The Predictive Value of Tumor Mutation Burden on Efficacy of Immune Checkpoint Inhibitors in Cancers: A Systematic Review and Meta-Analysis

Background: Despite an increasing understanding about tumor mutation burden (TMB) in cancer immunity and cancer immunotherapy, the comprehensive cognition between TMB and efficiency of immune checkpoint inhibitors (ICIs) is still lacking. A systematic review and meta-analysis was conducted to evaluate the predictive value of TMB on efficacy of ICIs. Methods: Systematic literature search was conducted on PubMed, EMBASE, Web of Science and Cochrane Library up to June 16, 2019. Pooled odds ratio (OR) of objective response rate (ORR), hazard ratio (HR) of progression-free survival (PFS) and overall survival (OS) were estimated by inverse variance weighted fixed-effects model (I(2) ≤ 50%) or DerSimonian-Laird random-effects model (I(2) > 50%). In addition, heterogeneity analysis, sensitivity analysis, publication bias and subgroup analysis were conducted. Moreover, fractional polynomial regression was conducted to investigate the dose-response relationship between TMB cutoffs and efficacy of ICIs. Furthermore, we assessed ORR by TMB and programmed cell death ligand 1 (PD-L1) expression after layering each other in studies which the two could be both acquired. Results: Three thousand six hundred fifty-seven records were retrieved through database searching, and 29 studies with 4,431 patients were finally included in the meta-analysis. TMB high group had significantly improved ORR (pooled OR 3.31, 95% CI 2.61, 4.19, P < 0.001), PFS (pooled HR 0.59, 95% CI 0.49, 0.71, P < 0.001) and OS (pooled HR 0.68, 95% CI 0.53, 0.89, P = 0.004). Sensitivity analyses illustrated the results were stable, and publication bias was identified in ORR. Subgroup analyses showed the predictive value of TMB was significant in non-small-cell lung cancer (except for the OS) and melanoma. In addition, heterogeneity was substantial in targeted next generation sequencing group but tiny in whole exome sequencing group. Furthermore, TMB and PD-L1 expression were capable to predict improved ORR of ICIs after stratification of each other, with tiny heterogeneity. Conclusions: High tumor mutation burden predicted improved efficacy of immune checkpoint inhibitors in cancers, and targeted next generation sequencing for estimating tumor mutation burden in clinic should be standardized to eliminate heterogeneity in the future. Moreover, tumor mutation burden and programmed cell death ligand 1 expression were independent factors on predicting efficacy of immune checkpoint inhibitors.