Can the model for end-stage liver disease be used to predict the prognosis in patients with Budd-Chiari syndrome?

The model for end-stage liver disease (MELD) is a widely accepted and objective scoring system for end-stage liver disease (ESLD) but has never been evaluated for Budd-Chiari syndrome (BCS). We investigated whether MELD can be used to predict survival in patients with BCS. Patients with BCS (n = 237) were obtained from a large international study. Patients with ESLD (n = 281) were used to compare the discriminative ability of MELD in BCS versus other chronic liver diseases. MELD and the Rotterdam BCS index, a recently developed prognostic index for BCS, were calculated with standard equations. Receiver operating characteristic curves and concordance statistics (c-statistics) were used to assess the models' ability to predict 1-year survival. The median MELD score was 12.5 (range = -7.4 to 43.4) for BCS and 11.3 (-3.0 to 49.5) for ESLD (P = 0.12). The c-statistic of MELD in BCS was 0.695 [95% confidence interval (CI) = 0.59-0.80], in contrast to 0.848 (95% Cl = 0.80-0.90) in ESLD. Survival was significantly poorer in ESLD than in BCS (P < 0.001). The c-statistic of the Rotterdam BCS index was 0.760 (95% Cl = 0.67-0.85). The correlation between MELD and the Rotterdam BCS index was 0.61, and most of the discrepancy existed in BCS patients with a high prevalence of ascites and encephalopathy and preserved liver function. The addition of ascites and encephalopathy to MELD improved the c-statistic to 0.751 (95% Cl = 0.65-0.85). In conclusion, MELD showed a suboptimal discriminative ability to predict survival in BCS. This was explained by the highly variable degree of liver dysfunction and hence clinical outcome in BCS in contrast to ESLD.