Willi Weber, Lutz Harnisch
Hoechst Marion Roussel, Biodynamics, 65926 Frankfurt/M, Germany
Introduction: The novel disease-modifying antirheumatic drug (DMARD) leflunomide has shown efficacy in rheumatoid arthritis. Leflunomide’s active metabolite, A77 1726 (M1), represses de novo pyrimidine synthesis by inhibition of the enzyme, dihydroorate dehydrogenase, which in turn is assumed to modulate the immune response causing rheumatoid arthritis.
Data collected during the clinical development of leflunomide, was used in a stepwise process to develop a population PK/PD model, which enabled us to predict the outcome of clinical trials and optimize therapy.
Objective: To illustrate the application of dose response modeling to anti-rheumatic therapy, the clinical success rates predicted for 10 and 20 mg leflunomide daily were compared to recommend a final dose regimen.
Methods:
- Population PK/PD model
An effect-compartment model was used pragmatically to describe the delay between active plasma concentrations and the occurrence of a clinically successful response. The empirical Hill model was used for the equilibrium relationship between the hypothetical effect site concentration (Ce) of the active compound and the probability of clinical success P(Y=1).
To describe the placebo effect, a virtual substance administered at the start of treatment was assumed. A separate equilibration time constant was introduced to independently modify the time course of the placebo effect. - Clinical Trial Simulation
A binomial probability-distribution function was used as a probability model for clinical success.
Results: 80% of the maximum effect was achieved at a Css of 13 mg/L (EC50=7.4 mg/L) which corresponds to a responder rate of 52.6 % (placebo response=24.8%). To compare the 10 and 20 mg dose regimens, Css values and responder rates were calculated for a half-year treatment. The simulation showed that only 70 % of the maximum effect will be reached in the 10 mg group, whereas in the 20 mg group, the maximum effect could be expected in nearly all patients (~99%).
The 20 mg dose regimen was therefore selected for the phase III trial, which confirmed the efficacy of the drug at that dose (RR=61%).
Discussion/Conclusion: The PK/PD modeling of Leflunomide’s active metabolite, A77 1726 allowed a final dose selection to achieve maximum efficacy and a minimum probability of adverse events, i.e. maximizing the benefit/risk ratio. The simulations performed using data from phase I-II could sensitively predict the clinical outcome in the confirmatory phase III trial.
Reference: PAGE 8 () Abstr 166 [www.page-meeting.org/?abstract=166]
Poster: poster