Abstract
Variants of Uncertain Significance (VUS) are genetic variants whose association with a disease phenotype has not been established. They are a common finding in sequencing-based genetic tests and pose a significant clinical challenge. The objective of this study was to assess the use of functional data to classify variants according to pathogenicity. We conduct functional analysis of a large set of BRCA1 VUS combining a validated functional assay with VarCall, a Bayesian hierarchical model to estimate the likelihood of pathogenicity given the functional data. The results from the functional assays were incorporated into a joint analysis of 214 BRCA1 VUS to predict their likelihood of pathogenicity (breast cancer). We show that applying the VarCall model (1.0 sensitivity; lower bound of 95% confidence interval (CI)=0.75 and 1.0 specificity; lower bound of 95% CI=0.83) to the current set of BRCA1 variants, use of the functional data would significantly reduce the number of VUS associated with the C-terminal region of the BRCA1 protein by ~87%. We extend this work developing yeast-based functional assays for two other genes coding for BRCT domain containing proteins, MCPH1 and MDC1. Analysis of missense variants in MCPH1 and MDC1 shows that structural inference based on the BRCA1 data set can aid in prioritising variants for further analysis. Taken together our results indicate that systematic functional assays can provide a robust tool to aid in clinical annotation of VUS. We propose that well-validated functional assays could be used for clinical annotation even in the absence of additional sources of evidence.
Highlights
Precision medicine approaches are based on the identification of molecular targets in the tumour or the host that can be used to identify at-risk individuals and inform treatment decisions resulting in improved outcomes
Following the classification scheme proposed by Plon et al.[12] that summarises the posterior probability in favour of a variant’s pathogenicity on a scale of 1 to 5 with specific probability thresholds, we propose using the posterior probability calculation of a variant being pathogenic in the TA assays (PrDel) output by VarCall to generate a functional classification scheme that would classify PrDel o0.001 as fClass 1, 0.001 oPrDel ⩽ 0.05 as fClass 2, 0.05 oPrDel ⩽ 0.95 as fClass 3, 0.95 oPrDel ⩽ 0.99 as fClass 4, and PrDel40.99 as fClass 5
The joint analysis described here provides the basis to assess the extent to which we can use variant annotation in one protein domain (i.e., BRCT domains of the BRCA1 protein) to annotate variants in other genes coding for proteins containing BRCT domains that are critical for the cellular response to DNA damage[25,26] with implications for cancer predisposition (e.g., BRCA1, BARD1 and NBN) and therapy (e.g., PARP1)
Summary
Precision medicine approaches are based on the identification of molecular targets in the tumour or the host that can be used to identify at-risk individuals and inform treatment decisions resulting in improved outcomes. Large initiatives focused on identifying DNA alterations linked to disease risk in germline DNA, and to cancer initiation and progression in somatic (tumour) tissue DNA have offered tantalising evidence that the goal of personalised medicine can be achieved in the near future. Newly discovered germline variants suspected of being pathogenic are assessed by tests applicable to all genes such as segregation analysis, family history, population frequency, loss of heterozygosity analysis and gene-specific tests such as the presence of a microsatellite instability phenotype in tumours. This labour-intensive work is further hampered by low minor-allele frequency in these susceptibility gene alleles.[1]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
