American Journal of Epidemiology Advance Access originally published online on December 10, 2007
American Journal of Epidemiology 2008 167(5):607-614; doi:10.1093/aje/kwm333
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ORIGINAL CONTRIBUTIONS |
Diabetes and Endometrial Cancer: An Evaluation of the Modifying Effects of Other Known Risk Factors
1 Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA
2 Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA
3 Department of Internal Medicine, School of Medicine, University of New Mexico, Albuquerque, NM
4 Department of Otolaryngology: Head and Neck Surgery, School of Medicine, University of Washington, Seattle, WA
Correspondence to Babette S. Saltzman, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, P.O. Box 19024 (M4-C308), Seattle, WA 98109-1024 (e-mail: bsiebold{at}fhcrc.org).
Received for publication December 4, 2006. Accepted for publication October 17, 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
To determine whether risk of endometrial cancer among women with type 2 diabetes differs with respect to other endometrial cancer risk factors, the authors used data from a population-based case-control study (1,303 cases and 1,779 controls) conducted in western Washington State during 1985–1999. History of type 2 diabetes was associated with endometrial cancer (odds ratio (OR) = 1.7, 95% confidence interval (CI): 1.2, 2.3), more strongly among women with a recent diabetes diagnosis (<5 years) (OR = 2.6, CI: 1.5, 4.7) than among those with a more distant diagnosis (
5 years) (OR = 1.3, CI: 0.8, 1.9). Type 2 diabetes was associated with endometrial cancer among women with a body mass index (BMI) (weight (kg)/height (m)2) less than 35 but not among women with a BMI of 35 or more. The observed associations persisted after finer adjustment for BMI to control for residual confounding. History of diabetes was associated with a twofold increased risk of endometrial cancer among hypertensive women, but no association was observed among nonhypertensive women. The risk associated with type 2 diabetes appeared not to vary greatly with respect to other endometrial cancer risk factors. These results support the hypothesis that type 2 diabetes is associated with endometrial cancer irrespective of the presence of other risk factors for this disease, except possibly hypertension and extreme obesity.
diabetes mellitus; diabetes mellitus, Type 2; endometrial neoplasms; risk factors
Abbreviations: BMI, body mass index; CARE, Contraceptive and Reproductive Experiences; CI, confidence interval; OR, odds ratio
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
A positive association has been observed in nearly all studies of type 2 diabetes in relation to the incidence of endometrial cancer. Given the adverse effect of obesity on the incidence of both diabetes and endometrial cancer, investigators have adjusted for obesity in a number of these studies. To varying degrees, all found that diabetes was independently associated with endometrial cancer (1–12). In the current analysis, we sought to explore the possible joint influence of type 2 diabetes and other endometrial cancer risk factors, including obesity, to better understand the means through which the elevation in endometrial cancer risk associated with type 2 diabetes occurs.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Cases and controls for this analysis were drawn from three population-based case-control studies of endometrial cancer which have been described in detail previously (13–15). The studies had similar study protocols and questionnaires and were conducted in western Washington State. To increase study power, we also included a subset of controls from the Women's Contraceptive and Reproductive Experiences (CARE) breast cancer study. The CARE Study had a similar study protocol and asked similar questions about exposures bearing on endometrial cancer risk as the three endometrial cancer case-control studies (16). All studies were approved by the institutional review board of the Fred Hutchinson Cancer Research Center (Seattle, Washington).
The case group included women aged 45–74 years who resided in King, Pierce, or Snohomish county and were diagnosed with histologically confirmed endometrial adenocarcinoma in 1985–1991, 1994–1995, or 1997–1999. Cases were identified through the Cancer Surveillance System, a population-based cancer registry serving western Washington that is affiliated with the Surveillance, Epidemiology, and End Results program of the National Cancer Institute (17). All cases diagnosed in 1985–1991 and cases aged 50–65 years diagnosed in 1994–1999 were limited to women residing in a household with a telephone at the time of their cancer diagnosis. Cases ages 66–69 years diagnosed in 1997–1999 were limited to those who had records in the Health Care Financing Administration (now the Centers for Medicare and Medicaid Services) data files. Overall, 128 cases died before we could ask them to participate, and another 304 declined to be interviewed or their physician instructed us not to contact them.
Controls were women aged 45–74 years with intact uteri and no prior history of endometrial cancer who were residents of King, Pierce, or Snohomish county during the same time periods as the cases. For reference years 1985–1991, all controls were identified using random digit dialing (18). For reference years 1994–1995 and 1997–1999, random digit dialing was used to identify control women aged 50–65 years and Health Care Financing Administration data were used to identify control women aged 66–69 years. Controls (excluding the CARE controls described below) were frequency-matched to cases by diagnosis year, age, and county. Table 1 presents the distribution of cases and controls for our three studies.
|
For the reference years 1994 and later, ascertainment of cases and controls was restricted to White women and Black women. Cases and controls were assigned reference dates. For cases, this date corresponded to the month and year of their cancer diagnosis. Reference years of controls were chosen so as to approximate the distribution of reference years for the cases, and reference months were assigned randomly.
An additional group of 252 controls was obtained from the CARE multisite study of breast cancer (16). The CARE Study identified control women using random digit dialing and included only White and Black women. Controls from the CARE Study were restricted to female residents of King County aged 50–64 years who had intact uteri and no history of endometrial cancer with reference dates in 1994–1995 and 1997–1998. The overall screening and interview response percentages for CARE controls were 83.6 percent and 88.3 percent, respectively.
In-person interviews were conducted with cases and controls according to a standard protocol, after informed consent had been obtained. Seventy-three (2.4 percent) of the interviews were conducted by telephone. Participants were asked only about events that had occurred prior to their reference date. A calendar of life events was constructed to facilitate recall. The data collected included: demographic information; height; weight at different ages; reproductive and menstrual history; family history of cancer; history of selected chronic diseases, including diabetes; and history of contraceptive and noncontraceptive hormone use. A total of 1,306 cases with endometrial cancer were interviewed. One interview was discarded because of poor-quality interview data, one interview was lost, and one case was excluded because she was later discovered not to have endometrial cancer. This left 1,303 cases in the analysis. A total of 1,779 controls were interviewed, and all were included in the analysis.
History of type 2 diabetes was ascertained by asking each study participant whether she had ever been diagnosed with diabetes by a physician or other health professional. Further questions established the year of diagnosis and whether the participant had ever been hospitalized or medicated for this condition. We excluded from the analysis women whose diagnosis was made before the age of 20 years in order to eliminate those who probably had type 1 diabetes (0 cases and 3 controls). To further minimize the possibility of misclassification, we restricted the definition of type 2 diabetes (henceforth called diabetes) to include only women who had been hospitalized or medicated for their condition. This resulted in 12 case women and seven control women with diabetes being excluded from the analyses. Similarly, women with hypertension were identified by asking each study participant whether she had ever been diagnosed with hypertension by a physician or other health professional. The group of women defined as having hypertension was restricted to include only women who had taken antihypertensive medication. The 41 case women and 45 control women who reported a history of hypertension but did not report a history of antihypertensive medication use were excluded from the analyses assessing diabetes-associated endometrial cancer risk according to the presence or absence of hypertension.
Data were analyzed by logistic regression using STATA (version 8.0 for Windows; Stata Corporation, College Station, Texas). Analyses were adjusted for the frequency-matching factors reference year, age, and county, as well as for body mass index (BMI) (weight (kg)/height (m)2; continuous linear variable) and menopausal hormone use. The latter was categorized into high, medium, and low risk for endometrial cancer as follows: 1) "high risk"—either use of unopposed estrogen for >4–8 years within 2 years of the reference date or use of unopposed estrogen for more than 8 years regardless of recency or use of sequential estrogen and progestin (with fewer than 10 days of progestin per month) for more than 12 years, regardless of recency; 2) "medium risk"—either use of unopposed estrogen for 0.5–4 years regardless of recency or use of unopposed estrogen for >4–8 years, ending more than 2 years prior to the reference date, or use of sequential estrogen and progestin (with fewer than 10 days of progestin per month) for 12 or fewer years; 3) "low risk"—never use of unopposed estrogen and never use of sequential estrogen plus progestin (with fewer than 10 days of progestin per month). Adjustment for other factors bearing on endometrial cancer risk (parity, age at first birth, number of births, and smoking status) did not substantially alter the relative risks (<10 percent), and the results presented here were not adjusted for them. We performed a likelihood ratio test to evaluate the presence of an interaction on a multiplicative scale between diabetes and other endometrial cancer risk factors. We did this by comparing a model containing diabetes and another endometrial cancer risk factor with a common reference group with a model containing the same variables as the previous model as well as an interaction term.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Table 2 shows the characteristics of cases and controls. Women with endometrial cancer were older, less likely to be Black or Asian, had fewer years of education, and were less likely to have ever smoked than were control subjects. Cases were relatively more likely to be overweight (BMI 25–<30), to have had a diagnosis of hypertension, and to be nulliparous. Cases were less likely than controls to have used oral contraceptives but were relatively more likely to have used high-risk hormone replacement therapy.
|
A higher proportion of women with endometrial cancer than controls had diabetes (odds ratio (OR) = 1.7, 95 percent confidence interval (CI): 1.2, 2.3) (table 3). The association was stronger for women who had been diagnosed with diabetes less than 5 years prior to their reference date (OR = 2.6, 95 percent CI: 1.5, 4.7) than in those diagnosed with diabetes 5 or more years prior to their reference date (OR = 1.3, 95 percent CI: 0.8, 1.9). The association between diabetes and endometrial cancer among women diagnosed within the 5 years before their reference date persisted when women who had been diagnosed with diabetes within 2 years of the reference date (done to rule out the possibility that increased medical surveillance following a diabetes diagnosis may have led to detection of endometrial cancer) were excluded from the analysis (OR = 2.9, 95 percent CI: 1.3, 6.5).
|
Diabetes was associated with endometrial cancer among women with a BMI less than 35, with odds ratios ranging from 1.6 to 2.5 (OR = 1.8 (95 percent CI: 0.8, 4.1), OR = 2.5 (95 percent CI: 1.3, 4.6), and OR = 1.6 (95 percent CI: 0.8, 3.1) among lean, overweight, and moderately obese women, respectively) (table 4). However, diabetes appeared not to be related to endometrial cancer among severely obese women (BMI
35; OR = 1.1, 95 percent CI: 0.6, 2.0). These stratum-specific associations persisted when we added BMI to the regression model as a continuous linear variable in order to control for any residual confounding (OR = 1.8 (95 percent CI: 0.8, 4.0), OR = 2.4 (95 percent CI: 1.3, 4.5), OR = 1.5 (95 percent CI: 0.7, 2.9), and OR = 0.9 (95 percent CI: 0.5, 1.8) among lean, overweight, moderately obese, and severely obese women, respectively; p = 0.25 for the dissimilarity of the odds ratios). Among women with no history of hypertension, those with diabetes had virtually no elevation in risk compared with those without diabetes (OR = 1.1, 95 percent CI: 0.6, 1.9), whereas among women with hypertension, the corresponding odds ratio was 2.1 (95 percent CI: 1.3, 3.2; p = 0.12 for the dissimilarity of the two odds ratios). The risk of endometrial cancer associated with diabetes appeared not to vary greatly with respect to menopausal hormone use, cigarette smoking, age, race, education, or parity (table 4).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Table 5 summarizes findings from prior studies that evaluated the association between diabetes and endometrial cancer risk; all attempted to adjust for adiposity. Those studies found that diabetes was related to endometrial cancer to at least some degree, with risk ratios ranging from 1.3 to 3.6 (1–12). These findings, along with those from the current study, suggest that there are other mechanisms besides adiposity through which diabetes is associated with endometrial cancer. The results from our study also suggest that the risk is particularly high among women with a recent diagnosis of diabetes. This observation holds even after the exclusion of women diagnosed with diabetes within 2 years of their reference date. It is consistent with the hypothesis that hyperinsulinemia gives rise to an elevated risk of endometrial cancer, given that 1) women in the prediabetic and early diabetic states have elevated levels of circulating insulin and 2) like other hormones (e.g., estrogens), insulin can influence endometrial cancer risk relatively quickly.
|
Several epidemiologic studies have sought to evaluate the relation between elevated levels of circulating insulin and endometrial cancer risk (19–21). One, based on the analysis of blood samples obtained at least 6.2 months prior to diagnosis (median elapsed time between sample collection and diagnosis, 4.7 years), observed that after adjustment for BMI, women in the highest fifth of C-peptide level (a marker of pancreatic insulin production whose levels mirror those of circulating insulin) had a 4.4-fold elevation (95 percent CI: 1.7, 11.7) in endometrial cancer risk compared with women in the lowest fifth (19). An association between circulating insulin levels and endometrial cancer risk also was present in two studies in which specimens were obtained following diagnosis (20, 21), but the association disappeared after adjustment for BMI and waist:hip ratio. The explanation for these discrepant findings in those studies that collected specimens before and after cancer diagnosis may lie in the fact that postdiagnosis levels of blood insulin among cancer cases may not mirror those present prior to the development of the tumor.
If elevated levels of circulating insulin were a mechanism through which diabetes imparted an excess risk of endometrial cancer, elevations in risk among diabetics in all BMI strata would be expected, since the diagnosis of type 2 diabetes is preceded by a period of hyperinsulinemia regardless of a person's BMI. Several studies have found that diabetes is associated with an elevation in risk (ranging from 1.4 to 2.9) among overweight and obese women (1, 8, 10, 11), and all but one (the study by Anderson et al. (1)) observed it to be associated with an elevation in risk among lean women. However, "lean" was defined broadly in these studies, probably in an attempt to maximize the power to detect an association between diabetes and endometrial cancer, since both are more common among overweight women. Therefore, although several studies did observe that the association between diabetes and endometrial cancer persisted among so-called "lean" women, the possibility remains that residual confounding by BMI within the analytic stratum may account for much of this association. The findings from the current study suggest that diabetes is associated with an elevation in endometrial cancer risk among women with a BMI lower than 35, even after controlling for possible residual confounding, although the association in lean women was weak and the confidence interval around the odds ratio was wide and included 1. Our finding that diabetes was not associated with endometrial cancer among women who were extremely obese may have resulted from chance or may have been due to a substantial prevalence of prediabetes (with associated hyperinsulinemia) or undiagnosed diabetes in these women.
We also found that diabetes was associated, to at least some extent, with an elevation in risk of endometrial cancer among both users and nonusers of unopposed estrogens. This observation suggests that diabetes confers an elevated risk of endometrial cancer by means other than its effect on circulating estrogen levels, since the adverse effect of estrogen on endometrial cancer risk probably plateaus after reaching some threshold of exposure intensity (22).
The strengths of the present study include its population-based design, its large sample size (1,303 cases and 1,779 controls), and the inclusion of only pathologically confirmed endometrial tumors. A limitation was our reliance on self-reports of diabetes diagnosis; however, past studies have found good agreement between self-report and medical record data for diagnoses of diabetes (23). Another limitation was the classification of diabetes as a single entity, without consideration of its severity or control. Similarly, since we had no information on diabetes treatment, we could not assess the possible impact of this on risk of endometrial cancer. We were unable to assess what proportion of women classified as not having diabetes had undiagnosed diabetes and whether this proportion differed between cases and controls or differed between groups of women defined by the presence of absence of other endometrial cancer risk factors. We also were unable to characterize persons with hypertension in terms of the severity of their condition or the treatment they received for it. Finally, although the overall sample was relatively large, risk estimates for some subgroups were based on small numbers and correspondingly were statistically imprecise.
In the current study, we observed that diabetes was particularly strongly associated with endometrial cancer among women who had been previously diagnosed with hypertension. However, in the only other study that examined this question, the magnitude of the association between diabetes and endometrial cancer was similar among women with hypertension and those without it (odds ratios of 3.0 and 3.2, respectively) (8). An analysis addressing this question in other data sets may shed some light on the presence of a possible diabetes-hypertension interaction in the genesis of endometrial cancer.
|
ACKNOWLEDGMENTS |
|---|
This work was supported by the following grants from the National Institutes of Health: R35 CA 39779 (N. S. W.), R01 CA 47749 (S. A. B.), R01 CA 75977 (N. S. W.), N01 HD 2 3166 (Women's Contraceptive and Reproductive Experiences Study), and K05 CA 92002 (N. S. W.).
Conflict of interest: none declared.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
- Anderson KE, Anderson E, Mink PJ, et al. Diabetes and endometrial cancer in the Iowa Women's Health Study. Cancer Epidemiol Biomarkers Prev (2001) 10:611–16.
[Abstract/Free Full Text] - Brinton LA, Berman ML, Mortel R, et al. Reproductive, menstrual, and medical risk factors for endometrial cancer: results from a case-control study. Am J Obstet Gynecol (1992) 167:1317–25.[Web of Science][Medline]
- Elwood JM, Cole P, Rothman KJ, et al. Epidemiology of endometrial cancer. J Natl Cancer Inst (1977) 59:1055–60.[Web of Science][Medline]
- Goodman MT, Hankin JH, Wilkens LR, et al. Diet, body size, physical activity, and the risk of endometrial cancer. Cancer Res (1997) 57:5077–85.
[Abstract/Free Full Text] - Kelsey JL, LiVolsi VA, Holford TR, et al. A case-control study of cancer of the endometrium. Am J Epidemiol (1982) 116:333–42.
[Abstract/Free Full Text] - La Vecchia C, Negri E, Franceschi S, et al. A case-control study of diabetes mellitus and cancer risk. Br J Cancer (1994) 70:950–3.[Web of Science][Medline]
- O'Mara BA, Byers T, Schoenfeld E. Diabetes mellitus and cancer risk: a multisite case-control study. J Chronic Dis (1985) 38:435–41.[CrossRef][Web of Science][Medline]
- Parazzini F, La Vecchia C, Negri E, et al. Diabetes and endometrial cancer: an Italian case-control study. Int J Cancer (1999) 81:539–42.[CrossRef][Web of Science][Medline]
- Salazar-Martinez E, Lazcano-Ponce EC, Lira-Lira GG, et al. Case-control study of diabetes, obesity, physical activity and risk of endometrial cancer among Mexican women. Cancer Causes Control (2000) 11:707–11.[CrossRef][Web of Science][Medline]
- Shoff SM, Newcomb PA. Diabetes, body size, and risk of endometrial cancer. Am J Epidemiol (1998) 148:234–40.
[Abstract/Free Full Text] - Weiderpass E, Persson I, Adami HO, et al. Body size in different periods of life, diabetes mellitus, hypertension, and risk of postmenopausal endometrial cancer (Sweden). Cancer Causes Control (2000) 11:185–92.[CrossRef][Web of Science][Medline]
- Xu WH, Matthews CE, Xiang YB, et al. Effect of adiposity and fat distribution on endometrial cancer risk in Shanghai women. Am J Epidemiol (2005) 161:939–47.
[Abstract/Free Full Text] - Beresford SA, Weiss NS, Voigt LF, et al. Risk of endometrial cancer in relation to use of oestrogen combined with cyclic progestagen therapy in postmenopausal women. Lancet (1997) 349:458–61.[CrossRef][Web of Science][Medline]
- Hill DA, Weiss NS, Beresford SA, et al. Continuous combined hormone replacement therapy and risk of endometrial cancer. Am J Obstet Gynecol (2000) 183:1456–61.[CrossRef][Web of Science][Medline]
- Doherty JA, Cushing-Haugen KL, Saltzman BS, et al. Long-term use of postmenopausal estrogen and progestin hormone therapies and the risk of endometrial cancer. Am J Obstet Gynecol (2007) 197(139):e1–7. (doi:10.1016/j.ajog.2007.01.019).
- Marchbanks PA, McDonald JA, Wilson HG, et al. The NICHD Women's Contraceptive and Reproductive Experiences Study: methods and operational results. Ann Epidemiol (2002) 12:213–21.[CrossRef][Web of Science][Medline]
- Hankey BF, Ries LA, Edwards BK. The Surveillance, Epidemiology, and End Results program: a national resource. Cancer Epidemiol Biomarkers Prev (1999) 8:1117–21.
[Free Full Text] - Waksberg J. Sampling methods for random digit dialing. J Am Stat Assoc (1978) 73:40–6.[CrossRef][Web of Science]
- Lukanova A, Zeleniuch-Jacquotte A, Lundin E, et al. Prediagnostic levels of C-peptide, IGF-I, IGFBP -1, -2 and -3 and risk of endometrial cancer. Int J Cancer (2004) 108:262–8.[CrossRef][Web of Science][Medline]
- Troisi R, Potischman N, Hoover RN, et al. Insulin and endometrial cancer. Am J Epidemiol (1997) 146:476–82.
[Abstract/Free Full Text] - Weiderpass E, Brismar K, Bellocco R, et al. Serum levels of insulin-like growth factor-I, IGF-binding protein 1 and 3, and insulin and endometrial cancer risk. Br J Cancer (2003) 89:1697–704.[CrossRef][Web of Science][Medline]
- Grady D, Gebretsadik T, Kerlikowske K, et al. Hormone replacement therapy and endometrial cancer risk: a meta-analysis. Obstet Gynecol (1995) 85:304–13.[CrossRef][Web of Science][Medline]
- Okura Y, Urban LH, Mahoney DW, et al. Agreement between self-report questionnaires and medical record data was substantial for diabetes, hypertension, myocardial infarction and stroke but not for heart failure. J Clin Epidemiol (2004) 57:1096–103.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||