JGH Open (Apr 2020)

Evaluation of the United Kingdom‐primary biliary cholangitis and global primary biliary cholangitis group prognostic models for primary biliary cholangitis patients treated with ursodeoxycholic acid in the U.S. population

  • Mohammad Alomari,
  • Fahrettin Covut,
  • Laith Al Momani,
  • Pravallika Chadalavada,
  • Asif Hitawala,
  • Mark F Young,
  • Carlos Romero‐Marrero

DOI
https://doi.org/10.1002/jgh3.12223
Journal volume & issue
Vol. 4, no. 2
pp. 132 – 139

Abstract

Read online

Background and Aim The United Kingdom‐primary biliary cholangitis (UK‐PBC) and global primary biliary cholangitis group (GLOBE) prognostic models have been recently developed to predict long‐term outcomes in primary biliary cholangitis (PBC). However, these predictive scores have not yet been well evaluated in the U.S. population. Methods We retrospectively reviewed newly diagnosed PBC patients at the Cleveland Clinic between November 1998 and February 2017. Adverse events were defined as liver transplantation, liver‐related mortality, and all‐cause mortality. Transplant‐free survival (TFS) was estimated using the Kaplan–Meier method. Predictive performances of all prognostic models were evaluated using the C‐statistic. Results We identified 352 patients who used ursodeoxycholic acid therapy. Of them, 311 (88.4%) only had PBC, while 41 (11.6%) were diagnosed with PBC‐autoimmune hepatitis overlap. A total of 22 (6%), 47 (13%), and 55 (16%) patients had adverse events within 5, 10, and 15 years after diagnosis, respectively. In patients with PBC only, the C‐statistic in predicting 15‐year adverse events was 0.75 per GLOBE compared to 0.74 per UK‐PBC (P = 0.94), 0.73 per Rotterdam (P = 0.44), 0.66 per Barcelona (P = 0.004), 0.65 per Paris 1 (P = 0.005), 0.62 per Paris 2 (P < 0.0001), 0.60 per Toronto (P < 0.0001), and 0.60 per Mayo (P < 0.0001) scores. Median follow‐up was 9.2 years. Ten‐year TFS for patients who had optimal versus suboptimal treatment response was 92 versus 74% per Paris 1 (P < 0.0001), 95 versus 79% per Paris 2 (P = 0.0002), 93 versus 65% per Barcelona (P < 0.0001), and 96 versus 68% per Rotterdam (P < 0.0001) risk scores, respectively. Conclusion In our cohort of PBC patients, the UK‐PBC and GLOBE scores were both accurate and reasonably valid prognostic models in the U.S. population.

Keywords