Title
Author
DOI
Article Type
Special Issue
Volume
Issue
1Department of Obstetrics & Gynecology (Division of Gynecologic Oncology), Stanford University School of Medicine, Palo Alto, CA 94305, USA
2Quantitative Sciences Unit, Stanford University School of Medicine, Palo Alto, CA 94305, USA
3Department of Medicine (Oncology), Palo Alto, CA 94305, USA
4Department of Epidemiology and Population Health, Stanford University School of Medicine, Palo Alto, CA 94305, USA
5Kaiser Permanente Northern California Division of Research, Oakland, CA 94305, USA
6Department of Medicine (Stanford Prevention Research Center), Palo Alto, CA 94305, USA
7Department of Obstetrics and Gynecology, Stanford University School of Medicine, Palo Alto, CA 94305, USA
*Corresponding Author(s):ssomaseg@stanford.edu (Sahana Somasegar)
| History | Submitted: 26 February 2025 | Accepted: 30 April 2025 | Published: 15 August 2025 |
| Copyright: | ©2025 The Author(s). Published by MRE Press. |
Background: Obesity, the major risk factor for endometrial cancer, has varying prevalence geographically. We aimed to examine patterns in obesity, demographic characteristics, lifestyle risk factors and histopathologic features among women with and without endometrial cancer and to evaluate their association with survival outcomes. Methods: We utilized data from the Women’s Health Initiative cohort study to analyze 93,676 postmenopausal women with a uterus enrolled between 1993 and 1998, followed through 2021. Demographic and lifestyle characteristics were compared between women with and without endometrial cancer using absolute standardized mean differences. Histologic subtype and stage at diagnosis were stratified by race and geographic region. Multivariable Cox proportional hazards models assessed associations between demographic and lifestyle factors and survival. Kaplan-Meier curves and log-rank tests compared survival by region. Results: Our cohort included 92,040 participants without and 1826 participants with incident endometrial cancer over a mean 16.7 (±7.36) years of follow-up. Women with endometrial cancer were more likely to be obese (BMI (body mass index) ≥30), non-Hispanic White, and have hypertension and diabetes. Obesity conferred a higher risk of endometrial cancer, but its effect was not modified by race. Histologic subtype and stage at diagnosis varied by race and region, with women in the South having a higher prevalence of aggressive histologies and advanced-stage cancers. Survival did not differ significantly by region (log-rank p = 0.187) and White vs. Black race did not modify this association (p = 0.857). Conclusions: These findings contribute to our understanding of endometrial cancer epidemiology and highlight the importance of considering racial disparities and histologic subtypes when studying the disease. While survival outcomes did not differ significantly by geographic region, notable disparities in tumor characteristics highlight the importance of a uniform focus on obesity prevention. Equitable access to prevention, early detection and treatment strategies can help inform health policy at the national level.
Cite this article
Sahana Somasegar, Haley Hedlin, Allison W. Kurian, Candyce Kroenke, Oliver Dorigo, Marcia L. Stefanick. Obesity, demographic, and lifestyle risk factor patterns associated with endometrial cancer diagnosis and survival: insights from the Women’s Health Initiative.European Journal of Gynaecological Oncology,2025,46(8):11-25 DOI:10.22514/ejgo.2025.105
1. Introduction
Endometrial cancer is the most common gynecologic cancer in the United States, with an estimated 69,120 new cases and 13,860 deaths in 2025; among these, 40% of cases and 67% of deaths are projected to be in women aged 65 and older [1, 2]. Despite advances in effective treatment options, the incidence of endometrial cancer has continued to increase in the United States [1, 2] and deaths from endometrial cancer have been steadily climbing, in sharp contrast to the drop in lung, breast and colorectal cancer deaths [1, 2].
Obesity is a well-established risk factor for endometrial cancer, with women with obesity, i.e., body mass index (BMI) ≥30 kg/m2, being two to four times more likely to develop endometrial cancer than normal-weight women [3]. Other risk factors include postmenopausal estrogen-only therapy use, other conditions accompanied by estrogens unopposed by progesterone, such as polycystic ovarian syndrome, early menarche, late menopause, hereditary cancer syndromes, including Lynch syndrome, older age, diabetes, hypertension, nulliparity or multiparity and family history [4].
Based on Centers for Disease Control and Prevention data, obesity rates differ by U.S. geographic region, being higher in the South (36.3%) and Midwest (35.4%) than the Northeast (29.9%) and West (28.7%) in 2021, and by race, with 56.2% of non-Hispanic Black U.S. women having obesity compared to 39.0% of non-Hispanic White women in 2018 [5]. Whereas the incidence of uterine corpus cancer was slightly higher in Black (28.4 per 100,000 women per year) than in White (27.8 per 100,000 women year) women, the mortality rate was nearly double in Black (9.1 per 100,000 women per year) versus White women (4.6 per 100,000 women per year) in 2023 [1].
While several studies show racial disparities in endometrial cancer outcomes [6, 7], data are limited for U.S. geographic variations in endometrial cancer in relation to demographic and lifestyle risk factors (particularly obesity), diagnoses and survival. We utilized data collected in the Women’s Health Initiative (WHI) cohort study, which has followed over 20,000 postmenopausal women in each U.S. geographic census region for decades, to compare baseline demographic and lifestyle factors between women who were or were not diagnosed with endometrial cancer over longitudinal follow-up, overall and by geographic region, to determine differences in baseline risks and whether there were any regional differences for cancer histology, grade, stage at diagnosis and overall survival. We also evaluated whether there were any differences in the relationships between region and survival for White versus Black women.
2. Methods
2.1 Study population
Postmenopausal women aged 50–79 years were enrolled at 40 clinical sites across the United States between 1993 and 1998 into one or more randomized clinical trials (WHI-CT) which excluded women with a history of breast cancer ever or any cancer other than non-melanoma skin cancer within the past 10 years (WHI-CT; n = 68,132) or an observational study (WHI-OS), if they were either ineligible or chose not to participate in one of the clinical trials (WHI-OS; n = 93,676) [8]. Only women who had a uterus and no history of endometrial cancer at enrollment were included in these analyses. The clinical trials included two WHI randomized, placebo-controlled menopausal hormone therapy (MHT) trials, a combined estrogen-progestin (E+P) trial for women with a uterus and an estrogen-only (E only) trial for women with prior hysterectomy, the later of whom were not eligible for the current analysis, and a randomized controlled low-fat diet modification (DM) trial [8]. The WHI E+P trial was stopped early due to overall risks of the combined estrogen and progestin therapy outweighing harms, with no significant effect on endometrial cancer incidence [9]. The WHI-DM and a placebo-controlled trial of calcium and vitamin D supplementation in WHI-CT participants who were invited to join at their first (and second) annual CT visit, and the WHI-OS were closed in 2004–2005, but participants were invited to reconsent to a 5-year follow-up by their respective clinical centers or the WHI Clinical Coordinating Center (CCC) (2005–2010). In 2010, active participants were invited to consent to continued follow-up by one of four regional centers or the CCC in a five-year WHI-Extension Study (WHI-ES, 2010–2015), and in 2015, those who were still active, were invited to consent again to ongoing follow-up with no specified end date [8, 9]. Written informed consent was obtained from all study participants. The Women’s Health Initiative (WHI) study protocols were approved by the Institutional Review Board of the Fred Hutchinson Cancer Research Center (Seattle, WA) and by the Institutional Review Boards of each participating institution at baseline (1993–1998) and in 2005, 2010 and 2015, as described above. The WHI study reference number at the Fred Hutchinson Cancer Research Center is IRB# 4451.
2.2 Data collection
At baseline of the WHI-CT and WHI-OS, participants completed self-administered questionnaires detailing demographic characteristics, race, geographic region, health behaviors, previous use of postmenopausal hormone therapy and selected medications, and medical histories. Regarding race, participants were able to check one or more boxes of the following racial categories: White, Black, American Indian, Asian, Hawaiian Pacific Islander or Other. Participants also completed a baseline clinic visit, during which trained staff measured each participant’s weight and height using a standardized protocol, and BMI (kg/m2) was calculated based on those measurements. BMI was then updated at annual clinic visits for WHI-CT participants through 2004–2005 and at the Year 3 clinic visit for OS participants.
2.3 Outcome
Adjudication and outcome ascertainment for the WHI have been described elsewhere [10, 11]. Briefly, cancer cases were self-reported in questionnaires, semi-annually in WHI-CT and annually in WHI-OS through 2010, with 93–96% completion rates and annually from 2010 onward. As reported previously [11], local physicians adjudicated endometrial cancer diagnoses through medical and pathology records review to finalize each cancer outcome, according to guidelines from Surveillance Epidemiology and End Results (SEER). Tumors were histopathologically classified based on pathology reports according to World Health Organization and International Society of Gynecological Pathology guidelines [11]. Information on tumor characteristics, including stage (according to 1988 or 2009 International Federation of Gynecologic Oncology (FIGO) criteria depending on year of diagnosis), histology (endometrioid, serous, carcinosarcoma, clear cell, mixed epithelial), and grade (1–3) were collected from pathology reports. Only endometrial cancer cases confirmed by adjudication were included in these analyses. Overall survival was defined as time from endometrial cancer diagnosis until death. Women who were lost-to-follow-up were right-censored at their last visit date and women who were alive at the time the database snapshot was taken (06 March 2021) were right-censored at their last follow-up visit prior.
2.4 Statistical analysis
Baseline demographic and lifestyle characteristics are displayed among eligible participants who did versus did not develop endometrial cancer during the follow-up period overall and within each geographic census region (Midwest, Northeast, South, West). For analyses of participants who developed endometrial cancer, women were excluded if informational on histopathological subtype was missing or if the diagnosis was not confirmed by local physician adjudication. Among the women with incident endometrial cancer, endometrial cancer history, including histology, grading and stage at diagnosis, were compared across the same geographic regions. Regarding race, women who only selected White or Black were categorized as these respective races in “Race Categories” for our analysis. Due to very small numbers of women of American Indian, Asian, Hawaiian Pacific Islander, Other or Unknown race, these women were categorized as “All Others”. Characteristics were compared between groups using the absolute standardized deviation (ASD), a measure of the difference between groups in units of standard deviations. The ASD is calculated as the difference in means or proportions divided by the pooled standard deviation and is not influenced by sample size, unlike p-values. ASD values provide a quantitative assessment of the magnitude of difference between groups: values <0.1 indicate a negligible difference, 0.1–0.2 represent small differences, 0.2–0.5 moderate differences and >0.5 suggest large differences, according to Cohen’s guidelines [12]. In the context of this study, ASDs help identify clinically relevant imbalances in baseline characteristics that may influence outcomes.
To compare overall survival after endometrial cancer diagnoses across the geographic regions, we fit a multivariable adjusted Cox proportional hazards model including all women to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) comparing overall survival between regions. The model was adjusted for age at diagnosis, race (White, Black or self-selected race other than White or Black, due to insufficient numbers for statistical analyses of other racial minority groups), Hispanic ethnicity, high school degree, BMI (kg/m2), total energy expended from recreational physical activity (metabolic equivalent (MET)-hours/week), total healthy eating index score-2015 [13], stage at diagnosis and histologic grade at diagnosis (measured at WHI baseline unless otherwise noted). To evaluate whether the association between region and overall survival after endometrial cancer differed by White versus Black race, we added a statistical interaction term between race and region and used a pooled Wald test to jointly test whether the statistical interaction terms were statistically significant, indicating the presence of effect modification.
In the subset of women who had endometrioid adenocarcinoma or serous/papillary serous cancers, we fit a multivariable adjusted logistic regression model comparing the odds of endometrioid adenocarcinoma by region. As endometrioid adenocarcinoma and serous/papillary serous cancers were the most common histologies in our study population, we compared them in this analysis as this comparison is likely to be most relevant to the population being studied. The adjustment variables were the same as those used in the survival analysis.
All statistical tests were two-sided and considered statistically significant at the α = 0.05 level. Statistical analyses were performed using R version 4.2.1. Given the number of comparisons performed, we acknowledge an increased risk of Type I error. These analyses were primarily exploratory in nature, and findings should be interpreted accordingly.
3. Results
Among the 161,808 women enrolled in the WHI-CT or OS, 93,899 participants had a uterus and no history of an endometrial cancer diagnosis at baseline, of whom 92,040 were not diagnosed with endometrial cancer during the follow up period for this analysis; whereas, 1859 participants were diagnosed with endometrial cancer, 33 of whom had missing cancer history data and were excluded from analyses specific to the endometrial cancer analytic cohort, leaving a total of 1826 cases (Fig. 1).
Fig. 1.Flow diagram illustrating WHI participant inclusion and exclusion criteria for current study analysis. Bold: Non-endometrial cancer analytic cohort and Endometrial Cancer Analytic Cohort were finally cohorts included in study analysis.
When comparing baseline characteristics of participants who were diagnosed with endometrial cancer versus those who were not (Table 1), a moderate difference was seen for BMI (ASD = 0.308), with 41.9% of those diagnosed with endometrial cancer having obesity, defined as BMI ≥30 kg/m2, compared to 27.5% of those with no endometrial cancer diagnosis. Most women in the entire study cohort, with and without endometrial cancer, were White. We evaluated any differences in the proportion of White and Black women among women with and without endometrial cancer. There were small differences in the racial distribution of women with and without endometrial cancer (ASD = 0.200), with 93% of women with endometrial cancer compared to 87% of women without endometrial cancer being White and 5% of women with endometrial cancer compared to 7% of women without endometrial cancer being Black. Among women included in the “All Others” race category, 1% of women with endometrial cancer identified, at least in part, as Asian compared to 3% of women without endometrial cancer. There were no differences between the groups with respect to mean baseline age (ASD = 0.055) or most other variables analyzed, such as completion of high school education, Hispanic ethnicity, smoking status, oral contraceptive use, parity, prior pregnancy, age of menarche, age of first birth, diabetes, hypertension, physical activity or diet.
| Endometrial Cancer | No Endometrial Cancer | ASD | ||
| n | 1859 | 92,040 | ||
| Age (mean (SD)) | 62.66 (6.93) | 63.05 (7.21) | 0.055 | |
| Race (n (%))* | ||||
| White | 1730 (93.6) | 80,790 (89.0) | 0.164 | |
| Black | 89 (4.8) | 6447 (7.1) | 0.097 | |
| American Indian | 6 (0.3) | 745 (0.8) | 0.066 | |
| Asian | 25 (1.4) | 2324 (2.6) | 0.087 | |
| Hawaiian Pacific Islander | 2 (0.1) | 195 (0.2) | 0.027 | |
| Other | 4 (0.2) | 828 (0.9) | 0.093 | |
| Unknown | 11 (0.6) | 1267 (1.4) | 0.080 | |
| Race Categories (n (%)) | ||||
| White | 1722 (92.6) | 80,035 (87.0) | 0.200 | |
| Black | 87 (4.7) | 6257 (6.8) | ||
| All Others | 50 (2.7) | 5748 (6.2) | ||
| Hispanic (n (%)) | 44 (2.4) | 4052 (4.4) | 0.114 | |
| BMI (n (%)) | ||||
| <24.9 | 545 (29.6) | 34,840 (38.2) | 0.308 | |
| 25.0–29.9 | 526 (28.5) | 31,276 (34.3) | ||
| ≥30.0 | 773 (41.9) | 25,067 (27.5) | ||
| Smoking Status (n (%)) | ||||
| Current | 85 (4.6) | 6381 (7.0) | 0.111 | |
| Former | 769 (41.7) | 38,690 (42.6) | ||
| Never | 988 (53.6) | 45,763 (50.4) | ||
| Oral Contraceptive Use (n (%)) | 758 (40.8) | 38,983 (42.4) | 0.032 | |
| Parity (n (%)) | ||||
| 0 | 424 (22.8) | 20,114 (21.9) | 0.108 | |
| 1 | 170 (9.2) | 7642 (8.3) | ||
| 2 | 430 (23.2) | 20,945 (22.8) | ||
| 3 | 432 (23.3) | 19,879 (21.7) | ||
| 4 | 230 (12.4) | 12,104 (13.2) | ||
| 5+ | 171 (9.2) | 11,108 (12.1) | ||
| Prior Pregnancy (n (%)) | 1628 (87.7) | 82,544 (89.9) | 0.072 | |
| Age of First Birth (n (%)) | ||||
| <20 | 157 (10.6) | 8973 (12.2) | 0.066 | |
| 20–29 | 1102 (74.3) | 54,096 (73.3) | ||
| 30+ | 160 (10.8) | 8155 (11.1) | ||
| Age of Menarche (n (%)) | ||||
| 9 or younger | 16 (0.9) | 1062 (1.2) | 0.188 | |
| 10–11 | 424 (22.9) | 17,989 (19.6) | ||
| 12–13 | 1097 (59.1) | 50,693 (55.3) | ||
| 14–16 | 308 (16.7) | 21,010 (10.0) | ||
| 17+ | 11 (0.6) | 917 (1.0) | ||
| History of Diabetes (n (%)) | 103 (5.5) | 4629 (5.0) | 0.023 | |
| Hypertension (n (%)) | ||||
| Never | 1193 (68.2) | 61,098 (70.0) | 0.046 | |
| Untreated | 131 (7.5) | 6664 (7.6) | ||
| On Medications | 424 (24.3) | 19,462 (22.3) | ||
| High School Degree (n (%)) | 1796 (97.1) | 87,175 (95.4) | 0.091 | |
| Weekly Exercise (median mins (IQR)) | 9.00 (2.50, 18.19) | 9.00 (2.50, 18.92) | 0.012 | |
| Healthy Eating Index 2015 (mean (SD)) | 65.25 (10.42) | 65.38 (10.44) | 0.013 | |
| Hormone Therapy Trial Arm | ||||
| E+P Control | 139 (7.5) | 7962 (8.7) | 0.146 | |
| E+P Intervention | 104 (5.6) | 8399 (9.1) | ||
| Observational Study | 1616 (86.9) | 75,679 (82.2) | ||
| Dietary Modification (DM) Trial Arm | ||||
| DM Control | 375 (20.2) | 16,143 (17.5) | 0.105 | |
| DM Intervention | 260 (14.0) | 10,821 (11.8) | ||
| Not Randomized to DM | 1224 (65.8) | 65,076 (70.7) | ||
| *Participants could select more than one race; ASD: absolute standardized difference, a measure of the difference between groups in units of standard deviations that can be interpreted using Cohen’s guidelines (d: 0.2 = small difference; 0.5 = moderate difference; 0.8 = large difference; d < 0.2 = trivial difference); BMI: body mass index, IQR: interquartile range (25th, 75th percentile); SD: standard deviation; E+P: estrogen-progestin. |
Similar patterns were observed within each of the four U.S. geographic census regions (Table 2) between participants who were diagnosed with endometrial cancer versus those who were not. In particular, moderate differences for BMI were seen between women who were versus were not diagnosed with endometrial cancer in all geographic regions (ASD = 0.364 in Midwest, ASD = 0.398 in Northeast, ASD = 0.254 in South, ASD = 0.228 in West), with the proportion of participants who had obesity being notably higher among those with endometrial cancer compared to those without in all geographic census regions (47.7% vs. 30.3% in Midwest, 46.7% vs. 28.1% in Northeast, 39.4% vs. 27.5% in South, 34.8% vs. 24.7% in West). In all geographic census regions, there were no differences in baseline age between those with and without endometrial cancer. While there was no difference in racial categories between those with and without endometrial cancer in the Midwest or Northeast, moderate differences were seen in the South (ASD = 0.236) and West (ASD = 0.308). There were no other substantial differences between participants who were versus were not diagnosed with endometrial cancer in other variables analyzed (Table 2).
| Midwest | Northeast | South | West | ||||||||||
| EC | No EC | ASD | EC | No EC | ASD | EC | No EC | ASD | EC | No EC | ASD | ||
| n | 417 | 21,093 | 496 | 23,854 | 428 | 21,092 | 518 | 26,001 | |||||
| Age (mean (SD)) | 61.86 (6.78) | 62.92 (6.99) | 0.154 | 62.80 (6.57) | 63.26 (6.91) | 0.068 | 61.95 (6.92) | 62.36 (7.29) | 0.057 | 63.76 (7.23) | 63.54 (7.52) | 0.031 | |
| Race (n (%))* | |||||||||||||
| White | 397 (95.2) | 19,274 (91.8) | 0.138 | 466 (94.1) | 22,008 (93.0) | 0.047 | 386 (90.8) | 17,548 (84.6) | 0.190 | 481 (94.1) | 21,960 (86.6) | 0.259 | |
| Black | 19 (4.6) | 1523 (7.3) | 0.115 | 22 (4.4) | 1359 (5.7) | 0.059 | 37 (8.7) | 2840 (13.7) | 0.159 | 11 (2.2) | 725 (2.9) | 0.045 | |
| American Indian | 1 (0.2) | 95 (0.5) | 0.036 | 2 (0.4) | 115 (0.5) | 0.012 | 2 (0.5) | 175 (0.8) | 0.046 | 1 (0.2) | 360 (1.4) | 0.137 | |
| Asian | 1 (0.2) | 114 (0.5) | 0.049 | 6 (1.2) | 148 (0.6) | 0.062 | 1 (0.2) | 152 (0.7) | 0.072 | 17 (3.3) | 1910 (7.5) | 0.186 | |
| Hawaiian Pacific Islander | 0 (0.0) | 1 (0.0) | 0.010 | 1 (0.2) | 7 (0.0) | 0.051 | 0 (0.0) | 7 (0.0) | 0.026 | 1 (0.2) | 181 (0.7) | 0.077 | |
| Other | 2 (0.5) | 107 (0.5) | 0.004 | 1 (0.2) | 178 (0.8) | 0.080 | 0 (0.0) | 170 (0.8) | 0.129 | 1 (0.2) | 373 (1.5) | 0.141 | |
| Unknown | 0 (0.0) | 98 (0.5) | 0.097 | 1 (0.2) | 184 (0.8) | 0.082 | 3 (0.7) | 356 (1.7) | 0.091 | 7 (1.4) | 629 (2.4) | 0.079 | |
| Race Categories (n (%)) | |||||||||||||
| White | 394 (94.5) | 19,164 (90.9) | 0.144 | 464 (93.5) | 21,886 (91.7) | 0.072 | 385 (90.0) | 17,403 (82.5) | 0.236 | 479 (92.5) | 21,582 (83.0) | 0.308 | |
| Black | 19 (4.6) | 1487 (7.0) | 20 (4.0) | 1301 (5.5) | 37 (8.6) | 2784 (13.2) | 11 (2.1) | 685 (2.6) | |||||
| All Others | 4 (1.0) | 442 (2.1) | 12 (2.4) | 667 (2.8) | 6 (1.4) | 905 (4.3) | 28 (5.4) | 3734 (14.4) | |||||
| Hispanic (n (%)) | 3 (0.7) | 183 (0.9) | 0.017 | 12 (2.4) | 645 (2.7) | 0.018 | 16 (3.7) | 1500 (7.1) | 0.150 | 13 (2.5) | 1724 (6.8) | 0.203 | |
| BMI (n (%)) | |||||||||||||
| <24.9 | 109 (26.4) | 7266 (34.9) | 0.364 | 121 (24.5) | 8638 (36.5) | 0.398 | 139 (32.8) | 8179 (39.1) | 0.254 | 176 (34.2) | 10,757 (41.7) | 0.228 | |
| 25.0–29.9 | 107 (25.9) | 7270 (34.9) | 142 (28.8) | 8371 (35.4) | 118 (27.8) | 6968 (33.4) | 159 (30.9) | 8667 (33.6) | |||||
| ≥30.0 | 197 (47.7) | 6305 (30.3) | 230 (46.7) | 6651 (28.1) | 167 (39.4) | 5746 (27.5) | 179 (34.8) | 6365 (24.7) | |||||
| Smoking Status (n (%)) | |||||||||||||
| Current | 20 (4.8) | 1511 (7.2) | 0.128 | 24 (4.8) | 1703 (7.2) | 0.101 | 19 (4.5) | 1529 (7.4) | 0.135 | 22 (4.3) | 1638 (6.4) | 0.094 | |
| Former | 159 (38.3) | 8560 (41.0) | 232 (46.9) | 11,000 (46.6) | 163 (38.4) | 8276 (39.9) | 215 (42.3) | 10,854 (42.4) | |||||
| Never | 236 (56.9) | 10811 (51.8) | 239 (48.3) | 10,914 (46.2) | 242 (57.1) | 10,914 (52.7) | 271 (53.3) | 13,124 (51.2) | |||||
| Oral Contraceptive Use (n (%)) | 186 (44.6) | 9924 (47.0) | 0.049 | 173 (34.9) | 8848 (37.1) | 0.046 | 168 (39.3) | 8099 (38.4) | 0.018 | 231 (44.6) | 12,112 (46.6) | 0.040 | |
| Parity (n (%)) | |||||||||||||
| 0 | 82 (19.7) | 4229 (20.1) | 0.145 | 126 (25.5) | 5111 (21.5) | 0.149 | 103 (24.1) | 5015 (23.9) | 0.235 | 113 (21.9) | 5759 (22.2) | 0.117 | |
| 1 | 33 (7.9) | 1570 (7.5) | 38 (7.7) | 1785 (7.5) | 43 (10.0) | 2036 (9.7) | 56 (10.8) | 2251 (8.7) | |||||
| 2 | 95 (22.8) | 4332 (20.6) | 102 (20.6) | 5147 (21.6) | 106 (24.8) | 5187 (24.7) | 127 (24.6) | 6279 (24.2) | |||||
| 3 | 99 (23.7) | 4548 (21.6) | 106 (21.4) | 5387 (22.6) | 115 (26.9) | 4359 (20.8) | 112 (21.7) | 5585 (21.5) | |||||
| 4 | 60 (14.4) | 3069 (14.6) | 73 (14.7) | 3216 (13.5) | 34 (7.9) | 2481 (11.8) | 63 (12.2) | 3338 (12.9) | |||||
| 5+ | 48 (11.5) | 3308 (15.8) | 50 (10.0) | 3164 (13.3) | 27 (6.2) | 1923 (9.1) | 46 (8.9) | 2713 (10.5) | |||||
| Prior Pregnancy (n (%)) | 370 (88.7) | 18,903 (89.8) | 0.034 | 427 (86.3) | 21,373 (89.8) | 0.108 | 376 (87.9) | 18,879 (89.9) | 0.065 | 455 (88.0) | 23,389 (90.2) | 0.071 | |
| Age of First Birth (n (%)) | |||||||||||||
| <20 | 37 (10.8) | 2056 (12.0) | 0.062 | 39 (10.4) | 1654 (8.7) | 0.109 | 37 (10.8) | 2598 (15.6) | 0.157 | 44 (10.4) | 2665 (12.8) | 0.126 | |
| 20–29 | 267 (77.8) | 12,949 (75.4) | 292 (77.9) | 14,647 (76.7) | 247 (72.2) | 11,515 (69.3) | 296 (70.0) | 14,985 (71.8) | |||||
| 30+ | 30 (8.7) | 1725 (10.0) | 33 (8.8) | 2242 (11.7) | 37 (10.8) | 1764 (10.6) | 60 (14.2) | 2424 (11.6) | |||||
| Age of Menarche (n (%)) | |||||||||||||
| 9 or younger | 2 (0.5) | 224 (1.1) | 0.193 | 7 (1.4) | 283 (1.2) | 0.219 | 3 (0.7) | 252 (1.2) | 0.303 | 4 (0.8) | 303 (1.2) | 0.180 | |
| 10–11 | 88 (21.2) | 4026 (19.2) | 117 (23.7) | 4906 (20.7) | 101 (23.6) | 4004 (19.1) | 118 (22.8) | 5053 (19.6) | |||||
| 12–13 | 248 (59.6) | 11,697 (55.6) | 292 (59.2) | 13,096 (55.1) | 256 (59.8) | 11,535 (54.9) | 301 (58.1) | 14,365 (55.5) | |||||
| 14–16 | 74 (17.8) | 4895 (23.3) | 76 (15.4) | 5264 (22.1) | 66 (15.4) | 5007 (23.8) | 92 (17.8) | 5844 (22.6) | |||||
| 17+ | 4 (1.0) | 183 (0.9) | 2 (0.4) | 242 (1.0) | 2 (0.5) | 188 (0.9) | 3 (0.6) | 304 (1.2) | |||||
| History of Diabetes (n (%)) | 18 (4.3) | 1012 (4.8) | 0.023 | 39 (7.9) | 1174 (4.9) | 0.120 | 22 (5.2) | 1153 (5.5) | 0.014 | 24 (4.6) | 1290 (5.0) | 0.015 | |
| Hypertension (n (%)) | |||||||||||||
| Never | 272 (68.7) | 13,991 (69.7) | 0.022 | 277 (59.4) | 15,447 (68.8) | 0.205 | 292 (71.9) | 14,064 (70.3) | 0.047 | 352 (73.3) | 17,596 (71.2) | 0.048 | |
| Untreated | 29 (7.3) | 1439 (7.2) | 40 (8.6) | 1787 (8.0) | 25 (6.2) | 1444 (7.2) | 37 (7.7) | 1994 (8.1) | |||||
| On Medications | 95 (24.0) | 4640 (23.1) | 149 (32.0) | 5226 (23.3) | 89 (21.9) | 4489 (22.4) | 91 (19.0) | 5107 (20.7) | |||||
| High School Degree (n (%)) | 408 (98.1) | 20,170 (96.4) | 0.105 | 471 (96.1) | 22,693 (95.7) | 0.020 | 413 (96.7) | 19,642 (94.0) | 0.129 | 504 (97.7) | 24,670 (95.4) | 0.123 | |
| Weekly Exercise (median mins (IQR)) | 8.50 (2.50, 18.62) | 9.00 (2.50, 18.17) | 0.001 | 8.25 (1.88, 17.23) | 8.50 (2.50, 18.25) | 0.024 | 9.75 (2.56, 18.08) | 8.25 (1.92, 17.67) | 0.053 | 9.50 (3.50, 19.50) | 10.50 (3.50, 21.00) | 0.060 | |
| Healthy Eating Index 2015 (mean (SD)) | 64.28 (10.62) | 65.42 (10.31) | 0.109 | 63.70 (10.72) | 64.56 (10.80) | 0.080 | 65.97 (10.23) | 65.10 (10.56) | 0.084 | 66.93 (9.82) | 66.34 (10.03) | 0.060 | |
| Hormone Therapy Trial Arm | |||||||||||||
| E+P Control | 37 (8.9) | 2013 (9.5) | 0.144 | 30 (6.0) | 1834 (7.7) | 0.154 | 38 (8.9) | 1909 (9.1) | 0.186 | 34 (6.6) | 2206 (8.5) | 0.123 | |
| E+P Intervention | 26 (6.2) | 2119 (10.0) | 24 (4.8) | 1946 (8.2) | 21 (4.9) | 2042 (9.7) | 33 (6.4) | 2292 (8.8) | |||||
| Observational Study | 354 (84.9) | 16,961 (80.4) | 442 (89.1) | 20,074 (84.2) | 369 (86.2) | 17,141 (81.3) | 451 (87.1) | 21,503 (82.7) | |||||
| Dietary Modification (DM) Trial Arm | |||||||||||||
| DM Control | 78 (18.7) | 3445 (16.3) | 0.099 | 105 (21.2) | 4337 (18.2) | 0.132 | 81 (18.9) | 3680 (17.4) | 0.074 | 111 (21.4) | 4681 (18.0) | 0.111 | |
| DM Intervention | 54 (12.9) | 2289 (10.9) | 77 (15.5) | 2947 (12.4) | 58 (13.6) | 2474 (11.7) | 71 (13.7) | 3111 (12.0) | |||||
| Not Randomized to DM | 285 (68.3) | 15,359 (72.8) | 314 (63.3) | 16,570 (69.5) | 289 (67.5) | 14,938 (70.8) | 336 (64.9) | 18,209 (70.0) | |||||
| *Participants could select more than one race; EC: endometrial cancer; ASD: absolute standardized difference, a measure of the difference between groups in units of standard deviations that can be interpreted using Cohen’s guidelines (d: 0.2 = small difference; 0.5 = moderate difference; 0.8 = large difference; d < 0.2 = trivial difference); BMI: body mass index; IQR: interquartile range (25th, 75th percentile); SD: standard deviation; E+P: estrogen-progestin. |
Among women who were diagnosed with endometrial cancer for whom cancer history details were known (Table 3), endometrioid adenocarcinoma was the most common histology, followed by serous/papillary serous cancers in all geographic regions and the proportion of endometrioid adenocarcinoma cases was higher in White women than in Black women in all geographic regions (Midwest: 81% versus 63%, Northeast: 80% versus 65%, South: 79% versus 75%, West: 81% versus 73%). In contrast, the proportion of serous/papillary serous cancers was lower in White compared with Black women in the Midwest (8% versus 21%) and the Northeast (9% versus 15%) but higher in the South (9% versus 6%) and the West (10% versus 9%) (Fig. 2).
| Midwest | Northeast | South | West | ASD | ||
| n | 412 | 490 | 424 | 511 | ||
| Age (mean (SD)) | 72.16 (7.54) | 72.28 (7.70) | 71.81 (7.83) | 73.03 (7.90) | 0.081 | |
| Race (n (%))* | ||||||
| White | 390 (95.1) | 457 (94.0) | 380 (90.9) | 474 (94.6) | 0.087 | |
| Black | 19 (4.6) | 22 (4.5) | 36 (8.6) | 11 (2.2) | 0.147 | |
| American Indian | 1 (0.2) | 2 (0.4) | 2 (0.5) | 1 (0.2) | 0.029 | |
| Asian | 1 (0.2) | 6 (1.2) | 1 (0.2) | 16 (3.2) | 0.137 | |
| Hawaiian Pacific Islander | 0 (0.0) | 1 (0.2) | 0 (0.0) | 1 (0.2) | 0.043 | |
| Other | 2 (0.5) | 1 (0.2) | 0 (0.0) | 1 (0.2) | 0.054 | |
| Unknown | 0 (0.0) | 1 (0.2) | 3 (0.7) | 7 (1.4) | 0.104 | |
| Race Categories (n (%)) | ||||||
| White | 389 (94.4) | 458 (93.5) | 381 (89.9) | 473 (92.6) | 0.212 | |
| Black | 19 (4.6) | 20 (4.1) | 37 (8.7) | 12 (2.3) | ||
| All Others | 4 (1.0) | 12 (2.4) | 6 (1.4) | 26 (5.1) | ||
| Hispanic (n (%)) | 3 (0.7) | 12 (2.5) | 16 (3.8) | 13 (2.6) | 0.107 | |
| BMI at Baseline (n (%)) | ||||||
| <24.9 | 109 (26.7) | 120 (24.6) | 139 (33.1) | 174 (34.3) | 0.179 | |
| 25.0–29.9 | 104 (25.5) | 141 (29.0) | 115 (27.4) | 156 (30.8) | ||
| ≥30.0 | 195 (47.8) | 226 (46.4) | 166 (39.5) | 177 (34.9) | ||
| Smoking Status at Baseline (n (%)) | ||||||
| Current | 19 (4.6) | 24 (4.9) | 19 (4.5) | 21 (4.2) | 0.111 | |
| Former | 159 (38.8) | 232 (47.4) | 160 (38.1) | 211 (42.1) | ||
| Never | 232 (56.6) | 233 (47.6) | 241 (57.4) | 269 (53.7) | ||
| Stage (n (%)) | ||||||
| In Situ | 4 (1.0) | 5 (1.0) | 3 (0.7) | 5 (1.0) | 0.138 | |
| Localized | 328 (79.6) | 390 (79.6) | 330 (77.8) | 394 (77.1) | ||
| Regional | 66 (16.0) | 59 (12.0) | 60 (14.2) | 80 (15.7) | ||
| Distant | 11 (2.7) | 27 (5.5) | 26 (6.1) | 28 (5.5) | ||
| Unknown | 3 (0.7) | 9 (1.8) | 5 (1.2) | 4 (0.8) | ||
| Grade (n (%)) | ||||||
| Anaplastic | 76 (18.4) | 66 (13.5) | 66 (15.6) | 83 (16.2) | 0.145 | |
| Moderately differentiated | 156 (37.9) | 201 (41.0) | 155 (36.6) | 182 (35.6) | ||
| Poorly differentiated | 85 (20.6) | 112 (22.9) | 94 (22.2) | 104 (20.4) | ||
| Unknown/Not done | 29 (7.0) | 28 (5.7) | 27 (6.4) | 25 (4.9) | ||
| Well differentiated | 66 (16.0) | 83 (16.9) | 82 (19.3) | 117 (22.9) | ||
| Histology (n (%)) | ||||||
| Carcinosarcoma | 9 (2.2) | 9 (1.8) | 11 (2.6) | 9 (1.8) | 0.136 | |
| Clear Cell | 5 (1.2) | 11 (2.2) | 14 (3.3) | 10 (2.0) | ||
| Endometrioid adenocarcinoma | 329 (79.9) | 389 (79.4) | 334 (78.8) | 410 (80.2) | ||
| Mixed malignant tumor | 15 (3.6) | 9 (1.8) | 12 (2.8) | 12 (2.3) | ||
| Mucinous | 9 (2.2) | 12 (2.4) | 11 (2.6) | 10 (2.0) | ||
| Serous/Papillary serous | 37 (9.0) | 47 (9.6) | 37 (8.7) | 53 (10.4) | ||
| Other rare histologies | 8 (1.9) | 13 (2.7) | 5 (1.2) | 7 (1.4) | ||
| High School Degree (n (%)) | 404 (98.3) | 464 (96.1) | 409 (96.7) | 499 (98.0) | 0.082 | |
| Weekly Exercise at Baseline (median mins (IQR)) | 8.75 (3.00, 18.71) | 8.29 (1.88, 17.23) | 9.50 (2.50, 18.08) | 9.75 (3.50, 20.17) | 0.045 | |
| Healthy Eating Index 2015 at Baseline (mean (SD)) | 64.42 (10.52) | 63.65 (10.76) | 66.12 (10.34) | 66.92 (9.83) | 0.186 | |
| Hormone Therapy Trial Arm | ||||||
| E+P Control | 37 (9.0) | 30 (6.2) | 38 (9.0) | 34 (6.7) | 0.093 | |
| E+P Intervention | 26 (6.3) | 24 (4.9) | 21 (5.0) | 33 (6.5) | ||
| Observational Study | 347 (84.6) | 433 (88.9) | 362 (86.0) | 441 (86.8) | ||
| Dietary Modification (DM) Trial Arm | ||||||
| DM Control | 75 (18.3) | 102 (20.9) | 80 (19.0) | 108 (21.3) | 0.074 | |
| DM Intervention | 53 (12.9) | 77 (15.8) | 56 (13.3) | 71 (14.0) | ||
| Not Randomized to DM | 282 (68.8) | 308 (63.2) | 285 (67.7) | 329 (64.8) | ||
| *Participants could select more than one race; ASD: absolute standardized difference, a measure of the difference between groups in units of standard deviations that can be interpreted using Cohen’s guidelines (d: 0.2 = small difference; 0.5 = moderate difference; 0.8 = large difference; d < 0.2 = trivial difference); BMI: body mass index; IQR: interquartile range (25th, 75th percentile); SD: standard deviation; E+P: estrogen-progestin. |
Fig. 2.Endometrial cancer histology distributions across United States census geographic regions.
Overall survival among women with endometrial cancer is displayed by region in a Kaplan-Meier plot (Fig. 3). Compared to endometrial cancer incidence in the Midwest, the hazard ratio for endometrial cancer incidence was 1.01 (95% CI 0.71, 1.26) in the Northeast, 1.16 (95% CI 0.92, 1.45) in the South, and 0.91 (95% CI 0.73, 1.14) in the West in a multivariable adjusted model (Table 4). The covariates included in this model were age of endometrial cancer diagnosis, geographic region, race category (White, Black, All Others), Hispanic ethnicity (regardless of race), receipt of high school degree, BMI, physical activity, diet, endometrial cancer stage and endometrial cancer grade. We found no evidence that region is associated with overall survival after endometrial cancer (p = 0.187, calculated using log-rank test comparing Kaplan-Meier survival curves). In a secondary analysis, we did not find any evidence that race category (White, Black, All Others) modified the association between region and overall survival (p = 0.857).
Fig. 3.Kaplan-Meier overall survival curve for endometrial cancer among United States census geographic regions. WHI: women’s health initiative.
| Midwest | Northeast HR (95% CI) | South HR (95% CI) | West HR (95% CI) |
| 1.00 (ref) | 1.01 (0.71, 1.26) | 1.16 (0.92, 1.45) | 0.91 (0.73, 1.14) |
| HR: hazard ratios; CI: confidence intervals. |
We fit a multivariable logistic regression model with serous/papillary serous cancer type as the outcome (vs. endometrioid adenocarcinoma), region as the primary independent variable, and the same covariates as above to determine whether the prevalence of endometrioid adenocarcinoma versus serous/papillary serous cancers differed by geographic region. We only included women with endometrioid adenocarcinoma or serous/papillary serous cancers and excluded all other women (n = 1625 included in the analysis). As compared to participants in the Midwest, the odds of being diagnosed with serous/papillary serous cancer was 1.26 (95% CI 0.71, 2.23) for the Northeast, 0.83 (95% CI 0.45, 1.51) for the South, and 1.17 (95% CI 0.67, 2.05) for the West (Table 5). We found no evidence of an association between region and being diagnosed with serous/papillary serous cancer (p = 0.486).
| Midwest OR (95% CI) | Northeast OR (95% CI) | South OR (95% CI) | West OR (95% CI) | |
| Endometrioid Adenocarcinoma | 1.00 (ref) | -- | -- | -- |
| Serous/papillary serous carcinoma | -- | 1.26 (0.71, 2.23) | 0.83 (0.45, 1.51) | 1.17 (0.67, 2.05) |
| OR: odds ratios; CI: confidence intervals. |
4. Discussion
Leveraging the Women’s Health Initiative, this is the largest study to date investigating regional variation in endometrial cancer risk factors, histologic subtypes and mortality in the United States. We observed significant differences in BMI between those with and without endometrial cancer, both in the overall analytic cohort and within each geographic region. While modest differences in tumor grade and stage were noted—such as slightly higher rates of moderately or poorly differentiated cancers and regional/distant disease in the Northeast—overall survival did not differ significantly by geographic regions. These findings suggest that despite geographic variation in risk factors, the clinicopathologic features and outcomes of endometrial cancer may be relatively consistent across the United States.
Due to the number of statistical comparisons made in this study, there is an inherent risk of Type I error. While we identified several significant associations, these results should be considered exploratory and hypothesis-generating rather than definitive. Replication in independent cohorts will be essential to validate these findings. We utilized standard epidemiologic approaches, including Cox proportional hazards models and Kaplan-Meier survival analyses, which were appropriate for our study objectives and the structure of the WHI dataset. These methods allowed us to assess associations between demographic, clinical and lifestyle factors and endometrial cancer outcomes in a transparent and interpretable manner. However, we acknowledge that more advanced statistical or machine learning methods—such as Bayesian modeling, classification trees, or neural networks—may offer additional insight into complex, nonlinear relationships and prediction of outcomes. Future work may benefit from incorporating such techniques, particularly when leveraging larger and more diverse datasets.
Our findings related to obesity are particularly notable. Obesity and high BMI are associated with elevated estrogen levels that can result in earlier menarche, menstrual irregularities during adolescence and adulthood, polycystic ovary syndrome, which can contribute to menstrual irregularities and abnormal uterine bleeding patterns, and suboptimal hormonal contraceptive efficacy when used to regulate uterine bleeding [14]. Obesity results in elevated levels of circulating endogenous estrogens arising from the conversion of adrenal androgens to estrogens, primarily androstenedione to estrone, by aromatase in the adipose tissue [15, 16]. In our WHI analytic cohort, the highest rates of obesity were observed in the Northeast and Midwest, which differs from the Center for Disease Control’s (CDC’s) Adult Obesity Prevalence Maps that consistently show the highest obesity rates in the South and Midwest [5].
Several factors may explain this discrepancy. While CDC estimates include both men and women, our WHI cohort consists exclusively of postmenopausal women with a uterus, which inherently excludes women who had previously undergone hysterectomy—a procedure more common among women with obesity and abnormal uterine bleeding, particularly in the South. The exclusion of this population may have disproportionately reduced obesity rates in Southern participants. Additionally, prior research has shown that Black women are more likely to undergo hysterectomy at younger ages, potentially contributing to lower enrollment of Black women in WHI and regional differences in racial distribution. Recruitment-related biases in WHI may also play a role; for example, eligibility criteria were more restrictive for White women (age 50–59 only) compared to minority participants, which may have skewed the racial and geographic composition of the cohort. Finally, WHI participants overall tend to have higher levels of education and income than the general U.S. population, factors associated with lower obesity rates. Regardless of these differences, our findings demonstrate that overall survival after endometrial cancer did not vary significantly by geographic region within the WHI analytic cohort, and this association was not modified by race. Future work could compare obesity rates in the WHI to those in other large, nationally representative cohorts such as the National Health and Nutrition Examination Survey (NHANES) or the Behavioral Risk Factor Surveillance System (BRFSS) to further contextualize these patterns and assess external validity.
An important consideration when interpreting our findings in the context of the diverse United States population is that the WHI analytic cohort may not accurately reflect the national demographic landscape, particularly with respect to regional and racial/ethnic differences. The racial distribution within the WHI cohort was imbalanced, with non-Hispanic White women comprising the majority of participants. As a result, analyses involving Black, Hispanic, Asian and other racial/ethnic subgroups were limited by smaller sample sizes, reducing statistical power to detect meaningful differences. While we included stratified analyses to explore potential disparities, these results should be interpreted with caution. This limitation may attenuate or obscure disparities that are more prominent in the general population. Additionally, the WHI’s recruitment strategy and inclusion criteria, such as the exclusion of women with prior hysterectomy, may have contributed to regional obesity prevalence patterns that differ from national trends, such as lower observed obesity rates in the South compared to CDC data [5]. Future studies should prioritize recruitment of more racially and ethnically diverse populations to better characterize risk factors, tumor characteristics and outcomes across all groups. National initiatives and consortia that integrate community-based research and population-level biobanks will be critical to addressing these gaps and advancing equitable cancer prevention strategies. While sensitivity analyses by race and region were limited by sample size in this cohort, we recognize this as a key direction for future work.
While there may be inherent differences between our study population and the broader United States population, there are important takeaways from our findings. Although disparities in risk factors and sociodemographic characteristics exist across the nation, the clinicopathologic features of endometrial cancer remain remarkably similar throughout the U.S. This consistency in the clinicopathologic aspects of endometrial cancer underscores the potential for a uniform and nation-wide approach to tackling this disease. However, it is increasingly recognized that endometrial cancer has a multifactorial etiology influenced not only by demographic and lifestyle factors, but also by genetic, molecular and environmental exposures. Our study did not include data on tumor genomics and environmental factors such as endocrine-disrupting chemicals or pollution, which may contribute to observed disparities. Future interdisciplinary research that integrates molecular profiling, germline and somatic genetic data, and environmental exposure assessment will be essential to better understand the biologic underpinnings of disease and to tailor prevention and treatment strategies more effectively across diverse populations.
Independent of race or geographic region, obesity is a key contributor to endometrial cancer. Reeves et al. [11] found that obesity increased endometrial cancer risk in WHI, independent of other factors but was not associated with stage or grade of disease. Our data extend these findings from 7–8 years of follow-up to a much longer mean follow-up and focus on geographic census region. The homogeneity in tumor characteristics and survival across regions in our cohort suggests a potential opportunity for unified national strategies. By targeting modifiable risk factors like obesity and improving access to timely diagnosis and treatment—particularly among high-risk and underserved populations—public health efforts may reduce the overall burden of endometrial cancer. These findings underscore the need for prevention and screening programs tailored to individuals with obesity, regardless of geographic location, and may inform equitable policy development on a national scale.
This study has several notable strengths. We leveraged a large and geographically diverse cohort of postmenopausal women from the Women’s Health Initiative (WHI), which provided robust statistical power to examine regional patterns in endometrial cancer risk, histologic subtypes and survival. The WHI’s rigorous data collection methods, long-term follow-up and comprehensive set of demographics, clinical and lifestyle covariates allowed for nuanced multivariable analyses across multiple subgroups. These design strengths enhance the internal validity and relevance of our findings.
However, several limitations should be considered when interpreting our results. First, while the WHI cohort is regionally diverse, it is not fully representative of the broader U.S. population. The majority of participants were non-Hispanic White women, and the cohort overall had higher educational attainment and income levels compared to national averages. As a result, key risk factors—such as obesity prevalence and access to care—may differ from those in more socioeconomically or racially diverse populations, limiting generalizability and potentially attenuating disparities that exist in the broader population.
Second, the study relied on self-reported data for several baseline variables, including hormone therapy use, smoking, physical activity and dietary intake. Although these are susceptible to recall bias, the WHI used validated instruments and standardized protocols for data collection, and prior studies have demonstrated acceptable validity and reproducibility of these measures.
Third, subgroup analyses involving less common histologic subtypes—such as serous and clear cell carcinomas—were limited by small sample sizes, reducing statistical power and our ability to detect meaningful differences. These aggressive subtypes are associated with worse outcomes and disproportionately affect certain racial and ethnic groups, but their rarity presents a challenge in observational cohorts not specifically designed to capture them. Future studies should consider targeted recruitment or pooled multi-institutional data to enable adequately powered analyses in these high-risk subgroups.
Fourth, our analysis focused on primary associations and included limited evaluation of complex interactions. Although we explored selected effect modification (e.g., BMI by race), we were not powered to assess higher-order interactions (e.g., BMI × physical activity or BMI × socioeconomic status). These multifactorial relationships may meaningfully influence risk and outcomes. Future work in larger and more heterogeneous cohorts may benefit from the use of advanced modeling techniques—including Bayesian approaches, interaction-based regression frameworks and machine learning—to better capture the interplay of behavioral, biological and social determinants.
Fifth, the WHI dataset did not include tumor molecular characteristics or environmental exposure data, both of which are increasingly recognized as important contributors to endometrial cancer risk and disparities. The integration of molecular profiling, genetic data and environmental exposures (e.g., endocrine-disrupting chemicals, pollution) in future studies will be critical to understanding the multifactorial etiology of endometrial cancer and informing precision prevention strategies.
Finally, although we included basic socioeconomic status indicators such as education and income, the WHI lacked more granular measures—such as insurance status, healthcare access, neighborhood-level deprivation and transportation barriers—that likely influence racial and geographic disparities. The absence of these data may have introduced residual confounding and limited our ability to fully characterize the social determinants of risk and outcomes. Future research should incorporate comprehensive SES metrics, including geocoded neighborhood disadvantage indices, longitudinal insurance and healthcare utilization data, and contextual social environment measures. Such integration is essential to addressing the structural inequities that underlie persistent disparities in endometrial cancer incidence and mortality.
5. Conclusions
This study provides a comprehensive evaluation of demographic, lifestyle and tumor characteristics associated with endometrial cancer in a large, geographically diverse cohort of postmenopausal women enrolled in the Women’s Health Initiative. Although we observed regional and racial variation in baseline factors such as BMI, as well as in histologic subtype and stage at diagnosis, these differences did not translate into statistically significant disparities in overall survival across U.S. Census regions. These findings offer important epidemiologic insights that can inform future prevention, screening and treatment strategies at a national level.
Given the observational nature of the study, these associations should be interpreted with caution and not as evidence of causality. Despite careful adjustment for confounding variables, the potential for residual confounding remains, and our results are best viewed as hypothesis-generating. The consistency of clinicopathologic features across regions reinforces the opportunity for standardized national approaches, while the disparities observed among racial subgroups highlight the persistent need for equitable access to early diagnosis and high-quality care. To move the field forward, future research must incorporate molecular, genetic, environmental, and granular socioeconomic data to better capture the multifactorial drivers of endometrial cancer risk and outcomes. Addressing these complex and intersecting determinants is essential to reduce disparities, guide personalized prevention strategies, and advance health equity in endometrial cancer care.
Availability of data and materials
The data are available from the Women’s Health Initiative. Restrictions apply to the availability of these data, which were used under license for this study. Data are available from the Women’s Health Initiative Coordinating Center with the permission of the WHI.
Author contributions
SS—conceptualization; methodology; formal analysis; writing—original draft. HH—data curation. HH, AWK, CK, OD—writing—review & editing. MLS—conceptualization; methodology; supervision; project administration.
Ethics approval and consent to participate
Written informed consent was obtained from all study participants. The Women’s Health Initiative (WHI) study protocols were approved by the Institutional Review Board of the Fred Hutchinson Cancer Research Center (Seattle, WA) and by the Institutional Review Boards of each participating institution at baseline (1993–1998) and in 2005, 2010 and 2015, as described above. The WHI study reference number at the Fred Hutchinson Cancer Research Center is IRB# 4451.
Acknowledgment
We thank the participants of the Women’s Health Initiative for their time and commitment to advancing women’s health. We also acknowledge the WHI investigators and staff for their contributions to data collection and management.
Funding
The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health, and Human Services through 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, 75N92021D00005. This manuscript is partially supported by the Biostatistics Shared Resource (BSR) of the NIH-funded Stanford Cancer Institute: P30CA124435.
Conflict of interest
The authors declare no competing financial interests or potential conflicts of interest.
