Tumor Vascular Features and Colorectal Cancer–Specific Mortality

Importance The phenotypic and morphologic heterogeneity of tumor vasculature in colorectal cancer (CRC) has been increasingly recognized; however, its clinical importance remains unclear. Objective To examine whether specific tumor vascular features are associated with CRC-specific mortality. Design, Setting, and Participants The prospective cohort incident-tumor biobank method (PCIBM) was used to analyze data from the Nurses’ Health Study and the Health Professionals Follow-up Study, in which 4476 incident CRC cases were documented and resected tumor tissue specimens were collected. Immunofluorescence staining was conducted from July 30 to August 14, 2024, and data analyses were performed between December 2024 and June 2025. Exposures Tumor vascular features were assessed by in situ multispectral immunofluorescence. The study used an in situ multispectral immunofluorescence assay targeting ACKR1 (atypical chemokine receptor 1), CD34 (cluster of differentiation 34), CD36 (cluster of differentiation 36), KDR (kinase insert domain receptor), LAMB1 (laminin subunit β1), MADCAM1 (mucosal addressin cell adhesion molecule 1), and KRT (keratin) combined with machine learning to characterize tumor vasculature. The vessels were morphologically classified as micro, collapsed, patent, and irregular. Multivariable-adjusted Cox proportional hazards regression models were used to assess CRC mortality. Main Outcomes and Measures CRC-specific mortality assessed using multivariable adjusted Cox proportional hazards regression models. Results Tumor vessels were successfully analyzed in 837 patients with CRC (median [IQR] age at diagnosis, 69 [63-75] years; 466 [56%] female). During a median (IQR) follow-up of 11.8 (8.7-14.7) years for censored cases, there were 628 all-cause deaths, including 263 CRC-specific deaths. Multivariable-adjusted CRC-specific mortality hazard ratios were 0.39 (95% CI, 0.25-0.60; P for trend <.001) for overall CD34 + vessel density in quartile 4 (vs quartile 1), 0.48 (95% CI, 0.31-0.74; P for trend = .004) for micro CD34 + vessel proportion in quartile 4, 1.58 (95% CI, 1.07-2.32; P for trend = .004) for CD34 + CD36 + vessel proportion highest category (vs lowest category), and 1.78 (95% CI, 1.22-2.61; P for trend <.001) for CD34 + LAMB1 + vessel proportion highest category. Conclusions and Relevance In this PCIBM-based study of patients with CRC, higher overall CD34 + vessel density was associated with better prognosis, whereas higher CD34 + CD36 + and CD34 + LAMB1 + vessel proportions were associated with worse prognosis. These findings suggest that tumor vasculature may have a role as a prognostic biomarker and potential therapeutic target.