e21178 Background: ICIs are cornerstones of therapy for advanced NSCLC. Despite dramatic and sometimes durable responses to therapy, most patients (pts) either (i) do not respond to therapy (intrinsic resistance), or (ii) subsequently progress after initial clinical benefit (acquired resistance). Currently, insights into the molecular mechanisms of resistance to ICIs in NSCLC are lacking. Methods: To investigate clinical and molecular features of pts progressing on ICIs, we identified pts who underwent repeat tumor biopsies on and/or after disease progression on ICIs and were included in the Stand Up 2 Cancer (SU2C)/Mark Foundation multi-institutional cohort. Biopsy specimens underwent whole-exome sequencing (WES) and/or whole transcriptome sequencing (RNAseq). Results: We identified 37 pts who underwent a total of 47 repeat biopsies on or after ICIs. Six pts underwent multiple post-ICI biopsies (range 2-4). Twenty-five pts (68%) received PD-(L)1 inhibitor monotherapy, 6 (16%) received PD-(L)1 plus CTLA-4 inhibitors, and 6 (16%) received other PD-1 inhibitor-based combinations. Overall, the objective response rate was 46% among pts undergoing repeat biopsies (complete response 2 [5%], partial response 15 [41%], stable disease 14 [38%], progressive disease 5 [14%] and not evaluable 1 [3%]). Median progression-free survival (PFS) was 8.1 months. In total, pre-ICI biopsy specimens were available in 20 pts. WES and RNAseq were performed on 67 and 44 specimens, respectively. Median tumor mutation burden (TMB) in pre-ICI specimens was 5.0 mutations/Mb versus 4.9 mutations/Mb in post-ICI specimens ( p= 0.3, Mann-Whitney U test). Among 20 paired pre/post-ICI specimens, there was no significant difference in TMB (pre-treatment median 3.9 mutations/Mb; post-treatment median 4.3 mutations/Mb; p= 0.7, Wilcoxon signed-rank test). One pt with a complete response acquired a nonsense mutation in B2M, and one pt with a partial response acquired a nonsense mutation in JAK1. Among 10 paired pre/post-ICI specimens that underwent RNAseq, we observed significant decreases in granzyme B and perforin in post-ICI specimens ( p= 4×10-5 and p= 2×10-3, respectively, limma-voom analysis). Conclusions: Genomic alterations impairing antigen presentation (e.g., B2M) or immune activation (e.g., JAK1) may enable resistance to ICIs in a small subset of cases. However, the majority of repeat biopsies obtained from pts progressing on ICIs lacked clear genetic mediators of resistance, suggesting the presence of additional tumor-intrinsic and/or tumor-extrinsic factors underlying resistance to ICIs in NSCLC.