Developing harmonized tumor microenvironment subtypes for patient stratification in clinical trials.

Abstract

3120 Background: Accurate patient stratification by tumor microenvironment (TME) is critical for optimizing immunotherapy and targeted therapy outcomes. Existing transcriptome-based classification methods often focus on limited TME aspects and tumor types, restricting their clinical utility. Here, we developed a harmonized (H) TME classification by integrating findings from published pan-cancer TME profiling with reported immune evasion mechanisms. Methods: We used agglomerative clustering of gene signature scores (ssGSEA) and pathway activities (PROGENy) encompassing immune cells, fibrosis, vascularization, hypoxia, and other signals to acquire 9 transcriptomic HTME subtypes (Table 1) independently on TCGA (n=7,362) and internal (n=5,186) pan-cancer cohorts with high reproducibility (Spearman r =0.94). Next, we validated the subtypes on 29,875 solid cancer samples from open-source transcriptomic datasets by applying the K-Nearest Neighbors classifier. Differential expression analysis followed by Spearman correlation and PCA were used for comparison among classifications. Adjusted Cox models for survival and logistic regressions for response were applied. Results: Correlative analysis between existing classifications and HTME subtypes revealed patient stratification along 3 principal biological axes: adaptive immune response vs. tumor cell activity, fibrosis vs. antigen-presenting activity, and proliferation vs. vascularization, with, HTME uniquely covering all three axes. When applied to the selected clinical cohorts, HTME effectively distinguished patients by response (logOR) or progression-free survival (logHR) in a diagnosis- and therapy-specific manner (Table 1; showing values with p≤0.05). Conclusions: The proposed HTME patient stratification framework, publicly available and applicable to any RNA-seq sample, constitutes a practical tool for biomarker discovery and trial design across solid tumors. logOR and logHR metrics of HTME subtypes based on diagnosis and treatment type. BRCA Basal-like, Luminal, Normal-like; ICIlogOR [CI] BRCA HER2+; anti-HER2logOR [CI] ccRCC; TKIPFS logHR [CI] ccRCC; anti-VEGFR+anti-PDL1PFS logHR [CI] ccRCC; anti-mTORPFS logHR [CI] SKCM; ICIlogOR [CI] NSCLC; ICIlogOR [CI] Lymphoid-Cell-Enriched B-Cell-Enriched/Angiogenic -0.9 [-1.5, -0.3]** -0.7 [-1.4, -0.1]* 1.1 [0.2, 2.0]* Immune-Enriched/Hypoxic -1.8 [-3.2, -0.4]* 0.6 [0.1, 1.0]** 1.1 [0.4, 1.8]** -1.3 [-2.5, -0.1]* Highly Immune-Enriched/Inflamed -2.9 [-5.3, -0.4]* -1.5 [-2.3, -0.6]*** 1.1 [0.2, 1.9]* -2.1 [-3.6, -0.5]** -1.1 [-1.9, -0.2]* Immune-Enriched/Fibrotic 1.0 [0.5, 1.6]*** 2.3 [0.6, 4.1]** Fibrotic/Angiogenic/Myeloid 0.9 [0.4, 1.3]*** Fibrotic/Hypoxic 1.4 [0.4, 2.5]** Immune Desert 0.8 [0.0, 1.5]* 1.2 [0.2, 2.2]* 0.6 [0.1, 1.2]* Desert/Angiogenic 1.1 [0.1, 2.1]* -0.6 [-1.1, -0.1]* 1.1 [0.1, 2.1]* *p≤0.05; **p≤0.01; ***p≤0.001.