Journal Pre-proof

The Future of CMC Regulatory Submissions: Streamlining Activities

Using Structured Content and Data Management

Kabir Ahluwalia , Michael J. Abernathy , Jill Beierle ,

Nina S. Cauchon , David Cronin , Sheetal Gaiki ,
Andrew Lennard , Pradeep Mady , Mike McGorry ,
Kathleen Sugrue-Richards , Gang Xue

PII: S0022-3549(21)00532-3
DOI: https://doi.org/10.1016/j.xphs.2021.09.046
Reference: XPHS 2556

To appear in: Journal of Pharmaceutical Sciences

Received date: 26 August 2021

Revised date: 29 September 2021
Accepted date: 29 September 2021

Please cite this article as: Kabir Ahluwalia , Michael J. Abernathy , Jill Beierle , Nina S. Cauchon ,
David Cronin , Sheetal Gaiki , Andrew Lennard , Pradeep Mady , Mike McGorry ,
Kathleen Sugrue-Richards , Gang Xue , The Future of CMC Regulatory Submissions: Stream-
lining Activities Using Structured Content and Data Management, Journal of Pharmaceutical Sciences
(2021), doi: https://doi.org/10.1016/j.xphs.2021.09.046

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2021 Published by Elsevier Inc. on behalf of American Pharmacists Association.

The Future of CMC Regulatory Submissions: Streamlining Activities Using Structured Content
and Data Management

Kabir Ahluwaliaa,b, Michael J. Abernathya, Jill Beierlea, Nina S. Cauchona,*, David Croninc, Sheetal Gaikid,
Andrew Lennarde, Pradeep Madyf, Mike McGorryg, Kathleen Sugrue-Richardsh, Gang Xuei

Author affiliations:
a
Department of Global Regulatory Affairs – CMC, Amgen, Inc., One Amgen Center Drive, Thousand Oaks, CA
91320, USA
b
University of Southern California, School of Pharmacy, 1985 Zonal Ave, Los Angeles, California 90089, USA
c
Cognition Corporation, 24 Hartwell Ave, Lexington, MA 02421, USA
d
Biotherapeutic Development & Supply, Janssen Pharmaceuticals, 1000 Route 202 South, Raritan, NJ 08807, USA
e
Department of Global Regulatory Affairs – CMC; Amgen Ltd, 1 Uxbridge Business Park, Sanderson Road, Uxbridge
UB8 1DH, United Kingdom
f
Product Quality Management, Janssen Pharmaceuticals, 1000 Route 202 South, Raritan, NJ 08807, USA
g
Biotherapeutic Development & Supply, Janssen Pharmaceuticals, Barnahely, Ringaskiddy, Co.Cork, Ireland
h
Department of Global Regulatory Affairs – CMC, Amgen Inc., 40 Technology Way West Greenwich, RI 02817
i
Biotherapeutic Development & Supply, Janssen Pharmaceuticals, 200 Great Valley Pkwy, Malvern, PA 10355, USA
*
Corresponding author: Nina S. Cauchon (Telephone: + 1 805-447-1569), Email address: [email protected]

Key Words:

Regulatory Science, Automation, Standards, Artificial Intelligence, Database(s), Machine Learning

Declaration of Competing Interest:

Kabir Ahluwalia (Summer Intern), Michael J. Abernathy, Jill Beierle, Nina S. Cauchon, Andrew Lennard,
and Kathleen Sugrue-Richards reports financial support was provided by Amgen Inc. Sheetal Gaiki,
Pradeep Mady, Mike McGorry, and Gang Xue reports financial support was provided by Janssen
Pharmaceuticals Inc. David M. Cronin reports financial support was provided by Cognition Corporation.

Abbreviations: AI, Artificial Intelligence; ANDA, Abbreviated New Drug Application; CBER, Center for
Biologics Evaluation and Research; CDER, Center for Drug Evaluation and Research; CDL,
Consolidation Layer; CDS, Chromatography Data System; CMC, Chemistry, Manufacturing, and Controls;
CMC-UDM, CMC specific Unified Data Model; CMO, Contract Manufacturing Organization; CPP, Critical
Process Parameters; CQA, Critical Quality Attributes; CRO, Contract Research Organization; CTD,
Common Technical Document (eCTD, electronic CTD); DMAP, Data Modernization Action Plan; DP,
Drug Product; EC, Established Condition; EDL, Enterprise Data Lake; eLN, electronic Laboratory
Notebooks; EMA, European Medicines Agency; ERP, Enterprise Resource Planning; ETL, Extract,
Transform, Load; FAIR, findable, accessible, interoperable, and reusable; FDA, Food & Drug
Administration; FHIR, Fast Healthcare Interoperability Resources; HL7, Health Level 7; HTS, High
Throughput Screening; ICH, International Council for Harmonisation of Technical Requirements for
Pharmaceuticals for Human Use; IDMP, Identification of Medicinal Products; ISO, International
Organisation for Standardisation; IT, Information Technology; JoS, Justification of Specification; JSON,
JavaScript Object Notation; KASA, Knowledge-Aided Assessment and Structured Application; LIMS,
Laboratory Information Management System; MES, Manufacturing Execution System; OPQ, Office of
Pharmaceutical Quality; PDF, Portable Document Format; PLCM, Product Lifecycle Management;
PQ/CMC, Pharmaceutical Quality/CMC; PQRI, Product Quality Research Institute; PQS, Pharmaceutical
Quality System; QOS, Quality Overall Summary SCA, Structured Content Authoring; SCDM, Structured
Content and Data Management; SPQS, Structured Product Quality Submissions; TMAP, Technology
Modernization Action Plan; XML, eXtensible Markup Language;
Table of Contents
Abstract ......................................................................................................................... 4

1. Introduction .......................................................................................................... 5

2. Regulatory Challenges from an Industry Perspective ....................................... 6

2.1 Brief Overview of Regulatory Submissions and Applicability of
SCDM ......................................................................................................... 7
2.2 Challenges in Regulatory Content Authoring ......................................... 8
2.3 Challenges in Responses to Questions from Health Authorities .......... 9
2.4 Challenges in CMC Data Management .................................................... 9

3. Current Regulatory Landscape ......................................................................... 10

3.1 FDA Initiatives ......................................................................................... 10
3.2 EMA Initiatives ........................................................................................ 11
3.3 ICH Efforts ............................................................................................... 12

4. Structured Content and Data Management and Authoring Based on a

CMC Unified Data Model (CMC-UDM) ............................................................... 13
4.1 The CMC-UDM and Semantics ............................................................... 14
4.2 Structured Content and Data Management (SCDM) ............................. 17
4.2.1 Multi-source Data Aggregation .............................................. 17
4.2.2 Data Pedigree .......................................................................... 18
4.3 Structured Content Authoring (SCA) .................................................... 18
4.3.1 Specifications.......................................................................... 19
4.3.2 Batch Analyses, Stability, and Justification of
Specifications ......................................................................... 20
4.4 Benefits of SCDM and SCA .................................................................... 21

5. Trends and Perspectives ................................................................................... 22

6. Conclusions........................................................................................................ 24

References .................................................................................................................. 26
Abstract

Recent advancements in data engineering, data science, and secure cloud storage can transform the
current state of global Chemistry, Manufacturing, and Controls (CMC) regulatory activities to automated
online digital processes. Modernizing regulatory activities will facilitate simultaneous global submissions
and concurrent collaborative reviews, significantly reducing global licensing timelines and variability in
globally registered product details. This article describes advancements made within the pharmaceutical
industry from theoretical concepts to utilization of structured content and data in CMC submissions. The
term Structured Content and Data Management (SCDM) outlines the end-to-end scientific data lifecycle
from capture in source systems, aggregation into a consolidated repository, and transformation into
semantically structured blocks with metadata defining relationships between scientific data and business
contexts. Automation of regulatory authoring (termed Structured Content Authoring) is feasible because
SCDM makes data both human and machine readable. It will offer health authorities access to the digital
data beyond the current standard of PDF documents and, for a review process, SCDM would “enrich the
effectiveness, efficiency, and consistency of regulatory quality oversight” (Yu et al., 2019). SCDM is a
novel solution for content and data management in regulatory submissions and can enable faster access
to critical therapies worldwide.
1. Introduction

Albert Einstein said that “the basis of all scientific work is the conviction that the world is an ordered and
comprehensible entity.”1 The current regulatory submission and review of drug applications is currently
not considered to be an ordered process as it requires significant manual and repetitive labor by both
sponsors and health authorities which delays the speed at which novel therapeutics become available to
patients. Pharmaceutical companies generate abundant volumes of data, content, and, ultimately,
electronic or paper documentation for regulatory submissions involving clinical trial applications, new drug
approvals, and post-approval lifecycle management activities. In turn, each health authority must receive,
review, and respond to these submissions, initiating further document generation between a health
2
authority and sponsor throughout the lifecycle of the product. In the process of developing Chemistry,
Manufacturing, and Controls (CMC) content for regulatory submissions, industry sponsors must
incorporate CMC data expectations that are aligned to globally harmonized guidance and regional
guidance into the regulatory strategy for each individual product. The CMC data are typically accessed
and manually transferred into the dossier from various systems, e.g., LIMS (Laboratory Information
Management System), electronic Laboratory Notebooks (eLN), batch records, certificates of analysis, or
from a data lake or other repository. From there, data and content flow into technical reports prepared by
the company, and then finally into the CMC sections of the regulatory filing dossier. Throughout this
process, significant redundancy is incurred when developing the original submissions, which is then
further adapted for different health authorities across the globe. Generally, the product-specific CMC
information remains the same as a single global product progresses through a commercialization
process. However, specific regional regulatory expectations add to the burden of information
management.

Digitization (the process of converting physical data into a digital format), digitalization (the conversion of
human-based processes to computer-operated processes), and automation are all expected to help with
management of region-specific documents by having a greater focus on the data and less on the
accompanying narrative.3 By integrating electronic narrative, data, and visuals, into a “single pane of
glass”, and by providing this information in a readily usable format outside of a PDF (Portable Document
Format), health authority reviewers and sponsors can draw conclusions in a more timely and efficient
manner. In addition, providing this information in a virtual cloud-based solution will enable multiple health
authorities to perform reviews collaboratively and in parallel across regions.4 Developing strategies to
automate and reduce redundant labor will allow sponsors and health authorities to focus on true risk-
benefit analyses, ease the burden of regional adaptation of regulatory expectations, limit the “flavors”
associated with variations in globally registered details, and accelerate approvals thus bringing new
medicines to patients faster.

One strategy to increase efficiency and speed of drug development is to restructure a sponsor‟s current
data management ecosystem and content authoring process to facilitate improved regulatory submission
and review. Technological maturation has now turned what was once theoretical into practical data
management with the advancements made to structured content management tools. This technology can
enable individually authored and approved content in a format that is both human and machine readable.5
Structured content is connected outside of a specific application or submission such that it can be
reapplied to any interface. This structured format retains the multiple layers of the data, which permits the
content to be readily used in multiple documents without the need to reauthor and reverify the
information, as opposed to the current rigid and unstructured formats, e.g., PDF. Simply put, structured
content is the individual building block of a document. The information block is fully defined and structured
which enables the block to be readily available in an approved state for use or reuse to build another
document. Structured Content and Data Management (SCDM) is an emerging field which is closely
related to structured content management. Here, SCDM is defined as the integration of structured content
with structured data and the management of those integrated components, currently specific to CMC
activities that involve high volumes of data used to author CMC submissions. At the heart of SCDM is a
core design principle which aims to shift a company‟s focus to managing data instead of managing
documents.

The current unscalable method of manual tracking and tracing data must be addressed to allow the
6
sponsor to pull reproducible and verified data for rapid regulatory strategic decision-making. This
transformative change is needed due to the growing size and complexity of company portfolios and the
corresponding increase in applications managed by health authorities. The pharmaceutical industry is
highly regulated and must abide by numerous, heterogeneous, and complex regulations across multiple
regions. While new regulatory guidances are continually in development across health authority interest
areas, the rate of change is often delayed as substantial evidence and data are needed to support
deviation from standard processes already in place. As a result, pharmaceutical companies and health
authorities are lagging in the implementation of technologies, such as SCDM, despite the advancement of
novel technologies and analytical capabilities that support crucial innovation. While there has been
significant discussion on the implementation of automation technologies over the last few years, there is
now an acute need for global industry participation.7-10

This article provides considerations on transforming the pharmaceutical industry‟s submission process
that could be realistically achieved within the next few years by utilizing intelligent automation and SCDM
to support the standardization of information pertaining to the submission process in terms of integrated
electronic narrative, data content and its analysis. This paper will address several topics of CMC data
management and the development of SCDM. First, it will define the challenges currently facing the
pharmaceutical industry regarding regulatory content authoring, responses to health authority information
requests, and CMC data management. It then describes the regulatory developments related to health
authority initiatives to standardize CMC data and automate the review process. This review of challenges
and initiatives is not meant to be exhaustive, but to illustrate key aspects being addressed by industry and
regulatory collaborators. This article will detail the current state of SCDM technology with use cases which
are being collaboratively developed to provide solutions addressing the deficiencies in the current data
collection and submission process. In addition, the benefits of SCDM and Structured Content Authoring
(SCA), as well as near and long-term objectives under development, are highlighted, thereby shedding
some light on what the future might hold relative to the ongoing efforts to transform the regulatory filing
and review landscape.

Regulatory Challenges from an Industry Perspective

The pharmaceutical industry lags behind other sectors and even other highly regulated industries in their
11-13
digital maturity. The need to address challenges associated with the current regulatory environment is
being discussed across the industry and by health authorities. Challenges in regulatory submissions,
health authority initiatives for data standardization, select cases of how SCDM is being implemented
internally in industry and challenges of SCDM implementation in CMC applications were previously
discussed in a 2020 publication.9 Recently in May 2021, another article was published as a review on
digital innovation in regulatory submissions in the pharmaceutical industry.14 The authors discussed
several challenges and opportunities in the regulatory submission process centering on the theme that
static, PDF-locked documents produce unique challenges for automation, dynamic access and review of
data, and intelligent analysis of data. Taking advantage of data and content formats, which are not as
restrictive as PDF formats, would digitally transform the pharmaceutical industry. This article expands on
the previous discussions and promotes the use of technology to modernize current methods of
coordinating and retrieving the required information from the immense volume of data produced and
reviewed in regulatory decision-making. Specifically, there is a focus on the possibilities and
developments in automating data for CMC sections of pharmaceutical regulatory submissions to health
authorities. CMC data are amenable for automation and this article will propose rules and processes
which can later be extended to other datasets such as pre-clinical and clinical data. Additionally,
digitization, digitalization, and automation provide an opportunity and platform by which true global
standardization of regulatory filing content and data can be realized. In the following sections, the
structure and challenges of these regulatory submissions will be described from an industry perspective.

Brief Overview of Regulatory Submissions and Applicability of SCDM

The International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human
Use (ICH) developed the Common Technical Document (CTD) describing the content required and where
the information should be located in a regulatory application using a standardized format.15 The CTD is
organized in five modules and CMC information is placed into Module 3 with a summary in Module 2. In
some regions this information is also transcribed and registered within Module 1 which contains regional-
specific administrative information. While the harmonization of the CTD across the ICH-participating
jurisdictions did improve the submission and review processes, the subsequent rapid acceleration of
therapeutic development, introduction of novel modalities, and accumulation and maturation of data
necessitates a further transformational change in the management and reporting of CMC data. The
dynamics of the pharmaceutical industry cannot sustain the outdated, isolated, and static data retrieval
methodologies utilized by both sponsor and health authorities in strategic decision-making and risk-based
analysis.

The information contained within the CMC modules is highly repetitive and heavily data driven, and,
therefore, represents a prime candidate for SCDM and automation. As shown in Figure 1, the submitted
data are defined by the CTD guidelines and the CMC data in Module 3 are highly interconnected and
dispersed into separate packages as needed. These separate submission packages account for region-
specific requirements, revisions based on health authority information requests, new product
presentations, and amendments throughout the product lifecycle. This requires significant tracking,
knowledge, and effort to ensure accurate and updated information is submitted to each health authority in
each separate package. The applicability of SCDM to CMC data is based on the feasibility to convert data
into a structured format, which would further enable automated content authoring including reuse and
traceability of submitted data.16 Targeting Module 3 data and narratives will provide a standardized
infrastructure enabling a sponsor to provide an accelerated, accurate data source to a health authority
review division and automate the authoring process in general. To meet the current standards of drug
applications outlined in the electronic CTD (eCTD), pharmaceutical companies have undergone data
digitization while several health authorities have developed electronic submission portals. However, in the
context of CMC data most companies and regulators are only in the beginning steps of digitalization
which includes the development and implementation of SCDM and SCA.
 Figure 1. Current CMC Data Management and Content Authoring Process.

To show the importance of digitalization in pharmaceutical submissions, the eCTD initially includes
approximately 45 independent granules (sections of the eCTD) that can be used for ICH and regionally
specific sections in Module 3. Dissecting these granules, there are typically an average of 5 to10
documents per granule, each of which are typically created via 3 authoring events, 3 review events, and 3
data verification events, totaling approximately 5,000 internal sponsor events in building the core Module
3. This immensely complex and time-consuming process only constitutes the first CMC module sent to a
health authority, following which regional customization results in other required variants for a product‟s
global approval which can include over 80 individual health authorities. All the previously described
manual processing, manipulation, and verification of data, in addition to global variation in registered
details, is for a product that is essentially the same for all global markets. These authoring and verification
efforts continue throughout the product lifecycle, which for many products can extend to 20 years on the
market, which further highlights the need for digitalization and global standardization. Despite the need for
digitization as a solution for streamlining data collection and regulatory authoring workflows, several
prominent regulatory challenges preclude broad implementation in the pharmaceutical industry and are
discussed further below.

Challenges in Regulatory Content Authoring

Figure 1 displays the multiple CMC submission packages that industry sponsors need to generate and
update throughout the lifecycle of a product. This content requires multiple rounds of review and data
verification for each authoring event to confirm accuracy, consistency or interpretation, and sufficient
content for regulators prior to its internal approval and submission. In addition, significant resources are
required to reauthor or repurpose the same section to meet the requirements or preferences for individual
countries or regions. This results in multiple variations of the same section containing slightly different
content because of differences in regulatory expectation, negotiations including specific requests, and
post-approval commitments for each region. Maintenance of these multiple variations is a challenge in
lifecycle management and in identifying and tracking the precise documentation submitted to each health
authority.

To add to the complexity, filings across regions are often staggered, sometimes by years, for a multitude
of reasons. During the time between submissions to different health authorities, lifecycle changes are
frequently implemented in the regions where the marketing authorization has been approved while an
original application is being prepared for a follow-on market. To prepare for the follow-on market, it is
critical to have access to the most current information including statistical assessments, which requires
knowing what is approved by region and having traceability back to what was revised, why, and for which
jurisdiction or health authority. It is imperative to be able to trace which information was included in the
approved submission to manage the lifecycle content properly for each region. The complexity of tracking
this for every CTD section in every country is laborious, manual, repetitive, and can lead to misalignment
of information if the correct approved document for the specific country is not used as the basis of a
variation. Another layer of intricacy is added when partnering with contract organizations as they may
utilize different internal document or data systems that feed into each CTD section. Even when leveraging
a health authority‟s initiative such as the Food & Drug Administration‟s (FDA) Project Orbis, which aims to
leverage a collaborative review format, variation in registered details still occur because of regional health
authority preferences.

Challenges in Responses to Questions from Health Authorities

Additional complexity occurs during the process of gaining health authority approvals with information
requests presenting additional challenges for both sponsors and health authorities. Following an initial
submission, responses to a health authority‟s information requests pose a challenge because the
turnaround time for responses varies by region and can, in many cases, be as little as a few days. In
addition, regulatory dossier approval timelines often depend on the response times being met. Agency
response times require that sponsors are diligent in efficiently utilizing and prioritizing resources to
compile responses, verify data, review, format, publish, and submit responses in a timely manner. Module
3 holds large datasets and the PDF format of the eCTD granules lack any automated traceability back to
the original data sources which makes new analysis, data retrieval, or updates to Module 3 that require
manual data input a time-consuming process. As discussed, there are redundant efforts involved in
developing documents for submissions to different health authorities. In turn, regulators each have their
own review methodologies and may ask the same or similar questions, creating repetitive efforts on the
side of both sponsors and reviewers. Ultimately, this extends the timeline for global product approvals.
Lastly, the static PDF format of the CTD is not efficient for data mining and information exchange
between the sponsor and reviewers because the raw data embedded within is not readily accessible or
deconstructable for further analyses. Regulatory reviewers have to manually deconstruct CMC narratives
and data from these rigid PDF files in order to populate information into their own data systems and
assessment reports, which involves additional cycles of copying and pasting data into other software
platforms.

Overall, the ability of sponsors to maintain efficient and user-friendly knowledge management systems is
degraded by the presence of multiple archives of documents throughout many storage systems and
repositories. This creates a complex array of submissions per region and further complicates subsequent
change controls involving agency information requests and updates to Module 3, as well as tracking
individual CMC commitments with each health authority. Throughout the product lifecycle, the discussions
and evaluations between sponsor and health authority demand fast access to data that is findable,
accessible, interoperable, and reusable (FAIR). However, industry document management standards
have not been universally adopted across companies because substantial resources and a cultural shift
17
in the industry are required for their development and implementation.

Challenges in CMC Data Management

The abundance of data contained within the thousands of pages being authored for regulatory
submissions is heavily reliant on narratives written by a large team of subject matter experts. There can
be many different authors, reviewers, and data verifiers per product submission which can lead to more
subjectivity with narrative-based submissions, resulting in inconsistency even from product to product
within a company. A further complication arises from the interrelationships between many of the granules
within a module of the CTD and even across modules. These interrelationships can result in different
authors using the same data with the potential for misalignment in interpretation between sections.

In addition to narratives, sponsors must generate data tables, process schematics, statistical plots and
visuals. These are often manually created and populated from a source document into Module 3 or „copy-
pasted‟ from an electronic source document. Given the vast amount of data provided in Module 3, this is
a resource-dense activity for both the author and the data verifier. If data change during lifecycle
management, it is difficult to trace the correct data to the submitted and approved version of the CTD.
Furthermore, the time lapse between regional filings often results in additional independent statistical
analyses based on changing data availability, which can potentially impact sponsor conclusions and
health authorities‟ assessments leading to further filing detail variations. Additionally, the heterogeneity of
data systems in use across departments or functions within a company has made it challenging to
implement more efficient document creation capabilities. Some important drivers of data variability include
source data origin (internal to the sponsor or from external partners), data gathered in an unstructured
format, regional inconsistencies (spelling, significant figures, formats), product modalities, and number of
manufacturing sites and batches. The lack of consistent structure in the data results in significant
resources placed in manual data transformation, aggregation, verification, and error correction.

While there have been improvements in the management of data, it is currently insufficient to
incrementally advance the authoring process. Some companies have begun compiling raw data into “data
lakes”, which allows different systems to store their data in a single repository. However, many of these
datasets are not easily accessible as it remains unstructured with minimal standardization. Unstructured
data create additional challenges for creation and timing of submissions as well as tracking the data
throughout the product lifecycle. Achieving the ultimate goal of submitting a single, harmonized global
filing with parallel reviews would be challenging because it relies on the implementation of data
standardization and automated processes by sponsors and regulators. The lack of digitalization towards
Structured Content Authoring (SCA) puts unnecessary burdens on both industry and regulators, delaying
regulatory approvals and the rate at which novel therapies can be delivered to patients. In the following
sections, the efforts being made by health authorities towards improving efficiencies in their review
processes are described.

Current Regulatory Landscape

Recently, industry and health authorities have begun to discuss and develop an intelligent automation-
based approach to modernize the regulatory submission and review process. This modernization includes
content authoring automation in parallel with data automation which will merge submission narratives with
data and images from internal disparate sources to drive submissions in a more efficient and accurate
manner. Additionally, health authorities are involved in digital transformation approaches to increase
information technology infrastructure and security for data management to meet the needs of their
research modernization efforts.18 Below, major developments from the FDA, European Medicines Agency
(EMA), and ICH initiatives are described.

FDA Initiatives

The FDA initiated several workshops and action plans to contribute to a SCDM-based approach for data
collection and regulatory review of CMC content and data. In 2018, the FDA established the Knowledge-
Aided Assessment and Structured Application (KASA) initiative, headed by the Office of Pharmaceutical
8
Quality (OPQ). The KASA initiative was created in response to increasing numbers of Abbreviated New
Drug Application (ANDA) submissions and included a public FDA meeting in September 2018 and a
discussion at a 2019 Product Quality Research Institute (PQRI)/FDA Conference.8,19-21 Structured
assessment templates and risk-ranking algorithms have been developed to support the large volume of
applications as part of the KASA initiative. In 2020, the OPQ developed, tested and began using
structured data interfaces for drug substance information and liquid dosage forms.22 In 2021, the KASA
initiative progressed further as the OPQ proceeded to evolve from text-based to increasingly data-based
22,23
quality reviews in developing the tool. Although the initial focus was on generics because of the large
24
number of ANDAs, the use is now being expanded to new drugs and biologics applications.

The FDA released the Pharmaceutical Quality/CMC (PQ/CMC) initiative on how to structure data with
eXtensible Markup Language (XML) and Health Level 7 (HL7) Fast Healthcare Interoperability Resources
25
(FHIR) data formats. PQ/CMC and KASA work synergistically; while KASA intends to change how
submissions are evaluated by the FDA, PQ/CMC aims to transform presentation of CMC data in
submissions by harnessing a structured data approach. To support KASA, PQ/CMC provides structure by
organizing data and standardization of terms to feed into structured applications. A PQ/CMC draft
guidance is proposed for 2022 and will likely be limited to data-driven sections such as Specification,
Batch Analysis and Stability.25 While these initiatives are specific to the FDA currently, there is increasing
discussion of global harmonization.

While older systems were once sufficient for enabling regulatory review, the current era of technological
innovation has demonstrated that significant improvements can be made to drive efficiency. The FDA has
acknowledged the need for newer technologies to meet increasing demand for a modern infrastructure by
piloting a variety of supportive initiatives. It recently announced two modernization plans for data
technology transformation. In 2019, the FDA provided the Technology Modernization Action Plan (TMAP)
for technology modernization in the agency‟s strategy for data and management of data. Additionally, the
Center for Biologics Evaluation and Research (CBER) is taking the first step to integrating the CBER‟s
26,27
Information Technology (IT) and data platforms by aligning with the FDA‟s TMAP. Following the
TMAP, in 2020, the Data Modernization Action Plan (DMAP) provided a strategy concentrating on newer
approaches, IT, and the process of using data to accelerate pathways to better therapeutics.28 The
DMAP focuses on identifying solutions and allowing the reviewer and sponsor to discuss the critical
scientific rationale and development of capabilities rather than focusing on collecting data first and then
looking for questions the data can answer.

In April 2021, the FDA Center for Drug Evaluation and Research (CDER) published an article on Industry
4.0 for pharmaceutical manufacturing. Industry 4.0 refers to the fourth industrial revolution which
combines advanced technologies including the Internet of Things , Artificial Intelligence (AI), robotics, and
advanced computing with the aim of revolutionizing the manufacturing landscape to create autonomous
7
and self-organizing systems which require little human involvement. However, the FDA acknowledges
that the current stage of pharmaceutical manufacturing is primarily, aligned with Industry 2.0, which
29
involves manufacturing with pre-determined and static settings. Reaching the level of digital maturity
necessitated by Industry 4.0 will require the integration of multiple data sources to allow for process
controls to be connected to process performance. As well, Industry 4.0 will identify control points to
ensure quality in drug development for both industry and health authorities. SCDM is among the key
enabling technologies which puts the industry on the path to Industry 4.0.

EMA Initiatives

The EMA and Heads of Medicines Agencies provide strategic direction in five-year strategy documents.
The EMA Network Strategy to 2025 identified six strategic focus areas including data analytics, digital
30
tools and digital transformation. The goals of this strategy include accessing and analysis of healthcare
and clinical trial data, building EU network capabilities to analyze big data, promoting dynamic regulation
and policy learning, leveraging bot technology, and ensuring data are managed and analyzed securely.
Importantly, the EMA intends to engage with the FDA and other regulatory agencies for global alignment.
The EMA has also developed an electronic format for product information in adopting the ISO IDMP
31
(International Organisation for Standardisation, Identification of Medicinal Products). Like the FDA‟s
FHIR standards, this EMA initiative is also based on HL7 for reliable exchange of product information by
using a common language for product name, ingredients, and pharmaceutical form for example. This
aligns with the FDA‟s PQ/CMC initiative as previously described. The purpose of the ISO IDMP and the
digital technology for implementation have many similarities to SCDM and, therefore, the IDMP is suitable
for future expansion of scope to CMC. Indeed, the scope is envisioned to soon be expanded for use in
GxP inspections. Both the IDMP and SCDM are designed for improved, accessible data exchange with
assured data integrity that can be reused for varied purposes; they allow for streamlined, simplified,
efficient regulatory filing and faster regulatory decision-making.32 The data may also be exchanged
between regulators in addition to between the applicant and agency. The data elements of the IDMP can
bring in source information from across the CTD and thereby has the potential to encompass all relevant
CMC information.

In 2017, the EMA held a workshop on the use of prior knowledge.14,33 In 2018, at a joint EMA/FDA
workshop, the application of prior knowledge to accelerate development of products in expedited review
procedures, such as Breakthrough Designation and PRIME, was discussed and led to an EMA draft
34,35
guideline. In 2020, the EMA also implemented a new regulatory submission portal (IRIS) that is being
piloted with selected types of submission such as Scientific Advice and Orphan Designation.36 IRIS
includes an electronic, cloud-based data lake repository of information intended to facilitate information
exchange between the EMA and applicant. Combining the IDMP and IRIS has the potential for a platform
to support SCDM, sponsor submissions, and health authority reviews. SCDM holds exciting potential for
the use of prior knowledge by pulling together cross-product information from a single data lake
repository. Furthermore, the use of prior knowledge requires justification in using data that could also be
appropriately structured within the data lake and by applying AI, it should be possible to filter prior
knowledge according to predetermined criteria that can learn differences and similarities between
products relative to a desired application.

ICH Efforts

In 2014, the ICH Steering Committee endorsed the development of a guideline to address post-approval
CMC challenges. The 2019 finalized guideline, ICH Q12: “Technical and Regulatory Considerations for
Pharmaceutical Product Lifecycle Management” provides a framework for an improved use of quality data
and a critical step towards harmonization and standardization of post approval changes.37 The core of the
endorsed guideline establishes a variety of related documents to be included with the original application
that describes planned changes to the manufacturing process with anticipation that these lifecycle
management plans simplify the review process for predicted post-approval changes. This provides an
excellent opportunity for the utilization of SCDM, which can efficiently facilitate and synergize with the
regulatory tools described in ICH Q12. For example, ICH Q12 defines Established Conditions (ECs) as
the elements in an application which would require a regulatory submission if changed post-approval.
With digitalization, binary decisions can be made relative to what is and what is not an EC, and through
automation details related to the specific determined ECs can be autopopulated into a standardized and
agreed upon format. Updates to ECs are then captured within the Product Lifecycle Management (PLCM)
document which is updated as needed throughout the product lifecycle. SCDM can use the ECs as rules
to flag and automate content authoring and submission of updates to the PLCM document which would
typically be a hands-on and time-consuming process. Thus, the proposed harmonized approach to
managing manufacturing and analytical procedure changes in ICH Q12 will encourage unification of post-
approval submission requirements across regions and enable more effective solutions, such as SCDM, to
provide streamlined access to higher-quality products by avoiding redundancies in manufacturing and
testing.
The ICH Assembly recently endorsed a proposal on revision of “The Common Technical Document For
The Registration Of Pharmaceuticals For Human Use: Quality – M4Q(R1),” and a proposal for a new
guideline on Structured Product Quality Submissions (SPQS).38 The ICH 2020 Annual Report conveyed
the revision of ICH M4Q(R1) as aiming to reorganize the application in a manner that would be
compatible with current quality assessment platforms in use in various regulatory agencies and would
facilitate a more highly structured and standardized assessment as compared to the current narrative
39
approach to filing submissions and regulatory review. The SPQS document identifies sections
containing CMC information and data that can be internationally standardized into a structured data
format in order to create a common set of data elements, vocabularies, and an electronic exchange
format for those sections of Module 3. Overall, the initiatives would minimize variability and duplication in
global data. This harmonization will also work well in parallel with SCDM to improve the efficiency of
collection, analysis, reporting, and review of CMC data.

Structured Content and Data Management and Authoring Based on a CMC Unified Data Model
(CMC-UDM)

Though the simplest of solutions would be to build a single CMC submission that is reviewed and
approved by a single global health authority, this scenario is highly unlikely because of geopolitical and
economic limitations. However, with the recent advancements in SCDM technology and solutions, a
single virtual submission housed within a cloud-based ecosystem is a real possibility. By leveraging
SCDM and SCA in an integrated fashion, the industry will likely see a shift from rigid filing formats to a
usable and much more efficient data exchange platform.

Three key pillars in using data in a knowledge-based management system for a submission process and
agency decision-making are: 1) data automation using SCDM, 2) electronic regulatory component or
content authoring automation and 3) leveraging standardized data from a structured cloud-based
platform. The relationship between these three pillars is illustrated in Figure 2. Implementing SCDM and
SCA will reduce manual and redundant labor throughout the product lifecycle. It avoids potential
replication errors when manually copying data to new submissions and allows automation of regulatory
content authoring. Additionally, SCDM has the potential to enable improved use of AI technologies, such
as machine learning, to analyze the collection of data as part of the company‟s Pharmaceutical Quality
System (PQS) and to provide further insights for development strategies, thereby improving decision-
making.40 Lastly, SCDM in combination with cloud technologies will enable the ability to construct
regulatory filings concurrently, submit filings simultaneously to multiple health authorities, allow
collaboration between agencies, and simplify the information requests from each health authority. The
following sections will describe technical aspects of a CMC specific Unified Data Model (CMC-UDM),
SCDM implementation, and examples of SCA.
Figure 2. Three pillars for a Structured Content and Data Management based Knowledge
Management Process.

The CMC-UDM and Semantics

A key element of SCDM and SCA is to establish the structure that would be consistently used to capture,
store, and consume data. In computer science, this structure is referred to as a UDM.41 An example for a
CMC-UDM that would support product specification and the analytical data of a pharmaceutical drug
product is shown in Figure 3. The core of the analytical data resides in the “Measurement”. However, to
enable the accurate and automated content authoring from the data, complete and consistent contextual
information (metadata) needs to be captured and logically linked to the “Measurement”. For example, and
in reference to the Module 3 Batch Analysis sections (S.4.4 and P.5.4 for drug substance and drug
product respectively), the Product (e.g., molecule, manufacturing stage, dosage form, etc.) and its
specific Material Lots must be selected to filter down to the corresponding batch release testing
measurement results. In addition, to support the conclusion of pass or fail for a batch release, the batch
should be compared against the Country/Geographic Region and its individual Specification for the
Product. Defining the data and metadata, as well as their semantic relationship (the basic meaning and
interrelationship between structured data) in a way that is unambiguously understandable by humans and
reasoned by machines is crucial to automate the data query and content authoring. In addition, controlled
vocabularies need to be established for each of the data elements such as Geographic Region, Quality
Attribute Class, Quality Attribute, Analysis, Measurement (parameter) as indicated with green boxes with
diagonal lines in Figure 3. In the computer-science domain, the controlled vocabulary and its hierarchical
structure is named taxonomy, while the relationships between the various taxonomies are defined as
ontology.42 The taxonomy and ontology collectively define the UDM accurately such that computers are
able to reason and support various data queries based on the data consumption and authoring needs. To
achieve effective SCDM and SCA, the CMC data domains, or collection of values that a data element
may contain within the CMC-UDM, must be defined.

In comparison to Batch Analyses, the Stability Data sections (S.7.3 and P.8.3) require similar data
elements but also need additional metadata related to stability studies such as stability time points,
storage conditions, etc. Instead of requiring a separate data model specifically for stability, the CMC-UDM
that supports the Batch Analysis data (typically also including „time zero‟ values in the stability protocol),
needs to be extended, as shown in Figure 3. The same data elements as well as semantic relationships
between Product, Process, Material Lot, Manufacturing, Specification, Instrument Class, Instrument,
Sample and Measurement can all be reused by the Stability section data query to automatically build the
stability lot summary as well as the stability data tables.

Figure 3. CMC-Unified Data Model for Product Specification and Analytical Data. Each green box
with diagonal lines represents a controlled vocabulary specific to a taxonomy domain; each blue outline or
blue shaded box represents a structured data container such as registries for Product, Material,
Equipment, Protocol, Results, etc.

To enable SCDM and SCA for the entire Module 3, the CMC-UDM needs to be extended to cover all the
various data elements that feed into these CMC sections. As the internationally harmonized standard for
regulatory filing, the CTD provides the complete content guideline to be extracted and semantically
organized to the CMC data model. To properly illustrate the entire scope of Module 3, the data model is
elevated to the data domain level as shown in the Figure 4 „honeycombs‟ scheme. Each hexagon
represents a data domain, which is semantically connected to the neighboring hexagon. For example, the
Material Lot in the center is a physical instance of a Product and produced by a Process. It then gets
aliquoted for release testing Measurement, Stability Study, or Product Characterization. Through these
connections, the Material Lot is further related to a Target and Specification via Product, and Raw
Material and Manufacturer via Process, as shown in the detailed data model in Figure 4. Compared to the
traditional rigid schema-based relational database, the knowledge graph, which uses graph-structured
data models, perfectly supports the extensibility of the semantic data model with great flexibility.

Figure 4. CMC Data Domains from CTD Module 3. Abbreviations: CPP, Critical Process Parameters;
DP, Drug Product; JoS, Justification of Specification.

A closer examination of the CTD sections would allow the construction and extension of the data
domains. For example, the product specification sections (S.4.1, P.5.1) clearly span across Product,
Country/Region and Specification, while the Stability sections (S.7.1, S.7.3, P.8.1, P.8.3) contain data
from Material Lot, Stability Study, Specification, and Measurement. On the other hand, the contents for
most data domains are used by many different sections such as Measurement which feeds into
Characterization (S.3.1, S.3.2), Justification of Specification (S.4.5, P.5.6), Batch Analysis (S.4.4, P.5.4),
and Stability (S.7.1, S.7.3, S.8.1, P.8.3), while Product definition and attribute data are required by
Nomenclature (S.1.1), Structure (S.1.2), General Properties (S.1.3), Formulation (P.2.1, P.2.2),
Specification (S.4.1, P.5.1), Analytical Procedure (S.4.2, P.5.2), and many more. For this reason, even
though Module 3 is complicated and data rich, the semantic data domains required to provide complete
coverage of Module 3 are manageable in quantity with careful semantic engineering. These same
principles can be applied to any product entities including combination products with devices.

Establishing an authoritative single data source for each logical data domain not only avoids the repeated
manual transcription of the same data, but also allows for a one-time only data integrity verification of the
input source information. While individual companies can build their own CMC-UDM, it would be more
efficient and impactful if the model could be harmonized across the pharmaceutical industry in partnership
with global harmonization efforts such as those within the ICH. Not only would a harmonized CMC-UDM
allow the collective expertise to build a more robust data model for SCA, but it would also allow the
submission of a digital data package for the statistical assessment, data trending, and analytical
comparison by a health authority.

Structured Content and Data Management (SCDM)

Once the CMC-UDM is established, SCDM becomes not only possible but optimal. When feasible, the
semantic data model and associated controlled vocabularies should be built into the laboratory
informatics ecosystems such as Compound Registry, eLN, and LIMS. The scientific data could be
captured in a FAIR manner with consistent and complete metadata from the beginning. Structuring data
in this manner enables the ability to pull verified data and visuals directly into authoring templates.
Although, it is recognized that in the highly regulated pharmaceutical industry it would typically be
expected to take years before such a sweeping change could be fully implemented, global agencies are
phasing in the standardization of key drug product terms through the ISO IDMP, which is a step in the
right direction. In the near future, especially considering the decades of legacy data, the Enterprise Data
Lake (EDL) approach, including multi-source aggregation and data pedigree, would be pragmatic in
implementation of SCDM and SCA tools.

Figure 5. Structured Content Authoring of Module 3 via a CMC-Unified Data Model (CMC-UDM)
Enabled by Structured Content and Data Management (SCDM). Abbreviations: CDS, Chromatography
Data System; CMO, Contract Manufacturing Organization; CRO, Contract Research Organization; eLN,
electronic Laboratory Notebooks; ERP, Enterprise Resource Planning; ETL, Extract, Transform, Load
process; HTS, High Throughput Screening; LIMS, Laboratory Information Management System; MES,
Manufacturing Execution System.

Multi-source Data Aggregation

As shown in Figure 5, the first step of SCDM is data aggregation via the established Extract, Transform,
Load (ETL) process into the EDL. Most pharmaceutical companies have established such consolidated
data warehouses either on the premises or in a private cloud environment. Typically, the data stored in
the EDL are consolidated from many disparate data sources and therefore follow the same data model as
the data sources. For this reason, it is often referred to as the Consolidation Layer (CDL). Due to the
heterogeneity of the data sources (typically over a dozen), the data in the CDL vary significantly in
structure and even nomenclature. Duplication and sometimes conflicting data are also common, which
makes it difficult to assemble and parse the data to support data analytics and automated report
authoring.

With the CMC-UDM, the aggregated data from the CDL can be re-engineered and the data elements
segregated into the various semantically connected data domains. In a way, the CDL resembles the
agricultural harvest process where produce is collected from various farms into warehouses after which a
semantic engineering tool would sort the produce into the different supermarket departments such as
vegetables, fruits, seafood, and meat. For example, the Enterprise Resource Planning (ERP) systems are
commonly used to manage the input and output materials for clinical batches. After the ERP data go
through the ETL process into the consolidation layer, they are parsed into the Raw Material, Material,
Process, and Manufacturer data domains. Here the Process domain would keep all the data domains
connected semantically, providing accurate lot genealogy and linkage to the process master batch record.
The semantic engineering process would also detect any duplication or conflicting data elements and
prompt for immediate remediation, some in the form of automated reconciliation by validated rules while
others may require human intervention. As a result, only one verified true copy of the same data element
would remain in the semantic layer, serving as the authoritative source for data analytics and SCA.

Data Pedigree

CMC data not only increase over time but also evolve and mature throughout the product lifecycle thus
increasing the complexity of data management. For this reason, one of the key elements of SCDM is to
manage the traceability (pedigree) throughout the data lifecycle from the initial data capture to
transformation, storage, and consumption, for both compliance and change management reasons. The
traceability must support bi-directional genealogy queries. Within the authored filing sections, each data
element should clearly point to the data element in the Semantic Layer as well as the raw data captured
in the original source system. In case there is any change in the source system, the SCDM and SCA
need to provide a summary of all impacted downstream data blocks, reports, and filing sections to allow
updates and version control. While the bi-directional query is feasible for SCDM because it has a
complete chain of custody of the data, proper breadcrumbs or data trails need to be built into the system
to allow the efficient pedigree search in both directions.

Structured Content Authoring (SCA)

Since the CMC-UDM is built by reverse engineering the eCTD contents, the SCA can be performed from
the SCDM by connecting the various data elements in the standardized filing templates to the
corresponding data domain in the Semantic Layer. As shown in Figure 5, the color-coded data elements
in the EDL are automatically authored into CTD sections based on the data pedigree defined by the
section templates. Product, Country/Region, and Specification get automatically fed into the Specification
sections (S.4.1, P.5.1). Additional metadata from data domains, such as Process which is semantically
related to Product and Method which in turn are related to Specification, are available to be populated in
the Specification sections (S.4.1, P.5.1) when needed. Similarly, Product, Process, Method, Material Lot,
Measurement and Specification may autopopulate the Batch Analysis section data tables (S.4.4, P.5.4),
while all these data domains plus the Stability Study data domain populate the Stability sections (S.7.1,
S.7.3, P.8.1, S.8.3). One distinct illustration is that the SCDM maintains a single true copy of a data block
for overlapped contents such as Product, Process, Specification, etc. The verified data blocks become
the common sources for many Module 3 sections, enabled by SCA templates with clear data pedigree for
straightforward compliance and change control management.

Figure 6 further illustrates the end-to-end SCDM and SCA lifecycle, which starts from the multi-source
aggregation of scientific data (eLN, LIMS, ERP, etc.) and semantic transformation, then autopopulates
the eCTD and technical report templates with optional data loaded into the SCA. The autogenerated
reports get published for compliance review and approval in the cloud before being released for
submission. Finally, any added content or interpretation results can be consumed back into the SCDM
layer. The SCA (step 3 in Figure 6) requires the digitization of the eCTD or technical report template
which semantically links each data column and data selection filter to the SCDM data elements defined
by the CMC-UDM as shown in the Specification and Stability examples below.

Figure 6. The Structured Content Authoring (SCA) Process Cycle. Abbreviations: eCTD, electronic
Common Technical Document; SCDM, Structured Content and Data Management.

Specifications

Specifications are set and monitored based on analysis of Critical Quality Attributes (CQAs) across the
life of the product, starting with the Target Product Profile prior to the initiation of First in Human clinical
trials. Specifications need to be clinically relevant, adequately justified (which may include leveraging prior
or platform knowledge), and supported by continued analysis. Specifications can vary from region to
region based on a multitude of factors such as regulatory requirements (e.g., peptide mapping is required
for specifications filed in Japan but not the US or EU), and negotiations during review (e.g., commitments
and responses to information requests). In the meanwhile, the specifications can change over the life of a
product for various reasons including new testing technology or „validating a test out‟ (e.g., sunsetting a
test because a CQA no longer needs to be monitored). Specification acceptance criteria are referenced
by multiple CTD sections within Module 3 including Specifications, Stability, Batch Analyses, and
Justification of Specifications, as well as other CTD modules (e.g., Module 2 Quality Overall Summary
[QOS]). This makes it challenging to track the individual specifications approved by each health authority
and difficult to lifecycle all relevant sections when there is a change that affects specifications.

For these reasons, the SCA Specification template is designed in three digital layers: the top layer links to
the Product and Process data domain, which defines the metadata for the product specification including
process and specification versions; the middle layer relates to the Country/Geographic Region and may
contain multiple entities, each defining the regional attribute for one specification section; and the bottom
layer is the core of the specification and may have one or multiple sections, each of which is a data table
containing Quality Attribute Class, Analysis, Method, Quality Attribute, and Acceptance Criteria as
discrete data columns based on the CMC-UDM (Figure 3). During authoring, the user would select the
Product and Process via the available dropdowns populated from the SCDM data query results. The SCA
would use the filtered Country/Regions to autogenerate one specification section for each
Country/Region. The Specification data contents then get populated into each specification section with
the product/process/country/region specific Quality Attributes and Acceptance Criteria. With accurate
metadata in the SCDM, the country/region variations could all be automatically reported with minimal user
input. When there is an update to all or part of the product specification, the SCA could be automatically
alerted of the impacted sections and prompted for version updates of either the entire specification
section or individual sections, leveraging the SCDM and SCA data pedigree. This creates traceability of
the lifecycle changes to the product specification and other parameters, including full audit trails and
justification for the changes.

Batch Analyses, Stability, and Justification of Specifications

Batch Analysis (S.4.5, P.5.4) and Stability (S.7.1, S.7.3, P.8.1, P.8.3) are all data-rich sections with
reasonably consistent tabular structures from ICH guidelines and minimal narrative. They also need to be
continuously updated as new data are obtained throughout the product lifecycle. Additionally, the
Justification of Specifications sections (S.4.5, P.5.6) largely reuse Batch Analyses and Stability data.
Therefore, these sections were selected as the next target for SCDM and SCA automation. Similar to
Specification, the Batch Analysis and Stability authoring follows the workflow in Figure 6, where the
Process and Material Lot metadata from the ERP systems and the Stability Study and Measurement data
from the eLN and LIMS systems are aggregated and semantically transformed using SCDM based on the
CMC-UDM as shown in Figure 3. Typically, these data are sourced across multiple data systems which
makes data compilation and verification time consuming, especially with the trend to outsource release
and stability testing to contract organizations. However, with the data consistently stored in a structured
format provided by SCDM, including data provided by contract organizations, SCA tools can directly
populate Batch Release and Stability templates.

The process flow for SCA to generate a stability report is shown in Figure 7. In Step 1, the user selects
filters for the Product, Material Lot, Specification, and Stability Study including the available stability time
points and analysis results. This will load the structured content and data from the EDL into the SCA
system. In Step 2, the methods and parameters of the loaded data are selected. The order can be
manually adjusted to establish a combination of lots and definition of how to group tables before
generating the tables. Finally, in Step 3, additional table formatting controls can be applied before
finalizing and exporting the tables. Narrative can be added to the footnote of the table to annotate some
observations or clarifications, but no reported data can be modified to assure the data integrity and
pedigree link back to the SCDM data elements. As the specifications may vary by country/region, different
versions of the Batch Release and Stability reports could be easily generated by selecting the different
Specification sections or more specifically the country/region metadata. When specifications require an
update by the company PQS, the data pedigree will trigger similar alerts to the impacted Batch Release
and Stability reports.

When new lot release data or additional stability time points are accrued, these reports may be easily
updated with the expanded data selection. In addition, trending analyses are typically required for stability
data over time and compared against the stability specification. A separate SCA template could use the
same selected dataset to autogenerate the Stability Summary and Conclusions sections (S.7.1, P.8.1).
Similarly, the Batch Analysis data is continuously monitored in the company PQS for any trends as new
lots are manufactured. Batch Analyses (S.4.4, P.5.4), Comparability (S.2.6, P.2.3) and Justification of
Specifications (S.4.5, P.5.6) all feed from the same SCDM data blocks to autogenerate the reports, some
involving additional data processing steps such as statistical analysis. As the SCA system links all these
sections back to the common source in the SCDM layer, it natively assures the consistency across the
variety of reports as well as complete traceability. Furthermore, since the data contents in the SCDM
system are established as a true copy of the raw data from various sources, the validated SCA system
eliminates the requirement for manual data verification for any of the automatically authored reports and
sections.

Figure 7. Structured Content Authoring (SCA) Workflow for a Stability Report. Abbreviations: EDL,
Enterprise Data Lake.

Benefits of SCDM and SCA

So far, the current challenges faced in data management, regulatory authoring, and lifecycle
management of CMC content have been described. SCDM has the ability to transition and transform the
current siloed data management systems and unstructured data lakes to an interactive and structured
data fabric. As well, SCDM allows for the creation, capture, and reuse of information as product and
process development progresses, addressing the manual and repetitive challenges associated with
constructing Module 3. It offers consistency and traceability of information across integrated data sets and
reduces the manual efforts of accurately reusing data across documents and data sources. Importantly,
increasing business efficiencies can be achieved via design and implementation of SCDM by enabling
process automation and content authoring tools. Finally, transitioning the CMC content and data from the
PDF file format to a functional, structured and exchangeable data format, such as JSON (JavaScript
Object Notation) or XML, allows sponsors and health authorities immediate access to usable content and
data, streamlining the submission and review processes and, most importantly, accelerating access to
therapies for patients.
To meet the challenges in the existing CMC regulatory landscape it is necessary to develop a more
templated approach to CMC content with the use of libraries and SCDM templates which will introduce a
standardized language and data presentation across all regions. Standardized language can be
integrated with PQ/CMC and similar initiatives. Within these libraries, sponsors can store the narratives as
components and multiple components can be compiled together to assemble each Module 3 CTD section
in a consistent, reproducible, and easily reusable fashion. If necessary, further “transcription” of Module 3
details into Module 2 QOS and Module 1 Regional Information becomes automated with SCA and SCDM,
avoiding the pitfalls associated with repetitive and tedious „copy and paste‟ activities. A SCDM-based
dossier preparation process could help to understand the dossier strategy, visualize the final output
earlier in the process, and identify gaps and conflicts among several variations of a narrative. SCDM
allows reviewers immediate access to supporting data which increases the efficiency of the internal
review processes and reduces inconsistencies as comments and edits are incorporated.

There is also the need to develop a system or tool capable of repetitive SCDM to allow real-time CMC
data automation, mapping, and authoring. Deployment of this tool would automate the flow of existing,
table-ready data from the EDL to finalized documents to health authorities‟ review tools. As such, the tool
enables simultaneous submission of data, electronic narrative, and visuals across multiple regions
allowing a sponsor to submit data once to all health authorities for data access and review. In this case,
health authorities could confer with each other and see the questions and decisions of other regulators.
This would reduce the amount of reauthoring and redundant activities for sponsors and health authorities.
SCA improves quality and efficiency of report deliverables and aids teams to establish a predictable
product development process, mitigating risks, improving quality, and decreasing time to market.

Extensive time is spent on manual sourcing and production of reports, leaving little time for analysis and
insights. If institutions synthesize their data to tell a story using analytics, they can cut through the noise
and provide clarity into trends and risk-based assessments. Data analytics platforms provide a time-
saving solution and effectively incorporate and assess massive data volumes, reducing effort required by
resources to create reports and increasing time for much-needed analysis. Removing the redundant
aspects of regulatory filings will create more time for scientific rationale discussions and true risk-benefit
analyses. Also, reducing redundant activities will increase job satisfaction for industry and health authority
employees. Overall, this improves the efficiency of global product submission, review and approval,
accelerating the access of new medicines for patients. This technology to reach digitalization in
Regulatory CMC is currently being developed and can transform these submission and review processes
in the coming years.

Trends and Perspectives

The pharmaceutical industry and health authorities are undergoing a transition towards digitization of data
and are at different stages of digitalization.43 The ultimate goal of SCDM implementation would be to
achieve a digital transformation of all data and information supporting regulatory filings from a global
perspective, supplying interconnectivity between CMC data in Module 3 and other modules of the eCTD.
Incorporating the aspects of digitization and digitalization with AI technologies will fundamentally change
strategies and operations for both industry and health authorities. In this article, examples have been
shown of how SCDM can be utilized in heavily data-driven CMC sections of the CTD to automate the
submission process.

SCDM has the potential to be expanded beyond data from the specific product under development by
incorporating prior knowledge of data obtained through product development of „like-molecules‟. Indeed,
combining SCDM across products and platforms, with advanced analytics including AI deep learning,
could aid the selection of applicable prior knowledge by automated access of all relevant information for
the designated criteria in order to achieve an objective assessment of the transferability of data across
„like-molecules‟. Moreover, deep learning may help identify differences in criteria between products that
are relevant to product stability and determine the acceptable limits for those criteria. In parallel to
submission and regulatory advancements, SCDM enables machine learning and similar technologies
which can transform the development process by providing novel insights from the data accumulated
throughout and between drug development platforms.

Over the next few years, additional SCDM-generated filings will be developed and tested in other regions
for developing harmonization efforts after successful internal industry incorporation efforts are
established. Within 10 years, SCDM can be extended to all modules of the CTD providing a standardized
data and submission format which is globally harmonized and accelerates the development of therapies.
However, the ultimate goal for submissions is to eventually create the opportunity for a single global
submission of a drug application that can be accessed and reviewed by various health authorities
simultaneously and reduce the time for a novel medicine to be accessible to patients in need. As stated
earlier, a single submission is challenging because of geo-political, socio-economical regional differences.
However, SCDM, SCA, and a CMC-UDM with proper rules can manage these differences and enable
such a concept. This is especially applicable for CMC data and content since typically a single global
product is characterized, manufactured, and tested in the same manner. SCDM provides the capability to
manage the regional differences of data reporting for the same product that is supplied globally. This
creates the basis to ground these technologies in CMC applications and branch out to other sections of
the CTD.

Information exchange between biopharmaceutical companies and health authorities and between multiple
agencies in a cloud environment containing the most recent data as they emerge will be a step towards
achieving the goal of a single global submission. To this end, Accumulus Synergy, a non-profit company
sponsored by leaders in the biopharmaceutical industry, is developing a cloud-based platform to facilitate
the information exchange to achieve real-time review in a global setting.44 Within the Accumulus platform
there will be separate and shared spaces for sponsors and health authorities to both work independently
and collaboratively. Sponsors can utilize SCDM and SCA technology to automate document generation
and push filings or have filings pulled to the Accumulus cloud to facilitate the submission and review
process (Figure 8). Additionally, there will be a strong focus on data privacy and cyber-security to protect
patients, sponsors, and agencies. Currently, Accumulus is testing their platform using a Parallel Review
Shared-Space and a CMC Data and Analytics use case. As discussed previously, parallel review will
reduce the total time to approve a drug in multiple regions while the Accumulus platform will facilitate the
collaboration between health authorities and communication between the sponsor and health authorities.
The CMC use case is focused on structured data submissions and real-time data exchange, as well as
leveraging the aspects associated with collaborative and parallel reviews. CMC data were chosen
because of the amenability to a structured approach, and once established the principles developed can
be extended to clinical and preclinical data. Accumulus aims to have developed their use cases and
additional components within three years and have a functioning cloud marketing application within a
decade.
Figure 8. Integration of Industry Structured Content and Data Management/Structured Content
Authoring (SCDM/SCA) with Global Information Exchange Cloud Platforms.

Conclusions

There are a number of drivers for establishing more efficient, reliable and automated SCDM ecosystems,
notably the direction health authorities are heading in evolving data standards and compliance
requirements.45 In addition, demands on pharmaceutical and biotechnology companies are increasing,
including cost, efficiency, speed, new regulatory requirements, novel modalities, new manufacturing
paradigms, and the vast amount of data which accompanies these changes. A further impetus is the
growing pressure on the life sciences and health industry to be more transparent, with readily available
answers to inquiries on demand and with more details. SCDM provides order and structure that will
enable concurrent, collaborative review of submissions among global health authorities, or within agency
committees and divisions. Reviewers can more quickly access data earlier as components are completed
and allocated to submissions with real-time dynamic and iterative assessments. Also, data provided for
multiple health authorities in the cloud will encourage unified submission requirements and collaboration
between regulatory agencies facilitating more efficient and timely assessments.

Data drive scientific and regulatory decision-making, and technology advances the interconnectivity of
data sources. These connections help identify, curate, and govern the data as well as automate the
sourcing of events for tracking on demand requests for specific data sets for any type of quality data
query in a product lifecycle. To match the initiatives taken by health authorities, SCDM and SCA will be
effective tools to address deficiencies in the current CMC regulatory submission and review processes
while improving data integrity, ensuring chain of evidence, and increasing health authority confidence in
sponsor submissions. This concept has moved from theoretical discussions to practical application in
ongoing testing phases as illustrated by the CMC-UDM, Specification and Stability examples described in
this article. Within the next 5 to 10 years, there is optimism that the concepts of SCDM and the supporting
technologies will expand beyond Module 3. The long-term aspiration and vision are to attain a single
cloud-based global regulatory submission for new drug applications utilizing SCDM, SCA, and a
seamless, yet secure information exchange cloud platform where health authorities can review
collaboratively and in parallel.
Acknowledgements:

The authors would like to thank Simon Hotchin, Rita Algorri, Gerd Kleeman, David Enck, and Thomas
DiFeo for their helpful suggestions throughout development of this review article.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-
for-profit sectors.
References

1. Einstein A, Shaw B. 1931. Cosmic religion : with other opinions and aphorisms. ed.

2. EFPIA. 2017. Optimising Post-Approval Change Management for Timely Access to

Medicines Worldwide. ed. https://efpia.eu/media/25953/efpia-post-approval-change-position-

paper_final_feb2017.pdf

3. Schmelzer R. 2020. AI Runs Into The Document And People Barrier: Digitization And

Digitalization. ed.: Forbes. https://www.forbes.com/sites/cognitiveworld/2020/06/23/ai-runs-into-

the-document-and-people-barrier-digitization-and-digitalization/

4. Robertson AS, Malone H, Bisordi F, Fitton H, Garner C, Holdsworth S, Honig P,

Hukkelhoven M, Kowalski R, Milligan S, O'Dowd L, Roberts K, Rohrer M, Stewart J, Taisey M,

Thakkar R, Van Baelen K, Wegner M 2020. Cloud-based data systems in drug regulation: an

industry perspective. Nat Rev Drug Discov 19(6):365-366. 10.1038/d41573-019-00193-7

5. DitaExchange. 2015. How Structured Content Management (SCM) Is Revolutionizing

the Life Sciences Industry. ed. https://ditaexchange.com/wp-content/uploads/2015/06/SCM-

White-Paper.pdf

6. Braun R. 2019. What does the future hold for regulatory information management? , ed.

https://www.pharma-iq.com/regulatorylegal/articles/what-does-the-future-hold-for-regulatory-

information-management

7. Arden NS, Fisher AC, Tyner K, Yu LX, Lee SL, Kopcha M 2021. Industry 4.0 for

pharmaceutical manufacturing: Preparing for the smart factories of the future. International

journal of pharmaceutics 602:120554. 10.1016/j.ijpharm.2021.120554

8. Yu LX, Raw A, Wu L, Capacci-Daniel C, Zhang Y, Rosencrance S 2019. FDA‟s new

pharmaceutical quality initiative: Knowledge-aided assessment & structured applications.

International Journal of Pharmaceutics: X 1:100010. https://doi.org/10.1016/j.ijpx.2019.100010

9. Algorri M, Cauchon NS, Abernathy MJ 2020. Transitioning Chemistry, Manufacturing,

and Controls Content With a Structured Data Management Solution: Streamlining Regulatory

Submissions. Journal of pharmaceutical sciences 109(4):1427-1438.

10.1016/j.xphs.2020.01.020

10. Cauchon NS, Oghamian S, Hassanpour S, Abernathy M 2019. Innovation in Chemistry,

Manufacturing, and Controls-A Regulatory Perspective From Industry. Journal of

pharmaceutical sciences 108(7):2207-2237. 10.1016/j.xphs.2019.02.007

11. Fox BP, Amit, Prevost M, Subramanian N. 2016. Closing the digital gap in pharma. ed.:

McKinsey & Company. https://www.mckinsey.com/industries/pharmaceuticals-and-medical-

products/our-insights/closing-the-digital-gap-in-pharma

12. Braun R. 2018. Why It‟s Time to Get Smarter about Content Management and

Document Production. ed.: @PharmExec. https://www.pharmexec.com/view/why-it-s-time-get-

smarter-about-content-management-and-document-production

13. Braun R. 2018. What Life Sciences Firms can Learn From Other Industries About

Optimizing Routine Document Production. ed.

https://www.pharmamanufacturing.com/articles/2018/what-life-sciences-firms-can-learn-from-

other-industries-about-optimizing-routine-document-production/

14. Macdonald JC, Isom DC, Evans DD, Page KJ 2021. Digital Innovation in Medicinal

Product Regulatory Submission, Review, and Approvals to Create a Dynamic Regulatory

Ecosystem-Are We Ready for a Revolution? Front Med (Lausanne) 8:660808.

10.3389/fmed.2021.660808

15. ICH. 2002. THE COMMON TECHNICAL DOCUMENT FOR THE REGISTRATION OF

PHARMACEUTICALS FOR HUMAN USE: QUALITY – M4Q(R1) ed.

https://database.ich.org/sites/default/files/M4Q_R1_Guideline.pdf

16. Benedictus J. 2019. Structured Content in Life Science. ed.

https://janacorp.com/structured-content-in-life-science/

17. Goran J, LaBerge L, Srinivasan R. 2017. Culture for a digital age. ed.: @mckinsey.

https://www.mckinsey.com/business-functions/mckinsey-digital/our-insights/culture-for-a-digital-

age

18. Cox B. 2020. Woodcock: The US FDA Sets The Stage For Global Quality Dossiers. ed.

https://pink.pharmaintelligence.informa.com/PS141465/Woodcock-The-US-FDA-Sets-The-

Stage-For-Global-Quality-Dossiers

19. Yu L. 2019. FDA's New Initiative: KASA. ed. https://pqri.org/wp-

content/uploads/2019/04/PQRI_KASA-Presentation_V4.pdf

20. Ahmed S. 2019. KASA To Support Generic Drug Review. ed.

https://www.contractpharma.com/issues/2019-04-01/view_fda-watch/kasa-to-support-generic-

drug-review/

21. Chatterjee B. 2019. Understanding The FDA's Knowledge-Aided Assessment &

Structured Application (KASA) Framework. ed.: BioProcess Online.

https://www.bioprocessonline.com/doc/understanding-the-fda-s-knowledge-aided-assessment-

structured-application-kasa-framework-0001
22. FDA. 2021. 2020 Annual Report. ed.

https://www.fdanews.com/ext/resources/files/2021/02-11-21-

OPQAnnualReport2020.pdf?1613087522

23. Eglovitch J. 2021. FDA gives generics updates at DIA town hall. ed.: RAPSorg.

https://www.raps.org/news-and-articles/news-articles/2021/6/dia-town-hall-fda-discusses-onsite-

inspections-int

24. Cox B. 2021. Newly Aligned Teams Sped US FDA‟s Drug Quality Reviews Over

Pandemic Hurdles. ed. https://pink.pharmaintelligence.informa.com/PS143801/Newly-Aligned-

Teams-Sped-US-FDAs-Drug-Quality-Reviews-Over-Pandemic-Hurdles

25. Fitzmartin R. 2021. Prescription Drug User Fee Act (PDUFA) VI: Electronic Submissions

and Data Standards. ed. https://www.fda.gov/media/147696/download

26. FDA. 2021. 2021-2025 Strategic Plan. ed. https://www.fda.gov/media/81152/download

27. FDA. 2019. FDA‟s Technology Modernization Action Plan. ed.

https://www.fda.gov/about-fda/reports/fdas-technology-modernization-action-plan

28. FDA. 2021. Data Modernization Action Plan | FDA. ed. https://www.fda.gov/about-

fda/reports/data-modernization-action-plan

29. Binggeli L, Heesakkers H, Wölbeling C, Zimmer T. 2018. Pharma 4.0™: Hype or

Reality? , ed.: ISPE. https://ispe.org/pharmaceutical-engineering/july-august-2018/pharma-

40tm-hype-or-reality

30. EMA. 2020. Network Strategy to 2025. ed.

https://www.ema.europa.eu/en/documents/report/european-union-medicines-agencies-network-

strategy-2025-protecting-public-health-time-rapid-change_en.pdf
31. EMA. 2021. Products Management Services - Implementation of International

Organization for Standardization (ISO) standards for the identification of medicinal products

(IDMP) in Europe. ed. https://www.ema.europa.eu/en/documents/regulatory-procedural-

guideline/products-management-services-implementation-international-organization-

standardization-iso-standards_en.pdf

32. EMA. 2016. Introduction to ISO Identification of Medicinal Products, SPOR programme

ed. https://www.ema.europa.eu/en/documents/other/introduction-iso-identification-medicinal-

products-spor-programme_en.pdf

33. EMA. 2017. Joint Biologics Working Party / Quality Working Party workshop with

stakeholders in relation to prior knowledge and its use in regulatory applications. ed.

https://www.ema.europa.eu/en/events/joint-biologics-working-party-quality-working-party-

workshop-stakeholders-relation-prior-knowledge

34. EMA. 2018. Workshop with stakeholders on support to quality development in early

access approaches (i.e. PRIME, Breakthrough Therapies). ed.

https://www.ema.europa.eu/en/documents/report/report-workshop-stakeholders-support-quality-

development-early-access-approaches-ie-prime_en.pdf

35. Oakes K. 2021. New toolbox available for EMA's PRIME designees. ed.: RAPSorg.

https://www.raps.org/news-and-articles/news-articles/2021/2/new-toolbox-available-for-emas-

prime-designees

36. EMA. 2020. New online platform for scientific advice. ed.

https://www.ema.europa.eu/en/news/new-online-platform-scientific-advice
37. ICH. 2019. TECHNICAL AND REGULATORY CONSIDERATIONS FOR

PHARMACEUTICAL PRODUCT LIFECYCLE MANAGEMENT Q12. ed.

https://database.ich.org/sites/default/files/Q12_Guideline_Step4_2019_1119.pdf

38. ICH. 2020. ICH Press Release. ed. https://admin.ich.org/sites/default/files/2020-

06/ICH40MayTC_PressRelease_2020_0603_FINAL_0.pdf

39. ICH. 2021. 2020 Annual Report. ed. https://admin.ich.org/sites/default/files/inline-

files/ICH_AnnualReport_2020_2021_0602.pdf

40. Kamiński R. 2021. AI in Pharma. What does Artificial Intelligence bring to the

pharmaceutical industry? - nexocode. ed.: @nexocode_com.

https://nexocode.com/blog/posts/ai-in-pharma/

41. IBM. 2021. IBM Unified Data Model for Healthcare. ed.

https://www.ibm.com/downloads/cas/KOV7YZPE

42. Breitman KK, Casanova MA, Truszkowski W. 2007. Ontology in Computer Science.

Semantic Web: Concepts, Technologies and Applications, ed., London: Springer London. p 17-

34.

43. McKinsey & Company. 2020. The state of AI in 2020. ed.: @mckinsey.

https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/global-survey-

the-state-of-ai-in-2020

44. Accumulus Synergy. 2020. Accumulus Synergy White Paper. ed.

https://www.accumulus.org/wp-content/uploads/2021/06/Accumulus_Synergy_White_Paper.pdf
45. Cwienczek A. 2020. Capitalising on structured content management in R&D/Regulatory

Affairs. ed. https://www.pharmiweb.com/article/capitalising-on-structured-content-management-

in-rdregulatory-affairs