Abstract 735: Immunologically active phenotype by gene expression profiling predicts clinical benefit from PD-1/PD-L1 inhibitors in a pancancer cohort

Background: Our team recently developed a scoring system called the HOT-score, derived from an unsupervised clustering analysis of 421 HPV-negative head and neck squamous cell carcinomas (HNSCC) from TCGA. The HOT-score is based on a 27-gene signature associated with cytotoxic CD8+ T cell activation, interferon-gamma signaling, and IDO1 expression. These genes were identified through differential expression analysis. The HOT-score is computed as an enrichment score (ES) using Gene Set Variation Analysis (GSVA), providing a bimodal ES distribution ranging from -1 to 1. Tumors are classified either as "hot" phenotypes (ES score > 0) or "cold" phenotypes (ES score < 0). We have previously shown an association between the HOT-score and “hot”/”cold” phenotypes with overall survival (OS) and progression-free survival (PFS) in retrospective cohorts of patients with HNSCC (n=102) and non-small cell lung cancer (NSCLC, n=82) treated with immune checkpoint blockade (ICB). Herein, we evaluated the HOT-score's association with clinical outcomes across a broader range of cancer types in the CRi-Atlas cohort. Methods: We analyzed pre-treatment tumor samples from the CRi-Atlas cohort (available at https://cri-iatlas.org/) that included cancers for which ICB therapies are approved. Inclusion criteria required samples to have annotations for progression-free survival, overall survival, or response evaluation by RECIST criteria. Associations with survival outcomes were assessed using Cox proportional hazards modeling, and comparisons of HOT-scores by response categories were performed using the Wilcoxon test. Results: The cohort consisted of 348 bladder carcinomas, 207 kidney clear cell carcinomas, 313 skin cutaneous melanomas, and 45 stomach adenocarcinomas. HOT-scores, analyzed as continuous variables, were significantly lower in patients with progressive disease compared to those with stable disease (P=3.6e-05), partial response (P=3.3e-07), and complete response (P=5.5e-07). HOT-scores were strongly associated with better survival outcomes, including progression-free survival (PFS; n=489, HR 0.67; 95% CI [0.53-0.85]) and overall survival (OS; n=913, HR 0.51; 95% CI [0.41-0.63]). Tumors with "hot" phenotypes exhibited improved survival outcomes compared to those with "cold" phenotypes, with PFS (HR 0.77; 95% CI [0.67-0.97]) and OS (HR 0.54; 95% CI [0.43-0.67]). Notably, in a multivariate analysis accounting for tumor types, both HOT-scores and "hot" phenotypes remained strongly associated with better PFS and OS (P<0.001). Conclusion: The HOT-score is associated with favorable clinical outcomes, including PFS, OS, and RECIST response, across multiple cancer types. Prospective studies are needed to validate its utility for routine clinical implementation. Citation Format: Mehdi Lamkhioued, Foy Jean-Philippe, Aurelie Swalduz, Sandra Ortiz-Cuaran, Andy Karabajakian, Kim-Arthur Baffert, Lucas Michon, Adrien Buisson, Maurice Perol, Sylvie Lantuejoul, Loic Verlingue, Jerome Fayette, Pierre Saintigny. Immunologically active phenotype by gene expression profiling predicts clinical benefit from PD-1/PD-L1 inhibitors in a pancancer cohort [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 735.