Impact of Complete Versus Incomplete Circumferential Lines Around the Pulmonary Veins During Catheter Ablation of Paroxysmal Atrial Fibrillation: Results From the Gap-Atrial Fibrillation–German Atrial Fibrillation Competence Network 1 Trial

Patients with symptomatic paroxysmal AF refractory to antiarrhythmic drug treatment (≥1 antiarrhythmic drug other than β-blockers) were eligible for inclusion in the study; they had to have no structural heart disease other than arterial hypertension. In all patients, an ECG documentation of at least 1 AF episode had to be available in combination with an average number of at least 1 episode per month. Patient age had to range between 50 and 85 years because of requirements of the German Federal Office for Radiation Protection (Bundesamt für Strahlenschutz). Patients were excluded from the study if they had previous PV ablation procedures, AF secondary to a reversible cause, known presence of intracardiac or other thrombi, evidence of obstructive lung disease requiring bronchodilator therapy, other medical conditions (ie, cancer and congestive heart failure) that may cause the patient to be noncompliant with the protocol, confound the data interpretation or be associated with limited life expectancy (ie, <1 year), bleeding diathesis or suspected procoagulant state, or contraindication to anticoagulation therapy.

Patients were randomly assigned to either incomplete or complete linear PVI to investigate the significance of complete linear PVI on the patients’ outcome. The primary hypothesis of the study was that complete linear PVI is superior to incomplete linear PVI with regard to the recurrence of AF. As a secondary hypothesis, the noninferiority of the complete linear PVI strategy was to be tested. Randomization was performed with the use of an automated central randomization system (Marvin).

After exclusion of the presence of left atrial thrombi by transesophageal echocardiography, invasive mapping and ablation were performed in a fasting state and under continuous sedation or as per institutional standard. After positioning of a multipolar catheter in the distal coronary sinus and transseptal puncture of the interatrial septum, intravenous heparin was administered by bolus injection to adjust the activated clotting time to 250 to 300 s throughout the procedure. A sequential reconstruction of the left atrium was performed using a 3D mapping system (CARTO; Biosense Webster, Inc, Diamond Bar, CA, or NavX; St. Jude Medical Inc, Minneapolis, MN) after initial PV angiography. Circumferential mapping catheters were placed in both upper and lower PVs of the respective side, to allow simultaneous recording of PV potentials throughout the entire ablation process. Subsequently, linear lesions were deployed to encircle the antra of the right (septal) and left (lateral) PVs in pairs attempting isolation by a single encircling line. Catheter ablation was performed using an irrigated radiofrequency current ablation catheter (maximum 40 W overall and maximum 30 W at the posterior wall).

All procedural data including documentation of the achieved end points in both groups were assessed by a separate core laboratory. Two independent experienced electrophysiologists evaluated the available recordings with regard to achievement of the attempted end point per protocol and the end point indicated by the respective investigator.

Oral anticoagulation was recommended for at least 3 months after ablation, regardless of the ablation success. All patients continued their last (ineffective) antiarrhythmic drug treatment throughout the hospital stay. All antiarrhythmic drugs were stopped before discharge. Other concomitant medication such as antihypertensive treatment and agents solely aiming for rate control were permitted (eg, digitalis, beta blockers, and calcium channel blockers).

Three months after the ablation procedures, all patients were scheduled to undergo an invasive reevaluation to investigate the presence of PVI irrespective of the presence or absence of potentially AF-related symptoms. After single transseptal puncture, PV conduction was assessed using a circumferential mapping catheter sequentially positioned in all PVs. Repeat ablation was permitted in cases of documented AF recurrence or when the patient’s symptoms seemed to be intolerable, most likely because of AF. At this time, investigators were not aware of the transtelephonic ECG (TTECG) analysis.

Additional substrate modification (linear lesions) or focal ablation was permitted in cases of AF recurrence despite isolation of all PVs. The initial findings of the repeat electrophysiology study and the achieved end point afterpotential reablation were also investigated by the core laboratory (Figure 1B).

All patients were provided with a TTECG allowing daily ECG recordings and transmission via telephone (RhythmCard; Instromedix, San Diego, CA). A total duration of 30 s can be recorded and transmitted for analysis to the Institut für Klinisch-Kardiovaskuläre Forschung (Institute for Clinical Cardiovascular Research) in Munich, Germany. Details of transmission and data analysis have been described previously.²² The TTECG was provided for 3 months after the index ablation, and patients were advised to record at least 1 ECG per day irrespective of any symptoms. The automated system asked the patient whether he/she had symptoms at the moment of the recording. In addition, patients were expected to transmit whenever they experienced symptoms potentially related to AF recurrence and were advised to subsequently have a 12-lead conventional ECG recording to correlate with the tele-ECG.

The primary end point was the time to first recurrence of symptomatic AF lasting >30 s on TTECG monitoring or the detection of asymptomatic AF defined as 2 consecutive recordings of AF during a minimum of 72 hours. Secondary end points were the time to first occurrence of any documented relapse of AF (including 12-lead ECG or Holter recording), the number of days sinus rhythm was documented in the patients’ TTECG recordings, and the number of hospitalizations because of AF. Furthermore, the number of visits without hospitalization and the number of SAEs of special interest (including death, resuscitation, syncope, stroke, transient ischemic events, prolonged reversible ischemic neurological deficits, major bleeding, tamponade, atrioesophageal fistula, and PV stenosis or occlusion) were predefined secondary end points. In addition, procedural complications were considered secondary end points.

The Web-based electronic data capture system Marvin was managed by the Institut für Klinisch-Kardiovaskuläre Forschung, which also served as Contract Research Organization on behalf of AFNET. The primary route of data was via the internet. The data cleaning process included automated range checks with change requests displayed immediately to the user (plausibility checks). A 2-step query process included automated queries related to single data fields (edit checks) and manually created queries. Data entry was finalized by an electronic signature. The software was managed to comply with data protection requirements and security standards (Food and Drug Administration requirements of part 11 of title 21 of the Code of federal regulation [21 CFR part 11]). All site information was confidential, and transmitted data were encrypted with a secure socket layer.

Predefined SAEs of special interest as defined above were recorded by the local investigators at regular follow-up visits. The information was entered into the Web-based database. Documentation of the events was sent to the SAE center in Brandenburg, Germany, where it was verified for completeness (eg, laboratory results, transesophageal echocardiography, and ECG). All SAEs were reviewed according to predefined criteria by a Critical Events Committee, consisting of 2 cardiologists and 1 neurologist, and assessed toward their definite, probable, possible, improbable, or no relation with AF or therapeutic interventions for AF.

The study investigated the hypothesis that complete linear PVI (group B) is superior to incomplete PVI (group A) with regard to the time to AF recurrence as defined above. Assumptions were based on results obtained from published data.⁵ Under the assumption of recurrence (event) rates of 30% in group B and 55% in group A 3 months after randomization, at least 2×86=172 patients were required to detect a difference between Kaplan–Meier curves by use of the log-rank test with a power of 90% if the type I error rate is fixed to 5%. The possibility of a sample size adaptation in an open interim analysis was taken into account (use of the 2-sided Freedman formula²³).

In addition to the sample size needed, an early drop-out rate of 10% in each arm was anticipated. Thus, at least 192 patients were planned to be randomized. An interim analysis was provided for sample size adaptation while the study was ongoing.

As prespecified for the primary end point, ≤2 ordered analyses were to be performed. The first (superiority) analysis was performed in the intention-to-treat population and consisted of the 2-sided log-rank test at an α level of 0.05, accompanied by Kaplan–Meier–based estimates of the 3-month event rates in the 2 groups. For the difference (ie, noncomplete minus complete linear PVI) of the 3-month event-free survival estimates, an approximate 95% confidence interval was calculated. The test of superiority was significant if the confidence interval for the difference fell completely above zero.

If the first analysis was not significant, a second (noninferiority) analysis was to be performed in the treatment-per-protocol population; it consisted of the calculation of an approximate 95% confidence interval for the difference in the 3-month survival estimates. Noninferiority was assumed if the difference between the event-free survival rates was <12% in favor of control (incomplete linear PVI). Because the test procedure is closed, the overall α level was 5%.

With respect to all other comparisons, continuous variables were compared using Student t test or Mann–Whitney U test, categorical variables using χ² likelihood-ratio tests, Jonckhere–Terpstra tests (if categories were ordered), or log-rank tests (censored data); the corresponding P values are given. Statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC).

Before the end of recruitment, the steering committee performed a sample size adaptation after a blinded interim analysis. After this analysis, a slight over-recruitment was tolerated because a higher drop-out rate was observed in the first part of the trial. A total of 233 patients were eventually enrolled, with 116 randomly assigned to incomplete PVI (group A) and 117 assigned to complete PVI (group B). Baseline characteristics are given for both groups in Table 1. There were only minor differences between the 2 groups with respect to sex and congestive heart failure/hypertension/age ≥75 years/diabetes mellitus/prior stroke or transient ischemic attack or thromboembolism (CHADS₂) score 0 or 3.

At the end of the ablation procedure, 11 group A patients (9.5%) retained complete PVI despite randomization to incomplete PVI; complete PVI could not be achieved in only 1 group B patient (0.9%).

The first analysis on the primary end point was performed in the intention-to-treat population for superiority. One hundred and sixty one patients reached the primary end point. Of these, 136 (84.5%) patients were symptomatic and 25 (15.5%) asymptomatic.

The superiority of complete PVI over incomplete PVI could be shown (Figure 2; 90-day event rates of 79.2% in group A and 62.2% in group B, difference 17.1% [95% CI, 5.3%–28.9%]; P<0.001). Because the superiority of complete PVI could be shown, the noninferiority analysis was not required and not performed. A correction of the P value for sample size adaptation was not required because the interim analysis was blinded.

The time free of any documented relapse of AF (symptomatic or asymptomatic episodes) was also significantly shorter in group A than in group B (P<0.001), as depicted in Figure 3. The number of days sinus rhythm was found in the TTECG recordings was significantly higher after complete PVI (median, 14.5 [interquartile range, 2–60] days versus 3.0 [1–15] days; P<0.001). There was no significant difference with respect to the number of hospitalizations for cardiac reasons (group A: median, 0 [0–0]; range 0–2 and group B: median, 0 [0–0]; range 0–4; P=0.62) and with regard to visits without hospitalization (A: median, 1 [0–2]; range, 0–20 and B: median, 1 [0–1]; range, 0–4; P=0.30).

SAEs occurred in 6 patients allocated to group A and 12 patients allocated to group B (P=0.15; Table 2). The overall fluoroscopy time during the initial ablation procedure was not significantly different between the groups (A: 33±19 minutes versus B: 37±21 minutes; P=0.13). However, procedure duration was shorter in group A (184±56 minutes) than in group B (213±78 minutes; P=0.0018).

Because a total of 37 patients, 13 of 116 (11%) group A and 24 of 117 (21%) group B, eventually withdrew their consent given at enrollment to undergo an electrophysiological restudy at 3 months, results were available from 103 group A patients and 93 group B patients (Figure 4). The reason for consent withdrawal given by 11 of the 13 group A patients and 21 of the 24 group B patients was absence of palpitations.

In the 103 group A patients, at least 1 gap was identified in 92 (89%); of the other 11 patients who showed no gap but completely isolated PVs, 6 already had PVI at the end of the ablation procedure despite randomization to incomplete ablation. Of the 92 patients with conduction gaps, AF recurred in 74 (80%) but not in the remaining 18 patients; in the 11 patients with isolated PVs at 3 months AF had also recurred in 9 (82%).

Of the 93 group B patients, conduction gaps were present at 3 months in 65 (70%), with persistent isolation of all PVs detected in only 28 (30%). AF had recurred in not only 40 (62%) of the 65 patients with conduction gaps at 3 months but also 16 (57%) of the 28 patients with persistent PVI. The difference in AF recurrence rates between group A and B patients with conduction gaps (80% [74/92] versus 62% [40/65], respectively) was statistically significant (P=0.0111). Conversely, freedom from AF recurrence in patients without conduction gaps at 3 months was statistically not different between groups A and B (18% [2/11] versus 43% [12/28], respectively; P=0.27).

A per-protocol analysis of group B patients with durable PVI (n=28) versus group A patients without PVI (n=92) at 3 months revealed a significantly higher AF recurrence rate in the latter patients (80% [74/92] versus 57% [16/28], respectively; P=0.023).

Repeat ablation during the 3-month repeat study in 142 of the 196 patients was performed significantly more often in group A patients (87/103 [84%]) than in group B patients (55/93 [59%]; P=0.0001). A total of 40 (28%) of the 142 patients underwent repeat ablation on the basis of the patients reporting symptoms that were later not documented as AF recurrence.

Complications encountered during the 3-month repeat studies were 1 major bleeding in the 54 patients without repeat ablation, and 1 tamponade, 2 major bleedings, and 1 PV stenosis in the 142 patients with repeat ablation.

A total of 202 (86.7%; 101 group A and 101 group B) of the 233 index procedures were evaluated by the core laboratory. In the remaining 13.3%, assessment by the core laboratory was not possible because of insufficient documentation of the respective ablation site by intracardiac ECG recording. Despite assignment to group A, complete PVI was present in 11 (9.9%) of 101 patients. On the contrary, in 1 of 101 patients, complete PVI was not achieved despite randomization to group B.

Gap-AF–AFNET 1 is the first prospective, randomized, multicenter study to prove that

Although the first and the second main findings may not be surprising, they do add scientific evidence from a multicenter randomized trial to current expert opinion.^8,9 The third finding is surprising. A 70% PV reconduction rate after 3 months in patients with acute complete PVI is extremely high. But it confirms data from previous observational studies showing that patients with AF recurrences almost always have conduction gaps and indicates the challenge of creating durable lesions by sequential radiofrequency current ablation.²⁴

The correlation between PV reconduction and AF recurrence underlines the important role of the PVs as shown in multiple observational studies^12–19 and in an experimental study.²⁵ In vitro data from optical mapping of a 2×2 cm² area of the PVs in isolated perfused dog hearts showed conduction slowing at the proximal part of the PVs, repolarization heterogeneity with the longest action potential duration at the PV endocardium when compared with the PV epicardium, and sustained focal discharge from the endocardial surface when stimulated with isoproterenol. In addition, nonsustained reentry that became sustained with isoproterenol could be initiated in 50%, with the complete reentrant loop within the 2×2 cm² mapped area.

The high PV reconduction rate also clearly indicates how limited irrigated radiofrequency current lesion formation still is and that we can only expect better results if durable contiguous and transmural lesions can be achieved. AF recurrences were frequent in patients with isolated veins at the time of restudy, namely 57% in group B and 82% in group A. This may indicate that other, most likely transient factors such as inflammation still play a role in this rather in the early phase after catheter ablation. Furthermore, the follow-up in this study, which required patients to transmit a TTECG at least once per day, was more intense than in almost all previous catheter ablation studies.

Recently, it has been shown that conduction gaps can be significantly reduced if RFC catheter ablation is guided by contact force,^26,27 and that a reduction in the number of gaps is associated with a reduction in AF recurrences.²⁸ Most recently, adenosine challenge after acute PVI leading to acute reisolation by radiofrequency current has been shown to be superior to no acute reisolation.²⁹ A reduction in the number of gaps may also be achieved if energy sources other than RFC are used for ablation. The laser balloon technology has been shown to create durable lesions with a low gap rate at 3 months of only 15%,³⁰ and the second generation of the cryoballoon also seems to reduce reconduction.³¹ Whether this translates into a higher clinical success rate needs further investigation in randomized clinical trials.

In this study, the gap rate at 3 months was 89% in patients randomized to incomplete circumferential lesions and thus significantly higher than in patients randomized to complete PVI (70%). This difference in gap rate explains the different clinical outcomes in the study and supports the hypothesis that gaps play a dominant role in AF recurrence after PVI.

The statistically significant difference in AF recurrence rates between group A and B patients with conduction gaps at 3 months (80% versus 62%, respectively) may be because of a different number of conduction gaps present after 3 months and different time courses in the development of conduction gaps. Because 70% of patients with initially complete PVI develop at least 1 conduction gap within 3 months, one may assume that the same mechanism is in effect in patients with 1 initial gap left intentionally, leading necessarily to a higher number of conduction gaps in the latter patients. Unfortunately, the actual number of conduction gaps at 3 months was not documented in this study. A single gap (in group A patients) versus no gap (in group B patients) present at the beginning of follow-up may justify the assumption that AF through the conduction gap recurs earlier in the former patients because the latter patients first had to wait for a conduction gap to develop.

According to the study protocol, the electrophysiology study after 3 months terminated the follow-up because the Ethical Board allowed conduction gaps to be closed only in symptomatic but not in asymptomatic patients. Because investigators were not aware of the TTECG recordings at that time, the decision to reablate was only guided by clinical information, namely recurrence of palpitations, provided by the patients or by palpitation-induced ECG documentation. In other words, 28% of patients without AF recurrences documented later underwent repeat ablation. This indicates, as shown in previous studies, that symptoms after AF ablation may not correlate with AF recurrence and that lack of symptoms does not exclude asymptomatic episodes of AF.²⁰

The complication rate of catheter ablation did not differ between the 2 groups. This is not unexpected because—according to the protocol—circumferential ablation around the PVs was performed also in group A patients and ablation was only stopped when PVI occurred, to allow recovery of PV reconduction. This occurred in all but 11 patients according to the core laboratory assessment. The overall complication rate of 7.7% does not differ from similar studies in patients undergoing PVI for paroxysmal AF.³²

Tamponade and bleedings at the puncture sites were the most common complications. No death or atrioesophageal fistula occurred in any patient. Stroke occurred in 1 group A patient, and PV stenosis in 1 group B patient.

The study results presented in this article were obtained only within 3 months. This time has generally been defined as the blanking period in which AF recurrences are not counted.^8,9 However, because reablation was allowed and intended at the time of the restudy, further follow-up would have been affected by the results of reablation. For the first time, this study systematically reassessed all ablated AF patients after 3 months by an electrophysiology study. First, we think that patients, in particular asymptomatic patients, would not have come back in sizeable numbers for an invasive study after a longer period of time of being asymptomatic, even if they had given informed consent. In the present study, 11% of group A patients and 21% of group B patients did not show up for the protocol-mandated 3-month repeat study. Furthermore, this study tested a hypothesis, namely, whether complete isolation of the PVs is superior to incomplete isolation of the PVs. Thus, the fact that there was a significant difference between the 2 study groups is more relevant than the actual number of AF recurrences. It is clear that acute PVI cannot be transferred into durable PVI, particularly not at the time when the study was performed. However, regardless of the actual AF recurrence rate, which may be debatable and needs further improvement, there is a significant difference among the 2 groups, which confirms that the strategy of complete PV isolation is superior to just PV ablation with 1 or multiple gaps.

This randomized, prospective, multicenter trial shows for the first time that isolation of the PVs is superior, with respect to AF recurrence within 3 months, to ablation around the PVs leaving 1 conduction gap. Nevertheless, the reconduction rate is high in patients with initially complete PVI and urgently requires improvement. Clearly, the functional end point of acute PVI needs to be supplanted in future studies by a morphological end point of lesion assessment with evidence of transmural necrosis.