O impacto das ferramentas in silico e recursos de bioinformática estrutural na determinação da patogenicidade de variantes genéticas encontradas em RAD51D

Abstract

Cancer is one of the main causes of morbidity and mortality. In Brazil, the most common types of cancer are breast, colorectal, prostate, and stomach cancer. The main causes of cancer are genetic and epigenetic alterations in oncogenes, tumor suppressor genes, miRNA genes, among others. At least 291 human genes are implicated in oncogenesis, all of which encode proteins that regulate cell proliferation, DNA repair, differentiation, and cell death. Consequently, the inclusion of genetic testing has become fundamental in the clinical management and genetic counseling of affected individuals and their families. The number of genes included in screening panels for genetic variants in cancer patients has increased and there is a tendency to replace these panels with the exome. However, genetic variants are very common, and it is necessary to evaluate their pathogenicity potential considering the gene, in which the variant was detected, and the phenotype. In this context, Bioinformatics provides many resources for the evaluation of variants, including evolutionary conservation of the residue, impact on protein or RNA structure and function, impact on post-synthesis modifications, etc. However, the number of resources used varies by laboratory. In addition, evidence from Bioinformatics is often not sufficient to allow the classification of a variant, requiring clinical and family data. Consequently, there are many genetic variants classified as of uncertain significance (VUS). The RAD51D gene has recently been implicated in susceptibility to breast and ovarian cancers. The objective of this work was to evaluate the efficiency of a detailed bioinformatics analysis in the reclassification of 20 germline variants identified in the RAD51D gene in Brazilian individuals with breast, ovarian or other cancers. These variants have already been deposited in databases, being classified as VUS, probably benign or pathogenic. In the present study, they were evaluated for allele frequency, impact on the protein (evolutionary conservation of the residue, location in protein functional domains) or on the RNA (splicing and nonsense mediated decay). Algorithms were also run to predict the effect of missense substitutions on tertiary protein structure and function. At the end of the analyses, the variants were reclassified according to the ACMG/AMP criteria. It was possible to reclassify RAD51D c.26G>C; (p.Cys9Ser), RAD51D c.137C>G; (p.Ser46Cys), and RAD51D c.140A>G; (p.Tyr47Cys) as pathogenic/probably pathogenic. RAD51D c.355T>C; (p.Cys119Arg) remains classified as a VUS, although evolutionary conservation in mammals and probable formation of disulfide bonds with Cys117 suggest pathogenicity. The other variants analyzed maintain the current classification as VUS or benign/probably benign. The main difficulty encountered, specifically in relation to the RAD51D protein, was the existence of many isoforms, which make the analysis of evolutionary conservation difficult. These results show that detailed in silico evaluation can help to solve the classification of certain variants or to select which would be priority targets for functional studies or association studies in families.