Evaluating the performance of models for predicting the BRCA germline mutations in Han Chinese familial breast cancer patients

Abstract

Purpose: Established models (Penn, Myraid and BRCApro) are useful of estimating the probability that a person has a BRCA mutation. But the value of these models in Chinese population is unclear. The aim of the study is to evaluate the performance of three models on the assisting in pre-test genetic risk counseling.

Methods: Three risk assessment models, Penn II, Myriad and BRCApro, were applied to 212 familial breast cancer patients who had undergone BRCA1/2 mutation analysis. Sensitivity, specificity, positive and negative predictive values, likelihood ratios and area under the receiver operator characteristic (ROC) curve were calculated for each model.

Results: Myriad showed a better ROC curve than BRCApro either for BRCA1 or BRCA1/2 combination mutation prediction, but BRCApro had a higher positive likelihood ratio when using 10% as the probability threshold. The performance of three models improved when they were evaluated in 66 patients from high risk families, presenting increased ROC and positive likelihood ratio. Especially that of BRCApro for BRCA2, the ROC was increased to 0.716 and its positive likelihood was 5.6.

Conclusion: Three models had the similar impact on the pre-test probability of BRCA mutation. But at a 10% cutoff point, BRCApro had the best BRCA mutation carrier prediction value. The performance of BRCApro for BRCA2 mutation prediction was improved when it was restricted in patients from high risk families.