Functional Analysis of Missense Variants in the Putative Breast Cancer Susceptibility Gene XRCC2.

XRCC2 genetic variants have been associated with breast cancer susceptibility. However, association studies have been complicated because XRCC2 variants are extremely rare and consist mainly of amino acid substitutions whose grouping is sensitive to misclassification by the predictive algorithms. We therefore functionally characterized variants in XRCC2 by testing their ability to restore XRCC2-DNA repair deficient phenotypes using a cDNA-based complementation approach. While the protein-truncating variants p.Leu117fs, p.Arg215*, and p.Cys217* were unable to restore XRCC2 deficiency, 19 out of 23 missense variants showed no or just a minor (<25%) reduction in XRCC2 function. The remaining four (p.Cys120Tyr, p.Arg91Trp, p.Leu133Pro, and p.Ile95Leu) had a moderate effect. Overall, measured functional effects correlated poorly with those predicted by in silico analysis. After regrouping variants from published case-control studies based on the functional effect found in this study and reanalysis of the prevalence data, there was no longer evidence for an association with breast cancer. This suggests that if breast cancer susceptibility alleles of XRCC2 exist, they are likely restricted to protein-truncating variants and a minority of missense changes. Our study emphasizes the use of functional analyses of missense variants to support variant classification in association studies.