Comparison of multiple concentration-QTc modeling approaches

The ICH E14/S7B Q&A guidance for in vivo preclinical studies recommends exposure-response modeling to support the assertion that there were no changes in QTc, to better characterize observed QTc prolongation, and/or when QTc changes are anticipated based on hERG assay results. While the guidance indicates that there are multiple acceptable approaches for concentration-QTc modeling, associated training documents and general industry views indicate that assessments should ideally occur in a separate phase of the study utilizing the same animals as those used for telemetry monitoring to generate a complete exposure profile. However, this “gold standard” approach is not necessarily ideal as it requires additional cost, time, and test material. Alternative options have been proposed including partial toxicokinetic profiles utilizing the same animals as those used for telemetry monitoring or by generating full profiles using data collected from a separate cohort of animals. However, it is unclear as to how concentration-QTc data estimates might differ with alternate approaches. The current study utilized moxifloxacin at 10, 30 and 90 mg/kg, PO in beagle dogs using a 4 × 4 cross-over design monitoring QTc via telemetry over 24 h. Plasma concentrations were determined at 0, 2, 4, 6, 8, 12, and 24 h postdose by using the same animals in a separate phase of the telemetry study or by using a separate cohort of animals. The 10 msec prolongation prediction values were estimated for the partial toxicokinetic profiles using data pairs to estimate plasma levels at each time point using the plasma samples collected during the telemetry phase. The actual exposure values were used for the independent groups. Additionally, simulations of possible toxicokinetic curves for the separate groups were used to estimate representative population values. The 10 msec prolongation prediction values varied minimally, with <300 ng/ml difference among the various estimation methods. Therefore, comparison of these exposure-response modeling methods illustrates the relatively small degree of impact that the different approaches have on QTc prediction values as compared to the gold standard methodology, thus providing multiple viable study design options for researchers to consider in their preclinical testing strategies.