High correlation between TMB, expressed TMB, and neoantigen load using tumor: Normal whole exome DNA and matched whole transcriptome RNA sequencing.

Abstract

e15238 Background: Tumor Mutation Burden (TMB) has emerged as a biomarker of response to immune checkpoint inhibitor therapy, independent of PDL1 expression. The optimal cutoff for high versus low TMB is unknown and neoantigen load has been proposed as an alternative biomarker. We have previously described neoantigen silencing at the RNA level as potential means of immune escape. We hypothesized that expressed TMB (eTMB) and neoantigen load would differ from TMB. Methods: Using data from matched tumor:normal DNA, whole exome sequencing at 150x average depth, and whole RNA transcriptome at ~200x106 reads per tumor from a commercial database of 1395 heterogeneous cancer patient samples, we measured TMB and eTMB ( > 2 alternate RNA reads supporting). Neoantigen load was measured as the count of unique 9mer peptides resulting from expressed somatic-specific variants with predicted binding affinity < 500nM by NetMHC. Results: High correlation between TMB and eTMB was seen (R2 = 0.94). No significant differences were seen between TMB, eTMB, and predicted neoantigen load. Using a cutoff of 200 non-synonymous variants (Rizvi 2015) to define TMB high versus highest quintile (TMB-20) (Samstein 2019) we saw high agreement of the 2 methods to assign TMB status as high v low. In the TMB-20 group with low TMB ( < 200), most often reclassified were soft tissue sarcoma (18%) and pancreatic cancers (17%). Conclusions: Surprisingly, eTMB, TMB-20, and neoantigen load showed no significant difference from TMB using 200 variants as a cutoff for high. Tissue-specific TMB may be useful in patient with Sarcoma and Pancreatic cancers. Additional data inputs such as microbiome, chemokine expression, and TME cell phenotyping may be required to improve upon TMB as a biomarker of immunotherapy response.