Multimodal therapeutic efficacy model for predicting early treatment response to TACE-HAIC combined with immune checkpoint inhibitors and tyrosine kinase inhibitors in unresectable hepatocellular carcinoma.

BACKGROUND: For patients with unresectable hepatocellular carcinoma (uHCC), the novel regimen of combining transarterial chemoembolization (TACE) with hepatic arterial infusion chemotherapy (HAIC), supplemented by immune checkpoint inhibitors (ICIs) and tyrosine kinase inhibitors (TKIs), has brought new hope for treatment. However, individual differences are significant. Some patients do not benefit from the treatment but instead bear the toxicity of the drugs and the economic burden. Accurate prediction of early treatment responses is key to optimizing strategies.

METHODS: A total of 205 patients with uHCC who underwent TACE-HAIC combined with ICIs and TKIs therapy were retrospectively enrolled from two independent institutions. A predictive model for early tumor response was developed based on enhanced CT imaging features, hepatic digital subtraction angiography (DSA) characteristics, and clinical parameters. The predictive performance of various models was systematically compared and evaluated.

RESULTS: Two clinical features, nine radiomics features, and eleven DSA features were included in the construction of the combined model. The prediction model exhibited robust performance, with an area under the receiver operating characteristic curve (AUC) of 0.944 (95%CI: 0.897-0.989) in the training cohort, 0.916 (95%CI: 0.832-0.997) in the internal validation cohort, and 0.902 (95%CI: 0.826-0.979) in the external validation cohort. This combined model has excellent calibration, and the analysis of the decision curve shows that it has extensive clinical practical value.

CONCLUSIONS: This study developed a multimodal predictive model integrating hepatic DSA, CT imaging, and clinical data for the prediction of early tumor response to TACE-HAIC combined with TKIs and ICIs in patients with uHCC.