Predictability of BRCA1/2 mutation status in patients with ovarian cancer: How to select women for genetic testing in middle-income countries

Objectives: To evaluate the accuracy of algorithms for predicting BRCAI/2 germ-line mutation carrier probability, and to identify factors that could improve their performance among Brazilian women with ovarian cancer (OC). Study design: In this cross-sectional study, we enrolled patients (unselected for family history of cancer) undergoing treatment or follow-up for OC in a single centre in Brazil. Clinical and demographic data, including family history of cancer, were obtained. Blood samples were collected for genetic testing. Main outcome measures: The entire coding sequence of BRCAI and BRCA2 was evaluated for mutations. Mutation carrier probability was calculated using BOADICEA, BRCAPRO, Myriad and the Manchester score. Sensitivity, specificity, positive and negative predictive values, and area under the receiver operating characteristic curves (AUC) were calculated for each algorithm. Logistic regression was used to detect additional factors associated with BRCA1/2 status, and these were added to the algorithms before recalculating the AUCs. Results: BRCAI/2 mutations were identified in 19 of the 100 included patients. BOADICEA outperformed other algorithms (sensitivity, 73.7%; specificity, 87.7%; AUC, 0.87, with a threshold of a 10% risk of mutation). Later menarche was associated with the presence of a BRCAI/2 mutation. Although adding age at menarche resulted in a larger AUC for all models, this increase was significant only for the Myriad algorithm. Conclusion: A BOADICEA risk evaluation of 10% or more most accurately predicted BRCA1/2 status, and the inclusion of age at menarche tended to improve the performance of all algorithms. Using these tools could reduce the number of tests, but at the expense of missing a significant proportion of mutation carriers