Abstract
Background: Careful assessment of QT interval prolongation is required before novel drugs are approved by regulatory authorities. The choice of the most appropriate method of electrocardiogram (ECG) acquisition and QT/RR interval measurement in clinical trials requires better understanding of the differences among currently available approaches. This study compared standard and Holter‐derived 12‐lead ECGs for utility in detecting sotalol‐induced QT/QTc and RR changes. Manual methods (digitizing pad and digital on‐screen calipers) were compared for precision of QT and RR interval measurement.
Methods and Results: Sixteen hundred pairs of serial 12‐lead digital ECGs were recorded simultaneously by standard resting ECG device and by continuous 12‐lead digital Holter over 3 days in 39 healthy male and female volunteers. No therapy was given on the 1st day followed by 160 mg and 320 mg of sotalol on the 2nd and 3rd day, respectively. Holter‐derived and standard ECGs produced nearly identical sotalol‐induced QT/QTc and RR changes from baseline, as did the manual digipad and on‐screen caliper measurements. The variability of on‐screen QT measurement in this study was greater than that of digipad.
Conclusions: Digital 12‐lead Holter and standard 12‐lead ECG recorders, as well as the manual digitizing pad and digital on‐screen calipers, are of equal utility for the assessment of drug‐induced change from baseline in QT and RR interval, although the variability of the on‐screen method in this study was greater than of the digipad.
Keywords: QT prolongation, electrocardiography, digital 12‐lead Holter, cardiac repolarization
Regulatory authorities require that the effect of novel drugs on cardiac repolarization be thoroughly investigated in clinical trials. 1 , 2 , 3 Regulatory actions consistently indicate that even small degrees of drug‐induced QT prolongation relative to placebo will raise safety concerns, necessitating that favorable risk–benefit ratio be demonstrated before marketing approval is granted. Several efficacious nonantiarrhythmic drugs shown to cause QT prolongation have been removed from the market (e.g., terfenadine, astemizole, cisapride, grepafloxacin, and sertindole). 4 , 5 , 6 , 7 Using a precise method of QT and RR interval measurement is critical for the detection of small but potentially clinically significant drug‐induced QT prolongation. The high resolution digitizing board is the standard approach for manual QT interval measurement on paper electrocardiogram (ECG), 8 , 9 although automated measurements by the ECG recorder's computer algorithm are also commonly used. The U.S. Food and Drug Administration and Health Canada, jointly proposed that for new drug approvals, digital ECGs should be obtained with annotation of the onset and offset of the prolonged QT intervals. 3 Digital on‐screen calipers would be the method of choice for manual QT interval measurement on digital ECGs. In addition to standard resting supine ECGs, new Holter technology allows for continuous ambulatory recording of 12‐lead digital ECGs for 24 hours. This Holter approach should prove useful for the assessment of cardiac repolarization at numerous discrete time points following drug administration or by continuous beat‐to‐beat analysis, as well as for detecting cardiac arrhythmia.
This study compared the utility for QTc risk assessment of ECGs recorded by standard or digital 12‐lead Holter devices, as well as the precision of QT and RR interval measurement by the manual digitized systems (digitizing board and digital on‐screen calipers) on standard and Holter‐derived ECGs.
METHODS
Study Design
This open‐label, nonrandomized study included a fixed treatment sequence administered on three successive days: 24‐hour baseline (Day –1), a single 160 mg dose of sotalol (Betapace® 80 mg tablets, Berlex Laboratories Montville, NY) (Day 1), and a single 320 mg dose of sotalol (Day 2). Subjects were dosed at 8 AM under fasting conditions and received standard meals at noon and 6 PM. Based on PK/PD modeling of published data on sotalol‐induced QTc prolongation, 10 , 11 subjects were withdrawn on Day 1 if predose QTc was >410 ms or any QTc within 6 hours after dosing was >450 ms (by ECG recorder's algorithm). Subjects with heart rate <50 bpm or abnormal serum potassium and magnesium predose on days 1 and 2 were also withdrawn. The study was conducted at Pharmacia's Clinical Research Unit (Kalamazoo, MI). All subjects gave written informed consent to the study protocol approved by an independent Institutional Review Board.
ECG Recordings
Standard digital 12‐lead ECGs were recorded for 10 seconds after 5 minutes supine rest (ELI‐200, Mortara Instrument (Milwaukee, WI); “10‐by‐2” rhythm strip format, paper speed 25 mm/s, amplitude 1 mV/10 mm). The recordings were acquired with a sampling frequency of 500 Hz (2 ms resolution) and with a 16‐bit amplitude (2.5 μV) resolution. Sixteen standard ECGs were recorded at identical times on Days 1 and 2 (immediately predose and at 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 6, 8, 10, 13, 16, and 22.5 hours after dosing) and at corresponding clock times at baseline (Day –1). Digital 12‐lead Holter ECGs were recorded continuously for 22.5 hours on Days –1, 1, and 2 (H12 Recorder, Mortara Instrument). Holter recordings were sampled at 180 Hz (5.6 ms resolution) and with a 16‐bit amplitude (2.5 μV) resolution. They were time‐stamped simultaneously with the start of each standard ECG. Sixteen discrete 12‐lead ECG digital files per study day were derived from Holter to the H‐Scribe analysis system and printed out (10‐s rhythm strip, paper speed 25 mm/s, amplitude 1 mV/10 mm) at the time points coinciding with standard ECGs. Holter and standard ECGs were recorded from the same location on each subject by dual snap electrodes (Nikomed USA, Inc. Doylestow, PA #4500) utilized for precordial leads. Single snap electrodes (Nikomed U.S.A. Inc. #4520) were utilized for limb leads. All study equipments were validated and recently serviced.
QT and RR Interval Measurement Systems
Two manual ECG interval measurement systems were used: first, a digitizing board (“digipad,” eResearch Technology, Inc., Philadelphia, PA) with magnification of the paper ECG coupled with digitization software whereby point‐to‐point determination of onset and offset points was made by two trained analysts; 9 second, an electronic caliper system (CalECG Software v1.0, AMPS, LLC New York, NY) applied on a computer screen to digital ECG files from the standard ECG recorder by the same trained analysts. At the time of the study, the on‐screen calipers could not be directly applied to the source digital ECG files from H12 Holter recorder, so the 16 Holter‐derived paper ECG printouts were scanned into a portable network graphics file at 300 dpi and calipers were applied to the scanned image. The RR and QT interval measurements by both manual methods were an average of three consecutive sinus rhythm beats. QT intervals were corrected for heart rate using Fridericia's (cube root) correction (QTcF).
The repeatability of the digipad method was examined to support its use as a standard for pairwise comparisons between measurement methods. The inter‐ and intraobserver variability in digipad measurements was tested by repeating the initial digipad analysis (ED‐1) twice within 5 and 6 months (ED‐2 and ‐3, respectively) on all standard ECGs by the same two analysts. In ED‐2, all ECGs were randomized between the two analysts. In ED‐3, all ECGs from the same subject were read by one analyst (50% by the same analyst from ED‐1 and 50% by the other).
Statistical Analysis
Primary statistical comparisons were made for manual measurements of QT, QTcF, and RR intervals in limb lead II of the standard and Holter‐derived 12‐lead ECGs. Statistical comparisons measurements made in precordial lead V5 produced similar results and are not presented. Measurement methods were compared in a pairwise fashion, whereby digipad was the reference for all comparisons. Descriptive statistics and the Bland–Altman plots 12 were used to compare methods, incorporating results for absolute QT/QTcF interval values and QT/QTcF changes from baseline from all subjects. Baseline was the average of all 16 ECGs obtained on Day–1. It is recognized that these do not represent independent data points, but support the primary objective of comparing ECG acquisition and measurement methods across a wide range of QT intervals.
RESULTS
Subject Disposition and ECG Acquisition
Thirty‐nine healthy adult volunteers (11 females) aged 18–45 years (mean 27 years), weighing 47–108 kg (mean 74 kg) with a Body Mass Index of 18.2–30.8 kg/m2 (mean 24.4 kg/m2) were evaluated. There were 33 Caucasians, 2 blacks, and 4 with unspecified race. All subjects received 160 mg dose of sotalol and 22 of them (all males) also received 320 mg dose. Seventeen subjects (11 females) were withdrawn on Day 1; 12 had QTc > 450 ms within 6 hours of dosing; 2 had predose heart rate <50 bpm; and one each had unstable T wave, difficulty with blood draws and urinary tract infection.
A total of 3200 ECG tracings were obtained from standard ELI200 and H12 Holter recorders (22 subjects × 48 ECGs + 17 subjects × 32 ECGs = 1600 per acquisition method). ECG results based on digipad measurements on standard ECGs are presented in Table 1.
Table 1.
ECG Results by Day‐Based Study on Digipad Measurements (ED‐1) in Limb Lead II of Standard ECG
| Day −1 Mean (Range) | Day 1 Mean (Range) | Day 2 Mean (Range) All Subjects* (n = 22) | |||||
|---|---|---|---|---|---|---|---|
| All Subjects (n = 39) | Males (n = 28) | Females (n = 11) | All Subjects (n = 39) | Males (n = 28) | Females (n = 11) | ||
| RR (ms) | 910.7 | 934.2 | 850.8 | 1030.6 | 1055.8 | 966.4 | 1090.1 |
| (737 to 1088) | (744 to 1088) | (737 to 1004) | (843 to 1176) | (876 to 1176) | (843 to 1103) | (945 to 1233) | |
| HR (bpm) | 67.2 | 65.4 | 71.7 | 59.5 | 57.9 | 63.4 | 55.8 |
| (56 to 82) | (56 to 81) | (60 to 82) | (51 to 72) | (51 to 69) | (55 to 72) | (49 to 64) | |
| QT (ms) | 355.9 | 354.7 | 359.0 | 393.6 | 388.3 | 407.3 | 412.9 |
| (318 to 395) | (318 to 395) | (337 to 386) | (357 to 432) | (357 to 424) | (378 to 432) | (379 to 446) | |
| QTcF (ms) | 368.1 | 363.6 | 379.7 | 390.8 | 382.1 | 412.9 | 401.9 |
| (343 to 402) | (343 to 293) | (358 to 402) | (362 to 431) | (362 to 407) | (386 to 431) | (374 to 430) | |
| ΔRR (ms) | N/A | N/A | N/A | 119.9 | 121.6 | 115.6 | 164.5 |
| (6 to 229) | (25 to 203) | (6 to 229) | (61 to 254) | ||||
| ΔHR (bpm) | N/A | N/A | N/A | –7.7 | –7.5 | –8.3 | –10.1 |
| (–15 to 0) | (–15 to –2) | (–15 to 0) | (–18 to 4) | ||||
| ΔQT (ms) | N/A | N/A | N/A | 37.8 | 33.6 | 48.3 | 60.9 |
| (20 to 66) | (20 to 51) | (34 to 66) | (41 to 84) | ||||
| ΔQTcF (ms) | N/A | N/A | N/A | 22.7 | 18.6 | 33.2 | 40.3 |
| (5 to 45) | (5 to 32) | (22 to 46) | (23 to 57) | ||||
*Subjects evaluated on Day 2 were males only.
ED‐1 = first measurement by digipad on standard ECG; QT = absolute QT interval; QTcF = absolute QT interval corrected for heart rate by Fridericia's formula; RR = absolute RR interval; HR = heart rate; Δ= change from baseline, where the baseline constitutes an average of all 16 measurements on Day –1; N/A = not applicable.
Reproducibility of the Digipad Measurement Method
Table 2 presents the agreement between the QT, QTcF, and RR intervals produced by the initial digipad measurement and the two repeated measurements performed by the same two analysts in limb lead II of all 1600 standard ECGs, 5 and 6 months apart.
Table 2.
Agreement Between Repeated QT, QTcF, and RR Interval Measurements by Digipad in Limb Lead II of Standard ECG
| QT Interval | QTcF Interval | RR Interval | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Absolute | Change from Baseline | Absolute | Change from Baseline | Absolute | Change from Baseline | |||||||
| Pairwise comparisons | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference | SD |
| ED‐1 vs ED‐2 | −11.6 | 9.8 | 1.0 | 8.7 | −11.5 | 9.8 | 1.6 | 8.6 | −1.0 | 6.3 | −0.1 | 6.5 |
| ED‐1 vs ED‐3 | −6.4 | 9.7 | 0.2 | 8.2 | −6.3 | 9.7 | 0.4 | 8.2 | −1.0 | 5.6 | −0.1 | 5.5 |
ED‐1 = first digipad measurement; ED‐2, ‐3 = digipad measurements repeated within 5 and 6 months after ED‐1 by the same two analysts; SD = standard deviation of difference; QTcF = QT interval corrected for heart rate by Fridericia's formula.
Comparability of ECG Recording and Measurement Methods
Table 3 presents the pairwise comparisons among QT, QTcF, and RR interval measurement methods on simultaneously recorded standard and Holter‐derived 12‐lead ECGs. Agreement among selected pairwise comparisons is depicted by Bland–Altman plots in Figure 1 (Parts 1A–3B). The mean difference indicates lack of agreement among the compared methods, 12 while the degree of variation about the mean difference is described by the range of ±2 standard deviations, expected to include 95% of differences in individual pairwise comparisons.
Table 3.
Agreement Between Manual Methods for the QT, QTcf, and RR Interval Measurement in Limb Lead II of Standard and Holter‐Derived Digital 12‐Lead ECGs
| QT Interval | QTcF Interval | R‐R Interval | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Absolute | Change from Baseline | Absolute | Change from Baseline | Absolute | Change from Baseline | |||||||
| Pairwise comparisons | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference (ms) | SD | Mean difference | SD |
| ED‐1 vs HD | −3.3 | 10.5 | 0.5 | 10.4 | −2.7 | 12.5 | −0.3 | 12.6 | −3.5 | 67.3 | 7.3 | 72.9 |
| EX vs HX | 5.2 | 19.1 | 2.4 | 18.2 | 6.0 | 20.1 | 1.4 | 19.8 | −5.0 | 70.3 | 6.8 | 74.0 |
| ED‐1 vs EX | −14.8 | 13.2 | −1.2 | 14.0 | −16.0 | 13.3 | −0.7 | 13.9 | 9.0 | 8.6 | 2.2 | 8.2 |
| HD vs HX | −6.1 | 15.6 | 0.9 | 12.8 | −7.2 | 15.2 | 1.1 | 12.0 | 7.6 | 19.0 | 1.7 | 18.6 |
ED‐1 = first measurement by digipad on standard ECG; EA = measurement by the standard ECG recorder's automated algorithm; EX = measurement by digital on‐screen calipers on standard ECG; HD = measurement by digipad on Holter‐derived ECG; HX = measurement by digital on‐screen caliper on Holter‐derived ECG; SD = standard deviation of difference; QTcF = QT interval corrected for heart rate by Fridericia's formula.
Figure 1.
Bland–Altman plots of agreement between digipad and on‐screen measurements on standard and Holter‐derived ECGs. Abbreviations as in Table 3. Part 1A: QT interval by digipad on standard and Holter‐derived ECG (ED‐1 vs HD). Part 1B: QT change from baseline by digipad on standard and Holter‐derived ECG (ED‐1 vs HD). Part 2A: QTcF interval by digipad on standard and Holter‐derived ECG (ED‐1 vs HD). Part 2B: QTcF change from baseline by digipad on standard and Holter‐derived ECG (ED‐1 vs HD). Part 3A: RR interval by digipad on standard and Holter‐derived ECG (ED‐1 vs HD). Part 3B: RR change from baseline by digipad on standard and Holter‐derived ECG (ED‐1 vs HD).
Outlier Analysis for QT Change from Baseline
Table 4 presents the outlier analysis, that is, the classification of individual QTcF changes from baseline determined by each measurement method on standard and Holter‐derived ECGs. Considering the initial digipad QT measurements on 1600 standard ECGs obtained on Days –1 to 2, (ED‐1), the absolute QTcF change from baseline of 30–60, 61–90, and >90 ms was observed in 36.3%, 9.1%, and 0.2% of all ECGs, respectively. Considering the initial digipad measurements on Day 1 ECGs, a change from baseline in QTc interval corrected by Fridericia's formula of 30–60, 61–90, and >90 m was observed in 41.0%, 12.4% and 0% of ECGs in women, and in 35.3%, 8.4%, and 0.3% of ECGs in men.
Table 4.
Classification of Individual QTcF Changes from Baseline Determined by Manual Measurement Methods in Limb Lead II of Standard and Holter‐derived ECGs
| QTcF Change From Baseline (ms) | ED‐1 N = 975 | HD N = 961 | EX N = 979 | HX N = 960 | ||||
|---|---|---|---|---|---|---|---|---|
| n | % | n | % | n | % | n | % | |
| 0 to 30 | 530 | 54.4 | 527 | 54.8 | 529 | 54.5 | 547 | 57.0 |
| 31 to 60 | 354 | 36.3 | 346 | 36.1 | 338 | 34.5 | 311 | 32.4 |
| 61 to 90 | 89 | 9.1 | 86 | 8.9 | 93 | 9.5 | 96 | 10.0 |
| >90 | 2 | 0.2 | 2 | 0.2 | 15 | 1.5 | 6 | 0.6 |
Abbreviations as in Table 3.
N = the total number of ECGs (Days 1 and 2) evaluated; n = the total number of ECGs (Days 1 and 2) where a given QTcF change was detected.
Comparability of Methods for Detecting QT and RR Change from Baseline Over Time
Figure 2 (parts 1–3) presents the mean changes from baseline in QT, QTcF, and RR intervals after dosing with sotalol on Days 1 and 2, as detected by each measurement method on standard and Holter‐derived ECGs.
Figure 2.
Pairwise method comparisons by mean QT, QTcF, and RR change from baseline over time after oral administration of 160 and 320 mg of Sotalol. (Legend: •‐•‐• ED‐1; ▪‐▪‐▪ HD; 
EX;
HX. Abbreviations as in Table 3.) Part 1: Mean QT change from baseline. Part 2: Mean QTcF change from baseline. Part 3: Mean RR change from baseline.
DISCUSSION
The regulatory requirement for preapproval characterization of the potential of new drugs to prolong the QT interval 1 , 2 , 3 warrants search for optimal methods of ECG acquisition and measurement. Inadequate acquisition or measurement of ECG data may lead to the approval of a drug with unacceptable risk–benefit ratio or to inappropriate discontinuation of a promising drug during development. 13 , 14 , 15 , 16
This is the first study to compare the standard 12‐lead ECGs to discrete ECGs derived from continuous digital 12‐lead Holter recorder for their utility in the assessment of drug‐induced changes in QT/QTc and RR intervals, measured on a large number of serial ECGs (1600 simultaneously recorded pairs). It is also the first study to compare the precision of QT/QTc and RR interval measurement by the two most relevant manual digitized methods (digipad and on‐screen calipers).
While substantial error was observed in repeated digitizing board measurement by untrained volunteers, 11 others reported low relative interobserver error in QT interval measurement when this method was used by trained observers. 17 , 18 A small mean QTc change from baseline of only 6 ms caused by terfenadine was detected by digipad. 19 We performed two repeated measurements by digipad on 1600 standard ECGs to examine the reproducibility of this method in our study. The repeated absolute QT interval measurements were on average 11.5 and 6.4 milliseconds longer than the initial values of 5 and 6 months earlier, indicating a small but systematic inter‐run bias in the method for the measurement of the absolute QT interval. We did not find differences between readers, nor when ECGs were grouped by subject versus being read at random, and we were unable to account for the drift in the repeated digipad measurements of the absolute QT interval over time. This implies that for any manual method, the variability of QT interval measurement over time on different prospectively obtained ECGs should be taken into account. If the shorter duration of the trial makes it feasible, it would be prudent to ensure that all ECGs from the same subject are manually read within a reasonably short time frame (e.g., 1 week). However, in this study both repeated digipad analyses showed only a negligible bias from the initial analysis for the mean differences in the change from baseline in QT (1.0 and 0.2 ms) and QTcF (1.6 and 0.4 ms). The mean differences between repeated and the initial digipad analyses were small for the absolute RR interval (–1.0 ms) and virtually nonexistent for the RR change from baseline (–0.1 ms). Such results indicate good reproducibility of digipad for detection of sotalol‐induced QT/QTc and RR changes from baseline in this trial and justify the use of digipad as a reference method for QT interval measurement to which other methods were compared.
Numerous serial ECGs recorded at baseline and upon repeated dosing with novel drugs and placebo are necessary for conclusive characterization of the potential of a new drug to prolong the QT interval. 14 , 15 Such intensive evaluation is generally performed in healthy volunteers hospitalized in well‐equipped Phase I units, and not in patients with the target disease indication enrolled in large multicentric Phase II/III clinical trials, where the vast majority of QT and RR data come from infrequently obtained standard ECGs with often a single baseline recording. If QT and RR intervals could be reliably obtained from ambulatory digital 12‐lead Holter recordings, this technology may become the method of choice for more intensive QT assessment in clinical trials. 15 In our study, QT, QTcF, and RR data produced by digipad on standard ECGs were essentially equivalent to those from digital 12‐lead Holter ECGs (Table 3). We have observed a small mean difference in the absolute QT and QTcF intervals (–3.3 and –2.7 ms) and a negligible mean difference in the QT and QTcF change from baseline (0.5 and –0.3 ms) when digipad was applied to standard ECGs and Holter‐derived paper ECG copies, respectively. There were only two ECGs where the individual QTcF changes by digipad differed from those by on‐screen by 60 ms or more (Fig. 1, Part 2B). Upon reinspection of these ECGs, we found that both had a flat T wave in limb lead II, making it technically difficult to identify the same end point of T wave by both manual methods. Digipad RR data from standard ECGs were similar to those from Holter‐derived ECGs (mean differences, −3.5 ms for absolute RR and 7.3 ms for RR change from baseline), although with a considerably large standard deviation of difference. The mean difference of 7 ms between the RR changes from baseline produced by digipad on standard and Holter ECGs is so small relative to the total duration of RR interval that it would not affect the QT/RR relationship. The mean differences among the absolute QT, QTcF, and RR data produced by on‐screen calipers applied to standard ECGs imported as digital files and to Holter ECGs scanned from paper were also small, although with greater standard deviation than digipad (Table 3).
Overall, despite somewhat different lead positions (12‐lead Holter ECG has electrodes only on the torso whereas the standard ECG also uses the limb leads) and slight differences in the sampling frequency of ECG recordings, our results validate the utility of digital 12‐lead Holter for the assessment of QT prolongation in clinical drug research. With its inherent ease of capturing ECG continuously over long periods of time, digital 12‐lead Holter would facilitate more intensive QT risk assessment in clinical trial patients who cannot comply with complex standard ECG schedules at baseline and on treatment. Holter will also make digital 12‐lead ECG recordings available for retrospective analysis at critical time points after dosing that are difficult or impossible to capture by advance scheduling (maximum plasma concentration of the parent drug and active metabolites, clinical adverse events possibly related to QT prolongation). Numerous QT and RR intervals derived from continuous Holter ECGs will allow for the employment of subject‐specific QT correction formulas, which might be more accurate than Bazett's or Fridericia's in evaluating the true drug‐induced QT effect at different heart rates. 15 , 20
We observed negligible bias between the digipad and on‐screen methods for QT, QTcF, and RR changes from baseline produced on standard and Holter‐derived ECGs (mean differences: –1.2 vs 0.9 ms for QT change, –0.7 vs 1.1 for QTcF change, and 2.2 vs 1.7 ms for RR change on standard and Holter ECGs, respectively), although the standard deviations of the difference between RR interval measurements on standard and Holter ECGs were quite high for both digipad and on‐screen methods. The absolute QT, QTcF, and R‐R intervals by on‐screen were on average longer than those by digipad (Table 3). As the only manual method currently available for true digital ECG files, on‐screen calipers are likely to replace the digitizing board as the standard manual measurement method in the digital ECG era. The absolute duration of QT/QTcF intervals up to 16 ms longer by on‐screen than by digipad might be an issue in drug trials utilizing the on‐screen method. However, the negligible mean differences between on‐screen and digipad in detection of sotalol‐induced QT or QTcF change from baseline (–0.7 to 1.2 ms) indicate that the digital on‐screen calipers method is a reliable method for the assessment of drug‐induced QT/QTc interval prolongation. Both manual methods applied to standard or Holter‐derived ECGs produced a similar proportion of QTcF changes from baseline between 0–30 ms, 31–60 ms, and 61–90 ms (Table 4). The incidence of extreme outliers for QTcF change (>90 ms) was higher with the on‐screen method than digipad on both standard and Holter‐derived ECGs. Lower resolution on a computer screen after paper ECGs are scanned into portable network graphics image files, in contrast to digipad's high power image magnification, might explain the greater variability of the on‐screen method than digipad on Holter‐derived ECGs. Importing true digital ECG files directly from the Holter recorder into the measurement system (available now but not during our study) would be a more appropriate way to perform manual on‐screen measurements than scanning the paper ECG copies. The sotalol‐induced changes in T‐wave morphology or amplitude by potassium channel blockade 21 could increase the variability of QT measurement by any manual method, and pure measurement errors can also occur.
Exposure to single 160 mg and 320 mg doses of sotalol had no adverse effects while causing a wide range of QT, QTcF, and RR changes from baseline in all study subjects (Table 1), which makes our results representative of a variety of clinical situations. In keeping with the published data, 22 female subjects showed greater mean and individual QT/QTcF changes from baseline than men after 160 mg of sotalol. All measurement methods, whether applied to standard and Holter‐derived ECGs, were comparable in describing the magnitude as well as the temporal dynamics of changes in QT, QTcF, and RR interval induced by 160 and 320 mg of sotalol (Fig. 2). Although the sotalol‐induced QT/QTc prolongation was larger in this study than those likely to be encountered in the assessment of nonantiarrhythmic drugs' effect on cardiac repolarization, the Bland–Altman plots (Fig. 1) show uniform variability of pairwise comparisons over a wide range of QT/QTcF prolongation, including the smaller degrees of <30 ms. Given this, our results would be applicable to the detection of smaller degrees of QT/QTc prolongation that must be ruled out during drug development.
CONCLUSIONS
ECGs derived at discrete time points from a digital 12‐lead Holter recorder provide QT/QTc and RR data of equal value to those from a standard resting 12‐lead ECG for the assessment of drug‐induced change from baseline in QT/QTc and RR intervals. Digital 12‐lead Holter may allow for a more comprehensive assessment of drug effects on cardiac repolarization. Manual digitized QT/QTc and RR interval measurement systems, whether on paper using a digitizing pad or by digital on‐screen calipers, produce comparable mean sotalol‐induced changes from baseline in QT/QTc and RR interval. The variability of QT/QTc and RR measurement by the on‐screen method in this study was greater than digipad.
Acknowledgments
Acknowledgments: The authors acknowledge significant contributions from: (1) medical and nursing staff of the Pharmacia Clinical Research Unit in Kalamazoo, MI, for conducting this study; (2) eResearch Technologies, Pennsylvania, PA, for ECG processing and measurement of ECG intervals; (3) Mortara Instrument, Milwaukee, WI, for assistance with the equipment used for ECG acquisition; (4) Dr. Roy Bullingham, Dr. Edward J. Antal, and Dr. Larry J. Schaaf for helpful and constructive comments.
This study was funded by Pharmacia Corp. (presently part of Pfizer, Inc.).
REFERENCES
- 1. Committee for Proprietary Medicinal Products (CPMP) . Points to consider: The assessment of the potential for QT interval prolongation by non‐cardiovascular medicinal products. The European Agency for the Evaluation of Medicinal Products, 17 December 1997. [Google Scholar]
- 2. Health Canada . Draft guidance “Assessment of the QT prolongation potential of non‐antiarrhythmic drugs.” March 15, 2001. Available at http://www.hc-sh.gc.ca/hpb-dgps/therapeut/htmleng/guidmain.html.
- 3. Food and Drug Administration and Health Canada . The clinical evaluation of QT/QTc interval prolongation and proarrhythmic potential for non‐antiarrhythmic drugs. Preliminary Concept paper, November 15, 2002. Available at http://www.fda.gov/cder/workshop.htm#upcoming.
- 4. Yap YG, Camm AJ. Arrhythmogenic mechanisms of non‐sedating antihistamines. Clin Exp Allergy 1999;29(Suppl. 3):174–181. [DOI] [PubMed] [Google Scholar]
- 5. Van Haarst AD, Van't Klooster GA, Van Gerven JM et al The influence of cisapride and clarithromycin on QT intervals in healthy volunteers. Clin Pharmacol Ther 1998;64: 542–546. [DOI] [PubMed] [Google Scholar]
- 6. Morganroth J, Talbot GH, Dorr MB, et al Effect of single, ascending, supratherapeutic doses of sparfloxacin on cardiac repolarization (QTc interval). Clin Therapeutics 1999;21: 818–828. [DOI] [PubMed] [Google Scholar]
- 7. Darpö B. Spectrum of drugs prolonging the QT interval and the incidence of torsades de pointes. Eur Heart J 2001;3(Suppl. K):K70–K80. [Google Scholar]
- 8. Malik M, Bradford A. Human precision of operating a digitizing board: Implications for electrocardiogram measurements. PACE 1998;21: 1656–1662. [DOI] [PubMed] [Google Scholar]
- 9. Morganroth J, Silber SS. How to obtain and analyze electrocardiograms in clinical trials: Focus on issues in measuring and interpreting changes in the QTc interval duration. Ann Noninvasive Electrocardiol 1999;4: 425–433. [Google Scholar]
- 10. Le Coz F, Funck‐Brentano C, Poirier JM, et al Prediction of sotalol‐induced maximum steady‐state QTc prolongation from single‐dose administration in healthy volunteers. Clin Pharmacol Ther 1992;52: 417–426. [DOI] [PubMed] [Google Scholar]
- 11. Barbey JT, Sale ME, Woosley RL, et al Pharmacokinetic, pharmacodynamic and safety evaluation of an accelerated dose titration regimen of sotalol in healthy middle‐aged subjects. Clin Pharmacol Ther 1999;66: 91–99. [DOI] [PubMed] [Google Scholar]
- 12. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Clin Res 1999;8: 135–160. [DOI] [PubMed] [Google Scholar]
- 13. Haverkamp W, Breithardt G, Camm AJ, et al The potential for QT prolongation and proarrhythmia by non‐antiarrhythmic drugs: Clinical and regulatory implications. Report on a Policy Conference of the European Society of Cardiology. Cardiovascular Res 2000;47: 219–233. [DOI] [PubMed] [Google Scholar]
- 14. Morganroth J. Focus on issues in measuring and interpreting changes in the QTc interval duration. Eur Heart J Suppl 2001;3(Suppl. K):K105–K111. [Google Scholar]
- 15. Malik M, Camm AJ. Evaluation of drug‐induced QT interval prolongation. Implications for drug approval and labeling. Drug Saf 2001;24: 323–351. [DOI] [PubMed] [Google Scholar]
- 16. Shah RR. The significance of QT interval in drug development. Br J Clin Pharmacol 2002;54: 188–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kautzner J, Gang Y, Camm AJ, et al Short‐ and long‐term reproducibility of QT, QTc, and QT dispersion measurement in healthy subjects. PACE 1994;17: 928–937. [DOI] [PubMed] [Google Scholar]
- 18. Kautzner J, Gang Y, Kishore AGR, et al Interobserver reproducibility of QT interval measurement and QT dispersion in patient after acute myocardial infarction. Ann Noninvasive Electrocardiol 1996;1: 363–374. [Google Scholar]
- 19. Pratt CM, Ruberg S, Morganroth J, et al Dose‐response relation between terfenadine (Seldane) and the QTc interval on the scalar electrocardiogram: Distinguishing a drug effect from spontaneous variability. Am Heart J 1996;131: 472–480. [DOI] [PubMed] [Google Scholar]
- 20. Malik M, Färbom P, Batchvarov V, et al Relation between QT and RR intervals is highly individual among healthy subjects: Implications for heart rate correction of the QT interval. Heart 2002;87: 220–228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Darpö B, Almgren O, Bergstrand R, et al Assessment of frequency‐dependency of the class III effects of almokalant. A study using programmed stimulation and recordings of monophasic action potentials and ventricular paced QT‐intervals. Cardiovasc Drug Ther 1996;10: 539–554. [DOI] [PubMed] [Google Scholar]
- 22. Makkar RR, Fromm BS, Steinman RT, et al Female gender as a risk factor for torsades de pointes associated with cardiovascular drugs. JAMA 1993;270: 2590–2597. [DOI] [PubMed] [Google Scholar]