S2821 BALAD-2 Emerges as the Most Accurate Prognostic Model in Hepatocellular Carcinoma: Results From a Biobank-Based Cohort Study

Introduction: Accurate prognostication in hepatocellular carcinoma (HCC) remains essential for treatment selection and risk stratification. This study aimed to compare the prognostic performance of individual serum biomarkers (AFP, AFP-L3%, and DCP) and composite scoring models—including GALAD, BALAD, BALAD-2, GAAP, ASAP, Doylestown algorithm, and aMAP—using data from a biobank-based HCC cohort. Methods: We enrolled 186 patients with confirmed HCC diagnoses with available serum, clinical and follow-up data, between 2019 and 2024 from Marmara University HCC biobank. Serum levels of AFP, AFP-L3% and DCP were measured and used to calculate composite scores using validated formulas. The primary endpoint was 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 by treatment intent (curative vs non-curative) and liver disease etiology (viral vs non-viral). Results: The median age was 65 years; 74.7% of patients were man, and 90.3% had underlying cirrhosis. HBV was the most common etiology (49.5%), followed by MASLD (29.0%) and HCV (14.0%). Among all models, BALAD-2 showed the top-performing prognostic accuracy (c-index = 0.737) with highest AUROCs at 1-year (0.827), 2-year (0.846), 3-year (0.781), and 5-year (0.716). BALAD-2 consistently showed superior discrimination in both patients treated with curative (c-index = 0.698) or non-curative therapies (0.716) with highest c-indexes, and AUROCs across most time points. BALAD-2 demonstrated the best prognostic discriminative accuracy in viral etiology by highest c-index (0.726), and AUROCs at 1-year (0.846), 2-year (0.853), and 5-year (0.758). In non-viral etiology subgroup, BALAD-2 remained among the top performers, although GAAP, ASAP, Doylestown and GALAD showed slightly higher c-index and AUROCs at different time points. Conclusion: BALAD-2 outperformed other biomarker-based and clinical composite prognostic models in predicting survival among patients with HCC, regardless of treatment type, with particularly strong performance in those with viral etiologies. These findings support its integration into clinical risk stratification and decision making in HCC management.