Daren J Austin and Ann Allen
Clinical Pharmacokinetics/Modelling & Simulation, GlaxoSmithKline, Greenford, UK.
Background: GSK123 is a potent and novel selective adenosine agonist under development for the inhaled treatment of asthma, chronic obstructive pulmonary disease (COPD) and allergic rhinitis. Studies with GSK123 via intravenous (IV), inhaled (IH) and intranasal (IN) routes in healthy volunteers and asthmatics have reported increases in heart rate (HR). Initial patient data suggest higher doses (>50mcg) of GSK123 would be required to provide clinical efficacy (FEV1 changes). A modelling strategy was utilised to address the tachycardia risk at higher doses of GSK123.
Objective: The objective of this analysis was to develop and validate a population PD model for GSK123 using data from studies in the Phase I program and to simulate a proposed study to quantify the risk of tachycardia at higher doses.
Methods: Model development and validation was performed using NONMEM V.
There is an apparent hysteresis of 10-20 min between systemic plasma concentrations, C(t), and HR changes. Rather than model PK profiles via three administration routes including an effect compartment, concentrations were transformed using an approximate effect compartment
E(t) = C(t) (1 – exp(– KEO(T– Tdose)))
which introduces a lag time for times less than ~4/KEO. HR changes were fitted to an Emax model with variability in individual responses for H0 (baseline), Hmax, EC50, n (slope), KEO (hysteresis).
The model was initially fitted to the IV data (189 C(t)-HR points) and independently validated using a Bayesian prediction of the IN and IH data (506 points). Finally, the model was used to simulate a repeat dose inhaled study of varying doses of GSK123 using a PK model of IH administration in healthy volunteers. Outcome measure was an estimate of the probability that a patient will experience HR increases greater than some threshold for a given dose.
Results: The intravenous infusion data fitted a sigmoid Emax model with the effect of hysteresis modelled using an approximate lag function where KEO corresponds to lag of ~25 min after Tmax. The model was validated by predicting conc-heart rate response from inhaled and intranasal healthy volunteer (HV)studies – Model confirmed asthmatic patients to have similar HR response as HV’s.
Results from the CTS also showed that the percentage of subjects predicted to experience at least one increase in HR above threshold during the trial (1, 2 or 14 doses) were as follows: 50% of subjects predicted to exceed 30 bpm increase at 75 mcg dose, rising to 75% of the subjects for 150 mcg. At a threshold of 40 bpm increase, 20% of subjects were predicted to exceed it at 75 mcg dose rising to 50% for 150 mcg dose. Results of the CTS were subsequently confirmed with actual patient data in COPD at one of the doses of GSK123 studied.
Conclusions: When corrected for hysteresis via an approximate lag time, HR changes were shown to be independent of administration route of GSK123 and disease status. The risk of tachycardia at proposed doses was shown to be unacceptable. The approximate effect compartment is a useful tool for rapidly generating PD models where hysteresis is evident via multiple administration routes.
Reference: PAGE 12 () Abstr 430 [www.page-meeting.org/?abstract=430]
Poster: poster