Combine radiomics models and multi-omics data stratified patients with AFPhigh - HCC Sensitivity to regorafenib.

Motivation: High expression of serum AFP (>400 ng/mL) in HCC patients predicts poor prognosis. Goal(s): Identify AFP-high HCC patients who are responsive to regorafenib. Approach: We developed a combined clinical-imaging-radiomics model to identify risk in AFP-high HCC patients post-surgery. Identify molecular differences between different risk groups through enrichment analysis and single-cell sequencing.Finally, validate the model using regorafenib treatment outcomes. Results: The model stratified patients into high- and low-risk groups with significant OS differences (P < 0.001). Enrichment analysis showed PI3K-AKT pathway upregulation. Among 26 patients treated with regorafenib, the DCR was 61.5% in the low-risk group and 30.8% in the high-risk group. Impact: The combined model based on clinical, imaging, and radiomics data can predict the postoperative survival status of patients. This model allows for the identification of a subgroup of patients who are at relatively low risk and responsive to regorafenib treatment.