Uterine Cancer After Risk-Reducing Salpingo-oophorectomy Without Hysterectomy in Women With BRCA Mutations

Key Points

Question  Are women with BRCA mutations who undergo risk-reducing salpingo-oophorectomy (RRSO) without hysterectomy at increased risk for uterine cancer compared with expected rates from the Surveillance, Epidemiology, and End Results database?

Findings  In this prospective cohort study, no increased risk for endometrioid carcinoma or sarcoma was seen. During a median 5.1 years of follow-up, 4 serous and/or serous-like carcinomas were seen in 627 women with BRCA1 mutations vs 0.18 expected (observed to expected ratio, 22.2), a significant difference.

Meaning  Women with BRCA1 mutations have an increased risk for serous and/or serous-like endometrial cancer that should be considered when discussing advantages and risks of hysterectomy at the time of RRSO.

Importance  The link between BRCA mutations and uterine cancer is unclear. Therefore, although risk-reducing salpingo-oophorectomy (RRSO) is standard treatment among women with BRCA mutations (BRCA+ women), the role of concomitant hysterectomy is controversial.

Objective  To determine the risk for uterine cancer and distribution of specific histologic subtypes in BRCA+ women after RRSO without hysterectomy.

Design, Setting, and Participants  This multicenter prospective cohort study included 1083 women with a deleterious BRCA1 or BRCA2 mutation identified from January 1, 1995, to December 31, 2011, at 9 academic medical centers in the United States and the United Kingdom who underwent RRSO without a prior or concomitant hysterectomy. Of these, 627 participants were BRCA1+; 453, BRCA2+; and 3, both. Participants were prospectively followed up for a median 5.1 (interquartile range [IQR], 3.0-8.4) years after ascertainment, BRCA testing, or RRSO (whichever occurred last). Follow up data available through October 14, 2014, were included in the analyses. Censoring occurred at uterine cancer diagnosis, hysterectomy, last follow-up, or death. New cancers were categorized by histologic subtype, and available tumors were analyzed for loss of the wild-type BRCA gene and/or protein expression.

Main Outcomes and Measures  Incidence of uterine corpus cancer in BRCA+ women who underwent RRSO without hysterectomy compared with rates expected from the Surveillance, Epidemiology, and End Results database.

Results  Among the 1083 women women who underwent RRSO without hysterectomy at a median age 45.6 (IQR: 40.9 - 52.5), 8 incident uterine cancers were observed (4.3 expected; observed to expected [O:E] ratio, 1.9; 95% CI, 0.8-3.7; P = .09). No increased risk for endometrioid endometrial carcinoma or sarcoma was found after stratifying by subtype. Five serous and/or serous-like (serous/serous-like) endometrial carcinomas were observed (4 BRCA1+ and 1 BRCA2+) 7.2 to 12.9 years after RRSO (BRCA1: 0.18 expected [O:E ratio, 22.2; 95% CI, 6.1-56.9; P < .001]; BRCA2: 0.16 expected [O:E ratio, 6.4; 95% CI, 0.2-35.5; P = .15]). Tumor analyses confirmed loss of the wild-type BRCA1 gene and/or protein expression in all 3 available serous/serous-like BRCA1+ tumors.

Conclusions and Relevance  Although the overall risk for uterine cancer after RRSO was not increased, the risk for serous/serous-like endometrial carcinoma was increased in BRCA1+ women. This risk should be considered when discussing the advantages and risks of hysterectomy at the time of RRSO in BRCA1+ women.

In 2002, Kauff et al¹ and Rebbeck et al² published the first 2 studies demonstrating that risk-reducing salpingo-oophorectomy (RRSO) decreased the risk for breast and ovarian or fallopian tube cancer in women with mutations in the BRCA1 (OMIM 113705) (BRCA1+ women) or BRCA2 (OMIM 600185) (BRCA2+ women) gene. Subsequent studies have confirmed that RRSO not only reduces the risk for ovarian or fallopian tube cancer by 80% to 90% and the risk for breast cancer by 40% to 70%^3-5 but also reduces disease-specific and overall mortality.^4,5

Although RRSO is now part of standard management for BRCA+ women, the role of concomitant hysterectomy remains controversial.^6,7 Some studies^8,9 have suggested that BRCA+ women may be at higher risk for uterine corpus cancer, while others¹⁰ have suggested that the risk is predominantly associated with tamoxifen citrate use. Further reports^11,12 have suggested that more aggressive, serous and/or serous-like (serous/serous-like) endometrial carcinoma is overrepresented in BRCA+ women, but other studies^13,14 have not confirmed this. Clarification of this issue is particularly relevant, because the serous/serous-like subtype accounts for only about 10% of uterine corpus cancer cases but more than 40% of deaths due to the disease.¹⁵

To provide data relevant to these questions, we conducted a multicenter, prospective evaluation of uterine corpus cancer risk after RRSO in BRCA1+ and BRCA2+ women. We also examined whether specific histologic subtypes were overrepresented.

Women were eligible for this study if they (1) had a deleterious BRCA mutation identified from January 1, 1995, to December 31, 2011; (2) underwent RRSO with their uterus left in situ; (3) did not have ovarian, fallopian tube, or uterine cancer before the later of genetic testing or RRSO; and (4) consented in writing to participation in one of several institutional review board–approved prospective follow-up studies conducted at the University of Pennsylvania, Memorial Sloan Kettering Cancer Center, and 7 academic centers participating in the Prevention and Observation of Surgical End Points Study (PROSE Consortium), including Baylor–Charles A. Sammons Cancer Center, City of Hope Medical Center, Dana-Farber Cancer Institute, Fox Chase Cancer Center, Lombardi Cancer Center, Royal Marsden Hospital, and UCLA Jonsson Comprehensive Cancer Center.

Women were followed up via structured questionnaire and medical record review as previously described.^1-3,16 Information regarding risk-reducing and therapeutic surgical procedures, medication exposures, and new malignant neoplasms was ascertained. Pathology reports were requested for all new uterine corpus cancers diagnosed during follow-up. No individuals were excluded owing to race or ethnicity. Race or ethnicity designations were self-reported and collected to assist in determining expected age- and race-specific cancer incidence.

Follow-up time began from the latest of date of ascertainment, receipt of BRCA testing results, or RRSO. The main outcome of interest was the diagnosis of uterine cancer. Censoring events were hysterectomy, last follow-up, or death. Women were considered to be at risk from 30 years and older, and women with less than 6 months of follow-up were excluded. Follow-up data available through October 14, 2014, were included in the analyses.

Expected incidence of uterine cancer was calculated by multiplying woman-years at risk by the age- and race-specific data from the Surveillance, Epidemiology, and End Results database (SEER*Stat database: Incidence—SEER 18 registries, November 2012 [submitted 2000-2010]), categorized by age into 5-year increments.¹⁷ Because 31% of US women have undergone hysterectomy by 60 years of age, expected incidences were adjusted for the age- and race-specific prevalence of hysterectomy, obtained using the 2010 Cross Tabulation Analysis Tool from the Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System (BRFSS),¹⁸ and grouped in 5-year increments of age as indicated in the following equation:

Adjusted Incidence = SEER Expected Incidence/(1 − BRFSS Hysterectomy Prevalence).

To determine the expected incidence of specific histologic subtypes, we used codes from the International Classification of Diseases for Oncology (ICD-O) to assign SEER cases into 1 of the following 5 histologic categories: (1) endometrioid endometrial carcinoma (eg, endometrioid, adenocarcinoma not otherwise specified); (2) serous/serous-like endometrial carcinoma (eg, serous, undifferentiated, carcinosarcoma); (3) clear cell carcinoma; (4) mucinous carcinoma; and (5) uterine sarcoma. Specific ICD-O codes assigned to each category are detailed in eTable 1 in the Supplement. Carcinosarcomas, which are epithelial cancers rather than sarcomas, were included in the serous/serous-like category, because these are frequently thought to have dedifferentiated from a serous precursor, and the carcinomatous component is the primary driver of tumor aggressiveness.¹⁹ Similarly, endometrial carcinomas with mixed histologic features were also classified as serous/serous-like if a serous component was present, because the Cancer Genome Atlas Research Network recently demonstrated that most of these cases cluster in the copy-number high (serous-like) group.²⁰ Observed cases were categorized in the same way based on pathologic reports, and in cases with available tumor specimens, review of primary tumor by the study pathologist (R.A.S.).

Exact 2-sided P values for observed to expected (O:E) ratios for cancers were calculated assuming a Poisson distribution of the observed values using the STATA statistical package (version 13; StataCorp). Confidence intervals were calculated for the lower and upper 2.5% limits compatible with the observed values. The cumulative risk through age 70 years for developing an incident uterine cancer, assuming the participant underwent RRSO at age 45 years, was estimated by 2 methods, first assuming a constant annual risk and second assuming a constant relative risk compared with SEER rates. In the first approach, annual risk was determined by dividing the number of uterine cancers observed by the number of woman-years of observation. This annual risk was then multiplied by the number of years at risk (ie, 25). In the second approach, age-specific SEER incidence rates were multiplied by the relevant O:E ratio. These absolute risks, starting at age 45 years, were then summed through age 70 years. Analyses stratified by BRCA1/2 mutation type, breast cancer history, and prior tamoxifen exposure were also performed.

For uterine cancer cases from BRCA1+ women, available tumor specimens were analyzed for BRCA1 protein expression, as previously described.²¹ Briefly, the study pathologist assessed immunohistochemical staining intensity relative to an internal positive control. Loss of BRCA1 protein expression was predefined as less than 5% of cells staining with a positive internal control, whereas staining of greater than 10% was considered retention of BRCA1 protein expression.

For uterine cancer cases with available tumor, microdissected tumor and normal DNA were extracted from archival material and amplified using mutation-specific primers. Loss of heterozygosity was determined by comparing fragment distributions between normal and tumor amplicons.²² Loss of heterozygosity was predefined as a reduction of greater than 50% of the wild-type allele at the site of the germline mutation in the tumor sample.

Of 1083 women who met the final eligibility criteria, 627 (57.9%) were BRCA1+, 453 (41.8%) were BRCA2+, and 3 (0.3%) had mutations in both genes. The median age at RRSO was 45.6 (interquartile range [IQR], 40.9-52.5) years and the median follow-up was 5.1 (IQR, 3.0-8.4) years. Seven hundred twenty-seven study participants (67.1%) had a history of breast cancer, and 273 of the 928 women (29.4%) with data available had used tamoxifen (eTable 2 in the Supplement).

During 6377 woman-years of follow-up, 8 women developed uterine cancer (O:E ratio, 1.9; 95% CI, 0.8-3.7; P = .09) (Table 1 and eTable 3 in the Supplement). Two women developed endometrioid carcinoma (1 with International Federation of Gynecology and Obstetrics grade 1 and 1 with grade 2) (O:E ratio, 0.6; 95% CI, 0.1-2.0; P = .88), both within 3 years of RRSO. One woman developed leiomyosarcoma (O:E ratio, 7.1; 95% CI, 0.2-39.4; P = .13) 1.4 years after RRSO. Five serous/serous-like endometrial carcinomas (1 high-grade carcinoma with serous and undifferentiated components; 1 high-grade carcinoma with serous and endometrioid features; 1 carcinosarcoma with a serous epithelial component; 1 serous carcinoma with an undifferentiated component; and 1 pure serous carcinoma) were observed (O:E ratio, 14.8; 95% CI, 4.8-34.6; P < .001), 7.2 to 12.9 years after RRSO.

Five of 627 BRCA1+ women developed uterine cancer (2.38 expected; O:E ratio, 2.1; 95% CI, 0.7-4.9; P = .09) during 3781 woman-years. Four serous/serous-like carcinomas (0.18 expected; O:E ratio, 22.2; 95% CI, 6.1-56.9; P < .001) (Table 2) and 1 sarcoma (0.08 expected; O:E ratio, 12.4; 95% CI, 0.3-69.3; P = .08) developed. With use of these data, the estimated risk for developing serous/serous-like carcinoma through age 70 years for a BRCA1+ woman undergoing RRSO at age 45 years was 2.6% (95% CI, 0.7%-6.8%), assuming a constant annual risk, and 4.7% (95% CI, 1.3%-12.1%), assuming a constant relative risk compared with SEER rates.

Three of 453 BRCA2+ women developed uterine carcinoma (1.91 expected; O:E ratio, 1.6; 95% CI, 0.3-4.6; P = .30) during 2580 woman-years. Two endometrioid cases (1.60 expected; O:E ratio, 1.2; 95% CI, 0.2-4.5; P = .48) and 1 serous/serous-like case (0.16 expected; O:E ratio, 6.4; 95% CI, 0.2-35.5; P = .15) developed.

Four of five serous/serous-like carcinomas occurred in women with prior breast cancer, 3 of whom used tamoxifen (eTable 3 in the Supplement). One woman with a history of breast cancer without tamoxifen use developed leiomyosarcoma. All women with serous/serous-like carcinoma or sarcoma received adjuvant chemotherapy and/or radiotherapy. Two of 5 women with serous/serous-like carcinoma have had disease recurrence, 1.6 and 2.0 years after diagnosis, and one ultimately died of the disease.

Tumor tissue from 3 of 5 serous/serous-like carcinomas and the leiomyosarcoma were available. All of these occurred in BRCA1+ women. When BRCA1 protein expression was examined, all 3 serous/serous-like carcinomas demonstrated loss of protein expression compared with intact internal control (Figure, A-C). The leiomyosarcoma, however, retained BRCA1 protein expression (Figure, D).

Two of 3 available serous/serous-like carcinomas demonstrated loss of the wild-type BRCA1 allele (patients A and B, eFigure in the Supplement). The other available serous/serous-like carcinoma and the leiomyosarcoma retained the wild-type BRCA1 allele (patients C and D, eFigure in the Supplement).

Given that only the serous/serous-like subtype was observed more frequently than expected, we limited the following analyses to the serous/serous-like subtype tumors. We observed 4 serous/serous-like carcinomas in 727 women with prior breast cancer (0.26 expected; O:E ratio, 15.5; 95% CI, 4.2-39.7; P < .001). One serous/serous-like carcinoma was seen in 356 women without prior breast cancer (0.08 expected; O:E ratio, 12.6; 95% CI, 0.3-70.3; P = .08). Three serous/serous-like carcinomas were seen in 273 tamoxifen-exposed women (0.12 expected; O:E ratio, 24.4; 95% CI, 5.0-71.3; P < .001). Two serous/serous-like carcinomas occurred in 655 women without prior tamoxifen use (0.18 expected; O:E ratio, 11.3; 95% CI, 1.4-40.8; P = .01) (Table 2).

These results suggest that BRCA1+ women undergoing RRSO without hysterectomy remain at increased risk for serous/serous-like endometrial carcinoma. The lack of BRCA1 protein expression in all 3 available BRCA1-associated serous/serous-like specimens and the loss of the wild-type BRCA1 allele in 2 of 3 available tumors support the biologic plausibility that the loss of BRCA1 function was important in the tumorigenesis of the serous/serous-like cancers seen in our cohort.

Whether uterine cancer is a BRCA-associated tumor has long been controversial. The Breast Cancer Linkage Consortium showed an increased risk for uterine cancer in BRCA1+ women (relative risk, 2.65; 95% CI, 1.69-4.16; P < .001), but not in BRCA2+ women.^8,9 Similarly, Segev et al¹⁰ found an increased risk for endometrial cancer for BRCA1+ women (standardized incidence ratio,1.91; 95% CI, 1.06-3.19; P = .03). However, this risk may have been limited to tamoxifen-exposed women. Importantly, neither study commented on the histologic subtypes of these cancers.

Several studies have also evaluated the potential association between serous endometrial carcinoma and BRCA mutations. In 2000, Goshen et al¹³ examined 56 patients with serous endometrial carcinoma and did not identify any BRCA mutations. However, only selective founder and protein-truncating mutations in exon 11 of BRCA1 and exons 10 and 11 of BRCA2 were examined. In 2001, Levine et al¹⁴ similarly did not find any founder BRCA mutations in 21 serous/serous-like endometrial carcinomas from individuals of Jewish heritage. However, 2 independent Israeli studies^11,12 since reported that 4 of 31 (12.9%) and 7 of 59 (11.9%) consecutive Jewish patients with serous endometrial carcinoma, respectively, carried a BRCA1 mutation. Also, a recent multi-institutional study identified nonfounder BRCA1 mutations in 3 of 151 unselected patients (2.0%) with serous endometrial carcinoma.²³ When all 5 studies to date are pooled, 3 of 207 serous/serous-like carcinomas in admixed US and Canadian patients (1.4%) and 11 of 111 serous/serous-like carcinomas in US and Israeli Jews (9.9%) had a detectable BRCA1 mutation (eTable 4 in the Supplement). This finding is in contrast to the 0.12% to 0.29% and 1.1% to 1.2% expected mutation rates in these respective populations.²⁴

In the only other prospective study analyzing the risk for uterine cancer after RRSO in BRCA mutation carriers with complete histologic information available, 2 endometrioid carcinomas developed during 6 years of follow-up in 315 BRCA+ women. No serous/serous-like carcinomas or sarcomas were seen.²⁵ However, patients could be ascertained and followed up before genetic testing, which could result in underestimation of risk in aggressive cancers such as serous carcinoma. Furthermore, that study included 70% fewer BRCA1+ women, with a median age at RRSO 3 years younger than in our report, which potentially limits the previous study’s power to detect an increased incidence of serous/serous-like endometrial carcinoma.

In our study, all 3 BRCA1-associated serous/serous-like carcinomas with available tissue showed clear loss of BRCA1 protein expression. In 2 cases, we also demonstrated that loss of the wild-type BRCA1 allele was the likely cause. In the third case, the mechanism for protein expression loss was not elucidated. This result may be analogous to the situation in BRCA-associated serous ovarian cancer, in which 19% to 28% of BRCA-associated ovarian cancers do not demonstrate loss of heterozygosity,²⁶ suggesting other mechanisms of BRCA silencing are important in a significant percentage of BRCA-associated disease.

Recent investigations have proposed that most serous ovarian cancers originate from the fallopian tubes.^27,28 Given this hypothesis, there is speculation that some serous endometrial cancers may be metastases from concomitant tubal primary tumors. However, for most cases with coexisting serous cancers in the fallopian tube and endometrium, the endometrial component appears to be superficial or noninvasive.²⁹ In contrast, all serous/serous-like carcinomas in our series presented with myoinvasive disease no fewer than 7 years after RRSO, arguing against these being tubal metastases. Further, the fallopian tubes and endometrium share an embryological precursor, the paramesonephric (Müllerian) ducts. It is therefore plausible the endometrium may be susceptible to similar BRCA-mediated carcinogenesis as the fallopian tube.

Possible confounding by a history of breast cancer and/or tamoxifen exposure remains a consideration. Several studies have suggested that women with serous endometrial carcinoma are more likely to have a personal or a family history of breast cancer than women without cancer³⁰ or women with endometrioid carcinoma.^31,32 The cause of this association is unclear, but as none of these studies have examined BRCA mutation status, the association may be explained in part by BRCA1 mutations. The relationship between tamoxifen exposure and serous/serous-like endometrial carcinoma is less clear³³; however, recent reports have suggested an association.^34,35 While tamoxifen exposure cannot explain the loss of the wild-type BRCA1 allele seen in 2 of the 3 available serous/serous-like tumors in our series, it may act as a risk modifier in the presence of decreased BRCA1 protein expression, as posited by Wen et al.³⁶ Given this possibility, tamoxifen exposure may account for some of the serous-like carcinoma risk seen in our report.

Although this report is, to our knowledge, the largest prospective study to date, relatively few events were observed. Given this, we have estimated the penetrance of serous/serous-like cancer through age 70 years in women with BRCA1 mutations using 2 approaches. In the first approach, we assumed a constant relative risk of 22.2 compared with SEER rates from ages 45 through 70 years. We also took a more conservative approach and assumed a constant annual risk for serous/serous-like endometrial cancer of 1.06 per 1000 woman-years (4 cases observed in 3781 woman-years). Although future studies will be required to determine which of these approaches is most appropriate, both approaches estimate a risk for serous/serous-like endometrial cancer through age 70 years of at least 2.6%, which is likely clinically important given the high morbidity and mortality rates for serous/serous-like disease.

Misclassification and/or underreporting of rare uterine cancer subtypes in SEER is also possible, which could lead to inflation of our O:E ratios. However, this possibility is unlikely to be a major effect, as 10.3% of uterine cancers reported in SEER are serous/serous-like, consistent with other epidemiologic studies.³⁷ In addition, bias could have been introduced because we excluded cases from SEER with unknown histologic subtypes from our calculated expected rates. Such cases only accounted for 0.77% of the total corpus cancers in SEER and therefore are unlikely to substantially alter our results. Because we only had 1 BRCA2-related serous event in our study, we cannot conclusively comment on whether BRCA2+ women have a significantly increased risk for serous/serous-like endometrial carcinoma, and the answer to this question awaits further studies.

Although abdominal hysterectomy with bilateral salpingo-oophorectomy is clearly associated with increased complications and costs compared with RRSO alone, comparison of complications and costs may be more appropriate when both hysterectomy and RRSO are performed via a minimally invasive (laparoscopic or robotic) approach. In 2 recent analyses of data from the Perspective database (Premier, Inc), 5.3% of 4971 women undergoing laparoscopic hysterectomy had an intraoperative, surgical site, or medical complication.³⁸ This rate was similar to the 6.0% complication rate of 3632 women undergoing laparoscopic oophorectomy.³⁹ Laparoscopic hysterectomy was more expensive to perform, with a median total cost of $6679 compared with $4737 for laparscopic oophorectomy.^38,39 Because the number of hysterectomies necessary to prevent 1 serous cancer case during 25 years of follow-up ranges from 21.2 to 37.9, an incremental cost ranging from $41 170 to $73 601 to prevent 1 serous cancer case would likely be considered reasonable. Other potential long-term complications of hysterectomy include urinary incontinence and pelvic prolapse, although the risk for these seems greatest in women with preexisting pelvic support defects.^40,41 It remains controversial whether hysterectomy has an adverse effect on sexual function.⁴² However, for BRCA1+ women undergoing premenopausal RRSO, the most important factor affecting sexual function, irrespective of uterine management, is almost certainly the induction of surgical menopause. Last, mortality due to hysterectomy is very rare and much lower than would be expected for serous uterine cancer in this cohort, having been reported as 0.03% to 0.06% in 3 population-based studies from Finland, the United Kingdom, and Australia, which together reported on more than 126 000 hysterectomies performed for nonmalignant indications.^43-45

Given the similar surgical risks, very low mortality, acceptable costs, and potential protection against serous-like endometrial cancer, if the present results are confirmed by future studies, hysterectomy with bilateral salpingo-oophorectomy may become the preferred risk-reducing surgical approach for BRCA1+ women. However, even if these results are confirmed, RRSO alone may still have a role for BRCA1+ women if strong reasons exist for uterine retention, such as dense pelvic adhesions or desire for future pregnancy using assisted reproductive approaches. For BRCA1+ women who have already undergone RRSO, the optimal approach is less clear. Whether a 25-year risk for serous/serous-like uterine cancer of 2.6% to 4.7% justifies the risks and costs of a second surgery will need to be addressed by future prospective studies.

Our results suggest that BRCA1+ women are at increased risk for serous/serous-like endometrial carcinoma. Although instability in the estimated magnitude of this risk remains, we believe that the possibility of this cancer should be considered when discussing the advantages and risks of hysterectomy at the time of RRSO in BRCA1+ women.

Corresponding Author: Noah D. Kauff, MD, Clinical Cancer Genetics Program, Duke Cancer Institute, Duke University Health System, Box 3607 DUMC, Durham, NC 27710 (noah.kauff@duke.edu).

Accepted for Publication: April 4, 2016.

Published Online: June 30, 2016. doi:10.1001/jamaoncol.2016.1820.

Author Contributions: Drs Shu and Kauff had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Shu, Pike, Kauff.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Shu, Pike, Jotwani, Kauff.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Shu, Pike, Jotwani, Kauff.

Obtained funding: Garber, Daly, Offit, Domchek, Rebbeck, Kauff.

Administrative, technical, or material support: All authors.

Study supervision: Offit, Domchek, Rebbeck, Kauff.

Conflict of Interest Disclosures: Dr Soslow reports receiving royalties from Cambridge University Press and Springer and serving as a consultant to EMD Serono, Inc. No other disclosures were reported.

Funding/Support: This study was supported in part by grant DAMD17-03-1-0375 from the Department of Defense Ovarian Cancer Research Program (Dr Kauff); grants R01-CA083855 and R01-CA102776 from the National Institutes of Health (NIH) (Dr Rebbeck); Cancer Center Support grants P30 CA008748, P30 CA016520, P30 CA51008, and P30 CA16042 from the NIH, National Cancer Institute; the Prevention, Control, and Population Research Program of Memorial Sloan Kettering Cancer Center; Project Hope for Ovarian Cancer Research and Education; the Eisenberg-Feinstein Fund for Gynecological Cancer Research and Treatment; the Chia Family Foundation; the Andrew Sabin Family Foundation; the Filomena M. D’Agostino Foundation; the National Institute for Health Research (Biomedical Research Centre at the Institute of Cancer Research); the Royal Marsden National Health Service Foundation Trust; Susan G. Komen for the Cure; Basser Center for BRCA; and the Breast Cancer Research Foundation.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.