The Integrated Prediction of Clinical and Pathological Factors on the Prognosis of Intrahepatic Cholangiocarcinoma

Objective The primary objective was to construct a high-performing prognostic risk model to accurately forecast the prognosis of patients diagnosed with intrahepatic cholangiocarcinoma (iCCA). Methods We retrospectively collected clinical data from the MSK database on 125 patients diagnosed with iCCA. Random sampling was utilized to divide patients into a training set and a validation set, maintaining a ratio of 7:3. Univariate and multivariate Cox proportional hazards regression models were utilized to identify independent prognostic factors influencing OS. Based on these independent factors, a model nomogram was established. The performance of the prognostic prediction models was assessed through calibration curves and C-index calculations. The Kaplan-Meier method was used to plot survival curves. Time-dependent ROC curve was used to evaluate the accuracy of the model. Results A nomogram was developed, incorporating hepatitis C, CA19, tumor extent, tumor size, LVI, positive lymph nodes, and TMB as predictive factors. The C-index for the training set was .78 and the validation set was .68. Using the riskscore derived from the nomogram, patients were stratified into high- and low-risk groups. The high-risk group exhibited considerably lower OS and RFS compared to the low-risk group in the training set (P < .05). However, no significant difference was detected in RFS among different risk groups in the validation set (P > .05). The AUC for 1-year, 3-year, and 5-year survival was .89, .69, and .69, respectively. Conclusion We successfully developed and validated a prognostic nomogram for iCCA, demonstrating its excellent accuracy in predicting patient outcomes and providing clinicians with a potential prognostic tool.