Abstract
Background
The causes of racial disparities in epithelial ovarian cancer (EOC) incidence remain unclear. Differences in the prevalence of ovarian cancer risk factors may explain disparities in EOC incidence among African American (AA) and White women.
Methods
We used data from 4 case-control studies and 3 case-control studies nested within prospective cohorts in the Ovarian Cancer in Women of African Ancestry Consortium to estimate race-specific associations of 10 known or suspected EOC risk factors using logistic regression. Using the Bruzzi method, race-specific population attributable risks (PAR) were estimated for each risk factor individually and collectively, including groupings of exposures (reproductive factors and modifiable factors). All statistical tests were 2-sided.
Results
Among 3244 White EOC cases and 9638 controls and 1052 AA EOC cases and 2410 controls, AA women had a statistically significantly higher PAR (false discovery rate [FDR] P < .001) for first-degree family history of breast cancer (PAR = 10.1%, 95% confidence interval [CI] = 6.5% to 13.7%) compared with White women (PAR = 2.6%, 95% CI = 0.8% to 4.4%). After multiple test correction, AA women had a higher PAR than White women when evaluating all risk factors collectively (PAR = 61.6%, 95% CI = 48.6% to 71.3% vs PAR = 43.0%, 95% CI = 32.8% to 51.4%, respectively; FDR P = .06) and for modifiable exposures, including body mass index, oral contraceptives, aspirin, and body powder (PAR = 36.0%, 95% CI = 21.0% to 48.8% vs PAR = 13.8%, 95% CI = 4.5% to 21.8%, respectively; FDR P = .04).
Conclusions
Collectively, the selected risk factors accounted for slightly more of the risk among AA than White women, and interventions to reduce EOC incidence that are focused on multiple modifiable risk factors may be slightly more beneficial to AA women than White women at risk for EOC.
Ovarian cancer is the fifth leading cause of cancer mortality among women in the United States, accounting for 5% of cancer deaths among women (1). Ovarian cancer is a heterogeneous disease composed of histotypes with unique pathologic, epidemiologic, molecular, and clinical features (2-5). The majority (90%) of ovarian tumors are epithelial and, of those, serous carcinoma is the most commonly diagnosed histotype (approximately 60%-70%) (6,7).
Although the incidence of epithelial ovarian cancer (EOC) is approximately 30% higher among White women than African American (AA) women (7,8), AA women have lower 5-year survival than White women (39% vs 48%, respectively) (1). Racial disparities in EOC likely involve a combination of interrelated factors, but few studies have been powered to examine racial differences in EOC risk. The studies with the largest number of AA cases observed racial differences in the prevalence of EOC risk factors (9,10), higher attributable fractions for some risk factors among AA women (11), and stronger associations for certain risk factors among AA women (10,11). Recent evidence also suggests that some risk factors are histotype specific (3,12-17). Thus, elucidating the causes of disparities for EOC overall and by histotype is essential for improving our understanding of EOC heterogeneity. Previous studies’ ability to examine risk by histotype was limited by sample size: each had less than 1000 AA EOC cases. Here, we used data from a large collaborative study of EOC in AA and White women, the Ovarian Cancer in Women of African Ancestry (OCWAA) Consortium, to examine whether racial differences in the incidence of EOC may be attributable, at least in part, to differences in the prevalence of known and suspected EOC risk factors, alone and in combination.
Methods
Study Population
The OCWAA Consortium has been described in detail elsewhere (18). In brief, questionnaire, medical record, and tumor or cancer registry data were harmonized for participants in 4 case-control studies: the African-American Cancer Epidemiology Study (AACES) (19), the Cook County Case-Control Study (CCCCS) (20,21), the Los Angeles County Ovarian Cancer Study (LACOCS) (22), and the North Carolina Ovarian Cancer Study (NCOCS) (23); and 4 case-control studies nested within prospective cohorts: the Black Women’s Health Study (BWHS) (24), the Multiethnic Cohort Study (MEC) (25), the Southern Community Cohort Study (26), and the Women’s Health Initiative (WHI) (27). WHI participants from both the clinical trial and the observational study were included. Researchers from each study obtained informed consent from participants, and the individual studies and the OCWAA Consortium were approved by the relevant institutional review boards. The present analyses include 1052 AA cases, 2410 AA controls, 3244 White cases, and 9638 White controls from 7 of the 8 contributing studies: AACES, BWHS, CCCCS, LACOCS, MEC, NCOCS, and WHI (Supplementary Table 1, available online).
Outcome
Eligible cases were diagnosed with EOC; tumor histotype was classified using a schema (6,18) based on the 2014 World Health Organization Classification of Tumors of Female Reproductive Organs (28) that combined morphology and grade into the following histotypes: high-grade serous, low-grade serous, endometrioid, clear cell, mucinous, carcinosarcoma, and other epithelial tumors.
Exposures
The exposures of interest included 10 known or suspected risk factors for EOC: recent body mass index (BMI) using self-reported height and weight from 1 to 5 years before diagnosis for cases or interview for controls (<30 kg/m2, ≥30 kg/m2), use of oral contraceptives (never, ever), parity (no full-term pregnancies, ≥1), history of tubal ligation (no, yes), first-degree family history of ovarian cancer (no, yes), first-degree family history of breast cancer (no, yes), use of aspirin (never, ever), use of body powder applied to genital areas (never, ever), education (<completed college, ≥college graduate), and a history of endometriosis (yes, no). Some sites did not inquire about certain exposures and are missing data for those variables: aspirin use in CCCCS, endometriosis in MEC and WHI, and body powder in BWHS, MEC, and WHI (clinical trial). Also, for body powder, participants diagnosed after 2013 were excluded to circumvent potential reporting bias due to talcum powder and ovarian cancer lawsuits (29).
Statistical Analyses
Detailed methods are provided in the Supplementary Methods (available online). Logistic regression was used to estimate pooled odds ratios (ORs) and 95% confidence intervals (CIs) for the association between each exposure and EOC risk by race. If site heterogeneity was detected, pooled odds ratios were calculated using multi-level regression (30-32). Models included the known and suspected risk factors and were additionally adjusted for study site, reference year, age at diagnosis or interview (continuous, years), age at menarche (continuous, years), menopausal status (premenopausal, postmenopausal), and history of hysterectomy 1 year before diagnosis or interview (no, yes).
Population attributable risk (PAR) for each exposure individually and collectively by race was estimated using the Bruzzi method (33-35), with confidence intervals calculated by a sample-based nonparametric bootstrap (36). PARs were also calculated for exposure groupings: reproductive factors (use of oral contraceptives, parity, and tubal ligation) and modifiable factors (BMI, use of oral contraceptives, use of aspirin, and use of body powder applied to genital areas). Because all risk associations were modeled in the positive direction for the PAR calculations, only positive values are plausible, and therefore PARs were truncated at 0.01%.
The WHI potentially constitutes a population of healthier women compared with the other OCWAA studies. Women with a competing risk or a health condition that could affect adherence or retention were ineligible for WHI hormone trials, and a variety of additional criteria (eg, myocardial infarction or stroke in the past 6 months, severe hypertension) excluded women from participating (27). Moreover, most OCWAA studies enrolled both pre- and postmenopausal women, unlike WHI, which included only postmenopausal women. Due to these potential differences between WHI and the other OCWAA studies, we repeated the main analyses excluding WHI.
Race-specific odds ratios and PARs were stratified by histotype and menopausal status to assess heterogeneity by these variables. For these analyses, high-grade serous carcinoma (HGSC) was compared with low-grade serous, endometrioid, clear cell, and mucinous carcinoma grouped together as non-HGSC due to the rarity of these histotypes individually. Carcinosarcoma and other epithelial tumors were not included as non-HGSC because these are rare tumors with fairly unknown etiology. As a sensitivity analysis, we repeated the analyses including carcinosarcoma and other epithelial tumors as non-HGSC.
P values were calculated from empirical bootstrap P values for 2-tailed tests of equality in PARs (37) and 2-tailed t tests for equality of odds ratios. Given the number of tests conducted in this analysis, we controlled the false discovery rate (FDR) within each set of a priori hypotheses (by table) using the adaptive Benjamini-Hochberg procedure (38). A cut point of .05 for statistical significance was used, which in the FDR interpretation allows for a 5% or less chance that statistically significant tests will be false positives. SAS version 9.3 was used for data management and analysis.
Imputation
Data were systematically missing by site for certain variables (eg, endometriosis, aspirin use) or sporadically missing due to participant nonresponse. Multilevel multiple imputation (39-44) using Blimp software V2.1 (45-47) was conducted separately on the 4 race or menopausal status combinations (see the Supplementary Methods, available online for further details). Primary findings include the imputed values for the systematically missing data by site but not the sporadically missing data. As a sensitivity analysis, the primary findings were compared with results from the complete case analysis (restricting to participants with data on all variables) and multiple imputation (including imputed values for all missing data).
Results
The prevalence of most exposures among the controls differed by race; AA women were more likely than White women to have a BMI of at least 30 kg/m2, to ever use oral contraceptives, to have a history of tubal ligation, and to use body powder applied to genital areas but less likely to ever use aspirin and to graduate college (Table 1). Statistically significant differences in the risk association for first-degree family history of breast cancer were noted by race (ORAA = 1.79, 95% CI = 1.45 to 2.22 and ORWhite = 1.18, 95% CI = 1.05 to 1.33; FDR P = .01). Likewise, although not statistically significant, the risk estimates for recent BMI of at least 30 kg/m2, never use of oral contraceptives, first-degree family history of ovarian cancer, ever use of body powder applied to genital areas, and endometriosis were slightly stronger in magnitude for AA women than White women.
Table 1.
Estimated odds ratiosa and 95% confidence intervals for the association between each exposure and ovarian cancer risk by race
| Exposure | AA |
White |
Raw P | FDR Pb | ||||
|---|---|---|---|---|---|---|---|---|
| Cases (n = 1052) | Controls (n = 2410) | OR (95% CI) | Cases (n = 3244) | Controls (n = 9638) | OR (95% CI) | |||
| No. (%) | No. (%) | No. (%) | No. (%) | |||||
| Recent BMIc | ||||||||
| <30 kg/m2 | 477 (45.8) | 1269 (53.3) | 1.00 (Referent) | 2455 (77.1) | 7139 (75.4) | 1.00 (Referent) | .23 | .76 |
| ≥30 kg/m2 | 564 (54.2) | 1111 (46.7) | 1.20 (1.02 to 1.41) | 728 (22.9) | 2327 (24.6) | 1.00 (0.77 to 1.31) | ||
| Oral contraceptive use | ||||||||
| Ever | 642 (61.0) | 1461 (60.6) | 1.00 (Referent) | 1670 (51.5) | 4508 (46.8) | 1.00 (Referent) | .06 | .30 |
| Never | 410 (39.0) | 949 (39.4) | 1.41 (1.18 to 1.70) | 1574 (48.5) | 5130 (53.2) | 1.04 (0.85 to 1.27) | ||
| Full-term pregnancies | ||||||||
| ≥1 pregnancy | 852 (81.0) | 2034 (84.4) | 1.00 (Referent) | 2489 (76.7) | 7882 (81.8) | 1.00 (Referent) | .81 | .90 |
| None | 200 (19.0) | 376 (15.6) | 1.41 (1.13 to 1.77) | 755 (23.3) | 1756 (18.2) | 1.49 (0.89 to 2.50) | ||
| Tubal ligation | ||||||||
| Yes | 315 (29.9) | 752 (31.2) | 1.00 (Referent) | 476 (14.7) | 1720 (17.8) | 1.00 (Referent) | .97 | .97 |
| No | 737 (70.1) | 1658 (68.8) | 1.28 (1.06 to 1.54) | 2768 (85.3) | 7918 (82.2) | 1.29 (1.14 to 1.46) | ||
| First-degree family history of breast cancer | ||||||||
| No | 756 (77.1) | 2026 (87.9) | 1.00 (Referent) | 2628 (83.4) | 7881 (85.7) | 1.00 (Referent) | .001 | .01 |
| Yes | 225 (22.9) | 280 (12.1) | 1.79 (1.45 to 2.22) | 524 (16.6) | 1316 (14.3) | 1.18 (1.05 to 1.33) | ||
| First-degree family history of ovarian cancer | ||||||||
| No | 907 (93.9) | 2221 (97.0) | 1.00 (Referent) | 2992 (95.3) | 8999 (97.5) | 1.00 (Referent) | .70 | .90 |
| Yes | 59 (6.1) | 69 (3.0) | 1.92 (1.30 to 2.84) | 147 (4.7) | 233 (2.5) | 1.76 (1.39 to 2.21) | ||
| Aspirin used | ||||||||
| Ever | 195 (22.4) | 674 (30.7) | 1.00 (Referent) | 763 (32.5) | 3308 (39.7) | 1.00 (Referent) | .79 | .90 |
| Never | 675 (77.6) | 1521 (69.3) | 1.12 (0.91 to 1.39) | 1582 (67.5) | 5026 (60.3) | 1.07 (0.69 to 1.65) | ||
| Body powder applied to genital arease | ||||||||
| Never | 416 (61.4) | 745 (66.5) | 1.00 (Referent) | 1931 (70.5) | 4515 (69.2) | 1.00 (Referent) | .61 | .90 |
| Ever | 262 (38.6) | 376 (33.5) | 1.36 (1.10 to 1.70) | 807 (29.5) | 2007 (30.8) | 1.28 (1.15 to 1.43) | ||
| Education | ||||||||
| ≥College graduate | 349 (33.2) | 870 (36.1) | 1.00 (Referent) | 1634 (50.4) | 4784 (49.6) | 1.00 (Referent) | .75 | .90 |
| High school graduate or partial college | 703 (66.8) | 1540 (63.9) | 1.20 (1.01 to 1.44) | 1610 (49.6) | 4854 (50.4) | 1.16 (0.93 to 1.43) | ||
| Endometriosisf | ||||||||
| Never | 849 (91.4) | 1596 (94.8) | 1.00 (Referent) | 1984 (90.8) | 2878 (94.3) | 1.00 (Referent) | .68 | .90 |
| Ever | 80 (8.6) | 88 (5.2) | 2.09 (1.47 to 2.96) | 200 (9.2) | 173 (5.7) | 1.92 (1.53 to 2.41) | ||
Adjusted for study site, reference year, age, full-term pregnancies, education, oral contraceptive use, age at menarche, imputed first-degree family history of breast cancer, imputed first-degree family history of ovarian cancer, tubal ligation, BMI, menopausal status, hysterectomy 1 year before diagnosis, education, imputed endometriosis, imputed aspirin use, and imputed body powder applied to genital areas. For AAs, study site interactions with age were included. For Whites, study site interactions with BMI, full-term pregnancies, aspirin use, education, and oral contraceptive use were included, and a pooled odds ratio was calculated using a multilevel logistic regression model for the aforementioned variables. AA = African American; BMI = body mass index; BWHS = Black Women’s Health Study; CCCCS = Cook County Case-Control Study; CI = confidence interval; FDR = false discovery rate; MEC = Multiethnic Cohort Study; OR = odds ratio; PAR = population attributable risk; WHI = Women’s Health Initiative.
Bootstrap P values for the 2-tailed test of no differences between race-specific PARs with adjustment for multiple comparisons using the adaptive Benjamini-Hochberg FDR procedure.
Recent BMI is based on height and weight 1 to 5 years before diagnosis (cases) or interview (controls). Women with a BMI less than 18.5 kg/m2 were excluded.
Excludes CCCCS because aspirin use was not asked in the questionnaire.
Excludes BWHS, MEC, and WHI (nonobservational group) because body powder exposure was not asked in the questionnaire. Due to potential reporting bias of the lawsuits, data were restricted to women diagnosed before 2014.
Excludes WHI and MEC because endometriosis was not asked in the questionnaire.
The highest PAR for both AA and White women was for not having a tubal ligation: 15.4% (95% CI = 4.5% to 25.8%) and 18.9% (95% CI = 9.8% to 26.8%), respectively (Table 2). The only PAR with a statistically significant difference by race was first-degree family history of breast cancer (PARAA = 10.1%, 95% CI = 6.5% to 13.7%; PARWhite = 2.6%, 95% CI = 0.8% to 4.4%; FDR P < .001). Collectively, the PAR for all risk factors was suggestively higher (FDR P = .06) among AA women (PAR = 61.6%, 95% CI = 48.6% to 71.3%) than White women (PAR = 43.0%, 95% CI = 32.8% to 51.4%). AA women had a statistically significantly higher PAR for the modifiable exposures compared with White women (PAR = 36.0%, 95% CI = 21.0% to 48.8% vs PAR = 13.8%, 95% CI = 4.5% to 21.8%, respectively; FDR P = .04), and no racial differences were observed for the reproductive exposures. These findings were consistent in sensitivity analyses using both the complete case analysis approach and multiple imputation (Supplementary Table 2, available online). After repeating the analyses excluding WHI (Table 2), the PARs in White women were generally higher than the findings in the entire OCWAA sample, whereas the PARs in AA women were consistent. The collective PAR remained higher among AA women (PAR = 63.5%, 95% CI = 51.6% to 73.0%) than White women (PAR = 52.1%, 95% CI = 40.9% to 60.6%), but the difference was not as marked as the results including WHI.
Table 2.
PAR percenta and 95% confidence intervals for each exposure individually and collectively by race
| Exposure | Entire OCWAA sample |
Raw P | FDR Pb | OCWAA excluding WHI |
Raw P | FDR Pb | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| AA (n = 3462) |
White (n = 12 882) |
AA (n = 3144) |
White (n = 6183) |
|||||||||
| Ca/Co | PAR% (95% CI) | Ca/Co | PAR% (95% CI) | Ca/Co | PAR% (95% CI) | Ca/Co | PAR% (95% CI) | |||||
| Recent BMIc ≥30 kg/m2 | 564/1111 |
9.1 (0.8 to 16.0) |
728/2327 |
0.6 (0.01 to 3.0) |
.03 | .07 | 540/983 |
8.7 (0.2 to 16.0) |
495/717 |
2.9 (0.01 to 5.8) |
.18 | .32 |
| Never use of oral contraceptives | 410/949 |
11.6 (5.8 to 17.6) |
1574/5130 |
2.2 (0.01 to 7.0) |
.02 | .06 | 376/764 |
10.7 (5.2 to 16.6) |
969/1431 |
2.6 (0.01 to 7.8) |
.05 | .14 |
| No full-term pregnancies | 200/376 |
5.5 (1.8 to 9.1) |
755/1756 |
7.6 (5.4 to 9.9) |
.35 | .47 | 186/305 |
5.3 (1.3 to 9.0) |
578/728 |
10.6 (7.8 to 13.3) |
.03 | .11 |
| No tubal ligation | 737/1658 |
15.4 (4.5 to 25.8) |
2768/7918 |
18.9 (9.8 to 26.8) |
.59 | .69 | 706/1447 |
17.0 (5.8 to 27.4) |
1976/3152 |
20.1 (9.4 to 30.1) |
.70 | .81 |
| First-degree family history of breast cancer | 225/280 |
10.1 (6.5 to 13.7) |
524/1316 |
2.6 (0.8 to 4.4) |
<.001 | <.001 | 217/263 |
9.8 (6.0 to 13.3) |
397/502 |
4.7 (2.5 to 6.8) |
.03 | .11 |
| First-degree family history of ovarian cancer | 59/69 |
2.9 (1.2 to 4.7) |
147/233 |
2.0 (1.1 to 2.8) |
.35 | .47 | 58/66 |
3.0 (1.2 to 4.9) |
122/120 |
2.6 (1.5 to 3.7) |
.73 | .81 |
| Never use of aspirind | 675/1521 |
8.4 (0.01 to 23.0) |
1582/5026 |
4.4 (0.01 to 12.1) |
.69 | .69 | 653/1344 |
13.6 (0.01 to 27.8) |
1097/1819 |
11.5 (0.01 to 23.2) |
.84 | .84 |
| Ever use of body powder applied to genital arease | 262/376 |
10.3 (3.1 to 17.1) |
807/2007 |
6.5 (3.3 to 9.3) |
.32 | .47 | 251/306 |
9.7 (2.0 to 16.8) |
561/564 |
6.1 (3.0 to 9.1) |
.34 | .54 |
| High school or partial college education | 703/1540 |
11.3 (0.01 to 21.3) |
1610/4854 |
4.7 (0.01 to 9.8) |
.28 | .47 | 675/1372 |
11.8 (0.1 to 22.0) |
1119/1669 |
7.4 (0.7 to 13.8) |
.52 | .71 |
| Endometriosisf | 80/88 |
4.4 (2.4 to 6.5) |
200/173 |
4.4 (2.9 to 6.0) |
.97 | .97 | 80/88 |
4.4 (2.4 to 6.5) |
200/173 |
4.4 (2.9 to 6.0) |
.97 | .97 |
| All exposures collectivelyg | — |
61.6 (48.6 to 71.3) |
— |
43.0 (32.8 to 51.4) |
.01 | .06 | — |
63.5 (51.6 to 73.0) |
— |
52.1 (40.9 to 60.6) |
.11 | .24 |
| All reproductive exposuresh | — |
28.5 (18.7 to 37.8) |
— |
26.0 (16.8 to 33.6) |
.67 | .69 | — |
29.0 (18.7 to 38.2) |
— |
29.6 (18.7 to 38.3) |
.91 | .91 |
| All modifiable exposuresi | — |
36.0 (21.0 to 48.8) |
— |
13.8 (4.5 to 21.8) |
.006 | .04 | — |
38.3 (23.9 to 51.0) |
— |
18.7 (7.7 to 29.2) |
.02 | .11 |
PAR percent risk models were adjusted for study site, reference year, age, full-term pregnancies, education, oral contraceptive use, age at menarche, imputed first-degree family history of breast cancer, imputed first-degree family history of ovarian cancer, tubal ligation, BMI, menopausal status, hysterectomy 1 year before diagnosis, education, imputed endometriosis, imputed aspirin use, and imputed body powder applied to genital areas. For AAs, study site interactions with age were included. For Whites, study site interactions with BMI, full-term pregnancies, aspirin use, education, and oral contraceptive use were included, and a pooled odds ratio was calculated using a multilevel logistic regression model for the aforementioned variables. AA = African American; BMI = body mass index; Ca = cases; BWHS = Black Women’s Health Study; CCCCS = Cook County Case-Control Study; CI = confidence interval; Co = controls; FDR = false discovery rate; MEC = Multiethnic Cohort Study; OCWAA = Ovarian Cancer in Women of African Ancestry; PAR = population attributable risk; WHI = Women’s Health Initiative.
Bootstrap P values for the 2-tailed test of no differences between race-specific PARs with adjustment for multiple comparisons using the adaptive Benjamini-Hochberg FDR procedure.
Recent BMI is based on height and weight 1 to 5 years before diagnosis (cases) or interview (controls). Women with a BMI less than 18.5 kg/m2 were excluded.
Excludes CCCCS because aspirin use was not asked in the questionnaire.
Excludes BWHS, MEC, and WHI (nonobservational group) because body powder exposure was not asked in the questionnaire. Due to potential reporting bias of the lawsuits, data were restricted to women diagnosed before 2014.
Excludes WHI and MEC because endometriosis was not asked in the questionnaire.
For the collective PAR, each exposure was set to the following reference values to indicate lowest risk: BMI less than 30 kg/m2, use of oral contraceptives, parous women, had a tubal ligation, no first-degree family history of breast or ovarian cancer, aspirin use, no use of body powder applied to genital areas, college education or higher, and no history of endometriosis.
Includes use of oral contraceptives, parity, and tubal ligation.
Includes BMI, use of oral contraceptives, aspirin use, and use of body powder applied to genital areas.
When assessing heterogeneity by histotype (Figure 1; Supplementary Table 3, available online) among 2678 HGSC cases and 900 non-HGSC cases, the PARs were higher for non-HGSC compared with HGSC for BMI, oral contraceptive use, parity, tubal ligation, and endometriosis but lower for body powder applied to genital areas, first-degree family history of breast cancer, and first-degree family history of ovarian cancer. The collective PARs were higher for non-HGSC than HGSC in both races: PARAA = 71.9%, 95% CI = 54.0% to 85.2% vs PARAA = 54.5%, 95% CI = 38.7% to 67.5%; PARWhite = 69.5%, 95% CI = 59.1% to 78.6% vs PARWhite = 35.1%, 95% CI = 21.9% to 46.5%, respectively. Likewise, higher PARs for all reproductive exposures were observed for non-HGSC compared with HGSC. Including carcinosarcoma and other epithelial tumors as non-HGSC (n = 1618) resulted in lower PARs than the more restricted non-HGSC definition (PARAA = 70.4%, 95% CI = 55.9% to 81.3% and PARWhite = 53.7%, 95% CI = 43.0% to 63.7%; Supplementary Table 4, available online), particularly for White women because there was a higher proportion of these tumors in White vs AA women.
Figure 1.
Population attributable risk (PAR) percent and 95% confidence intervals (CI) for each exposure individually by race and histotype. This analysis includes 2678 high-grade serous carcinoma (HGSC) cases (662 African American [AA] and 2016 White women), 900 non-HGSC cases (218 AA and 682 White women), and 12 048 controls (2410 AA and 9638 White women). Low-grade serous, endometrioid, clear cell, and mucinous carcinoma were included as non-HGSC. The PARs were calculated from models adjusted for study site, reference year, age, full-term pregnancies, education, oral contraceptive use, age at menarche, imputed first-degree family history of breast cancer, imputed first-degree family history of ovarian cancer, tubal ligation, body mass index (BMI), menopausal status, hysterectomy 1 year before diagnosis, education, imputed endometriosis, imputed aspirin use, and imputed body powder applied to genital areas. For AAs, study site interactions with age were included. For Whites, study site interactions with BMI, full-term pregnancies, aspirin use, education, and oral contraceptive use were included, and a pooled odds ratio was calculated using a multilevel logistic regression model for the aforementioned variables. The PAR for certain factors (HGSC: recent BMI ≥30 kg/m2 among White women; non-HGSC: family history of ovarian cancer among White women, never use of aspirin among AA women) was not provided because there was no association with ovarian cancer risk, resulting in a negative and implausible value for the PAR. For the collective PAR, each exposure was set to the following reference values to indicate lowest risk: BMI less than 30kg/m2, use of oral contraceptives, parous women, had a tubal ligation, no first-degree family history of breast or ovarian cancer, aspirin use, no user of body powder applied to genital areas, college education or higher, and no history of endometriosis.
Compared with postmenopausal women, the PARs were higher among premenopausal women for parity, first-degree family history of breast cancer, aspirin use (only among AA women), body powder applied to genital areas, and education (Figure 2; Supplementary Table 5, available online). Premenopausal women had a higher PAR for all exposures collectively than postmenopausal women: PARAA = 80.5%, 95% CI = 63.5% to 91.1% vs PARAA = 54.3%, 95% CI = 37.7% to 67.2%; PARWhite = 69.2%, 95% CI = 50.6% to 80.6% vs PARWhite = 39.2%, 95% CI = 28.2% to 49.2%, respectively. Compared with postmenopausal women, premenopausal women, particularly White premenopausal women, had higher PARs for all reproductive exposures, and premenopausal AA women had higher PARs for all modifiable exposures.
Figure 2.
Population attributable risk (PAR) percent and 95% confidence intervals (CI) for each exposure individually by race and menopausal status. This analysis includes 934 premenopausal African American (AA) women (267 cases, 667 controls), 1735 premenopausal White women (563 cases, 1172 controls), 2528 postmenopausal AA women (785 cases, 1743 controls), and 11 147 postmenopausal White women (2681 cases, 8466 controls). Models were adjusted for study site, reference year, age, full-term pregnancies, education, oral contraceptive use, age at menarche, imputed first-degree family history of breast cancer, imputed first-degree family history of ovarian cancer, tubal ligation, body mass index (BMI), menopausal status, hysterectomy 1 year before diagnosis, education, imputed endometriosis, imputed aspirin use, and imputed body powder applied to genital areas. For AAs, study site interactions with age were included. For Whites, study site interactions with BMI, full-term pregnancies, aspirin use, education, and oral contraceptive use were included, and a pooled odds ratio was calculated using a multilevel logistic regression model for the aforementioned variables. Among premenopausal women, the PAR for recent BMI ≥30 kg/m2 and never use of aspirin among White women was not provided as there was no association with ovarian cancer risk, resulting in a negative and implausible value for the PAR. For the collective PAR, each exposure was set to the following reference values to indicate lowest risk: BMI less than 30 kg/m2, use of oral contraceptives, parous women, had a tubal ligation, no first-degree family history of breast or ovarian cancer, aspirin use, no user of body powder applied to genital areas, college education or higher, and no history of endometriosis.
Discussion
In the OCWAA Consortium, we found that known or suspected risk factors accounted for more of the risk among AA women than White women, although few differences in the PARs for individual risk factors by race reached statistical significance after controlling for multiple comparisons. Compared with White women, AA women had a statistically significantly higher PAR for first-degree family history of breast cancer and for all modifiable exposures. Stratifying by histotype and menopausal status revealed higher collective PARs among non-HGSC vs HGSC tumors and among premenopausal vs postmenopausal women.
Racial differences in EOC risk associations (9-11,48,49) have primarily been from studies included in the OCWAA Consortium. AACES, NCOCS, and LACOCS represented 91% of the AA cases and 85% of the AA controls in a large analysis of racial or ethnic differences in EOC risk associations using data from 12 studies in the Ovarian Cancer Association Consortium (10). In that study, the only statistically significant difference across race or ethnicity was a more pronounced positive association between hysterectomy and EOC risk for AA women compared with the other racial or ethnic groups. Because there are currently controversies regarding the direction of the association between hysterectomy and EOC risk (50,51), we did not focus on this exposure and instead included hysterectomy as a covariate in all regression models. The 2 other studies comparing PARs between AA and White women, LACOCS (9) and NCOCS (11), were included in this analysis.
In this study, the only individual risk factor PAR found to be statistically significantly different by race was first-degree family history of breast cancer, where a greater proportion of EOC cases were attributable to this exposure among AA women compared with White women. Breast and ovarian cancer that clusters in families is likely due to BRCA mutations. Germline BRCA mutations are more prevalent among AA vs White women with ovarian cancer (52), yet there is no indication that AA women have a higher prevalence of these genetic mutations in the general population (53). Women with a family history of breast or ovarian cancer and/or BRCA mutation carriers can undergo prophylactic surgery to remove tissues to reduce the risk of developing EOC. However, studies have found that AA BRCA mutation carriers are less likely to receive prophylactic salpingo-oophorectomy than White mutation carriers (54,55). Our finding of a higher PAR among AA women may reflect an underuse of genetic testing and risk-reducing prophylactic surgery among AA women with a family history of breast or ovarian cancer compared with White women and potentially a stronger genetic contribution to ovarian cancer in AAs relative to Whites. Although genetic testing and subsequent prophylactic intervention represents 1 plausible explanation for this finding, other factors may contribute and additional studies investigating this association are needed.
The higher PARs for parity, oral contraceptive use, tubal ligation, BMI, and endometriosis for non-HGSC tumors, and the higher PARs for body powder applied to genital areas and a first-degree family history of breast and ovarian cancer for HGSC tumors observed in this study reflected known histotype-specific associations. Previous reports have shown that reproductive factors, endometriosis, and BMI are more strongly associated with risk of endometrioid and clear cell tumors (non-HGSC), and family history of ovarian or breast cancer is more strongly associated with risk of HGSC (3,12,14). There were also more substantive differences in the PARs by race for HGSC than non-HGSC, likely driven by family history; however, no statistically significant differences by race were observed after correction for multiple comparisons. Likewise, PARs for parity were higher among premenopausal than postmenopausal women, which is consistent with previous studies (56,57). The collective group of exposures accounted for more of the incidence of non-HGSC than HGSC and among premenopausal vs postmenopausal women. Because HGSC is the most commonly diagnosed histotype and most women are diagnosed with EOC in the postmenopausal period, it is critical to uncover additional risk factors that are associated with risk in these subgroups of women.
Due to possible differences in WHI participant selection from the other OCWAA studies (27), we excluded WHI and repeated the analyses, revealing appreciable differences in the PARs for White women but minimal differences for AA women. This likely reflects the large proportion of White cases and controls contributed by WHI to our analysis. We noted higher collective PARs for White women after excluding WHI, likely due to the removal of a large proportion of postmenopausal women as the 10 exposures under study accounted for less of the incidence of EOC among postmenopausal than premenopausal women. This also resulted in more similar PARs across race, although the trends remained the same as the analyses including WHI.
In this work, the PAR describes the proportion of cases in the population that would not have occurred if the exposure(s) were absent, but does not provide an estimate of the probability of causation or an estimate of the fraction of cases accelerated in time because of the exposure(s) (58-60). The PARs empirically suggest which exposures could be addressed with public health intervention. Because few racial differences were observed in this study, our findings imply that similar approaches to reduce ovarian cancer risk could be used among both AA and White women. Reducing the prevalence of modifiable risk factors should be combined with targeted interventions toward women with characteristics that are unmodifiable.
The OCWAA Consortium is a unique resource bringing together 7 studies to amass the largest sample size, to date, of AA women with EOC. Because the studies in OCWAA geographically cover most of the contiguous United States, our findings are likely generalizable to AA and White women in the United States. However, limitations can result from pooling and harmonizing data from multiple studies. Differences in the way responses are categorized across studies can result in a loss of detail after harmonization. A few of the studies did not collect information on exposures of interest (eg, body powder, endometriosis), resulting in systematically missing data by study. Multiple imputations were used to circumvent this limitation, and a complete case analysis showed comparable odds ratios and PARs with the imputed findings. All exposure information was self-reported and may be subject to recall bias, particularly for the case-control studies that were not nested within cohorts. All exposures were dichotomized, which may result in a loss of information and power. A diagnosis of endometriosis was not confirmed through medical records; however, other studies have shown good agreement between self-reported and medical record–abstracted endometriosis (61,62). Other well-established risk factors, including menopausal hormone therapy, breastfeeding, and smoking status, were not included because these exposures are associated with EOC in specific subpopulations (hormone therapy among postmenopausal women, breastfeeding among pregnant women) or are associated with risk of specific EOC histotypes (smoking and risk of mucinous tumors). Physical inactivity, a suspected risk factor for ovarian cancer (63), was not included in these analyses because different data collection instruments for this exposure across studies prevented data harmonization. Despite the relatively large sample size, we had limited power to investigate the less common histotypes individually.
Our findings suggest that racial differences in EOC incidence are partially driven by both known and suspected risk factors. The studied exposures accounted for more of the EOC risk among AA vs White women, of non-HGSC than HGSC, and among premenopausal vs postmenopausal women. Interventions to reduce EOC incidence focused on modifiable exposures may be slightly more beneficial among AA than White women at risk for EOC, although given that few racial differences were observed, both AA and White women are likely to benefit from similar approaches. It is critical to investigate different combinations of risk factors that could inform and improve prevention strategies among different subpopulations, especially as scientific understanding of EOC histotypes grows.
Funding
The OCWAA Consortium is funded by the National Cancer Institute (R01CA207260), and the individual studies in the OCWAA Consortium received funding from the National Institutes of Health: R01CA142081 for AACES; R01CA058420, UM1CA164974, and U01CA164974 for BWHS; P60MD003424 for CCCCS; N01CN025403, P01CA17054, P30CA14089, R01CA61132, N01PC67010, and R03CA113148, R03CA115195 for LACOCS; U01CA164973 for MEC; R01CA76016 for NCOCS; and R01CA092447 and U01CA202979 for the Southern Community Cohort Study. The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, US Department of Health and Human Services through contracts HHSN268201600001C, HHSN268201100046C, HHSN268201100001C, HHSN268201100002C, HHSN268201100003C, HHSN268201100004C, and HHSN271201100004C. The WHI Cancer Survivor Cohort is funded by 5UM1CA173642–05. Harris and Peres are additionally supported by the National Cancer Institute (K22 CA193860 [Harris]; R00 CA218681 [Peres]).
Notes
Role of the funder: The funders did not play a role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.
Disclosures: Patricia Moorman has received compensation for work related to litigation in regard to talc and ovarian cancer. All other authors report no conflicts of interest.
Acknowledgements: The authors thank the WHI investigators and staff for their dedication, and the study participants for making the program possible. A full listing of WHI investigators can be found at: https://www.whi.org/doc/WHI-Investigator-Long-List.pdf. Some of the studies obtained data on ovarian cancer cases from several state cancer registries (AZ, CA, CO, CT, DE, DC, FL, GA, IL, IN, KY, LA, MD, MA, MI, NJ, NY, NC, OK, PA, SC, TN, TX, VA) and results reported do not necessarily represent their views.
Author contributions: Conceptualization: JMS, LR, LCP, TNB. Data curation: TNB, ABF, HRH, CEJ, PGM, EM, HMO, VWS, AHW, LR, JMS. Formal Analysis: TFC, DLC, WR. Funding acquisition: JMS, LR. Methodology: JMS, LR, TFC, LCP, TNB. Writing – original draft: LCP, TNB, TFC. Writing – review & editing: All authors.
Data Availability
The data underlying this article will be shared on reasonable request to the OCWAA Consortium.
Supplementary Material
References
- 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA A Cancer J Clin. 2020;70(1):7-30. [DOI] [PubMed] [Google Scholar]
- 2. Köbel M, Kalloger SE, Boyd N, et al. Ovarian carcinoma subtypes are different diseases: implications for biomarker studies. PLoS Med. 2008;5(12):e232. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Wentzensen N, Poole EM, Trabert B, et al. Ovarian cancer risk factors by histologic subtype: an analysis from the Ovarian Cancer Cohort Consortium. J Clin Oncol. 2016;34(24):2888-2898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. McCluggage WG. Morphological subtypes of ovarian carcinoma: a review with emphasis on new developments and pathogenesis. Pathology. 2011;43(5):420-432. [DOI] [PubMed] [Google Scholar]
- 5. Soslow RA. Histologic subtypes of ovarian carcinoma. Int J Gynecol Pathol. 2008; 27(2):161-174. [DOI] [PubMed] [Google Scholar]
- 6. Peres LC, Cushing-Haugen KL, Kobel M, et al. Invasive epithelial ovarian cancer survival by histotype and disease stage. J Natl Cancer Inst. 2019;111(1):60-68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Torre LA, Trabert B, DeSantis CE, et al. Ovarian cancer statistics, 2018. CA Cancer J Clin. 2018;68(4):284-296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Park HK, Ruterbusch JJ, Cote ML. Recent trends in ovarian cancer incidence and relative survival in the United States by race/ethnicity and histologic subtypes. Cancer Epidemiol Biomarkers Prev. 2017;26(10):1511-1518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wu AH, Pearce CL, Tseng CC, et al. African Americans and Hispanics remain at lower risk of ovarian cancer than non-Hispanic whites after considering nongenetic risk factors and oophorectomy rates. Cancer Epidemiol Biomarkers Prev. 2015;24(7):1094-1100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Peres LC, Risch H, Terry KL, et al. ; Australian Ovarian Cancer Study Group. Racial/ethnic differences in the epidemiology of ovarian cancer: a pooled analysis of 12 case-control studies. Int J Epidemiol. 2018;47(2):460-472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Moorman PG, Palmieri RT, Akushevich L, et al. Ovarian cancer risk factors in African-American and white women. Am J Epidemiol. 2009;170(5):598-606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Olsen CM, Nagle CM, Whiteman DC, et al. Obesity and risk of ovarian cancer subtypes: evidence from the Ovarian Cancer Association Consortium. Endocr-Relat Cancer. 2013;20(2):251-262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Faber MT, Kjær SK, Dehlendorff C, et al. ; The Australian Cancer Study (Ovarian Cancer). Cigarette smoking and risk of ovarian cancer: a pooled analysis of 21 case-control studies. Cancer Causes Control. 2013;24(5):989-1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Pearce CL, Templeman C, Rossing MA, et al. Association between endometriosis and risk of histological subtypes of ovarian cancer: a pooled analysis of case-control studies. Lancet Oncol. 2012;13(4):385-394. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Collaborative Group on Epidemiological Studies of Ovarian Cancer:Beral V, Gaitskell K, Hermon C, et al. Ovarian cancer and smoking: individual participant meta-analysis including 28,114 women with ovarian cancer from 51 epidemiological studies. Lancet Oncol. 2012;13(9):946-956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Gates MA, Rosner BA, Hecht JL, et al. Risk factors for epithelial ovarian cancer by histologic subtype. Am J Epidemiol. 2010;171(1):45-53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Gaitskell K, Green J, Pirie K, et al. ; Million Women Study Collaborators. Histological subtypes of ovarian cancer associated with parity and breastfeeding in the prospective Million Women Study. Int J Cancer. 2018;142(2):281-289. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Schildkraut JM, Peres LC, Bethea TN, et al. Ovarian Cancer in Women of African Ancestry (OCWAA) consortium: a resource of harmonized data from eight epidemiologic studies of African American and white women. Cancer Causes Control. 2019;30(9):967-978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Schildkraut JM, Alberg AJ, Bandera EV, et al. A multi-center population-based case-control study of ovarian cancer in African-American women: The African American Cancer Epidemiology Study (AACES). BMC Cancer. 2014;14(1):688. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Kim S, Dolecek TA, Davis FG. Racial differences in stage at diagnosis and survival from epithelial ovarian cancer: a fundamental cause of disease approach. Soc Sci Med. 2010;71(2):274-281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Peterson CE, Rauscher GH, Johnson TP, et al. The association between neighborhood socioeconomic status and ovarian cancer tumor characteristics. Cancer Causes Control. 2014;25(5):633-637. [DOI] [PubMed] [Google Scholar]
- 22. Wu AH, Pearce CL, Tseng CC, et al. Markers of inflammation and risk of ovarian cancer in Los Angeles County. Int J Cancer. 2009;124(6):1409-1415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Schildkraut JM, Moorman PG, Halabi S, et al. Analgesic drug use and risk of ovarian cancer. Epidemiology. 2006;17(1):104-107. [DOI] [PubMed] [Google Scholar]
- 24. Bethea TN, Palmer JR, Adams-Campbell LL, et al. A prospective study of reproductive factors and exogenous hormone use in relation to ovarian cancer risk among Black women. Cancer Causes Control. 2017;28(5):385-391. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Kolonel LN, Henderson BE, Hankin JH, et al. A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol. 2000;151(4):346-357. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Signorello LB, Hargreaves MK, Steinwandel MD, et al. Southern Community Cohort Study: establishing a cohort to investigate health disparities. J Natl Med Assoc. 2005;97(7):972-979. [PMC free article] [PubMed] [Google Scholar]
- 27. Hays J, Hunt JR, Hubbell FA, et al. The Women's Health Initiative recruitment methods and results. Ann Epidemiol. 2003;13(9):S18-S77. [DOI] [PubMed] [Google Scholar]
- 28. Kurman RJ, Carcangiu ML, Herrington CS, et al. WHO Classification of Tumours of Female Reproductive Organs. Lyon, France: International Agency for Research on Cancer; 2014.
- 29. Schildkraut JM, Abbott SE, Alberg AJ, et al. Association between body powder use and ovarian cancer: The African American Cancer Epidemiology Study (AACES). Cancer Epidemiol Biomarkers Prev. 2016;25(10):1411-1417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Bangdiwala SI, Bhargava A, O’Connor DP, et al. Statistical methodologies to pool across multiple intervention studies. Behav Med Pract Policy Res. 2016;6(2):228-235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Turner RM, Omar RZ, Yang M, et al. A multilevel model framework for meta-analysis of clinical trials with binary outcomes. Stat Med. 2000;19(24):3417-3432. [DOI] [PubMed] [Google Scholar]
- 32. Thompson SG, Turner RM, Warn DE. Multilevel models for meta-analysis, and their application to absolute risk differences. Stat Methods Med Res. 2001;10(6):375-392. [DOI] [PubMed] [Google Scholar]
- 33. Bruzzi P, Green SB, Byar DP, et al. Estimating the population attributable risk for multiple risk factors using case-control data. Am J Epidemiol. 1985;122(5):904-914. [DOI] [PubMed] [Google Scholar]
- 34. Benichou J. Methods of adjustment for estimating the attributable risk in case-control studies: a review. Stat Med. 1991;10(11):1753-1773. [DOI] [PubMed] [Google Scholar]
- 35. Benichou J. A review of adjusted estimators of attributable risk. Stat Methods Med Res. 2001;10(3):195-216. [DOI] [PubMed] [Google Scholar]
- 36. Llorca J, Delgado-Rodriguez M. A comparison of several procedures to estimate the confidence interval for attributable risk in case-control studies. Stat Med. 2000;19(8):1089-1099. [DOI] [PubMed] [Google Scholar]
- 37. Davison AC, Hinkley DV. Bootstrap Methods and Their Application. Cambridge, UK: Cambridge University Press; 1997. [Google Scholar]
- 38. Benjamini Y, Krieger AM, Yekutieli D. Adaptive linear step-up procedures that control the false discovery rate. Biometrika. 2006;93(3):491-507. [Google Scholar]
- 39. van Buuren S, Brand JPL, Groothuis-Oudshoorn CGM, et al. Fully conditional specification in multivariate imputation. J Stat Comput Simul. 2006;76(12):1049-1064. [Google Scholar]
- 40. Schafer JL, Graham JW. Missing data: our view of the state of the art. Psychol Methods. 2002;7(2):147-177. [PubMed] [Google Scholar]
- 41. Sterne JA, White IR, Carlin JB, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ. 2009;338(jun29 1):b2393. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Horton NJ, Kleinman KP. Much ado about nothing: a comparison of missing data methods and software to fit incomplete data regression models. Am Stat. 2007;61(1):79-90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Jolani S. Hierarchical imputation of systematically and sporadically missing data: an approximate Bayesian approach using chained equations. Biom J. 2018;60(2):333-351. [DOI] [PubMed] [Google Scholar]
- 44. Jolani S, Debray TP, Koffijberg H, et al. Imputation of systematically missing predictors in an individual participant data meta-analysis: a generalized approach using MICE. Stat Med. 2015;34(11):1841-1863. [DOI] [PubMed] [Google Scholar]
- 45. Enders CK, Du H, Keller BT. A model-based imputation procedure for multilevel regression methods with random coefficients, interaction effects, and nonlinear terms. Psychol Methods. 2020;25(1):88-112. [DOI] [PubMed] [Google Scholar]
- 46. Enders CK, Keller BT, Levy R. A fully conditional specification approach to multilevel imputation of categorical and continuous variables. Psychol Methods. 2018;23(2):298-317. [DOI] [PubMed] [Google Scholar]
- 47. Keller BT, Enders CK. Blimp User’s Manual (Version 2.1). Los Angeles, CA; 2019. http://www.appliedmissingdata.com/blimpusermanual-2-1.pdf. [Google Scholar]
- 48. John EM, Whittemore AS, Harris R, et al. ; Collaborative Ovarian Cancer Group. Characteristics relating to ovarian cancer risk: collaborative analysis of seven U.S. case-control studies. Epithelial ovarian cancer in Black women. J Natl Cancer Inst. 1993;85(2):142-147. [DOI] [PubMed] [Google Scholar]
- 49. Ness RB, Grisso JA, Klapper J, et al. Racial differences in ovarian cancer risk. J Natl Med Assoc. 2000;92(4):176-182. [PMC free article] [PubMed] [Google Scholar]
- 50. Jordan SJ, Nagle CM, Coory MD, et al. Has the association between hysterectomy and ovarian cancer changed over time? A systematic review and meta-analysis. Eur J Cancer. 2013;49(17):3638-3647. [DOI] [PubMed] [Google Scholar]
- 51. Peres LC, Alberg AJ, Bandera EV, et al. Premenopausal hysterectomy and risk of ovarian cancer in African-American women. Am J Epidemiol. 2017;186(1):46-53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Kurian AW, Ward KC, Howlader N, et al. Genetic testing and results in a population-based cohort of breast cancer patients and ovarian cancer patients. J Clin Oncol. 2019;37(15):1305-1315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53. Kurian AW. BRCA1 and BRCA2 mutations across race and ethnicity: distribution and clinical implications. Curr Opin Obstet Gynecol. 2010;22(1):72-78. [DOI] [PubMed] [Google Scholar]
- 54. Cragun D, Weidner A, Lewis C, et al. Racial disparities in BRCA testing and cancer risk management across a population-based sample of young breast cancer survivors. Cancer. 2017;123(13):2497-2505. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55. Bradbury AR, Ibe CN, Dignam JJ, et al. Uptake and timing of bilateral prophylactic salpingo-oophorectomy among BRCA1 and BRCA2 mutation carriers. Genet Med. 2008;10(3):161-166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56. Moorman PG, Calingaert B, Palmieri RT, et al. Hormonal risk factors for ovarian cancer in premenopausal and postmenopausal women. Am J Epidemiol. 2008;167(9):1059-1069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57. Tung KH, Wilkens LR, Wu AH, et al. Effect of anovulation factors on pre- and postmenopausal ovarian cancer risk: revisiting the incessant ovulation hypothesis. Am J Epidemiol. 2005;161(4):321-329. [DOI] [PubMed] [Google Scholar]
- 58. Greenland S. Attributable fractions: bias from broad definition of exposure. Epidemiology. 2001;12(5):518-520. [DOI] [PubMed] [Google Scholar]
- 59. Greenland S. Concepts and pitfalls in measuring and interpreting attributable fractions, prevented fractions, and causation probabilities. Ann Epidemiol. 2015;25(3):155-161. [DOI] [PubMed] [Google Scholar]
- 60. Levine B. What does the population attributable fraction mean? Prev Chronic Dis. 2007;4(1):A14. [PMC free article] [PubMed] [Google Scholar]
- 61. Missmer SA, Hankinson SE, Spiegelman D, et al. Reproductive history and endometriosis among premenopausal women. Obstet Gynecol. 2004;104(5 Pt 1):965-974. [DOI] [PubMed] [Google Scholar]
- 62. Saha R, Marions L, Tornvall P. Validity of self-reported endometriosis and endometriosis-related questions in a Swedish female twin cohort. Fertil Steril. 2017;107(1):174-178.e2. [DOI] [PubMed] [Google Scholar]
- 63. Cannioto R, LaMonte MJ, Risch HA, et al. ; on behalf of The Australian Ovarian Cancer Study Group. Chronic recreational physical inactivity and epithelial ovarian cancer risk: evidence from the Ovarian Cancer Association Consortium. Am Soc Prev Oncol. 2016;25(7):1114-1124. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article will be shared on reasonable request to the OCWAA Consortium.