Immune checkpoint inhibitors combined with anti-angiogenic targeted agents have become the first-line standard therapeutic regimen for advanced hepatocellular carcinoma (HCC). Nevertheless, the high incidence rate of therapy-related diarrhea severely impairs patients' quality of life and treatment adherence, and may even result in treatment discontinuation. Current studies mostly focus on symptomatic intervention after diarrhea onset, while few tools are available for upfront risk screening prior to treatment initiation. According to the theory of traditional Chinese medicine (TCM) constitutional types, individuals with different constitutions exhibit discrepant disease susceptibility and drug responses. Based on a prospective cohort design, this study aims to explore the correlation between TCM constitutional types and immune-targeted therapy-related diarrhea in HCC patients, and further construct a risk prediction model integrated with both western medical and TCM indicators.

This study is designed as a multicenter prospective observational cohort study. A total of 350 primary HCC patients receiving combined immunotherapy and targeted therapy at the Affiliated Hospital of Chengde Medical University from January 2027 to December 2028 will be enrolled into the model development cohort. Another 150 patients from Chengde Central Hospital and Chengde Hospital of Traditional Chinese Medicine between January 2029 and January 2030 will be recruited as the external validation cohort. Baseline data will be collected before treatment, including sociodemographic and general clinical information, tumor and treatment-related data, laboratory test results, TCM constitutional assessment, nutritional risk screening, physical function and psychological status evaluation. All participants will be followed up regularly for one year to record the occurrence of diarrhea.

Samples in the development cohort will be randomly divided into a training set and an internal validation set at a stratified ratio of 7:3. Lasso regression will be applied to screen predictive factors, and multivariate Logistic regression will be adopted to establish a nomogram prediction model. Receiver operating characteristic (ROC) curves, calibration curves and decision curve analysis (DCA) will be used to assess the discrimination, calibration and clinical net benefit of the established model, respectively. The independent multicenter data of the external validation cohort will adopt consistent variable criteria, outcome definitions and the prediction model from the development cohort to evaluate the cross-center generalizability, stability and clinical applicability of the model for comprehensive efficacy verification.

The study protocol has been approved by the Medical Ethics Committee of the Affiliated Hospital of Chengde Medical University. Written informed consent will be obtained from all subjects before enrollment. The research findings will be disseminated via academic journal publications, web-based assessment tools and clinical promotion. The established TCM constitution risk screening tool can be directly applied to routine admission assessment in the hepatology department.