Background Although QRS duration >180 ms has prognostic value in adults with tetralogy of Fallot (TOF), its sensitivity to predict mortality is low. Fragmented QRS complexes, a simple measurement on ECG, are related to myocardial fibrosis and dysfunction in patients with TOF. Our objective was to determine whether QRS fragmentation predicts major outcomes in TOF.
Methods This multicentre study included adult patients with TOF from a prospective registry. Notches in the QRS complex in ≥2 contiguous leads on a 12-lead ECG, not related to bundle branch block, were defined as QRS fragmentation, which was classified as none, moderate (≤4 leads) or severe (≥5 leads). The primary and secondary outcomes were all-cause mortality and clinical ventricular arrhythmia, respectively.
Results A total of 794 adult patients with TOF (median age 27 years, 55% male; 52% no QRS fragmentation, 32% moderate, 16% severe) were included. During long-term (median 10.4 years) follow-up, 46 (6%) patients died and 35 (4%) patients had ventricular arrhythmias. Overall, 10-year survival was 98% in patients without fragmented QRS complexes, 93% in patients with moderate QRS fragmentation and 81% in patients with severe QRS fragmentation. In multivariable Cox hazards regression analysis, extent of QRS fragmentation (HR: 2.24/class, 95% CI 1.48 to 3.40, p<0.001) remained independently predictive for mortality, whereas QRS duration was not predictive (p=0.85). The extent of QRS fragmentation was also independently predictive for ventricular arrhythmia (HR: 2.00/class, 95% CI 1.26 to 3.16, p=0.003).
Conclusions The extent of QRS fragmentation is superior to QRS duration in predicting mortality in adult patients with TOF and may be used in risk stratification.
Statistics from Altmetric.com
Despite improved survival and surgical techniques, adults born with congenital heart disease (CHD) remain at risk for life-threatening arrhythmias, heart failure and sudden death.1–6 In patients with tetralogy of Fallot (TOF), right ventricular (RV) dysfunction is thought to contribute to late complications and has been related with QRS duration.1 ,2 ,7 Severe QRS prolongation has been widely recognised as a risk factor for mortality in adults with TOF.1–3 ,7–10 However, sensitivity of QRS >180 ms for mortality was lower than 50% in recent studies.3 ,8
In patients with coronary artery disease, the fragmentation of QRS (fQRS) complexes has been identified as an independent predictor of cardiac events.11 Recent studies revealed that fQRS is strongly related with regional RV dysfunction and myocardial fibrosis in adult patients with TOF.12 ,13 Myocardial fibrosis is associated with deterioration of myocardial contractility, and may serve as a substrate for arrhythmias.14–16 However, the prognostic significance of fQRS in adults with TOF remains unknown. Our primary objective was to determine whether the presence and extent of fQRS predict all-cause mortality in adults with TOF.
We used the prospective Dutch nationwide congenital corvitia (CONCOR) registry17 to identify patients with TOF, in one of the six participating centres. Since 2001, patients were included by two permanently employed research nurses. CONCOR registry was approved by the ethics board of all participating centres and all patients provided informed consent. Clinical data such as diagnosis, surgery and clinical events were obtained from medical records and classified according to the European Pediatric Cardiac Code Short List scheme.
The standard 12-lead ECG (25 mm/s, 10 mm/mV) at time closest to CONCOR inclusion was retrieved. All ECGs were analysed by a single observer (JPB) blinded to patient characteristics and clinical data. The rhythm, rate, morphology and duration of QRS and JT were determined. QRS complexes were assessed on the presence of fragmentation on each lead. Most patients had right bundle branch block (RBBB). In these patients, QRS fragmentation was defined as ≥3 R-waves/notches in the R/S complex (more than typical 2 in RBBB) in ≥2 contiguous leads (right sided/septal: aVR, V1, V2; anterior: V2–V5; lateral: I, aVL, V5, V6; or inferior: II, aVF, III) (figure 1A,B). In paced QRS and premature ventricular complexes, QRS fragmentation was defined as ≥3 notches in the R/S complex. In patients with QRS <120 ms, QRS fragmentation was defined as an additional R wave (R’) or notch in the nadir of the S wave. The assessment of fQRS was according to previously reported protocols and can be performed within a minute per ECG,11–13 ,18 To allow easy interpretation of extent of fQRS on clinical impact, extent of fQRS was classified as none, moderate (2–4 leads with fQRS) or severe (5–12 leads). A random sample of 40 ECGs was assessed on fQRS by a second independent observer (JTV).
The primary outcome variable was all-cause mortality. Date and cause of death were obtained from hospital databases and general practitioners and verified by linking CONCOR with the Dutch Central Bureau of Statistics. The secondary outcome was clinical ventricular arrhythmia (VA). Hospital databases were reviewed for clinical VA, which was defined as: (1) documented (eg, on Holter or pacemaker) symptomatic and/or recurrent non-sustained ventricular tachycardia (VT), requiring intervention such as cardioversion or ablation therapy, (2) documented sustained VT, lasting ≥30 s or requiring cardioversion and (3) ventricular fibrillation or out-of-hospital cardiac arrest, with successful resuscitation.
Categorical data were described as numbers with percentage. Continuous data were described as median with IQR or mean with SD, as appropriate. Differences on baseline variables were assessed using independent samples t-test, χ2 test or Wilcoxon rank-sum test, as appropriate. A univariable Cox proportional hazards analysis was used to identify variables predictive for the primary and secondary outcomes. A multivariable conditional forward-selected Cox proportional hazards analysis was performed with all variables with a p value <0.10 in the univariable analysis. The assumption of proportional hazards was checked with log-minus-log plots and by adding a time-dependent covariate. The C-statistic of predictive variables was calculated, which reflected the area under the curve for binary outcome variables with censored survival time data. The C-statistic for the multivariable model was calculated, and reflected the area under the curve for binary outcome variables with censored survival time data. The C-statistic for the multivariable model was calculated, internally validated and corrected for optimism by using 1000 bootstrap samples constructed with replacement of cases. The predictive variables were tested head-to-head for superiority by determining the difference in C-statistic with 95% CI in all bootstrap samples. The predictive value of QRS fragmentation within different subgroups was assessed in the multivariable analysis, which included all variables of the final risk model. The intraclass correlation coefficient (ICC) was used to determine interobserver variability in the observed number of leads with fQRS. Analyses were performed with SPSS V.21.0 (IBM, Armonk, New York, USA) or R V.3.1.0 (R Foundation for Statistical Computing, Vienna, Austria). A p value <0.05 was considered statistically significant.
Out of a total of 820 patients with TOF included in CONCOR within participating centres, an ECG could be obtained for 794 patients (97%). The median age was 27 years (IQR: 20–38), 438 (55%) were male, 247 (31%) had a previous shunt procedure and 773 (97%) had undergone surgical correction. Baseline patient characteristics are listed in table 1. ECG was obtained within 1 year of CONCOR inclusion in 89% of patients. The interobserver ICC of fQRS was 0.85 (95% CI 0.72 to 0.92). Patients with fQRS were older (median: 32 years vs 23 years, p<0.001) and had longer QRS duration (155 (SD: 24) ms vs 127 (SD: 26) ms, p<0.001) than patients without fQRS.
During a median follow-up duration of 10.7 years (IQR: 6.8–11.9) and a total of 7120 patient years, 46 patients died. Causes of death were: heart failure in 20 (43%), sudden cardiac in 10 (22%), perioperative in 1 (2%), vascular in 3 (7%), non-cardiac in 8 (17%) and unknown in 4 (9%). Deceased patients were older, more likely had a previous shunt procedure, had more ECG abnormalities and had many other complications prior to inclusion (table 1). Ten-year survival was 97.9% (95% CI 96.1% to 99.7%) in patients without fQRS (n=414), 92.5% (95% CI 88.9% to 96.1%) in patients with moderate fQRS (n=257) and 81.4% (95% CI 74.0 to 88.8) in patients with severe fQRS (n=123) (figure 2). Any fQRS (≥2 leads) and severe fQRS (≥5 leads) had a sensitivity of 87% and 46%, respectively, a specificity of 55% and 86%, respectively, a negative predictive value of 99% and 96%, respectively, and a positive predictive value of 11% and 17%, respectively, for death during follow-up. Compared with a sensitivity of only 28% for QRS >180 ms. The extent of fQRS was superior to QRS duration in predicting mortality (C-statistic 0.75 vs 0.66 (95% CI difference: 0.01 to 0.17)).
Independent predictors of mortality identified in multivariable analysis were (table 2): extent of fQRS (HR: 3.11, moderate vs none; HR: 5.84, severe vs none, HR: 2.24/class), p<0.001), age at inclusion (HR: 1.91/decade, 95% CI 1.53 to 2.37, p<0.001), previous shunt procedure (HR: 2.20, 95% CI 1.22 to 3.97, p=0.009) and previous pacemaker implantation (HR: 3.12, 95% CI 1.48 to 6.57, p=0.003). QRS duration was not predictive in the multivariable analysis (p=0.85). In a sensitivity analysis, excluding patients without surgical correction, similar variables remained predictive in multivariable analysis. A risk model was constructed in which 1 point was attributed for each class of fQRS (0–2 points), 1 for each decade older age (1 for age 30–40, 2 for age 40–50 and so on), 1 for previous shunt and 1 for pacemaker. The ten-year survival ranged from 100% in patients without any points (n=201) to 52% in patients with ≥6 points (n=19) (figure 3). The corrected C-statistic of the final risk model was 0.84 (95% CI 0.77 to 0.89).
The extent of fQRS remained predictive for all-cause mortality (HR: >1.7/class for all, p<0.05 for all) in the multivariable analysis within all 12 major subgroups (figure 4). Interaction terms between subgroups and fQRS extent were not significant (p>0.05 for all) (figure 4).
Clinical VAs occurred in 35 patients (13: symptomatic/recurrent non-sustained VT, 12: sustained VT and 10: aborted ventricular fibrillation or out-of-hospital cardiac arrest). In the multivariable analysis (see online supplementary table S1), extent of fQRS (HR: 2.00/class, 95% CI 1.26 to 3.16), atrial fibrillation on ECG (HR: 4.78, 95% CI 1.45 to 14.8) and QRS duration >180 ms (HR: 2.63, 95% CI 1.12 to 6.18) remained predictive for clinical VA.
This is the first study to demonstrate the prognostic value of QRS fragmentation in adult patients with TOF. The extent of fQRS strongly predicted both all-cause mortality and clinical VAs in multivariable models. The analysis of fQRS on a standard 12-lead ECG is easily performed, is inexpensive and may be implemented in risk stratification for adult patients with TOF.
The medical management of adults with TOF requires assessment and, if necessary, treatment of haemodynamic residual lesions such as pulmonary regurgitation.2 In addition, there remain concerns about sudden cardiac death, for which appropriate risk stratification strategies are needed.19 ,20 A potential substrate for life-threatening VAs is myocardial fibrosis, which may be caused by previous surgery and childhood cyanosis, and is proposed to negatively affect myocardial function.13–15 ,21 ,22 We hypothesise that fibrotic scars may serve as substrates for VT and cause fragmented QRS complexes by delayed or zigzag conduction, similar to causing late potentials that may be identified on signal-averaged ECGs in TOF.23 Considering RV outflow tract fibrosis is most typically found in adult patients with TOF, patients with RBBB may have the most suitable conduction to identify RV fibrosis by fQRS.14 ,15 However, we found QRS fragmentation remained predictive for mortality in the subgroups, such as patients without RBBB configuration. These findings confirm the prognostic value of QRS fragmentation in patients without complete RBBB, as previously found in patients with coronary artery disease.11 ,18
QRS duration >180 ms had a sensitivity of only 28%, was inferior to fQRS in predicting mortality and lost all predictive values after correcting for fQRS in the multivariable analysis. These findings indicate that patients with shorter, fragmented, QRS complexes have higher mortality risk compared with patients with longer, unfragmented, QRS complexes. Long-term prospective follow-up data were used to create a simple mortality risk score for adults with TOF. Importantly, the internally validated and corrected C-statistic (0.84) revealed an excellent predictive value and was comparable to the C-statistic (0.83) from a study by Valente et al,3 which also included cardiovascular magnetic resonance (CMR) imaging data to predict death or clinical VA but lacked information of fQRS. Our risk score can be used as an initial step to differentiate low-risk from high-risk patients. The high-risk patients could be referred for additional risk assessment (including more costly and invasive investigations such as gadolinium-enhanced CMR imaging, catheterisation and/or programmed ventricular stimulation) to identify those patients who should be referred for implantable cardioverter defibrillator implantation for primary prevention.3 ,9 ,10 ,14 ,15 ,19 ,21 ,24 In patients without fQRS or other major risk factors, patient counselling can be matched on the excellent prognosis and follow-up may be limited to periodic assessment of residual haemodynamic lesions. The risk score can also be easily applied in centres with limited availability of imaging modalities. Finally, the predictive value of fQRS was not restricted to all-cause mortality; a second multivariable analysis confirmed that fQRS also predicted potentially life-threatening VAs.
The findings of this study warrant future studies to explore the potential benefits of assessing fQRS in patients with TOF and other CHD. For instance, serial fQRS data could be linked with serial CMR-derived RV fibrosis data to determine whether changes in fQRS reflect progression of RV fibrosis. In addition, the use of fQRS to identify patients at risk for RV deterioration and heart failure and to select patients for interventions should be explored in future studies.
Our study used prospective data from the Dutch CONCOR registry. There was some variation in time between ECG and CONCOR inclusion. However, 97% of patients had a baseline ECG, which was performed within a year of inclusion in 89% of patients. All obtained ECGs were analysed retrospectively by a single observer blinded to clinical outcomes. Furthermore, CONCOR database did not acquire echocardiographic or CMR imaging data and signal-averaged ECGs were not performed. No data on RV myocardial fibrosis were collected. Finally, we did not investigate changes of fQRS during follow-up to assess whether fQRS can both mediate and reflect disease progression.
The fragmentation of QRS complexes is a simple, inexpensive parameter, which can be used in risk stratification for both all-cause mortality and VAs in adults with TOF.
What is already known on this subject?
Adults with tetralogy of Fallot are at risk for life-threatening arrhythmias and mortality. QRS duration is used in risk stratification despite limited sensitivity to predict these outcomes. The fragmentation of QRS complexes is related with myocardial fibrosis and may improve risk stratification.
What might this study add?
The extent of QRS fragmentation is superior to QRS duration in predicting all-cause mortality. In addition, this easily measured parameter is also a predictor of ventricular arrhythmias.
How might this impact on clinical practice?
Clinicians should assess QRS fragmentation on standard 12-lead ECGs. The extent of QRS fragmentation can guide risk stratification and help select high-risk patients for additional risk assessment (including more costly and invasive investigations).
The authors thank Lia Engelfriet and Sylvia Mantels for their dedicated work on CONCOR registry and Wybo Hoekstra for his technical assistance.
Contributors All authors contributed to the conception and design of the study, and critical revision and final approval of this manuscript. JPB analysed and interpreted the data and drafted the manuscript under supervision of senior authors BJMM and BJB.
Funding This work was supported by the Netherlands Heart Institute and Nuts Ohra foundation. The work described in this study was carried out in the context of the Parelsnoer Institute. The Parelsnoer Institute is part of and funded by the Netherlands Federation of University Medical Centres.
Competing interests None declared.
Ethics approval Academic Medical Center, Amsterdam.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data from this study are available to the authors of this manuscript. Unpublished data are available upon request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.