Bayesian Forecasting of Ribavirin in Hepatitis C patients: Population model and validation.

Background: Dual therapy of ribavirin and pegalated interferon is the first line therapy for Hepatitis C. Ribavirin has a narrow therapeutic index with high plasma concentrations resulting in haemolytic anaemia. The aim of ribavirin therapy is to reduce serum viral load whilst minimizing side effects. Strategies such as dosing on body weight or according to renal function have been used previously with some success. Target concentration intervention (TCI) using Bayesian Forecasting is another approach, which may help achieve both safe and effective plasma concentrations of ribavirin.

Aims: ( 1) To develop a population pharmacokinetic model for ribavirin;  (2) To evaluate if the population PK model can predict week 4 area-under-plasma-concentration-time curve (AUC) using a peak and a trough concentration taken either at week 1 or week 2 of treatment.

Methods: Data for ribavirin were available from 2 clinical studies. The first study (Caucasian, n=47; Merck, Sharpe & Dhome North America) included 10 samples over 12 h (single dose and multiple dosing at week 4). The second study (Japanese, n=27; Merck, Sharpe & Dhome Japan) consisted of 7 concentrations (over 48 h) from a single dose and multiple dosing (week 24), 12-hour troughs at weeks 1, 4, 8, 12, 16, 20, and weekly concentrations to week 28 after the dose was ceased (week 24). A population PK model was developed using NONMEM. The final ribavirin PK model was transferred to TCIWorks ( The evaluation set consisted of 10 patients with hepatitis C who were being treated with ribavirin and pegalated interferon. Plasma concentrations were collected at week 1 (2, 12 h), week 2 (2, 12 h) and week 4 (1, 2, 4, 12 h). All concentrations for each patient were entered and the AUC at week 4 was determined. Concentrations from either week 1 or week 2 were then entered separately and used to predict week 4 AUC.

Results: Ribavirin plasma concentrations were best described by a two-compartment model with a first-order absorption. The mean (coefficient of variation, CV%) apparent clearance of ribavirin was 16.6 L/h (31%), apparent central volume of distribution was 524 L (41%), apparent peripheral volume was fixed to 3070 L, apparent inter-compartmental clearance was 39.9 L/h (62%); the rate constant of absorption was fixed to 1.58 /h. Inter-occasion variability (CV%) of clearance and central volume of distribution was 6% and 24%, respectively. Total body weight allometrically scaled was a significant covariate of apparent clearance and apparent central volume of distribution. The predicted week 4 AUC using peak and trough concentrations of ribavirin at week 1 or week 2 in TCIWorks had a bias (precision) of 2.8% (15%) and -3.6% (15.2%), respectively, compared to the observed full data set AUC.

Conclusions: Peak and trough concentrations from weeks 1 and 2 can predict week 4 AUC with reasonable bias and precision. This study lends support to the use of TCI to optimise ribavirin therapy in hepatitis C patients.