Andanson M.(1,2), Girard P.(1) , Chabaud S.(1) , Charpiat B.(2), Fourcade N.(3), Ducerf C.(4)
1) Clinical Pharmacology Department, Claude Bernard University, Lyon, France.; 2) Department of Pharmacy, Croix-Rousse Hospital, Lyon, France.; 3) Department of Anesthesiology, Croix-Rousse Hospital, Lyon, France.; 4) Department of Surgery, Croix-Rousse Hospital, Lyon, France.
Tacrolimus is an immunosuppressive agent who presents great intraindividual and interindividual variability. The goal of the present population PK analysis is to characterise tacrolimus PK in liver transplanted patients, using clinical routine data, and more specifically how clearance evolves with time after the transplantation. Forty patients treated by tacrolimus after liver transplantation were studied. The drug was administered twice daily (at 8:00 AM and 8:00 PM). Approximately, one determination of tacrolimus trough levels (just before morning administration) on whole blood was done per day. The population data set comprised 982 concentration measurements. Details about number of concentration measurements (on each week after transplantation) were as follows:
| Week 1 | Week 2 | Week 3 | Week 4 | >Week 4 | |
| Total number of concentrations | 217 | 235 | 209 | 137 | 184 |
| Min and max of concentrations per patient | 3-8 | 3-7 | 1-7 | 1-7 | 1-18 |
| Number of patients | 40 | 40 | 39 | 36 | 25 |
For each patient, the covariates: indices of hepatic function [bilirubine, AST, ALT], haematocrit, weight, temperature of the human body, creatinine concentration were determined approximately each day. Cholesterol concentration and triglycerides concentration were determined sparsely. Moreover, all comedication able to interact with tacrolimus was collected.
We used NONMEM with FOCE centered method to fit the models. In a first step, we compared two models : a one and a two compartment model with first-order absorption. Since no points were available to describe absorption and distribution, neither IV data, Ka and F were fixed at litterature values 4.5h-1 and 25% respectively. In the model, interindividual variability of systemic CL and central volume was described by an exponential model, while peripheral volume and intercompartmental clearance were considered as fixed for the two compartment model. Intraindividual error was a combined additive and proportional error model. The results were as follows :
| Parameters | One compartment | Two compartment |
| V central (CV) | 454 L (65%) | 370 L (83%) |
| V peripheral | 73.1 L | |
| CL | 19 L/h (43%) | 18.9 L/h (42%) |
| OF | 4890.0 | 4889.7 |
(CL=clearance, V=volume of distribution, OF=objective function)
Further analysis was performed with 1 and 2 compartments. In a second step, CL was divided in five CL varying within time (one CL per week until the fourth week and only one CL after). Best result was obtain with the two compartment model. The difference between the two OF (one and two compartment model) was statistically significant. We obtained following results: Vcentral = 157 L, Vperipheral = 116 L, CL1 (week 1)= 12.5 L/h; CL2 (week 2)= 19.4 L/h; CL3 (week 3) = 19 L/h, CL4 (week 4)= 18.8 L/h; CL5 (week 5) = 15.8 L/h.
In a third step, we analysed graphically relationship between clearances (determined in the second step) and some covariates as a function of time in order to determine which covariates could influence clearance. One graphic was done per patient. We found that bilirubine, transaminases (AST, ALT) were correlated with clearance (clearance was small when bilirubine, AST, ALT were elevated particularly in the immediate post-operative period).
Some authors have found clearance and volume of distribution mean values in liver transplant patients treated by tacrolimus but, at our knowledge, none have done population pharmacokinetic study in adults. So, we couldn?t compare our results. Moreover, this model would be improve by integrating covariates like bilirubine, AST and ALT. This work is ongoing : next step will consist in using a parametric (or non parametric) function for describing time varying CL and including the time varying covariates in the model.
Reference: PAGE 8 (1999) Abstr 140 [www.page-meeting.org/?abstract=140]
Poster: poster