Transcriptomic profiling identifies immunotherapy-responsive phenotypes in microsatellite-stable metastatic colorectal cancer.

Conventional immune checkpoint inhibitors (ICIs) remain largely ineffective in microsatellite-stable metastatic colorectal cancer (MSS mCRC), where low tumor immunogenicity and molecular heterogeneity across metastatic sites underpin therapeutic resistance. We present a comprehensive transcriptomics analysis of metastatic and primary tumor biopsies from MSS mCRC patients treated with botensilimab (BOT; Fc-enhanced anti-CTLA-4) ± balstilimab (BAL; anti-PD-1). Self-organizing map (SOM) machine learning stratified tumors into four molecular types, including a liver-like (LIV) subtype characterized by metabolic reprogramming and immunosuppressive signatures, and proliferative (PRO), inflammatory (INF), and mesenchymal (MES) types concordant with pan-cancer classifications. PRO, INF, and MES types were enriched for epithelial tumor cells, immune cells, and fibroblasts, respectively, defining immune-depleted, immune-enriched, and fibrotic states along a plasticity gradient. We observed treatment-related transcriptomic shifts toward immune-enriched states via upregulation of antigen presentation, T cell recruitment, and cytotoxicity pathways. INF and MES tumor types exhibited improved clinical responses and survival vs PRO and LIV types. This study identified distinct tumor microenvironment states that align along an immunophenotype axis marked by CD74, interferon-γ, and APOBEC3 expression identified previously for primary CRC. Our findings provide novel insights into molecular correlates of immunotherapy response in MSS mCRC, potentially informing future therapeutic strategies to expand ICI efficacy to historically unresponsive tumors.