Background/aims: Liver transplantation remains the only definitive treatment for patients with end-stage primary biliary cirrhosis, although the optimal timing of the procedure remains uncertain. The aim of the study was to use prognostic modelling to determine the optimal timing of transplantation for patients with primary biliary cirrhosis.
Methods: A prognostic model for predicting the survival of patients after transplantation was generated using the Cox regression model with data from 312 patients transplanted for primary biliary cirrhosis at the Queen Elizabeth Hospital, Birmingham. The prognosis after transplantation was compared to that without transplantation (using a previously published prognostic index for non-transplantation) both in these patients and in 98 non-transplanted primary biliary cirrhosis patients dying from the liver disease, in order to establish at what stage the prognosis with transplantation was better than without transplantation.
Results: The prognostic index for transplantation included the following significant prognostic variables: serum bilirubin, serum albumin, age, year of transplantation, and the presence of ascites or treatment with diuretics. Comparison of prognosis with and without transplantation showed that the predicted gain in survival after transplantation becomes increasingly positive when the 6-month survival probability in the absence of transplantation falls below 0.85. In the non-transplanted patients this occurs on average about 8 months before death.
Conclusions: Comparison of the prognosis with and without transplantation provides a rational method for determining the optimum timing of the procedure which occurs approximately when the predicted 6-month survival probability without transplantation falls below 0.85.