First- and second-degree family history of ovarian and breast cancer in relation to risk of invasive ovarian cancer in African American and white women.

Details about the OCWAA Consortium have been published previously ( 5 ). In brief, the OCWAA Consortium includes pooled and harmonized data from four case-control studies – African American Cancer Epidemiology Study (AACES) ( 6 ), Cook County Case-Control Study (CCCCS) ( 7 ), Los Angeles County Ovarian Cancer Study (LACOCS) ( 8 ), North Carolina Ovarian Cancer Study (NCOCS) ( 9 ) – and four nested case-control studies within prospective cohorts – Black Women’s Health Study (BWHS) ( 10 ), Multiethnic Cohort Study (MEC) ( 11 ), Southern Community Cohort Study (SCCS) ( 12 ), Women’s Health Initiative (WHI) ( 13 ) to examine differences between AAs and Whites in ovarian cancer incidence and survival, with particular attention to histotype. Each study collected risk factor and covariate data via self-administered questionnaires and/or interviews. Data on cancer diagnoses were abstracted from medical records and cancer registry reports. Only invasive epithelial ovarian cancer cases, defined using International Classification of Diseases for Oncology, Version 3 ( 14 ), were eligible for inclusion. Accounting for both morphology and grade ( 15 , 16 ), histotype was classified as high-grade serous, low-grade serous, endometrioid, clear cell, mucinous, carcinosarcoma, and other epithelial tumors. Each study obtained informed consent from its participants; the individual studies and the OCWAA Consortium were approved by the relevant Institutional Review Boards. For the present analyses, data were available from AACES, BWHS, CCCCS, LACOCS, MEC, and NCOCS. Thus, 1,052 AA cases, 2,328 AA controls, 2,380 White cases, and 3,982 White controls were included in the present analyses. Each study collected data on first-degree FH of ovarian cancer and of breast cancer via questionnaire or interview. First-degree family history of ovarian cancer was defined as a diagnosis of ovarian cancer in the mother, any sister, and/or any daughter and first-degree family history of breast cancer was defined as a diagnosis of breast cancer in any first-degree relative (i.e., mother, father, sister, brother, daughter, son). Self-reported first-degree FH of ovarian cancer and first-degree FH of breast cancer was dichotomized as no or yes for analysis. We also examined self-reported second-degree FH of ovarian (or breast) cancer, defined as a diagnosis of ovarian (or breast) cancer in any grandmother and/or aunt, in studies where these data were collected. The four case-control studies (AACES, CCCCS, LACOCS, NCOCS) collected data on second-degree family history of both ovarian and breast cancer; the BWHS collected data on second-degree family history of breast cancer. Self-reported second-degree FH of ovarian (or breast) cancer was dichotomized as no or yes for analysis. We considered the following covariates in logistic regression models: age (continuous, years), study, menopausal status and hormone therapy use (premenopausal, postmenopausal/never, postmenopausal/ever), oral contraceptive use (never, <60 months, ≥60 months), tubal ligation (no, yes), hysterectomy at least one year prior (no, yes), parity (0, 1, 2, 3, ≥4 full-term pregnancies), age at menarche (<12, 12-14, ≥14 years), recent body mass index (BMI, <25.0, 25.0-29.9, 30.0-34.9, ≥35.0 kg/m 2 ), education (high school or less, some college, college graduate, graduate education), and cigarette smoking (never, ever). Descriptive statistics were calculated to characterize the frequencies of several key variables for the analysis stratified by both race and case-control status. Race-specific multi-level logistic regression models were used to test for overall associations of first- and second-degree FH of ovarian and breast cancer with risk of ovarian cancer separately. We restricted analyses for second-degree FH of cancer to participants without first-degree FH. To investigate potential histotype-specific associations, analyses were stratified by high-grade serous carcinoma (HGSC) and all other histotypes combined, referred to as non-HGSC. SAS 9.4 (SAS Institute, Inc., Cary, NC) was used for regression analyses. Race- and study-specific associations were estimated with control for covariates and combined in a multilevel random-effects meta-analysis in SAS 9.4 ( 17 – 19 ), where the association between main exposure and outcome was allowed to vary by study by including a random coefficient. As no significant heterogeneity by study was detected ( 18 , 20 ), random-effects coefficients were removed from the models and associations were assumed constant across studies. Forest plots were created using R statistical software (version 3.6.2). We used the Bruzzi method to calculate population attributable risk percentages (PARs) and 95% CIs for first- and second-degree FH of ovarian and breast cancer ( 21 ). Multivariable models used in the calculation included the previously listed covariates, as well as endometriosis (yes/no), pelvic inflammatory disease (yes/no), and study site interactions for oral contraceptive use, BMI, full-term pregnancies, endometriosis, reference time, and pelvic inflammatory disease.

Table 1 describes characteristics of the cases and controls in OCWAA stratified by race. Among both cases and controls, AA women were less likely to have graduated college and to report ever using hormone therapy. AA women were more likely to have had a tubal ligation, report a hysterectomy, have had ≥3 full-term pregnancies, and have a recent BMI ≥30 kg/m 2 compared to White women. Among the cases, AA women were slightly younger than White women (58.2 vs 60.0 years). The distribution of the five major ovarian cancer histotypes was similar for AA and White women, and the majority of cases presented with a high-grade serous tumor (63.1% for AAs and 59.6% for Whites). The next most common major histotypes were endometrioid (8.7%) for AAs and both endometrioid (7.9%) and clear cell (7.9%) for Whites. The prevalence of first-degree FH of ovarian cancer was the same for AA and White controls (3.2%) ( Table 1 ). First-degree FH of breast cancer in controls was similar by race (12.8% for AAs and 13.3% for Whites). Second-degree FH of ovarian cancer was more prevalent in AA controls than in White controls (7.6 vs. 4.5%), while second-degree FH of breast cancer was less prevalent among AA controls than White controls (20 vs 28%). Cases had a higher prevalence of FH of ovarian and breast cancer than controls ( Table 1 ). Among cases, 6.3% of AA women and 5.3% of White women reported a first-degree FH of ovarian cancer. AA cases had a higher prevalence of first-degree FH of breast cancer compared to White cases (22.9 vs 17.3%). For second-degree FH, AA cases had a higher prevalence of FH of ovarian cancer than White cases (11 vs 7%), but the prevalence of second-degree FH of breast cancer was similar by race (29% for AAs and 30% for Whites). Table 2 shows associations between first-degree FH and ovarian cancer, stratified by race and histotype. The OR for ovarian cancer in women with a first-degree FH of ovarian cancer was similar for AAs (OR=1.93, 95% CI: 1.30, 2.85) and Whites (OR=2.04, 95% CI: 1.54, 2.70). First-degree FH of ovarian cancer was associated with increased odds of HGSC for both AA (OR=2.32, 95% CI: 1.50, 3.59) and White women (OR=2.48, 95% CI: 1.82, 3.38). First-degree FH of ovarian cancer was not significantly associated with non-HGSC among AA women; however, for White women, we observed a positive association (OR=1.55, 95% CI: 1.04, 2.33). No statistically significant study heterogeneity was observed ( Figure 1A and 1B , p heterogeneity >0.9). First degree FH of breast cancer was associated with higher odds of ovarian cancer, with ORs of 1.77 (95% CI: 1.43, 2.20) and 1.30 (95% CI: 1.11, 1.51) for AA and White women, respectively. When stratified by histotype, the ORs for risk of high-grade serous tumors were similar to the overall ORs ( Table 2 ). For non-HGSC, first-degree FH of breast cancer was associated with higher odds of ovarian cancer among AA women, but not White women. Results were similar across the studies ( Figures 2A and 2B , p heterogeneity >0.9). Table 3 shows the association of second-degree FH of cancer with odds of ovarian cancer among OCWAA participants without first-degree FH of cancer. Among AA women, the odds of ovarian cancer was significantly higher in women who reported a second-degree FH of ovarian cancer, with an OR of 1.74 (95% CI: 1.03, 2.94) for HGSC. Among White women, the odds of both overall ovarian cancer and high-grade serous tumors was increased in women reporting a second-degree FH of ovarian cancer, with ORs of 1.49 (95% CI: 1.05, 2.11) and 1.63 (95% CI: 1.08, 2.47), respectively. For second-degree FH of breast cancer, we observed no significant associations for overall ovarian cancer or high-grade serous carcinoma in AA or White women. Second degree family history of breast cancer was marginally associated with non-HGSC for AAs: 1.50 (95% CI: 1.00, 2.24). We estimated population attributable risks (PARs) for ovarian cancer for first- and second-degree FH of ovarian and breast cancer. The PAR for first-degree FH of ovarian cancer was 2.9 (95% CI: 1.2, 4.7) for AA women and 2.0 (95% CI: 1.1, 2.8) for White women. The PARs for first-degree FH of breast cancer were 10.1 (95% CI: 6.5, 13.7) for AA women and 2.6 (95% CI: 0.8, 4.4) for White women. The PAR for second-degree FH of ovarian cancer was 3.1 (95% CI: −0.5, 6.8) for AA women and 3.0 (95% CI: 1.2, 4.8) for White women. The PAR for second-degree FH of breast cancer was 8.3 (95% CI: 2.6, 13.8) for AA women and 2.3 (95% CI: −2.6, 6.7) for White women.

First-degree FH of ovarian cancer and first-degree FH of breast cancer was associated with higher odds of ovarian cancer in both AA and White women. Our findings suggest that first-degree FH of ovarian cancer is more strongly associated with risk of HGSC than with non-HGSC in both racial groups. First-degree FH of breast cancer was associated with risk of HGSC and non-HGSC among AAs, but with HGSC only among Whites. Second-degree FH of ovarian cancer was associated with increased risk of HGSC in both AA and White women, whereas second-degree FH of breast cancer was marginally associated with non-HGSC in AAs only. A pooled study that included AACES, LACOCS, NCOCS, and nine other studies from the Ovarian Cancer Association Consortium (some of which are also included in this OCWAA consortium) provided evidence that first-degree FH of ovarian or breast cancer is associated with epithelial ovarian cancer risk in White and AA women ( 4 ). AA ovarian cancer cases were slightly more likely to report a first-degree FH of breast or ovarian cancer compared to White cases, at 26% and 21%, respectively and AA participants were at slightly higher risk (OR=1.77, 95% CI: 1.42, 2.21) compared to Whites (OR=1.35, 95% CI: 1.25, 1.45). For HGSC, the ORs were also slightly higher for African Americans compared to Whites: OR=1.72 (95% CI 1.35, 2.19) and OR=1.47 (95% CI: 1.36, 1.59), respectively. In the Ovarian Cancer Cohort Consortium (OC3), a pooled study that included MEC, first-degree FH of ovarian cancer was associated with both epithelial ovarian cancer overall (RR=1.48, 95% CI: 1.26, 1.75) and high-grade serous tumors (RR=1.60, 95% CI: 1.26, 2.03) ( 22 ). These findings are similar to those in our study that was focused on evaluating the associations in AAs and Whites specifically. In the present study, ORs were consistently higher for both overall and high-grade carcinoma in both AAs and Whites. Our findings suggest that family history may be a strong risk factor for both overall and HGSC irrespective of race. Notably, the magnitude of association and the PAR for FH of breast cancer was higher among AAs than among Whites, though the confidence intervals overlapped. OC3 also reported on ovarian cancer risk in the presence of first-degree FH of breast cancer; the OR was 1.48 (95% CI: 1.00, 1.19) for overall ovarian cancer and 1.13 (95% CI: 1.00, 1.29) for HGSC ( 22 ). Our findings for overall and high-grade serous tumors in OCWAA are reflective of a stronger association of a FH of breast cancer with ovarian cancer risk among both AAs and Whites, similar to the findings from OC3. There was a stronger association for serous carcinoma compared to the overall ovarian cancer in the present study, which differs from the OC3 finding. Several studies have examined the PAR for first-degree FH of ovarian cancer and ovarian cancer incidence. In a case-control study conducted in Italy, the PAR for having a first-degree FH of ovarian or breast cancer was 4.0% (95% CI: 2.0%, 6.0%) ( 23 ). These findings in European populations are similar to the PAR among White and AA participants in OCWAA. While the CIs overlapped, in LACOCS, the only study to present PARs for both AA and White women( 24 ), the PARs for first-degree FH of ovarian cancer in White and AA participants – 2.9% (95% CI: 1.9%, 3.6%) and 3.9% (95% CI: 2.6%, 4.9%), respectively – were slightly higher than among participants in OCWAA. In the present study, 8.0% of African American participants and 46.9% of White participants were from LACOCS and the difference in PARs could be due to the differences in the populations or the covariates included in the analyses. Although most of the covariates were similar, the LACOCS analysis adjusted for income and the OCWAA analysis adjusted for hysterectomy, cigarette smoking, pelvic inflammatory disease, and study site interactions. Family history of ovarian cancer is recognized as a strong risk factor for ovarian cancer and represents an underlying genetic susceptibility for the disease. There is ample evidence from both family and population-based studies that several alleles, ranging from the rare and highly penetrant variants to the common and low risk alleles, increase risk ( 1 ) and that genetic risk may differ by histotype ( 25 ). For example, mutations in the BRCA genes account for 5-15% of ovarian cancers ( 25 , 26 ). Ovarian tumors among women with BRCA1 and BRCA2 mutations are more likely to be serous tumors and to have a higher grade ( 25 , 27 ). However, only a portion of the estimated heritable risk of ovarian cancer is accounted for by alleles at the extreme ends of the allele frequency spectrum ( 1 ); the rest of the variability is due to genes or genetic factors that are still unknown. This highlights the need to continue to study family history among first-degree relatives and other family members in epidemiologic studies. The present study is the largest to-date to examine the relation of FH of ovarian cancer and breast cancer to risk of ovarian cancer among African American women. The sample size enabled us to control for a variety of important confounders, stratify by histotype, and calculate PARs for both African American and White participants. However, our study was unable to assess low-grade serous, endometrioid, clear cell, mucinous, carcinosarcoma, and other epithelial tumors individually due to small sample sizes. To our knowledge, OCWAA is only the second study to present the PAR for first-degree FH of ovarian cancer among African American women and is the first study to present the PAR for first-degree FH of breast cancer among African American women. FH of ovarian and breast cancer was self-reported in each of the contributing OCWAA studies. Although first-degree FH of breast cancer tends to be accurately reported by both African American and White women ( 28 ), first-degree FH of ovarian cancer is less well-reported ( 29 , 30 ). The OCWAA Consortium has incomplete data on family size in order to account for differences in having first- and second-degree relatives at risk of ovarian and/or breast cancer. The findings for second-degree FH of ovarian and breast cancer must be interpreted with caution. Data on second-degree FH were limited as not all studies collected data on second-degree FH and, among the studies with these variables, data about all second-degree relatives at risk (e.g., nieces) were not collected. These FH data may be underreported, particularly if participants were unaware of cancer diagnoses among second-degree relatives. In addition, FH of other cancers could be important confounders of an association between second-degree FH of ovarian and breast cancer and ovarian cancer risk ( 31 ), but we were unable to assess FH of other cancer sites. First-degree FH of ovarian cancer may be more strongly associated with high-grade serous carcinoma than with other histotypes in African American and White women. The association of second-degree FH of cancer with risk of ovarian cancer may differ by race; however, replication in studies with valid classification of affected second degree relatives in both African American and White populations is needed.

Family history (FH) of ovarian cancer is an important risk factor for ovarian cancer ( 1 – 3 ), yet there are still unanswered questions about this relationship, such as how FH is associated with ovarian cancer risk in African Americans and whether FH associations are specific to certain ovarian cancer histotypes within this group. For example, one recent study found that the association of first-degree family history of ovarian or breast cancer was greater among African American women than among White women, though the difference was not statistically significant, and the magnitude of association was stronger for high-grade serous ovarian carcinoma ( 4 ). Less is known about the relation of second-degree family history of ovarian and breast cancer to ovarian cancer risk, overall and by histotype, among African Americans. In the largest sample of African American ovarian cancer cases and controls in the U.S., assembled via the Ovarian Cancer in Women of African Ancestry (OCWAA) Consortium, we examined the associations of first- and second-degree FH of ovarian and breast cancer to risk of ovarian cancer by race and histotype.