Hemorrhage-coagulation immune phenotype is associated with CD163/HO-1-enriched heme-processing macrophage remodeling and predicts recurrence in colorectal cancer: a real-world retrospective cohort study.

Background: Coagulation activation and tumor hemorrhage are common in colorectal cancer, but their combined immunological meaning remains incompletely defined. We developed a prespecified hemorrhage-coagulation immune phenotype (HCIP) integrating local hemorrhagic pathology and systemic coagulation activation, and evaluated its association with recurrence-free survival (RFS), CD163/HO-1-enriched immune remodeling, and incremental prognostic value. Methods: This retrospective cohort study included patients who underwent radical resection for colorectal adenocarcinoma at The First People's Hospital of Jiashan County between January 2016 and December 2025. HCIP was defined from a local hemorrhagic axis, based on H&E-assessed intratumoral hemorrhage and hemosiderin/Prussian blue evidence, and a systemic coagulation axis, based on preoperative fibrinogen, D-dimer, and platelet count. The primary outcome was RFS. A balanced FFPE subset was used for exploratory immunohistochemical assessment of CD8, CD68, CD163, HO-1, fibrinogen/fibrin, and tissue factor. Cox models, prespecified sensitivity analyses, and validation-cohort prediction analyses with bootstrap uncertainty estimates were performed. Results: The final analytic cohort included 752 patients. HCIP-poor tumors were enriched for advanced AJCC stage, larger tumor size, ulceration, and necrosis. In the balanced FFPE subset, HCIP-poor tumors showed lower CD8+ stromal T-cell density, higher CD163+ macrophage density, stronger HO-1, fibrinogen/fibrin, and tissue factor staining, a lower CD8/CD163 ratio, and a higher macrophage-heme-coagulation score. In multivariable Cox analysis, HCIP-poor status was associated with worse RFS (adjusted HR 1.91, 95% CI 1.33-2.73). The association remained directionally consistent in stage II/III-only, T/N-stage-adjusted, MSI/MMR-stratified, IPTW, and recurrence-only sensitivity analyses. Among individual hemorrhage/coagulation components, fibrinogen was the most stable marker associated with RFS. In the validation cohort, adding fibrinogen, HCIP, or hemorrhage/coagulation components produced modest incremental discrimination and lower prediction error beyond clinical variables. Conclusion: HCIP identifies a clinically accessible hemorrhage-coagulation state associated with CD163/HO-1-enriched heme-processing macrophage features and coagulation-associated immune remodeling in colorectal cancer. Fibrinogen remained the most practical single coagulation marker, whereas HCIP provided a biologically interpretable composite phenotype linking local hemorrhagic pathology, systemic coagulation activation, and recurrence risk.