Author + information
- Received June 6, 2016
- Revision received August 11, 2016
- Accepted August 18, 2016
- Published online September 28, 2016.
- Daniel Cortez, MDa,b,c,d,
- Waseem Barham, MDc,
- Emily Ruckdeschel, MDe,
- Nandita Sharma, MDd,
- Anthony C. McCanta, MDf,
- Johannes von Alvensleben, MDa,b,
- William H. Sauer, MDb,
- Kathryn K. Collins, MDa,b,
- Joseph Kay, MDa,b,
- Sonali Patel, MDa,b and
- Duy T. Nguyen, MDb,∗ ()
- aDepartment of Cardiology, Children’s Hospital of Colorado, Aurora, Colorado
- bDepartment of Cardiology, University of Colorado, Aurora, Colorado
- cDivision of Cardiovascular Sciences, Lund University, Lund, Sweden
- dDepartment of Electrophysiology, Penn State Milton S. Hershey Medical Center, Hershey, Pennsylvania
- eDepartment of Adult and Pediatric Cardiology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania
- fDepartment of Cardiology, CHOC Children’s, Orange, California
- ↵∗Reprint requests and correspondence:
Dr. Duy T. Nguyen, University of Colorado, Denver, Anschutz Medical Campus, 12401 East 17th Avenue, B-132, Aurora, Colorado 80045.
Objectives This study sought to test the hypothesis that a vectorcardiographic parameter, the QRS vector magnitude (QRSVm), can risk stratify those patients at risk for sustained spontaneous ventricular arrhythmias (VAs) or ventricular arrhythmia inducibility (VAI) in a large cohort of patients with tetralogy of Fallot (TOF).
Background Patients with TOF have an increased risk of VAs, but predicting those at risk can often be challenging.
Methods Blinded retrospective analyses of 177 TOF patients undergoing pulmonary valve replacement (PVR) between 1997 and 2015 were performed. VAI was evaluated by programmed electrical stimulation in 48 patients. QRS intervals and QRSVm voltage measurements were assessed from resting 12-lead electrocardiograms, and risk of VA was determined. Clinical characteristics, including imaging and cardiac catheterizations, were used for other modality comparisons.
Results Sustained spontaneous VA occurred in 12 patients and inducible VA in 18 patients. Age and QRSVm were significant univariate predictors of VA. QRSVm was the only independent predictor of VAI (p < 0.001). Using a root mean square QRS value of 1.24 mV, the positive and negative predictive values were 47.9% and 97.8%, respectively, for spontaneous sustained VA. For VAI, using a QRSVm cutoff of 1.31 mV, positive and negative predictive values were 63.0% and 95.3%, respectively.
Conclusions In TOF patients undergoing PVR, older age was associated with increased spontaneous VA risk. Lower QRSVm predicted spontaneous VA or VAI risk with high negative predictive values. QRSVm is the only independent predictor of VAI. These clinical features may help further risk stratify TOF patients requiring therapies to prevent sudden death.
Patients with tetralogy of Fallot (TOF) have a significant burden of arrhythmias post-operatively, reported to be as high as 43.3% in some series (1). The clinical history remains important but may be insufficient for predicting the risk of ventricular arrhythmias (VAs). Invasive risk stratification via programmed electrical stimulation (PES) in TOF patients has been shown to have diagnostic and prognostic value (2). Noninvasive measures of the right ventricle (RV), such as QRS duration (QRSd), QRS fragmentation, increased RV volumes, as well as right ventricular ejection fraction (RVEF) and left ventricular ejection fraction (LVEF), have demonstrated association with increased risk of arrhythmias (3–7).
Vectorcardiographic principles provide additional clinical information to the 12-lead electrocardiogram (ECG) and have yielded further diagnostic (8–10) and prognostic (10–15) value to the ECG in the traditional 12-lead configuration. In a small cohort of adult TOF patients, increased risk of sustained spontaneous VA and inducible ventricular arrhythmia has been demonstrated by a measure of QRS dispersion called the QRS vector magnitude (QRSVm, or magnitude of the 3-dimensional QRS vector), with improved predictive value over QRSd or spatial QRS-T angle. The predictive value of QRSVm was independent of magnetic resonance imaging (MRI) RV volume, gadolinium enhancement, or hemodynamics measured via cardiac catheterization (16).
QRSVm denotes the magnitude of the maximum 3-dimensional QRS vector and is calculated as the root mean square of the QRS wave in 3-dimentional space. It is calculated from the equation:
We hypothesized that QRSVm, based on a sinus rhythm ECG recorded just before a TOF patient undergoes pulmonary valve replacement (PVR), will predict risk of spontaneous VA as well as risk of inducible VA during PES in a large cohort of TOF patients. We chose to evaluate QRSVm on ECGs at the time of PVR because of several clinical considerations: 1) the risk of VA and ventricular arrhythmia inducibility (VAI) is high during this time; 2) other clinical parameters (cardiac catheterizations and MRIs) are usually measured during this period so that correlations can be made; and 3) predicting VA or VAI risks before PVR may allow for changes in management at the time of PVR, such as cryoablation during open surgery or consideration of further risk stratification such as electrophysiological (EP) studies around the time of PVR.
Patients and Methods
This study was approved by the institutional review board at the University of Colorado.
A blinded retrospective analysis was performed on available electrocardiograms and associated imaging data from 362 TOF patients from 1977 to 2015 at the University of Colorado Hospital systems (including the Children's Hospital of Colorado). This group included 177 TOF patients who were undergoing PVR via transcatheter insertion utilizing Melody valves (Medtronic, Minneapolis, Minnesota) or via cardiac surgery. Patients were excluded if they did not undergo a procedure for PVR or if they did not have an interpretable ECG with adequate baseline measurement recorded within the 6 months before PVR. Patients also were excluded if a diagnosis of TOF or TOF-like physiology was not certain or if they had left-sided obstruction at the time of PVR. ECG assessment near the time of PVR was chosen because cryoablation during this period might be able to prevent recurrence of arrhythmia (7).
Patients who had sustained spontaneous VA burden (seen by ECG, Holter, exercise stress test, pacemaker, or telemetry monitoring) were identified. Sustained spontaneous VA was defined as ≥30 s of VA or a VA that was associated with hemodynamic instability. For these patients, the arrhythmia had to have occurred within the 6 months before or 6 months after PVR, and their last sinus rhythm ECG before arrhythmia identification/treatment and before PVR was used for evaluation. Thus, at the time of ECG assessment, no patients were taking ion-channel inhibiting medications, had not undergone ablations, and had no new medical devices placed.
Comparisons were made between those patients with spontaneous sustained VA versus all patients without VA, including those undergoing PES (N = 177); for those with either VA or induced sustained VA (VAI) versus no VA or VAI; and for only those undergoing PES (n = 48), comparisons were made between those who had inducible sustained VA (VAI) versus those without VAI.
EP Studies and ECGs
Some patients had undergone EP studies before PVR. Inducibility was evaluated from EP studies using PES, as previously described (2). Briefly, PES was performed at 2 drive trains and at 2 sites, with up to triple extrastimuli and decrementing to ventricular effective refractory period or 180 ms. If a patient was noninducible, the protocol was repeated on isoproterenol. Inducibility was defined as any VA (monomorphic VA, polymorphic VA, or ventricular fibrillation [VF]) that lasted ≥30 seconds or was hemodynamically unstable requiring pace termination or defibrillation.
Sinus rhythm ECGs (GE, Milwaukee, Wisconsin, or Phillips, Andover, Massachusetts) were recorded at 25 mm/s speed with 10 mm/mV for limb and precordial leads. ECGs were analyzed within 6 months of PVR. Measurements of QRSVm (Figure 1) and QRSd intervals were performed. QRSVm was performed blinded by 1 author (D.C.) on the full-scale view (10 mm/mV) PDF file of the ECG using electronic calipers. Intraobserver and interobserver variability as described in the Statistical Analysis section were measured by 2 authors (D.C., N.S.). QRSd was reported on the ECG device. Pre-operative and post-operative differences were assessed based on the pre-operative ECGs and the post-operative ECGs recorded within 1 month of PVR.
MRI and Cardiac Catheterizations
MRI data within 6 months of the patient’s PVR were included in data analysis. MRI measurements of left ventricular and RV volumes were indexed to body surface area. Presence of gadolinium enhancement, as well as RVEF, LVEF, and pulmonary regurgitant fractions, also were assessed. Hemodynamic measurements from cardiac catheterization were included if they were recorded within 6 months of PVR. Cardiac catheterization data included left-sided phasic and mean pressures as well as oxygen saturation data.
Data were assessed for normality using Shapiro-Wilk testing. Non-normally distributed continuous data are presented as median and interquartile range (IQR) (first and third quartiles). Normally distributed data are presented as mean ± SD. Student t test, Mann-Whitney U test, and contingency table testing were used to identify significant differences between groups. A value of p < 0.05 was considered significant. Odds ratios (ORs) were calculated to estimate risk for parameters identified as significantly different by comparative analysis.
Univariate analysis was performed to identify predictors associated with VAs or VAI. Multivariate analysis was performed using logistic regression analysis to identify independent risk factors for VAs or VAI. Selection of variables for inclusion in the multivariate analysis was made using a stepwise approach (p removal = 0.10).
Receiver-operating characteristic (ROC) curves were generated to determine optimal cutoff values for variables, particularly QRSVm, in the prediction of VAs during the follow-up period. The curve point with the highest sum of sensitivity and specificity was labeled as the optimal cutoff point and was used in OR, sensitivity, and specificity analyses.
Pearson and Spearman correlation coefficients were used as appropriate for parametric and nonparametric data. Intraobserver and interobserver variability were estimated by intraclass correlation coefficients based on a 10% sample of the population. Repeatability was performed by 2 of the authors (D.C., N.S.). Data analysis was performed using SPSS (IBM, Chicago, Illinois).
One hundred seventy-seven patients met the inclusion criteria (median age 13 years; IQR: 8 to 29 years). The cohort was 52.0% male. Nineteen Melody valves were inserted percutaneously, and 158 open-heart surgeries were performed to replace the pulmonary valves. Twelve patients (6.8%) had documented sustained VAs. Forty-eight patients (27.1%) had undergone PES studies before PVR. Of those patients, 18 (37.5%) had sustained VAI (12 polymorphic ventricular tachycardia (VT)/VF and 6 monomorphic VT without difference in parameters between these subgroups). Sixty-seven MRIs and 70 cardiac catheterizations were performed before PVR.
Risk of Spontaneous VAs (Including Patients Undergoing PES Testing)
Age and QRSVm were significantly different in patients with spontaneous sustained VA (n = 12) compared to all patients without spontaneous VA, including those patients undergoing PES, regardless of whether they had VAI or did not have VAI (n = 165) (Table 1). QRSVm for those patients with spontaneous VA was 1.05 mV (IQR: 0.81 to 1.18 mV), whereas all other patients had a median QRSVm of 1.62 mV (IQR: 1.23 to 2.28 mV). Using a QRSVm cutoff of 1.24 mV (Figure 2A), the positive and negative predictive values were 42.9% and 97.8%, respectively (OR: 33.7; 95% confidence interval [CI]: 8.1 to 141.6). Area under the ROC curve was 0.81 (0.70 to 0.91) (Figure 3). Although age and QRSVm were significant univariate predictors, with multivariate analysis no parameters were significant predictors (Table 2). Syncope may have been an independent predictor of VA; however, it was not included in univariate or multivariate analyses because of the limited number of patients with syncope thus underpowered comparisons.
Risk of Spontaneous VA or VAI
Only age and QRSVm significantly differentiated TOF patients with VA or VAI (n = 30) from those without VA or VAI (n = 30). QRSVm significantly differentiated patients with spontaneous VA (median 1.05 mV; IQR: 0.79 to 1.24 mV) compared to those without VA (median 1.50 mV; IQR: 1.10 to 1.80 mV), with p < 0.001 (Table 3, Figure 2B). The area under the ROC curve was 0.83 (95% CI: 0.73 to 0.92). The positive and negative predictive values, using a cutoff of 1.24 mV, were 84.9% and 92.6%, respectively. The area under the ROC curve was 0.83 (95% CI: 0.73 to 0.92). Patients with a QRSVm of 1.24 mV had OR OF 70.0 for spontaneous VA or VAI (95% CI: 12.5 to 393.4). Age significantly differentiated VA (median age 34.0 years; IQR: 24.3 to 53.0 years) from no VA (median age 17.0 years; IQR: 13.0 to 22.0 years; p = 0.002). QRSVm remained the only independent predictor of VA or VAI (p = 0.011) on multivariate analysis (Table 4).
Risk of Inducible VA
QRSVm was the only parameter that significantly differentiated (p < 0.001) patients with VAI (0.98 mV; IQR: 0.76 to 1.24 mV) from those without sustained VAI (1.51 mV; IQR: 1.1 to 1.8 mV). For QRSVm, the positive and negative predictive values, using a cutoff of 1.31 mV (Figure 2C), were 63.0% and 95.2%, respectively, with OR of 34.0 for those with VAI (95% CI: 3.9 to 293.3). Age, history of syncope, QRSd, MRI volumes, gadolinium enhancement, and catheter-measured hemodynamics were not significantly different between those with VAI and those without VAI (Table 5). Of those with VAI, there was no difference in characteristics between those who had monomorphic VT versus those who had polymorphic VT (Table 6). QRSVm remained an independent predictor on multivariate analysis when gender (p = 0.061) and RVEF (p = 0.101) were taken into account (p = 0.005) (Table 7).
Correlations With Other Measurements
QRSVm had significant correlation coefficients when compared to age, QRSd, and left ventricular end-diastolic volume (LVEDV), with r values of –0.420, –0.190, and 0.280, respectfully. QRSd had significant correlation coefficients when compared to age, spatial QRS-T angle, QRSVm, and RVEF at 0.480, 0.350, –0.190, and –0.490, respectively. Repeat PVR was associated with lower RVEF and lower LVEF, with correlation coefficients of –0.40 and –0.30, respectively.
Higher LVEDV was inversely associated with age and right ventricular end-diastolic volume (RVEDV), with correlation coefficients of –0.33 and –0.20, respectfully.
RVEF significantly correlated with repeat PVR, QRSd, RVEDV, and LVEF, with correlation coefficients of –0.370, –0.490, –0.490, and 0.580, respectfully. Left ventricular end-diastolic pressure (LVEDP) significantly correlated with age and right ventricular end-diastolic pressure (RVEDP), with correlation coefficients of 0.420 and 0.530, respectively.
Repeatability of Measures Tested and Change Over Time
Intraclass correlation coefficients for QRSVm for interobserver and intraobserver variability were 0.94 and 0.92, respectively, for 10% of the sample.
For patients without sustained VA, the difference in QRSVm was a median of 0.15 mV (IQR: –0.06 to 0.37 mV) before PVR versus 0.12 mV (IQR: –0.09 to 0.25 mV) after PVR. There was no significant difference for QRSvm between values taken before PVR and QRSvm values measured within 1 month after PVR.
Risk of Spontaneous VA
Of the clinical characteristics in this TOF cohort, only age and QRSVm were univariate predictors with significant ORs for predicting spontaneous sustained VA. On multivariate analyses, no single parameter was a significant predictor for spontaneous VA when all patients, including those undergoing PES, were included in the comparison. However, there was a trend toward significance for LVEDV, age, and QRSVm.
RVEDV by MRI was not helpful in predicting risk of VA. QRSd has had equivocal results in terms of predicting VA in published reports. Similar to previous findings, we found significant correlations between RV volumes and QRSd (1,3). Given the younger mean age of patients in our study, this may be one reason why QRSd and RV volumes were not independent predictors for VA compared to results of previous studies with older patient populations (4,17).
A decrease in the peak magnitude of the QRS complex (as measured by QRSVm) is a measure of increased dispersion of depolarization or less unidirectional absolute depolarization, which may be a potential cause of increased arrhythmogenic risk (14). By measuring a derived 3-dimensional voltage, one is assessing overall summative depolarizing vector forces in the X-, Y-, and Z-directions. As summative vectors begin to disperse in different directions, the overall absolute magnitude in 1 particular direction decreases. Thus, increased dispersion of depolarization, as measured in our study by a lower QRSVm, was found to be associated with an increased risk of spontaneous VA. Similar findings in signal-averaged ECG measurements have been demonstrated in a smaller cohort (17). QRSVm has excellent negative predictor value. If a patient had QRSVm > 1.24 mV, his or her risk for spontaneous VA was very low, regardless of whether the patient had an inducible VA on PES.
In our study, gadolinium enhancement in the RV (including around the outflow tract conduit or patch and outside of the ventricular septal patch) was not associated with an increased risk of VAs. Older age was associated with an arrhythmogenic risk, as previously reported (1,18). Increased LVEDV, although not statistically significant, approached significance (univariate p = 0.053, multivariate p = 0.058). Given the trend toward a significant difference in LVEDV, likely ventricular-to-ventricular interactions likely play a role in patients with TOF and may be related to their arrhythmia burden (19).
As with other arrhythmic syndromes, a history of syncope should raise concern and prompt treatment and evaluation. However, given the low numbers of patients with syncope, we were not able to analyze this parameter in univariate or multivariate analyses. In previous studies, syncope has been shown to be an independent predictor of VAs (1).
Risk of Inducible VA
QRSVm was the only independent predictor in either the VAI versus no inducibility comparison group as well as the VA or inducibility versus no arrhythmia or inducibility comparison group. This relationship has been previously reported (14). Induced VAs included polymorphic and monomorphic VT and VF. There were no differences in QRSVm among the arrhythmic subgroups, but the number of patients was small. The risk of VAI or spontaneous arrhythmia in these groups was independent of QRSd, hemodynamics, RV volume, or presence of gadolinium enhancement. RVEF was not a predictor of VAI, unlike for other diseases of the RV for which RVEF has been found to be a risk predictor (20).
Correlations of Parameters Measured
Correlation coefficients showed QRSVm to be inversely proportional to age, QRSd, and LVEDV (which was inversely proportional to RVEDV). QRSd showed a significant positive correlation with age, spatial QRS-T angle, and decreased RVEF. It was inversely associated with QRSVm and RV volume. Increasing age was significantly correlated with increased QRSd, decreased QRSVm, RVEF, and LVEF, and increased LVEDP. These may be scar-mediated, as there was no direct correlation with RV volume. Thus, as the patient population became older, they had wider QRSd, lower QRSVm (thus greater time and voltage dispersion of depolarization), lower RVEF/LVEF, and increased LVEDP, likely due to impaired filling via RV–LV interactions. These changes each predispose the patient to an increased arrhythmia risk (5). QRSd was inversely correlated to RVEF and positively correlated to RVEDP; thus, as conduction time increases, fractional output from the RV decreases and end-diastolic pressure remains higher. This finding is similar to that previously reported by Koestenberger et al. (21). In addition, the investigators found a significant positive correlation with QRSd and MRI-measured RV volume. However, our cohort differs from that reported by Koestenberger et al. (21) in that our study population included adults.
Of interest, of all the parameters tested, the highest correlation coefficients were found for LVEDP with RVEDP (0.513) and LVEF with RVEF (0.580). These findings likely indicate ventricular-to-ventricular interactions given the significant and highly positive correlations (18). There were no significant correlations for either RVEDP or LVEDP with QRSVm, indicating that end-diastolic pressures may not be associated with voltage dispersion of depolarization.
The retrospective nature of this study is a limitation. Furthermore, not all patients with post-repair TOF were evaluated; only those undergoing PVR were studied. MRI and catheterization data were not available for all patients. ECG at one time point may not be reflective of a dynamic disease process. Temporal changes in vectorcardiographic parameters may impact the association with VA risk over longer periods of follow-up. Other limitations include the fact that first and repeat PVR patients were included; that not all patients had undergone VAI testing; that signal-averaged ECG, heart rate variability, and QRS fragmentation were not measured; and that both transcatheter and surgical PVR patients were included, although no significant differences were found between these 2 cohorts of patients.
Patients with TOF undergoing PVR are at greater risk for VA burden at older ages. Dispersion of depolarization, as measured by QRSVm, was the best ECG predictor of VA burden with a high negative predictive value. Furthermore, it was the only independent predictor of VAI. QRSVm may help noninvasively risk stratify patients with TOF undergoing PVR.
COMPETENCY IN MEDICAL KNOWLEDGE: Patients with TOF have an increased risk of VAs. With current clinical criteria, identifying those at risk for VAs can be challenging. QRSVm is a vectorcardiographic measure of ventricular dispersion of depolarization that can be used to further risk stratify those patients at risk for VAs, particularly those undergoing PVR.
TRANSLATIONAL OUTLOOK: QRSVm is a vectorcardiographic measure of ventricular dispersion of depolarization. Abnormal ventricular dispersion of depolarization may be associated with an increased risk of VAs. By assessing QRSVm, those patients at highest risk for VAs may be identified, but larger prospective studies in other cohort populations are needed before this concept can be translated to clinical practice.
Drs. Sauer and Nguyen receive significant research grants from Biosense Webster and CardioNXT; and educational grants from St. Jude Medical, Boston Scientific, and Medtronic. Dr. Nguyen has a provisional patent on partially insulated focused catheter ablation; and has non–public equity interests/stock options in CardioNXT. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- interquartile range
- left ventricular end-diastolic pressure
- left ventricular end-diastolic volume
- left ventricular ejection fraction
- magnetic resonance imaging
- odds ratio
- programmed electrical stimulation
- pulmonary valve replacement
- QRS duration
- QRS vector magnitude
- receiver-operating characteristic
- right ventricle
- right ventricular end-diastolic pressure
- right ventricular end-diastolic volume
- right ventricular ejection fraction
- tetralogy of Fallot
- ventricular arrhythmia
- ventricular arrhythmia inducibility
- ventricular fibrillation
- ventricular tachycardia
- Received June 6, 2016.
- Revision received August 11, 2016.
- Accepted August 18, 2016.
- American College of Cardiology Foundation
- Khairy P.,
- Aboulhosn J.,
- Gurvitz M.,
- et al.
- Khairy P.,
- Landzberg M.,
- Gatzoulis M.,
- et al.
- Bassareo P.P.,
- Mercuro G.
- Valente A.M.,
- Gauvreau K.,
- Assenza G.E.,
- et al.
- Udink ten Cate F.E.,
- Sreeram N.,
- Brokcmeier K.
- Sabate Rotes A.,
- Connolly H.M.,
- Warnes C.A.,
- et al.
- Triola B.,
- Olson M.B.,
- Reis S.E.,
- et al.
- Cortez D.,
- Sharma N.,
- Cavanaugh J.,
- et al.
- Borleffs C.J.,
- Scherptong R.W.,
- Man S.C.,
- et al.
- Kardys I.,
- Kors J.A.,
- van der Meer I.M.,
- Hofman A.,
- van der Kuip D.A.,
- Witteman J.C.
- Cortez D.,
- Ruckdeschel E.,
- McCanta A.C.,
- et al.
- Bokma J.P.,
- Winter M.M.,
- Oosterhof T.,
- et al.
- Davlouros P.A.,
- Kilner P.J.,
- Hornung T.S.,
- et al.
- Turrini P.,
- Angelini A.,
- Thiene G.,
- et al.