Changes In Global Electrical Heterogeneity Associated With Dofetilide, Quinidine, Ranolazine, And Verapamil

Abstract

Background:

ECG markers of antiarrhythmic drug (AAD) activity could be used to optimize efficacy and minimize toxicity. Vectorcardiographic (VCG) global electrical heterogeneity (GEH) parameters are associated with ventricular arrhythmias and sudden death, but it is unclear how measurements of GEH change in response to AADs.

Objective:

To characterize acute effects of AADs on GEH measurements.

Methods:

We analyzed publicly available double-blind placebo-controlled trial data from healthy volunteers given 1 dose of placebo, dofetilide, quinidine, ranolazine, or verapamil on subsequent visits. Serial ECGs and plasma drug levels were collected. VCG GEH parameters (spatial ventricular gradient [SVG], spatial QRST angle, sum absolute QRST integral, and SVG-QRS peak angle) were measured. Placebo-corrected change from baseline was regressed on drug level stratified by sex using linear mixed effects models.

Results:

Among 22 persons (50% male, age 27 ± 5 years), 5232 ECGs were analyzed. Dofetilide and quinidine were associated with significant changes in more GEH parameters (5) compared to verapamil (2) and ranolazine (1). The most notable change occurred in SVG azimuth, with largest changes (degrees per unit normalized drug level) in dofetilide (6.1, 95%CI 4.28.0) and quinidine (9.4, 95%CI 6.712.0), and smaller effects in verapamil (4.4, 95%CI 2.95.9) and ranolazine (5.4, 95%CI 3.57.3). AAD induced GEH changes significantly differed in men and women.

Conclusion:

AADs change GEH measurements. These changes, which differ in men and women, are likely driven by alterations in ion channel function and dispersion of depolarization/repolarization. GEH measurement may allow early assessment of favorable or adverse AAD effects.

Introduction

Antiarrhythmic drugs (AADs) are important for the treatment of both atrial and ventricular arrhythmias. However, AAD therapy is often limited by modest efficacy, non-cardiac side effects, and adverse electrophysiologic effects, including QT prolongation and proarrhythmia, which can be fatal [1]. Unfortunately, it can be difficult to predict which patients will have a favorable response to AAD therapy, and which patients will experience adverse effects. Dofetilide, a Class III AAD with potent hERG (I_Kr) blocking effects, causes excessive QT prolongation and torsade de pointes (TdP) in 1–3% of patients [2–4]. Quinidine, which blocks I_Kr as well as multiple other ion channels (including I_Na, I_Ks, and I_CaL) to varying degrees at different drug concentrations, is associated with a risk of TdP that is independent of the degree of QT prolongation [5]. Other AADs which affect multiple ion channels including I_Na and I_Kr, such as dronaderone and amiodarone, cause QT prolongation but are not associated with a significantly elevated risk of TdP compared to other AADs [6]. The QT interval is therefore not a specific marker of AAD-related proarrhythmia for all AADs.

A likely mechanism of AAD induced proarrhythmia is development of excessive myocardial electrical heterogeneity. The myocardium does not depolarize and repolarize simultaneously or in the same sequence; spatial and temporal myocardial electrical heterogeneity is required for normal cardiac function and is responsible for the genesis of the normal QRST complex [7–9]. Excessive myocardial electrical heterogeneity, however, is associated with an increased risk of ventricular arrhythmias in both animal models and humans [10–12]. AADs affect action potential duration and the temporal dispersion of action potential duration, and have been used to induce increased electrical heterogeneity in animal studies [10–12]. A non-invasive assessment of the degree to which an AAD alters myocardial electrical heterogeneity might therefore identify patients at risk for adverse proarrhythmic effects.

Although many ECG parameters have been proposed as markers of electrical heterogeneity, many previously studied parameters lack sensitivity and specificity, and efforts are ongoing to develop and test ECG parameters that are more specifically linked to the mechanisms underlying ventricular arrhythmias and sudden death. Global electrical heterogeneity (GEH) is a set of noninvasive vectorcardiographic (VCG) measurements which quantifies electrical heterogeneity. GEH parameters are derived from the mathematical theory underlying the generation of ECG waveforms [13] and their clinical use is supported by associations with various clinical outcomes in humans [14]. Recent studies in the general population have demonstrated that GEH parameters are specifically associated with sudden death, that GEH differs in men and women, and that temporal changes in GEH over time predict risk of sudden death and adverse cardiac structural changes. [15, 16].

The effect of AAD therapy on GEH is unclear. We hypothesized that changes in electrical heterogeneity caused by AAD therapy may be detectable as acute changes in GEH measurements that correlate with AAD plasma levels. Since GEH parameters [15, 17] as well as AAD tolerability and safety [18, 19] vary by gender, we also hypothesized that some AAD-induced changes in GEH would differ in men and women.

Methods

Data Source

We analyzed publicly available double-blind crossover placebo-controlled trial data collected by the United States Food and Drug Administration (FDA) and available on the Physionet website (http://www.physionet.org) [20–22]. Healthy volunteers were given 1 dose of placebo, dofetilide (500 mcg), quinidine sulfate (400 mg), ranolazine (1500 mg), or verapamil HCl (120 mg) on serial visits one week apart. Over the course of the subsequent 24 hours, serial 12-lead ECGs (3 per time point, 5232 in total) and plasma drug levels were collected at 0.5 hours pre-drug administration, and then at 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 5, 6, 7, 8, 12, 14, and 24 hours post drug administration. For all analyses, the independent variable was plasma drug concentration normalized to the mean maximum plasma drug concentration observed across all participants in the study:

$$x_i=\frac{\text{plasma concentration at time}\hspace{0.17em}i}{〈\text{maximum plasma concentration}〉}.$$ (1)

For further details we refer to references [20, 21] and supplements therein.

Electrocardiographic Data

We used median beat VCGs provided in the FDA data set, which were derived from 12-lead ECGs sampled at 1 kHz transformed into the Frank X, Y, and Z leads using Guldenring’s transformation [23]. The X, Y, and Z leads were then baseline corrected so that the flattest part of the TP segment, rather than the onset of the QRS complex (as was used previously [20, 24]), was the zero reference point for both area calculations and the origin of the VCG (see Supplemental Figure 1). We used the fiducial points (QRS_onset, QRS_end, and T_end) that were provided with the data set for consistency and reproducibility. We followed previous conventions for orientation of axes with the positive X axis towards the left, the positive Y axis towards the feet, and the positive Z axis posterior. QT interval was corrected for heart rate using the Fridericia correction. T_peak-T_end interval was provided in the original dataset.

GEH Calculations

We assessed the effect of AADs on 5 previously defined GEH parameters (illustrated in Figure 1): Spatial ventricular gradient (SVG) is defined as the vector sum of the area QRS- and area T-vectors which are obtained by integrating the QRST complex. The SVG has magnitude and orientation in 3-dimensional space expressed as azimuth (angle in the XZ/transverse plane) and elevation (angle in the XY/frontal plane). Spatial QRST angle is the 3-dimensional angle between the area QRS- and area T-vectors. Sum absolute QRST integral (SAI QRST) is defined as the sum of areas under the absolute values of the X, Y, and Z leads. In addition to these established GEH parameters, we measured the SVG-QRS peak angle defined as the 3-dimensional angle between the SVG and peak QRS vector. Details regarding these definitions/calculations may be found in the Supplement. ECG/VCG processing and parameter calculations were performed using Matlab R2017a (Mathworks, Natick, MA).

Figure 1:

Illustration of GEH parameters. A: The spatial ventricular gradient (SVG) is defined as the vector sum of the QRS-area vector and the T-area vectors (see Eq. 3). These vectors are distinct from the “peak” QRS and T vectors which points towards the point on the QRS vector loop and T vector loop respectively that are furthest from the origin. The SVG vector has magnitude (vector length), elevation, and azimuth. Elevation can have values between 0 and 180 degrees, with 0 pointing downwards towards the feet, and 180 degrees pointing upwards towards the head. Azimuth can have values between −180 and 180 degrees; positive azimuth values between 0 to +180 degrees are oriented posteriorly, and negative azimuth values between 0 to −180 degrees are oriented anteriorly. QRST angle is the 3-dimensional angle between the QRS-area vector and the T-area vector. B: Sum absolute QRST integral (SAI QRST) is defined as the sum of areas under the absolute values of the X, Y, and Z leads. See methods for additional details.

Statistical Analysis

Placebo-corrected change from baseline (known as “∆∆”) for each measurement G and time point i was defined as:

$$\mathrm{\Delta }\mathrm{\Delta }{G}_i=(G_i-G_0)-(P_i-P_0),$$ (2)

where G₀ and P₀ are the baseline measurements calculated from the ECG 30 minutes before drug or placebo administration, respectively. Each of 6 G_i were regressed on normalized plasma drug level using a linear mixed effects model in order to estimate the slope of parameter change vs drug level. The interaction between drug, gender, and dose were the fixed effects. The random effects were random slopes by participant and drug; covariance between random effects for different drugs was found to be small and was omitted from the final model. Because of the baseline dose correction, constant terms and random intercepts were not included in the model, as ∆∆G₀ = 0 by definition at zero dose x₀ = 0.

Stability of baseline GEH measurements at each of the 5 study visits was assessed using repeated-measures ANOVA. All angular variables did not cross 180 degrees, so circular statistical methods were not required for analyses.

All analyses were performed with Stata 15 (StataCorp LP, College Station, TX). Two-sided P-values were computed without adjustment for multiple comparisons.

Results

Participant Characteristics

Table 1 summarizes the baseline characteristics of the study participants as previously described [20]. Among 22 participants, 50% were male and the median age was 27 ± 5 years. Most participants were white (77%). Baseline ECG/GEH parameter values and gender differences were consistent with prior studies in healthy individuals; SVG magnitude and SAI QRST were smaller and SVG azimuth was more posterior in women compared to men. [17, 25] There was a non-significant trend of the QTc interval being longer in women.

Table 1:

Baseline characteristics of the study population (N = 22), given as mean and standard deviation (σ) for continuous variables, and mean and % of the total for catagorical variables. Abbreviations: BMI=body mass index, SVG=spatial ventricular gradient, SAI QRST=sum absolute QRST integral, Mag=magnitude, Az=azimuth, El=elevation.

Demographics
	N	σ/%
White	17	77.3
Male	11	50.0
Age (y)	26.9	5.5
Weight (kg)	68.3	8.2
BMI (kg/m2)	23.1	2.7
ECG/GEH Parameters

	Female		Male
	Mean	σ	Mean	σ	p
SVG Mag (mV·ms)	80.51	21.79	132.00	28.38	0.00
SVG Az (deg)	−18.89	10.12	−33.77	11.34	0.00
SVG El (deg)	59.84	7.06	56.63	7.34	0.31
QRST Angle (deg)	35.38	8.28	39.88	12.37	0.33
SVG-QRS Angle (deg)	6.65	3.01	9.78	5.08	0.09
SAI QRST (mV·ms)	147.83	35.46	251.74	52.45	0.00
Median beat QTc (ms)	402.54	15.59	389.25	16.58	0.07

GEH Changes In Response to AAD Administration

Figure 2 summarizes how baseline-adjusted, placebo-corrected measurements (∆∆) of GEH change as a function of increasing normalized drug concentration. The corresponding columns in Table 2 show the numerical values of the best-fit slopes in the model. Dofetilide and quinidine were associated with significant changes in SVG magnitude, SVG azimuth, QRST angle, SVG-QRS peak angle, and SAI QRST. Ranolazine and verapamil were associated with generally smaller or non-significant changes in GEH. SVG elevation was not significantly changed by any drug.

Figure 2:

Placebo-corrected change from baseline (∆∆) of global electrical heterogeneity parameters vs drug level normalized to the mean maximum dose across the population.

Table 2:

Model slopes of baseline-adjusted, placebo-corrected global electrical heterogeneity parameters vs normalized plasma drug concentration. Units are: mV·ms per normalized concentration for SVG magnitude and SAI QRST, and degrees per normalized concentration for all other quantities, where a unit change in drug level is normalized to the mean maximum plasma drug concentration.

Drug	SVG Mag	SVG Az	SVG El	QRST Angle	SVG-QRS Angle	SAI QRST
Ranolazine	0.11 [−6.10,6.33]	5.39*** [3.47,7.30]	0.78 [−0.38,1.94]	2.09 [−0.30,4.48]	−0.41 [−1.16,0.34]	−0.43 [−10.07,9.20]
Dofetilide	8.75** [2.37,15.12]	6.11*** [4.23,7.99]	0.86 [−0.19,1.90]	−3.23* [−5.78,−0.67]	−1.97*** [−2.71,−1.24]	11.00* [0.69,21.31]
Verapamil	−4.33 [−10.31,1.66]	4.39*** [2.93,5.85]	0.26 [−1.16,1.67]	5.28* [0.54,10.03]	−0.70 [−1.67,0.28]	−6.44 [−14.92,2.03]
Quinidine	−12.83*** [−18.81,−6.84]	9.38*** [6.71,12.04]	0.96 [−0.46,2.38]	6.43*** [2.62,10.23]	−2.52*** [−3.77,−1.26]	−13.71*** [−21.87,−5.55]

Asterisks *, **, or *** denote p < 0.05, 0.01, or 0.001 for the null hypothesis β = 0. No adjustment for multiple comparisons. Brackets contain 95% confidence intervals. Abbreviations are the same as in Table 1.

Standardized (by per-drug standard deviation) slopes of ∆∆ vs. normalized plasma drug concentration for GEH parameters and other ECG parameters (QTc, QRS duration, and T_peak-T_end) are summarized in Supplemental Table 1. ∆∆ vs. normalized plasma drug concentration for QTc, QRS duration, and T_peak-T_end are shown in Supplemental Figure 2. As previously shown, dofetilide and quinidine caused the largest changes in QTc and T_peak-T_end [20, 21]. Changes in QRS duration were very small for all drugs.

Influence of Gender on AAD Induced GEH Changes

Figure 3 shows ∆∆ of GEH as a function of increasing normalized drug concentration, stratified by gender. Table 3 shows the corresponding male-female differences in model slopes. Dofetilide had a smaller effect on the spatial QRST angle, while quinidine had a larger effect on SVG magnitude and SAI QRST, in men compared to women. Aside from a larger slope in men for SVG-QRS peak angle with ranolazine (p = 0.03), there were no differences by gender in the effects of ranolazine and verapamil. Changes in SVG azimuth in response to dofetilide and quinidine tended to be larger for women in our study, although this did not reach statistical significance. SVG elevation was not significantly different by gender for any drug.

Figure 3:

Placebo-corrected change from baseline (∆∆) of global electrical heterogeneity parameters vs drug level, stratified by gender.

Table 3:

Difference in model slopes for males vs females. Units are the same as in Table 2.

	Difference in slopes male vs female
Drug	SVG Mag	SVG Az	SVG El	QRST Angle	SVG-QRS Angle	SAI QRST
Ranolazine	1.69 [−10.72,14.10]	−3.27 [−6.83,0.29]	0.45 [−1.86,2.75]	3.83 [−0.67,8.33]	1.49* [0.12,2.86]	6.27 [−12.82,25.37]
Dofetilide	−8.13 [−20.40,4.13]	−1.89 [−5.56,1.79]	−1.25 [−3.27,0.77]	6.13** [1.70,10.57]	0.06 [−1.40,1.53]	−8.62 [−28.90,11.66]
Verapamil	2.00 [−10.08,14.08]	−1.15 [−4.09,1.79]	0.31 [−2.50,3.13]	−3.17 [−12.60,6.27]	−1.25 [−3.14,0.64]	0.86 [−16.21,17.93]
Quinidine	−11.91* [−22.72,−1.10]	−4.35 [−9.35,0.65]	−1.69 [−4.44,1.05]	2.06 [−5.51,9.63]	−1.02 [−3.49,1.46]	−17.66* [−32.06,−3.25]

Asterisks *, **, or *** denote p < 0.05, 0.01, or 0.001 for the null hypothesis of equal slopes in males and females β_M − β_F = 0. Brackets contain 95% confidence intervals. Units are the same as in Table 2. Abbreviations are the same as in Table 1.

Participant-specific GEH Changes In Response to AAD Adminstration

Table 4 shows the standard deviations of per-participant random slopes for each drug. For SVG azimuth, the standard deviations are smallest compared to the size of the corresponding average effect from Table 2. In comparison, the other standard deviations are larger when compared to their corresponding average slopes, indicating that per-participant fluctuations around the average response to drug are large for these measures. A graphical representation of these fluctuations can be found in Supplemental Figure 3, which shows histograms of the fitted slope values for each participant that include both the fixed (i.e., average) and random (i.e., per-participant) effects.

Table 4:

Standard deviations of per-participant (i.e., random effect) slopes for each measurement across participants by drug. Null hypothesis of zero random effect excluded at p < 0.05 level in all cases. Units are the same as in Table 2. Abbreviations are the same as in Table 1.

	Per-Participant Random-Slope Standard Deviation
Drug	SVG Mag	SVG Az	SVG El	QRST Angle	SVG-QRS Angle	SAI QRST
Ranolazine	13.783	3.843	2.548	4.826	1.557	21.294
Dofetilide	13.970	4.127	2.256	4.929	1.705	23.313
Verapamil	12.553	2.649	3.054	10.800	2.154	17.438
Quinidine	11.874	5.657	3.093	8.635	2.866	15.555

GEH Stability Over Time

Supplemental Table 2 shows mean baseline GEH measurements and QTc prior to drug/placebo administration at each of the 5 study visits. There were no significant changes in baseline GEH measurements or QTc over time.

Discussion

Our results demonstrate that AAD administration alters measurements of GEH. These changes occur rapidly and correlate with AAD plasma drug levels. Acute GEH changes are more marked for dofetilide and quinidine than for ranolazine and verapamil. Some changes in GEH differ by gender, and there is considerable individual variation in GEH changes after AAD administration.

Among the GEH measurements affected by the AADs were SVG magnitude, SVG azimuth, SAI QRST, and QRST angle. Abnormal values of and long-term changes in these measurements were shown to correlate with the risk of sudden cardiac death independent of multiple other known sudden cardiac death risk factors including left ventricular ejection fraction [15, 16]. Our analysis demonstrates that significant changes in GEH parameters can also be observed over the course of hours after a single dose of AAD, without the need to reach steady state drug levels. Observed changes in GEH parameters are therefore likely reflecting changes in electrical heterogeneity induced by AADs, as previously seen in experimental models [12].

The largest changes in GEH parameters were seen with administration of dofetilide and quinidine, the two drugs with the most potent hERG (I_Kr) blockade. Dofetilide and quinidine had opposite effects on SVG magnitude, QRST angle, and SAI QRST. We hypothesize this is because dofetilide is a pure I_Kr blocker, while quinidine affects multiple ion channels, including the late sodium current, transient outward potassium current (I_to), and L-type calcium channels, and has variable ion channel blocking effects at different concentrations. Overall, there were no significant differences in GEH parameters when verapamil and ranolazine were compared (p > 0.24 for each). Although verapamil is primarily a calcium channel blocker, it does have mild I_Kr blocking effects, as well as weak sodium channel blocking activity. The more potent I_Kr blockade caused by ranolazine was likely offset by more potent concomitant block of the late sodium current. The smaller changes in GEH parameters with verapamil and ranaolazine compared to dofetilide and quinidine also mirror the lower risk of drug induced arrhythmias with verapamil and ranolazine compared to dofetilide and quinidine.

AADs are useful in the treatment of atrial and ventricular arrhythmias, but overall there is only a modest rate of success, and drug-specific risk of proarrhythmia varies widely and differs by gender [1, 18, 19]. If future, larger studies of AAD induced GEH changes which are powered to assess adverse outcomes establish a link between GEH changes and proarrhythmia, assessment of GEH on a 12-lead ECG may be useful in the assessment of drug efficacy and safety. SVG azimuth had the smallest per-participant variation when compared to average response to drug, and so may be particularly useful in this regard. For example, patients with rapid and extreme changes in GEH measurements might be at high risk for arrhythmia, whereas minimal changes after starting an AAD may suggest that the drug is likely to be ineffective. Either scenario might prompt the clinician to consider an alternative AAD.

Optimal AAD treatment strategies may differ for men and women. Women tend to exhibit larger changes in QT interval in response to dofetilide [19] and quinidine [18], and are known to be more prone to TdP. In our study these drugs caused changes in SVG azimuth which tended to be larger in women, although this did not reach statistical significance. Changes in SVG magnitude and SAI QRST with quinidine were larger for men, whereas QRST-angle changes were larger for women with dofetilide. These findings are interesting when viewed in the context that when compared to healthy men, healthy women have smaller average SVG azimuth, SAI QRST, and QRST angle (see Table 1 and [17]). The factors that contribute to gender-specific variations in baseline GEH measurements and AAD-induced changes in GEH are unclear; they may be mediated by variations in sex hormones or gene expression [26].

A prior analysis of this dataset [21] also assessed SVG magnitude and QRST angle. The authors found no change in SVG magnitude with any drug, and similar results for QRST angle. We believe that the differences between our results and these prior results might be explained by differences in how the isoelectric point of the VCG was defined [21]. As noted in the methods, the prior study used the onset of the QRS complex as the zero reference, while we used the flattest segment of the TP segment (a more physiologic choice) as the zero reference. Correct choice of the isoelectric point has important implications for GEH calculations (see Supplemental Figure 1 and [24]). This prior analysis of these data also demonstrated, with a different modeling approach, that dofetilide and quinidine caused the largest changes in QTc and T_peak-T_end interval. It is critical to recognize that a large drug induced change in an ECG/GEH parameter does not necessarily mean that the parameter will be useful in predicting AAD-induced proarrhythmia; for example, both dofetilide and quinidine caused significant QTc prolongation, but although QTc has been linked to the risk of TdP when using dofetilide [2–4], the risk of TdP when using quinidine is independent of QTc prolongation [5]. Further studies of GEH parameters using clinical outcomes are needed to determine if changes in GEH correlate with clinical outcomes, and whether they perform better than previously identified markers such as QTc and T_peak-T_end interval.

SVG-QRS peak angle has likewise not been previously measured in persons receiving AAD therapy. SVG-QRS peak angle is a novel measurement that attempts to correct for variations in cardiac anatomy and QRS morphology by comparing the SVG angle to the principal electrical axis of the heart during depolarization. Significant changes in SVG-QRS peak angle were only seen with dofetilide and quinidine, the drugs with the most I_Kr blockade, although the clinical significance of this will require further study.

Strengths and Limitations

Strengths of our study include the crossover and placebo-controlled design of the original trial, which should eliminate several confounders, and the high quality of ECGs that were collected. Measurements were adjusted for pre-drug baseline which also simplifies the analysis and interpretation of results. Limitations of our study include the small sample size, and a young, healthy population, which limits generalizability to patients who are the usual recipients of AADs. Our study was not designed to determine whether GEH changes after AAD administration are associated with clinical outcomes, but to generate hypotheses for future testing.

Conclusion

AADs change GEH measurements. These changes, which differ in men and women, are likely driven by alterations in ion channel function and dispersion of depolarization or repolarization. GEH measurement may allow early assessment of favorable or adverse AAD effects.