Cancer Causes & Control (2018) 29:125–133

https://doi.org/10.1007/s10552-017-0984-x

BRIEF REPORT

Meat consumption and pancreatic cancer risk among men and women

in the Cancer Prevention Study-II Nutrition Cohort
Marjorie L. McCullough1 · Eric J. Jacobs1 · Roma Shah1 · Peter T. Campbell1 · Ying Wang1 · Terryl J. Hartman2 ·
Susan M. Gapstur1

Received: 12 July 2017 / Accepted: 17 November 2017 / Published online: 28 November 2017
© Springer International Publishing AG, part of Springer Nature 2017

Abstract
Purpose  Prospective cohort studies suggest that red and processed meat consumption is associated with increased risk of
pancreatic cancer among men, but not women. However, evidence is limited, and less evidence exists for other types of meat.
Methods  Cox proportional hazards regression was used to estimate multivariable-adjusted hazard ratios (HR) for the asso-
ciation of meat consumption, by type, with pancreatic cancer risk among 138,266 men and women in the Cancer Prevention
Study-II Nutrition Cohort. Diet was assessed at baseline in 1992, and 10 years earlier, at enrollment into the parent CPS-II
mortality cohort. 1,156 pancreatic cancers were verified through 2013.
Results  Red meat, processed meat, and fish intake at baseline were not associated with pancreatic cancer risk. However,
for long-term red and processed meat consumption (highest quartiles in 1982 and 1992, vs. lowest quartiles), risk appeared
different in men [hazard ratio (HR) 1.32, 95% confidence interval (CI) 0.90, 1.95] and women (HR 0.72, 95% CI 0.47, 1.10,
p heterogeneity by sex = 0.05). Poultry consumption in 1992 was associated with increased pancreatic cancer risk (HR 1.27,
95% CI 1.04, 1.55, p trend = 0.01, top vs. bottom quintile).
Conclusions  The associations of meat consumption with pancreatic cancer risk remain unclear and further research, par-
ticularly of long-term intake, is warranted.

Keywords  Diet · Meat · Red meat · Processed meat · Pancreatic cancer · Prospective cohort study

Introduction be relevant. Established risk factors for pancreatic cancer

Pancreatic cancer is the fourth leading cause of cancer death
[1], with 1- and 5-year survival rates of 29 and 7%, respec-
tively. There are no reliable methods for early detection,
and symptoms typically do not occur until the cancer has
progressed. Analyses of the genetic sequence of metastatic
pancreatic tumors suggest that the time course between the
initiating mutation and diagnosis is approximately 17 years
[2]. Causal factors may act by creating the initial genomic
alteration or by promoting progression and metastases. Thus,
exposures occurring in the 20 years prior to diagnosis may
* Marjorie L. McCullough
marji.mccullough@cancer.org
1
Epidemiology Research Program, American Cancer Society,
250 Williams St. NE, Atlanta, GA 30303‑1002, USA
2
Department of Epidemiology, Rollins School of Public
Health, Winship Cancer Institute, Emory University, Atlanta,
GA, USA
include smoking, obesity, type 2 diabetes, chronic pancrea-
titis and certain inherited genetic syndromes [1, 3]. Identi-
fication of additional risk factors and prevention strategies
is important for minimizing morbidity and mortality from
this highly fatal cancer.
In the 2012 World Cancer Research Fund/American
Institute for Cancer Research (WCRF/AICR) Continuous
Update Project specifically on pancreatic cancer [3], no
dietary factors were rated as “Convincing” or “Probable,”
but the evidence was considered “Limited-suggestive” for
red meat, processed meat, alcohol (heavier drinking), foods
and beverages containing sucrose, and saturated fatty acids.
Moreover, the 2015 International Agency for Research on
Cancer (IARC) monograph on red and processed meat noted
that positive associations between red meat and pancreatic
cancer have been observed, but data were considered limited
and inconclusive [4]. More recently, a 2016 meta-analysis
including 16 cohort studies of red meat, and 20 of processed
meat in relation to pancreatic cancer [5], reported significant

13
Vol.:(0123456789)

126
positive associations among men (RR 1.21, 95% CI 1.07,
1.37 for red and RR 1.18, 95% CI 1.06, 1.31 for processed
meat), but not among women. The relationship of other
types of meat with pancreatic cancer risk is inconsistent, but
positive associations with poultry intake have been observed
[6–13].
Using data from the Cancer Prevention Study (CPS)-II
Nutrition Cohort, a large U.S. prospective cohort study of
cancer incidence and mortality, we examined associations
of baseline and long-term red and processed meat, poultry,
and fish intake with incident pancreatic cancer, and potential
differences in associations by sex.
Materials and methods
Study participants
The CPS-II Nutrition Cohort is a subset of the parent CPS-
II mortality cohort, a study of cancer mortality including
1.2 million men and women who completed a brief question-
naire at the time of enrollment in 1982 [14]. In 1992, CPS-II
mortality cohort participants from 21 states were invited
to enroll in the CPS-II Nutrition Cohort [14]. At baseline
in 1992, CPS-II Nutrition Cohort participants completed a
mailed self-administered questionnaire providing informa-
tion on demographic, medical, dietary, and lifestyle factors.
Follow-up questionnaires to ascertain newly diagnosed can-
cers and to update exposure information were sent biennially
starting in 1997; the response rate to follow-up surveys is
consistently 80% or greater. The Emory University Insti-
tutional Review Board approves all aspects of the CPS-II
Cohorts.
Of the 184,185 men and women who returned the 1992
baseline survey, we excluded those who were lost to fol-
low-up (n = 6,206), reported a history of pancreatic cancer
(n = 42) or cancer other than non-melanoma skin cancer
(n = 22,826) at baseline, or whose pancreatic cancer was
unverifiable (n = 4), or whose diagnosis date was > 6 months
after self-report (n = 1). We also excluded participants
diagnosed with pancreatic endocrine tumors, sarcomas,
and lymphomas (histology types 8150–8155, 8240–8246)
(n = 23), as the causes of these uncommon pancreatic
tumors may differ from causes of exocrine tumors. Addi-
tionally, we excluded those who reported total daily energy
intake < 650 kcal or > 4,000 kcal (men) or < 500 kcal or
> 3,500 kcal (women) (n = 14,749) or who were missing
answers to ≥ 8 of the 14 meat questions at baseline (n = 213).
We also excluded participants who could not contribute
follow-up time to the analysis (n = 1,855) because they
died or were diagnosed with pancreatic cancer during the
first 2 years of follow-up, which were not included in this
13
Cancer Causes & Control (2018) 29:125–133
analysis to minimize reverse causation. After these exclu-
sions, a total of 138,266 participants comprised the analytic
cohort.
Of the 1,156 participants with incident pancreatic cancer
[International Classification of Disease for Oncology, Third
Edition (Code C250–C259)] in this analysis, 205 (17.7%)
cases were initially identified by self-report and subse-
quently verified by obtaining medical records (115, 56.1%),
or by linkage with state cancer registries (90, 43.9%) [14].
We additionally obtained medical records on three partici-
pants (0.3% total) via proxy report and consent. An addi-
tional 941 (81.4%) cases had pancreatic cancer listed as a
primary or contributory cause of death on their death certifi-
cate and were identified through computerized linkage with
the National Death Index [10]; additional information on
524 (55.7%) of these cases was obtained through subsequent
linkage with state cancer registries. The remaining cases (7,
0.6% total) were identified as a pancreatic cancer case during
verification of another reported cancer. Mean (SD) follow-up
was 15.7 (4.5) years.
Dietary assessment
Diet was assessed at baseline (i.e., 1992) using a validated,
modified brief Block FFQ [15]. Participants were asked to
report average frequency of consumption over the previous
year, ranging from never to two or more times per day, and
whether their typical portion was small, medium or large
compared to the medium portion size listed. Meat types
included fresh red meat (i.e., hamburgers/ground beef, steak/
roast, beef stew or pot pie, liver, and pork), processed meat
(i.e., hot dogs, lunch meats, sausage, and bacon), poultry
(i.e., chicken or turkey, fried chicken), and fish (i.e., tuna
and other fish). In 1982, participants reported dietary intake
on a brief questionnaire as part of the parent CPS-II mortal-
ity cohort. The survey asked “how many days per week do
you eat the following foods?” including the following meat
items: hamburgers, beef, pork, ham, liver; smoked meats,
sausage, fried bacon; chicken, and fish. Individuals whose
diet questionnaire was considered incomplete (missing the
entire right or left column of the survey or ≥ 24 of the 28
total line items) were excluded from the analyses using 1982
dietary data (n = 63 cases, n = 7,483 total).
Statistical analyses
Follow-up time began 2 years following completion of the
baseline survey and continued until the date of pancreatic
cancer diagnosis or pancreatic cancer death if verified only
through linkage with NDI, the date of censoring due to loss
to follow-up, other cause of death, or June 30, 2013, which-
ever came first. Individuals who self-reported pancreatic
Cancer Causes & Control (2018) 29:125–133 127
cancer that could not be verified were censored at the last
cancer-free survey.
Meat exposures were categorized using sex-specific
quintiles in 1992, and quartiles in long-term analyses due
to smaller numbers. Cox proportional hazards models
were used to evaluate multivariable-adjusted associations
of each exposure with pancreatic cancer risk. All models
adjusted for age at baseline using the stratified Cox pro-
cedure with 1-year age strata, and models including men
and women controlled for sex. We included the following
established risk factors, assessed from the baseline survey,
in all models: smoking status [never, former (≤ 5, > 5–10,
> 10–15, and > 15 years since quit), current (< 20 or ≥ 20
cigarettes/day, missing); body mass index (BMI) in kg/m2
(< 18.5, 18.5–<25, 25–<30, 30+, missing/unknown); his-
tory of diabetes (no, yes); and ethanol in g per day (0/miss-
ing, > 0–10, > 10–15, > 15); saturated fat (% kilocalories
per day, in quintiles); and total energy (kilocalories per day,
in quintiles) were also included]. We examined associations
including other types of meat in the models. We also con-
ducted substitution analyses by including total meat in the
model, where the beta reflected substitution of the exposure
of interest for other types of meat in the diet; however, none
of these additional models changed the results. Education,
race, recreational physical activity, fruits, vegetables, total
sugar, and whole grains were considered as covariates but
adjustment for these factors did not change the risk estimates
and were therefore not included.
P for trend was calculated using a continuous variable
created from medians within quantile categories. We exam-
ined heterogeneity in associations by several factors, includ-
ing sex, smoking status (never, ever), hormone replacement
therapy (never/ever, women only), </≥ median age at base-
line, median BMI (sex-specific, kg/m2), and median etha-
nol intake (sex-specific, g per day), using continuous meat
intake variables (servings per week).
Likelihood ratio tests [16] were used to test violations
of the Cox proportional hazards assumption and heteroge-
neity of associations by modeling multiplicative interac-
tions between meat and age, sex, smoking history, hormone
replacement therapy use, BMI, and alcohol. All analyses
were conducted using SAS version 9.2 (SAS Institute, Cary,
NC). p values were two-sided and considered statistically
significant if < 0.05.
Results
At baseline, women consumed red and processed meat less
often, and poultry slightly more often, than men (Table 1).
The difference in meat servings per week between the
highest and lowest quintiles varied 10- and 14-fold in men
and women, respectively. Men and women with higher
consumption had lower educational attainment, higher BMI,
and prevalence of diabetes, were more likely to be current
smokers, and were less physically active than participants
who ate less red and processed meat. On average, partici-
pants with high consumption of red and processed meat were
more likely to be heavy alcohol drinkers, to consume more
calories, and to consume a higher proportion of calories
from saturated fat. Poultry and fish intake were only slightly
lower across higher quintiles of red and processed meat con-
sumption. Consumption of fruits and vegetables, and the
average dietary glycemic index were not markedly different
across quintiles. However, glycemic load was considerably
lower with a higher meat intake.
Baseline red and processed meat consumption was not
associated with pancreatic cancer incidence among men
and women examined separately, or combined (Table 2).
In analyses of men and women combined, baseline poultry
consumption was associated with higher pancreatic cancer
risk (highest vs. the lowest quintile of consumption HR 1.27,
95% CI 1.04–1.55, p trend = 0.01); however, in sex-specific
analyses, this association was statistically significant only
among women (HR 1.45, 95% CI 1.07, 1.95, p trend = 0.03,
p heterogeneity by sex = 0.51). There was no consistent pat-
tern of association between fish intake and pancreatic cancer
risk overall or when stratified by sex. Including other types
of meat, or total meat, in the models did not change the risk
estimates (not shown).
When meat intake consistently in the top quartile 10 years
before baseline and at study baseline (relative to intake con-
sistently in the bottom quartile) was considered, no meat
types were statistically significantly related to risk of devel-
oping pancreatic cancer (Fig. 1). For red meat, processed
meat or both combined, associations among men and women
were qualitatively different, with point estimates were above
1.0 for men and below 1.0 for women (p heterogeneity by
sex for red and processed meat combined = 0.05). These
results were similar when we controlled for BMI in 1982
and change in BMI from 1982 to 1992. A positive associa-
tion with poultry intake in women only was not statistically
significant.
None of the associations with meat intake violated the
Cox Proportional Hazards assumption, and no significant
heterogeneity of associations was observed by age, smok-
ing history, hormone replacement therapy use, alcohol con-
sumption, or BMI.
Discussion
In this prospective cohort study of 138,266 U.S. adults, we
found no evidence of an association between baseline con-
sumption of unprocessed red meat or processed meat and
pancreatic cancer risk in both sexes. When we considered
13

128 Cancer Causes & Control (2018) 29:125–133

Table 1  Baseline characteristics by red and processed meat intake among 138,266 men and women in the Cancer Prevention Study II Nutrition
Cohort, 1992–2013
Red/processed meat intake quintile (servings/week)
Men Women
Q1: ≤2.4 Q3: >4.2–6.2 Q5: >9.1 Q1: ≤1.3 Q3: >2.6–3.9 Q5: >5.9

Sociodemographic characteristic
 N per category 13,060 12,905 12,965 14,767 14,641 14,681
 Mean (SD) age (years) 64.1 (6.2) 63.8 (6.0) 63.0 (5.8) 62.3 (6.6) 61.8 (6.5) 61.4 (6.5)
 Race
  White, white-Hispanic 96.9 97.7 97.7 96.9 97.7 97.6
  Black, black-Hispanic 1.0 1.1 1.2 1.4 1.2 1.5
  Unknown, other or not reported 2.0 1.2 1.0 1.7 1.1 0.9
 Education
  Less than high school 4.4 7.0 11.4 3.1 4.5 6.2
  High school graduate 12.9 18.1 26.0 24.8 31.7 37.2
  Some college, trade school 23.2 25.8 28.4 33.4 31.4 31.5
  College graduate 59.5 49.1 34.2 38.7 32.4 25.1
 Body mass index (kg/m2)
  <18.5 0.5 0.4 0.4 2.8 1.4 1.1
  18.5–<25 47.6 34.4 25.6 61.6 50.9 37.4
  25–<30 43.2 50.9 50.9 24.9 31.9 35.5
 ≥30 7.5 13.1 21.5 9.1 14.2 24.3
 Missing 1.3 1.3 1.6 1.5 1.5 1.6
  History of diabetes (% yes) 7.2 8.9 12.0 5.2 5.7 8.7
 Smoking status
  Never 37.6 33.3 27.5 52.9 55.0 56.0
  Current, < 20 cigarettes/day 2.3 3.9 5.5 3.4 4.6 5.8
  Current, ≥ 20 cigarettes/day 1.5 4.2 9.1 1.8 3.5 5.9
  Former, ≤ 5 years since quit 4.2 6.1 7.8 4.3 5.1 5.3
  Former, > 5–10 years since quit 5.1 6.2 6.9 4.6 4.6 4.1
  Former, > 10–15 years since quit 6.2 6.5 6.9 4.9 4.3 3.5
  Former, > 15 years since quit 40.7 37.7 34.1 25.7 20.7 17.5
  Missing 2.3 2.1 2.3 2.4 2.3 1.9
 Physical activity (MET-h/week) a
  0 7.0 11.2 17.9 6.3 8.5 12.7
  >0–<17.5 56.4 58.7 54.3 60.7 66.4 64.7
  ≥17.5 35.9 29.1 26.6 32.0 24.2 21.6
  Missing 0.7 1.0 1.2 1.0 0.9 1.0
Dietary characteristic
 Ethanol intake (g/day)
  0 37.0 33.1 34.8 50.0 45.8 49.3
  >0–10 35.3 33.6 31.6 36.7 37.7 34.4
  >10–15 9.4 9.2 8.8 6.8 7.5 6.7
  >15 18.3 24.1 24.8 6.6 9.0 9.6
Mean (SD)

Energy intake (kcal/day) 1,403.9 (458.4) 1,730.4 (490.6) 2,377.5 (608.9) 1,115.3 (382.3) 1,305.9 (398.9) 1,752.2 (497.8)
Red and processed meat intake 1.2 (0.8) 5.2 (0.6) 12.9 (3.9) 0.6 (0.4) 3.2 (0.4) 8.7 (2.9)
(servings/week)
Red meat (servings/week) 0.9 (0.6) 3.1 (1.1) 6.6 (3.1) 0.5 (0.4) 2.2 (0.7) 5.1 (2.4)
Processed meat (servings/week) 0.3 (0.4) 2.0 (1.1) 6.3 (3.6) 0.1 (0.2) 1.0 (0.7) 3.5 (2.4)
Poultry intake (servings/week) 2.4 (2.2) 2.0 (1.6) 2.2 (1.8) 2.3 (2.0) 2.0 (1.6) 2.2 (1.7)

13
Cancer Causes & Control (2018) 29:125–133 129
Table 1  (continued)
Mean (SD)
Fish intake (servings/week) 2.0 (2.0) 1.6 (1.4)
Saturated fat (as % of energy) 8.6 (3.0) 11.8 (2.7)
Fruit intake (servings/day) 1.5 (1.2) 1.2 (1.0)
Vegetable intake (servings/day) 2.1 (1.4) 1.9 (1.1)
Energy-adjusted glycemic index 76.3 (9.3) 75.2 (8.3)
Energy-adjusted glycemic load 164.1 (34.9) 142.0 (28.9)
All values are percentages unless otherwise indicated; some percentages do not add up to 100% due to missing data or rounding
MET metabolic equivalent, SD standard deviation
a
 Metabolic equivalents (METS) are defined for each type of exercise-related physical activity as a multiple of metabolic equivalent of sitting
quietly for 1 h
red and processed meat intake at baseline and 10 years ear-
lier, qualitative differences in risk estimates by sex were
observed, but associations did not reach statistical signifi-
cance. For poultry intake, men and women in the highest
vs. lowest quintile at baseline had a statistically significant
27% higher risk of pancreatic cancer; this association was
stronger in women than men. Neither baseline nor long-term
fish consumption was associated with pancreatic cancer risk.
Red and processed meats are considered probable and
definite carcinogens, respectively, by IARC based on a
review of mechanistic and human evidence for colorectal
cancer [4]. However, the evidence for a positive associa-
tion of red and processed meat with pancreatic cancer risk
was considered inconclusive [4]. Similarly, the WCRF/
AICR concluded that the evidence was “limited/suggestive”
[3]. Our overall findings do not support an association of
red and processed meat consumption and pancreatic can-
cer risk. The direction of associations in men and women
observed for long-term red and processed meat consumption
(p heterogeneity by sex = 0.05) was in directions consistent
with prior literature, although numbers were limited in this
analysis and results were not statistically significant. Two
previous meta-analyses reported significant heterogeneity
by sex in the association of red or processed meat and pan-
creatic cancer risk [5, 17]. A 2016 meta-analysis [5] of 16
prospective cohort studies reported statistically significant
increased pancreatic cancer risk for “high” versus “low”
red meat consumption among men (RR 1.21, 95% CI 1.07,
1.37) but not women (RR 1.06, 95% CI 0.85,1.31); similar
findings were reported for processed meat among men and
women from 20 cohorts (RR 1.18, 95% CI 1.06, 1.31 and
RR 0.99, 95% CI 0.84, 1.16, respectively). Although four
of the studies included in these meta-analyses were of dif-
ferent meat exposures from the NIH-AARP cohort [10, 13,
18, 19], with different periods of follow-up, an earlier meta-
analysis including only one publication from NIH-AARP
also observed a statistically significant positive association
in men, but not women [17].
1.5 (1.4) 1.7 (1.8) 1.4 (1.4) 1.5 (1.4)
14.1 (2.7) 8.0 (3.0) 10.7 (3.0) 13.2 (2.9)
1.2 (1.0) 1.5 (1.1) 1.3 (1.0) 1.2 (0.9)
2.1 (1.2) 2.3 (1.5) 2.0 (1.2) 2.2 (1.2)
76.0 (8.3) 74.2 (8.7) 74.2 (7.4) 74.9 (7.1)
131.8 (27.4) 127.6 (24.3) 112.9 (20.9) 102.2 (18.9)
A potential role of red and processed meat consumption
in pancreatic cancer etiology is plausible because 95% of
pancreatic cancers arise in exocrine cells, which produce
enzymes for digestion [1]. Fat and protein in the diet stimu-
late the release of cholecystokinin in the duodenum, which
in turn stimulates the release of pancreatic enzymes [20]. In
animal experiments, a high-fat diet causes overproduction of
pancreatic juices, leading to pancreatitis [20], an established
cause of pancreatic adenocarcinoma [21]. Other potential
mechanisms for meat intake may involve DNA-damaging
heterocyclic amines and polycyclic aromatic hydrocarbons
in cooked meat, heme iron, and exogenously or endog-
enously formed nitrosamines [19, 22].
The reasons for qualitatively different associations by sex
for long-term red and processed meat intake with pancre-
atic cancer risk observed in this and other studies [5, 17]
are unclear. In CPS-II, men reported consuming more red
and processed meat compared to women, but it is unknown
whether relevant thresholds of exposure exist. Differential
lifetime iron stores in men and women, which influences free
serum iron levels and percent transferrin saturation, were
previously suggested as a potential explanation for increased
risk in men but not women in the NIH-AARP cohort [10].
Higher serum iron and iron saturation were associated
with subsequent pancreatic cancer risk in an exploratory
case–control study of San Francisco Bay area residents who
received health check-ups [23]; additional research is needed
to confirm, or refute, a sex difference in this association,
and to identify potential underlying mechanisms if it exists.
In the current study, high poultry intake (i.e., a median
of 5.3 servings/week) was associated with a 27% increased
risk of pancreatic cancer as compared to low intake (i.e.,
a median of 0.6 servings/week), which appeared to be
driven mainly by a 45% increased risk of pancreatic cancer
among women (p heterogeneity by sex = 0.51). While this
finding was unexpected, it is consistent with the published
literature. Of eight prospective studies that examined the
associations of poultry, chicken and/or “white meat” with
13


Table 2  Multivariable-adjusted hazard ratios and 95% confidence intervals for incident pancreatic cancer by 1992 meat intake among men and women in the Cancer Prevention Study II Nutri-
130

tion Cohort, 1992–2013

Meat intake quintile Combined Men Women

13
(servings/week) a a
Cases Person-years HR (95% CI) Median (serv- Cases Person-years HR (95% CI) Median (serv- Cases Person-years HR (95% CI)a
ings/week) ings/week)

Red & processed m­ eatb

 ≤2.4/1.3 261 442,410 1.00 (Referent) 1.3 139 198,949 1.00 (Referent) 0.7 122 243,461 1.00 (Referent)
 >2.4/1.3–4.2/2.6 236 437,309 0.90 (0.75, 1.08) 3.3 128 195,111 0.91 (0.71, 1.17) 2.0 108 242,198 0.88 (0.68, 1.15)
 >4.2/2.6–6.2/3.9 219 433,827 0.84 (0.70, 1.02) 5.2 112 192,323 0.81 (0.61, 1.06) 3.2 107 241,504 0.89 (0.67, 1.17)
 >6.2/3.9–9.1/5.9 232 433,236 0.89 (0.72, 1.09) 7.5 123 192,184 0.87 (0.65, 1.15) 4.7 109 241,052 0.91 (0.68, 1.22)
 >9.1/5.9 208 426,857 0.81 (0.64, 1.02) 11.8 116 189,596 0.82 (0.58, 1.14) 7.8 92 237,260 0.80 (0.57, 1.12)
 p ­trendc 0.21 0.32 0.28
Red ­meatb
 ≤1.4/0.9 275 452,683 1.00 (Referent) 0.8 147 205,474 1.00 (Referent) 0.5 128 247,209 1.00 (Referent)
 >1.4/0.9–2.3/1.6 217 429,310 0.84 (0.70, 1.00) 1.9 103 184,402 0.77 (0.60, 1.00) 1.3 114 244,908 0.90 (0.69, 1.16)
 >2.3/1.6–3.5/2.5 216 429,572 0.83 (0.69, 1.00) 2.9 123 196,728 0.87 (0.68, 1.12) 2.0 93 232,844 0.78 (0.59, 1.03)
 >3.5/2.5–5.2/3.8 238 442,927 0.91 (0.75, 1.10) 4.2 128 192,926 0.93 (0.72, 1.21) 3.1 110 250,001 0.89 (0.67, 1.17)
 >5.2/3.8 210 419,148 0.86 (0.69, 1.07) 6.7 117 188,635 0.87 (0.65, 1.18) 5.1 93 230,514 0.84 (0.62, 1.15)
 p ­trendc 0.56 0.79 0.40
Processed ­meatb
 ≤0.5/0.1 263 463,861 1.00 (Referent) 0 140 200,956 1.00 (Referent) 0 123 262,905 1.00 (Referent)
 >0.5/0.1–1.3/0.6 238 450,254 0.91 (0.76, 1.09) 0.9 122 193,014 0.88 (0.68, 1.13) 0.4 116 257,240 0.94 (0.73, 1.22)
 >1.3/0.6–2.4/1.2 226 415,596 0.91 (0.76, 1.10) 1.8 124 195,193 0.87 (0.67, 1.12) 0.9 102 220,403 0.97 (0.74, 1.28)
 >2.4/1.2–4.2/2.2 222 420,271 0.89 (0.73, 1.08) 3.2 121 190,983 0.85 (0.64, 1.12) 1.6 101 229,289 0.93 (0.70, 1.23)
 >4.2/2.2 207 423,658 0.82 (0.66, 1.02) 6.2 111 188,020 0.78 (0.58, 1.06) 3.5 96 235,638 0.86 (0.63, 1.17)
 p ­trendc 0.14 0.18 0.36
Poultryb
 ≤0.8/0.8 244 471,200 1.00 (Referent) 0.6 134 203,684 1.00 (Referent) 0.5 110 267,516 1.00 (Referent)
 >0.8/0.8–1.2/1.2 204 422,255 0.94 (0.78, 1.14) 1.0 98 189,020 0.79 (0.61, 1.03) 1.0 106 233,235 1.15 (0.88, 1.50)
 >1.2/1.2–2.0/2.0 261 489,108 1.05 (0.88, 1.25) 2.0 143 217,149 1.01 (0.79, 1.28) 2.0 118 271,960 1.12 (0.86, 1.46)
 >2.0/2.0–3.5/3.5 252 481,673 1.05 (0.87, 1.26) 3.0 134 211,944 0.97 (0.76, 1.24) 3.5 118 269,729 1.17 (0.89, 1.53)
 >3.5/3.5 195 309,403 1.27 (1.04, 1.55) 5.3 109 146,367 1.16 (0.89, 1.52) 5.3 86 163,036 1.45 (1.07, 1.95)
 p ­trendc 0.01 0.06 0.03
Fishb
 ≤0.6/0.5 242 490,795 1.00 (Referent) 0.2 135 241,326 1.00 (Referent) 0.2 107 249,470 1.00 (Referent)
 >0.6/0.5–1.0/0.9 198 395,984 1.04 (0.86, 1.25) 0.9 102 165,512 1.09 (0.85, 1.42) 0.7 96 230,472 0.97 (0.74, 1.29)
 >1.0/0.9–1.6/1.4 267 430,572 1.28 (1.07, 1.53) 1.4 137 189,278 1.29 (1.01, 1.64) 1.2 130 241,295 1.28 (0.99, 1.65)
 >1.6/1.4–2.6/2.2 225 435,599 1.07 (0.89, 1.29) 2.0 110 181,974 1.07 (0.83, 1.38) 1.8 115 253,625 1.08 (0.83, 1.42)
Cancer Causes & Control (2018) 29:125–133
 Cancer Causes & Control (2018) 29:125–133 131

pancreatic cancer risk [6, 7, 9–13, 24], seven point estimates

 Multivariable model adjusted for age in 1992, gender (combined models only), 1992 body mass index, 1992 history of diabetes, 1992 smoking status, 1992 ethanol intake, 1992 energy intake
0.94 (0.70, 1.26)
were positive [6–13], and two of these were statistically sig-
HR (95% CI)a nificant [11, 13]. Among men and women in the European
Prospective Investigation of Cancer, Rorhmann et al. [11]

0.60
reported a statistically significant 24% higher risk of pan-
creatic cancer per 50 g per day increase in poultry intake
Person-years

(about 2 oz), which strengthened (to 72%) after calibrat-

ing intake estimates across countries using 24-h diet recall
230,613

data in subsets of each country (this corrects for systematic

under- and over-estimates of meat intake). The most recent
analysis from the U.S. NIH-AARP cohort reported a statisti-
Median (serv- Cases

cally significant 33% higher risk of pancreatic cancer among

90

women in the top vs. bottom quintile of poultry consump-

tion; the association in men was null [13]. In contrast, among
ings/week)

women in a Swedish cohort with 172 cases over a 17-year

Women

 Meat intake quintiles in all models use gender-specific cut-points (men/women); p value for heterogeneity by gender > 0.05 for all meat types

follow-up, poultry consumption of 1/2 serving or more per

3.2

week compared to none was associated with a statistically

significant 64% lower risk of pancreatic cancer [8]. Finally,
1.24 (0.97, 1.59)

the Multiethnic Cohort with 482 cases observed no asso-

a
HR (95% CI)

ciation between poultry consumption and risk of pancreatic

cancer in both sexes combined [24].
0.15

A possible, though speculative, explanation for the posi-

tive association of poultry consumption with pancreatic can-
cer risk is potentially carcinogenic contaminants in poultry
Person-years

feed, medications, or livestock conditions during the years

190,075

of this study (1982–2013). For example, arsenic-containing

Roxarsone, a medication used to control intestinal diseases
in poultry was discontinued in 2011 after a study detected
 p values for trend calculated using the median meat intake in each quintile, specific to gender
Median (serv- Cases

low doses of inorganic arsenic in the livers of chickens

134

receiving the drug [25]. Higher inorganic arsenic exposure

was recently documented in the urine of humans in the high-
ings/week)

est quartile of poultry consumption in the National Health

and Nutrition Examination Survey 2003–2010, compared to
Men

(continuous), and 1992 saturated fat [as percentage of energy] (continuous)

3.6

non-consumers [26]. An ecological study in Florida found

that pancreatic cancer cases living within 1 mile of known
1.09 (0.90, 1.32)

arsenic-contaminated drinking water wells had a twofold

a
HR (95% CI)

increase in pancreatic cancer risk [27]. As arsenic is ubiq-

uitous in water, air, and soil [28], the relative importance
0.49

of poultry consumption is unclear. In addition, although

inorganic arsenic is an established cause of cancers of the
lung, urinary bladder and skin [29], neither a 2012 IARC
Person-years

monograph on arsenic [29] nor a meta-analysis of occupa-

420,688

tional exposures [30] identified arsenic as a potential cause

CI confidence interval, HR hazard ratio

of pancreatic cancer.
Combined

The current study is among the largest prospective

Cases

cohorts of both men and women to examine the relation-

224

ship between meat intake and incident pancreatic cancer.

Other strengths include detailed adjustment for important
Table 2  (continued)
Meat intake quintile

pancreatic cancer risk factors, and data on long-term diet,

(servings/week)

potentially relevant given the long time-frame for pancreatic

carcinogenesis [2]. However, measurement error from self-
 >2.6/2.2
 p ­trendc

reported dietary intake is likely; in most cases, this would

attenuate associations with pancreatic cancer.
b
a

13

132 Cancer Causes & Control (2018) 29:125–133

Meat Type No. Cases HR (95% CI)

62 1.34 (0.91, 1.97)
Red/Processed Meat (*)
39 0.70 (0.46, 1.07)

63 1.19 (0.81, 1.73)

Red Meat
40 0.76 (0.51, 1.15)

54 1.12 (0.76, 1.65)

Processed Meat
41 0.83 (0.55, 1.24)

32 0.96 (0.61, 1.51)

Poultry
50 1.40 (0.90, 2.19)

42 0.76 (0.49, 1.18)

Fish
51 0.72 (0.48, 1.08)

0.1 1 10
HR (95% CI) Men Women

Fig. 1  Multivariable-adjusted hazard ratios (HR) and 95% Confi-
dence Intervals (CI) for incident pancreatic cancer among men (filled
diamond) and women (filled circle) with consistently high versus low
long-term (1982 and 1992) Meat Intake, Cancer Prevention Study
II Nutrition Cohort, 1992–2013. Results presented are for highest
sex-specific quartile in both 1982 and 1992 compared to the refer-
ence category of lowest quartile in both time periods. Multivariable
models adjusted for age in 1992, history of diabetes in 1992, smok-
ing status in 1992, ethanol intake in 1992, energy intake and % satu-
rated fat intake in 1992, and BMI in 1992. Numbers of cases provided
are those with consistently high meat intake. For the reference group
(quartile 1  at both time points), case numbers were as follows: red/
processed meat (n = 75 men, n = 81 women); red meat (n = 69 men,
n = 84 women); processed meat (n = 75 men, n = 84 women); poultry
(n = 50 men, n = 34 women); fish (n = 42 men, n = 48 women). *p het-
erogeneity by sex = 0.05

In conclusion, these findings do not support a delete-
rious effect of red and processed meat consumption on
pancreatic cancer risk; however, we cannot rule out an
association in men as suggested by previous studies. The
positive association observed with higher poultry con-
sumption deserves further study.
Acknowledgments  The authors express sincere appreciation to all
Cancer Prevention Study-II participants, and to each member of the
study and biospecimen management group. The authors would like to
acknowledge the contribution to this study from central cancer regis-
tries supported through the Centers for Disease Control and Preven-
tion’s National Program of Cancer Registries and cancer registries
supported by the National Cancer Institute’s Surveillance Epidemiol-
ogy and End Results Program. The American Cancer Society supports
the maintenance and follow-up of the Cancer Prevention Studies. The
American Cancer Society also supported all investigators during the
conduct of this analysis.
Compliance with ethical standards  systematic review and meta-analysis. Clin Gastroenterol Hepatol
Conflict of interest  The authors declare that they have no conflict of
interest.
Disclaimer  The views expressed here are those of the authors and do
not necessarily represent the American Cancer Society or the American
Cancer Society—Cancer Action Network.
References
1. American Cancer Society (2017) Cancer facts & figures, 2017.
American Cancer Society, Atlanta
2. Yachida S, Jones S, Bozic I et al (2010) Distant metastasis occurs
late during the genetic evolution of pancreatic cancer. Nature
467:1114–1117
3. WCRF/AICR (2012) Continuous update project report: diet, nutri-
tion, physical activity, and pancreatic cancer 2012. WCRF/AICR,
Washington DC
4. Bouvard V, Loomis D, Guyton KZ et al (2015) Carcinogenicity of
consumption of red and processed meat. Lancet 16(16):1599–1600
5. Zhao Z, Yin Z, Pu Z, Zhao Q (2016) Association between con-
sumption of red and processed meat and pancreatic cancer risk-a
15(4):486–493
6. Zheng W, McLaughlin JK, Gridley G et  al (1993) A cohort
study of smoking, alcohol consumption, and dietary factors

13
Cancer Causes & Control (2018) 29:125–133 133
for pancreatic cancer (United States). Cancer Causes Control
4:477–482
7. Stolzenberg-Solomon RZ, Pietinen P, Taylor PR, Virtamo J,
Albanes D (2002) Prospective study of diet and pancreatic cancer
in male smokers. Am J Epidemiol 155:783–792
8. Larsson SC, Hakanson N, Permert J, Wolk A (2006) Meat, fish,
poultry and egg consumption in relation to risk of pancreatic can-
cer: a prospective study. Int J Cancer 118:2866–2870
9. Lin Y, Kikuchi S, Tamakoshi A et al (2006) Dietary habits and
pancreatic cancer risk in a cohort of middle-aged and elderly Japa-
nese. Nutr Cancer 56:40–49
10. Stolzenberg-Solomon RZ, Cross AJ, Silverman DT et al (2007)
Meat and meat-mutagen intake and pancreatic cancer risk
in the NIH-AARP cohort. Cancer Epidemiol Biomark Prev
16:2664–2675
11. Rohrmann S, Linseisen J, Nothlings U et al (2013) Meat and fish
consumption and risk of pancreatic cancer: results from the Euro-
pean Prospective Investigation into Cancer and Nutrition. Int J
Cancer 132:617–624
12. Ghorbani Z, Pourshams A, Fazeltabar Malekshah A, Sharafkhah
M, Poustchi H, Hekmatdoost A (2016) Major dietary protein
sources in relation to pancreatic cancer: a large prospective study.
Arch Iran Med 19:248–256
13. Taunk P, Hecht E, Stolzenberg-Solomon R (2016) Are meat and
heme iron intake associated with pancreatic cancer? Results
from the NIH-AARP diet and health cohort. Int J Cancer
138:2172–2189
14. Calle EE, Rodriguez C, Jacobs EJ et al (2002) The American
Cancer Society Cancer Prevention Study II Nutrition Cohort—
rationale, study design, and baseline characteristics. Cancer
94:2490–2501
15. Flagg EW, Coates RJ, Calle EE, Potischman N, Thun MJ (2000)
Validation of the American Cancer Society Cancer Prevention
Study II Nutrition Survey Cohort Food Frequency Questionnaire.
Epidemiology 11:462–468
16. Kleinbaum DG, Kupper LL, Muller KE (1988) Applied regres-
sion analysis and other multivariable methods. Duxbury Press,
Belmont
17. Larsson SC, Wolk A (2012) Red and processed meat consump-
tion and risk of pancreatic cancer: meta-analysis of prospective
studies. Br J Cancer 106:603–607
18. Jiao L, Stolzenberg-Solomon R, Zimmerman TP et al (2015)
Dietary consumption of advanced glycation end products and
pancreatic cancer in the prospective NIH-AARP Diet and Health
Study. Am J Clin Nutr 101:126–134
19. Aschebrook-Kilfoy B, Cross AJ, Stolzenberg-Solomon RZ et al
(2011) Pancreatic cancer and exposure to dietary nitrate and
nitrite in the NIH-AARP Diet and Health Study. Am J Epidemiol
174:305–315
20. Matters GL, Cooper TK, McGovern CO et al (2014) Cholecys-
tokinin mediates progression and metastasis of pancreatic cancer
associated with dietary fat. Dig Dis Sci 59:1180–1191
21. Ryan DP, Hong TS, Bardeesy N (2014) Pancreatic adenocarci-
noma. New Engl J Med 371:2140–2141
22. Anderson KE, Mongin SJ, Sinha R et al (2012) Pancreatic can-
cer risk: associations with meat-derived carcinogen intake in the
Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial
(PLCO) cohort. Mol Carcinog 51: 128–137
23. Friedman GD, van den Eeden SK (1993) Risk factors for pancre-
atic cancer: an exploratory study. Int J Epidemiol 22:30–37
24. Nothlings U, Wilkens LR, Murphy SP, Hankin JH, Henderson
BE, Kolonel LN (2005) Meat and fat intake as risk factors for
pancreatic cancer: the multiethnic cohort study. J Natl Cancer Inst
97:1458–1465
25. Food US and Drug Administration (2015) Arsenic-based animal
drugs and poultry. U.S. Department of Health and Human Ser-
vices, New York
26. Nigra AE, Nachman KE, Love DC, Grau-Perez M, Navas-Acien
A (2017) Poultry consumption and arsenic exposure in the U.S.
Population. Environ Health Perspect 125:370–377
27. Liu-Mares W, Mackinnon JA, Sherman R et al (2013) Pancreatic
cancer clusters and arsenic-contaminated drinking water wells in
Florida. BMC Cancer 13:111
28. Nachman KE, Ginsberg GL, Miller MD, Murray CJ, Nigra AE,
Pendergrast CB. (2017) Mitigating dietary arsenic exposure:
Current status in the United States and recommendations for an
improved path forward. Sci Total Environ 581–582:221–236
29. IARC (2012) Arsenic, metals, fibres, and dusts. IARC, Lyon
30. Ojajarvi IA, Partanen TJ, Ahlbom A et al (2000) Occupational
exposures and pancreatic cancer: a meta-analysis. Occup Environ
Med 57:316–324
13