Disease-Drug Model of Methylphenidate (MPH) in Children with Attention Deficit Hyperactivity Disorder Via Longitudinal Meta-analysis
Holly Kimko (1), Ekaterina Gibiansky (2), Leonid Gibiansky (2), H. Lynn Starr (1), Joris Berwaerts (1), Joseph Massarella (1), Robert Armstrong (1), Frank Wiegand (3)
(1) Johnson & Johnson Pharmaceutical Research & Development LLC, NJ, USA; (2) QuantPharm LLC, MD, USA, (3) Janssen Global Services LLC, NJ, USA
Objectives: To develop disease (placebo)-pharmacokinetic-pharmacodynamic (D-PK-PD) population models that relate MPH concentration-time profiles from formulations with various release patterns and placebo with time courses of PD measures (i.e., SKAMP-Composite, PERMP-Attempted, and PERMP-Correct), using a meta-analytic approach to predict the time course of PD measures of any MPH formulation in a pediatric ADHD population.
Methods: Mean MPH concentration-time data from adult healthy volunteer studies of four MPH formulations (Concerta®, Ritalin LA®, Focalin XR® Metadate CD®); mean PD measures versus time data (and the number of subjects represented at each point) from pediatric efficacy trials of the MPH formulations; individual PD data for Concerta® (OROS methylphenidate) from 3 pediatric trials were used.
The pediatric PD data were used to build mixed effects models from the mean data. The observed SKAMP-Composite score was modeled as the sum of placebo and drug effects. The daily variations of the score for the placebo treatment were described by an indirect response model with the score linearly related to B, which is being produced with the constant rate kout and is eliminated with the time-dependent rate α(t) kout B where α(t)=1 at night, and it has different values at two (final model) stretches of the day between waking up and going to bed. The drug effect was described by the Emax function of MPH concentrations with acute tolerance. Acute tolerance was accounted for by dependence of EC50 on time from the morning dose. Similarly, PERMP-Attempted and -Correct scores were jointly modeled.
The final meta-analysis models were used to construct the patient-level population models. Covariate and inter-individual random effects, and the residual variability were estimated.
Results: In total, 11 meta-patients were available. The estimated EC50 value was approximately 7.6 ng/mL, which is close to the MPH Cmax at an intermediate (36-40 mg) dose of extended release MPH in adults. Modeling supports presence of time-dependent tolerance.
Conclusions: The models of the three PD measures have been developed using the meta-analysis and the patient-level population analysis. Internal and external evaluation demonstrated the ability of the models to predict mean time-course of the PD measures and their variability in the target population.
