Author + information
- Received September 1, 2016
- Revision received January 5, 2017
- Accepted January 12, 2017
- Published online March 29, 2017.
- Takatsugu Kajiyama, MD∗ (, )
- Shinsuke Miyazaki, MD,
- Junji Matsuda, MD,
- Tomonori Watanabe, MD,
- Takayuki Niida, MD,
- Takamitsu Takagi, MD,
- Hiroaki Nakamura, MD,
- Hiroshi Taniguchi, MD,
- Hitoshi Hachiya, MD and
- Yoshito Iesaka, MD
- ↵∗Address for correspondence:
Dr. Takatsugu Kajiyama, Cardiology Division, Cardiovascular Center, Tsuchiura Kyodo Hospital, 4-1-1 Otsuno, Tsuchiura, Ibaraki 300-0028, Japan.
Objectives This study sought to identify anatomic parameters predicting procedural difficulty in achieving pulmonary vein isolation (PVI) and in-procedural predictors of successful applications during second-generation cryoballoon (CB) ablation.
Background PV anatomies vary and influence the procedural difficulty during CB PVI.
Methods In total, 408 initial freezes among 110 patients undergoing PVI for paroxysmal atrial fibrillation using 28-mm second-generation CBs with single 3-min freeze techniques were included. The anatomic parameters were obtained from pre-procedural cardiac computed tomography. The nadir balloon temperature and temperature at the start of the plateau phase were recorded during each freeze.
Results Acute PVI was achieved by initial applications in 335 pulmonary veins (PVs) (82.1%) and touch-up was required in 13 (3.2%). A multivariate analysis revealed that a thinner left lateral ridge (<4.7 mm), higher ovality (>50.5%), and longer PV ostium-bifurcation distance (>26.1 mm) required multiple applications for a successful left superior PVI. Older age (>68 years), and shorter PV ostium-bifurcation distance (<10.4 mm) required multiple applications for a successful right superior and right inferior PVI, respectively. Shorter PVTLs were also associated with requiring touch-up of the RIPV. Balloon temperatures were lower for successful than failed PVI applications. Successful PVIs were predicted using the nadir balloon temperature at 33.0 ± 2.6 s, 33.0 ± 2.5 s, 33.6 ± 2.5 s, and 33.0 ± 2.5 s from the initiation of freezes with positive predictive values of 87.7%, 88.5%, 98.5%, and 81.6% using cutoff temperatures of -34°C, -33°C, -37°C, and -33°C in the left superior, left inferior, right superior, and right inferior PVs, respectively.
Conclusions The anatomic information might predict procedural difficulty and the balloon temperature a successful PVI during the early CB ablation freezing phase.
Pulmonary vein (PV) isolation (PVI) is a firmly evidenced therapeutic intervention for paroxysmal atrial fibrillation (AF) (1–3). Although radiofrequency catheter ablation has been the standard technology for achieving PVI for decades, PVI based on cryoballoon (CB) technology is an emerging and spreading alternative because of the less complicated technique, shorter procedure time, and higher durability of the PVI when compared with radiofrequency ablation (4,5). Indeed, a large, prospective, randomized study clarified the noninferiority in terms of the efficacy and safety of PVI by a CB when compared with radiofrequency (6). The recently introduced second-generation CB has exhibited a significantly higher performance than the first-generation CB owing to the improved cooling effect (7).
In the CB ablation procedure, the pulmonary vein (PV) anatomy has a definite influence on the procedural difficulty, because the circumferential contact of the CB with the PV ostium is essential for a complete vein occlusion to achieve an acute PVI and for the durability (8). Although the device is designed to achieve PVI ideally with a single shot, due to the high degree of PV anatomic variability among patients, an optimal occlusion may hardly be achieved for any given individual PV, especially the right inferior pulmonary vein (RIPV) (4–6,8–10). In this context, it is expected that pre-procedural imaging of the left atrium (LA) and PVs might be helpful to predict procedural difficulty. In contrast, the in-procedural parameters predicting a successful PVI aid in avoiding prolonged unnecessary applications, especially if real-time PV potential monitoring is indeterminate by distally placed electrodes. Currently, some data (9,10) are available regarding the anatomic and in-procedural predictors for a successful PVI in first-generation CB procedures, but not in second-generation CB procedures. Consequently, we carried out this study to identify the anatomic parameters predicting the procedural difficulty in achieving PVI and the association between the balloon temperature during freezing and a successful PVI during 28-mm second-generation CB ablation.
We enrolled 110 patients with paroxysmal AF who underwent their first PVI using second-generation CBs after obtaining pre-procedural cardiac enhanced multidetector computed tomography (MDCT) scanning at our institute. Patients with a left common PV were not included. PVI was performed with a single 3-min freeze technique, without a routine bonus application, using exclusively large (28-mm) CBs (8,11). AF was classified according to the latest guidelines (3). All patients gave written informed consent. The study protocol was approved by the hospital’s institutional review board. The study complied with the Declaration of Helsinki.
All antiarrhythmic drugs were discontinued for at least 5 half-lives before the procedure. The surface electrocardiogram and bipolar intracardiac electrograms were continuously monitored and stored on a computer-based digital recording system (LabSystem PRO, Bard Electrophysiology, Lowell, Massachusetts). The bipolar electrograms were filtered from 30 to 500 Hz. A 7-F 20-pole 3-site mapping catheter was inserted through the right jugular vein for pacing, recording, and internal cardioversion.
The procedure was performed under conscious sedation obtained with dexmedetomidine. Heparin (100 IU/kg body weight) was administered immediately after venous access, and heparinized saline was additionally infused to maintain the activated clotting times at 250 to 350 s. A single transseptal puncture was performed using a radiofrequency needle (Baylis Medical., Montreal, Quebec, Canada) and 8-F sheath (SL0, AF Division, SJM, Minneapolis, Minnesota). The transseptal sheath was exchanged over a guidewire for a 15-F steerable sheath (Flexcath Advance, Medtronic, Dublin, Ireland). A 20-mm circular mapping catheter (Lasso, Biosense-Webster, Diamond Bar, California) was used for mapping all PVs before and after the cryoablation to confirm electrical isolation. A spiral mapping catheter (Achieve, Medtronic) was used to advance the CB into the PV for support and mapping the PV potentials. After sealing at the PV antrum, a complete occlusion was confirmed by injecting contrast medium. No 23-mm CBs were used. This was followed by a freeze cycle of 180 s. No additional applications were performed after the isolation. To avoid bilateral phrenic nerve injury, all CB applications were applied while monitoring the ipsilateral diaphragmatic compound motor action potentials during phrenic nerve pacing (11). When the balloon nadir temperatures exceeded –60°C, the ablation was terminated (12). Ablation was also terminated upon a significant reduction in the maximal diaphragmatic compound motor action potential amplitude from baseline, using a double-stop technique. The procedural endpoint was defined as electrical PVI verified by the 20-mm circular mapping catheter. Additional touch-up freezes with an 8-mm tip cryocatheter (Freezor MAX, Medtronic) were performed with 2-min applications each. To analyze the technical difficulty in isolating each PV by a CB, the number needed to disconnect (NND) the PV was defined as the total number of freezes required to achieve an acute PVI.
Anatomic parameters obtained by MDCT
All patients underwent MDCT before the intervention to obtain the contour of the LA cavity. The slice data of the image was reconstructed into a 3-dimensional volume rendering by computer software (Synapse Vincent, Fujifilm, Tokyo). The 4 major PVs, consisting of the right superior pulmonary vein (RSPV), RIPV, left superior pulmonary vein (LSPV), and left inferior pulmonary vein (LIPV), were identified. The right middle lobe branch was always regarded as a branch of the RSPV or RIPV. The following anatomic parameters were measured on the volume rendering: area of the pulmonary vein ostium (PVOA), length of the PV trunk from the ostium to the bifurcation (PVTL), diameter of the medial branch of the PV, distance between bilateral carinas, thickness of the left lateral ridge (LLRT) between the left PVs and left atrial appendage, distance between the interatrial septum and RIPV ostium, and length of the left mitral isthmus (Figure 1). The ovality index of the PV ostia was also calculated using the formula: 2 × (major diameter – minor diameter)/(major diameter + minor diameter) (9).
Analysis of the balloon thermal dynamics
The builtin thermosensor of the CB catheter automatically measured the temperature of nitroperoxide gas circulated through the balloon at the exit point of the balloon. The alteration of the temperature every second during the balloon inflation was recorded automatically and stored in the console of the cryoablation system. To avoid any confounding effects of repeated applications, only the initial freeze at each PV was analyzed. The lowest balloon temperature (temperature at the starting point of slow cooling phase [Tm]) and temperature at the transitional time point between the rapid and slow cooling phase (nadir balloon temperature [Ti]) were examined. The Ti was defined as the earliest time point at which the temperature change was <1°C/s averaged over a 3-s period (Figure 2).
All statistical analyses were performed using R version 3.2.2 software (R Foundation for Statistical Computing, Vienna, Austria). Continuous variables were reported as mean ± SD and were compared using the Student t test. Differences between proportions were compared using Fisher's exact tests. Differences in the mean values between ≥3 groups were evaluated by a 1-way ANOVA. All statistical analyses were separately performed for the LSPV, LIPV, RSPV, and RIPV. A multivariate analysis was performed by a logistic regression analysis (backward hierarchical elimination method) to identify the parameters (clinical characteristics, and echocardiographic and anatomic parameters) associated with a successful PVI by a single freeze, and the in-procedural parameters associated with a successful PVI. The optimal cutoff point was chosen as the combination with the highest sensitivity and specificity using a receiver-operating characteristic curve. A 95% CI was presented with the area under the curve (AUC). All p values were 2-sided, and statistical significance was established at p < 0.05.
The patient characteristics are shown in Table 1. Among 440 PVs, 427 were isolated successfully using exclusively 28-mm CBs, and the remaining 13 PVs (2 LSPVs, 4 LIPVs, 1 RSPVs, and 6 RIPVs) required touch-up ablation. Among the 440 initial freezes, 32 freezes that were terminated prematurely before reaching 180 s, mainly due to a reduction in the compound motor action potential amplitude and were excluded from the analysis; the remaining 408 initial freezes were classified into 4 PVs and further analyzed separately.
An acute PVI was achieved by a single application (NND: 1) in 87 of 107 (81.3%), 84 of 106 (79.2%), 88 of 95 (92.6%), and 76 of 100 (76.0%) and by a second application (NND: 2) in 15 of 107 (14.0%), 14 of 106 (13.2%), 4 of 95 (4.2%), and 9 of 100 (9.0%) LSPVs, LIPVs, RSPVs, and RIPVs, respectively (Figure 3). All 408 PVs were isolated successfully by cryothermal ablation. The mean NND significantly differed among the 4 PVs (LSPV: 1.26 ± 0.65; LIPV 1.33 ± 0.79; RSPV: 1.12 ± 0.52; RIPV: 1.50 ± 1.03; p < 0.001), and the proportion of an NND of 1 was highest in the RSPV and lowest in the RIPV (Figure 3). It was significantly higher in the superior PVs than the inferior PVs (176 of 203 vs. 159 of 205 PVs; p = 0.016), and was similar between the left and right PVs (170 of 212 vs. 165 of 196 PVs; p = 0.293). Procedure-related complications occurred in 4 patients, including cardiac tamponade requiring pericardiocentesis in 1 and transient right phrenic nerve injury that remained on the next day of the procedure in 3 patients.
Anatomic parameters predicting procedural difficulty
The parameters measured on MDCT are presented in Table 1. The PVOA, PVTL, diameter of the medial branch of the PV, and ovality index significantly differed among the 4 PVs (p < 0.001 in each). The results of the multivariate analyses to identify the clinical, echocardiographic, and anatomic parameters associated with a higher NND in the 4 PVs are shown in Table 2.
For the LSPV, a multivariate analysis revealed that a thinner LLRT, higher ovality index, and longer PVTL were the independent predictors of an NND ≥2. The optimal cutoff of the LLRT was 4.7 mm (AUC: 0.676; sensitivity: 0.750; specificity: 0.597), and a thinner LLRT (OR: 4.86; 95% CI: 1.43 to 15.60; p = 0.011) predicted a higher NND. The optimal cutoff of the ovality index was 50.5% (AUC: 0.637; sensitivity: 0.350; specificity: 0.931), and a higher ovality index (OR: 9.44; 95% CI: 2.19 to 40.7; p = 0.003) predicted a higher NND. The optimal cutoff of the PVTL was 26.1mm (AUC: 0.674; sensitivity: 0.700; specificity: 0.586), and a longer PVTL (OR: 5.98; 95% CI: 1.65 to 21.7; p = 0.006) predicted a higher NND. In the LIPV, no predictors could be identified. In the RSPV, a multivariate analysis revealed that an older age was the independent predictor of an NND of ≥2. The optimal cutoff age was 68 years (AUC: 0.821; sensitivity: 1.000; specificity: 0.663), and age >68 years (OR: 16.20; 95% CI: 1.86 to 142.0; p = 0.012) predicted a higher NND. In the RIPV, a multivariate analysis revealed that a shorter PVTL was the sole independent predictor of an NND of ≥3. The optimal cutoff of the PVTL was 10.4 mm (AUC: 0.869; sensitivity: 0.800; specificity: 0.800), and a PVTL <10.4 mm (OR: 10.10; 95% CI: 3.0 to 34.30; p < 0.001) was associated with a higher NND.
Because the RIPV required touch-up ablation most frequently as well as repeat applications for an acute PVI, further analysis was performed to identify the factors associated with the requirement for touch-up ablation. A multivariate analysis revealed that a PVTL of <10.4 mm was still the sole predictor for the requirement of touch-up ablation (OR: 16.8; 95% CI: 1.81 to 156.0; p = 0.013). Figure 4 shows representative cases with a short and long PVTL.
Thermal dynamics and successful PVI
The balloon temperature sharply dropped and reached the beginning of the Ti at a mean of 33.2 ± 2.5 s from the start of the application. The mean Ti and Tm of the initial freeze were -34.9°C ± 5.4°C and -49.8°C ± 7.1°C, respectively. The Ti, time to the Ti (from the start of the freeze), and Tm during the initial freeze for the 4 PVs are shown in Table 3. Interestingly, the time to the Ti was similar among the 4 PVs despite both the Ti and Tm having had significantly differed among the 4 PVs. Both the Ti and Tm were lowest at the RSPV and highest at the LIPV, and were significantly lower in the superior PVs when compared with the inferior PVs. In the multivariate analyses, a smaller LA volume index (p = 0.03), successful PVI (p < 0.0001), larger PVOA (p < 0.0001), and longer PVTL (p < 0.0001) were associated with a lower Tm. A successful PVI (p < 0.0001), larger PVOA (p < 0.0001), and longer PVTL (p = 0.045) were associated with a lower Ti.
Both the Ti and Tm were significantly lower in CB applications that resulted in a successful PVI than a failed PVI. To identify the best cutoff of the CB temperature to discriminate between effective and failed CB applications, receiver-operating characteristic curves of the Ti and Tm were constructed. The results of the receiver-operating characteristic statistics are shown in Table 4. Successful PVI could be predicted by the Ti with a positive predictive value of 87.7%, 88.5%, 98.5%, and 81.6% using a cutoff temperature of -34°C, -33°C, -37°C, and -33°C in the LSPV, LIPV, RSPV, and RIPV, respectively. Similarly, a successful PVI could be predicted by the Tm with a positive predictive value of 92.2%, 86.5%, 98.7%, and 89.5% using a cutoff temperature of -51°C, -43°C, -52°C, and -48°C in the LSPV, LIPV, RSPV, and RIPV, respectively.
To the best of our knowledge, this is the first report to investigate the anatomic parameters predicting procedural difficulty and the in-procedural balloon temperature predicting a successful PVI in each of the 4 individual PVs separately during a 28-mm second-generation CB ablation. We found that: 1) procedural difficulty in achieving PVI and the thermal dynamics during the freezing significantly differed among the 4 PVs; 2) the anatomic parameters obtained from the pre-procedural MDCT might aid in predicting the procedural difficulty in achieving a PVI; and 3) the balloon temperature during the freezing might be useful to discriminate between successful and failed applications, and it was especially predictable using the Ti during the early phase of the freeze.
Although the majority of the PVs could be isolated by a single shot owing to the improved cooling effect of the second-generation CB (7), a large anatomic variation in the LA and PVs resulted in difficulty of achieving an acute PVI by a single shot. Identification of the pre-procedural predictors would be helpful to build a procedural strategy. In the present study, the RSPV showed the lowest NND among the 4 PVs, and older age was associated with an higher NND, presumably because those populations might have a deformed anatomy and/or operators were especially vigilant about damage to the phrenic nerve, which would predispose to incomplete CB contact. It is well-known that the RIPV is the most challenging target PV among the 4 PVs (8–13). Touch-up ablation is required most frequently for the isolation and reconnections are observed most frequently during the chronic phase (8,13). Ostia of the RIPV is located not only close, but also posterior, to the interatrial septum, and the inserted catheter is easily bent too sharp and impairs the pushability. It is also notable that the RIPV had the shortest PVTL among the 4 PVs. In our study, a shorter PVTL was associated with requiring both multiple applications and touch-up ablation. Knecht et al. (9) elegantly reported that early branching was associated with failure during RIPV ablation in the first-generation CB procedure, and our results agree with that report. Because the length of the nose of the CB catheter was 11 mm, optimal alignment could not be obtained and the freezing surface was predisposed to floating from the PV ostia in cases with a short PVTL. The LSPV generally exhibited an upright take-off and large trunk, similar to the RSPV. However, catheter pushability might be impaired rather more than with the RSPV due to the longer distance from the interatrial septum. We found that a thinner LLRT was related to an higher NND. Knecht et al. (9) also reported that continuous sharp ridges were significant predictors of an acute failed PVI due to a lower contact area between the CB surface and myocardium, which was in accordance with our study data. We also found that high ovality was associated with an higher NND, which is supported by the data showing an inverse association between the degree of the PV occlusion during ablation and the ovality of the left-sided PV ostia (14). A longer PVTL might force the balloon into a more horizontal position and affect the catheter pushability, which might synergistically increase the difficulty of balloon contact due to the former 2 factors. In the LIPV, no anatomic factors predicting a higher NND were identified. The LIPV was always targeted after the LSPV in this study, and the impact of a preceding LSPV ablation might have had an influence on this result. In addition, the procedural-related anatomic aspect, such as a transseptal location, should also be noted. The location of the insertion point into the LA might influence the distance and angle to each PV ostia, especially the RIPV. Although we revealed the anatomic factors predicting procedural difficulty, we believe that the presence of anatomic variants of the PVs should not discourage the referral of patients with paroxysmal AF for CB ablation, as shown in previously published data (15,16). More prospective data is necessary for the pre-procedural patient selection using cardiac MDCT.
Thermal dynamics during cryoapplications
Although the best way to evaluate the efficacy of a CB application during freezing seems to be real-time PV potential monitoring, this is not available in about one-half of the applications, especially during an RIPV ablation (8,10,13). Angiography cannot be used to evaluate the contact and location of the CB during freezing. In this context, the balloon temperature provides useful real-time information for the assessment of efficacy. Theoretically, an incomplete occlusion should result in a higher balloon temperature, because the remaining PV blood flow has a rewarming effect on the CB. Fürnkranz et al. (10) reported data on the thermal dynamics during the first-generation CB ablation; however, the cooling effect is considerably enhanced with the second-generation CB (7). Indeed, the proportion of an NND of 1 was much higher in the present study than their study (80.5% vs. 60%), suggesting different thermal dynamics.
In addition, we evaluated the thermal dynamics of the 4 PVs separately because both the Ti and the Tm significantly differed among the 4 PVs. The superior PVs had a lower Ti and Tm than the inferior PVs, presumably because of a higher contact force due to better alignment of the sheath/balloon system to the superior PVs and a larger PVOA. Indeed, both the Ti and Tm were significantly associated with the PVOA in our study. Because the surrounding flow interferes with the freezing of the balloon surface, a larger contact area might result in a lower balloon temperature. In the lower PVs, it may be more difficult to fit the spherical CB to the PV ostium and to obtain a larger contact area, resulting in a higher temperature. In addition, much more surrounding flow close to the mitral valve might rewarm the CB when the balloon is placed at the LIPV ostium.
As expected, balloon temperatures were significantly lower with CB applications that resulted in a successful PVI than a failed PVI. It should be noted that a balloon temperature difference was already apparent at the beginning of the plateau phase (30-40 s from freeze initiation) between efficient and failed applications. This provides an opportunity to interrupt inefficient freezing to shorten the procedure duration and avoid overtreatment. When a sufficient temperature decrease was not observed at the beginning of plateau phase, repositioning the balloon should be considered after termination of the application. The present study suggested that the efficacy of the initial freeze might be predictable during the early phase of the freeze using different cutoff temperatures in the 4 PVs, which might be useful to decide whether to continue further freezing. A further prospective, multicenter study is necessary to confirm our study results.
First, this study was a single-center, retrospective analysis. Second, the sequence of targeting the PVs might have had some impact on the results. Third, the patients with a left common PV were excluded because of our limited experience and a large variation in the size and configuration of left common PVs. Fourth, an incidence of needing touch-up of 3.2% was relatively higher than the reported data from highly experienced centers.
Procedural difficulty in achieving PVI and the thermal dynamics during the cryoapplications significantly differed among the 4 PVs during the second-generation CB ablation. The anatomic information obtained from MDCT might aid in predicting the procedural difficulty pre-procedurally. In addition, the balloon temperature during the early phase of the initial freezing might be useful to discriminate between successful and failed PVIs with a high positive predictive value. This information might be helpful for operators to predict the procedural difficulty and to avoid any prolonged, inefficient freezing during the procedure.
COMPETENCY IN MEDICAL KNOWLEDGE: Pre-procedural estimation of the individual difficulty of PVI of each PV is helpful for operators. Although it is ambiguously recognized that the balloon temperature is associated with a success isolation, there are few data available that quantify the correlation between them in using second-generation CBs. Our present investigation provided a significant cut-off value to predict a successful cryoapplication during the early CB ablation freezing phase.
TRANSLATIONAL OUTLOOK: A specific method of PVI using a second-generation CB regarding the unique anatomic features of each patient as well as the in-procedural balloon temperature should be established and evaluated in a prospective multicenter trial.
The authors thank Mr. John Martin for his linguistic assistance in the preparation of this manuscript.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- area under the curve
- left atrium
- left inferior pulmonary vein
- thickness of left lateral ridge
- left superior pulmonary vein
- multidetector computed tomography
- number needed to disconnect
- pulmonary vein
- pulmonary vein isolation
- area of the pulmonary vein ostium
- length of the pulmonary vein trunk from ostium to the bifurcation
- right inferior pulmonary vein
- right superior pulmonary vein
- nadir balloon temperature
- temperature at the starting point of slow cooling phase
- Received September 1, 2016.
- Revision received January 5, 2017.
- Accepted January 12, 2017.
- 2017 American College of Cardiology Foundation
- Nault I.,
- Miyazaki S.,
- Forclaz A.,
- et al.
- Calkins H.,
- Kuck K.H.,
- Cappato R.,
- et al.
- Kojodjojo P.,
- O'Neill M.D.,
- Lim P.B.,
- et al.
- Packer D.L.,
- Kowal R.C.,
- Wheelan K.R.,
- et al.,
- STOP AF Cryoablation Investigators
- Miyazaki S.,
- Taniguchi H.,
- Hachiya H.,
- et al.
- Miyazaki S.,
- Hachiya H.,
- Taniguchi H.,
- et al.
- Heeger C.H.,
- Wissner E.,
- Mathew S.,
- et al.
- Sorgente A.,
- Chierchia G.B.,
- de Asmundis C.,
- et al.
- Chun K.R.,
- Schmidt B.,
- Metzner A.,
- et al.
- Khoueiry Z.,
- Albenque J.P.,
- Providencia R.,
- et al.