Concentration-Response Modeling of ECG Data From Early-Phase Clinical Studies as an Alternative Clinical and Regulatory Approach to Assessing QT Risk - Experience From the Development Program of Lemborexant.

Lemborexant is a novel dual orexin receptor antagonist being developed to treat insomnia. Its potential to cause QT prolongation was evaluated using plasma concentration–response (CR) modeling applied to data from 2 multiple ascending-dose (MAD) studies. In the primary MAD study, placebo or lemborexant (2.5 to 75 mg) was administered for 14 consecutive nights. In another MAD study designed to “bridge” pharmacokinetic and safety data between Japanese and non-Japanese subjects (J-MAD), placebo or lemborexant (2.5, 10, or 25 mg) was administered for 14 consecutive nights. QT intervals were estimated using a high-precision measurement technique and evaluated using a linear mixed-effects CR model, for each study separately and for the pooled data set. When each study was analyzed separately, the slopes of the CR relationship were shallow and not statistically significant. In the pooled analysis, the slope of the CR relationship was –0.00002 milliseconds per ng/mL (90%CI, –0.01019 to 0.01014 milliseconds). The highest observed Cmax was 400 ng/mL, representing a margin 8-fold above exposures expected for the highest planned clinical dose. The model-predicted QTc effect at 400 ng/mL was 1.1 milliseconds (90%CI, –3.49 to 5.78 milliseconds). In neither the J-MAD study nor the pooled analysis was an effect of race identified. CR modeling of data from early-phase clinical studies, including plasma levels far exceeding those anticipated clinically, indicated that a QT effect >10 milliseconds could be excluded. Regulatory agreement with this methodology demonstrates the effectiveness of a CR modeling approach as an alternative to thorough QT studies.