Methods

In this retrospective study, we evaluated consecutive patients with persistent AF admitted to our service from November 2007 to April 2010 for initiation of dofetilide. We also ensured that no patient had a creatinine clearance of <20 ml/min. Finally, no patient had taken any medication known to be absolutely contraindicated within the past 72 hours of dofetilide initiation (i.e., cimetidine, hydrochlorothiazide, ketoconazole, megestrol, prochlorperazine, trimethoprim [with or without sulfamethoxazole], and verapamil). Amiodarone was discontinued for at least 30 days in all patients before initiation of dofetilide. The study was approved by the Institutional Review Board.

Dofetilide was not initiated in any patient in whom the QTc was >460 ms during ongoing AF (≤500 ms in patients with concomitant intraventricular conduction delay). The electrocardiograms (ECGs) used for QT measurement and correction were obtained 2 hours after each dofetilide dose, at peak plasma concentration. The QT was measured in lead II from the beginning of the QRS complex to the point the T wave reached the baseline, as previously described. If lead II did not have a distinct T wave or if the end of the T wave could not be discerned because of underlying fibrillation waves, then the longest QT in any lead was used. In any given patient, the same electrocardiographic lead was used to assess the QT interval throughout the study. For the first 20 subjects, the QTs in AF were averaged over 10 beats. No significant difference was obtained compared with only 3 beats; thus, for subsequent subjects, the latter method was applied. Two investigators (MA and DLM) independently assessed the QT interval on all ECGs; there was a very high degree of reproducibility as reflected by a correlation coefficient of 0.94.

The QT measurements were corrected for heart rate using Bazett’s formula (QT/2√RR)(QT/RR−−−√2)
( QT / RR 2 )
. In patients with underlying AF, the average RR interval was calculated over a 10-second window. The same method was used to obtain the average RR interval for ECGs in SR to account for possible sinus arrhythmia. The QT was then corrected using 2 other accepted formulas: (1) Fridericia formula (QT/3√RR)
( QT / RR 3 )
and (2) the Framingham linear formula [QT + 0.154 (1 − RR)]. If dofetilide restored SR, an ECG was obtained and the QTc was compared with the QTc on the last ECG obtained during ongoing AF. If AF remained present 2 hours after the sixth dose of dofetilide, electrical cardioversion was performed to restore SR. Whether SR was restored by dofetilide alone or electrical cardioversion, an ECG to assess the QT and RR intervals was obtained after SR had been present for at least 5 minutes to allow for resolution of any QT/RR hysteresis.

The initial dofetilide dose was chosen depending on the patient’s renal function, which was estimated using the Cockcroft formula ([creatinine clearance (CrCl) = (140 − patient age) × body weight/(serum creatinine × 72)]; for women, the resulting value was multiplied by 0.85). When the CrCl was >60 ml/min, dofetilide was initiated at a dose of 500 μg every 12 hours. The starting dose was 250 μg q12 hours if the CrCl was 40 to 60 ml/min and 125 μg q12 hours if the CrCl was 20 to 40 ml/min. An ECG was obtained 2 hours after each dose of dofetilide or immediately after spontaneous or electrical cardioversion to SR. If the QTc increased by 15% from baseline after the first dose or >500 ms (550 ms in patients with an intraventricular conduction delay) at any time, the next dose was held. After the QTc normalized, dofetilide was resumed at 1/2 the dose. If the same degree of QTc prolongation was observed with reduced dose, dofetilide was reduced to 125 μg bid or permanently discontinued. All decisions regarding dose titration or drug stoppage were made based on QTc calculated by Bazett’s formula.

Continuous variables are represented as mean ± SD. A paired Student’s t test was used to compare the average QTc intervals in AF and SR as defined by the 3 formulas. We corrected the QT interval using each of the 3 formulas during ongoing AF, and using Pearson’s correlation, we assessed the relation of the QTc to RR interval. Lastly, we identified 32 patients, who after receiving a dose of dofetilide (e.g., dose number 1, 2, 3, 4, 5, or 6), initially remained in AF (such that an ECG was obtained and a QT interval calculated) but converted to SR (such that a second ECG was obtained and a QT interval calculated) before the next dose of dofetilide was administered or underwent electrical cardioversion and ECGs were obtained before and after cardioversion. Thus, the comparison of QTc between AF and SR reflects that attributable to the change in rhythm alone, that is, were performed during the same dosing interval. A Bland-Altman analysis was performed in these 32 patients and used to evaluate the correlation of QTc intervals for individual patients using each method in AF to SR. A p <0.05 was considered statistically significant.