Large scale case-control analyses of rare variant data; application to BRCA1 and BRCA2.

Abstract

Background/Objectives. Clinical genetic testing of high-risk cancer predisposition genes often leads to the identification of variants of uncertain significance, which complicate management of carriers and families. Case-control data can be used to inform variant interpretation. We developed a novel rare variant casecontrol likelihood ratio (ccLR) method, that incorporates gene- and age-specific penetrance, which we applied to a large breast cancer case-control dataset. Methods. Our method was used for the analysis of 1553 rare BRCA1 and BRCA2 variants (with >=2 carriers and MAF < 0.001), identified by panel sequencing of 35,674 breast cancer cases and 32,532 controls from the Breast Cancer Association Consortium (BCAC). Resulting LRs were categorised as weights for (LR >= 18.70) or against (LR < 0.48) pathogenicity following recommendations of the American College of Medical Genetics and Genomics/ Association for Molecular Pathology (ACMG/AMP) variant classification framework. Results. Approximately 10% of variants (PVS1 ACMG/AMP criterion) were predicted pathogenic. For the remainder 1,399 variants with unknown consequence (missense, intronic, UTRs, inframe indels, synonymous), LRs provided evidence in favour of pathogenicity for 16 variants (5 intronic and 11 missense) and evidence against pathogenicity for 952 variants. For a set of missense variants with (likely) benign/pathogenic ClinVar class, we observed a 97.8% consistency with the calculated casecontrol LRs. Conclusion. Our analysis provided evidence relevant for variant classification for 70% of BRCA1 and BRCA2 rare variants of unknown consequence. We demonstrate the utility of the ccLR method to provide weighted information to aid interpretation of a large proportion of rare variants identified by sequencing of casecontrol datasets.