Abstract
Tumor mutational burden (TMB) is considered a prototypic feature of tumor foreignness and has been established as a tumor-agnostic FDA-approved biomarker at a threshold of 10 mut/Mb for immune checkpoint inhibitors (ICI). Despite its clinical utility as a companion diagnostic for ICI across cancers, a high TMB does not consistently predict response due to technical and biological limitations. Tumor heterogeneity and purity, blood versus tissue sampling, variation in next-generation sequencing, and algorithmic evaluation attenuate the predictive value of TMB. In addition to technical standardization and moving beyond TMB as a numeric or binarized value, it is of paramount importance to consider the underlying biology and the differential contribution of mutation subsets to tumor foreignness and immunogenicity. The importance of consideration of mutations within the overall TMB that are unlikely to be immunoedited together with the density of immunogenic "quality" mutation-associated neoantigens introduces the concept of biological calibration of TMB that may enhance its clinical utility. Mutagenic processes such as microsatellite instability and ultra-mutation and cancer lineage-dependent co-mutation patterns also represent biological modifiers that enable the interpretation of the overall TMB in different contexts. In this perspective, we dissect TMB on a biological and technical level, followed by a critical assessment of the predictive role of TMB in capturing ICI response in the setting of clinical trials across human cancers. The standardization of technical methodologies, together with the interpretation of TMB on the basis of the tumor genomic landscape, represents key steps toward maximizing the predictive value of TMB for cancer immunotherapy.