JCO Precis Oncol. 2026 Jul;10(7):
e2600349
PURPOSE: Given the high postoperative recurrence rates and poor prognosis of cholangiocarcinoma (CCA), this study aimed to evaluate cytochrome C oxidase subunit 6B2 (COX6B2) as a potential biomarker for recurrence prediction and to develop a composite risk stratification model integrating clinicopathologic features for personalized treatment planning.
METHODS: We first performed an unbiased recurrence-oriented transcriptomic screening using TCGA-CHOL and GSE107943 data sets. Recurrence-upregulated genes were subjected to log-rank survival testing, univariate Cox regression, and cross-cohort intersection to identify candidate recurrence-associated genes. COX6B2 was then evaluated at the protein level by immunohistochemistry in a training cohort (N = 115) and an independent validation cohort (N = 48). A composite recurrence risk stratification model incorporating COX6B2 and routine clinicopathologic variables, including carbohydrate antigen 19-9, carcinoembryonic antigen, vascular invasion, and tumor size, was developed and validated.
RESULTS: Bioinformatics analysis revealed significant COX6B2 upregulation in recurrent CCA (hazard ratio, 1.351; P = .034). COX6B2 alone demonstrated robust predictive performance with AUC values of 0.87 (training cohort) and 0.81 (validation cohort). The composite model achieved superior predictive accuracy (AUC, 0.93) and effectively stratified patients into high-risk (5-year recurrence rate: 90.62%-95.59%) and low-risk (recurrence-free rate: 85.11%-87.50%) groups. The model's negative predictive value improved by 11% compared with COX6B2 alone.
CONCLUSION: COX6B2 may serve as a clinically useful tissue-based biomarker for postoperative recurrence prediction in CCA. By integrating COX6B2 with standard clinicopathologic features, the composite model provides a practical approach for recurrence risk assessment and individualized postoperative management.