Beyond mismatch repair in colorectal cancer liver metastases: rethinking immunotherapy through the hepatic metastatic niche

Immune checkpoint blockade has made deficient mismatch repair/microsatellite instability-high (dMMR/MSI-H) status the principal biomarker for immunotherapy in metastatic colorectal cancer. With more than 5 years of follow-up, KEYNOTE-177 continues to show durable benefit from first-line pembrolizumab, and CheckMate 8HW has established nivolumab plus ipilimumab as a highly active option in the same molecular subgroup (1)(2)(3). In contrast, checkpoint inhibitors are not standard therapy for unselected proficient mismatch repair/microsatellite-stable (pMMR/MSS) disease (4). MMR/MSI is therefore a clinically established proxy for tumor-intrinsic immunogenic potential, but effective immunity additionally depends on antigen presentation, persistence of tumor-reactive lymphocytes, and their access to malignant cells.Clinical series have associated liver involvement with inferior outcomes during checkpoint-based treatment in pMMR/MSS colorectal cancer and in dMMR/MSI-H disease, although findings are inconsistent (5,6). We therefore use a two-domain model: MMR/MSI status and bona fide POLE/POLD1 proofreading deficiency estimate tumor-intrinsic immunogenic potential, whereas active hepatic disease may constrain priming, effector-cell survival, and tumor-cell access (Figure 1). The framework is intended to organize testable hypotheses rather than serve as a clinical decision rule. Its central test is whether hepatic disease status adds predictive information beyond genotype, treatment backbone, total disease burden, and patient selection. Hepatic involvement should be classified as active CRLM, previously treated CRLM without active hepatic disease, or no history of liver involvement. Mechanistic assessment should extend beyond bulk immune-cell density to include dendritic-cell priming, myeloid-fibroblast organization, spatial CD8+ T-cell access, and the tumor-liver interface. Prospective studies should prespecify hepatic-disease stratification and integrate tissue, imaging, and circulating tumor DNA analyses. This framework is conceptual and does not constitute a validated clinical decision algorithm. Abbreviations: CRLM, colorectal cancer liver metastases; MMR, mismatch repair; MSI, microsatellite instability.We regard active hepatic disease as an adverse clinical context and a potential modifier of immunotherapy efficacy rather than a contraindication to treatment. We focus on mechanisms and study designs that can separate a treatment-specific liver interaction from prognosis, tumor burden, and treatment selection.The benefit of checkpoint blockade in dMMR/MSI-H disease supports the view that mutation-derived neoantigens and pre-existing immune recognition provide a therapeutically exploitable substrate (1)(2)(3). Nonetheless, dMMR/MSI-H tumors are not uniformly responsive. A pooled retrospective analysis of 104 patients receiving first-line checkpoint inhibition associated liver metastasis with lower response and shorter progression-free survival (5). A smaller series reported high response rates despite liver involvement and instead highlighted lesion number, peritoneal disease, and tumor mutational burden (6). The discordance argues against withholding immunotherapy solely because of hepatic disease.In pMMR/MSS disease, several datasets show a similar association, but most are single-arm or retrospective. A cohort study and a post hoc analysis of CCTG CO.26 linked liver metastases to lower disease control or less checkpoint-related benefit (7,8). Responses in regorafenib-nivolumab and regorafenib-ipilimumab-nivolumab studies were concentrated in patients without active liver disease (9)(10)(11). Botensilimab-balstilimab likewise produced no responses among 24 efficacy-evaluable patients with active liver metastases, whereas responses occurred after prior liver resection or ablation (12). However, patients whose hepatic disease had already been completely treated constitute a selected subgroup, so the observed difference cannot be attributed to removal of the hepatic niche alone.Tumor burden is an essential alternative explanation. A liver-specific effect is more credible if hepatic status remains informative after accounting for burden and hepatic lesions fare worse than comparable nonhepatic lesions. Burden should include lesion count, summed diameter or volume, extrahepatic burden, and the hepatic fraction of disease; within-patient lesion-level response offers an additional control, although lesion size and prior local therapy still matter. Existing evidence only partly meets this standard. In a nonhypermutated cohort receiving PD-1-based combinations, liver metastasis and baseline lesion number were independently associated with durable clinical benefit, and hepatic lesions were less responsive than lung, nodal, or peritoneal lesions (13). Pakvisal et al. analyzed 132 MSS/pMMR mCRC patients treated with ICI-based regimens, including 93 with active liver metastases at ICI initiation (14). Clinical benefit was 16.1% with versus 46.2% without active liver disease; median PFS was 2.1 versus 2.5 months (HR 1.68, 95% CI 1.13-2.51) and median OS was 6.17 versus 11.53 months (HR 2.03, 95% CI 1.35-3.06). In multivariable models using active liver metastases as reference, no history of liver metastases remained associated with longer PFS (adjusted HR 0.04, 95% CI 0.006-0.23) and OS (adjusted HR 0.006, 95% CI 0.00-0.07), whereas prior complete metastasectomy did not (PFS adjusted HR 1.51; OS adjusted HR 0.73). The study used multivariable Cox models but did not report matched or propensity-score sensitivity analyses; small subgroups and retrospective treatment selection leave substantial residual confounding. Randomized evidence is mixed: AtezoTRIBE showed no treatment-by-liver-metastasis interaction, whereas LEAP-017 was negative overall (15)(16)(17). Active CRLM thus remains a candidate treatment-effect modifier, not an established predictive biomarker. Rare pMMR tumors with bona fide POLE/POLD1