Genes involved in the homologous recombination repair pathway-as exemplified by BRCA1, BRCA2, PALB2, ATM and CHEK2-are frequently associated with hereditary breast and ovarian cancer syndrome. Germline mutations in the loci of these genes with loss of heterozygosity or additional somatic truncation at the wildtype allele, lead to the development of breast cancers with characteristic clinicopathological features and prominent genomic features of homologous recombination deficiency, otherwise referred to as “BRCAness.” Whereas clinical genetic testing for these and other genes has increased the chances of identifying pathogenic variants, there has also been an increase in the prevalence of variants of uncertain significance, which poses a challenge to patient care because of the difficulties associated with making further clinical decisions. To overcome this challenge, we sought to develop a methodology to re-classify the pathogenicity of these unknown variants using statistical modeling of BRCAness. The model was developed with LASSO logistic regression by comparing 116 genomic attributes derived from 37 BRCA1/2 biallelic mutant and 32 homologous recombination-quiescent breast cancer exomes. The model exhibited 95.8% and 86.7% accuracies in the training cohort and the Cancer Genome Atlas validation cohort, respectively. Through application of the model for variant re-classification of homologous recombination-associated hereditary breast and ovarian cancer causal genes and further assessment with clinicopathological features, we finally identified one likely pathogenic and five likely benign variants. As such, the BRCAness model developed from the tumor exome was robust and provided a reasonable basis for variant re-classification.
This article is protected by copyright. All rights reserved.

Author