Autonomic and Cardiac Repolarization Lability in Long QT Syndrome Patients

Abstract

Objective:

Long QT-Syndrome (LQTS) patients are at risk of arrhythmias and seizures. We investigated whether autonomic and cardiac repolarization measures differed based on LQTS genotypes, and in LQTS patients with vs. without arrhythmias and seizures.

Methods:

We used 24-hour ECGs from LQTS1 (n=87), LQTS2 (n=50), and LQTS genotype negative patients (LQTS⁽⁻⁾, n=16.) Patients were stratified by LQTS genotype, and arrhythmias/seizures. Heart rate variability (HRV) and QT variability index (QTVI) measures were compared between groups during specific physiological states (minimum, middle, & maximum sympathovagal balance, LF/HF.) Results were further tested using logistic regression for each ECG measure, and all HRV measures in a single multivariate model.

Results:

Across multiple physiological states, total autonomic (SDNN) and vagal (RMSSD, pNN50) function were lower and repolarization dynamics (QTVI) were elevated in LQTS⁽⁺⁾, LQTS1, and LQTS2, compared to LQTS⁽⁻⁾. Many measures remained significant in the regression models. Multivariate modeling demonstrated that SDNN, RMSSD, and pNN50 were independent markers of LQTS⁽⁺⁾ vs. LQTS⁽⁻⁾, and SDNN and pNN50 were markers for LQTS1 vs. LQTS⁽⁻⁾. During sympathovagal balance (middle LF/HF), RMSSD and pNN50 distinguished LQTS1 vs. LQTS2. LQTS1 patients with arrhythmias had lower total (SDNN) and vagal (RMSSD and pNN50) autonomic function, and SDNN remained significant in the models. In contrast, ECG measures did not differ in LQTS2 patients with vs. without arrhythmias, and LQTS1 and LQTS2 with vs. without seizures.

Conclusion:

Autonomic (HRV) and cardiac repolarization (QTVI) ECG measures differ based on LQTS genotype and history of arrhythmias in LQTS1. SDNN, RMSSD, and pNN50 were each independent markers for LQTS genotype.

Introduction:

Long QT Syndrome (LQTS) is due to mutations in ion channel and interacting proteins. LQTS1 and LQTS2, which are the focus of this study, are due to loss-of-function mutations in K_v7.1 and K_v11.1 potassium channel proteins, respectively. Patients exhibit QT prolongation on the cardiac electrocardiogram (ECG), and a high risk of cardiac arrhythmias. Additionally, our group and others have reported an increased prevalence of epileptic seizures in LQTS patients^(1–3).

It remains incompletely understood whether there are LQTS-genotype specific differences in basal autonomic nervous system (ANS) function. Sympathetic stimulation, particularly exercise and stress, provides a trigger for arrhythmias in LQTS1 and LQTS2⁽⁴⁾. The ANS provides the neural connection between the brain and heart, and it is unknown whether ANS function differs in LQTS patients with vs. without a history of arrhythmias and seizures, separately. Additionally, we compared cardiac repolarization dynamics between groups. Using 24-hour ECG recordings and quantification of beat-to-beat variability in heart rate (HRV) and repolarization index (QTVI), we assessed LQTS-genotype specific ANS function and repolarization lability, respectively^{(5, 6)}. As HRV and QTVI are useful measures for predicting arrhythmias and mortality in various forms of acquired heart disease^(7–9), we assessed whether these measures differ in LQTS genotype positive and negative patients, and in those with vs. without arrhythmias and seizures, separately. We hypothesized that these non-invasive ANS and repolarization ECG measures are independent predictors of LQTS genotype, and these measures differ in LQTS patients with neuro-cardiac electrical disturbances.

Methods:

Patient Population

The study included 24-hour Holter monitor recordings from genotype positive LQTS1 (n=87) and LQTS2 (n=50) patients, as well as LQTS genotype negative LQTS⁽⁻⁾ (n=16) family members. All clinical, genetic, pharmacologic, and demographic information was available in the Rochester-based LQTS Registry⁽¹⁰⁾. Follow-up was from initial enrollment to last contact, death, or March 2016, whichever came first. Information was gathered from physician reports or direct contact with patients, relatives, and medical providers. There were no race, ethnicity, or sex restrictions for inclusion. No one in the study had a pacemaker or left cervical sympathetomy. For those patients with an implanted cardioverter defibrillator (LQTS1: n=15, LQTS2: n=8), none of the recordings included paced beats. Outcome measures included time and frequency domain HRV and QTVI parameters. We tested for differences between LQTS genotype (LQTS⁽⁻⁾ vs. LQTS⁽⁺⁾ & LQTS⁽⁻⁾ vs. LQTS1 vs LQTS2), and within each LQTS1 and LQTS2 group we compared with vs. without a history of arrhythmias or seizures (separate analyses).

Definitions

LQTS⁽⁺⁾ patients had a single known mutation in KCNQ1 (LQTS1) or KCNH2 (LQTS2.) Cardiac arrhythmias were defined as ventricular tachycardia, torsades de pointes, ventricular fibrillation, aborted cardiac arrest, and sudden cardiac death⁽¹⁾.

8-lead Mason Likar configuration Holter monitor recordings were converted to a three orthogonal Frank lead system. Analyses was performed on the lead with the highest T-wave amplitude during a 5-minute epoch of stable heart rate and free of arrhythmias/seizures. The heart rate was similar between LQTS groups (Table 1). Commercial software (Mortara H-Scribe) was used for automatic QRS detection, which was manually adjudicated by an expert ECG analyst. Sinus beats directly before and after artifacts and non-sinus beats were not used for HRV analysis, and the appropriate RR interval in the cycle was interpolated. As we only considered epochs that had <1% non-sinus beats, and where ≥95% of the beats could be used for analysis, the number of recordings within each group varied between analyses. Time-domain HRV analyses included standard deviation of the normal sinus beat-to-beat (NN) interval (SDNN), which reflects total ANS function^{(5, 6)}. We also calculated the root mean square of successive differences in paired NN intervals (RMSSD) and percentage of successive NN intervals that differed more than 50ms (pNN50), which are both measures of vagal function. Frequency domain analyses (using fast fourier transformation) was performed and the ratio of the low (LF, 0.04–0.15 Hz) and high frequency (HF, 0.15–0.4 Hz) domains provided an assessment of sympathovagal balance.

Table 1:

Baseline Characteristics.

Clinical Characteristics	LQTS(−)	LQTS1	LQTS2	p-value all*	p-value 1 vs 2**
N	16	87	50
Females (%)	13 (81)	61 (70)	34 (85)	0.595	0.799
Age at Holter (yr)	31 ± 10	35 ± 11	36 ± 10	0.170	0.427
Follow up (yr)	34 ± 7	38 ± 5	38 ± 4	0.070	0.408
Beta Blocker (%)	0 (0)	41 (47)	19 (38)	0.002	0.303
History of Cardiac Events
History of Arrhythmias (%)***	0 (0)	7 (8)	8 (16)	0.123	0.154
History of Seizure (%)***	2 (13)	10 (11)	9 (18)	0.563	0.292
ECG (minimum HR)
Heart Rate (bpm)	61.27 ± 15.49	59.85 ± 1	58.33 ± 10.34	0.697	0.409
QTc (ms)	475.7 ± 31.15	543.86 ± 81.42	579.27 ± 112.87	<0.001	0.045

^*p-values were calculated using a Kruskal-Wallis Test.

^**p-values were calculated using a two-sided Wilcoxon Rank Sum test with normalized approximation.

^***p-values were calculated using an Exact Wilcoxon test due to small sample size

Beat-to-beat changes in the QT duration reflect temporal fluctuations in the ventricular activation recovery process, and is influenced by ANS function (i.e., HR and HRV) ^{(5, 6)}. Increased QTVI is associated with an increased risk of malignant ventricular arrhythmias and sudden death⁽⁸⁾. QT variability analysis included the variance in the QT (QT_v) and RR (RR_v), as well as the mean QT (QT_m) and RR (RR_m) $\mathit{\text{QTVI}}=log10\left[\frac{QTv/QT{m}^2}{RRv/RR{m}^2}\right]$ ^{(8, 9)}. Variability was quantified by fitting (squeezing/stretching) each S-T_end ECG waveform to a template waveform morphology for that epoch^{(8, 9)}.

Timepoints for Analyses

For each participant we selected a 5-minute epoch of minimum heart rate from the 24-hour recording. Next, using the SuperECG output, we created a 24-hour rolling average of LF/HF ratio (5 minute epochs, 1 minute steps), which provides a measure of sympathovagal balance. LF/HF was used to identify specific 5-minute epochs of differing physiological states. We analyzed HRV and QTVI measures during 5-minute timepoints of vagal dominance (minimum LF/HF), sympathovagal balance (LF/HF=1), and sympathetic dominance (maximum LF/HF.) MatLab version 2015b (MathWorks, Massachusetts) was used to perform the analyses⁽⁹⁾.

Study Approval:

The LQTS Registry is approved by the University of Rochester Research Subject Review Board (RSRB00025305). HIPAA requirements for accounting of consent, withdrawal of consent, and disclosures were followed. The data was handled in accordance with applicable laws, and all information and ECG recordings were de-identified. Written informed consent was obtained from all participants/guardians in the study.

Statistics:

Non-parametric comparisons between groups were performed by Wilcoxon Rank Sum and Kruskal-Wallis Tests. Additionally, we assessed the strength of these differences through univariate and multivariate logistical regression models, with the LQTS genotype as the outcome measures. We adjusted for beta-adrenergic blocker use, sex, age, RR, and QT_c duration where appropriate. As RR and QT_c are part of the QTVI equation, they were not included in the QTVI adjusted models. For the multivariate models, all HRV measures were included in a single model, and these same adjustments were applied. QTVI could not be included in the multivariate model, as the QTVI denominator includes RR variance, which is mathematically linked to SDNN. Logistic regression models that included measurements at minimum, middle, and maximum LF/HF (i.e., overall across multiple physiological states), accounted for repeated measures and included an adjustment for LF/HF. 48% of the LQTS1 and 35% of the LQTS2 patients were taking beta adrenergic blockers at the time of the recording (Table 1.) For analysis that was limited to LQTS1 and LQTS2 patients, the regression models included adjustments for beta blocker use (Figure 3 & e-Figure 4.) In contrast, when assessing ANS and repolarization function in LQTS⁽⁻⁾ and LQTS⁽⁺⁾ patients, as none of the LQTS⁽⁻⁾ patients were taking beta blockers, we could not adjust for this in our statistical models. Thus, to remove this potential confounder and to increase the level of rigor, analysis in Figures 1 and 2 was limited to people not taking beta blockers; albeit, to assess these measures in the general LQTS population, we include results from the entire cohort without adjustments for beta blockers (e-Figure 2 & 3.) Results were similar in both analytical groups, and often the level of significance was larger when excluding those on beta blockers.

Figure 3: HRV and QTVI measures in LQTS1 and LQTS2 Patients With vs. Without Arrhythmias:

Assessment of the overall (minimum, middle, & maximum LF/HF) difference in HRV (SDNN, RMSSD and pNN50) and QTVI. Bars denote significance (p<0.0125) by nonparametric analyses, and “*” denote significance for univariate regression models adjusted for beta blockers, baseline QT_c, RR, age, sex, and LF/HF. QT_c and RR were not adjusted for in the QTVI model because they are connected.

Figure 1: Reduced ANS Function and Increased QT Variability at Times of Minimum Heart Rate in LQTS Patients:

Time (A-C) and frequency domain (D) HRV measures, and QTVI (E) during a 5-minute period of minimum heart rate in LQTS⁽⁻⁾ (n=15), LQTS1 (n=42), and LQTS2 (n=26) patients not on beta blockers. Bars denote significance (p<0.01) by nonparametric analysis, and “*” denote significance for univariate regression models adjusted for baseline QT_c, RR, sex, and age (A-E). QT_c and RR were not adjusted for in the QTVI model because they are mathematically connected. In (F), SDNN, RMSSD, pNN50, and LF/HF were run in the same model with adjustments for baseline QT_c, RR, sex, and age. Odds Ratios, 95% Confidence intervals, and p-values are shown.

Figure 2: LQTS Genotype Specific Differences in ANS and Cardiac Repolarization Function During Specific Physiological States:

HRV (A-C) and QTVI (D) during 5-minute periods of minimum, middle (LF/HF=1), maximum, and all LF/HF combined in LQTS⁽⁻⁾ (n=16), LQTS1 (n=39–43), and LQTS2 (n=22–26) patients not on beta blockers. Bars denote significance (p<0.0125) by nonparametric analyses, and “*” denote significance for univariate regression models with adjustments for baseline QT_c, RR, sex, and age (LF/HF was adjusted for in the overall LF/HF models) in LQTS⁽⁺⁾, LQTS1, and LQTS2, compard to LQTS⁽⁻⁾. QT_c and RR were not adjusted for in the QTVI model because they are connected.

For convenience, in each figure the bars indicate significance through nonparametric analyses. Bonferroni correction was applied to the non-parametric tests, and the corresponding p-values are listed in the figure legends. “*” indicates significance after adjustment for potential confounders in the univariate models. Odds ratios and p values for univariate and multivariate analysis models are listed in e-tables 1 and 2. Significance (two-tailed) was defined as p<0.05. All results were plotted using GraphPad Prism (version 7, mean ± standard error), and statistical analyses was performed using SAS version 9.4 (SAS Institute Inc. Cary, NC.)

Results:

Patient Population

Table 1 outlines the baseline characteristics of the study population. There were no differences between groups in terms of sex, age at Holter ECG recording, and years of follow-up. There was a higher prevalence of beta-adrenregic blocker use and longer QT_c duration, but not heart rate in LQTS1 and LQTS2, compared to LQTS⁽⁻⁾ patients. There were no differences in heart rate, QT_c, and the prevalence of beta blocker use, arrhythmias, and seizures in LQTS1 vs. LQTS2 patients.

LQTS Genotype Specific Differences in HR and QT Dynamics at Times of Minimum HR

Despite no differences in heart rate between groups at times of minimum heart rate, total ANS function (SDNN) was lower in LQTS⁽⁺⁾, LQTS1, and LQTS2, compared to LQTS⁽⁻⁾ patients (Figure1A.) Additionally, RMSSD and pNN50, which are indicative of vagal function, were lower in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients (Figure 1B & C.) These differences in vagal function remained significant after adjustment for potential confounders (eTable 1.) There were no LQTS genotype specific differences in sympathovagal balance (LF/HF, Figure 1D.) QTVI was larger in LQTS⁽⁺⁾ and LQTS1, compared to LQTS⁽⁻⁾ patients (Figure 1E), and these results remained significant after adjustment for potential confounders. There were no differences in LQTS1 vs. LQTS2 ANS and repolarization measurements.

Next, we investigated whether HRV measures provide independent markers of LQTS genotype. In a single logistic regression model we included the above described HRV measures (i.e., SDNN, RMSSD, pNN50, LF/HF), as well as adjustments for potential confounders (see methods.) pNN50 was an independent marker of LQTS⁽⁺⁾ (p=0.028, Figure 1F.)

As 3 LQTS⁽⁻⁾ patients had QT_c > 500ms, we conducted sensitivity analysis, in which we performed HRV and QTVI measurement in patients that were LQTS genotype and phenotype positive (LQTS mutation & QT_c >500) vs. LQTS genotype and phenotype negative (no LQTS mutation & QT_c < 500 ms.) As seen in e-Figure 1, similar results were seen in the refined cohort. Furthermore, HRV and QTVI measures were similar in LQTS⁽⁻⁾ patients with a QT_c < 500 ms vs. QT_c > 500 ms, and LQTS⁽⁺⁾ with a QT_c < 500 ms vs. QT_c > 500 ms.

To make the results of this study more generalizable to all LQTS patients, we assessed ANS and repolarization function in all patients regardless of whether they were on or off beta-adrenergic blockers at the time of the recording (e-Figure 2) Similar to analysis in patients off beta blockers, in the full cohort, ANS measures were lower and QTVI higher in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients compared to the LQTS⁽⁻⁾ group (see e-Figure 2 and eTable 1 for the specific ECG measures that reached significance.) Multivariate regression modeling indicated that pNN50 was an independent marker of LQTS⁽⁺⁾ (p=0.036.)

Next, we examined ANS function and cardiac repolarization lability in LQTS patients during specific physiological states. We selected 5-minute periods of minimum, middle, and maximum LF/HF. Heart rates were significantly different between each LF/HF timepoint, except for between LQTS⁽⁻⁾ timepoints and LQTS2 at middle vs. maximum (Table 2). Except for LQTS2 at max LF/HF, total (SDNN) and vagal (RMSSD & pNN50) ANS function were significantly reduced in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients, compared to LQTS⁽⁻⁾ at all physiological states and overall across all timepoints (Figure 2A–C.) In separate models for each HRV measure, we incorporated data from all three physiological states, adjusted for LF/HF and potential confounders, as well as accounted for repeated measures. SDNN and pNN50 differed in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients, compared to LQTS⁽⁻⁾. As an indication of more advanced disease, QTVI was significantly elevated in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients, compared to LQTS⁽⁻⁾ overall across all physiological states combined, as well as at minimum and middle LF/HF (Figure 2D.) These QTVI results and minimum and middle LF/HF remained significant even after adjusting for potential confounders. When including all HRV measures in a single multivariate regression model, SDNN, RMSSD, and pNN50 were independent markers for LQTS⁽⁺⁾ (p-values: 0.002, 0.047, 0.017), and SDNN and pNN50 (p-values: <0.001, 0.032) were independent markers of LQTS1. Additionally, RMSSD and pNN50 (p-values: 0.03, 0.014) were independent markers that distinguished LQTS1 vs. LQTS2 patients during periods of middle LF/HF (Figure 2E.)

Table 2:

Group Differences.

Group Differences	LQTS(−)	LQTS1	LQTS2	p-value all*	p-value 1 vs 2**
Minimum LF/HF
HR (bpm), mean ± SD	61.04 ± 2.56	60.33 ± 8.58	59.86 ± 10.74	0.431	0.725
QTc (ms), mean ± SD	438.76 ± 26	529.86 ± 51.52	562.84 ± 95.72	<0.001	0.124
Middle LF/HF
HR (bpm), mean ± SD	63.31 ± 9.32	72.17 ± 10.45	66.96 ± 11.22	<0.001	0.008
QTc (ms), mean ± SD	420.63 ± 37.5	518.25 ± 53.27	520.04 ± 73.69	<0.001	0.970
Maximum LF/HF
HR (bpm), mean ± SD	70.48 ± 17.30	78.43 ± 10.14	70.06 ± 13.17	0.002	0.004
QTc (ms), mean ± SD	398.15 ± 43.89	516.16 ± 47.41	541.89 ± 68.76	<0.001	0.058
Minimum vs. Middle vs. Maximum LF/HF			p-value LQTS(−)*	p-value LQTS1*	p-value LQTS2*
HR (Min. vs. Mid.)			0.985	<0.001	0.006
HR (Mid. vs. Max.)			0.25	<0.001	0.281
HR (Min. vs. Max.)			0.181	<0.001	0.001
QTc (Min. vs. Mid.)			0.213	0.186	0.048
QTc (Mid. vs. Max.)			0.086	0.493	0.164
QTc (Min. vs. Max.)			0.007	0.063	0.474

^*p-values were calculated using a Kruskal-Wallis Test.

^**p-values were calculated using a two-sided Wilcoxon Rank Sum test with normalized approximation.

Similar results were seen when performing the analysis using the entire cohort (patients on & off beta blockers.) Total (SDNN) and vagal (RMSSD & pNN50) function were lower and QTVI higher in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients during most physiological states and overall across all timepoints (see e-Figure 3 & e-Table 2 for the specific ECG measures that reached significance.)

HRV and QTVI Measures in LQTS Patients With vs. Without a History of Arrhythmias & Seizures

We assessed whether HRV and QTVI measures differed in LQTS patients with vs. without a history of arrhythmias (e-Table 5.) We selected epochs of minimum, middle, and maximum LF/HF, and assessed these ANS and repolarization measures overall across all three physiological states. Total (SDNN) and vagal (RMSSD & pNN50) ANS function was lower in LQTS1 patients with a history of arrhythmias (Figure 3A–C.) Despite the apparent higher QTVI in LQTS1 patients with a history of arrhythmias, it did not reach significance (Figure 3D.) SDNN remained significant in the univariate logistic regression model that adjusted for LF/HF, and potential confounders, including beta blocker use. None of the ECG measures were independent markers for arrhythmias in the multivariate regression model. In contrast, none of the HRV and QTVI measures differed in LQTS2 patients with vs. without a history of arrhythmias, or LQTS1 and LQTS2 patients with vs. without seizures, separately (Figure 3E–H & e-Figure 4). These results suggest that reduced ANS function, particularly SDNN, identifies LQTS1, but not LQTS2, patients with a history of arrhythmias.

Discussion:

Summary

Results from this study demonstrate that there is depressed ANS function and increased ventricular repolarization lability in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients. During periods of minimum HR, decreased pNN50, which is a marker of vagal function, was an independent marker for LQTS⁽⁺⁾. Additionally, in multivariate regression models that assessed ANS function across numerous physiological states combined, measures of total ANS (SDNN) and vagal function (RMSSD & pNN50) were all independent marker that distinguished LQTS⁽⁺⁾ vs. LQTS⁽⁻⁾ patients. SDNN and pNN50 were independent markers of LQTS1, compared to LQTS⁽⁻⁾. RMSSD and pNN50 were independent markers that distinguished LQTS1 vs. LQTS2 patients. ECG measures of total (SDNN) and vagal (RMSSD & pNN50) function differed in LQTS1 patients with vs. without a history of arrhythmias, and SDNN remained significant after adjustment for potential covariates. No differences were seen in LQTS2 patients with vs. without arrhythmias. Additionally, while genetic forms of epilepsy are associated with mutations in genes expressed in both the brain and heart^(11–13), and there is an increased risk of seizures in LQTS⁽⁺⁾ patients, this study supports the null hypothesis that, during non-seizure periods, measures of ANS function do not delineate LQTS patients with vs. without a history of seizures. This study stimulates future hypotheses about the LQTS genotype specific triggers for arrhythmias and the mechanisms for differences in ANS function and repolarization lability.

HRV Measures Provide a Marker for LQTS⁽⁺⁾ and are Reduced in LQTS1 Patients With Arrhythmias

In the normal heart, sympathetic stimulation leads to action potential and QT shortening, as well as a decrease in transmural dispersion of repolarization^{(14, 15)}. Yet, in ischemic heart disease, heart failure, and LQTS, sympathetic stimulation provides a trigger for arrhythmias⁽¹⁴⁾. HRV measures are useful for risk stratification for mortality, arrhythmias, and sudden cardiac death^{(5, 7)}.

LQTS1 patients are prone to arrhythmias upon sympathetic stimulation, such as during exercise and swimming⁽⁴⁾. The action potential duration and QT duration do not adapt or even prolong upon sympathetic stimulation^{(14, 16)}. LQTS2 patients often develop arrhythmias upon startle, stress, or arousal⁽⁴⁾. In LQTS patients with arrhythmias that are not controlled by beta blocker therapy, left cervicothoracic sympathetic ganglionectomy reduces the QT and many patients become free of arrhythmias^{(17, 18)}. In contrast, LQTS3 patients often develop arrhythmias during rest and bradycardia, which are times of increased voltage-gated sodium channel availability^{(4, 12)}. Unfortunately, as this is a retrospective dataset, we did not have any LQTS3 patients with Holter monitor ECG recordings. This would be a great future direction and an important group for a prospective study.

HRV analyses has not been thoroughly examined in LQTS. However, Perkiomaki et al. (2001)⁽¹⁹⁾ reported that at baseline, none of the HRV measures differed in LQTS⁽⁺⁾ vs. LQTS⁽⁻⁾ patients, nor between LQTS1 vs. LQTS2 vs. LQTS3 carriers⁽¹⁹⁾. We demonstrated in a larger patient cohort, HRV measures associated with total and vagal ANS function were lower in LQTS⁽⁺⁾, LQTS1, and LQTS2 patients. These results remained significant even after applying adjustments for potential confounders, such as QT_c duration, which is greatly prolonged in the LQTS⁽⁺⁾ group. Second, detailed analyses during specific physiological states indicated differences in ANS function in LQTS⁽⁺⁾, LQTS1, and LQTS2, compared to LQTS⁽⁻⁾. HRV measures of total ANS and vagal function were independent markers to distinguish LQTS⁽⁺⁾ and LQTS1 patients from LQTS⁽⁻⁾ patients, and vagal measures were independent indicators of LQTS1 vs. LQTS2. Third, HRV analyses provided non-invasive measures to identify LQTS1, but not LQTS2, patients with arrhythmias. Consistent with our results, Porta et al (2015)⁽²⁰⁾, reported that vagal function and QT variability distinguish symptomatic vs. asymptomatic LQTS1 patients with the same mutation⁽²⁰⁾.

LQTS Genotype and Arrhythmia Specific Differences in QTVI

QT variability is due to fluctuations in the ventricular activation, conduction, and repolarization patterns, which are influenced by ANS function and heart rate⁽⁸⁾. In patients with dysautonomia and cardiac disease (e.g., coronary artery disease, ischemia, myocardial infarction, left ventricular hypertrophy, & hypertension), increased QTVI is associated with arrhythmic and cardiovascular death; QTVI provides a predictor for appropriate shock^{(9, 21, 22)}. QTVI is elevated in patients with spinal cord injuries above T5–T6⁽⁸⁾.

Our results align with previous studies that indicate that reduced repolarization reserve (i.e., LQTS mutations) are associated with increased QT variability^{(23, 24)}. Bilchick et al (2004) reported that during periods of rest, there was increased QTVI in LQTS2 vs. LQTS⁽⁻⁾ family members⁽²⁵⁾. We observed similar differences, as well as differences in LQTS1 vs. LQTS⁽⁻⁾ patients. We extended these results by assessing QTVI during various physiological states, and in LQTS⁽⁺⁾ patients with vs. without arrhythmias and seizures. Even after adjusting for potential confounders, QTVI was higher in LQTS1 and LQTS2 patients. During blockade of the slow delayed rectifier current (I_Ks, mutant current in LQTS1), beta adrenergic stimulation leads to increased repolarization variability^{(26, 27)}. In a canine model of LQTS1, I_Ks blockade was associated with elevated QT variability⁽²⁶⁾, and increased QT variability was a predictor for arrhythmogenesis⁽²⁸⁾. Anti-depressant medications are associated with increased QT variability⁽⁸⁾.

Limitations

Beta-adrenergic blockers alter ANS function. In addition to blockade of sympathetic function, during various physiological states, beta adrenergic blocker therapy leads to increased vagal activity⁽²⁹⁾. As this is a retrospective collection of ECG recordings, and beta blockers are the main line of therapy in LQTS⁽⁺⁾, there were differences in the prevalence of beta blocker use in LQTS⁽⁺⁾ vs. LQTS⁽⁻⁾.Since none of the LQTS⁽⁻⁾ patients were on beta blockers during the recordings, we were unable to adjust for beta blocker use in models that included LQTS⁽⁻⁾ patients. Thus, in Figures 1 and 2, we limited assessments of HRV and QTVI measures to patients not on beta blockers. Yet, similar effects were seen when also including patients on beta blockers (e-Figures 2 & 3.) There were no differences in beta-blocker use in LQTS1 vs. LQTS2, as well as in LQTS patients with vs. without arrhythmias/seizures (e-Figure 5). We attempted to account for this possible confounder by performing univariate and multivariate logistical regression modeling that included an adjustment for beta blocker use (Figure 3 & e-Figure 4.) Unfortunately, the dose of beta-blocker was not captured in the database.

We examined ANS function and repolarization dynamics during basal non-seizure and non-arrhythmia states, as we did not have recordings surrounding these events. As ANS changes are often reported surrounding arrhythmias and seizures^{(6, 7, 20)(30–32)}, future studies need to investigate the temporal evolution of HRV and QTVI measures in LQTS patients leading up to and following these events.

Conclusion:

Our study investigated ANS function and repolarization lability in LQTS1, LQTS2, and LQTS⁽⁻⁾ patients during specific physiological states. LQTS⁽⁺⁾, LQTS1, and LQTS2 patients exhibit reduced total (SDNN) and vagal (RMSSD, pNN50) ANS function, as well as increased QTVI. Reduced SDNN, RMSSD, and pNN50 were independent markers for LQTS⁽⁺⁾ across various physiological states. Also, SDNN and pNN50 were independent markers for LQTS1 vs. LQTS⁽⁻⁾ patients. During periods of sympathovagal balance, vagal function (RMSSD & pNN50) were independent markers that distinguished LQTS1 vs. LQTS2 patients. LQTS1, but not LQTS2, patients with a history of arrhythmias exhibited reduced ANS function (SDNN, RMSSD, & pNN50), and SDNN remained significant after adjustment for confounders. Despite the ANS providing a direct neural connection between the brain and heart, a history of seizures in LQTS was not associated with altered basal ANS function.

Highlights.

• HRV and QTVI are non-invasive measures of autonomic and cardiac electrical function.

• These ECG measures differ in LQTS⁽⁻⁾ vs LQTS⁽⁺⁾, LQTS1, and LQTS2.

• SDNN, RMSSD, and pNN50 are independent markers of LQTS-genotype.

• HRV measures are depressed in LQTS1 patients with arrhythmias.

• Basal autonomics are similar in LQTS1 and LQTS2 with vs. without seizures.

List of Abbreviations:

LQTS

Long QT Syndrome (LQTS⁽⁺⁾, Type 1: LQTS1, Type 2: LQTS2; Negative, LQTS⁽⁻⁾)

HR

heart rate

HRV

heart rate variability

QTVI

QT variability index

SUDEP

Sudden Unexpected Death in Epilepsy

ANS

autonomic nervous system

NN

normal sinus beat-to-beat (NN) interval

SDNN

standard deviation of NN

RMSSD

root mean square of successive differences in paired NN intervals

pNN50

percentage of successive NN intervals that differed more than

LF

low frequency

HF

high frequency