Cancer Causes & Control

, Volume 18, Issue 8, pp 809–819

Pancreatic cancer and medical history in a population-based case–control study in the San Francisco Bay Area, California

Authors

  • Andrew H. Ko
    • Division of Hematology/OncologyUniversity of California San Francisco
  • Furong Wang
    • Department of Epidemiology and Biostatistics, Division of Cancer Epidemiology, School of MedicineUniversity of California San Francisco
    • Department of Epidemiology and Biostatistics, Division of Cancer Epidemiology, School of MedicineUniversity of California San Francisco
    • Department of Health Research and Policy, School of MedicineStanford University
Original Paper

DOI: 10.1007/s10552-007-9024-6

Cite this article as:
Ko, A.H., Wang, F. & Holly, E.A. Cancer Causes Control (2007) 18: 809. doi:10.1007/s10552-007-9024-6

Abstract

Objective

To determine the association between pancreatic cancer and medical conditions.

Methods

A large population-based case–control study identified pancreatic cancer cases in the San Francisco Bay Area between 1995 and 1999. A total of 1,701 controls were randomly selected from the same population and were frequency-matched to 532 cases by sex and age. In-person interviews were conducted with no proxy interviews.

Results

Prior history of gallbladder disease was associated with increased risk of pancreatic cancer (OR = 1.6, 95% CI = 1.2–2.2), with the highest risk occurring for gallbladder disease not caused by cholelithiasis (OR = 2.1, 95% CI = 1.1–3.7). Risk was associated with cholelithiasis only for participants diagnosed within the year before their pancreatic cancer (OR = 15, 95% CI = 6.2–34), and for those with cholelithiasis and cholecystectomy within the same time frame (OR = 28, 95% CI = 8.2–96). Gastric and/or duodenal ulcers were associated with increased risk of pancreatic cancer for individuals with ulcers of ≤two years duration (OR = 2.3, 95% CI = 1.1–4.6). Hyperthyroidism (OR = 2.1, 95% CI = 1.0–4.2) and “other” thyroid conditions (OR = 2.2, 95% CI = 1.1–4.2) were associated with increased risk. Participants with at least one first-degree relative with pancreatic cancer had an increased risk (OR = 1.6, 95% CI = 1.1–2.5).

Conclusion

History of recent gallbladder conditions, gastric and/or duodenal ulcers may represent an early manifestation of pancreatic cancer rather than an independent risk factor. These results warrant further investigation in pooled analyses.

Keywords

Pancreatic neoplasmsGallbladder diseasesGastric ulcersDuodenal ulcersThyroid diseasesCase–control studies

Introduction

Pancreatic cancer is the fourth leading cause of cancer-associated mortality in the United States, with nearly 37,170 new patients estimated in the year 2007 and almost the same number of deaths each year [1]. Most patients with pancreatic cancer are diagnosed with advanced disease and once surgery is ruled out, other therapeutic options such as chemotherapy are of limited effectiveness. Consequently, stage for stage, pancreatic cancer is associated with the lowest survival rate of any cancer site [1]. The imperative clearly exists to determine modifiable risk factors, to develop methods to detect pancreatic cancer at earlier stages in the general population, and to identify individuals at high risk who would benefit from more intensive screening.

Previously published epidemiologic studies have examined a variety of factors that may influence the risk of pancreatic cancer, including cigarette use, diet, environmental exposures, body mass index, and medical conditions [214] and as reviewed in reference number 15 [15]. Several factors have consistently emerged in these studies as increasing the risk of pancreatic cancer. Cigarette smokers have an approximate 2-fold greater risk of pancreatic cancer compared with non-smokers, with cigarette smoking accounting for approximately 25% of pancreatic cancer incidence [11, 13, 14]. A history of pancreatic cancer among first-degree relatives also consistently has been associated with excess risk, especially as the number of affected first-degree relatives increases [1622], although hereditary pancreatic cancer accounts for only 5%–10% of all cases [23].

Other medical conditions have not been consistently found to be associated with pancreatic cancer. Pre-existing diabetes mellitus has been studied extensively as a potential risk factor for pancreatic cancer with inconsistent results [6, 8, 12, 2427]. One literature review and meta-analysis suggested that individuals with longstanding diabetes (i.e. greater than five years’ duration) had about a 2-fold increased risk of pancreatic cancer [6, 12]. A recently published meta-analysis that reviewed many of the same studies, as well as several new studies, reported that patients with diabetes for ≤four years had a 2-fold increased risk, but the risk was attenuated to 1.5 for patients who had had their diabetes for ≥five years [24]. Results from our study that evaluated the association between history of diabetes and pancreatic cancer have been published and showed similar results to those of the second meta-analysis, whereby risk was attenuated for those with long-term diabetes [28]. Several other potentially relevant medical conditions commonly seen in Western populations, including gallbladder diseases, gastric and/or duodenal ulcers, pernicious anemia, hypertension, and thyroid diseases may represent risk factors for pancreatic cancer, although prior population-based studies have produced inconclusive results. The purpose of this analysis was to determine the relation between pancreatic cancer risk and a history of predefined medical conditions including family history.

Materials and methods

Study population

A population-based case–control study of pancreatic cancer was conducted in the San Francisco Bay Area. Details on study design and selection of the study population have been published [5, 9, 2833]. In brief, between 1995 and 1999, newly diagnosed patients with adenocarcinoma of the pancreas in the San Francisco Bay Area were identified using rapid case ascertainment through the Surveillance, Epidemiology and End Results (SEER) tumor registry. Eligible cases were 21–85 years of age at diagnosis, resided in one of six counties (Alameda, Contra Costa, Marin, Santa Clara, San Francisco, San Mateo) of the San Francisco Bay Area, were alive when contacts were attempted and could complete an interview in English. We also recruited out-of-area cases to increase the number of pancreatic cancer patients for this study due to the high mortality rate of this cancer. Out-of-area cases (n = 65 interviewed), who were residents of counties adjacent to the six Bay Area counties, were identified through clinical records at the University of California San Francisco and had the same eligibility requirements as in-area cases except for residency. Among 797 eligible cases, 532 (67%, 291 men and 241 women) completed an interview. Diagnoses of pancreatic cancer were confirmed by participants’ physicians and by the SEER abstracts. Control participants were population-based, were selected from the target population using random digit dial (RDD), and were frequency-matched to cases by sex and by age within five-year categories. Control participants older than 65 years were supplemented by random selection from the Centers for Medicare & Medicaid Services lists that covered the same six San Francisco Bay Area counties. 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. Among 2,525 eligible controls, 1,701 (67%, 883 men and 813 women) completed an interview and their data were included in the final analyses.

Eligible cases and controls were interviewed in-person by trained interviewers using the main questionnaire and a supplementary food-frequency questionnaire. No proxy interviews were conducted in this study. The study was reviewed and approved by the University of California San Francisco institutional review board. Written informed consent was obtained from each participant prior to interview. Items in the main questionnaire included demographic factors, smoking, alcohol consumption, medical, and occupational history, and family history of pancreatitis, diabetes, and pancreatic cancer.

Participants were asked to report their medical history prior to their diagnosis (cases) or interview (controls), including diabetes, pancreatitis, gallbladder diseases, gastric and/or duodenal ulcers, pernicious anemia, hypertension, thyroid diseases (women only), cancer history, chemotherapy for cancer and family history of pancreatic cancer.

Statistical methods

Odds ratios (OR) and 95% confidence intervals (CI) were computed from unconditional logistic regression models and were used as estimates of the relative risk (hereafter called risk) of pancreatic cancer by exposure group. Effect modification and potential confounding were assessed for cigarette smoking, body mass index (BMI), race, education, and history of diabetes using stratification and modeling. All statistical tests were two-sided and considered statistically significant when p ≤ 0.05. Statistical analyses were conducted using SAS software version number 9.1 (SAS Institute, Inc., Cary, NC).

Age at pancreatic cancer diagnosis for cases was matched on age at interview for controls and was categorized in groups as <50, 50–54, 55–59, 60–64, 65–69, 70–74, 75–79, and 80–85 years. BMI was estimated from adult weight and height (weight (kg) / height (m)2) as a measure of body size. BMI was examined using the World Health Organization (WHO) weight categories: <25, 25.0–29.9, ≥30 kg/m2. Smokers were defined as participants who had smoked more than 100 cigarettes in their lifetime, or a pipe or cigar for at least once a month for ≥six months. Smoking was examined using the following categories: never smokers, former cigarette smokers who quit smoking >15 years ago, former cigarette smokers who quit ≤15 years ago, current cigarette smokers, and pipe/cigar smokers. History of physician-diagnosed diabetes mellitus was self-reported by participants as yes, or no. Race included white, black or African American, Asian or Pacific Islander, American Indian or Alaskan Native, or other. Hispanic ethnicity was self-reported as yes, or no. Education level was self-reported by participants as ≤high school (1–12 years), some college (13–16 years), or graduate school (≥17 years). None of these factors was an effect modifier or a confounder and no individual item changed the risk estimates by >10%. Two multivariable analyses were presented. One model adjusted for the matching factors of sex and age and thus was the most parsimonious model. The second model additionally adjusted for the putative risk factors of race, education, BMI, history of diabetes, and smoking status as defined above.

Results

The demographic characteristics of the pancreatic cancer patients and controls are presented in Table 1. Table 2 highlights the increased risk of pancreatic cancer associated with gallbladder diseases, both cholelithiasis-related and non-cholelithiasis-related. Whereas the excess risk associated with cholelithiasis was statistically significant, temporal analyses revealed that only individuals who had been diagnosed with cholelithiasis within the past one year before their diagnosis with pancreatic cancer or interview for controls were at increased risk (Table 2A). By contrast, those who reported having had their diagnosis with cholelithiasis ≥two years were not at increased risk. Patients who underwent cholecystectomy for any indication also were at increased risk of pancreatic cancer (Table 2A). Again, there was a clear temporal association, with the increased risk limited exclusively to individuals who had undergone cholecystectomy within the past year (Table 2A). Patients with cholelithiasis without cholecystectomy did not have a statistically significant excess risk of pancreatic cancer (Table 2A), although the odds ratio was elevated 3.5 fold for cholelithiasis duration of one year or less (Table 2B). Gallbladder diseases not related to cholelithiasis were associated with increased risk of pancreatic cancer. This association was found in both participants who had undergone a cholecystectomy and those who had not (Table 2A). Further, those with cholelithiasis and cholecystectomy within one year prior to pancreatic cancer diagnosis had a 28-fold risk (Table 2B).
Table 1

Demographic characteristics of pancreatic cancer patients and control participants in a population-based case-control study in the San Francisco Bay Area, California

Characteristics

Cases (N = 532)

Controls (N = 1,701)

No.

%

No.

%

Agea, years

    <50

46

9

164

10

    50–54

54

10

226

13

    55–59

66

12

212

12

    60–64

83

16

169

10

    65–69

89

17

304

18

    70–74

98

18

299

18

    75–79

60

11

199

12

    80–85

36

7

128

8

Sex

    Men

291

55

883

52

    Women

241

45

818

48

Race

    White

442

83

1,471

86

    Black or African American

46

9

78

5

    Asian or Pacific Islander

35

7

119

7

    American Indian or Alaskan Native or Other

9

2

33

2

Hispanic Ethnicity

    Yes

25

5

114

7

    No

507

95

1585

93

Body Mass Index (WHO), kg/m2

    <25.0

280

53

993

58

    25.0–29.9

197

37

552

33

    ≥30

52

10

147

9

Body Mass Index (quartiles), kg/m2

    ≤22.1

107

20

425

25

    22.2–24.2

124

23

426

25

    24.3–26.5

142

27

419

25

    ≥26.6

159

30

431

25

Smoking

    Never

163

31

652

38

    Cigarette

        Former, quit > 15 years ago

133

25

508

30

        Former, quit ≤ 15 years ago

107

20

271

16

        Current

113

21

194

11

    Pipe/cigar

16

3

73

4

Education

    ≤High school

235

44

534

31

    Some college

200

38

754

44

    Graduate school

97

18

413

24

History of diabetes mellitus

    No

455

86

1538

90

    Yes

76

11

161

10

Numbers may not add to 100% due to missing values

aAge at pancreatic cancer diagnosis for cases or age at interview for controls

Table 2

Gallbladder conditions in patients with pancreatic cancer and in control participants in a population-based case-control study in the San Francisco Bay Area, California

A. Gallbladder conditions

Cases

Controls

Odds Ratioa (95% Confidence Interval)

Odds Ratiob (95% Confidence Interval)

No.

No.

Gallbladder conditions

    No

432

1,493

1.0 (referent)

1.0 (referent)

    Yes

99

207

1.7 (1.3, 2.3)

1.6 (1.2, 2.2)

Cholelithiasis

    Yes

75

172

1.6 (1.2, 2.1)

1.5 (1.1, 2.1)

    Other gallbladder diseases, no cholelithiasisc

21

34

2.3 (1.3, 4.0)

2.1 (1.1, 3.7)

Age at first cholelithiasis diagnosis, years

    <50

28

84

1.2 (0.78, 1.9)

1.2 (0.73, 1.8)

    50–59

22

37

2.0 (1.2, 3.4)

1.8 (1.0, 3.2)

    ≥60

25

48

1.7 (1.1, 2.9)

1.8 (1.1, 2.9)

Years between cholelithiasis diagnosis and pancreatic cancer diagnosis/interview

    ≤1

26

7

13 (5.7, 31)

15 (6.2, 34)

    ≥2

49

167

1.1 (0.76, 1.5)

1.0 (0.70, 1.4)

Cholecystectomy, Yes

    For cholelithiasis

62

132

1.7 (1.2, 2.4)

1.7 (1.2, 2.3)

    For other gallbladder disease

13

23

2.1 (1.0, 4.2)

1.9 (0.89, 3.9)

No cholecystectomy

    For cholelithiasis

12

40

1.0 (0.54, 2.0)

0.93 (0.47, 1.8)

    For other gallbladder disease

9

12

2.8 (1.1, 6.6)

2.5 (1.0, 6.3)

Years between cholecystectomy and pancreatic cancer diagnosis/interview

    ≤1

31

7

16 (6.9, 36)

16 (6.8, 37)

    ≥2

43

148

1.1 (0.73, 1.5)

1.0 (0.69, 1.5)

B. Cholecystectomy for cholelithiasis

Cholecystectomy, Yes

    Years between cholelithiasis diagnosis and pancreatic cancer diagnosis/interview

        ≤1

21

3

26 (7.7, 88)

28 (8.2, 96)

        ≥2

41

126

1.2 (0.80, 1.7)

1.1 (0.76, 1.6)

No cholecystectomy

    Years between cholelithiasis diagnosis and pancreatic cancer diagnosis/interview, years

       ≤1

4

4

3.3 (0.82, 14)

3.5 (0.81, 15)

       ≥2

8

36

0.77 (0.36, 1.7)

0.68 (0.31, 1.5)

aAdjusted for age (five years) and sex; n may not add to totals due to missing values

bAdditonally adjusted for race, education, body mass index (<25, 25.0–29.9, ≥30 kg/m2), 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)

cCholelithiasis may have been passed prior to ultrasound

No overall excess risk was found for a history of gastric and/or duodenal ulcers (Table 3). Participants who first developed ulcers between the ages of 50 and 59 years showed an increased risk for pancreatic cancer, although confidence intervals overlapped unity for all age groups after multivariable-adjustment for putative risk factors. Participants whose duration of ulcer disease was two years or less also had an increased risk of pancreatic cancer. Neither the use of anti-ulcer medication nor a prior history of surgery for ulcer disease was associated with increased risk.
Table 3

Gastric and/or duodenal ulcers in patients with pancreatic cancer and in control participants in a population-based case-control study in the San Francisco Bay Area, California

Gastric and/or duodenal ulcers

Cases

Controls

Odds Ratioa (95% confidence interval)

Odds Ratiob (95% confidence interval)

No.

No.

Ulcer

    No

447

1,462

1.0 (referent)

1.0 (referent)

    Yes

84

238

1.1 (0.87, 1.5)

1.0 (0.75, 1.3)

Age at first ulcer, years

    <50

48

168

0.93 (0.66, 1.3)

0.81 (0.57, 1.2)

    50–59

17

25

2.2 (1.2 , 4.1)

1.7 (0.89, 3.3)

    ≥60

19

44

1.4 (0.81, 2.5)

1.2 (0.70, 2.2)

Ulcer duration, years

    ≤2

17

20

2.7 (1.4, 5.3)

2.3 (1.1, 4.6)

    3–10

10

35

0.93 (0.46, 1.9)

0.80 (0.38, 1.7)

    >10

57

182

1.0 (0.74, 1.4)

0.89 (0.64, 1.2)

Ulcer medication use

    Yes

62

186

1.1 (0.81, 1.5)

0.93 (0.67, 1.3)

    No

20

51

1.2 (0.72, 2.1)

1.2 (0.67, 2.0)

Ulcer surgery

    Yes

11

26

1.3 (0.63, 2.7)

1.2 (0.57, 2.5)

    No

72

211

1.1 (0.83, 1.5)

0.96 (0.71, 1.3)

aAdjusted for age (five years) and sex; n may not add to totals due to missing values

bAddtionally adjusted for race, education, body mass index (<25, 25.0–29.9, ≥30 kg/m2), 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)

Neither pernicious anemia nor a history of hypertension that required medical intervention was associated with an increased risk for pancreatic cancer (Table 4). Thyroid conditions were examined among women only due to lower incidence among men. Hypothyroidism was not associated with any increased risk. A history of hyperthyroidism was associated with a greater than 2-fold increased risk as was the “other” category. A history of partial thyroidectomy was associated with a 5-fold increased risk, although no excess risk was observed in individuals requiring a total thyroidectomy. Among hyperthyroid patients, total years of use of thyroid medication showed increasing risk with longer duration of medication use.
Table 4

Other medical conditions in patients with pancreatic cancer and in control participants in a population-based case-control study in the San Francisco Bay Area, California

Other medical conditions

Cases

Controls

Odds Ratioa (95% confidence interval)

Odds Ratiob (95% confidence interval)

No.

No.

Pernicious anemia

    No

518

1,641

1.0 (referent)

1.0 (referent)

    Yes

13

56

0.77 (0.41, 1.4)

0.70 (0.37, 1.3)

Hypertension treatment

    No

405

1,326

1.0 (referent)

1.0 (referent)

    Yes

121

365

1.1 (0.85, 1.4)

0.94 (0.73, 1.2)

        First age treatment, year

            <50

33

112

0.95 (0.63, 1.4)

0.85 (0.55, 1.3)

            50–59

39

96

1.3 (0.85, 1.9)

1.1 (0.75, 1.7)

            ≥60

47

153

1.0 (0.71, 1.5)

0.88 (0.62, 1.3)

        Total years of treatment

            ≤1

29

90

1.0 (0.67, 1.6)

0.92 (0.58, 1.4)

            2–5

33

92

1.2 (0.77, 1.8)

0.97 (0.62, 1.5)

            6–10

20

59

1.1 (0.66, 1.9)

0.99 (0.58, 1.7)

            >10

37

122

0.97 (0.66, 1.4)

0.89 (0.60, 1.3)

Thyroid conditions, women

    No

153

562

1.0 (referent)

1.0 (referent)

    Yes: Hypothyroid

55

200

0.99 (0.69, 1.4)

1.1 (0.75, 1.6)

            Hyperthyroid

15

25

2.3 (1.2, 4.5)

2.1 (1.0, 4.2)

            Otherc

17

29

2.2 (1.2, 4.1)

2.2 (1.1, 4.2)

    Thyroid removed

            Yes

6

21

1.0 (0.40, 2.6)

1.0 (0.40, 2.7)

            No

73

228

1.2 (0.84, 1.6)

1.2 (0.88, 1.7)

            Partially removed

8

6

5.0 (1.7, 15)

5.3 (1.8, 16)

    Thyroid medication use

            Hypothyroid

53

191

0.99 (0.70, 1.4)

1.1 (0.74, 1.6)

            Hyperthyroid

13

20

2.5 (1.2, 5.1)

2.3 (1.1, 5.1)

            Other thyroid problems

10

22

1.6 (0.74, 3.5)

1.6 (0.71, 3.5)

    Hypothyroid medication, years

        ≤5

27

93

1.0 (0.65, 1.7)

1.1 (0.64, 1.7)

        >5

25

97

0.94 (0.58, 1.5)

1.1 (0.66–1.8)

    Hyperthyroid medication, years

        ≤5

6

12

1.8 (0.67, 5.0)

1.6 (0.57, 4.7)

        >5

6

8

3.0 (1.0, 8.9)

3.2 (1.0, 10)

aAdjusted for age and sex; n may not add to totals due to missing values, e.g. 5 cases and 2 controls had unknown hyperthyroid or hypothyroid medication use

bAddtionally adjusted for race, education, body mass index (<25, 25.0–29.9, ≥30 kg/m2), 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)

cBoth hypothyroid and hyperthyroid problems at different times; thyroid problem other than hypothyroid and hyperthyroid; or could not remember whether hypothyroid or hyperthyroid

Overall, a history of any prior malignancy was not associated with risk of subsequently developing pancreatic cancer. A history of any one of the most commonly identified site-specific cancers in this study also was not associated with risk of pancreatic cancer, although the risk estimate was somewhat elevated for breast cancer and melanoma. Chemotherapy for prior malignancy was not associated with risk of pancreatic cancer (Table 5).
Table 5

Previous malignancies in patients with pancreatic cancer and in control participants in a population-based case-control study in the San Francisco Bay Area, California

 

Cases

Controls

Odds Ratioa (95% confidence interval)

Odds Ratiob (95% confidence interval)

No.

%

No.

%

Prior malignancy

No

457

86

1,477

87

1.0 (referent)

1.0 (referent)

Yes: All cancersc

75

14

223

13

1.1 (0.81, 1.4)

1.2 (0.89, 1.6)

    Breast cancerd

17

7

45

6

1.3 (0.74, 2.4)

1.6 (0.86, 2.9)

    Prostate cancere

13

5

53

6

0.74 (0.39, 1.4)

0.80 (0.42, 1.5)

    Melanoma

10

2

28

2

1.1 (0.52, 2.3)

1.3 (0.63, 2.8)

    Colon cancer

6

1

35

2

0.59 (0.24, 1.4)

0.65 (0.27, 1.6)

    Uterusd

6

3

9

1

    Ovaryd

3

1

3

0.4

    Non-Hodgkin lymphoma

3

0.6

3

0.2

    Soft and connective tissue

3

0.6

0

0

    Lung

2

0.4

12

0.7

    Bladder

2

0.4

10

0.6

    Thyroid

2

0.4

4

0.2

    Kidney

2

0.4

3

0.2

    Cervixd

1

0.4

5

0.6

    Testise

1

0.3

2

0.2

    Eye, inc. ocular melanoma

1

0.2

1

0.1

    Bone, inc. jawbone

1

0.2

0

0

    Tongue

1

0.2

0

0

    Salivary gland

1

0.2

0

0

Chemotherapy for prior malignancy

No

64

12

183

11

1.1 (0.81, 1.5)

1.2 (0.89, 1.7)

Yes

11

2

38

2

0.96 (0.49, 1.9)

1.1 (0.54, 2.1)

Family history of pancreatic cancer

No

482

90

1,590

93

1.0 (referent)

1.0 (referent)

Yes

37

7

70

4

1.7 (1.1, 2.6)

1.6 (1.1, 2.5)

    Mother/father/brothers/sisters

32

5

58

3

1.8 (1.1, 2.8)

1.7 (1.1, 2.6)

    Other relativesf

5

2

12

0.7

1.3 (0.44, 3.6)

1.3 (0.45, 3.9)

    Unknown

12

2

39

2

1.0 (0.54, 2.0)

0.88 (0.44, 1.8)

aAdjusted for age (five years) and sex; n may not add to totals due to missing values

bAdditionally adjusted for race, education, body mass index (<25, 25.0–29.9, ≥30 kg/m2), 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)

c3 cases reported two cancer events at the same site, 12 controls reported two types of cancers, 6 controls reported two cancer events at the same site including 2 controls who reported two separate primary breast cancers. If a cancer site was reported as ≥2 events, for these analyses, it was coded as one cancer site

0 cases and ≤4 controls had cancers on each of the following sites: 4 stomach, 2 pituitary gland, 2 adrenal gland, 2 Hodgkin lymphoma, 2 throat or pharyngeal, 2 rectum, 1 peritoneum/retroperitoneal tissue, 1 other neural system, 1 hairy-cell leukemia, 1 chronic-lymphocytic leukemia, 1 leukemia, NOS, 1 essential thrombocythemia, 1 Waldenstrom’s macrogloblinemia, 1 labia, 1 nasopharynx

dPercentage among women (total 241 cases and 818 controls)

ePercentage among men (total 291 cases and 838 controls)

fGrandparents (n = 4), aunt/uncle (n = 8), son (n = 1), cousin (n = 1), nephew/niece (n = 3)

Self-reported family history of pancreatic cancer was associated with a 1.6-fold increased risk of pancreatic cancer (Table 5). Participants who reported first-degree relatives with this cancer (7% cases, 4% controls) had an increased risk of pancreatic cancer, whereas positive history among other relatives conferred no elevated risk. Too few patients had more than one affected relative to ascertain whether a greater number of affected relatives was associated with a greater increase in pancreatic cancer risk.

Discussion

In this large population-based case–control study, several medical conditions were evaluated to determine their association with pancreatic cancer. These study results and our temporal analyses suggest that history of some medical conditions may be an early manifestation of pancreatic cancer.

Gallbladder diseases

A history of gallbladder disease, including both cholelithiasis and non-cholelithiasis-related, was associated with an increased risk of pancreatic cancer in our study, although examination of temporal trends suggests this increased risk was a consequence of the cancer rather than an independent risk factor. Specifically, the increased risk was seen only among participants in whom cholelithiasis was diagnosed within a year of the time of diagnosis or interview for the controls.

Epidemiologic studies that have investigated the potential association between cholelithiasis/cholecystectomy and risk of pancreatic cancer have produced inconsistent results [8, 2527, 3452]. Several early case–control studies showed no association between gallbladder disease and pancreatic cancer [25, 26, 49]. However, subsequent case–control studies have reported an increased risk of pancreatic cancer ranging from 1.4- to 2.8-fold [27, 41, 52, 53]. Those studies that examined a temporal association, as we did in our current study, found that differences in pancreatic cancer risk depended upon the interval between cholelithiasis diagnosis or cholecystectomy and the diagnosis of pancreatic cancer. While some studies found an excess risk for pancreatic cancer that continued up to two decades following cholecystectomy [8, 43], this risk commonly diminished with time. In one large case–control study, the odds ratio for pancreatic cancer was 58 (95% CI, 27–123) for individuals who had undergone cholecystectomy within one year before the cancer diagnosis, falling to 1.7 (95% CI = 1.0–3.0) for those whose cholecystectomy was ≥20 years previously [8]. Another large case–control study showed a positive association for patients who had a history of cholelithiasis (OR = 1.4; 95% CI = 1.0–1.9), but noted that this association was diminished to the null in patients whose diagnosis of cholethiasis had been made more than one year previously [52]. These results, similar to ours, suggested that cholelithiasis and the need for cholecystectomy are reflective of the presence of pancreatic cancer rather than representing true risk factors.

In addition to case–control studies, large cohort studies from the United States, Europe, and Japan that addressed this question have produced mixed results [25, 3436, 3840, 43, 44, 47, 48, 54]. Analysis of combined cohorts from the Nurses’ Health Study and the Health Professionals Follow-Up Study (n = 153,184) did not report cholelithiasis or prior cholecystectomy as significant risk factors for pancreatic cancer (RR = 1.1) [34]. Conversely, other large population-based studies have suggested that cholecystectomy increased pancreatic cancer risk up to 2-fold [39, 40, 43, 44]. Whereas we found no difference by sex, the positive association between cholelithiasis/cholecystectomy and pancreatic cancer was sex-specific in other studies, with an increased risk in men in one study and an increased risk in women in another [35, 39].

Gastric and/or duodenal ulcers

Prior epidemiologic studies that have examined history of peptic ulcer disease and pancreatic cancer have found no association between the two [10, 25, 26, 36, 5456]. Epidemiologic data are mixed regarding the association between peptic ulcer surgery and pancreatic cancer. A cohort study of 2,633 patients who underwent gastrectomy at the University of Amsterdam over a 30-year period reported an overall increased risk for pancreatic cancer of 1.8 (95% CI = 1.3–2.6) [57]. Another large cohort study reported a 4-fold increased risk following gastrectomy, although this risk was detected only after a 20-year latency period [58]. Similar to our results, other studies have failed to find a positive association, with no excess risk regardless of whether study participants had undergone surgery for ulcer disease [10, 47, 59]. However, it should be noted that the number of participants who had surgery for their ulcer disease was small. Also, our study did not distinguish between gastric vs. duodenal ulcers, a possibly important distinction, as ulcer location may produce different degrees of risk for pancreatic cancer [60]. It is unlikely that the subgroups that reflected positive associations in our study (e.g. patients with ulcer disease in the sixth decade of life or duration of ulcers for less than two years) represent unique populations at particularly high risk. Most likely, these short-term risks noted in close proximity to the cancer indicate that ulcer disease may have been a result of the pancreatic cancer rather than an independent risk factor, similar to our findings for gallbladder disease, diabetes [28] and pancreatitis.

Pernicious anemia

Some epidemiologic studies, primarily from Sweden, have suggested a positive association between pernicious anemia and risk of pancreatic cancer [61, 62]. One study that examined a population-based cohort of more than 4,500 Swedish patients with pernicious anemia reported more cases of pancreatic cancer (n = 34) over a 20-year follow-up period than expected in the general population (standardized incidence rate 1.7, 95% CI = 1.2–2.4) [62]. Increased risk was greatest in patients in whom the diagnosis of pernicious anemia was made at a younger age [61]. A plausible biologic mechanism could explain an association between pernicious anemia and pancreatic cancer, whereby patients secrete excess gastrin, a trophic factor for pancreatic cancer cells [63, 64]. In contrast to these findings, no excess risk of pancreatic cancer was identified in our study in association with pernicious anemia.

Hypertension

No increased risk of pancreatic cancer was seen with any duration of antihypertensive medication use in our study or for age at first treatment. Few studies have examined this question [36, 45, 51]. One cohort study of Finnish male smokers reported a positive association between hypertension measured at baseline and subsequent development of pancreatic cancer, although subsequent self-reported history of hypertension was not associated with increased risk [36].

Thyroid disease

We found that hyperthyroidism, but not hypothyroidism, was suggestive of excess pancreatic cancer risk. This result is consistent with several other reports [4951, 65, 66]. One follow-up study of 1,762 women treated for hyperthyroidism suggested a 2-fold risk of pancreatic cancer (n = 10) over an average follow-up period of 17 years (95% CI = 1.0–3.7) [66]. These same researchers reported a 2.6-fold increased risk of pancreatic cancer-associated deaths (n = 14) in hyperthyroid women treated for a minimum of one year (95% CI = 1.4–4.3) [65]. This association requires replication in other studies because of the small number of pancreatic cancer patients in the cohort studies and a poorly defined mechanistic link between overactive thyroid and pancreatic tumorigenesis. Interestingly, a history of partial thyroidectomy also was associated with increased pancreatic cancer risk in our study, although the number of such procedures was small in both cases and controls. Earlier case–control studies also have observed a somewhat higher rate of thyroidectomies in cases than controls [49, 51], although numbers in these studies were too small to draw firm conclusions.

Other prior malignancy

Cancers known to be associated with pancreatic cancer based on well-defined genetic syndromes were examined with special interest, including breast cancer (BRCA2 carriers), colon cancer (hereditary non-polyposis colorectal cancer) and melanoma (familial atypical multiple mole-melanoma). While most of these cancers occur sporadically, individuals who harbor germline mutations predisposing them to these cancers have an increased risk of pancreatic cancer ranging from 3.5- to 22-fold [6771]. However, we did not observe a statistically significant increased risk of pancreatic cancer in individuals who had had a prior history of any of these site-specific malignancies. This is not surprising given that the vast majority of our cases were likely to have had sporadic pancreatic cancer and only 5% of pancreatic cancer cases fall into one of the known hereditary cancer syndromes [72].

Family history of pancreatic cancer

Similar to our study results, it has been reported that 5% to 10% of patients with pancreatic cancer have one or more first-degree relatives with this disease [23]. Several population-based case–control studies have demonstrated that a family history of pancreatic cancer represents a risk factor for the disease, with an increased risk ranging from 2.5- to 13-fold [8, 18, 2022]. Since familial pancreatic cancer is rare, we had too few familial patients in our sample to determine the degree of risk for pancreatic cancer with an increasing number of affected family members.

Strengths and potential limitations of our study should be noted. Since our study relied exclusively on in-person interviews with study participants, we were able to avoid some of the potential inaccuracies associated with proxy interviews. Our questionnaire was designed to obtain a wide variety of information to evaluate the impact of potential confounders and effect modifiers on risk estimates. We used rapid-case ascertainment to minimize possible selection bias inherent in case–control studies of pancreatic cancer, a disease characterized by short survival. However, it is conceivable that interviewed patients may have had different degrees of risk and different tumor characteristics compared with individuals who died before they could be interviewed. Information from SEER abstracts showed that non-interviewed patients included more women and minorities, tended to be slightly older, and had more unknown tumor characteristics.

As in all case–control studies, potential for recall bias may have influenced some of our study results. Persons with pancreatic cancer could be more likely to ask their relatives about familial history of their disease than would persons in the control group. None of the medical conditions or family history was confirmed using medical records. Small numbers of participants in some subgroups may be a concern and will need confirmation in pooled analyses with a larger number of participants. These study results and our temporal analyses suggest that history of recent gallbladder conditions may be an early manifestation of pancreatic cancer in some individuals. Pancreatic cancer should be considered in the differential diagnosis of high-risk individuals with recent episodes of apparent gallbladder problems of undetermined origin.

We have begun an additional large NCI-funded five-year clinic-based case–control study to address pancreatic cancer risk, to confirm and refine our results, and to examine new molecular and other risk factors. Our results when pooled with results from our earlier investigation and those from other investigators should help to clarify populations at risk for pancreatic cancer in whom more intensive screening programs may be warranted.

Acknowledgments

Grant support: National Institutes of Health, National Cancer Institute grants CA59706, CA108370, CA109767, CA121846, the Rombauer Pancreatic Cancer Research Fund and David J. Hasbun Pancreatic Cancer Research Fund.

Copyright information

© Springer Science + Business Media B.V. 2007