Models for predicting BRCA1 and BRCA2 mutations in Han Chinese familial breast and/or ovarian cancer patients

Abstract

Our aim was to find an appropriate method to estimate the likelihood that a family history of cancer was a result of a mutation in the BRCA1 or BRCA2 genes. We also compared the performance of the established method with three different methods (Couch, Sh-E and BRCApro) to identify an alternative strategy for genetic council targeted to the specified population. The family history as well as individual information of two hundred unrelated probands who had completed BRCA1 and BRCA2 mutation screening was analyzed to assess the likelihood of a pathogenic mutation. A model was developed by empirical method. The performance of this model was validated in a separate patient cohort compared with BRCApro. Several factors were associated with mutations in univariate analysis and a logistic model was devised to estimate the probability for a proband of harboring a mutation in BRCA1 and/or BRCA2. Using a greater than 10% probability threshold, the highest accuracy was achieved by the established model when compared to other three models, presenting the highest sensitivity, PPV, NPV and area under ROC curve. The empirical model showed a better ROC curve compared to BRCApro in the verification cohort. A probability model targeted to Han Chinese population should be a useful tool in the genetic counseling for the specified ethnic. Its ability to predict BRCA2 mutation carriers needs to be improved.