Computer Simulation: A Useful Tool for Trial Dosage Selection

There is often limited data, and thus great risk, in choosing a specific dose for a pivotal efficacy trial. Recently computer simulations of pharmacokinetic (PK) and pharmacodynamic (PD) models developed with data from two phase I trials of sodium dichloroacetate (Ceresine)(DCA), a pyruvate dehydrogenase kinase inhibitor, were used to guide in dosage selection of a pivotal efficacy trial in TBI. The PD effect of DCA is reduced lactate concentrations.

37 human volunteers received cumulative (2 doses over 8 hours) i.v doses of placebo, 45 mg/kg, 90 mg/kg or 150 mg/kg DCA. An additional 25 patients with TBI received a single i.v. dose of placebo, 60 mg/kg, 100 mg/kg or 200 mg/kg DCA in a dose-escalating, randomized, placebo controlled study. An additional 12 TBI patients received multiple doses of DCA over a 3 day period. NONMEM version IV software was used to develop PK and PD models, and evaluate the impact of covariate inclusion in the model. Serum and CSF lactate concentrations were related to serum and CSF DCA concentrations by an indirect physiologic response model with the stimulation of Kout (Type IV model). These PK and PD models were then used with the BioMed simulation program to provide predicted concentrations of DCA in serum and CSF, as well as predicted concentrations of lactate.

48 concentrations were simulated for 1000 subjects at each dosing level. The simulated predicted concentrations of DCA and lactate were plotted versus time to aid in determining the optimal dose. Based on observations of plots generated, it appeared that the optimal dose to reach near maximal suppression of CSF lactate was 150 mg/kg every 12 hours x 4 doses, followed by 75 mg/kg every 12 hours x 6 doses. No apparent increase in PD effect was predicted by increasing the dose.

Computer simulation, when combined with PK and PD models, provided a useful tool for deciding on a dose for a proposed phase III pivotal study of DCA.