Carcinogenesis vol.34 no.1 pp.

170–175, 2013
doi:10.1093/carcin/bgs315
Advance Access publication October 4, 2012

Phytanic acid and the risk of non-Hodgkin lymphoma

Nicholas J.Ollberding1,*, Briseis Aschebrook-Kilfoy1, Greater consumption of red meat and processed meat (4–6), and
Donne Bennett D.Caces2, Margaret E.Wright3, Dennis dairy products(7–9) has been associated with an increased risk of NHL,
D.Weisenburger4, Sonali M.Smith2 and Brian C.-H.Chiu1,5 in several, but not all epidemiologic studies (4,7,9–11). Phytanic acid
(3,7,11,15-tetramethylhexadecanoic acid) is a 3-methyl, saturated,
1
Department of Health Studies, University of Chicago, Chicago, IL, USA, branched-chain fatty acid formed in the gut of ruminant animals via
2
Division of Hematology/Oncology, Department of Medicine, University the degradation of chlorophyll to phytol, and it is primarily obtained
of Chicago, Chicago, IL, USA, 3Department of Pathology, College of in the human diet through the consumption of ruminant meat, dairy
Medicine, University of Illinois at Chicago, Chicago, IL, USA, 4Department products and certain species of fatty fish (12,13). The hypothesis that
of Pathology and Microbiology, University of Nebraska Medical Center,
Omaha, NE, USA and 5University of Chicago Comprehensive Cancer Center,
phytanic acid in meat and dairy products is associated with the risk of
Chicago, IL USA lymphoma has not been tested. The concentration of phytanic acid in
*
the serum (14,15), as well as phytanic acid intake as estimated from
To whom correspondence should be addressed. Tel: +773 702 1741; a food frequency questionnaire (16), has been positively associated
Fax: +773 702 0068; with the risk of prostate cancer. This increased risk may be due to
Email: [email protected]
the generation of reactive oxygen species during the β-oxidation of
Greater consumption of red meat, processed meat and dairy phytanic acid (17). High intake of phytanic acid may also confer a
products has been associated with an increased risk of non-Hodg- growth advantage to lymphoid neoplasms relative to normal cells, as
kin lymphoma (NHL) in several previous reports. Phytanic acid, the peroxisomal enzyme α-methylacyl-CoA racemase (AMACR),
a saturated fatty acid obtained primarily through the consump- which is critical to the metabolism of phytanic acid, has been shown
tion of ruminant meat and dairy products, may offer a potential to be overexpressed in several tumor lines including lymphoma (18).
underlying mechanism for these associations. In a population- In addition, phytanic acid is a natural ligand for the peroxisome pro-
based case–control study of 336 cases and 460 controls conducted liferator-activated receptor (PPAR)-α shown to regulate cellular pro-
in Nebraska during 1999–2002, we examined whether phytanic liferation, differentiation, apoptosis and inflammatory signaling (19);
acid-containing foods or total phytanic acid intake, estimated properties potentially involved in the development of NHL.
from a food frequency questionnaire and the published phytanic In this study, we report our findings from the first investigation of
acid values of 151 food items, were associated with increased the association between dietary intake of phytanic acid and the risk
NHL risk. Unconditional logistic regression was used to calcu- of NHL. In a secondary analysis, we evaluated possible associations
late odds ratios (ORs) and 95% confidence intervals for overall with common NHL subtypes.
NHL and the common NHL histologic subtypes. In multivariable
models, higher intakes of density-adjusted beef [ORT3 vs. T1 = 1.5
(1.1–2.2); Ptrend = 0.02], total dairy products [OR = 1.5 (1.1–2.2); Materials and methods
Ptrend  =  0.02) and milk [OR  =  1.6 (1.1–2.3); Ptrend  =  0.01] were Study population
associated with an increased risk of NHL. Intake of total phy- Detailed information on the study population and the methods for this
tanic acid was positively associated with NHL risk [OR  =  1.5 case–control study have been reported previously (20,21). Briefly, the cases
(1.0–2.1); Ptrend  =  0.04]. In analyses stratified by NHL subtype, included men and women, age 20–75 years, residing in the 66 counties of east-
greater consumption of beef was associated with an increased ern Nebraska and who had a primary diagnosis of NHL made between January
risk of diffuse large B-cell lymphoma, and greater consumption 1999 and December 2002. The NHL cases were identified by the Nebraska
of milk was associated with an increased risk of follicular lym- Lymphoma Study Group using a rapid case ascertainment system and con-
phoma (FL). Total phytanic acid intake was associated with an tacted within 2  months of diagnosis to participate in the study. Individuals
increased risk of FL and small lymphocytic lymphoma/chronic with any prior malignancy, other than cutaneous squamous cell carcinoma
lymphocytic leukemia. Our results provide support that total or basal cell carcinoma human immunodeficiency virus infection, and who
were deceased at initial contact or not mentally competent to participate were
phytanic acid and phytanic acid-containing foods may increase excluded. All cases were reviewed by an expert hematopathologist (D.D.W.)
NHL risk. and classified according to the World Health Organization (WHO) classifi-
cation of NHL (22). Of the 529 eligible cases, 387 (73%) participated in the
study. Controls were selected by random digit dialing from the same 66 coun-
ties in eastern Nebraska and frequency matched to the cases by gender and
Introduction 5-year age groups. Of the 697 eligible controls, 535 (77%) participated in the
study. Informed consent was obtained from all participants prior to the inter-
Non-Hodgkin lymphoma (NHL) represents a heterogeneous group of view. The study protocol was approved by the Institutional Review Board of
malignant neoplasms arising from the lymphoid tissue of the immune the University of Nebraska Medical Center.
system. Temporal shifts and geographic variations in the incidence Dietary assessment and estimation of phytanic acid intake
of NHL suggest a role for environmental factors in the etiology of Information on participant demographic characteristics and medical his-
this disease; however, well-characterized immunosuppressive states tory, as well as relevant lifestyle and environmental factors, were collected
such as primary immunodeficiency, human immunodeficiency virus using a structured telephone interview administered concurrently for cases
infection and iatrogenic immunosuppression do not fully account for and controls. Following completion of the telephone interview, participants
the increase in recent decades (1,2). Although factors influencing the were mailed a modified version of the Block 1995 Revision of the Health
risk of NHL in the general population remain largely unknown, the Habits and History Questionnaire (23). This 117-item quantitative food fre-
associations between dietary factors and the risk of NHL have been quency questionnaire (FFQ) assessed the frequency and the quantity of food
examined given the actions of dietary factors in immune system func- items consumed during the previous year. Items on the FFQ were the min-
tion (3). imum set that could capture at least 90% of the intake of specific nutrients
in the National Health and Nutrition Examination Survey II (24). The Health
Habits and History Questionnaire has been validated against dietary records
with correlation coefficients in the range of 0.5–0.6 for most nutrients (25).
Abbreviations:  AMACR, alpha-methylacyl-CoA racemase; CI, confidence The FFQ was completed and returned by 348 (90%) cases and 470 (88%)
interval; DLBCL, diffuse large B-cell lymphoma; FFQ, food frequency ques- controls. Participants completing the FFQ were slightly older, but were similar
tionnaire; FL, follicular lymphoma; MZL, marginal zone lymphoma; NHL, with respect to educational attainment, marital status, smoking status and body
non-Hodgkin lymphoma; OR, odds ratio; SLL/CLL, small lymphocytic mass index to participants who did not return the FFQ. In addition, 12 cases
lymphoma/chronic lymphocytic leukemia; WHO, World Health Organization. and 10 controls were excluded from this analysis for reporting implausible

© The Author 2012. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected] 170
 Phytanic acid and non-Hodgkin lymphoma

total energy intakes of <800 kcal or >6000 kcal for men and <600 kcal or Table I.  Top 10 food items contributing to total phytanic acid intake in the
>5000 kcal for women or for leaving >20% of the items blank on the FFQ. Nebraska case–control study of non-Hodgkin lymphoma, 1999–2002
In total, 336 cases and 460 controls provided complete dietary data and were
included in the present analysis. Average intake (g/ Assigned phytanic Contribution to
Dietary intake of beef, sausage, dairy products and fish was calculated from day) Mean ± SD acid content (mg/g total phytanic
items on single foods and from mixed dishes using the Health Habits and food)a acid intake (%)
History Questionnaire food composition tables (23). Following the approach
of Wright et  al. (16), phytanic acid intake was estimated from FFQ items Cheese and 11.7 0.734 18.1
on single foods and mixed dishes using the phytanic acid content values for cheese spreads
151 foods commonly used in home cooking and manufactured products (12). Hamburger/ 22.0 0.331 17.4
Daily total phytanic acid intake was computed by multiplying the amounts of cheeseburger
food items consumed (grams per day) by the reported phytanic acid content 2% milk 97.3 0.049 11.3
for each food item (milligrams per gram of food) and then summing across all Butter 2.4 1.767 10.2
food items identified to contribute phytanic acid to the diet. The phytanic acid Whole milk 42.6 0.097 9.8
content of several cream-based products and species of fish, as well as for Beef steak 12.3 0.331 9.7
sausages and hot dogs, was approximated because specific values for these Ice cream 22.9 0.110 6.0
food items were not available in the literature (12). As the phytanic acid con- Beef (roast or 3.8 0.331 3.0
tent of ruminant products is proportional to the total fat content, we assigned sandwiches)
a value for cream that was 25% of the value of that reported for double cream. Fried fish 5.4 0.211 2.7
Accordingly, low-fat dairy products were assigned a phytanic acid value half Cottage cheese 6.3 0.136 2.0
that of their whole-fat counterparts when specific low-fat values were not a
available. Sausages and hot dogs with unknown beef content were assigned Values obtained or estimated using the reported phytanic acid content for
conservative values of 15 and 50% of the phytanic acid value assigned to commonly consumed foods as reported by Brown et al. (12).
beef, respectively. As specific types of broiled or fried fish were not queried
by the FFQ, we estimated the frequency of consumption for specific types
of fish using data for non-Hispanic whites participating in the 1999–2000
National Health and Nutrition Examination Survey II (26) because the data hamburgers/cheeseburgers, 2% milk, butter, whole milk, beef steak,
from this nationally representative sample best approximated the period of ice cream, beef (roast or sandwiches), fried fish and cottage cheese
exposure assessment for the case–control study. Phytanic acid values for spe- were the top 10 food items contributing to total phytanic acid intake.
cific types of fish were then assigned using the frequency weights obtained Together, these items accounted for >90% of the total estimated
from the 1999–2000 National Health and Nutrition Examination Survey II in intake of phytanic acid. Conversely, sausage, yogurt and broiled fish
conjunction with phytanic acid values from the literature. For species of fish in each contributed <2% to total phytanic acid intake.
which no phytanic values were reported, we assigned values based on similar- The mean age at diagnosis was 58.6  years for the cases, and the
ities in the type and fat content of fish species. mean age at recruitment was 58.0  years for the controls (Table II).
Statistical analysis The majority of cases and controls reported educational attainment
Characteristics of the cases and controls were compared using chi-squared tests in excess of 16  years and were non-smokers. No statistically sig-
for categorical variables and t-tests for continuous variables. Unconditional nificant differences in age, sex, educational attainment, first-degree
logistic regression was used to calculate odds ratios (ORs) and 95% confidence family history of cancer, smoking status, alcohol consumption, body
intervals (CIs) for overall NHL and for six major subtypes of NHL according mass index, total energy intake or total estimated phytanic acid intake
to the WHO classification: follicular lymphoma (FL), diffuse large B-cell lym- were detected between cases and controls (P > 0.05). The majority of
phoma (DLBCL), small lymphocytic lymphoma/chronic lymphocytic leuke- NHL cases were of B-cell origin (>94%) and classified as SLL/CLL
mia (SLL/CLL), marginal zone lymphoma (MZL), other miscellaneous B-cell
lymphomas and T-cell lymphomas. Tertiles for food items and estimated phy-
(7.4%), FL (31.3%), DLBCL (26.5%), MZL (8.9%) or other B-cell
tanic acid levels were examined in all models using cut-points based on the lymphomas (20.2%), with only a small number of T-cell lymphomas
exposure distribution among controls with the lowest exposure group serving (5.7%).
as the referent in all models. Linear trends were tested by entering tertile medi- In models adjusting for age, sex, educational attainment and
ans as continuous variables in regression models. For all dietary exposures, total energy (kcal), there was an increased risk of overall NHL
total energy intake was adjusted for using the energy density method (27). for participants in the highest tertile of density-adjusted beef
Covariates included in the final models were age (continuous), sex, education intake [ORT3 vs. T1 = 1.5 (1.1–2.2); Ptrend = 0.02], total dairy intake
(<12 years, 12–15 years and 16+ years) and total energy intake (continuous). [OR = 1.5 (1.1–2.2); Ptrend = 0.02] and milk intake [OR = 1.6 (1.1–
Alcohol consumption, smoking status, body mass index, physical activity,
2.3); Ptrend  =  0.01] when compared with those in the lowest tertile
farming status, use of hair dye, history of blood transfusion and fruit and veg-
etable intake were examined as potential confounders, but were not included (Table III). A  positive linear trend was detected for sausage intake
in the final models as they were not found alone, or in combination, to change (Ptrend = 0.04). No association with overall NHL risk was detected for
the risk estimates by >10% (28). cheese, ice cream, yogurt, butter or total fish consumption. In analy-
We assessed whether the risk estimates for overall NHL were similar for ses stratified by NHL subtype, the positive association between beef
men and women by Wald tests of the cross-product terms for dietary factors intake and the risk of DLBCL [OR = 2.3 (1.2–4.4); Ptrend = 0.02] and
and sex, and by sex-stratified analyses. As no departures from additivity in the positive association between milk intake and FL [OR = 1.9 (1.1–
the log-odds of disease were detected for the cross-product terms, and point 3.2); Ptrend = 0.02] were statistically significant. No associations were
estimates were similar for men and women in stratified analyses, the combined detected for sausage or total dairy intake in stratified analyses, although
results are presented here. In addition, based on our a priori hypothesis that the ORs were in the direction of an increased risk for both food items
the association between phytanic acid and the risk of NHL may be modified by
factors contributing to oxidative stress, we examined associations for phytanic
across all subtypes examined. No associations were detected for any
acid in models stratified by body mass index (<26.8 versus ≥26.8, median), phytanic acid-containing food and the risk of MZL, other miscellane-
alcohol consumption (<0.3% kcal/day versus ≥0.3% kcal/day, median) and fruit ous B-cell lymphomas or T-cell lymphomas (data not shown).
and vegetable intake (<3.9 servings/day versus ≥3.9 servings/day, median). All Total phytanic acid intake, as well as phytanic acid from beef
tests were two-sided with P  <  0.05 considered statistically significant. Data sources, was positively associated with overall NHL risk (Ptrend < 0.05)
analyses were performed using SAS 9.2 statistical software (SAS Institute Inc, (Table IV). Phytanic acid from dairy products was in the direction of
Cary, NC). an increased risk for overall NHL; however, the risk estimates did not
reach statistical significance. No association was detected for phytanic
Results acid from fish. In analyses stratified by NHL subtype, total phytanic
acid was associated with an increased risk of FL [OR  =  2.0 (1.1–
A list of top 10 foods contributing to total phytanic acid intake in 3.7); Ptrend = 0.02] and SLL/CLL [OR = 3.0 (1.1–8.1); Ptrend = 0.02],
our study population, as well as their average intake and assigned phytanic acid from beef with an increased risk of DLBCL [OR = 2.3
phytanic acid value, are given in Table I. Cheese and cheese spreads, (1.2–4.5); Ptrend = 0.02] and phytanic acid from dairy products with an
171
N.J.Ollberding et al.

Table II.  Characteristics of eligible non-Hodgkin lymphoma (NHL) cases and controls in Nebraska, 1999–2002

Cases (n = 336) Controls (n = 460) P value*

Age (years), mean (SD) 58.6 (12.7) 58.0 (12.8) 0.54

Male, n (%) 185 (55.1) 236 (51.3) 0.29
Education, n (%)
 <12 years 15 (4.5) 9 (2.0)
 12–15 years 126 (37.8) 191 (41.5)
  16+ years 192 (57.7) 260 (56.5) 0.09
Positive first-degree family history of cancer, n (%)
 None 159 (47.3) 224 (48.7)
 Non-hematopoietic 140 (41.7) 196 (42.6)
 Hematopoietic 37 (11.0) 40 (8.7) 0.55
Smoking status, n (%)
 Never 169 (52.2) 221 (50.3)
 Former 108 (33.3) 140 (31.9)
 Current 47 (14.5) 78 (17.8) 0.48
Percent calories from alcohol, mean (SD) 2.2 (5.2) 2.6 (5.7) 0.31
Body mass index (kg/m2), mean (SD) 27.9 (5.2) 27.3 (5.2) 0.15
Dietary
  Total energy (kcal), mean (SD) 1984.5 (791.0) 1911.1 (706.3) 0.18
  Total phytanic acid (mg/1000 kcal/day), mean (SD) 30.7 (12.4) 29.2 (12.3) 0.08
WHO-defined NHL histologic subtypes, n (%)
 SLL/CLL 25 (7.4)
  Follicular lymphoma 105 (31.3)
  Diffuse large B cell 89 (26.5)
  Marginal zone 30 (8.9)
  Other B cell 68 (20.2)
  T-cell lymphomas 19 (5.7)

Numbers may not sum to total due to missing data.

*P value for the t-test for continuous variables and the chi-squared test for categorical variables when comparing the means and the proportions between cases
and controls.

increased risk of FL [OR = 2.0 (1.1–3.6); Ptrend = 0.02]. No association dairy fat consumption (14–16,31), phytanic acid provides a potential
was detected between phytanic acid and the risk of MZL, other mis- underlying mechanism for the positive association of dairy foods with
cellaneous B-cell lymphomas or T-cell lymphomas (data not shown). the risk of prostate cancer (32,33) and NHL (7) observed in epidemio-
The associations for total phytanic acid and phytanic acid from logical studies. In this study, cheese and butter were not associated
beef, dairy products and fish sources were similar in analyses strati- with the risk of NHL. These findings may, in part, reflect limitations in
fied at the median values of body mass index, alcohol consumption our assessment of these food items, or suggest that other dietary fac-
and fruit and vegetable intake (data not shown). tors contained in these foods may be mitigating the potential adverse
effects of phytanic acid. In addition, previous studies, in agreement
Discussion with our findings, have generally reported non-statistically significant
inverse associations between fish intake and the risk of NHL (3). The
In this population-based case–control study conducted in Nebraska, finding that no association for phytanic acid from fish with NHL risk
higher intake of phytanic acid and phytanic acid-containing foods was detected may, in part, reflect its low contribution to total phytanic
including beef, total dairy products, milk and sausage were associated acid, or that other dietary factors in fatty fish, which are highly corre-
with an increased risk of overall NHL. In analyses stratified by NHL lated with phytanic acid, may be inversely associated with NHL risk.
subtype, total phytanic acid intake was associated with an increased In the secondary analyses, total phytanic acid was associated with
risk of FL and SLL/CLL, and beef intake and milk intake were associ- an increased risk of FL and SLL/CLL. The risk of FL was strongest
ated with an increased risk of DLBCL and FL, respectively. No asso- for phytanic acid estimated from dairy products and was consistent
ciation with overall or subtype-specific NHL risk was detected for with findings in the direction of increased risk of FL for total dairy
cheese, ice cream, yogurt, butter or total fish consumption. products, milk, ice cream and butter. Phytanic acid estimated from
To the best of our knowledge, no previous study has investigated beef, as well as beef consumption, was associated with an increased
the association between phytanic acid and the risk of NHL. Consistent risk of DLBCL. Beef, sausage and dairy product intakes were also in
with our study hypothesis, we detected a positive association between the direction of an increased risk for SLL/CLL; however, the associa-
total phytanic acid levels estimated from a FFQ and the reported phy- tions did not depart from unity for any food item examined, reflect-
tanic acid content values for 151 commonly consumed food items. In ing, in part, our limited ability to detect associations for the subgroup
line with the findings for estimated phytanic acid, phytanic acid-con- analyses. NHL can originate from different pathogenetic processes
taining foods including beef, sausage, total dairy products and milk including an oncogenic drive on cell proliferation and defective apop-
were associated with an increased risk of overall NHL. These findings tosis mechanisms (34). Our finding that phytanic acid from different
are in agreement with previous reports for red meat and processed food sources was associated with different subtypes of NHL suggests
meat (4–6), total dairy products (7) and milk (7–9); however, null find- the potential for both etiologic commonality and heterogeneity across
ings have also been reported for meat (7,9–11), total dairy products NHL subtypes.
(4,5) and milk (6,10,17,29,30). Higher total phytanic acid intake, as Phytanic acid is derived from phytol, the side chain of chlorophyll,
estimated from a FFQ, has been associated with an increased risk of by microorganisms in the gut of ruminant animals (13). Phytanic acid
prostate cancer in a cohort of Finnish smokers (16) and higher plasma accumulates in adipose tissue and milk of these animals and is pri-
levels of phytanic acid have been associated with an increased risk marily obtained in the human diet through their consumption (12,35).
of prostate cancer overall (14) and among men fasting for >3 h (15). Although humans can convert free phytol into phytanic acid, they
Given the high correlations reported for plasma phytanic acid with do not accumulate significant amounts of phytanic acid as a result

172
 Phytanic acid and non-Hodgkin lymphoma

Table III.  Odds ratios and 95% confidence intervals for non-Hodgkin lymphoma according to tertiles of major food sources of phytanic acid

Controls Overall NHL Follicular lymphoma Diffuse large B-cell SLL/CLL

lymphoma

Cases OR (95% CI)a Cases OR (95% CI)a Cases OR (95% CI)a Cases OR (95% CI)a

Beef (g/1,000 kcal/day)

 <30.6 154 87 1.0 25 1.0 16 1.0 8 1.0
 30.6–52.4 153 117 1.4 (1.0–2.0) 39 1.6 (0.9–2.8) 36 2.3 (1.2–4.3) 5 0.7 (0.2–2.3)
 >52.4 153 129 1.5 (1.1–2.2) 40 1.6 (0.9–2.9) 36 2.3 (1.2–4.4) 12 1.8 (0.7–4.7)
  Ptrend* 0.02 0.10 0.02 0.20
Sausage (g/1000 kcal/day)
 0.0 166 97 1.0 33 1.0 27 1.0 7 1.0
 0.1–2.7 147 108 1.2 (0.8–1.7) 35 1.2 (0.7–2.0) 23 0.9 (0.5–1.6) 9 1.6 (0.6–4.6)
 >2.7 147 128 1.5 (1.0–2.1) 36 1.2 (0.7–2.1) 38 1.5 (0.8–2.6) 9 1.5 (0.5–4.3)
  Ptrend* 0.03 0.46 0.15 0.47
Total dairy (g/1000 kcal/day)
 <41.2 154 84 1.0 28 1.0 23 1.0 4 1.0
 41.2–109.8 153 118 1.4 (1.0–2.0) 33 1.2 (0.7–2.0) 31 1.3 (0.7–2.4) 9 2.3 (0.7–7.6)
 >109.8 153 131 1.5 (1.1–2.2) 43 1.5 (0.9–2.6) 34 1.4 (0.8–2.6) 12 3.0 (0.9–9.5)
  Ptrend* 0.02 0.10 0.24 0.07
Milk (g/1000 kcal/day)
 0.0 179 100 1.0 30 1.0 23 1.0 7 1.0
 0.1–72.4 141 104 1.3 (0.9–1.8) 30 1.3 (0.7–2.2) 31 1.6 (0.9–2.9) 5 0.9 (0.3–3.0)
 >72.4 140 129 1.6 (1.1–2.3) 44 1.9 (1.1–3.2) 34 1.8 (1.0–3.1) 13 2.3 (0.9–6.0)
  Ptrend* 0.01 0.02 0.06 0.08
Cheese (g/1000 kcal/day)
 <2.7 154 111 1.0 33 1.0 32 1.0 7 1.0
 2.7–8.7 153 119 1.1 (0.8–1.5) 38 1.1 (0.7–1.9) 28 0.9 (0.5–1.6) 9 1.5 (0.5–4.1)
 >8.7 153 103 0.9 (0.7–1.3) 33 1.0 (0.6–1.7) 28 0.9 (0.5–1.6) 9 1.5 (0.5–4.2)
  Ptrend* 0.75 0.93 0.74 0.43
Ice cream (g/1000 kcal/day)
 <3.9 154 90 1.0 30 1.0 21 1.0 6 1.0
 3.9–13.2 153 112 1.2 (0.9–1.8) 34 1.1 (0.7–1.9) 27 1.3 (0.7–2.4) 11 1.8 (0.6–5.1)
 >13.2 153 130 1.4 (1.0–2.0) 40 1.3 (0.8–2.3) 40 1.8 (1.0–3.2) 8 1.1 (0.4–3.4)
  Ptrend* 0.06 0.28 0.05 0.88
Yogurt (g/1000 kcal/day)
 0.0 260 190 1.0 64 1.0 53 1.0 11 1.0
 0.1–16.1 100 87 1.2 (0.9–1.8) 23 0.9 (0.5–1.5) 24 1.3 (0.7–2.3) 10 4.2 (1.6–10.9)
 >16.1 100 56 0.8 (0.5–1.2) 17 0.6 (0.4–1.2) 11 0.6 (0.3–1.2) 4 1.6 (0.5–5.6)
  Ptrend* 0.49 0.16 0.33 0.16
Butter (g/1000 kcal/day)
 0.0 239 171 1.0 46 1.0 52 1.0 12 1.0
 0.1–1.5 111 80 1.0 (0.7–1.4) 27 1.2 (0.7–2.1) 17 0.7 (0.4–1.2) 6 1.3 (0.5–3.6)
 >1.5 110 82 1.0 (0.7–1.4) 31 1.4 (0.9–2.4) 19 0.7 (0.4–1.3) 7 1.3 (0.5–3.5)
  Ptrend* 0.91 0.16 0.18 0.59
Total fish (g/1000 kcal/day)
 <4.7 154 121 1.0 35 1.0 30 1.0 11 1.0
 4.7–10.4 153 111 0.9 (0.6–1.3) 41 1.1 (0.7–1.9) 27 0.9 (0.5–1.5) 4 0.4 (0.1–1.2)
 >10.4 153 101 0.8 (0.6–1.2) 28 0.8 (0.4–1.4) 31 1.1 (0.6–1.9) 10 0.9 (0.3–2.2)
  Ptrend* 0.35 0.41 0.85 0.76

Cut-points based on the exposure distribution among controls. Due to the proportion of controls reporting no consumption of sausage, milk, yogurt or butter,
approximate tertiles were created by contrasting participants reporting no intake to those below and above the median intake for consumers.
a
Odds ratios estimated from unconditional logistic regression and adjusted for age (continuous), sex, education (<12 years, 12–15 years and 16+ years) and total
energy intake (continuous).
*P value for the Wald test when modeling tertile medians as a continuous variable.

of consuming plant materials. This is thought to be due to differ- our sample. Phytanic acid has also been hypothesized to increase the
ences in the digestive systems and the microbiomes of humans and risk of cancer by influencing AMACR expression. In prostate cancer
ruminant animals (35). Certain species of fatty fish, such as salmon, cell lines, phytanic acid has been shown to increase AMACR expres-
also accumulate appreciable amounts of phytanic acid; however, the sion (36), potentially allowing for more efficient metabolism of fatty
exact mechanism as to how this occurs remains unclear, but may be acids, and that blocking AMACR expression leads to the impaired
due to the ingestion of phytoplankton or related species that accu- proliferation of tumor cells (37). As AMACR has also been shown
mulate chlorophyll or chlorophyll degradation products. In humans, to be overexpressed in lymphoma cell lines (18), greater consump-
phytanic acid is first oxidized by α-oxidation to pristanic acid and tion of phytanic acid may influence the progression of lymphomas
then by multiple rounds of β-oxidation in the peroxisome to carnitine overexpressing AMACR by conferring a growth advantage to these
(13). High phytanic acid intake may increase β-oxidation and over- neoplasms relative to normal cells. However, as phytanic acid has
whelm antioxidant capacity leading to increased oxidative stress and been shown to inhibit the growth of prostate cancer cells in vitro (38),
the potential for malignant DNA transformation (16). However, the the potentially deleterious effects of phytanic acid consumption may
risk estimates for overall NHL were similar in analyses stratified by result from the buildup of downstream products in the α-oxidation
proxies of oxidative stress including body mass index, alcohol con- pathway arising from the interaction of high phytanic acid intake and
sumption and fruit and vegetable intake at their median values in high AMACR expression (39). In addition, phytanic acid is a natural

173
N.J.Ollberding et al.

Table IV.  Odds ratios and 95% confidence intervals for non-Hodgkin lymphoma according to tertiles of estimated phytanic acid intake

Controls Overall NHL Follicular lymphoma Diffuse large B-cell lymphoma SLL/CLL
a a a
Cases OR (95% CI) Cases OR (95% CI) Cases OR (95% CI) Cases OR (95% CI)a

Total phytanic acid

 <22.5 154 86 1.0 21 1.0 26 1.0 6 1.0
 22.5–33.4 153 119 1.4 (1.0–2.0) 41 2.0 (1.1–3.5) 28 1.1 (0.6–1.9) 4 0.7 (0.2–2.6)
 >33.4 153 128 1.5 (1.0–2.1) 42 2.0 (1.1–3.7) 34 1.2 (0.7–2.2) 15 3.0 (1.1–8.1)
  Ptrend* 0.04 0.02 0.45 0.02
Phytanic acid from beef
 <10.7 154 90 1.0 27 1.0 16 1.0 8 1.0
 10.7–17.8 153 113 1.3 (0.9–1.8) 36 1.4 (0.8–2.4) 36 2.3 (1.2–4.3) 5 0.7 (0.2–2.3)
 >17.8 153 130 1.5 (1.0–2.2) 41 1.6 (0.9–2.7) 36 2.3 (1.2–4.5) 12 1.8 (0.7–4.6)
  Ptrend* 0.03 0.12 0.02 0.20
Phytanic acid from dairy
 <7.1 154 86 1.0 21 1.0 27 1.0 5 1.0
 7.7–14.8 153 130 1.5 (1.1–2.1) 41 1.9 (1.1–3.4) 30 1.1 (0.6–1.9) 9 1.9 (0.6–5.7)
 >14.8 153 117 1.3 (0.9–1.9) 42 2.0 (1.1–3.6) 31 1.1 (0.6–1.9) 11 2.3 (0.8–6.9)
  Ptrend* 0.15 0.02 0.81 0.14
Phytanic acid from fish
 <0.4 154 118 1.0 34 1.0 28 1.0 10 1.0
 0.4–1.2 153 108 0.9 (0.6–1.3) 35 1.0 (0.6–1.7) 35 1.2 (0.7–2.1) 5 0.5 (0.2–1.5)
 >1.2 153 107 0.9 (0.6–1.3) 35 1.0 (0.6–1.8) 25 0.9 (0.5–1.6) 10 0.9 (0.3–2.2)
  Ptrend* 0.57 0.92 0.64 0.75

Cut-points were based on the exposure distribution among controls. Due to the proportion of controls reporting no consumption of sausage, milk, yogurt or butter,
approximate tertiles were created by contrasting participants reporting no intake to those below and above the median intake for consumers. Estimated phytanic
acid values reported in milligram per 1000 kilocalories per day).
a
Odds ratios estimated from unconditional logistic regression and adjusted for age (continuous), sex, education (<12 years, 12–15 years and 16+ years), and total
energy intake (continuous).
*P value for the Wald test when modeling tertile medians as a continuous variable.

ligand for the peroxisome proliferator-activated receptor-α and the more precise estimates of phytanic acid in the diet. Further, differ-
retinoid X receptor (39) and, therefore, may also modulate transcrip- ences in the associations between estimated phytanic acid intake and
tional activities potentially related to the development of cancer. the risk of common NHL subtypes underscores the need for pooled
This study has several strengths including the confirmation of NHL analyses of NHL subtypes to address etiologic heterogeneity. Given
diagnoses by expert pathology review, high response rates for cases the biological plausibility that phytanic acid intake may operate syn-
(73%) and controls (77%), recruitment of randomly selected, popula- ergistically with AMACR in the etiology of lymphoma, future studies
tion-based controls sampled from the same source population giving examining differences in the association between phytanic acid and
rise to the cases, use of a rapid case ascertainment system to mini- the risk of NHL by AMACR expression levels are warranted.
mize the potential for survival bias and use of a validated FFQ. There
were also limitations. First, as phytanic acid values are not currently
Funding
included in any US database, we relied on the published values for 151
commonly consumed foods in the United Kingdom (12). Therefore, American Institute for Cancer Research (99B083); National Cancer
we had to approximate values for several cream-based products, Institute (CA94770 and CA100555).
certain types of fish, sausages and hot dogs, possibly leading to an
underestimation of total phytanic acid intake (12,13). In addition,
butter consumption was probably underestimated in our sample due Acknowledgements
to the limited information on the contribution of butter from mixed
The authors thank Mr Martin Bast of the Nebraska Lymphoma Study Group
dishes. Second, we estimated phytanic acid using a single FFQ, which for coordinating the case identification, and for obtaining the physician con-
may not adequately reflect circulating concentrations and may have sents for the study cases.
allowed for some random misclassification of diet; however, previous
studies have shown strong correlations between phytanic acid meas- Conflict of Interest Statement: None declared.
ured in the serum and the consumption of phytanic acid-containing
foods (14,15,31). Third, similar to other questionnaire-based studies
of diet and disease, given the high correlations between phytanic acid References
and total fat (r = 0.40, P < 0.001) and animal fat (r = 0.64, P < 0.001)
1. Ekström-Smedby,K. (2006) Epidemiology and etiology of non-Hodgkin
in our sample, we cannot rule out the potential confounding effects lymphoma—a review. Acta Oncol., 45, 258–271.
of dietary factors highly correlated with phytanic acid intake. Fourth, 2. Alexander,D.D. et al. (2007) The non-Hodgkin lymphomas: a review of the
the small number of cases limited our ability to detect associations epidemiologic literature. Int. J. Cancer, 120 (suppl 12), 1–39.
for specific histologic subtypes of NHL. Fifth, we cannot rule out the 3. Skibola,C.F. (2007) Obesity, diet and risk of non-Hodgkin lymphoma.
possibility of residual confounding and recall bias. Cancer Epidemiol. Biomarkers Prev., 16, 392–395.
In conclusion, our findings provide support that greater con- 4. Chiu,B.C. et al. (1996) Diet and risk of non-Hodgkin lymphoma in older
sumption of total phytanic acid and phytanic acid-containing foods women. JAMA, 275, 1315–1321.
including beef, total dairy, milk and sausage may increase the risk of 5. Zhang,S. et  al. (1999) Dietary fat and protein in relation to risk of non-
NHL. Additional studies, ideally allowing for prediagnostic exposure Hodgkin’s lymphoma among women. J. Natl. Cancer Inst., 91, 1751–1758.
6. Purdue,M.P. et  al. (2004) Dietary factors and risk of non-Hodgkin lym-
assessment and the inclusion of phytanic acid measured in the serum, phoma by histologic subtype: a case-control analysis. Cancer Epidemiol.
are required to confirm these findings. In addition, the development of Biomarkers Prev., 13, 1665–1676.
a detailed food composition table providing the phytanic acid content 7. Chang,E.T. et al. (2005) Dietary factors and risk of non-hodgkin lymphoma
values for foods commonly consumed in the USA is needed to obtain in men and women. Cancer Epidemiol. Biomarkers Prev., 14, 512–520.

174
 Phytanic acid and non-Hodgkin lymphoma

8. Tavani,A. et al. (1997) Diet and risk of lymphoid neoplasms and soft tissue 25. Block,G. et al. (1990) Validation of a self-administered diet history ques-
sarcomas. Nutr. Cancer, 27, 256–260. tionnaire using multiple diet records. J. Clin. Epidemiol., 43, 1327–1335.
9. Zheng,T. et al. (2004) Diet and nutrient intakes and risk of non-Hodgkin’s 26. Centers for Disease Control and Prevention (CDC). National Center for
lymphoma in Connecticut women. Am. J. Epidemiol., 159, 454–466. Health Statistics (NCHS). (1999–2000) National Health and Nutrition
10. Ward,M.H. et al. (1994) Dietary factors and non-Hodgkin’s lymphoma in Examination Survey Data. U.S. Department of Health and Human Services,
Nebraska (United States). Cancer Causes Control, 5, 422–432. Centers for Disease Control and Prevention, Hyattsville, MD. http://www.
11. Cross,A.J. et  al. (2006) Meat and meat-mutagen intake and risk of non- cdc.gov/nchs/nhanes/nhanes1999-2000/nhanes99_00.htm.
Hodgkin lymphoma: results from a NCI-SEER case-control study. 27. Willett,W.C. et al. (1997) Adjustment for total energy intake in epidemio-
Carcinogenesis, 27, 293–297. logic studies. Am. J. Clin. Nutr., 65, 1220S–1228S.
12. Brown, P.J. et al. (1993) Diet and Refsum’s disease. The determination of 28. Mickey,R.M. et al. (1989) The impact of confounder selection criteria on
phytanic acid and phytol in certain foods and the application of this knowl- effect estimation. Am. J. Epidemiol., 129, 125–137.
edge to the choice of suitable convenience foods for patients with Refsum’s 29. De Stefani,E. et al. (1998) Tobacco, alcohol, diet and risk of non-Hodg-
disease. J. Hum. Nutr. Diet, 6(4): 295–305. kin’s lymphoma: a case-control study in Uruguay. Leuk. Res., 22, 445–452.
13. Verhoeven,N.M. et  al. (2001) Human metabolism of phytanic acid and 30. Rohrmann,S. et al. (2011) Consumption of meat and dairy and lymphoma
pristanic acid. Prog. Lipid Res., 40, 453–466. risk in the European Prospective Investigation into Cancer and Nutrition.
14. Xu,J. et al. (2005) Serum levels of phytanic acid are associated with pros- Int. J. Cancer, 128, 623–634.
tate cancer risk. Prostate, 63, 209–214. 31. Allen,N.E. et al. (2008) Phytanic acid: measurement of plasma concentra-
15. Price,A.J. et  al. (2010) Plasma phytanic acid concentration and risk of tions by gas-liquid chromatography-mass spectrometry analysis and asso-
prostate cancer: results from the European Prospective Investigation into ciations with diet and other plasma fatty acids. Br. J. Nutr., 99, 653–659.
Cancer and Nutrition. Am. J. Clin. Nutr., 91, 1769–1776. 32. Gao,X. et  al. (2005) Prospective studies of dairy product and calcium
16. Wright,M.E. et al. (2012) Estimated phytanic acid intake and prostate can- intakes and prostate cancer risk: a meta-analysis. J. Natl. Cancer Inst., 97,
cer risk: a prospective cohort study. Int. J. Cancer, 131, 1396–1406. 1768–1777.
17. Tamatani,T. et al. (1999) Neoplastic conversion of human urothelial cells 33. Qin,L.Q. et al. (2004) Milk consumption is a risk factor for prostate cancer:
in vitro by overexpression of H2O2-generating peroxisomal fatty acyl CoA meta-analysis of case-control studies. Nutr. Cancer, 48, 22–27.
oxidase. Int. J. Oncol., 15, 743–749. 34. Magrath,I.T. (1997) Introduction: concepts and controversies in lymphoid
18. Zhou,M. et  al. (2002) Alpha-Methylacyl-CoA racemase: a novel tumor neoplasia. In Magrath, I.T. (ed.) The Non-Hodgkin’s Lymphomas. Oxford
marker over-expressed in several human cancers and their precursor University Press, New York, pp. 3–46.
lesions. Am. J. Surg. Pathol., 26, 926–931. 35. Watkins,P.A. et al. (2010) Identification of differences in human and great
19. Zomer,A.W. et al. (2000) Pristanic acid and phytanic acid: naturally occur- ape phytanic acid metabolism that could influence gene expression profiles
ring ligands for the nuclear receptor peroxisome proliferator-activated and physiological functions. BMC Physiol., 10, 19.
receptor alpha. J. Lipid Res., 41, 1801–1807. 36. Mobley,J.A. et al. (2003) Branched fatty acids in dairy and beef products
20. Chiu,B.C. et al. (2005) Association of NAT and GST polymorphisms with markedly enhance alpha-methylacyl-CoA racemase expression in prostate
non-Hodgkin’s lymphoma: a population-based case-control study. Br. cancer cells in vitro. Cancer Epidemiol. Biomarkers Prev., 12, 775–783.
J. Haematol., 128, 610–615. 37. Zha,S. et al. (2003) Alpha-methylacyl-CoA racemase as an androgen-inde-
21. Chiu,B.C. et  al. (2007) Obesity and risk of non-Hodgkin lymphoma pendent growth modifier in prostate cancer. Cancer Res., 63, 7365–7376.
(United States). Cancer Causes Control, 18, 677–685. 38. Tang,X.H. et al. (2007) Cell proliferation inhibition and alterations in ret-
22. Jaffe,E.S. et al. (2001) World Health Organization Classification of Tumors. inol esterification induced by phytanic acid and docosahexaenoic acid. J.
Pathology and Genetics of Tumors of Hematopoietic and Lymphoid Tissues. Lipid Res., 48, 165–176.
IARC Press, Lyon. 39. Hellgren,L.I. (2010) Phytanic acid–an overlooked bioactive fatty acid in
23. Block,G. et al. (1994) Revision of dietary analysis software for the Health dairy fat? Ann. N. Y. Acad. Sci., 1190, 42–49.
Habits and History Questionnaire. Am. J. Epidemiol., 139, 1190–1196.
24. Block,G. et al. (1986) A data-based approach to diet questionnaire design Received June 26, 2012; revised September 12, 2012; accepted September
and testing. Am. J. Epidemiol., 124, 453–469. 28, 2012

175