A retrospective review of the effects of moxifloxacin in NHP at four different test sites

With adoption of the ICH E14/S7B Q&A best practice guidance, nonclinical CV data sets can be used in place of clinical data to support of a TQT waiver application. However, as part of this guidance each laboratory is required to demonstrate adequate sensitivity of their specific in vivo model using a known positive control agent (e.g., moxifloxacin) or through retrospective review of historical data sets. These requirements raise questions around how often a positive control study should be conducted and how stable the effects of moxifloxacin are across multiple laboratories. In the current review, we assessed the effects of moxifloxacin in non-human primates (NHP) from four test sites. All four laboratories administered moxifloxacin at 10, 80 and 175 mg/kg by oral administration and evaluated cardiovascular parameters for 24 h post administration. The study designs between labs were highly comparable with each site using a Williams Latin square 4*4 crossover design in male animals, with the exception of site 1 which used female animals. Concentration-QTc modeling was also conducted at all sites with minor site-to-site variances in the blood sample collection methods and timing observed. All sites illustrated dose dependent increases in QTc prolongation. Administration at 175 mg/kg resulted in QTc prolongation of up to 21.9, 25.3, 16.9 and 45.1 msec, relative to control, at test sites 1, 2, 3 and 4 respectively. Whilst a review of study specific least significant difference, ranged from 8.4 to 10.2 msec, and residual error ranged from 4.8 to 5.9 msec; with conc-QT modeling, indicating that concentration ranges of 2444–3938 ng/ml would be expected to induce a 10-msec prolongation. These results indicate that when following best practice methodologies, NHP CV models are highly consistent, independent of laboratory. This suggests that repeat conductance across of positive control studies, across multiple labs is not required and reliance on historical data sets alone should suffice to confirm model suitability.