Radiopathomics for prediction of early recurrence after curative resection in patients with hepatocellular carcinoma.

[BACKGROUND] The early recurrence (ER) of hepatocellular carcinoma (HCC) following curative liver resection is closely associated with poor clinical outcomes. This study aimed to develop a radiopathomics signature for predicting the risk of ER after curative-intent resection in patients with HCC.

[METHODS] This study comprised 287 patients with HCC who underwent curative-intent resection and divided them into a development cohort (n = 200) and an internal validation cohort (n = 87). An independent external validation cohort was obtained from The Cancer Genome Atlas Liver Hepatocellular Carcinoma (TCGA-LIHC) database. Radiomics features were extracted from portal venous-phase CT images, and pathomics features were derived from digitized whole slide images (WSIs). Feature selection was performed using support vector machine-recursive feature addition (SVM-RFA) to construct the radiopathomics signature. For comparison, three unimodal models, namely radiomics-only, pathomics-only, and clinical-only, were developed using the same computational framework. Model performance was comprehensively evaluated across both validation cohorts. Furthermore, gene set enrichment analysis (GSEA) was conducted to investigate the biological pathways associated with the radiopathomics signature.

[RESULTS] The radiopathomics signature outperformed all unimodal models in predicting ER, yielding area under the curve (AUC) values of 0.887 (95% CI: 0.840-0.935), 0.824 (95% CI: 0.737-0.911), and 0.762 (95% CI: 0.612-0.912) in the development, internal validation, and external validation cohorts, respectively. Stratification based on the radiopathomics score successfully identified the high- and low-risk subgroups with significantly different recurrence-free survival (RFS) outcomes. GSEA further revealed that the high-risk group was enriched in cancer-related and metabolic pathways.

[CONCLUSIONS] The proposed radiopathomics signature provides a promising approach for predicting postoperative ER in patients with HCC, potentially informing individualized clinical management strategies.