Journal of Electrocardiology 47 (2014) 155–157www.jecgonline.com
Drug-induced QT prolongation when QT interval is measured in each ofthe 12 ECG leads in men and women in a thorough QT studyGopi Krishna Panicker, BHMS, PGDCR,⁎ Vaibhav Salvi, MBBS, PGDCR,
In clinical pharmacology studies assessing drug-inducedrepolarization changes, such as thorough QT (TQT) studies,the QT interval is conventionally measured in a single ECGlead (commonly lead II) by the threshold or the tangentmethod though, in recent times superimposition of medianbeats (SMB) constructed from the 12 leads is also beingincreasingly used [1]. Moxifloxacin, which produces modestdegrees of QT prolongation, has been extensively used as apositive control to demonstrate assay sensitivity in thoroughQT studies. Assay sensitivity is demonstrated when thelower 90% two-sided confidence limit of mean QTcprolongation with moxifloxacin exceeds 5 ms at one ormore time points close to the time of maximum concentra-tion level (Cmax) is achieved by moxifloxacin, which rangesfrom 0.75 hours to 3.5 hours. [2,3]. The selection of lead IIfor QT measurement in most clinical studies was based onthe assumption that the vectors of repolarization in lead II
author. Quintiles Cardiac Safety Services, ResearchLeela Business Park, M.V. Road, Andheri (East),400059, [email protected]
usually result in a long single wave rather than discrete T andU waves and the presence of U wave or notched T waveconfound the accurate measurement of QT interval [4,5].Some authors suggest that lead II most frequently containsthe longest QT interval although recent evaluation has shownthat maximum QT interval occurs very frequently in otherleads [5]. This raises the question what if QT interval ismeasured in any of the other 11 leads in all ECGs in athorough QT study? We studied moxifloxacin-induced QTprolongation in each of the 12 leads in healthy subjects in aTQT study.
Subject and methods
ECG data
Triplicate ECGs were recorded at pre-dose baseline and4 post-dose time points (1, 1.5, 2, and 4 hours) close to theCmax of moxifloxacin from a TQT study conducted in97 healthy subjects (54 males and 43 females) afteradministration of moxifloxacin (400-mg single oral dose)or placebo in an open-label, cross-over design on Day 1 orDay 3 (49 subjects received moxifloxacin on Day 1 and
156 G.K. Panicker et al. / Journal of Electrocardiology 47 (2014) 155–157
placebo on Day 3, while 48 subjects received placebo onDay 1 and moxifloxacin on Day 3). The age distributions ofthe male (age 20–54 years; mean age 34 ± 10 years) andfemale subjects in the study (age 19–56 years; mean age34 ± 12 years) were comparable. Subjects were in a supineposition and at rest for 10 minutes prior to the recording ofECGs at each time point. Digital ECGs were acquired at1000 Hz. (Eli 250, Mortara Instruments) and manuallyanalyzed in a core ECG laboratory using CalECG 3.0 (AMPSLLC) computer software by a team of four physicians whohad undergone training in on-screen measurement of digitalECGs. The readers were blinded to subject demographics,treatment group and time points. The ECGs were allocated toeach reader such that all ECGs of a single subject were read bythe same reader. QT interval was manually measured bythreshold method in temporally aligned five complexes ineach of the 12 leads and averaged for each lead. RR intervalswere measured in the preceding complexes in lead II. T-waveamplitudes in each of the 12 leads were also measured in thecomplexes selected for QT measurement. QT interval wascorrected for the effect of heart rate using the Fridericia'sformula (QTcF). QT interval values were excluded for thoseleads with flat T waves, i.e., T-wave amplitude between −0.1and 0.1 mV [4,6]. For complexes with upright, inverted,biphasic or notched T waves, point of return of the terminallimb of T wave to the isoelectric baseline was considered asthe end of T wave.
Statistical analysis
Data were analyzed using SAS (version 9.2). The placebo-subtracted “change from baseline” values (ΔΔQTcF) for thepost-dose time points in the moxifloxacin arm were calculatedfrom the QT measurements for each of the 12 leads using alinear mixed-effects ANCOVA model with baseline QTcF asa covariate and gender, time (categorical), treatment,treatment-by-time interaction as fixed effects and subject asa random effect. The number and percent of ECGs with flatT waves (−0.1 to 0.1 mV) in each of the 12 leads was noted.
Results
The frequency of ECGs with flat T waves (−0.1 to0.1 mV) for male and female subjects is shown in Table 1.Of the 12 ECG leads, flat T waves were seen in leads aVL,III and V1 in the maximum number of ECGs, which wereexcluded from calculation of ΔΔQTcF.
Lead II showed the typical pattern of moxifloxacin-induced QT prolongation with maximum prolongation seenbetween 1.5 and 4 hours. The lower 90% two-sidedconfidence bound for the mean change in QTcF exceeded5 ms at all four post-dose time points (Fig. 1).
It also exceeded 5 ms at one or more time points in all theother leads in men and women except in leads III, aVL, andV1 in men due to wide confidence limits. In female subjects,the 90% two-sided lower confidence bound of ΔΔQTcFexceeded 5 ms in one or more time points in these leads aswell, despite wide confidence limits because of a greaterdegree of mean ΔΔQTcF.
Discussion
Assay sensitivity in detecting moxifloxacin-induced QTprolongation was demonstrated in all ECG leads except leadsIII, aVL and V1 in men due to wide confidence limits inΔΔQTcF. The confidence limits were wide in women too,but the lower confidence bound exceeded 5 ms at one ormore time points because of greater mean ΔΔQTcF inwomen possibly due to greater sensitivity of women to drug-induced QT prolongation [7].
In most thorough QT studies, QT interval is not measuredin leads with a flat T wave as there is a large variability inidentifying the end of the T wave when T-wave amplituderanges between −0.1 and 0.1 mV [3,5]. The proportion ofECGs with flat T waves was highest in leads III, aVL and V1.As a result, 60% of ECGs in aVL, 45% in lead III and 42% inV1 had to be excluded from statistical analysis. This may beone of the reasons for the wide confidence limits for theΔΔQTcF in these leads.
The high proportion of flat T waves in lead III, aVL andV1 is unlikely to affect QT measurements by the super-imposed median beat method since the limitation of theindistinct T end will be overcome by overlapping complexesfrom leads with more prominent T waves, making itrelatively easy to identify the latest offset of the T wave.On the other hand, it would severely compromise QTmeasurement by the tangent method [1].
Clinical implications
This study confirms the validity of lead II as thepreferred lead for measurement of the QT interval in serialECGs in clinical pharmacology studies and in clinicalpractice. Leads III, aVL and V1 did not demonstrate assaysensitivity for ΔΔQTcF. Leads I, aVR, V5, aVF, V6 and V4
(in order of preference) showed mean estimates, width of90% confidence intervals as well as the time course ofthe moxifloxacin effect that were similar to that in leadII, across all time points and could be considered asalternatives for QT measurement in serial ECGs, when itcannot be measured in lead II.
ALead I Lead II Lead III
Lead aVR Lead aVL Lead aVF
BLead V1 Lead V2 Lead V3
Lead V4 Lead V5 Lead V6
Fig. 1. Mean placebo-adjusted change from baseline in QTcF (ΔΔQTcF) and its 90% two-sided confidence limits obtained in each of the 12 leads in 97 healthysubjects (54 males and 43 females) after administration of 400 mg moxifloxacin. Mean ΔΔQTcF and its 90% confidence limits are shown in red (dark) in menand blue (light) in women. Panel A shows limb leads and Panel B shows precordial leads. Color illustration online.
157G.K. Panicker et al. / Journal of Electrocardiology 47 (2014) 155–157
Conclusions
Moxifloxacin-induced QT prolongation is not satisfacto-rily demonstrated when QT intervals are measured in leadsIII, aVL and V1. These leads should be avoided whenmeasuring QT interval in thorough QT studies.
References
[1] Salvi V, Karnad DR, Panicker GK, et al. Comparison of 5 methods ofQT interval measurements on electrocardiograms from a thoroughQT/QTc study: effect on assay sensitivity and categorical outliers.J Electrocardiol 2011;44(2):96–104.
[2] International Conference on Harmonization of Technical Requirementsfor Registration of Pharmaceuticals for Human Use. The Clinical Eval-uation of QT/QTc Interval Prolongation and Proarrhythmic Potential forNon-Antiarrhythmic Drugs: E14. Geneva, Switzerland: InternationalConference on Harmonization of Technical Requirements for Registra-tion of Pharmaceuticals for Human Use; 2005. http://www.ich.org/fi-
leadmin/Public_Web_Site/ICH_Products/Guidelines/Efficacy/E14/E14_Guideline.pdf. Accessed December 18, 2013.
[3] Rodvold KA, Neuhauser M. Pharmacokinetics and pharmaco-dynamics of fluoroquinolones. Pharmacotherapy 2001;21(10 Pt 2):233S–52S.
[4] Shetty S, Khan M, Salvi S, et al. Do ECG characteristicspredict variability in QT measurements in clinical trials? Indian HeartJ 2005;57:302.
[5] Salvi V, Karnad DR, Kerkar V, et al. Choice of an alternative lead forQT interval measurement in serial ECGs when lead II is not suitable foranalysis. Indian Heart J 2012;64(6):535–40.
[6] Rautaharju PM, Surawicz B, Gettes LS, et al. AHA/ACCF/HRSrecommendations for the standardization and interpretation of theelectrocardiogram: part IV: the ST segment, T and U waves, and the QTinterval. Endorsed by the International Society for ComputerizedElectrocardiology. J Am Coll Cardiol 2009;53(11):982–91.
[7] Darpo B, Karnad DR, Badilini F, et al. Are women more susceptiblethan men to drug-induced QT prolongation? Concentration-QTcmodeling in a phase 1 study with oral rac-sotalol. Br J Clin Pharmacol2013, http://dx.doi.org/10.1111/bcp.12201 [Epub ahead of print].
