Detailed Abstract
[BP Oral Presentation 2 - Biliary & Pancreas (Biliary Disease/Surgery)]
[BP OP 2-2] PREOPERATIVE RISK STRATIFICATION IN HILAR CHOLANGIOCARCINOMA USING A NOVEL DECISION TREE MODEL
Yuze YANG 1, Yongkang DIAO 2, Zixuan WANG 1, Lu SUN 1, Han LIU 1, Mingda WANG 2, Chao LI 2, Lihui GU 2, Han WU 2, Guoyue LV 1, Feng SHEN 2, Ping ZHANG 1, Tian YANG 2
1 Department of Hepatobiliary And Pancreatic Surgery, General Surgery Center, First Hospital of Jilin University, Changchun, China, 2 Department of Hepatobiliary Surgery, Eastern Hepatobiliary Surgery Hospital, Naval Medical University, Shanghai, China
Background : Current prognostic assessment for hilar cholangiocarcinoma (hCCA) predominantly relies on postoperative pathological findings, limiting preoperative decision-making capabilities. This study aimed to develop a practical preoperative risk stratification model using readily available biomarkers to predict surgical outcomes.
Methods : A retrospective analysis was performed on 435 patients who underwent curative resection for hCCA across two high-volume hepatobiliary centers. The cohort was randomly divided into training (n=305) and validation (n=130) groups. A decision tree model was constructed using preoperative biomarkers identified through LASSO regression and multivariate Cox analysis. Model performance was assessed using Harrell's C-index and calibration plots.
Results : Three independent prognostic factors were identified: CA199 ≥39 U/L (HR=1.96, 95%CI:1.24-3.08), CA125 ≥35 U/L (HR=2.62, 95%CI:1.78-3.85), and CRP ≥5 mg/L (HR=1.79, 95%CI:1.31-2.45). The decision tree model effectively stratified patients into low- and high-risk groups. In the training cohort, low-risk patients demonstrated significantly better 3-year overall survival (56.1% vs 23.0%, P<0.0001) and recurrence-free survival (40.5% vs 21.9%, P<0.0001) compared to high-risk patients. The model maintained robust discriminative performance in the validation cohort, with consistent survival differences between risk groups.
Conclusions : This novel decision tree model, incorporating three readily available preoperative biomarkers, provides reliable risk stratification for patients with resectable hCCA. The model’s simplicity and effectiveness make it a practical tool for preoperative risk assessment and surgical decision-making, potentially facilitating personalized treatment strategies.
Methods : A retrospective analysis was performed on 435 patients who underwent curative resection for hCCA across two high-volume hepatobiliary centers. The cohort was randomly divided into training (n=305) and validation (n=130) groups. A decision tree model was constructed using preoperative biomarkers identified through LASSO regression and multivariate Cox analysis. Model performance was assessed using Harrell's C-index and calibration plots.
Results : Three independent prognostic factors were identified: CA199 ≥39 U/L (HR=1.96, 95%CI:1.24-3.08), CA125 ≥35 U/L (HR=2.62, 95%CI:1.78-3.85), and CRP ≥5 mg/L (HR=1.79, 95%CI:1.31-2.45). The decision tree model effectively stratified patients into low- and high-risk groups. In the training cohort, low-risk patients demonstrated significantly better 3-year overall survival (56.1% vs 23.0%, P<0.0001) and recurrence-free survival (40.5% vs 21.9%, P<0.0001) compared to high-risk patients. The model maintained robust discriminative performance in the validation cohort, with consistent survival differences between risk groups.
Conclusions : This novel decision tree model, incorporating three readily available preoperative biomarkers, provides reliable risk stratification for patients with resectable hCCA. The model’s simplicity and effectiveness make it a practical tool for preoperative risk assessment and surgical decision-making, potentially facilitating personalized treatment strategies.
SESSION
BP Oral Presentation 2
Room 5 3/27/2025 3:30 PM - 4:30 PM