Abstract 7140: Non-synonymous mutations in the IFN-γ pathway are predictive of response to immune checkpoint inhibition in NSCLC

Abstract Introduction: Baseline disruptions in interferon gamma (IFN-γ) signalling enhance tumor immune elimination post-checkpoint blockade in murine models. In the Stand Up to Cancer-Mark Foundation (SU2C-MF) NSCLC clinical cohort, non-synonymous mutations in IFN-γ pathway genes (IFNGR1, IFNGR2, JAK1, JAK2, STAT1, IRF1 & IRF9; n=26, 8%) associated with better outcomes following PD-(L)1 inhibitors (ICI), including improved radiological responses, progression-free survival, and overall survival (OS) (AACR 2024). Here, we aimed to validate these results in an independent dataset and examine predictive vs. prognostic significance. Methods: The Caris Life Sciences (CARIS) dataset includes NSCLC patients with standard-of-care biopsies and multi-omic testing. Outcomes were tracked using claims. IFN-γ pathway disruption was defined as non-synonymous mutations in the genes available from CARIS’s curated whole exome sequencing data (WES) and deep panel of 592 cancer-associated whole-genes: IFNGR1, JAK1, JAK2 & IRF1. Cox proportional-hazards models (CoxPH) assessed OS from ICI initiation. Co-variables had to be dichotomized: age (<65/≥65), PDL1 TPS (0%/>0%), smoking history (non-smoker/ever smoker), and continuous, including TMB, by median. Results: Among 1070 patients with pretreatment WES, ICI treatment, and OS data, 88 (8.2%) had IFN-γ pathway mutations. Disruption was associated with significantly improved OS (HR 0.63, p=0.028*). Median OS for patients with disrupted vs. intact pathways was 60.0 months (95% CI: 27.6-NA) vs. 26.2 months (95% CI: 20.5-29.7). Clinical multivariable CoxPH models incorporating TMB and PDL1 TPS confirmed pathway disruption as independently significant (HR 0.27, p=0.030*); this persisted when including sex, age and smoking (OS: HR 0.66, p=0.05). Biological models, including TMB, PDL1 TPS, and immune biomarkers (CXCL9, CD8A, GZMA, PRF1, IFNG), supported survival benefits (HR 0.65, p=0.046*). Correlation coefficients between pathway disruption and covariates were low (e.g., IFNG, 0.005; PDL1 TPS, -0.04; TMB, 0.14). To assess prognostic effects, in the CARIS dataset, we evaluated analogous patients with NSCLC who had pretreatment WES, no prior or concurrent immunotherapy, but were treated with chemotherapy (disrupted: n=15; intact: n=143). Pathway disruption did not associate with OS [univariable: HR 1.64, p=0.42; multivariable with TMB and PDL1: HR 1.86, p=0.22]. TCGA lung data also failed to suggest a prognostic impact on OS (disrupted: n=40; intact: n=902) [univariable: HR 0.96, p=0.88; multivariable - continuous TMB, histology, stage, purity, age and sex: HR 1.86, p=0.70]. Conclusion: In the SU2C-MF and CARIS NSCLC cohorts, non-synonymous IFN-γ pathway mutations predict improved ICI outcomes, independently of TMB and other variables, but are not prognostic. IFN-γ pathway disruption could guide immunotherapy stratification in NSCLC. Citation Format: Alexander Azizi, Arvind Ravi, Marzia Capelletti, Nicholas Haradhvala, Natalie Vokes, Stephen-John Sammut, Steven Ressler, Milan Radovich, Justin Gainor, Gad Getz. Non-synonymous mutations in the IFN-γ pathway are predictive of response to immune checkpoint inhibition in NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 7140.