into the environment of genetically
modified organisms (90/220/EEC). Off J Eur Communities L117:15.
EEC. 1994. European Parliament and Council Directive 94/34/EC of 30 June 1994 amending Directive 89/107/
EEC on the approximation of the laws of the Member States concerning food additives authorized for use in
foodstuffs intended for human consumption. Off J Eur Communities L237:1.
FDA (Food and Drug Administration). 1996. Current good manufacturing practice, quality control procedures,
quality factors, notification requirements, and records and reports, for the production of infant formula:
Proposed rule. Fed Regist 61:36153–36219.
FSAI (Food Safety Authority of Ireland). 1999. Recommendations for a National Infant Feeding Policy. Dublin:
FSAI. Pp. 94–121.
Hediger ML, Overpeck MD, Ruan WJ, Troendle JF. 2000. Early infant feeding and growth status of US-born
infants and children aged 4–71 mo: Analyses from the Third National Health and Nutrition Examination
Survey, 1988–1994. Am J Clin Nutr 72:159–167.
Koletzko B, Ashwell M, Beck B, Bronner A, Mathioudakis B. 2002. Characterisation of infant food modifications
in the European Union. Ann Nutr Metab 46:231–242.
Lo CW, Kleinman RE. 1996. Infant formula, past and future: Opportunities for improvement. Am J Clin Nutr
63:646S–650S.
LSRO (Life Sciences Research Office). 1998. Assessment of Nutrient Requirements for Infant Formulas. Bethesda,
MD: LSRO.
Owen CG, Whincup PH, Odoki K, Gilg JA, Cook DG. 2002. Infant feeding and blood cholesterol: A study in
adolescents and a systematic review. Pediatrics 110:597–608.
Ryan AS, Wenjun Z, Acosta A. 2002. Breastfeeding continues to increase into the new millennium. Pediatrics
110:1103–1109.
SMDMC (Society for Medical Decision Making Committee). 1992. Proposal for clinical algorithm standards.
Society for Medical Decision Making Committee on Standardization of Clinical Algorithms. Med Decis
Making 12:149–154.
Vanderveen JE. 1991. The role of the Food and Drug Administration in regulating food products for children. Ann
N Y Acad Sci 623:400–405.

Defining Safety for Infants
ABSTRACT
“Safety” refers to a reasonable certainty of no harm and is described by noting
a range along a continuum, rather than as an absolute point or value (Food Addi-
tives Amendment, P.L. 85-929 of the Federal Food, Drug and Cosmetic [FD&C]
Act, 21 U.S.C. §301). The relationship between biology and safety is mediated
through the concepts of harm, benefit, and risk.
Manufacturers may propose the addition of a new ingredient to infant formulas
by demonstrating the safety, not the efficacy (the capacity to produce an intended
effect under the realistic situation of product use), of the proposed ingredient. Foods
are generally considered to be inherently efficacious (with inherent sensory proper-
ties and nutrition) and, thus, efficacy is not a consideration in their safety assess-
ment. In the case of infant formulas, this assumption is modified to some degree
because it has been proposed that the products must be capable of sustaining physi-
cal growth for a specified period of time. Currently, however, manufacturers are not
required to demonstrate the benefits of an individual ingredient in the product. In-
fancy is a uniquely vulnerable period that complicates the interpretation of safety
guidelines. Not all organ systems are fully mature at birth, and as they undergo
further development they are highly susceptible to nutritional inputs, illnesses, care
practices, and other environmental inputs. Early infancy represents a period of
growth and development when a successful outcome depends on the timely emer-
gence of critical structures and developmental processes. The gastrointestinal, renal,
and immune systems, as well as brain and neurological functions, could be affected
by exposure to potentially harmful substances contained in infant formulas. Optimi-
zation of nutrition and minimization of exposure to potentially harmful substances
in the food supply is of heightened importance during infancy. The committee con-
cluded that there are six issues that must be considered as important safety issues
when regulating infant formulas: (1) infant formulas are the sole or predominant
source of nutrition for many infants, (2) formulas are fed during a sensitive period of
29

development and may therefore have short- and long-term consequences for infant
health, (3) animals may not be the most appropriate model on which to base deci-
sions of safety, (4) “one size fits all” food safety models may not work for all new
additions to formulas, (5) infant formulas could be considered as more than just
food, and (6) potential benefits, along with safety, should be considered when add-
ing a new ingredient to formulas.
INTRODUCTION
A discussion on guidelines to ensure the safety of ingredients new to infant formulas
requires a broad understanding of the concepts and models of safety regulations and an in-
depth review of infancy as a unique period that requires unique safety measures. This
chapter describes fundamental concepts of safety regulations, statistical considerations in
assessing food safety, models of safety assessment (including the Novel Foods model of
Health Canada), and special considerations for ensuring the safety of infants and regulating
infant formulas.
U.S. REGULATORY AGENCIES

Several federal, state, and local agencies monitor the safety and quality of substances
that are ingested and inhaled by the general public. Each agency operates within its own
domain using its own set of rules and procedures, but the agencies work collaboratively on
some issues to ensure the safety and quality of food, water, and air. The Food and Drug
Administration (FDA) has the primary responsibility for regulating ingredients new to infant
formulas and other agencies serve in peripheral roles.
FUNDAMENTAL CONCEPTS OF SAFETY REGULATIONS

The terms safety, hazard, risk, harm, and benefit are central to the full understanding of
safety assessment. Safety refers to a reasonable certainty of no harm (21 U.S.C §348) (e.g.,
for food ingredients in the U.S. regulatory system) or, in some systems (e.g., for pharmaceu-
tical drugs in the U.S. regulatory system), a reasonable balance between costs (harm) and
benefits. Safety is an intellectual concept; it is not an inherent biological property of a
substance. Safety is described by noting a range along a continuum, not an absolute point or
value. The relationship between biology and safety is mediated through concepts of harm,
benefit, and risk. As such, the concept of safety and its assessment is influenced by many
different organizations, individuals, and intellectual disciplines.
A hazard generally refers to some substance or combination of substances (organic or
inorganic, or in some cases psychological) that may, under some circumstances or for some
individuals, produce undesired health-related outcomes. Nearly any substance can have the
potential to produce an undesired health outcome to some individual under some circumstance.
Exposure to a hazard, however, does not guarantee that an undesired outcome will occur.
The likelihood that such an outcome will occur given exposure to a hazard is what is
referred to as the risk of that hazard. The risk of a hazard is not a general (main effect)
property of the substance, but rather it is an interaction between the nature of the hazard,
the circumstances of the exposure (e.g., the magnitude, the timing, and the specific features
of the individual, such as health status, age, and susceptibility), and other such conditions
that moderate the actual health impact of the hazard.
Harm refers to the nature of the undesired outcome (usually a health outcome) associ-

DEFINING SAFETY FOR INFANTS 31
ated with a hazard and is often expressed in terms such as cost. Not all harm is the same, and
not all individuals would assess the same outcome as having equivalent harm.
Opposite harms are the benefits of the addition of substances. The ratio of costs to
benefits is a critical unit in some safety assessment systems. Cost-benefit ratios may apply to
individuals or groups. For example, an individual may benefit from a treatment but may
experience side effects. Another example is the case of iron supplementation to reduce infant
risk of anemia. Some or all of the infants within a group may benefit from receiving
additional iron, while some may be harmed (e.g., experience constipation); on average,
however, the population that benefits from iron fortification will be larger than the popula-
tion that experiences harm.
Members of the general public, special interest advocates, lawmakers, scientists, leaders of
government agencies, and industry representatives play significant roles in establishing safety
guidelines. The public makes certain demands for lowering food safety risks (whether real or
perceived risks as a result of misinformation) and expresses its concerns either through con-
sumer organizations or through individual contact with appropriate government agencies.
Consumer organizations may voice such concerns in a focused manner. Lawmakers weigh
these concerns and, where appropriate, engage in debates that may result in new laws and
regulations. In the process of establishing formal policy, other individuals and organizations
often enter into the debate to influence the final statements of safety regulation. For example,
scientists or professional organizations may contribute important information that derives
from scientific studies, economists may provide information about the costs of implementing
certain safety standards, and industry representatives may describe the impact of the regula-
tion on manufacturers. Once laws are enacted, regulatory agencies are entrusted with the
responsibility of developing, implementing, and enforcing regulations.
STATISTICAL CONSIDERATIONS IN FOOD SAFETY DETERMINATION

Among adults, guidance concerning the clinical relevance of differences in some physi-
ological parameter (e.g., blood pressure) is derived from a body of accumulated evidence
that provides a rationale for a clear definition of pathological state (e.g., hypertension). In
testing an ingredient new to an infant formula, it is unlikely that investigators would detect
any clear evidence of disease (which should have been ruled out by preclinical testing).
Instead, more subtle differences in physiology or development may appear that lack suffi-
cient evidence to inform clinical judgment. Investigators must determine whether a differ-
ence (e.g., level of growth) has an immediate health consequence for the infant and the level
of difference that matters for the long term (e.g., a growth deficit associated with a particular
ingredient rapidly disappears when other foods are added to the diet). In the absence of
sufficient evidence for clinical judgment, investigators may be forced to utilize statistical or
analytical approaches as the basis for making judgments about safety.
The process of establishing the safety of food products, especially infant formulas, is
complex and requires empirical evidence from many disciplines. Each step in the process
requires the application of the highest standards, whether using methods of bioassay, nutri-
tional analysis, or basic chemistry. Eventually studies involving human subjects (particularly
in the case of infants) must be conducted in order to demonstrate the product’s safety for the
human consumer. Studies involving humans are almost always conducted as randomized
clinical trials and standard methods of design and analysis are followed.
The most typical analytical approach to interpreting the data from scientific studies, includ-
ing clinical trials, is the statistical significance test, also known as the null hypothesis significance
test (NHST). This approach, which has recently come under much scrutiny and debate, formally

applies a set of principles for establishing a “rule of evidence” in scientific inquiry. In the simplest
terms, NHST provides a set of rules for decision making under uncertainty.
First, a set of two mutually exclusive alternative conditions is specified. For example, the
addition of substance X to an infant formula either: (a) hinders the ability of the infant to
maintain proper physical growth, or (b) it does not hinder proper physical growth. Second,
a set of risk probabilities are specified (the “alpha level” of the test and the “power” of the
test), which allows the researcher to control the probabilities of drawing an incorrect infer-
ence. Third, a set of assumptions is specified that, taken together with the null and alterna-
tive hypotheses, allow the complete specification of the behavior of some statistical index.
From this model one can specify a set of decision rules to draw some conclusions based
on the empirical results of the experiment. The result of such a statistical test procedure does
not establish with certainty the “true state of nature,” but rather it expresses a degree of
confidence that one of the two states is not likely to occur. The randomized clinical trial and
associated NHST are the mainstays of certain safety and efficacy approaches, such as the
FDA drug trials described later, but they have certain potential limitations in their applica-
tion to the safety of ingredients new to infant formulas.
The first limitation is that NHST lacks a certain degree of direct applicability. The basic
concept underlying the safety of an ingredient added to infant formulas is the “reasonable
certainty of no harm” concept without a requirement for the demonstration of benefit. NHST,
however, is generally formulated to demonstrate the superiority of one condition versus an-
other. The fundamental idea in the formulation of reasonable certainty of no harm is one of
equivalence, not of difference. While one can manipulate the null and alternative hypothesis in
this circumstance (e.g., to make the “no difference” condition the alternative hypothesis), the
resulting formulation is awkward and shifts the probabilistic control of the important error
rate to that of power rather than to the more direct alpha level of the resulting test. Due to its
highly unusual nature, it is likely that this test’s results will not be properly understood and
interpreted.
The second limitation is that NHST is concerned with demonstrating a difference rather
than with the size or importance of the difference. In a number of scientific disciplines, most
notably psychology, there has been a shift in emphasis from statistical significance to clinical
significance. When formulated as clinical significance, the question becomes whether the
difference that is detected by NHST is one that has any practical health consequences from
the perspective of the individual receiving the treatment. Thus it is recognized that while a
very small (potentially clinically inconsequential) difference can be detected by NHST (par-
ticularly in very large samples), the most important question is to determine whether the
difference has any health implications.
In order to address these different questions, a number of approaches to testing have
been adopted that include the following features:
• a greater focus on effect size estimates,

• the use of confidence intervals as the primary way of reporting the results of experi-
ments, and
• the use of alternative statistical tests that are not based on the sample mean as the
primary way of characterizing populations.
One approach to determine clinical significance is the use of dominance statistics (Cliff,
1993). In this approach it is asked, in a probabilistic manner, how likely is it that an
individual chosen at random from the group receiving treatment A will score better than an
individual chosen at random from the group receiving treatment B. Using the dominance

statistic approach, the question of weight maintenance might be approached as determining

the probability that an infant drawn at random from a group consuming the new formula
would weigh less than an infant drawn at random from a group fed the existing formula. If
this probability were not too extreme, it might argue for the safety of the new addition
relative to that of a formula without that addition.
An alternative approach to NHST may also be considered. For example, Seaman and
Serlin (1998) developed methodologies that allow the direct demonstration of equivalence
(as opposed to the subtly different approach of NHST that is oriented to showing differ-
ences). Such approaches may be applied to establish rules to determine the safety of ingredi-
ents new to infant formulas.
In assessing the safety of ingredients new to infant formulas, investigators must consider
whether the NHST approach is the most appropriate method to analyze empirical data or
whether an approach that more directly assesses effect sizes, clinical significance, domi-
nance, or equivalency is preferred. Of particular importance in the selection of a proper
analytical technique is the question of whether any decrease in infant growth rate can be
tolerated. If slower growth rates or other minor differences in physiology or function cannot
be tolerated, then investigators may wish to demonstrate equivalency, rather than merely
detect differences. In addition, the error rate that one is willing to tolerate must be carefully
considered rather than routinely adopting the traditional 0.05 alpha level. It is the alpha
level that indicates the “societal” values for the acceptance of risk and harm.
MODELS OF SAFETY ASSESSMENT

Virtually all other models used to assure the safety of ingested or inhaled substances are
variations of the food and drug safety models (described below): the nutrients model based
on dose-response relationships; the in-market monitoring and surveillance model; the novel
foods and food additives models, based on a reasonable certainty of no harm; and the drug
model, based on risk-benefits assessments.
Food Models
Nutrients Model
The Dietary Reference Intakes (DRIs) are a set of quantitative reference values for
nutrient intake to be used for planning and assessing diets, and they are based on risk
assessments. One of the DRIs, the Tolerable Upper Intake Level (UL), uses risk assessment
approaches. The UL uses a substantially different approach from the one used for food
additives in that it does not address any particular food product or ingredient, yet it provides
useful information on the magnitude of intake of nutrients for individuals as a function of
age, gender, pregnancy and lactation status, and other such factors. At the heart of the DRI
methodology is a dose-response relationship (e.g., the examination of a targeted health
outcome as a function of the intake of the nutrient in question). The UL is the level of intake
at which one would expect virtually no risk of an adverse health outcome for almost all
healthy individuals in the population. There would be an increased risk of an adverse health
outcome if more than the UL were consumed. The actual risk assessment methodology of
the DRI approach is quite complex; it is described in Dietary Reference Intakes: A Risk
Assessment Model for Establishing Upper Intake Levels for Nutrients (IOM, 1998) and in
each of the nutrient-specific DRI reports, most recently in Dietary Reference Intakes for
Water, Potassium, Sodium, Chloride, and Sulfate (IOM, 2004).

In-Market Monitoring and Surveillance

A different approach to ensuring food safety is in-market monitoring and surveillance.
For instance, once an infant formula is released for general consumption, FDA requires that
the manufacturer maintain and analyze records of consumer complaints. Further, the manu-
facturer is required to report to FDA cases where the evidence suggests a reasonable possibil-
ity of a link between consumption of the formula and an infant illness, death, or a reduction
in intake of required nutrients. This is currently a passive system that merely requires that a
caregiver report a suspected case, but it is one more mechanism by which the safety of infant
formulas can be monitored. Details of in-market monitoring and surveillance are discussed
in Chapter 7.
Novel Foods Model

In 1994 Health Canada issued the Guidelines for the Safety Assessment of Novel Foods
(Health Canada, 1994) to assist in the design of premarket notifications for novel foods. The
manufacturer must submit a request to Health Canada 45 days prior to the sale or advertis-
ing for sale of any novel food under specific requirements laid out in Division 28 of the Food
and Drug Regulations (Canada, 2001). Where it is determined that data of a scientific nature
are required to support the safety of the product, a safety assessment may include an
evaluation of microbiological, molecular, chemical, nutritional, and toxicological endpoints
in preclinical and clinical studies. Specifically, the assessment may evaluate the process used
to develop the novel food (molecular, chemical, and microbiological), the comparison of the
characteristics of the novel food with that of its traditional counterpart (nutritional and
toxicological), safety issues related to a source with no history of use in the food supply
(nutritional and toxicological), and the potential allergenicity from proteins introduced into
the food (nutritional). Health Canada may request additional information to assess the
safety of the novel food and, if satisfied, will notify the manufacturer that the information is
sufficient and the product is safe for consumption.
Food Ingredients Model

The food safety evaluation of food ingredients and associated regulations in the United
States is based on “reasonable certainty of no harm.” Sections 201(s), 201(z), 409, and 412
of the FD&C Act and FDA’s Toxicological Principles for the Safety Assessment of Direct
Food Additives and Color Additives Used in Food, also known as the Redbook1 (OFAS,
2001, 2003), are the regulations and guidelines that are used to assess the safety of food
ingredients and infant formulas.
Regulatory oversight of the addition of new ingredients to infant formulas is governed
largely by two processes: the Food Additive Petition and the Generally Recognized as Safe
(GRAS) Notification. It is important to note that these procedures are to ensure the safety (a
reasonable certainty of no harm)—not the benefits—of the proposed ingredient. A food
additive is any product added to a food that is not generally recognized as safe by qualified
experts (see Box 4-1 in Chapter 4 for the complete definition). The food additive petition

process requires that the manufacturer file a petition with FDA that provides all available
data on the safety of the product and proposes the conditions under which such an additive
may be safely used. FDA may issue a safety declaration upon review of the petition. In this
case it is FDA that makes the affirmative declaration of the safety of the additive in the
context of its proposed use.
By contrast, the GRAS Notification process requires that the manufacturer make the
initial declaration of the safety of the product based on consensus by qualified experts. FDA
then reviews the notification and, if all of its questions are satisfactorily answered, the
agency issues a letter of no objection. Because this process has become the primary route of
introduction of new ingredients to infant formulas, the GRAS Notification process is re-
viewed in greater detail in Chapter 4.
Infant formulas are the sole source of nutrition for many infants and, therefore, one step
is required for the approval of modifications to formulas that is not required for other foods.
Manufacturers seeking to market a new infant formula need to comply with regulations
under Section 412 of the FD&C Act. New regulations under that section of the FD&C Act
have been proposed that would require manufacturers to demonstrate that the formula
containing the new ingredient in the matrix in which the product is delivered is capable of
sustaining physical growth and development over 120 days, the period when the formula is
likely to be the sole source of infant nutrition.
In 1982 FDA issued, and later updated, the Redbook (OFAS, 2001, 2003).2 These
guidance documents were prepared to assist in the design of protocols for animal studies
conducted to test the safety of food ingredients and include detailed guidelines for testing the
effects of food ingredients on mothers and their developing fetuses. However, due to the
special conditions surrounding infancy described below, special considerations need to be
taken into account when applying the Redbook in the case of infant formulas.
As mentioned previously, since virtually all models of safety determination are based on
empirical evidence, only minor variations of the basic models seen in the FDA food safety
determination system are possible. These differences are based more upon emphasis and
implementation than on any profound differences in methodology. One major exception to
this uniformity is the food safety model based upon the “reasonable certainty of no harm”
concept. As opposed to other models where benefits from the addition of the new ingredient
need to be demonstrated, in that model safety is seen as no harm and no proof of beneficial
effects is needed. As discussed further in Chapter 4, the committee believes that for infant
formulas, the concepts of efficacy (benefit) and safety are not always mutually exclusive
because of the uniqueness of the infant population and, therefore, potential benefit should
be considered when allowing new ingredients to be added to infant formulas.
Drug Safety Model

Structurally the FDA drug approval process closely resembles the food additive model of
food safety. A manufacturer that wishes to market a new drug submits a petition to FDA,
2The original Redbook (Redbook I) was published in 1982, revised in 1993, and updated in 2001 (draft
Redbook II). In 2000, FDA released a revised version of the publication as Redbook 2000: Toxicological Prin-
ciples for the Safety Assessment of Food Ingredients. However, some chapters in Redbook 2000 have not yet been
revised, so both the draft Redbook II and Redbook 2000 are used as guidance documents when conducting animal
studies. Appendix C lists the contents of the draft Redbook II and Redbook 2000 and indicates which chapters in
Redbook 2000 have been updated.

along with a specified set of empirical evidence concerning the product. FDA (with advice
from an established panel of experts) evaluates the evidence and, if satisfied that the drug
meets its regulatory criteria, approves the drug for commercial use. In the drug approval
process, however, the evidentiary basis for the decision is quite different from that employed
in the food additive model where evidence of benefit is not necessary.
First, the applicant must offer evidence through one or more clinical trials of the efficacy
of the drug. That is, there must be clear and cogent evidence that the drug does what it
claims to do. In addition, it must be shown to have the same effect as the current standard
treatment of the condition being studied. Second, side effects (e.g., adverse reactions) of the
drug must be carefully studied and reported. The criterion for approval of the applicant
product is then based upon an assessment of the benefit:risk ratio.
Other Safety Models

As noted above, virtually all other models used to assure the safety of substances
ingested or inhaled are variations of the food and drug safety models. Given the number of
regulated substances (e.g., air, water, lead) and the number of agencies charged with regulat-
ing them at the federal, state, and local level, it is not surprising that numerous variations in
safety determination practices exist. Most of these differences, however, are at the technical
level rather than at the conceptual level (e.g., setting the standard for lead is conceptually
very much like the establishment of upper boundaries in the DRI process).
SPECIAL CONSIDERATIONS FOR INFANT FORMULAS
Infancy as a Vulnerable Period

Dealing with infancy makes interpretation of safety guidelines particularly difficult.
Infancy is a period of very rapid development, and change is the rule rather than the
exception. Infants are nonverbal and cannot report their own experiences, and parents may
not be able to accurately interpret the signals that infants provide. It is reasonable to expect
that any harmful effects from an ingredient might be subtle. Throughout this report, the
committee describes the period of infancy as so special and the consequences of any harm so
great that the public should be unwilling to tolerate any harm due to the addition of an
ingredient new to infant formulas, even if there are benefits to most who would be exposed
to the new ingredient.
At birth, infants make a dramatic transition from a highly controlled prenatal environ-
ment where oxygen and nutrients are delivered directly from the mother to infant via the
blood supply, to one that involves enteral feeding and thus requires an efficient gastrointes-
tinal system and coordination of a wide range of functions. Not all organ systems are fully
mature at birth, and as they undergo further development, they are highly susceptible to
environmental inputs. Early infancy represents a period of growth and development when a
successful outcome depends on the timely emergence of critical structures and developmen-
tal processes. At the same time, infancy is a period of heightened vulnerability to nutritional
inputs, illnesses, care practices, and other environmental influences. Depending on the tis-
sues involved and the timing and severity of insults, effects may be irreversible or the ability
to compensate for differences may be limited. Furthermore, infants have a higher food
consumption rate than adults when expressed on a per kilogram body-weight basis, and they
typically rely on a sole nutrient source (human milk or milk substitutes) in the early months
of life. Thus optimization of nutrition and minimization of exposure to potentially harmful

substances in the food supply is of heightened importance during infancy. Examples of the
various systems that could be affected by exposure to potentially harmful substances con-
tained in infant formulas are described below.
Gastrointestinal and Renal Function

With the exception of exocrine pancreatic function and bile acid metabolism, the gas-
trointestinal tract is anatomically and functionally fully developed in infants born at term,
and dietary contents do not influence their development. Trophic factors can be found in
human milk and in the gastrointestinal tract, but their function is unclear. As these factors
are more fully identified and their structure, composition, and physiological effects under-
stood, it is possible that one or several may have effects on infant organ systems.
Similar to the gastrointestinal tract, development of renal function is not influenced by
dietary content. Glomular filtration does not approximate adult values until about 3 years of
age, and tubular reabsorption and urine acidification reach normal values at several months
of age. However both are sufficient for healthy term infants, but contribute to fluid and
electrolyte abnormalities in infants who are ill and in those who are fed an inappropriate
diet.
Immune Function
The infant immune system is not fully mature at birth; it has deficits in the ability to
prevent invasion of pathogens and to respond to antigens. Of particular concern in the
context of ingredients new to infant formulas is the increased permeability of the gut
mucosal barrier in the presence of inflammation or infection or if the integrity of the
epithelial cell layer is disrupted. The increased permeability allows macromolecules to be
absorbed, which stimulates allergic responses to food proteins.
Brain and Behavior

There is rapid development of both brain and behavior during the first year of life. Both
subcortical and cortical central nervous system (CNS) structures mature and allow the
appearance of critical developmental functions, such as visually guided reaching, face recog-
nition and orientation toward faces, explicit and working memory, focused attention, and
inhibitory control (Johnson, 2001; Nelson, 1995). The quality of the young infant’s nutri-
tional intake can be an important influence on CNS development and function (Rao and
Georgieff, 2000; Wauben and Wainwright, 1999). For example, iron deficiency anemia in
the first year interferes with the development of functional CNS processes, such as alteration
in the number of dopamine receptors or degree of myelination (Lozoff et al., 1998; Wauben
and Wainwright, 1999). These functional changes in turn can have long-term cognitive and
behavioral consequences that cannot be compensated for by subsequent nutritional inter-
ventions (Lozoff et al., 1998).
A similar process is also seen with regard to behavioral development. The infant’s
increasing behavioral competencies and patterns of social-emotional functioning during the
first year can impact upon critical developmental mediators, such as patterns of parent-child
relations, stress reactivity, initiation of self-regulation skills, and extent of infant interactions
with the larger environment (Gunnar, 2000; Ruff and Rothbart, 1996). Interference with the
development of these critical mediators also can have long-term consequences for later
functional competence (Wachs, 2000). Since a large proportion of the infant’s nutrition

during the first year may be supplied by formulas, changes to formulas that impact upon
normally occurring brain and behavioral developmental patterns can have potential long-
term consequences.
Important Safety Considerations

Most of the principles that govern the safety of ingredients new to infant formulas derive
from the same principles that govern food safety for older children and adults. The commit-
tee concluded that the six issues described below and summarized in Box 2-1 must be
considered as important safety issues when regulating infant formulas.
1. Infant formulas are the sole or predominant source of nutrition for many infants.
Even for infants for whom formulas are a supplement to human milk, formulas can consti-
tute a major source of nutrition. For infants, formulas represent a much larger percentage of
the total nutritional intake than is any single substance for the older child, adolescent, or
adult. Given the extremely high dependence on this one source of dietary input, safety rules
based upon models deemed appropriate for persons with a wide range of dietary inputs may
be inappropriate.
2. Formulas are fed during a sensitive period of development and may therefore have
short- and long-term consequences for infant health. The first 12 to 18 months of life is a
period of extremely rapid growth and development for the human infant that is only begin-
ning to be understood. The brain and neural system change dramatically during this period,
as do other organ, cognitive, and social-emotional systems. These changes are thought to
have long-range implications for the human—not all of which are expressed directly during
infancy. The current safety models for infant formulas only look at relatively short-term
outcomes and are narrowly limited to the maintenance of physical growth.
3. Animals may not be the most appropriate model on which to base decisions of safety.
In the case of infants, limitations of using experimental animals as models are due not only
to species differences, as in the case of adults, but also to developmental differences in
animals versus infants. The differences in growth rates and the resulting differences in
biological effects of ingredients could therefore accentuate the concerns that already exist
when using animal models. Given these differences, animal models may not uncover possible
threats to the long-term well-being of infants due to the addition of new ingredients; this
issue is of critical significance if the models used are developmentally inappropriate.
4. “One size fits all” food safety models may not work for all new additions to formu-
BOX 2-1 Important Safety Considerations When Regulating Infant Formulas
1. Infant formulas are the sole or predominant source of nutrition for many infants.
2. Formulas are fed during a sensitive period of development and may therefore have short- and
long-term consequences for infant health.
3. Animals may not be the most appropriate model on which to base decisions of safety.
4. “One size fits all” food safety models may not work for all new additions to formulas.
5. Infant formulas could be considered as more than just food.
6. Potential benefits, along with safety, should be considered when adding a new ingredient to
formulas.

las. The basic models of food safety are based upon the assumption that nearly all individu-
als function basically in similar ways with regard to the safety of substances in their diet
(although there are gross subtypes with different dietary needs, such as premature infants
and infants with food allergies). However it may be speculated that there are more subtle
underlying subgroups (e.g., genetic subtypes) for which new substances may or may not
have the same safety characteristics as they would for individuals not in those subgroups. In
clinical trials the minimum sample size required to detect systematic group differences in
response to a new ingredient is likely to be entirely inadequate for subgroup analyses.
5. Infant formulas could be considered as more than just food. The rules governing the
safety of infant formulas are regulated under the logic of infant formulas being a “food
product,” and there is no doubt that the primary purpose of infant formulas is nutrition.
However infant formulas can also be seen as a potential delivery system for non-nutritional
agents. This potential as a delivery system should cause one to consider the question of what
are the appropriate (both legal and ethical) boundaries for additions to infant formulas.
Consideration must be given to constructs that will appropriately set the limits and provide
the definitions for an addition being “nutritional.” The presence of the substance in human
milk may not be a sufficient definition of a nutritional substance, and other factors, such as
whether the substance is produced from genetically modified sources, need to be considered.
In addition, the regulatory steps to follow to ensure the safety of a substance whose purpose
is non-nutritional (e.g., a substance whose purpose is to reduce infant discomfort) need to be
determined.
6. Potential benefits, along with safety, should be considered when adding a new ingre-
dient to formulas. Food products are generally considered to be inherently beneficial or
efficacious (with inherent sensory properties and nutrition) and thus efficacy (i.e., health
benefits) has not been a consideration in the safety assessment of foods. In the case of infant
formulas, this starting assumption may be modified through the additional requirement in
proposed regulations (FDA, 1996) that the product be capable of sustaining physical growth
for a specified period of time, that is, it is already required that the final formulated product
be demonstrated to be efficacious from the point of view of physical development. (This
proposed requirement, however, is usually stated as a safety criterion rather than as an
efficacy criterion, that is, a lack of efficacy would be a risk to the organism and therefore a
safety concern.)
Other than this proposed requirement for sustaining physical growth, however, the
manufacturer would not need to demonstrate the benefits of any proposed new ingredient to
infant formulas. Given the special circumstances of the use of infant formulas and the fact
that is it virtually impossible to understand the long-term outcomes of adding any given
ingredient (the clinical trials are, of necessity, of relatively short duration), prudence would
seem to dictate that an ingredient not be added to an infant formula unless it can be shown
that some benefit is accrued to the infant through its addition. Since the committee was not
charged with addressing efficacy, it recommends that consideration be given to convening a
group of experts to explore if benefit could be an appropriate requirement for adding a new
ingredient to infant formulas.
SUMMARY
Infancy is a uniquely vulnerable period that complicates the interpretation of safety
guidelines. Manufacturers may propose the addition of a new ingredient to infant formulas
by demonstrating the safety of the proposed ingredient. They are not required to demon-
strate the benefit of an individual ingredient in the product. Although the committee believes

that, in the case of infant formulas, efficacy is an important consideration, the committee did
not discuss this issue because it was beyond its charge.
The six safety considerations discussed in this chapter are important when developing
safety guidelines for adding ingredients new to infant formulas. In making recommendations
in the chapters that follow, the committee considers these six special issues that set infant
nutrition apart from that of toddlers, children, and adults.
REFERENCES
Ottawa: Minister of Public Works and Government Services Canada.
Cliff N. 1993. Dominance statistics: Ordinal analyses to answer ordinal questions. Psychol Bull 114:494–509.
quality factors, notification requirements, and records and reports, for the production of infant formula.
Gunnar MR. 2000. Early adversity and the development of stress reactivity and regulation. In: Nelson CA, ed. The
Effects of Early Adversity on Neurobehavioral Development. The Minnesota Symposia on Child Psychology.
Vol. 31. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 163–200.
Health Canada. 1994. Guidelines for the Safety Assessment of Novel Foods. Ottawa: Food Directorate, Health
Protection Branch, Health Canada.
IOM (Institute of Medicine). 1998. Dietary Reference Intakes: A Risk Assessment Model for Establishing Upper
Intake Levels for Nutrients. Washington, DC: National Academy Press.
IOM. 2004. Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate. Washington, DC:
The National Academies Press.
Johnson MH. 2001. Functional brain development in humans. Nat Rev Neurosci 2:475–483.
Lozoff B, Klein NK, Nelson EC, McClish DK, Manuel M, Chacon ME. 1998. Behavior of infants with iron-
deficiency anemia. Child Dev 69:24–36.
Nelson CA. 1995. The ontogeny of human memory: A cognitive neuroscience perspective. Dev Psychol 31:723–
738.
OFAS (Office of Food Additive Safety). 2001. Toxicological Principles for the Safety Assessment of Direct Food
Additives and Color Additives Used in Food. Redbook II-Draft. Washington, DC: OFAS, Center for Food
Safety and Applied Nutrition, Food and Drug Administration.
OFAS. 2003. Redbook 2000. Toxicological Principles for the Safety of Food Ingredients. Online. Center for Food
Safety and Applied Nutrition, Food and Drug Administration. Available at http://www.cfsan.fda.gov/
~redbook/red-toca.html. Accessed November 19, 2003.
Rao R, Georgieff MK. 2000. Early nutrition and brain development. In: Nelson C, ed. The Minnesota Symposia
on Child Psychology. The Effects of Adversity on Neurobehavioral Development. Vol. 31. Mahway, NJ:
Lawrence Erlbaum Associates. Pp. 1–30.
Ruff HA, Rothbart MK. 1996. Attention in Early Development. Themes and Variations. New York: Oxford
University Press.
Seaman MA, Serlin RC. 1998. Equivalence confidence intervals for two-group comparisons of means. Psychol
Methods 3:403–411.
Wachs TD. 2000. Necessary but Not Sufficient. The Respective Roles of Single and Multiple Influences on
Individual Development. Washington, DC: American Psychological Association.
Wauben IP, Wainwright PE. 1999. The influence of neonatal nutrition on behavioral development: A critical
appraisal. Nutr Rev 57:35–44.

Comparing Infant Formulas with Human Milk
ABSTRACT
The vast majority of infants in the United States are fed human-milk substitutes
by 6 months of age. This food source, although inferior to human milk in multiple
respects, promotes more efficient growth, development, and nutrient balance than
commercially available cow milk.
Manufacturers often add new ingredients to infant formulas in an attempt to
mimic the composition or performance of human milk. However the addition of
these ingredients is not without risks as a result of a range of complex issues, such as
bioavailability, the potential for toxicity, and the practice of feeding formula and
human milk within the same feeding or on the same day.
Assessing the safety of ingredients new to infant formulas by comparing the
proposed formulas with human milk also presents both regulatory and research
issues. From a research standpoint, clinical studies that assess the effects of new
ingredients are difficult to design because infants cannot be randomized to consume
formulas or human milk. Furthermore, there may be significant non-nutritional con-
founding variables between the groups, including factors related to which mothers
choose to breastfeed. Finally, human-milk composition varies considerably among
and within individuals over time, while the content of infant formulas generally
remains constant.
From a regulatory standpoint, the effect of an ingredient new to infant formu-
las is usually driven by the manufacturer’s desire to produce a product that mimics
the advantages of breastfeeding. This motivation implies that formulas in their
current state are less efficacious (e.g., neurologically or immunologically), although
not necessarily unsafe, when compared with human milk. Thus the safety of any
addition of an ingredient new to infant formulas will need to be judged against two
controls: the previous iteration of the formulas without the added ingredient and
human milk.
41

BACKGROUND
Multiple health organizations, including the World Health Organization (WHO, 2002),
the American Academy of Pediatrics (AAP, 1997), the American Academy of Family Physi-
cians (AAFP, 2003), the American Dietetic Association (ADA, 2001), the Institute of Medi-
cine (IOM, 1991), the Life Sciences Research Organization (LSRO, 1998), the U.S. Depart-
ment of Health and Human Services (HHS/OWH, 2000), Health Canada, and the Canadian
Pediatric Society (Canadian Paediatric Society, 1998) endorse breastfeeding as the optimal
form of nutrition for infants for the first year of life. Nevertheless the vast majority of infants
in the United States are fed human milk substitutes by 6 months of age (Ryan et al., 2002).
This food source, although inferior to human milk in multiple respects, promotes more
efficient growth, development, and nutrient balance than commercially available cow milk.
The American Academy of Pediatrics recommends that infants who are not breastfed should
consume iron-fortified infant formulas rather than cow or goat milk until 12 months of age
(AAP, 1997).
HISTORY OF THE DEVELOPMENT OF INFANT FORMULAS
Milk-Based Formulas
Human-milk substitutes existed before the modern age of formulas. Because some in-
fants could not be fed by their mothers, humans adopted two methods for substitute feedings.
The most obvious was the utilization of a surrogate mother (e.g., wet nurse), who would
feed the child human milk. The alternative was to feed the child milk obtained from another
mammal. The most frequently used sources were the cow, sheep, and goat (Fomon, 1993).
Until the end of the nineteenth century, the use of a wet nurse was by far the safest way to
feed infants who could not be breastfed by their mothers. As general sanitation measures
improved during the latter part of the nineteenth century, and as differences in composition
between human milk and that of other mammals were defined, feeding animal milk became
more successful. However few infants survived until infant formulas based on cow milk with
added water and carbohydrate were introduced. Box 3-1 lists the main landmarks in the
BOX 3-1 History of Commercially Available Infant Formulas

in the United States
Cow-milk-based formulas
1867 – Formula contained wheat flour, cow milk, malt flour, and potassium bicarbonate
1915 – Formula contained cow milk, lactose, oleo oils, and vegetable oils; powdered form
1935 – Protein content of formula considered
1959 – Iron fortification introduced
1960 – Renal solute load considered; formula as a concentrated liquid
1962 – Whey:casein ratio similar to human milk
1984 – Taurine fortification introduced
Late 1990s – Nucleotide fortification introduced
Early 2000s – Long-chain polyunsaturated fatty-acid fortification introduced
Noncow-milk-based formulas
1929 – Introduction of commercially available soy formula (soy flour)
Mid 1960s – Isolated soy protein introduced

COMPARING INFANT FORMULAS WITH HUMAN MILK 43
history of the development of infant formulas. Liebig’s food for infants was marketed in
1867 and consisted of wheat flour, cow milk, malt flour, and potassium bicarbonate (Fomon,
2001). In 1915 a formula called “synthetic milk adapted” was developed with nonfat cow
milk, lactose, oleo oils, and vegetable oils. This was the basis for modern commercially
prepared formulas (Fomon, 1993).
The limitations of using nonhuman-mammalian milks as substitutes became clear. As
early as 1545, people were concerned with the feeding of animal milks to babies. The Boke of
Chyldren stated that “If children be fed the milk of sheep, then their hair will be soft as that of
a lamb, but if they be fed the milk of the goat, the hair will be coarse” (Phaire, 1955, P. 18).
There are, of course, far greater concerns about feeding animal milk to infants, such as folate
deficiency (goat milk) and early onset hypocalcemic seizures and azotemia (cow milk).
By the early twentieth century it was clear that cow milk was most likely the best
animal-milk base to work from, but that certain modifications were needed to make it safe
and palatable for human infants. These modifications included:
• removing animal fat and substituting vegetable oils,

• diluting the protein content for the newborn’s relatively immature renal tubular
system, and
• adding or balancing minerals and vitamins (e.g., adding iron, adjusting the calcium:
phosphorus ratio).
The process of modifying cow milk for large-scale production in the 1920s repre-
sented the birth of the infant formula industry. Since then new ingredients have been
added for a variety of reasons. For example, iron was added in 1959 to reduce the risk of
iron deficiency in formula-fed infants (Fomon, 1993), and long-chain polyunsaturated
fatty acids (LC-PUFAs) were recently added in an effort to improve infant visual and
cognitive development.
The protein content of formulas was a consideration from about 1935 onward. Early
estimates of human-milk protein levels were higher than is now known, and it was believed
that cow-milk protein was far inferior to human-milk protein. Formulas thus included high
levels of protein (3.3–4.0 g/100 kcal). In the 1960s renal solute load began to be considered
in the design of infant formulas, although infant formula regulations permit higher loads
than are currently recommended by expert panels (no greater than 30 mosm/100 kcal)
(Fomon, 2001).
Based on the recognition that human milk contains a predominance of whey proteins,
while in cow milk, caseins are higher, formulas with a whey:casein ratio similar to human
milk were introduced in 1962. By 2000 whey-predominant formulas were the most widely
used milk-based formulas. These changes were made primarily based on composition rather
than on functional measures (Fomon, 2001).
In 1984 taurine was added to infant formulas, based on at least a decade of studies that
included composition, provisional essentiality, safety, and function in mammals (MacLean
and Benson, 1989). Nucleotides were added to formulas with both compositional and
efficacy claims in the late 1990s. They may act as growth factors and may have immuno-
modulating effects on immune defenses (Carver et al., 1991).
When considering new ingredients, manufacturers analyze every step in the production
process, including raw materials (availability, source, and purity), processing methods, pack-
aging, storage conditions and shelf life, methods of home preparation, and potential for
misuse. Chapter 4 provides a discussion of these manufacturing considerations and their
relevance to the regulatory process.

These considerations continue today as manufacturers attempt to alter infant formulas

to imitate human milk in either composition or performance and to address the nutritional
needs of specific infant populations (e.g., those with cow-milk allergy, metabolic abnormali-
ties, and prematurity) (Benson and Masor, 1994). This chapter is concerned with infant
formulas that are being altered to mimic composition or performance of human milk; it does
not address the nutritional needs of specific infant populations.
Nonmilk-Based Formulas
Soy-based formulas were developed for infants perceived to be intolerant of cow-milk
protein. The first soy formulas were commercially available in 1929 (Abt, 1965). These
formulas were made with soy flour and were not well accepted by parents, who complained
of loose, malodorous stools, diaper rash, and stained clothing. In the mid-1960s isolated soy
protein was introduced into formulas. These formulas were much more like milk-based
formulas in appearance and acceptance. However the preparation of isolated soy protein
resulted in the elimination of most of the vitamin K in the soy, and a few cases of vitamin K
deficiency were reported. The occurrence of nutrient deficiencies in infants fed milk-free
formulas contributed to the development of federal regulations concerning the nutrient
content of formulas (Fomon, 1993). Soy formulas now account for about 40 percent of
formula sales in the United States. Some parents want to avoid cow-milk protein in the diet
and thus wean directly to soy without any reported intolerance to cow-milk formulas. While
formulas containing extensively hydrolyzed protein have long been available for infants with
allergy to intact cow-milk protein, formulas with protein that is not as completely hydro-
lyzed have recently been introduced for normal-term infants.
CHALLENGES OF MATCHING HUMAN-MILK COMPOSITION AND

BREASTFEEDING PERFORMANCE
Infant formula manufacturers have made changes to formulas in order to match either
human milk composition or breastfeeding performance (Benson and Masor, 1994). The
term “breastfeeding performance” is used because, with the exception of one study of
preterm infants (Lucas et al., 1994), all other studies comparing human milk with formulas
involved breastfeeding—not providing human milk from a bottle.
Matching Human-Milk Composition

Historically one approach to match human-milk composition is to add new ingredients
(see Appendix B for the composition of formulas and human milk). This turns out to be a
quixotic quest since human milk is a complex body fluid that is variable not only among
individuals, but within an individual over time. In addition, it contains components, such as
live cells and bioactive compounds, that either cannot be added to formulas or cannot
survive a shelf life. Finally, not all human-milk constituents are essential; some, like LC-
PUFAs, docosahexaenoic acid (DHA), and arachidonic acid (ARA), can be synthesized by
term and preterm infants born at 33 weeks gestation (Uauy et al., 2000).
Manufacturers who wish to add some, but not all, ingredients found in human milk may
defeat the purpose of the added nutrients or may potentiate negative interactions. Examples
include the deleterious effect on growth when eicosapentaenoic acid is added without ad-
equate DHA (Carlson et al., 1996) and the potential negative effect of adding polyunsatu-

rated fats and large amounts of iron without adding adequate antioxidants (Halliwell and
Chirico, 1993; McCord, 1996).
The issue of the context or matrix in which nutrients are provided in milk remains a
challenge to infant formula manufacturers as they try to match human-milk composition
and breastfeeding performance (Benson and Masor, 1994). The matrix can highly influence
the bioavailability of nutrients. In the simplest example, nutrients that are present in both
milks may be present in different ratios. For many nutrients that do not interact chemically
or compete for enzymatic or receptor binding sites, the relative amounts may not be impor-
tant. However in situations where there is competition for enzymes (e.g., among n-3 and n-
6 PUFAs) (Brenner, 1974) or receptor binding sites in the intestine (e.g., for zinc, iron, and
copper), the relative proportions may have biological significance.
Manufacturers must also consider the form of the molecule in which a nutrient is
presented to the intestine and its bioavailability. For example, the high bioavailability of iron
from lactoferrin in human milk allows for a much lower concentration of iron in human
milk (0.2–0.4 mg/L) compared with infant formulas (4.0–12 mg/L) and thereby decreases
competition between iron and other divalent cations, such as copper and zinc (Lonnerdal
and Hernell, 1994).
In the case of LC-PUFAs, care must be taken to ensure no toxicity from these com-
pounds. Manufacturers must study the effects of fats, minerals, enzymes, or other factors on
LC–PUFA bioavailability and processing. For example, newborn fat absorption can be
highly variable because of the immaturity of several lipases, including pancreatic lipase (for
review, see Hamosh, 1988). Human milk contains lipases that compensate for the lack of
pancreatic lipases. Thus human-milk fat is more bioavailable than the vegetable oils found in
infant formulas.
Finally, manufacturers must examine the effects of infant formulas in the context of mixed
feedings (Ryan et al., 2002). Throughout the course of the day, an infant in the United States
may consume both human milk and infant formulas in any number of combinations. For
example, some infants of working mothers are breastfed during the morning and evening and
fed formula during the day by a caregiver. Here the nutrients and their respective matrixes are
kept quite separate and less interaction may be expected than in the situation where an infant
is supplemented with formula directly after each nursing. In the latter case there is a theoretical
concern that certain nutrients found in high concentration in infant formulas (e.g., iron) may
interfere with the intended matrix delivery system found in human milk (e.g., lactoferrin). The
nutritional consequence of mixed-feeding paradigms has not been adequately investigated, but
should be targeted in future studies of the performance of infant formulas.
Matching Breastfeeding Performance

The alternative to matching human-milk composition is to match breastfeeding perfor-
mance (Benson and Masor, 1994). Initially the goal of infant formulas was to match the
growth rate of the breastfed infant. However over time it was recognized that breastfeeding
may confer several other potential advantages to the infant (for review, see AAP, 1997),
including:
• prevention of infectious diseases (Beaudry et al., 1995; Dewey et al., 1995),

• neurodevelopment (Mortensen et al., 2002), and
• protection from chronic diseases in childhood (Saarinen and Kajosaari, 1995; Shu et
al., 1995).

These perceived and potential advantages of breastfeeding are the impetus behind many
of the proposed addition of ingredients to infant formulas. Not all of these advantages are
necessarily attributable to the nutritional content of human milk. Advantages resulting from
a fundamentally different interaction between the nursing mother and her infant or to a
selection bias of mothers who choose to breastfeed cannot be matched by simply adding
nutrients to cow milk. It has been difficult to sort out which of the performance factors of
breastfeeding are due to nutritional components and which are accounted for by social and
psychological factors. Obviously, randomized trials assigning infants to breastfeed or for-
mula feed are not ethically feasible.
Breastfeeding also confers certain risks to the developing infant, including potential
nutrient deficiencies (Kreiter et al., 2000; Pisacane et al., 1995) and exposure to toxins
secreted by the mother into her milk. Advantages and risks are discussed in detail below.
PERFORMANCE ADVANTAGES OF BREASTFEEDING

Breastfed infants have different growth characteristics compared with formula-fed in-
fants. They grow at slightly different rates and have a different body composition (Butte et
al., 1990; Heinig et al., 1993) and may have a lower risk for later obesity (Gillman et al.,
2001; Singhal et al., 2002). (These characteristics are discussed in greater detail in Chapter
6.) Given the great interest in the effect of early nutrition on metabolic setpoints that may
affect the child’s risk for adult diseases (e.g., the early origins of chronic disease hypothesis)
(Barker et al., 2002) and the increasing incidence of early insulin resistance, obesity, and
type II diabetes in teenagers, future research should concentrate on whether breastfeeding is
protective.
As discussed earlier, breastfed infants absorb fat better than formula-fed infants due to
the presence of lipases in human milk that are not present in cow milk (Hamosh, 1988). The
healthy breastfed infant consumes less milk (approximately 85 kcal/kg body weight/day)
during the first months of life than the same infant given ad libitum infant formula (100
kcal/kg/day; Heinig et al., 1993). The breastfed infant continues to consume approximately
10 fewer kcal/kg/body weight calories than the formula-fed infant. The breastfed infant has
a lower total energy expenditure (Butte et al., 1990) and a slower growth rate (Butte et al.,
1990; Heinig et al., 1993). In addition, there is less gastro-esophageal reflux in breastfed
infants, most likely due to a more rapid gastric emptying time, resulting in less loss of intake.
Some of the trophic and metabolic factors that promote the characteristic nutrient handling
and growth of the breastfed infant are listed in Table 3-1.
Breastfed infants, compared with formula-fed infants, have a lower incidence of infec-
tious diseases, such as diarrhea (Popkin et al., 1990), otitis media (Duncan et al., 1993), and
lower respiratory tract illness (Wright et al., 1989). The effect is particularly profound in the
developing world, but studies show clear advantages in the developed world as well (Wright
et al., 1989). The effect extends beyond breastfeeding itself to when human milk is adminis-
tered without the infant nursing from the mother. For example, preterm infants fed human
milk by nasogastric tube in the newborn intensive care unit have a lower rate of necrotizing
enterocolitis (Lucas and Cole, 1990). Moreover, the presence of the close contact between
the mother and child stimulates the mother to make antibodies against bacteria colonized in
the infant and to secrete these antibodies in her milk.
Human milk has multiple components that likely mediate this anti-infectious, immuno-
logically enhancing effect. These include secretory immunoglobulin A, lactoferrin, lysozymes,
intact cellular components, and oligosaccharides. A comprehensive list of compounds found
in human milk by class of ingredient is shown in Table 3-2.

TABLE 3-1 Unique Factors in Human Milk That Positively Affect Nutritional Status
and Somatic Growth
Ingredient Class of Ingredient Function Reference
Amylase Enzyme Polysaccharide digestion Howell et al., 1986
Epidermal growth Growth factor/hormone Gastrointestinal growth/ Donovan and Odle,
factor differentiation 1994; Dvorak et al.,
2003; Howell et al.,
1986
Erythropoietin Growth factor/hormone Red cell production; possible Kling, 2002
growth factor for gut and
central nervous system
Insulin Growth factor/hormone Anabolic hormone promotes Donovan and Odle,
carbohydrate, protein, and 1994
fat accretion
Insulin-like growth Growth factor/hormone Primary growth hormone of Donovan and Odle,
factor-I late fetal/neonatal period 1994
Insulin-like growth Growth factor/hormone Unknown function; thought Donovan and Odle,
factor-II to be weak growth hormone 1994
Lactoferrin Carrier protein Increases efficiency of iron Howell et al., 1986
delivery
Lipase Enzyme Triglyceride hydrolysis Howell et al., 1986
Nerve growth factor Growth factor/hormone Neuronal growth/ Donovan and Odle,
differentiation 1994
Proteases Enzyme Unknown if active in protein Howell et al., 1986
hydrolysis
Relaxin Growth factor/hormone Regulates morphological Donovan and Odle,
development of the nipple 1994
Transforming growth Growth factor/hormone Gastrointestinal growth Donovan and Odle,
factor-alpha 1994; Dvorak et al.,
2003
TABLE 3-2 Unique Factors in Human Milk with Anti-Infective or Immunological

Properties
Class of
Ingredient Ingredient Function Reference
Antiproteases (e.g., Enzyme Inhibits breakdown of anti- Howell et al., 1986;
secretary immuno- infective immunoglobulins IOM, 1991
globulin A and and enzymes
trypsin inhibitor)
Arylsulfatase Enzyme Degrades leukotrienes Hanson et al., 1988
Catalase Enzyme Degrades hydrogen peroxide; Lindmark-Mansson
protects against bacterial and Akesson, 2000
breeches of intestinal barrier
Fibronectin Opsonin May present debris to macrophages IOM, 1991;
Mestecky et al.,
1990
Free fatty acids Lipids Antiviral (coronavirus); antiparasitic Mestecky et al., 1990
(Giardia, Entamoeba)
Granulocyte-colony Cytokine Causes endothelial cell migration Wallace et al., 1997
stimulating factor and proliferation
Hemagglutinin inhibitor Opsonin Prevents bacterial adherence Neeser et al., 1988
Continued

TABLE 3-2 continued

Class of
Ingredient Ingredient Function Reference
Histaminase Enzyme Degrades histamine Hanson et al.,1988
Immunoglobulin G Immunoglobin Immune protection Howell et al., 1986;
IOM, 1991
Interleukin-1-beta Cytokine Activates T-cells Mestecky et al., 1990
Interleukin-6 Cytokine Enhances immunoglobulin A Mestecky et al., 1990
and C-reactive protein production
Interleukin-8 Cytokine Chemotaxis Maheshwari et al.,
2002
Interleukin-10 Cytokine Decreases inflammatory cytokine Goldman et al., 1996
synthesis
Lactadherin Protein Prevents rotavirus binding Peterson et al., 2001
Lactoferrin Carrier Anti-infective; may prevent iron Howell et al., 1986;
from being bioavailable to IOM, 1991
microbes
Leukocytes Live cell Cytokine production by T-cells; IOM, 1991;
direct in vivo roles of B-cells, Mestecky et al.,
macrophages, and neutrophils 1990
Lipases Enzyme Releases bacteriostatic and Howell et al., 1986;
bacteriocidal free fatty acids IOM, 1991
Lysozyme Enzyme Bactericidal Howell et al., 1986;
IOM, 1991
Macrophage colony Cytokine Macrophage proliferation Goldman et al., 1986
stimulating factor
Mucin Protein Inhibits E. coli binding Peterson et al., 2001
to gut epithelium
Oligosaccharides, Carbohydrates, Receptor analogs block Coppa et al., 1999;
polysaccharides, glycoconjugates binding of enteric bacteria; IOM, 1991;
gangliosides growth promoters for Rivero-Urgell and
Lactobacillus Santamaria-
Orleans, 2001
Peroxidases Enzyme Bactericidal Howell et al., 1986;
IOM, 1991
Platelet activating Enzyme Catabolizes platelet Furukawa et al.,
acetyl hydrolase activator factor 1993
factor
Prostaglandin E2, Prostaglandin Intestinal cytoprotection Howell et al., 1986
F2-alpha
Ribonuclease Enzyme Prevents viral replication Nevinsky and
Buneva, 2002
Secretory Immunoglobulin Immune protection (broad Howell et al., 1986;
immunoglobulin A spectrum antiviral, antibacterial, IOM, 1991
antiparasitic)
Soluble intracellular Cytokine Alters adhesion of viral or Xyni et al., 2000
adhesion molecule-1 other molecules to intestinal
epithelium
Transforming growth Cytokine Produces immunoglobulin A Bottcher et al., 2000
factor-beta and activates B-cells
Tumor necrosis Cytokine Mobilizes amino acids Mestecky et al., 1990
factor-alpha
Uric acid Small molecular- Antioxidant Van Zoeren-Grobben
weight et al., 1994
nitrogenous
compound

The neurodevelopmental advantages of breastfeeding or supplying infants with human

milk have received significant amounts of attention (Lucas et al., 1998; Morrow-Tlucak et
al., 1988; Mortensen et al., 2002; Wang and Wu, 1996). Indeed, the primary impetus for
adding LC–PUFAs to infant formulas is their postulated effect on brain development. The
general research on breastfeeding, human milk, and neurodevelopment is fraught with
confounding variables that have prevented pinpointing specific nutrients that are responsible
for the putative effect. Overall it appears that breastfed infants have modest improvements
in cognitive, motor, and visual status up to the age of 8 years, but it is unclear whether any
early effects disappear over time (for review, see Grantham-McGregor et al., 1999). The
degree of neurodevelopmental advantage is directly related to duration of breastfeeding
(Mortensen et al., 2002). However critics of the literature point out that there may be
fundamental differences not only between mothers who do or do not choose to breastfeed,
but also between those who choose to breastfeed for a longer rather than shorter time
period. These selection biases may be based on characteristics (e.g., maternal IQ, education,
and socioeconomic status) that may confer independent positive effects on the neuro-
development of the infant. Furthermore, patterns of parent-child interactions may be differ-
ent in those who breastfeed longer; these interactions may have effects on development.
Just as it is difficult to separate out the confounding social factors among those who do
and do not choose to breastfeed, it is also difficult to isolate the role of nutrition alone in the
assessment of the positive effects. This is because very few individuals bottle-feed their
infants human milk and, when this is done, it is frequently for medically extenuating circum-
stances (e.g., prematurity). Thus one cannot expect to rely on randomized trials of breast-
feeding versus formula feeding or breastfeeding versus bottle feeding of human milk to sort
out the nutritional effects of human milk on the developing brain. The only trial that
approached this issue was conducted by Lucas and coworkers (1994), where preterm infants
received either human milk or term infant formula by gavage tube during their early weeks.
Infants fed bottled human milk had higher mental and psychomotor development indices 18
months after hospital discharge. However it should be reiterated that these were premature
infants and that they were not randomized to their particular diets.
Nevertheless there are reasons to think that the provision of human milk, based on its
composition, is good for the human brain. Human milk contains LC–PUFAs (e.g., DHA and
ARA) that are important for synaptogenesis in the visual system. However studies assessing
the addition of these ingredients to cow-milk formula have not resulted in consistent effects.
Some demonstrated enhanced visual acuity and speed of processing in infants fed the supple-
mented formulas (Uauy et al., 1990; for review, see Uauy-Dagach and Mena, 1995). The
positive effects on visual acuity have been found most often in premature infants, who are
arguably more deficient of these fats. There may be effects on cognitive outcome, although
the effects are inconsistent, particularly in term infants (Auestad et al., 2001; Wroble et al.,
2002). The reason for these inconsistent effects might be that these compounds do not work
alone; rather the matrix of human milk includes general growth factors and specific neural
growth factors (see Table 3-3). If there is a positive effect on neurodevelopment, it is likely
that these factors work in concert with each other.
Finally, there is epidemiological evidence that breastfeeding protects infants from cer-
tain childhood diseases at older ages, including atopy/allergy (Kull et al., 2002; Saarinen and
Kajosaari, 1995), obesity (Gillman et al., 2001; Singhal et al., 2002), and childhood leuke-
mia/lymphoma (Shu et al., 1995). The biological mechanisms of the positive effects are not
always clear, but may relate to avoidance of exposure to antigenic proteins found in cow
milk, particularly in relation to allergy. The lack of clear biological mechanisms makes it
more difficult to resolve conflicting results, such as those recently indicating an increased

TABLE 3-3 Unique Factors in Human Milk That May Positively Affect
Neurodevelopment
Ingredient Class of Ingredient Function Reference
Choline Amino acid Neurotransmitter Zeisel et al., 1986
synthesis
Insulin-like growth factor-1 Growth factor/hormone Neuronal growth/ Cheng et al., 2003;
differentiation Donovan and Odle,
1994
Long-chain polyunsaturated Essential/semiessential fat Visual acuity Uauy-Dagach and
fatty acids Mena, 1995
Nerve growth factor Growth factor/hormone Neuronal growth/ Donovan and Odle, 1994
differentiation
Oligosaccharides (fucose, Carbohydrates Neuronal cell-cell Hynes et al., 1989
mannose, n-acetylglucosa- communication
mine, sialic acid)
risk of atopy (Sears et al., 2002) and eczema (Bergmann et al., 2002) in large cohorts of
breastfed infants.
RISKS OF BREASTFEEDING
Breastfeeding is not without potential nutritional risks. The best documented risks
include iron deficiency (Duncan et al., 1985; Pisacane et al., 1995), vitamin D deficiency
(Kreiter et al., 2000), and exposure to environmental toxins. The inability to sustain growth
due to the low energy density of milk is relatively rare in the first 4 months of life in the
breastfed infant. However there is great variability in the protein-energy density of human
milk. Energy values may range from 15 to 24 kcal/oz. Most infants can overcome a lower-
density milk by consuming a greater volume.
Iron deficiency is approximately twice as common in breastfed infants; up to 30 percent
have iron deficiency anemia, and more than 60 percent of the anemic infants are also iron
deficient at 12 months of age (Pisacane et al., 1995), although the etiology is unclear. The
iron content of human milk is low: 0.5 mg/L compared with 10 to 12 mg/L in supplemented
cow-milk formulas. The absorption rate, however, is considerably higher. Breastfed infants
absorb up to 50 percent of consumed iron, compared with a 7- to 12-percent absorption rate
for formula-fed infants (Fomon et al., 1993). The risk of iron deficiency increases after 4
months of age since most full-term infants are born with adequate iron stores to support
hemoglobin synthesis through the first 4 months after birth.
There have been increasing reports of nutritional rickets in breastfed infants, particu-
larly in northern climates (Kreiter et al., 2000). This is likely due to lack of sunlight expo-
sure, which is increasingly common with the use of sunscreens and the tendency to cover
infants for health or cultural reasons. Human milk, like cow milk, is very low in vitamin D,
with average concentrations of 24 to 68 IU/L. Since infants consume less than 0.5 L of milk/
day in the first months of life, breastfed infants have vitamin D intake well below the
Adequate Intake of 200 IU/day. With sun exposure this is not likely to be a problem.
However infants born to mothers with vitamin D deficiency are at increased risk for rickets,
as are those who are not exposed to the sun. The American Academy of Pediatrics and the
Canadian Paediatric Society recently recommended supplementing all breastfed infants with
200 IU of vitamin D by 2 months of age (AAP, 2003; Canadian Paediatric Society, 1998).

In addition to transplacental passage of environmental allergens and dietary antigens, it

is possible that susceptible infants may be sensitized to such agents by exposure to maternal
milk. Although dietary antigens have been recovered in human milk, and allergen-specific
IgE antibodies have been demonstrated in cord blood (Fälth-Magnusson, 1995; Lilja et al,
1988), available evidence suggests little or no role for breastmilk-associated food antigens
in the development of food allergy (Businco et al., 1983; Fälth-Magnusson, 1995; Fälth-
Magnusson and Kjellman, 1987).
Breastfed infants can be exposed to environmental toxins (e.g., lead and polychlorinated
biphenyls), legal and illegal drugs, and infectious pathogens that the mother may harbor
(e.g., Human Immunodeficiency Virus [HIV]). A discussion of all of the potential environ-
mental toxins, drugs, and infectious agents is beyond the scope of this chapter. However it
is important to note the effect of increasing rates of HIV infection worldwide and the
potential for human milk to be both a vector of transmission of the virus from mother to
infant and to contain protective anti-infective factors that may decrease the risk of vertical
transmission. These risks and benefits must be weighed against the potential risks of formula
feeding, not the least of which is preparation of formula with water contaminated with
infectious agents (Humphrey and Iliff, 2001; Mbori-Ngacha et al., 2001; WHO, 1992).
SUMMARY
This chapter affirms that breastfeeding is the standard by which all other infant-feeding
methods should be judged. This position has been taken by numerous professional bodies
and reflects the fact that human milk is species specific and thus uniquely suited for human
infant nutrition. It must be recognized, however, that using a human-milk composition or
breastfeeding performance standard presents both regulatory and research issues when as-
sessing the addition of ingredients new to infant formulas.
From a research standpoint, clinical studies that assess the effects of new ingredients will
be difficult to design because infants cannot be randomized to be formula fed or breastfed.
Furthermore, there may be significant non-nutritional confounding variables between the
groups, including, but not limited to, factors related to which mothers breastfeed. Finally,
human-milk composition varies considerably among individuals and within individuals over
time, while infant formula content remains constant.
The committee anticipates that manufacturers will wish to add both ingredients that are
currently contained in human milk, but not in formulas (e.g., LC-PUFAs), and those not
found in human milk (e.g., prebiotics) to enhance the performance of formulas to a level at
or nearer to human milk. Thus a breastfed control group should be part of experimental
designs to assess the addition of ingredients new to infant formulas in order to provide a
performance standard.
From a regulatory standpoint, the effect of an ingredient new to infant formulas is
usually driven by a manufacturer’s desire to produce products that mimic the advantages of
breastfeeding. This motivation implies that formula in its current state is inferior (e.g.,
relatively neurologically or immunologically less beneficial, although not necessarily unsafe)
when compared with human milk. Thus the safety (and efficacy) of any addition of an
ingredient new to infant formulas will need to be judged against two control groups: one fed
the previous iteration of the formula without the added ingredient, and one breastfed.

REFERENCES
AAFP (American Academy of Family Physicians). 2003. Breastfeeding (Position Paper). Online. Available at http:
//www.aafp.org/x6633.xml?printxml. Accessed February 5, 2003.
AAP (American Academy of Pediatrics). 1997. Breast feeding and the use of human milk. Pediatrics 100:1035–
1039.
AAP. 2003. Prevention of rickets and vitamin D deficiency: New guidelines for vitamin intake. Pediatrics 111:908–
910.
Abt IA, ed. 1965. Allergy. In: History of Pediatrics. Philadelphia: W.B. Saunders. Pp. 259–260.
ADA (American Dietetic Association). 2001. Position of the American Dietetic Association: Breaking the barriers
to breastfeeding. J Am Diet Assoc 101:1213–1220.
Auestad N, Halter R, Hall RT, Blatter M, Bogle ML, Burks W, Erickson JR, Fitzgerald KM, Dobson V, Innis SM,
Singer LT, Montalto MB, Jacobs JR, Qiu W, Bornstein MH. 2001. Growth and development in term infants
fed long-chain polyunsaturated fatty acids: A double-masked, randomized, parallel, prospective, multivariate
study. Pediatrics 108:372–381.
Barker DJP, Eriksson JG, Forsen T, Osmond C. 2002. Fetal origins of adult disease: Strength of effects and
biological basis. Int J Epidemiol 31:1234–1239.
Beaudry M, Dufour R, Marcoux S. 1995. Relation between infant feeding and infections during the first six
months of life. J Pediatr 126:191–197.
Benson JD, Masor ML. 1994. Infant formula development: Past, present and future. Endocr Regul 28:9–16.
Bergmann RL, Diepgen TL, Kuss O, Bergmann KE, Kujat J, Dudenhausen JW, Wahn U. 2002. Breastfeeding
duration is a risk factor for atopic eczema. Clin Exp Allergy 32:205–209.
Bottcher MF, Jenmalm MC, Garofalo RP, Bjorksten B. 2000. Cytokines in breast milk from allergic and nonallergic
mothers. Pediatr Res 47:157–162.
Brenner RR. 1974. The oxidative desaturation of unsaturated fatty acids in animals. Mol Cell Biochem 3:41–52.
Businco L, Marchetti F, Pellegrini G, Perlini R. 1983. Predictive value of cord blood IgE levels in “at risk”
newborn babies and influence of type of feeding. Clin Allergy 13:503–508.
Butte NF, Wong WW, Ferlic L, Smith EO, Klein PD, Garza C. 1990. Energy expenditure and deposition of breast-
fed and formula-fed infants during early infancy. Pediatr Res 28:631–640.
Canadian Paediatric Society, Dietitians of Canada and Health Canada. 1998. Nutrition for Healthy Term Infants.
Ottawa: Minister of Public Works and Government Services.
Carlson SE, Werkman SH, Tolley EA. 1996. Effect of long-chain n-3 fatty acid supplementation on visual acuity
and growth of preterm infants with and without bronchopulmonary dysplasia. Am J Clin Nutr 63:687–697.
Carver JD, Pimentel B, Cox WI, Barness LA. 1991. Dietary nucleotide effects upon immune function in infants.
Pediatrics 88:359–363.
Cheng CM, Mervis RF, Niu SL, Salem N, Witters LA, Tseng V, Reinhardt R, Bondy CA. 2003. Insulin-like growth
factor 1 is essential for normal dendritic growth. J Neurosci Res 73:1–9.
Coppa GV, Pierani P, Zampini L, Carloni I, Carlucci A, Gabrielli O. 1999. Oligosaccharides in human milk during
different phases of lactation. Acta Paediatr Suppl 430:89–94.
Dewey KG, Heinig MJ, Nommsen-Rivers LA. 1995. Differences in morbidity between breast-fed and formula-fed
infants. J Pediatr 126:696–702.
Donovan SM, Odle J. 1994. Growth factors in milk as mediators of infant development. Annu Rev Nutr 14:147–
167.
Duncan B, Schifman RB, Corrigan JJ Jr, Schaefer C. 1985. Iron and the exclusively breast-fed infant from birth to
six months. J Pediatr Gastroenterol Nutr 4:421–425.
Duncan B, Ey J, Holberg CJ, Wright AL, Martinez FD, Taussig LM. 1993. Exclusive breast-feeding for at least 4
months protects against otitis media. Pediatrics 91:867–872.
Dvorak B, Fituch CC, Williams CS, Hurst NM, Schanler RJ. 2003. Increased epidermal growth factor levels in
human milk of mothers with extremely premature infants. Pediatr Res 54:15–19.
Fälth-Magnusson K. 1995. Dietary restrictions during pregnancy. In: de Weck AL, Sampson HA, eds. Intestinal
Immunology and Food Allergy. Nestlé Nutrition Workshop Series. Vol. 34. New York: Raven Press. Pp.
191–201.
Fälth-Magnusson K, Kjellman NI. 1987. Development of atopic disease in babies whose mothers were receiving
exclusion diet during pregnancy—A randomized study. J Allergy Clin Immunol 80:868–875.
Fomon SJ. 1993. Nutrition of Normal Infants. St. Louis: Mosby-Year Book.
Fomon SJ. 2001. Infant feeding in the 20th century: Formula and Beikost. J Nutr 131:409S–420S.
Fomon SJ, Ziegler EE, Nelson SE. 1993. Erythrocyte incorporation of ingested 58Fe by 56-day-old breast-fed and
formula-fed infants. Pediatr Res 33:573–576.


Furukawa M, Narahara H, Yasuda K, Johnston JM. 1993. Presense of platelet-activating factor-acetylhydrolase in
milk. J Lipid Res 34:1603–1609.
Gillman MW, Rifas-Shiman SL, Camargo CA Jr, Berkey CS, Frazier AL, Rockett HR, Field AE, Colditz GA. 2001.
Risk of overweight among adolescents who were breastfed as infants. J Am Med Assoc 285:2461–2467.
Goldman AS, Chheda S, Garofalo R, Schmalstieg FC. 1996. Cytokines in human milk: Properties and potential
effects upon the mammary gland and the neonate. J Mammary Gland Biol Neoplasia 1:251–258.
Grantham-McGregor SM, Fernald LC, Sethuraman K. 1999. Effects of health and nutrition on cognitive and
behavioral development in children in the first three years of life. Part 1: Low birthweight, breastfeeding, and
protein-energy malnutrition. Food Nutr Bull 20:53–75.
Halliwell B, Chirico S. 1993. Lipid peroxidation: Its mechanism, measurement, and significance. Am J Clin Nutr
57:715S–725S.
Hamosh M. 1988. Fat needs for term and preterm infants. In: Tsang RC, Nichols BL, eds. Nutrition During
Infancy. Philadelphia: Hanley and Belfus. Pp. 133–159.
Hanson LA, Carlsson B, Jalil F, Hahn-Zoric M, Hermodson S, Karlberg J, Mellander L, Khan SR, Lindblad B,
Thiringer K, Zaman S. 1988. Antiviral and antibacterial factors in human milk. In: Hanson LA, ed. Biology
of Human Milk. Nestlé Nutrition Workshop Series. Vol. 15. New York: Raven Press. Pp. 141–157.
Heinig MJ, Nommsen LA, Peerson JM, Lonnerdal B, Dewey KG. 1993. Energy and protein intakes of breast-fed
and formula-fed infants during the first year of life and their association with growth velocity: The DAR-
LING Study. Am J Clin Nutr 58:152–161.
HHS/OWH (U.S. Department of Health and Human Services/Office on Women’s Health). 2000. HHS Blueprint
for Action on Breastfeeding. Washington, DC: DHHS.
Howell RR, Morriss FH, Pickering LK, eds. 1986. Human Milk in Infant Nutrition and Health. Springfield, IL:
Charles C. Thomas.
Humphrey J, Iliff P. 2001. Is breast not best? Feeding babies born to HIV-positive mothers: Bringing balance to a
complex issue. Nutr Rev 59:119–127.
Hynes MA, Buck LB, Gitt M, Barondes S, Dodd J, Jessell TM. 1989. Carbohydrate recognition in neuronal
development: Structure and expression of surface oligosaccharides and beta-galactoside-binding lectins. Ciba
Found Symp 145:189–210.
IOM (Institute of Medicine). 1991. Nutrition During Lactation. Washington, DC: National Academy Press.
Kling PJ. 2002. Roles of erythropoietin in human milk. Acta Paediatr Suppl 91:31–35.
Kreiter SR, Schwartz RP, Kirkman HN Jr, Charlton PA, Calikoglu AS, Davenport ML. 2000. Nutritional rickets
in African American breast-fed infants. J Pediatr 137:153–157.
Kull I, Wickman M, Lilja G, Nordvall SL, Pershagen G. 2002. Breast feeding and allergic diseases in infants—A
prospective birth cohort study. Arch Dis Child 87:478–481.
Lilja G, Dannaeus A, Fälth-Magnusson K, Graff-Lonnevig V, Johansson SGO, Kjellman N-IM, Öman H. 1988.
Immune response of the atopic woman and foetus: Effects of high- and low-dose food allergen intake during
pregnancy. Clin Allergy 18:131–142.
Lindmark-Mansson H, Akesson B. 2000. Antioxidative factors in milk. Br J Nutr 84:S103–S110.
Lonnerdal B, Hernell O. 1994. Iron, zinc, copper and selenium status of breast-fed infants and infants fed trace
element fortified milk-based infant formula. Acta Paediatr 83:367–373.
LSRO (Life Sciences Research Office). 1998. LSRO report: Assessment of nutrient requirements for infant formu-
las. J Nutr 128:2059S–2293S.
Lucas A, Cole TJ. 1990. Breast milk and neonatal necrotizing enterocolitis. Lancet 336:1519–1523.
Lucas A, Morley R, Cole TJ, Gore SM. 1994. A randomised multicentre study of human milk versus formula and
later development in preterm infants. Arch Dis Child 70:F141–F146.
Lucas A, Morley R, Cole TJ. 1998. Randomised trial of early diet in preterm babies and later intelligence quotient.
Br Med J 317:1481–1487.
MacLean WC Jr, Benson JD. 1989. Theory into practice: The incorporation of new knowledge into infant for-
mula. Semin Perinatol 13:104–111.
Maheshwari A, Lu W, Lacson A, Barleycorn AA, Nolan S, Christensen RD, Calhoun DA. 2002. Effects of
interleukin-8 on the developing human intestine. Cytokine 20:256–267.
Mbori-Ngacha D, Nduati R, John G, Reilly M, Richardson B, Mwatha A, Ndinya-Achola J, Bwayo J, Kreiss J.
2001. Morbidity and mortality in breastfed and formula-fed infants of HIV-1-infected women: A randomized
clinical trial. J Am Med Assoc 286:2413–2420.
McCord JM. 1996. Effects of positive iron status at a cellular level. Nutr Rev 54:85–88.
Mestecky J, Blair C, Ogra PL, eds. 1990. Immunology of milk and the neonate. Adv Exp Med Biol 310:1–177.
Morrow-Tlucak M, Haude RH, Ernhart CB. 1988. Breastfeeding and cognitive development in the first 2 years of
life. Soc Sci Med 26:635–639.

Mortensen EL, Michaelsen KF, Sanders SA, Reinisch JM. 2002. The association between duration of breastfeeding
and adult intelligence. J Am Med Assoc 287:2365–2371.
Neeser JR, Chambaz A, Del Vedovo S, Prigent MJ, Guggenheim B. 1988. Specific and nonspecific inhibition of
adhesion of oral actinomyces and streptococci to erythrocytes and polystyrene by caseinoglycopeptide deriva-
tives. Infect Immun 56:3201–3208.
Nevinsky GA, Buneva VN. 2002. Human catalytic RNA- and DNA-hydrolyzing antibodies. J Immunol Methods
269:235–249.
Peterson JA, Scallan CD, Geriani RL, Hamosh M. 2001. Structural and functional aspects of three major glyco-
proteins of the human milk fat globule membrane. Adv Exp Med Biol 501:179–187.
Phaire T. 1955. In: Neale AV, Wallis HRE, eds. The Boke of Chyldren. Edinburgh, London: E & S Livingstone.
P. 18.
Pisacane A, De Vizia B, Valiante A, Vaccaro F, Russo M, Grillo G, Giustardi A. 1995. Iron status in breast-fed
infants. J Pediatr 127:429–431.
Popkin BM, Adair L, Akin JS, Black R, Briscoe J, Flieger W. 1990. Breast-feeding and diarrheal morbidity.
Rivero-Urgell M, Santamaria-Orleans A. 2001. Oligosaccharides: Application in infant food. Early Hum Dev
65:S43–S52.
Ryan AS, Wenjun Z, Acosta A. 2002. Breastfeeding continues to increase into the new millennium. Pediatrics
110:1103–1109.
Saarinen UM, Kajosaari M. 1995. Breastfeeding as prophylaxis against atopic disease: Prospective follow-up study
until 17 years old. Lancet 346:1065–1069.
Sears MR, Greene JM, Willan AR, Taylor DR, Flannery EM, Cowan JO, Herbison GP, Poulton R. 2002. Long-
term relation between breastfeeding and development of atopy and asthma in children and young adults: A
longitudinal study. Lancet 360:901–907.
Shu XO, Clemens J, Zheng W, Ying DM, Ji BT, Jin F. 1995. Infant breastfeeding and the risk of childhood
lymphoma and leukaemia. Int J Epidemiol 24:27–32.
Singhal A, Farooqi IS, O’Rahilly S, Cole TJ, Fewtrell M, Lucas A. 2002. Early nutrition and leptin concentrations
in later life. Am J Clin Nutr 75:993–999.
Uauy-Dagach R, Mena P. 1995. Nutritional role of omega-3 fatty acids during the perinatal period. Clin Perinatol
22:157–175.
Uauy RD, Birch DG, Birch EE, Tyson JE, Hoffman DR. 1990. Effect of dietary omega-3 fatty acids on retinal
function of very-low-birth-weight neonates. Pediatr Res 28:485–492.
Uauy R, Mena P, Wegher B, Nieto S, Salem N Jr. 2000. Long chain polyunsaturated fatty acid formation in
neonates: Effect of gestational age and intrauterine growth. Pediatr Res 47:127–135.
van Zoeren-Grobben D, Lindeman JHN, Houdkamp E, Brand R, Schrijver J, Berger HM. 1994. Postnatal changes
in plasma chain-breaking antioxidants in healthy preterm infants fed formula and/or human milk. Am J Clin
Nutr 60:900–906.
Wallace JMW, Ferguson SJ, Loane P, Kell M, Millar S, Gillmore WS. 1997. Cytokines in human breast milk. Br J
Biomed Sci 54:85–87.
Wang YS, Wu SY. 1996. The effect of exclusive breastfeeding on development and incidence of infection in
infants. J Hum Lact 12:27–30.
WHO (World Health Organization). 1992. Consensus statement from the consultation on HIV transmission and
breastfeeding. J Hum Lact 8:173–174.
WHO. 2002. The Optimal Duration of Exclusive Breastfeeding. Report of an Expert Consultation. Geneva:
WHO.
Wright AL, Holberg CJ, Martinez FD, Morgan WJ, Taussig LM. 1989. Breast feeding and lower respiratory tract
illness in the first year of life. Group Health Medical Associates. Br Med J 299:946–949.
Wroble M, Mash C, Williams L, McCall RB. 2002. Should long chain polyunsaturated fatty acids be added to
infant formula to promote development? Appl Dev Psychol 23:99–112.
Xyni K, Rizos D, Giannaki G, Sarandakou S, Phocas I, Creatsas G. 2000. Soluble form of ICAM-1, VCAM-1, E-
and L-selectin in human milk. Mediators Inflamm 9:133–140.
Zeisel SH, Char D, Sheard NF. 1986. Choline, phosphatidylcholine and sphingomyelin in human and bovine milk
and infant formulas. J Nutr 116:50–58.

Strengthening the Current Processes to Evaluate

New Ingredients for Infant Formulas
ABSTRACT
In the United States the Food and Drug Administration (FDA) is charged with
monitoring the safety of infant formulas. If a manufacturer wishes to add a new
ingredient to infant formulas, it must either file a Food Additive Petition or declare
it a Generally Recognized as Safe (GRAS) substance through a GRAS Notification
process. The GRAS Notification has become the route of choice for the introduction
of new ingredients because it is scientifically rigorous, far more efficient, and equally
or more transparent than the Food Additive Petition route. The committee found
that the current regulatory system is limited by the following factors: (1) the com-
plexity of the system has resulted in regulations that are not well understood by the
scientific or medical community; (2) the system does not appear to adequately ad-
dress the uniqueness of infancy and infant nutrition; (3) if a panel of experts is used
to show consensus, there is a lack of guidelines for selecting a qualified and unbiased
expert panel to evaluate the safety of the proposed new ingredient; and (4) formal
guidelines for in-market surveillance do not exist for infant formulas.
The committee recommends that a set of guidelines be developed to provide a
hierarchy of decision-making steps for manufacturers wishing to add new ingredi-
ents to infant formulas. In addition, elements of the safety assessments of infant
formulas need to be standardized. Along with growth and development, other qual-
ity factors to be considered for these safety assessments are tolerance, allergenicity,
impact of gastrointestinal flora, interference with bioavailability of other nutrients,
and possible nutrient imbalances (if ratios or cofactors are important). To review the
existing data and identify additional studies needed, the manufacturer should estab-
lish balanced, qualified expert panels in consultation with the regulatory agency.
Also, the manufacturer should implement an appropriate in-market surveillance
strategy that is based on findings from preclinical and clinical studies and the poten-
tial for harm to infants.
55

INTRODUCTION
Existing guidelines and regulations for evaluating the safety of conventional food ingre-
dients (e.g., vitamins and minerals) added to infant formulas have been adequate in the past;
however they were not designed to address the unique needs and vulnerabilities of infants
and potential new types of ingredients. This chapter provides an overview of the current
regulatory system with a discussion of the existing laws; definitions of safety, quality, food,
drugs, and dietary supplements; and routes to add new ingredients to infant formulas in the
United States. The committee then describes four major limitations of the current system.
Finally, an approach is presented to assess the safety of new ingredients to be added to infant
formulas, using the current GRAS Notification process as a starting model. The committee
suggests that these guidelines could be applied in other countries as regulation parameters.
THE CURRENT U.S. REGULATORY SYSTEM

In the United States infant formulas and ingredients added to infant formulas fall within
the purview of the Office of Food Additive Safety and the Office of Nutritional Products,
Labeling and Dietary Supplements of FDA’s Center for Food Safety and Applied Nutrition.
The safety of new ingredients added to infant formulas is regulated under Section 409 of the
Federal Food, Drug and Cosmetic (FD&C) Act (21 U.S.C. §348), which was the primary
focus of the committee.
The Infant Formula Acts of 1980 (P.L. 96-359) and 1986 (P.L. 99-570) were incorpo-
rated into the FD&C Act as Section 412 (21 U.S.C §350a), which deals with proper manu-
facturing, formulation, and quality factors of infant formulas. This chapter will focus mainly
on safety issues of ingredients (Section 409), but it should be noted that some aspects of
Section 412 (quality factors) also play a critical role in the safety assessment process.
Foods, Dietary Supplements, and Drugs as Defined by

the Federal Food, Drug and Cosmetic Act
The definition of foods, dietary supplements, and drugs and their respective safety
standards are essential to the understanding of the regulatory differences between these
classes of bioactive materials. It is beyond the scope of this chapter to delve into the rationale
behind the regulations, but the concept of intended use is a critical differentiator between
foods, dietary supplements, and drugs, which are defined in the following manner:
• Foods are “articles or components of articles used for food or drink for humans or
animals” (21 U.S.C. §321 (f)(1)).
• Drugs are “articles intended for the diagnosis, cure, mitigation, treatment, or preven-
tion of disease in man or other animals” (21 U.S.C. §321 (g)(1)(b)).
• Dietary Supplements are products “. . . intended to supplement the diet that bear or
contain one or more of the following dietary ingredients: a vitamin; a mineral; an herb or
other botanical; an amino acid; a dietary substance for use by man to supplement the diet by
increasing the total dietary intake; or a concentrate, metabolite, constituent, extract or
combination of any ingredient described above” (21 U.S.C. §321 (ff)(2)).
Table 4-1 provides a comparison of the safety standards for foods, drugs, and dietary
supplements. Infant formulas are considered food and, thus, a risk-benefit analysis would be
inappropriate because foods are safe for everyone—young or old, male or female, healthy or

STRENGTHENING THE CURRENT PROCESSES 57
TABLE 4-1 Comparison of Safety Standards Included in Federal Regulations

Article Safety Standard Safety Decision
Foods or Food Additives Reasonable certainty of no harm Food Additive Petition—
(Federal Food, Drug and Food and Drug Administration
Cosmetic Act) (FDA) must approve that the
material is safe
Generally Recognized as Safe—
Notifier makes safety
determination; FDA reviews
notification
Drugs (Federal Food, Drug Risk-benefit FDA must prove the material is
and Cosmetic Act) safe and efficacious
Dietary Supplements Reasonable assurance that ingredient FDA must prove the material
(Dietary Supplement does not present a significant is not safe
Health and Education or unreasonable risk of illness
Act) or injury
ill. Their purchase and consumption are unsupervised, unlike drugs for which access is
carefully controlled and use is supervised by a physician. However because of the special role
of infant formulas in the diet of infants, FDA requires that these products meet certain
standards developed under the Infant Formula Act.
Quality Factors and Safety Aspects of Infant Formula

Section 412 of the FD&C Act and associated regulations issued by FDA deal with good
manufacturing practices, quality control procedures, quality factors, notification require-
ments, and records and reports for the production of infant formulas. It is important to
understand the concept of “quality factors” when considering the safety of ingredients, even
though they are not considered as an integral part of the safety assessment of new substances
under Section 409.
Quality factors are discussed in the report of the House Committee on Interstate and
Foreign Commerce that accompanied the 1980 Infant Formula Act. An ingredient can only be
used commercially (incorporated into infant formulas for sale) after the requirements of
Section 412 (e.g., nutrient requirements and quality factors) are satisfied. This means that the
new ingredient must be incorporated into the formula and the formula must then be tested for
each required nutrient (i.e., protein, fat, essential fatty acids, vitamins, and minerals) under
21 U.S.C. 350a, Section 412(i). However the FD&C Act also mandates that the Secretary of
the U.S. Department of Health of Human Services establish requirements for quality factors
consistent with the scientific knowledge and states that the nutrient and nutrient level required
by the Act may be revised. Under current regulations, the concept of quality factors has not
been developed; the requirements focus on meeting the level of specific nutrients.
FDA has proposed to revise several aspects of its infant formula regulations, including
requirements for quality factors and Good Manufacturing Practices (GMPs). FDA has pro-
posed to revise “Quality Factors for Infant Formulas” (FDA, 1996), the subpart that defines
the minimum quality factors for infant formulas. In these proposed regulations, FDA iden-
tifies two factors for which there is scientific basis to define them as quality factors, namely
that the formula is “capable of supporting normal physical growth of infants” and “protein
is of sufficient biological quality to meet the protein requirements of infants.” FDA, in
proposed rule Section 106.3(o), has defined quality factors in a manner that encompasses

several basic concepts, including bioavailability and healthy growth (FDA, 1996). The pro-
posed rule states, “The quality factors, therefore provide a means of evaluating whether a
nutrient has become less bioavailable than would be expected, so that it is not sufficiently
effective to meet its normal nutritive functions, or whether its bioavailability has been
enhanced to a level that raises safety concerns” (FDA, 1996, P. 36179). Furthermore, “FDA
considers the concept of healthy growth to be broad, encompassing all aspects of physical
growth and normal maturational development, including maturation of organ systems and
achievement of normal functional development of motor, neurocognitive, and immune sys-
tems” (FDA, 1996, P. 36179). These parameters are also considered in the safety evaluation
of any new ingredient (Section 409). Thus the safety aspects and the quality factors of the
FD&C Act overlap in that they consider the safety of new ingredients by themselves (Section
409) and as part of the matrix (formula) (Section 412).
Routes to Add New Ingredients to Infant Formulas

Infant formulas are regulated as food. If a manufacturer wishes to add a new ingredient
to an infant formula, it must follow one of three routes: it may determine a substance is
GRAS without formally notifying FDA, it may file a GRAS Notification with FDA, or it may
file a Food Additive Petition (see Box 4-1 for the definition of a food additive) with FDA.
The GRAS Notification and the Food Additive Petition processes are described below.
GRAS Notification
GRAS status is based on common knowledge about the safety of the ingredient (the
substance and its impurities) throughout the scientific community that is knowledgeable
in food toxicology and related disciplines specific to the safety and intended use of the
ingredient under consideration. A GRAS evaluation through scientific procedures is based
on “generally available and accepted scientific data, information, methods, or principles,
which ordinarily are published and may be corroborated by unpublished scientific data,
information or methods” (FDA, 1997, P. 18960). There must be a “consensus among
qualified experts about the safety of the substance for its intended use.”
If the manufacturer believes the potential new ingredient is GRAS, the manufacturer will
proceed to make the safety assessment and GRAS determination. Under the proposed GRAS
Notification Rule (FDA, 1997), the manufacturer must declare that a substance is GRAS on
the basis of scientific consensus by qualified experts. A manufacturer may convene a panel of
BOX 4-1 Definition of a Food Additive
“. . . any substance the intended use of which results or may reasonably be expected to result, directly
or indirectly, in its becoming a component of food or otherwise affecting the characteristics of any food
. . . if such substance is not generally recognized, among experts qualified by scientific training and
experience to evaluate its safety, as having been adequately shown through scientific procedures* to
be safe under conditions of its intended use.”
*(or, in the case of a substance used in food prior to January 1, 1958, through either scientific proce-
dures or experience based on common use in food.)
SOURCE: Federal Food, Drug and Cosmetic Act (21 U.S.C. §301).

experts. If the manufacturer concludes that the ingredient is safe for its intended use, the
manufacturer notifies FDA. FDA reviews the notification (which includes a summary of the
scientific evidence and historic use), and if FDA has no questions, it issues a letter of no
objection. If FDA has questions concerning the safety of the ingredient, the manufacturer
must satisfactorily answer them. In general, the quantity and quality of the data used to
support a GRAS determination is comparable with the quantity and quality of the data used
to support a Food Additive Petition.
Food Additive Petition

If a potential new ingredient cannot be determined to be GRAS, the manufacturer must
file a petition proposing the issuance of a regulation prescribing the conditions under which
the proposed additive may be safely used. The manufacturer supplies FDA with all pertinent
data, especially safety data. The agency then conducts a comprehensive review of all the
safety data and determines if the ingredient is safe for its intended use. In other words, FDA
“owns” the safety decision after it makes an exhaustive, rigorous scientific evaluation of all
appropriate safety studies. The process is generally lengthy because FDA has limited re-
sources to review all the data and render a decision across a broad spectrum of ingredients
and foods. This brief description of FDA’s safety assessment process should not be inter-
preted to mean the process is simple; quite the contrary, the process is complex and rigorous.
Limitations of the Current Process

The GRAS Notification has become the route of choice for the introduction of new
ingredients because it is scientifically rigorous, far more efficient, and equally or more
transparent than the Food Additive Petition route. Still, the current process to assess the
safety of ingredients new to infant formulas is complex and presents a number of limitations.
The committee reviewed the current process and raised a number of issues that need to be
addressed. Some are addressed here, some are addressed elsewhere in this report, and some
need further study.
The regulatory scheme for the safety assessment of ingredients new to infant formulas is
complex and spread out over several sections of the FD&C Act (Sections 409 and 412) and
proposed regulations. In reality, the GRAS Notification is a scientifically rigorous and
transparent determination of safety. It is flexible enough to accommodate the broad spec-
trum of issues, yet rigorous enough to ensure safety of the ingredient. However, for infant
formulas, the complexity of the system has resulted in regulations that are not well under-
stood by the scientific or medical community and misconceptions about the safety of GRAS
ingredients for infant formulas may exist.
A second limitation of the current system is that it does not appear to adequately address
the uniqueness of infancy and infant nutrition. Human milk or formulas are the sole source
of nutrition for the first 4 to 6 months of life. Animal models to evaluate the safety of food
ingredients in the period from birth to weaning are not addressed in FDA’s Toxicological
Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in
Food (also known as the Redbook1; see Appendix C) (OFAS, 2001, 2003a). Conducting the

types of toxicological studies that are discussed in the Redbook is a significant challenge (see
Chapter 5).
Under current regulations, no measures of efficacy (i.e., health benefits) for the sub-
stance in question are required. Section 409 deals with safety of food additives independent
of efficacy because efficacy has traditionally been considered a property of drugs and inap-
propriate for foods unless accompanied by an authorized health claim. Interestingly, many
of the components of human milk have drug-like effects (e.g., prevention of disease) and are
not considered classic nutrients.
The committee’s primary charge was to assess safety of new ingredients added to infant
formulas under U.S. regulations, with possible international applications. Safety and efficacy
are clearly separated under U.S. regulations, but may not be so clearly delineated elsewhere.
The committee was fully cognizant that efficacy is not a consideration of the GRAS process
(Section 409) and focused its attention on matters related to safety as delineated in its
charge. However safety and efficacy are not always mutually exclusive attributes and, in the
case of infant formulas, which serve as a sole source of nutrition for a vulnerable population,
there are overlaps of these parameters in Section 412. Although efficacy was outside the
charge to the committee and an in-depth discussion was not attempted, the committee
recommends consideration be given to convening a scientific expert committee to explore if
benefit could be an appropriate requirement under Section 412.
A third limitation is the lack of guidelines for selecting a qualified and unbiased expert
panel that may be used to evaluate the safety of the ingredient. The proposed GRAS regula-
tions do not provide guidance about the panel selection or composition for manufacturers
that use this mechanism to achieve consensus about the safety of the ingredient under
consideration. Since only a few ingredients specifically intended for use in infant formulas
have been evaluated by this route, manufacturers may be uncertain about selecting appropri-
ate panel members until they have more experience with the process.
Nearly 25 percent of GRAS Notifications proposed since 1997 were rejected by FDA
due to the inability of the notifier to satisfactorily answer FDA’s questions regarding the
substance and its health effects (OFAS, 2003b). The expert panel should have the appropri-
ate experts to ask the right questions and form an opinion that is robust and of the highest
scientific integrity. Guidelines for selecting a panel early in the process could improve the
efficiency and objectivity of the process.
A fourth limitation is that formal guidelines for in-market surveillance do not exist for
infant formulas. Infant formula manufacturers routinely conduct passive surveillance via
toll-free calls, contact with health care professionals, and reports from their field sales force.
Since infants are unable to communicate verbally, any adverse effects must be observed and
reported through the parents or caregiver, thus special attention must be paid to detect
adverse or unusual reactions. GRAS status is a time- and exposure-dependent judgment that
requires the frequent monitoring of toll-free calls during the first 6 to 9 months after
introduction of the product to the market. If the original safety determination was properly
conducted, adverse outcomes should be rare and it should take a significant period of time
to collect sufficient data in order to reaffirm GRAS status of the ingredient. A review of all
pertinent data published and unpublished since the GRAS determination should be con-
ducted approximately 2 to 4 years after introduction of the product. (Chapter 7 provides
more details on current practices for in-market surveillance.)
In summary, the current GRAS review process was not designed specifically to address
possible concerns for new ingredients in infant formulas or for situations where the sole
intended use of the substance is in the only dietary product provided to an individual, as is
the case for formula-fed infants.

PROPOSED SAFETY ASSESSMENT PROCESS
Use of a Hierarchical Approach

A hierarchical approach is recommended to assist in determining the appropriate level
of assessment by considering: (1) the harm (e.g., toxicity), and (2) the potential adverse
effects of a new ingredient. This hierarchical approach will guide the level of assessments to
be applied to the new ingredient by considering the following factors:
• the reversibility of potential harmful effects,

• the severity and consequences of adverse effects,
• the time of onset of manifestations of the adverse effects,
• the likelihood that a new ingredient could adversely affect a specific system, and
• whether the effect would be common or rare.
This approach to evaluating the safety of new ingredients to be added to infant formulas
was based on the uniqueness and vulnerability of the infant population. Therefore each step
in the process requires empirical evidence from many disciplines and the application of the
highest standards, whether using methods of bioassay, nutritional analysis, or basic chemis-
try. This approach is valuable in determining the relative importance of potential adverse
effects for each specific new ingredient by providing generic templates for different steps in
the safety assessment process rather than specific recommendations for each compound. It is
neither realistic nor desirable to design individual templates for each new ingredient; rather
expert panels can refine the generic templates as needed. This approach is designed for a
broad spectrum of ingredients and could be applied to new ingredients to be added to infant
formulas regardless of the regulatory process used.
The hierarchical approach is graphically presented in Figure 4-1 and by algorithms
throughout the report. It is also applied in Appendix D to long-chain polyunsaturated fatty
acids and probiotics. Each algorithm is a step-by-step decision tree that depicts the logic of
the process, but it does not denote a particular chronology. For example, a manufacturer
may initiate several different studies and procedures at the onset of the process, the results of
which could be assessed at different steps in the algorithm. Any new ingredient considered
for use in infant formulas must be considered in the context of its form, the matrix, and
other ingredients with which it may interact.
RECOMMENDATION: Since infants have distinct needs and vulnerabilities, a set of

guidelines should be developed to provide a hierarchy of decision-making steps for
manufacturers seeking to add new ingredients to infant formulas. Because specific safety
assessments will need to be targeted according to the nature of the ingredient, the set of
guidelines should allow for flexibility in the approach, while being rigorous and scien-
tifically based (see Figure 4-1).
Proposed Elements of the Safety Assessment

While each submission is unique and a reflection of the ingredient and its intended use,
there is an opportunity to standardize the format and approach to the submission that
would streamline its preparation and evaluation. The committee recommends amplifying
the current approach without being prescriptive. It should be noted that some of the ele-
ments of the system proposed here are currently in place. The committee recognizes that
some of its recommendations may require statutory changes.

PROPOSED PROCESSES
1
formula
2
infant formula

- history of safe use and neurological safety.
(See Chapter 5)
4
(See Sidebar A)
5 (See Chapter 6)
Clinical Studies
(See Sidebar B)
Establish expert panel to evaluate in-market monitoring,
consultation with regulatory agency to review
literature reviews. Determine necessary follow-up studies.
results and determine safety of new (See Chapter 7)
ingredient
ingredient 9

Y es
other components?
No

Yes Yes
of new ingredient
No
No
15 14 REGULATORY AGENCY 12 REGULATORY AGENCY

DISCONTINUE DOES NOT APPROVE APPROVES INFANT
PROCESS INFANT FORMULA WITH FORMULA WITH NEW
NEW INGREDIENT INGREDIENT
13
(See Sidebar C)
FIGURE 4-1 Proposed process for evaluating the safety of ingredients new to infant formulas algo-
rithm. In-market assessment should be planned in conjunction with preclinical and clinical testing.
This algorithm is modeled after the U.S. Generally Recognized as Safe Notification process; similar
schemes can be adapted to other regulatory processes. = a state or condition, =

The committee proposes the following elements, described in detail below, for infant
formula-related submissions to the regulatory agency:
1. Introduction
2. Background
3. Chemical or Biological Composition
4. Production Methods
5. Intended Use in Foods
6. Level of Exposure
7. Chemical Structure/Activity Relationships
8. Safety Assessment
9. Other Related Biological Activity or Interactions
10. Hazard Identification and Risk Assessment
11. Expert Panel Findings
12. General Availability of the Data
RECOMMENDATION: Elements of the safety assessments of infant formulas need to

be standardized.
Element 1: Introduction
The introduction of a new ingredient submission should include the regulatory back-
ground to the submission and a summary of the submission.
Element 2: Background
The special nutritional requirements of infants and the role of formulas as a potential
sole source of nutrients should be discussed in the context of the rationale for the addition of
the proposed ingredient. A comparison with the composition of human milk should be
provided as a reference point. While efficacy and risk-benefit should not be a consideration
in the safety determination of an ingredient intended to be used in food, these factors cannot
be totally ignored for biologically active substances, as they can be for the more traditional
food ingredients (e.g., color, flavor, and preservatives). It must be recognized that it is
unlikely that a manufacturer would add an ingredient solely because it is a component of
human milk. Thus the rationale for the addition of the ingredient must be provided.
A targeted review of the literature and relevant commercial application within the
United States or other countries could be helpful in understanding the novelty of the sub-
stance and its use (see Figure 4-1, Box 3). The extent and level of assessments will, in part,
be decided upon by an expert panel’s review of the history of use.
Element 3: Chemical or Biological Composition

Many of the substances that can be considered for addition to infant formulas are
natural components of human milk, but their source is plants or animals, or they are
synthesized. The safety issues therefore often revolve around the purity of the ingredients, as
most are components of complex mixtures and may contain other biologically active or
possibly toxic substances. The group of active components and the associated impurities
must be well characterized (see Figure 4-1, Box 4). If the ingredient is produced syntheti-
cally, any by-products and residues must also be well characterized or removed.

Whether natural or synthetic, any complex ingredient must be examined to determine if

its other components are toxic or in any way affect the function of the final product;
byproducts shown to be of concern should be removed. If appropriate preclinical and
clinical evaluations are conducted, this should lessen the safety concern. Sources of new
ingredients must be determined to be free of dangerous impurities, such as pesticide residues,
heavy metals, toxins, or pathogenic microorganisms. Allergenicity related to the ingredient
or byproducts also should be addressed.
For new ingredients not present in human milk, a full review of safety will be required
in the submission since there will likely be no history of safe use by infants. Reproducibility
of the manufacturing process must be adequately demonstrated and tolerances established
for the critical parameters. Complete specifications and analytical methods must be publicly
available to allow independent validation of the characterization of the ingredient and
associated tolerances. Stability of the ingredient during storage and any subsequent process-
ing should be demonstrated.
Nontraditional sources, such as transgenic products and those produced by new meth-
ods, must have programs of evaluation designed to answer concerns related to these meth-
ods. For example, a review that includes adequate characterization and documentation of
the genetic origins of the starting material and the characteristics of the process should be
provided. The active components of the process (e.g., culture, recombinant deoxyribonucleic
acid), along with the potential for introducing levels of undesirable compounds (e.g., toxins
or allergens), should also be stated. As with other ingredients, the absence of substances that
may present a hazard should be demonstrated.
Element 4: Production Methods

A detailed description of the methods used to manufacture the ingredient and the
formula into which the ingredient is incorporated must be provided. Trade secret infor-
mation should be clearly identified. Manufacturing and quality specifications should be
established.
Most infant formulas are subjected to some form of heat treatment in order to provide
a microbiologically safe product. These may include retorted or overpressure retorted, dried,
aseptic, and pasteurized-refrigerated methods. New ingredients may require innovative pro-
cessing methods to preserve the biological activity for the shelf life of the product. In some
cases shelf life may be quite brief, refrigeration may be required, or separate components
may need to be mixed at the time of feeding. Any changes in the processing and storage must
be evaluated for safety and bioavailability. Individual ingredients have different rates of
degradation (e.g., ascorbic acid degrades rapidly). The shelf life of the product depends on
the ingredient with the most rapid degradation.
A wide variety of packaging (cans, bottles, composite canisters, pouches) is currently
used for infant formulas, and current regulations define methods to assure safety and
bioavailability. If new ingredients require novel package forms, the safety assessment must
account for this.
GMPs (Good Manufacturing Practices) for infant formulas must also be addressed in
this area of the submission to ensure that no unusual safety concerns arise.
Element 5: Intended Use in Foods

The technical, functional, or nutritional rationale for adding the substance to the infant
formula should be clearly stated. If label claims will be made (e.g., “improves cognitive

performance,” “lowers cholesterol”), the claim and references to support the claim should
be cited. The intended use level should also be provided.
Element 6: Level of Exposure

The estimated daily intake of a substance in the context of the entire diet is of
paramount importance in assessing the safety of any food ingredient. In the case of infant
formulas, this determination should be straightforward as long as formulas are the sole
source of nutrition. When solid food is introduced, a more detailed analysis must be
conducted based on dietary records and panel data if the substance naturally occurs in the
diet. For example, a careful record (diary) of all food consumed for a period of a few
weeks should be kept and then analyzed for the substance in question. Another approach
is to analyze available consumption data for foods containing the substance and calculat-
ing an estimated level of intake. These values are then combined with the estimated intake
from the formula, appropriate safety factors are applied, and a safety determination is
conducted.
Element 7: Chemical Structure/Activity Relationships

The analysis of the chemical structure/activity relationships is a useful approach to
correlating molecular structure with biological activity or toxicity. In the absence of specific
information about the toxicity of a compound, similarity in chemical structure to known
toxicants can be used to estimate toxicity. Based on molecular structure, ingredients can be
placed in one of the following categories (OFAS, 2001, 2003a):
• Category A – low potential for toxicity,

• Category B – intermediate potential for toxicity, or
• Category C – high potential for toxicity.
Categories, coupled with exposure levels, lead to the classification of ingredients into
Concern Levels. Concern Levels are relative measures of the degree to which the use of an
ingredient may present a hazard to human health. For example, an ingredient whose struc-
ture places it in Category C (high potential) and has a high level of exposure would result in
assignment to Concern Level III (high concern), while an ingredient in Category A and low
exposure would be placed in Concern Level I. The Concern Levels are used as a starting
point to recommend toxicity tests, and the committee recommends using the same criteria
for new ingredients added to infant formulas. Table 4-2 summarizes the toxicity tests
recommended based on the Concern Levels.
Element 8: Safety Assessment

FDA’s Redbook and the classic principles of toxicology are an excellent starting point
for assessing the safety of new ingredients, but blindly applying the Redbook would be
clearly inappropriate in the case of infant formulas. Many of the ingredients being consid-
ered for addition to infant formulas are naturally occurring nutrients in human milk, yet
simply because an ingredient is a component of human milk does not mean it is safe. The
matrix effect, ratios of other components, and interactions with other ingredients must be
taken into account before a safety conclusion can be drawn for an infant formula. For

TABLE 4-2 Summary of Toxicity Tests Recommended for Different Levels of Concern
of Food Ingredients
Concern Levels
Toxicity Tests a I II III
Short-term tests for genetic toxicity X X X
Metabolism and pharmacokinetic studies X X
Short-term toxicity tests with rodents Xb
Subchronic toxicity tests with rodents Xb Xb
Subchronic toxicity tests with nonrodents Xb
Reproduction study with teratology phase Xb Xb
One-year toxicity tests with nonrodents X
Carcinogenicity study with rodents Xc
Chronic toxicity/carcinogenicity study with rodents X c,d
aNot including dose range-finding studies, if appropriate.
bIncluding neurotoxicity and immunotoxicity screens.
cAn in utero phase is recommended for one of the two recommended carcinogenicity studies with rodents, prefer-
ably the study with rats.
dCombined study may be performed as separate studies.
SOURCE: OFAS (2001).
ingredients that are already a part of human milk, the amount of toxicological testing
required should be reduced, but the following must be thoroughly considered:
• What is the level of addition and interaction with other nutrients, processing, and
storage?
• Are there any differences in absorption, distribution, metabolism, and excretion?
• Is the added ingredient chemically identical to the substance contained in human
milk?
• Were contaminants introduced via the manufacturing process?
• Will the ingredient be safe for the shelf life of the product?
In addition to these considerations, the classic issues of subchronic toxicity, repeat-dose

target organ toxicity, genotoxicity, carcinogenicity, reproductive and developmental toxic-
ity, and neurotoxicity should be considered and, where appropriate, in vitro and in vivo
animal testing should be conducted (see Figure 4-1, Box 4). Animal models must be chosen
appropriately to extrapolate results to humans (see Chapter 5). Dose, bioavailability, nutri-
tional requirements, and developmental stage must be considered to prevent confounding
the toxicology results.
Currently clinical trials for infant formulas with added new ingredients, although rec-
ommended, are not required by FDA, and the manufacturer decides when and if clinical
trials are needed. The required tests focus on nutrient requirements, and only in cases of
minor changes to the formulation can the manufacturer request a waiver of the tests require-
ments. However in the proposed rule (Section 106.3(o), subpart E, “Quality Factors for
Infant Formulas”), quality factors are defined in terms of growth and development of the
infant and therefore would require clinical trials as evidence of normal growth. The commit-
tee recommends that those clinical trials that address safety of the ingredient, metabolites,
and matrix be included as part of the submission (see Figure 4-1, Box 5).

Further complicating the safety evaluation is the fact that most of the new ingredients
are nutrients or biologically-active molecules that supposedly confer health benefits—not
classic toxicants. In addition, some of the new ingredients are macronutrients, which require
the application of different safety factors, experimental designs, and interpretation of re-
sults. In most cases micronutrients fed at the high doses typically used in toxicological
testing would be toxic, thus safety factors of 10 or less are more appropriate than safety
factors of 100.
Element 9: Other Related Biological Activity or Interactions

Early studies of the isolated ingredient may be helpful to identify biomarkers and to help
in the design of more complex products. In addition to identifying any toxic or untoward
reactions, preclinical studies may help identify appropriate levels of use. New ingredients
should be evaluated in the matrix expected to be used in the final product in order to
determine absorption, utilization, and activity when other ingredients are present. For ex-
ample, soy formulas contain phytate, which can bind zinc, iron, and other divalent cations
and make them unavailable. To allow for this, soy formulas are fortified with zinc, iron, and
calcium. They also provide relatively large amounts of iron to allow for decreased absorption.
Ingredients are often available in more than one form. For example, selenium may be
added to formulas as an inorganic salt or as an organic compound. Selection of the form
most desirable for biological availability may present product-development challenges, but
clear bioavailability is critical (MacLean and Benson, 1989). For infants, human milk and
formulas are the sole source of nutrition for the first 4 to 6 months of life and an important
source for the next 8 months. Although it is assumed that many new ingredients will be
components of human milk or already accepted as safe for adults and children, the matrix in
which the new ingredient is incorporated is very relevant in the safety evaluation (currently
covered under Section 412). The importance of the matrix can be simply illustrated by the
differences in cow-milk and soy-based formulas. Soy-based formulas may contain phyto-
estrogens, which could impact on the biological activity of a new ingredient. In addition to
biological activity, factors in soy-based formulas also might impact bioavailability. For
example, the fatty-acid profile of the matrix may differ, and this could impact on the
metabolism of fatty acids added to provide a certain benefit. Since infants are a vulnerable
population, the rationale for adding any new ingredient must be clearly stated so that
qualified experts can make the safety evaluation with all factors known.
As these examples show, any new ingredient considered for use in infant formulas must
be considered in the context of its form, the matrix, and other ingredients with which it may
interact. Specific procedures for assuring bioavailability will depend on the compound under
consideration.
RECOMMENDATION: Bioavailability is of special importance to infants and should

be specifically addressed in any evaluation of the safety of infant formulas. Other factors
that should be considered are: tolerance, allergenicity, impact of gastrointestinal flora,
and possible nutrient imbalances (if ratios or cofactors are important).
Element 10: Hazard Identification and Risk Assessment

As stated previously, both the subject ingredient and associated contaminants or impu-
rities may be hazardous to the target population, in this case term infants. The risk assess-
ment for the ingredient versus the impurities are likely to be quite different. A no-observed-

adverse-effect level, safety factor, and estimated daily intake need to be determined, taking
into account the differences in risk posed by the ingredient and possible contaminants. This
will need to be determined for each new ingredient by each new expert panel.
Element 11: Expert Panel Findings

The findings of the expert panel are the heart of the safety evaluation submission. A
properly constituted panel must be objective and consist of the appropriate types of scientific
experts. In other words, the panel should have the appropriate experts to ask the right
questions and form an opinion that is robust and of the highest scientific integrity. The
manufacturer, in consultation with the regulatory agency, should determine an appropriate
expert panel early in the process (see Figure 4-1, Box 6) to:
• guard against the theoretical possibility of bias,

• ensure an efficient process, and
• minimize the chance of rejection after submission.
The experts serving on the panel need to be free of any conflicts of interest (e.g., former
employees or current stockholders of the manufacturer). Although there is no evidence of
biased panels in the past, the potential for bias is a theoretical concern that needs to be
addressed when selecting panel members.
The committee strongly recommends that in selecting appropriate experts to analyze
ingredients new to infant formulas, the expert panel should include a physician with experi-
ence in clinical assessment, preferably a pediatrician. The composition of the rest of the
panel should be determined in consultation with the regulatory agency and will depend on
the nature of the ingredient (e.g., if dealing with probiotics, a microbiologist and immunolo-
gist should be included on the panel). A subpanel of specific experts may be needed in certain
instances (e.g., when certain levels of in-market surveillance are needed; see Chapter 7). The
panel members must be recognized experts in their field of expertise and highly regarded by
their peers and the regulatory agency. They must conduct an independent critical evaluation
of the information that is publicly available and conclude that they, as well as other experts
in the field, would generally recognize that the ingredient is safe for its intended use.
The regulatory agency would review the panel’s safety assessment and other data in the
submission and if satisfied, would issue a letter of no objection. If the regulatory agency has
questions and the manufacturer does not answer them satisfactorily, the agency can reject
the submission, ask the company to withdraw it, or suggest consultation with additional
experts qualified to opine on the specific safety concerns.
RECOMMENDATION: In seeking to add new ingredients to infant formulas, manu-

facturers should establish balanced, qualified expert panels in consultation with the
regulatory agency. The panel members should review existing data and may identify a
need for additional studies.
Element 12: General Availability of the Data

The requirement for all the pivotal data to be generally available for scrutiny by the
scientific community at large makes the process transparent and robust. The data on which
the safety determination is based must be published in peer-reviewed scientific journals and
could be supplemented by secondary scientific literature, such as review articles and text-

books. For example, FDA’s letter of no objection summarizing the basis for the GRAS
Notification is made available on FDA’s website (OFAS, 2003b), and the full notification
should be available through the Freedom of Information Act. The responsibility for the
safety determination rests with the company filing the notification, and it is the company’s
continuing responsibility to ensure that the food ingredients they market are safe and in
compliance with all applicable legal and regulatory requirements.
SUMMARY
The safety assessment of infant formulas is complex and not fully standardized. While
the current processes that regulate the addition of new ingredients to infant formulas are
both flexible and scientifically rigorous, they do not adequately address the uniqueness of
infants and infant nutrition. Also, they do not provide either enough guidelines on the
selection of an appropriate expert panel that may be used to show consensus or guidelines
for in-market surveillance. There is an opportunity to address these limitations and stan-
dardize the elements of the safety assessment of ingredients new to infant formulas without
being overly prescriptive. The recommendations described here are meant to address the
specific needs of infants in improving the regulatory process for this potentially vulnerable
population group.
REFERENCES
FDA. 1997. Substances Generally Recognized as Safe. Proposed rule. Fed Regist 62:18937–18964.
OFAS. 2003a. Redbook 2000. Toxicological Principles for the Safety of Food Ingredients. Online. Center for
Food Safety and Applied Nutrition, Food and Drug Administration. Available at http://www.cfsan.fda.gov/
OFAS. 2003b. Summary of all GRAS Notices. Online. Center for Food Safety and Applied Nutrition, Food and
Drug Administration. Available at http://www.cfsan.fda.gov/~rdb/opa-gras.html. Accessed November 19,
2003.

Testing Ingredients with Preclinical Studies
ABSTRACT
Preclinical studies are a vital first step to assess the safety and quality of ingredi-
ents new to infant formulas. They must be performed before an ingredient can be
considered for clinical studies in humans in order to determine the potential toxicity
of the ingredient, its metabolites, and its matrix. Guidelines for these studies to as-
sess the safety of infant formulas must be based on considerations of the diversity of
potential new ingredients and the ingredients’ source and matrix. In the United
States, the Food and Drug Administration’s (FDA) Redbook provides comprehen-
sive guidelines for conducting preclinical studies to test the safety of food and color
additives, but it often does not take the many special needs and vulnerabilities of
infants into consideration. In Canada, there are no comprehensive guidelines for
conducting preclinical studies, and decisions are made on a case-by-case basis using
internationally accepted guidelines.
The committee recommends implementing a flexible, two-level assessment ap-
proach to help guide the expert panel’s decisions on the appropriate preclinical stud-
ies (including preanimal tests; absorption, distribution, metabolism, and excretion
studies; toxicity studies; and neurological studies) to assess the safety of ingredients
new to infant formulas. Level 1 assessments include standard measures for each
organ system required for all new ingredients (e.g., commonly used screening tests of
cell and organ composition and function). Level 2 assessments include in-depth mea-
sures of organ systems that would be used to explicate equivocal level 1 findings or
specific theoretical concerns not typically addressed by level 1 tests.
In addition to following established guidelines, a distinct set of procedures using
appropriate cellular and animal models at relevant developmental stages should be
included in studies to assess safety. The most commonly used animal models for
general toxicological studies are the rat and mouse, but they are of limited use for
developmental studies involving ingredients new to infant formulas because of the
70

TESTING INGREDIENTS WITH PRECLINICAL STUDIES 71
difficulty of feeding formulas to a preweanling rodent. The nonhuman primate and

the piglet are more amenable for these types of studies.
INTRODUCTION
This chapter describes the importance and the unique aspects of conducting preclinical
studies to assess the safety of infant formulas. Current regulatory guidelines for preclinical
studies are described, and a two-level assessment process is proposed. The committee’s
recommended two-level process is a flexible approach that can accommodate a variety of
potential ingredients. For example, safety assessment might not be as intense for an ingredi-
ent that is not absorbed systemically. The manufacturer, in consultation with an expert
panel, determines the types of tests conducted. The extent of this investigation will be
determined on the basis of previous experience with the ingredient in other populations and
on theoretical concerns based on the putative biological effects of the ingredient, its metabo-
lite, and its matrix. Finally, the committee recommends the following preclinical tests de-
scribed in detail below:
1. preanimal tests (including structure, stability, and solubility tests of the ingredients;
genetic tests; and cellular studies)
2. absorption, distribution, metabolism, and excretion studies
3. toxicity studies (including acute, subchronic, chronic, developmental, and organ
studies)
4. neurological studies
Neurological studies are described in a section separate from other toxicity studies
because the scope of work defined for this project placed special emphasis on the potential
effect of ingredients on the rapidly developing infant brain.
THE IMPORTANCE OF APPROPRIATE PRECLINICAL STUDIES

Preclinical studies must be performed before an ingredient can be considered for clinical
studies in humans in order to determine the potential toxicity of the ingredient and its
metabolites and their effects in the matrix. Preclinical studies allow researchers to expose cell
cultures and experimental animals to doses of ingredients not normally encountered in
human consumption. Results of such assessments are used to determine the threshold of
toxicity for the given ingredient (i.e., the margin of safety).
Guidelines for preclinical studies to assess the safety of infant formulas must be based on
considerations of:
• The diversity of the potential new ingredients. The new ingredients will possess
different chemical characteristics, nutritional contributions, pharmacological activities, and
physiological activities. The ingredients may be a conventional synthetic or an extracted
single component, a plant extract, or a complex mixture. The ingredients may also be
derived from novel sources or processes (e.g., products of fermentation or biotechnology).
Such diversity requires preclinical guidelines that are clear but not overly prescriptive be-
cause of the disparity in the issues that each class of ingredient may represent. Nevertheless,
given the vulnerability of the population to receive the ingredient (infants), it is incumbent
upon the manufacturer, in consultation with the expert panel, to be overly conscientious in
considering potential safety issues.

• The ingredient’s source and matrix. The approach to evaluate ingredients new to
infant formulas should be driven by the class of the functional substances and by the full
characterization of the ingredient in the matrix of the infant formulas. In other words, the
general approach is to deal with the targeted ingredient, as well as the nontargeted com-
pounds, such as metabolites of the targeted ingredient, the vehicle required for delivery of
the targeted ingredients, and impurities introduced in the manufacturing process.
CURRENT REGULATORY GUIDELINES FOR PRECLINICAL STUDIES

In considering how to construct guidelines for preclinical studies of ingredients new
to infant formulas, the committee drew on existing guidelines for the testing of food
ingredients and infant formulas. In the United States these are included in Sections 409
and 412 of the Federal Food Drug and Cosmetic (FD&C) Act (21 U.S.C. §301) and in
the FDA Redbook (OFAS, 2001, 2003). The committee suggests that the risks of adding
ingredients to sole nutrient sources, such as infant formulas, are not adequately ad-
dressed; however some aspects of the more stringent guidelines applied to new ingredi-
ents as articulated in the Redbook are appropriate for some, if not all, ingredients new
to infant formulas.
Sections 409 and 412 of the Federal Food, Drug and Cosmetic Act
There are no explicit requirements for preclinical testing of infant formulas specified
under Section 409 of the FD&C Act. The section stipulates that a petition to establish safety
of a food additive shall contain “all relevant data bearing on the physical or other technical
effect such additive is intended to produce. . .” (21 U.S.C. §301), but it does not dictate a
specific type of preclinical study.
Under Section 412, which applies to infant formulas, a formula shall be deemed to be
adulterated if it does not meet the quality factor requirements prescribed by the Secretary
of Health under Subsection (b)(1). Subsection (b)(1) then states, “The secretary shall by
regulation establish requirements for quality factors for infant formulas to the extent
possible consistent with current scientific knowledge, including quality factor require-
ments for the nutrients required by subsection (i).” Currently, only protein quality is
named as a quality factor in FDA regulations but there are no specific requirements to be
met regarding quality factors. Subsection (i) specifies levels of certain nutrients (e.g., pro-
tein, fat, and specific vitamins) that are required to be met. There is no other requirement
for any specific preclinical studies.
FDA Redbook
The FDA Redbook II and Redbook 2000 (OFAS, 2001, 2003) provide comprehensive
guidelines for conducting preclinical studies to test the safety of food and color additives.
Chapters IV and V in the Redbook II and 2000 describe:
• general guidelines for designing, conducting, and reporting results of toxicity studies,
• special considerations in toxicity studies (e.g., pathology, statistics), and
• guidelines for specific toxicity studies (e.g., genetic, acute, subchronic, chronic, im-
munotoxicity, and neurotoxicity).
These guidelines, however, do not consider the unique characteristics of having infants
as consumers.

Canada’s Food and Drug Regulations

There are no explicit requirements for preclinical testing of infant formulas specified
under Canada’s Food and Drug Regulations in Division 16 (Food Additives), Division 25
(Infant Formulas), or Division 28 (Novel Foods) (Canada, 2001). Divisions 16 and 28
require that a notification be submitted to Health Canada that includes information used to
establish the safety of a food additive or a novel food. Health Canada refers manufacturers
to internationally accepted guidelines for preclinical testing or asks to be consulted because
decisions are made on a case-by-case basis.
OVERVIEW OF RECOMMENDED LEVELS OF ASSESSMENT
RECOMMENDATION: A hierarchy of two levels of preclinical assessment, using tech-

niques from cellular-molecular studies through whole-animal studies, should be imple-
mented to assess the safety of ingredients new to infant formulas on developing organ
systems:
• Level 1 assessments. These are suggested standard measures for each organ sys-
tem (e.g., gastrointestinal, blood, kidney, immune, endocrine, and brain) and are re-
quired for any new ingredient.
• Level 2 assessments. In-depth measures of organ systems or functions that would
be performed to explain abnormalities found in level 1 assessments and specific theoreti-
cal concerns not typically addressed by level 1 tests. These are suggested measures to
assess any new ingredient that primarily interacts with an organ system, has a metabo-
lite that interacts with an organ system, or stimulates or changes the synthesis of factors
(e.g., hormones, cytokines, immunoglobulins, endotoxin) that interact with an organ
system.
Figure 5-1 describes the overall flow of proposed preclinical studies, from preanimal
tests (Boxes 8 and 4) to toxicity tests (Box 4) to neurological tests (Box 7). Figure 5-2
describes the two assessment levels that should be considered when conducting preclinical
studies, and it refers to tables in the text for specific measurements that could be conducted.
CONDUCTING PREANIMAL TESTS
Structure, Stability, and Solubility

Chemical and physical characterization is warranted if the targeted and nontargeted
ingredients to be added to infant formulas are not well known and if there is no adequate
literature about them to determine their safety. Thus the complete chemical structure and
functional groups and the purity and stability of the intended and nontargeted ingredients
present in the matrix must be determined using well-established physical methods (see
Figure 5-1, Box 8). These methods include high performance liquid chromatography (HPLC),
liquid chromatography-mass spectrometry (LC-MS), and thin layer chromatography (TLC).
LC-MS is a particularly versatile technique. It can be used to assess the structure, molecular
mass, and purity of most classes of compounds because derivatization is not necessary. The
stability to temperature and ultraviolet light and the solubility properties of the ingredients
should be established by standard methodology. The percentage of the unidentifiable mate-
rials should be stated in the notification or petition to the regulatory agency. The kind of


1
infant formula
2
AND Yes
determine their safety?
Sidebar A: Chemical and Physical
Characterization
3 Structure, Stability, and Solubility
Initiate preclinical studies to evaluate toxicity - High performance liquid chromatography,
No liquid chromatography-mass spectrometry, and
and neurological safety (See Figure 5-2)
- The stability to temperature, ultraviolet light as
8 - The percentage of the unidentifiable materials
in ingredient.
Chemical and Physical Characterization
- The kind of so lvents, suspending agents,
(See Sidebar A)
emulsifiers, or other materials that will be used
in administering the ingredient whether in
in vitro or animal studies.
- Ingredient should bestored under condit ions
9 4 that maintain its stability, quality, and purity
Are chemical and Toxicity Assessment until the subsequent studies are complete.
Yes - These studies should be repeated with the
1. Genetic tests
2. Cellular studies
- Acute,subchr onic, and chro nic
toxicity
- Absorption, distribution,
No
- Hepatic
- Renal
- Hematology
- Immune
- Endocr ine
- Neurolo gic
5
safety?
No
6
10 7

PROCESS (See Figure 5-2) PROCESS
FIGURE 5-1 Proposed preclinical assessment algorithm. = a state or condition, =

solvents, suspending agents, emulsifiers, or other material that will be used in administering
the new ingredients during testing, whether using in vitro or animal studies, should be
disclosed. The targeted ingredient should then be stored under conditions that maintain its
stability, quality, and purity. The stability and solubility studies should be performed with
the ingredient both in the solution and in the matrix that would be fed to human infants.

P
1

2
No
7 Standard measures of genetic tests,
(See Sidebar A) major organ systems, and neurological
(See Tables 5-1 through 5-10)

Any evidence of cellular studies, animal toxicity studies,
Lev el 2 Assessment
adverse Yes major organ systems, and neurological
(See Sidebar B)
effect/event? preclinical measures.
No
4
for safety?
No
9 6 5
FIGURE 5-2 Proposed levels of preclinical assessment algorithm. = a state or condition,

statement.

Genetic Tests
Numerous genetic studies, such as those provided in the Redbook, are available, and the
committee recommends evaluating all ingredients, their metabolites, or their secondary
effectors for their ability, if any, to cause molecular changes in the deoxyribonucleic acid
(DNA) or to cause structural changes in the chromosomes of cells (see Figure 5-1, Box 4).
These changes may include forward and reverse mutations, point mutations, deletion muta-
tions, chromosomal aberrations, micronuclei deletions, polymorphisms, DNA strand breaks,
or unscheduled DNA synthesis. In vitro assessments use microorganisms and cells from
multicellular animals; examples are listed in Table 5-1.
Cellular Studies
It is often most efficient to perform in vitro studies of metabolism before whole-animal
(oral dosing) studies to provide information about future in vivo studies and estimate
dosages to be used in preclinical animal studies. In vitro work and pharmacokinetic model-
ing can be used to predict the potential toxicity and in vivo kinetics of the ingredients and the
matrix. The in vitro studies can also provide initial evidence of the level of clinical safety or
risk.
Because in vitro studies are generally less complex than whole-animal studies, elucida-
tion of an ingredient’s metabolic pathway and toxicity characteristics may be facilitated.
Different doses of the solubilized ingredient can be incubated with either confluent or
preconfluent cells cultured in 10-cm dishes or 96-wells plates to establish uptake and toxic-
ity levels of both the ingredient and its metabolic byproducts. Time-course and dose-response
experiments should be conducted to check the growth characteristics of the cells and the
toxicity of the ingredient. The production of metabolites can also be followed upon adding
TABLE 5-1 Examples of Potential in vitro Tests to Assess Genetic Toxicity

Test Name Function of Test Reference
Ames test Microsome reverse mutation Aeschbacher et al., 1983
Mouse lymphoma Genetic forward mutations McGregor et al., 1987, 1988a,
thymidine kinase gene 1988b; Myhr and Caspary,
mutation assay 1988; Myhr et al., 1990
Mammalian erythrocyte Micronuclei deletions, chromosomal Schlegel and MacGregor, 1982;
micronucleus test aberrations Schlegel et al., 1986
Polymerase chain reaction Changes in gene expression and Innis, 1990
deoxyribonucleic (DNA) sequence,
polymorphisms, point mutations
DNA microarray a Identifies genes that are up or down Cohen et al., 2002; Daniel, 2002;
regulated DeRisi and Iyer, 1999;
Guengerich, 2001; Lee et al.,
2000; Moreno-Aliaga et al.,
2001; Perou et al., 1999;
Wang, 2000; Williams, 1999
Proteonomics a Identifies proteins that are altered Anderson and Anderson, 1998;
after exposure to the ingredient Bogyo and Hurley, 2003;
Govorun and Archakov, 2002;
Hochstrasser, 1998; Jungblut
et al., 1999; MacBeath, 2002
Bioinformaticsa ? Vihinen, 2001
aThese techniques are relatively new and are not standardized for clinical use.

the ingredient directly to the cellular extracts. In this case, different concentrations of the
solubilized material are incubated with the cellular extract, and the production of metabo-
lites as a function time is determined by analytical methods (e.g., HPLC, LC-MS, TLC).
Radioactive and stable isotope studies are useful in tracing nutrient uptake by cells and
intracellular metabolic pathways.
The in vitro systems can also be used to measure binding adduct and conjugate forma-
tion, transport across membranes, enzyme activity and concentration, enzyme substrate
specificity, and other specific objectives. In addition to providing information on the effects
of the ingredient and its metabolites on target receptors, cell-culture studies can also indicate
subsequent downstream effects. In the future, DNA microarray technologies may be comple-
mented by the capabilities of proteomics and metabolomics to assess the expression of
expected and unexpected proteins and metabolites after ingredient exposure to the cells.
However, at this point, these techniques are relatively new and are not standardized for
clinical use. The principal problems that these advances confront lie in assessing the func-
tional significance of the results of the analyses. Numerous algorithms in dealing with such
analyses are available (Lander, 1999; Lee et al., 2000; Soukas et al., 2000) and should be
consulted.
CONDUCTING ANIMAL TOXICITY STUDIES

Several toxicity studies must be performed in animals to ensure the safety of ingredients
new to infant formulas (see Figure 5-1, Box 4). This section reviews the following issues:
• considerations when choosing animal models,

• general considerations when conducting animal studies, and
• considerations when conducting the following specific animal toxicity studies:
– acute, subchronic, and chronic toxicity studies,
– developmental toxicity studies,
– absorption, distribution, metabolism, and excretion studies, and
– organ-level studies.
RECOMMENDATION: In addition to established guidelines and regulations (e.g., the

Redbook and regulations under Sections 409 and 412 of the FD&C Act), a distinct set
of procedures using appropriate animal models at relevant developmental stages should
be included in studies to assess safety.
Choosing Animal Models

Animal models are used as a tool in initial toxicology studies before human clinical trials
are conducted. The end points of animal studies include:
• repeat dose toxicity (target organ evaluation),

• carcinogenicity (mutagenicity screen),
• developmental toxicity,
• reproductive toxicity, and
• neurotoxicity.
Animal models should be chosen appropriately, and it is especially important to look at

the comparative development between the animal model and humans and to choose the

model that can be developmentally extrapolated to humans. For example, the animal model
chosen must have digestive and metabolic similarities to the human infant and must be at the
same point in development as the human infant who consumes formulas. Investigators
should also ensure that the animal model and the reason for the studies have been justified
such that there is an understanding of what can and cannot be extrapolated and how the
preclinical data generated will be used to interpret the safety profile. All of the following
questions must be answered before selecting animal models and moving ahead with a
submission to the regulatory agency:
• What are the appropriate species?

• What are their ages?
• Should juvenile animals be assessed, or is age irrelevant for the animal models being
considered?
• What safety factors are needed with these models, and what are going to be the
relevant scientific disciplines?
At least two animal models should be selected, keeping in mind that the more animal
models used, the better (OFAS, 2001, 2003), because converging lines of biological proof of
cause and effect can be established. There are several additional factors to consider when
conducting animal studies:
• The intended ingredient should be added to the animal’s diet during the time in
development that corresponds to when the average human infant will consume it.
• The bioavailability of the ingredient in the human infant must be known.
• The study must be designed to prevent differences in pharmacokinetics, handling of
the ingredient, and dietary imbalance from competing ingredients.
The most commonly used animal models for general toxicological studies are the rat and
mouse. However the rat and the mouse are of limited use for developmental studies involv-
ing ingredients new to infant formulas because of the difficulty of feeding formulas to a
preweanling rodent. The nonhuman primate and the piglet are more amenable for these
types of studies because they readily accept infant formulas as a nutrient source. The advan-
tages and disadvantages of using each of these animal models is discussed below and sum-
marized in Table 5-12.
Nonhuman Primate
Nonhuman primates are the closest analog to humans. Their diet is similar to humans
and they can be fed infant formulas and followed developmentally. Humans and nonhuman
primates are also comparable at a neural systems level. The animals display a wide array of
sophisticated cognitive and motor behaviors (for review, see Bachevalier, 2001; Golub and
Gershwin, 1984; Overman and Bachevalier, 2001). The nonhuman primate’s neurodevelop-
mental trajectory is well described, with about a 4:1 ratio in relative time of development
(e.g., 4 months of human brain development is equal to 1 month of development in the
nonhuman primate).
Nonhuman primates can be more thoroughly assessed than humans through catheters in
the blood stream, direct cerebrospinal fluid sampling, deep implanted electrodes, and re-
peated neuroimaging. These approaches allow a greater correlation of form and function
changes.

Nonhuman primates have made substantial contributions to human nutrition studies,

including assessment of the neurobiological effects of long-chain polyunsaturated fatty-acid
(LC-PUFA) deficiency (Neuringer et al., 1986; Reisbick et al., 1994), iron deficiency (Kriete
et al., 1995), and zinc deficiency (Golub et al., 1985). This model could be used to study
changes in general behavior and speed of neural processing in response to the addition of
ingredients with neurological effects, such as LC-PUFAs. Furthermore, neuroimaging with
magnetic resonance imaging (MRI) may reveal differences in myelination or regional brain
volumes.
The main disadvantages of using nonhuman primates are that they are very expensive to
maintain, and most researchers are reluctant to sacrifice the animals to obtain brain tissue
for a single analysis.
Piglet
Piglets are comparable in size to neonatal humans and metabolize fatty acids like hu-
mans do. Piglets can be raised on a relatively high-fat infant formulas and are an excellent
choice for studies of hypoxia, endocrinology, and lipid metabolism. Piglets can be sacrificed
and, thus, tissue can be obtained for analysis.
Pig models have been used extensively to study the role of added LC-PUFAs on the
developing brain (Arbuckle et al., 1994; Craig-Schmidt et al., 1996; Hrboticky et al., 1990).
These studies generally show that infant formulas enriched with LC-PUFAs and fed to
neonatal pigs result in increased incorporation of these fatty acids into neuronal membranes
and myelin. However piglets are difficult to work with in behavioral/learning paradigms,
and neurodevelopmental form-function relationships are difficult to assess. Thus the inabil-
ity to test these animals functionally to determine if the fatty acid incorporation results in
beneficial or toxic effects is a significant drawback.
Rat
The rat is the most versatile model for toxicological testing for the following reasons:
• There is a wealth of information available on the rat model, including the nutrient
effects on biochemistry, cell biology, neurophysiology, anatomy, and behavior.
• The developmental trajectory of the rat brain has been extensively compared with the
human brain. For example, a postnatal day-7 rat has the structural maturity of a 34-week
premature human, a postnatal day-12 animal is similar to a term human (Rice and Barone,
2000), and a postnatal day-28 rat approximates a human toddler.
• The rat’s short reproductive cycle (21 days gestation) allows for rapid cycle studies
and assessment of generational effects.
• Rats are excellent learners on basic cognitive (medial temporal lobe) tasks. Learning
paradigms in rats have been correlated with human behavioral tests, allowing extrapolation
of behavioral data between species (for review, see Overman and Bachevalier, 2001).
However the rat also has several limitations. Rats normally consume low-fat diets (ap-
proximately 5 percent fat), while human infants consume 45 to 55 percent of their calories as
fat. Therefore the rat is a poor model for assessing the effects of altering or supplementing fats
(including LC-PUFAs). Studies of LC-PUFA supplementation in the developing rat pup in-
volved supplementing the mother and thus enriching the composition of rat milk (Haubner et

al., 2002; Pasquier et al., 2001). The matrix and source of these LC-PUFAs delivered to the rat
pup were therefore different than the matrix and source of LC-PUFAs added to formulas and
fed directly to the pup. Although a feeding system for artificial rat-milk substitute (formula)
has been devised, it is invasive (i.e., it involves gastrostomy placement), it is difficult to
maintain long term, and it is likely nonphysiological in nature.
Rats have been used extensively in brain efficacy and toxicity studies involving LC-
PUFAs (Delion et al., 1994; Vazquez et al., 1994; Wainwright et al., 1997). For example, the
GRAS Notification by Martek for the addition of LC-PUFAs to infant formulas lists mul-
tiple toxicity protocols involving rats (Hahn, 2000). Since it is difficult to feed infant formu-
las to a preweanling rat, the developmental tenets of timing, dose, and duration cannot be
addressed. The literature cited in Martek’s notification provides no mention of neuro-
toxicological effects in either the developing or the mature rat.
An appropriate approach to study the impact of LC-PUFAs on development would be to
assess the animals for evidence of adequate synaptogenesis, myelin biochemistry, and myelin
quantity. More specific probes for the effects of these ingredients on regulation of myelina-
tion (e.g., microtubule associated protein-2 or synaptophysin messenger ribonucleic acid)
would represent targeted approaches to potential physiological processes likely to be af-
fected by LC-PUFAs.
In some instances, such as when rats may possess different tissue metabolism and
cortical function compared with humans, the rat model may be inappropriately extrapolated
to humans. For example, the newborn human has relatively more cortical activity than the
rat, so nutrient effects on cortical structures may be underestimated in the rat model. In
contrast, the rat has a large and metabolically active hippocampus at birth relative to the
human. Thus nutrient perturbations, such as iron deficiency, protein-energy malnutrition, or
hypoglycemia, which profoundly affect the rat’s hippocampus, may overestimate the effect
in humans.
Mouse
The mouse is an excellent model because of the ability to manipulate its genetics and to
link alterations in genotype to metabolic phenotypes. The mouse genome sequence is almost
complete, so it provides a major resource to link gene expression to disease. For example, the
ability to manipulate the genome to alter the uptake and processing of nutrients provides
valuable information on mechanistic insight. A particularly powerful emerging technology is
the conditional knock-out or knock-in model where nutrient dependent genes can be altered
in specific regions of the brain at specific times of development (Lee et al., 1999).
Mice have been widely used in LC-PUFA research with respect to regulation of genes of
lipid metabolism. In these studies it has been demonstrated that LC-PUFAs are potent
inhibitors of lipogenic gene expression (Clarke, 2001; Ntambi, 1999; Shimomura et al.,
1998).
The main disadvantages of the mouse model are:
• The mouse, like the rat, consumes a low-fat diet and cannot be fed infant formulas,
making developmental effects difficult to assess.
• The mouse has different tissue metabolism compared with humans, particularly with
respect to adipose tissue and liver lipid metabolism.
• The mouse has a limited neurobehavioral repertoire compared with the rat, making it
more difficult to detect subtle neurocognitive effects after nutrient manipulations.

General Considerations When Conducting Animal Studies

Animal studies should include consideration of bioavailability, nutritional requirements
and limitations, metabolic parameters, and developmental stage. They must be conducted in
accordance with recommendations for the care and use of laboratory animals. The source
and strain of the animals should be justified and stated in the submission. Investigators
should choose:
• healthy animals,
• animals that have never been exposed to any experimental procedure,
• male and female animals, and
• animals that are capable of completing the entire study.
Testing should be performed on suckling animals and should continue after weaning
and acclimation. Depending on the test and objective, investigators should choose several
animals per sex for each experimental and control group. The animals should be assigned
to the experimental and control groups in a randomized manner to minimize bias.
The diets to be used in the study should also be justified and stated in the submis-
sion. If the ingredient is to be fed, it must meet the nutritional requirements of the
species. The diet fed to the control groups should be equivalent in nutritional value to
the diets of the dosed groups. Because the amount of food consumed by each animal in
the study cannot be determined when more than one animal is housed in each cage, it is
recommended that the test animals be single-caged. Cross-fostering should be consid-
ered if a significant effect on feeding or nursing behavior is a potential effect of the
added ingredient. For example, all studies of ingredients with potential neurological
effects (e.g., LC-PUFAs, choline, oligosaccharides) should be evaluated in this manner.
The bedding materials in the cages, ambient temperature, humidity, and lighting condi-
tions also should be stated.
Acute, Subchronic, and Chronic Toxicity Studies

The FDA Redbook (OFAS, 2001, 2003) provides criteria for toxicity assessment of
direct food additives and color additives used in food. These criteria are very stringent and
it is debatable whether this degree of stringency is necessary for each ingredient new to
infant formulas. Nevertheless the Redbook provides a scientifically conservative approach
for the riskiest of new ingredients.
For all levels of toxicity studies, the route of administration of the supplement should
approximate that of normal human exposure as closely as possible (e.g., through the diet in
the case of infant formulas). The diets to be used in toxicological studies have to be properly
selected. Natural-ingredient diets, as opposed to purified diets, are preferred because they do
not usually affect nutrient balances. However each diet may have advantages and disadvan-
tages that need to be considered.
If there is no information that can be used to determine the appropriate dose levels for
short-term or subchronic toxicity levels, toxicity studies should begin with tests of acute
toxicity, followed by subchronic, and finally chronic assessments. Table 5-2 provides guide-
lines for conducting these tests. The committee proposes the addition of developmental
toxicity studies to the assessment of ingredients new to infant formulas in addition to the
recommended guidance in the Redbook.

TABLE 5-2 Examples of Acute, Subchronic, Chronic, and Developmental Toxicity

Studies
Study Example
Acute toxicity Single dose known to be toxic to the species, followed by observation of the
animals for at least 2 weeks and establishment of the lethal dose for 50% of
the animals (LD50) for the ingredient, known bioactive metabolites, and
biomass (source)
Subchronic toxicity Generally conducted for 90 days in rats using doses established with the acute
toxicity studies
Chronic toxicity Can follow the subchronic and are usually carried out beyond the 90-day
period and perhaps to adulthood
Developmental toxicity Multigenerational study endpoints: generation toxicity F0 (parental
generation) and F1 (second generation)
Reproductive toxicity study endpoints: fertility, live born, weaning, viability
indices, and male reproductive indices (e.g., testicular spermatid numbers)
SOURCE: OFAS (2001, 2003).
Developmental Toxicity Studies

The purpose of developmental toxicity studies is to evaluate the effects of the ingredient
on developing fetuses that result from exposure of either parent prior to conception or to
mothers during gestation. In the evaluation of ingredients new to infant formulas, the latter
consideration is less relevant since pregnant mothers do not consume these ingredients. The
main manifestations of an effect on the developing organism are death, structural abnormal-
ity, altered or retarded growth, and functional deficiency. Table 5-2 provides preclinical
endpoints for developmental toxicity studies.
Absorption, Distribution, Metabolism, and Excretion Studies

The purpose of conducting absorption, distribution, metabolism, and excretion studies
is to address the biological activity or availability of the ingredient when given to the infant.
The toxicokinetics of the ingredient should be studied via the following:
• Absorption studies to assess the possible points of entry into the body (e.g., gas-
trointestinal tract, nose, mouth, and lung). Many, but not all (e.g., probiotics), new ingredi-
ents will be absorbed from the intestinal tract and have the potential to have important
positive or negative biological effects. (Even probiotics can stimulate the synthesis of com-
pounds that have important positive or negative biological effects.) It is incumbent on the
manufacturer to assay for these effects, identify them, and assess their potential risk to the
developing animal and human.
• Distribution studies to assess the subsequent transport and deposition throughout the
body. Some tissues, such as bone, adipose, brain, kidney, liver, and hemotopeotic, may act
as reservoirs for the new ingredient.
• Metabolism studies to assess the organ and cellular response to the presence of the
ingredient. A relatively inert ingredient can be metabolized to a biologically potent com-
pound that has extreme toxicity.
• Excretion studies to assess the removal of the ingredient from both systemic and
nonsystemic stores. Organs of excretion include the liver, kidneys, gastrointestinal tract,
lungs, and skin.

TABLE 5-3 Examples of Tests for Absorption, Distribution, Metabolism, and Excretion
Assessment
Function Assessmenta
Absorption Everted gut sacs and isolated intestinal loops, fecal material analysis, analysis of
material in large intestine, radiolabel
Distribution Whole body and organ autoradiography
Metabolism Radioactive and stable isotopes, LC-MS, HPLC, TLC, DNA microarray,b
metabolomics,b proteomics,b bioinformaticsb
Excretion Urine and other body fluid chemical analysis, LC-MS, HPLC, scintillation counting
if ingredient is radioactive
NOTE: The petitioner (or manufacturer), in consultation with the expert panel, determines which tests are re-
quired based on a thorough analysis of the potential effects of the new ingredient.
aLC-MS = liquid chromatography-mass spectrometry, HPLC = high performance liquid chromatography, TLC =
thin layer chromatography, DNA = deoxyribonucleic acid.
bThese techniques are relatively new and are not standardized for clinical use.
Unlike the organ system analyses described below, absorption, distribution, metabo-
lism, and excretion studies are relatively uniform for any new ingredient and should follow
the basic guidelines of the Redbook (OFAS, 2001, 2003). Table 5-3 provides examples of
tests that could be conducted.
Organ-Level Studies
Examination of organs from the selected animal models can reveal important informa-
tion concerning the effects of ingredients at the organ level. Different compounds will have
different effects on the different organs and, as discussed earlier in this chapter, the general
approach to evaluate the effect of new compounds added to infant formulas, as well as other
material, should therefore be driven by the class of the functional substance and by the full
characterization of the ingredient. The toxicity studies could be organ driven. All organs
should be screened with their appropriate level 1 assessments, while more specific concerns
with respect to a particular ingredient can be investigated through level 2 assessments (see
Figure 5-2, Box 3).
All test animals should be subjected to complete gross necropsy, including examination
of external surfaces, orifices, cranial cavity, carcass, and all organs, in the presence of a
qualified pathologist. The ratio of organ weight to body weight should be documented. For
a complete list of organs that should be examined and weighed, the reader should consult
the Redbook (Chapter IV.B.1.) (OFAS, 2001). This list is extensive, but it is important
because long-term adverse effects of the ingredient could later be found to be associated with
organs that were not conspicuous or were ignored, and therefore unchecked at the time of
the evaluation.
All organs should undergo a general screen or assessment, and then specific screens or
assessments should be conducted based on expected biological effects of the new ingredient
to be added to infant formulas. The evaluations of some of these key organs systems are
detailed in the subsequent sections. Certain organs (e.g., liver, kidney, immune system, bone
marrow, and brain) probably deserve greater scrutiny than others because of their functional
or metabolic significance. For example, certain histological abnormalities, such as scattered
focal mononuclear cell infiltrates in nonlymphoid tissues (e.g., liver and kidney), may indi-
cate autoimmune disease. Also, if the ingredient is shown to either stimulate cell prolifera-

tion or to cause atrophy and cell depletion in any lymphoid organ, the effect is likely to be
viewed as potentially immunotoxic and requires more definitive testing.
Gastrointestinal Tract Function

As Chapter 6 describes in detail, the development of the infant’s gastrointestinal tract is
essentially complete at birth and, therefore, assessments of its proper development will
involve ensuring that its functions (e.g., digestion, absorption, secretion) have not been
impaired by the addition of an ingredient new to infant formulas. Using a two-level ap-
proach for assessment, level 1 tests will include weighing the organ, performing histology
examinations of the organ tissues, and other screening tests. Level 2 tests should be used to
explicate equivocal level 1 findings or specific theoretical concerns not typically addressed
by level 1 tests. Table 5-4 provides several examples of the types of tests that could be used
in level 1 and level 2 assessments of the gastrointestinal tract.
Hepatic Function
The liver is involved in synthesis, metabolism, and excretion. Therefore, along with the
above mentioned histology evaluation, tests that account for each of these functions must be
performed as part of the level 1 assessment of liver health. Level 2 tests should be used to
explicate equivocal level 1 findings or specific theoretical concerns not typically addressed
by level 1 tests. Table 5-5 provides several examples of the types of tests that could be used
in level 1 and level 2 assessments of liver health.
Renal Function
The kidney is the major excretory organ and also serves a role in blood pressure
homeostasis. Ingredients new to infant formulas or their metabolites may be excreted by
the kidney and may potentially damage the glomerular or reno-vascular function of the
organ. Glomerular health is assessed by serum and urine chemical profiles in addition to
the above mentioned histology. Abnormalities in level 1 assessments would lead to level 2
assessments, which should be tailored to the issues raised in level 1. Table 5-6 provides
several examples of the types of tests that could be used in level 1 and level 2 assessments
of kidney health.
TABLE 5-4 Gastrointestinal Assessment: Examples of Tests in Level 1 and Level 2

Level Assessment
Level 1 Absorption
Cell culture
Organ weight/histology
Level 2 Isotopic absorption tests
Microarray/proteonomics
Receptor expression
Specific histology stains
Permeability tests

TABLE 5-5 Hepatic Assessment: Examples of Tests in Level 1 and Level 2

Level Assessmenta
Level 1 Liver weight/histology, cell culture/mutagenicity
Assessment of synthetic function: serum ALAT, ASAT, ornithine carbamyl transferase,
albumin:globulin, coagulation profile, prothrombin time, partial thromboplastin time,
radioactive amino acids, electrophoresis techniques for serum proteins
Assessment of excretion function: gamma-glutyml transferase, LDH, bilirubin, alkaline
phosphatase
Assessment of metabolic function: total protein, albumin, fasting glucose, urea nitrogen,
triglycerides (LDL, VLDL), cholesterol (HDL and LDL)
Level 2 Metabolism assessments
Protein electrophoresis
Special clotting factor levels
Special imaging studies
Special stains on histology
aALAT = alanine amino transferase, ASAT = aspartate amino transferase, LDH = lactate dehydrogenase, LDL =
low-density lipoprotein, VLDL = very low-density lipoprotein, HDL = high-density lipoprotein.
Hematological Function
Ingredients new to infant formulas or their metabolites may have profound effects on
the bone marrow. Numerous tests are available for level 1 assessments in addition to bone
marrow histology. Abnormalities in level 1 assessments should lead to level 2 assessments of
the relevant system that was perturbed by the addition of the new ingredient. Table 5-7
provides several examples of the types of tests that could be used in level 1 and level 2
assessments of hematological function.
Immunological Function
The immunological system is highly complex and has been shown to be sensitive to
nutritional manipulation (Miles and Calder, 1998). The various effects of nutrients in the
TABLE 5-6 Renal Assessment: Examples of Tests in Level 1 and Level 2

Level Assessment
Level 1 Cell culture/mutagenicity
Kidney weight/histology
Serum electrolytes, acid-base status, and blood urea nitrogen/creatinine
Urinalysis
Blood pressure
Level 2 Microarray/proteonomics
Insulin clearance
Special histology stains
Special imaging studies
NOTE: The petitioner (or manufacturer), in consultation with the expert panel, determines which tests are
required based on a thorough analysis of the potential effects of the new ingredient.

TABLE 5-7 Hematology Assessment: Examples of Tests in Level 1 and Level 2

Level Assessment
Level 1 Bone marrow histology
Assessment of hematopoietic system: whole blood hemoglobin, hematocrit, mean
corpuscular volume, mean corpuscular hemoglobin, total red cell count, red cell
morphology
Assessment of thrombopoietic system: platelet count, platelet morphology
Assessment of white cells: total white cell count and the differential
Assessment of the clotting system: prothrombin time, partial thromboplastin time,
bleeding time
Level 2 Colony forming units
Special bone marrow stains
immunological system can be divided into those mediated by antigen-specific immunoglobu-

lin (Ig) E (allergic reactions), other antibodies, T-cells, cytokines, and chemokines, and those
mediated by nonimmunological mechanisms. The effects of nutrients, such as LC-PUFAs, on
immune response have been extensively examined in laboratory animals and humans (for
review, see Miles and Calder, 1998). One could argue that the value of immune system
assessments for such nonproteinaceous food ingredients may be questionable since they are
unlikely to affect the system or to provoke an allergic reaction. However any residue present
in the matrix of an ingredient may be of protein nature and may cause an allergic reaction.
Therefore even when assessing nonproteinaceous material, an assessment for allergenicity
may be needed depending on the presence and source of impurities.
In addition to histopathological evaluation of the lymphoid tissues, level 1 assessments
should include in vivo and in vitro tests with animal models that examine unusual incidences
of maternal infections in order to evaluate potential developmental indicators of immuno-
logical toxicity (see Table 5-8).
As part of the assessment of the effect of a new ingredient on the immune system, the
possibility of a new ingredient being an allergen (i.e., able to provoke an allergic reaction,
mostly through IgE mediation) for an infant should also be evaluated. In the case of pro-
teinaceous material with unknown allergenic properties, level 1 and level 2 assessments
should be performed to assess their potential for allergenicity (see Table 5-8). A decision tree
developed by Metcalfe and colleagues (1996), and later modified by a Food and Agriculture
Organization of the United Nations/World Health Organization Expert Consultation (FAO/
WHO, 2001), should serve as the basis for the level 1 assessment. In addition, the committee
believes that further recommendations for in vitro evaluation of allergenicity by the Codex
Ad Hoc Intergovernmental Task Force on Safety Assessment of Genetically Modified Foods
(Codex Alimentarius Commission, 2003) should be followed.
It should be pointed out that development of allergic or immunologically mediated
hypersensitivity relative to ingested foods represents a complex spectrum of symptomatol-
ogy, which is also influenced by a large number of environmental and cultural factors
(Sampson, 2002). Because the immune system of the infant is essentially complete, animal
models that predict adult allergic responses can also serve as models for infants. Although
some animal models, such as the brown Norway rat model, may provide a valuable tool in
the future (Knippels and Penninks, 2003), appropriately validated animal models for assess-

TABLE 5-8 Immunology Assessment: Examples of Tests in Level 1 and Level 2

Level Assessmenta
Level 1 T-/B-cell quantitation and function (immunological analysis of B- and T-lymphocytes
and T-lymphocytes subsets [Th+Ts or CD4 and CD8])
Thymus, spleen, bone marrow, lymph nodes, tissue histology
Electrophoresis (e.g., for changes in levels of γ-globulin fractions [IgG, IgM, IgA, IgE])
Total serum complement and components of complement (e.g., C3 from CH50
determinations)
Levels of prostoglandin E2, balance of LTB4 and LTB 5
Immunochemical assays of γ-interferons and serum autoantibodies (antinuclear,
antimitochondrial, antiparietal antibodies of B-lymphocytes)
In vitro assays of activity of natural killer cells
Mitogenic stimulation assay of B- and T-lymphocytes
Macrophage activity assays
Stem cell assays
In vitro assays to assess allergenicity (source of the protein, amino acid sequence
homology analysis, physicochemical properties)
Stimulation tests of immune function using the T-dependent or independent antigens or
human vaccines
Cell-mediated immune reactivity and host-resistance assays
aTh = T helper cells, Ts = T suppressor cells, CD4 = cell differentiation antigen 4, CD8 = cell differentiation
antigen 8, IgG = immunoglobulin G, IgM = immunoglobulin M, IgA = immunoglobulin A, IgE = immunoglobulin
E, LTB4 = leukotriene B4, LTB5 = leukotriene B5.
ment of human allergic disease are not currently available. The committee believes that it is
of critical importance that animal models for allergenicity assessment in humans be devel-
oped. Until such models are validated, the assessment of allergic potential should be ap-
proached in vitro on a molecular level (e.g., utilizing sequence homologies, comparative
digestive stabilities, functional antigenic determinants, and B- and/or T-cell epitope charac-
teristics), and the nature of the demonstrated immune response to the known allergic prod-
ucts should be determined.
Endocrine Function
Growth abnormalities of the test animal is an important early indication of a possible
effect of a new ingredient on the endocrine system. Because endocrine effects may not be
immediately apparent in growth changes, nor in other metabolic functions, some screening
tests are indicated. Table 5-9 provides several examples of the types of tests that could be
used in level 1 and level 2 assessments of endocrine function.
CONDUCTING NEUROLOGICAL TESTS
Background
As explained in detail in Chapter 6, there are important reasons to include neurological
tests in safety assessments of new ingredients for infant formulas, including the sensitivity of

TABLE 5-9 Endocrine Assessment: Examples of Tests in Level 1 and Level 2

Level Assessmenta
Level 1 Hormone levels (e.g., T4, TSH, LH, FSH, GH, CCK, NPY, cortisol, leptin), blood
sugar, insulin
Organ weight/histology (e.g., adrenals, ovaries, testes, pancreas, thyroid)
Provocative endocrine tests (e.g., cosyntropin stimulation)
aT4 = thyroxine, TSH = thyrotropin, LH = luteinizing hormone, FSH = follicle-stimulating hormone, GH = growth
hormone, CCK = cholecystokinin, NPY=neuropeptide Y.
growth and development to toxic substances and the long-term predictive value of behav-
ioral measures. Therefore the scope of work defined for this project placed special emphasis
on the potential effects of ingredients to be added to infant formulas on the rapidly develop-
ing infant brain. This section discusses the preclinical assessment tools and models that can
be utilized to assess ingredients with either positive or negative neurological effects. The
approach is decidedly developmental because the effects of added ingredients (or any envi-
ronmental stressor) in the developing animal are highly dependent on the timing, dose, and
duration of ingredient exposure (Kretchmer et al., 1996).
Traditionally, two metrics—composition and performance—are applied as infant for-
mula manufacturers continue to refine their products to match the gold standard of breast-
feeding (Benson and Masor, 1994; MacLean and Benson, 1989). With respect to the latter,
the impact of infant formulas on neurodevelopment has come to the forefront. It is widely
accepted that breastfed infants demonstrate more advanced neurodevelopment (Lucas et al.,
1998; Morrow-Tlucak et al., 1988; Mortensen et al., 2002; Wang and Wu, 1996). One
potential explanation is that ingredients found in human milk, but not in infant formulas
(e.g., LC-PUFAs, nucleotides, growth factors, and oligosaccharides), may be responsible
(DeLucchi et al., 1987; Innis, 1992; MacLean and Benson, 1989). As these compounds are
identified and added to infant formulas, valid and stringent assessments of their impacts on
the developing brain should be undertaken.
Nutrients and growth factors regulate brain development during prenatal and postnatal
life. Late fetal and early neonatal life is a period of rapid brain growth and development in
most mammals, including humans (Rice and Barone, 2000; for review, see Dobbing, 1990).
There is a wealth of information on the critical events that take place in the brain’s develop-
ment in the early neonatal period. At this stage of development, cell migration is complete,
but myelination, synaptogenesis, dendritic arborization and pruning, and apoptosis are
highly active. The rapidly growing brain is more vulnerable to restriction or loss of nutrients
when compared with the more mature, less actively growing brain in later childhood and
adulthood. Arguably, the rapidly developing brain is also more amenable to repair following
nutritional perturbations and may be more highly influenced by nutrient supplements
(Kretchmer et al., 1996). Researchers have argued that the persistence of neurological ab-
normalities after repletion of a nutrient deficiency acquired early in life provides evidence
that early neurological vulnerability outweighs potential central nervous system (CNS) plas-
ticity in the developing human (Lozoff et al., 2000; Rao, 2000).
Nutrients may have variable effects on the developing brain. Nutrient deficiencies may
produce negative effects or may have no effect at all depending on the stage of brain

development (for review, see Georgieff and Rao, 2001). Similarly, nutrient overabundance
or supplementation may produce positive effects, negative effects, or no effects, and the
magnitude of the effects can be regionally different. For example, iron deficiency in the
young rat brain can cause severe energy and structural deficits in the hippocampus (de
Ungria et al., 2000; Rao et al., 1999) and can cause abnormalities in dopamine metabo-
lism without energy deficits in the striatum (Erikson et al., 2000). At an older age, the
same degree of iron deficiency causes no structural deficits. Thus any nutrient’s effect on
the developing brain will be based on the timing, the dose, and the duration of the
exposure.
All nutrients are important for neuronal cell growth and development, but manipulation
of some nutrients appears to cause more effects than others. Protein, energy, iron, zinc,
selenium, iodine, folate, vitamin A, choline, and LC-PUFAs are nutritional components that
influence early brain development with measurable clinical effects in humans and animal
models (Georgieff and Rao, 2001). A nutrient that promotes normal brain development at
one time and concentration may be toxic at another time or concentration. Many nutrients,
such as iron, are regulated within a very narrow range because deficiency and toxicity have
profound effects in brain development.
The potential effects of nutrients on brain development will determine the types of
preclinical testing that are needed. Effects can be neuroanatomical, that is, affecting neu-
ronal cell division and cell growth (e.g., size, complexity, synaptogenesis, dendritic arboriza-
tion). Furthermore, nutrients can affect developing non-neuronal cells, such as oligodendro-
cytes, astrocytes, and microglia, in turn influencing myelination, other nutrient delivery, and
cell trafficking. Nutrients can affect neurochemistry through alterations of neurotransmitter
and receptor expression. They can affect neurophysiology and neurometabolism with subse-
quent alterations of signal propagation.
The fundamental question that must be answered in the assessment of nutrient effects on
neurodevelopment is whether nutritionally induced alterations in brain development result
in changes in brain function, loosely defined as behavior. Specific considerations include:
• Is this effect transient (e.g., only during the time when the nutrient is supplemented or
deficient) or are there long-term gains or losses beyond the time of deficiency?
• How close is the linkage for each nutrient and each time period of development?
• What accounts for the individual variability in behavioral outcome of nutritional-
based alterations in brain development?
• Does a lack of behavioral effect in spite of a measurable biochemical effect imply
plasticity or alternate circuit development?
• Are there genetic polymorphisms from a change in nutrient status that confer greater
or lesser risk or benefit to the host?
Many of these questions can be answered experimentally before a nutrient is added to

formulas, but a multilevel, integrated approach is required because of the limitations of
human brain analysis (Nelson et al., 2002). The multiple levels of approach are shown in
Table 5-10. The integrated approach (see Table 5-11) chosen should be complementary and
obey the laws of timing, dose, and duration (Kretchmer et al., 1996). The animal models
must be developmentally appropriate with respect to timing of brain events and must
coincide with the likely time of nutrient supplementation or deficiency in the human. In
other words, for brain-behavior associations to be relevant, the time at which a given
nutrient alters brain developmental processes in the animal model should coincide with the
time that the nutrient deficit or supplement occurs in the infant population.

TABLE 5-10 Neurological Assessment: Examples of Tests in Level 1 and Level 2

Level Assessment
Level 1 Neuronal/glial cell culture
Histological examination
General behavior
Level 2 Electrophysiology (implanted/surface)
Magnetic resonance imaging
Magnetic resonance spectroscopy
Neurotransmitter concentrations/receptor expression
Targeted behavior assessments (e.g., learning paradigms)
The nonhuman primate, pig, rat, and mouse are the traditional models for neurological
research. Table 5-12 and the information provided earlier in this chapter describe the
advantages and disadvantages to using each model for neurological research.
Preclinical studies should assess the neurological safety of the nutrients by measuring the
physiological and pharmacological levels of the nutrients. The goal is to match structural
and biochemical alterations to changes in function (e.g., behavior). It must be recognized,
however, that there can be a dissociation between biochemical and cellular changes in the
brain and neurodevelopment (defined as behavior).
Environmental effects on the developing brain have been historically divided into effects
on the motor and nonmotor systems. The motor system is far better understood than the
nonmotor system because it is more discretely localized in the CNS and has a longer history
of investigation. Principles learned from the motor system serve as a model for the current
approaches to experimentation in the nonmotor system. The nonmotor system includes
cognitive and noncognitive behaviors. Within cognition, a number of behaviors can be
assessed, including memory and emotion. Memory is further divided into declarative or
recognition memory and implicit or procedural memory (Squire, 1992). The neural circuits
that subserve these memory systems are quite distinct and are differentially vulnerable to
nutrient insults. The difference in ontogeny of the memory systems has been extensively
reviewed by Nelson (1995).
It should be noted that many of the neural systems have mixed components. For ex-
ample, there are connections between structures underlying recognition memory (e.g., hip-
pocampus) and emotion (e.g., amygdala). Thus, for example, it may be difficult to know
from behavioral testing alone whether the enhanced ability to perform a recognition memory
task following a nutritional supplement is due to direct effects on the hippocampus or
through attentional and motivational effects mediated by the amygdala. Certain structures,
such as the cerebellum and basal ganglia, are involved in both motor and cognitive events.
TABLE 5-11 Integrated Approach to the Assessment of Neurodevelopment

Neurological Function System of Assessment
Animal behavior (fit to human) Nonhuman primate, rat
Biochemistry (neurochemistry) Human, animal model
Brain structure Human, animal model
Human behavior Human
Physiology, cellular/molecular Animal model, cell culture

TABLE 5-12 Summary of Animal Models Used in Preclinical Studies

Animal Advantages Disadvantages
Chicken Immunology No relevance to human nutrition,
genetics
Dog Metabolism, immunology, organs Behavior, genetics
Hamster Lipid metabolism Immunology, genetics
Mouse Genetics, molecular analysis, mechanisms, Learning paradigms,
organs developmental
Nonhuman primate Functional/behavioral relationship to Basic biochemistry, histopathology,
humans, similar diet as humans, genetics; expense and ethical
immunological studies, dermatological concerns for neural tests
studies, renal function, kidney biopsy,
invasive assessment
Pig Comparable size to neonatal humans, Learning paradigms, poor model
lipid biochemistry, organs for genetics
Rabbit Biochemistry, immunology No relevance to human metabolism
Rat Cellular, molecular analysis, behavior, Higher-level learning, genetics,
organs, physiological studies; brain developmental
development is similar to human infant
The optimal use of an integrated methods approach should allow assessment of nutrient
effects on behavior, neuroanatomy, neurochemistry, and form-function relationships (e.g.,
time-locked combinations of methods, such as functional MRI). It is important to have
assessments that tap similar brain areas across species (from human to rodent), allowing
extrapolation of CNS effects at the tissue level (rodents) to species where tissue is unavail-
able (humans, nonhuman primates). Table 5-13 provides examples of the cross-referencing
of behavioral tasks across three species.
Neurological Assessment Techniques

Overall there must be a systematic approach to this important aspect of safety effects of
ingredients added to infant formulas. Preclinical assessment must include whole-animal
work with specific attention to regional brain effects. Here, neuroimaging through his-
tochemistry at autopsy and through MRI in the living animal provides information. Neuro-
chemistry experiments should target plausible biological mechanisms, particularly as to how
nutrients might affect neurotransmission. Electrophysiology, using either deep-implanted
electrodes or scalp-surface electrodes, provides information about neuronal population func-
tions (Bachevalier, 2001; Fuster and Alexander, 1971) and can be cross-referenced to the
assessment of humans conducted by evoked or event-related potentials (see Chapter 6).
The following sections describe various techniques for using preclinical studies to assess
the effect of ingredients new to infant formulas on the developing brain. As with the non-
TABLE 5-13 Examples of Cross-Species Developmental-Neural Assessment Tools

Measure Infant Monkey Rat
A-not-B X X
Black-white discrimination X X X
Delayed nonmatch to sample test X X X
Maze X X
Paired comparison X X X
Object discrimination X X X

neurological organs, the committee proposes a two-level hierarchy with respect to applica-
tion of these techniques in the preclinical studies. Level 1 assessments are required of all
proposed ingredients and are designed to survey the neurological system of the developing
brain. Level 2 assessments are performed if the ingredient to be added could theoretically
affect brain development through a biologically plausible mechanism or if level 1 assess-
ments have previously demonstrated a safety concern for the brain (Table 5-10).
Neuronal Genetics/Mutagenicity
In the future, evaluations should include neuronal and glial-cell culture work to assess
the effect of the added ingredient on genomics, metabolomics, and proteomics with the
purpose of evaluating mutagenicity and further functional consequences. Genomic tech-
niques (e.g., DNA microarray analysis) are new and promising techniques that, once refined,
will likely have great utility in assessing whether there are inductions of desirable or undesir-
able genes after exposure to an ingredient. These methodologies are still limited because of
difficulties in interpreting the functionality of observed changes in gene expression. The
parallel fields of proteomics and metabolomics complement the genomics methods by as-
sessing the production of expected and unexpected proteins after nutrient exposures. Fur-
thermore, cell culture work can inform about nutrient effects on target receptors (for the
nutrient or its metabolite) (Alcantara et al., 1994) and the subsequent downstream effect on
signaling cascades in the neurons (Gietzen and Magrum, 2001; Magrum et al., 1999). An
excellent overview of this type of approach can be found in the work of Gietzen (2000) on
imbalanced amino acid diet effects on feeding behavior as mediated through the neurons of
the anterior piriform cortex. Gietzen’s work serves as a model system for assessing nutrient
effects on signaling cascades in the relevant neuronal systems. As noted earlier in this
chapter, virtually every new ingredient has the potential to advertently or inadvertently alter
basic cellular processes. Therefore neuronal-cell culture techniques should be considered
level 1 assessments.
Neuronal- and glial-cell culture techniques were not reported in the GRAS Notification
for LC-PUFAs. These techniques may have identified unwanted effects of these ingredients
on gene expression through microarray screening analysis. The effect of these added ingredi-
ents on the signaling cascade involved in long-term potentiation could have been assessed in
hippocampal CA1 cell recordings.
Neuroanatomy
There are a number of techniques that assess neuroanatomy, ranging from direct histo-
logical assessment by light, confocal, and electron microscopy to neuroimaging (e.g., MRI).
Histological assessment allows a direct view of the brain and should be used as a level 1 tool
to assess cytological effects of all proposed ingredients. The approach is used in the pig and
rodent models and can visualize general neuronal structure (e.g., myelin, synapses, dendritic
arborization, neuronal number), effector proteins (e.g., transporters, receptors), apoptosis/
cell death, and regulatory elements (e.g., iron regulatory proteins). The limitation of the
approach is that it only shows structure, and alterations of function may not always follow
structural changes.
MRI can be used in the developing human and all typical animal models to assess
nutrient effects on total brain volume, regional volumes, myelination, and the visualization
of some nutrients (for review, see Casey et al., 2001). In spite of great technical advances,
this method is generally insensitive and assesses only relatively large differences in protein-

energy or fatty acid status. Therefore it has poor predictive value for future subtle disability
of neurological development in humans. At this time, neuroimaging should be categorized as
a level 2 assessment.
Neurochemistry
Neurochemistry can be assessed either directly in the tissue once it is harvested, in the
tissue of the living brain through microdialysis catheters (Chen et al., 1995), or in vivo
through MRI spectroscopy (Tkáč et al., 2003). The goal is to assess the potential adverse
effect of ingredients on neurotransmitters and intermediate metabolism. The advent of MRI
spectroscopy allows for the in vivo, real time, repeatable assessment of the chemistry of the
developing brain. The technique uses proton, phosphorous, or 13-carbon tracer nuclear
magnetic resonance, and it can be used in humans and in animals as small as mice. Its main
limitation is the inability to target compounds prospectively; the compounds that can be
assessed are dependent on the major peaks that are visualized on the spectra. If the nutri-
tional effect is on compounds that are in a concentration that is too low to result in a major
peak, the technique is not useful. This technique is innovative and as yet not standardized.
Therefore it should be considered a level 2 assessment in the evaluation of ingredients likely
to affect neurochemistry.
The measurement of nutrient effects on neurotransmitter systems is an important part of
nutritional research on brain development. The response of neurotransmitters to changes in
nutrient availability can be directly measured in the rodent (Chen et al., 1995) and in neuronal
cell culture. It is important with the glutamine system that cell cultures contain both neurons
and glia since there is an important shuttle of glutamate and glutamine that takes place. It is
also important to assess compensatory feedback mechanisms when transmitter concentrations
are affected by nutrients. In response to increased transmitter release, there can be alterations
in receptor number and affinity and in presynaptic reuptake mechanisms. The effect of iron
deficiency on dopamine provides an excellent example (Chen et al., 1995). The assessment of
neurotransmitter systems will be a level 2 assessment for most ingredients. However many
compounds being considered for addition to infant formulas (e.g., choline) are likely to have a
significant impact on these systems and will require this type of assessment.
Neurophysiology
Nutrients have effects on the electrophysiology of neuronal populations. A fundamental
question that must be addressed in any nutrient-additive experiment is the plausible biologi-
cal mechanism by which neuronal output is changed, which could result in a behavioral
change. Examples include changes in calcium gating through alpha-amino-3-hydroxy-5-
methylisoxazole-4-proprionic acid and N-methyl-D-aspartate receptors in brain areas in-
volved in learning, which could alter long-term potentiation and learning behavior (Magrum
et al., 1999). If nutrient supplements have a putative biological effect on myelination, there
should be demonstrable effects on the speed of processing of myelinated circuits (Birch et al.,
1992; Uauy-Dagach and Mena, 1995). Indirect effects must be considered as well. For
example, steroids have significant effects on hippocampal anatomy, physiology, and gene
expression (for review, see Sapolsky, 1994). Therefore if a nutrient supplement presents a
stress to the developing organism with increased output of corticosteroids or estrogens,
secondary effects of these steroids on the hippocampus must be considered. Electrophysi-
ological testing will typically be warranted as a level 2 assessment when there is a reasonable
concern that a new ingredient will affect neuronal function.

Behavior
Ultimately the preclinical assessment must shed light on any positive or negative effects
on behavior since it is the summation of the efferent expression of brain activity. An
assessment of general behavior of the animal model should be performed with any added
ingredient. This is part of standard animal care during experiments. Particular attention
should be paid to changes in typical behaviors, such as activity level, feeding, and comfort
seeking. In the case of suspected specific regional or neurotransmitter effects, a more com-
prehensive second level evaluation targeting the potential neuropathology is required. Ex-
amples include specific motor, cognitive, and behavioral paradigms that map on the brain
systems at risk. With any behavioral assessment, cross-referenced behavioral assessments in
the motor and cognitive domain are important during the period of nutrient delivery to
assess acute effects. Just as importantly, down-stream, long-term effects must also be evalu-
ated to assess whether the added nutrient has permanent or transient effects and whether
there are any “sleeper” effects.
SUMMARY
Preclinical studies are a vital first step to assess the safety and quality of ingredients new
to infant formulas. Regulatory guidelines for preclinical studies must be based on consider-
ations of the diversity of the potential new ingredients and the ingredients’ source and
matrix. In the United States, the FDA Redbook provides comprehensive guidelines for
conducting preclinical studies to test the safety of food and color additives, but it often does
not take the many special needs and vulnerabilities of infants into consideration.
The committee recommends a two-level preclinical approach with respect to assessing
the safety of ingredients new to infant formulas. The approach should take into consider-
ation model systems that are appropriate for the developing infant. All proposed new
ingredients require assessments using level 1 techniques (Tables 5-1 through 5-10) at the
discretion of the expert panel. Any proposed new ingredient with expected effects on struc-
ture or function, either on a theoretical basis or as a result of level 1 tests, will require
appropriately targeted level 2 evaluations (Tables 5-1 through 5-10).
REFERENCES
Aeschbacher HU, Finot PA, Wolleb U. 1983. Interactions of histidine-containing test substances and extraction
methods with Ames mutagenicity test. Mutat Res 113:103–116.
Alcantara O, Obeid L, Hannun Y, Ponka P, Boldt DH. 1994. Regulation of protein kinase C (PKC) expression by
iron: Effect of different iron compounds on PKC-beta and PKC-alpha gene expression and role of the 5'-
flanking region of the PKC-beta gene in the response to ferric transferrin. Blood 84:3510–3517.
Anderson NL, Anderson NG. 1998. Proteome and proteomics: New technologies, new concepts and new words.
Electrophoresis 19:1853–1861.
Arbuckle LD, MacKinnon MJ, Innis SM. 1994. Formula 18:2(n-6) and 18:3(n-3) content and ratio influence long-
chain polyunsaturated fatty acids in the developing piglet liver and central nervous system. J Nutr 124:289–
298.
Bachevalier J. 2001. Neural bases of memory development: Insights from neuropsychological studies in non-
human primates. In: Nelson CA, Luciana M, eds. Handbook of Developmental Cognitive Neuroscience.
Cambridge, MA: MIT Press. Pp. 365–379.
Benson JD, Masor ML. 1994. Infant formula development: Past, present and future. Endocr Regul 28:9–16.
Birch DG, Birch EE, Hoffman DR, Uauy RD. 1992. Retinal development in very-low-birth-weight infants fed diets
differing in omega-3 fatty acids. Invest Ophthalmol Vis Sci 33:2365–2376.
Bogyo M, Hurley JH. 2003. Proteomics and genomics. Curr Opin Chem Biol 7:2–4.
(Food and Drugs Act). Ottawa: Minister of Public Works and Government Services Canada.


Casey BJ, Thomas KM, McCandliss B. 2001. Applications of magnetic resonance imaging to the study of develop-
ment. In: Nelson CA, Luciana M, eds. Handbook of Developmental Cognitive Neuroscience. Cambridge,
MA: MIT Press. Pp. 137–147.
Chen Q, Beard JL, Jones BC. 1995. Abnormal rat brain monoamine metabolism in iron deficiency anemia. J Nutr
Biochem 6:486–493.
Clarke SD. 2001. Polyunsaturated fatty acid regulation of gene transcription: A molecular mechanism to improve
the metabolic syndrome. J Nutr 131:1129–1132.
Codex Alimentarius Commission. 2003. Codex Principles and Guidelines on Foods Derived from Biotechnology.
Prepublication. Online. Food and Agriculture Organization, World Health Organization. Available at
ftp://ftp.fao.org/codex/2003Prepublicationtexts/Codex_Texts_on_Biotechnology(E).pdf. Accessed January 12,
2004.
Cohen P, Miyazaki M, Socci ND, Hagge-Greenberg A, Liedtke W, Soukas AA, Sharma R, Hudgins LC, Ntambi
JM, Friedman JM. 2002. Role for stearoyl-CoA desaturase-1 in leptin-mediated weight loss. Science 297:240–
243.
Craig-Schmidt MC, Stieh KE, Lien EL. 1996. Retinal fatty acids of piglets fed docosahexaenoic and arachidonic
acids from microbial sources. Lipids 31:53–59.
Daniel H. 2002. Genomics and proteomics: Importance for the future of nutrition research. Br J Nutr 87:S305–
S311.
de Ungria M, Rao R, Wobken JD, Luciana M, Nelson CA, Georgieff MK. 2000. Perinatal iron deficiency de-
creases cytochrome c oxidase (CytOx) activity in selected regions of neonatal rat brain. Pediatr Res 48:169–
176.
Delion S, Chalon S, Herault J, Guilloteau D, Besnard JC, Durand G. 1994. Chronic dietary alpha-linolenic acid
deficiency alters dopaminergic and serotoninergic neurotransmission in rats. J Nutr 124:2466–2476.
DeLucchi C, Pita ML, Faus MJ, Molina JA, Uauy R, Gil A. 1987. Effects of dietary nucleotides on the fatty acid
composition of erythrocyte membrane lipids in term infants. J Pediatr Gastroenterol Nutr 6:568–574.
DeRisi JL, Iyer VR. 1999. Genomics and array technology. Curr Opin Oncol 11:76–79.
Dobbing J. 1990. Vulnerable periods in developing brain. In: Dobbing J, ed. Brain, Behaviour, and Iron in the
Infant Diet. New York: Springer-Verlag. Pp. 1–25.
Erikson KM, Jones BC, Beard JL. 2000. Iron deficiency alters dopamine transporter functioning in rat striatum. J
Nutr 130:2831–2837.
FAO/WHO (Food and Agriculture Organization of the United Nations/World Health Organization). 2001. Evalu-
ation of Allergenicity of Genetically Modified Foods. Joint FAO/WHO Expert Consultation on Allergenicity
of Foods Derived from Biotechnology. Rome: FAO.
Fuster JM, Alexander GE. 1971. Neuron activity related to short-term memory. Science 173:652–654.
Georgieff MK, Rao R. 2001. The role of nutrition in cognitive development. In: Nelson CA, Luciana M, eds.
Handbook of Developmental Cognitive Neuroscience. Cambridge, MA: MIT Press. Pp. 491–504.
Gietzen DW. 2000. Amino acid recognition in the central nervous system. In: Berthoud HR, Seeley RJ, eds. Neural
and Metabolic Control of Macronutrient Intake. New York: CRC Press. Pp. 339–357.
Gietzen DW, Magrum LJ. 2001. Molecular mechanisms in the brain involved in the anorexia of branched-chain
amino acid deficiency. J Nutr 131:851S–855S.
Golub MS, Gershwin ME. 1984. Standardized neonatal assessment in the rhesus monkey. In: Nathanielsz PW,
Parer JT, eds. Research in Perinatal Medicine (I). Ithaca, NY: Perinatology Press. Pp. 55–86.
Golub MS, Gershwin ME, Hurley LS, Hendrickx AG, Saito WY. 1985. Studies of marginal zinc deprivation in
rhesus monkeys: Infant behavior. Am J Clin Nutr 42:1229–1239.
Govorun VM, Archakov AI. 2002. Proteomic technologies in modern biomedical science. Biochemistry 67:1109–
1123.
Guengerich FP. 2001. Functional genomics and proteomics applied to the study of nutritional metabolism. Nut
Rev 59:259–263.
Hahn MJ. 2000. Letter to Linda Kahl on February 29, 2000. GRAS notification on behalf of Martek Biosciences
Corporation.
Haubner LY, Stockard JE, Saste MD, Benford VJ, Phelps CP, Chen LT, Barness L, Wiener D, Carver JD. 2002.
Maternal dietary docosahexanoic acid content affects the rat pup auditory system. Brain Res Bull 58:1–5.
Hochstrasser DF. 1998. Proteome in perspective. Clin Chem Lab Med 36:825–836.
Hrboticky N, MacKinnon MJ, Innis SM. 1990. Effect of a vegetable oil formula rich in linoleic acid on tissue fatty
acid accretion in the brain, liver, plasma, and erythrocytes of infant piglets. Am J Clin Nutr 51:173–182.
Innis MA. 1990. PCR Protocols: A Guide to Methods and Applications. San Diego: Academic Press.
Innis SM. 1992. Human milk and formula fatty acids. J Pediatr 120:S56–S61.

Jungblut PR, Zimny-Arndt U, Zeindl-Eberhart E, Stulik J, Koupilova K, Pleissner KP, Otto A, Muller EC,
Sokolowska-Kohler W, Grabher G, Stoffler G. 1999. Proteomics in human disease: Cancer, heart and infec-
tious diseases. Electrophoresis 20:2100–2110.
Knippels LMJ, Penninks AH. 2003. Assessment of the allergic potential of food protein extracts and proteins on
oral application using the brown Norway rat model. Environ Health Perspect 111:233–238.
Kretchmer N, Beard JL, Carlson S. 1996. The role of nutrition in the development of normal cognition. Am J Clin
Nutr 63:997S–1001S.
Kriete MF, Champoux M, Suomi SJ. 1995. Development of iron deficiency anemia in infant rhesus macaques. Lab
Anim Sci 45:15–21.
Lander ES. 1999. Array of hope. Nat Genet 21:3–4.
Lee CK, Weindruch R, Prolla TA. 2000. Gene-expression profile of the ageing brain in mice. Nat Genet 25:294–
297.
Lee KH, Calikoglu AS, Ye P, D’Ercole AJ. 1999. Insulin-like growth factor-I (IGF-I) ameliorates and IGF binding
protein-1 (IGFBP-1) exacerbates the effects of undernutrition on brain growth during early postnatal life:
Studies in IGF-I and IGFBP-1 transgenic mice. Pediatr Res 45:331–336.
Lozoff B, Jimenez E, Hagen J, Mollen E, Wolf AW. 2000. Poorer behavioral and developmental outcome more
than 10 years after treatment for iron deficiency in infancy. Pediatrics 105:E51.
Lucas A, Morley R, Cole TJ. 1998. Randomised trial of early diet in preterm babies and later intelligence quotient.
Br Med J 317:1481–1487.
MacBeath G. 2002. Protein microarrays and proteomics. Nat Genet 32:526–532.
Magrum LJ, Hickman MA, Gietzen DW. 1999. Increased intracellular calcium in rat anterior piriform cortex in
response to threonine after threonine deprivation. J Neurophysiol 81:1147–1149.
McGregor DB, Martin R, Cattanach P, Edwards I, McBride D, Caspary WJ. 1987. Responses of the L5178Y tk+/
tk– mouse lymphoma cell forward mutation assay to coded chemicals. I: Results for nine compounds. Environ
Mutagen 9:143–160.
McGregor DB, Brown A, Cattanach P, Edwards I, McBride D, Caspary WJ. 1988a. Responses of the L5178Y tk+/
tk– mouse lymphoma cell forward mutation assay to coded chemicals. II: 18 coded chemicals. Environ Mol
Mutagen 11:91–118.
McGregor DB, Brown A, Cattanach P, Edwards I, McBride D, Riach C, Caspary WJ. 1988b. Responses of the
L5178Y tk+/tk– mouse lymphoma cell forward mutation assay to coded chemicals. III: 72 coded chemicals.
Environ Mol Mutagen 12:85–154.
Metcalfe DD, Astwood JD, Townsend R, Sampson HA, Taylor SL, Fuchs RL. 1996. Assessment of the allergenic
potential of foods derived from genetically engineered crop plants. Crit Rev Food Sci Nutr 36:S165–S186.
Miles EA, Calder PC. 1998. Modulation of immune function by dietary fatty acids. Proc Nutr Soc 5:277–292.
Moreno-Aliaga MJ, Marti A, Garcia-Foncillas J, Alfredo Martinez J. 2001. DNA hybridization arrays: A powerful
technology for nutritional and obesity research. Br J Nutr 86:119–122.
Morrow-Tlucak M, Haude RH, Ernhart CB. 1988. Breastfeeding and cognitive development in the first 2 years of
life. Soc Sci Med 26:635–639.
Mortensen EL, Michaelsen KF, Sanders SA, Reinisch JM. 2002. The association between duration of breastfeeding
and adult intelligence. J Am Med Assoc 287:2365–2371.
Myhr BC, Caspary WJ. 1988. Evaluation of the L5178Y mouse lymphoma cell mutagenesis assay: Intralaboratory
results for sixty-three coded chemicals tested at Litton Bionetics, Inc. Environ Mol Mutagen 12:103–194.
Myhr B, McGregor D, Bowers L, Riach C, Brown AG, Edwards I, McBride D, Martin R, Caspary WJ. 1990.
L5178Y mouse lymphoma cell mutation assay results with 41 compounds. Environ Mol Mutagen 16:138–
167.
Nelson CA. 1995. The ontogeny of human memory: A cognitive neuroscience perspective. Devel Psychol 31:723–
738.
Nelson CA, Bloom FE, Cameron JL, Amaral D, Dahl RE, Pine D. 2002. An integrative, multidisciplinary approach
to the study of brain-behavior relations in the context of typical and atypical development. Dev Psychopathol
14:499–520.
Neuringer M, Connor WE, Lin DS, Barstad L, Luck S. 1986. Biochemical and functional effects of prenatal and
postnatal omega 3 fatty acid deficiency on retina and brain in rhesus monkeys. Proc Natl Acad Sci USA
83:4021–4025.
Ntambi JM. 1999. Regulation of stearoyl-CoA desaturase by polyunsaturated fatty acids and cholesterol. J Lipid
Res 40:1549–1558.


OFAS (Office of Food Additive Safety). 2003. Redbook 2000. Toxicological Principles for the Safety of Food
Ingredients. Online. Center for Food Safety and Applied Nutrition, Food and Drug Administration. Available
at http://www.cfsan.fda.gov/~redbook/red-toca.html. Accessed November 19, 2003.
Overman WH, Bachevalier J. 2001. Inferences about the functional development of neural systems in children via
the application of animal tests of cognition. In: Nelson CA, Luciana M, eds. Handbook of Developmental
Cognitive Neuroscience. Cambridge, MA: MIT Press. Pp. 109–124.
Pasquier E, Ratnayake WM, Wolff RL. 2001. Effects of delta5 polyunsaturated fatty acids of maritime pine (Pinus
pinaster) seed oil on the fatty acid profile of the developing brain of rats. Lipids 36:567–574.
Perou CM, Jeffrey SS, van de Rijn M, Rees CA, Eisen MB, Ross DT, Pergamenschikov A, Williams CF, Zhu SX,
Lee JCF, Lashkari D, Shalon D, Brown PO, Botstein D. 1999. Distinctive gene expression patterns in human
mammary epithelial cells and breast cancers. Proc Natl Acad Sci USA 96:9212–9217.
Reisbick S, Neuringer M, Hasnain R, Connor WE. 1994. Home cage behavior of rhesus monkeys with long-term
deficiency of omega-3 fatty acids. Physiol Behav 55:231–239.
Rao R GM. 2000. Early nutrition and brain development. In: Nelson C, ed. The Minnesota Symposia on Child
Psychology. The Effects of Adversity on Neurobehavioral Development. Vol. 31. Mahway, NJ: Lawrence
Erlbaum Associates. Pp. 1–30.
Rao R, de Ungria M, Sullivan D, Wu P, Wobken JD, Nelson CA, Georgieff MK. 1999. Perinatal brain iron
deficiency increases the vulnerability of rat hippocampus to hypoxic ischemic insult. J Nutr 129:199–206.
Rice D, Barone S. 2000. Critical periods of vulnerability for the developing nervous system: Evidence from humans
and animal models. Environ Health Perspect 108:511–533.
Sampson HA. 2002. Peanut allergy. New Engl J Med 346:1294–1299.
Sapolsky RM. 1994. The physiological relevance of glucocorticoid endangerment of the hippocampus. Ann N Y
Acad Sci 746:294–304.
Schlegel R, MacGregor JT. 1982. The persistence of micronuclei in peripheral blood erythrocytes: Detection of
chronic chromosome breakage in mice. Mutat Res 104:367–369.
Schlegel R, MacGregor JT, Everson RB. 1986. Assessment of cytogenetic damage by quantitation of micronuclei in
human peripheral blood erythrocytes. Cancer Res 46:3717–3721.
Shimomura I, Shimano H, Korn BS, Bashmakov Y, Horton JD. 1998. Nuclear sterol regulatory element-binding
proteins activate genes responsible for the entire program of unsaturated fatty acid biosynthesis in transgenic
mouse liver. J Biol Chem 273:35299–35306.
Soukas A, Cohen P, Socci ND, Friedman JM. 2000. Leptin-specific patterns of gene expression in white adipose
tissue. Genes Dev 14:963–980.
Squire LR. 1992. Declarative and nondeclarative memory: Multiple brain systems supporting learning and memory.
J Cogn Neurosci 4:232–243.
Tkáč I, Rao R, Georgieff MK, Gruetter R. 2003. Developmental and regional changes in the neurochemical profile
of the rat brain determined by in vivo 1H NMR spectroscopy. Magn Reson Med 50:24–32.
Uauy-Dagach R, Mena P. 1995. Nutritional role of omega-3 fatty acids during the perinatal period. Clin Perinatol
22:157–175.
Vazquez E, Herrero I, Miras-Portugal MT, Sanchez-Prieto J. 1994. Role of arachidonic acid in the facilitation of
glutamate release from rat cerebrocortical synaptosomes independent of metabotropic glutamate receptor
responses. Neurosci Lett 174:9–13.
Vihinen M. 2001. Bioinformatics in proteomics. Biomol Eng 18:241–248.
Wainwright PE, Xing HC, Mutsaers L, McCutcheon D, Kyle D. 1997. Arachidonic acid offsets the effects on
mouse brain and behavior of a diet with a low (n-6):(n-3) ratio and very high levels of docosahexaenoic acid.
J Nutr 127:184–193.
Wang J. 2000. From DNA biosensors to gene chips. Nucleic Acids Res 28:3011–3016.
Wang YS, Wu SY. 1996. The effect of exclusive breastfeeding on development and incidence of infection in
infants. J Hum Lact 12:27–30.
Williams KL. 1999. Genomes and proteomes: Towards a multidimensional view of biology. Electrophoresis
20:678–688.

Going Beyond Current Clinical Studies
ABSTRACT
Clinical studies are essential to ensure the safety of infant formulas and any
systematic deviation from normal physical growth and development attributable to
a new ingredient should be considered a safety threat. Growth studies, currently a
centerpiece of clinical evaluation of infant formulas, should include precise and reli-
able measurements of weight and length velocity and head circumference. Appropri-
ate measures of body composition also require assessment. Duration of follow-up
measurements should at least cover the period when infant formula remains the sole
source of nutrients in the diet of the infant. However the committee believes that
growth studies are not sufficient on their own to assess ingredients new to infant
formulas. Specific guidelines are needed to determine “normal” growth and to estab-
lish what represents a biologically meaningful difference among groups of infants
consuming different formulas. Specific recommendations are needed to establish a
level of difference that represents a safety concern.
Regulatory guidelines should ensure that infant outcomes encompass, as the
Food and Drug Administration (FDA) has proposed, “all aspects of physical growth
and normal maturational development.” Any systematic differences in clinical out-
comes that can be attributed to an ingredient new to infant formulas should be
considered a safety concern that requires careful evaluation and, if needed, further
clinical study to identify the pathway through which the infant has been affected.
The committee recommends that a hierarchy of two levels of clinical assessment be
implemented with regard to growth and organ systems. Level 1 assessments should
include checking for signs of all adverse laboratory indicators of the major organ
systems. Level 2 assessments should include in-depth measures of organ systems or
functions that would be performed to explain abnormalities found in level 1 assess-
ments or specific theoretical concerns not typically addressed by level 1 tests.
98

GOING BEYOND CURRENT CLINICAL STUDIES 99
There are a number of reasons why it is equally important to include develop-

mental-behavioral outcomes in future studies of the safety of ingredients new to
infant formulas: the measures are sensitive to exposure to toxic substances, they can
have long-term predictive value, and bidirectional brain-behavior links exist. There-
fore, assessment of clinical endpoints should include measurement of infant sensory-
motor, cognitive, affectual, and neural function with instruments that follow recom-
mended criteria. The committee recommends that a hierarchy of three levels of
clinical assessment be developed and implemented to determine what levels are ap-
propriate to apply with regard to developmental-behavioral-neural outcomes. The
levels of assessment are: level 1 assessments, including developmental screening mea-
sures; level 2 assessments, including in-depth measures of infant functions in major
developmental areas (single assessment for each area with one instrument); and level
3 assessments, including in-depth measures of infant functions in major developmen-
tal areas (repeated assessment with multiple instruments).
The instruments used for these assessments should satisfy the following criteria:
be age appropriate, have predictive value for long-term consequences, be adequately
sensitive, have documented brain-behavior links, have cross-species generalizability,
assess specific function, and be easy to administer. In addition, the committee con-
siders that certain design features (e.g., adequate statistical power) are essential in all
clinical studies.
INTRODUCTION
This chapter provides an overview of clinical studies and a brief overview of the
current regulatory requirements for them. The first part of the chapter includes a rationale
for clinical assessment of growth, specific recommendations on what should be measured,
and guidelines for interpretation of results. In the second part, the committee describes
more specific clinical endpoints in each of the organ systems likely to be affected by
ingredients new to infant formulas. In the last part of the chapter considerable attention
is paid to behavioral and developmental endpoints because of the young infant’s height-
ened sensitivity to potentially toxic substances and the long-term consequences of such
exposures.
THE IMPORTANCE OF CLINICAL STUDIES

While preclinical laboratory and animal studies have substantial value for identifying
potential safety concerns, they are limited in their ability to predict what may happen in
human infants. Clinical studies in human infants are needed for several reasons. First,
extrapolation from animal studies may be limited by differences between animal and
human structure, physiology, and development. Second, extrapolation from isolated tissue
studies is limited by the inability of such models to assess functions in the context of whole
organ systems where coordination and integration are the rule. For example, the digestion
and absorption of nutrients requires coordination of numerous gastrointestinal functions.
Third, there may be no available animal or tissue models to test specific functions. For
example, it is not possible to use animal models to duplicate clinically relevant allergic
reactions to foreign proteins, to determine the effects of a substance on acceptance or
tolerance of an infant formula, or to test some of the higher cognitive functions found only
in humans.

CURRENT REGULATORY GUIDELINES FOR CLINICAL STUDIES
Canada’s Food and Drug Regulations

There are no specific requirements for clinical testing of infant formulas set out under
Canada’s Food and Drug Regulations in Division 16 (Food Additives), Division 25 (Infant
Formula), or Division 28 (Novel Foods) (Canada, 2001). Division 25 of the Regulations
requires that a premarket submission with respect to a new infant formula or an infant
formula that has undergone a major change in composition, manufacturing, or packaging
include the evidence relied on to establish that the infant formula is nutritionally adequate to
promote acceptable growth and development in infants when consumed in accordance with
the directions for use. Divisions 16 and 28 require that data be submitted to Health Canada
that include information used to establish the safety of a food additive or a novel food,
respectively. Health Canada refers manufacturers to internationally accepted guidelines for
clinical testing or asks to be consulted because decisions are made on a case-by-case basis.
Sections 409 and 412 of the Federal Food, Drug and Cosmetic Act
There are no explicit requirements for clinical testing of infant formulas specified under
Section 409 of the Food, Drug and Cosmetic (FD&C) Act. Section 409 stipulates that a
petition to establish safety of a food additive shall contain “all relevant data bearing on the
physical or other technical effect such additive is intended to produce . . . ,” but it does not
dictate a specific type of clinical study.
Current regulations for infant formulas under Section 412 of the FD&C Act do not
define quality factor requirements, such as physical growth, but only describe required
nutrient levels, without considering bioavailability. This gap is addressed in a proposed rule
(FDA, 1996), where assessment of physical growth, using anthropometry, is proposed “as
an integrative indicator of net overall nutritional quality of the formula.” The proposed rule
further states, “as the science evolves, FDA anticipates being able to progress beyond gener-
alized, nonspecific indicators of overall nutritional intakes (e.g., measures of physical growth)
to more specific and sensitive measures of biochemical and functional nutritional status”
(FDA, 1996, P. 36181). Thus neither the current nor the proposed rules identify specific
requirements for other clinical studies.
FDA Redbook
FDA does not require petitioners to conduct human clinical studies to support the safety
of food additives or color additives used in food, but, if deemed necessary, it recommends
that the studies conform to guidelines presented in section VI.A. of the Redbook (OFAS,
2001, 2003). These guidelines are comprehensive and relevant for the clinical testing of
ingredients new to infant formulas.
General guidance is provided to identify the scientific and ethical principles for clinical
studies, including the need for presentation of a defensible rationale for human studies. The
Redbook states that this rationale should be based on:
• adequate preclinical investigations,

• results of clinical studies conducted elsewhere,
• consideration of the organs and organ systems that may be affected, and

• careful attention to the qualifications of investigators and the safety and ethical
treatment of subjects in clinical trials.
The Redbook suggests the sequence of and subjects for clinical studies. Early clinical
studies are to determine the “metabolism and level of the food or food additive that gives an
adverse or toxic response in man” (specifically physiological studies of the additive’s dispo-
sition, its potential to induce enzyme levels or increase activity, and its interactions with
other nutrients) (OFAS, 2001, P. 183). In general children are to be excluded from these
early (typically acute or shorter duration) clinical studies. However tolerance studies, which
are to be included among early studies, need to be conducted in infants because of the special
nature of infant formulas.
Infants are more likely to be included in what the Redbook describes as chronic intake
studies, which are to be conducted once general safety in humans is established in the early
adult studies. Here, the Redbook provides specific guidance on protocol design, study
population, and statistical analyses, as well as on how reports of clinical studies should be
presented. Box 6-1 lists questions that should be answered when conducting studies to
determine the safety of a proposed additive.
GENERAL APPROACH TO CONDUCTING CLINICAL STUDIES

In the conceptualization of the range of infant health concerns, the committee was
guided by the following: “FDA considers the concept of ‘healthy growth’ to be broad,
encompassing all aspects of physical growth and normal maturational development, includ-
ing maturation of organ systems and achievement of normal functional development of
motor, neurocognitive, and immune systems. All of these growth and maturational devel-
opmental processes are major determinants of an infant’s ability to achieve his/her biolog-
ical potential, and all can be affected by the nutritional status of an infant” (FDA, 1996,
P. 36179).
The committee proposes the use of a multilevel approach to establish more comprehensive
guidelines to ensure that infant outcomes encompass “all aspects of physical growth and
normal maturational development.” Figure 6-1 illustrates the three different types of clinical
studies recommended by the committee, including assessment of growth, organ systems, and
development and behavior. Figures 6-2 and 6-3 further explain the clinical studies through the
proposed two-level approach to organ systems and the three-level approach to development-
BOX 6-1 Questions That Should Be Answered

When Conducting Clinical Studies
• How is the food or food additive absorbed, metabolized, deposited in tissue, and excreted?
• What is the half-life of the food or food additive in the human body?
• How may interactions between the food or food additive and nutrients or medications compro-
mise the availability of any of these substances (including the consideration of the matrix)?
• How does the food or food additive affect the function of human organs and organ systems
(including infant growth and development)?
• What are the possible adverse reactions to the food or food additive in the general population of
individuals who are likely to use the substance and in special (more sensitive) populations?
SOURCE: OFAS (2001, 2003).

PROPOSED CLINICAL ASSESSMENT

1
formula

development
Assess:
- Weight velocity
3 - Length velocity
Grow th Studies - Head circumference
(See Sidebar A) - Body composition
and
Clinical Endpoints
(See Sidebar B and Figure 6-2)
- Kidney
- Blood
4
Abnormal growth or
Yes
specific organ, immune, or ingredient where appropriate
endocrine systems
Sidebar C:
Assessment
6
Assess:
(See Sidebar C and Figure 6-3)
- T emperament
7 5

Yes
No
8
IS SAFE
FIGURE 6-1 Proposed clinical assessment algorithm. = a state or condition, =



1
form ul a

li nk to major organ system s?
Y es Sidebar A: Level 1 Assessment
Major organ systems screening

No
measures in the following:
6 - Gastrointestinal tract
- Liver
Adverse effect/event documented - Kidney
in precli ni cal trials? - Blood
OR - Immune
Y es
Evi dence of signifi cant individual - Endocrine
difference i n susceptibil ity to the (See Tables 6-3, 6-5, 6-6, 6-8, and 6-9)
ingredient?
No

Lev el 1 Assessment in the foll owi ng:
- Liver
- Kidney
- Blood
8 - Immune
- Endocrine
Evi dence of
(See Tabl es 6-3, 6-5, 6-6, 6-8, and 6-9)
adverse Yes
effect/event?
3
Lev el 2 Assessment
(See Sidebar B)
No 4
Evi dence of
adverse Yes
effect/event?
No
9 5
Continue to
DISCONTINUE
neurobehavioral clinical
PROCESS
studi es
FIGURE 6-2 Proposed levels of clinical assessment of major organ, immune, and endocrine systems
algorithm. = a state or condition, = a decision point, = an action,
sidebar = an elaboration of recommendation or statement.


1

formula
Sidebar A: Developmental-Behavioral
Assessment
Cri teria for choosing neural -behavioral
2 assessment measures:
- Age appropri ateness
Known or theoreti cal li nk to - Predictive value
Yes
neurobehavior? - Sensi tivi ty
- Cross-species general izabil ity
No
Adverse effect/event Study design requirements:
documented in preclinical - Adequate stati stical power
trial s? - Avoi d over-control of mediator variabl es
OR - Use measurement aggregation
Yes
Evidence of significant - Use repeated measures
ingredient?
No
11 measures admi nistered duri ng a routine
Known or theoreti cal well-baby physical exam or through
indi rect l ink to other Yes parent reports. (See Tabl e 6-10)
organ systems?
No Sidebar C: Level 2 Assessment

12
Detai led measures of function in maj or
Lev el 1 Assessment child developmental areas. Si ngle
(See Sidebars A and B) assessment for each area using one
instrument.
(See T ables 6-11 through 6-15)
13
Sidebar D: Level 3 Assessment
8
Evidence of
adverse Yes
(See Sidebars A and C) child developmental areas on at least two
effect/event?
separate occasions using two
9 3
Evidence of
Lev el 3 Assessment
adverse Yes
effect/event?
No
4
Evidence of
effect/adverse Yes
event?
No
No
14 10 6 5
MANUFACTURER/REGU
AGENCY DETERMINES
INGREDIENT IS SAFE
INGREDIENT IS SAFE
FIGURE 6-3 Proposed levels of clinical assessment of development and behavior algorithm.
= a state or condition, = a decision point, = an action, sidebar = an
elaboration of recommendation or statement.

behavior. There are decision-making points within each of these three types of clinical studies
that will be discussed in detail in subsequent sections of this chapter. (In keeping with the
charge to the committee, proposed guidelines focus on the health and well-being of term
infants only.)
The committee recognizes that all clinical studies would need to be reviewed and
approved by human-subject research review boards. Because the clinical studies to deter-
mine the safety of new ingredients will be carried out in healthy infants, the committee
does not recommend the use of highly invasive tests, such as tissue biopsies or gastrointes-
tinal incubations.

RECOMMENDATION: Any adverse systematic differences in clinical outcomes that
can be attributed to an ingredient new to infant formulas should be considered a safety
concern that requires careful evaluation and, if needed, further clinical study to identify
the pathway through which the infant has been affected.
A hierarchy of two levels of clinical assessment should be implemented for organ systems:
• Level 1 assessments. Check of signs for all adverse laboratory indicators.

• Level 2 assessments. In-depth measures of organ systems or functions that would
be performed to explain abnormalities found in level 1 assessments or specific theoreti-
cal concerns not typically addressed by level 1 tests.
A hierarchy of three levels of clinical assessment should be implemented for develop-

mental-behavioral measures:
• Level 1 assessments. Developmental screening measures.

• Level 2 assessments. In-depth measures of infant functions in major developmen-
tal areas (single assessment for each area with one instrument).
tal areas (repeated assessment with multiple instruments).
GROWTH
Growth is well recognized as a sensitive, but nonspecific, indicator of the overall health
and nutritional status of an infant. Monitoring infant growth has always been an integral
part of pediatric care and is particularly important for young infants. Growth and nutrient
requirements per kilogram of body weight are higher during the first few months of infancy
than during any other period of life. Furthermore, the greatest percentage of dietary intake
is devoted to supporting growth at this time, and thus nutritional imbalances are likely to be
reflected in growth rates.
The committee believes that the inability of a formula to support normal growth repre-
sents a significant harm to infants and therefore growth is an essential endpoint for all safety
assessments of an ingredient new to infant formulas. Any systematic deviation from normal
physical growth attributable to a new ingredient should be considered a safety threat.
Under current regulations the core of the requirements focuses on meeting certain levels
of specific nutrients. The concept of quality factors has not been defined, but proposed

regulations include a subsection on quality factors, with a focus on physical growth. Despite
the absence of quality factors in current legislation, there appears to be a strong consensus
that growth should be a quality factor for infant formulas. In the United States FDA
recognized the need for clear guidelines on the assessment of growth and commissioned a
report from the American Academy of Pediatrics’ (AAP) Committee on Nutrition Task
Force on clinical testing of infant formulas with respect to nutritional suitability for term
infants (AAP, 1988). The task force identified the following types of clinical studies as useful
in the premarket evaluation of formulas: acceptance or tolerance studies, gains in weight and
length, food intake, body composition, serum chemical indices, and metabolic balance
studies. Most of the recommendations of the task force were incorporated into the proposed
changes to the infant formula act (FDA, 1996).
Currently clinical studies tend to follow the proposed rule, the 120-day growth study
being the main method used to assess the ability of an infant formula to sustain normal
infant growth. The proposed rule would codify standards for clinical growth studies by
specifying methods (controlled clinical trials), duration (4 months), measurements (weight,
recumbent length, and head circumference), and ages at measurement (at 2 and 4 weeks,
then at least monthly thereafter), with a further requirement that individual infant data be
plotted against Centers for Disease Control and Prevention (CDC) reference curves for
weight and length.1
The AAP task force concluded that “rate of gain in weight gain is the single most
valuable component of the clinical evaluation of infant formula” (AAP, 1988, P. 7). Further,
it judged that length assessment is unnecessary because significant differences in length gain
would not occur in the absence of differences in weight gain, and that there is a higher
potential for measurement error and thus misclassification of growth in length. While the
committee concurs with the centrality of weight gain in clinical assessment, it also believes
that length and head circumference should be measured in growth studies in order to
evaluate the effects of substances on other aspects of growth, such as skeletal growth and
body proportions.
Notably absent from existing and proposed requirements are specific guidelines on what
constitutes “normal” growth, or what represents a biologically meaningful difference among
groups of infants consuming different formulas. Recommendations are needed both to
define the most relevant comparison groups for clinical studies and to establish a level of
difference that represents a safety concern. These are challenging and critical questions that
will be discussed in later sections.
In addition, the committee recommends that guidelines go beyond growth studies to
assess the safety of ingredients new to infant formulas. Deficits in brain function and effects
of specific micronutrients may occur in the absence of differences in physical growth. Fur-
thermore, while a “decrease in the growth rate during infancy is the earliest indication of
nutritional failure” (Fomon, 1993, P. 48), growth deficits are likely to appear only second-
ary to effects on specific organs or tissues, and they may not appear for some time after
nutritional insult. Thus growth studies should be considered a necessary, but not sufficient,
part of human clinical studies of the safety of ingredients new to infant formulas (see Figure
6-1, Box 3).
1Proposed changes to 21 C.F.R. Parts 106 and 107 specify the reference charts to be used. Since CDC has
published updated references for use in the United States (Kuczmarski et al., 2000), the requirement should be
updated to specify the new reference values.

Measuring Growth
Ascertainment of growth status typically relies on anthropometric assessment, which is
noninvasive and highly practical, requires relatively little training to achieve reliability, and
is accomplished with low-cost, low-tech tools. Further, there are ample descriptions of
standard anthropometric methods and reference data for the interpretation of measurements
(Kuczmarski et al., 2000; Lohman et al., 1988). Although each has limitations and advan-
tages (Table 6-1), the committee recommends the following measures of infant growth for
clinical studies (see Figure 6-1, Box 3):
• Weight is an overall measure of body size and is responsive to acute insults, such as
infectious morbidity or changes in nutrient intakes. Attained weight is hard to interpret in
the absence of length data since an underweight child could be well proportioned or thin,
with different implications for morbidity risk.
• Recumbent length is an overall indicator of linear or bone growth. Length reflects
genetic factors and growth history. It is less responsive to acute insults, and the response of
length to varying nutrition levels typically lags behind the response in weight.
• Weight for length is an indicator of relative weight (thinness or overweight). These
measures are typically expressed as a Z-score or a percentile based on comparison with
national reference data.
• Head circumference is often used in clinical settings as an overall, nonspecific indica-
tor of brain growth. It has limited usefulness in screening for potential developmental or
neurological disabilities, but it is useful in comparison with other anthropometrics to assess
proportionality. The ratio of mid-arm to head circumference is a less commonly used index
of proportionality.
• Body composition is a more sensitive indicator of infant nutritional status than
measures of size. Depending on the method used, measurements can provide the mass of
lean tissue, fat tissue, total body water, and bone. Methods vary greatly in terms of invasive-
ness, feasibility, cost, technology, need for trained personnel, accuracy, reliability, and pre-
cision. The most feasible methods for assessing infant body composition include anthropom-
etry (e.g., skinfold measurements), dual X-ray absorptiometry (DEXA), and isotope dilution.
A recent review concluded that for intergroup comparisons, skinfold thicknesses were use-
ful, but for individual infant assessments, DEXA was recommended (Koo, 2000). In the
absence of reference data based on a large sample of infants, the interpretation of body
TABLE 6-1 Limitations and Advantages of Recommended Growth Assessments

Recommended
Assessment Limitations Advantages
Rate of weight gain Nonspecific Good global measure of infant growth and
health, easy to measure reliably
Rate of length gain Difficult to measure accurately, Provides important additional information
deficits less likely unless weight about linear/skeletal growth and
is also compromised proportionality
Head circumference Nonspecific Easy to measure accurately, adequate global
measure of head and brain growth and
proportionality
Body composition Difficult to measure accurately, More precise information about possible
best method requires expensive metabolic effects of ingredients, possible
equipment (dual-energy X-ray better long-term predictor of health
absorptiometry) outcomes

TABLE 6-2 Limitations and Advantages of Common Measurements of Body

Composition
Method Relevant Papers/Measurement Limitations Advantages
Skinfold Schmelzle and Fusch (2002); Can be inaccurate Rapid, low cost
body fat in neonates and young
infants: validation of skinfold
thickness versus dual-energy
X-ray absorptiometry
Dual-energy X-ray Butte et al. (1999); fat mass in Requires expensive Rapid, precisely
absorptiometry infants and toddlers: compara- equipment estimates bone
bility of total body water, total mineral content,
body potassium, total body fat mass, and
electrical conductivity, and dual- lean body mass
energy X-ray absorptiometry
Isotope dilution Expensive and needs Noninvasive, safe
specialized equipment
composition outcomes should rest on the comparison of groups in randomized controlled

trials. Additional information on the methods used to assess body composition is provided
in Table 6-2.
RECOMMENDATION: Growth studies should include precise and reliable measure-

ments of weight and length velocity and head circumference. Duration of measurements
should cover at least the period when infant formula remains the sole source of nutrients
in the infant diet. Appropriate measures of body composition also require assessment.
Defining Normal Growth

The purpose of growth assessment is to determine whether a child is growing “nor-
mally.” The definition of normal, inadequate, or excess growth rests largely on comparison
of individual measurements with reference data that represent the distribution of sizes found
in healthy infants of a given age and sex. While there is no clear cut point to define a size at
which there is an abrupt elevation in risk of poor outcomes, measurements that fall above
the 95th or below the 5th percentiles of an accepted reference are typically cause for
concern. While short periods of abnormal growth rate may not be of concern, low or high
rates over several months may be related to increased morbidity risk, both in the long and
short term. Therefore a single measurement of attained size at a given age is not a sufficient
measure of growth. Repeated, appropriately spaced measurements are needed to calculate
growth rate. A clinical assessment of infant growth for the purpose of determining the safety
of an ingredient new to infant formulas must therefore be based on a longitudinal study,
with repeated measures at relatively frequent intervals during the period when growth is
most rapid and during the time period when formula serves as the sole source of infant
nutrition.
Identifying Appropriate Comparison Groups

As discussed in Chapter 3, there are challenges in selecting appropriate comparison
groups for clinical studies to assess the safety of infant formulas. The gold standard design—
the double-blind, randomized, controlled trial—randomly assigns comparable groups of

infants to receive either the formula containing the new ingredient or a previously approved
formula. Implicit in this design is the assumption that infants fed the approved formula form
the appropriate comparison group. However when testing for deviation from optimal infant
growth, the appropriate comparison group should be one that demonstrates optimal growth.
Since growth of healthy breastfed infants is considered optimal, then exclusively breastfed
infants form the most appropriate comparison group. The committee recommends using
dual control groups—breastfed infants and infants fed the previously approved formula
without the new ingredient—in order to ensure a thorough analysis.
Breast versus formula feeding cannot ethically be randomly assigned, nor could these
feeding conditions be blinded. Instead, reference data from healthy breastfed infants, mea-
sured at comparable intervals using identical methods, can be used for comparative pur-
poses. This would allow multiple intergroup comparisons that would put differences be-
tween two infant formulas in perspective relative to formula-breastfeeding differences, as
has been done in trials of formula containing long-chain polyunsaturated fatty acids (LC-
PUFAs) (Auestad et al., 2001). The World Health Organization is currently working to
create a growth reference for breastfed infants that should be suitable for such comparisons
in the future (Garza and de Onis, 1999).
Estimating Intake
All clinical trials must include an estimation of daily formula intake in order to deter-
mine which effects are the result of different levels of intake and which are to the result of the
specific ingredient. For example, if an ingredient alters taste or palatability, it may change
the level of intake.
Specifying a Level of Difference in Growth That Represents a Safety Concern

This is important for interpreting average group differences, as well as deviations from
normal growth in an infant that are attributable to being fed a different formula. There is
very little scientific evidence to establish a level of difference associated with long- or short-
term health consequences. The AAP task force (AAP, 1988) recommended that a weight
gain difference greater than 3 g/day over 3 to 4 months should be considered nutritionally
significant. Over 3.5 months, this would represent a difference of about 320 g. This is less
than the difference between the 25th and 50th percentile and equivalent to the difference
between the 90th and 95th percentile of weight at age 3.5 months for boys based on the
CDC growth charts (Kuczmarski et al., 2002). No specific evidence to support this level of
difference was provided. In a clinical setting, a diagnosis of failure to thrive is based on rate
of weight gain and weight-for-length status interpreted in the context of growth history.
Fomon (1993) defines failure to thrive in comparison with U.S. reference data for weight
increments and weight-for-length status. He uses two standard deviations relative to 2-
month increment data for infants under 6 months of age and a weight-for-length below the
5th percentile as cutpoints.
For perspective, it may be useful to compare reported differences in growth rates of
healthy breastfed and formula-fed infants. Data from Fomon’s infant growth studies (Nelson
et al., 1989) show differences of 2.4 g/day in boys and 1.3 g/day in girls who were breastfed
versus formula fed from 8 to 112 days of life. This would result in a 250-g difference in boys
and a 135-g difference in girls over the 104 days. Based on more recent data, Dewey and
colleagues (1992) compared growth of infants in the DARLING study and found consis-
tently higher weight velocities (g/mo) in formula-fed versus breastfed boys in the first 6

months of life, but no significant effects of feeding mode on weight velocity in girls. The
cumulative effect of the differences in weight velocity among boys amounted to about 284 g
over a period of 4 months. Furthermore, infants who were breastfed for 12 months or more
were leaner, had smaller skinfolds, and had a lower percent body fat. These differences
persisted into the second year of life (Dewey et al., 1993).
Kramer and colleagues (2002) recently reported results of a large study of Belarus
infants and found that infants who were exclusively breastfed for at least 3 months had
weight and length Z scores that were about 0.2 standard deviations above those of infants
who were weaned in the first month of life. Using data from the Third National Health and
Nutrition Examination Survey, Hediger and colleagues (2000) found no differences in weight
status by feeding method in 4- to 7-month-old children, but between 8 and 11 months of
age, infants who had been exclusively breastfed had weights that were about one-fifth of a
standard deviation below the U.S. reference median. This would represent a 200-g difference
in growth associated with breastfeeding among infants of average size.
In comparison with published growth velocity reference data from the Fels Longitudinal
Study (Guo et al., 1991), 3 g/day roughly represents the differences between the major
percentile lines in the 3-month increment data (e.g., for boys, the 25th, 50th, and 75th
percentiles were 23, 27, and 31 g/day, respectively). The clinical or functional significance of
such differences is not well established.
Body composition is not typically assessed as a part of normal well-child care in clinical
settings. The interpretation of body composition measures has been particularly challenging
because extensive reference data on infants is lacking, and few studies have attempted to
identify specific health risks associated with levels of body fat or lean tissue. More than a
decade ago, the AAP task force concluded that methods to determine body water, body fat,
and bone mass had “not reached the stage of precision, noninvasiveness and convenience
that would make them feasible as a part of routine clinical testing of infant formulas” (AAP,
1988). However the state of the art has changed dramatically since then, and it is now
possible to assess body composition using a variety of minimally invasive and precise meth-
ods. Furthermore, Butte and colleagues (2000) recently published reference data for infant
body composition using a four-compartment model to estimate fat and fat-free mass, a
deuterium dilution to measure total body water, and DEXA to measure bone mineral
content.
It is important to evaluate body composition in the context of safety. Ultimately the goal
of assessment is to identify levels of difference in body composition that are associated with
immediate or long-term disease risk. The relevant component of body composition to mea-
sure will depend on the nature of the added ingredient. For example, if an ingredient is likely
to have metabolic effects, it will be important to assess the relative contribution of fat and
fat-free mass since these components may differentially reflect underlying factors related to
energy and protein balance. In contrast, for other ingredients, bone mineral content may be
more relevant.
Interpreting Inadequacies and Excesses in Growth Outcomes

The health implications of inadequate growth increments are not well described except
in the context of severe undernutrition, which is an event unlikely to occur in closely
monitored infant-feeding trials. There is evidence, primarily from populations with high
poverty levels, of an association between more severe length and weight deficits (stunting
and wasting) and impaired immune function (Forse et al., 1994), increased risk of morbidity
and mortality (Pelletier et al., 1993), and poor developmental outcomes (for review, see

Grantham-McGregor et al., 2000). Mild growth deficits tend not to be strongly related to
specific health outcomes independent of other nutritional risk factors. In comparisons of
breastfed and formula-fed infants, despite differences in growth patterns, Dewey and col-
leagues (1991) found no differences in behavior or activity levels of breastfed and formula-
fed infants.
While the focus in the past has been on nutritional inadequacies or growth deficits
associated with formula feeding, it is important to also assess the potential for a new
ingredient to cause excess growth. Aside from the risks associated with macrosomia in
newborns, there is a lack of information on the immediate consequences of excess weight or
of differences in body composition during infancy. While not related to feeding, infants with
macrosomia associated with maternal gestational diabetes are at increased risk of postnatal
obesity. This would seem to be an effect of the mother’s diabetes on body composition, with
macrosomic infants having a significantly higher percent body fat (Fee and Weil, 1960). Of
greater concern is the long-term consequences of excess infant growth, particularly in light
of the worldwide epidemic of child and adult obesity. There is inconsistent evidence that
fatness or excess weight gain in infancy predicts later obesity. When associations between
excess infant growth and later outcomes do exist, they could reflect genetic factors or
common behaviors, such as a consistent tendency of parents to overfeed.
The best evidence supporting an association between rapid infant weight gain and later
risk of overweight comes from a prospective cohort study of more than 19,000 participants
in the National Perinatal Collaborative Study. Researchers assessed the relationship of
weight gain in the first 4 months of life to overweight at age 7 years (defined as body mass
index [BMI] > 95th percentile of the CDC growth charts). After adjusting for birthweight,
gestational age, sex, race, firstborn status, maternal BMI, and maternal education, they
found that for each 100-g weight gain increase per month, the risk for overweight at age 7
increased by about 30 percent (Stettler et al., 2002b). Furthermore, nearly one-fifth of
overweight status at age 7 could be attributed to infancy weight gain above the highest
quintile. Stettler and colleagues (2002a) also found that weight gain in the first year of life
was strongly associated with overweight and obesity in the school years among children
living in the Seychelles. In a large British cohort born in the 1990s, more rapid weight gain
in the first 2 years of life, evidenced by an increase in a weight-for-age Z score of greater than
0.67, was associated with higher BMI, percent body fat, and total fat mass in later childhood
(Ong et al., 2000). In Pima Indian children, a population with a very high prevalence of
obesity and type 2 diabetes, Lindsay and colleagues (2002) found two periods characterized
by excess weight gain relative to the CDC growth reference. These were from 1 to 6 months
and 2 to 11 years of age. Although Dietz (1994) did not identify infancy as one of the three
critical periods in childhood for the development of obesity, the more recent findings sum-
marized above have led researchers to suggest that infancy represents another critical period
for the development of obesity later in life.
In contrast, there are a number of studies that find no evidence that overweight babies
are destined to become overweight adults, unless they have obese parents. For example,
Whitaker and colleagues (1997) found that risk of obesity in young adults was not increased
by obesity at age 1 to 2 years unless at least one parent was also obese. Infancy is character-
ized by a substantial capacity for compensatory growth following a period of failure to
achieve growth potential or a period of excess growth, thus limiting the long-term conse-
quences of relatively short periods of abnormal growth. Butte and colleagues (2000) com-
pared multiple dimensions of body composition among breastfed and formula-fed infants.
Despite significant differences in early infancy, they found no persistent difference by feeding
method beyond age 12 months.

There remains controversy over the extent to which deficits or excesses in overall
growth, growth of specific organs and tissues, or differences in fat versus lean tissue have
long-term effects on physiological functioning and disease risk. Again, the evidence of long-
term effects tends to focus on the extremes in child size. For example, stunting in childhood
is associated with short stature in adults, which is in turn associated with lower work
capacity among adults engaged in physically demanding jobs and increased risk of poor
obstetric outcomes in women. However a recent study of a cohort of Finnish children was
the first to show that infant obesity was significantly associated with later development of
type 1 diabetes (Hypponen et al., 2000). The hypothesized mechanism is hyperinsulinemia
and damage to beta cells associated with early excess body fat. Based on research among
Indian infants, Yajnik (2001) has hypothesized that deficits in skeletal muscle in infancy may
contribute to insulin insensitivity and risk of type 2 diabetes later in life.
There is increasing evidence that growth deficits in utero and in the early postnatal
period have important long-term health consequences owing to “programming” of structure
or metabolic functioning by nutritional inadequacies. The focus of most of the research has
been on fetal programming (Godfrey and Barker, 2001), but there is also evidence of effects
of postnatal growth deficits resulting in small size at 1 year of age (Vijayakumar et al.,
1995). Furthermore, there is evidence to suggest that feeding mode during infancy has long-
term effects on lipid profiles (Cowin and Emmett, 2000; Plancoulaine et al., 2000), risk of
later obesity (Armstrong and Reilly, 2002; for a review of effects on obesity, see Butte,
2001), and risk of other diseases (Leeson et al., 2001). However there is no particular
substance in milk to which these effects may be attributed.
Evaluating the 120-Day Growth Study

Although not currently a requirement, manufacturers often provide 120-day growth
studies to demonstrate healthy growth.2 The committee was specifically asked to evaluate
the adequacy of the currently used 120-day growth study. Conceptual issues on the measure-
ment of growth are discussed above, so the following section is confined to a discussion of
the duration of the study. The first 120 days of life is a period during which infant formula
is most likely to be the sole source of nutrition for the infant and a period of high growth
rates and, thus, a period of high susceptibility to dietary intake. From a practical perspective,
it may be difficult to recruit infants whose parents are willing to forego the introduction of
other foods until after 6 months of age, and there is no reason to think that an adverse effect
of an ingredient new to formulas would be detected only between 4 and 6 months of age.
However a study length of 120 days may be insufficient for several reasons. First, human
milk is recommended as the sole nutrient source for infants for at least the first 4 months
(AAP, 1997; IOM, 1991) and preferably for the first 6 months of life (ADA, 2001; IOM,
1991; WHO, 2002). When intended as a human-milk substitute, exclusive formula feeding
should be recommended for the same period of time. Ideally formula should be tested for the
entire period for which it is intended to be fed as the sole source of infant nutrition,
consistent with breastfeeding guidelines. This is particularly true since intake of infants
receiving only formula will be greater in the period from 4 to 6 months of age.
2FDA published a proposed rule that would change several aspects of the infant formula regulations. FDA is
proposing to revise “Quality Factors for Infant Formulas” (FDA, 1996). Among other requirements, FDA
is proposing to establish healthy growth as a quality factor. To ensure that the new infant formula supports
normal physical growth, the proposed rule would require that growth studies be performed for 120 days.

Second, serious limitations of the 120-day growth study are that it does not allow for the
determination of delayed effects or for understanding longer-term effects of early perturba-
tions in growth. Longer-term follow-up of participants in clinical studies should be recom-
mended, with the duration to cover at least the period when infant formulas remain a
substantial source of nutrients in the infant diet.
In summary, the committee recognizes that to establish levels of growth that indicate a
safety concern is a difficult endeavor. However the committee concludes that any systematic
and statistically significant difference in size or growth rate between infants fed a formula
containing a new ingredient versus human milk or a previously approved formula should be
a safety concern.
SPECIFIC ORGAN SYSTEMS
The Importance of Assessing Specific Organ Systems

As described in the previous section, the committee recommends that growth studies
should remain the centerpiece of clinical testing of ingredients new to infant formulas.
However growth deficits are likely to appear only secondary to effects on specific organs or
tissues and may not appear for some time after nutritional insult. The major organ systems
should also be studied when assessing the safety of an ingredient new to infant formulas (see
Figure 6-1, Box 3 and Figure 6-2).
The gastrointestinal tract is the first organ that encounters ingested ingredients. It serves
to protect the infant from environmental pathogens, antigens, toxins, and other noxious
agents (Walker, 2002). In healthy infants, gastrointestinal tract functions assure that normal
growth and development occur, provided the infant is offered the necessary nutrients. Any
of these functions may be affected by the diet or ingredients in the diet, and impaired
function can lead to inadequate nutrient availability.
The immune system is also important during infancy because of its regulatory effects of
a substance on immune competence and the potential for an inflammatory or allergic
response to a new ingredient. Food allergy and other adverse reactions to food are more
common in infants than in any other age group. This is partly a reflection of the relative
immaturity of the infant immune system. Compared with older children and adults, young
infants have low immunoglobulin A concentrations and thus reduced binding of antigens in
the gut. The infant immune system is not fully mature at birth, with deficits in the ability to
prevent invasion of pathogens and to respond to antigens. Of particular concern in the
context of ingredients new to infant formulas is the increased permeability of the gut
mucosal barrier in the presence of inflammation or infection or if the integrity of the
epithelial cell layer is disrupted. The increased permeability allows macromolecules to be
absorbed and stimulates allergic responses to food proteins. Furthermore, immature lysoso-
mal function in mucosal cells and limited intracellular proteolysis may result in further
intestinal damage and increased permeability.
Finally, the endocrine system is not fully developed until after puberty has occurred. It is
possible that an ingredient new to infant formulas could affect endocrine development or
expression of endocrine function. An example of such a possibility is the phytoestrogen
content of an infant formula having an effect on the development or expression of estrogen-
responsive tissues. Therefore clinical studies should include additional assessments to ensure
that infants:
• grow and develop according to standards,

• consistently display normal vital signs,

• feed and stool normally,
• do not vomit (aside from infantile reflux),
• demonstrate consistently normal laboratory values (blood count, blood chemistries,
liver function, renal function, and urine analysis),
• do not present immunologically related injuries, and
• do not present signs of endocrine disruption.
RECOMMENDATION: Assessment of clinical endpoints should include signs or ad-

verse laboratory indicators of the major organ systems, including the gastrointestinal
tract, kidneys, blood, and immunological and endocrinological systems.
Assessing the Gastrointestinal Tract

The gastrointestinal tract consists of the hollow organs (mouth, pharynx, esophagus,
stomach, small bowel, and colon) and the solid organs (liver and pancreas). The functions of
the hollow organs are as follows:
• Motility. The propulsion of lumenal contents through the gastrointestinal tract oc-
curs as the result of contractions of two layers of perpendicularly oriented smooth muscle.
Beginning in the esophagus, the movement of the walls of the hollow tube mixes and propels
lumenal contents. Nutrients and water are absorbed and waste is extruded during the
passage of substances through the tubular gastrointestinal tract.
• Digestion and absorption. Digestion begins in the mouth with salivary enzymes and
continues through the colon, where some digestion of carbohydrate can occur, especially in
infants. A relative fat malabsorption occurs in infants compared with adults (Fomon et al.,
1970). Similarly, pancreatic secretion of amylase and starch digestion is less in infants than
in adults. Absorption of nutrients occurs throughout the gastrointestinal tract, beginning in
the small bowel. Absorption can be passive (diffusion), active, or carrier mediated.
• Secretion. Secretion of substances, such as acid, pepsin, bile, and enzymes, is neces-
sary to digest nutrients. In addition to the digestive material, the gastrointestinal tract
secretes hormones and paracrine substances that modulate the function of other cells.
With the exception of pancreatic exocrine function and bile acid synthesis and compo-
sition, development of the gastrointestinal tract is essentially complete at birth for infants
born after a 34-week gestation (Antonowicz and Lebenthal, 1977; Auricchio et al., 1965;
Fredrikzon et al., 1978; Hadorn et al., 1968; Hamilton, 2000; Lindberg, 1966; Montgomery
et al., 1999; Norman et al., 1972; Watkins, 1985; Watkins et al., 1973).
Clinically relevant tools are available to assess each function of the gastrointestinal tract
and some tools are more specific than others. For example, normal growth and development
occurs only when the gastrointestinal tract is functioning optimally. But slowed or inad-
equate growth, as the common denominator of impaired gastrointestinal function, does not
identify the function that is impaired. Table 6-3 lists the functions of the gastrointestinal
tract and general (level 1 assessments) and specific (level 2 assessments) clinical outcome
measures that can be used to assess whether a specific function has been impaired. Table 6-
4 lists the advantages and disadvantages of each of the outcome measures.
Motility can be assessed by measurement of esophageal, antroduodenal, small bowel,
and rectal contractions (Scott, 2000); gastric emptying time (Di Lorenzo et al., 1987); and

TABLE 6-3 Gastrointestinal Tract Clinical Endpoints

Function Level 1 Assessments Level 2 Assessmentsa
Absorption Growth velocity Stool fat, stool protein, stool carbohydrate, stool
alpha-1-antitrypsin, stool pH, balance studies,
blood levels of specific nutrients
Allergic Vomiting, diarrhea, irritability, Serum IgE, quantitate peripheral eosinophils,
colitis skin tests, RAST, challenge tests
Barrier Not applicable Polyethylene glycol, lactulose/mannitol, 51Cr EDTA,
proteins, stool cultures, serum antigens
Biotransformation Serum liver enzyme levels Blood levels of bile acids or specific drugs and
their metabolites, serum liver enzyme levels,
isotope excretion scans
Digestion Growth velocity Stool: fat, protein, sugars, pH
Homeostasis Not applicable Balance studies, blood/tissue levels of specific
nutrients
Immunological Serum antibodies Response to oral vaccinations, stool cultures
Motility Absence of vomiting, stool Measurements of transit times (e.g., charcoal,
pattern, growth velocity hydrogen breath tests)
Secretory Growth velocity Measure specific hormones, stool chymotrypsin,
elastase, alpha-1-antitrypsin
Metabolism of
macronutrients
Protein Growth velocity, serum liver Urine and serum amino acid levels, serum glucose
enzyme levels, liver size by serum proteins (e.g., albumin, prealbumin,
examination clotting factors, alpha-1-antitrypsin, transferrin),
serum liver enzyme levels, liver size by
examination and ultrasound
Lipid Liver size by examination, Serum lipid levels (cholesterol, phospholipids,
serum liver enzyme levels triglycerides), serum glucose, liver size, and fat
content by examination and ultrasound, urine
organic acids
Carbohydrate Liver size by examination, Serum glucose levels, serum liver enzymes, liver
serum liver enzyme levels size by examination and ultrasound
NOTE: The petitioner (or manufacturer), in consultation with the expert panel, will determine which tests are
apH = potential of hydrogen, IgE = immunoglobulin E, RAST = radioallergosorbent test, EDTA = ethylene-
diaminetetraacetic acid.
transit time (Scott, 2000). The muscle fibers, nervous tissue, and some neurotransmitters can
be evaluated with biopsies. Generally, however, motility is functionally normal if there is no
vomiting, if the stool pattern is normal, and if growth velocity is normal. If measurements of
motility are needed they should be carried out by noninvasive techniques.
Digestion and absorption can be monitored by quantifying stool fat (Fomon et al., 1970;
van de Kamer et al., 1949); protein, such as alpha-1-antitrypsin (Dinari et al., 1984);
carbohydrate content (Grant et al., 1989); and breath tests (Fernandes et al., 1978; Maffei
et al., 1977; Perman et al., 1978; Robb and Davidson, 1981; Thomas et al., 1981). The
amount of specific nutrients can be quantified in blood. However if growth velocity remains
normal, it is unlikely that digestion is adversely affected by dietary intake.
Some secretory functions of the gastrointestinal tract can be assessed by quantifying levels
of specific hormones or enzymes (e.g., gastrin, cholecystokinin, trypsin, lipase, or motolin) in
the blood or stool. Growth velocity falters if the secretory functions are impaired.
For some nutrients the gastrointestinal tract regulates absorption based on nutrient

TABLE 6-4 Advantages and Disadvantages of Various Organ Clinical Endpoints

Outcome Measure Advantages Disadvantages
Balance studies Accurate assessment of specific Requires admission to a clinical
nutrients research center
Bleeding time Easy, safe, accurate Painful, requires a small incision,
does not identify specific clotting
abnormality
Blood pressure Noninvasive, used in routine health Requires personnel with some
assessment, standards available training
for children
Dual-energy X-ray Accurate for bone mineralization, Normative data not available for
absorptometry scan fat mass, and lean body mass infants, may require sedation,
(bone mineralization) requires expensive equipment
Fecal fat–72 hour Noninvasive Collection starts and ends with
marker, in-home collection may be
difficult, collection in a clinical
research center is more accurate
Growth velocity Established normal values, Does not identify specific function
noninvasive, inexpensive, readily that is deficient or impaired
available technology
Motility Specific, can identify area of the Invasive, limited to centers, time
gastrointestinal tract where an intensive in infants who cannot
abnormality occurs cooperate, but can measure some
aspects by noninvasive techniques
(e.g., hydrogen breath tests)
Permeability studies Easy to use probes Requires urine or blood collection
(polyethylene glycol,
sugars, antigens)
Serum levels of nutrients, Accurate measure of circulating Requires blood draw, some assays
hepatic enzymes, bile nutrient levels, liver function, available only in centers
acids, chemistries, hematological function
blood gas, blood
counts, specific proteins
Stool components Noninvasive, relatively easy to Depending on the test, varying
(fat, enzymes, collect in infancy specificity and sensitivity
protein, pH,
reducing substances,
cultures)
Urine analysis, Noninvasive, accurate Collection may be difficult in infants
quantitation of
nutrients, estimate
of glomular filtration
Ultrasound Noninvasive, gold standard to size Relatively expensive
abdominal organs and assess
cirrhosis, fatty infiltration in liver,
and inflammation in bowel
Vaccine response Accurate measure of B-cell function Requires injection of vaccine,
blood draw
levels. This regulation is often complex and involves other organs, such as the liver and
kidney for calcium homeostasis (IOM, 1997), and the liver, spleen, and bone marrow for
iron (IOM, 2001). Balance studies, stable isotopes, levels of specific nutrients in blood or
tissue, and storage forms of specific nutrients can be assessed.
The gastrointestinal tract is the site at which interaction with a food allergen occurs.

Different factors predispose for the development of food allergy, such as family history,
immune deficiency, or early exposure to antigens. Food allergy can consist of type I, III, or
IV reactions. Allergic reactions of this organ include enteropathy, colitis, and nonspecific
reactions, such as recurrent vomiting, bowel edema, obstruction, constipation, occult bleed-
ing, and colic. The manifestations of food allergy vary with age and site of food antigen
exposure. In infancy food allergy can be assessed by evaluating the infant for vomiting,
diarrhea, malabsorption, gastrointestinal loss of blood or protein, and constipation, and by
performing challenge tests.
It is unlikely that the human term-infant gastrointestinal tract is more permeable than
that of older infants and children (Sanderson and Walker, 1993). One study using human
α-lactalbumin as a marker of permeability showed that serum concentration of this protein
was increased in term breastfed infants for the first several months of life (Jakobsson et al.,
1986). However others, using bovine β-lactoglobulin in formula-fed infants, did not show a
change in gastrointestinal permeability over the first several months of life (Roberton et al.,
1982). In healthy term infants, gastrointestinal permeability may be increased by allergy
(Boehm et al., 1992; Dupont et al., 1989; Falth-Magnusson et al., 1986; Heyman et al.,
1988; Juvonen et al., 1990; Schrander et al., 1990), infection (Holm et al., 1992), and
perhaps colic (Lothe et al., 1990). Permeability can be assessed by using the inert carbohy-
drates (e.g., lactulose and mannitol), polyethelene glycol 4000, 51Cr ethylenediaminetetra-
acetate, and heterologous proteins (e.g., bovine β-lactoglobulin) or homologous proteins
(e.g., human α-lactalbumin) (Bjarnason et al., 1995; Sanderson and Walker, 1993).
The pancreas serves as a secretory- (exocrine) and hormone- (endocrine) producing
organ. Exocrine functions are difficult to assess in clinical studies in healthy infants but, as
noted above, can be assessed by directly quantifying lumenal concentrations of enzymes and
bicarbonate before and after a stimulus. It is only when pancreatic exocrine secretion is
dramatically decreased that a deceleration of growth velocity occurs (Huynh and Couper,
2000). Severe pancreatic insufficiency can be monitored by measuring fat or certain enzymes
(e.g., trypsin) in stools. Pancreatic endocrine dysfunction is most often manifested as diabe-
tes, which can be assessed by obtaining serum insulin concentrations, blood, and urine
glucose (Huynh and Couper, 2000).
The liver plays a central role in the metabolic adaptation of the fetus to extrauterine life
through glucogenolysis, gluconeogenesis, and the regulation of amino acid and fat metabo-
lism (Karpen and Suchy, 2001). These functions of the liver can be assessed by quantifying
urine and blood amino acid levels; urine organic acid levels; blood proteins, lipids, ammo-
nia, and bicarbonate; liver fat; and ultrasound. However if the liver is unable to function
normally with respect to carbohydrate, protein, or lipid metabolism, normal growth velocity
will not be maintained.
In addition, the liver synthesizes and excretes bile acids (Setchell and O’Connell, 2001).
Bile acid synthesis, the bile acid pool, and intralumenal bile acid concentrations gradually
increase during the first year of life. Bile acid secretion is maximal at birth and cannot be
further stimulated. This function of the liver can be assessed by quantifying blood levels of
liver-derived enzymes as a marker of hepatocyte integrity, by quantifying blood levels of bile
acids and isotope excretion scans as a marker of hepatic excretory function, and by measur-
ing serum levels of specific drugs and their metabolites (Batres and Maller, 2001).
Assessing the Kidneys

The kidneys perform vital functions, including filtration of plasma, reabsorption of
water and electrolytes, excretion of wastes, and the production of hormones that control

TABLE 6-5 Kidney Clinical Endpoints

Function Level 1 Assessments Level 2 Assessments
Filtration Growth velocity, urine analysis, serum Glomular filtration rate
creatinine
Reabsorption Growth velocity, serum acid base, serum Plasma acid base, serum electrolytes,
electrolytes, urine analysis urine analysis
Endocrine Blood pressure, urine analysis, bone Vitamin D, serum calcium, phosphorus,
mineralization assessment, serum magnesium, erythropoetin, prosta-
calcium, phosphorus glandins, renin, angiotensin, kallikrein,
kinin, bone mineralization assessment
blood pressure, calcium homeostasis, and red cell production (Binley et al., 2002; Gleim,
2000; McMurray and Hackney, 2000). Specific tests (level 2 assessments) can be performed
to identify each of these functions, but general assessments (level 1 assessments) of blood
pressure, urinary analysis, growth velocity, serum creatinine, blood urea nitrogen, calcium,
bicarbonate, and a complete blood count will establish if renal function is abnormal or
adversely affected by a component of the diet (Table 6-5).
Specific functions of the kidney that can be assessed include glomerular filtration rate
(GFR), which can be measured by quantifying the clearance of a substance that is freely
filtered across the capillary wall and is neither reabsorbed nor secreted by the tubules. The
optimal measurement of GFR is insulin clearance (Arant et al., 1972). Clinically, however,
GFR can be estimated by the clearance of endogenous creatinine. At serum levels of creati-
nine exceeding 2.0 mg/dL, changes in renal function can be monitored by the serum creati-
nine concentration. GFR is adequate for healthy term infants, but it does not approximate
adult rates until about 3 years of age. Renal tubular reabsorption and urine acidification is
less at birth and for several months thereafter than it is for adults. This function is adequate
for healthy infants, but contributes to fluid and electrolyte abnormalities in infants who are
ill or are fed an inappropriate diet (Goldsmith and Novello, 1992).
The kidney also serves as an endocrine organ, synthesizing and degrading prostaglan-
dins, kallikrein-kinin, and renin-angiotensin, which control blood pressure. Hydroxylation
of vitamin D creates the hormone that controls calcium homeostasis, which occurs in the
kidney. Erythropoetin, the glycoprotein that regulates both steady-state and accelerated red
blood cell production, is governed by oxygen availability to the kidney. Blood levels of these
hormones and the substances they regulate can be quantified.
Assessing the Blood

The hematological system consists of red blood cells, white blood cells, platelets, and
proteins. The function of the red blood cell is to transport oxygen to tissues. This function is
performed by hemoglobin, which combines reversibly with oxygen, allowing the red blood
cells to transport oxygen from the lungs and deliver it to tissues. Hemoglobin accounts for
more than 95 percent of the total protein and about 90 percent of the dry weight of the red
blood cell. Red blood cell function can be assessed by quantifying the number of cells, the
hemoglobin concentration, and the hematocrit. Membrane lipid analysis, fragility studies,
observation of the blood smear, and a reticulocyte count can also be performed (Brugnara
and Platt, 1998).

The capacity of white blood cells to produce antibodies to antigens is intact at birth. In
general, white blood cell function can be assessed by quantifying the total white cell count,
the absolute count of specific cells, skin tests, and immunoglobulin levels. The specific func-
tion of phagocytic cells, such as chemotaxis, ingestion, and oxidative metabolism, can be
assessed in isolated cells. Some products of these functions, such as myeloperoxidase, can be
quantified in blood or stool. Specific lymphocyte function can also be assessed in isolated
cells and by quantifying inflammatory mediators in blood. Abnormalities in white blood cell
function can be suspected clinically by occurrence of frequent infections or infections caused
by low-virulence pathogens.
Platelets are important in homeostasis. Platelet activity is assessed by quantifying the
number and morphology of platelets in a blood sample and by assessing platelet aggregation
and specific platelet functions. Platelet function can also be assessed by performing a bleed-
ing time (Handin, 1998).
Several clotting factor concentrations in blood are lower during the neonatal period
than in adulthood (Esmon, 1998). The lower level of clotting factors is associated with
prolonged prothrombin and partial thromblastin time. After the neonatal period, coagula-
tion is the same as that of adults. Coagulation function can be assessed by quantifying each
of the following factors: I through XII, plasminogen, antithrombin III, prekallikrein, and
high molecular weight kininogen. Coagulation can also be assessed by performing a throm-
bin time and a partial thromboplastin time or by noting if abnormal bleeding is present (see
Table 6-6).
Assessing Immunological and Allergic Activity

A newborn’s digestive system is fairly mature, but may only incompletely break down
food proteins. Thus infants are especially susceptible to allergic sensitization. The immune
response of newborn infants is predominantly associated with the Th2-type of the helper
T-cell population, possibly because of in utero priming of fetal T cells by transplacental
passage of common environmental allergens and dietary antigens. In general normal infants
exhibit a low-grade immunological response to subsequent exposure to such environmental
TABLE 6-6 Blood Clinical Endpoints

Function Level 1 Assessments Level 2 Assessments
Oxygenation Complete blood count, clinical Red blood cell number, hemoglobin/hematocrit,
color (pallor, ruddiness, membrane lipids, fragility, reticulocyte count,
cyanosis) red cell fragility
Immune defense Frequency of natural infections White blood cell number, differential count,
phagocytosis chemotaxis, ingestion, oxidative metabolism,
absolute polymorphonuclear count
Immune defense Frequency of natural infections White blood cell number, differential count,
lymphocytes delayed sensitivity, skin tests, immunoglobulin
levels, absolute lymphocyte count
Bleeding Bleeding time Platelet numbers, in vitro bleeding, platelet
aggregation, clotting factors, thromboplastin
time, partial thromboplastin time

agents after birth, which is limited to immunoglobulin (Ig) G and IgM isotypes and to the
Th1-type of the helper T-cell population (Holt et al., 1999). During further development the
neonatal immune system continues to shift towards Th1-type response. It has been proposed
that alterations in the neonatal mucosal environment (e.g., a change in microflora), the use
of formulas (and lack of breastfeeding), antibiotics, mucosal infections, and a highly hy-
gienic environment in early infancy may lead to further increase in the Th2-type of helper
T cells that were primed in utero (Holt et al., 1999). Th2-type helper T-cell expression is
currently considered the hallmark of allergic immunopathology (Kay, 2001).
Many of the proteins added to infant formulas are functional proteins (i.e., proteins that
are added for their function—not as a source of amino acids) and to maintain their function,
they must be resistant to digestion, a property shared with allergens. The addition to infant
formulas of novel proteins (including glycoproteins or lipoproteins), which by their nature
can induce allergic or other adverse reactions, requires clinical testing. Human-milk proteins
are not expected to be allergenic in humans since they are produced by the human mammary
gland. Possible exceptions are the proteins that originate from maternal dietary components,
such as cow-milk proteins. Also, as has been recognized in the production of biotech crops
(Kok and Kuiper, 2003), the commercial production of milk proteins using recombinant
technologies may produce unintended and unexpected side effects. For instance, one milk
protein produced in recombinant microorganisms may differ from native proteins in level of
glycosylation, posttranslational modifications, or minor amino acid sequences, which may
change the allergenicity potential. Furthermore, there is potential for contamination with
compounds deriving from the genetically modified organism used as the protein source.
The central aspect of clinical testing in infants should include the evaluation of a diverse
spectrum of immune functions in response to an added substance. To develop the appropri-
ate tests for assessing the safety of the immunological responses to new substances, it is
useful to first identify the target tissues affected by the interaction of ingested substances
with the host immune system (Table 6-7).
TABLE 6-7 Target Tissues and Signs Derived from Interactions of a New Ingredient
with a Host Immune System Target Tissue
Immunoglobulin Mediated by Other
Target Tissue E-Mediated Immunological Mechanisms
Gastrointestinal Infantile colic Food-induced enterocolitis and proctocolitis
Eosinophilic gastroenteritis Allergic gastroenteritis, eosinophilic
(postprandial nausea, weight loss)
Oral allergy (angioedema) Celiac-like disease
Gastrointestinal tract anaphylaxis
(nausea, chronic diarrhea)
Celiac disease
Airway Rhinitis-conjunctivitis Heiner syndrome (pulmonary hemosiderosis)
Laryngeal edema-obstruction
Acute bronchospasm
Skin, joint, Urticaria Dermatitis herpetiforms
blood vessels Atopic dermatitis Contact sensitivity
Contact irritation (acidic fruits and vegetables)
Migraine
Arthritis
SOURCE: Sampson (1996, 2002).

When measuring immunocompetence, clinical assessment of allergic response should in-

clude evaluation of specific signs, laboratory testing for evidence of specific immune responses,
inflammatory cytokines, and allergen-specific response to challenges. IgE isotype is present at
birth and, therefore, specific signs of IgE-mediated allergic reactions, such as urticaria, vomit-
ing, diarrhea, respiratory signs, and anaphylaxis, can occur in the neonate. Clinical signs of
atopic dermatitis, including erythema, edema, crusts, excoriations, lichenification, dryness,
degree of itch, and loss of sleep, should be evaluated. The latter are evaluated using a scoring
system for the extent and intensity of dermatitis (European Task Force on Atopic Dermatitis,
1993). Gastrointestinal responses, particularly the chronic type, are generally presumed to be
T-cell mediated, but may also be associated with specific IgE-mediated immunological interac-
tions. These include enterocolitis, proctocolitis, enteropathy, and a subset of allergic eosino-
philic esophagitis/enteropathy (e.g., vomiting and diarrhea) with eosinophil infiltration of the
affected portion of the gastrointestinal tract (Sicherer et al., 2001).
Laboratory testing for evidence of specific immune responses should include the determi-
nation of serum and mucosal antibody profile, including IgE, and measurement of T-cell-
mediated immune responses and of specific proinflammatory and immunoregulatory cytokines
synthesized after exposure to the new ingredient. Specific IgE responses to allergens can be
assessed by measuring serum IgE concentrations by the radioallergosorbent test for binding of
IgE. These tests can be used as level 1 and level 2 assessments, as indicated in Table 6-8, based
on the criteria specified by the committee. Specific level 2 assessments should be performed
when there is evidence of adverse effects from the more general level 1 assessments.
However because allergic sensitization is a rare event, level 1 assessments in unselected
infants may not have the power to detect such responses. The evaluation of subpopulations
of infants selected for pre-existing allergies will also not be helpful because they will not be
sensitized to the novel protein under consideration. Thus the potential allergenicity of such
ingredients must be carefully evaluated in the preclinical studies described in Chapter 5.
TABLE 6-8 Available and Potential Tools for Assessment of Immnological and Allergic
Outcomes
Current
Assessment Use in Potential Value for Safety Endpoint Testing
Tool Level Assessment Advantage Disadvantage
Potential for 1 Yes Screening to demonstrate Nonspecific, does not
antigenicity and the ability to induce an predict potential for
immunogenicity immune response disease
Serologic evidence
of prior exposure
to ingredient
Serum antibody 1 Yes Provides specific evidence Does not predict
of prior exposure potential for disease;
to specific ingredient; requires peripheral
can differentiate blood samples
between recent
and past exposure
Cellular immunity 2 Yes Sensitive marker for Difficult to perform
immunoregulatory vs routinely; requires
proinflammatory cellular or tissue
immune response sample; not
standardized
Continued

TABLE 6-8 continued

Current
Assessment Use in Potential Value for Safety Endpoint Testing
Tool Level Assessment Advantage Disadvantage
Homocytotropic 1 Yes Strong correlation with Variable sensitivity

response disease and anaphylaxis, and specificity
(including helpful tool for follow
immunoglobulin E) up of clinical course
Mucosal immune
responses
Salivary immuno- 2 Yes Marker for mucosal Not available routinely
globulin A, immunity
immunoglobulin E
Cellular response 2 No Can identify with more Not available routinely;
specificity Th1 vs Th2 may not predict
helper T-cell responses potential for disease
Pre- and post- 2 No Marker for immunologic Not available for
exposure cytokine down regulation routine use
and chemokine
profiles
Induction of tolerance 2 No May be useful in treatment May not predict
of allergic disorders potential for disease
Change in mucosal
environment
Microflora 2 No Simple and easy approach May not predict
to identify potential potential for or
effect of ingredient on existence of
microflora; useful for active disease
follow up for treatment
modalities
Inflammation 2 No Provides histological Requires mucosal
evidence of disease tissues; spectrum of
inflammation may
vary considerably
within different
ingredients
Clinical evaluation
Skin-prick test 1 Yes Sensitive marker of May not predict
primary exposure; potential for disease;
easy to perform; does false positive results
not require blood with cross-reactive
samples responses
Food elimination 1 Yes Highly effective for Effective elimination
specific diagnosis or often difficult for
treatment of disease ubiquitous ingredients;
prolonged time lag
before clinical effect
and elimination
Food challenge 2 Yes High specificity for Risk of increased
diagnosis of disease; possible
ingredient- (but rare)
related disease anaphylaxis
on reintroduction
of the ingredient

Sensitized cells release specific cytokine and chemokine mediators that can be quanti-
fied. These include substances such as histamine and tryptase, as well as markers of inflam-
mation in the gastrointestinal tract. The latter may be of value in the evaluation of allergic
response since inflammation is a risk factor for increased sensitization. Markers of intestinal
inflammation include eosinophil cationic protein in serum and feces, α-1 antitrypsin, and
tumor necrosis factor alpha (Majamaa et al., 1996). Skin prick and patch tests have good
negative predictive value, but poor positive predictive value, and therefore are of more
limited use in clinical testing.
As with other clinical studies, the most definitive clinical assessments are accomplished
by double-blind, controlled trials. However investigators should consider that oral provoca-
tion of sensitive subjects could result in severe reactions and therefore study conditions
should be carefully designed and controlled.
Assessing Endocrinological Activity

The endocrine system consists of multiple organs that secrete a wide variety of hormones
that are responsible for maintaining the proper biochemical milieu of the body. Hormones are
biochemicals that are secreted by the various glands (e.g., pituitary, thyroid, parathyroid,
pancreas, adrenal, testes, and ovary) and act at other sites within the body. These hormones
usually act by signaling biochemical reactions at cell membranes or intracellularly. Changes in
endocrine function may either be intrinsic (e.g., a decrease in thyroid function because iodine
is missing from the diet) or extrinsic because of chemicals (including nutrients) in the formula
that may act as endocrine effectors or disruptors. Changes in hormone function may first be
evident in growth changes. Most changes may take months or years to become evident; for
example, sex hormone disruption may not be obvious until puberty.
Breastfeeding does not eliminate the concern for infant exposure to hormones. Oral
contraceptive hormones are excreted in milk, and cohorts have been followed long term to
ensure that the concentrations seen do not change the onset or course of puberty. Any
substance added to an infant formula that may change secretion or function of growth or sex
hormones may require follow-up through adolescence.
Exposure to endocrine disruptors in the environment may be brief in relation to the life
span. Changes may thus be transient or not measurable. Some parameters that may show an
immediate effect upon disruption of the endocrine function are thyroid-stimulating hormone
(TSH), triiodothyronine (T3), thyroxine blood glucose (T4), blood calcium, phosphorus, and
urine-specific gravity. Measurement of possible long-term effects include ovarian or testicu-
lar function, obesity, and pituitary function. Most endocrine measurements can be con-
ducted by collecting either the blood or the urine. It is important to consider the age, sex,
dietary status, body size, and medications during the interpretation of the numerical value of
any clinical test for the endocrine system, especially during the gestational age. Level 1
screening should contain at least one test from all of the major substances, organs, and
outcomes of the endocrine system. They include growth (e.g., insulin-like growth factor-1),
the thyroid (TSH, T3, T4), the adrenals (cortisol, adrenocorticotropic hormone), the para-
thyroid (calcium, parathyroid hormone), antidiuretic hormone (urine osmolality), and glu-
cose/insulin (Sperling, 1996). If there are any adverse events detected, this leads to a level 2
assessment, which includes immunoassays, binding proteins, and imaging techniques (e.g.,
body imaging, radionuclide imaging) (Sperling, 1996). A summary of the clinical endpoints
for the endocrine system is provided in Table 6-9.

TABLE 6-9 Endocrine Clinical Endpoints

Function Level 1 a Level 2b
Growth IGF-1 IGFBP-3
Thyroid TSH immunoassays, T3, T 4 Binding proteins, radionuclide imaging
Adrenal ACTH, cortisol CRH, plasma rennin, plasma
aldosterone, body imaging
Parathyroid Calcium, PTH Phosphorus
Antidiuretic hormone Urine osmolality Urine sodium excretion
Glucose/insulin Plasma insulin, glucose concentration
aIGF-1 = insulin growth factor-1, TSH = thyroid-stimulating hormone, T3 = triiodothyronine, T4 = thyroxine,
ACTH = adrenocorticotropic hormone, PTH = parathyroid hormone.
bIGFBP-3 = insulin growth factor-binding protein-3, CRH = corticotrophin-releasing hormone.
SOURCE: Sperling (1996).
DEVELOPMENTAL-BEHAVIORAL OUTCOMES
“. . . subtle behavioral effects can appear well in advance of
clear neurological dysfunction.”
The Importance of Assessing Developmental-Behavioral Outcomes

There are a number of reasons why it is essential to include developmental-behavioral
outcomes in future studies of the safety of ingredients new to infant formulas (see Figure 6-
1, Box 6 and Figure 6-3). First, behavioral outcome measures are sensitive to exposure to
toxic substances, particularly at low exposure levels. There are potential consequences to the
infant of deviations from normal developmental pathways. Therefore when evaluating the
addition of ingredients new to infant formulas, avoidance of type II errors (failure to detect
a real effect) may be more critical then avoidance of type I errors (accepting a spurious
finding as significant). In order to minimize the likelihood of failing to detect developmen-
tally meaningful consequences associated with the addition of ingredients new to infant
formulas, investigators must go beyond traditional toxicological and morphological assess-
ments. As Weiss (1995) has argued, focusing only on easily observed physical malformations
may seriously underestimate the actual impact of toxic exposure at either an individual or a
population level. Supporting this position, evidence from a number of studies indicates that
following exposure to toxins, subtle behavioral effects can appear well in advance of clear
neurological dysfunction (Evangelista de Duffard and Duffard, 1996; Sobotka et al., 1996).
Inclusion of developmental-behavioral deficits may be particularly critical when investi-
gating low-level exposure to toxins (Gaylor et al., 1998). It has been hypothesized that when
multiple outcomes can be affected by exposure to toxic substances, behavioral outcomes
may be among the most sensitive to toxic effects. When expressed in terms of level of
exposure required to produce an effect, going from highest to lowest required level, the
order of outcome sensitivity would be: mortality > malformations > physical growth >
behavior (Vorhees, 1986). This hypothesized hierarchical ordering has been documented
with regard to exposure in utero to vitamin A, salicylates, mercury, and alcohol (Adams,
1993; Jacobson and Jacobson, 2000). While the structural or functional domains affected by
exposure to toxins will vary depending on the developmental course of different organ and
functional systems (Shaheen, 1984), in the first year of life there is rapid development of
these systems. What this means is that during the time period of maximum human exposure

to infant formulas, subtle, but important, developmental consequences may not be detected
in nonhuman-based preclinical studies that primarily focus on toxicity or morphological
changes and that may not include potentially more sensitive developmental-behavioral end-
points that are comparable with those assessed at the human level.
Second, developmental-behavioral measures can have long-term predictive value. Little
direct evidence is available comparing the relative long-term consequences of outcomes such
as physical malformations versus behavioral deviations. However it has been hypothesized
that the long-term consequences of alterations in some components of behavioral develop-
ment may be more critical than physical consequences in terms of affecting the individual’s
ability to adapt to environmental demands (Russell, 1992). For example, both facial changes
and cognitive deficits are associated with fetal alcohol syndrome. The most likely candidate
to influence the individual’s ability to succeed in school would be the cognitive consequences
of fetal alcohol syndrome (Vorhees, 1986).
A third reason to include developmental-behavioral outcomes in studies of the safety of
ingredients new to infant formulas is that bidirectional brain-behavior links exist (e.g., brain
development mediates changes in behavioral competence, but the child’s interactions with
his or her environment also can influence brain development). Although this report discusses
neural and behavioral development separately from each other, these two areas of develop-
ment are closely interlinked. Obviously changes in central nervous system (CNS) structure
and function act as critical mediators for infant developmental-behavioral changes (Johnson,
2001; Lozoff et al., 1998; Nelson, 1994, 1995). However the converse is also true. There is
increasing evidence showing that brain development can reflect changes in the child’s envi-
ronment (Greenough and Black, 1992; Nelson and Bloom, 1997; Schore, 1994). For ex-
ample, systematic differences in infant brain electrical activity have been related to differ-
ences in rearing styles between depressed and nondepressed mothers (Dawson and Ashman,
2000). Environmental changes can be driven by changes in the child’s behavioral patterns, as
seen in studies showing systematic changes in maternal independence- and dependence-
fostering behaviors as their infant shows increased levels of functional competence (Kinder-
mann, 1993). This means that exposure to toxic substances that initially impact upon brain
development will result in synergistic bidirectional influences upon brain and behavior.
Criteria for Including Developmental-Behavioral Outcome Measures
Outcome Domains
When developmental-behavioral outcomes have been used, much of the research in
behavioral teratology and behavioral toxicology was designed to investigate whether there
were cognitive deficits associated with exposure to potentially toxic substances, such as lead
(e.g., Bellinger, 1995; Cory-Slechta, 1990). However it is important to recognize that devel-
opment in the early years of life is characterized by changes across multiple domains (Masten
and Coatsworth, 1998; Wachs, 1999). Because infant development is multifaceted, it is
important to go beyond a relatively narrow focus on cognitive outcomes in order to fully
evaluate the potential adverse effects of the addition of ingredients new to infant formulas
(Struthers and Hansen, 1992; Vorhees, 1986). Normality of development in one develop-
mental domain does not necessarily mean that there will be normality across all domains
(Lester et al., 1995). At a minimum, assessment of the potential developmental-behavioral
consequences of the addition of ingredients new to infant formulas should encompass out-
comes in the domains of: (1) sensory and motor development, (2) cognitive development,
(3) affect and temperament, and (4) neural integrity (see Table 6-10).

RECOMMENDATION: Assessment of clinical endpoints should include measurement

of infant sensory-motor, cognitive, affectual, and neural function with instruments that
follow recommended criteria.
Level of Assessment Required

Ideally the level and type of assessment utilized to evaluate the safety of an ingredient
new to infant formulas would be driven by known links between the class of the added
ingredient and the specific aspects of brain and behavioral development. Within this frame-
work, knowledge of which and to what degree specific neural structures, processes, or
behavioral functions were influenced by a given ingredient would tell the researcher which
brain areas or processes and which behavioral functions should be tested to determine
possible adverse consequences associated with the new ingredient. There is a solid body of
knowledge on nutrient-brain-behavior links for a few substances, such as iron, and a grow-
ing body of knowledge on such links for polyunsaturated fatty acids (Rao and Georgieff,
2000). Unfortunately, for the most part, the knowledge is insufficient to base the choice of
neural-behavioral measures on the empirical evidence alone. For the majority of new ingre-
dients added to infant formulas, choice of the level of assessment and the domains assessed
will require balancing the empirical evidence on substance-brain-behavior links with inte-
gration of results from the preclinical studies on a given substance, with knowledge of the
long-term relevance of different outcome measures. To this end, the committee proposes a
hierarchical set of criteria to determine the level of assessment that is needed in future studies
of the potential developmental-behavioral-neural consequences of new ingredients added to
infant formulas (see Figure 6-3).
Level 1 assessments are based on using behavioral and neural screening measures that
can be easily administered using parental report or during a routine well-baby physical
exam, which can be carried out as part of a clinical trial using a standardized examination
protocol (Table 6-10). Because these measures are lower in sensitivity than the measures
sufficient for use in the other levels described here, what appears to reflect normal develop-
ment on these measures does not necessarily mean that an ingredient added to infant
formulas is safe. Because of the lower sensitivity of level 1 assessment measures, more
stringent criteria must be met to justify use of such measures. Hence level 1 assessments are
recommended only if all of the following decision criteria occur:
• There is no existing evidence indicating a direct link between the new ingredient,
metabolites, secondary effectors, or source and impairments in either neural or behavioral
functions in infancy. In the absence of empirical evidence, there is no accepted theory that
would suggest a plausible link between the new ingredient or new ingredient source and
either neural or behavioral functions in infancy.
• There is no existing evidence for significant individual differences in susceptibility to
the ingredient, metabolites, secondary effectors, or source (e.g., there is little evidence sup-
porting a link between intake of food coloring or sugar and attention deficit disorder in
children, but there is evidence that a small proportion of children with attention deficit
disorder may have adverse reactions to food coloring or sugar that can accentuate their
hyperactive-inattentive behavior [Christensen, 1996]).
• There are neither empirical nor theoretical links between the ingredient, metabolites,
secondary effectors, or source and the functioning of other organ systems that might indirectly
influence brain or behavioral development (e.g., given existing evidence on relations between
nutrition and brain development [Rao and Georgieff, 2000], a new ingredient, metabolite,

secondary effector, or source that had an adverse influence on the digestive system could well
result in subtle nutritional deficits that could in turn influence brain development).
• There is no evidence of adverse effects in preclinical studies, including adverse effects
with potentially plausible alternative explanations (e.g., the effects are viewed as the result of
random chance or the reviewers believe that there may be methodological or statistical
problems in the studies). This means that even if potentially plausible alternative explana-
tions are offered to explain adverse findings, level 1 assessments would not be warranted
since adverse effects were found.
If all of the above criteria are met, then level 1 assessment instruments can be used in
clinical studies of the safety of new ingredients added to infant formulas. Recommended
screening instruments appropriate for level 1 neural and behavioral assessments in the first
year of life are shown in Table 6-10. The screening instruments are low cost and, therefore,
the committee recommends studying behavior and development from each of the domains
listed in the table (visual development, audition, motor development, cognition, tempera-
ment, and neural integrity).
Level 2 and 3 assessments involve detailed measures of child function in major develop-
mental areas. For level 2 assessments, one instrument is used for each area of function, and
only a single assessment occasion is required. Three of the criteria discussed above are also
involved in decisions as to whether level 2 assessments should be required. They are required
when there is not a plausible direct empirical or theoretical link between the new ingredient
or new ingredient source and impairments in either neural or behavioral functions in in-
fancy, but one or more of the following criteria occur:
• Based on either existing evidence or accepted theory, there are plausible links be-
tween the ingredient and other organ systems that might indirectly influence neural or
behavioral development (e.g., recombinant proteins that had an adverse influence on the
digestive system or pre- or probiotics that changed the nature of intestinal flora/fauna, which
could also influence digestive processes; in either case, one result could be subtle nutritional
deficits that in turn could act to influence brain development).
• There is evidence of adverse effects in preclinical studies for organ systems that could
influence neural or behavioral development. This means that even if potentially plausible
alternative explanations are offered to explain adverse findings, level 2 assessments would
be required since adverse effects did occur. For example, given current knowledge linking
iron deficiency in infancy to adverse cognitive and neural development (Rao and Georgieff,
2000), recent evidence showing lower-than-predicted iron retention by young infants (Fomon
et al., 2000) would be a preclinical adverse effect that would require use of at least level 2
assessment procedures in clinical trials involving new ingredients that either included iron or
could influence iron absorption.
• There is existing evidence for significant individual differences in susceptibility to the
ingredient.
Level 3 assessments also require detailed measures of child function in major develop-
mental areas. In addition, if more than one recommended instrument is available for a given
function, then a second recommended instrument that assesses either converging or comple-
mentary functions should be used on each assessment occasion. Assessment on at least two
separate occasions is required within the first year. Level 3 assessments are required when
the criteria for level 2 assessments are met and/or the following additional criterion occurs:

128
TABLE 6-10 Screening Measures Used in Level 1 Assessment of Infants Exposed to Formulas Containing New Ingredients
Nature of Deviation
Function Measure Assessment Source from Expected Development
Visual Visual attention to stimuli (Lester and Tronick, 2001) Standardized medical examination Infant not tracking or responding
development carried out as part of a to visual stimuli (e.g., direction
clinical trial if gaze not follow moving
visual stimuli)
Audition 1. Observation of infant orientation to sounds Parent report or standardized Infant not responding to auditory
(Cobo-Lewis and Eilers, 2001) medical examination carried stimulation or showing
2. Given documented links between hearing acuity and out as part of a clinical trial language development that is
early language development (Cobo-Lewis and Eilers, below what would be expected
2001), an alternative measurement would be to for a infant at a given
assess the infant’s acquisition of language landmarks chronological age
in the first year, such as appearance of syllabic
contours at around 3 mo, onset of babbling between
6–9 mo, and recognizing familiar words between
7–8 mo (Bloom, 1998; Posner, 2001)
Motor Age at which infant achieves motor landmarks like Standardized medical examination Degree of deviation from expected
development postural control, eye-hand coordination, sitting, carried out as part of a clinical development can be assessed
crawling trial using standard norms such as
those found in the Revised
Bayley motor scales (1993)

Cognition Norm-referenced parent report measures of develop- Parent report or standardized Level of developmental
mental progress on adaptive behavior scales, such as medical examination carried competence two or more
the survey form of the Vineland Adaptive Behavior out as part of a clinical trial standard deviations below
Scales (Sparrow et al., 1984) the norm
Temperament Short but validated parent report measure of infant Parent report or standardized Infant rated as consistently
temperament, such as the 24-item Infant Character- medical examination carried expressing a high intense
istics Questionnaire (Bates and Bayles, 1984) out as part of a clinical trial negative mood
Neural Neural screening exam assessing dimensions, such as Standardized medical examination Clinicians judgment of infant’s
integrity cranial nerve function, sensory reactivity, age carried out as part of a clinical level of neural integrity
appropriate reflexes, and autonomic nervous system trial
function, which can be administered by a pediatric
nurse (Slota, 1983) or a more detailed clinical neural
assessment administered by a physician (Herskowitz
and Rosman, 1985)
Standardized brief neurological assessment instrument Standardized medical examination Infant falls below global neural
with cut-off points for normal versus abnormal carried out as part of a clinical integrity cut-off score or cut-
neural functioning, for example, the Infant Neuro- trial or as part of an overall off score for specific neural
logical International Battery (INIB) (Ellison et al., screening battery in a research function
1985; Ellison, 1994) or the Neurologic Evaluation of assessment during the first 6
the Newborn and the Infant (NENI) (Harris and wk after birth (NNNS) or
Brady, 1986); or summary scores, for example, NICU across the first year (INIB,
Network Neurobehavioral Scale: (NNNS) (Lester and NENI)
Tronick, 2001)
NOTE: The petitioner (or manufacturer), in consultation with the expert panel, will determine which tests are required based on a thorough analysis of the potential
effects of the new ingredient.

129
• Based on either existing evidence or accepted theory, there is a plausible direct link
between the new ingredient or new ingredient source and impairments in either neural or
behavioral functions in infancy.
RECOMMENDATION: A hierarchy of three levels of clinical assessment should be

implemented:
• Level 1 assessments. Developmental measures that are easily administered but
low in sensitivity.
tal areas (single assessment for each area with one instrument).
tal areas (repeated assessments with multiple instruments).
Identifying Appropriate Measurement Instruments

There are a large number of potential instruments available to assess outcomes in the
multiple domains in which development occurs during the first year of life. Therefore it is
important to specify criteria that would allow the identification of measurement instruments
that are more appropriate for use in level 2 and level 3 assessments of the potential develop-
mental-behavioral consequences of the addition of new ingredients to infant formulas. Based
on a review of the literature, the committee has identified seven criteria to rank the utility of
instruments used to assess infant sensory-motor, cognitive, affectual, and neural function in
level 2 and level 3 studies. The criteria are:
• Age appropriateness. Because infant intake of formulas is at a maximum during the

first year of life, the committee recommends concurrent measures that can be administered
during this time period.
• Potential long-term consequences. It is essential that assessment measures are not
only reliable, but also that they either have predictive value for functional competence
beyond the first year of life or are precursors for critical mediators of later functional
competence. An example of the latter type of measure would involve the assessment of
infant behaviors that interfere with the establishment of adequate caregiver-infant relations
(Chasnoff et al., 1987).
• Sensitivity. An instrument may be appropriate for the first year of life and may have
long-term predictive value, but still be relatively insensitive to exposure to toxic substances.
Given the consequences to the developing infant, avoidance of type II errors (failure to detect
a real effect) may be more critical then avoidance of type I errors when evaluating addition
of new ingredients to infant formulas. To maximize the ability to detect adverse conse-
quences, measures should demonstrate sensitivity to exposure to toxic substances during
infancy. In recommending a sensitivity criterion, the committee fully recognizes that new
ingredients for infant formulas will have been extensively evaluated during the preclinical
trials, making it highly unlikely that substances that are clearly toxic will be tested in clinical
trials. Thus the committee does not expect the assessment instruments described below to be
used in clinical trials involving known toxic substances. However if an assessment instru-
ment has not been shown to be sensitive to known toxins in previous research studies, it is
unlikely that it will be sensitive to potentially subtle adverse effects of those new ingredients
that are tested in clinical trials. As a result, the committee requires, where possible, that the

instruments used must have previously demonstrated the ability to detect adverse conse-
quences associated with children’s exposure to known toxins.
• Brain-behavior links. To maximize the ability to understand the mechanisms that
underlie deleterious effects associated with the addition of ingredients new to infant formu-
las, outcome measures at the human level should have documented links to CNS structural
and functional development. Knowledge of the brain-behavior link allows investigators to
select which behavioral outcomes are most likely to be affected when toxins are known to
impact on a given set of neural functions and which neural functions to select when toxins
are known to impact on a specific set of behaviors (Jacobson and Jacobson, 2000).
• Cross-species generalizability. To maximize generalizability from preclinical nonhu-
man research to human clinical research, outcome measures used in human studies should
have behavioral analogues at the nonhuman level. (See Table 5-11 for information on inte-
grated tests across species.)
• Function specificity. When there is a choice of instruments it is desirable to avoid
using global outcome measures that combine multiple outcome dimensions into a single
score. The effects on a specific outcome dimension of exposure to a toxin may be lost when
the score on this dimension is combined with scores on other nonaffected dimensions
(Grandjean et al., 1996; McCall and Appelbaum, 1991). Alternatively, children exposed to
different toxins may end up with the same global score, but may arrive at this score via
different developmental paths, thus masking the differential impact of different toxins
(Jacobson, 1998).
• Ease of administration. Developmental-behavioral measures used in the infancy pe-
riod should be noninvasive and, all other criteria being satisfied, they should be relatively
easy to administer.
Utilizing Appropriate Study Design

In addition to the criteria for instrument selection, a review of the research literature
also indicated that certain requirements should be an essential part of any human study that
examines the potential consequences of the addition of new ingredients to infant formulas.
The committee recommends that the following four requirements be met.
• Establish adequate statistical power. As noted earlier, there is a critical need to avoid
type II errors when evaluating safety. Investigators should document that there is sufficient
statistical power to detect adverse effects for all studies in this area. The importance of
establishing the adequacy of statistical power is illustrated in studies investigating the devel-
opmental-behavioral consequences of the addition of LC-PUFAs to infant formulas. In a
recent review of this question, Wroble and colleagues (2002) cited seven studies involving
term infants. The committee assessed the statistical power of these studies plus two addi-
tional studies with term infants that were not cited in this review. In computing statistical
power, it was assumed that studies should be able to detect at least moderate effect sizes. The
analysis indicated that the average level of statistical power across these nine studies was
0.56 (median power = 0.48, range across studies = 0.34–0.97). Not surprisingly, in their
review Wroble and colleagues (2002) concluded that the evidence for developmental-behav-
ioral benefits for term infants from adding LC-PUFAs to infant formulas was inconsistent at
best. While adverse consequences of such additions were reported in two of the nine studies
on which the committee did power calculations (Jensen et al., 1997; Scott et al., 1998), for
the majority of studies reviewed statistical power was inadequate to detect either beneficial

or adverse consequences of the addition of LC-PUFAs to infant formulas. Demonstration

that statistical power is sufficient to detect at least moderate-effect sizes is strongly recom-
mended for all future clinical studies on the safety of the addition of new ingredients to
infant formulas. Documentation of sufficient power is the responsibility of those seeking to
add a new ingredient.
• Avoid control of mediator variables. A typical design strategy is to statistically con-
trol for confounding variables that may covary with the predictor (toxin) or the outcome
variable (e.g., statistically control for the influence of sociodemographic covariates of lead
exposure in infants). However it is critical to distinguish between a confounder and a
mediator variable. Confounder variables are independent risk factors that occur at a higher
level among those exposed to a toxic substance than those not exposed. In contrast, media-
tor variables are developmental risk factors whose occurrence is caused by exposure to a
toxic substance (Neuspiel, 1995). Mediator variables covary with the predictor and the
outcome variable because they are the mechanism through which the predictor influences
the outcome. If mediating variables are statistically controlled, overcorrection may occur
and a real risk effect may be missed (Jacobson and Jacobson, 1996). For example, prenatal
alcohol exposure has been shown to influence infant temperament, which influences mother-
child interaction, which influences child cognitive performance (O’Connor et al., 1993). In
this case it would be a mistake to covary out either infant temperament or mother-infant
interaction since these are mediators of the effects of prenatal alcohol exposure (for a recent
review of design and statistical procedures used to assess the impact of mediator variables,
see Shrout and Bolger, 2002).
• Use measurement aggregation. In developmental-behavioral studies not all measures
of a construct have equal sensitivity. Therefore it is important to use converging multiple
measures of the construct so a real link is not missed due to reliance on a single, less-sensitive
measure.
• Use repeated measures. It is important to look at performance on outcome measures
across time. This is partly because the impact of exposure to a toxin may vary depending on
age level (neural maturation) of the infant (Shaheen, 1984). In addition, the sensitivity of
various measures of the same dimension may vary depending on the age of the child (e.g.,
intramodal visual recognition memory is a sensitive predictor at 7 months of age but not at
12 months; crossmodal transfer, which taps the same cognitive dimension, but at a more
complex level, is not a sensitive predictor at 7 months of age but is at 12 months (Rose et al.,
1989).
Assessing Sensory and Motor Functions
Visual and Auditory Function

Instruments that could be used to evaluate infant sensory competencies in studies on the
developmental effects of adding new ingredients to infant formulas are shown in Table 6-11.
Even though the instruments described in the table have not been specifically linked to CNS
development, sensory functions tapped by these instruments clearly have specific associa-
tions with brain structure and function. While preferential-looking procedures to test visual
function and visual reinforcement audiometry to test auditory function have not been used
in studies with infants exposed to toxins, they have been extensively used in clinical studies.
These two procedures are thus recommended as being “state of the art” in the behavioral
assessment of visual and auditory function. They could be easily combined with the func-

tional neurological measures described later in this section, such as brain stem-evoked
response, to obtain converging measurements of infant sensory competence.
Motor Function
In contrast to the areas of visual and auditory assessment, where there are relatively few
behavioral measures available in the first year of life, there are far more choices available for
the measurement of early motor function. Also in contrast to assessment of sensory function,
none of the motor measures can be described as being the “gold standard.” A list of
available measures for the assessment of infant motor function are shown in Table 6-12.
Even though specific instruments described in this table have not been linked to CNS
development, links between motor and brain development have been well established. Ab-
normal motor behavior has been reported as characteristic of infants exposed to toxins
(Schneider et al., 1989), but studies linking such exposure to performance on the instru-
ments described in Table 6-12 have been relatively rare. In the absence of more satisfactory
measures, two instruments are provisionally recommended to assess adequate motor devel-
opment in studies involving the safety of addition of new ingredients to infant formulas: the
Alberta Infant Motor Scale and the Movement Assessment of Infants Scale. While the long-
term predictive validity of these specific scales has yet to be established, early motor compe-
tence per se is an important precursor for later critical developmental functions, such as
exploration, goal-directed behavior, and spatial learning (Bertenthal et al., 1984; Bushnell
and Boudreau, 1993; Smitsman, 2001). In addition, evidence from several studies has in-
dicated that these instruments are sensitive to early exposure of human infants to toxic
substances.
Assessing Cognitive Development

The majority of behavioral teratology and behavioral toxicology research studies that
assessed infant cognitive development as an endpoint utilized the Bayley Scales of Infant
Development as an outcome measure (Jacobson and Jacobson, 2000; Kaltenbach and
Finnegan, 1989). There are a variety of reasons to question the heavy reliance on the Bayley
scales, including poor predictive validity of performance when administered to infants less
than 2 years of age (Bendersky and Lewis, 2001; Colombo, 1993), and the fact that even in
the revised version of the scale, the first-year mental scale is still heavily weighted with motor
items. These problems are particularly critical given the fact that it is during the first year
when infant formulas are most likely to be consumed. In addition, the earlier (Bayley, 1969)
version of the Bayley scale, used in the majority of previous research on this topic, yields a
global mental development score. One consequence of using a global score is that the
significant impact of developmental risk factors on a specific cognitive dimension may be
lost when the depressed score on the affected dimension is combined with scores on other
nonaffected cognitive dimensions (Grandjean et al., 1996; Jacobson, 1998; McCall and
Appelbaum, 1991). While the most recent revision of the Bayley scales does allow scoring of
dimensional subscale scores, the heterogeneous factor structure of these subscales still in-
creases the likelihood that a specific effect on a single cognitive dimension could well be
attenuated (Bendersky and Lewis, 2001). Thus the global score problem has not been
completely solved, even in the revised edition of this scale (Bendersky and Lewis, 2001).
The importance of going beyond global cognitive assessments in studies on the impact of
potential developmental risk factors has been strengthened by both conceptual and empiri-

TABLE 6-11 Behavioral Measures Available to Test Sensory Functions in Clinical Studies on the Safety of New Additions to
134
Infant Formulas
Description Selection Criteria Met Rating Comments
Tests of visual function
Preferential looking procedures Can be administered during the first Recommended for use Current standard measure of preferen-
under controlled conditions using year (Mayer and Arendt, 2001) in either level 2 or tial looking is the Teller Acuity
stimuli with different levels of Documented links to central nervous level 3 assessments Card Procedure (Teller et al., 1986)
visual contrast to assess visual system (CNS) structure or function Early acuity deficits can have long-
acuity and contrast sensitivity (Slater, 2001) term consequences (Posner, 2001),
(Posner, 2001) Analogous measures available at the but little evidence on the specific
nonhuman level (Banks and Salapatek, predictive validity of this procedure
1983; Jacobs and Blakemore, 1988) A lack of response may be due to
Assesses specific functions (Posner, infant fatigue or reduced attention
2001) and examiner experience essential
Relative ease of administration for younger infants (Mayer and

(Mayer and Arendt, 2001) Arendt, 2001)
Tests of auditory function

Visual reinforcement audiometry, Can be administered during the first Recommended for use Can be taught quickly since the
where the child’s head turning year (Cobo-Lewis and Eilers, 2001) in either level 2 or behavioral response is very clear
toward different sounds at different Documented links to CNS structure or level 3 assessments and computers can be used to
frequency levels are reinforced by an function (Fernald, 2001) control the recording and
interesting visual stimulus (Moore Assesses specific functions (Cobo-Lewis reinforcements
et al., 1977) and Eilers, 2001) Early hearing problems have long-
Relative ease of administration (Cobo- term consequences, but little
Lewis and Eilers, 2001) evidence on the predictive value of
this procedure per se
Except under highly controlled
laboratory conditions, best used

with infants at or above 5 mo of
age (Cobo-Lewis and Eilers, 2001)
Observer-based psychoacoustic Can be administered during the first Not recommended for Can be used with infants as young
procedures, based on the same year (Cobo-Lewis and Eilers, 2001) level 2 assessments, as 2 mo of age, but requires
principle as visual reinforcement Documented links to CNS structure or but can be used as specialized and extensive examiner
audiometry except that observers function (Fernald, 2001) the alternate training (Cobo-Lewis and Eilers,
utilize multiple, often subtle, cues Assesses specific functions (Cobo-Lewis instrument for level 2001)
to judge if an infant is orienting and Eilers, 2001) 3 assessments
towards a sound stimulus (Olsho
et al., 1987)
Habituation procedures based on Can be administered during the first Not recommended for Can be used with relatively young
sounds being played when an infant year (Fernald, 2001) level 2 assessments, infants, but requires sophisticated
displays a particular behavior, such Assesses specific functions (Fernald, but can be used as equipment
as high intensity sucking; when the 2001) the alternate
infant behavior declines a new instrument for level
sound is played and if the infant 3 assessments
can discriminate, then the behavior
should increase in frequency
(Jusczyk, 1985)

135
TABLE 6-12 Behavioral Measures Available to Test Motor Functions in Clinical Studies on the Safety of New Additions to
136
Infant Formulas
Alberta Infant Motor Scale Can be administered during the first In the absence of more Range from birth–18 mo (Piper and
Norm-referenced, observational- year (Piper and Darrah, 1994) adequate measures, Darrah, 1994); has satisfactory
based assessment of gross motor Has shown sensitivity to exposure to recommended as the test-retest and interobserver
and postural development (Piper toxic substances during the first year best option available reliability and discriminative
and Darrah, 1994) (Fetters and Tronick, 1996) for use in either level validity; little assessment of fine
Documented links to central nervous 2 or level 3 studies motor skills
system (CNS) structure or function Background in infant motor
(Tanner, 1970) development important for valid
Assesses specific functions (Piper and observations
Darrah, 1994)
Movement Assessment of Infants Scale Can be administered during the first In the absence of more Range from birth–12 mo; inconsistent
Assesses muscle tone, reflexes, and year (Case-Smith and Bigsby, 2001) adequate measures, evidence on predictive validity
functional movement; 65 items with Has shown sensitivity to exposure to recommended as the (Case-Smith and Bigsby, 2001)
risk points for each item (Harris toxic substances during the first best option available Experience with infant motor
et al., 1984) year (Arendt et al., 1998; Fetters for level 2 or level development essential for valid
and Tronick, 1996) 3 assessments administration (Chandler et al.,
Documented links to CNS structure or 1980)
function (Tanner, 1970)
Assesses specific functions (Harris
et al., 1984)
Revised Bayley Psychomotor Scale Can be administered during the first Meets few selection 84% item overlap from 1969 version;
Revised version of earlier Bayley year (Bayley, 1993) criteria; use only better predictive validity for clinical
motor scale that assesses gross and Documented links to CNS structure or under limited or than nonclinical populations
fine motor skills (Bayley, 1993) function (Tanner, 1970) special circumstances (Bendersky and Lewis, 2001)

Data inconsistent in regard to
sensitivity of Bayley motor scale to
exposure to toxic substances (e.g.,
Mayes and Cicchetti, 1995; Singer
et al., 1997 vs. Hurt et. al., 1995;
Jacobson et al., 1996)
Infant Neurological International Can be administered during the first Meets few selection Range from birth–18 mo; satisfactory
Battery year (Case-Smith and Bigsby, 2001) criteria; use only inter-rater, test-retest, and
20-item, criterion-referenced scale under limited or discriminative validity (Case-Smith
assessing muscle range, reflexes, and special circumstances and Bigsby, 2001)
motor milestones (Ellison, 1994)
Milani-Comparetti Development Can be administered during the first Meets few selection Range 1–16 mo (Case-Smith and
Screening Test-Revised year (Milani-Comparetti and Gidoni, criteria; use only Bigsby, 2001); can be administered
27-item, norm-referenced measure 1967) under limited or in a relatively brief period of time;
that assesses reflexes, postural Documented links to CNS structure or special circumstances satisfactory inter-rater and test-
control, and movement (Milani- function (Tanner, 1970) retest reliability

Comparetti and Gidoni, 1967)
Test of Infant Motor Performance Can be administered during the first Meets few selection Restricted age range from birth–4 mo;
Consists of 27 observational items year (Campbell et al., 1993) criteria; use only satisfactory inter-rater reliability
and 25 items that are elicited by the Documented links to CNS structure or under limited or and internal consistency
examiner; norm-referenced measure function (Tanner, 1970) special circumstances Requires highly trained examiner
that assesses posture and movements (Case-Smith and Bigsby, 2001)
(Campbell et al., 1993)
Toddler and Infant Motor Evaluation Can be administered during the first Meets few selection Well standardized with satisfactory
Norm-referenced measure based on year (Miller and Roid, 1994) criteria; use only inter-rater and test-retest reliability
parent-elicited behaviors and Documented links to CNS structure or under limited or and discriminative validity
observation; assesses motor mobility, function (Tanner, 1970) special circumstances Requires skilled examiner and needs
stability, motor organization, and to be scored from videotape (Case-
functional performance (Miller and Smith and Bigsby, 2001)
Roid, 1994)
Visually Guided Reaching Can be administered during the first Use only under limited Not a formal test, but there are
A normal developmental function year (Bushnell, 1985) or special laboratory-based procedures to
that shows an increase in amount Documented links to CNS structure or circumstances assess the changes in visually guided
and coordination from 4–8 mo; function (Johnson, 2001; von Hofsten reaching over time

typically after 8 mo the infant no and Fazel-Zandy, 1984) Based on developmental norms;
longer needs ongoing visual guidance Analogous measures available at the would expect increases in this
for accurate reaching (Bushnell, nonhuman primate level (Johnson, behavior from 4–8 mo with
1985; von Hofsten, 1991) 2001) decreases thereafter if the infant is
Assesses specific functions (Bushnell, developing normally
1985)
137
cal advances in the study of early cognitive development. Specifically, there is increasing
agreement among developmental researchers that early intellectual development can be
characterized by performance on three specific dimensions, which the committee recom-
mends as the preferred dimensions of early cognitive performance for future clinical research
studies. These are:
• Speed of processing, as assessed on tasks like habituation and visual expectancy

performance (Colombo, 1993; Dougherty and Haith, 1997). Tasks of this type may be
sensitive to nutritional intakes related to neural developmental processes involving myelina-
tion, synaptogenesis, and pruning (Rao and Georgieff, 2000).
• Recognition memory, as assessed on tasks like novelty preference and conjugate
reinforcement (Fagen and Ohr, 2001; Rovee-Collier and Barr, 2001). Tasks of this type may
be sensitive to intake of energy-glucose, iron, and zinc, which may relate to hippocampal
development (Nelson et al., 2000; Pinero et al., 2001; Takeda, 2001).
• Behavioral inhibition, as assessed on measures like the A-not-B task (Fagen and Ohr,
2001; McCall, 1994). Tasks of this type may be sensitive to nutrients involved in monoam-
ine synthesis (Lozoff et al., 1998).
Table 6-13 summarizes the available evidence that the committee reviewed in order to
recommend these specific measures of cognitive function during the first year of life. While
the table does not include an exhaustive list, it clearly illustrates the utility of assessment
procedures that tap the three dimensions in studies on the potential cognitive consequences
associated with the addition of new ingredients to infant formulas. Evidence suggests that
toxic substances may differentially impact on different dimensions of cognitive performance
(e.g., prenatal alcohol exposure is related to slower processing speed in infancy, but not to
memory performance, and prenatal exposure to polychlorinated biphenyls adversely affects
infant memory, but not processing speed [Jacobson, 1998]). This pattern of specificity may
reflect differential sensitivity of the exposure of various brain areas to toxic substances,
underlying different dimensions of early information processing (Jacobson and Jacobson,
2000; Mayes and Bornstein, 1995; Rao and Georgieff, 2000). The likely possibility of
specificity of effects due to different potential toxins underlies the recommendation to assess
at least two different dimensions of infant cognitive performance in level 3 studies on the
cognitive impact of the addition of new ingredients to infant formulas.
Assessing Infant Temperament

Over the past 15 years there has been an increasing focus on noncognitive aspects of
early development, with specific reference to the domain of infant temperament (Kohnstamm
et al., 1989). Temperament refers to early-appearing, biologically rooted individual differ-
ences in behavioral tendencies that are relatively stable across contexts and time (Bates,
1989). The biological roots of individual differences in temperament include genetic influ-
ences (Goldsmith et al., 1994), exposure to toxic substances (Alessandri et al., 1995), and
nutritional deficiencies (Wachs, 2000). While researchers have identified a number of differ-
ent domains of infant temperament, recent formulations have indicated that these different
domains converge on two major dimensions: reactivity and self-regulation (Rothbart and
Bates, 1998). Reactivity refers to the speed, quality, and strength of the infant’s reaction to
stimulation (e.g., stimulus sensitivity, negative affect). Self-regulation refers to the quality
and level of the infant’s ability to control emotional responses (e.g., soothability, attentional
allocation). Individual differences in temperament-driven behavioral patterns, like tenden-

cies towards approach and inhibition, are based on the balance between individual self-
regulation and reactivity.
There have been two major approaches used to assess infant temperament. The first
involves parent report measures using standardized scales. While parent report measures
have been criticized for assessing parental emotional qualities rather than child temperament
characteristics, reviews indicate that such measures are predictive in ways that would not
occur if they were just measuring parental characteristics (Wachs and Bates, 2001). The
second approach utilizes structured laboratory-based assessments of infant temperament.
These assessments involve presenting a series of structured situations to an infant, videotap-
ing the infant’s behavior during the situations, and then coding the observed behaviors into
temperament-related dimensions. Laboratory-based assessment procedures have been criti-
cized for only sampling a restricted set of child behaviors. However the strengths of this
approach include high intercoder reliability when using trained coders and the fact that these
procedures do predict later developmental outcomes (Wachs and Bates, 2001).
At present, few of the standard parent report or laboratory-based assessment approaches
for assessing infant temperament have been utilized in behavioral teratology or behavioral
toxicology studies. However there are a number of reasons that support the need for use of
such measures in future studies on questions involving the safety of new ingredients added to
infant formulas. Specifically, evidence indicates that measures of temperament in the first
year of life predict later personality (Rothbart et al., 2000), as well as attention and behav-
ioral problems in preschool and school-age children (Bates, 2001; Posner, 2001; Robson and
Pederson, 1997). The predictive function of early temperament problems may be partly due
to an increased likelihood of problems in parent-child reactivity or with impairments in
children’s ability to self-regulate (Bendersky et al., 1996; O’Connor et al., 1993; Rossetti
Ferreira, 1978; Rothbart and Bates, 1998; Schneider et al., 1989).
In addition, individual differences in early temperament have been linked to structural
and functional development of the CNS (Nelson, 1994; Posner, 2001; Rothbart et al.,
1994), and nonhuman analogues to human temperament measures have been reported
(Higley and Suomi, 1989; Laughlin et al., 1991; Spear, 1995). Finally, and perhaps most
critically, although standard temperament instruments have rarely been used in studies of
human infants exposed to toxic substances, a number of studies from the behavioral toxicol-
ogy and teratology literature have reported that such infants do show behavioral impair-
ments that are clearly temperament driven, including problems in reactivity (e.g., increased
irritability, hypo- or hyper-reactivity to stimuli [Hill et al., 1989; Lester et al., 1995]), and in
self-regulation (e.g., poor state control, low consolability, and poor impulse control [Chas-
noff et al., 1987; Mayes et al., 1995]). One consequence of toxin-driven differences in infant
temperament is that infants exposed to toxic substances are harder to test and are less likely
to complete testing due to temperament characteristics, such as high activity, high distract-
ibility, high negative emotionality, hypersensitivity, or low consolability (Fagen and Ohr,
2001; Mayes et al., 1995; Struthers and Hansen, 1992). Infants who fail to complete testing
due to temperament characteristics may be at increased risk for later developmental prob-
lems (Bathurst and Gottfried, 1987; Sebris et al., 1984).
Table 6-14 lists recommended indices of infant temperament for future studies of the
safety of the addition of new ingredients to infant formulas. Based on available evidence, the
committee strongly recommends using converging laboratory and parent report measures of
temperament rather than just laboratory or just parent report measures. In addition, the
committee recommends obtaining both mother and father reports for parent report mea-
sures (Wachs and Bates, 2001). Recommended parent report measures include the Infant
Behavior Questionnaire or the Infant Characteristics Questionnaire. Either the LAB-TAB

TABLE 6-13 Measures Available to Test First-Year Cognitive Function in Clinical Studies on the Safety of New Additions to
140
Infant Formulas
Habituation Can be administered during the first Recommended for use The most sensitive predictors of later
Repeated presentation of a visual year (Colombo, 1993). in either level 2 or cognitive function are total looking
or auditory stimulus to an infant; Has predictive value beyond the first level 3 assessments time and duration of longest
observers code length of fixation year of life (McCall and Carriger, fixation, both of which decline with
to the stimulus and decline in 1993; Ruff and Rothbart, 1996; age (Fagen and Ohr, 2001); infants
attentional behavior with repeated Slater, 1997) whose fixation times do not decline
presentations (Colombo, 1993) Has shown sensitivity to exposure to with age are showing slower
toxic substances during the first year processing speed and would be
(Hill et al., 1989; Jacobson and considered as being at risk
Jacobson, 2000) (Colombo, 1993)
Documented links to central nervous Sensitivity of habituation procedure
system (CNS) structure (Nelson, to toxic exposure may be higher in

1995) first several months of life (Fagen
Assesses specific functions (Colombo, and Ohr, 2001)
1993)
Relative ease of administration (Fagen
and Ohr, 2001)
Recognition memory Can be administered during the first Recommended for use Less than 53% preference for novel
Based on the preference of infants year (Colombo, 1993) in either level 2 or stimuli is used as the cut-off to
for novel stimuli; infants are Has predictive value beyond the first level 3 assessments distinguish at-risk from not-at-risk
familiarized with a stimulus and year of life (Rose and Feldman, infants (Fagan et al., 1986;
then the familiar stimulus is paired 1995; Rose et al., 1989, 1991; Jacobson et al., 1996)
with a novel stimulus; over repeated Slater, 1997) Intramodal memory tasks are a
trials with different stimuli, observers Has shown sensitivity to exposure to stronger predictor of later cognitive
code whether the infant orients toxic substances during the first year performance below 12 mo;

toward the familiar or the novel (Jacobson et al., 1996; Rose and crossmodal tasks are a stronger
stimulus (Rose and Orlian, 2001). Orlian, 2001); Struthers and Hansen, predictor at 12 mo (Rose and
In intramodal tasks the familiar and 1992) Feldman, 1995)
novel stimuli are in the same Documented links to CNS structure Crossmodal transfer can be seen in
modality; cross-modal transfer is a (Johnson, 2001; Nelson, 1995) infants as young as 6 mo, but is
more complex task in that it involves Analogous measures available at the much more characteristic of infants
the child being familiarized with the nonhuman primate level (Grant- who are closer to 12 mo (Rose and
stimulus in one sensory modality Webster et al., 1990; Rose and Orlian, 2001)
(e.g., tactual) and tested in a different Orlian, 2001)
modality for novelty preference Assesses specific functions (Colombo,
(e.g., visual) with the same familiar 1993)
stimulus and a novel stimulus (Rose Relative ease of administration (Rose
and Feldman, 1990) and Orlian, 2001)
Conjugate reinforcement learning tasks Can be administered during the first Not recommended for Speed of acquisition is not predictive
Infant actions (e.g., leg kicking, arm year (Rovee-Collier and Barr, 2001) level 2 assessments; of later intelligence, but retention
waving, hitting a lever) activate an Has predictive value beyond the first can be used as the of what has been learned is
interesting audio-visual stimulus year of life (Colombo, 1993; Fagen alternate instrument predictive (Fagen and Ohr, 2001)
(e.g., mobile moves, clown face and Ohr, 2001) for level 3 Tasks can be computer driven, which
lights up, toy train moves); by Has shown sensitivity to exposure to assessments increases cost, but then requires
repeating this procedure across toxic substances during the first little examiner training; other tasks
sessions, measures of speed of year (Alessandri et al., 1993) are nonautomated but require a
acquisition, level of retention, and Documented links to CNS structure moderate amount of examiner
speed of extinction can be obtained (Nelson, 1995) training (Fagen and Ohr, 2001)
(Rovee-Collier and Barr, 2001) Assesses specific functions (Rovee-

Collier and Barr, 2001)
Relative ease of administration (Fagen
and Ohr, 2001)
A-not-B task Can be administered during the first Not recommended for Past 6 mo of age, as infants get older,
Infant sees an interesting object year (Ruff and Rothbart, 1996) level 2 assessments; there is an increase in the amount
hidden at a specific location; after Documented links to CNS structure can be used as the of delay time infants can tolerate
infant retrieves the object, infant (Nelson, 1995; Posner, 2001) alternate instrument and still search successfully; older
then sees the object hidden at a Analogous measures available at the for level 3 infants who search where the object
different location; does the infant nonhuman primate level (Diamond, assessments was hidden only with no delay or
go to the location where he/she 1990) who do not show increased
previously found the object hidden Assesses specific functions (Ruff and tolerance for delay over time would
or the location where he/she last saw Rothbart, 1996) be considered as being at risk
the object hidden; the task can be Relative ease of administration (Ruff (Ruff and Rothbart, 1996)
made more difficult by increasing and Rothbart, 1996)

the time between when the infant
saw the object hidden and when
he/she is allowed to search for it
(Ruff and Rothbart, 1996)

141
TABLE 6-13 continued
142

Visual expectation Can be administered during the first Not recommended for Stimulus anticipation is a less-strong
Infant seated before a stimulus year (Colombo, 1993) level 2 assessments; predictor of later cognitive
panel that has lights that flash on Has predictive value beyond the first can be used as the performance than reaction time
and off in a regular sequence; year of life (Dougherty and Haith, alternate instrument (Colombo, 1993)
assesses whether child orients to the 1997) for level 3 Visual expectation performance either
next light in the sequence before it Assesses specific functions (Colombo, assessments does not distinguish between infants
comes on; can obtain measures of 1993) who were or were not exposed to
reaction time and number of correct toxic substances (Fagen and Ohr,
anticipations (Colombo, 1993) 2001), or findings are inconsistent
with regard to the question of
whether exposed infants show
slower or faster reaction times
(Jacobson, 1998)
Requires both sophisticated equipment
and a high level of examiner
training (Fagen and Ohr, 2001)
Focused attention Can be administered during the first Not recommended for Infants showing less than 2% of time
Usually assessed during free play; year (Ruff and Rothbart, 1996) level 2 assessments; spent in focused attention during
using standardized rating criteria, Has predictive value beyond the first can be used as the free-play task at 7 mo may be at
observers code the amount of time year of life (Lawson and Ruff, 2001) alternate instrument developmental risk (Lawson and
the infant is attending to properties Documented links to CNS structure for level 3 Ruff, 2001)
of the object they are playing with (Lawson and Ruff, 2001) assessments Requires substantial training to learn
or exploring the object to see what Assesses specific functions (Ruff and to recognize focused attention and
can be done with it (Lawson and Rothbart, 1996) to maintain examiner calibration
Ruff, 2001) over time (Lawson and Ruff, 2001)

Bayley Mental Development Scale: Can be administered during the first Meets few selection Consistent evidence indicating low
2nd ed. (Bayley, 1993) year (Bayley, 1993) criteria; use only predictive validity of Bayley scale
under limited or performance when administered
special circumstances below 2 y (Bendersky and Lewis,
2001; Colombo, 1993);
inconsistency of findings when
comparing Bayley mental
development scores of infants
exposed to toxic substances versus
nonexposed control infants (e.g., see
Arendt et al., 1998; O’Connor et
al., 1993; Singer et al., 1997 vs.
Hurt et al., 1995; Jacobson et al.,
1996; Mayes and Cicchetti, 1995)
Most recent revision of the Bayley
scales does allow scoring of
dimensional subscale scores, but the
heterogeneous factor structure of

these subscales still increases the
likelihood that a specific effect on a
single cognitive dimension could
well be attenuated (Bendersky and
Lewis, 2001)

143
or the Louisville Temperament Assessment Battery are recommended for laboratory-based

measures.
Neurological Function
Knowledge of the associations among toxin, brain, and behavior increases the ability to
detect the neural precursors of developmental-behavioral consequences of early exposure to
toxic substances; knowledge of toxin-affected neural precursors helps to select which behav-
ioral outcomes are most likely to be affected by exposure to toxins; and knowledge of brain-
behavior relations can aid the researcher in selecting which areas of the brain to investigate
in children who display specific developmental deficits as a result of exposure to toxic
substances (Jacobson and Jacobson, 2000). With the explosion in neuroimaging techniques,
there have been major advances in the study of brain metabolism and electrical activity as a
window to CNS structure and function in children (Nelson and Bloom, 1997; Posner, 2001).
Neuroimaging techniques can provide a more objective assessment of brain activity than can
neurobehavioral measures. Further, many of these new techniques offer the promise of
allowing researchers to detect relatively subtle neural deficits that may result from exposure
to toxic substances.
Table 6-15 lists measures of CNS structure and function that may be useful in the
study of developmental consequences of exposure to new ingredients in infant formulas.
Electroencephalographic assessment and measures based on assessment of event-related
potential are specifically recommended. While the instruments listed in Table 6-15 fit most
of the selection criteria presented earlier, it is important to recognize that almost all of
these instruments are expensive and require highly trained staff to operate them and to
interpret their results. The committee has noted in this table which instruments are rela-
tively less costly or require relatively lower staff expertise. It should also be noted that not
all existing measures of CNS structure and function are listed in the table. Certain mea-
sures, such as positron emission tomography (Bookheimer, 2000), X-ray computed to-
mography (Singer, 2001) and assessment of homovanillic acid levels as a marker for
dopamine status (Needlman et al., 1995), are invasive procedures and thus contraindi-
cated for normal infants. Other potentially promising measures, such as magnetic diffu-
sion tensor imaging (Posner, 2001), magnetic encephalography, and functional near-infra-
red spectroscopy (Nelson et al., 2002), have only recently been developed and more needs
to be known before their utility in studies on neural consequences of the addition of new
ingredients to infant formulas can be determined.
SUMMARY
Infancy is a particularly vulnerable period of development. A difference in growth is
likely to signal some alteration in underlying biological or physiological processes. The
committee took the conservative position that any systematic difference in growth that could
be attributed to a new formula ingredient rather than to chance alone should represent a
safety concern and should require explanation and further study. The major organ systems
should be studied because growth deficits are likely to appear only secondary to effects on
specific organs or tissues and may not appear for some time after nutritional insult. The
committee therefore recommends implementing a hierarchy of two levels of clinical assess-
ments for organ evaluations.
It is essential to include developmental-behavioral outcomes in future studies of the
safety of ingredients new to infant formulas because such measures are sensitive to exposure

TABLE 6-14 Measures Available to Test First-Year Temperament in Clinical Studies on the Safety of New Additions to
Infant Formulas
Parent report measures
Infant Behavior Questionnaire Can be administered during the first Recommended for use Infants rated as higher on negative
87-item, parent report measure; year (Rothbart, 1986) in either level 2 or affect, inhibition, and intensity
can be administered starting at Has shown sensitivity to exposure to level 3 assessments (especially when linked to negative
age 3 mo; assesses 5 dimensions toxic substances during the first year affect) and low on sociability,
of temperament (Rothbart, 1986) (Alessandri et al., 1995) approach, positive affect, or sooth-
Documented links to CNS structure ability would be considered as being
(Goldsmith et al., 2000) at greater risk for developmental
Assesses specific functions (Rothbart problems.
and Bates, 1998) Relatively long measure and parents
need to be literate in English
Infant Characteristics Questionnaire Can be administered during the first Recommended for use Infants rated as high on intense
24-item, parent report measure; can year (Bates, 2001; Bates and Bayles, in either level 2 or negative mood would be at greater
be administered starting at age 4 mo; 1984) level 3 assessments risk for later developmental
assesses level of high-intense negative Assesses specific functions (Bates and problems (Bates, 2001)
mood (Bates and Bayles, 1984) Bayles, 1984)
Relative ease of administration (Bates
and Bayles, 1984)
Revised Infant Temperament Can be administered during the first Not recommended for Infants rated as higher on negative
Questionnaire year (Carey and McDevitt, 1978) level 2 assessments, affect, inhibition, and intensity
95-item, parent report measure; can Has predictive value beyond the first but can be used as (especially when linked to negative
be administered starting at age 4 mo; year of life (Sanson et al., 1996) the alternate affect) and low on sociability,
assesses 9 dimensions of temperament Assesses specific functions (Martin instrument for level approach, positive affect, or sooth-
(Carey and McDevitt, 1978) et al., 1994) 3 assessments ability would be considered as being

at greater risk for developmental
problems
Relatively long measure and parents
need to be literate in English

145
146

Laboratory-based assessments
LAB-TAB Can be administered during the first Recommended for use Infants rated as higher on negative
Laboratory-based temperament year (Goldsmith and Rothbart, 1991) in either level 2 or affect, inhibition, and intensity
measure; assesses 5 dimensions of Has predictive value beyond the first level 3 assessments (especially when linked to negative
temperament, starting at age 6 mo year of life (Rothbart et al., 2000) affect) and low on sociability,
(Goldsmith and Rothbart, 1991) Assesses specific functions (Goldsmith approach, positive affect, or sooth-
and Rothbart, 1991) ability would be considered as being

at greater risk for developmental
problems
Test requires a high level of examiner
skill and coder competence
The Behavior Rating Scale on the Can be administered during the first Not recommended for Predictive validity has been demon-
1993 Revision of the Bayley Scales year (Bayley, 1993) level 2 assessments, strated for the 1969 Bayley Infant
of Infant Development-II Analogous measures available at the but can be used as Behavior Rating Scale (Matheny,
Used to rate infant behavior during nonhuman primate level (Bayley, the alternate 1989)
cognitive testing on 4 empirically 1993) instrument for level Cut-off scores for each score have
derived factor scores, 3 of which 3 assessments been developed, with scores below
involve temperament: attention/ the 10th percentile being considered
arousal, orientation/engagement, and in the risk range (Bendersky and
emotional regulation (Bayley, 1993) Lewis, 2001)
Requires extensive examiner training
(Bendersky and Lewis, 2001)
Louisville Temperament Assessment Can be administered during the first Recommended for use Infants rated as higher on negative
Battery year (Matheny, 1991) in either level 2 or affect, inhibition and intensity
Laboratory-based temperament Has predictive value beyond the first level 3 assessments (especially when linked to negative
measure; assesses 7 areas of year of life (Matheny and Phillips, affect) and low on sociability,

temperament, starting at age 3 mo 2001) approach, positive affect, or sooth-
(Matheny, 1991) Assesses specific functions (Matheny, ability would be considered as being
1991) at greater risk for developmental
problems
Test requires high level of examiner
competence and coder skill
TABLE 6-15 Measures Available to Test First-Year Neural Function in Clinical Studies on the Safety of the Addition of
Ingredients New to Infant Formulas
Event-related potential Can be administered during the first Recommended for use Noninvasive measure that is less
Assessment of latency and amplitude year (Posner, 2001) in either level 2 or sensitive to movement artifact then
of electrical changes in different Has predictive value beyond the first level 3 assessments other neural measurements
regions in the brain in response to year of life (Molfese and Molfese, Requires substantial investment in
specific sensory stimulation (Nelson 2001) equipment and examiner training

and Bloom, 1997) Documented links to central nervous (Posner, 2001)
system (CNS) structure or function Spatial resolution is less adequate
(Posner, 2001) than temporal resolution, but still
Assesses specific functions (Nelson and cannot assume 1 to 1 correspon-
Bloom, 1997) dence between brain electrical
activity and what is recorded
(Nelson et al., 2002)

In contrast to evoked potentials,
which assess the integrity of
primary sensory pathways, measures
appear to be assessing aspects of
higher order cognitive processing
such as attention
Electroencephalograph (EEG) Can be administered during the first Recommended for use Measures are noninvasive and are
Based on computerized analysis of year (Marshall and Fox, 2001) in either level 2 or relatively inexpensive compared
electrical activity in different areas Has predictive value beyond the first level 3 assessments with other functional measures of
of the brain; of particular relevance year of life (Fox et al., 2001) the brain, but extensive training is
are studies of relative electrical Has shown sensitivity to exposure to required to avoid artifact and to
activity in different hemispheres of toxic substances during the first year interpret results correctly (Marshall
the brain (Marshall and Fox, 2001) (Kaneko et al., 1996; Needlman and Fox, 2001)

et al., 1995) Cannot assume a 1 to 1 correspon-
Documented links to CNS structure or dence between brain electrical
function (Posner, 2001) activity and what is recorded
Analogous measures available at the (Nelson et al., 2002)
nonhuman level (Needlman et al.,
1995)
Assesses specific functions (Marshall
and Fox, 2001)
147
148
Cardiac variability-vagal tone Can be administered during the first Not recommended for Individual differences in vagal tone
Changes in heart rate as a function year (Porter, 2001) level 2 assessments, have been linked to differences in
of stimulation are related to changes Has predictive value beyond the first but can be used as attention and temperament, but this
in respiratory sinus arrhythmia that year of life (Doussard-Roosevelt, the alternate measure may primarily reflect
reflect changes in the parasympa- et al., 2001; Porges et al., 1996) instrument for level general reactivity (Posner, 2001).
thetic nervous system (Posner, 2001) Has shown sensitivity to exposure to 3 assessments Vagal tone has been shown to
toxic substances during the first year distinguish between breast- and
(DiPietro et al., 1995) formula-fed infants (DiPietro et al.,
Documented links to CNS structure or 1987).
function (Porter, 2001) Can be used in the first year, but
lower stability early in the first year
suggests this measure may be more

useful after 6 mo (Porter, 2001)
Requires extensive instrumentation
and extensive training for valid
interpretation of results (Porter,
2001)
Cortisol Can be administered during the first Not recommended for Can be easily obtained from infant
A hormonal measure based on the year (Gunnar, 2000) level 2 assessments, saliva, but since cortisol production
functioning of the hypothalamic- Has shown sensitivity to exposure to but can be used as follows a circadian rhythm and level
pituitary-adrenocortical axis; cortisol toxic substances during the first year the alternate of salivary cortisol can be influ-
level can be viewed as the level of (Gunnar and White, 2001) instrument for level enced by recent intake of milk or
reactivity of the organism to stress Documented links to CNS structure or 3 assessments milk products, assessment requires
(Gunnar, 2000) function (Gunnar, 2000) controls for time of day and milk
Analogous measures available at the exposure (Posner, 2001)
nonhuman level (Needlman et al., Infants with consistently under- or

1995) overproduction of cortisol are
Relative ease of administration (Gunnar considered as being at risk, but
and White, 2001) there are no specific norms as to
what constitutes under- or over-
reaction (Gunnar and White, 2001);
might be useful for assessing impact
of new ingredients affecting sterol
synthesis (including estrogen/
phytoestrogen), along with its role
as a stress marker
Structural magnetic resonance imaging Can be administered during the first Not recommended for Excellent spatial resolution of brain
Assesses size and changes in brain year (Singer, 2001) level 2 assessments, structure, but requires a heavy
volume for different regions of the Has shown sensitivity to exposure to but can be used as investment in equipment and
brain (Posner, 2001) toxic substances during the first year the alternate training (Posner, 2001)
(Mattson and Riley, 1995) instrument for level There would have to be a large impact
Documented links to CNS structure or 3 assessments of a new ingredient to reduce global
function (Posner, 2001) or regional brain volume; an impact
Analogous measures available at the of this degree should have been
nonhuman primate level (Hopkins detected in preclinical studies
and Rilling, 2000)

Assesses specific functions (Posner,
2001)
Functional magnetic resonance imaging Documented links to CNS structure or Meets few selection Because of the need to lie quietly and
When a brain region is activated function (Posner, 2001) criteria; use only the high noise levels, it is not
to deal with stimulation or task Analogous measures available at the under limited or applicable for children under 6 y;
demands, there is increased blood nonhuman primate level (Nakahara special circumstances however some recent studies using
and oxygen flow to that region; et al., 2002; Sereno, 1998) sedation of infants and passive
magnetic changes associated with Assesses specific functions (Nelson and presentation of stimulation have
increased hemoglobin flow to a Bloom, 1997) reported success with this procedure
specific brain region can be recorded in infancy (Bookheimer, 2000)
as an index of increased activation especially in napping postprandial
of the region involved (Nelson and babies (< 2 mo of age), but sedation
Bloom, 1997) would not be appropriate because
of effects on cognitive processing
(e.g., chloral hydrate)
Brain stem-evoked response Can be administered during the first Use only under High number of false negatives and
EEG response of auditory brainstem year (Cobo-Lewis and Eilers, 2001) limited or special positives limit the utility (Molfese
responses to sound stimuli; allows Has shown sensitivity to exposure to circumstances and Molfese, 2001)
assessment of the functional level of toxic substances during the first year Does allow potential assessment of

noncortical areas involved in hearing (Needlman et al., 1995) conduction speed of neural circuits
(Cobo-Lewis and Eilers, 2001) Analogous measures available at the involved in auditory processing
nonhuman level (Needlman et al., (Roncagliolo et al., 1998)
1995)
Assesses specific functions (Cobo-Lewis
and Eilers, 2001)
149
to toxic substances and can have long-term predictive value. These measures also are impor-
tant because bidirectional brain-behavior links exist. In the case of neurological and behav-
ioral assessment, the committee recommends that a hierarchy of three levels of clinical
assessment be applied.
REFERENCES
AAP (American Academy of Pediatrics). 1988. Clinical Testing of Infant Formulas with Respect to Nutritional
Suitability for Term Infants. Report to the FDA. Committee on Nutrition. Elk Grove Village, IL: AAP.
AAP. 1997. Breast feeding and the use of human milk. Pediatrics 100:1035–1039.
ADA (American Dietetic Association). 2001. Position of the American Dietetic Association: Breaking the barriers
to breastfeeding. J Am Diet Assoc 101:1213–1220.
Adams J. 1993. Structure-activity and dose-response relationships in the neural and behavioral teratogenesis of
retinoids. Neurotoxicol Teratol 15:193–202.
Alessandri SM, Sullivan MW, Imaizumi S, Lewis M. 1993. Learning and emotional responsivity in cocaine-
exposed infants. Dev Psychol 29:989–997.
Alessandri SM, Sullivan MW, Bendersky M, Lewis M. 1995. Temperament in cocaine-exposed infants. In: Lewis
M, Bendersky M, eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale, NJ:
Antonowicz I, Lebenthal E. 1977. Developmental pattern of small intestinal enterokinase and disaccharidase
activities in the human fetus. Gastroenterology 72:1299–1303.
Arant BS Jr, Edelmann CM Jr, Spitzer A. 1972. The congruence of creatinine and inulin clearances in children: Use
of Technicon Autoanalyzer. J Pediatr 81:559–561.
Arendt R, Singer L, Angelopoulos J, Bass-Busdiecker O, Mascia J. 1998. Sensorimotor development in cocaine-
exposed infants. Infant Behav Dev 21:627–640.
Armstrong J, Reilly JJ. 2002. Child Health Information Team. Breastfeeding and lowering the risk of childhood
obesity. Lancet 359:2003–2004.
Auricchio S, Rubino A, Murset G. 1965. Intestinal glycosidase activities in the human embryo, fetus and newborn.
Banks MS, Salapatek P. 1983. Infant visual perception. In: Mussen PH, ed. Handbook of Child Psychology.
Formerly Carmichael’s Manual of Child Psychology. 4th ed, vol 2. New York: John Wiley & Sons. Pp. 435–
571.
Bates JE. 1989. Concepts and measures of temperament. In: Kohnstamm GA, Bates JE, Rothbart MK, eds.
Temperament in Childhood. Chichester, Eng: John Wiley & Sons. Pp. 3–26.
Bates JE. 2001. Adjustment style in childhood as a product of parenting and temperament. In: Wachs TD,
Kohnstamm GA, eds. Temperament in Context. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 173–200.
Bates JE, Bayles K. 1984. Objective and subjective components in mothers’ perceptions of their children from age
6 months to 3 years. Merrill Palmer Q 30:111–130.
Bathurst K, Gottfried AW. 1987. Untestable subjects in child development research: Developmental implications.
Child Dev 58:1135–1144.
Batres LA, Maller ES. 2001. Laboratory assessment of liver function and injury in children. In: Suchy FJ, Sokol RJ,
Balistreri WF, eds. Liver Disease in Children. 2nd ed. Philadelphia: Lippincott Williams & Wilkins. Pp. 155–
169.
Bayley N. 1969. Manual for the Bayley Scales of Infant Development. New York: Psychological Corporation.
Bayley N. 1993. The Bayley Scales of Infant Development. 2nd ed. New York: Psychological Corporation.
Bellinger DC. 1995. Interpreting the literature on lead and child development: The neglected role of the “experi-
mental system.” Neurotoxicol Teratol 3:201–212.
Bendersky M, Lewis M. 2001. The Bayley scales of infant development. Is there a role in biobehavioral assess-
ment? In: Singer LT Zeskind P, eds. Biobehavioral Assessment of the Infant. New York: Guilford Press. Pp.
443–459.
Bendersky M, Alessandri SM, Lewis M. 1996. Emotions in cocaine-exposed infants. In: Lewis M, Sullivan MW,
eds. Emotional Development in Atypical Children. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 89–108.


Berenthal BI, Campos JJ, Barrett KC. 1984. Self-produced locomotion. In: Emde RN, Harmon RJ, eds. Continu-
ities and Discontinuities in Development. New York: Plenum Press. Pp. 175–210.
Binley K, Askham Z, Iqball S, Spearman H, Martin L, de Alwis M, Thrasher AJ, Ali RR, Maxwell PH, Kingsman
S, Naylor S. 2002. Long-term reversal of chronic anemia using a hypoxia-regulated erythropoietin gene
therapy. Blood 100:2406–2413.
Bjarnason I, Macpherson A, Hollander D. 1995. Intestinal permeability: An overview. Gastroenterology 108:1566–
1581.
Bloom L. 1998. Language acquisition in its developmental context. In: Kuhn D, Siegler RS, eds. Handbook of
Child Psychology. 5th ed, vol 2. New York: John Wiley & Sons. Pp. 309–370.
Boehm G, Jakobsson I, Raiha N. 1992. Macromolecular absorption in small-for-gestational-age infants. Acta
Paediatr 81:864–867.
Bookheimer SY. 2000. Methodological issues in pediatric neuroimaging. Ment Retard Dev Disabil Res Rev 6:161–
165.
Brugnara C, Platt OS. 1998. The neonatal erythrocyte and its disorders. In: Nathan DG, Orkin SH, eds. Nathan
and Oski’s Hematology of Infancy and Childhood. 5th ed. Philadelphia: WB Saunders. Pp. 19–52.
Bushnell EW. 1985. The decline of visually guided reaching during infancy. Infant Behav Dev 8:139–155.
Bushnell EW, Boudreau JP. 1993. Motor development and the mind: The potential role of motor abilities as a
determinant of aspects of perceptual development. Child Dev 64:1005–1021.
Butte NF. 2001. The role of breastfeeding in obesity. Pediatr Clin North Am 48:189–198.
Butte N, Heinz C, Hopkinhson J, Wong W, Shypailo R, Ellis K. 1999. Fat mass in infants and toddlers: Compara-
bility of total body water, total body potassium, total body electrical conductivity, and dual-energy x-ray
absorptiometry. J Pediatr Gastroentrol Nutr 29:184–189.
Butte NF, Wong WW, Hopkinson JM, O’Brian Smith E, Ellis KJ. 2000. Infant feeding mode affects early growth
and body composition. Pediatrics 106:1355–1366.
Campbell SK, Osten ET, Kolobe THA, Fisher AG. 1993. Development of the test of infant motor performance.
Phys Med Rehabil Clin N Am 4:541–550.
Canada. 2001. Departmental Consolidation of the Food and Drugs Act and the Food and Drug Regulations (Food
and Drugs Act). Ottawa: Minister of Public Works and Government Services Canada.
Carey WB, McDevitt SC. 1978. Revision of the infant temperament questionnaire. Pediatrics 61:735–739
Case-Smith J, Bigsby R. 2001. Motor assessment. In: Singer LT, Zeskind P, eds. Biobehavioral Assessment of the
Infant. New York: Guilford Press. Pp. 423–441
Chandler LS, Andrews MS, Swanson MW. 1980. Movement Assessment of Infants. A Manual. Rolling Bay, WA:
Chandler, Andrews, and Swanson.
Chasnoff IJ, Burns KA, Burns WJ. 1987. Cocaine use in pregnancy: Perinatal morbidity and mortality. Neurotoxicol
Teratol 9:291–293.
Christensen L. 1996. Diet-Behavior Relationships: Focus on Depression. Washington, DC: American Psychologi-
cal Association.
Cobo-Lewis AB, Eilers R. 2001. Auditory assessment in infancy. In: Singer LT, Zeskind P, eds. Biobehavioral
Assessment of the Infant. New York: Guilford Press. Pp. 95–106.
Colombo J. 1993. Infant Cognition: Predicting Later Intellectual Functioning. Vol. 5. Newbury Park, CA: Sage
Publications.
Cory-Slechta DA. 1990. Bridging experimental animal and human behavioral toxicology studies. In: Russell RW,
Flattau PE, Pope AM, eds. Behavioral Measures of Neurotoxicity. Report of a Symposium. Washington, DC:
National Academy Press. Pp. 137–158.
Cowin I, Emmett P. 2000. Cholesterol and triglyceride concentrations, birthweight and central obesity in pre-school
children. ALSPAC Study Team. Avon Longitudinal Study of Pregnancy and Childhood. Int J Obes Relat
Metab Disord 24:330–339.
Dawson G, Ashman SB. 2000. On the origins of a vulnerability to depression: The influence of the early social
environment on the development of psychobiological systems related to risk for affective disorder. In: Nelson
CA, ed. The Minnesota Symposia on Child Psychology. Vol 31: The Effects of Early Adversity on Neuro-
behavioral Development. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 245–279.
Dewey KG, Heinig MJ, Nommsen LA, Lonnerdal B. 1991. Adequacy of energy intake among breast-fed infants in
the DARLING study: Relationships to growth velocity, morbidity, and activity levels. Davis Area Research
on Lactation, Infant Nutrition and Growth. J Pediatr 119:538–547.
Dewey KG, Heinig MJ, Nommsen LA, Peerson JM, Lonnerdal B. 1992. Growth of breast-fed and formula-fed
infants from 0 to 18 months: The DARLING Study. Pediatrics 89:1035–1041.
Dewey KG, Heinig MJ, Nommsen LA, Peerson JM, Lonnerdal B. 1993. Breast fed infants are leaner than formula
fed infants at 1 y of age: The DARLING study. Am J Clin Nutr 57:140–145.

Diamond A. 1990. The development and neural bases of memory functions as indexed by the AB and delayed
response tasks in human infants and infant monkeys. Ann N Y Acad Sci 608:267–309.
Dietz WH. 1994. Critical periods in childhood for the development of obesity. Am J Clin Nutr 59:955–959.
Di Lorenzo C, Piepsz A, Ham H, Cadranel S. 1987. Gastric emptying with gastroesophageal reflux. Arch Dis
Child 62:449–453.
Dinari G, Rosenbach Y, Zahavi I, Sivan Y, Nitzan M. 1984. Random fecal β1-antitrypsin excretion in children
with intestinal disorders. Am J Dis Child 138:971–973.
DiPietro JA, Larson SK, Porges SW. 1987. Behavioral and heart rate pattern differences between breast-fed and
bottle-fed neonates. Dev Psychol 4:467–474.
DiPietro JA, Suess PE, Wheeler JS, Smouse PH, Newlin DB. 1995. Reactivity and regulation in cocaine-exposed
neonates. Infant Behav Dev 18:407–414.
Dougherty TM, Haith MM. 1997. Infant expectations and reaction times as predictors of childhood speed and
processing and IQ. Dev Psychol 33:146–155.
Doussard-Roosevelt JA, McClenny BD, Porges SW. 2001. Neonatal cardiac vagal tone and school-age develop-
mental outcome in very low birth weight infants. Dev Psychobiol 38:56–66.
Dupont C, Barau E, Molkhou P, Raynaud F, Barbet JP, Dehennin L. 1989. Food-induced alterations of intestinal
permeability in children with cow’s milk-sensitive enteropathy and atopic dermatitis. J Pediatr Gastroenterol
Nutr 8:459–465.
Ellison PH. 1994. The INFANIB. A Reliable Method for the Neuromotor Assessment of Infants. Tucson, AZ:
Therapy Skill Builders.
Ellison PH, Horn JL, Browning CA. 1985. Construction of an Infant Neurological International Battery (INFANIB)
for the assessment of neurological integrity in infancy. Phys Ther 65:1326–1331.
Esmon CT. 1998. Blood coagulation. In: Nathan DG, Orkin SH, eds. Nathan and Oski’s Hematology of Infancy
and Childhood. Philadelphia: WB Saunders. Pp. 1531–1556.
European Task Force on Atopic Dermatitis. 1993. Severity scoring of atopic dermatitis: The SCORAD index.
Consensus Report of the European Task Force on Atopic Dermatitis. Dermatology 186:23–31.
Evangelista de Duffard AM, Duffard R. 1996. Behavioral toxicology, risk assessment, and chlorinated hydrocar-
bons. Environ Health Perspect 104:353–360.
Fagan JF 3rd, Singer LT, Montie JE, Shepherd PA. 1986. Selective screening device for the early detection of
normal or delayed cognitive development in infants at risk for later mental retardation. Pediatrics 78:1021–
1026.
Fagen JW, Ohr PS. 2001. Learning and memory in infancy. Habituation, instrumental conditioning, and expect-
ancy formation. In: Singer L, Zeskind P, eds. Biobehavioral Assessment of the Infant. New York: Guilford
Press. Pp. 233–273.
Falth-Magnusson K, Kjellman N-IM, Odelram H, Sundqvist T, Magnusson K-E. 1986. Gastrointestinal perme-
ability in children with cow’s milk allergy: Effect of milk challenge and sodium-cromoglycate as assessed with
polyethyleneglycols (PEG 400 and 1000). Clin Allergy 16:543–551.
quality factors, notification requirements, and records and reports, for the production of infant formula;
Fee B, Weil WM. 1960. Body composition of a diabetic offspring by direct analysis. Am J Dis Child
100:718.
Fernald A. 2001. Hearing, listening, and understanding: Auditory development in infancy. In: Bremner G, Fogel A,
eds. Blackwell Handbook of Infant Development. Malden, MA: Blackwell Publishers. Pp. 35–70.
Fernandes J, Vos CE, Douwes AC, Slotema E, Degenhart HJ. 1978. Respiratory hydrogen excretion as a param-
eter for lactose malabsorption in children. Am J Clin Nutr 31:597–602.
Fetters L, Tronick EZ. 1996. Neuromotor development of cocaine-exposed and control infants from birth through
15 months: Poor and poorer performance. Pediatrics 98:938–943.
Fomon SJ. 1993. Nutrition of Normal Infants. St. Louis: Mosby.
Fomon SJ, Ziegler EE, Thomas LN, Jensen RL, Filer LJ Jr. 1970. Excretion of fat by normal full-term infants fed
various milks and formulas. Am J Clin Nutr 23:1299–1313.
Fomon SJ, Nelson SE, Ziegler EE. 2000. Retention of iron by infants. Ann Rev Nutr 1:273–290.
Forse RA, Bell SJ, Blackburn GL, Kabbash LG, eds. 1994. Diet, Nutrition, and Immunity. Boca Raton, FL: CRC
Press.
Fox NA, Henderson HA, Rubin KH, Calkins SD, Schmidt LA. 2001. Continuity and discontinuity of behavioral
inhibition and exuberance: Psychophysiological and behavioral influences across the first four years of life.
Child Dev 72:1–21.


Fredrikzon B, Hernell O, Blackberg L, Olivecrona T. 1978. Bile salt-stimulated lipase in human milk: Evidence of
activity in vivo and of a role in the digestion of milk retinal esters. Pediatr Res 12:1048–1052.
Garza C, de Onis M. 1999. A new international growth reference for young children. Am J Clin Nutr 70:169S–
172S.
Gaylor DW, Bolger PM, Schwetz BA. 1998. U.S. Food and Drug Administration perspective of the inclusion of
effects of low-level exposures in safety and risk assessment. Environ Health Perspect 106:391–394.
Gleim GW. 2000. Renal responses to exercise and training. In: Garrett WE, Krikendall DT, eds. Exercise and
Sport Science. Philadelphia: Lippincott Williams & Wilkins. Pp. 217–225.
Godfrey KM, Barker DJ. 2001. Fetal programming and adult health. Public Health Nutr 4:611–624.
Goldsmith DI, Novello AC. 1992. Clinical laboratory evaluation of renal function. In: Edelmann CM, ed. Pediat-
ric Kidney Disease. Boston: Little, Brown. P. 471.
Goldsmith HH, Rothbart MK. 1991. Contemporary instruments for assessing early temperament by questionnaire
and in the laboratory. In: Strelau J, Angleitner A, eds. Explorations in Temperament. International Perspec-
tives on Theory and Measurement. New York: Plenum Press. Pp. 249–272.
Goldsmith HH, Losoya SH, Bradshaw DL, Campos JJ. 1994. Genetics of personality: A twin study of the five-
factor model and parent-offspring analysis. In: Halverson CF, Kohnstamm GA, Martin RP, eds. The Devel-
oping Structure of Temperament and Personality from Infancy to Adulthood. Hillsdale, NJ: Lawrence
Goldsmith HH, Lemery KS, Aksan N, Buss KA. 2000. Temperamental substrates of personality development. In:
Molfese VJ, Molfese DL, eds. Temperament and Personality Development Across the Lifespan. Mahwah, NJ:
Grandjean P, White RF, Weihe P. 1996. Neurobehavioral epidemiology: Application in risk assessment. Environ
Health Perspect 104:397–400.
Grant JD, Bezerra JA, Thompson SH, Lemen RJ, Koldovsky O, Udall JN. 1989. Assessment of lactose absorption
by measurement of urinary galactose. Gastroenterology 97:895–899.
Grantham-McGregor SM, Walker SP, Chang S. 2000. Nutritional deficiencies and later behavioural development.
Proc Nutr Soc 59:47–54.
Grant-Webster KS, Gunderson VM, Brubacher TM. 1990. Behavioral assessment of young nonhuman primates:
Perceptual-cognitive development. Neurotoxicol Teratol 12:543–546.
Greenough WT, Black JE. 1992. Induction of brain structure by experience: Substrates for cognitive development.
In: Gunnar MR, Nelson CA, eds. Developmental Behavioral Neuroscience. The Minnesota Symposia on
Child Psychology. Vol. 24. Hillsdale, NJ: Lawrence Erlbaum Associates. Pp. 155–200.
Gunnar MR. 2000. Early adversity and the development of stress reactivity and regulation. In: Nelson CA, ed. The
Effects of Early Adversity on Neurobehavioral Development. The Minnesota Symposia on Child Psychology.
Vol. 31. Mahwah, NJ: Lawrence Erlbaum Associates, Publishers. Pp. 163–200.
Gunnar MR, White BP. 2001. Salivary cortisol measures in infant and child assessment. In: Singer LT, Zeskind PS,
eds. Biobehavioral Assessment of the Infant. New York: Guilford Press. Pp. 167–189.
Guo SM, Roche AF, Fomon SJ, Nelson SE, Chumlea WC, Rogers RR, Baumgartner RN, Ziegler EE, Siervogel
RM. 1991. Reference data on gains in weight and length during the first two years of life. J Pediatr 119:355–
362.
Hadorn B, Zoppi G, Shmerling DH, Prader A, McIntyre I, Anderson CM. 1968. Quantitative assessment of
exocrine pancreatic function in infants and children. J Pediatr 73:39–50.
Hamilton J. 2000. The pediatric patient: Early development and the digestive system. In: Walker WA, Durie PR,
Hamilton JR, Walker-Smith JA, Watkins JB, eds. Pediatric Gastrointestinal Disease. 3rd ed. Hamilton, Ont:
BC Decker. Pp. 1–11.
Handin RI. 1998. Blood platelets and the vessel wall. In: Nathan DG, Orkin SH, eds. Nathan and Oski’s Hematol-
ogy of Infancy and Childhood. Philadelphia: WB Saunders. Pp 1511–1529.
Harris SR, Brady DK. 1986. Infant neuromotor assessment instruments: A review. Phys Occup Ther Pediatr
6:121–153.
Harris SR, Swanson MW, Andrews MS, Sells CJ, Robinson NM, Bennett FC, Chandler LS. 1984. Predictive
validity of the “movement assessment of infants.” J Dev Behav Pediatr 5:336–342.
Hediger ML, Overpeck MD, Ruan WJ, Troendle JF. 2000. Early infant feeding and growth status of US-born
infants and children aged 4–71 mo: Analyses from the Third National Health and Nutrition Examination
Survey, 1988–1994. Am J Clin Nutr 72:159–167.
Herskowitz J, Rosman NP. 1985. Pseudoseizure in a child with epilepsy. Am J Psychiatry 142:390–391.
Heyman M, Grasset E, Ducroc R, Desjeux JF. 1988. Antigen absorption by the jejunal epithelium of children with
cow’s milk allergy. Pediatr Res 24:197–202.

Higley JD, Suomi SJ. 1989. Temperamental reactivity in non-human primates. In: Kohnstamm GA, Bates JE,
Rothbart MK, eds. Temperament in Childhood. Chichester, Eng: John Wiley & Sons. Pp. 153–167.
Hill RM, Hegemier S, Tennyson LM. 1989. The fetal alcohol syndrome: A multihandicapped child. Neuro-
toxicology 10:585–596.
Holm S, Andersson Y, Gothefors L, Lindbert T. 1992. Increased protein absorption after gastroenteritis in chil-
dren. Acta Paediatr 81:585–588.
Holt PG, Macaubas C, Stumbles PA, Sly PD. 1999. The role of allergy in the development of asthma. Nature
402:B12–B17.
Hopkins WD, Rilling JK. 2000. A comparative MRI study of the relationship between neuroanatomical asymme-
try and interhemispheric connectivity in primates: Implication for the evolution of functional asymmetries.
Behav Neurosci 4:739–748.
Hurt H, Brodsky NL, Betancourt L, Braitman LE, Malmud E, Giannetta J. 1995. Cocaine-exposed children:
Follow-up through 30 months. J Dev Behav Pediatr 16:29–35.
Huynh H, Couper R. 2000. Pancreatic function tests. In: Walker WA, Durie PR, Hamilton JR, Walker-Smith JA,
Watkins JB, eds. Pediatric Gastrointestinal Disease. 3rd ed. Hamilton, Ont: BC Decker. Pp. 1515–1528.
Hypponen E, Virtanen SM, Kenward MG, Knip M, Akerblom HK. 2000. Obesity, increased linear growth, and
risk of type 1 diabetes in children. Diabetes Care 23:1755–1760.
IOM (Institute of Medicine). 1991. Nutrition During Lactation. Washington, DC: National Academy Press.
IOM. 1997. Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride. Washing-
ton, DC: National Academy Press.
IOM. 2001. Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine,
Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc. Washington, DC: National Academy
Press.
Jacobs DS, Blakemore C. 1988. Factors limiting the postnatal development of visual acuity in the monkey. Vision
Res 28:947–958.
Jacobson JL, Jacobson SW. 1996. Prospective, longitudinal assessment of developmental neurotoxicity. Environ
Health Perspect 104:275–283.
Jacobson SW. 1998. Specificity of neurobehavioral outcomes associated with prenatal alcohol exposure. Alcohol
Clin Exp Res 22:313–320.
Jacobson SW, Jacobson JL. 2000. Teratogenic insult and neurobehavioral function in infancy and childhood. In:
Nelson CA, ed. The Minnesota Symposia on Child Psychology. Vol 31: The Effects of Early Adversity on
Neurobehavioral Development. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 61–112.
Jacobson SW, Jacobson JL, Sokol RJ, Martier SS, Chiodo LM. 1996. New evidence for neurobehavioral effects of
in utero cocaine exposure. J Pediatr 129:581–590.
Jakobsson I, Lindberg T, Lothe L, Axelsson I, Benediktsson B. 1986. Human alpha-lactalbumin as a marker of
macromolecular absorption. Gut 27:1029–1034.
Jensen C, Prager T, Fraley J, Chen H, Anderson R, Heird W. 1997. Effects of dietary linoleic/alpha-linolenic acid
ratio or growth and visual function of term infants. J Pediatr 131:200–209.
Johnson MH. 2001. Functional brain development in humans. Nat Rev Neurosci 2:475–483.
Jusczyk PW. 1985. The high-amplitude sucking technique as a methodological tool in speech perception research.
In: Gottlieb G, Krasnegor NA, eds. Measurement of Audition and Vision in the First Year of Postnatal Life: A
Methodological Overview. Norwood, NJ: Ablex Publishing. Pp. 195–222.
Juvonen P, Jakobsson I, Lindberg T. 1990. Macromolecular absorption and cows milk allergy. Arch Dis Child
65:300–303.
Kaltenbach KA, Finnegan LP. 1989. Prenatal narcotic exposure: Perinatal and developmental effects. Neuro-
toxicology 10:597–604.
Kaneko WM, Phillips EL, Riley EP, Ehlers CL. 1996. EEG findings in Fetal Alcohol Syndrome and Down Syn-
drome children. Electroencephalogr Clin Neurophysiol 98:20–28.
Karpen SJ, Suchy FJ. 2001. Structural and functional development of the liver. In: Suchy FJ, Sokol RJ, Balistreri
WF, eds. Liver Disease in Children. Philadelphia: Lippincott Williams & Wilkins. Pp. 3–21.
Kay AB. 2001. Allergy and allergic diseases. First of two parts. N Engl J Med 344:30–37.
Kindermann TA. 1993. Fostering independence in mother-child interactions: Longitudinal changes in contingency
patterns as children grow competent in developmental tasks. Int J Behav Dev 16:513–535.
Kohnstamm GA, Bates JE, Rothbart MK, eds. 1989. Temperament in Childhood. Chichester, Eng: John Wiley &
Sons.
Kok EJ, Kuiper HA. 2003. Comparative safety assessment for biotech crops. Trends Biotechnol 21:439–444.
Koo WW. 2000. Body composition measurements during infancy. Ann N Y Acad Sci 904:383–392.


Kramer MS, Guo T, Platt RW, Shapiro S, Collet JP, Chalmers B, Hodnett E, Sevkovskaya Z, Dzikovich I,
Vanilovich I. 2002. Breastfeeding and infant growth: Biology or bias? Pediatrics 110:343–347.
Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, Flegal KM, Guo SS, Wei R, Mei Z, Curtin LR, Roche AF,
Johnson CL. 2000. CDC growth charts: United States. Adv Data 314:1–27.
Laughlin NK, Bushnell PJ, Bowman RE. 1991. Lead exposure and diet: Differential effects on social development
in the rhesus monkey. Neurotoxicol Teratol 13:429–440.
Lawson KR, Ruff HA. 2001. Focused attention: Assessing a fundamental cognitive process in infancy. In: Singer
LT, Zeskind PS, eds. Biobehavioral Assessment of the Infant. New York: Guilford Press. Pp. 3–17.
Leeson CP, Kattenhorn M, Deanfield JE, Lucas A. 2001. Duration of breast feeding and arterial distensibility in
early adult life: Population based study. Br Med J 322:643–647.
Lester BM, Tronick EZ. 2001. Behavioral assessment scales. The NICU Network Neurobehavioral Scale, the
Neonatal Behavioral Assessment Scale, and the Assessment of the Preterm Infant’s Behavior. In: Singer LT,
Zeskind PS, eds. Biobehavioral Assessment of the Infant. New York: Guilford Press. Pp. 363–380.
Lester BM, Freier K, LaGasse L. 1995. Prenatal cocaine exposure and child outcome: What do we really know? In:
Lewis M, Bendersky M, eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale,
NJ: Lawrence Erlbaum Associates. Pp. 19–39.
Lindberg T. 1966. Intestinal dipeptidases: Characterization, development and distribution of intestinal dipepti-
dases of the human foetus. Clin Sci 30:505–515.
Lindsay RS, Cook V, Hanson RL, Salbe AD, Tataranni A, Knowler WC. 2002. Early excess weight gain of
children in the Pima Indian population. Pediatrics 109:E33.
Lohman TG, Roche AF, Martorell R. 1988. Anthropometric Standardization Reference Manual. Champaign, IL:
Human Kinetics Books.
Lothe L, Ivarsson SA, Ekman R, Lindberg T. 1990. Motilin and infantile colic. A prospective study. Acta Paediatr
Scand 7:410–416.
Lozoff B, Klein NK, Nelson EC, McClish DK, Manuel M, Chacon ME. 1998. Behavior of infants with iron-
deficiency anemia. Child Dev 69:24–36.
Maffei HV, Metz G, Bampoe V, Shiner M, Herman S, Brook CG. 1977. Lactose intolerance, detected by the
hydrogen breath test, in infants and children with chronic diarrhoea. Arch Dis Child 52:766–767.
Majamaa H, Miettinen A, Laine S, Isolauri E. 1996. Intestinal inflammation in children with atopic eczema: Faecal
eosinophil cationic protein and tumour necrosis factor-alpha as non-invasive indicators of food allergy. Clin
Exp Allergy 26:181–187.
Marshall DH, Fox NA. 2001. Electroencephalographic assessment and human brain maturation. A window into
emotional and cognitive development in infancy. In: Singer LT, Zeskind PS, eds. Biobehavioral Assessment of
the Infant. New York: Guilford Press. Pp. 341–356.
Martin RP, Wisenbaker J, Huttunen M. 1994. Review of factor analytic studies of temperament measures based
on the Thomas-Chess structural model: Implications for the big five. In: Halverson CF Jr, Kohnstamm GA,
eds. The Developing Structure of Temperament and Personality from Infancy to Adulthood. Hillsdale, NJ:
Masten AS, Coatsworth JD. 1998. The development of competence in favorable and unfavorable environments.
Lessons from research on successful children. Am Psychol 53:205–220.
Matheny AP Jr. 1989. Temperament and cognition: Relations between temperament and mental test scores. In:
Kohnstamm GA, Bates JE, Rothbart MK, eds. Temperament in Childhood. Chichester, Eng: John Wiley &
Sons. Pp. 263–282.
Matheny AP Jr. 1991. Play assessment of infant temperament. In: Schaefer CE, Gitlin K, Sandgrund A, eds. Play
Diagnosis and Assessment. New York: John Wiley & Sons. Pp. 39–63.
Matheny AP Jr, Phillips K. 2001. Temperament and context: Correlates of home environment with temperament
continuity and change, new born to 30 months. In: Wachs TD, Kohnstamm GA, eds. Temperament in
Context. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 81–101.
Mattson SM, Riley EP. 1995. Prenatal exposure to alcohol. What the images reveal. Alcohol Health Res World
19:273–278.
Mayer DL, Arendt RE. 2001. Visual acuity assessment in infancy. In: Singer LT, Zeskind PS, eds. Biobehavioral
Assessment of the Infant. New York: Guilford Press. Pp. 81–94.
Mayes LC, Bornstein MH. 1995. Developmental dilemmas for cocaine-abusing parents and their children. In:
Lewis M, Bendersky M, eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale,
Mayes LC, Cicchetti DV. 1995. Prenatal cocaine exposure and neurobehavioral development: How subjects lost to
follow-up bias study results. Child Neuropsychol 1:128–139.

Mayes LC, Bornstein MH, Chawarska K, Granger RH. 1995. Information processing and developmental assess-
ments in 3-month-old infants exposed prenatally to cocaine. Pediatrics 95:539–545.
McCall RB. 1994. What process mediates predictions of childhood IQ from infant habituation and recognition
memory? Speculations on the roles of inhibition and rate of information processing. Intelligence 18:107–125.
McCall RB, Appelbaum MI. 1991. Some issues of conducting secondary analyses. Dev Psychol 27:911–917.
McCall RB, Carriger MS. 1993. A meta-analysis of infant habituation and recognition memory performance as
predictors of later IQ. Child Dev 64:57–79.
McMurray RG, Hackney AC. 2000. Endocrine responses to exercise and training. In: Garrett WE, Krikendall DT,
eds. Exercise and Sport Science. Philadelphia: Lippincott Williams & Wilkins. Pp. 135–161.
Milani-Comparetti A, Gidoni EA. 1967. Routine developmental examination in normal and retarded children.
Dev Med Child Neurol 9:631–638.
Miller LJ, Roid GH. 1994. The T.I.M.E. Toddler and Infant Motor Evaluation. A Standardized Assessment. San
Antonio: Therapy Skill Builders.
Molfese DL, Molfese VJ. 2001. Cortical electrophysiology and language processes in infancy. In: Singer LT,
Zeskind PS, eds. Biobehavioral Assessment of the Infant. New York: Guilford Press. Pp. 323–338.
Montgomery RK, Mulberg AE, Grand RJ. 1999. Development of the human gastrointestinal tract: Twenty years
of progress. Gastroenterology 116:702–731.
Moore JM, Wilson WR, Thompson G. 1977. Visual reinforcement of head-turn responses in infants under 12
months of age. J Speech Hear Disord 42:328–334.
Nakahara K, Hayashi T, Konishi S, Miyashita Y. 2002. Functional MRI of macaque monkeys performing a
cognitive set-shifting task. Science 295:1532–1536.
Needlman R, Frank DA, Augustyn M, Zuckerman BS. 1995. Neurophysiological effects of prenatal cocaine expo-
sure: Comparison of human and animal investigations. In: Lewis M, Bendersky M, eds. Mothers, Babies, and
Cocaine: The Role of Toxins in Development. Hillsdale, NJ: Lawrence Erlbaum Associates. Pp. 229–250.
Nelson CA. 1994. Neural bases of infant temperament. In: Bates JE, Wachs TD, eds. Temperament: Individual
Differences at the Interface of Biology and Behavior. Washington, DC: American Psychological Association.
Pp. 47–82.
Nelson CA. 1995. The ontogeny of human memory: A cognitive neuroscience perspective. Dev Psychol 31:723–
738.
Nelson CA, Bloom FE. 1997. Child development and neuroscience. Child Dev 5:970–987.
Nelson CA, Wewerka S, Thomas KM, Tribby-Walbridge S, deRegnier R, Georgieff M. 2000. Neurocognitive
sequelae of infants of diabetic mothers. Behav Neurosci 114:950–956.
Nelson CA, Bloom FE, Cameron JL, Amaral D, Dahl RE, Pine D. 2002. An integrative, multidisciplinary approach
to the study of brain-behavior relations in the context of typical and atypical development. Dev Psychopathol
14:499–520.
Nelson SE, Rogers RR, Ziegler EE, Fomon SJ. 1989. Gain in weight and length during early infancy. Early Hum
Dev 19:223–239.
Neuspiel DR. 1995. The problem of confounding in research on prenatal cocaine effects on behavior and develop-
ment. In: Lewis M, Bendersky M, eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development.
Hillsdale, NJ: Lawrence Erlbaum Associates. Pp. 95–109.
Norman A, Strandvik B, Ojamae O. 1972. Bile acids and pancreatic enzymes during absorption in the newborn.
Acta Paediat Scand 61:571–576.
O’Connor MJ, Sigman M, Kasari C. 1993. Interactional model for the association among maternal alcohol use,
mother-infant interaction, and infant cognitive development. Infant Behav Dev 16:177–192.
Olsho LW, Koch EG, Halpin CF, Carter EA. 1987. An observer-based psychoacoustic procedure for use with
young infants. Dev Psychol 23:627–640.
Ong KK, Ahmed ML, Emmett PM, Preece MA, Dunger DB. 2000. Association between postnatal catch-up growth
and obesity in childhood: Prospective cohort study. Br Med J 320:967–971.
Pelletier DL, Frongillo EA Jr, Habicht JP. 1993. Epidemiologic evidence for a potentiating effect of malnutrition
on child mortality. Am J Public Health 83:1130–1133.
Perman JA, Barr RG, Watkins JB. 1978. Sucrose malabsorption in children: Noninvasive diagnosis by interval
breath hydrogen determination. J Pediatr 93:17–22.


Pinero D, Jones B, Beard J. 2001. Variations in dietary iron alter behavior in developing rats. J Nutr 131:311–318.
Piper MC, Darrah J. 1994. Alberta Infant Motor Scale: Construction of a Motor Assessment Tool for the Develop-
ing Infant. Philadelphia: WB Saunders Company. Pp. 25–35.
Plancoulaine S, Charles MA, Lafay L, Tauber M, Thibult N, Borys JM, Eschwege E, Laventie Ville Sante Study
Group. 2000. Infant-feeding patterns are related to blood cholesterol concentration in prepubertal children
aged 5–11 y: The Fleurbaix-Laventie Ville Santé Study. Eur J Clin Nutr 54:114–119.
Porges SW, Doussard-Roosevelt JA, Portales AL, Greenspan SI. 1996. Infant regulation of the vagal ‘brake’
predicts child behavior problems: A psychobiological model of social behavior. Devel Psychobiol 29:697–
712.
Porter FL. 2001. Vagal tone. In: Singer LT, Zeskind PS, eds. Biobehavioral Assessment of the Infant. New York:
Guilford Press. Pp. 109–124.
Posner MI. 2001. The developing human brain. Dev Sci 4:253–387.
Rao R, Georgieff MK. 2000. Early nutrition and brain development. In: Nelson C, ed. The Minnesota Symposia
on Child Psychology. The Effects of Adversity on Neurobehavioral Development. Vol 31. Mahway, NJ:
Robb TA, Davidson GP. 1981. Advances in breath hydrogen quantitation in paediatrics: Sample collection and
normalization to constant oxygen and nitrogen levels. Clin Chim Acta 111:281–285.
Roberton DM, Paganelli R, Dinwiddie R, Levinsky RJ. 1982. Milk antigen absorption in the preterm and term
neonate. Arch Dis Child 57:369–372.
Robson AL, Pederson DR. 1997. Predictors of individual differences in attention among low birth weight children.
J Dev Behav Pediatr 18:13–21.
Roncagliolo M, Garrido M, Walter T, Peirano P, Lozoff B. 1998. Evidence of altered central nervous system
development in infants with iron deficiency anemia at 6 mo: Delayed maturation of auditory brainstem
responses. Am J Clin Nutr 68:683–690.
Rose SA, Feldman JF. 1990. Infant cognition: Individual differences and developmental continuities. In: Colombo
J, Fagen J, eds. Individual Differences in Infancy: Reliability, Stability, Prediction. Hillsdale, NJ: Lawrence
Rose SA, Feldman JF. 1995. Prediction of IQ and specific cognitive abilities at 11 years from infancy measures.
Dev Psychol 31:685–696.
Rose SA, Orlian EK. 2001. Visual information processing. In: Singer LT, Zeskind PS, eds. Biobehavioral Assess-
ment of the Infant. New York: Guilford Press. Pp. 274–292.
Rose SA, Feldman JF, Wallace IF, McCarton C. 1989. Infant visual attention: Relation to birth status and
developmental outcome during the first 5 years. Dev Psychol 25:560–576.
Rose SA, Feldman JF, Wallace IF, McCarton C. 1991. Information processing at 1 year: Relation to birth status
and developmental outcome during the first 5 years. Dev Psychol 5:723–737.
Rossetti Ferreira MC. 1978. Malnutrition and mother-infant asynchrony: Slow mental development. Int J Behav
Dev 1:207–219.
Rothbart MK. 1986. Longitudinal observation of infant temperament. Dev Psychol 22:356–365.
Rothbart MK, Bates JE. 1998. Temperament. In: Eisenberg N, ed. Handbook of Child Psychology. 5th ed, vol 3.
New York: John Wiley & Sons. Pp. 105–176.
Rothbart MK, Derryberry D, Posner MI. 1994. A psychobiological approach to the development of temperament.
In: Bates JE, Wachs TD, eds. Temperament: Individual Differences at the Interface of Biology and Behavior.
Washington, DC: American Psychological Association. Pp. 83–116.
Rothbart MK, Derryberry D, Hershey K. 2000. Stability of temperament in childhood: Laboratory infant assess-
ment to parent report at seven years. In: Molfese VJ, Molfese DL, eds. Temperament and Personality Devel-
opment Across the Life Span. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 85–118.
Rovee-Collier C, Barr R. 2001. Infant learning and memory. In: Bremner G, Fogel A, eds. Blackwell Handbook of
Infant Development. Malden, MA: Blackwell Publishers. Pp. 139–168.
Ruff HA, Rothbart MK. 1996. Attention in Early Development. Themes and Variations. New York: Oxford
University Press.
Russell RW. 1992. “Malnutrition” and the vulnerable brain. What a difference a molecule makes. In: Isaacson RL,
Jensen KF, eds. The Vulnerable Brain and Environmental Risks. Vol 1. Malnutrition and Hazard Assessment.
New York: Plenum Press. Pp. 109–126.
Sampson HA. 1996. Epidemiology of food allergy. Pediatr Allergy Immunol 7:42–50.
Sampson HA. 2000. Peanut allergy. N Engl J Med 346:1294–1299.
Sanderson IR, Walker WA. 1993. Uptake and transport of macromolecules by the intestine: Possible role in
clinical disorders (an update). Gastroenterology 104:622–639.

Sanson A, Pedlow R, Cann W, Prior M, Oberklaid F. 1996. Shyness ratings: Stability and correlates in early
childhood. Int J Behav Dev 19:705–724.
Schmelzle HR, Fusch C. 2002. Body fat neonates and young infants: Validation of skinfold thickness versus dual-
energy x-ray absorptimetry. Am J Clin Nutr 76:1096–1100.
Schneider JW, Griffith DR, Chasnoff IJ. 1989. Infants exposed to cocaine in utero: Implications for developmental
assessment and intervention. Infants Young Child 2:25–36.
Schore AN. 1994. Affect Regulation and the Origin of the Self. The Neurobiology of Emotional Development.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Schrander JJP, Unsalan-Hooyen RWM, Forget PP, Jansen J. 1990. 51Cr EDTA intestinal permeability in children
with cow’s milk intolerance. J Pediatr Gastroenterol Nutr 10:189–192.
Scott DT, Janowsky JS, Carroll RE, Taylor JA, Auestad N, Montalto MB. 1998. Formula supplementation with
long-chain polyunsaturated fatty acids: Are there developmental benefits? Pediatrics 102:1203–1204.
Scott RB. 2000. Motility disorders. In: Walker WA, Durie PR, Hamilton JR, Walker-Smith JA, Watkins JB, eds.
Pediatric Gastrointestinal Disease. 3rd ed. Hamilton, Ont: BC Decker. Pp. 103–115.
Sebris SL, Dobson V, Hartmann EE. 1984. Assessment and prediction of visual acuity in 3- to 4-year-old children
born prior to term. Hum Neurobiol 3:87–92.
Sereno MI. 1998. Brain mapping in animals and humans. Curr Opin Neurobiol 8:188–194.
Setchell KDR, O’Connell NC. 2001. Disorders of bile acid synthesis and metabolism: A metabolic basis for liver
disease. In: Suchy FJ, Sokol RJ, Balistreri WF, eds. Liver Disease in Children. Philadelphia: Lippincott
Williams & Wilkins. Pp. 701–733.
Shaheen SJ. 1984. Neuromaturation and behavior development: The case of childhood lead poisoning. Develop
Psychol 20:542–550.
Shrout PE, Bolger N. 2002. Mediation in experimental and nonexperimental studies. New procedures and recom-
mendations. Psychol Methods 7:422–445.
Sicherer SH, Noone SA, Koerner CB, Christie L, Burks, AW, Sampson HA. 2001. Hypoallergenicity and efficacy
of an amino acid based formula in children with cow’s milk and multiple food hypersensitivities. J Pediatr
138:688–693.
Singer LT. 2001. General issues in infant assessment and development. In: Singer LT, Zeskind PS, eds. Bio-
behavioral Assessment of the Infant. New York: Guilford Press. Pp. 3–17.
Singer L, Arendt R, Farkas K, Minnes S, Huang J, Yamashita T. 1997. Relationship of prenatal cocaine exposure
and maternal postpartum psychological distress to child developmental outcome. Dev Psychopathol 9:473–
489.
Slater A. 1997. Can measures of infant habituation predict later intellectual ability? Arch Dis Child 77:474–476.
Slater A. 2001. Visual perception. In: Bremner G, Fogel A, eds. Blackwell Handbook of Infant Development.
Malden, MA: Blackwell Publishers. Pp. 5–34.
Slota MC. 1983. Neurological assessment of the infant and toddler. Crit Care Nurse 3:87–92.
Smitsman MW. 2001. Action in infancy—Perspectives, concepts, and challenges: The development of reaching and
grasping. In: Bremner G, Fogel A, eds. Blackwell Handbook of Infant Development. Malden, MA: Blackwell
Publishers. Pp. 71–98.
Sobotka TJ, Ekelman KB, Slikker W Jr, Raffaele K, Hattan DG. 1996. Food and Drug Administration proposed
guidelines for neurotoxicological testing of food chemicals. Neurotoxicology 17:825–836.
Sparrow SS, Balla DA, Cicchetti DV. 1984. Vineland Adaptive Behavioral Scales. Circle Pines, MN: American
Guidance Service.
Spear LP. 1995. Alterations in cognitive function following prenatal cocaine exposure: Studies in an animal model.
In: Lewis M, Bendersky M, eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale,
Sperling MA. 1996. Pediatric Endocrinology. Philadelphia: WB Saunders. Pp. 549–584.
Stettler N, Bovert P, Shamlaye H, Zemel BS, Stallings VA, Paccaud F. 2002a. Prevalence and risk factors for
overweight and obesity in children from Seycheles, a country in rapid transition: The importance of early
growth. Int J Obes Relat Metab Disord 26:214–219.
Stettler N, Zemel BS, Kumanyika S, Stallings VA. 2002b. Infant weight gain and childhood overweight status in a
multicenter, cohort study. Pediatrics 109:194–199.
Struthers JM, Hansen RL. 1992. Visual recognition memory in drug-exposed infants. J Dev Behav Pediatr 13:108–
111.
Takeda A. 2001. Zinc homeostasis and functions of zinc in the brain. Biometals 14:343–351.
Tanner JM. 1970. Physical growth. In: Mussen PH, ed. Carmichael’s Manual of Child Psychology. 3rd ed, vol 1.
New York: John Wiley & Sons. Pp. 77–155.


Teller DY, McDonald MA, Preston K, Sebris SL, Dobson V. 1986. Assessment of visual acuity in infants and
children: The acuity card procedure. Dev Med Child Neurol 28:779–789.
Thomas DW, Sinatra FR, Merritt RJ. 1981. Random fecal alpha-1-antitrypsin concentration in children with
gastrointestinal disease. Gastroenterology 80:776–782.
van de Kamer JH, ten Bokkel Huinink H, Weyers HA. 1949. Rapid method for the determination of fat in feces. J
Biol Chem 177:347–355.
Vijayakumar M, Fall CH, Osmond C, Barker DJ. 1995. Birth weight, weight at one year, and left ventricular mass
in adult life. Br Heart J 73:363–367.
von Hofsten C. 1991. Structuring of early reaching movements: A longitudinal study. J Mot Behav 23:280–292.
von Hofsten C, Fazel-Zandy S. 1984. Development of visually guided hand orientation in reaching. J Exp Child
Psychol 38:208–219.
Vorhees CV. 1986. Principles of behavioral teratology. In: Riley EP, Vorhees CV, eds. Handbook of Behavioral
Teratology. New York: Plenum Press. Pp. 23–48.
Wachs TD. 1999. The nature and nurture of child development. Food Nutr Bull 20:7–22.
Wachs TD. 2000. Linking nutrition and temperament. In: Molfese VJ, Molfese DL, eds. Temperament and Per-
sonality Development Across the Life Span. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 57–84.
Wachs TD, Bates JE. 2001. Temperament. In: Bremner G, Fogel A, eds. Blackwell Handbook of Infant Develop-
ment. Malden, MA: Blackwell Publishers. Pp. 465–501.
Walker WA. 2002. Development of the intestinal mucosal barrier. J Pediatr Gastroenterol Nutr 34:S33–S39.
Watkins JB. 1985. Lipid digestion and absorption. Pediatrics 75:151–156.
Watkins JB, Ingall D, Szczepanik P, Klein PD, Lester R. 1973. Bile-salt metabolism in the newborn. Measurement
of pool size and synthesis by stable isotope technique. New Engl J Med 288:431–434.
Weiss B. 1995. Incipient hazards of cocaine: Lessons from environmental toxicology. In: Lewis M, Bendersky M,
eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale, NJ: Lawrence Erlbaum
Associates. Pp. 41–55.
Whitaker RC, Wright JA, Pepe MS, Seidel KD, Dietz WH. 1997. Predicting obesity in young adulthood from
childhood and parental obesity. N Engl J Med 337:869–873.
WHO (World Health Organization). 2002. The Optimal Duration of Exclusive Breastfeeding. Report of an
Expert Consultation. Geneva: WHO.
Wroble M, Mash C, Williams L, McCall R. 2002. Should long chain polyunsaturated fatty acids be added to
infant formula to promote development? J App Dev Psychol 23:99–112.
Yajnik CS. 2001. The insulin resistance epidemic in India: Fetal origins, later lifestyle, or both? Nutr Rev 59:1–9.

Selecting an In-Market Surveillance Plan
ABSTRACT
In-market surveillance is an essential element of regulating the safety of ingredi-
ents new to infant formulas and should be included in new safety assessment guide-
lines. There are two components of in-market surveillance. The first, the “monitor-
ing component,” involves procedures to detect adverse effects upon infants after a
formula has been put on the market. The second, the “follow-up component,” con-
centrates on possible long-term adverse effects after the period of maximum formula
usage. Although formal regulatory guidelines for in-market surveillance do not exist
in the United States or Canada, infant formula manufacturers routinely conduct
passive surveillance via toll-free calls, contact with health care professionals, and
reports from their field sales force.
Satisfactory completion of the appropriate preclinical and clinical studies dimin-
ishes the likelihood of systematic adverse reactions; however the risks for adverse
reactions cannot be ignored because adverse effects may not be detected in preclini-
cal studies if the wrong animal model was chosen, if the assessment instrument cho-
sen measured a function other than the one adversely affected by the new ingredient,
or if a subpopulation of individuals who are highly sensitive to the new ingredient
added to infant formula was not sufficiently represented in clinical studies.
The committee believes that there is a crucial need for follow-up strategies to
ensure safety and normal development of the infant population (e.g., brain areas that
are adversely affected by new ingredients may not functionally become apparent
until later in development, or early exposure to a toxin may increase susceptibility to
later exposure to toxins). The committee recommends that a systematic plan for
continued in-market monitoring and long-term surveillance become an essential part
of all submissions for regulatory agency review seeking to add a new ingredient to
infant formula. The in-market surveillance strategies the committee proposes are
specifically designed for evaluation of the safety of new ingredients added to infant
160

SELECTING AN IN-MARKET SURVEILLANCE PLAN 161
formulas and do not necessarily apply to the safety of new ingredients not intended
for infant formulas.
Infant formula manufacturers should select in-market strategies by imple-
menting a hierarchy of three levels of assessment, including passive surveillance
(level 1 assessment), expert panel reviews of the literature (level 2 assessment),
and active surveillance (level 3 assessment). Active surveillance is the most com-
plex assessment and includes options such as studying specific populations, con-
ducting retrospective studies, using the Pediatric Research in Office Settings pro-
gram of the American Academy of Pediatrics, and conducting clinical follow-up
studies of the original study populations. Selection of the appropriate level of
assessment is based upon conditions under which potential adverse effects of a
new ingredient added to infant formulas might have been missed in preclinical or
clinical studies. The length of the follow-up period will depend on the targeted
area (organ system), preclinical and clinical studies, and the nature of the added
ingredient.
BACKGROUND
The Importance of Systematic In-Market Surveillance

Satisfactory completion of the appropriate preclinical and clinical studies diminishes the
likelihood of adverse reactions to infant formulas that contain new ingredients once they
have been placed on the market. However completion of those studies does not ensure
absolute safety, and the probability of adverse reactions cannot be overlooked for several
reasons.
First, adverse effects may not be detected in preclinical studies if the wrong animal
model is chosen to map onto a function at the human level (see Chapter 5). Second, in
clinical studies a less-sensitive instrument may fail to detect an adverse reaction due to poor
initial outcome measurement (Clarke and Clarke, 1981; see Chapter 6). Third, adverse
effects could be missed during clinical studies if the instrument chosen, even if highly
sensitive, was measuring a function other than the one adversely affected by the new ingre-
dient. A similar point holds in regard to predictor measures. Analytical procedures have
been developed that are capable of measuring compounds at concentrations as low as
picograms per liter. Concurrent appreciation of potential biological changes has not kept
pace with this degree of laboratory sophistication. Finally, the sampling design of clinical
trials could have been inappropriate. Given what is known about variability in individual or
subpopulation sensitivities to exposure to dietary ingredients (Beaton, 1986; Rutter and
Pickles, 1991) or toxic substances (Bellinger, 1995; Ruff, 1999), if a subpopulation of
individuals who are highly sensitive to the new ingredient added to an infant formula are not
sufficiently represented in the clinical studies, adverse effects may not be detected until the
formula is marketed to a wider population.
These reasons support the need for a systematic plan to include both types of in-market
surveillance into every submission to the regulatory agency for the addition of an ingredient
new to infant formulas. The two types of in-market surveillance are discussed below.
RECOMMENDATION: A systematic plan for continued in-market monitoring and

long-term surveillance should be an essential part of submissions for regulatory agency
review in assessing the safety of ingredients new to infant formulas.

Monitoring
The first component of in-market surveillance, the monitoring component, involves
procedures to detect adverse effects to infants after a formula has been put on the market. As
discussed above, the probability that negative effects will emerge during in-market monitor-
ing is likely to be quite small if the appropriate preclinical and clinical studies detected no
negative effects associated with the introduction of a new ingredient to an existing formula.
However the number of infants enrolled in clinical trials is small in relation to in-market use,
and the trials may not detect a full range of variations. For this reason it is important to
integrate in-market monitoring procedures into the evaluation process to judge the safety of
new ingredients introduced into infant formulas.
Follow-up
The second component of in-market surveillance, the follow-up component, is one
that is less often considered during clinical trials to assess the safety of new ingredients
added to infant formulas. In contrast to in-market monitoring, which focuses on adverse
effects occurring during the period of maximum formula usage, in-market follow-up
concentrates on possible long-term adverse effects after the period of maximum formula
usage. The length of a follow-up study will depend upon the nature of the ingredient, so
the expert panel should define it on a case-by-case basis. As described later in this chapter,
higher levels of assessments should be performed when the ingredient may affect slow-
developing brain regions, hormone or neurotransmitter function, or behavior. In addition,
follow-up during critical life transitions, such as entry into school or onset of puberty,
should be emphasized.
Most clinical studies are confined to a short amount of time for the period of maximum
exposure to the formulas, and they track adverse patterns only during the time of maximum
exposure. However there is also the possibility that the new ingredient’s negative effects on
the growth and development of children may have delayed onset and only appear later in
life. Evidence from a number of sources supports the validity of this statement. First, there
are both preclinical and clinical studies that document long-term cognitive and behavioral
effects of early exposure to toxic substances (Galler and Tonkiss, 1998; Jacobson and
Jacobson, 2000; Leech et al., 1999; Leviton et al., 1993; Richardson, 1998; Richardson et
al., 1996; Romano and Harvey, 1998; Wasserman et al., 2000).
More critically, although the overall pattern of evidence is not totally consistent, there
are a number of examples from both the clinical and preclinical research literature where
negative effects of early exposure to toxic substances were not found upon initial testing, but
did appear during follow-up assessment (Weiss, 1995; Winneke, 1990). Toxic substances
that do not show their effects until well after the period of exposure to the substance have
been labeled as chemical or neurobehavioral “time bombs” (Russell, 1990; Spencer, 1990).
Delayed effects have been shown for prenatal or early postnatal exposure to drugs, alcohol
(Griffith et al., 1994; Singer et al., 2002), and lead (Bellinger et al., 1991). Although not
directly related to toxic exposure, epidemiological studies have also indicated that there may
be delayed long-term adult biomedical consequences as a result of the quality of very early
nutrition (Barker et al., 1993; Jackson, 2000) or of the level of morbidity in the first year of
life (Bengtsson and Lindstrom, 2000).
There are a number of mechanisms through which a delayed impact of early exposure to
toxic substances might occur. One involves cumulative effects. Biologically, cumulating
intake of a low-level toxin can result in the gradual replacement of a specific neurotransmit-

ter by a less efficient molecular substitute; only over time will the detrimental impact of the
substitute become manifest (Russell, 1990). Another example is the long-chain polyunsatu-
rated fatty acids (LC-PUFAs) that have been determined as GRAS for addition to infant
formulas. These LC-PUFAs are derived from genetically selected algae that produce triglyc-
erides with a high content of either arachidonic acid or docosahexaenoic acid onto two or
three of the fatty acid chains of the triacylglycerol molecule. This contrasts to the tri-
acylglycerol makeup of human milk, where LC-PUFAs rarely form two-thirds of the fatty
acids of a single triacylglycerol molecule. In this case, diglycerides resulting from hydrolysis
of the algae triglycerides in infant formula would be different than those produced from
human milk and different biological effects could occur (e.g., specific diglycerides have very
different effects in cell signaling pathways and could have different “triggering” effects on
certain cellular pathways).
Such a cumulative effect scenario is particularly likely to occur when a toxic substance
is contained in formula, which may be the main nutrient source for an infant over an
extended period of time. A cumulative effect process, wherein initial small deficits cumulate
and become significant only over time, appears to be a particularly likely scenario when the
critical mechanisms are biobehavioral in nature (Wachs, 2000). Such a process could occur
when toxic exposure leads to changes in infant temperament characteristics, such as an
increased level of irritability or a reduced level of responsivity. Toxin-driven changes in
temperament could adversely impact upon critical influences on later development, such as
patterns of parent-child relations, which over time can lead to long-term developmental-
behavioral deficits (Bendersky et al., 1996; Chasnoff et al., 1987; Fried, 1989; O’Connor et
al., 1993; Schneider et al., 1999).
Alternatively, the area of the brain that is damaged by a toxic substance may be one in
which the functions mediated by this area become apparent only later in development, or it
could be one in which connections go from a damaged area to a later-developing area (Lyon
and Gadisseux, 1991; Weiss, 1995). In these cases, the consequences of early exposure to
toxic substances could occur only after sufficient time has elapsed for the brain-mediated
function to appear in a normal developmental sequence. For example, early damage to areas
of the brain associated with motor control may not be seen until that time period when
infants would be expected to acquire voluntary movement sequences (Lyon and Gaddisseux,
1991).
Finally, there is evidence from both preclinical (Spear et al., 1998) and clinical studies
(Mayes et al., 1998) that early exposure to toxic substances may increase the organism’s
sensitivity or vulnerability to later biological or psychosocial stressors. A similar argument
for greater susceptibility has also been made in regard to mechanisms underlying the long-
term biomedical consequences of early nutritional deficiencies (Waterland and Garza, 1999).
If increased susceptibility to later stresses occurs, fewer developmental-behavioral conse-
quences of early exposure to toxins would appear during the relatively sheltered years of
infancy, with increased developmental risk when environmental demands increase as the
child gets older. Intraventricular hemorrhage in infancy is one example of increased suscep-
tibility to later stress. The impact of intraventricular hemorrhage on cognitive performance
seems to fade through the preschool years, but reappears when the child enters grade school
(e.g., a high-demand situation) (Sostek, 1992).
The available evidence supports the importance of a program of systematic and contin-
ued monitoring of potential consequences of the addition of new ingredients to infant
formulas and systematic in-market follow-up past the infancy period. Such a program is
especially important for ingredients with putative or potential biological effects.

Limitations of In-Market Surveillance

Although there is ample justification for a systematic program of in-market monitoring
and follow-up surveillance, many factors make such a program difficult to implement, and
there are many questions and issues that need to be considered before setting up such a
program. There are both logistical problems (e.g., tracking enrolled subjects, continuity of
the research team, record retention in follow-up studies, knowing or anticipating what
elements to track) and methodological problems (e.g., length of follow-up, confounding
factors, reconstructing causality in monitoring studies, having sufficient statistical power to
detect what may be subtle, yet clinically significant, adverse effects).
One major issue is the problem of confounding factors. It is obviously difficult, years
later, to differentiate effects related to new ingredients from effects due to later exposure to
other risk factors. Even with longitudinal studies there are significant logistical problems
with associating a single added ingredient in infant formula with any observed outcome.
Small differences in the amount or type of a carbohydrate, protein, or fat source may not
have the implications of the addition of a bioactive substance completely new to infant
formulas.
Throughout life the infant, then the child, and finally the adult, is exposed to a wide
variety of food substances with potentially unknown composition. The recent explosion in
the use of dietary supplements is an example. Study participants can have difficulty with
retrospective recall of prior events, as in the case of whether there was a concurrent use of
breastfeeding, even if partial, along with formula feeding. Lifestyle factors may not be
obvious, and perhaps not admitted, by the participant. Developmental risk factors not
related to infant formulas, such as family stress, also can result in long-term deficits. The
longer the follow-up interval, the greater the influence of these possible confounding factors.
The length of time that investigators should study children who were placed on a
formula containing a new ingredient presents another set of problems. The length of a study
may depend on the nature of the substance added. Formulas and solid infant foods may
change over the years of the study so that different cohorts may ingest different nonstudy
foods, making comparison difficult.
Defining an adverse event may also be difficult in conducting long-term studies (e.g.,
beyond 1 year). Investigators must define the duration (permanent or not) and nature of an
adverse event. As noted in Chapter 6, there also is a question of what functional domains
would need to be assessed when conducting in-market surveillance. Ideally the type of
compound added to the formula would dictate what form of follow-up would be most
important. However sufficient data may not be available to precisely specify associations
between a compound and an affected function.
Finally, as outlined in Chapter 2, sample sizes must be large enough to ensure sufficient
statistical power in follow-up studies. As noted in Chapter 6, analysis indicated that the
majority of studies investigating the cognitive effects of the addition of LC-PUFAs to infant
formulas had insufficient statistical power. The problem of sufficient statistical power to
detect adverse consequences is particularly critical when follow-up studies are conducted
years after the child has ceased consumption of infant formula. In this situation adverse
consequences linked to prior consumption of infant formula may be indirect rather than
direct, and thus more subtle, so larger sample sizes will be needed to ensure that there is
sufficient statistical power. For example, using a statistical power level of 0.80 as the goal
and comparing two types of formula with new ingredients versus a control group (standard
formula or breastfed), a sample size of 52 children per group will be required to detect a
moderate-effect size difference. In contrast, under the same conditions, a sample size of 140

children per group will be required to detect an intermediate-effect size (i.e., between small
and moderate). Unless there are compelling reasons to do otherwise, the committee recom-
mends having sufficient power to detect differences between groups of 0.20 standard devia-
tions or less when estimating sample-size needs in follow-up studies. As is noted later in this
chapter, at a population level, even effect sizes of this magnitude can have important clinical
implications.
The cost to ensure sufficient power and demonstrate statistical significance of long-term
follow-up studies may be a factor to consider when designing such studies. It is beyond the
scope of the committee’s charge to set cost limits or recommendations.
Current Regulatory Guidelines

As discussed in Chapter 4, formal regulatory guidelines for in-market surveillance do
not exist for infant formulas. Section 412 of the Federal Food, Drug and Cosmetic Act states
that manufacturers must retain “. . . all complaints and the maintenance of files with respect
to, and the review of, complaints concerning infant formulas which may reveal the possible
existence of a hazard to health.”
As a matter of current practice, infant formula manufacturers routinely conduct passive
surveillance via toll-free calls, contact with health care professionals, and reports from their
field sales force. Since infants are unable to verbally communicate, any adverse effects must
be observed and reported through the parents or caregiver, thus special attention must be
paid to detect adverse or unusual reactions when feeding infant formulas containing new
ingredients.
Similarly, in Canada there are no explicit guidelines or requirements for in-market
surveillance of infant formulas specified under Canada’s Food and Drug Regulations in
Divisions 16 (Food Additives), 25 (Infant Formula), or 28 (Novel Foods).
RECOMMENDATION: Determine in-market surveillance strategies by implementing

a hierarchy of three levels of assessment:
• Level 1 assessment: passive surveillance. Toll-free or Internet reporting to for-

mula manufacturers or to the regulatory agency. For example, in August 2002 the U.S.
Food and Drug Administration (FDA) established a passive surveillance system for food,
cosmetics, and dietary supplements (OSAS, 2002). This system, the Adverse Events
Reporting System, will be part of the MedWatch program. Formula manufacturers
could add this Internet web site’s address to the labeling on each can of formula next to
the toll-free reporting number.
• Level 2 assessment: panel review. In-market panels to review existing data (both
published and proprietary). Issues with regard to the selection and composition of such
panels have been discussed extensively in Chapter 4. The same selection and composi-
tion recommendations presented earlier also hold with regard to in-market panels.
• Level 3 assessment: active surveillance. The following are examples of three
different active surveillance options:
A. Use of follow-up surveillance populations where there is known use of the
formula containing the new ingredient or ingredient source (e.g., by using databases

from the Special Supplemental Nutrition Program for Women, Infants and Children
or health maintenance organizations).
B. Use of existing pediatric networks as a data source (e.g., Pediatric Research
in Office Settings Program of the American Academy of Pediatrics).
C. Clinical follow-up of the original study populations. Such studies would be
part of the original study design—not an add-on study.
Based upon testimony to the committee, companies that market infant formulas often
rely on level 1 assessment for in-market surveillance. However there are inherent flaws in
this assessment level. One such flaw is the risk of underestimating actual negative occur-
rences given that not all caregivers whose infants experience problems will call in a report.
Underestimation is an even greater problem for long-term follow-up since caregivers are not
likely to link a child’s current problems to intake of infant formula years ago, even when
such a linkage may occur. Because the committee does not believe that passive surveillance
(level 1 assessment) is sufficient for monitoring all ingredients that might be added to infant
formulas, a hierarchy of the options was developed.
Level 1, 2, and 3A assessments would seem most appropriate for in-market monitoring
studies, whereas level 2 and 3 assessments would be appropriate for in-market follow-up
studies. These levels should not be considered as either-or alternatives. In some cases results
from one level could lead to application of a higher level of assessment, as in the case when
potential negative effects emerging from in-market panels would suggest that more data be
gathered using methods from the higher levels. Figure 7-1 provides an overview of the
decision-making process.
In addition to surveillance, a level 2 review of all pertinent data, both published and
unpublished, available since the submission to the regulatory agency should be conducted
approximately 2 to 4 years after introduction into the market of the product containing the
new ingredient.
CRITERIA AND METHODS FOR IN-MARKET SURVEILLANCE
In-Market Monitoring Programs in the First Year

The criteria used to determine the choice of level of assessment and strategies to be
utilized are based on the previous discussion in this chapter of conditions under which
potential adverse effects of a new ingredient added to infant formula might have been missed
in preclinical or clinical trials. Regardless of the level that is chosen, it is expected that
systematic data collection procedures (level 1 assessment) or systematic review procedures
(level 2 assessment) will ensure that in-market monitoring information will be assessed for
each area of function reviewed in Chapter 6.
The choice of level 1 assessment for in-market monitoring is recommended only when all of
the following conditions occur:
• There is no evidence (presented in the submission) that indicates significant indi-
vidual or population differences in susceptibility to the ingredient, metabolites, secondary
effectors, or source.
• There is no evidence of adverse effects in preclinical or clinical studies, including
adverse effects with potentially plausible alternative explanations (e.g., the effects are viewed
as the result of random chance or the reviewers believe that there may be methodological or

PROPOSED IN-M ARKET SURVEILLANCE

1

Did any of the follow ing exist prior to - Slower developing brain regions,
marketing: - Endocrine or neurotransmitter action, and
- Evidence for significant individual - Earl y beha vior th at can impact the quality of
ongoing parent-child relations.
or population differences in
suscept ibility to the new ingredient?
the individual's growth or other outcomes,
preclinical or clinical trials? circum ference per centiles; skin or hair
- Scientific evidence linking the new changes; muscle atrop h y; mood changes;
ingredient or new ingredientsource to anorexia; vomiting; and diarrhea.
infant function (See Chapter 6) ?
The following issues should be considered in level 3
No assessment:
9 instruments to be used will var y depending on
the organ or fu nctional systems that are most
PASSIVE SURVEILLANCE
likel y to be affected b y the ingredient and
1-800 line or Internet web site
results from in-mar ket, preclinical, or clinical
studies.
10 make major life transitions, such as entry into
Surveillance data indicates problems school, the onset of puberty, high-school
graduation, and vocational choice. E ven further
in a function area beyond
follow-up in adult l i fe ma y be desirable with
expectation based upon previous Yes
transitions points, such as post-high-school
survey, clinical studies, or population
educa tion an d vocation status.
clinical trials should be tracke d a fter the trials
are over.
No
3
reported in literature Yes
published after marketing ? published or proprietary data, submitted
evidence, surveillance data, and ongoing
literature reviews

YYes

No No market

No
(See Sidebar A)
Yes
8
Expert panel makes
7
recommendations to
12 Initiate studies at level and type needed long-term safety status of
Continue surveillance to establish safety of formula formula with new ingredient
follow-up is needed
FIGURE 7-1 Proposed in-market surveillance algorithm. = a state or condition,

= a decision point, = an action, sidebar = an elaboration of recommendation
or statement.

statistical problems in the studies). This means that even if potentially plausible alternative
explanations are offered to explain adverse findings, level 1 in-market monitoring would not
be warranted since adverse effects were reported.
• A review of the relevant scientific literature indicates that there is no link between the
ingredient, metabolites, secondary effectors, or source, and the development of any of the
areas of infant function described in Chapter 6.
The choice of level 2 assessment for in-market monitoring must occur when any one of
the following conditions occurs:
• There is existing evidence for significant individual or population differences in sus-

ceptibility to the ingredient, metabolites, secondary effectors, or source.
• There is evidence of some type of adverse effect detected during preclinical or clinical
studies. This means that even if potentially plausible alternative explanations are offered to
explain adverse findings, level 2 in-market monitoring is warranted since adverse effects
were reported.
• A review of the relevant scientific literature indicates there is a link between the
ingredient, metabolites, secondary effectors, or source and the development of any of the
areas of infant function described in Chapter 6.
Level 2 assessments also would be automatically activated if level 1 assessments (in-

market toll-free or Internet reporting strategies) are utilized and information indicates prob-
lems in a function area over and above what might be expected based upon previous survey
or clinical information or population base rates. If the level 2 assessment panel review
indicates that existing published or proprietary data show deviations over and above what
might be expected based on population base rates, then a study using one or more of the
databases in the level 3 assessment would be required. A systematic re-evaluation of safety
issues (level 3 assessment studies) is necessary even when individual risk levels are of small
effect size, given that even small individual deficits can be magnified when cumulated at a
population level (Lester et al., 1998; Rice, 1990; Scott et al., 1994, 1999).
The choice of domains investigated during such a study would depend on the conclu-
sions of the review panel. As noted in Chapter 4, it is essential that each review panel include
qualified scientists from all of the domain areas discussed in Chapter 6. If the potential
detrimental effects appear to be restricted to a specific organ or functional system, then a
detailed investigation of this system with screening assessments of other systems or functions
potentially linked to the functioning of the specific organ system would be appropriate. For
example, growth could be easily measured during physical examinations of children in
surveillance populations by plotting height, weight, head circumference, and perhaps body
composition, especially body fat. If the organ or function system involved had known direct
links to other organs or functions, then a detailed investigation of the linked systems would
be warranted (e.g., given known links between brain and cognitive function, the potential
impact of a new ingredient on brain development would require detailed assessment of both
brain and cognitive function using the level 3 instruments described in Chapter 6).
In-Market Follow-up Assessments

Potential long-term adverse consequences associated with the addition of ingredients
new to infant formulas may not be detected in clinical studies. Therefore, for every new
ingredient to be added to an infant formula, a plan for a systematic approach for in-market

follow-up of the new formula should be required as part of the submission to the regulatory
agency. The studies should specially look for effects at the times when children make major
life transitions, such as entry into school. If evidence from the first transition period indi-
cated adverse effects, future studies would be warranted at later transition periods, such as
the onset of puberty, high-school graduation, post-high-school education, and vocational
choice.
The level of follow-up studies would depend on a number of specific criteria. For
example, the level of follow-up surveillance may be influenced by the type of substance
added to formulas. For ingredients that change only the flavor, color, or texture, minimal
concern may exist for long-term changes, especially if such compounds have a long history
of use in food. In contrast, a greater concern may be warranted in the case of nutritional
substances. For example, soy-based infant formulas are widely used and account for ap-
proximately 40 percent of formula sales in the United States. Most studies of growth and
development of infants consuming soy formula follow the infants only to age 1 year. Mendez
and coworkers (2002) emphasize the need for follow-up of these infants in the areas of
reproduction, immune function, thyroid function, visual acuity, and cognitive function. In
addition, evidence indicating that exposure to toxins may increase the organism’s re-
sponsivity to stress suggests that existing follow-up data should be evaluated or new follow-
up data should be collected at time periods when children make life transitions that would
increase the level of stress or demand upon the child, such as entry into school.
Based on the evidence cited earlier in this chapter, higher levels of follow-up assessment
levels would be particularly critical when any one of the following situations occurs:
• The action of the new ingredient relates to the development of slower developing
brain regions.
• The action of the new ingredient could affect endocrine or neurotransmitter action.
• The action of the new ingredient may have affected early child behavioral changes
(e.g., temperament) that can impact upon the quality of ongoing parent-child relations (see
Chapter 6).
• Primary-care physicians identify changes in the individual’s growth or other out-
comes, such as changes in weight, height, and head circumference percentiles; skin or hair
changes; muscle atrophy; mood changes; anorexia; vomiting; or diarrhea.
• The action of the new ingredient may have an effect only when individuals are
exposed to excess calories or other specific situations.
In the submission to the regulatory agency, the following criteria should be referred to
when justifying the strategy proposed for in-market follow-up of new ingredients added
to infant formulas.
The choice of level 2 assessment (review panel) for in-market follow-up is recommended
when any one of the following conditions occurs:
• In-market monitoring reveals adverse effects reported for the new ingredient or ingre-
dient source.
• There is existing evidence for significant individual or population differences in sus-
ceptibility to the ingredient, metabolites, secondary effectors, or source.
• There is evidence of adverse effects in preclinical or clinical studies, including adverse
effects with potentially plausible alternative explanations (e.g., the effects are viewed as the
result of random chance or the reviewers believe that there may be methodological or

statistical problems in the studies). This means that even if potentially plausible alternative
explanations are offered to explain adverse findings, level 2 in-market monitoring would not
be warranted since adverse effects were reported.
• A review of the relevant scientific literature indicates that there is existing evidence
linking the new ingredient, metabolites, secondary effectors, or source to the growth and
development of organ systems whose functions become apparent after the period of maxi-
mum exposure to infant formula, have known long-term direct consequences, or impact
upon developmental outcomes that could result in cumulative adverse effects over time.
The choice of level 3 assessments (options A through C) for in-market follow-up must
occur when any one of the following conditions occurs:
• In-market monitoring reveals a greater than expected number of adverse effects

reported for the new ingredient or ingredient source above what might be expected based on
population base rates, and an expert panel (level 2 assessment) concludes that there is
potential harm for the population. A systematic re-evaluation of long-term safety issues
(level 3 assessment studies) is necessary even when individual risk levels are of small effect
size, given that even small individual deficits can be magnified when cumulated at a popula-
tion level (Lester et al., 1998; Rice, 1990; Scott et al., 1994, 1999).
• There is evidence of some type of adverse effect detected during preclinical or clinical
trials and an expert panel concludes that there is potential for harm. This means that even if
potentially plausible alternative explanations are offered to explain adverse findings, level 2
in-market monitoring is not warranted since adverse effects were reported.
• A review of the relevant scientific literature by an expert panel (level 2 assessment)
indicates that new evidence exists linking the ingredient, metabolites, secondary effectors, or
source to the growth and development of organ systems whose functions become apparent
after the period of maximum exposure to infant formula, have known long-term direct
consequences, or a link to developmental outcomes that could result in cumulative adverse
effects over time.
The choice between options A through C in level 3 assessment (see earlier section,
“Recommended Levels of Assessment”) will depend on which populations with known
intakes of the formula under question are most available. However the committee recom-
mends as a matter of procedure that efforts be made to keep track of the location of
individuals in the original clinical trials (option C), at least through the grade-school years,
since this level offers the highest probability of accurately assessing intake of the formula
under question. One implication of this recommendation is the importance of ensuring
that there is a sufficient sample size in the original clinical trials so that subject attrition
does not compromise the level of statistical power in follow-up studies using the original
trial population. As discussed earlier in this chapter, sufficient statistical power is espe-
cially crucial in follow-up studies since effect sizes may be smaller than in the original
clinical trials.
Regardless of which option is chosen, the choice of domains to be investigated in follow-
up evaluations and the instruments to be used for each domain will depend on a variety of
factors. If the potential detrimental effects appear to be restricted to a specific organ or
functional system, then a detailed investigation of this system with screening assessments of
other systems or functions potentially linked to the functioning of the specific organ system
would be appropriate. For example, anti-infective ingredients (probiotics, prebiotics, lacto-
ferrin) may need follow-up of intestinal function and flora, while immune modulators may

be assessed with both T-cell and B-cell function. As discussed earlier, if the organ or function
system involved had known direct links to other organs or functions, then a detailed inves-
tigation of the linked systems would be warranted. It also is essential to include qualified
scientists from all of the domain areas discussed in Chapter 6 throughout the decision-
making process, including the level of follow-up assessment used and the domains where
more detailed follow-up will be needed.
Specific instruments chosen must be age appropriate, have documented sensitivity to
toxic exposure, and require the least level of invasiveness that still allows sufficient sensitiv-
ity. For example, Delaney-Black and colleagues (1998) used routine achievement testing
conducted by the child’s school system as measures of cognitive functioning in their investi-
gation of the long-term effects upon school-age children who had been exposed to cocaine in
the prenatal period. Measures of child behavior were based on parent and teacher responses
to standardized rating scales that assessed different dimensions of child adjustment. Follow-
up assessments could also look for evidence of referrals to special education, medical ser-
vices for achievement-related disorders (e.g., attention deficit hyperactivity disorder or learn-
ing disabilities), or mental health treatment for associated disorders (e.g., conduct disorder).
These sorts of evaluations will need to address confidentiality issues in the evaluation
protocol. Again, the decision of which specific measures should be used will require the
participation of expert scientists in the domain under consideration (see Chapter 6).
Existing Models for In-Market Surveillance

Despite the difficulties inherent in drawing useful conclusions from long-term, in-market
surveillance studies, possible research models for such studies exist. For example, an oppor-
tunity to study long-term effects of nutritional changes may be possible with the beginning
of the National Children’s Study, authorized by Congress in 2000. This study proposes to
follow 100,000 children from birth to age 21, with a specific focus on environmental
influences on growth and development (National Children’s Study, 2003). Nutrition will
obviously be a major focus of this study. Data from this type of longitudinal follow-up study
might be used to determine if there were increases in specific adverse events, over what
would be expected at a population level, after a formula containing a new ingredient was
introduced to the general marketplace.
In terms of actual research models, a recent analysis of the outcomes of infants placed on
soy formula is an example of extending what was originally a short-term study of growth of
infants between the age of 9 days and 16 weeks (Strom et al., 2001). The original study was
performed from 1965 to 1978. The investigators returned to their original population to
investigate possible long-term endocrinological and reproductive outcomes of infants who
had received soy formula containing phytoestrogens. Of the 952 subjects in the original
cohort, the investigators had to reject 48 subjects because of exposure to formulas other
than soy. Of the remaining 904 subjects, 51 were lost to follow-up and 42 refused to
participate. Overall, 90 percent of the original cohort was located and data were available
from 85 percent. This study was unable to identify any effect on endocrinological or general
health. While this type of study is difficult to conduct because both patients and investiga-
tors move, it shows that formula follow-up studies are possible when there is tenacity and
continuity of investigators.
Other research models also exist. An efficacy and safety study of an infant formula with
added LC-PUFAs was carried out for 18 months (Lucas et al., 1999). Another study of the
addition of LC-PUFAs with a follow-up period of 14 months (Auestad et al., 2001) was
extended, with the original study group of patients, to a follow-up period of 39 months

(Auestad et al., 2003). In 1978 and 1979, two soy-based formulas were released that
contained markedly decreased chloride content. The result was a large number of cases of
what was termed the “chloride deficiency syndrome” diagnosed between 1 and 6 months of
age (Roy, 1984). These infants presented with failure to thrive, lethargy, muscular weakness,
and loss of appetite. Laboratory analysis revealed metabolic alkalosis, hypochloremia, hy-
pokalemia, and hyponatremia. Nine- and 10-year follow-up of some of these infants showed
no measurable deficits in cognitive development (Willoughby et al., 1990).
The Strom and colleagues (2001) soy study, as well as other longitudinal studies, such as
the Framingham Heart Study (NHLBI, 2002) and the Harvard Physicians Health Study
(Steering Committee of the Physicians’ Health Study Research Group, 1989), demonstrates
that drawing valid conclusions from long-term surveillance studies can be accomplished,
even considering the multiple methodological problems inherent in such studies.
SUMMARY
The two components of in-market surveillance include monitoring for adverse effects in
infants after a formula has been introduced and long-term follow-up to ensure that there are
no delayed effects. There are a variety of logistical and methodological problems associated
with in-market surveillance, especially in regard to long-term follow-up (e.g., tracking sub-
jects, reconstructing causality). However there is a crucial need for such long-term surveil-
lance (e.g., brain areas that are adversely affected by new ingredients may not functionally
become apparent until later in development, or early exposure to a toxin may increase
susceptibility to later exposure to toxins). The committee recommends a systematic plan for
both continued in-market monitoring and long-term surveillance as an essential part of each
safety evaluation seeking to add new ingredients to infant formulas. The committee provides
three possible levels of assessment for in-market surveillance, along with criteria to decide
which surveillance level is appropriate for a new ingredient.
REFERENCES
Auestad N, Scott DT, Janowsky JS, Jacobsen C, Carroll RE, Montalto MB, Halter R, Qiu W, Jacobs JR, Connor
WE, Connor SL, Taylor JA, Neuringer M, Fitzgerald DM, Hall RT. 2003. Visual, cognitive, and language
assessments at 39 months: A follow-up study of children fed formulas containing long-chain polyunsaturated
fatty acids to 1 year of age. Pediatrics 112:e177–e183.
Barker DJ, Gluckman PD, Godfrey KM, Harding JE, Owens JA, Robinson JS. 1993. Fetal nutrition and cardiovas-
cular disease in adult life. Lancet 341:938–941.
Beaton GH. 1986. 1986 EV McCollum International Lectureship in Nutrition. Toward harmonization of dietary,
biochemical, and clinical assessments: The meanings of nutritional status and requirements. Nutr Rev 44:349–
358.
Bellinger DC. 1995. Interpreting the literature on lead and child development: The neglected role of the “experi-
mental system.” Neurotoxicol Teratol 17:201–212.
Bellinger D, Sloman J, Leviton A, Rabinowitz M, Needleman HL, Waternaux C. 1991. Low-level lead exposure
and children’s cognitive function in the preschool years. Pediatrics 87:219–227.
Bendersky M, Alessandri SM, Lewis M. 1996. Emotions in cocaine-exposed infants. In: Lewis M, Sullivan MW,
eds. Emotional Development in Atypical Children. Hillsdale, NJ: Lawrence Erlbaum Associates. Pp. 89–108.
Bengtsson T, Lindstrom M. 2000. Childhood misery and disease in later life: The effects on mortality in old age of
hazards experienced in early life, southern Sweden, 1760–1894. Popul Stud (Camb) 54:263–277.


Chasnoff IJ, Burns KA, Burns WJ. 1987. Cocaine use in pregnancy: Perinatal morbidity and mortality. Neurotoxicol
Teratol 9:291–293.
Clarke ADB, Clarke AM. 1981. “Sleeper effects” in development: Fact or artifact? Dev Rev 1:344–360.
Delaney-Black V, Covington C, Templin T, Ager J, Martier S, Compton S, Sokol R. 1998. Prenatal coke: What’s
behind the smoke? Ann N Y Acad Sci 846:277–288.
Fried PA. 1989. Cigarettes and marijuana: Are there measurable long-term neurobehavioral teratogenic effects?
Neurotoxicology 10:577–584.
Galler JR, Tonkiss J. 1998. The effects of prenatal protein malnutrition and cocaine on the development of the rat.
Ann N Y Acad Sci 846:29–39.
Griffith DR, Azuma SD, Chasnoff IJ. 1994. Three-year outcome of children exposed prenatally to drugs. J Am
Acad Child Adolesc Psychiatry 33:20–28.
Jackson AA. 2000. Nutrients, growth, and the development of programmed metabolic function. Adv Exp Med
Biol 478:41–55.
Jacobson SW, Jacobson JL. 2000. Teratogenic insult and neurobehavioral function in infancy and childhood. In:
Nelson CA, ed. The Minnesota Symposia on Child Psychology. Vol 31: The Effects of Early Adversity on
Neurobehavioral Development. Mahwah, NJ: Lawrence Erlbaum Associates. Pp. 61–112.
Leech SL, Richardson GA, Goldschmidt L, Day NL. 1999. Prenatal substance exposure: Effects on attention and
impulsivity of 6-year-olds. Neurotoxicol Teratol 21:109–118.
Lester BM, LaGasse LL, Seifer R. 1998. Cocaine exposure and children: The meaning of subtle effects. Science
282:633–634.
Leviton A, Bellinger D, Allred EN, Rabinowitz M, Needleman H, Schoenbaum S. 1993. Pre- and postnatal low-
level lead exposure and children’s dysfunction in school. Environ Res 60:30–43.
Lucas A, Stafford M, Morley R, Abbott R, Stephenson T, MacFadyen U, Elias-Jones A, Clements H. 1999.
Efficacy and safety of long-chain polyunsaturated fatty acid supplementation of infant-formula milk: A
randomised trial. Lancet 354:1948–1954.
Lyon G, Gadisseux J-F. 1991. Structural abnormalities of the brain in developmental disorders. In: Rutter M,
Casaer P, eds. Biological Risk Factors for Psychosocial Disorders. New York: Cambridge University Press.
Pp. 1–19.
Mayes LC, Grillon C, Granger R, Schottenfield R. 1998. Regulation of arousal and attention in preschool children
exposed to cocaine prenatally. Ann N Y Acad Sci 846:126–143.
Mendez MA, Anthony MS, Arab L. 2002. Soy-based formula and infant growth and development: A review. J
Nutr 132:2127–2130.
National Children’s Study. 2003. What is the National Children’s Study? Available at http://nationalchildrensstudy.
gov/about/overview.cfm. Accessed November 21, 2003.
NHLBI (National Heart, Lung, and Blood Institute). 2002. Framingham Heart Study. 50 Years of Research
Success. Online. Available at http://www.nhlbi.nih.gov/about/framingham/index.html. Accessed July 1, 2002.
O’Connor MJ, Sigman M, Kasari C. 1993. Interactional model for the association among maternal alcohol use,
mother-infant interaction, and infant cognitive development. Inf Behav Dev 16:177–192.
OSAS (Office of Scientific Analysis and Support). 2002. Letter to Stakeholders. Announcing CAERS—The CFSAN
Adverse Event Reporting System. Online. Center for Food Safety and Applied Nutrition, Food and Drug
Administration. Available at http://www.cfsan.fda.gov/~dms/caersltr.html. Accessed January 21, 2003.
Rice DC. 1990. The health effects of environmental lead exposure: Closing Pandora’s box. In: Russell RW, Flattau
PE, Pope AM, eds. Behavioral Measures of Neurotoxicity: Report of a Symposium. Washington, DC: Na-
tional Academy Press. Pp. 243–267.
Richardson GA. 1998. Prenatal cocaine exposure. A longitudinal study of development. Ann N Y Acad Sci
846:144–152.
Richardson GA, Conroy ML, Day NL. 1996. Prenatal cocaine exposure: Effects on the development of school-age
children. Neurotoxicol Teratol 18:627–634.
Romano AG, Harvey JA. 1998. Prenatal cocaine exposure: Long-term deficits in learning and motor performance.
Ann N Y Acad Sci 846:89–108.
Roy S III. 1984. The chloride depletion syndrome. Adv Pediatr 31:235–257.
Ruff HA. 1999. Population-based data and the development of individual children: The case of low to moderate
lead levels and intelligence. J Dev Behav Pediatr 20:42–49.
Russell RW. 1990. Neurobehavioral time bombs: Their nature and their mechanisms. In: Russell RW, Flattau PE,
Pope AM, eds. Behavioral Measures of Neurotoxicity: Report of a Symposium. Washington, DC: National
Academy Press. Pp. 206–225.

Rutter M, Pickles A. 1991. Person-environment interactions: Concepts, mechanisms, and implications for data
analysis. In: Wachs TD, Plomin R, eds. Conceptualization and Measurement of Organism Environment
Interaction. Washington, DC: American Psychological Association. Pp. 105–141.
Schneider ML, Roughton EC, Koehler AJ, Lubach GR. 1999. Growth and development following prenatal stress
exposure in primates: An examination of ontogenetic vulnerability. Child Dev 70:263–274.
Scott KG, Shaw KH, Urbano JC. 1994. Developmental epidemiology. In: Friedman SL, Haywood HC, eds.
Developmental Follow-up. Concepts, Domains, and Methods. San Diego: Academic Press. Pp. 351–374.
Scott KG, Avchen RN, Hollomon HA. 1999. Epidemiology of child developmental problems: The extent of the
problems of poor development in children from deprived backgrounds. Food Nutr Bull 20:34–44.
Singer LT, Arendt R, Minnes S, Farkas K, Salvator A, Kirchner HL, Kliegman R. 2002. Cognitive and motor
outcomes of cocaine-exposed infants. J Am Med Assoc 287:1952–1960.
Sostek AM. 1992. Prematurity as well as intraventricular hemorrhage influence development outcome at 5 years.
In: Friedman SL, Sigman MD, Sigel IE, eds. The Psychological Development of Low-Birthweight Children.
Annual Advances in Applied Developmental Psychology. Vol 6. Norwood, NJ: Ablex Publishing. Pp. 259–
274.
Spear LP, Campbell J, Snyder K, Silveri M, Katovic N. 1998. Animal behavior models. Increased sensitivity to
stressors and other environmental experiences after prenatal cocaine exposure. Ann N Y Acad Sci 846:76–88.
Spencer PS. 1990. Chemical time bombs: Environmental causes of neurodegenerative diseases. In: Russell RW,
Flattau PE, Pope AM, eds. Behavioral Measures of Neurotoxicity: Report of a Symposium. Washington, DC:
National Academy Press. Pp. 268–284.
Steering Committee of the Physicians’ Health Study Research Group. 1989. Final report on the aspirin component
of the ongoing Physicians’ Health Study. N Engl J Med 321:129–135.
Strom BL, Schinnar R, Ziegler EE, Barnhart KT, Sammel MD, Macones GA, Stallings VA, Drulis JM, Nelson SE,
Hanson SA. 2001. Exposure to soy-based formula in infancy and endocrinological and reproductive out-
comes in young adulthood. J Am Med Assoc 286:807–814.
Wachs TD. 2000. Necessary but Not Sufficient. The Respective Roles of Single and Multiple Influences on
Individual Development. Washington, DC: American Psychological Association.
Wasserman GA, Liu X, Popovac D, Factor-Litvak P, Kline J, Waternaux C, LoIacono N, Graziano JH. 2000. The
Yugoslavia Prospective Lead Study: Contributions of prenatal and postnatal lead exposure to early intelli-
gence. Neurotoxicol Teratol 22:811–818.
Waterland RA, Garza C. 1999. Potential mechanisms of metabolic imprinting that lead to chronic disease. Am J
Clin Nutr 69:179–197.
Weiss B. 1995. Incipient hazards of cocaine: Lessons from environmental toxicology. In: Lewis M, Bendersky M,
eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale, NJ: Lawrence Erlbaum
Associates. Pp. 41–55.
Willoughby A, Graubard BI, Hocker A, Storr C, Vietze P, Thackaberry JM, Gerry MA, McCarthy M, Gist NF,
Magenheim M, Berendes H, Rhoads GG. 1990. Population-based study of the developmental outcome of
children exposed to chloride-deficient infant formula. Pediatrics 85:485–490.
Winneke G. 1990. Neurobehavioral toxicity of selected environmental chemicals: Clinical and subclinical aspects.
In: Russell RW, Flattau PE, Pope AM, eds. Behavioral Measures of Neurotoxicity: Report of a Symposium.
Washington, DC: National Academy Press. Pp. 226–242.

Acronyms and Glossary
AAP American Academy of Pediatrics

ADME Absorption, distribution, metabolism, and excretion
ALAT Alanine amino transferase
Algorithm “A step-by-step procedure for solving a problem that contains con-
ditional logic (if/then) statements” (SMDMC, 1992)
ARA Arachidonic acid
ASAT Aspartate amino transferase
CDC U.S. Centers for Disease Control and Prevention

CFSAN U.S. Center for Food Safety and Applied Nutrition
CNS Central nervous system
DEXA Dual X-ray absorptiometry

DHA Docosahexaenoic acid
DNA Deoxyribonucleic acid
DRI Dietary Reference Intakes
Efficacy The capacity to produce an intended effect under the realistic situ-
ation of product use
FDA U.S. Food and Drug Administration

FDA Redbook Toxicological Principles for the Safety Assessment of Direct Food
Additives and Color Additives Used in Food. Redbook II-Draft
(OFAS, 2001) and Redbook 2000. Toxicological Principles for the
Safety of Food Ingredients (OFAS, 2003)
FD&C Act Federal Food, Drug and Cosmetic Act
175

Food additive “Any substance the intended use of which results or may reason-
ably be expected to result, directly or indirectly, in its becom-
ing a component or otherwise affecting the characteristics of any
food . . . , if such substance is not generally recognized, among
experts qualified by the scientific training and experience to evalu-
ate its safety, as having been adequately shown through scientific
procedures (or, in the case as a substance used in food prior to
January 1, 1958, through either scientific procedures or experience
based on common use in food) to be safe under the conditions of its
intended use” (FD&C Act, Section 201(s))
GFR Glomerular filtration rate

GMP Good Manufacturing Practices
GRAS Generally Recognized as Safe
Harm The nature of the undesired outcome (usually a health outcome)

associated with a hazard; often expressed in terms such as “cost”
Hazard Some substance or combination of substances (organic or inor-
ganic, or in some cases psychological) that may, under some cir-
cumstances and/or for some individuals, produce undesired health-
related outcomes.
HDL High-density lipoprotein
HHS U.S. Department of Health and Human Services
HIV Human immunodeficiency virus
HPLC High-performance liquid chromatography
IgA, IgE, IgG, IgM Immunoglobulin A, E, G, M

Infant formula A food which purports to be or is represented for special dietary
use solely as a food for infants by reason of its simulation of human
milk or its suitability as a complete or partial substitute for human
milk (FD&C Act, Section 201(z))
IOM Institute of Medicine
IVH Intraventricular hemorrhage
LC-MS Liquid chromatography-mass spectrometry

LC-PUFAs Long-chain polyunsaturated fatty acids
LD50 Lethal dose for 50 percent of the animals used in a study
LDH Lactate dehydrogenase
LDL Low-density lipoprotein
LSRO Life Sciences Research Office
MRI Magnetic resonance imaging
NHST Null hypothesis significance test

Novel food “(a) a substance, including a microorganism, that does not have a
history of safe use as a food; (b) a food that has been manufac-
tured, prepared, preserved or packaged by a process that (i) has not
been previously applied to that food, and (ii) causes the food to

APPENDIX A 177
undergo a major change; and (c) a food that is derived from a

plant, animal or microorganism that has been genetically modified
such that (i) the plant, animal or microorganism exhibits character-
istics that were not previously observed in that plant, animal or
microorganism, (ii) the plant, animal or microorganism no longer
exhibits characteristics that were previously observed in that plant,
animal or microorganism, or (iii) one or more characteristics of the
plant, animal or microorganism no longer fall within the antici-
pated range for that plant, animal or microorganism” (Canada,
2001)
Prebiotic “a non-digestible food ingredient that beneficially affects the host

by selectively stimulating the growth and/or activity of one or a
limited number of bacteria in the colon, and thus improves host
health” (Gibson and Roberfroid, 1995)
Probiotic “a live microbial feed supplement which beneficially affects the
host animal by improving its intestinal microbial balance” (Fuller,
1989)
Quality factors Factors necessary to demonstrate that the infant formula, as pre-
pared for market, provides nutrients in a form that is bioavailable
and safe as shown by evidence that demonstrates that the formula
supports the healthy growth when fed as the sole source of nutri-
tion (FDA, 1996)
Risk Likelihood that an adverse event will occur given exposure to a

hazard
Safety A reasonable certainty of no harm or, in some systems, a reason-

able balance between costs (harm) and benefits
Term infants Infants delivered between gestational age of 37 to 42 weeks with

birth weights of 2.5 kg or more; safety parameters and commit-
tee reviews were limited to healthy term infants with no chronic
disease
TLC Thin layer chromatography
Type I error The probability of falsely rejecting the null hypothesis (i.e., the
chance of concluding that a systematic relationship exists in the
population when it does not)
Type II error Failure to detect a real effect; occurs when one fails to reject a false
null hypothesis
UL Tolerable Upper Intake Level
VLDL Very low-density lipoprotein
WHO World Health Organization

REFERENCES
(Food and Drugs Act). Ottawa: Minister of Public Works and Government Services Canada.
Fuller R. 1989. Probiotics in man and animals. J Appl Bacteriol 66:365–378.
Gibson GR, Roberfroid MB. 1995. Dietary modulation of the human colonic microbiota. Introducing the concept
of prebiotics. J Nutr 125:1401–1412.
SMDMC (Society for Medical Decision Making Committee). 1992. Proposal for clinical algorithm standards.
Society for Medical Decision Making Committee on Standardization of Clinical Algorithms. Med Decis
Making 12:149–154.

Composition of Infant Formulas and

Human Milk for Feeding Term Infants
in the United States
179

TABLE B-1 Composition of Selected Formulas Marketed for Feeding to Term Infants in
the United States
Milk-Based
Whey Contain DHA
Component Lactose Free Predominant and ARA a
Protein equivalent (g) 14 14–16 14
Fat (g) 31–37 36–37 36–37
Fatty acids
DHA (%) NAb NA 0.15
ARA (%) NA NA 0.40
Polyunsaturated (%) — 22 —
Saturated (%) — 45 —
Monounsaturated (%) — 33–38 —
Linoleic (mg) 8,784 3,360–6,205 5,810–6,757
Carbohydrate (g) 72–74 71–81 73–74
Minerals
Calcium (mg) 550–568 423–527 527
Phosphorus (mg) 370–378 263–358 284–358
Magnesium (mg) 41–540 47–54 41–54
Iron (mg) 12 11–12 12
Zinc (mg) 5–7 4–7 5–7
Manganese (µg) 34–101 51–101 34–101
Copper (µg) 510–608 470–584 527–608
Iodine (µg) 61–101 11–68 41–68
Selenium (µg) 15–19 14–19 12–19
Sodium (mg) 200–203 148–182 162–182
Potassium (mg) 723–740 558–730 709–730
Chloride (mg) 439–450 376–431 426–439
Vitamins
Vitamin A (IU) 2,000–2027 2,016–2,190 2,027
Vitamin D (IU) 405–410 403–438 228–405
Vitamin E (IU) 14–20 9–15 10–13
Vitamin K (µg) 54 54–60 54
Thiamin (µg) 540–676 438–672 541–676
Riboflavin (µg) 950–1,014 946–1,008 946–1,014
Niacin (µg) 6,800–7,095 5,475–6,757 6,757–7,095
Vitamin B 6 (µg) 405–410 405–548 405
Folic Acid (µg) 101–108 50–110 101–108
Vitamin B 12 (µg) 2 1.3–2 2
Pantothenic acid (µg) 3,041–3,400 2,117–3,378 3,040–3,378
Biotin (µg) 20–30 15–35 20–30
Vitamin C (mg) 61–81 57–81 61–81
Other nutrients
Choline (mg) 81–108 81–101 81–108
Inositol (mg) 29–115 28–131 32–41
Nucleotides (mg) — 28–34 —
Potential renal solute load 132–180 99–132 270
(mOsm/L)
NOTE: Nutrient unit/L, unless otherwise noted.

aDHA = docosahexaenoic acid, ARA = arachidonic acid.
bNA = not applicable.
c — = not available.
SOURCE: Melanie Fairchild-Dzanis, personal communication, Nestle (August 26, 2002); IOM (1997, 1998,
2000, 2001, 2002, 2004); Mead Johnson Nutritionals (1999); Ross Products Division (2001), Anna Skulimowski,
personal communication, Wyeth Nutrition (August 26, 2002, and December 11, 2002).

APPENDIX B 181
Human Milk Adequate Intake (per day)

Isolated Soy- 7–12
Protein Based 0–6 Months 7–12 Months 0–6 Months Months
17–20 11.7 12.1 9.1 9.9
33–37 40 40 31 31
NA 1.58 1.58 0.5 g 0.5 g

NA —c — — —
24 15.6 15.6 4.9 g 5.1 g
43 — — — —
33 — — — —
3,360–6,757 560 560 440 460
68–74 74 74 60 95
605–710 264 210 210 270

409–560 124 124 100 275
50–74 34 34 30 75
12 0.35 0.35 0.27 —
4–8 2.5 0.85 2.0 —
169–228 3.5 3.5 3 600
470–804 250 200 200 220
60–101 146 146 110 130
14–19 18 18 15 20
202–297 160 130 120 370
706–810 500 500 400 700
376–540 — — 180 570
2,000–2,077 1,616 1,616 1,333 1,666

402–410 15.9 ± 8.6 15.9 ± 8.6 200 200
9–20 7.3 7.3 6 7.5
52–74 250 250 2 2.5
402–672 21 ± 0.04 21 ± 0.04 20 30
608–1,008 35 35 30 40
5,040–9,122 180 180 200 400
402–420 13 13 10 30
50–108 65 65 65 80
2–3 0.4 0.4 0.4 0.5
3,024–5,068 220 220 170 180
20–52 6 6 5 6
56–107 50 50 40 50
54–87 160 160 125 150

28–121 — — — —
— — — — —
130–180 — — — —

REFERENCES
IOM (Institute of Medicine). 1997. Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D,
and Fluoride. Washington, DC: National Academy Press.
IOM. 1998. Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin B12, Pan-
tothenic Acid, Biotin, and Choline. Washington, DC: National Academy Press.
IOM. 2000. Dietary Reference Intakes for Vitamin C, Vitamin E, Selenium, and Carotenoids. Washington, DC:
National Academy Press.
IOM. 2001. Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine,
Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc. Washington, DC: National Academy
Press.
IOM. 2002. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein,
and Amino Acids. Washington, DC: National Academy Press.
IOM. 2004. Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate. Washington, DC:
The National Academies Press.
Mead Johnson Nutritionals. 1999. Pediatric Products Handbook. Evansville, IN: Mead Johnson.
Ross Products Division. 2001. Pediatric Nutritionals Product Guide. Columbus, OH: Ross Products Division/
Abbott Laboratories.

Redbook Table of Contents
Chapter I. Introduction
A. Major Changes in the Revised Guidelines
1. Introduction
2. Changes in Determining Concern Levels and Recommended Toxicity Studies
for Food Ingredients
3. Changes in Toxicity Testing Guidelines
4. Other Changes
B. Flexibility and Consistency in Guidelines for Toxicity Testing
C. Applicability of These Guidelines to the Safety Evaluation of all Food Ingredients
Chapter II. Agency Review of Toxicology Information Submitted in Support of the Safety of
Food Ingredients
A. Introduction
B. Expediting Review of Toxicology Information
C. Evaluating Toxicology Information
Chapter III. Concern Levels and Recommended Toxicity Studies

A. Introduction
B. Concern Concentration Levels
C. Recommended Toxicity Tests
Chapter IV. Guidelines for Preclinical Toxicity Studies

A. Introduction
B. General Recommendations for Toxicity Studies
1. General Guidelines for Designing and Conducting Toxicity Studies
2. Guidelines for Reporting Results of Toxicity Studies
3. Pathology Considerations in Toxicity Studies
183

4. Statistical Considerations in Toxicity Studies

5. Diets for Toxicity Studies
a. Types of Diets
i. Natural Ingredient Diets
ii. Purified Diets
b. Issues to Consider when Selecting Diets for Animals in Toxicity Studies
C. Guidelines for Specific Toxicity Studies
1. Short-Term Tests for Genetic Toxicity
a. Bacterial Reverse Mutation Test
b. In Vitro Mammalian Chromosome Aberration Test
c. In Vitro Mouse Lymphoma TK+/– Gene Mutation Assay
d. In vivo Mammalian Erythrocyte Micronucleus Test
2. Acute Oral Toxicity Tests
3. Short-Term Toxicity Tests with Rodents and Non-Rodents
a. Short-Term Toxicity Studies with Rodents
b. Short-Term Toxicity Studies with Non-Rodents
4. Subchronic Toxicity Tests with Rodents and Non-Rodents
a. Subchronic Toxicity Studies with Rodents
b. Subchronic Toxicity Studies with Non-Rodents
5. One-Year Long-Term Toxicity Tests with Non-Rodents
6. Carcinogenicity Studies with Rodents
7. Combined Chronic Toxicity/Carcinogenicity Studies with Rodents
8. In Utero Exposure Phase for Addition to Carcinogenicity Studies with Rodents
9. Reproduction and Developmental Toxicity Studies
a. Guidelines for Reproduction Studies
b. Guidelines for Developmental Toxicity Studies
10. Neurotoxicity Studies
Chapter V. Additional Recommended Studies

A. Introduction
B. Metabolism and Pharmacokinetic Studies
1. Recommended Metabolism and Pharmacokinetic Studies
2. Considerations in the Design of Pharmacokinetic Studies
a. Test Substance
b. Animals
c. Route of Administration
d. Dosage Regimen
e. Sampling
f. In Vitro Studies: Dose Response, Mechanism
g. Analysis of Data
h. Pregnancy/Lactation/Reproductive Studies
3. Analysis and Use of Data from Pharmacokinetic Studies
a. Data Reporting and Parameter Estimation
b. Pharmacokinetic Models: Data Interpretation and Predicting Effects
4. Use of Pharmacokinetic Results for Study Design and Risk Assessment
a. Design of Toxicity Studies
b. Setting Dose Levels
c. Determining Mechanisms of Toxicity

APPENDIX C 185
d. Improving the Risk Assessment Process/Safety Assessment

5. References
C. Immunotoxicity Studies
1. Immunity: A Brief Review
2. Key Concepts in Immunotoxicity Testing
3. Indicators of Possible Immune Toxicity
4. Expanded Type 1 Immunotoxicity Tests
5. Type 2 Immunotoxicity Tests
6. Relevance of Primary Indicators of Immune Toxicity to Health
7. Adequacy and Reliability of Primary Indicators of Immune Toxicity
8. Recommendations for Further Immunotoxicity Testing when Primary Indica-
tors are Positive
9. Animal Models for Immunotoxicity Tests
10. Recommended Strategy for Assessing the Immunotoxic Potential of Food
Ingredients
11. Conclusion
Chapter VI. Human Studies

A. Clinical Evaluation of Food Ingredients
1. General Considerations for Clinical Studies of Food Ingredients
2. Specific Considerations for Clinical Studies of Food Ingredients
3. Sequence of Clinical Studies for Food Ingredients
4. Submitting Reports of Clinical Studies on Food Ingredients
5. Appendix A — Principles of Institutional Review and Informed Consent
B. Epidemiology Studies
Chapter VII. Emerging Issues

A. Introduction
B. Macro-Additives
C. Safety of Food Ingredients Developed by Biotechnology
D. Enzymes
E. Microbially Derived Food Ingredients
F. Advances in the Development of Alternatives to Toxicity Testing
G. Heritable and Somatic Genetic Toxicity
Chapter VIII. Glossary: Acronyms and Definitions
NOTE: Food ingredients include: direct food additives, color additives used in food, Generally Recognized as Safe
substances, food contact substances, and constituents or impurities of any of the above. Bolded text denotes a
section that has been finalized and is found only online in the Redbook 2000 (OFAS, 2003); the other information
is found only in Redbook II-Draft (OFAS, 2001).
REFERENCES

Applying the Recommended Approaches
The special needs and vulnerabilities of infants require a clear and complete set of
guidelines to assess the safety of infant formulas. Guidelines must also provide an appropri-
ate level of flexibility to address the multitude and diversity of possible new ingredients. It is
not realistic or desirable to provide specific recommendations for each potential new ingre-
dient. Thus the committee recommends that the manufacturer or notifier and an expert
review panel establish the relative importance of potential adverse effects for each new
ingredient and determine the types of preclinical and clinical studies and in-market surveil-
lance needed to accurately assess safety.
The committee was requested to apply the recommended tools and approaches to the
specific situation of adding long-chain polyunsaturated fatty acids (LC-PUFAs) to infant
formulas and to consider another example of a specific ingredient, if appropriate. Probiotics
was selected as the second case study to demonstrate the flexibility of the algorithms when
analyzing a “complex” ingredient comprised of a variety of different components that
would be contained in a microorganism or microorganism mix. This appendix describes the
steps in the committee’s recommended approaches that should be considered when assessing
the safety of LC-PUFAs and probiotics.
As discussed in Chapter 1, algorithms diagram the safety assessment process into a step-
by-step decision tree. The algorithms presented in Chapters 4 through 7 are provided to
summarize the appropriate level of assessment by considering the harm and the potential
adverse effects of a new ingredient. The approaches presented in this appendix are not meant
to provide all of the information, events, or tests that need to be assessed to ensure safety,
but rather to exemplify needed steps in the review process. The corresponding chapters
provide more information on specific levels and tests.
The committee emphasizes that the purpose of these case studies is not to make conclu-
sions about the completeness of the information or the safety of the new ingredient, but to
point out assessment processes or steps that need to be considered. The algorithms that
follow utilize an asterisk (*) with corresponding text underlined to indicate steps that the
committee concluded could have been included if the notifier had used the committee’s
186

APPENDIX D 187
recommended algorithms to guide decisions about the type of safety assessments to apply.
The asterisk with corresponding text underlined does not imply that the step was not
performed or considered by the qualified experts, but that the information was not available
for review by the committee. Note that under the proposed recommendation (Chapter 4),
the expert panel can choose to include or not include certain tests in the submission.
LONG-CHAIN POLYUNSATURATED FATTY ACIDS

For the case of LC-PUFAs, the committee applied the information given to the Food and
Drug Administration available through the Freedom of Information Act in Generally Recog-
nized as Safe (GRAS) Notices 000041 and 000080 to its recommended algorithms, compar-
ing the steps taken by the manufacturer, notifier, or expert panel to determine the safety of
ARASCO (arachidonic acid-rich single-cell oil) and DHASCO (docosahexaenoic acid-rich
single-cell oil) in infant formulas. Figure D-1 provides an overview of the proposed process.
The sources of ARASCO and DHASCO have no prior use in foods in the United States, but
they are used in infant formulas in Europe.
Preclinical Studies
As shown in Figure D-2, structure, stability, and solubility characterization studies are
an important part of preclinical assessment. Figure D-3 illustrates that level 2 assessments
would have been applied in determining the safety of the LC-PUFAs. These in-depth mea-
sures of the organ and neurological systems would further investigate abnormalities and/or
are theoretically related to structure or function. In this case it appeared that there were
some minor effects on organ systems. It is not obvious which of the proposed testing regimes
in Chapter 5 were followed. The nonhuman primate studies were limited. These are consid-
ered an appropriate model to study changes in general behavior and speed of neural process-
ing in response to the addition of LC-PUFAs. Chapter 5 provides more details on structure,
stability, and solubility characterization, as well as the committee’s recommendations for
level 2 assessments.
Clinical Studies
As seen in the overview of the proposed clinical guidelines (Figures D-4, D-5, and D-6),
it is not clear whether assessments of body composition, immune response, auditory func-
tion, and temperament were conducted. Several of these tests (to be determined by expert
panels), applied at level 2 or level 3, are especially important to determine the safety of LC-
PUFAs because theoretical safety concerns exist. For example, LC-PUFAs affect immune
response, and they have been linked to neural development. Chapter 6 provides the com-
mittee’s findings and recommendations on body composition and immune, auditory, and
temperament assessment.
In-Market Surveillance
Figure D-7 illustrates the levels of proposed in-market surveillance. Selection of an
appropriate type of in-market surveillance should be based on theoretical concerns about the
new ingredient and/or results from preclinical and clinical studies. As long as preclinical and
clinical studies are properly conducted, adverse outcomes should be rare and it would take
a considerable period of time to collect sufficient data in order to reaffirm the GRAS status

PROPOSED PROCESSES
1
formula
2
infant formula

- history of safe use and neurological s afety.
(See Chapter 5)
4
(See Sidebar A)
5 (See Chapter 6)
Clinical Studies
(See Sidebar B )
*
*
(See Chapter 7)
ingredient
ingredient 9

Y es
other components?
No

Yes
of new ingredient
No
No
15 14 REGULATORY AGENCY 12 REGULATORY AGENCY

DISCONTINUE DOES NOT APPROVE APPROVES INFANT
PROCESS INFANT FORMULA WITH FORMULA WITH NEW
NEW INGREDIENT INGREDIENT
13
* (See Sidebar C)
FIGURE D-1 Proposed process for evaluating the safety of ingredients new to infant formulas
algorithm: Application by using the long-chain polyunsaturated fatty acid Generally Recognized as
Safe (GRAS) Notifications 000041 and 000080 as a case study. An asterisk (*) along with the
corresponding text underlined indicate steps that were either not apparent or not carried out within
the GRAS notifications 000041 and 000080 provided to the committee. In-market assessment should
be planned in conjunction with preclinical and clinical testing. This algorithm is modeled after the
U.S. Generally Recognized as Safe Notification process; similar schemes can be adapted to other
regulatory processes. = a state or condition, = a decision point, = an
action, sidebar = an elaboration of recommendation or statement.

APPENDIX D 189
PPROPOSED PRECLINICAL ASSESSMENT

1
infant formula
2
AND Yes
determ ine their safety? Sidebar A: Chemical and Physical
Characterization

3
- High performance liquid chromatography,
No Initiate preclinical studies to evaluate toxicity liquid chromatography-mass spectrometry, and
and neuro logical safety (See Figure D-3) thin layer chromatography.
- T he stability to temperature, ultraviolet light as
8
* - T he percentage of the unidentifiable mater ials

in ingredient.
*
Chemical and Physical Characterization - T he kind ofsolvents,suspending agents,
(See Sidebar A) emulsifiers, or other materials that will be used
in a d minister ing the ingredient whether in
in vitro or an imal studies.
- Ingredient should be stored under conditions
that maintain its stability, quality, and purity
9 4
until the subsequent studies are complete.
Are chemical and Toxicity Assessment - T hese studies should be repeated with the
Yes
physical purity assured? (See Sidebar B) ingredient in the solution or matrix that would

1. Genetic tests
2. Cellular studies
- Acute, subchronic, and chronic
toxicity
- Absorpt ion, distribution,
- Gastrointest inal tract
No
- Hepatic
- Renal
- Hematology
- Immune
- Endocrine
- Neurolo gic
5
safety?
No
6
10 7

PROCESS (See Figure D-3) PROCESS
FIGURE D-2 Proposed preclinical assessment algorithm: Application by using the long-chain poly-
unsaturated fatty acid Generally Recognized as Safe (GRAS) Notifications 000041 and 000080 as a
case study. An asterisk (*) along with the corresponding text underlined indicate steps that were
either not apparent or not carried out within the GRAS notifications 000041 and 000080 provided to
the committee. = a state or condition, = a decision point, = an action,

P
1

2
No
(See Sidebar A) major organ systems, and neurological

8
Any evidence of
adverse Yes
3
Lev el 2 Assessment
* Detailed measures of genetic tests,
cellular studies, animal toxicity studies,
major organ systems, and neurological
(See Sidebar B)
No
4
for safety?
No
9 6 5
FIGURE D-3 Proposed levels of preclinical assessment algorithm: Application by using the long-
chain polyunsaturated fatty acid Generally Recognized as Safe (GRAS) Notifications 000041 and
000080 as a case study. An asterisk (*) along with the corresponding text underlined indicate steps that
were either not apparent or not carried out within the GRAS notifications 000041 and 000080 provided
to the committee. = a state or condition, = a decision point, = an

APPENDIX D 191
CPROPOSED CLINICAL ASSESSMENT

1
formula

development
Assess:
- Weight velocity
3
Grow th Studies
(See Sidebar A)
and
* - Length velocity
- Head circumference
- Body composition
* Clinical Endpoints
(See Sidebar B and Figure D-5)
- Kidney
4
Abnormal growth or
* - Blood
Yes
endocrine systems
Sidebar C:
Assessment
6
*
Assess:
*
(See Sidebar C and Figure D-6)
- T emperament
7 5

Yes
No
8
IS SAFE
FIGURE D-4 Proposed clinical assessment algorithm: Application by using the long-chain polyun-
saturated fatty acid Generally Recognized as Safe (GRAS) Notifications 000041 and 000080 as a case
study. An asterisk (*) along with the corresponding text underlined indicate steps that were either not
apparent or not carried out within the GRAS notifications 000041 and 000080 provided to the
committee. = a state or condition, = a decision point, = an action,


1
form ul a

Y es
li nk to major organ system s? Major organ systems screening
No
- Liver
6
- Kidney
Adverse effect/event documented - Blood
in precli ni cal trials? - Immune
OR - Endocrine
Y es (See Tabl es 6-3, 6-5, 6-6, 6-8, and 6-9)
Evi dence of signifi cant individual
difference i n susceptibil ity to the
ingredient?

No
Major organ systems detail ed m easures
7
in the foll owi ng:
Lev el 1 Assessment - Gastrointestinal tract
(See Sidebar A) - Liver
8
* - Kidney
- Blood
- Imm une
- Endocrine
Evi dence of (See Tabl es 6-3, 6-5, 6-6, 6-8, and 6-9)
effect/adverse Yes
event?
*
Lev el 2 Assessment
(See Sidebar B)
No 4
Evi dence of
effect/adverse Yes
event?
No
9 5
Continue to
DISCONTINUE
neurobehavioral clinical
PROCESS
studi es
FIGURE D-5 Proposed levels of clinical assessment of major organ, immune, and endocrine systems
algorithm: Application by using the long-chain polyunsaturated fatty acid LC-PUFA Generally Recog-
nized as Safe (GRAS) Notifications 000041 and 000080 as a case study. An asterisk (*) along with
the corresponding text underlined indicate steps that were either not apparent or not carried out
within the GRAS notifications 000041 and 000080 provided to the committee. = a state or
condition, = a decision point, = an action, sidebar = an elaboration of recom-
mendation or statement.

APPENDIX D 193

1

Assessment
2
assessment measures:
Known or theoreti cal li nk to - Age appropri ateness
Yes
neurobehavior? - Predictive value
- Sensi tivi ty
No - Cross-species general izabil ity
7
- Ease of ad ministration
Adverse effect/event
Study design requirements:
documented in preclinical
- Adequate stati stical power
trial s?
- Avoi d over-control of mediator variabl es
OR Yes - Use measurement aggregation
Evidence of significant
- Use repeated measures
ingredient?
No
11 Neural and behavioral screening

Known or theoreti cal measures admi nistered duri ng a routine
indi rect l ink to other well-baby physical exam or through
Yes
organ systems? parent reports. (See Tabl e 6-10)
No Sidebar C: Level 2 Assessment

12
Detai led measures of function in maj or
Lev el 1 Assessment child developmental areas. Si ngle
(See Sidebars A and B) assessment for each area using one
instrument.
13
8
Evidence of
adverse Yes
(See Sidebars A and C) child developmental areas on at least two
effect/event?
* separate occasionsusing two

9 3
Evidence of
Lev el 3 Assessment
No
adverse
effect/event?
Yes
* (See Sidebars A and D)
4
Evidence of
adverse Yes
effect/event?
No
No
14 10 6 5
MANUFACTURER/REGU
AGENCY DETERMINES
INGREDIENT IS SAFE
INGREDIENT IS SAFE
FIGURE D-6 Proposed levels of clinical assessment of development and behavior algorithm: Appli-
cation by using the long-chain polyunsaturated fatty acid Generally Recognized as Safe (GRAS)
Notifications 000041 and 000080 as a case study. NOTE: An asterisk (*) along with the correspond-
ing text underlined indicate steps that were either not apparent or not carried out within the GRAS
notifications 000041 and 000080 provided to the committee. = a state or condition,
statement.


1

- Evidence for signi ficant individual -Ea rly behavior t hat can impact the q ual i ty of
or pop ulatio n differences in ongoing parent-child relations.
susceptib ility to the new ingredient?
No assessment:
the organ or functional systems that are most
1-800 li ne or Internet web site
result s from in-market, preclinical, or clinical
studies.
Surveill ance data indicates problems school, the onset of puberty, high-school
survey, cl ini cal studi es or population
are over.
No
3
*
11

YYes

No No market

No
(See Sidebar A)
Yes
8
Expert panel makes
7
recommendations to
follow-up is needed
FIGURE D-7 Proposed in-market surveillance algorithm: Application by using the long-chain poly-
unsaturated fatty acid Generally Recognized as Safe (GRAS) Notifications 000041 and 000080 as a
case study. NOTE: An asterisk (*) along with the corresponding text underlined indicate steps that
were either not apparent or not carried out within the GRAS notifications 000041 and 000080 provided

APPENDIX D 195
of the ingredient. Chapter 7 provides more details on the committee’s recommendations for
in-market surveillance. If needed, a selection of a qualified and unbiased expert panel is
important to evaluate surveillance data and ongoing literature reviews to determine if follow-
up studies are necessary.
PROBIOTICS
For the case of probiotics, the committee reviewed GRAS Notice 000049 to analyze the
addition of probiotics to infant formula using its recommended algorithms. Figure D-8
provides an overview of the proposed process. Probiotics have a history of safe use in infant
formulas in other countries, and a review of the scientific literature showed no significant
adverse events linked to these ingredients.
Preclinical Studies
A probiotic is a complex ingredient (e.g., a microorganism) and thus, stability and
solubility studies should be performed with the ingredient in solution, as well as in the
matrix that would be fed to human infants (Figure D-9). Each probiotic should be tested
separately, as well as in the combination that will be used in the infant formula. This is
important because different probiotics may possess different chemical characteristics, nutri-
tional contributions, and pharmacological and physiological activities, and they may be
derived from novel sources or processes.
A comprehensive preclinical level 1 assessment also should be conducted. As shown in
Figures D-9 and D-10, preclinical studies are important in assessing the safety of probiotics
since changing intestinal flora may lead to production of atypical components in the intes-
tines. Chapter 5 provides more details on the committee’s recommendations for preclinical
studies.
Clinical Studies
As seen in the overview of the proposed clinical guidelines (Figures D-11 and D-12), it is
important to include appropriate measures of body composition and hepatic and endocrine
function. The addition of probiotics could lead to formation of certain molecules at high
levels not commonly present in the intestines. This could theoretically affect hepatic and
endocrine function and other systems. Finally, clinical studies should include a comprehen-
sive level 1 assessment of behavioral and neural screening measures (see Figures D-11, D-12,
and D-13). Chapter 6 provides more information about these assessments.
In-Market Surveillance
Figure D-14 illustrates proposed in-market surveillance guidelines. Assuming that the
recommended preclinical and clinical tests using probiotics detected no adverse effects,
passive surveillance for in-market monitoring and level 2 long-term follow-up strategies are
recommended. Chapter 7 provides more information about the committee’s recommenda-
tions on in-market surveillance.

PROPOSED PROCESSES
1
formula
2
infant formula

- history of safe use
* and neurological safety.
(See Chapter 5)
4
* (See Sidebar A)
*
metabolism, and developmental and behavioral outcomes .
5 (See Chapter 6)
Clinical Studies
* (See Sidebar B )
*
*
(See Chapter 7)
ingredient
ingredient 9

Y es
other components?
No

Yes
of new ingredient
No No
15
14 REGULATORY AGENCY
DISCONTINUE 12 REGULATORY AGENCY
PROCESS DOES NOT APPROVE
APPROVES INFANT
INFANT FORMULA WITH
NEW INGREDIENT FORMULA WITH NEW
INGREDIENT
13
* (See Sidebar C)
FIGURE D-8 Proposed process for evaluating the safety of ingredients new to infant formulas
algorithm: Application by using the probiotics Generally Recognized as Safe (GRAS) Notification
000049 as a case study. An asterisk (*) along with the corresponding text underlined indicate steps
that were either not apparent or not carried out within the GRAS notification 000049 provided to the
committee. In-market assessment should be planned in conjunction with preclinical and clinical test-
ing. This algorithm is modeled after the U.S. Generally Recognized as Safe Notification process;
similar schemes can be adapted to other regulatory processes. = a state or condition,
statement.

APPENDIX D 197
1
infant formula
2
AND Yes Sidebar A: Chemical and Physical
Is adequate literature available to Characterization
determ ine their safety?
- High performance liquid chromatography,
3 liquid chromatography-mass spectrometry, and
Initiate preclinical studies to evaluate toxicity
*
No - T he stability to temperature, ultraviolet light as
and neurological safety (See Figure D-10) well as the solubility properties of the ingredient.
- T he percentage of the unidentifiable mater ials
in ingredient.
8
Chemical and Physical Characterization * - T he kind of solvents, suspending agents,

emulsifiers, or other materials that will be
*
used in adm inister ing the ingredient whether in
(See Sidebar A) in vitro or a nimal studies.
- Ingredient should be stored under conditions
that maintain its stability, quality, and purity
until the subsequent studies are complete.
9 4 - T hese studies should be repeated with the
Are chemical and Toxicity Assessment
*
Yes be fed to the human infants.

1. Genetic tests
2. Cellular studies
- Acute, subchronic, and chronic
toxicity
* - Absorpt ion, distributi on,

- Gastrointest inal tract
- Hepatic
No
- Renal
- Hematology
- Immune
- Endocrine
- Neurologic
5
safety?
No
6
10 7

PROCESS
* (See Figure D-10) PROCESS
FIGURE D-9 Proposed preclinical assessment algorithm: Application by using the probiotics Gener-
ally Recognized as Safe (GRAS) Notification 000049 as a case study. An asterisk (*) along with the
corresponding text underlined indicate steps that were either not apparent or not carried out within
the GRAS notification 000049 provided to the committee. Many of the steps of chemical and physical
characterization cannot be applied to probiotics. = a state or condition, = a
decision point, = an action, sidebar = an elaboration of recommendation or statement.

P
1

2
No
* (See Sidebar A)
* major organ systems, and neurological

Any evidence of cellular studies, animal toxicity studies,
Lev el 2 Assessment
adverse Yes major organ systems, and neurological
(See Sidebar B)
No
4
for safety?
No
9 6 5
FIGURE D-10 Proposed levels of preclinical assessment algorithm: Application by using the probi-
otics Generally Recognized as Safe (GRAS) Notification 000049 as a case study. An asterisk (*) along
with the corresponding text underlined indicate steps that were either not apparent or not carried out
within the GRAS notification 000049 provided to the committee. = a state or condition,
statement.

APPENDIX D 199
PROPOSED CLINICAL ASSESSMENT

1
formula

development
Assess:
3
Grow th Studies
(See Sidebar A)
* - Weight velocity
- Length velocity
- Head circumference
- Body composition
and
* Clinical Endpoints
(See Sidebar B and Figure D-12)
- Kidney
4
Abnormal growth or
* - Blood
Yes
endocrine systems
Sidebar C:
Assessment
6
Assess:
*
(See Sidebar C and Figure D-13) * - Sensory and motor function
- T emperament
7 5

Yes
No
8
IS SAFE
FIGURE D-11 Proposed clinical assessment algorithm: Application by using the probiotics Gener-
ally Recognized as Safe (GRAS) Notification 000049 as a case study. An asterisk (*) along with the
corresponding text underlined indicate steps that were either not apparent or not carried out with-
in the GRAS notification 000049 provided to the committee. = a state or condition,
statement.


1
form ul a

Y es
li nk to major organ system s? Major organ systems screening
No
- Liver
6
Adverse effect/event documented

in precli ni cal trials?
OR
* - Kidney
- Blood
- Immune
- Endocrine
Y es (See Tables 6-3, 6-5, 6-6, 6-8, and 6-9)
Evi dence of signifi cant individual
difference i n susceptibil ity to the
ingredient?
No

in the foll owi ng:
*
Lev el 1 Assessment
- Liver
- Kidney
- Blood
8 - Immune
- Endocrine
Evi dence of
(See Tabl es 6-3, 6-5, 6-6, 6-8, and 6-9)
adverse Yes
effect/event?
3
Lev el 2 Assessment
(See Sidebar B)
No 4
Evi dence of
adverse Yes
effect/event?
No
9 5
Continue to DISCONTINUE
neurobehavioral studi es PROCESS
FIGURE D-12 Proposed levels of clinical assessment of major organ, immune, and endocrine sys-
tems algorithm: Application by using the probiotics Generally Recognized as Safe (GRAS) Notifica-
tion 000049 as a case study. An asterisk (*) along with the corresponding text underlined indicate
steps that were either not apparent or not carried out within the GRAS notification 000049 provided

APPENDIX D 201

1

Assessment
2 assessment measures:
- Age appropri ateness
Known or theoreti cal li nk to - Predictive value
Yes
neurobehavior? - Sensi tivi ty
- Cross-species general izabil ity
No
Adverse effect/event Study design requirements:
documented in preclinical - Adequate stati stical power
trial s? - Avoi d over-control of mediator variabl es
OR - Use measurement aggregation
Yes
Evidence of significant - Use repeated measures
ingredient?
No
*
11 measures admi nistered duri ng a routine
Known or theoreti cal well-baby physical exam or through
indi rect l ink to other Yes parent reports. (See Tabl e 6-10)
organ systems?
Sidebar C: Level 2 Assessment

No
12 Detai led measures of function in maj or
child developmental areas. Si ngle
Lev el 1 Assessment
* (See Si debars A and B)

assessment for each area using one
instrument.

13 8
Evidence of Detai led measures of function in maj or
Lev el 2 Assessment
adverse Yes child developmental areas on at least two
(See Sidebars A and C)
effect/event? separate occasionsusing two
9 3
Evidence of
Lev el 3 Assessment
adverse Yes
effect/event?
No
4
Evidence of
adverse Yes
effect/event?
No
No
14 10 6 5
MANUFACTURER/REGU
AGENCY DETERMINES
INGREDIENT IS SAFE
INGREDIENT IS SAFE
FIGURE D-13 Proposed levels of clinical assessment of development and behavior algorithm: Appli-
cation by using the probiotics Generally Recognized as Safe (GRAS) Notification 000049 as a case
study. An asterisk (*) along with the corresponding text underlined indicate steps that were either
not apparent or not carried out within the GRAS notification 000049 provided to the committee.
= a state or condition, = a decision point, = an action, sidebar = an
elaboration of recommendation or statement.


1

- Evidence for signi ficant individual or -Ea rly behavior t hat can impact the q ual i ty of
populatio n differences in suscepti bility ongoing parent-child relations.
to the new ingredient?
No assessment:
the organ or functional systems that are most
* 1-800 li ne or Internet web site

result s from in-market, preclinical, or clinical
studies.
Surveill ance data indicates problems school, the onset of pube rty, high-school
survey, cl ini cal studi es, or population
are over.
No
3

YYes

No No market

No
(See Sidebar A)
Yes
8
Expert panel makes
7
recommendations to
follow-up is needed
FIGURE D-14 Proposed in-market surveillance algorithm: Application by using the probiotics Gen-
erally Recognized as Safe (GRAS) Notification 000049 as a case study. An asterisk (*) along with the
corresponding text underlined indicate steps that were either not apparent or not carried out with-
in the GRAS notification 000049 provided to the committee. = a state or condition,
statement.

APPENDIX D 203
SUMMARY
These two case studies demonstrate the flexibility and utility of the proposed processes.
They also indicate an important potential role for expert panels in determining the types and
levels of assessment to ensure the safety of two very different ingredients. Their application
also shows the importance of standardized elements and frameworks when considering the
safety of new ingredients added to infant formulas.

Biographical Sketches of Committee Members
Richard J. Deckelbaum, M.D. (chair), is the Robert R. Williams Professor of Nutrition, a

professor of pediatrics, and director of the Institute of Human Nutrition at Columbia
University, College of Physicians and Surgeons. His expertise and research interests include
regulatory mechanisms for cell-lipid particle interaction and cell cholesterol and triglyceride
metabolism. He also coordinates programs related to the effects of varying nutrient intakes
on expression of cardiovascular risk factors in children of different genetic backgrounds and
the impact of nutritional status on expression of enteric infections. Dr. Deckelbaum received
his M.D. from McGill University. He was a member of the U.S. Department of Agriculture/
U.S. Department of Health and Human Services’ 2000 Dietary Guideline Advisory Commit-
tee, chair of the March of Dimes Task Force on Nutrition and Optimal Health, and is a
member of the National Institutes of Health Nutrition Study Section.
Linda Adair, Ph.D., is a professor of nutrition at the School of Public Health, University of
North Carolina at Chapel Hill, and fellow of the Carolina Population Center. Her research
interests include the determinants of early childhood feeding and growth patterns and
obesity in childhood and adolescence. She has designed and implemented population-based
health, demographic, and nutrition surveys with a special emphasis on longitudinal model-
ing. She received her Ph.D. from the University of Pennsylvania. Dr. Adair is a member of
the Society for International Nutrition Research and the American Society for Nutritional
Sciences.
Mark Appelbaum, Ph.D., is a professor in the Department of Psychology at the University of

California, San Diego. His research interests involve the application of quantitative and data
analytic methods to a wide variety of problems in psychology and the behavioral sciences.
Of particular interest to him are the development of quantitative methods for the study of
small samples for dealing with problems of variability and for understanding growth and
change. Dr. Appelbaum also deals with problems of data structure and analysis in large,
multisite studies. He received his Ph.D. from the University of Illinois.
204

APPENDIX E 205
George L. Baker, M.D., retired in 1999 from Mead Johnson Nutritionals where he was vice
president and medical director. In this role he provided regulatory, technical, and medical
oversight for the launch of new ingredients for infant formulas and drugs. Prior to joining
Mead Johnson in 1983, Dr. Baker was associate dean of the College of Medicine and a
professor in the Department of Pediatrics at the University of Iowa. Dr. Baker holds an A.B.
and an M.D. from the University of Missouri and completed his pediatric residency at the
University of Iowa Hospitals and Clinics.
Susan S. Baker, M.D., Ph.D., is a professor of pediatrics at the State University of New York
at Buffalo and codirector of the Digestive Diseases and Nutrition Center at Children’s
Hospital of Buffalo. Her research interests are in pediatrics, general nutrition, and the
barrier function of the gastrointestinal tract. Dr. Baker received her M.D. from Temple
University School of Medicine and her Ph.D. from Massachusetts Institute of Technology.
She recently completed service as chair of the American Academy of Pediatrics Committee
on Nutrition and chair of the American Board of Pediatrics Gastroenterology Sub Board.
Presently, Dr. Baker is chair of the North American Society of Pediatric Gastroenterology
and Nutrition Patient Care Committee.
Cheston M. Berlin Jr., M.D., is a university professor of pediatrics and professor of pharma-
cology at the Milton S. Hershey Medical Center at the Pennsylvania State University College
of Medicine, Penn State Children’s Hospital. His expertise and research interests are in
pediatric nutrition, lactation, breastfeeding, drugs in human milk, failure to thrive, phe-
nylketonuria, Tourette syndrome, drugs and nutrition, and caffeine. Dr. Berlin received his
M.D. from Harvard Medical School. He is a member of the American Academy of Pediat-
rics, the American Society for Nutritional Sciences, and the American Society for Clinical
Nutrition and is chair of the U.S. Pharmacopeia Immunizing Agents Expert Committee.
William C. Franke, Ph.D., is an associate director of the Center for Advanced Food Technol-
ogy at Rutgers University. He provides technical expertise in the area of product develop-
ment and food regulations, especially as related to nutraceuticals, and develops new oppor-
tunities for technology transfer to small and large companies. He also contributes to
administration, marketing, and strategic planning. Previously he spent 28 years at Lipton/
Unilever, where he served in a number of senior management positions in product develop-
ment, quality assurance, and regulatory affairs. Most recently he was Vice President for
Scientific and Regulatory Affairs with Unilever United States. He is a member of the boards
of the Cancer Institute of New Jersey and the Institute of Food Technologists Foundation.
Michael K. Georgieff, M.D., is a professor in the Departments of Pediatrics and Child

Development and codirector of the Center for Neurobehavioral Development at the Univer-
sity of Minnesota School of Medicine. His expertise and research interests are in fetal and
neonatal nutrition and neurodevelopment, with special emphasis on the effect of fetal/
neonatal iron nutrition on brain development and neurocognitive function and the effect of
illness on neonatal protein-energy metabolism. Dr. Georgieff received his M.D. from the
Washington University Medical School. He is a member of the Perinatal Research Society,
the American Academy of Pediatrics, the Society for Pediatric Research, and the American
Pediatric Society.
James M. Ntambi, Ph.D., is the Steenbock Professor in the Department of Biochemistry and
Nutritional Sciences at the University of Wisconsin, Madison. His expertise and research

interests are in the genetic regulation of lipid and carbohydrate metabolism. Dr. Ntambi’s
experimental work on the genetic regulation of the stearoyl-CoA desaturase enzyme has
recently led to many new insights into the importance of this enzyme in metabolism and in
disease states, such as obesity, diabetes, atherosclerosis, and cancer. His pioneering work
will help to explain the complex aspects of the “Metabolic Syndrome” and to advance our
understanding of nutrient-gene interactions. Dr. Ntambi serves on several university com-
mittees and National Institute of Health study sections and is a member of the National
Institute on Alcohol Abuse and Alcoholism’s Board of Scientific Counselors. Dr. Ntambi
received his Ph.D. from Johns Hopkins University School of Medicine and is a member of
the American Society for Nutritional Sciences.
Theodore D. Wachs, Ph.D., is a professor of psychological sciences at Purdue University.

His research interests are in two major areas: the role of early physical and social environ-
ments upon subsequent development and the relation of chronic, mild nutritional deficits to
infant and toddler cognition, temperament, and parent-child relations. Dr. Wachs received
his B.A. from Muhlenberg College and his M.S. and Ph.D. in clinical psychology from
George Peabody College. He is a member of the Society for Research in Child Development
and the International Society for the Study of Behavioral Development and is a fellow of the
American Psychological Association.

Infant Formula - NBK 2004

Uploaded by

Copyright:

Available Formats

Infant Formula - NBK 2004

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Infant Formula - NBK 2004

Uploaded by

Copyright:

Available Formats

http://www.nap.edu/catalog/10935.

Infant Formula: Evaluating the Safety of New

This book plus thousands more are available at http://www.nap.edu.

Committee on the Evaluation of the Addition of Ingredients

Food and Nutrition Board

THE NATIONAL ACADEMIES PRESS

Copyright © National Academy of Sciences. All rights reserved.

Library of Congress Cataloging-in-Publication Data

Copyright 2004 by the National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

The National Academy of Sciences is a private, nonprofit, self-perpetuating society of distinguished

Copyright © National Academy of Sciences. All rights reserved.

COMMITTEE ON THE EVALUATION OF THE ADDITION OF INGREDIENTS

RICHARD J. DECKELBAUM (Chair), Institute of Human Nutrition and Pediatric

Copyright © National Academy of Sciences. All rights reserved.

FOOD AND NUTRITION BOARD

CATHERINE E. WOTEKI (Chair), College of Agriculture, Iowa State University, Ames

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Richard J. Deckelbaum, Chair

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

Copyright © National Academy of Sciences. All rights reserved.

2 INFANT FORMULA: EVALUATING THE SAFETY OF NEW INGREDIENTS

COMMITTEE CHARGE AND APPROACH

Copyright © National Academy of Sciences. All rights reserved.

FINDINGS AND RECOMMENDATIONS

Approach for Evaluating Safety

Copyright © National Academy of Sciences. All rights reserved.

4 INFANT FORMULA: EVALUATING THE SAFETY OF NEW INGREDIENTS

Important Safety Considerations When Regulating Infant Formulas

Use of a Hierarchical Approach

• the reversibility of potential harmful effects,

Copyright © National Academy of Sciences. All rights reserved.

Additional Elements of a Safety Assessment

Copyright © National Academy of Sciences. All rights reserved.

6 INFANT FORMULA: EVALUATING THE SAFETY OF NEW INGREDIENTS

Sidebar A: Preclinical Studies

8 Regulatory agency raises

10 Regulatory agency has 11

Copyright © National Academy of Sciences. All rights reserved.

Preclinical Studies for Evaluating Safety

Clinical Tests for Evaluating Safety

Copyright © National Academy of Sciences. All rights reserved.

8 INFANT FORMULA: EVALUATING THE SAFETY OF NEW INGREDIENTS

PROPOSED PRECLINICAL ASSESSM ENT

3 Structure, Stability, and Solubility

Sidebar B: Toxicity Assessment

FIGURE ES-2 Proposed preclinical studies algorithm. = a state or condition, =

Copyright © National Academy of Sciences. All rights reserved.

Initiate preclinical studies to evaluate

Sidebar B: Level 2 Assessment

Copyright © National Academy of Sciences. All rights reserved.