Abstract 6360: A tumor agnostic composite gene expression signature identifies three groups of patients treated with immune checkpoint blockade with distinct clinical outcome

Abstract Background: Immune checkpoint blockade (ICB) therapies have transformed cancer treatment, providing durable responses and survival benefits across diverse cancer types. However, current biomarkers, such as PD-L1 expression and tumor mutational burden (TMB), have limitations in predictive accuracy and routine clinical application. Advances in RNA sequencing have facilitated the development of gene expression signatures (GES) that characterize the tumor microenvironment, including CD8+ T cell infiltration and interferon-gamma pathway activation. Nonetheless, most GES are either tumor type- or ICB-specific and none of them has been implemented to stratify patients in the routine care. Methods: We designed a three-step approach to develop a composite gene expression signature (cGES) predictive of clinical outcome in ICB-treated patients regardless of cancer types. (1) First, we tested 39 previously published GES associated with clinical outcome in patients treated with ICB, and 50 hallmark gene sets (HGS) from the Broad Institute GSEA, for their association with overall survival (OS) in a publicly available pancancer (bladder carcinomas, kidney clear cell carcinomas, skin cutaneous melanomas, and stomach adenocarcinomas) meta-cohort of 913 ICB-treated patients and available at https://cri-iatlas.org/. (2) Second, we applied machine learning cox boost analysis to identify the most predictive transcripts and construct a cGES. (3) Third, we validated cGES in an independent pancancer cohort of 97 patients (19 tumor types) included in our molecular screening program PROFILER (NCT01774409) and treated with ICB. Results: (1) In the meta-cohort, 21 GES and 9 HGS were associated with improved overall survival (OS), allowing to identify 1, 324 genes, while 7 HGS associated with poorer OS led to the identification of 879 genes. (2) Cox-boost analysis retained 22 favorable and 18 unfavorable genes, enabling to define two distinct signatures and stratifying patients into three distinct risk groups. Intermediate- and low-risk groups showed improved OS compared to the high-risk group, with hazard ratios (HR) of 0.50 (95% CI: 0.39-0.63) and 0.20 (95% CI: 0.15-0.29) respectively. (3) Validation in our independent pancancer cohort (n=97) confirmed the cGES predictive value for OS and progression-free survival (PFS). Intermediate- and low-risk groups exhibited respectively better OS (HR 0.72, 95% CI: 0.44-1.18; HR 0.30, 95% CI: 0.13-0.67) and PFS (HR 0.40, 95% CI: 0.23-0.69; HR 0.25, 95% CI: 0.10-0.67) compared to the high-risk group. Notably, in a multivariate analysis accounting for tumor types, our cGES remained strongly associated with better PFS and OS (P<0.001). Conclusion: We developed and validated a robust tumor agnostic cGES predictive of OS and PFS in ICB-treated patients, suggesting its potential utility in clinical practice. Citation Format: Mehdi Lamkhioued, Guillaume Robert-Siegwald, Roxane Pommier, Deborah Tricoli, Camille Brodin, Valery Attignon, Nicolas Gadot, Severine Tabone-Eglinger, Fayette Jerome, Aurelie Swalduz, Isabelle Ray-Coquard, Eve-Marie Neidhardt, Clélia Coutzac, Loic Verlingue, Olivier Tredan, Jean-Yves Blay, Christophe Caux, Alain Viari, Nicolas Poirier, Aurore Morello, Isabelle Girault, Pierre Saintigny. A tumor agnostic composite gene expression signature identifies three groups of patients treated with immune checkpoint blockade with distinct clinical outcome [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 6360.