Tumor methylation subtypes as a predictor of clinical outcome to immunotherapy in pleural mesothelioma patients from the NIBIT-EPI-MESO study.

8052 Background: Pleural mesothelioma (PM) is an aggressive malignancy with a poor prognosis. The clinicalefficacy of standard therapy with immune checkpoint inhibitors (ICI) is limited and heterogeneous acrossPM subtypes. Tumor-intrinsic characteristics (i.e., inflammatory phenotype, molecular features, DNAmethylation) may influence immune responsiveness, but predictive biomarkers of ICI therapy efficacy inPM are still lacking. Methods: NIBIT-EPI-MESO is a retrospective, multicenter study, sponsored by theNIBIT Foundation, evaluating biological correlates of clinical outcomes in PM patients (pts) treated withICI (i.e., anti-CTLA-4 plus anti-PD-1, anti-CTLA-4 plus anti-PD-L1, or anti-CTLA-4 monotherapy). Pre-ICI therapy FFPE tumor samples were analyzed by RRBS methylation (n = 83 pts) and RNA-seq (n = 82pts), with methylation subtypes defined by consensus clustering of the top 1% most variable CpGs.Tumor microenvironment (TME) was characterized by multiplex immunofluorescence analysis of CD4,CD8, CD20, CD68, CD163 (n = 35 pts). Integrated multi-omics analyses were used to associate tumorbiology with clinical outcome of PM pts. Results: Unsupervised methylation profiling identified four PMsubsets with increasingly global DNA methylation levels: demethylated (DEM), LOW, intermediate (INT),and CpG island methylator phenotype (CIMP). Methylation subtypes were significantly associated withresponse to ICI, with LOW/DEM enriched among responder (R) pts and INT/CIMP in non-R pts (p =0.002); no association of response to ICI was found with PM histotype (p = 0.33). The LOW subsetexhibited the longest median overall survival (mOS) and the highest 3-year OS rate, expressed genesinvolved in pathways associated with innate and adaptive immune responses, and showed an “inflamed”TME (i.e., CD8+ T cells, CD20+ B cells). Conversely, the CIMP subtype had the shortest mOS and OSrate, was characterized by genes enriched in developmental, morphogenetic and cell cycle-relatedprocesses, along with a “desert” TME. Functional characterization of the identified methylation classes ofPM was validated in the MESOMICS dataset. Accordingly, a PM methylation subtype classifier wasdeveloped to predict response to ICI therapy. Conclusions: Tumor DNA methylation defines biologicallyand clinically distinct immune phenotypes in PM and robustly predicts clinical response and long-termsurvival in ICI-treated PM patients, regardless of tumor histology.