Summary
Background
Catheter ablation (CA) of persistent atrial fibrillation (AF) is challenging, and reported results are capable of improvement. A better patient selection for the procedure could enhance its success rate while avoiding the risks associated with ablation, especially for patients with low odds of favorable outcome. CA outcome can be predicted non-invasively by atrial fibrillatory wave (f-wave) amplitude, but previous works focused mostly on manual measures in single electrocardiogram (ECG) leads only.
Aim
To assess the long-term prediction ability of f-wave amplitude when computed in multiple ECG leads.
Methods
Sixty-two patients with persistent AF (52 men; mean age 61.5 ± 10.4 years) referred for CA were enrolled. A standard 1-minute 12-lead ECG was acquired before the ablation procedure for each patient. F-wave amplitudes in different ECG leads were computed by a non-invasive signal processing algorithm, and combined into a mutivariate prediction model based on logistic regression.
Results
During an average follow-up of 13.9 ± 8.3 months, 47 patients had no AF recurrence after ablation. A lead selection approach relying on the Wald index pointed to I, V1, V2 and V5 as the most relevant ECG leads to predict jointly CA outcome using f-wave amplitudes, reaching an area under the curve of 0.854, and improving on single-lead amplitude-based predictors.
Conclusion
Analysing the f-wave amplitude in several ECG leads simultaneously can significantly improve CA long-term outcome prediction in persistent AF compared with predictors based on single-lead measures.
Résumé
Contexte
L’ablation par cathéter de la fibrillation atriale (FA) persistante demeure un problème difficile et les résultats rapportés sont perfectibles. L’affinement de la sélection des patients pour cette procédure pourrait améliorer le taux de succès en évitant les risques liés à la procédure, notamment chez les patients à faible probabilité de réussite. Le suivi au décours d’une ablation par cathéter peut être prédit de façon non invasive par l’analyse de l’amplitude des ondes de FA mais des travaux précédents se sont focalisés principalement sur des mesures manuelles sur une seule dérivation électrocardiographique (ECG).
Objectifs
Ce travail évalue la prédiction à long terme du succès d’une ablation par cathéter d’une FA par analyse numérique de l’amplitude des ondes de FA sur des dérivations ECG multiples.
Méthode et résultats
Soixante-deux patients présentant une FA persistante ont été inclus, 52 hommes, d’âge moyen 61,5 ± 10,4 ans. Pendant un suivi moyen de 14 ± 8 mois, 47 patients n’ont pas présenté de récidive de FA au décours de cette procédure. Un ECG standard a été acquis avant la procédure chez chaque patient. L’amplitude des ondes de FA sur différentes dérivations ECG a été numérisée et un algorithme non invasif utilisé, permettant d’établir un modèle multivarié à partir d’une analyse par régression logistique. Une analyse basée sur l’index de Wald sur les dérivations D1, V1, V2 et V5 permet de prédire au mieux le succès d’une ablation par cathéter à partir de l’amplitude des ondes de FA, fournissant une surface sous la courbe ROC à 0,854, améliorant donc la performance d’une prédiction basée sur l’analyse d’une seule dérivation ECG.
Conclusion
L’analyse de l’amplitude des ondes de FA sur plusieurs dérivations ECG peut améliorer de façon significative la prédiction du succès d’une ablation par cathéter chez des patients souffrant d’une FA persistante.
Background
Atrial fibrillation (AF) is the most common sustained arrhythmia encountered in clinical practice . Radiofrequency catheter ablation (CA) of persistent AF is a well-established therapy, with proven efficacy in maintaining sinus rhythm during follow-up . Despite recent significant progress, CA for this form of AF yields less than perfect results, as it remains a costly, time-consuming intervention, with risk of complications for the patient. Hence, accurate selection of long-term responders to CA is crucial for improved patient-tailored management of this cardiac condition.
The patient characteristics that correlate most to CA outcome are unclear . The atrial fibrillatory waves (f-waves) observed in the surface electrocardiogram (ECG) reflect the electrical behaviour of the atria in a non-invasive fashion, and their analysis in time or frequency domains , or using more elaborate complexity indices , has been shown to correlate with CA outcome. In previous studies, f-wave amplitude has been measured manually in single leads separately (such as II, aVF or V1), thus neglecting information from the remaining leads that may be relevant for AF characterization. Indeed, the link between f-wave amplitude and long-term outcome has not been clearly established or has been demonstrated with limited accuracy only .
To overcome these drawbacks, the present study analysed whether the consideration of multiple ECG leads simultaneously could improve CA long-term outcome prediction based on automated f-wave amplitude measures.
Methods
Study population
All patients underwent radiofrequency ablation for persistent and long-standing persistent AF at Princess Grace Hospital (Monaco). The study was approved by the Institutional Committee on Human Research, and all patients gave written informed consent. Anti-arrhythmic drugs (except amiodarone) were withdrawn at least five half-lives before the study. Rate control drugs were interrupted just before CA. Before the procedure, all patients had echocardiographic assessment of the left atrium (LA; anteroposterior diameter in the parasternal long-axis view; two-dimensional surface in the apical four-chamber view) and the left ventricular ejection fraction by Simpson’s biplane method. A computed tomography scan acquisition of the LA was also performed for each patient before the procedure. The LA three-dimensional volume was calculated and reconstructed on the computed tomography scan.
Signal acquisition
For every patient, a 1-minute standard 12-lead ECG was recorded at a sampling rate of 977 Hz immediately before the start of ablation. ECG signals were acquired on a digital electrophysiological recording system (Prucka Engineering, Inc., Houston, TX, USA), including 0.05 to 40 Hz bandpass and 50 Hz notch filters. For patients who underwent repeat procedures, only the ECG recorded before the first intervention was considered in our prediction analysis.
Ablation procedure
The procedural approach for LA arrhythmia ablation in the study’s institution has been described elsewhere . In short, we performed coronary sinus catheterization using a decapolar diagnostic catheter, double transseptal puncture, systemic anticoagulation with heparin, with a target activated clotting time > 350 seconds, and electroanatomical mapping of the LA with the Carto system (Biosense-Webster Inc., Diamond Bar, CA, USA). Mapping and ablation catheters were inserted transseptally via a non-steerable (Fast-Cath SL1; St. Jude Medical, Minnetonka, MN, USA) or steerable (Agilis, St. Jude Medical; or V-Cas Deflect, Stereotaxis, St. Louis, MO, USA) sheath. A 20-pole circular mapping catheter (Lasso 2512; Biosense-Webster Inc.) was used to assess pulmonary vein (PV) potentials.
LA shell for anatomical definition was done with either standard electroanatomical (Carto XP; Biosense-Webster Inc.) or adjusted fast anatomical (Carto 3; Biosense-Webster Inc.) techniques. Image integration with the LA computed tomography scan reconstruction was always used. Detailed mapping of electrical activation of the LA (and, in selected cases, of the right atrium) was performed, with visual annotation of complex fractionated atrial electrograms (CFAEs) . Ablations were carried out in a stepwise manner, with endpoints of circumferential PV disconnection, ablation at CFAE sites and block across the lines (if performed). In all procedures, the operator systematically delivered point-by-point ablation lesions for at least 30 seconds to create a contiguous antral circumferential line around the PV pairs. Catheter–tissue contact was optimized before each radiofrequency delivery, using catheter motion on fluoroscopy, near-field electrogram (EGM) stability, impedance drop during radiofrequency delivery and morphological EGM changes suggestive of lesion creation . Whenever available, contact force was also taken into account to optimize catheter–tissue contact. If a significant lesion (based on local EGM modification) was not obtained, reablation at the same site, with further optimization of contact (at times requiring the use of a steerable sheath) and energy increase, was performed. Irrigated radiofrequency was delivered with a Stockert 70 generator (Biosense-Webster Inc.), a 42 °C limiting temperature and 30–40 W for the endocardial part of the line. Baseline irrigation flow was 17 to 30 mL/min, with an increase to 60 mL in case of excessive heating. If PV isolation (entry block) was not obtained at the end of the circular lesion, the lines were remapped and the gaps reablated. If needed, further lesions guided by the circular catheter were delivered.
In case of ongoing AF after PV isolation, additional lesions targeting fractionated EGMs in the LA as well as the roof and, in a few cases, the left isthmus lines (with endpoints of block across the line) were performed. Further lesions were delivered, in selected cases, within the coronary sinus (20 W) and, in some cases without AF termination, in the right atrium, targeting fractionated EGMs.
AF termination during ablation was defined as sinus rhythm resumption or its change to a stable atrial tachyarrhythmia. Nevertheless, this was not a procedural endpoint, and operators ended the procedures after PV isolation, ablation of the annotated CFAE sites and, when appropriate, block across the lines. Associated atrial tachyarrhythmia ablation was performed, and the critical isthmus or focal origin was specifically targeted up to sinus rhythm resumption and reconfirmation of PV isolation before catheter withdrawal.
In cases without AF or atrial tachyarrhythmia termination after catheter withdrawal, a loading dose of amiodarone (30 mg/kg) was administered, unless contraindicated. An electrical cardioversion (150 to 200 J, repeated up to three times under general anaesthesia) was performed if the arrhythmia was ongoing 24 to 48 hours afterwards.
Follow-up
After the 3-month blanking period recommended by current guidelines , patients were followed for clinical and asymptomatic recurrences. Follow-up was performed in a “real-life” setting, by regular visits to the treating cardiologist, with repeated ECG and 24-hour Holter monitoring in all cases (every 3 months during the first year after ablation; every 6 months afterwards). Supplementary documentation by ECG or Holter was sought in case of recurring symptoms suggestive of arrhythmia. Any recurring, sustained (> 30 seconds), symptomatic AF or flutter was considered for a repeat procedure. Absence of any AF recurrence during follow-up defined the CA success group of our study, while patients with documented AF recurrences after the last procedure constituted the CA failure group.
Signal-processing and statistical analysis
F-wave amplitude computation
The signal-processing algorithm used to compute the f-wave amplitude in each lead is illustrated in Fig. 1 . First, ECG fiducial points were detected to properly segment TQ intervals, where atrial activity can be measured free from QRST complexes of ventricular interference ( Fig. 1 , top). R-wave time instants were located on lead V1 by applying the Pan-Tompkins algorithm , and Q-wave onsets were simply obtained by subtracting 40 ms, a typical ventricular activation time. From the lead where the most prominent T-waves could be visually identified, T-wave offsets were estimated by an adapted Woody’s method ; then, the segmented intervals were mean centred and concatenated ( Fig. 1 , middle). In the concatenated TQ segments, the local maxima were detected, and an upper envelope was estimated by interpolation; the lower envelope was estimated in an analogous manner from the local minima ( Fig. 1 , bottom). Finally, the average difference between both envelopes along time was computed as an estimate of the f-wave mean amplitude in the lead examined. The mathematical details of this amplitude measurement algorithm have been described by Meo et al. .
Of the six frontal leads, only two provide linearly independent voltages . Hence, to avoid redundancy, only leads I, II and V1–V6 were considered in subsequent analyses. Amplitude computation was performed using MATLAB, version 2011a (MathWorks, Natick, MA, USA).
For each patient, the f-wave amplitude used for prediction was computed in every lead, using the available duration of atrial activity signal after TQ segment concatenation. To validate the temporal stability of this measurement method, the f-wave mean amplitude was also computed on initial 10-second and 30-second segments of atrial activity in the recordings, where such lengths were available, and Pearson correlation coefficients were then determined for every lead.
Univariate analysis
Distribution normality was first checked for the variables under examination by the Kolmogorov–Smirnov test. Levene’s correction was applied when homoscedasticity (homogeneity of variance) could not be assumed. Under data normality, groups were compared by a parametric Student’s t -test, whereas a non-parametric Mann–Whitney U -test was used when the variables did not show a normal distribution. Proportion analysis was based on the χ 2 test. For each univariate predictor of CA outcome, receiver operating characteristic (ROC) curves were computed to find the cut-off point providing the optimal trade-off between sensitivity and specificity, and the area under the ROC curve (AUC) was used as a prediction performance index. This statistical analysis and the logistic regression (LR) model described next were performed with SPSS software, version 13.0 (IBM, Armonk, NY, USA).
Multivariate analysis
An LR model was constructed from a linear combination of f-wave amplitudes computed in the ECG leads, acting as multiple predictor variables. Optimal linear combination coefficients were determined by maximum likelihood estimation from the available dataset . After estimating the model coefficients, the LR score was computed from the f-wave amplitude set of each patient. The numerical value of the LR score is directly related to the estimated odds of CA success (not a voltage), and was then used as a CA outcome predictor. Using the LR score, ROC-based indices were derived to quantify the multivariate model predictive performance, as in the univariate analysis above. A backward elimination technique based on the Wald index was employed to select the ECG leads whose f-wave amplitudes contributed to the LR prediction score in a statistically significant manner.
Results
Study population
Sixty-two consecutive patients (52 men; mean age 61.5 ± 10.4 years) were included in the study. Patient characteristics are summarized in Table 1 . Patients had a mean AF history of 61.5 ± 56.1 months. AF was persistent in 54 patients (87.1%) and long-standing persistent in eight patients (12.9%). Duration of the actual AF episode (ongoing at the time of CA) was 7.3 ± 11.1 months. AF was idiopathic in 26 patients (41.9%).
| Patient characteristics | Overall ( n = 62) | CA success ( n = 47; 75.8%) | CA failure ( n = 15; 24.2%) | P |
|---|---|---|---|---|
| Men | 52 (83.9) | 40 (85.1) | 12 (80.0) | 0.64 a |
| Age (years) | 61.5 ± 10.4 | 60.3 ± 10.4 | 65.1 ± 9.7 | 0.12 b |
| Hypertension | 22 (35.5) | 19 (40.4) | 3 (20.0) | 0.15 a |
| Sleep apnoea syndrome | 5 (8.1) | 4 (8.5) | 1 (6.7) | 0.82 a |
| Diabetes | 8 (12.9) | 8 (17.0) | 0 (0) | 0.09 a |
| Body mass index | 27.6 ± 4.4 | 27.6 ± 4.6 | 27.8 ± 3.7 | 0.90 b |
| Obesity | 14 (22.6) | 10 (21.3) | 4 (26.7) | 0.66 a |
| Coronary heart disease | 7 (11.3) | 7 (14.9) | 0 (0) | 0.11 a |
| Hypertensive cardiomyopathy | 6 (9.7) | 2 (4.3) | 4 (26.7) | 0.011 a,d |
| Dilated cardiomyopathy | 8 (12.9) | 4 (8.5) | 4 (26.7) | 0.07 a |
| Valvular heart disease | 4 (6.5) | 2 (4.3) | 2 (13.3) | 0.21 a |
| Hypertrophic cardiomyopathy | 2 (3.2) | 2 (4.3) | 0 (0) | 0.42 a |
| AF history (months) | 61.5 ± 56.1 | 54.7 ± 53.5 | 87.0 ± 61.0 | 0.047 c,d |
| Current AF episode | 7.3 ± 11.1 | 6.8 ± 12.0 | 8.8 ± 7.3 | 0.06 c |
| LA anteroposterior diameter (mm) | 47.2 ± 7.1 | 47.1 ± 7.3 | 47.5 ± 6.4 | 0.86 b |
| LA surface (cm 2 ) | 29.1 ± 5.9 | 28.4 ± 5.8 | 31.5 ± 5.5 | 0.09 b |
| LVEF (%) | 59.4 ± 15.6 | 61.5 ± 13.5 | 53.1 ± 19.7 | 0.07 b |
| LA CT scan maximal volume (mL) | 141.0 ± 44.9 | 135.7 ± 42.7 | 158.6 ± 49.1 | 0.11 b |
| Patients on amiodarone before CA | 26 (41.9) | 19 (40.4) | 7 (46.7) | 0.67 a |
| Patients with repeat procedure | 5 (8.1) | 4 (8.5) | 1 (6.7) | 0.82 a |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
