gms | German Medical Science

Introduction: Thorough QT (TQT) studies are performed in accordance with the ICH E14 guidelines in order to demonstrate that new investigational drugs do not alter the corrected QT interval (QTc) of the electrocardiogram (ECG). We previously proposed a new crossover design for TQT studies, which used two placebo periods instead of one [1], [2].

This design was expected to be statistically more efficient in estimating the placebo-corrected QTc interval changes from baseline (CfB), under the assumption of a low correlation of the data between both placebo periods. Theoretically, the corresponding confidence intervals (CI) of the CfB estimates were expected to be about 1 - √(3/4) =13.4% smaller when using the double placebo design, compared to the conventional single placebo design. This would allow for a reduction of the sample size from 40 to 30 subjects while maintaining 90% power for the primary study objectives [1], compared with the conventional single placebo design [1], [2].

This novel design was implemented in a TQT study of the sodium glucose cotransporter-2 (SGLT-2) inhibitor empagliflozin for the treatment of patients with type 2 diabetes [3]. Here we present the outcome of the statistical efficiency in this trial.

Methods: The actual width of the CIs for the estimated placebo-corrected QTc interval mean CfB (primary analysis) and the estimated placebo-corrected QTc interval CfB per time point (secondary analysis using the repeated measures cross-over [RMC] mixed model [4]) were compared using a) the double placebo design, with b) the single placebo design (by leaving out the data of one placebo period).

In addition, the outcomes of various sensitivity analyses were compared to those of the primary and secondary analyses. This included analyses which accounted for different baseline covariates (period baseline only vs. period baseline and global mean baseline) into the mixed model for the QTc CfB [2], [5].

Results: In the primary analysis, the average point estimate of the actual reduction of CI widths was 18.9%. For the repeated measurements in the secondary analysis, the average point estimate of the actual reduction was 11.8%; both values were similar to the expected value. Hence in this trial, the assumption of a low correlation between the CfB values of the 2 placebo periods was fulfilled. Additional sensitivity analyses led to similar results as the primary analysis (specifically the inclusion of the global mean baseline as an additional covariate [5]) and did not lead to major changes in the estimates of the placebo-corrected mean QTc CfB and its standard error.

Conclusion: All primary, secondary and sensitivity analyses demonstrated that empagliflozin was not associated with QTc interval prolongation. In this TQT trial, expectations for the reduction of the standard errors of the point estimates were met; hence, the double placebo design proved to be statistically more efficient. This makes it possible to reduce the overall sample size by 25% compared to the single placebo design.