Co-occurring genomic alterations and immunotherapy efficacy in NSCLC

Abstract

AbstractAn oncogene-centric molecular classification paradigm in non-small cell lung cancer (NSCLC) has been established. Of note, the heterogeneity within each oncogenic driver-defined subgroup may be captured by co-occurring mutations, which potentially impact response/resistance to immune checkpoint inhibitors (ICIs). We analyzed the data of 1745 NSCLCs and delineated the landscape of interaction effects of common co-mutations on ICI efficacy. Particularly in nonsquamous NSCLC, KRAS mutation remarkably interacted with its co-occurring mutations in TP53, STK11, PTPRD, RBM10, and ATM. Based on single mutation-based prediction models, adding interaction terms (referred to as inter-model) improved discriminative utilities in both training and validation sets. The scores of inter-models exhibited undifferentiated effectiveness regardless of tumor mutational burden and programmed death-ligand 1, and were identified as independent predictors for ICI benefit. Our work provides novel tools for patient selection and insights into NSCLC immunobiology, and highlights the advantage and necessity of considering interactions when developing prediction algorithms for cancer therapeutics.