Concentration-response analysis of antipsychotic drug effects using an indirect response model
Ortega, Ignacio , An Vermeulen, Vladimir Piotrovsky
Advanced PK/PD modeling & Simulation, Johnson & Johnson Pharmaceutical R&D, Beerse, Belgium
Objectives: The effect of antipsychotics is typically assessed using various scales reflecting the clinical status of patients. The Positive and Negative Syndrome Scale (PANSS) is the most often used. Generally speaking, scores are ordered as categorical variables, however, when the number of categories is large enough they can be considered as continuous, though they are constrained between lower and upper limits, 30 and 210 in the case of PANSS scores. The aim of our study was to apply a recently proposed indirect efficacy model [1], to link the response to antipsychotic treatment with plasma concentrations in three clinical Phase 3 trials.
Methods: Overall, 1252 patients randomized to receive either placebo or one of the five active doses, were included in the analysis. The overall duration of the trials was 7 weeks. A substantial number of patients dropped out before the trial end. The model describes the patient's clinical status as a balance between deterioration (rD) and amelioration (rA) processes. The nature of these processes cannot be described as of yet, so the model is empirical. However, some mechanistically sound elements can be added. The general description of the model can be found in [1]. A new element added to perform the current analysis was the impact of the time to last observation (TTLO) on the ultimate score, RP. The final expression for RP was as follows:
RP = 180*exp(logit(R0)+DP+EFF+TT)/(1+exp(logit(R0)+DP+EFF+TT))+30
DP, EFF and TT stand for disease progression, placebo and drug effects, and the effect of the time to last observation (TTLO), respectively. It was also assumed that EFF decreased with time (long-term “tolerance”): EFF = (EP+E)*exp(-KT*T), where EP and E are placebo and drug effects, respectively, KT is a tolerance rate constant and T is time.
The steady-state plasma concentration of the drug, Css, was used as a predictor of the drug effect through an Emax model:
E = Emax*Css/(C50 + Css)
where Emax is the maximal drug effect in the logit domain, and C50 is the plasma concentration corresponding to half-maximal effect. The model incorporated also individual random effects on K, R0 and DP.
Results: The equilibration rate constant (K) was 0.0252 1/day and PANSS at baseline (R0) was 93.1. Both placebo (EP = -3.48) and drug effects (Emax= -1.98, C50 = 1.46 ng/mL), reduce ultimate PANSS. On the other hand, the effect of the time to last observation, participates in the global effect, deteriorating the overall response that results in dropouts. Finally, due to "tolerance" the drug and/or placebo effect diminishes, and an increase in the PANSS score with time occurs. Since the model is still under development these estimates should be consider as preliminary.
Conclusion: The model presents a platform for analysis and simulation of clinical efficacy trials in schizophrenia. It provides unequivocal parameters that can potentially be related to biomarker responses thereby enabling early prediction of clinical efficacy.
References:
[1] Piotrovsky V. Indirect-response model for the analysis of concentration-effect relationships in clinical trials where response variables are scores. 14th PAGE Meeting, Pamplona, Spain, 2005.