Abstract 4889: Comparison of tumor mutational burden using the Ion Oncomine™ TML and FoundationOne™ assays with routine clinical FFPE tissue samples to predict durable clinical benefit in lung cancer and melanoma patients - a multivariate analysis integrating PD-L1 and CD8+ evaluation

Background and objective: Tumor mutational burden (TMB) has emerged in recent clinical trials as a novel biomarker for the stratification of non-small cell lung cancer (NSCLC) patients treated with anti-PD1/PD-L1 antibodies. TMB in trials has been mainly assessed using the FoundationOne™ test (Foundation Medicine, Cambridge, MA, USA), which requires samples to be sent to the authorized testing center for the analysis. Therefore, our aim was to compare an inhouse assessment of TMB using the Ion Oncomine™ TML (Thermo Fisher Scientific, Waltham, MA, USA) and the FoundationOne™ tests. Furthermore, we evaluated PD-L1 expression in tumor cells and the level of tumor infiltrating CD8+ lymphocytes by immunohistochemistry to find the best combination of predictive biomarkers. Materials and Methods: Late stage lung adenocarcinoma (51) and melanoma (25) patients with first- or second-line treatment with checkpoint inhibitors were included prospectively and selected randomly. Tissue FFPE sections were sent out for TMB and genetic assessment using the FoundationOne™ test, and parallel in-house analysis using the Ion Oncomine™ TML assay and the Ion AmpliSeq™ Cancer Hotspot Panel v2. IHC for PD-L1 (22C3 pharmDx assay; Dako Agilent, Santa Clara, CA), in addition to CD8+ lymphocyte expression analyses (clone SP57; Roche Ventana, Tucson, AZ) were performed. Results: TMB was assessed by the Ion Oncomine™ TML assay in comparison to FoundationOne™, demonstrating a high correlation in melanoma (R² = 0.95), but a lower correlation in NSCLC patients (R² = 0.88). Prediction of a durable clinical response by receiver operating characteristic (ROC) was comparable between Oncomine™ TML (AUC = 0.70) and FoundationOne™ (AUC = 0.74) and also similar to the CD8+ score (AUC = 0.76) and PD-L1 expression (AUC = 0.72). Combining different biomarkers did not improve the predictive value. In addition, detection of actionable mutations exhibited a good correlation between the different gene panels. TMB analysis using the Ion Oncomine™ TML assay was initially ineffective as elevated deamination rates from formalin fixed samples interfered with the analysis but this could be successfully overcome with bioinformatic filtering and UDG repair. Stringent sample requirements for FoundationOne™ testing prevented the analysis in 22% and 8% of lung cancer and melanoma samples, respectively. Conclusion: TMB can be assessed using the Ion Oncomine™ TML assay since its predictive value is comparable to that