Sensitivity of latin square and ascending dose designs for in vivo QT moxifloxacin positive control to support ICH E14/S7B studies

The Latin square crossover is recommended as best practice for in-vivo QT studies. However, an escalating design may instead be required due to compound long half-life, insufficient toxicity data to inform dose level selection, a need to build up tolerance to the test article to enable safe dosing, or logistics (e.g. avoiding use of multiple simultaneous inhalation dose generation systems thereby reducing risk of contamination). We have performed ICH E14/S7B Q&A compliant in vivo QT validation studies with oral moxifloxacin using both a Latin square and an ascending dose design in primate (n = 4), dog (n = 4 for Latin square and n = 6 for ascending dose design) and minipig (n = 8 for Latin square and n = 5 for ascending dose design). The objective of this work was to compare the sensitivity of the dosing designs and to generate positive moxifloxacin QTc data in support of ICH E14/S7B studies when using an ascending dose design. Increases in QTc at the highest doses for the Latin square and ascending dose designs were + 30 and + 26 msec in primate, +40 and + 34 msec in dog, +86 and + 67 msec in minipig, respectively. Root mean square error (RMSE) values ranged from 5.9 to 10.8 msec for Latin square and 4.1 to 7.3 msec for ascending dose in all species. The smallest statistically detectable difference (SSDD) for the Latin square and ascending dose designs were 10 and 6.6 msec in primate, 10.2 and 5.4 msec in dog, and 11.4 and 10 msec in minipig, respectively. While the SSDD values cannot directly be compared due to the differing N values for dog and minipig, SSDD values were generally low for all species and study designs (6.6–11.4 msec). In conclusion, these data demonstrate that the sensitivity to detect a moxifloxacin-induced increase in QTc was similar with use of either an ascending dose or Latin square design. However, it is important to note that day effect is confounded with treatment effect in an ascending dose design, risking bias in the estimation of a treatment-related effect. Therefore, while sensitivity values were similar between designs, when feasible, Latin square is preferred over ascending dose.