Site-specific immune remodeling underlies resistance to dual immune checkpoint blockade in metastatic colorectal cancer

Immune checkpoint inhibitors (ICIs) have limited efficacy in metastatic colorectal cancer (mCRC) with proficient mismatch repair and microsatellite stability (pMMR/MSS), with liver metastases emerging as a dominant determinant of resistance. Whether this reflects a failure to activate antitumor immunity or a failure to direct it remains unclear. We investigated this in the phase II SERPENTINE trial of dual PD-L1 and CTLA-4 blockade by integrating paired pre-treatment and on-treatment single-cell transcriptomic and T-cell receptor (TCR) profiling (388,196 cells; 95,617 clonotypes) metastatic biopsies with deep TCR sequencing of matched peripheral blood (57,902,418 clonotypes). In MSS mCRC, clinical benefit was restricted to patients without liver metastases, whereas all patients with liver metastases progressed. Dual ICI induced systemic T-cell priming regardless of metastatic site, mobilizing therapy-expanded clones that trafficked to both lung and liver lesions, but the fate of this response was determined locally. In lung metastases, therapy-specific clones focused into tissue-resident, proliferative, tumor-reactive CD8+ T-cell states. In liver metastases, the same activation was dispersed: regulatory T cells and suppressive myeloid populations expanded, TCR repertoires diversified, and exhaustion-associated T-cell gene networks fragmented. Cross-lineage co-expression analysis further revealed that tumor proliferation increased in concert with cytotoxic CD8+ T-cell programs, indicating effector engagement that fails to restrain tumor growth. These findings show that resistance to ICI in liver metastases arises from tissue-specific misdirection of immune activation away from productive antitumor immunity, identifying multicellular coordination of the immune response in the metastatic microenvironment as a determinant of therapeutic efficacy.