Three-Fold Overestimation Of Tumor Mutation Burden Using 248 Gene Panel Versus Whole Exome.

12117 Background: Next generation sequencing (NGS) Gene panel testing is used to imputed tumor mutational burden (iTMB) and has shown rough correlation with TMB derived from whole exome sequencing (WES). TMB is used to estimate immune checkpoint inhibitor (ICT) response based on potential neoantigen load. We hypothesized that actual TMB (aTMB), consisting of mutations across the exome, and expressed TMB (eTMB), consisting of expressed genes, would differ substantially from iTMB. Methods: Retrospective analysis of a database from a commercial DNA tumor:normal and RNAseq platform was carried out. We analyzed 890 clinical samples composing of both primary and metastatic disease by whole genome sequencing (WGS) or WES and RNA sequencing (RNA-Seq), and compared true tumor mutational burden to a predicted tumor mutational burden from a list of 248 genes thought to drive cancer. (COSMICv76) Results: Estimated tumor mutational burden based only on the list of 248 genes had an average of 15.79 mutations per megabase (14.16--17.43, 95% CI) whereas WGS/WES derived TMB had an average of 5.09 mutations per megabase of coding DNA (4.22--5.96). The relationship between gene panel size, aTMB and eTMB will be presented. Conclusions: In this retrospective analysis using a 248 gene list as a panel to impute TMB, we observed a roughly 3-fold over-estimate of TMB. This may impact ICT prescription and expectation of clinical benefit.