Integrated genomic and tumor microenvironment subtyping refines risk stratification in primary central nervous system lymphoma

Abstract Purpose: Primary CNS lymphoma (PCNSL) remains a therapeutic challenge due to limited precision medicine approaches and suboptimal risk stratification. We aimed to establish an integrated classification framework by combining genomic and tumor microenvironment (TME) profiling. Experimental Design: Whole-genome sequencing (WGS) and multiplex immunofluorescence (mIF) were performed on 68 treatment-naïve PCNSL patients. Unsupervised clustering of 271 genomic features (mutations, structural variants, and copy number variations) defined molecular subtypes. TME classification leveraged CD8⁺T/M2 macrophage ratios. Results: We identified four genomic subtypes with distinct driver alterations: C1 (DNA repair defects: BRIP1/MTAP mutations and 3p/16q amps), C2 (apoptotic dysregulation: BCL2/IRF4 mutations), C3 (BCR- MAPK hyperactivation: CD79A/CARD11/EGFR mutations), and C4 (epigenetic dysregulation and mismatch repair deficiency: KMT2A/TET2/MSH6 mutations). These subtypes stratified overall survival (OS, p=0.02), with C4 exhibiting the poorest median OS (26 months). TME classification by CD8⁺T/M2 ratio revealed three groups: High (>1.5), Intermediate (0.8–1.5), and Low (<0.8). Unexpectedly, the Intermediate group showed the worst outcomes (5-year OS: 10% vs. 75% in High and 43% in Low; p=0.005). Integration exposed a lethal C4/Intermediate-TME subgroup (n=6, 9.8%) with catastrophic median OS of 3.0 months (HR=7.24, p=0.006). Multivariate analysis confirmed both systems as independent prognostic factors. Nomograms incorporating genomic/TME subtypes, Han's classification, and progression of disease within 24 months (POD24) achieved robust prediction (C-index >0.78). Conclusions: Our integrated classification reveals a high-risk biology in C4/Intermediate-TME PCNSL and provides a clinically deployable framework for risk-adapted therapy. The CD8⁺T/M2 ratio refines TME evaluation beyond conventional paradigms, enabling biomarker-driven trials.