Abstract
PURPOSE: Identification of the 10% to 15% of patients with ovarian cancer who have germline BRCA1 or BRCA2 mutations is important for management of both patients and relatives. The BRCAPRO model, which estimates mutation likelihood based on personal and family cancer history, can inform genetic testing decisions. This study's purpose was to assess the accuracy of BRCAPRO in women with ovarian cancer. METHODS: BRCAPRO scores were calculated for 589 patients with ovarian cancer referred for genetic counseling at three institutions. Observed mutations were compared with those predicted by BRCAPRO. Analysis of variance was used to assess factors impacting BRCAPRO accuracy. RESULTS: One hundred eighty (31%) of 589 patients with ovarian cancer tested positive. At BRCAPRO scores less than 40%, more mutations were observed than expected (93 mutations observed v 34.1 mutations expected; P < .001). If patients with BRCAPRO scores less than 10% had not been tested, 51 (28%) of 180 mutations would have been missed. BRCAPRO underestimated the risk for high-grade serous ovarian cancers but overestimated the risk for other histologies (P < .001), underestimation increased as age at diagnosis decreased (P = .02), and model performance varied by institution (P = .02). CONCLUSION: Patients with ovarian cancer classified as low risk by BRCAPRO are more likely to test positive than predicted. The risk of a mutation in patients with low BRCAPRO scores is high enough to warrant genetic testing. This study demonstrates that assessment of family history by a validated model cannot effectively target testing to a high-risk ovarian cancer patient population, which strongly supports the recommendation to offer BRCA1/BRCA2 genetic testing to all patients with high-grade serous ovarian cancer regardless of family history.