Modeling and Simulation of the Asenapine Exposure-Response and Drop-Out in Acute Schizophrenia
Lena E. Friberg* (1), Rik de Greef (2), Thomas Kerbusch (2), Mats O. Karlsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (2) Clinical Pharmacology and Kinetics, N.V. Organon, a part of Schering-Plough Corporation, Oss, The Netherlands
Objectives: To characterize the efficacy time-course and drop-out in schizophrenic patients following placebo or asenapine treatment, to substantiate, by simulations, the efficacious dose range of asenapine and to improve the understanding of trial outcomes.
Methods: In three Phase 2 and three Phase 3 placebo-controlled 6-week trials in patients with schizophrenia asenapine was administered sublingually in doses ranging from 0.2 to 10 mg bid. In schizophrenia trials, substantial drop-out rates and variable placebo responses are commonly observed, and both may affect traditional LOCF analysis outcomes. Therefore a pooled population exposure-response analysis was performed in which both the time-course of the primary efficacy parameter Total PANSS as well as drop-out were analyzed using NONMEM.
Results: The placebo effect (EFFPlacebo) and the asenapine exposure response relationship (EFFAsenapine) were proportional to the underlying Total PANSS:
Total PANSS=PANSSbaseline*(1-EFFPlacebo)*(1-EFFAsenapine)
EFFPlacebo =PMAX*(1- e -(TIME/TD)^POW) )
EFFAsenapine=EMAX*AUC/(AUC+AUC50)*f(TIME)
f(TIME)=TIME/42 if TIME≤42 Days and 1 if TIME>42 Days
The maximum placebo effect (PMAX) was 69% higher and PANSSbaseline was 3.5 units lower in Phase 3 studies compared with Phase 2 studies. Patients who had active schizophrenia for more than one month had a 3.4% lower baseline than those with a shorter duration of the present episode.
The most important factors in the drop-out model were related to the observed PANSS scores; a high PANSS score on the preceding observation, an increase between the two last scores, and an increase in score from baseline increased the probability to drop out.
The observed LOCF time-courses were well described in simulations from the combined PANSS and drop-out model. The original trial outcomes were within the 95% prediction interval for all placebo arms and the majority of the treatment arms. Simulated success rates indicated that in the applied study setting the probability of detecting a statistically significant change in PANSS LOCF from placebo at Day 42 for 5 and 10 mg bid asenapine was 24-64%.
Conclusions: The model-based analysis demonstrated a significant exposure-response for asenapine. Simulations indicated that the post-hoc probability of success of the performed trials was relatively low. Overall, this analysis supported the dose range of 5-10 mg bid asenapine as effective in the treatment of schizophrenia.