Immune Checkpoint Blockade (ICB) has revolutionized cancer treatment, however mechanisms determining patient response remain poorly understood. Our recent work has identified inherited variants related to immune genes as a source of inter-individual variability in anti-tumor immunity. Here we used immune eQTLs to investigate individual differences in response to ICB treatment.  We then trained machine learning classifiers to predict ICB response jointly from germline and somatic biomarkers and used Shapley values to interpret the learned model toward uncovering putative mechanisms driving superior outcomes. Patients with higher T follicular helper infiltrates were robust to somatic defects in the class-I Major Histocompatibility Complex (MHC-I). Further investigation uncovered different ICB responses in MHC-I versus MHC-II neoantigen reliant tumors across patients. Despite similar response rates, MHC-II reliant responses were associated with significantly longer durable clinical benefit (Discovery: Median OS=63.6 vs. 34.5 months P=0.0074; Validation: Median OS=37.5 vs. 33.1 months, P=0.040). Characteristics of the tumor immune microenvironment reflected MHC neoantigen reliance, and analysis of immune checkpoints revealed LAG3 as a potential target in MHC-II but not MHC-I reliant responses. This study highlights the value of interpretable machine learning models in elucidating the biological basis of therapy responses.