Prediction of tumor progression in intermediate and advanced hepatocellular carcinoma undergoing TACE combined with...

Transcatheter arterial chemoembolization (TACE) combined with targeted immunotherapy have become the standard first-line treatment strategy for intermediate and advanced hepatocellular carcinoma (HCC), but some patients still cannot benefit from this treatment. This study aimed to construct a prediction model based on a cohort of HCC patients to guide individualized treatment decisions.

A total of 243 intermediate and advanced HCC patients who received TACE combined with targeted immunotherapy from January 1, 2019 to March 31, 2024 were retrospectively enrolled. The optimal prognostic factors were screened by Cox regression analysis and least absolute shrinkage and selection operator (LASSO) regression. A nomogram model for predicting the probability of radiologic progression-free survival at 6-, 12-, and 24-month was constructed based on the screened risk factors, and the model performance was evaluated by calibration curve, decision curve analysis, and restricted cubic spline analysis.

Multivariate COX analysis showed that total bilirubin, D-dimer (DD) and portal vein tumor thrombosis (PVTT) were independent risk factors affecting the tumor progression of patients. LASSO regression screened out 10 key prognostic factors: aspartate aminotransferase, total bilirubin, albumin (ALB), white blood cell count, DD, alpha-fetoprotein, CD4+ T cell count, PVTT, tumor number (≥3) and lymph node invasion. Compared with the Cox regression model, significant advantages in screening prognostic factors were demonstrated by the LASSO regression model. Therefore, the risk factors identified by LASSO regression were chosen to construct a nomogram prediction model for subsequent analysis. The model showed good discrimination ability (Log rank P<0.05), calibration ability and net benefit of clinical strategy in both the training and validation group. Restricted cubic spline analysis found that ALB level and DD were significantly nonlinearly correlated with patient prognosis.

This study constructed a multidimensional prediction model for tumor progression in patients with intermediate and advanced HCC. It was helpful to screen the best benefit population and optimize the treatment strategy.