Keywords: BRCA2, poly(ADP-ribose) polymerase (PARP) inhibitor, variants of unknown 22 significance (VUSs), companion diagnosis, Bayesian hierarchical model 23 24 ABSTRACT 44Numerous nontruncating missense variants of the BRCA2 gene have been identified, but there 45 is a lack of convincing evidence, such as familial data, demonstrating their clinical relevance 46 and they thus remain unactionable. To assess the pathogenicity of variants of unknown 47The BRCA1 and BRCA2 (BRCA) genes encode key proteins in the homology-directed DNA 60 break repair (HDR) pathway, and their inactivation predisposes individuals to cancer 61 development 1 . Germline loss-of-function variants in BRCA markedly increase the risk of early-62 onset breast and ovarian cancer; in such cases, prophylactic oophorectomy and mastectomy 63 and genetic testing for at-risk relatives must be considered 2, 3, 4 . Tumors with pathogenic 64 mutations within BRCA and defective HDR have been shown to be particularly sensitive to 65 platinum-based chemotherapies and PARP inhibitors, the efficacy of which is mediated 66 through synthetic lethality in cancer cells with BRCA loss-of-function 5, 6 . 67
68The American College of Medical Genetics and Genomics (ACMG) standards and guidelines 69 for the interpretation of sequence variants recommend a process for variant classification based 70 on criteria using population, computational, functional, and segregation data 7 . Family-based 71 studies including a multifactorial model of pathology, a cosegregation profile, and the 72 cooccurrence and family history of cancer may exemplify the most reliable methods for 73 classifying BRCA2 gene variants 8, 9 . Nonsense or frameshift mutations within the coding exons 74 of BRCA markedly alter the structures of the protein products and are presumed to confer loss-75 of-function. However, the vast majority of missense variants are individually rare in both 76 general populations and cancer patients, and case-control studies may not have sufficient 77 statistical significance to classify these variants as pathogenic or benign 10, 11, 12 . No current in 78 silico computational prediction algorithm for missense variants is accurate enough when used 79 alone 13 . Thus, the functional evaluation of missense variants of unknown significance (VUSs) 80 is urgently required to improve the interpretation of variants identified by genetic testing and 81 to support clinical decision-making for their carriers 14 . 82Page 5 of 63 While thousands of BRCA1 VUSs have been assessed by recently developed high-throughput 83 functional assays 15, 16, 17 , a few hundred BRCA2 variants have been evaluated by a conventional 84 functional assay for BRCA2, the HDR assay 18, 19 . The HDR assay has three issues requiring 85 improvement: (i) the throughput is quite low, (ii) it uses a hamster cell line with transient 86 expression of BRCA2 cDNA, and most importantly, (iii) it can evaluate only variants in the 87 DNA-binding domain 14, 20 . To overcome these limitations, we propose herein a high-8...