Intratumoral immune heterogeneity drives divergent outcomes to PD-(L)1 blockade in lung cancer

Clinical outcomes with immune checkpoint inhibitors (ICIs) in non-small cell lung cancer (NSCLC) range from durable disease control to rapid progression. Current biomarkers have limited ability to predict these divergent outcomes. Experimental design: Spatially-resolved immune subtyping in the TRACERx421 multiregion cohort was used to assess intratumoral immune heterogeneity (ITIH), defined as coexisting immune-enriched and immune-depleted regions. A computational model was used to infer ITIH from single samples, classifying tumors as homogenously immune-enriched (Hom-IE), heterogeneously immune-enriched (Het-IE), heterogeneously immune-depleted (Het-D), and homogenously immune-depleted (Hom-D). ITIH was validated by digital pathology and evaluated in three independent cohorts (n=1,085) of advanced NSCLC patients treated with ICIs. Results: In TRACERx421, ITIH was associated with differences in immune programs among tumors otherwise classified within the same immune subtype. Single-sample ITIH inference recapitulated these patterns, as supported by gradients of tumor-infiltrating lymphocytes detected via digital pathology. Across cohorts, Hom-IE tumors demonstrated longer OS compared to other groups (34.6 vs. 12.1-16.6 months in cohort A; 21.9 vs. 8.4-13.3 months in cohort B; 22.6 vs. 11.3-13.4 months in cohort C). Hom-IE tumors were associated with longer OS regardless of PD-L1 status. Consistently, among PD-L1 low/negative tumors, Hom-IE tumors showed significantly longer OS compared to the non-Hom-IE counterpart. Conversely, the detrimental effects of KEAP1STK11 mutations persisted in the Hom-IE background. Conclusions: These findings demonstrate that ITIH can be assessed from individual tumor samples, identifies a microenvironment state associated with long-term benefit from immunotherapy, and provides an additional layer of resolution to established and emerging biomarkers.