Digital spatial profiling for identification of prognostic genes and molecular subgroups in pleural mesothelioma.

Le résumé fourni par la source

8084 Background: Pleural mesothelioma (PM) is an aggressive malignancy that harbors significant inter- and intra-tumoral heterogeneity. Spatial transcriptomics enables the dissection of the tumor's molecular architecture by facilitating compartment-specific gene expression profiling. We performed high-resolution RNA-seq analysis of tumor (Tm) and stroma (St) compartments in PM samples to identify gene expression patterns and their association with clinical outcomes. Methods: Formalin-fixed paraffin-embedded (FFPE) tumor samples from untreated PM patients (pts) across three institutions were analyzed using the NanoString GeoMx Digital Spatial Profiling (DSP) platform. Regions of interest (ROIs) were selected based on histopathological features and fluorescently labeled antibodies for tumor and stromal areas. RNA expression of >1800 genes from selected ROIs was analyzed using GeoMx Cancer Transcriptome Atlas (CTA). Differential gene expression was assessed utilizing R “limma” package. A cutoff of absolute fold change ≥1 and p-value <0.05, with the Benjamini-Hochberg false discovery rate method, was applied to identify significant differentially expressed genes (DEGs). Elastic Net regression optimized through cross-validation methods was employed to identify genes associated with overall survival (OS) outcomes. Data from the TCGA PanCancer Atlas was utilized for external validation. Results: A total of 72 pts, 80.3% male, median age of 71y (range: 44-94) were identified for the analysis. Among them, 87.5% (63/72) were epithelioid (Ep) and 12.5% (9/72) non-epithelioid (NEp). After quality control, RNA data was available from 71 and 67 pts in Tm and St compartments, respectively. Across 132 ROIs in the Tm compartment, we identified 4 significantly DEGs between NEp (upregulated COL5A2, THBS1 ; downregulated CLU, KRT19 ) and Ep subgroups. No DEG between Ep and NEp subgroups were identified in 115 ROIs from the St compartment. Unsupervised clustering identified four molecular subgroups with distinct gene expression in the Tm compartment, of which Cluster 1 showed significantly decreased OS (6.3m vs. 16.4m; HR 3.3, p =0.001). Elastic Net regression identified 31 genes predictive of OS (R² = 0.43, Harrell’s c-index = 0.87), including nine genes ( IFNGR2, FCER1G, MFGE8, CKLF, CBL, HLA-DRB3, HK1, PLAT, CD163 ) associated with worse prognosis. Tumors in Cluster 1 demonstrated higher expression of these genes. External validation using the TCGA cohort confirmed four genes IFNGR2 ( p = 0.02), CBL ( p = 0.01), HK1 ( p <0.001), PLAT ( p <0.001), as significantly associated with decreased OS in PM. Conclusions: Spatially resolved transcriptomic profiling suggests Tm-enriched regions as the primary drivers of PM subtype and aggressiveness, identifying nine genes and a molecular subgroup associated with poorer survival outcomes. Further validation and functional studies are warranted.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.