Cancer Causes & Control

, Volume 18, Issue 10, pp 1153–1167

Pancreatic cancer, animal protein and dietary fat in a population-based study, San Francisco Bay Area, California

Authors

  • June M. Chan
    • Department of Epidemiology and BiostatisticsUniversity of California San Francisco
    • Department of UrologyUniversity of California San Francisco
  • Furong Wang
    • Department of Epidemiology and BiostatisticsUniversity of California San Francisco
    • Department of Epidemiology and BiostatisticsUniversity of California San Francisco
    • Department of Health Research and PolicyStanford University
Original Paper

DOI: 10.1007/s10552-007-9054-0

Cite this article as:
Chan, J.M., Wang, F. & Holly, E.A. Cancer Causes Control (2007) 18: 1153. doi:10.1007/s10552-007-9054-0

Abstract

Objective

The associations between animal protein or fat and risk of pancreatic cancer have been reported previously with inconsistent results. A population-based case–control study of pancreatic cancer was conducted in the San Francisco Bay Area to examine these associations.

Methods

A semi-quantitative food-frequency questionnaire was administered to 532 cases and 1,701 controls between 1995 and 1999. Odds ratios (OR) and 95% confidence intervals (CI) were computed as estimates of the relative risk of pancreatic cancer.

Results

When comparing highest versus lowest levels of intake in multivariable adjusted models, positive associations were observed for several beef/lamb and individual animal protein items, including beef/lamb as a main dish (OR = 2.2, 95% CI: 1.0–4.5), regular hamburger (OR = 1.7, 95% CI: 1.2–2.4), whole eggs (OR = 1.6, 95% CI: 1.0–2.4), butter (OR = 2.4, 95% CI: 1.6–3.5), and total dairy not including butter (OR = 2.6, 95% CI: 1.8–3.7). Some high-fat/processed-meat products (i.e., sausage, salami, bacon), but not all (i.e., beef, pork, or poultry hot dogs), also were positively associated with risk. An inverse association was noted for greater chicken/turkey consumption (OR = 0.7, 95% CI: 0.5–1.0). The risk comparing the highest versus lowest quartiles for fats and cholesterol consumption were: total fat (OR = 1.6, 95% CI: 1.2–2.1); animal fat (OR = 1.9, 95% CI: 1.4–2.5); saturated fat (OR = 1.9, 95% CI: 1.4–2.6); monounsaturated fat (OR = 1.3, 95% CI: 1.0–1.8); and dietary cholesterol (OR = 1.5, 95% CI: 1.1–2.0, all p-trends ≤ 0.02).

Conclusions

These data provide some evidence that beef or lamb, eggs, dairy, fat, or cholesterol may increase the risk of pancreatic cancer.

Keywords

Pancreatic neoplasmCase–control studyMeatDairyFat

Introduction

Pancreatic cancer is the fourth leading cause of cancer death among men and among women with a five-year relative survival rate of <4% [1]. The American Cancer Society estimates that 37,170 new cases of pancreatic cancer will be diagnosed in the United States in 2007 and 33,370 deaths will occur, making it the most fatal cancer within the US [1]. Identifying modifiable risk factors to prevent pancreatic cancer could have substantial public health impact [2, 3]. Smoking consistently has conferred an increased risk [46]. Several other modifiable risk factors have been suggested for pancreatic cancer, including diabetes, obesity, chronic pancreatitis, exercise, and diet [2, 3, 7].

It has been hypothesized that greater intake of cholesterol, fat, saturated fat, red meat, grilled meats, or eggs may be associated with an elevated risk of pancreatic cancer [2]. Greater vegetable and fruit consumption and antioxidant nutrients may be inversely associated with the disease [2, 810]. The data on fat and meat and pancreatic cancer risk is inconsistent. Studies that have reported positive associations generally have observed two- to three-fold elevations in risk [1113], although some studies have reported no association [9, 1419]. To explore the relationship between fat, animal protein, and pancreatic cancer, we conducted a large population-based case–control study that included no proxy interviews, in the San Francisco Bay Area, California. We hypothesized that red meat, processed meat, and fat would be positively associated with risk of pancreatic cancer.

Materials and methods

Study population

Details of the study design and selection of the study population have been published previously [8, 2024]. Briefly, between 1995 and 1999, individuals with incident adenocarcinoma of the exocrine pancreas were identified in six counties of the San Francisco Bay Area (in-area cases) by the Northern California Cancer Center (NCCC) using rapid case ascertainment (RCA). The RCA personnel contacted us approximately one to five weeks after they first obtained the diagnosis, depending upon when they were able to confirm physician contact information. Patients or their physicians were contacted within one week after we first received information about their diagnosis from RCA. Pancreatic cancer diagnoses were confirmed by participants’ physicians and by the Surveillance, Epidemiology and End Results (SEER) abstracts. Eligible in-area cases were: 21–85 years old, residents of one of the six Bay Area counties, alive, and could complete an interview in English. Additionally, 81 eligible University of California San Francisco clinic out-of-area cases met the same criteria as in-area cases except that they were residents of counties adjacent to the six Bay Area counties. A total of 735 in-area patients died before we could interview them, and 84 patients were ineligible because they could not complete an interview in English [21, 22]. Among 798 eligible cases, 532 completed an interview for a 67% response rate; 8% of the patients who were contacted refused to participate. In-area control participants were frequency-matched to in-area cases by sex and age within five-year categories and selected from the target population using random digit dial (RDD). Out-of-area controls also were identified using RDD and were frequency matched to out-of-area cases by telephone area-code and prefix, and by sex and age. Controls over 65 years of age were supplemented by random selection from Health Care Finance Administration (now the Centers for Medicare & Medicaid Services) lists that covered the same six San Francisco Bay Area counties. Among 2,525 eligible controls, 1,701 eligible controls completed an interview for a 67% response rate. The study was reviewed and approved by the University of California San Francisco institutional review board and written informed consent was obtained from each participant prior to interview.

Intake of food and nutrients

Eligible cases and controls were interviewed in-person by trained interviewers using a structured epidemiologic- and a food-frequency questionnaire. No proxy interviews were conducted. Participants self-reported their frequency of food intake one year before their cancer diagnosis for cases or interview for controls using a previously validated 131-item semi-quantitative food-frequency questionnaire [2527]. This study examined dietary patterns one year before the cancer diagnosis or interview for controls, thereby avoiding recent assessment of dietary changes due to pancreatic cancer. In a comparison of mean nutrient intake based on the questionnaire versus two one-week diet records administered six-months apart to 127 men, most nutrients were similar. Correlation coefficients between the calorie-adjusted nutrient intake measured by diet records and the questionnaire generally were high after adjusting for week-to-week variation in diet-record intake (mean r = 0.7; range: 0.3–0.9) [26]. Intra-class correlation coefficients for nutrient intake assessed by questionnaires administered one-year apart ranged from 0.5 for vitamin E without supplements to 0.8 for vitamin C with supplements [26]. In our study, the correlation between individual fats and meat products ranged from −0.006 (for chicken without skin and monounsaturated fat intake) to 0.45 (for beef/lamb as a main dish and saturated fat). In general, all individual beef/lamb or pork items were correlated with total, saturated, or monounsaturated fat (r range: 0.24–0.45, mean r = 0.34), with the strongest correlations existing for individual beef/lamb or pork items and saturated fat. Chicken or turkey food items were less positively (r range: 0.10–0.25; mean r = 0.13) or inversely correlated (i.e., chicken without skin) with fat intake.

Intake of seasonal food was reported as the average over the entire year. The frequency of food intake in general was asked as: never, <1/month, 1–3/month, 1/week, 2–4/week, 5–6/week, 1/day, 2–3/day, 4–5/day, and 6+/day. The specified portion size of each food item was considered as one serving, so that the food frequency could be transformed into number of servings per day. Total consumption of meat, fish, eggs, and dairy products was computed by combining servings per day of individual items.

All meats combined contained all items in both white meat and red meat. All red meat included: bacon; other processed meat; processed meat sandwiches; pork, beef, or lamb as a main dish; beef, pork, or lamb as a sandwich or mixed dish; beef or pork hotdogs; and hamburger. White meats included: chicken or turkey with skin; chicken or turkey without skin; chicken or turkey hotdogs; and chicken or turkey sandwiches. We conducted initial analyses for white meat in two models, with and without chicken or turkey hotdogs, and observed no meaningful difference in the risk of pancreatic cancer. Thus, we included chicken or turkey hotdogs in the broader white-meat group in subsequent analyses. Total fish/shellfish included the following items: canned tuna fish; shrimp, lobster, scallops, clams as a main dish; dark fish; other fish; and breaded fish cakes, pieces, or fish sticks. We conducted separate analyses for total fish/shellfish with and without breaded fish cakes, pieces, or fish sticks in the model, and observed no substantial difference in the risk of pancreatic cancer. Therefore, we included breaded fish cakes, pieces, or fish sticks in the total fish/shellfish group in subsequent models. Dairy products included: skim milk; 1% or 2% milk; whole milk; cream; frozen yogurt; sherbet or non-fat ice cream; ice cream; flavored yogurt without Nutrasweet©; yogurt, plain or with Nutrasweet©; cottage or ricotta cheese; cream cheese; other cheese; and butter. The model that included total dairy did not include butter due to the high concentration of fat in butter and the low concentration of calcium, which was different from the other sources of dairy products in total dairy. Participants also were asked about dietary patterns 10 years ago compared with their food-consumption patterns of 1-year ago. They reported the differences of intake for various foods as being the “same amount,” “much more,” or “much less” compared with their dietary habits 10 years ago. Six participants were excluded whose dietary questionnaires had responses for fewer than 10 items.

Daily consumption of total and subgroups of fat was computed based on the food-frequency questionnaire data. Subgroups of fat included: animal fat; dietary cholesterol; saturated fat; monounsaturated fat; vegetable fat; and polyunsaturated fat.

Statistical methods

Odds ratios (OR) and their 95% confidence intervals (CI) were computed by unconditional logistic regression to estimate the relative risk (hereafter called risk) of pancreatic cancer. The energy-adjusted residual model was used to estimate the risk of pancreatic cancer and nutrients [28]. The tests for linear trend were based upon the effect estimates for the factor of interest when included as an ordinal variable in the adjusted unconditional logistic-regression model. Potential interaction and confounding effects were investigated for energy intake in total calories, sex, education, body mass index (BMI), cigarette smoking, race, and history of diabetes. When there was no interaction effect observed, we further investigated potential confounding due to these factors using stratified analyses. A factor was considered as a confounder if it changed the OR between pancreatic cancer and the dietary variable by ≥10%. Of these factors, only total energy intake affected the associations for meat, dairy, fish/seafood, eggs, and risk of pancreatic cancer. However, for comparability with other studies, we provided data for both a parsimonious model that was adjusted for total energy intake and the matching factors of sex and age at diagnoses for cases or interview for controls, and a full multivariable model that adjusted for all of the putative risk factors mentioned above. Total energy intake was divided into quartiles based on the consumption among control participants: <1,439, 1,439–1,810, 1,811–2,265, >2,265 kcal per day. We found no substantial difference when comparing ORs for pancreatic cancer among study participants when we included or excluded extremely low (<500 kcal/day) and extremely high (>3,500 kcal/day) total caloric intake. These participants were retained in the final analyses. Education was grouped as ≤high school, college, and graduate work. BMI was estimated from usual adult weight and height (weight (kg)/height (m)2) as a measure of total adiposity. BMI was divided into three groups: <25, 25.0–29.9, ≥30 kg/m2 based on World Health Organization criteria. Smokers were defined as participants who had smoked >100 cigarettes in their lifetime, or a pipe or cigar for at least once a month for ≥6 months. Cigarette smokers were grouped as former smokers who quit >15 years ago, former smokers who quit ≤15 years, and current smokers. Participants self-reported race as white, black/African American, Asian/Pacific islanders or other. Hispanic ethnicity was reported as yes or no.

Results for total meat and animal product food groups and fats and the risk of pancreatic cancer among all participants and men and women separately are reported in Tables 2 and 3. Because the risk of pancreatic cancer was not appreciably different by sex, we subsequently presented individual food item risk estimates for total participants in Tables 46. There were essentially no differences between the results for whites only and all races combined, and these results are presented with all races combined. All statistical tests were two-sided and considered statistically significant when p ≤ 0.05. Statistical analyses were conducted using SAS software V9.1 (SAS Institute, Inc., Cary, NC).

Results

Demographic characteristics of study participants are presented in Table 1. Most participants were white non-Hispanics, with similar proportions in cases and controls. As anticipated, cases were more likely to have had a smoking history than controls. Participants were otherwise similar except that cases also were less well educated compared to controls.
Table 1

Socio-demographic characteristics of incident cases of pancreatic cancer and controls in a population-based study, San Francisco Bay Area, California

Factors

Pancreatic cancer cases

Control participants

No.

%

No.

%

Age

    <50

46

9

164

10

    50–59

120

23

438

26

    60–69

172

32

473

28

    70–79

158

30

498

29

    ≥80

36

7

128

8

Sex

    Men

291

55

883

52

    Women

241

45

818

48

Race

    White

442

83

1471

86

    Black/African American

46

9

78

5

    Asian/Pacific Islander

35

7

119

7

    Others

9

2

33

2

Hispanic origin

    Yes

25

5

114

7

    No

507

95

1585

93

Education

    ≤High school

235

44

534

31

    College

200

38

754

44

    Graduate work

97

18

413

24

Body mass index, kg/m2

    <25.0

280

53

993

59

    25.0–29.9

197

37

552

33

    ≥30

52

10

147

9

History of diabetes

    Yes

76

11

161

10

    No

455

86

1538

90

Smoking

    Never

163

31

652

38

    Pipe/cigar

16

3

73

4

    Cigarette

      Former, quit >15 years ago

133

25

508

30

      Former, quit ≤15 years ago

107

20

271

16

      Current

112

21

194

11

Table 2 provides associations between quartiles of meat/animal product food items and pancreatic cancer risk among all participants and men and women separately. In the full multivariable analyses, total meat, red meat, and total fish/seafood were not associated with risk of pancreatic cancer. Dairy products were associated with more than a doubling in risk, and eggs also were associated with an increased risk, although eggs showed no clear dose–response association. There was some suggestion of an inverse association for total chicken/turkey consumption. The results for men and women separately were very similar, except that only men appeared to experience an elevated risk of pancreatic cancer associated with egg consumption.
Table 2

Odds ratios (OR) and 95% confidence intervals (CI) for pancreatic cancer risk related to the consumption of meat, fish, eggs, and dairy in a population-based case–control study, San Francisco Bay Area, California

Quartiles (Q) of intake (median, serving/day)

Men and women

Men

Women

Case

Control

ORa

ORb

Case

Control

ORa

Case

Control

ORa

No. (%)

No. (%)

(95% CI)

(95% CI)

No. (%)

No. (%)

(95% CI)

No. (%)

No. (%)

(95% CI)

All meats

    Q1 (0.4)

98 (19)

426 (25)

1.0 (ref.)

1.0 (ref.)

33 (11)

156 (18)

1.0 (ref.)

65 (27)

270 (33)

1.0 (ref.)

    Q2 (0.7)

126 (24)

425 (25)

1.2 (0.9–1.7)

1.2 (0.9–1.7)

57 (20)

203 (23)

1.1 (0.7–1.9)

69 (29)

222 (27)

1.3 (0.9–2.1)

    Q3 (1.1)

129 (25)

428 (25)

1.2 (0.8–1.6)

1.1 (0.8–1.6)

74 (26)

238 (27)

1.1 (0.6–1.9)

55 (23)

190 (23)

1.2 (0.7–2.0)

    Q4 (1.7)

173 (33)

422 (25)

1.4 (1.0–2.0)

1.3 (0.9–1.9)

124 (43)

286 (32)

1.2 (0.7–2.2)

49 (21)

136 (17)

1.4 (0.8–2.6)

    Trend p

  

p = 0.01

p = 0.2

  

p = 0.3

  

p = 0.5

All red meatc

    Q1 (0.1)

85 (16)

427 (25)

1.0 (ref.)

1.0 (ref.)

29 (10)

161 (18)

1.0 (ref.)

56 (24)

266 (33)

1.0 (ref.)

    Q2 (0.4)

128 (24)

429 (25)

1.4 (1.0–2.0)

1.3 (0.9–1.7)

55 (19)

201 (23)

1.3 (0.8–2.3)

73 (31)

228 (28)

1.3 (0.9–2.0)

    Q3 (0.7)

134 (25)

426 (25)

1.4 (1.1–2.0)

1.1 (0.8–1.6)

76 (26)

231 (26)

1.1 (0.7–1.9)

58 (24)

195 (24)

1.1 (0.7–1.8)

    Q4 (1.2)

179 (34)

419 (25)

1.8 (1.3–2.5)

1.3 (0.9–1.8)

128 (44)

290 (33)

1.2 (0.7–2.1)

51 (21)

129 (16)

1.5 (0.9–2.5)

    Trend p

  

p < 0.0001

p = 0.1

  

p = 0.3

  

p = 0.3

Chicken, Turkey

    Q1 (0.1)

143 (27)

422 (25)

1.0 (ref.)

1.0 (ref.)

77 (27)

209 (24)

1.0 (ref.)

66 (28)

213 (26)

1.0 (ref.)

    Q2 (0.2)

138 (26)

430 (25)

0.9 (0.7–1.2)

0.9 (0.7–1.2)

73 (25)

221 (25)

0.8 (0.5–1.2)

65 (27)

209 (26)

1.1 (0.7–1.7)

    Q3 (0.4)

125 (24)

409 (24)

0.9 (0.6–1.1)

0.9 (0.6–1.2)

66 (23)

204 (23)

0.8 (0.5–1.2)

59 (25)

205 (25)

0.9 (0.6–1.5)

    Q4 (0.7)

120 (23)

440 (26)

0.7 (0.5–0.9)

0.7 (0.5–1.0)

72 (25)

249 (28)

0.6 (0.4–1.0)

48 (20)

191 (23)

0.8 (0.5–1.3)

    Trend p

  

p = 0.02

p = 0.06

  

p = 0.1

  

p = 0.2

Total fish/shellfish

    Q1 (0.1)

134 (25)

425 (25)

1.0 (ref.)

1.0 (ref.)

72 (25)

197 (22)

1.0 (ref.)

62 (26)

228 (28)

1.0 (ref.)

    Q2 (0.2)

95 (18)

328 (19)

0.9 (0.6–1.2)

1.0 (0.7–1.4)

51 (18)

157 (18)

0.9 (0.6–1.5)

44 (18)

171 (21)

1.3 (0.8–2.0)

    Q3 (0.3)

162 (31)

532 (31)

0.9 (0.7–1.2)

1.1 (0.8–1.4)

83 (29)

293 (33)

0.8 (0.5–1.2)

79 (33)

239 (29)

1.5 (1.0–2.4)

    Q4 (0.5)

135 (26)

416 (25)

0.9 (0.6–1.1)

1.2 (0.8–1.6)

82 (28)

236 (27)

1.0 (0.7–1.6)

53 (22)

180 (22)

1.4 (0.9–2.3)

    Trend p

  

p = 0.3

p = 0.4

  

p = 0.3

  

p = 0.8

Eggs

    Q1 (0.07)

122 (23)

584 (34)

1.0 (ref.)

1.0 (ref.)

40 (14)

285 (32)

1.0 (ref.)

82 (34)

299 (37)

1.0 (ref.)

    Q2 (0.1)

93 (18)

306 (18)

1.4 (1.0–1.9)

1.4 (1.0–1.9)

41 (14)

133 (15)

2.1 (1.2–3.5)

52 (22)

173 (21)

1.1 (0.7–1.7)

    Q3 (0.4)

232 (44)

611 (36)

1.7 (1.3–2.2)

1.6 (1.2–2.1)

151 (52)

342 (39)

2.7 (1.8–4.0)

81 (34)

269 (33)

1.1 (0.7–1.6)

    Q4 (0.8)

79 (15)

200 (12)

1.6 (1.1–2.2)

1.4 (1.0–2.0)

56 (19)

123 (14)

2.2 (1.3–3.7)

23 (10)

77 (9)

0.9 (0.5–1.6)

    Trend p

  

p = 0.006

p = 0.1

  

p = 0.04

  

p = 0.6

Dairy without butter

    Q1 (0.4)

74 (14)

417 (25)

1.0 (ref.)

1.0 (ref.)

23 (11)

218 (25)

1.0 (ref.)

41 (17)

199 (24)

1.0 (ref.)

    Q2 (1.1)

119 (23)

430 (25)

1.6 (1.1–2.2)

1.9 (1.4–2.7)

70 (24)

234 (27)

2.3 (1.4–3.8)

49 (21)

196 (24)

1.5 (0.9–2.5)

    Q3 (1.7)

153 (29)

429 (25)

2.0 (1.4–2.7)

2.5 (1.8–3.5)

74 (26)

208 (24)

2.6 (1.6–4.4)

79 (33)

221 (27)

2.4 (1.5–4.0)

    Q4 (3.5)

180 (34)

425 (25)

2.2 (1.6–3.0)

2.6 (1.8–3.7)

111 (39)

223 (25)

3.1 (1.8–5.1)

69 (29)

202 (25)

2.1 (1.2–3.6)

    Trend p

  

p < 0.0001

p < 0.0001

  

p = 0.0001

  

p = 0.006

aParsimonious model adjusted for age, sex and total energy intake (kcal/day, quartiles)

bAdditionally adjusted for body mass index (<25, 25.0–29.9, ≥30 kg/m2), race (white, black/African American, Asian/Pacific Islander, others), education (≤high school, college, graduate work), smoking (never smoker, former cigarette smoker who quit smoking >15 years ago, former cigarette smoker who quit smoking ≤15 years, current cigarette smoker, pipe and/or cigar smoker), and history of diabetes (yes, no), and grain (serving/day, quartiles), vegetables and fruits (serving/day, quartiles), eggs (serving/day, quartiles), dairy (serving/day quartiles), total fish (serving/day quartiles) and chicken and turkey (serving/day quartiles) and red meats (serving/day quartiles) as appropriate

cAll red meat included processed and non-processed red meat

Increased risks of 1.3–1.9 also were observed for highest versus lowest quartiles of total, animal, saturated, monounsaturated fats, and cholesterol for men and women combined (Table 3). There was no association for overall vegetable or polyunsaturated fat consumption. The magnitude of the associations and the trends also were stronger for men versus women. Among women total fat and cholesterol consumption were not associated with risk of pancreatic cancer, while associations for saturated and animal fat persisted.
Table 3

Odds ratios (OR) and 95% confidence intervals (CI) for pancreatic cancer and dietary intake of total and specific fats in a population-based study, San Francisco Bay Area, California

Quartiles (Q) of intake (median, gm/day)

Men and women

Men

Women

Case

Control

ORa

ORb

Case

Control

ORa

Case

Control

ORa

No. (%)

No. (%)

(95% CI)

(95% CI)

No. (%)

No. (%)

(95% CI)

No. (%)

No. (%)

(95% CI)

Total fat

    Q1 (32.1)

94 (18)

426 (25)

1.0 (ref.)

1.0 (ref.)

46 (16)

229 (26)

1.0 (ref.)

48 (20)

197 (24)

1.0 (ref.)

    Q2 (46.2)

110 (21)

425 (25)

1.2 (0.86–1.6)

1.1 (0.8–1.5)

63 (22)

219 (25)

1.4 (0.9–2.1)

47 (20)

206 (25)

0.9 (0.6–1.4)

    Q3 (61.9)

152 (29)

425 (25)

1.6 (1.2–2.2)

1.5 (1.1–2.1)

86 (30)

223 (25)

1.8 (1.2–2.7)

66 (28)

202 (25)

1.3 (0.8–2.0)

    Q4 (89.7)

169 (32)

425 (25)

1.8 (1.4–2.4)

1.6 (1.2–2.1)

93 (32)

212 (24)

1.9 (1.2–2.8)

76 (32)

213 (26)

1.4 (0.9–2.1)

Trend p

  

p < 0.0001

p = 0.0005

  

p = 0.002

  

p = 0.07

Animal fat

    Q1 (15.8)

84 (16)

426 (25)

1.0 (ref.)

1.0 (ref.)

37 (13)

205 (23)

1.0 (ref.)

47 (20)

221 (27)

1.0 (ref.)

    Q2 (25.4)

112 (21)

424 (25)

1.3 (1.0–1.8)

1.3 (0.9–1.7)

62 (22)

230 (26)

1.4 (0.9–2.2)

50 (21)

194 (24)

1.3 (0.8–2.0)

    Q3 (34.3)

147 (28)

426 (25)

1.7 (1.3–2.3)

1.6 (1.2–2.1)

79 (27)

217 (25)

1.8 (1.2–2.8)

68 (29)

209 (26)

1.3 (0.9–2.1)

    Q4 (51.2)

182 (35)

425 (25)

2.2 (1.6–2.9)

1.9 (1.4–2.5)

110 (38)

231 (26)

2.3 (1.5–3.5)

72 (30)

194 (24)

1.6 (1.0–2.4)

    Trend p

  

p < 0.0001

p < 0.0001

  

p < 0.0001

  

p = 0.05

Saturated fat

    Q1 (11.0)

93 (18)

425 (25)

1.0 (ref.)

1.0 (ref.)

44 (15)

226 (26)

1.0 (ref.)

49 (21)

199 (24)

1.0 (ref.)

    Q2 (16.3)

103 (20)

426 (25)

1.1 (0.8–1.5)

1.0 (0.8–1.4)

63 (22)

220 (25)

1.4 (0.9–2.1)

40 (17)

206 (25)

0.8 (0.5–1.2)

    Q3 (22.1)

128 (24)

424 (25)

1.4 (1.0–1.9)

1.3 (0.9–1.7)

66 (23)

221 (25)

1.4 (0.9–2.2)

62 (26)

203 (25)

1.2 (0.8–1.8)

    Q4 (32.7)

201 (38)

426 (25)

2.2 (1.6–2.9)

1.9 (1.4–2.6)

115 (40)

216 (24)

2.4 (1.6–3.7)

86 (36)

210 (26)

1.5 (1.0–2.3)

    Trend p

  

p < 0.0001

p < 0.0001

  

p < 0.0001

  

p = 0.01

Dietary Cholesterol

    Q1 (122.8)

92 (18)

425 (25)

1.0 (ref.)

1.0 (ref.)

42 (15)

212 (24)

1.0 (ref.)

50 (21)

213 (26)

1.0 (ref.)

    Q2 (192.6)

125 (24)

426 (25)

1.3 (1.0–1.8)

1.3 (0.9–1.7)

64 (22)

216 (24)

1.4 (0.9–2.1)

61 (26)

210 (26)

1.2 (0.8–1.9)

    Q3 (257.6)

144 (27)

425 (25)

1.6 (1.2–2.1)

1.4 (1.1–2.0)

84 (29)

235 (27)

1.6 (1.1–2.5)

60 (25)

190 (23)

1.4 (0.9–2.1)

    Q4 (368.9)

164 (31)

425 (25)

1.8 (1.3–2.4)

1.5 (1.1–2.0)

98 (34)

220 (25)

1.9 (1.2–2.9)

66 (28)

205 (25)

1.2 (0.8–1.9)

    Trend p

  

p < 0.0001

p = 0.006

  

p = 0.003

  

p = 0.3

Monounsaturated fat

    Q1 (13.1)

111 (21)

425 (25)

1.0 (ref.)

1.0 (ref.)

53 (18)

215 (24)

1.0 (ref.)

58 (25)

210 (26)

1.0 (ref.)

    Q2 (19.5)

108 (21)

426 (25)

1.0 (0.7–1.3)

0.9 (0.7–1.2)

62 (22)

221 (25)

1.0 (0.7–1.6)

46 (19)

205 (25)

0.8 (0.5–1.3)

    Q3 (25.9)

141 (27)

425 (25)

1.3 (0.9–1.7)

1.2 (0.9–1.6)

86 (30)

228 (26)

1.4 (0.9–2.1)

55 (23)

197 (24)

1.0 (0.6–1.5)

    Q4 (38.6)

165 (31)

425 (25)

1.5 (1.1–2.0)

1.3 (1.0–1.8)

87 (30)

219 (25)

1.4 (0.9–2.1)

78 (33)

206 (25)

1.3 (0.8–1.9)

    Trend p

  

p = 0.001

p = 0.02

  

p = 0.05

  

p = 0.2

Vegetable fat

    Q1 (15.0)

121 (23)

426 (25)

1.0 (ref.)

1.0 (ref.)

70 (24)

236 (27)

1.0 (ref.)

51 (22)

190 (23)

1.0 (ref.)

    Q2 (24.1)

144 (27)

425 (25)

1.2 (0.9–1.6)

1.3 (1.0–1.7)

77 (27)

227 (26)

1.2 (0.8–1.8)

67 (28)

198 (24)

1.3 (0.9–2.1)

    Q3 (33.3)

126 (24)

425 (25)

1.0 (0.8–1.4)

1.1 (0.8–1.5)

76 (26)

214 (24)

1.2 (0.8–1.7)

50 (21)

211 (26)

0.9 (0.6–1.4)

    Q4 (52.3)

134 (26)

425 (25)

1.1 (0.8–1.5)

1.1 (0.8–1.5)

65 (23)

206 (23)

1.0 (0.7–1.5)

69 (29)

219 (27)

1.2 (0.8–1.8)

    Trend p

  

p = 0.7

p = 0.8

  

p = 0.9

  

p = 0.9

Polyunsaturated fat

    Q1 (6.7)

113 (22)

425 (25)

1.0 (ref.)

1.0 (ref.)

70 (24)

235 (27)

1.0 (ref.)

43 (18)

190 (23)

1.0 (ref.)

    Q2 (9.9)

133 (25)

425 (25)

1.2 (0.9–1.6)

1.2 (0.9–1.5)

72 (25)

218 (25)

1.0 (0.7–1.5)

61 (26)

207 (25)

1.3 (0.8–2.0)

    Q3 (13.3)

149 (28)

426 (25)

1.3 (1.0–1.8)

1.2 (0.9–1.6)

75 (26)

221 (25)

1.1 (0.7–1.6)

74 (31)

205 (25)

1.5 (0.9–2.3)

    Q4 (20.2)

130 (25)

425 (25)

1.2 (0.9–1.5)

1.0 (0.8–1.4)

71 (25)

209 (24)

1.0 (0.7–1.5)

59 (25)

216 (26)

1.1 (0.7–1.8)

    Trend p

  

p = 0.2

p = 0.7

  

p = 1.0

  

p = 0.5

aParsimonious model adjusted for age, sex using energy-adjusted residual model

bAdditionally adjusted for body mass index (<25, 25.0–29.9, ≥30 kg/m2), race (white, black/African American, Asian/Pacific Islander, others), education (≤high school, college, graduate work), smoking (never smoker, former cigarette smoker who quit smoking >15 years ago, former cigarette smoker who quit smoking ≤15 years, current cigarette smoker, pipe and/or cigar smoker), and history of diabetes (yes, no) using energy-adjusted residual model

Tables 46 present the associations of individual meat, fish/shellfish, and dairy food items and the risk of pancreatic cancer. Most beef/lamb, processed, or high-fat meat products were positively associated with risk of pancreatic cancer in the full multivariable models (i.e., sausage, salami/sandwich meats, bacon, beef, or lamb as a main dish, beef/pork/lamb as a mixed dish, regular hamburger), with the exception of beef/pork hotdogs and pork as a main dish that were not associated with risk (Table 4). Chicken or turkey food items were not consistently associated with pancreatic cancer risk. Whole eggs, but not egg whites were positively associated with the risk of pancreatic cancer. Most fish items were not associated with the risk of pancreatic cancer (Table 5). Skim milk, ice cream, cottage or ricotta cheese, other cheese, and butter were positively associated with the risk of pancreatic cancer and associations were slightly stronger for some higher-fat dairy items (e.g., ice cream, other cheese, and butter) (Table 6). 1% or 2% milk, whole milk, cream, frozen or other yogurt, and cream cheese were not associated with pancreatic cancer risk.
Table 4

Odds ratios (OR) and 95% confidence intervals (CI) for pancreatic cancer and individual meats/eggs in a population-based study, San Francisco Bay Area, California

Meats and eggs (serving size)

Case

Control

ORa (95% CI)

ORb (95% CI)

No. (%)

No. (%)

Processed meats

Sausage, kielbasa, etc. (2 oz. or 2 small links)

    <1/month

258 (49)

1,023 (60)

1.0 (ref.)

1.0 (ref.)

    1–3/month

141 (27)

403 (24)

1.3 (1.0–1.7)

1.2 (1.0–1.6)

    1/week

48 (9)

148 (9)

1.2 (0.8–1.7)

1.0 (0.7–1.5)

    ≥2/week

79 (15)

127 (7)

2.2 (1.6–3.0)

1.8 (1.3–2.6)

    Trend p

  

p < 0.0001

p = 0.003

Salami, bologna, other processed meat sandwichesc (1 whole)

    <1/month

145 (46)

781 (66)

1.0 (ref.)

1.0 (ref.)

    1–3/month

70 (13)

217 (18)

1.7 (1.2–2.4)

1.6 (1.2–2.3)

    1/week

49 (16)

87 (7)

2.8 (1.9–4.2)

2.3 (1.5–3.5)

    ≥2/week

49 (16)

102 (9)

2.4 (1.6–3.6)

1.9 (1.3–3.0)

    Trend p

  

p < 0.0001

p < 0.0001

Bacon (2 slices)

    <1/month

228 (43)

867 (51)

1.0 (ref.)

1.0 (ref.)

    1–3/month

111 (21)

405 (24)

1.0 (0.8–1.3)

0.9 (0.7–1.2)

    1/week

92 (17)

257 (15)

1.2 (0.9–1.6)

1.1 (0.8–1.4)

    2–4/week

74 (14)

146 (9)

1.7 (1.2–2.3)

1.3 (0.9–1.9)

    ≥5/week

21 (4)

26 (2)

2.7 (1.5–4.9)

1.9 (1.0–3.5)

    Trend p

  

p = 0.0002

p = 0.04

Beef or pork hotdogs (1 whole)

    <1/month

302 (57)

1,069 (63)

1.0 (ref.)

1.0 (ref.)

    1–3/month

137 (26)

428 (25)

1.1 (0.8–1.3)

1.0 (0.7–1.2)

    ≥1/week

87 (17)

203 (12)

1.3 (1.0–1.7)

1.1 (0.8–1.4)

    Trend p

  

p = 0.1

p = 0.9

Chicken or turkey hotdogsc (1 whole)

    <1/month

261 (83)

949 (80)

1.0 (ref.)

1.0 (ref.)

    ≥1/month

52 (17)

238 (20)

0.7 (0.5–1.0)

0.6 (0.4–0.9)

    Trend p

  

p = 0.06

p = 0.008

Chicken or turkey sandwichc (1 whole)

    <1/month

86 (27)

418 (35)

1.0 (ref.)

1.0 (ref.)

    1–3/month

85 (27)

310 (26)

1.3 (0.9–1.9)

1.4 (1.0–1.9)

    1/week

82 (26)

236 (20)

1.7 (1.2–2.4)

1.8 (1.3–2.6)

    ≥2/week

60 (19)

223 (19)

1.3 (0.9–1.9)

1.4 (0.9–2.0)

    Trend p

  

p = 0.04

p = 0.02

Non-processed meat

Beef or lamb as a main dish (4–6 oz)

    <1/month

107 (20)

453 (27)

1.0 (ref.)

1.0 (ref.)

    1–3/month

175 (33)

573 (34)

1.2 (1.0–1.6)

1.2 (0.9–1.6)

    1/week

127 (24)

401 (24)

1.2 (0.9–1.6)

1.1 (0.8–1.5)

    2–4/week

102 (19)

253 (15)

1.4 (1.0–2.0)

1.4 (1.0–2.0)

    ≥5/week

14 (3)

21 (1)

2.4 (1.2–4.9)

2.2 (1.0–4.5)

    Trend p

  

p = 0.01

p = 0.03

Pork as a main dishc (4–6 oz)

    <1/month

132 (42)

528 (44)

1.0 (ref.)

1.0 (ref.)

    1–3/month

113 (36)

434 (37)

1.0 (0.7–1.3)

0.9 (0.7–1.2)

    1/week

57 (18)

177 (15)

1.0 (0.7–1.5)

0.9 (0.6–1.4)

    ≥2/week

11 (4)

48 (4)

0.7 (0.3–1.3)

0.6 (0.3–1.1)

    Trend p

  

p = 0.6

p = 0.2

Beef, pork, lamb as a sandwich or mixed dish (1 whole)

    <1/month

124 (24)

534 (31)

1.0 (ref.)

1.0 (ref.)

    1–3/month

175 (33)

584 (34)

1.2 (0.9–1.6)

1.2 (0.9–1.6)

    1/week

137 (26)

349 (21)

1.5 (1.1–2.0)

1.5 (1.1–2.0)

    ≥2/week

90 (17)

234 (14)

1.4 (1.0–1.9)

1.4 (1.0–2.0)

    Trend p

  

p = 0.01

p = 0.01

Regular hamburger (1 patty)

    <1/month

230 (44)

890 (52)

1.0 (ref.)

1.0 (ref.)

    1–3/month

134 (25)

454 (27)

1.1 (0.9–1.5)

1.1 (0.8–1.4)

    1/week

92 (17)

236 (14)

1.4 (1.1–1.9)

1.3 (1.0–1.7)

    ≥2/week

70 (13)

121 (7)

2.0 (1.4–2.8)

1.7 (1.2–2.4)

    Trend p

  

p < 0.0001

p = 0.005

Lean or extra-lean hamburgerc (1 patty)

    <1/month

125 (40)

540 (45)

1.0 (ref.)

1.0 (ref.)

    1–3/month

93 (30)

371 (31)

1.1 (0.8–1.5)

1.1 (0.8–1.4)

    1/week

63 (20)

201 (17)

1.2 (0.9–1.7)

1.2 (0.8–1.8)

    ≥2/week

32 (10)

75 (6)

1.6 (1.0–2.6)

1.4 (0.9–2.3)

    Trend p

  

p = 0.05

p = 0.1

Chicken or turkey with skin (4–6 oz)

    <1/month

331 (63)

1,110 (65)

1.0 (ref.)

1.0 (ref.)

    1–3/month

76 (14)

255 (15)

1.0 (0.8–1.3)

0.9 (0.7–1.2)

    1/week

73 (14)

198 (12)

1.2 (0.9–1.6)

1.1 (0.8–1.4)

    ≥2/week

46 (9)

138 (8)

1.0 (0.7–1.5)

0.8 (0.6–1.2)

    Trend p

  

p = 0.5

p = 0.6

Chicken or turkey without skin (4–6 oz)

    <1/month

112 (21)

312 (18)

1.0 (ref.)

1.0 (ref.)

    1–3/month

106 (20)

332 (20)

1.0 (0.7–1.3)

1.1 (0.8–1.5)

    1/week

131 (25)

451 (27)

0.8 (0.6–1.1)

1.0 (0.7–1.3)

    ≥2/week

177 (34)

606 (36)

0.8 (0.6–1.0)

0.9 (0.7–1.2)

    Trend p

  

p = 0.06

p = 0.5

Whole eggs (1)

    <1/month

60 (11)

305 (18)

1.0 (ref.)

1.0 (ref.)

    1–3/month

84 (16)

361 (21)

1.2 (0.8–1.7)

1.2 (0.8–1.7)

    1/week

97 (18)

326 (19)

1.5 (1.0–2.1)

1.4 (1.0–2.1)

    2–4/week

221 (42)

571 (34)

1.8 (1.3–2.5)

1.7 (1.2–2.3)

    ≥5/week

64 (12)

138 (8)

2.0 (1.3–3.0)

1.6 (1.0–2.4)

    Trend p

  

p < 0.0001

p = 0.002

Egg beaters or egg whitesc

    <1/month

272 (87)

1,010 (85)

1.0 (ref.)

1.0 (ref.)

    ≥ /month

41 (13)

177 (15)

0.9 (0.6–1.2)

0.9 (0.6–1.4)

    Trend p

  

p = 0.4

p = 0.7

aParsimonious model adjusted for age, sex and total energy intake (kcal/day, quartiles)

bAdditionally adjusted for body mass index (<25, 25.0–29.9, ≥30 kg/m2), race (white, black/African American, Asian/Pacific Islander, others), education (≤high school, college, graduate work), smoking (never smoker, former cigarette smoker who quit smoking >15 years ago, former cigarette smoker who quit smoking ≤15 years, current cigarette smoker, pipe and/or cigar smoker), and history of diabetes (yes, no)

c733 participants answered a questionnaire without these food items

Table 5

Odds ratios (OR) and 95% confidence intervals (CI) for risk of pancreatic cancer and fish and shellfish in a population-based study, San Francisco Bay Area, California

Fish and shellfish (serving size)

Case

Control

ORa (95% CI)

ORb (95% CI)

No. (%)

No. (%)

Dark meat fish (3–5 oz)

    <1/month

271 (52)

817 (48)

1.0 (ref.)

1.0 (ref.)

    1–3/month

171 (33)

580 (34)

0.9 (0.7–1.1)

1.0 (0.8–1.2)

    ≥1/week

84 (16)

304 (18)

0.8 (0.6–1.0)

0.9 (0.7–1.2)

    Trend p

  

p = 0.04

p = 0.5

Canned tuna fish (3–4 oz)

    <1/month

139 (26)

557 (33)

1.0 (ref.)

1.0 (ref.)

    1–3/month

201 (38)

670 (39)

1.2 (0.9–1.5)

1.2 (0.9–1.5)

    1/week

127 (24)

305 (18)

1.6 (1.2–2.1)

1.6 (1.2–2.1)

    ≥2/week

59 (11)

169 (10)

1.2 (0.9–1.7)

1.3 (0.9–1.8)

    Trend p

  

p = 0.02

p = 0.02

Breaded fish cakes, pieces, or fish sticks (1 serving)

    <1/month

261 (83)

1,012 (85)

1.0 (ref.)

1.0 (ref.)

    ≥ /month

52 (17)

175 (15)

1.0 (0.7–1.4)

0.9 (0.6–1.3)

    Trend p

  

p = 1.0

p = 0.6

Shrimp, lobster, scallops, clams as main dish (1 serving)

    <1/month

290 (55)

879 (52)

1.0 (ref.)

1.0 (ref.)

    1–3/month

173 (33)

621 (37)

0.8 (0.6–1.0)

0.8 (0.6–1.0)

    ≥1/week

63 (12)

201 (12)

0.8 (0.6–1.1)

0.9 (0.6–1.2)

    Trend p

  

p = 0.04

p = 0.1

Other fish (3–5 oz)

    <1/month

209 (40)

725 (43)

1.0 (ref.)

1.0 (ref.)

    1–3/month

213 (40)

671 (39)

1.1 (0.9–1.3)

1.1 (0.9–1.4)

    1/week

85 (16)

231 (14)

1.2 (0.9–1.7)

1.3 (1.0–1.8)

    ≥2/week

19 (4)

74 (4)

0.8 (0.5–1.4)

0.9 (0.5–1.6)

    Trend p

  

p = 0.6

p = 0.3

aParsimonious model adjusted for age, sex and total energy intake (kcal/day, quartiles)

bAdditionally adjusted for body mass index (<25, 25.0–29.9, ≥30 kg/m2), race (white, black/African American, Asian/Pacific Islander, others), education (≤high school, college, graduate work), smoking (never smoker, former cigarette smoker who quit smoking >15 years ago, former cigarette smoker who quit smoking ≤15 years, current cigarette smoker, pipe and/or cigar smoker), and history of diabetes (yes, no)

Table 6

Odds ratios (OR) and 95% confidence intervals (CI) for pancreatic cancer risk and dairy products in a population-based study, San Francisco Bay Area, California

Dairy (serving size)

Case

Control

ORa (95% CI)

ORb (95% CI)

No. (%)

No. (%)

Skim milk (8 oz glass)

    <1/month

285 (54)

974 (57)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

29 (6)

152 (9)

0.7 (0.4–1.0)

0.7 (0.4–1.1)

    2/week–1/day

160 (30)

456 (27)

1.2 (1.0–1.5)

1.4 (1.1–1.8)

    ≥2/day

52 (10)

119 (7)

1.4 (1.0–2.0)

1.5 (1.1–2.2)

    Trend p

  

p = 0.03

p = 0.001

1% or 2% milk (8 oz glass)

    <1/month

160 (51)

637 (54)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

34 (11)

177 (15)

0.8 (0.6–1.3)

0.9 (0.6–1.3)

    2/week–1/day

102 (33)

305 (26)

1.3 (1.0–1.8)

1.4 (1.0–1.8)

    ≥2/day

17 (5)

68 (6)

0.9 (0.5–1.6)

0.8 (0.5–1.5)

    Trend p

  

p = 0.3

p = 0.3

Whole milk (8 oz glass)

    <1/month

409 (78)

1,465 (86)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

44 (8)

88 (5)

1.7 (1.2–2.6)

1.6 (1.1–2.3)

    2/week–1/day

54 (10)

115 (7)

1.5 (1.1–2.1)

1.2 (0.8–1.7)

    ≥2/day

19 (4)

33 (2)

1.7 (0.9–3.0)

1.2 (0.6–2.2)

    Trend p

  

p = 0.001

p = 0.1

Cream (1 tablespoon)

    <1/month

344 (65)

1,159 (68)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

96 (18)

320 (19)

1.0 (0.7–1.3)

1.0 (0.8–1.3)

    2/week–1/day

59 (11)

148 (9)

1.2 (0.9–1.7)

1.3 (0.9–1.8)

    ≥2/day

27 (5)

74 (4)

1.1 (0.7–1.8)

1.1 (0.7–1.7)

    Trend p

  

p = 0.4

p = 0.4

Frozen yogurt, sherbet or non-fat ice cream (1/2 cup)

    <1/month

348 (66)

1,129 (66)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

114 (22)

405 (24)

0.9 (0.7–1.2)

1.0 (0.8–1.3)

    ≥2/week

64 (12)

167 (10)

1.2 (0.8–1.6)

1.3 (0.9–1.8)

    Trend p

  

p = 0.7

p = 0.2

Ice cream (1/2 cup)

    <1/month

157 (30)

757 (45)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

173 (33)

625 (37)

1.3 (1.0–1.7)

1.3 (1.0–1.6)

    2/week–1/day

176 (33)

295 (17)

2.8 (2.1–3.6)

2.8 (2.1–3.7)

    ≥2/day

20 (4)

22 (1)

3.6 (1.9–6.9)

3.5 (1.8–6.9)

    Trend p

  

p < 0.0001

p < 0.0001

Flavored yogurt no Nutrasweet© (1 cup)

    <1/month

357 (68)

1,148 (68)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

96 (18)

337 (20)

0.9 (0.7–1.2)

1.0 (0.8–1.4)

    ≥2/week

73 (14)

214 (13)

1.1 (0.8–1.4)

1.2 (0.9–1.6)

    Trend p

  

p = 0.9

p = 0.3

Yogurt, plain or with Nutrasweet© (1 cup)

    <1/month

260 (83)

952 (80)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

30 (10)

143 (12)

0.8 (0.5–1.2)

0.8 (0.5–1.3)

    ≥2/week

23 (7)

91 (8)

0.9 (0.5–1.4)

1.0 (0.6–1.7)

    Trend p

  

p = 0.3

p = 0.7

Type of yogurt usually eatenc

    None

156 (50)

493 (42)

1.0 (ref.)

1.0 (ref.)

    Regular

31 (10)

96 (8)

1.0 (0.6–1.6)

1.1 (0.7–1.8)

    Low fat

78 (25)

316 (27)

0.8 (0.6–1.1)

0.9 (0.7–1.3)

    Non-fat

47 (15)

274 (23)

0.6 (0.4–0.8)

0.7 (0.4–1.0)

    Trend p

  

p = 0.004

p = 0.06

Cottage or ricotta cheese (1/2 cup)

    <1/month

247 (47)

950 (56)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

191 (36)

549 (32)

1.3 (1.1–1.6)

1.4 (1.1–1.7)

    ≥2/week

87 (17)

201 (12)

1.6 (1.2–2.1)

1.7 (1.2–2.3)

    Trend p

  

p = 0.001

p = 0.0003

Cream cheese (1 oz)

    <1/month

333 (63)

1,174 (69)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

153 (29)

410 (24)

1.3 (1.0–1.6)

1.4 (1.1–1.7)

    ≥2/week

40 (8)

117 (7)

1.1 (0.8–1.7)

1.2 (0.8–1.8)

    Trend p

  

p = 0.1

p = 0.03

Other cheese (1 slice or 1 oz)

    <1/month

50 (10)

240 (14)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

190 (36)

639 (38)

1.4 (1.0–2.0)

1.4 (1.0–2.0)

    2–4/week

168 (32)

545 (32)

1.4 (0.9–1.9)

1.4 (1.0–2.1)

    ≥5/week

118 (22)

277 (16)

1.8 (1.2–2.6)

1.8 (1.2–2.7)

    Trend p

  

p = 0.04

p = 0.02

Type of cheese usually eatenc

    None

8 (3)

33 (3)

1.0 (ref.)

1.0 (ref.)

    Non-fat

16 (5)

65 (5)

1.1 (0.4–2.9)

0.9 (0.3–2.4)

    Low fat or lite

32 (10)

199 (17)

0.7 (0.3–1.7)

0.6 (0.3–1.6)

    Regular

257 (82)

889 (75)

1.3 (0.6–2.9)

1.1 (0.5–2.6)

    Trend p

  

p = 0.06

p = 0.05

Butter (1 small pat or teaspoon)

    <1/month

246 (47)

891 (52)

1.0 (ref.)

1.0 (ref.)

    1/month–1/week

95 (18)

318 (19)

1.1 (0.8–1.4)

1.2 (0.9–1.5)

    2/week–1/day

124 (24)

409 (24)

1.0 (0.8–1.3)

1.1 (0.8–1.4)

    ≥2/day

61 (12)

82 (5)

2.3 (1.6–3.4)

2.4 (1.6–3.5)

    Trend p

  

p = 0.004

p = 0.002

aParsimonious model adjusted for age, sex and total energy intake (kcal/day, quartiles)

bAdditionally adjusted for body mass index (<25, 25.0–29.9, ≥30 kg/m2), race (white, black/African American, Asian/Pacific Islander, others), education (≤high school, college, graduate work), smoking (never smoker, former cigarette smoker who quit smoking >15 years ago, former cigarette smoker who quit smoking ≤15 years, current cigarette smoker, pipe and/or cigar smoker), and history of diabetes (yes, no)

c733 participants answered a questionnaire without these food items

We also considered combined dietary patterns of meat and fruits/vegetables. In general, the highest risks were observed among those who ate more red meat and fewer fruits and vegetables (data not shown), although there was no statistically significant multiplicative interaction (p = 0.9). We examined potential interaction effects of smoking on the associations between meat intake and risk of pancreatic cancer and found that there were no statistically significant interactions. There was also no evidence of trends (the OR’s (95% CI’s) of the fourth versus first quartile of red meat intake were 1.8 (0.9–3.9) among former smokers who quit more than 15 years ago, 1.3 (0.5–3.4) among former smokers who quit <15 years ago, and 1.4 (0.4–4.2) among current smokers) (data not shown in table). Compared with 10 years before the interview, the dietary pattern of participants one year before their cancer diagnosis or interview changed such that consumption of red meat, eggs, cheese, and butter decreased, and consumption of poultry and fish increased. Among the 36% of cases and 37% of controls whose consumption of red meat stayed the same as 10 years before, an increased risk of pancreatic cancer was observed for increasing amounts of red meat consumption. Persons in the highest quartile of red meat intake had an associated crude OR = 1.4 (1.0–1.9). However, with additional consideration for all of the factors in the full model, the OR was 1.1. Among participants who ate more red meat 10 years before the interview, those in the second to the fourth quartile of red-meat intake had age, sex and calorie-adjusted risks of pancreatic cancer of 1.7 (1.2–2.5), 1.8 (1.2–2.4) and 2.7 (1.5–3.4), (trend p = 0.004) compared with those in the lowest quartile of red-meat consumption (data not shown in table). However, after consideration for all of the factors in the full model, risk estimates dropped to 1.4, 1.5, and 1.5.

Discussion

Greater intake of meat and dairy products were positively associated with pancreatic cancer risk. The association for meat appeared to be due primarily to positive associations for beef/lamb products, or processed or higher fat red meat items (e.g., sausage, salami, sandwich meats, and bacon); hot dogs were the exception and were unassociated with risk. There were no consistent associations for chicken, turkey, total fish/shellfish. Pork was not independently associated with risk when assessed as a solo main dish, and this may account for some of the inconsistent associations for red meat or processed red meat as a food category because beef/lamb and pork were grouped together on several of the questionnaire items and could not be distinguished individually (e.g., sausage, sandwich meat, and mixed dish).

For dairy products, there was some suggestion that higher-fat items (i.e., ice cream, butter, and cheese) were more strongly positively associated with risk; although skim milk and not higher-fat milk was positively associated with risk. Total fat, in particular animal fat, saturated fat, and to a lesser extent monounsaturated fat, also were positively associated with pancreatic cancer risk. Cholesterol was positively associated with risk, although due to the high correlation between fat and cholesterol intake, it is unclear whether this was an independent association. There was some indication that cholesterol intake might be an independent indictor of pancreatic cancer risk based on the positive association for whole eggs and not egg whites. Results generally were similar for men and women, although among men the associations for fats and risk of pancreatic cancer were somewhat stronger.

These results are consistent with other case–control and some cohort studies. Some prospective cohort studies have observed positive associations for pancreatic cancer risk and intake of meat or animal protein, especially red meat [1112, 29, 30], processed meat [12], butter [13], or fat, especially saturated fat [13]. Another cohort study observed a borderline elevated risk of fatal pancreatic cancer associated with meat intake [31]. In contrast, other prospective studies have reported no association for meat, dairy products, fat, cholesterol, eggs [11, 15], a Western dietary pattern characterized by high meat and high-fat intake, or red meat specifically [15, 16]. Similar to our study, a few other cohort studies showed no association with fish consumption [11, 13, 15].

Several case–control studies have reported positive associations for some measure of meat, in particular red meat [10, 3236] or fat intake [34, 3739] and increased risk of pancreatic cancer. In some studies associations were observed differentially for men and women [36, 38]. At least one case–control study has observed a strong positive association with higher fish intake and risk of pancreatic cancer [38], whereas another observed an inverse association for poultry/fish among women only [17].

One case–control study estimated that the attributable risk percent for a high meat/low fruit diet was approximately 18–25% [33], although this study was conducted in an Italian population and may not be generalizable to the San Francisco Bay Area populations. A few case–control studies have observed positive associations for cholesterol intake [4042], or eggs [10, 42] and pancreatic cancer risk, and some have reported no association for meat [9, 17, 19], fats [18, 19], or a Western dietary pattern [14]. A minority of studies have reported inverse associations for saturated fat [17] or polyunsaturated fat [18].

The positive associations for beef/lamb and higher fat meat and dairy items with pancreatic cancer risk were consistent with the similar elevated risks observed for total, animal, saturated, and (to a lesser extent) monounsaturated fats and cholesterol. Hypothesized biological mechanisms underlying such associations are speculative, but include hormonal effects and carcinogenic heterocyclic amines (HCAs). Greater fat consumption may increase circulating levels of sex hormones [43, 44], and in vitro studies and animal models indicate that pancreatic tumors are sensitive to testosterone and estrogen, although data from human populations are conflicting [4552]. For example, women have a lower incidence of pancreatic cancer compared to men [1, 2, 53]. Also, among women, studies have suggested that higher parity may be associated with reduced risk [46, 54] although at least one study reported an increased risk [55]. Of note, in the current study population, there was no association between risk of pancreatic cancer and age at menarche, parity, oral contraceptive use, estrogen replacement therapy, or history of oophorectomy [56].

Studies in animal models also have indicated that dietary fat acts as a promoter in the carcinogenic process [57, 58]. Hamsters fed high-fat diets had a three- to four-fold higher risk of pancreatic carcinogenesis than did those fed a low-fat diet with equivalent calorie intake [58]. Fat consumption also may stimulate release of cholecystokinin that may contribute to the promotion of pancreatic cancer [59]. It is also possible that diets high in fat affect insulin resistance, and diabetes has been positively linked to pancreatic cancer risk, although it is uncertain whether there is a substantial risk of pancreatic cancer among individuals with long-term diabetes [20, 60, 61]. Of interest, a large recent case–control study reported a strong inverse association for statin use and pancreatic cancer risk that would be consistent with the observed positive associations between pancreatic cancer risk and fat and cholesterol consumption [62].

HCAs are generated when animal proteins are cooked at high temperatures, such as those achieved by grilling, frying, or barbecuing. Some studies have reported positive associations between pancreatic cancer and deep-fried, grilled, or barbecued meats [9, 32, 63], but not necessarily other types of cooked meats. In the current study, cooking methods were not assessed, and thus we have no data to indicate whether HCA content of some of the meat items may have been related to the observed associations.

The lack of associations for pork as a main dish or beef/pork hot dogs were inconsistent with our a priori hypotheses that red meat, processed meat, and fat might be positively associated with risk of pancreatic cancer. As several of the questions grouped pork and beef/lamb items together (e.g., sandwich meats, sausages, and hot dogs), it was difficult to further assess whether this difference between beef/lamb and pork persisted across different foods, and future studies may benefit from separating these food items on the questionnaire. The results for skim milk and not fattier milks were also unexpected and warrant further study.

Differences in results across studies also may be due to differences in study design, residual confounding, measurement error, bias, or variation in dietary patterns of the various study populations. Generally, case–control studies may be limited by recall or selection bias and the use of proxy interviews. On the other hand, prospective investigations often have a limited number of observed cases. Both study designs may be affected by measurement error that could further restrict their ability to observe smaller associations.

Our study had several strengths. It was a large population-based study. Our control participants were randomly selected from the general population in the geographic area, providing reasonable generalization to the general population of this age group. Because pancreatic cancer is relatively rare, the case–control study design is appropriate to identify an adequate number of patients to provide a large sample size within a reasonable time frame. In-person interviews also were conducted with participants, and measurement error due to proxy interviews was avoided. We used a validated semi-quantitative food-frequency questionnaire that has been used in many other populations. This study examined dietary patterns one year before the cancer diagnosis or interview for controls, thereby avoiding recent assessment of dietary changes due to pancreatic cancer. The refusal rate was low at 8% and the primary reason we lost patients was the aggressive process of pancreatic cancer and the subsequent rapid high mortality rate.

Our study also had several limitations. Like any case–control study, our investigation may have been affected by recall bias due to differential reporting of past diet by cases versus controls. While it is generally considered a strength that no proxy interviews were used, we must consider the possibility that some cases may have been experiencing physical discomfort due to their disease at the time of interview. However, our analyses of cases and controls with regard to interviewer perception of questionnaire reliability showed that there was a 1.0% difference in response between cases and controls, wherein 2.4% of interviewed cases were reported to have had some difficulty with the questionnaire due to illness, and 1.4% of controls were reported by the interviewers to have had this difficulty. Further, we observed that controls rather than cases were more likely to have changed their diets in the previous decade with control participants reporting having changed their diets towards more healthy choices over time [8]. This argues against the likelihood that the observed associations were due to recall bias among the cases. Additionally, because risk factors generally are not known for pancreatic cancer, we expect that differential recall between cases and controls was less likely to have played a substantial role. Due to the high mortality rate for pancreatic cancer, many patients died before they could be contacted for interview. However, rapid-case ascertainment was used to locate patients quickly. Given the differences in education level between cases versus controls, we explored the possibility of volunteer bias due to the participating controls being more highly educated and potentially healthier than the overall population that gave rise to the cases. We examined associations for the various meat groups and fat among only those participants with a college education or higher and results were similar to those found for all combined. Additionally, adjustment for education level alone did not substantially affect results in the multivariable models. Multiple comparisons were made in this study and some results could be due to chance. However, this study was designed to test specific hypotheses related to food and food groups, and the outcome of the analyses were in general consistent with our overall a priori hypotheses [64].

In conclusion, results from this large population-based case–control study have provided some support for positive associations between beef/lamb, processed or higher fat red-meat items, eggs, dairy, and fats and risk of pancreatic cancer. In contrast, chicken, turkey, total fish/shellfish, and egg whites were not associated with increased risk and may be preferred sources of protein when considering pancreatic cancer prevention. Earlier results from this study also indicated that higher vegetable and fruit intake was associated with a lower estimated risk of pancreatic cancer [8]. These findings warrant further study and confirmation given the high fatality rate of this disease and the lack of many identified modifiable risk factors. Particular attention should be paid to meat preparation methods in future work on pancreatic cancer and diet.

Acknowledgments

Dr. Holly is principal investigator of this project, designed and implemented the study, and oversaw and worked closely with Drs. Chan and Wang on the analyses and writing of this report. Dr. Chan participated in the development of the analysis plan, provided expertise in nutritional epidemiology of cancer, and led the background research/reporting/writing of this project. Dr. Wang conducted all statistical analyses/programming and participated in the writing of this report.

Grant support: National Institutes of Health, NCI grants CA59706, CA89726, CA09889, CA108370, CA121846, the Rombauer Pancreatic Cancer Research Fund and David J. Hasbun Pancreatic Cancer Research Fund.

Copyright information

© Springer Science + Business Media B.V. 2007