Development and validation of a novel decision tree-based stratification model using preoperative biomarkers predicts survival in hilar cholangiocarcinoma.
The current prognostic assessment of hilar cholangiocarcinoma (hCCA) relies heavily on postoperative pathological parameters, which limit preoperative clinical decision-making. This study aimed to develop and validate a decision tree-based prognostic model for risk stratification among patients with resectable hCCA using preoperative biomarkers.
A bicentric retrospective analysis was conducted on patients who underwent curative resection for hCCA. The decision tree model was developed in a training cohort and validated in an independent cohort. The primary endpoints were overall survival (OS) and recurrence-free survival (RFS).
Among 435 patients (training cohort: 305 patients; validation cohort: 130 patients), multivariate analysis identified a carbohydrate antigen 19-9 level of ≥37 U/mL (hazard ratio [HR], 1.96 [95% CI, 1.24-3.08]), a carbohydrate antigen 12-5 level of ≥35 U/mL (HR, 2.62 [95% CI, 1.78-3.85]), and a C-reactive protein level of ≥5 mg/L (HR, 1.79 [95% CI, 1.31-2.45]) as independent predictors for survival and recurrence. The decision tree model was used to stratify patients into low- and high-risk groups with significantly different outcomes. In the training cohort, compared with high-risk patients, low-risk patients demonstrated superior 3-year OS (23.0% vs 56.1%, respectively) and RFS (21.9% vs 40.5%, respectively) (both P <.0001). The model maintained robust performance in the validation cohort.
This novel biomarker-based decision tree model effectively stratified patients with resectable hCCA into distinct prognostic groups. By integrating readily available preoperative parameters, this practical tool could facilitate personalized treatment strategies and improve surgical decision-making.
A bicentric retrospective analysis was conducted on patients who underwent curative resection for hCCA. The decision tree model was developed in a training cohort and validated in an independent cohort. The primary endpoints were overall survival (OS) and recurrence-free survival (RFS).
Among 435 patients (training cohort: 305 patients; validation cohort: 130 patients), multivariate analysis identified a carbohydrate antigen 19-9 level of ≥37 U/mL (hazard ratio [HR], 1.96 [95% CI, 1.24-3.08]), a carbohydrate antigen 12-5 level of ≥35 U/mL (HR, 2.62 [95% CI, 1.78-3.85]), and a C-reactive protein level of ≥5 mg/L (HR, 1.79 [95% CI, 1.31-2.45]) as independent predictors for survival and recurrence. The decision tree model was used to stratify patients into low- and high-risk groups with significantly different outcomes. In the training cohort, compared with high-risk patients, low-risk patients demonstrated superior 3-year OS (23.0% vs 56.1%, respectively) and RFS (21.9% vs 40.5%, respectively) (both P <.0001). The model maintained robust performance in the validation cohort.
This novel biomarker-based decision tree model effectively stratified patients with resectable hCCA into distinct prognostic groups. By integrating readily available preoperative parameters, this practical tool could facilitate personalized treatment strategies and improve surgical decision-making.
Authors
Yang Yang, Diao Diao, Qi Qi, Sun Sun, Liu Liu, Zhao Zhao, Chen Chen, He He, Xu Xu, Wang Wang, Li Li, Wu Wu, Gu Gu, Kow Kow, Zhang Zhang, Yang Yang
View on Pubmed