Trajectory-Based Prognostic Modeling in Hepatocellular Carcinoma

Integrating Longitudinal Biomarker Dynamics With Clinical Features for Prognostic Prediction in Hepatocellular Carcinoma

To evaluate whether tracking changes in key blood markers over time, together with clinical features, can improve the prediction of outcomes in patients with hepatocellular carcinoma (HCC). By developing a trajectory-based prognostic model, we aim to provide more accurate risk assessment and support personalized treatment decisions.

This study retrospectively analyzed patients with advanced hepatocellular carcinoma (HCC) who were treated at Shenzhen Third People's Hospital between 2018 and 2024. Eligible participants received systemic therapy with molecular targeted agents or PD-(L)1 inhibitors, with or without interventional procedures such as transarterial chemoembolization (TACE) or hepatic arterial infusion chemotherapy (HAIC).

Clinical, radiological, and laboratory data were collected at baseline and during follow-up. Blood tests were performed at regular intervals, and imaging evaluations with contrast-enhanced CT or MRI were conducted every 6-12 weeks in accordance with standard practice. Demographic information, disease stage, comorbidities, and treatment- or disease-related complications were also recorded.

The study focuses on the analysis of longitudinal biomarker changes, with the aim of developing prognostic models that integrate biomarker trajectories with clinical features. The ultimate goal is to improve dynamic risk stratification and support personalized treatment decisions for patients with HCC.

Inclusion Criteria:

1. pathologically or radiologically confirmed HCC;
2. availability of pre- treatment clinical record, radiological and hematological data and more than 2 cycles of post-treatment data;
3. receipt of treatment during the study period;
4. Age rather than 18 years old.

Exclusion Criteria:

1. baseline data missed;
2. Non-primary liver cancer;
3. incomplete follow-up data;
4. Unevaluable lesions.