meat intake and cooking techniques: associations with pancreatic cancer

Mutation Research 506–507 (2002) 225–231

Meat intake and cooking techniques: associationswith pancreatic cancer

Kristin E. Andersona,∗, Rashmi Sinhab, Martin Kulldorff c, Myron Grossa,Nicholas P. Langd, Cheryl Barbera, Lisa Harnacka, Eugene DiMagnoe,

Robin Blissa, Fred F. Kadlubarfa University of Minnesota, Minneapolis, MN 55454, USAb National Cancer Institute, Rockville, MD 20852, USA

c School of Medicine, University of Connecticut, Farmington, CT 06030, USAd University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA

e Mayo Clinic, Rochester, MN 55905, USAf National Center for Toxicological Research, Jefferson, AR 72079, USA

Received 18 January 2002; received in revised form 25 May 2002; accepted 2 June 2002

Abstract

Heterocyclic amines (HCAs), and polycyclic aromatic hydrocarbons (PAHs), formed in temperature and time-dependentmanners during cooking of meat, may increase the risk of certain cancers. As these compounds could be carcinogenic forthe pancreas, we assessed meat intake, preparation methods, and doneness preferences as risk factors for exocrine pancreaticcancer.

In a case-control study (cases= 193, controls= 674), subjects provided information on their usual meat intake and howit was cooked, e.g. fried, grilled or barbecued (BBQ), etc. Meat doneness preferences were measured using photographs thatshowed internal doneness and external brownness with a numerical scale. Data were analyzed with unconditional logisticregression. Odds ratios (ORs) increased with increased intake of grilled/BBQ red meat in an analysis adjusted for age, sex,smoking, education, race, and diabetes. Based on amount of BBQ meat consumed, the OR and 95% confidence interval (CI) forthe fifth quintile relative to the reference group (quintiles 1 and 2) was 2.19 (1.4, 3.4). Findings were not substantively changedby further adjustment for calories, total fat, fruit and vegetables, or alcohol consumption (from a food frequency questionnaire(FFQ)). Other meat variables did not show statistically significant associations with risk nor did they substantively alter thefindings for BBQ. These included total meat, processed meat, total red meat, total white meat, total broiled meat, total friedmeat, or total meat cooked by means other than grilling. We conclude that grilled red meat intake is a risk factor for pancreaticcancer and that method of meat preparation in addition to total intake is important in assessing the effects of meat consumptionin epidemiologic studies.© 2002 Elsevier Science B.V. All rights reserved.

Keywords: Epidemiology; Pancreas cancer; Meat intake; Cooking practices

∗ Corresponding author. Tel.:+1-612-626-8568;fax: +1-612-624-0315.E-mail address: [email protected] (K.E. Anderson).

1. Introduction

Pancreatic cancer is rapidly fatal in the majority ofcases. Currently, there are no screening tests for early

0027-5107/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved.PII: S0027-5107(02)00169-0

226 K.E. Anderson et al. / Mutation Research 506–507 (2002) 225–231

detection and about 90% of cases are diagnosed withmetastatic disease. There are very few therapeutic op-tions, thus prevention of the disease, through identifi-cation and modification of risk factors, is a potentialmeans to reduce mortality from this cancer.

Most, but not all epidemiologic studies of pancreascancer have reported increased risks associated withmeat intake or correlated variables such as protein andfat (reviewed by Anderson et al.[1] and Howe andBurch [2]) [3,4].

Numerous potential carcinogens are formed inmeat cooked at high temperatures[5–7] includingheterocyclic amines (HCAs) and polycyclic aromatichydrocarbons (PAHs). Formation of these compoundsdepends on cooking method and temperature and thedegree of doneness[8–12]. While baked and stewedmeats do not contain these compounds, well-donebarbecued (BBQ) and pan-fried meats typically con-tain high levels[12].

Few studies have specifically assessed meat cook-ing methods, as well as meat intake, with respect torisk of pancreatic cancer. We undertook to assess meatintake, cooking methods, and doneness preferencesin a population-based case-control study of pancreascancer.

2. Methods

A case-control study of cancer of the exocrinepancreas was conducted using incident cases diag-nosed between April 1994 and September 1998. Thesource cohort—the target population—was The Up-per Midwest of the United States. Incident cases ofpancreatic cancer were recruited from two sources: allhospitals in the Twin Cities seven-county metropoli-tan area of Minneapolis and St. Paul, MN, USA;and the Mayo Clinic (where cases were restrictedto those who lived in The Upper Midwest). TheInstitutional Review Boards of the University ofMinnesota and the Mayo Clinic approved the study,and informed consent was obtained from all sub-jects. Investigators were notified of potential casesthrough tumor registrars at local hospitals and TheUpper Midwest Oncology Registry System (TU-MORS). Cases from the Twin Cities were contactedfor the study initially through two NCI-funded pro-grams, the Community Clinical Oncology Program

and the Community Group Outreach Program andsubsequently through the Minnesota Cancer Surveil-lance System (in the final phase of the study). MayoClinic cases were recruited from patients attendinggastroenterology clinics from October 1996 throughSeptember 1998. Of 460 cases deemed eligible forthe study, 258 participated; 79 cases refused, 85 casesdied either before contact or before interview; 31cases were not invited to participate because theirphysicians would not allow it; and 7 cases were notapproached prior to the end of the study. This consti-tutes a participation rate among those notified to studyinvestigators of 56%. Through the use of recordsfrom The Minnesota Cancer Surveillance System, apathology-based cancer registry, we were able to es-timate our case finding rate for the Twin Cities casesat approximately 56%, allowing an estimate of totalparticipation for eligible cases of approximately 30%.

2.1. Diagnostic criteria

Cases were subjects with cancer of the exocrine pan-creas confirmed by histology or cytology or subjectswith clinical symptoms consistent with pancreas can-cer plus a finding of the “double duct sign” with endo-scopic retrograde cholangiopancreatography (ERCP)or findings by computed tomography (CT) of a mass inthe pancreas with evidence of liver metastases. Prior toanalysis, all pathology reports of potential cases werereviewed by a pathologist or confirmed through theMinnesota Cancer Surveillance System for eligibility.All cases had to be 20 years of age or older, Englishspeaking and mentally competent.

2.2. Controls

Controls were selected from drivers’ license listsfor individuals 20–64 years of age and from USHealth Care Financing Administration records forthose aged 65 years and above using stratified ran-dom sampling from the seven-county metropoli-tan area. Controls were frequency matched by age,sex, and race with the cases. To be eligible for thestudy they needed to speak English and be men-tally competent and have no history of pancreaticcancer. Of 1141 potentially eligible controls, 676were enrolled in the study yielding a response rateof 59%.

K.E. Anderson et al. / Mutation Research 506–507 (2002) 225–231 227

2.3. Survey and sample collection

In-person interviews were conducted with subjectsto obtain information on basic demographic informa-tion; a history of cigarette smoking, dietary intake,medical and family history, and occupational history.Blood and urine samples were also collected as pre-viously described[13].

2.4. Assessment of meat consumption andcooking practices

Meat intake was assessed as described previously[12]. Briefly, the assessment tool was a food frequencyquestionnaire (FFQ) of meat consumed. For each typeof meat studied, subjects were asked about the fre-quency of consumption, the usual portion size, andpreparation methods. Choices for consumption fre-quency were: never, less than one time per month,one time per month, two–three times per month, onceper week, twice per week, three–four times per week,five–six times per week, once per day, or two or moretimes per day.

Food models were used to help subjects estimateportion size. Preparation methods inquired about werepan-fried, oven-broiled, baked or roasted, grilled orBBQ, or “other method”.

For analysis purposes, red meat was classified intotwo subgroups: (1) “red meat” which included ham-burger patties, beef-steak, bacon, sausage, pork (chopsand ham) and (2) “other red meat” which includedmeat in stews and sauces, etc.

We measured meat doneness preferences with ascale that incorporated photographs that show inter-nal doneness and external brownness accompanied bya numerical scale from 0.5 to 5.5[14]. Participantscould choose to report doneness level between the in-teger values associated with the photographs. Partici-pants were directed to pay special attention to “the wayhamburger and steaks look inside as well as outside”.

This study also included a semi-quantitative FFQsimilar to the Willett FFQ[15]. Intake of specifiedportions of 153 food items was ascertained, and nu-trient intake was estimated using a nutrient databasedeveloped for Minnesota Colon Cancer PreventionResearch Unit studies. For these particular analyses,reported frequencies of consumption were used toestimate usual intake of fruits, vegetables, and alco-

hol while nutrient values were used to estimate totalcalories and total fat intake.

Odds ratios (ORs) and 95% confidence intervals(CIs) were estimated by unconditional logistic regres-sion analysis (SAS Institute, Carey, NC). All ORswere adjusted for age, sex, race, education, cigarettesmoking (pack-years and pack-years squared), and ahistory of diabetes for more than 2 years prior to thedate of cancer diagnosis in cases or the selection datefor controls. Additional adjustment for physical activ-ity or for consumption of total calories, total fat, totalfruits, total vegetables, or alcohol did not substan-tively alter the findings and they were not includedin final models. The ORs for meat consumption vari-ables were modeled using a linear relationship be-tween the amount of reported meat intake and the logodds of disease. The linear relationship was checkedby adding a quadratic term to the regression model,which was not statistically significant for any of themeat variables. Tests for trend were based on contin-uous data. Dietary intake quintiles were determinedfrom the distribution among control subjects.

3. Results

This analysis was restricted to subjects who com-pleted the meat module (193 cases and 674 controls).The study population analyzed was 97% caucasian.The mean age of the cases and controls was 65.4and 66.0 years, respectively. Sixty-one percent of thecases and 56.4% of the controls were males. Morecases than controls reported current cigarette smok-ing (17% versus 11%) and fewer cases than controlsreported never smoking (36% versus 49%). Caseswere more likely than controls to report a history ofdiabetes (24% versus 8%).

Compared to controls, cases reported higher meanlevels of intake for total meat, total red meat (includingboth subgroups of red meat), and processed meat, andlower mean levels of intake for white meat (Table 1).

Examining red meat by doneness preferences,we found higher levels of daily intake reported bycases compared to controls for both rare/mediumand well/very well-done red meat, but the relativeamounts (%) of meat intake by doneness levels werenot different between cases and controls (Table 2).Cases reported higher levels of fried and grilled/BBQ


Table 1Meat consumption in pancreatic cancer cases and controls

Meat intake(grams per day)

Cases(n = 193;Mean (S.D.))

Controls(n = 674;Mean (S.D.))

Total meat 100.5 (61.2) 91.5 (56.3)Total red meat 69.4 (52.2) 56.9 (46.6)

Red meat (five items) 40.5 (38.3) 32.2 (29.6)Other red meat 29.0 (24.2) 24.8 (22.5)

White meat 31.1 (24.4) 34.6 (26.7)Processed meats 5.9 (9.4) 5.8 (9.7)

Five meat items are hamburgers, steak, bacon, sausage, and pork(chops and ham).

meat intake and lower levels of broiled meat intakerelative to controls (Table 2).

In logistic regression models that were adjusted forage, sex, smoking, education, race, and diabetes, BBQmeat intake (as a continuous variable) was statisticallysignificant as a predictor of pancreatic cancer risk. Thisfinding was robust with further adjustment for whitemeat, other red meat, and red meat not grilled. The ORand 95% CI for BBQ meat intake (gram per day) was1.03 (1.01, 1.04). The test for trend was statisticallysignificant (P < 0.001).

Using quintiles of BBQ meat consumed, estimatesof cancer risk increased with reported increases in in-take (Table 3). The OR and 95% CI for the fifth quin-tile relative to the reference group (quintiles 1 and 2)were 2.2 (1.4, 3.4). Findings were not substantivelychanged by further adjustment for total calories, totalfat, total fruit and vegetables, or alcohol consumption(from a FFQ).

Although the findings with respect to fried red meatconsumption were not statistically significant, the risksassociated with increasing intake were elevated, and

Table 2Red meat consumption by doneness preferences and preparationmethods in pancreatic cancer cases and controls

Red meat intake(grams per day)

Cases(n = 193;Mean (S.D.))

Controls(n = 674;Mean (S.D.))

Rare/medium 20.8 (33.2) 16.2 (22.1)Well/very well-done 19.6 (20.0) 16.0 (19.8)Fried 16.8 (19.4) 14.9 (20.4)Grilled 14.1 (32.1) 6.4 (11.4)Broiled 5.7 (9.2) 6.6 (11.6)

Table 3Pancreas cancer risk and intake of grilled/barbecued red meata


Quintile Cases (n) Odds ratiob 95% CI

0–0 Q1,2 77 1.0 (reference)0.9–3.5 Q3 14 1.4 0.7–2.73.7–10.7 Q4 36 1.2 0.7–1.910.8–88.0 Q5 66 2.2 1.4–3.4

a P−trend< 0.001.b Adjusted for age, sex, smoking (pack-years and pack-years

squared), education, race, diabetes, white meat, red meat notgrilled, and other red meat.

Table 4Pancreas cancer risk and intake of fried red meata



0–1.1 Q1 25 1.0 (reference)1.2–4.6 Q2 26 1.1 0.6–2.04.7–11.5 Q3 55 1.9 1.1–3.311.7–24.1 Q4 44 1.6 0.9–2.824.2–192.6 Q5 43 1.4 0.7–2.6

a P−trend= 0.90.b Adjusted for age, sex, smoking (pack-years and pack-years

squared), education, race, diabetes, white meat, red meat not fried,and other red meat.

thus consistent with our a priori hypothesis (Table 4).In contrast, intake of broiled meat, which does notcontain high levels of carcinogens, showed no patternof increased risk (Table 5). These findings were alsoconsistent with our predictions.

Other meat variables did not show statisticallysignificant associations with risk nor did they sub-stantively alter the findings for BBQ meat intake.These included total meat, processed meat, total red

Table 5Pancreas cancer risk and intake of broiled red meata



0–0 Q1,2 102 1.0 (reference)0.5–11.7 Q3 31 0.9 0.5–1.45.0–11.7 Q4 28 0.7 0.4–1.112–171.7 Q5 32 0.7 0.4–1.2

a P−trend= 0.08.b Adjusted for age, sex, smoking (pack-years and pack-years

squared), education, race, diabetes, white meat, red meat notbroiled, and other red meat.


meat, total broiled meat, total fried meat, or total meatcooked by means other than grilling.

4. Discussion

In this population-based case-control study, wefound that reported consumption of grilled/BBQ redmeat intake was associated with pancreatic cancer.Risk estimates increased with increased intake whenmodeled as either a continuous variable or in quin-tiles. We had predicted that we would see an increasedrisk associated with grilled/BBQ meat intake andfried meat intake. Elevated risk was associated withfried meat intake, although the findings were not sta-tistically significant. Associations between pancreascancer and other meat variables were not statisti-cally significant, nor did they alter the findings forgrilled/BBQ or fried red meat intake.

It is of interest to compare our findings with those ofother epidemiologic studies. Few studies of pancreaticcancer have collected information on the method ofmeat preparation, however, positive associations havebeen found for fried and grilled meat[16] and othershave reported associations with fried and or grilledfoods[3,17].

A number of epidemiologic studies have analyzedmeat intake (and fat intake which is closely correlatedwith meat in the diet). These include positive associ-ations for the following: fat[18–20]; beef and bacon[21]; pork and beef[22,23]; pork and fish, but notbeef[24]; beef, chicken, and pork[25]; red meat andsalted/smoked meat[26]; total meat, liver, ham andsausages[4]; and daily meat consumption[27].

Null and inverse associations have also been re-ported. La Vecchia et al.[28] reported inconsistentfindings with respect to meat and fat intake, with ele-vated risk for ham and a marginally increased risk formeat, and decreased risk associated with fish. Millset al.[29] reported an increased risk with consumptionof meat that disappeared after controlling for smoking,while Gold et al.[30] reported null associations withmeat. Baghurst et al.[31] found no consistent findingsfor meat intake. Raymond et al.[32] found that meatintake did not alter the risk except for lean pork, whichwas associated with a decreased risk. Mizuno et al.[33] found no elevated risks associated with meat con-sumption in a case-control study in Japan, but found

that daily fish consumption was inversely associatedwith the disease.

Our findings may help to explain, in part, the in-consistencies in the literature with respect to meatconsumption. Meat cooking practices vary betweenindividuals and between populations. The grilling/barbecuing and frying of meat can produce high lev-els of carcinogens such as HCAs and PAHs whileother methods of preparation, such as broiling, bakingor stewing form negligible levels of these compounds[12,14]. If the carcinogens formed during the cook-ing of meat are casually related to pancreatic cancer,true associations might be obscured in studies wheredetailed meat preparation data were not collectedand analyzed together with meat intake. In our study,although total meat consumption and red meat con-sumption were higher in cases than controls, theywere not significant predictors of risk. A positive androbust association was observed for BBQ red meat in-take and a positive, though not significant associationwas observed with fried red meat intake.

Case-control studies of human pancreatic cancerpresent numerous challenges. Most cases present atadvanced stages of the disease, and median survivalis 3 months. As a consequence, many cases are eithernot approached for study because they die before no-tification channels bring them to the attention of studyinvestigators, or they are unable or unwilling to par-ticipate because of the severity of their disease. Thepotential for selection bias in cases is thus very great.This is true in our study. Of cases we actually spokewith, approximately 75% took part in the study, but lo-gistical constraints lowered the total participation ratesubstantially.

However, the control response rate in this study isnearly identical to that achieved in a population-basedcase-control study of colon cancer conducted in thesame region[34]. Some eligible controls may have re-fused to participate because our study entailed variouscomponents including an in-person interview, bloodsample, and urine-based caffeine assay.

Formal sensitivity analyses can help to quantify themagnitude of potential biases in epidemiologic stud-ies and the potential impact they might have on causalparameter estimates[35]. Even without a formal anal-ysis, in this study, we can model scenarios with selec-tion bias that would abrogate our findings with respectto meat intake. Thus, our conclusions carry caveats.


However, it is hard to imagine why selection biaswould result in over-reporting by cases of particularpreparation methods: namely, frying and grilling.

Several strengths of this study should also be noted.This study was designed to address the hypothesis thatmeat intake, as a source of dietary carcinogens, is as-sociated with risk of pancreas cancer. We collecteddetailed information, from direct interviews, on cook-ing practices and doneness levels for specific types ofcommonly consumed meats. We are not aware of anyother studies of pancreatic cancer that have collectedthis type of data. In addition, we collected detailed in-formation on various possible confounders.

Our findings suggest that individuals may decreasetheir risk of pancreatic cancer by altering their meatcooking techniques. HA and PAH levels in meats,which may be responsible for the association withBBQ and fried meat intake that we see in this study,can be minimized by cooking methods that do notproduce high levels of carcinogens—such as baking,broiling and stewing[10–12]. When meats are pre-pared by barbecuing or pan-frying, they should becooked sufficiently to kill bacteria, while avoiding ex-cessive external brownness or charring. Consumptionof well- and very well-done meats should be avoided.Adopting safer meat cooking practices may reduce therisk of pancreatic and other cancers.

Acknowledgements

The authors wish to thank the following individualsfor their contributions to this study: Lois Murphy, Car-olee DeSotel, Tami Dahl, for study coordination; FayeImker-Witte, Ann Fitzpatrick, and Jennifer Hayes forlaboratory aspects of the study; Jose Jesserun, forpathology review; John Potter and Richard K. Sever-son, for collaboration in the early phases of the study;Phyllis Pirie and Karen Virnig and their staff membersfor control ascertainment; Paul McGovern for statisti-cal advice; Gail Jolitz (TUMORS) and the tumor reg-istrars in the Twin Cities seven-county metropolitanarea of Minnesota for case identification; Judy Punykoand Sally Bushhouse, Minnesota Cancer SurveillanceSystem for case identification and verification; CharlesMurray, for aspects of case ascertainment and studydesign; Ann Deshler (CCOP), Sally Fraki (CGOP), thelate Sharon Dahlen (CGOP), and their staffs for case

ascertainment. David Dunn, Gary Grammens, AnneJoseph, and the many physicians, nurses, and otherstaff members of hospitals in the Twin Cities area andthe Mayo Clinic who helped to identify cases; PatBrothen, Trista Johnson, Jane Curtin, Susie Reyes fordata base management and analysis; and Lisa Fosdickfor assistance with the nutrient database. Finally, wethank the cases and controls who generously partic-ipated in this study with the often expressed hopethat their efforts might help others. Support for theseanalyses included a grant from the National CancerInstitute (K. Anderson, PI, RO1-CA58697).

References

[1] K.E. Anderson, J.D. Potter, T. Mack, Pancreas, in: D.Schottenfeld, J.F. Fraumeni Jr. (Eds.), Cancer Epidemiologyand Prevention, Oxford Press, New York, 1996, pp. 725–771.

[2] G.R. Howe, J.D. Burch, Nutrition and pancreatic cancer,Cancer Causes Control 7 (1996) 69–82.

[3] B.T. Ji, W.H. Chow, G. Gridley, J.K. McLaughlin, Q. Dai, S.Wacholder, M.C. Hatch, Y.T. Gao, J.F. Fraumeni Jr., Dietaryfactors and the risk of pancreatic cancer: a case-control studyin Shanghai, China, Cancer Epidemiol. Biomarkers Prev. 4(1995) 885–893.

[4] M. Soler, L. Chatenoud, C. La Vecchia, S. Franceschi, E.Negri, Diet, alcohol, coffee and pancreatic cancer: final resultsfrom an Italian study, Eur. J. Cancer Prev. 7 (1998) 455–460.

[5] T. Sugimura, Carcinogenicity of mutagenic heterocyclicamines formed during the cooking process, Mutat. Res. 150(1985) 33–41.

[6] T. Sugimura, S. Sato, Mutagens–carcinogens in foods, CancerRes. 43 (1983) 2415s–2421s.

[7] B. Commoner, A.J. Eithayathil, P. Dolara, S. Nair, P.Madyastha, G.C. Cuca, Formation of mutagens in beef andbeef extract during cooking, Science 201 (1978) 913–916.

[8] J.S. Felton, M.G. Knize, Occurrence, identification andbacterial mutagenicity of heterocyclic amines in cooked food,Mutat. Res. 259 (1991) 205–217.

[9] M.G. Knize, R. Sinha, C.P. Salmon, S.S. Metha, K.P.Dewhirst, J.S. Felton, Formation of heterocyclic amine muta-gens/carcinogens during cooking of muscle meat, J. MuscleFood 7 (1996) 271–290.

[10] D.W. Layton, K.T. Bogen, M.G. Knize, F.T. Hatch, V.M.Johnson, J.S. Felton, Cancer risk of heterocyclic amines incooked foods: an analysis and implications for research,Carcinogenesis 16 (1995) 39–52.

[11] R. Sinha, N. Rothman, Exposure assessment of heterocyclicamines (HCAs) in epidemiologic studies, Mutat. Res. 376(1997) 195–202.

[12] R. Sinha, N. Rothman, Role of well-done, grilled red meat,heterocyclic amines (HCAs) in the etiology of human cancer,Cancer Lett. 143 (1999) 189–194.


[13] M. Gross, T. Kruisselbrink, K. Anderson, N. Lang, P.McGovern, R. Delongchamp, F.F. Kadlubar, The distributionand concordance ofN-acetyltransferase genotype and pheno-type in an American population, Cancer Epidemiol. Bio-markers Prev. 8 (1999) 683–692.

[14] R. Sinha, N. Rothman, C.P. Salmon, M.G. Knize, E.D.Brown, C.A. Swanson, D. Rhodes, S. Rossi, J.S. Felton,O.A. Levander, Heterocyclic amine content in beef cooked bydifferent methods to varying degrees of doneness and gravymade from meat drippings, Food Chem. Toxicol. 36 (1998)279–287.

[15] W. Willett, R. Reynolds, S. Cottrell-Hoehner, L. Sampson,M. Brown, Validation of a semi-quantitative food frequencyquestionnaire: comparison with a one-year diet record, J. Am.Diet. Assoc. 87 (1987) 43–47.

[16] S.E. Norell, A. Ahlbom, R. Erwald, G. Jacobson, I. Lindberg-Navier, R. Olin, B. Törnberg, K.-L. Wiechel, Diet andpancreatic cancer: a case-control study, Am. J. Epidemiol.124 (1986) 894–902.

[17] P. Ghadirian, J. Baillargeon, A. Simard, C. Perret, Foodhabits and pancreatic cancer: a case-control study of theFrancophone Community in Montreal, Canada, Cancer Epide-miol. Biomarkers Prev. 4 (1995) 895–899.

[18] J.P. Durbec, G. Chevillotte, J.M. Bidart, P. Berthezene, H.Sarles, Diet, alcohol, tobacco and risk of cancer of the pan-creas: a case-control study, Br. J. Cancer 47 (1983) 467–470.

[19] R. Goto, H. Masuoka, K. Yoshida, M. Mori, H. Miyake, Acase-control study of cancer of the pancreas, Jpn. J. CancerClin. 36 (1990) 344–350.

[20] J.L. Lyon, M.L. Slattery, A.W. Mahoney, L.M. Robison, Die-tary intake as a risk factor for cancer of the exocrine pancreas,Cancer Epidemiol. Biomarkers Prev. 2 (1993) 513–518.

[21] T.M. Mack, M.C. Yu, R. Hanisch, B.E. Henderson, Pancreascancer and smoking, beverage consumption and past medicalhistory, JNCI 76 (1986) 49–60.

[22] R.T. Falk, L.W. Pickle, E.T. Fontham, P. Correa, J.F.Fraumeni, Life-style risk factors for pancreatic cancer inLouisiana: a case-control study, Am. J. Epidemiol. 128 (1988)324–336.

[23] G.W. Olsen, J.S. Mandel, R.W. Gibson, L.W. Wattenberg,L.M. Schuman, A case-control study of pancreatic cancer andcigarettes, alcohol, coffee and diet, Am. J. Public Health 79(1989) 1016–1019.

[24] H.B. Bueno de Mesquita, P. Maisonneuve, S. Runia, C.J.Moerman, Intake of foods and nutrients and exocrinecarcinoma of the pancreas: a population-based case-controlstudy in The Netherlands, Int. J. Cancer 48 (1991) 540–549.

[25] D.C. Farrow, S. Davis, Diet and risk of pancreatic cancer inmen, Am. J. Epidemiol. 132 (1990) 423–431.

[26] W. Zheng, J.K. McLaughlin, F. Gridley, E. Bjelke, L.M.Schuman, D.T. Silverman, S. Wacholder, H.T. Co-Chien, W.J.Blot, J.F. Fraumeni Jr., A cohort study of smoking, alcoholconsumption and dietary factors for pancreatic cancer (UnitedStates), Cancer Causes Control 4 (1993) 477–482.

[27] T. Hirayama, Epidemiology of pancreatic cancer in Japan,Jpn. J. Clin. Oncol. 19 (1989) 208–215.

[28] C. La Vecchia, E. Negri, B. D’Avanzo, M. Ferraroni, A.Gramenzi, R. Savoldelli, P. Boyle, S. Franceschi, Medicalhistory, diet and pancreatic cancer, Oncology 47 (1990) 463–466.

[29] P.K. Mills, L. Beeson, D.E. Abbey, G.E. Fraser, R.L. Phillips,Dietary habits and past medical history as related to fatalpancreas cancer risk among adventists, Cancer 61 (1988)2578–2585.

[30] E.B. Gold, L. Gordis, M.D. Diener, R. Seltser, J.K. Boitnott,T.E. Bynum, D.F. Hutcheon, Diet and other risk factors forcancer of the pancreas, Cancer 55 (1985) 460–467.

[31] P.A. Baghurst, A.J. McMichael, A.H. Slavotinek, K.I.Baghurst, P. Boyle, A.M. Walker, A case-control study of dietand cancer of the pancreas, Am. J. Epidemiol. 134 (1991)167–179.

[32] L. Raymond, F. Infante, A.J. Tuyns, M. Voirol, A.B. Lowefels,Alimentation et cancer du pancreas, Gatroenterol. Clin. Biol.11 (1987) 488–492.

[33] S. Mizuno, S. Watanabe, K. Nakamura, M. Omata, H. Ogushi,K. Ohashi, H. Ohanagi, T. Fujuki, K. Motojima, A multi-institute case-control study on the risk factors of developingpancreatic cancer, Jpn. J. Clin. Oncol. 22 (1992) 286–291.

[34] R.A. Kerber, M.L. Slattery, J.D. Potter, B.J. Caan, S.L.Edwards, Risk of colon cancer associated with a familyhistory of cancer or colorectal polyps: the diet, activity, andreproduction in colon cancer study, Int. J. Cancer 78 (1998)157–160.

[35] S. Greenland, Basic methods for sensitivity analysis of biases,Int. J. Epidemiol. 25 (1996) 1107–1116.

meat intake and cooking techniques: associations with pancreatic cancer

Documents