Comparison of platforms for determining tumor mutational burden (TMB) from blood samples in patients with non-small cell lung cancer (NSCLC)

Abstract

Background: TMB has been associated with response to immune checkpoint inhibitors in clinical studies and has been investigated as a biomarker for first-line (1L) treatment with nivolumab + ipilimumab (N+I) in patients with NSCLC. TMB can be measured by whole exome sequencing (WES) or estimated using targeted gene panels. Clinical implementation of TMB assays can be facilitated by standardization of methods and concordance across platforms. We compared TMB results in commercial and clinical samples from patients with NSCLC using WES, the in vitro diagnostic FoundationOne CDx™ (F1CDx), and the research use only assay TruSight™ Oncology 500 (TSO500). Methods: TMB was evaluated in serial, formalin-fixed, paraffin-embedded sections from 98 procured NSCLC samples using WES (captures coding regions from all genes; ∼30 megabases [Mb]), F1CDx (324 genes; ∼0.8 Mb), and TSO500 (523 genes; ∼1.3 Mb). TMB scores were reported as mutations/Mb (mut/Mb) and compared by Spearman's correlation. Association of TMB with response was assessed in a subset of 59 patients who had NSCLC treated with 1L N+I (CheckMate 568 [NCT02659059]) and had data from both F1CDx and TSO500. Results: TMB estimates from F1CDx and TSO500 correlated highly with WES (r = 0.83 and 0.86, respectively; P < 0.001 for both). Accounting for different calculation methods used for each assay, the clinically relevant cutoff for high TMB (TMB-H) for 1L N+I in NSCLC patients of 10.0 mut/Mb by F1CDx projected to 7.7 mut/Mb (95% CI, 7.1–8.3) by WES and 12.3 mut/Mb by TSO500. Overall percentage agreements around these cutoffs were 86% for WES vs F1CDx and 83% for WES vs TSO500. TMB estimates from F1CDx and TSO500 correlated well (r = 0.84; P < 0.001). In patients treated with N+I, 8/10 responders and 16/38 nonresponders were identified as TMB-H by F1CDx and 7/10 responders and 12/38 nonresponders by TSO500. Conclusions: TMB data from WES, F1CDx, and TSO500 correlated well in commercial and clinical NSCLC samples. More responders than nonresponders to N+I were identified as TMB-H but with similar frequencies when assessed by F1CDx or TSO500. This study demonstrates the feasibility of TMB harmonization and bridging of TMB cutoffs across testing platforms. Clinical trial identification: (NCT02659059). Editorial acknowledgement: Amrita Dervan, PhD, and Jay Rathi, MA, of Spark Medica Inc, funded by Bristol-Myers Squibb. Legal entity responsible for the study: Bristol-Myers Squibb. Funding: Bristol-Myers Squibb. Disclosure: J. Baden: Shareholder/Stockholder/Stock options, Full/Part-time employment: Bristol-Myers Squibb; Shareholder/Stockholder/Stock options: Johnson & Johnson. C. Zhao: Shareholder/Stockholder/Stock options, Full/Part-time employment: Illumina Inc. J. Pratt: Shareholder/Stockholder/Stock options, Full/Part-time employment: Bristol-Myers Squibb. S. Kirov: Shareholder/Stockholder/Stock options, Full/Part-time employment: Bristol-Myers Squibb. S. Pant: Shareholder/Stockholder/Stock options, Full/Part-time employment: Bristol-Myers Squibb. A. Seminara: Shareholder/Stockholder/Stock options, Full/Part-time employment: Bristol-Myers Squibb. G. Green: Shareholder/Stockholder/Stock options, Full/Part-time employment: Bristol-Myers Squibb. S. Bilke: Shareholder/Stockholder/Stock options, Full/Part-time employment: Illumina Inc. I. Deras: Shareholder/Stockholder/Stock options, Full/Part-time employment: Illumina Inc; Shareholder/Stockholder/Stock options: Bristol-Myers Squibb. D.A. Fabrizio (co-senior author): Shareholder/Stockholder/Stock options: Roche; Full/Part-time employment: Foundation Medicine Inc. T. Pawlowski (co-senior author): Shareholder/Stockholder/Stock options, Full/Part-time employment: Illumina Inc.