136P Evaluation of a regularly updated knowledge base and automated variant interpretation tool for somatic mutations detected in solid tumours

Abstract

Routine clinical interpretation of somatic variants from next-generation sequencing (NGS) testing of cancer samples remains challenging due to evolving medical guidelines and increasingly complex multi-variant results. We assessed the ability of a CE-IVD somatic variant interpretation tool, NAVIFY® Mutation Profiler*, to provide accurate and timely clinical content to aid in the interpretation of somatic variants in 3810 samples from The Cancer Genome Atlas (TCGA) across ten solid tumor types. * This product is not for diagnostic purposes in the United States and is not commercially available in the United States. Whole exomes from TCGA projects in lung (LUAD and LUSC), colorectal (COAD and READ), breast (BRCA), melanoma (SKCM), head and neck (HNSC), stomach (STAD), bladder (BLCA), and prostate (PRAD) were analyzed. Mutation Annotation Format (MAF) files were converted to individual Variant Call Format (VCF) files containing SNVs and indels for each case to upload to NAVIFY Mutation Profiler. The software automatically reported tier classifications of variants based on consensus recommendations from AMP, ASCO, CAP, and ACMG. Actionable mutations were interpreted according to a highly curated and regularly updated knowledge base. Over 1.59 million somatic mutations across 20,635 genes were assessed with NAVIFY Mutation Profiler. 78% of cases had variants of strong (Tier I-A or I-B) or potential (Tier II-C or II-D) clinical significance. This percentage of cases with potentially actionable mutations was highest in lung and colorectal cancers (91%) and lowest in prostate cancer (16%). Overall, a third of the cases had Tier I classifications while 9% had potentially actionable variant-variant interactions. The tool also identified appropriate tier classifications by geographic region in accordance with regional medical guidelines. To benchmark NAVIFY Mutation Profiler against other tools, we utilized open-access data from TCGA to assess the yield of potentially actionable annotation and found a higher percentage of cases with clinically significant variant interpretations than previously published methods.