Identifying optimal candidates for post-TIPS patients with HCC undergoing TACE: a multicenter observational study

Area of Science:

• Hepatobiliary surgery
• Interventional radiology
• Oncology

Background:

• Hepatocellular carcinoma (HCC) patients beyond Milan criteria often require advanced treatments like transjugular intrahepatic portosystemic shunt (TIPS) followed by transarterial chemoembolization (TACE).
• Accurate prognostic models are crucial for managing these complex cases and guiding treatment selection.

Purpose of the Study:

• To develop and validate a novel prognostic model for patients with HCC beyond the Milan criteria who have undergone TACE after TIPS.
• To compare the performance of the new model against existing prognostic tools.

Main Methods:

• Retrospective analysis of 512 patients from 15 centers (2013-2020) with HCC beyond Milan criteria treated with TACE post-TIPS.
• Development of a prognostic model using multivariate Cox regression, incorporating factors like vascular invasion, AFP, creatinine, extrahepatic spread, and ALT.
• Stratification into risk groups and comparison of predictive performance (AUROC, C-index) with existing models.

Main Results:

• The VACEA score, incorporating vascular invasion, AFP, creatinine, extrahepatic spread, and ALT, significantly stratified patients into four risk groups with distinct survival outcomes.
• The VACEA model demonstrated superior predictive accuracy and discrimination compared to seven other established prognostic models in both training and validation sets.
• Median overall survival ranged from 25.2 months (lowest risk) to 6.2 months (highest risk).

Conclusions:

• The VACEA score provides accurate individual-level prognosis for post-TIPS TACE-treated HCC patients beyond Milan criteria.
• This model effectively stratifies patients into distinct risk categories, aiding in the identification of candidates who may benefit most from this treatment strategy.
• The VACEA score offers improved prognostic discrimination over current TACE/TIPS-specific models.