MHC class II genotypes are independent predictors of anti-PD1 immunotherapy response in melanoma.

Abstract

Immune checkpoint blockade is a highly successful anti-cancer immunotherapy. Both CTLA4 and PD1 checkpoint blockers are clinically available for melanoma treatment, with anti-PD1 therapy reaching response rates of 35-40%. These responses, which are mediated via neoantigen presentation by the polymorphic MHC complex, are hard to predict and the tumor mutation burden is currently one of the few available biomarkers. While MHC genotypes are expected to determine therapy responses, association studies have remained largely elusive. We developed an overall MHC genotype binding score (MGBS), indicative of a patient's MHC class I (MHC-I) and class II (MHC-II) neoantigen binding capacity and solely based on the germline MHC-I (MGBS-I) and MHC-II (MGBS-II) genotypes. These scores were then correlated to survival and clinical responses following anti-PD1 immunotherapy in a previously published dataset of 144 melanoma patients. We demonstrate that MGBS scores are TMB-independent predictors of anti-PD1 immunotherapy responses in melanoma. Opposite outcomes were found for both MHC classes, with high MGBS-I and MGBS-II predicting good and bad outcomes, respectively. Interestingly, high MGBS-II is mainly associated with treatment response failure in a subgroup of anti-CTLA4 pretreated patients. Our results suggest that MGBS, calculated solely from the MHC genotype, has clinical potential as a non-invasive and tumor-independent biomarker to guide anti-cancer immunotherapy in melanoma.