991 BALAD and BALAD-2 Predict Survival of Hepatocellular Carcinoma Patients: A North American Cohort Study

Abstract

INTRODUCTION: The discontinuous BALAD and continuous BALAD-2 scores derived from bilirubin, albumin, alpha-fetoprotein (AFP), Lens culinaris agglutinin-reactive AFP (AFP-L3), and des-gamma-carboxyprothrombin (DCP) are effective in predicting HCC patient mortality, but have not been validated in North America. We validated the BALAD and BALAD-2 scores in North America. METHODS: Data from 148 HCC patients seen from 2000–2015 with results for all five biomarkers at diagnosis were abstracted. Hazard ratios (HR) and C-statistics were calculated. RESULTS: 75 of the 148 patients died during a median follow-up of 33 months (0.3-145 months),. The 1-year and 3-year survival rates were 70.8% and 47.6%, respectively. 114 (77%) patients had cirrhosis with a median MELD of 9.5 (range 6.4-29.5). HCC etiologies were HCV (34%), NASH (34%), alcohol (14%), and HBV (10%). Primary treatments were liver transplant, 21 (14%), surgical resection, 32 (22%), local ablation, 10 (7%), transarterial chemoembolization or radioembolization, 47 (32%), chemotherapy, 11 (7%), best supportive care, 9 (6%), and unknown treatment, 18 (12%). MELD, tumor size (cm), ≥2 tumors, tumor number, and neutrophil-lymphocyte ratio >4 significantly increased mortality with HR (95%CI) of 1.12 (1.06–1.18) per unit increase in MELD, 1.06 (1.0004–1.12) per cm increase in size, 1.21 (1.03–1.43) per number of tumor, and 1.93 (1.16–3.24). For the BALAD score analysis, 12, 49, 40, 31, 16 and 0 patients had BALAD scores of 0–5 respectively. The HR (95%CI) for death were 1.24 (0.42–3.67), 1.79 (0.61–5.26), 2.83 (0.95–8.38), and 7.19 (2.26–22.91) for BALAD scores 1, 2, 3, and 4 vs. BALAD 0. For the BALAD-2 analysis, 58, 35, 33, and 22 patients were BALAD-2 classes 1, 2, 3, and 4. The HR (95%CI) for death were 1.25 (0.65–2.40), 1.75 (0.94–3.23), and 6.20 (3.29–11.68) for BALAD-2 classes 2, 3, and 4 vs. BALAD-2 class 1. A multivariate model incorporating maximal tumor diameter, tumor number, neutrophil-lymphocyte ratio and BALAD had HR of 1.43 (1.14–1.81) per increase of 1 BALAD score. A similar multivariate model with BALAD-2 had HR of 1.50 (1.18–1.90) per increase of 1 BALAD-2 class. BALAD, BALAD-2 and BCLC had similar C-statistics of 0.69, 0.68, and 0.67. CONCLUSION: The BALAD score and BALAD-2 class at HCC diagnosis can predict survival of HCC patients in North America. AFP, AFP-L3, and DCP in the BALAD models incorporate HCC tumor biology and make the BALAD model distinguish from other previous predictive models.