Problems of heart rate correction in assessment of drug-induced QT interval prolongation

Introduction: Estimation of QT interval prolongation belongs to safety assessment of every drug. Among unresolved issues, heart rate correction of the QT interval may be problematic. This article proposes a strategy for heart rate correction in drug safety studies and demonstrates the strategy using a study of ebastine, a nonsedating antihistamine. Methods and Results: Four-way cross-over Phase I study investigated 32 subjects on placebo, ebastine 60 mg once a day, 100 mg once a day, and terfenadine 180 mg twice a day. Repeated ECGs were obtained before each arm and after 7 days of treatment. The changes in heart rate-corrected QTc interval were investigated using (A) 20 published heart rate correction formulas, (B) a correction formula optimized by QT/RR regression modeling in all baseline data, and (C) individual corrections optimized for each subject by drug-free QT/RR regression modeling. (A) Previously published correction formulas found QTc interval increases on terfenadine. The results with ebastine were inconsistent. For instance, Bazett's and Lecocq's correction found significant QTc increase and decrease on ebastine, respectively. The results were related (\r\ > 0.95) to the success of each formula (independence of drug-free QTc and RR intervals). (B) The pooled drug-free QT/RR regression found an optimized correction QTc = QT/RR0.314. QTc interval changes on placebo, ebastine 60 mg, ebastine 100 mg, and terfenadine were -1.95 +/- 6.87 msec (P = 0.18), -3.91 +/- 9.38 msec (P = 0.053), 0.75 +/- 8.23 msec (P = 0.66), and 12.95 +/- 14.64 msec (P = 0.00025), respectively. (C) Individual QT/RR regressions were significantly different between subjects and found optimized corrections QTc = QT/RRalpha with alpha = 0.161 to 0.417, Individualized QTc interval changes on placebo, ebastine 60 mg, ebastine 100 mg, and terfenadine were -2.76 +/- 5.51 msec (P = 0.022), -3.15 +/- 9.17 msec (P = 0.11), -2.61 +/- 9.55 msec (P = 0.19), and 12.43 +/- 15.25 msec (P = 0.00057, respectively. Drug-unrelated QTc changes up to 4.70 +/- 8.92 msec reflected measurement variability. Conclusion: Use of published heart rate correction formulas in the assessment of drug-induced QTc prolongation is inappropriate, especially when the drug might induce heart rate changes. Correction formulas optimized for pooled drug-free data are inferior to the formulas individualized for each subject, Measurement imprecision and natural variability can lead to mean QTc interval changes of 4 to 5 msec in the absence of drug treatment.