This is an ambispective study including all consecutive patients with rTOF who were referred for an electrophysiological study (EPS) and an RV EAM at a single tertiary referral center from July 2006 to July 2017. Most patients (n=50) were recruited prospectively at the time of EPS (November 2012 to July 2017); additionally, 6 patients were recruited retrospectively (March 2006 to October 2012). Following our standard clinical protocol, indications for EPS and RV EAM in rTOF patients were documented spontaneous VT, clinical suspicion of VT, and risk stratification (with potential ablation) prior to pulmonary valve replacement. The study followed the national and international guidelines (2013 Declaration of Helsinki) and was approved by the Ethics Committee of the Vall d’Hebron Research Institut (VHIR). All patients provided specific informed consent.

Clinical, echocardiographic, cardiac magnetic resonance, baseline (nonpaced) electrocardiogram, EPS and EAM data were collected. Information on nonsustained ventricular tachycardia (NSVT) or high-frequency of premature ventricular contractions (Hi-PVC) (> 30 per hour)11 was retrieved from electrocardiogram recordings, 24-hour Holter monitoring or implantable cardiac defibrillator interrogation.

Antiarrhythmic drugs were stopped 5 half-lives before the procedure. EPS were performed under superficial conscious sedation and through femoral vein access. Atrioventricular conduction was evaluated at baseline. Programmed electrical stimulation for VT inducibility was performed at 3 paced cycle lengths (600/500/400ms) followed by up to 3 extra stimuli down to the refractory period or 200ms, delivered at the RV apex and the infundibular septum, before and during isoproterenol infusion (2-10μg/min). Programmed electrical stimulation was considered positive if sustained VT (monomorphic or polymorphic) or ventricular fibrillation were induced.

RV 3D-EAM was performed during sinus rhythm, using the QRS as reference, with Ensite Velocity (n=11), Ensite Precision (n=4) (Abbott), Carto XP (n=4) or Carto 3 (n=37) (Biosense Webster). Mapping catheters used were NaviStar (n=41) (Biosense Webster), Cool-Flex (n=14) and Tacticath Quartz (n=1) (Abbott). Electrogram characteristics (peak-to-peak bipolar amplitude and local activation time, defined as sharp peak deflection of the local bipolar electrograms that coincides with the maximum downstroke of the local unipolar signal) at each contact point were recorded and displayed color-coded, for voltage and activation sequence over the RV shell (figure 1). Points were acquired in stable sites and adequate contact with a 15-mm threshold setting. According to previous literature,12 electrograms with a peak-to-peak voltage <1.5mV were defined as low voltage, and areas with at least 3 adjacent points with peak-to-peak voltage <0.5mV were considered scar. Scar areas in the whole RV and RV outflow tract (RVOT) were measured and the percentage of scar was calculated with respect to the total area of the RV. Anatomical isthmi with a voltage amplitude > 0.5 mV between scar areas and the valvular annulus were quantified (figure 1). Fragmented electrograms (FEGM), defined as those with ≥ 4 deflections,> 40ms duration and <1mV voltage,13 and double potentials (DP), defined as those with 2 components separated by an isoelectric line, were labeled (figure 1). To achieve uniform localization of FEGM and DP, the RV was segmented into 7 anatomical areas (RV septum, RV anterior wall, RV posterior wall, ventricular septum defect, septal RVOT, anterior RVOT, and posterior RVOT), and the number of areas with FEGM and DP was calculated. Total RV activation time, defined as the time elapsed between the earliest and the latest electrograms, and infundibular activation time, defined as the time between the septal and lateral edge of the ventricular septal defect were determined.

Figure 1.

A: right ventricle outflow tract scar area (in red) on the voltage map of a patient repaired with a transannular patch. B: posterior view of a right ventricle voltage map showing an area of scar at the VSD. Double potentials and fragmented electrograms are labeled in the map with blue and pink dots respectively. C: activation map of a patient with a right ventricle activation time of 108ms. VSD, ventricular septal defect.

(0.42MB).

Variables are reported as mean±SD, median [p25-p75] or frequency (%) where appropriate. Differences between groups were compared using the Fisher exact test, Student t-test or Mann-Whitney U-test where appropriate. The primary endpoint was the presence of spontaneous or induced VA at any time. For better clinical applicability, a secondary endpoint evaluating exclusively the presence of spontaneous VA was also defined. Variables associated with both endpoints were explored using logistic regression analyses. Continuous variables were categorized or stratified for simplification. RV activation time strata were set every 10ms from the entire number closest to the minimum value (60ms). Those showing an association with a P <.1 or thought to be clinically relevant were included in the multivariable model. Several models were tested to select the most parsimonious model containing the smallest number of variables that maximized the discrimination ability assessed by the area under the receiver operating characteristic curve (AUC). Significance for all analyses was set at the level of .05. Statistical analyses were performed with IBM SPSS 20.0 software and RStudio.

A total of 56 patients with rTOF were included. Baseline characteristics of the study population are summarized in table 1. A ventricular approach was used in all reparative surgeries. Thirty-eight patients underwent implantation of a transannular patch to expand the RVOT, 3 a RV-to-pulmonary artery conduit, and 15 a ventriculotomy alone. At the time of EPS, 8 patients had already undergone pulmonary valve replacement for severe pulmonary regurgitation and 5 patients a previous ineffective focal ablation. Thirty-nine patients (69%) were symptomatic for palpitations (n=24), presyncope (n=8) or syncope (n=7). Overall, the functional status assessed by the treating clinician was good (NYHA class I in 40 patients, II in 14, and III-IV in 2). Imaging studies confirmed that almost 80% of patients had severe pulmonary regurgitation, although, in the vast majority, left ventricular and RV systolic function were preserved or only mildly depressed.

Details of EPS and EAM parameters obtained within the same procedure are shown in table 2. At the time of the EAM, a total of 17 sustained monomorphic VT were induced in 14 patients with a mean cycle length of 291.9±68.5ms (from 200.0 to 410.0ms). No episodes of polymorphic VT or ventricular fibrillation were induced. A pathological infra-Hisian conduction was documented in 21 patients (38%).

High-density EAM (mean 469 points) revealed that all patients had a RV scar area embracing a mean of 8% of the RV tissue (scar extension could not be calculated in 3 cases due to technical reasons). In most patients, scar areas were located in the regions affected by the reparative surgery (RVOT, ventricular septal defect, and RV anterior wall). Other locations of scar (inferior wall and other septum locations) were infrequent (11%). All patients showed at least 1 anatomical isthmus, with a mean of 2.5 isthmi per patient.

Mean RV activation time and infundibular activation time were 124.1±35.7ms and 51.5±31.8ms, respectively. RV activation time was correlated with both anatomical and functional parameters of the RV (figure 2), exhibiting a positive correlation with scar size (r=0.455, P=.001) and a negative correlation with RV ejection fraction (r=-0.475, P <.001). Likewise, scar size was similarly correlated with the HV interval (r=0.307, P=.027, figure 2).

Figure 2.

Scatter plots showing the correlation between electrophysiological, anatomical and functional parameters of the right ventricular (RV).

(0.17MB).

The prevalence of DP and FEGM for the overall population was high (71% and 79%, respectively) and only 3 patients did not have pathological electrograms. For the overall population, DP and FEGM were found in a mean of 0.9±0.8 and 1.3±1.0 anatomical areas, most frequently at RVOT and surrounding ventricular septal defect.

Twenty-one out of the 56 patients met the primary endpoint of spontaneous or induced VA: 11 had clinical VT (one of whom also had an episode of recovered sudden cardiac death), and in 10 additional patients, with no prior VA, a sustained VT was inducible in the EPS (figure 3). Four of the 11 patients with clinical VT were also inducible at the time of the EAM, and 2 additional patients had been inducible in a previous EPS. An implantable cardiac defibrillator had been implanted prior to EPS in 5 of the patients. The remaining 35 patients comprised the group without VA.

Figure 3.

Study groups. EAM, electroanatomical mapping; rTOF, repaired tetralogy of Fallot; RV, right ventricle; VT, ventricular tachycardia. *At any electrophysiological study.

(0.19MB).

Univariate analysis (table 1) showed that patients with VA were older at the time of surgical repair (median age: 7 vs 3 years, P=.047) and had a higher prevalence of Hi-PVC/NSVT (56% vs 24%, P=.032). RV systolic function measured by cardiac magnetic resonance tended to be lower in the group with VA, although RV dimensions were similar between groups.

No significant differences were observed between the 2 groups regarding age at the time of evaluation, time since intervention, left ventricular function, history of palliative surgery, use of transannular patch, arrhythmia-related symptoms, and baseline electrocardiogram parameters.

Infra-Hisian conduction time was significantly longer in the group of patients with VA (HV interval 60.4±13.8 vs 49.9±9.0ms, P=.001), with a higher proportion of patients with infra-Hisian conduction disturbance (67% vs 21%, P=.001, table 2). RV and RVOT scar extensions were similar in the 2 groups. RV activation time was significantly longer in the group with VA (140.2±41.5 vs 114.8±22.8ms, P=.014), although no significant differences between groups were observed in infundibular activation time (52.74±41.23 vs 50.76±35.51, P=.543). Patients with VA had a significantly higher prevalence of DP (95% vs 57%, P=.002) and a higher number of anatomical areas with DP (1.24±0.7 vs 0.71±0.75, P=.013). However, no differences were observed in the prevalence of FEGM and the mean number of areas with FEGM (table 2).

From those variables showing an association with the primary endpoint in the univariable analysis, the most parsimonious multivariate model was built using the minimum number of variables with the highest discrimination ability. The final model included the following variables (table 3): an HV interval> 55ms (OR, 21.20; 95%CI, 3.11-144.14; P=.002); the presence of Hi-PVC/NSVT (OR, 11.33; 95%CI, 1.50-85.97; P=.019); and RV activation time (OR [per 10ms interval], 1.34; 95%CI, 1.02-1.75; P=.035). A model including this information had a discrimination ability, assessed by the AUC, of 0.884 (95%CI, 0.79-0.97; P <.001). Other factors such as age at surgical repair, the presence of DP and the number of areas with DP, associated with VA in the univariate analysis, were not confirmed to be independently associated in the multivariate analysis.

As a secondary endpoint, and to provide more clinically relevant guidance, we also evaluated the variables potentially associated exclusively with clinical VA (n=11). Of note, VT inducibility in this case was considered as one of the predictors of interest. The results of the univariate analysis are shown in the table 1 of the supplementary data. In the multivariate model (table 3), only Hi-PVC/NSVT (OR, 13.14; 95%CI, 1.95-88.54; P=.008), and an HV interval> 55ms (OR, 9.65; 95%CI, 1.41-66.14; P=.021) proved to be independently associated with spontaneous VT. The model including these variables had a discrimination power, evaluated by the AUC, of 0.836 (95%CI, 0.663-1.000; P=.002).

Our sample size was small. However, the study provides preliminary data on potential novel parameters associated with VA in patients with rTOF with good discrimination ability, setting the basis for further research, most likely through multicenter studies including larger patient populations. The present study included a selected population of rTOF patients referred for EPS with documented or suspected VT, or prior to pulmonary valve replacement. Therefore, the applicability of our results to the overall population of patients with rTOF should be evaluated in future specific studies. By the time the EPS was performed, 5 patients had undergone a previous ineffective focal ablation. These procedures were performed without navigation and mapping systems and, on the other hand, the approaches were limited to focal ablation that was clinically ineffective and, we believe, had no impact on the RV activation time. Importantly, a sensitivity analysis excluding these patients showed no major changes in the logistic regression models (see Tables 2 and 3 of the supplementary data).