BALAD-2 Emerges as the Most Accurate Prognostic Model in Hepatocellular Carcinoma: Results from a Biobank-Based...

Background/Objectives: Accurate prognostication of hepatocellular carcinoma (HCC) remains essential for treatment selection and risk stratification. This study aimed to compare the prognostic performance of individual serum biomarkers and composite scoring models, including GALAD, BALAD, BALAD-2, GAAP, ASAP, the Doylestown algorithm, and aMAP, using data from a biobank-based HCC cohort. Methods: This study enrolled 186 patients with confirmed HCC diagnosed between 2019 and 2024. Serum biomarkers (AFP, AFP-L3%, DCP) and composite models were evaluated for their association with overall survival (OS). Prognostic performance was assessed using time-dependent area under the receiver operating characteristic curve (AUROC) at 1-, 2-, 3-, and 5-year intervals and Harrel's concordance index (c-index). Subgroup analyses were performed based on treatment intent and liver disease etiology. Results: All three biomarkers and composite models were independently associated with OS in multivariate analyses (all p < 0.05). Among all models, BALAD-2 demonstrated the best overall performance (c-index: 0.737), with the highest AUROCs at 1 year (0.827), 2 years (0.846), 3 years (0.781), and 5 years (0.716). BALAD-2 consistently showed superior discrimination in patients treated with curative or noncurative therapies and in the viral etiology subgroup. In the non-viral etiology subgroup, BALAD-2 remained among the top performers, although the GAAP, ASAP, and Doylestown algorithms showed slightly higher metrics. Conclusions: BALAD-2 demonstrated consistent and robust prognostic performance compared with other biomarker-based and clinical models across different patient subgroups, particularly among those receiving curative therapy and viral etiologies. These findings support its integration into clinical risk stratification and decision-making for HCC management.