Smoking and the risk of endometrial cancer: Results from the Nurses' Health Study
Article first published online: 11 JAN 2005
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 114, Issue 6, pages 996–1001, 10 May 2005
How to Cite
Viswanathan, A. N., Feskanich, D., De Vivo, I., Hunter, D. J., Barbieri, R. L., Rosner, B., Colditz, G. A. and Hankinson, S. E. (2005), Smoking and the risk of endometrial cancer: Results from the Nurses' Health Study. Int. J. Cancer, 114: 996–1001. doi: 10.1002/ijc.20821
- Issue published online: 11 MAR 2005
- Article first published online: 11 JAN 2005
- Manuscript Accepted: 15 OCT 2004
- Manuscript Received: 7 APR 2004
- Gillette Women's Cancer Center
- American Cancer Society Research Scholarship. Grant Number: RSG-00-061-04CCE
- American Cancer Society Clinical Research Professorship
- National Institutes of Health. Grant Numbers: CA82838, CA87969
- endometrial cancer;
- prospective cohort
An inverse association between smoking and endometrial cancer has generally been observed, primarily among current smokers. To assess this association, we analyzed data from the prospective Nurses' Health Study. From 1976 to 2000, 702 cases of invasive endometrial cancer were identified during 1.8 million person-years of follow-up. Smoking status was assessed in 1976 and updated every 2 years. Cox proportional hazards models were used to calculate multivariate relative risks (RRs), controlling for endometrial cancer risk factors. Compared to never smokers, the multivariate RR of endometrial cancer was significantly lower among both current smokers (RR = 0.63; 95% CI = 0.50–0.79) and past smokers (RR = 0.73; 95% CI = 0.62–0.87). When additionally adjusting for body mass index (BMI), the RR for current smokers was attenuated (RR = 0.72; 95% CI = 0.57–0.90), but the RR for past smokers did not change. Risk was lower among women who smoked 35 or more cigarettes a day (RR = 0.60; 95% CI = 0.39–0.91) and among those who smoked for 40 or more years (RR = 0.63; 95% CI = 0.45–0.87). Tests for trend, which excluded never smokers, were not statistically significant for any of the smoking variables analyzed. These data indicate that both current and past smoking are associated with a lower risk of endometrial cancer. The findings provide insight into disease etiology and suggest that the influence of smoking on endometrial cancer risk occurs even in early adulthood, is long-lasting, and may not be attributed solely to short-term hormonal modulation. © 2004 Wiley-Liss, Inc.
Endometrial cancer is the most common gynecologic cancer in Western societies. In the United States, approximately 40,100 cases of endometrial cancer occur annually,1 and the lifetime risk for this malignancy in women is approximately 2.4%.2 The major etiologic hypothesis for the development of endometrial cancer is exposure to high levels of estrogen in conjunction with inadequate progesterone.3 The use of unopposed estrogen replacement therapy in postmenopausal women has been associated with an increased risk of endometrial cancer.4, 5, 6, 7, 8 Other factors shown to increase risk include obesity and nulliparity.9, 10 Cigarette smoking has been hypothesized to lower the risk of endometrial cancer by reducing levels of circulating estrogens through weight reduction, an earlier age at menopause, or by altering hormone metabolism.11, 12, 13
Three prospective studies have reported nonsignificantly lower risks of endometrial cancer in both past and current smokers compared to never smokers.10, 14, 15 The largest of these studies, with 403 cases diagnosed over 10.6 years of follow-up, found no significant association for number of years smoked, pack-years, or years since quitting smoking for either current or past smokers,15 though risk was significantly lower for women who were current smokers of > 20 cigarettes per day compared to never smokers. At least 5 case-control studies have reported a significantly lower risk of endometrial cancer for current smokers compared to never smokers,16, 17, 18, 19, 20 while only 1 hospital-based case-control study showed a significant inverse risk for past smokers.21 Several other case-control studies have reported nonsignificantly lower risks of endometrial cancer in both past and current smokers.22, 23, 24, 25, 26, 27, 28
To address the association between smoking and risk of endometrial cancer, we analyzed data from the Nurses' Health Study, which has accumulated detailed information on smoking and endometrial cancer risk factors biennially over a 24-year period.
Material and methods
Study population and design
The Nurses' Health Study (NHS) is a prospective cohort study of 121,700 registered nurses who were between the ages of 30 and 55 years and lived in 11 states in the United States when they completed an initial questionnaire on their medical history and lifestyles in 1976. Since then, follow-up questionnaires have been mailed every 2 years to update information on risk factors and major medical events. The follow-up rate through 2000, as a percent of total possible person-years, was 95%. Deaths in the cohort are identified by reports from family members and the U.S. Postal Service, as well as a search of the National Death Index; at least 98% of deaths are ascertained.29 The Human Research Committee of the Brigham and Women's Hospital (Boston, MA) approved the protocol.
For this study, we excluded women who had had a hysterectomy (n = 22,968), did not report their smoking status (n = 262), or reported any type of cancer (excluding nonmelanoma skin cancer; n = 1,584) at baseline in 1976. Women who reported endometrial cancer but for whom we were unable to establish a date of diagnosis were also excluded at baseline. A total of 96,704 women remained for analysis.
At baseline, nurses were asked whether they smoked cigarettes currently and whether they had ever smoked in the past. Current smokers were asked to quantify the number of cigarettes typically smoked per day, and past smokers were asked to report how many cigarettes were typically smoked per day when they last smoked regularly. Baseline information also included the age at which they first started to smoke regularly, and past smokers provided the age at which they stopped smoking. Smoking status and current cigarettes per day were updated on every subsequent biennial questionnaire. From 1982 forward, nurses were no longer asked to write a specific number of cigarettes smoked but rather marked a category of usage (1–4, 5–14, 15–24, 25–34, 35–44, 45 or more cigarettes per day).
To calculate pack-years, smoking duration in years was multiplied by packs of cigarettes smoked per day (20 cigarettes per pack). Pack-years smoked before age 30 was calculated in order to determine whether early exposure had an impact on endometrial cancer risk.
Endometrial cancer cases
Beginning in 1978 and on all subsequent questionnaires, participants were asked to report any diagnosis of endometrial cancer. A total of 2,043 women with an affirmative response were asked for permission to review their medical records. Medical records were obtained for 1,674 (82%) of these reports. Physicians, masked to the participant's exposure status, confirmed the diagnosis, histologic type, presence of invasion and stage in 1,321 medical records. Of these, we identified 702 cases of invasive adenocarcinoma defined by the International Federation of Gynecology and Obstetrics (FIGO) as stage IB to IVB.
Information on most potential confounding factors, including menopausal status, age at menopause, postmenopausal hormone (PMH) use, weight, diabetes and hypertension, was collected on the baseline questionnaire and in 2-year updates. If no data were available for any updated covariate (except BMI) in a given 2-year period, those women were assigned to a missing category for that period. A nurse was classified as postmenopausal from the time she returned a questionnaire reporting natural menopause (women reporting a hysterectomy were excluded from subsequent follow-up). Self-report of menopausal status is valid in this cohort.30 PMH use was first assessed in 1976; women were queried about current and past postmenopausal hormone therapy use and duration. From 1978, information on the type of hormone used was collected. For 1976 to 1978, we assigned women to the type of hormone therapy reported on the 1978 questionnaire.
Body mass index (weight in kg/height in m2) was calculated from height at baseline and from the updated report of current weight. Since weight is an important potential confounder in this analysis, we carried forward the weight reported in the prior questionnaire cycle if it was missing in the current cycle. If weight was not reported for 2 consecutive time periods, these women were excluded from follow-up until an updated weight was reported. In a validation study among 140 NHS members in 1986, self-reported height and weight measures correlated highly with standardized measures by a technician (r = 0.97).31
Several risk factors were collected during only part of the follow-up period. A previously validated semiquantitative food frequency questionnaire was first included in 1980, and dietary intake was updated about every 4 years.32, 33 Information on parity and oral contraceptive use was collected through 1982 when the youngest woman was 36, and few reported current use of oral contraceptives. Data on height and age at menarche were collected at baseline only. Information on menstrual irregularity between ages 20 and 35 (assessed as very regular, usually regular, usually irregular, very irregular) was collected in 1982.
Person-years of follow-up accrued from the date of return of the 1976 baseline questionnaire until either the date of diagnosis of endometrial cancer, the date of death, a report of other cancer, a report of hysterectomy, or the end of follow-up (1 June 2000). Person-time, equal to the number of months between the return of successive questionnaires, was allocated for each variable on the basis of the updated exposure/covariate status at the beginning of each questionnaire cycle. Women did not contribute person-time in follow-up cycles in which they were missing data for smoking status or body weight.
Incidence rates were calculated by dividing the number of events by the number of person-years of follow-up. Relative risks (RRs) were calculated by dividing the incidence rate in a smoking exposure category by the corresponding rate in the reference category (i.e., never smokers) and then adjusted for age using 5-year age strata. To adjust the RRs for multiple covariates, we used Cox proportional hazard models conditioned on age and follow-up cycle. All multivariate models included menopausal status, age at menarche, oral contraceptive use, PMH use, parity, diabetes and hypertension (see footnote to Table II for details). Because BMI is likely an intermediate in the smoking/endometrial cancer association, yet the risk relationship after accounting for BMI is also of interest, we present models both controlling and not controlling for BMI (modeled as a categorical variable). Tests for linear trend were calculated excluding the reference category of never smokers and using continuous values for smoking exposure. Multivariate models were rerun with adjustment for other smoking variables and excluding never smokers.
To assess whether the relationship between smoking and endometrial cancer risk varied across categories of other risk factors, we performed stratified analyses by PMH use (never vs. ever; never, current and past), menopausal status (premenopausal vs. postmenopausal) and BMI (< 30 vs. ≥ 30 kg/m2). We also calculated an interaction term for each dichotomized variable by each smoking variable as ordered categories and evaluated the Wald statistic and the likelihood ratio test.
Diet was first queried in 1980; therefore, statistical models including dietary variables used follow-up from 1980 to 2000 only. Participants who did not respond to the 1980 semiquantitative dietary questionnaire were excluded from this subanalysis. Dietary covariates included animal fat, vegetable fat and alcohol intake in one model, as well as saturated, trans, polyunsaturated and monounsaturated fats in another model, in addition to the previously described nondietary variables.
A total of 702 cases of invasive endometrial adenocarcinoma were identified during almost 1.8 million person-years of follow-up. Characteristics of the study population at the midpoint of follow-up (1988) are shown in Table I. Current smokers were less likely than never or past smokers to be obese. The prevalence of other endometrial cancer risk factors was similar across categories of smoking status. Over the follow-up period, 23% of participants were currently smoking, 33% had previously smoked and 44% had never smoked.
|Never smokers||Past smokers||Current smokers|
|Age first smoked (years)1||19||19|
|Smoking duration (years)1||18||34|
|Smoked > 25 cigarettes/day2||17%||26%|
|Time since quit (years)1||15|
|Body mass index (kg/m2)1||25.7||25.8||24.6|
|Body mass index > 30||16%||16%||10%|
|Age at menarche (years)1||12.6||12.6||12.5|
|Parity (among parous)1||3.1||3.1||3.1|
|Ever used oral contraceptives||46%||50%||50%|
|Age at menopause (years)||49.7||49.4||48.4|
|Past PMH use3||12%||14%||13%|
|Current PMH use3||27%||29%||21%|
|Estrogen only use4||20%||17%||26%|
|Estrogen/progesterone or progesterone only use4||62%||63%||57%|
Both current and past smoking status were associated with a significantly lower risk of endometrial cancer compared to never smoking. The age-adjusted relative risk for current and past smokers was 0.61 and 0.77, respectively. When adjusted for all other risk factors except BMI, the RRs for current and past smokers were 0.63 (95% CI = 0.50–0.79) and 0.73 (95% CI = 0.62–0.87), respectively. Additional adjustment for BMI attenuated the RR for current smokers (0.72; 95% CI = 0.57–0.90) but did not alter the RR of past smokers (0.73; 95% CI = 0.62–0.87; Table II).
|Number of cases||Age-adjusted RR||Multivariate|
|RR11 (95% CI)||RR22 (95% CI)||RR33 (95% CI)|
|Past||235||0.77||0.73 (0.62–0.87)||0.73 (0.62–0.87)|
|Current||98||0.61||0.63 (0.50–0.79)||0.72 (0.57–0.90)|
|Amount smoked4 (cigarettes/day)|
|1 to 4||51||0.97||0.92 (0.68–1.23)||0.96 (0.71–1.29)||1.0|
|5 to 14||97||0.66||0.67 (0.53–0.84)||0.70 (0.56–0.88)||0.72 (0.51–1.03)|
|15 to 24||114||0.69||0.68 (0.55–0.85)||0.72 (0.58–0.89)||0.77 (0.55–1.08)|
|25 to 34||41||0.76||0.72 (0.52–0.99)||0.74 (0.53–1.02)||0.79 (0.52–1.21)|
|35+||23||0.68||0.64 (0.42–0.98)||0.60 (0.39–0.91)||0.64 (0.39–1.06)|
|p for trend5||0.39||0.25||0.11||0.22|
|Amount smoked among past smokers only (cigarettes/day)|
|1 to 14||115||0.79||0.76 (0.62–0.94)||0.78 (0.63–0.96)||1.0|
|15 to 24||75||0.73||0.72 (0.56–0.93)||0.72 (0.56–0.93)||0.95 (0.71–1.28)|
|25+||41||0.81||0.73 (0.52–1.00)||0.68 (0.49–0.95)||0.91 (0.63–1.32)|
|p for trend5||0.92||0.73||0.43||0.59|
|Amount smoked among current smokers only (cigarettes/day)|
|1 to 14||33||0.62||0.66 (0.46–0.95)||0.76 (0.53–1.08)||1.0|
|15 to 24||39||0.61||0.62 (0.44–0.87)||0.72 (0.52–1.01)||0.99 (0.62–1.61)|
|25+||23||0.62||0.63 (0.41–0.96)||0.68 (0.44–1.04)||0.93 (0.54–1.61)|
|p for trend5||0.93||0.53||0.37||0.58|
|Smoking duration4 (years)|
|> 0 to 10||62||0.82||0.80 (0.61–1.04)||0.81 (0.62–1.07)||1.0|
|> 10 to 20||56||0.68||0.67 (0.50–0.88)||0.68 (0.52–0.91)||0.86 (0.59–1.24)|
|> 20 to 30||74||0.77||0.74 (0.57–0.95)||0.74 (0.57–0.95)||0.91 (0.64–1.30)|
|> 30 to 40||91||0.73||0.69 (0.55–0.88)||0.72 (0.57–0.92)||0.88 (0.61–1.27)|
|> 40||42||0.54||0.55 (0.40–0.77)||0.63 (0.45–0.87)||0.75 (0.47–1.19)|
|p for trend5||0.10||0.09||0.19||0.34|
|Smoking duration among past smokers only4 (years)|
|> 0 to 20||113||0.75||0.72 (0.59–0.90)||0.74 (0.60–0.92)||1.0|
|> 20 to 30||54||0.77||0.71 (0.54–0.95)||0.69 (0.52–0.92)||0.93 (0.66–1.29)|
|> 30||62||0.77||0.75 (0.57–0.99)||0.73 (0.56–0.97)||0.99 (0.71–1.37)|
|p for trend5||0.59||0.54||0.39||0.83|
Both the intensity (number of cigarettes per day) and duration of smoking were inversely related to risk; women in the top categories of each were at a significantly lower risk of endometrial cancer compared to never smokers. However, there were no significant trends of decreasing risk with increasing intensity or duration. These associations were relatively unchanged when smoking intensity and duration were evaluated among ever smokers in the same statistical model. Similar inverse associations were also noted when amount smoked was evaluated in past and current smokers separately. The RR associated with smoking 15–24 cigarettes per day, relative to the risk in never smokers, was 0.72 (95% CI = 0.56–0.93) for past smokers and 0.72 (95% CI = 0.52–1.01) for current smokers. The duration of smoking was not evaluated separately in current smokers because of the somewhat limited range in age of smoking initiation and due to the high correlation between duration and current age.
Because of the inverse associations observed for amount and duration, we next evaluated the association with pack-years of smoking, a variable that combines these 2 components (Table III). Pack-years were inversely associated with risk, though a dose-response effect was not observed (p for trend = 0.10). Women smoking more than 50 pack-years had an approximately 50% lower risk of endometrial cancer relative to that of never smokers. To evaluate more explicitly smoking early in life, we assessed the relationship between pack-years smoked prior to age 30, controlling for pack-years smoked after age 30, and subsequent risk of endometrial cancer. A significant inverse association was observed for those with 10 or more pack-years before age 30 (RR = 0.66; 95% CI = 0.48–0.91) compared to never smokers. The analysis of pack-years smoked after age 30 showed a significantly decreased risk in those smoking over 40 pack-years (RR = 0.47; 95% CI = 0.24–0.92).
|Number of cases||Age-adjusted RR||Multivariate|
|RR1 (95% CI)||RR22 (95% CI)|
|1 to 10||108||0.81||0.79 (0.64–0.98)||0.82 (0.66–1.01)|
|> 10 to 20||48||0.58||0.57 (0.42–0.78)||0.59 (0.44–0.80)|
|> 20 to 30||58||0.87||0.86 (0.65–1.14)||0.87 (0.66–1.15)|
|> 30 to 40||52||0.79||0.80 (0.59–1.07)||0.84 (0.63–1.13)|
|> 40 to 50||26||0.58||0.55 (0.36–0.82)||0.57 (0.38–0.86)|
|> 50||28||0.48||0.47 (0.32–0.69)||0.50 (0.34–0.74)|
|p for trend3||0.04||0.09||0.10|
|Pack years before age 304|
|1 to < 5||129||0.75||0.82 (0.65–1.04)||0.85 (0.67–1.07)|
|5 to < 10||117||0.70||0.78 (0.59–1.03)||0.79 (0.60–1.05)|
|10+||64||0.63||0.67 (0.48–0.92)||0.66 (0.48–0.91)|
|p for trend3||0.25||0.16||0.06|
|Pack years after age 305|
|1 to 10||113||0.76||0.65 (0.36–1.16)||0.63 (0.35–1.13)|
|> 10 to 20||49||0.69||0.60 (0.32–1.13)||0.57 (0.30–1.09)|
|20+||121||0.66||0.58 (0.31–1.06)||0.58 (0.31–1.07)|
|p for trend3||0.25||0.37||0.45|
Age first smoked was unrelated to risk of endometrial cancer (p for trend = 0.94; Table IV). A linear association was not observed for time since quitting (p for trend = 0.19). Findings for age first smoked and time since quitting were somewhat attenuated when both variables plus cigarettes per day were included in the same model.
|Number of cases||Age-adjusted RR||Multivariate|
|RR11 (95% CI)||RR22 (95% CI)||RR33 (95% CI)|
|Age first smoked (years)|
|13 to < 18||65||0.79||0.78 (0.60–1.02)||0.79 (0.60–1.03)||1.0|
|18 to < 20||110||0.68||0.68 (0.55–0.84)||0.70 (0.57–0.87)||0.89 (0.64–1.22)|
|20 to < 22||93||0.71||0.69 (0.55–0.87)||0.73 (0.58–0.92)||0.92 (0.66–1.27)|
|≥ 22||60||0.72||0.70 (0.53–0.92)||0.74 (0.56–0.98)||0.92 (0.64–1.32)|
|p for trend4||0.88||0.99||0.94||0.94|
|Years since quit|
|20+ years||100||0.78||0.74 (0.59–0.93)||0.76 (0.61–0.95)||1.0|
|15 to < 20||35||0.92||0.87 (0.61–1.23)||0.86 (0.60–1.22)||1.13 (0.76–1.67)|
|10 to < 15||37||0.89||0.86 (0.61–1.20)||0.82 (0.58–1.16)||1.06 (0.72–1.56)|
|5 to < 10||31||0.66||0.64 (0.44–0.92)||0.61 (0.43–0.89)||0.78 (0.51–1.19)|
|< 5 years||30||0.60||0.59 (0.40–0.86)||0.59 (0.40–0.87)||0.83 (0.55–1.25)|
|p for trend4||0.28||0.27||0.19||0.23|
Further analyses were run with BMI modeled as a continuous instead of categorical variable and with additional covariates for age at menopause (as a continuous variable) and menstrual irregularity. In all cases, results were essentially unchanged. Similarly, analyses controlling for PMH use by both duration and type of formulation (estrogen vs. estrogen plus progesterone) did not alter the smoking estimates. Two models with data from 1980–2000 that included either alcohol and animal and vegetable fat or alcohol, trans fatty acids and saturated, polyunsaturated and monounsaturated fats did not result in any significant change in relative risk for any of the smoking variables. Although statistical power was limited, no substantial effect modification was observed in analyses stratified by BMI, menopausal status, or PMH use, and no test for heterogeneity was statistically significant (all p-values for heterogeneity > 0.1).
This study represents the largest prospective cohort study of smoking and endometrial cancer to date and the first with repeated assessments of smoking status over 24 years, resulting in more accurate classification of current and past smoking status during follow-up. We observed a statistically significant decrease in the risk of endometrial cancer among both current and past smokers compared to never smokers, although trends with increasing dose and duration of smoking were not significant.
Several case-control and 3 prospective cohort studies have been previously published.16, 17, 18, 19, 20 In general, these studies have shown an inverse association, but few of the relative risks were statistically significant.12 The largest prior prospective cohort study found a significantly decreased risk for women currently smoking more than 20 cigarettes a day but no association in heavy past smokers. In that study, neither exposure information nor hysterectomy status was updated during follow-up (average 10.6 years); hence, some potential for misclassification and bias was present. In our study, biennial updates of both exposure information and study eligibility minimize these potential limitations.
The biologic mechanism whereby smoking reduces the risk of endometrial cancer is likely to be multifactorial. Prior evidence has shown that endometrial cancer risk in postmenopausal women is influenced both by current exposures, such as weight or estrogen replacement therapy, and perhaps to a lesser extent by past exposures, such as the duration of oral contraceptive use or parity. We found that the associations of current and past smoking to endometrial cancer risk were both significantly decreased and quite similar after controlling for BMI. These analyses indicate that the lower BMI of current smokers contributed to their lower cancer risk; however, the inverse associations observed even after careful control for BMI suggest that additional mechanisms contribute to this risk reduction. In addition, exposure early in life, defined as pack-years smoked before age 30, was inversely related to risk. Thus, smoking appears to reduce the risk of endometrial cancer at least in part by a prolonged rather than solely by an immediate hormonal effect.
Smoking-induced alterations in steroid production and metabolism may play a role in the reduced risk of endometrial cancer observed in cigarette smokers. Estrogen may contribute to the development of endometrial cancer by initiating mutations and promoting cellular proliferation in endometrial glands.34 Progestins and androgens decrease estrogen-induced cellular proliferation in endometrial glands35 and protect against the development of endometrial cancer. In premenopausal and postmenopausal women, cigarette smoking does not appear to alter endogenous circulating levels of estrone and estradiol36, 37, 38, 39, 40, 41 but does increase circulating androgens, including serum dehydroepiandosterone sulfate, androstenedione, testosterone and progesterone.36, 39, 42, 43, 44 The mechanisms by which smoking influences circulating progesterone and androgens have not been fully characterized, but may include the following. One, nicotine increases pituitary release of adrenocorticotropic hormone, thereby causing increased adrenal progesterone and androgen secretion.45, 46 Two, components of cigarette smoke, including cotinine, block the adrenal 11- and 21-hydroxylase enzymes, thereby increasing adrenal steroid production.47, 48 Three, smoking slows metabolism of progesterone and androgens, thereby raising circulating concentrations.40 Smoking also shifts estradiol metabolism toward 2-hydroxylation pathways, resulting in an increased level of 2-hydroxyestrone instead of the more biologically active 16α-hydroxyestrone.11, 49
Smoking may also have a direct toxic effect on the ovaries. Tobacco smoke contains polycyclic aromatic hydrocarbons, which cause oocyte destruction and ovarian failure in exposed mice.50, 51 Recent experiments demonstrate that proapoptotic Bax expression is increased in oocytes exposed to polycyclic aromatic hydrocarbons.51 Both premenopausal women who have received chemotherapy and smokers undergo a significant decrease in the number of ovarian follicles, resulting in an earlier menopause.52 Smokers (even past smokers) also have been reported to have higher levels of follicle-stimulating hormone, an indicator of menopause.37, 53
When we divided smoking status into categories by amount and duration, there was a decrease in the relative risks in several categories, but the trend across categories was not statistically significant. The dose-response relationship for the biologic mechanisms described above are not established, and thus the effects may not increase linearly with increasing exposure. Alternatively, the lack of trend may be related to the relatively small number of exposed cases in some of the analyses; this can be reassessed after additional follow-up. Stratification by menopausal status (pre- vs. post-), BMI and PMH use showed no significant differences from the overall analysis, a finding that also concurs with previously published results.
Limitations of this study include possible misclassification and possible exclusion of some early invasive cancers from the analyses. The primary analysis was limited to invasive endometrial cancer. Although we evaluated only endometrial adenocarcinoma, other histologies are rare (clear cell, papillary serous) and may be estrogen-independent.54 Finally, our cohort is predominantly white and hence associations may not be generalizable to all ethnic groups.
In summary, this study represents the largest prospective cohort examination to date of the risk of endometrial cancer based on smoking status. Our results are the first to show clearly that endometrial cancer risk may be influenced by past as well as current smoking exposure.
Supported by the Gillette Women's Cancer Center (to A.N.V.), an American Cancer Society Research Scholarship (RSG-00-061-04CCE to I.D.V.) an American Cancer Society Clinical Research Professorship (to G.A.C.), and the National Institutes of Health CA82838 and CA87969.
- 1Society AC. Cancer facts and figures: ACS statistics for 2003. Atlanta, GA: American Cancer Society, 2003.
- 2Cancer of the female genital tract: cancer manual of the American Cancer Society. Atlanta, GA: American Cancer Society, 1995., ,
- 13Surgeon General. Women and smoking: a report of the Surgeon General. Washington, DC: Public Health Service, 2001.