Limited evidence of tumour mutational burden as a biomarker of response to immunotherapy

Abstract

AbstractCancer immunotherapy by immune checkpoint blockade (ICB) is effective for several cancer types 1, however, its clinical use is encumbered by a high variability in patient response. Several studies have suggested that Tumour Mutational Burden (TMB) correlates with patient response to ICB treatments 2–6, likely due to immunogenic neoantigens generated by novel mutations accumulated during cancer progression 7. Association of TMB and response to checkpoint inhibitors has become widespread in the oncoimmunology field, within and across cancer types 7–11, and has led to the development of commercial TMB-based biomarker platforms. As a result, patient prioritization for ICB based on individual TMB level was recently approved by the FDA 12. Here we revisit the association of mutational burden with response to checkpoint inhibitors by aggregating the largest pan-cancer dataset with more than 2500 ICB-treated patients with sequencing data and clinical annotation. Surprisingly, we find little evidence that TMB is predictive of patient response to immunotherapy. Our analysis suggests that previously reported associations arise from a combination of confounding disease subtypes and incorrect statistical testing. We show that using a TMB threshold for clinical decisions regarding immunotherapy could skew access to treatment for patients who may benefit from these therapies. Finally, we present a simple mathematical model that extends the neoantigen theory, is consistent with the lack of association between TMB and response to ICB and highlights the role of immunodominance. Our analysis calls for caution in the use of TMB as a biomarker and emphasizes the necessity of continuing the search for other genetic and non-genetic determinants of response to immunotherapy.