Background: Identifying individuals harbouring germline mutations in hereditary cancer genes provides opportunities for tumour surveillance programs, disease-specific treatment and cascade testing in family members but is reliant on accurate variant interpretation, which may be confounded by imprecise methods for ascribing pathogenicity. Where insufficient evidence supports a definitive classification, variant of uncertain significance (VUS) status is applied, which often leads to clinical uncertainty. Here, population-level genetic data and high-throughput computational tools were employed to aid VUS stratification relevant to monogenic endocrine tumour syndromes.
Methods: Thirteen genes were selected and all rare (allele frequency <0.05%) non-synonymous single nucleotide variants (SNVs) occurring in the GnomAD non-cancer cohort identified (n=134, 187 individuals). Variants were evaluated using the CharGer bioinformatic pipeline, which applies individual ACMG variant interpretation criteria into a single score, facilitating classification into benign, VUS, or pathogenic categories. Variant pathogenicity scores were derived using the ensemble prediction tool REVEL and gene-specific violin plots generated. These were compared to the distribution of REVEL scores for all possible missense SNVs for each gene as well as known pathogenic missense SNVs reported in mutation databases.
Results: High cumulative frequencies of rare missense SNVs were observed in the GnomAD control cohort for each of the 13 genes, with the great majority allocated VUS status by ACMG criteria. REVEL pathogenicity prediction scores for these variants revealed gene-specific distributions. Notably, for the majority of genes (e.g. MEN1, NF1, RET), the control population variants demonstrated negative selection against the most pathogenic REVEL scores, whilst distributions for known pathogenic variants were markedly skewed towards higher scores.
Conclusions: Gene-specific metrics encompassing the cumulative frequency of VUS variants in the background population, together with CharGer and REVEL scores can enhance the risk stratification of SNVs allocated VUS status during genetic testing. This information may be used to develop gene-specific thresholds to guide clinical management.