61 Tumor mutational burden assessments by two commercial targeted sequencing assays

Background Tumor mutational burden (TMB) is a key biomarker for immune checkpoint inhibitor across several cancer types. While TMB as calculated from whole exome sequencing of the tumor tissue is still the gold standard, enabling TMB in clinical labs requires targeted sequencing panels for faster turnaround time and low input requirements. Herein, we assess two commercially available, targeted sequencing (research-use-only) TMB assays for the possibility of offering in the Medgenome labs. Methods Two assays, Oncomine Tumor Mutation Load Assay (or Oncomine) by Thermo Fisher Scientific and QIAseq Tumor Mutational Burden Panel (or QIAseq) by Qiagen, were studied. One negative control (NA12878), five positive control (A549, lung; T47D, breast; SKMEL2, skin; HCT-15, large intestine; HCT116, large intestine) cell lines, and 18 FFPE (13 colon, 1 lung, 1 testicular, and 1 oral cancer; 2 healthy) samples were ran on both assays. Sample QC was performed through measuring DNA fragmentation on TapeStation and concentration on Qubit. Failure rates on FFPE samples were investigated. TMB values by both assays were compared on all samples, as well as with expected TMB on cell line samples. Expected TMB on the negative control was considered zero; expected TMB for positive cell lines was calculated by restricting somatic mutations (from cBioPortal.com) to each panel, normalizing by panel size, and averaging. TMB values of 3 samples with known MSI were evaluated and signature patterns of relatively high TMB samples were studied. Results On cell line samples, high correlation (r2 = 0.9994) was observed between TMB values by both assays. TMB values were consistently zero on negative control by both assays. Both assays estimated lower than expected TMB on positive control samples. 6/18 FFPE samples failed on both assays, with Oncomine’s error mode was high deamination (i.e., number of C:G>T:A mutations at low allelic frequency) and QIAseq’s was low confidence (i.e., 50 by both assays, but 14 by QIAseq and 33 by Oncomine on one MSI FFPE sample. Four out of five FFPE samples with > 25 TMB by both assays displayed MSI signature patterns from COSMIC or incorporated a pathogenic mutation in MLH1 gene. Conclusions Preliminary analyses showed comparative accuracy and failure rates on FFPE samples. Future analyses will aim at comparison with WES based TMB on reference cell line and FFPE material.