Author + information
- Received December 23, 2016
- Revision received March 21, 2017
- Accepted March 23, 2017
- Published online October 16, 2017.
- Waqas Ullah, PhDa,b,∗ (, )
- Ross J. Hunter, PhDa,
- Malcolm C. Finlay, PhDa,
- Ailsa McLean, RNa,
- Mehul B. Dhinoja, MBBSa,
- Simon Sporton, MDa,
- Mark J. Earley, MDa and
- Richard J. Schilling, MDa
- aCardiovascular Biomedical Research Unit, Barts Heart Centre, St. Bartholomew’s Hospital, Barts Health National Health Service Trust, London, United Kingdom
- bCardiology Department, University Hospital Southampton National Health Service Foundation Trust, Southampton, United Kingdom
- ↵∗Address for correspondence:
Dr. Waqas Ullah, Cardiovascular Biomedical Research Unit, Barts Heart Centre, St. Bartholomew's Hospital, Barts Health National Health Service Trust, West Smithfield, London EC1A 7BE, United Kingdom.
Objectives This study sought to assess the impact of ablation power and catheter irrigation during clinical radiofrequency ablation using impedance drop.
Background In preclinical studies, ablation power and catheter irrigation are determinants of ablation efficacy.
Methods Static 30-s left atrial ablations were delivered in patients undergoing their first atrial fibrillation ablation. Impedance drop during ablation (as a measure of efficacy) was compared using the following: the force time integral (FTI); the FTI-P (a cumulative multiple FTI and ablation power), and ablation index (AI), a weighted algorithm including contact force, power, and duration. Comparison was also made between a conventionally irrigated (SmartTouch [ST]) versus surround flow (STSF) contact force–sensing catheter.
Results We analyzed 1,013 ablations. For both catheters, the Spearman correlation was higher between impedance drop and AI (rho = 0.89 ST, 0.84 STSF) than FTI-P (rho = 0.71 ST, 0.53 STSF) or FTI (rho = 0.77 ST, 0.52 STSF); p < 0.0005 for each. STSF ablations had lower minimum catheter tip temperatures (25°C [interquartile range (IQR): 25°C to 27°C] vs. 35°C [IQR: 34°C to 36°C]; p < 0.005), and lesser impedance drop per FTI or AI (p < 0.005 for both). For STSF, impedance drop plateaued sooner than for ST with respect to FTI (184g.s vs. 463g.s) and AI (370 AI vs. 430 AI).
Conclusions AI is a more complete ablation descriptor than is FTI or FTI-P, reflected by a stronger correlation with impedance drop. STSF ablations have lower impedance drop per AI or FTI than ST ablations do, suggesting different targets should be used if ablating guided by impedance drop with STSF. With ST, ablation beyond 430 AI provides minimal additional biophysical efficacy, suggesting an upper limit to use for clinical ablation.
- ablation index
- atrial fibrillation ablation
- contact force
- force time integral
- surround flow irrigation
Preclinical work has demonstrated that factors including ablation duration, tissue contact force (CF), power, and catheter irrigation are determinants of radiofrequency lesion size (1–3). Clinical studies have demonstrated relationships between force time integral (FTI), the product of ablation duration and CF, and ablation efficacy as judged by arrhythmia recurrence (4), pulmonary vein reconnection (4), and impedance drop (5). An FTI-based analysis though does not include ablation power, and one would presume that including power would be a more complete description of the delivered ablation and so correlate better with efficacy.
Efficacy is readily judged for an individual ablation preclinically using histological lesion parameters, but these parameters are not available clinically. In animal studies, a correlation has been observed between lesion dimensions and impedance drop during ablation (2,6). Consequently, impedance drop has been used as a target for ablation (7) and to assess individual ablation efficacy clinically (5). In this study, our primary hypothesis was that the inclusion of power in the description of an ablation would lead to a greater correlation with the impedance drop than FTI would. The former was achieved using a simple cumulative multiple of FTI and power (FTI-P) and also, for comparison, a novel measure, the ablation index (AI) (8,9), which includes these factors in a weighted algorithm described in the methods section.
Recently, a CF-sensing catheter, which has surround flow (SF) irrigation (SmartTouch Surround Flow [STSF], Biosense Webster, Diamond Bar, California), has been introduced (Figure 1). Whereas non-CF sensing SF catheters have been studied in humans in small studies (10,11), it is unknown how the STSF catheter compares clinically with the conventionally irrigated SmartTouch (ST) catheter (Biosense Webster). A secondary objective for the study was to compare the impedance drop during ablation between the ST and STSF catheters (based on FTI and also ablation descriptors including power). The hypothesis here was that irrigation mode significantly affects impedance drop (biophysical efficacy) during ablation.
All participants gave informed consent to participate in the study, and the study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the institution's human research committee and the U.K. National Research Ethics Service. Consecutive patients undergoing their first catheter ablation procedure for atrial fibrillation (AF) were enrolled. Procedures were performed with patients under moderate (“conscious”) sedation. All data were collected prospectively. CF was measured using a ST or STSF catheter. In a subset of cases, at the discretion of the operator, remote robotic navigation (Sensei Robotic Catheter System, Hansen Medical, Mount View, California) was used, in which case, the transseptal sheath used for the ablation catheter was an Artisan sheath (Hansen Medical), otherwise a Mullins sheath (Medtronic, Minneapolis, Minnesota) was used. Carto3 (Biosense Webster) was used to record CF and biophysical data, and LabSystem Pro (Boston Scientific, Marlborough, Massachusetts) to record electrogram data.
Study ablations were performed during wide area circumferential ablation, and, for persistent AF patients, additionally during complex fractionated atrial electrogram ablation. All study ablations were of 30-s duration, with the catheter in a stable position for at least 2.5 s prior to energy delivery and at least 8 s after cessation. In the ST cohort, ablations were performed in temperature-controlled mode with temperature limited to 48°C and power to 30 W. The irrigation flow rate was 2 ml/min during mapping and 17 ml/min during ablation. In the STSF cohort, ablations were in power-controlled mode with power limited to 25 W and flow of 8 ml/min during ablation. The power adjustment using the STSF and irrigation flow during ablation for both catheters were per the manufacturer’s recommendations. All ablations were performed at a CF of 5 g to 40 g. Locations where there was visual macro-displacement were discarded. Location markers were manually applied pre- and post-ablation on Carto3, as well as lesion markers during ablation. Automated lesion tagging was not used for study ablations. Study ablations were nonoverlapping.
Impedance during ablation was measured between the tip of the ablation catheter and the ground patch (on the patient’s left thigh), using a 50 kHz current, recorded at 10 Hz.
All data were processed and analyzed using custom-written Matlab (MathWorks, Natick, Massachusetts) scripts.
Impedance and FTI data were processed using previously described methods (5,12,13). To include power in the assessment of the delivered ablation, 2 methods were used. In the first case, a simple cumulative multiple of FTI and power was used, with the FTI determined in 100-ms increments and multiplied by the measured ablation power for that period. These values were then summated to give the FTI-P at a point in time. In the second case, a commercial algorithm (Biosense Webster) was used (8,9): , where CF is contact force, P is radiofrequency power, T is application time, and a, b, c, and k are constants. The constants are proprietary and unpublished. This nonlinear formula is weighted unequally with power receiving a higher weighting than CF and generates a value termed the AI. Impedance data were processed to remove noise using a Savitzky-Golay filter. FTI was determined by the area under the CF curve derived using trapezoidal integration.
To investigate the relationships among FTI, the simple multiple, AI, and impedance drop, an incremental analysis (5,12) was performed by dividing all ablations into consecutive, cumulative intervals of 10 g·s FTI, 100 g·s/W for the simple multiple, or 5 AI. The maximum impedance drop was then determined by comparing the minimum impedance during that part of the ablation to the impedance at ablation onset. This was converted to a percentage to minimize any influence from the initial impedance. As an example, where there was a 600g.s ablation, the maximum impedance drop was determined at 0g.s to 10g.s, 0g.s to 20g.s, and so on to 0g.s to 600g.s (producing 60 measurements for the ablation).
In the case of the FTI curves, the plateau with the impedance drop relationship was determined quantitatively by determining the point in the fitted curve for these relationships where the impedance drop was 0.5% per 100g.s of FTI (5). The first derivative of the fitted curve formula was used to determine where the instantaneous gradient of the curve fell to this value. For the AI curves, the onset of the plateau phase was determined qualitatively.
Statistical analysis was performed using SPSS (version 20 [IBM SPSS Statistics, Armonk, New York]) and Matlab (version 8.6, Statistics Toolbox version 10.1, MathWorks, Natick, Massachusetts). A p value of <0.05 was taken to indicate statistical significance. The strength of correlation between variables was assessed using Pearson correlation where the distributions appeared normal and Spearman rank correlation otherwise. Normally distributed data were analyzed using a Student t test or analysis of variance and non-normally distributed data using a Mann-Whitney U test or Kruskal-Wallis test. To account for any differences in ablation efficacy due to factors identified as significant in previous work (12), hierarchical multivariate analyses were conducted incorporating the patient number as well as these factors: catheter orientation, drift, navigation mode, and atrial rhythm. Data are presented as mean ± SD or median (interquartile range).
Baseline characteristics are presented in Table 1.
This analysis included 1,013 study ablations. Ablation parameters are compared in Table 2. The STSF catheter ablations had a lower temperature at the start of ablation than the ST catheter, as well as lower impedance. In both cases, the catheter tip temperature fell during radiofrequency ablation (though for longer and to a lower temperature with the STSF). After this fall, the temperature tended to rise in the ST but not STSF group.
There was a strong correlation between FTI and percentage impedance drop in the ST and STSF groups (Spearman rho = 0.77 and 0.52, respectively; p < 0.005 for both). The curves for impedance drop versus FTI are shown in Figure 2. The curves were compared statistically using a hierarchical multivariate analysis taking into account other factors that could have an impact on the impedance drop (12) to allow for a purer comparison of irrigation mode; impedance drop was significantly higher where the ST rather than STSF catheter was used: p < 0.005; unstandardized β = 2.2 (95% confidence interval: 2.13 to 2.28); adjusted R2 = 0.7. The ST curve plateaued from 463g.s and the STSF curve from 184g.s.
When power was included in the ablation description, the correlations became stronger than for FTI, especially for the STSF catheter, but only where power was included as part of the AI calculation. For the simple multiple, the correlations were: Spearman rho = 0.71 (ST) and 0.53 (STSF), p < 0.005 both, with the curve demonstrating a broadly similar shape to the FTI/impedance drop curve. The correlations between AI and percentage impedance drop in both the ST and STSF groups were the strongest for the cohort: Spearman rho = 0.89 and 0.84, respectively; p < 0.005 for both.
The AI/impedance drop curve was markedly different from the logarithmic FTI/impedance drop curves: there was a lag phase before the impedance started to drop, following which, there was a linear phase and then the curve plateaued (Figures 3 and 4). In the case of the ST catheter, the linear phase was 80 to 430 AI, whereas for the STSF catheter, the initial lag phase and onset of plateau phase were sooner, with the linear portion between 160 and 370 AI. To allow for comparison between groups, the AI range where the relationships were linear for both curves (160 to 370 AI) was compared statistically by a hierarchical multivariate analysis; impedance drop was significantly higher where the ST rather than STSF catheter was used: p < 0.005; unstandardized β = 2.45 (95% confidence interval: 2.37 to 2.53); adjusted R2 = 0.6.
The inclusion of power in the description of an ablation using the AI formula had a stronger correlation with impedance drop during ablation than FTI and also FTI-P. The AI relationship, after a lag in the impedance rise, appeared linear after which it plateaued. The impedance drop relationships plateaued earlier with the STSF catheter than the ST catheter with respect to both FTI (184g.s vs. 463g.s, respectively) and AI (370 AI vs. 430 AI, respectively).
In animal studies, a correlation has been observed between lesion dimensions and impedance drop during ablation (2,6). In clinical work, impedance drop has been used as a surrogate for tissue heating, and so lesion size, during radiofrequency ablation (14), and has been used as a target for clinical ablation (7). In using it as a measure of biophysical ablation efficacy, comparison can be drawn regarding the impact of ablation parameters such as catheter irrigation mode on an actual outcome measure that can be related to individual ablation applications.
Previous work has demonstrated a weaker relationship between impedance drop and FTI (15). In the current study, impedance data was sampled at a high frequency (10 Hz), the resulting waveform filtered, percentage rather than absolute impedance drop examined, and lastly, the results were averaged across a very large number of measurements in a large number of ablations. This combination of methods was used to improve the reliability of assessing impedance drop over the course of an averaged ablation but clearly could not be used for an individual radiofrequency application. An ablation parameter that is more strongly predictive of impedance drop though, such as AI, therefore represents a useful target for clinical ablation (until a reliable real-time outcome parameter such as individual lesion visualization becomes available).
In animal studies, a plateau is observed between lesion size and delivered ablation energy (16) and also CF (6), with higher lesion size plateaus at higher energies or CF. These plateaus therefore represent the maximum lesion size that can be attained for a particular set of ablation parameters. FTI correlates with lesion size (17) and plateaus in the impedance drop–FTI relationship have been observed during clinical ablation (5).
FTI does not include ablation power, and increasing ablation power results in larger lesions (2,3). The correlation with impedance drop between measures including ablation power was strongest for AI. This suggests an advantage to including power, not merely by simple multiplication but in the weighted AI formula. This is especially the case for STSF, which was used at a lower power setting. When power is not included (the FTI analysis), the correlation with impedance drop is weaker for STSF than ST, it improves slightly when power is included as a simple multiple, but when the weighted formula is used, the correlation becomes of a similar magnitude for the 2 catheters. This suggests that the greater weighting given to power in the AI formula over CF is appropriate. In a beating canine heart model, AI has been found to prospectively predict lesion depth in the ventricles with high accuracy (8). AI has also been found to be predictive of reconnection sites in pulmonary vein isolation lines (9).
The impedance drop versus AI and FTI curves had different shapes. The latter was a logarithmic curve while the former started with a lag before becoming linear and then plateauing. This relationship between AI and impedance drop is only partially consistent with the preclinical work where a linear relationship was observed (8). One possible reason for this is that the current work is in a human atrium rather than a canine ventricle. The lag likely represents the influence of catheter-tip irrigation cooling the tissue. This cooling likely plays a greater role in the thin-walled atrium compared with the thicker ventricle and has to be overcome prior to effective heating of the tissue by radiofrequency (hence the lag). The plateau likely represents the lesion attaining maximal size. Not only is the correlation with impedance drop stronger with AI than FTI, but also the linear portion and more obvious plateau suggest it should be a more useful measure clinically.
SF catheters have been shown to cause more effective catheter tip cooling previously in vitro (18) and in the current study (Table 2). By cooling the catheter tip more efficiently, the expectation would be that one could ablate more effectively and safely with reduced thrombus formation. In prior studies investigating non-CF sensing catheters, pulmonary vein reconnection at 30 min in paroxysmal AF cases was lower with SF catheters (10,11), although at 6 months, no difference in recurrence rates was observed (10). Non-CF sensing work in vitro demonstrated that at 20 W, lesion surface diameter was significantly smaller with the SF catheter, whereas at 35 W, there was no difference (18). In another study, with ablation at 30 W, no significant difference in lesion volumes was found (19).
In the current study, there was a significantly lower impedance drop with STSF than ST during ablation. In the case of the FTI-based analysis, this could be explained by the use of manufacturer-recommended lower power settings with STSF. Therefore, for the comparison, AI is important as it takes power settings into account. This is also important due to the different ablation modes used and is highlighted by the significantly shorter time to achieve maximum power in the power-controlled STSF ablations than the temperature-controlled ST ablations (Table 2). Whereas in the first few seconds of an ablation, if the CF is the same, the FTI could be identical between a power- and temperature-controlled ablation, the AI would be very different as the power will increase to maximum faster in a power-controlled ablation. In fact, for the ST catheter, simply multiplying in the power gives a slightly worse correlation with impedance drop than the FTI, suggesting the importance of the weighting of power in the AI formula, whereby the lower powers at the start of the temperature-controlled ablations are given greater significance. On comparing the 2 catheters based on AI, the impedance drop is still lower with the STSF catheter. Based on the catheter tip temperature measurements in Table 2, the STSF catheter irrigation is very efficient at cooling the catheter tip: a lower temperature is reached and catheter tip cooling continues for longer during the ablation compared with ST and does not really rise during ablation. The atrium is a thin-walled structure, with maximal wall thickness by region of 2.79 ± 0.91 in 1 autopsy study (20). It may be that the STSF catheter, by being so efficient at cooling the catheter tip, is also excessively cooling the adjacent atrial tissue, counteracting radiofrequency energy-related heating and so reducing efficacy. This may explain the longer lag compared with the ST catheter before the tissue impedance starts to drop in the AI curves. This effect of SF irrigation would not have been appreciated in preclinical studies where thicker ventricular tissue was examined (8,18).
It may be the case that the STSF catheter measures impedance differently per se. Supportive of the notion that the STSF catheter affects impedance measurements is the finding of lower impedances at the start of radiofrequency delivery. These lower impedance values are unexpected as the catheter tip temperature is lower compared with the ST catheter at the start of ablation, and, if this is associated with a lower tissue temperature, one would expect it to be associated with a higher rather than lower impedance: previous work though has not examined the impedance relationship below 37°C (2,21). It could be that at lower tissue temperatures, such as those related to the STSF catheter, the relationship is different. In a preclinical study using non-CF sensing Thermocool and Thermocool SF catheters (Biosense Webster), no significant differences were found in the starting or final impedances (19)—so whether any differences observed are unique to human atrial ablation or the STSF catheter itself are unclear. In view of the differences noted in this study between the ST and STSF catheter with respect to impedance, it is therefore important to appreciate that previous impedance drop targets for ablation validated with the ST catheter are unlikely to be applicable to the STSF catheter.
Based on the impedance drop plateaus, it appears that maximal biophysical efficacy is achieved at a lower FTI and AI in the STSF cohort (as the curves plateau earlier). How the actual lesion sizes compare between catheters is unknown from the current data, but the plateaus do represent values for each catheter beyond which there appears to be minimal further biophysical benefit to continuing ablation (as there will be no further impedance drop gains). The plateau AI value of 430 for ST catheters observed here is similar to the value in a recent study investigating the mean AI in wide area circumferential ablation segments that did not reconnect (422 AI), compared with segments where reconnection was observed, which had been ablated at a significantly lower mean AI (9).
Histological lesion parameters were not available and so an alternative, biophysical, measure of efficacy (the impedance drop) was assessed, taking this to be a surrogate of lesion parameters. This appears reasonable on the basis of prior preclinical work (2,6). If though the catheter affects the way a surrogate relates to lesion parameters, then comparisons between catheters based on that surrogate are no longer valid. In this case, this has made efficacy comparisons based on tissue impedance between the ST and STSF catheters less conclusive.
The focus of the current study was at the individual ablation level. Once the protocolized study ablations had been delivered, operators were permitted to pursue whichever ablation strategy they wished. This meant that comparisons beyond the study ablations, such as relating contact force parameters to clinical outcomes were beyond the design of the study.
The catheters were used with different ablation settings: the ST in temperature-controlled power limited mode; the STSF in power-control mode. The latter is recommended by the manufacturer whereas the former reflects how the ST catheter is used in many labs, including our own (22–24). The comparison therefore represents a “real world” one for the 2 catheters, reflecting how they would be used in practice. A key impact of these different ablation settings is that while the pre-set maximum power is reached with both ablation settings, it is attained more rapidly during power-controlled ablation (Table 2), a factor that would affect FTI-based comparison where power is not included in the algorithm (and may have contributed to the weaker relationship between impedance drop and FTI for the STSF catheter). The analyses including power would not have been affected by this difference though.
AI has been developed by Biosense Webster and, as such, the presented data only applies to the catheters from that manufacturer used in the study: different irrigation strategies used by other catheter manufactures may lead to different results.
AI, which is generated by a weighted nonlinear formula, has a stronger relationship with impedance drop during ablation than a simple multiple of CF, power, and ablation duration or FTI. This suggests that AI is a better descriptor of an ablation and favors its adoption for clinical ablation over the other measures.
STSF catheter ablations are associated with a lesser impedance drop per FTI and AI (though with earlier plateaus in both relationships, suggesting maximal biophysical efficacy is achieved sooner with STSF). At the very least, this suggests the impedance drop relationship is altered with the STSF catheter, which is of relevance to operators targeting an impedance drop with ablation, as different targets would be needed compared with ST. Whether the differences in the impedance drop with ablation observed represent difference in histological lesion parameters is not known.
In the AI–impedance drop relationship there is a linear phase making the effect of AI on biophysical efficacy more predictable. In the left atrium, the AI–impedance relationship plateaus from 430 AI for the ST, suggesting ablation beyond this value has minimal additional biophysical benefit. Prospective, AI-guided ablation studies assessing clinical outcomes are merited to explore this further.
COMPETENCY IN MEDICAL KNOWLEDGE: The addition of ablation power to the contact force and duration into the description of a radiofrequency ablation is associated with a stronger correlation with an outcome measure of the ablation and impedance drop, compared with measures not including power. This improvement in the relationship is through the inclusion of power in a weighted algorithm (the AI) rather than a simple multiple of the 3 constituents. The use of a catheter that has SF rather than conventional irrigation changes the nature of the relationship between ablation and impedance drop and suggests that ablation targets established for catheters using the latter irrigation modality may not be applicable when SF irrigation is used.
TRANSLATIONAL OUTLOOK: Further studies are warranted to assess the impact on clinical outcomes for AF ablation of using ablation index to guide radiofrequency ablation prospectively, based on the targets suggested in this study, as well as whether these outcomes are affected when a contact force sensing SF (rather than conventional irrigation) catheter is used for the procedure.
Supported by Investigator Initiated Study funding agreements (IIS 146, IIS 371) from Biosense Webster Inc. Dr. Finlay is a shareholder and founder of Epicardio; has received travel and lecture fees from Biotronik; and, through Epicardio, has received travel fees from St. Jude Medical. Drs. Sporton and Earley have received lecture honoraria from Biosense Webster. Dr. Schilling has received research grants and fellowship support from Biosense Webster. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- atrial fibrillation
- ablation index
- contact force
- force time integral
- simple multiple of force time integral and power
- surround flow
- SmartTouch surround flow
- Received December 23, 2016.
- Revision received March 21, 2017.
- Accepted March 23, 2017.
- 2017 American College of Cardiology Foundation
- Ullah W.,
- Hunter R.J.,
- Baker V.,
- et al.
- Nakagawa H.,
- Ikeda A.,
- Govari A.,
- et al.
- Das M.,
- Loveday J.J.,
- Wynn G.J.,
- et al.
- Ullah W.,
- Hunter R.J.,
- Baker V.,
- et al.
- Wittkampf F.H.,
- Hauer R.N.,
- Robles de Medina E.O.