A Computational Method to Classify Variants of Uncertain Significance Using Functional Assay Data with Application to BRCA1

Abstract

Background Besides revealing cancer–predisposition variants or the absence of any changes, genetic testing for cancer predisposition genes can also identify variants of uncertain clinical significance (VUS). Classifying VUSs is a pressing problem as ever more patients seek genetic testing for disease syndromes and receive non–informative results from those tests. In cases like the breast–ovarian cancer syndrome where prophylactic options can be severe and life changing, having information on the disease relevance of the VUS that a patient harbors can be critical. Methods We describe a computational approach for inferring the disease relevance of VUSs in disease genes from data derived from an in vitro functional assay. It is based upon a Bayesian hierarchical model that accounts for sources of experimental heterogeneity. Results The functional data correlate well with the pathogenicity of BRCA1 BRCT VUSs, thus providing evidence regarding pathogenicity when family and genetic data are absent or uninformative. Conclusions We demonstrate the utility of the model by using it to classify 76 VUSs located in the BRCT region of BRCA1. The approach is both sensitive and specific when evaluated on variants previously classified using independent sources of data. While the functional data are very informative, they will need to be combined with other forms of data in order to meet the more stringent requirements of clinical application. Impact Our work will lead to improved classification of VUSs and will aid in the clinical decision making of their carriers.

Reclassification of BRCA1 and BRCA2 variants of uncertain significance: a multifactorial analysis of multicentre prospective cohort

“…The Leu1780 residue is located on the α-helix of the BRCT domain, involved in the folding of the BRCT domain, and the amino acid substitution from leucine to proline destabilises the structure 15 17. For the variant BRCA1 c.5089T>C (p.Cys1697Arg), our results were comparable to those obtained using a computational approach by Iversen et al , who reported a posterior probability for this variant of >0.99 18. This variant also destabilises the BRCT structure via the substitution of a charged residue arginine in the protein core 19 .…”

Probing Structure-Function Relationships in Missense Variants in the Carboxy-Terminal Region of BRCA1

“…Note that because this tool uses a Bayesian approach every variant, even the 82 variants assessed previously [18] are re-evaluated given the new data. This analysis (Figure 5) reveals interesting insights about the architecture of the C-terminus.…”

Collective evidence supports neutrality of BRCA1 V1687I, a novel sequence variant in the conserved THV motif of the first BRCT repeat

“…The usefulness of indirect evidence contributed by sources of data other than genetic (e.g., in silico and in vitro analyses) to achieving VUS classification is a well-accepted notion. Recently, substantial progress has been made [5] to meet the requirement of rigorous validation, which is mandatory to warrant integration of functional assay results into VUS classification models. In this context, low-throughput analyses like ours, focused on a single or a few variants and aimed at elucidating yet unclear structure-function relationships of BRCA1, might provide valuable qualitative evidence to complement scanty genetic information and, hence, contribute to VUS categorization.…”

“…Although genetic evidence remains the gold standard for variant classification, the inherent rarity of individual VUS and consequent scarcity of epidemiological data often limit the statistical power of the analysis thus impeding an accurate estimate of disease causality. Ongoing efforts are aimed at widening the range of information evaluated in the model by including data derived from in silico and in vitro analyses [2,5].…”