Applicability of BCLC stage for prognostic stratification in comparison with other staging systems: single centre experience from long-term clinical outcomes of 1717 treatment-naive patients with hepatocellular carcinoma

Background The most informative staging system regarding survival outcomes for treatment-naive hepatocellular carcinoma (HCC) remains debated. We evaluated prognostic values of Barcelona Clinic Liver Cancer (BCLC) stage compared with other staging systems, and identified discrepancies between treatment options chosen in Korean clinical practice and BCLC guidelines. Methods Between 2003 and 2008, 1717 prospectively enrolled patients with treatment-naive HCC were analysed. Prognostic ability of each staging system was assessed using time-dependent receiver-operating characteristic (ROC) curves. Results The most common aetiology was hepatitis B virus (1238, 72.1%); 167 (9.8%) patients were classified as BCLC stage 0, 526 (30.6%) as A, 333 (19.4%) as B, 608 (35.4%) as C and 83 (4.8%) as D. Median overall survival was 22.5months, and 1-, 2-, 3-, 4-, and 5-year survival rates were 62.6, 48.3, 39.9, 34.7, and 29.3% respectively. Of six staging systems, BCLC had the highest area under ROC (AUROC; 0.821) for overall survival, followed by JIS (0.809), Tokyo score (0.771), CLIP (0.746), CUPI (0.701) and GRETCH (0.685) system. In both subgroups stratified according to treatment strategy (curative vs. palliative), BCLC also showed the best AUROCs (curative, 0.708/palliative, 0.807) for overall survival. Regarding discrepancies between treatment options chosen in our cohort and BCLC guidelines, more than half with very early/early-stage HCC underwent transarterial chemoembolization, rather than resection or local ablative therapy; most of those with advanced-stage HCC received intra-arterial chemotherapy-based treatments rather than sorafenib. Conclusion BCLC was the best long-term prognostic model for treatment-naive HCC in a large-scale Korean cohort. However, treatment modalities did not exactly match BCLC paradigm.