Simplified 6-month prediction scores for primary biliary cholangitis patients treated with ursodeoxycholic acid

Abstract

OBJECTIVES: To develop a prognostic score evaluating treatment response at 6 months after ursodeoxycholic acid (UDCA) initiation in primary biliary cholangitis (PBC) patients. METHODS: Adult PBC patients who were newly prescribed UDCA at our institution (n = 292) were included. Significant determinants of liver-related adverse events in the multivariable Cox model were used for score development, weighted by β-coefficients. Discrimination ability was assessed using Harrell's C-statistic. The performance of our model was compared to the previous models. RESULTS: Our model included the following variables evaluated at 6 months: (1) alkaline phosphatase decline of less than 50% from baseline and >upper limit normal (ULN) (2 points); (2) bilirubin >ULN (2 points); (3) albumin <lower limit normal (1 point). The score ranged from 0 to 5 points. C-statistic estimates were 0.87 (overall cohort), 0.87 (no cirrhosis) and 0.77 (cirrhosis), indicating good discrimination of treatment response. Patients with scores ≥3 points had significant shorter transplant-free survival (TFS) than scores <3 points (P < 0.001). The TFS rates for patients with score ≥3 points at 5, 10 and 15 years were 52, 26 and 7%, and for patients with scores <3 points were 96, 92 and 82%, respectively. There was no significant difference between the performance of our 6-month model and the previous models (Paris I, Paris II, Barcelona, Rotterdam and GLOBE scores evaluated at 12 months) in predicting liver-related outcomes (all P = NS). CONCLUSION: This novel 6-month prognostic model showed good prognostic performance. Utilization of this score would identify patients with suboptimal responses to UDCA earlier.