Brugada syndrome is a used to describe the combination of specific ECG changes, the Brugada pattern, in addition to life threatening arrhythmias and sudden cardiac death (SCD). Traditionally, it has been considered a congenital ion channelopathy linked to abnormalities in the cardiac sodium channel. Recently, pathogenic mutations in other ion channels have been described. Mechanisms of arrhythmogenesis can be broadly divided into triggered activity and re-entry. Of these, re-entry is thought to be the predominant mechanism underlying increased arrhythmogenicity in Brugada syndrome requiring an increased spatial dispersion of repolarization. Such re-entrant activity may involve direct electrotonic activation during phase 2 of the cardiac action potential, as shown in pre-clinical studies using arterially perfused, canine wedge preparations, or circus-type/spiral wave activity around an anatomical or functional obstacle. Regardless of the precise underlying mechanism for re-entry, this transmural dispersion of repolarization can be quantified electrocardiographically by the interval from the peak to the end of the T-wave (Tpeak-Tend interval), (Tpeak-Tend)/QT ratio and Tpeak-Tend dispersion.
However, not all studies have shown an association between higher Tpeak-Tend intervals, (Tpeak-Tend)/QT ratio or Tpeak-Tend dispersion with an arrhythmogenic phenotype in Brugada Syndrome. Recently, Mugnai and colleagues conducted one of the largest retrospective studies to date, including a total of 448 patients with spontaneous or drug induced type 1 Brugada pattern. They found no statistically significant difference in all three indices between asymptomatic subjects and patients with syncope and malignant arrhythmias. Morita and colleagues also found in 471 patients no difference in Tpeak-Tend intervals between patients with syncope or VT/VF and those who were asymptomatic. These findings contrast with a meta-analysis published previously by some members of our group, which extracted and pooled odds or hazard ratios for the relationship between Tpeak-Tend and arrhythmic and/or mortality outcomes in various clinical conditions, including Brugada Syndrome. This demonstrated prolonged Tpeak-Tend interval was associated with an increased risk of ventricular arrhythmias and SCD in Brugada Syndrome.
However, our previous study did not determine the absolute mean values for Tpeak-Tend, nor was it possible to include the largest dataset from Mugnai and colleagues. Moreover, it did not investigate the utility of other indices such as (Tpeak-Tend)/QT ratio or Tpeak-Tend dispersion. Therefore, we conducted a systematic review with meta-analysis into the relationships between Tpeak-Tend interval, (Tpeak-Tend)/QT ratio and Tpeak-Tend dispersion and arrhythmic and/or mortality endpoints in Brugada Syndrome.
METHODS Search strategy, inclusion and exclusion criteriaThis study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISM) statement. PubMed and Embase were searched for studies that investigated the association between Tpeak-Tend or Tpeak-Tend /QT with arrhythmic or mortality endpoints in Brugada syndrome. The following search terms were used for both databases: [“Tpeak-Tend” or “Tpeak-end” or “Tp-e” AND Brugada]. The databases were searched until 1 May 2018 without language restrictions. The following inclusion criteria were used: (a) the study was a case-control, prospective or retrospective cohort study in human subjects with a Brugada phenotype, (b) Tpeak-Tend intervals or (Tpeak-Tend) /QT ratios were provided; (c) predefined adverse events (appropriate implantable cardioverter-defibrillator therapy [ICD], syncope, ventricular tachycardia/fibrillation [VT/VF], SCD, cardiovascular death [CVD], major adverse cardiac events [MACE]) or all-cause mortality were reported. In cases of incomplete data from the published studies, the original authors were contacted, but no replies were received.
The Newcastle-Ottawa Quality Assessment Scale (NOS) was used for quality assessment of the included studies. The NOS system evaluated the categories of study participant selection, results comparability, and quality of the outcomes. Specifically, the following characteristics were assessed: (a) representativeness of the exposed cohort; (b) selection of the non-exposed cohort; (c) ascertainment of exposure; (d) demonstration that outcome of interest was not present at the start of study; (e) comparability of cohorts based on study design or analysis; (f) assessment of outcomes; (g) follow-up periods that were sufficiently long for outcomes to occur; and (h) adequacy of follow-up of cohorts. This scale varied from zero to nine stars, which indicated that studies were graded as poor quality if the score was <5, fair if the score was 5-7, and good if the score was >8. Studies with a score equal to or higher than six were included. The details of the NOS quality assessment are shown in Tables S1 and S2.
Data extraction and statistical analysisData from the different studies were entered in pre-specified spreadsheets in Microsoft Excel. All potentially relevant studies were retrieved as complete manuscripts, which were assessed fully to determine their compliance with the inclusion criteria. We extracted the following data from the included studies: (a) publication details: last name of first author, publication year and locations; (b) study design; (c) endpoint(s); (d) quality score; and (e) characteristics of the population including sample size, gender, age and number of subjects. Two reviewers (GT and MG) reviewed each included study independently. Disagreements were resolved by adjudication with input from a third reviewer (TL).
Adverse events were defined as ventricular arrhythmias (VT/VF), SCD, cardiovascular death, MACE or all-cause mortality. If more than one mortality endpoint was described, then SCD was preferentially used for analysis, followed by cardiovascular and all-cause mortality in this order. Mean differences between event-positive and event-negative groups, with 95% confidence intervals (CIs) for Tpeak-Tend interval, (Tpeak-Tend)/QT ratio and Tpeak-Tend dispersion were extracted and subsequently combined to generate a pooled estimate.
Heterogeneity between studies was quantified using The Cochran's Q value and the I2 statistic from the standard chi-square test, which describes the percentage of the variability in effect estimates resulting from heterogeneity. I2 > 50% was considered to reflect significant statistical heterogeneity. A fixed effects model was used if I2 < 50%. The random-effect model using the inverse variance heterogeneity method was used when I2 > 50%. To locate the origin of the heterogeneity, sensitivity analysis by excluding one study at a time, and subgroup analyses based on different disease conditions and different endpoints were performed. Funnel plots, Begg and Mazumdar rank correlation test and Egger's test were used to detect publication bias.
RESULTSFigure shows a flow diagram detailing the above search terms with inclusion and exclusion criteria. A total of 29 and 57 entries were retrieved from PubMed and Embase, respectively. Nine studies met the inclusion criteria and were included in our final meta-analysis. In this meta-analysis, a total of 1740 subjects with Brugada Syndrome were included (mean age 45 years old, 80% male). The mean follow-up duration was 68 ± 27 months. Of the entire cohort, 40% had a spontaneous Type 1 pattern and 19% were positive for SCN5a mutation. The baseline characteristics of these studies and of the study populations are shown in Table .
Characteristics of the nine studies included in this meta-analysis
| First author/year | Sample size (n) | Tpeak-Tend measurement: method and leads | Age (SD) | No. of males (%) | No. of Sp. type 1 patients (%) | No. of SCN5a positive patients (%) | Endpoints | Comparisons | No. of patients with adverse events /without adverse events/%/% per year | Follow-up duration (months) | Quality score | Ref. |
| Morita 2017 | 471 | Tangent method; V1, V2, V3, V5 | 47 (19) | 447 (95) | 118 (25) | 27 (15) | Syncope or VT/VF | Syncope/VT/VF vs asymptomatic | 145/326/31/4.09 | 91 | 7 | |
| Mugnai 2017 | 448 | End of the T-wave; V1 to V6 | 45 (16) | 273 (61) | 96 (21) | 55 (22) | Spontaneous VF or SCD | AT/SD vs asymptomatic | 43/290/13/1.67 | 93 | 6 | |
| Kawazoe 2016 | 143 | Tangent method; V1 to V6 | 46 (12) | 140 (98) | 84 (59) | – | VF | VF vs no VF | 35/108/24/1.9 | 105 | 7 | |
| Zumhagen 2016 | 78 | Tangent method; V1 | 45 (14) | 57 (73) | 22 (28) | 17 (22) | Spontaneous VT/VF | VT/VF/aborted SCD vs asymptomatic/syncope | 22/54/29/- | – | 6 | |
| Maury 2015 | 325 | Tangent method; V1 to V4 | 47 (13) | 260 (80) | 143 (44) | 43 (13) | Spontaneous VT/VF | AT/SD vs asymptomatic | 26/226/10/2.50 | 48 | 7 | |
| Letsas 2010 | 23 | End of the T-wave; V2, V6 | 43 (15) | 19 (83) | 10 (43) | – | Inducible VT/VF | Inducible VT vs no inducible VT | 17/6/74/16.15 | 55 | 6 | |
| Junttila 2008 | 200 | End of the T-wave; V2, II | 40 (16) | 143 (72) | 200 (100) | 25 (50) | Syncope, VT/VF, SCD | Syncope/VT/VF/aborted SCD vs asymptomatic | 66/134/33/- | – | 7 | |
| Wang 2007 | 23 | End of the T-wave; Max from V1 to V6 | 45 (8) | 23 (100) | – | – | Spontaneous VT/VF | Syncope/VT/VF/inducible VT vs asymptomatic | 11/9/55/5.12 | 43 | 8 | |
| Castro Hevia 2006 | 29 | Tangent method, Max from V1 to V6 | 41 (12) | 25 (86) | 15 (52) | – | Spontaneous VT/VF | Presyncope/syncope/aborted SCD vs asymptomatic | 12/17/41/3.81 | 43 | 8 |
SCD: sudden cardiac death; VT: ventricular tachycardia; VF: ventricular fibrillation; Sp.: spontaneous.
Tpeak-TendFor determining Tend, the tangent method and the return of the voltage to baseline method were used. Tpeak-Tend intervals from different leads and the maximum of these measurements have been presented by most studies. Regarding maximum Tpeak-Tend intervals, the mean value for the event-positive group was 98.9 ms (95% CI: 90.5-107.2 ms) (Figure A) and event-negative group was 87.7 ms (95% CI: 80.5-94.9 ms) (Figure B). Five studies reported longer values in the event-positive compared to event-negative groups, whereas four studies reported no significant difference (Figure C). Tpeak-Tend intervals were 11.9 ms longer (95% CI: 3.6-20.2 ms, P = 0.005) in event-positive patients than in event-negative patients. The Cochran's Q value was greater than the degrees of freedom (56 vs 8), indicating that the true effect size was different between studies. I2 took a value of 86%, suggesting the presence of substantial heterogeneity. A funnel plot plotting standard errors against differences in means is shown in Figure S1. Begg and Mazumdar rank correlation analysis demonstrated that Kendall's Tau took a value of 0.3 with P = 0.30, which suggests no significant publication bias. Egger's test demonstrated no significant asymmetry (intercept 2.4, t-value 1.2; P = 0.25). To identify the source of the heterogeneity, sensitivity analysis was performed by removing one study at a time, but this did not significantly influence the mean difference (Figure S2), suggesting that no single study was responsible for the heterogeneity observed in this meta-analysis. Subgroup analysis based on the method of Tend determination was performed. For the tangent method, the Tpeak-Tend mean difference was 15.5 ms (95% CI: 3.9-27.2 ms; P = 0.009) and I2 remained high at 90%. For full recovery of voltage to baseline, the mean difference was 6.0 ms (95% CI: 0.7-11.4 ms; P = 0.006) and I2 remained high at 76%. Therefore, different methods of Tend determination did not introduce significant heterogeneity to the pooled effect estimate.
Enlarge this image.Forest plot demonstrating Tpeak-Tend intervals obtained from event-positive (A) and event-negative (B) groups and the mean difference between both groups (C) in Brugada Syndrome
Regarding maximum (Tpeak-Tend)/QT ratio, the mean value for the event-positive group was 0.221 (95% CI: 0.208-0.234) (Figure A) and event-negative group was 0.210 (95% CI: 0.205-0.214) (Figure B). Two studies reported higher values in Brugada subjects with positive events compared to those without such events, whereas four studies demonstrated no significance between the groups (Figure C). Pooling of the mean values demonstrated significantly higher (Tpeak-Tend)/QT ratios in the event-positive group than in the event-negative group (mean difference = 0.019, 95% CI: 0.003-0.036, P = 0.024). The Cochran's Q value was greater than the degrees of freedom (19 vs 5), indicating that the true effect size was different between studies. I 2 took a value of 74%, suggesting significant heterogeneity. A funnel plot plotting standard errors against differences in means is shown in Figure S3. Begg and Mazumdar rank correlation analysis demonstrated that Kendall's Tau took a value of 0.07 with P = 1, which suggested no significant publication bias. Egger's test demonstrated no significant asymmetry (intercept 3.5, t-value 1.1; P = 0.31). To identify the source of the heterogeneity, sensitivity analysis was performed by removing one study at a time, but this did not significantly influence the mean difference (Figure S4), suggesting that no single study was responsible for the heterogeneity observed in this meta-analysis. Subgroup analysis based on the method of Tend determination was performed. For the tangent method, the mean difference of (Tpeak-Tend)/QT ratio was 0.03 (95% CI: 0.01-0.05; P < 0.05) and I2 was lowered to 55%. For full recovery of voltage to baseline, the mean difference was only 0.004 (95% CI: −0.03 to 0.03 ms; P = 0.81) and I2 remained high at 74%. Therefore, different method of Tend determination appeared to contribute partially to the heterogeneity of the pooled effect estimate. Moreover, statistical significance was achieved when the tangent method was used, but was lost when the return to baseline method was used, which may suggest the former approach may be more sensitive.
Enlarge this image.Forest plot demonstrating Tpeak-Tend/QT ratios obtained from event-positive (A) and event-negative (B) groups and the mean difference between both groups (C) in Brugada Syndrome
Regarding maximum Tpeak-Tend dispersion, the mean value for the event-positive group was 40.8 ms (95% CI: 26.9-54.8 ms) (Figure A) and event-negative group was 29.7 ms (95% CI: 24.5-34.8 ms) (Figure B). Regarding Tpeak-Tend dispersion, two studies reported longer values in event-positive group compared to event-negative groups, whereas three studies found no significant difference (Figure C). Overall, pooling of the data showed that Tpeak-Tend dispersion was significantly higher in the event-positive than in the event-negative groups (mean difference = 7.8 ms, 95% CI: 2.1 to 13.4 ms, P = 0.007). The Cochran's Q value was greater than the degrees of freedom (20 vs 4), indicating that the true effect size was different between studies. I2 took a value of 80%, suggesting significant heterogeneity. A funnel plot plotting standard errors against differences in means is shown in Figure S5. Begg and Mazumdar rank correlation analysis demonstrated that Kendall's Tau took a value of −2 with P = 0.62, which suggests no significant publication bias. Egger's test demonstrated no significant asymmetry (intercept −5.4, t-value 0.8; P = 0.48). To identify the source of the heterogeneity, sensitivity analysis was performed by removing one study at a time, but this did not significantly influence the mean difference between event-positive and event-negative groups (Figure S6), suggesting that no single study was responsible for the heterogeneity observed in this meta-analysis. Subgroup analysis based on the method of Tend determination was performed. For the tangent method, the mean difference of Tpeak-Tend dispersion was 16.2 ms (95% CI: 7.9-24.5 ms; P < 0.0001) and I2 was 65%. For full recovery of voltage to baseline, the mean difference was 0.4 ms (95% CI: −7.3 to 8.2 ms; P = 0.91) and I2 was reduced to 19%. Therefore, different method of Tend determination contributed heterogeneity to the pooled effect estimate. Moreover, statistical significance was achieved when the tangent method was used, but was lost when the return to baseline method was used, which may suggest the former approach may be more sensitive.
Enlarge this image.Forest plot demonstrating Tpeak-Tend dispersion obtained from event-positive (A) and event-negative (B) groups and the mean difference between both groups (C) in Brugada Syndrome
SCN5A is the commonest ion channel gene that is mutated in Brugada syndrome. Separate meta-analyses were conducted to compare the different Tpeak-Tend parameters between patients with and without SCN5A mutations. Two of the included studies provided sufficient information for such analyses. No significant difference in Tpeak-Tend (mean difference = 8.2 ms, 95% CI: −6.7 to 23.2 ms, P = 0.28; I2 = 59%; Figure S7), Tpeak-Tend/QT ratio (mean difference = −0.006 ms, 95% CI: −0.023 to 0.011 ms, P = 0.47; I2 = 24%; Figure S8) or Tpeak-Tend dispersion (mean difference = 5.2 ms, 95% CI: −2.9 to 13.2 ms, P = 0.21; I2 = 31%; Figure S9) was observed between patients with and without SCN5A mutations.
DISCUSSIONThe main findings of our meta-analysis, which included 1597 Brugada subjects, are (a) Tpeak-Tend intervals, (b) (Tpeak-Tend)/QT ratio and (c) Tpeak-Tend dispersion are higher in Brugada subjects with adverse cardiac events (ventricular tachy-arrhythmias and SCD) when compared to Brugada subjects free from such events.
The presence of pre-existing electrophysiological heterogeneities is important for mediating the normal, unidirectional spread of action potentials in the heart. These are attributed to differences in repolarization times of the different cell types, which are responsible for generation of the T-wave on the electrocardiogram (ECG). However, exacerbation of such differences has been associated with ventricular tachy-arrhythmias in different conditions, thereby generating a pro-arrhythmic phenotype. These include congenital ion channelopathies such as long QT syndrome and Brugada syndrome and acquired cardiac diseases such as myocardial infarction. These heterogeneities can occur locally or across the myocardial wall, potentially causing arrhythmias by inducing unidirectional conduction block and therefore circus-type or spiral wave re-entry. Moreover, a greater epicardial-endocardial repolarization time difference may increase the propensity of phase 2 re-entry, which is hypothesized to generate extrasystolic activity in Brugada syndrome. This occurs when sites with an action potential dome to sites which a dome morphology, leading to direct depolarization of the downstream sites. Once an extrasystole is generated, together with a favorable re-entrant substrate, ventricular tachycardia and fibrillation can result.
A number of electrocardiographic indices have been proposed for stratification of arrhythmic or mortality risk. Of these, Yan and Antzelevitch were the first to propose the use of the difference between the peak and the end of the T-wave (the Tpeak-Tend interval) as a measure of transmural dispersion of repolarization. Subsequent clinical studies have demonstrated that, confirmed recently in a systematic review and meta-analysis from our group, that Tpeak-Tend prolongation significantly elevated the risk of ventricular tachy-arrhythmias and/or SCD in heart failure, ischemic heart disease, Brugada syndrome, hypertension, and the general population. Recently, Mugnai and colleagues in a total of 448 subjects found no significant differences Tpeak-Tend intervals, (Tpeak-Tend)/QT ratio or Tpeak-Tend dispersion between patients with VT/VF requiring anti-tachycardia pacing or with sudden death, and those who were asymptomatic. Similarly, in a separate population of 471 subjects, Morita and colleagues found no significance difference in Tpeak-Tend intervals between patients with syncope or VT/VF and asymptomatic patients. Publication of these two studies prompted us to conduct this meta-analysis, which confirms the value of Tpeak-Tend interval, (Tpeak-Tend)/QT ratio and Tpeak-Tend dispersion, in distinguishing high-risk patients from low-risk patients.
In the Mugnai study, the largest study to date, the percentage of patients with adverse events were the lowest at 13%. Male gender, a spontaneous Type 1 Brugada pattern and SCN5a mutation positive status were significantly associated with ventricular arrhythmias. Therefore, the lower percentage of patients with adverse events can be explained by the lower percentage of Type 1 Brugada patients (21% vs 28%-100% in the remaining studies) and lower percentage male patients (61% vs 72%-100%) despite similar percentage with SCN5a positive status (22% vs 13%-50%). While these differences in patient characteristics affect the likelihood of adverse events occurring, they should not explain the lack of difference in Tpeak-Tend intervals between event-positive and event-negative groups in the Morita study or the Mugnai study. Interestingly, Mugnai and colleagues found a non-statistically significant lower Tpeak-Tend intervals in event-positive groups. Of the remaining six studies, five studies had reported significantly higher Tpeak-Tend intervals and one study reported no difference. A recent epidemiological study reported a U-shaped relationship between Tpeak-Tend intervals and increased mortality. Autonomic modulation, which is part of Coumel's triad for arrhythmogenesis, is known to modulate the re-entrant substrate. Increased activity of the parasympathetic nervous system may reduce Tpeak-Tend intervals, which may also be pro-arrhythmic. By contrast, exercise, during which sympathetic activity is increased, can exacerbate pre-existing heterogeneities, such as producing conduction slowing and increasing the dispersion of repolarization.
In our previous meta-analysis pooling together studies that reported odds ratios or hazard ratios, the average cut-off for Tpeak-Tend was 95.8 ms across different clinical conditions. The present meta-analysis pooling mean values for event-positive and -negative groups clearly indicates that the 100 ms cut-off is too high for Brugada syndrome. Our data would support a lower cut-off value between 88 and 99 ms to be used. This cut-off will also be method-dependent for determining Tend in the case of the Tpeak-Tend intervals. Previously, it was shown that in a cohort of high-risk Brugada subjects, only 10 of 16 studies reported a Tpeak-Tend longer than 100 ms, supporting our notion that this cut-off value may be too high. Moreover, different studies measured Tpeak-Tend from different leads. Some had measured it from all 12 leads and taken the mean values while others have done so for V1 to V3 only. While there is no consensus as to which leads are most appropriate for measurement, obtaining it from all 12 leads is likely to be less useful clinically due to the time-consuming nature. To simplify Tpeak-Tend determination, we would thus propose measuring it from the right precordial leads given BrS is primarily a right ventricular disorder.
While it may appear that the difference in Tpeak-Tend between high-risk and low-risk Brugada patients was only small, at around 12 ms, it should be emphasized that increased transmural dispersion of repolarization is only one mechanism by which re-entrant arrhythmogenesis is generated. Other mechanisms, such as reduced conduction velocity, increased dispersion of conduction or dynamic substrates such as steep action potential restitution, in which normal Tpeak-Tend interval, Tpeak-Tend/QT ratio or Tpeak-Tend dispersion may be observed, also contribute to arrhythmogenesis in Brugada syndrome. Therefore, better risk stratification scores will need to incorporate a combination of repolarization and conduction indices. Moreover, some of these dynamic changes may not be detectable on the ECG and may require additional tests such as non-invasive ECG imaging (ECGi), or only becomes detectable only under stressful conditions such as exercise.
LimitationsThe following limitations of this meta-analysis should be noted. First, there is marked heterogeneity between the included studies. The method of Tpeak-Tend determination across the studies was split even between the tangent method and full recovery of the voltage to baseline. Subgroup analysis based on the method used did not reduce the heterogeneity observed. Therefore, measurement method was unlikely to have significantly contributed to the heterogeneity observed. Moreover, the Letsas 2010 study used a different endpoint of inducible VT compared to the remaining studies, but its exclusion did not significant affect the mean Tpeak-Tend values for event-positive group, event-negative group, and mean difference between these groups. Second, retrospective studies may have more bias than prospective studies. Finally, it should be acknowledged that there is overlap between event-postiive and event-negative groups irrespective of the method of measuring Tend. This would suggest as a single measurement, Tpeak-Tend is unlikely to be useful in its own right. Indeed, accurate risk stratification will require a composite scoring system assessing not only dispersion of repolarization, but that of conduction, clinical symptoms, family history, the type of Brugada pattern, genetic background, electrical and drug provocation testing as well as electrophysiological mapping.
CONCLUSIONSTpeak-Tend interval, Tpeak-Tend/QT ratio and Tpeak-Tend dispersion were higher in high-risk than low-risk Brugada subjects, and thus offer incremental value for risk stratification.
ACKNOWLEDGEMENTSGT thanks the Croucher Foundation of Hong Kong for the support of a clinical assistant professorship.
CONFLICTS OF INTERESTSAuthors declare no Conflict of Interests for this article.
© 2018. This work is published under http://creativecommons.org/licenses/by-nc/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Gong, Mengqi 2 ; Christien Ka Hou Li 3 ; Keith Sai Kit Leung 4 ; Georgopoulos, Stamatis 5 ; Bazoukis, George 5
; Letsas, Konstantinos P 5 ; Sawant, Abhishek C 6 ; Mugnai, Giacomo 7
; Martin C.S. Wong 8 ; Gan, Xin Yan 9 ; Brugada, Pedro 7 ; Gian-Battista Chierchia 7 ; de Asmundis, Carlo 7
; Baranchuk, Adrian 10
; Liu, Tong 2 1 Department of Medicine and Therapeutics, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China; Li Ka Shing Institute of Health Sciences, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China; Shenzhen Research Institute, The Chinese University of Hong Kong, Shenzhen, China
2 Tianjin Key Laboratory of Ionic-Molecular Function of Cardiovascular Disease, Department of Cardiology, Tianjin Institute of Cardiology, Second Hospital of Tianjin Medical University, Tianjin, China
3 Department of Medicine and Therapeutics, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China; Li Ka Shing Institute of Health Sciences, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China; Shenzhen Research Institute, The Chinese University of Hong Kong, Shenzhen, China; Faculty of Medicine, Newcastle University, Newcastle, UK
4 Department of Medicine and Therapeutics, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China; Li Ka Shing Institute of Health Sciences, Faculty of Medicine, Chinese University of Hong Kong, Hong Kong, China; Shenzhen Research Institute, The Chinese University of Hong Kong, Shenzhen, China; Aston Medical School, Aston University, Birmingham, UK
5 Second Department of Cardiology, Laboratory of Cardiac Electrophysiology, Evangelismos General Hospital of Athens, Athens, Greece
6 Division of Cardiology, Department of Internal Medicine, State University of New York at Buffalo, Buffalo, New York
7 Heart Rhythm Management Center, Postgraduate Program in Cardiac Electrophysiology and Pacing, Universitair Ziekenhuis Brussel-Vrije Universiteit Brussel, Brussels, Belgium
8 JC School of Public Health and Primary Care, The Chinese University of Hong Kong, Hong Kong, China
9 Lankenau Institute for Medical Research and Lankenau Medical Center, Wynnewood, Pennsylvania; Beijing Anzhen Hospital, Capital Medical University, Beijing, China
10 Division of Cardiology, Kingston General Hospital, Queen's University, Kingston, ON, Canada