- Email: [email protected]
Hormonal Factors and Pancreatic Cancer in Women
Hormonal Factors and Pancreatic Cancer in Women NANCY KREIGER, MPH, PhD, JEANIE LACROIX, MPA, AND MARGARET SLOAN, BA
PURPOSE: The aim of the study was to evaluate hormonal risk factors for carcinoma of the exocrine pancreas among postmenopausal women. METHODS: Mailed questionnaire data from 52 cases and 233 population-based controls in Ontario were used to assess parity, age at first birth, and other hormonal factors on pancreatic cancer risk. RESULTS: Reduced risk was seen with three or more pregnancies [adjusted odds ratio (OR) 0.22, 95% confidence interval (CI) 0.07–0.65] and use of oral contraceptives (adjusted OR 0.36, 95% CI 0.13–0.96), whereas no significant associations were found for age at menarche or menopause, or estrogen replacement therapy. Among parous women, later age at first full-term pregnancy significantly increased the risk of this cancer (adjusted OR 4.05, 95% CI 1.50–10.92 for ages 25–29 years, adjusted OR 3.78, 95% CI 1.02–14.06 for ages 30 years). CONCLUSIONS: Our data support the hypothesis that pancreatic cancer is, at least in part, an estrogen-dependent disease; there is growing epidemiological evidence that aspects of reproductive history and hormonal exposure are associated with risk of this disease. Ann Epidemiol 2001;11:563-567. © 2001 Elsevier Science Inc. All rights reserved. KEY WORDS:
Pancreatic Cancer, Hormonal Factors, Case-Control Study.
INTRODUCTION
METHODS
Although cancer of the pancreas is associated with a substantial disease burden, both in morbidity and mortality (1), very little is known about its etiology. Aside from cigarette smoking (2–7), there are no well-substantiated risk factors, and little evidence that points to preventive measures. In the last few years, focus has shifted to an examination of the role that female reproductive factors might play in the onset of pancreatic cancer. While not completely consistent, the literature suggests that age at menarche, aspects of pregnancy history, and exogenous hormone exposure all may be associated with altered risk, particularly in postmenopausal women (8–11). Within the context of a large, population-based casecontrol study, we analyzed the reproductive data for postmenopausal women with a new diagnosis of primary carcinoma of the exocrine pancreas, in order to provide further clarification of the relationship between reproductive factors and this disease.
Data were derived from the case-control component of the Enhanced Cancer Surveillance project (ECS), a multi-centre collaborative study of cancer and the environment, conducted in eight of the ten Canadian provinces. Analyses presented here are restricted to the Ontario data as most of the reproductive information was not included in the questionnaires used by the other provinces. Eligible Ontario cases in Enhanced Cancer Surveillance were diagnosed between January 1995 and September 1996. Individuals were aged 20 to 74 years old, were residents of Ontario, and were diagnosed with one of ten cancers. Controls were a sample of the 1995 Ontario Property Assessment file, which contains name, address, sex, and year and month of birth for all residents of the province. It is approximately 95% complete, with no systematic exclusions. In each sex and five-year age stratum, the number of controls was determined by the largest number of cases, among all ten cancer sites, expected in that stratum. This ensured a minimum of 1:1 frequency matching in each stratum, regardless of cancer site being analyzed. The response rate for all female pancreatic cancer cases (including surrogate respondents) was 71.4%, and the response rate for female controls was 80%. A mailed questionnaire focused on a wide array of potential risk factors for the cancers covered by the ECS, including reproductive history for women, cigarette smoking, physical activity, diet history using a food frequency list, occupational and residential histories, and sociodemographic information.
From the Division of Preventive Oncology, Research Unit, Cancer Care Ontario (NK, MS), Toronto, Ontario, Canada; Department of Public Health Sciences, University of Toronto (NK), Toronto, Ontario, Canada; and Trauma Registries, Clinical Data Services, Canadian Institute for Health Information (JL), Toronto, Ontario, Canada. Address reprint requests to: Nancy Kreiger, Division of Preventive Oncology, Cancer Care Ontario, 620 University Avenue, Toronto, Ontario M5G 2L7 Canada. Received April 21, 2000; revised December 19, 2000; accepted December 28, 2000. © 2001 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
1047-2797/01/$–see front matter PII S1047-2797(01)00219-8
564
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
Selected Abbreviations and Acronyms ECS Enhanced Cancer Surveillance OR odds ratio CI confidence interval BMI body mass index
For the analysis presented here, cases of pancreatic cancer were identified under the ECS protocol described above. Full diagnostic histories available in the Ontario Cancer Registry were reviewed, and cases were excluded if there was an earlier pathology report (i.e., diagnosis prior to 1995) or if on further examination it was determined that the correct diagnosis was not primary cancer of the pancreas. Only women who were determined to be postmenopausal (i.e., reported cessation of menses prior to diagnosis) were included. A sample of postmenopausal female controls was selected from the total pool of control subjects to reflect the age distribution of the postmenopausal cases; a 4.5:1 ratio was the maximum possible given that age distribution. Furthermore, although the ECS in Ontario used surrogate respondents for subjects who were too ill or had died, these surrogate respondents (n 56) were excluded from the analysis here, due to the high proportion of missing data on reproductive variables (especially age at menarche) provided by them. Thus, the analysis is based on 52 postmenopausal cases and 233 postmenopausal controls. The hormonal variables examined included: age at menarche, parity, age at first full-term birth, age at menopause, oral contraceptive use, and estrogen and estrogen/progesterone replacement therapy. Ever use of these medications was defined as at least six months of use; we could not assess years of oral contraceptive or estrogen replacement use because very few women reported short term use of these medications, so that ever and long term users were the same individuals. Furthermore, there were too few women who had used combined estrogen and progesterone preparations (n 28) to analyze these data. Potential confounding variables included body mass index (BMI; weight in kg/height in m2), dietary fat, tofu (a soy-based product containing phytoestrogens), coffee, alcohol (all from the food frequency questions), education, and cigarette smoking status (current, former, never smokers), all of which were associated with risk of disease (p-value 0.15). BMI was based on height and weight two years ago. Dietary fat was based on fat intake from the consumption of 26 food items which account for approximately 80% of total fat intake in women. Grams of fat in each of the food items were estimated based on published values (12). Missing values ( 3% for any single food item) were replaced with the median response for the applicable age and case
AEP Vol. 11. No. 8 November 2001: 563–567
group. Total estimated intake was summed across the 26 food items and categorized into quartiles based on the distribution in the controls. Tofu consumption was categorized as ever/never (there were too few users to further categorize); coffee as 1 cup/day, 1 cup/day, 2–3 cups/day, 4 cups/day, and alcohol (which included total consumption of beer, wine, and liquor) as 1 drink/month, from 1/month to 1/day, 1 drink/day. Risks of pancreatic cancer associated with reproductive variables were estimated by odds ratios (ORs) and 95% confidence intervals (CIs), using multivariable logistic regression models fitted by the method of maximum likelihood (13). All ORs were adjusted for the potentially confounding factors identified above. A variable was considered to be confounding, and kept in the final model, if its addition to the model altered the risk estimate for any level of the factor of interest by 10% or more. A separate analysis was conducted among postmenopausal parous women, in order to examine the effect of age at first birth. Education did not act as a confounding variable in either analysis; coffee and alcohol consumption were confounders in the analysis of age at first birth among parous women.
RESULTS Table 1 displays the frequency distribution of cases and controls, and unadjusted odds ratios, for the reproductive and other variables. Neither age at menarche nor age at menopause were associated with increased or decreased risk of pancreatic cancer. Parity was associated with reduced risk (OR 0.38, 95% CI 0.16–0.87 for 3 births) and age at first birth was associated with increased risk; this latter effect did not change monotonically with increasing age. Oral contraceptive use was associated with decreased risk (OR 0.55, 95% CI 0.25–1.19), whereas estrogen replacement therapy was associated with increased risk (OR 1.70, 95% CI 0.89–3.25), although these risk estimates are not statistically significant. Risk estimates for the potentially confounding variables, including smoking, Body Mass Index, and aspects of dietary intake, did not vary substantially nor consistently from 1.0. Adjusted OR estimates and 95% CIs for the reproductive variables are shown in Table 2. Both parity and use of oral contraceptives were associated with decreased risk of pancreatic cancer. Having had three or more children yields an OR of 0.22 (95% CI 0.07–0.65) compared with nulliparity, and ever use of oral contraceptives an OR of 0.36 (95% CI 0.13–0.96) compared with never use. Among parous women, age at first birth showed the expected increased risk with later age (OR 4.05, 95% CI 1.50–10.92 for age 25–29 years; OR 3.78, 95% CI 1.02– 14.06 for age 30 years).
AEP Vol. 11. No. 8 November 2001: 563–567
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
565
TABLE 1. Distributiona, by case/control status, unadjusted odds ratios (OR), 95% confidence intervals (95% CI), for variables of interest and potential confounders for all post-menopausal females Cases Variable Age at menarche 9–11 yrs 12–13 yrs 14 yrs Age at menopause 45 46–54 55 Parity Nulliparous 1–2 3 Age at first full term birth Less than 25 25–29 30 Nulliparous Oral contraceptive use 6 months 6 months Estrogen replacement therapy 6 months 6 months Age group (years) 25–59 60–69 70–74 Smoking status Non-smoker Former smoker Current smoker BMI 20–24 20 25–29 30 Coffee 1 cup/day 1 cup 2–3 cups 4 cups Alcohol 1 month 1 week 1 day Tofu Never Ever Dietary fat 24 g 25–36 g 37–49 g 50 g Education High school High school a
Totals may vary due to missing observations.
Controls
N
%
9 26 11
19.6 56.5 23.9
26 23 3
N
%
OR
95% CI
40 98 80
18.3 45.0 36.7
1.00 1.18 0.61
— 0.51–2.74 0.23–1.60
50.0 44.2 5.8
94 107 27
41.2 46.9 11.8
1.00 0.78 0.40
— 0.42–1.45 0.11–1.43
11 20 21
21.1 38.5 40.4
27 68 137
11.6 29.3 59.1
1.00 0.72 0.38
— 0.31–1.71 0.16–0.87
17 18 6 11
32.7 34.6 11.5 21.2
139 47 15 27
61.0 20.6 6.6 11.8
1.00 3.13 3.27 3.33
— 1.49–6.57 1.12–9.56 1.40–7.90
41 9
82.0 18.0
160 64
71.4 28.6
1.00 0.55
— 0.25–1.19
34 18
65.4 34.6
177 55
76.3 23.7
1.00 1.70
— 0.89–3.25
16 19 17
30.8 36.5 32.7
72 85 76
30.9 36.5 32.6
1.00 1.01 1.01
— 0.48–2.10 0.47–2.14
27 15 9
52.9 29.4 17.7
133 65 35
57.1 27.9 15.0
1.00 1.14 1.27
— 0.57–2.28 0.55–2.94
21 6 15 10
40.4 11.5 28.9 19.2
104 15 79 34
44.8 6.5 34.0 14.7
1.00 1.98 0.94 1.46
— 0.69–5.70 0.46–1.94 0.62–3.40
14 11 19 8
26.9 21.2 36.5 15.4
60 48 95 30
25.7 20.6 40.8 12.9
1.00 0.98 0.86 1.14
— 0.41–2.36 0.40–1.84 0.43–3.02
23 18 10
45.1 35.3 19.6
107 86 34
47.1 37.9 15.0
1.00 0.97 1.37
— 0.49–1.92 0.59–3.16
46 6
88.5 11.5
203 19
91.4 8.6
1.00 1.39
— 0.53–3.68
7 13 14 18
13.5 25.0 26.9 34.6
58 59 58 58
24.9 25.3 24.9 24.9
1.00 1.83 2.00 2.57
— 0.68–4.90 0.75–5.32 1.00–6.62
32 20
61.5 38.5
150 83
64.4 35.6
1.00 1.13
— 0.61–2.10
566
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
TABLE 2. Odds ratios (OR)a, 95% confidence intervals (CI), likelihood ratio statistic (LR) and LR p-value for all postmenopausal females for parity, age at menopause, oral contraceptive use, estrogen replacement therapy, and age at menarche Variable Age at menarche 9–11 yrs 12–13 yrs 14 yrs Age at menopause 45 46–54 55 Parity Nulliparous 1–2 3 Oral contraceptive use 6 months 6 months Estrogen replacement therapy 6 months 6 months Age at first full term pregnancyb Less than 25 25–29 30
OR
95% CI
LR
p-value
1.00 1.92 0.85
— 0.71–5.17 0.29–2.52
3.965
0.138
1.00 0.87 0.49
— 0.40–1.87 0.09–2.60
0.807
0.668
1.00 0.69 0.22
— 0.24–2.00 0.07–0.65
11.409
0.003
1.00 0.36
— 0.13–0.96
4.662
0.031
1.00 1.66
— 0.72–3.82
1.387
0.239
1.00 4.05 3.78
— 1.50–10.92 1.02–14.06
9.538
0.008
a
Adjusted for the variables in the table as well as for age, smoking status, BMI, tofu, and dietary fat. b Includes parous women only.
DISCUSSION Our data provide some support for the hypothesis that pancreatic cancer may be, in part, an estrogen-dependent disease. Previous epidemiological studies have suggested a role for female sex hormones and reproductive factors in the development of pancreatic cancer. The inverse relationship between parity and pancreatic cancer, observed in this study, has been reported in previous case-control studies (8, 14, 15) but is not in agreement with other studies which reported a reverse effect with the number of pregnancies (9, 10, 11, 16). The association of oral contraceptive use is also reflected in previous studies (8, 11). Despite earlier studies linking excess risks to early age at menarche and late age at menopause (11, 15) we found no association with age at menarche or menopause. Support for the hypothesis of the role of estrogens in the development of pancreatic cancer comes from other data as well. First, the male: female incidence ratio for pancreatic cancer is high in the younger ages: 2:1 or greater before age 50 (1, 17, 18). Among older age groups, however, this ratio approaches 1:1. The possible explanations for this changing male to female ratio include a promotional effect by male sex hormones in early life (19) and an inhibitory effect of
AEP Vol. 11. No. 8 November 2001: 563–567
estrogen (18, 20). Second, estrogen and progesterone receptors have been identified in the pancreas. Not all studies examining pancreatic tissue for steroid receptors have found them (21), but others have (22–27). Estrogen was first discovered in the exocrine pancreas in the 1960’s (28), and serum estrogen levels are reported to differ between pancreatic cancer cases and controls (26). Furthermore, experimental and human studies have shown anti-estrogens such as tamoxifen to have an inhibitory effect on the early stages of pancreatic carcinogenesis (19, 20). Thus, the literature provides a rationale for the risk of pancreatic cancer among women being associated in some part with female hormone levels. In fact, the pattern of increased and decreased risk seen for pancreatic cancer in women is somewhat similar to that which has been known for many years for breast cancer in women (29). We recognize a limitation of this study in the small sample size, yielding somewhat inconsistent point estimates and wider confidence intervals for some variables (e.g., age at menarche). By limiting our analyses to index respondents only, it is also possible that “survivor bias” was introduced, whereby the results obtained reflect the probability of surviving a certain length of time after diagnosis and not the risk associated with the disease itself. For our subjects, however, the time between diagnosis and contact was shorter for index respondents than for proxies, reflecting a delay in obtaining physician consent. When this delay is taken into account, survival time for index respondents was not significantly longer than for those cases for whom a proxy responded. Our risk estimates for smoking, the only well-established risk factor for pancreatic cancer, were not inconsistent with those found in previous studies [see, for example, Harnack and coworkers (2)], which further supports the validity of the results. Although the point estimate for current smokers is somewhat lower than values generally reported, it is in the same direction as, and the confidence interval overlaps with, these values. The study was presented as one of environmental risk factors and cancer generally, so the likelihood of the reproductive variables being affected by recall bias is small. Furthermore the long term recall of reproductive histories has been reported to be reasonably reliable (30, 31). Still, relatively few epidemiological studies have investigated the relationship between hormonal factors and pancreatic cancer, and those that have did not adjust for the potential confounding of other factors. In particular, our data exhibit confounding by smoking (the only well-established risk factor for pancreatic cancer) as well as certain aspects of nutrition (e.g., BMI, intake of dietary fat, and intake of tofu, a soy product). The literature indicates adjustment for smoking in some (8, 10, 11, 14, 15), but not all (9, 16) studies of parity and age at first birth. While some authors report adjusting for other reproductive variables
AEP Vol. 11. No. 8 November 2001: 563–567
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
567
[see, for example, Ji and coworkers (10)], dietary factors which may be related to serum estrogen levels have not been examined as potential confounders. It is clear that if dietary factors may confound the relationship between hormonal variables and pancreatic cancer, then the opposite is also true, i.e., studies of dietary intake and pancreatic cancer which have not considered these reproductive factors are quite likely subject to bias due to the lack of control for confounding. It is possible that reanalysis of these studies, accounting for the effect of female reproductive variables, would modify our understanding of the role of nutritional factors as well.
11. Bueno De Mesquita HB, Maisonneuve P, Moerman CJ, Walker AM. Anthropometric and reproductive variables and exocrine carcinoma of the pancreas: A population based case-control study in the Netherlands. Int J Cancer. 1992;52:24–29.
This research was performed within the context of the Enhanced Cancer Surveillance project, sponsored in part by the Laboratory Centre for Disease Control, Health Canada (contract no. H4078–3-C119/01-SS). The authors would like to thank Gayle Morrison for the reviewing of the pathology reports, Michelle Cotterchio for her input on estimation of fat consumption, and Lori-Ann Larmand for her able assistance in the preparation of the manuscript.
16. Kvale G, Heuch I, Nilssen S. Parity in relation to mortality and cancer incidence: A prospective study of Norwegian women. Int J Epidemiol. 1994;23:691–699.
REFERENCES 1. World Health Organization, International Agency for Research on Cancer and International Association of Cancer Registries. Cancer Incidence in Five Continents. v. VI. Scientific Publications No. 120. Lyon: IARC Scientific Publications; 1992. 2. Harnack LJ, Anderson KE, Zheng W, Folsom AR, Sellers TA, Kushi LH. Smoking, alcohol, coffee, and tea intake and incidence of cancer of the exocrine pancreas: The Iowa Women’s Health Study. Cancer Epidemiol Biomarkers Prev. 1997;6:1081–1086.
12. National Health and Welfare Canada. Nutrient value of some common foods. Catalogue H58–28/1988E. Ottawa, ON: Minister of Supply and Services Canada; 1988. 13. EGRET. Statistics and Epidemiology Research Corporation and Cytel Software Corporation. Cambridge, MA: Cytel Software Corporation; 1999:1985–1993. 14. La Vecchia C, Negri E, Franceschi S, Parazzini F. Long-term impact of reproductive factors on cancer risk. Int J Cancer. 1993;53:215–219. 15. Kalapothaki V, Tzonou A, Hsieh CC, Toupadaki N, Karakatsani A, Trichopoules D. Tobacco, ethanol, coffee, pancreatitis, diabetes mellitus, and cholelithiasis as risk factors for pancreatic carcinoma. Cancer Causes Control. 1993;4:375–382.
17. Levi F, La Vecchia C, Lucchini F, Negri E. Trends in cancer mortality sex ratios in Europe, 1950–1989. World Hlth Stat Q. 1992;45:117– 164. 18. Bourhis J, Lacaine F, Augusti M, Huguier M. Protective effect of oestrogen in pancreatic cancer. Lancet. 1987;2:977. 19. Sumi C, Longnecker DS, Roebuck BD, Brinck-Johnsen T. Inhibitory effects of estrogen and castration on the early stage of pancreatic carcinogenesis in Fischer rats treated with azaserine. Cancer Res. 1989;49:2332–2336. 20. Benz C, Hollander C, Miller B. Endocrine-responsive pancreatic carcinoma: Steroid binding and cytotoxicity studies in human tumor cell lines. Cancer Res. 1986;46:2276–2281. 21. Taylor OM, Teasdale J, Cowen PN, McMahon MJ, Benson EA. Classical oestrogen receptor is not detectable in pancreatic adenocarcinoma. Br J Cancer. 1992;66:503–506.
3. Falk RT, Pickle LW, Fontham ET, Correa P, Fraumeni JF Jr. Life-style risk factors for pancreatic cancer in Louisiana: A case-control study. Am J Epidemiol. 1988;128:324–336.
22. Targarona EM, Pons MD, Gonzalez G, Boix L, Marco V, Marco C. Is exocrine pancreatic cancer a hormone-dependent tumor? A study of the existence of sex hormone receptors in normal and neoplastic pancreas. Hepatogastroenterology. 1991;38:165–169.
4. Ghadirian P, Simard A, Baillargeo J. Tobacco, alcohol, and coffee and cancer of the pancreas. A population-based case-control study in Quebec, Canada. Cancer. 1991;67:2664–2670.
23. Singh S, Baker PR, Poulsom R, Wright NA, Sheppard MC, Langman MJ, et al. Expression of oestrogen receptor and oestrogen-inducible genes in pancreatic cancer. Br J Surg. 1997;84:1085–1089.
5. Howe GR, Jain M, Burch JD, Miller AB. Cigarette smoking and cancer of the pancreas: Evidence from a population-based case-control study in Toronto, Canada. Int J Cancer. 1991;47:323–328.
24. Ollayos CW, Riordan GP, Rushin JM. Estrogen receptor detection in paraffin sections of adenocarcinoma of the colon, pancreas, and lung. Arch Pathol Lab Med. 1994;118:630–632.
6. Mack TM, Yu MC, Hanisch R, Henderson BE. Pancreas cancer and smoking, beverage consumption and past medical history. J Natl Cancer Inst. 1986;76:49–60.
25. Longnecker DS. Hormones and pancreatic cancer. Int J Pancreatol. 1991;9:81–86.
7. Zatonski WA, Boyle P, Przewozniak K, Maisonneuve P, Drosik K, Walker AM. Cigarette smoking, alcohol, tea and coffee consumption and pancreas cancer risk: A case-control study from Opole, Poland. Int J Cancer. 1993;53:601–607. 8. Fernandez E, La Vecchia CL, Avanzo BD, Negri E. Menstrual and reproductive factors and pancreatic cancer risk in women. Int J Cancer. 1995;62:11–14.
26. Greenway B, Iqbal MJ, Johnson D, Williams R. Oestrogen receptor proteins in malignant and fetal pancreas. Brit Med J. 1981;283:751–753. 27. Andrén-Sandberg A. Estrogens and pancreatic cancer: Some recent aspects. Scand J Gastroenterol. 1986;21:129–133. 28. Ullberg S, Bengtsson G. Autoradiographic studies with natural oestrogens. Acta Endocrinol. 1963;43:75–86. 29. Kelsey JL, Gammon MD, John EM. Reproductive Factors and Breast Cancer. Epidemiol Rev. 1993;15:36–47.
9. Karlson BM, Wuu J, Hsieh CC, Lambe M, Ekbom A. Parity and the risk of pancreatic cancer: A nested case-control study. Int J Cancer. 1998;77:224–227.
30. Den Tonkelaar I. Validity and reproducibility of self-reported age at menopause in women participating in the DOM-project. Maturitas. 1997;27:117–123.
10. Ji BT, Hatch MC, Chow WH, McLaughlin JK, Dai Q, Howe GR, et al. Anthropometric and reproductive factors and the risk of pancreatic cancer: A case-control study in Shanghai, China. Int J Cancer. 1996;66:432–437.
31. Hunter DJ, Manson JE, Colditz GA, Chasan-Taber L, Troy L, Stampfer MJ, et al. Reproducibility of oral contraceptive histories and validity of hormone composition reported in a cohort of US women. Contraception. 1997;56:373–378.
PURPOSE: The aim of the study was to evaluate hormonal risk factors for carcinoma of the exocrine pancreas among postmenopausal women. METHODS: Mailed questionnaire data from 52 cases and 233 population-based controls in Ontario were used to assess parity, age at first birth, and other hormonal factors on pancreatic cancer risk. RESULTS: Reduced risk was seen with three or more pregnancies [adjusted odds ratio (OR) 0.22, 95% confidence interval (CI) 0.07–0.65] and use of oral contraceptives (adjusted OR 0.36, 95% CI 0.13–0.96), whereas no significant associations were found for age at menarche or menopause, or estrogen replacement therapy. Among parous women, later age at first full-term pregnancy significantly increased the risk of this cancer (adjusted OR 4.05, 95% CI 1.50–10.92 for ages 25–29 years, adjusted OR 3.78, 95% CI 1.02–14.06 for ages 30 years). CONCLUSIONS: Our data support the hypothesis that pancreatic cancer is, at least in part, an estrogen-dependent disease; there is growing epidemiological evidence that aspects of reproductive history and hormonal exposure are associated with risk of this disease. Ann Epidemiol 2001;11:563-567. © 2001 Elsevier Science Inc. All rights reserved. KEY WORDS:
Pancreatic Cancer, Hormonal Factors, Case-Control Study.
INTRODUCTION
METHODS
Although cancer of the pancreas is associated with a substantial disease burden, both in morbidity and mortality (1), very little is known about its etiology. Aside from cigarette smoking (2–7), there are no well-substantiated risk factors, and little evidence that points to preventive measures. In the last few years, focus has shifted to an examination of the role that female reproductive factors might play in the onset of pancreatic cancer. While not completely consistent, the literature suggests that age at menarche, aspects of pregnancy history, and exogenous hormone exposure all may be associated with altered risk, particularly in postmenopausal women (8–11). Within the context of a large, population-based casecontrol study, we analyzed the reproductive data for postmenopausal women with a new diagnosis of primary carcinoma of the exocrine pancreas, in order to provide further clarification of the relationship between reproductive factors and this disease.
Data were derived from the case-control component of the Enhanced Cancer Surveillance project (ECS), a multi-centre collaborative study of cancer and the environment, conducted in eight of the ten Canadian provinces. Analyses presented here are restricted to the Ontario data as most of the reproductive information was not included in the questionnaires used by the other provinces. Eligible Ontario cases in Enhanced Cancer Surveillance were diagnosed between January 1995 and September 1996. Individuals were aged 20 to 74 years old, were residents of Ontario, and were diagnosed with one of ten cancers. Controls were a sample of the 1995 Ontario Property Assessment file, which contains name, address, sex, and year and month of birth for all residents of the province. It is approximately 95% complete, with no systematic exclusions. In each sex and five-year age stratum, the number of controls was determined by the largest number of cases, among all ten cancer sites, expected in that stratum. This ensured a minimum of 1:1 frequency matching in each stratum, regardless of cancer site being analyzed. The response rate for all female pancreatic cancer cases (including surrogate respondents) was 71.4%, and the response rate for female controls was 80%. A mailed questionnaire focused on a wide array of potential risk factors for the cancers covered by the ECS, including reproductive history for women, cigarette smoking, physical activity, diet history using a food frequency list, occupational and residential histories, and sociodemographic information.
From the Division of Preventive Oncology, Research Unit, Cancer Care Ontario (NK, MS), Toronto, Ontario, Canada; Department of Public Health Sciences, University of Toronto (NK), Toronto, Ontario, Canada; and Trauma Registries, Clinical Data Services, Canadian Institute for Health Information (JL), Toronto, Ontario, Canada. Address reprint requests to: Nancy Kreiger, Division of Preventive Oncology, Cancer Care Ontario, 620 University Avenue, Toronto, Ontario M5G 2L7 Canada. Received April 21, 2000; revised December 19, 2000; accepted December 28, 2000. © 2001 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
1047-2797/01/$–see front matter PII S1047-2797(01)00219-8
564
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
Selected Abbreviations and Acronyms ECS Enhanced Cancer Surveillance OR odds ratio CI confidence interval BMI body mass index
For the analysis presented here, cases of pancreatic cancer were identified under the ECS protocol described above. Full diagnostic histories available in the Ontario Cancer Registry were reviewed, and cases were excluded if there was an earlier pathology report (i.e., diagnosis prior to 1995) or if on further examination it was determined that the correct diagnosis was not primary cancer of the pancreas. Only women who were determined to be postmenopausal (i.e., reported cessation of menses prior to diagnosis) were included. A sample of postmenopausal female controls was selected from the total pool of control subjects to reflect the age distribution of the postmenopausal cases; a 4.5:1 ratio was the maximum possible given that age distribution. Furthermore, although the ECS in Ontario used surrogate respondents for subjects who were too ill or had died, these surrogate respondents (n 56) were excluded from the analysis here, due to the high proportion of missing data on reproductive variables (especially age at menarche) provided by them. Thus, the analysis is based on 52 postmenopausal cases and 233 postmenopausal controls. The hormonal variables examined included: age at menarche, parity, age at first full-term birth, age at menopause, oral contraceptive use, and estrogen and estrogen/progesterone replacement therapy. Ever use of these medications was defined as at least six months of use; we could not assess years of oral contraceptive or estrogen replacement use because very few women reported short term use of these medications, so that ever and long term users were the same individuals. Furthermore, there were too few women who had used combined estrogen and progesterone preparations (n 28) to analyze these data. Potential confounding variables included body mass index (BMI; weight in kg/height in m2), dietary fat, tofu (a soy-based product containing phytoestrogens), coffee, alcohol (all from the food frequency questions), education, and cigarette smoking status (current, former, never smokers), all of which were associated with risk of disease (p-value 0.15). BMI was based on height and weight two years ago. Dietary fat was based on fat intake from the consumption of 26 food items which account for approximately 80% of total fat intake in women. Grams of fat in each of the food items were estimated based on published values (12). Missing values ( 3% for any single food item) were replaced with the median response for the applicable age and case
AEP Vol. 11. No. 8 November 2001: 563–567
group. Total estimated intake was summed across the 26 food items and categorized into quartiles based on the distribution in the controls. Tofu consumption was categorized as ever/never (there were too few users to further categorize); coffee as 1 cup/day, 1 cup/day, 2–3 cups/day, 4 cups/day, and alcohol (which included total consumption of beer, wine, and liquor) as 1 drink/month, from 1/month to 1/day, 1 drink/day. Risks of pancreatic cancer associated with reproductive variables were estimated by odds ratios (ORs) and 95% confidence intervals (CIs), using multivariable logistic regression models fitted by the method of maximum likelihood (13). All ORs were adjusted for the potentially confounding factors identified above. A variable was considered to be confounding, and kept in the final model, if its addition to the model altered the risk estimate for any level of the factor of interest by 10% or more. A separate analysis was conducted among postmenopausal parous women, in order to examine the effect of age at first birth. Education did not act as a confounding variable in either analysis; coffee and alcohol consumption were confounders in the analysis of age at first birth among parous women.
RESULTS Table 1 displays the frequency distribution of cases and controls, and unadjusted odds ratios, for the reproductive and other variables. Neither age at menarche nor age at menopause were associated with increased or decreased risk of pancreatic cancer. Parity was associated with reduced risk (OR 0.38, 95% CI 0.16–0.87 for 3 births) and age at first birth was associated with increased risk; this latter effect did not change monotonically with increasing age. Oral contraceptive use was associated with decreased risk (OR 0.55, 95% CI 0.25–1.19), whereas estrogen replacement therapy was associated with increased risk (OR 1.70, 95% CI 0.89–3.25), although these risk estimates are not statistically significant. Risk estimates for the potentially confounding variables, including smoking, Body Mass Index, and aspects of dietary intake, did not vary substantially nor consistently from 1.0. Adjusted OR estimates and 95% CIs for the reproductive variables are shown in Table 2. Both parity and use of oral contraceptives were associated with decreased risk of pancreatic cancer. Having had three or more children yields an OR of 0.22 (95% CI 0.07–0.65) compared with nulliparity, and ever use of oral contraceptives an OR of 0.36 (95% CI 0.13–0.96) compared with never use. Among parous women, age at first birth showed the expected increased risk with later age (OR 4.05, 95% CI 1.50–10.92 for age 25–29 years; OR 3.78, 95% CI 1.02– 14.06 for age 30 years).
AEP Vol. 11. No. 8 November 2001: 563–567
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
565
TABLE 1. Distributiona, by case/control status, unadjusted odds ratios (OR), 95% confidence intervals (95% CI), for variables of interest and potential confounders for all post-menopausal females Cases Variable Age at menarche 9–11 yrs 12–13 yrs 14 yrs Age at menopause 45 46–54 55 Parity Nulliparous 1–2 3 Age at first full term birth Less than 25 25–29 30 Nulliparous Oral contraceptive use 6 months 6 months Estrogen replacement therapy 6 months 6 months Age group (years) 25–59 60–69 70–74 Smoking status Non-smoker Former smoker Current smoker BMI 20–24 20 25–29 30 Coffee 1 cup/day 1 cup 2–3 cups 4 cups Alcohol 1 month 1 week 1 day Tofu Never Ever Dietary fat 24 g 25–36 g 37–49 g 50 g Education High school High school a
Totals may vary due to missing observations.
Controls
N
%
9 26 11
19.6 56.5 23.9
26 23 3
N
%
OR
95% CI
40 98 80
18.3 45.0 36.7
1.00 1.18 0.61
— 0.51–2.74 0.23–1.60
50.0 44.2 5.8
94 107 27
41.2 46.9 11.8
1.00 0.78 0.40
— 0.42–1.45 0.11–1.43
11 20 21
21.1 38.5 40.4
27 68 137
11.6 29.3 59.1
1.00 0.72 0.38
— 0.31–1.71 0.16–0.87
17 18 6 11
32.7 34.6 11.5 21.2
139 47 15 27
61.0 20.6 6.6 11.8
1.00 3.13 3.27 3.33
— 1.49–6.57 1.12–9.56 1.40–7.90
41 9
82.0 18.0
160 64
71.4 28.6
1.00 0.55
— 0.25–1.19
34 18
65.4 34.6
177 55
76.3 23.7
1.00 1.70
— 0.89–3.25
16 19 17
30.8 36.5 32.7
72 85 76
30.9 36.5 32.6
1.00 1.01 1.01
— 0.48–2.10 0.47–2.14
27 15 9
52.9 29.4 17.7
133 65 35
57.1 27.9 15.0
1.00 1.14 1.27
— 0.57–2.28 0.55–2.94
21 6 15 10
40.4 11.5 28.9 19.2
104 15 79 34
44.8 6.5 34.0 14.7
1.00 1.98 0.94 1.46
— 0.69–5.70 0.46–1.94 0.62–3.40
14 11 19 8
26.9 21.2 36.5 15.4
60 48 95 30
25.7 20.6 40.8 12.9
1.00 0.98 0.86 1.14
— 0.41–2.36 0.40–1.84 0.43–3.02
23 18 10
45.1 35.3 19.6
107 86 34
47.1 37.9 15.0
1.00 0.97 1.37
— 0.49–1.92 0.59–3.16
46 6
88.5 11.5
203 19
91.4 8.6
1.00 1.39
— 0.53–3.68
7 13 14 18
13.5 25.0 26.9 34.6
58 59 58 58
24.9 25.3 24.9 24.9
1.00 1.83 2.00 2.57
— 0.68–4.90 0.75–5.32 1.00–6.62
32 20
61.5 38.5
150 83
64.4 35.6
1.00 1.13
— 0.61–2.10
566
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
TABLE 2. Odds ratios (OR)a, 95% confidence intervals (CI), likelihood ratio statistic (LR) and LR p-value for all postmenopausal females for parity, age at menopause, oral contraceptive use, estrogen replacement therapy, and age at menarche Variable Age at menarche 9–11 yrs 12–13 yrs 14 yrs Age at menopause 45 46–54 55 Parity Nulliparous 1–2 3 Oral contraceptive use 6 months 6 months Estrogen replacement therapy 6 months 6 months Age at first full term pregnancyb Less than 25 25–29 30
OR
95% CI
LR
p-value
1.00 1.92 0.85
— 0.71–5.17 0.29–2.52
3.965
0.138
1.00 0.87 0.49
— 0.40–1.87 0.09–2.60
0.807
0.668
1.00 0.69 0.22
— 0.24–2.00 0.07–0.65
11.409
0.003
1.00 0.36
— 0.13–0.96
4.662
0.031
1.00 1.66
— 0.72–3.82
1.387
0.239
1.00 4.05 3.78
— 1.50–10.92 1.02–14.06
9.538
0.008
a
Adjusted for the variables in the table as well as for age, smoking status, BMI, tofu, and dietary fat. b Includes parous women only.
DISCUSSION Our data provide some support for the hypothesis that pancreatic cancer may be, in part, an estrogen-dependent disease. Previous epidemiological studies have suggested a role for female sex hormones and reproductive factors in the development of pancreatic cancer. The inverse relationship between parity and pancreatic cancer, observed in this study, has been reported in previous case-control studies (8, 14, 15) but is not in agreement with other studies which reported a reverse effect with the number of pregnancies (9, 10, 11, 16). The association of oral contraceptive use is also reflected in previous studies (8, 11). Despite earlier studies linking excess risks to early age at menarche and late age at menopause (11, 15) we found no association with age at menarche or menopause. Support for the hypothesis of the role of estrogens in the development of pancreatic cancer comes from other data as well. First, the male: female incidence ratio for pancreatic cancer is high in the younger ages: 2:1 or greater before age 50 (1, 17, 18). Among older age groups, however, this ratio approaches 1:1. The possible explanations for this changing male to female ratio include a promotional effect by male sex hormones in early life (19) and an inhibitory effect of
AEP Vol. 11. No. 8 November 2001: 563–567
estrogen (18, 20). Second, estrogen and progesterone receptors have been identified in the pancreas. Not all studies examining pancreatic tissue for steroid receptors have found them (21), but others have (22–27). Estrogen was first discovered in the exocrine pancreas in the 1960’s (28), and serum estrogen levels are reported to differ between pancreatic cancer cases and controls (26). Furthermore, experimental and human studies have shown anti-estrogens such as tamoxifen to have an inhibitory effect on the early stages of pancreatic carcinogenesis (19, 20). Thus, the literature provides a rationale for the risk of pancreatic cancer among women being associated in some part with female hormone levels. In fact, the pattern of increased and decreased risk seen for pancreatic cancer in women is somewhat similar to that which has been known for many years for breast cancer in women (29). We recognize a limitation of this study in the small sample size, yielding somewhat inconsistent point estimates and wider confidence intervals for some variables (e.g., age at menarche). By limiting our analyses to index respondents only, it is also possible that “survivor bias” was introduced, whereby the results obtained reflect the probability of surviving a certain length of time after diagnosis and not the risk associated with the disease itself. For our subjects, however, the time between diagnosis and contact was shorter for index respondents than for proxies, reflecting a delay in obtaining physician consent. When this delay is taken into account, survival time for index respondents was not significantly longer than for those cases for whom a proxy responded. Our risk estimates for smoking, the only well-established risk factor for pancreatic cancer, were not inconsistent with those found in previous studies [see, for example, Harnack and coworkers (2)], which further supports the validity of the results. Although the point estimate for current smokers is somewhat lower than values generally reported, it is in the same direction as, and the confidence interval overlaps with, these values. The study was presented as one of environmental risk factors and cancer generally, so the likelihood of the reproductive variables being affected by recall bias is small. Furthermore the long term recall of reproductive histories has been reported to be reasonably reliable (30, 31). Still, relatively few epidemiological studies have investigated the relationship between hormonal factors and pancreatic cancer, and those that have did not adjust for the potential confounding of other factors. In particular, our data exhibit confounding by smoking (the only well-established risk factor for pancreatic cancer) as well as certain aspects of nutrition (e.g., BMI, intake of dietary fat, and intake of tofu, a soy product). The literature indicates adjustment for smoking in some (8, 10, 11, 14, 15), but not all (9, 16) studies of parity and age at first birth. While some authors report adjusting for other reproductive variables
AEP Vol. 11. No. 8 November 2001: 563–567
Kreiger et al. HORMONAL FACTORS AND PANCREATIC CANCER
567
[see, for example, Ji and coworkers (10)], dietary factors which may be related to serum estrogen levels have not been examined as potential confounders. It is clear that if dietary factors may confound the relationship between hormonal variables and pancreatic cancer, then the opposite is also true, i.e., studies of dietary intake and pancreatic cancer which have not considered these reproductive factors are quite likely subject to bias due to the lack of control for confounding. It is possible that reanalysis of these studies, accounting for the effect of female reproductive variables, would modify our understanding of the role of nutritional factors as well.
11. Bueno De Mesquita HB, Maisonneuve P, Moerman CJ, Walker AM. Anthropometric and reproductive variables and exocrine carcinoma of the pancreas: A population based case-control study in the Netherlands. Int J Cancer. 1992;52:24–29.
This research was performed within the context of the Enhanced Cancer Surveillance project, sponsored in part by the Laboratory Centre for Disease Control, Health Canada (contract no. H4078–3-C119/01-SS). The authors would like to thank Gayle Morrison for the reviewing of the pathology reports, Michelle Cotterchio for her input on estimation of fat consumption, and Lori-Ann Larmand for her able assistance in the preparation of the manuscript.
16. Kvale G, Heuch I, Nilssen S. Parity in relation to mortality and cancer incidence: A prospective study of Norwegian women. Int J Epidemiol. 1994;23:691–699.
REFERENCES 1. World Health Organization, International Agency for Research on Cancer and International Association of Cancer Registries. Cancer Incidence in Five Continents. v. VI. Scientific Publications No. 120. Lyon: IARC Scientific Publications; 1992. 2. Harnack LJ, Anderson KE, Zheng W, Folsom AR, Sellers TA, Kushi LH. Smoking, alcohol, coffee, and tea intake and incidence of cancer of the exocrine pancreas: The Iowa Women’s Health Study. Cancer Epidemiol Biomarkers Prev. 1997;6:1081–1086.
12. National Health and Welfare Canada. Nutrient value of some common foods. Catalogue H58–28/1988E. Ottawa, ON: Minister of Supply and Services Canada; 1988. 13. EGRET. Statistics and Epidemiology Research Corporation and Cytel Software Corporation. Cambridge, MA: Cytel Software Corporation; 1999:1985–1993. 14. La Vecchia C, Negri E, Franceschi S, Parazzini F. Long-term impact of reproductive factors on cancer risk. Int J Cancer. 1993;53:215–219. 15. Kalapothaki V, Tzonou A, Hsieh CC, Toupadaki N, Karakatsani A, Trichopoules D. Tobacco, ethanol, coffee, pancreatitis, diabetes mellitus, and cholelithiasis as risk factors for pancreatic carcinoma. Cancer Causes Control. 1993;4:375–382.
17. Levi F, La Vecchia C, Lucchini F, Negri E. Trends in cancer mortality sex ratios in Europe, 1950–1989. World Hlth Stat Q. 1992;45:117– 164. 18. Bourhis J, Lacaine F, Augusti M, Huguier M. Protective effect of oestrogen in pancreatic cancer. Lancet. 1987;2:977. 19. Sumi C, Longnecker DS, Roebuck BD, Brinck-Johnsen T. Inhibitory effects of estrogen and castration on the early stage of pancreatic carcinogenesis in Fischer rats treated with azaserine. Cancer Res. 1989;49:2332–2336. 20. Benz C, Hollander C, Miller B. Endocrine-responsive pancreatic carcinoma: Steroid binding and cytotoxicity studies in human tumor cell lines. Cancer Res. 1986;46:2276–2281. 21. Taylor OM, Teasdale J, Cowen PN, McMahon MJ, Benson EA. Classical oestrogen receptor is not detectable in pancreatic adenocarcinoma. Br J Cancer. 1992;66:503–506.
3. Falk RT, Pickle LW, Fontham ET, Correa P, Fraumeni JF Jr. Life-style risk factors for pancreatic cancer in Louisiana: A case-control study. Am J Epidemiol. 1988;128:324–336.
22. Targarona EM, Pons MD, Gonzalez G, Boix L, Marco V, Marco C. Is exocrine pancreatic cancer a hormone-dependent tumor? A study of the existence of sex hormone receptors in normal and neoplastic pancreas. Hepatogastroenterology. 1991;38:165–169.
4. Ghadirian P, Simard A, Baillargeo J. Tobacco, alcohol, and coffee and cancer of the pancreas. A population-based case-control study in Quebec, Canada. Cancer. 1991;67:2664–2670.
23. Singh S, Baker PR, Poulsom R, Wright NA, Sheppard MC, Langman MJ, et al. Expression of oestrogen receptor and oestrogen-inducible genes in pancreatic cancer. Br J Surg. 1997;84:1085–1089.
5. Howe GR, Jain M, Burch JD, Miller AB. Cigarette smoking and cancer of the pancreas: Evidence from a population-based case-control study in Toronto, Canada. Int J Cancer. 1991;47:323–328.
24. Ollayos CW, Riordan GP, Rushin JM. Estrogen receptor detection in paraffin sections of adenocarcinoma of the colon, pancreas, and lung. Arch Pathol Lab Med. 1994;118:630–632.
6. Mack TM, Yu MC, Hanisch R, Henderson BE. Pancreas cancer and smoking, beverage consumption and past medical history. J Natl Cancer Inst. 1986;76:49–60.
25. Longnecker DS. Hormones and pancreatic cancer. Int J Pancreatol. 1991;9:81–86.
7. Zatonski WA, Boyle P, Przewozniak K, Maisonneuve P, Drosik K, Walker AM. Cigarette smoking, alcohol, tea and coffee consumption and pancreas cancer risk: A case-control study from Opole, Poland. Int J Cancer. 1993;53:601–607. 8. Fernandez E, La Vecchia CL, Avanzo BD, Negri E. Menstrual and reproductive factors and pancreatic cancer risk in women. Int J Cancer. 1995;62:11–14.
26. Greenway B, Iqbal MJ, Johnson D, Williams R. Oestrogen receptor proteins in malignant and fetal pancreas. Brit Med J. 1981;283:751–753. 27. Andrén-Sandberg A. Estrogens and pancreatic cancer: Some recent aspects. Scand J Gastroenterol. 1986;21:129–133. 28. Ullberg S, Bengtsson G. Autoradiographic studies with natural oestrogens. Acta Endocrinol. 1963;43:75–86. 29. Kelsey JL, Gammon MD, John EM. Reproductive Factors and Breast Cancer. Epidemiol Rev. 1993;15:36–47.
9. Karlson BM, Wuu J, Hsieh CC, Lambe M, Ekbom A. Parity and the risk of pancreatic cancer: A nested case-control study. Int J Cancer. 1998;77:224–227.
30. Den Tonkelaar I. Validity and reproducibility of self-reported age at menopause in women participating in the DOM-project. Maturitas. 1997;27:117–123.
10. Ji BT, Hatch MC, Chow WH, McLaughlin JK, Dai Q, Howe GR, et al. Anthropometric and reproductive factors and the risk of pancreatic cancer: A case-control study in Shanghai, China. Int J Cancer. 1996;66:432–437.
31. Hunter DJ, Manson JE, Colditz GA, Chasan-Taber L, Troy L, Stampfer MJ, et al. Reproducibility of oral contraceptive histories and validity of hormone composition reported in a cohort of US women. Contraception. 1997;56:373–378.