Rotors in Human Atrial Fibrillation

Published on 20/06/2015 by admin

Filed under Cardiovascular

Last modified 22/04/2025

Print this page

This article have been viewed 1486 times

Rotors in Human Atrial Fibrillation

Sanjiv M. Narayan and Wouter-Jan Rappel

Chapter Outline

Triggers Interact With Sustaining Mechanisms to Cause Human Atrial Fibrillation

Mapping Rotors During Human Atrial Fibrillation

Sustaining Rotors and Focal Impulses for Human Atrial Fibrillation

Clinical Implications

Conclusions

Mechanistic concepts for human atrial fibrillation (AF) have evolved substantially. In the past 15 years, meticulous mapping has established that human AF may be triggered by ectopic focal beats, while the combination of bioengineering analyses with detailed mapping has recently provided compelling evidence that AF, once triggered, is maintained by relatively few stable electrical rotors and focal sources that lie in patient-specific locations. These advances in our mechanistic understanding have already led to advances in therapy and improved outcomes for patients with AF.

In seminal work, Haïssaguerre ¹ reported that localized ectopy from the pulmonary veins (PVs) may trigger AF. This discovery launched the field of potentially curative AF ablation, with PV isolation as its cornerstone.² Nevertheless, the mechanisms that perpetuate AF, once triggered, remained undefined.^3,4 The multiwavelet hypothesis proposed that multiple spatially meandering electrical waves cause AF.⁵ However, this did not explain consistent patterns of propagation observed in patients with AF,^6,7 the termination of AF after localized ablation,^2,8 or the opposite finding—that extensive ablation to constrain wavelets may have little acute impact.^2,9 The alternative localized source hypothesis is based on animal and in silico experiments in which rapid localized spiral waves (rotors)^4,10 or focal sources⁷ directly cause AF. Although rotors in human AF have been disputed,⁵ and direct evidence has been lacking,⁴ rotors have been indirectly suggested at sites of high dominant frequency where ablation terminates paroxysmal AF^11,12 and, if interatrial rate gradients are abolished, improves outcome,¹¹ and by AF mapping.¹³

Recent data from our ¹⁴ and other laboratories¹⁵ show that human AF is indeed perpetuated by a small number of stable rotors or focal sources^14,15 in individuals with paroxysmal, persistent, and long-standing persistent AF. The mechanistic role of stable AF source rotors has been supported by the acute termination and subsequent non-reinducibility of AF by brief targeted ablation (focal impulse and rotor modulation [FIRM]) before any other intervention. Clinically, this approach has been shown to substantially improve the long-term elimination of AF compared with conventional ablation alone.¹⁶

In this chapter, we summarize the evidence for localized electrical rotors and focal sources for human AF, in the context of the historical background in this area. We then describe approaches to, and results from, mapping rotors and focal sources for human AF, and conclude with current clinical applications of this approach.

Triggers Interact With Sustaining Mechanisms to Cause Human Atrial Fibrillation

The fact that AF may be triggered by ectopic beats ¹ draws an intuitively attractive parallel with the mechanisms for “simple” supraventricular (SVT) and ventricular (VT) tachycardias, which are also triggered by ectopic beats. In those arrhythmias, triggers engage specific mechanisms of dual atrioventricular (AV) nodal physiology to maintain AV node reentry,¹⁷ an accessory pathway for maintaining AV reciprocating tachycardia,¹⁸ or a slowly conducting isthmus to maintain ventricular tachycardia.¹⁹

We hypothesized that AF may be caused when triggers engage specific AF-maintaining mechanisms that may be created dynamically via conduction block, leading to localized reentry, or by triggering repetitive focal beats. To address this hypothesis, we studied the rate dependence of monophasic action potential duration (APD) in the left and right atria, and bi-atrial patterns of rate-dependent bi-atrial conduction slowing in the left and right atria.20,21 and propagation en route to AF using multipolar basket catheters.

In a series of studies, we recently demonstrated that human AF onset is consistently preceded by alternans and complex oscillations in APD, which create a milieu of heightened repolarization dispersion immediately before AF onset. In patients with persistent and paroxysmal AF, we have found that sustained rapid rates produce marked alternans and complex oscillations in APD ²² preceding AF,²³ independently of APD restitution. In parallel, bi-atrial conduction velocity slows dynamically (restitution) just before AF onset at the location where AF is initiated.²⁴

In summary, these observations show that the onset of human AF is preceded by dynamic localized conduction slowing and exaggerated repolarization dispersion, both of which may be subtle or concealed at baseline. These conditions favor reentry and have provided the functional underpinnings for our studies of human AF mechanisms.

Mapping Rotors During Human Atrial Fibrillation

Prior Mapping Studies of Human AF

Several mechanisms for the maintenance of human AF have been proposed, including disorganized multiwavelet reentry,5,25 spatially localized reentrant²⁶ or focal⁷ sources, and mixed patterns.¹³ Divergence in these hypotheses in large part may reflect mapping that has not always met “classical” requirements: to broadly map chambers of interest, at sufficient spatiotemporal resolution to identify varying patterns, over long enough periods to capture variability, then to use interventions to demonstrate that proposed mechanisms are causal and are not bystanders. Failure to apply these criteria even to simple supraventricular^17,18,27 and ventricular¹⁹ arrhythmias is a well-recognized cause of incorrect diagnosis and potentially undesirable therapy.

Many clinical studies over the past decade show that human AF is spatially non-uniform. For instance, human AF exhibits consistent activation patterns,²⁸ consistent rate, or dominant frequency gradients within and between atria,^6,29 as well as electrocardiographic (ECG) spectra suggesting conserved global spatiotemporal organization for at least days³⁰ within and between patients. These data have long supported the notion that human AF is maintained by spatially localized mechanisms, further supported by the fact that human AF may terminate with ablation at defined triggers, drivers,¹ and other localized regions.^12,31 Such regions may arise in either atrium³² and are difficult to identify a priori, but can be ablated in both atria by a systematic stepwise approach.⁸

Higher-resolution mapping, on the other hand, has produced surprisingly inconsistent results. In seminal intraoperative mapping studies in AF patients, Cox, Schuessler, et al 33,34 found stable reentry within disordered AF that were interrupted by lesions that formed the basis for the Maze procedure³⁵ and that still underpin many current ablation lesion sets.² Conversely, in separate human AF studies,³⁶ Allessie et al found no consistency within disordered AF and concluded that AF was attributed to multiple reentrant waves (as in early computer models³⁷) with “focal” events reflective of transmural breakthrough. However, these studies mapped <20% of the dilated atria in these patients, and did not apply interventions to prove causality of disordered activation. More recent epicardial mapping has revealed localized high-frequency regions in AF patients consistent with sources.^7,38 Schilling et al³⁹ and more recently Cuculich et al¹³ used mathematical inverse solutions to map the atria via noncontact approaches (Ensite 3000™, St Jude Medical, Minneapolis, Minnesota, and EcVue™, Cardioinsight, Cleveland, Ohio, respectively) in AF.¹³ However, such studies have yet to use focused ablation to establish causality of each proposed mechanism and exclude bystanders.

Requirements to Map Human AF

Without a priori knowledge of the spatiotemporal organization of human AF, it has been difficult to propose “design” requirements for mapping. In 2001, we set out to study human AF by designing an approach suitable for mapping spatially meandering multiple wavelets or for identifying potential localized sources (rotors or focal sources). Our initial hypothesis was that localized sources do not exist in human AF.

Detection Design Requirements

We reasoned that the factor most relevant to mapping spatial resolution is the length scale of the mapped event. Figure 43-1 illustrates a rotor with limited movement of the rotor tip (in a locus of migration) and wavebreak to fibrillation, generated in a computer simulation. Electrophysiological model parameters were chosen such that the spiral breaks down far from the migration locus.^3,4,23 The rotor controls activation in a “coherent domain” in 1 : 1 fashion with length scale R_rotor, beyond which activation breaks down into complex spatiotemporal patterns. A coherent rotor domain surrounded by incoherent activity can also be generated in simulations of heterogeneous tissue.^40,41 To map the rotor core, the required spatial resolution is comparable with the length scale of the reentrant path R_locus. It is important to note, however, that to simply detect rotational activity around the core (the rotor), the required spatial resolution is coarser and is comparable with its wavelength, λ, which is much larger than R_locus.

Figure 43-1 Snapshot of a Computer Simulation in Homogeneous Tissue
A rotor, the locus of migration of its tip, and its coherent domain of rotational activity (i.e., tissue controlled in a 1 : 1 fashion by the rotor). The model parameters were chosen such that beyond this region of tissue, the rotor breaks down into other spatiotemporal structures as the result of discordant alternans. The relevant length scales in the problem are the size of the domain of locus migration of the rotor tip, R_locus, the coherent domain (of detectable rotational activation), R_rotor, and the wavelength of the rotor, λ.

Figure 43-1 also illustrates the importance of mapping a sufficiently large field-of-view. Attempting to map a rotor with a field-of-view smaller than its trajectory of migration (using a small mapping plaque) may lead to results that are difficult to interpret. Hence, if the locations of putative AF sources are unknown, then as much of the atrial surface as possible should be mapped. Finally, the temporal resolution must be able to distinguish activation between neighboring recording sites and thus can be found by dividing the spatial resolution by the dynamic conduction velocity. For example, for a spatial resolution of 5 mm and a range of conduction velocity of 50 to 150 cm/s,⁴² the required temporal resolution is 3.3 to 10 ms.

We estimated the length scales λ, R_locus, and R_rotor, and thus the required spatial resolution, to map potential localized sources of human AF, based on animal studies and observations of human AF. Animal models of AF show varying mechanisms, including localized spiral waves (rotors),43,44 focal sources,⁴⁵ or nonlocalized waves.¹⁰ In experiments showing rotors, the length scale for the reentrant path (R_locus) ranges from 1 cm to >3 cm,^10,46,47 requiring a minimum resolution of 0.5 cm for mapping. Spatial organization in some models controlled tissue areas >5 cm²,⁴⁴ corresponding to a rotor length scale and wavelength >2 cm. Thus, once the location of a rotor is identified, reentry requires mapping a field-of-view of at least 2.5 × 2.5 cm, with a resolution of ≈1 cm.

In humans, the length scale of the reentrant path may be estimated from the concept of tissue wavelength ⁴⁸ as the product of minimum conduction velocity and the shortest refractory period. In AF patients, minimum (dynamic) conduction velocity in left and right atria is ≈40 cm/s⁴² and the minimum atrial refractory period ≈100 to 110 ms,^20,23 resulting in a minimum wavelength of ≈4 to 5 cm, diameter of ≈1.5 cm, and minimum required spatial resolution of ≈1 cm.

Numeric Validation of Design Requirements for AF Mapping

To test these design criteria for mapping human AF, we performed in silico data validation using an electrophysiological model. The model simulates wave propagation using the monodomain equation:

where V is the membrane voltage, C_m represents the membrane capacitance, D is the diffusion tensor, and I_ion represents the membrane currents. For the purposes of illustrating propagation in silico, we computed membrane currents using the 3- and 4-variable Fenton-Karma (FK) model.^49,50

Figure 43-2 illustrates spiral wave reentry in a 200 × 200-node simulation area with a physical size of 5 × 5 cm, corresponding to a spatial resolution Δx of 0.25 mm (Figure 43-2, A). The spiral is single-armed with a wavelength larger than L_rotor/2 and a period of 90 ms. The computed locus of migration of the rotor tip is illustrated in red in Figure 43-2, A, and consists of a complex meandering trajectory with a length scale of ≈1 cm. Thus, this simulated rotor has L_locus ≈1 cm, L_rotor >5 cm, and λ >2.5 cm.

Figure 43-2 Effect of Mapping Resolution on a Simulated Clockwise Rotor
Simulation on a 200 × 200 grid. A, Spatial resolution Δx = 0.25 mm, corresponding to a 5 × 5-cm domain. The activation is plotted using a gray scale, with white corresponding to depolarized and black corresponding to repolarized tissue. The tip of the rotor meanders as shown in red. The green symbols are isochrones superimposed onto the snapshots. These isochrones are 20 ms apart and are computed as all grid points activated within 1 ms. Scalebar = 1 cm. B, Isochrones computed on a 20 × 20 grid obtained by coarsening the original grid. Isochrones are again 20 ms apart with the order green, blue, purple, and red; scalebar represents spatial resolution (Δx = 2.5 mm). C, Isochrones computed on an 8 × 8 grid providing spatial resolution Δx = 6.25 mm. D, Isochrones computed on a 4 × 4 grid providing spatial resolution Δx = 12.5 mm. Note that rotational activity (organized domain) of the rotor remains detectable at all resolutions.

Activation times for each node were determined using a voltage threshold (10% maximal) and were stored at temporal resolution Δt = 1 ms. Activation times were used to compute isochrones separated by 20 ms (Figure 43-2, B, green). To simulate coarser recording resolution, we coarsened stored activation time intervals but did not rerun simulations. Figure 43-2, B-D illustrates isochrones for the same rotor at resulting spatial resolutions of Δx = 2.5 mm (20 × 20 grid), Δx = 6.25 mm (8 × 8 grid), and Δx = 12.5 mm (4 × 4 grid), corresponding to isochronal intervals of 3 ms, 7 ms, and 18 ms, respectively. Notably, all values of Δx preserved rotational activity of the organized domain of the rotor.

Summary of Design Requirements for Human AF Mapping

These estimates and numeric simulations suggest that a spatial resolution of ≈0.5 cm may be able to resolve rotor migration, and a spatial resolution of ≈1 cm may capture rotational activity of a rotor in human AF. The corresponding temporal resolution should be at least 6 ms. These considerations laid the foundation for our direct contact focal impulse and rotor mapping (FIRM) approach,¹⁴ which we recently described to identify patient-specific sustaining rotors and focal sources for human AF. Localized AF sources were identified at electrophysiological study and their causal role verified by AF termination by patient-specific targeted ablation alone.

Sustaining Rotors and Focal Impulses for Human Atrial Fibrillation

Recently, our group ¹⁶ and others¹⁵ showed that human paroxysmal, persistent, and long-standing persistent AF are predominantly sustained by localized rotors or focal sources. In a multicenter experience that currently includes more than 200 patients, stable rotor or focal sources were identified in >98% of patients, in diverse bi-atrial locations that were stable for prolonged periods in each individual. Direct targeted ablation was applied to each mapped source (FIRM) before any other intervention, which led acutely to the termination or substantial organization of AF in the vast majority of cases. Patients who received FIRM together with conventional ablation had substantially greater freedom from AF on rigorous long-term follow-up compared with those receiving conventional ablation alone.

Identification of Rotors and Focal Impulses in Human AF: Focal Impulse and Rotor Mapping

We designed FIRM mapping on the basis of the above considerations including the requirement to apply ablation to prove proposed AF mechanisms.¹⁴ Accordingly, FIRM mapping is performed during clinical electrophysiology study by advancing 64-pole basket catheters to the right atrium and, after transseptal puncture, to the left atrium. Contact electrodes provide spatial resolution of 4 to 6 mm along each spline, and ≈4 to 10 mm between splines (higher resolution at the poles than at the equator). Heparin is infused to maintain activated clotting time >350 ms. In early work, we mapped both atria simultaneously (Figure 43-3), but we now use one basket: first in the right atrium while performing transseptal puncture, then in the left atrium. Multi-site electrograms are recorded with a temporal resolution of 1 ms as unipoles or as overlapping bipoles to reduce far-field artifact, and are filtered at 0.05 to 500 Hz. AF data are exported digitally as multiple 1-minute epochs over a period of tens of minutes.

Figure 43-3 Anatomical Reference and Nomenclature for FIRM Mapping
A, Fluoroscopic view of 64-pole basket catheters in the right and left atria. An ablation catheter and a coronary sinus catheter are shown, along with an esophageal temperature probe. The image shows a subcutaneous continuous ECG monitor (Reveal XT™, Medtronic, Minneapolis, Minnesota) used to categorically establish elimination of AF after ablation in clinical series. B, Sinus rhythm map on bi-atrial schematic. Activation at basket electrodes, shown as dots, is displayed as a color-coded map from the sinus node to the lateral inferior left atrium. The right atrium is opened between its poles, with the tricuspid annulus opened laterally and medially; the left atrium is opened along its equator, with the mitral annulus opened superiorly and inferiorly. The pulmonary vein ostia are indicated by dashed lines.

A novel system (RhythmView™, Topera Inc., Palo Alto, California) is then used to analyze multipolar AF signals.14,16,51 Electrograms are filtered to exclude noise and far-field signals guided by the reported rate dynamics of human right and left atrial APD^20,22,23,52 to estimate minimum activation time, and conduction velocity^20,24 to map propagation paths. The system produces isopotential movies of AF over successive cycles, as the primary display modality used to guide ablation, as discussed later. However, for illustration, single-cycle isochronal snapshots are provided throughout the chapter.

Demonstration of AF Rotors and Focal Sources

Propagation can be mapped in various rhythms. For orientation, Figure 43-3 illustrates bi-atrial propagation in sinus rhythm. The atria are projected onto grids, with the right atrium cut vertically through the tricuspid valve, opening its lateral and medial halves. The left atrium is cut horizontally through the mitral valve, opening its superior and inferior halves. On these projections, sinus activation is represented by color-coded isochrones emanating from the sinoatrial node (in red), crossing Bachmann’s bundle to the left atrium (in blue).

In AF, isopotential movies were analyzed to identify rotors as rotational activity around a center, or focal impulses as centrifugal activation from a point of origin, or laminar activation or disorganized patterns that fell outside the three categories previously described. Rotors and focal sources were diagnosed only if stable and sustained throughout several recording epochs (>100 cycles), to exclude transient rotational and/or focal activations that represent fibrillatory conduction.⁵

We observed sustained rotors and focal sources in nearly all patients (98/101, or 98%) with paroxysmal, persistent, and long-standing persistent AF.¹⁶ Subjects exhibited 2.1 ± 1.0 sources concurrently that were more prevalent in individuals with persistent than paroxysmal AF (2.2 ± 1.0 vs. 1.7 ± 0.9; P = .03). AF sources were sustained for hundreds of cycles analyzed in “time lapse” fashion, representing tens of minutes (i.e., thousands of cycles). Sources lay in diverse atrial locations, and it was surprising to note that 24% were found in the right atrium. The prevalence of rotors was greater than the prevalence of focal beats.¹⁴ In patients in whom both spontaneous AF and induced AF were observed, AF propagation movies converged once AF had been sustained for >10 minutes in either case. No complications were noted during mapping.¹⁶

Figure 43-4 illustrates rotors in the (A) right and (B) left atria in patients during AF, showing head-to-tail (red-to-blue) activation in the coherent domain, and peripheral disorganization and/or collision. In each case, electrograms in the coherent domain exhibit sequential 1 : 1 activation.

Figure 43-4 Right Atrial and Left Atrial Spiral Waves (Rotors) During Human AF on FIRM Mapping
A, Counterclockwise rotor in posterolateral right atrium, with collision (white double lines) beyond the coherent domain of the spiral arm. Electrograms around the right atrium rotor site indicate sequential activation of counterclockwise rotation with variability at cycle length ≈177 ms. B, Counterclockwise rotor in midposterior left atrium, with collision beyond the spiral arm (double lines). Pulmonary vein ostia indicated by dashed lines. ABL, Ablation electrogram; AF, atrial fibrillation; CS, coronary sinus.

Figure 43-5, A illustrates a right atrial AF rotor with disorganization peripherally and in the left atrium. Figure 43-5, B illustrates, for comparison, right atrial clockwise reentry in reverse typical atrial flutter, showing 1 : 1 propagation in both atria. Figure 43-5, C shows a repetitive focal source driving AF with peripheral disorganization. For comparison, Figure 43-5, D illustrates the distinct bi-atrial propagation maps of a “simple” focal atrial tachycardia in a patient without AF, in which propagation is 1 : 1 throughout ipsilateral and contralateral atria.

Figure 43-5 AF Sources Are Distinct from Simple Rhythms on FIRM Mapping
A, Clockwise right atrial rotor in AF, showing noncoherent (non-1 : 1) propagation in left atrium. Conversely, (B) clockwise reverse typical flutter with 1 : 1 propagation to the left atrium. C, Repetitive focal beat during paroxysmal AF (in low septal left atrium), with activation to remaining left atrium and fibrillatory conduction to right atrium (CL ≈100 ms). In contrast, (D) focal atrial tachycardia (nonfibrillatory) from the high posterior left atrium differs from AF by showing 1 : 1 activation centrifugally to the ipsilateral then the contralateral atria (CL, 300 ms).

Electrophysiological Evidence That Rotors and Focal Sources Are Primary Sustaining Mechanisms for AF

Two lines of evidence were used to prove that rotors and focal sources are causal mechanisms for human AF and not mere bystanders. First, we carefully mapped activation to show that activation emanates from sustained sources to the remaining atria for each cycle of AF. Second, we targeted rotors and/or focal sources directly for ablation to acutely modulate AF.

Mapping Activation Sequence

Rotors exhibited 1 : 1 activation within their coherent domains, and focal sources showed 1 : 1 activation from their origins, with distal disorganization and/or collision from isochronal analyses (see Figures 43-4 through 43-7) and isopotential movies. Directionality was analyzed to further establish that propagation emanated from rotors or focal sources during AF and not was directed to them. Directionality⁵³ was computed by analyzing isochronal AF snapshots in 10-ms bins, then calculating vectors between successive isochronal contours at the spatial boundaries of each bin. Figure 43-6 shows (A) a left atrial AF rotor with (B) activation emanating to remaining atria, and (C) a left atrial focal AF source with (D) activation emanating to remaining atria. Analysis of consecutive cycles revealed that directionality was conserved close to the center of each source but not farther away. These data support the notion that sustained sources control and drive fibrillatory activity over multiple cycles.

Figure 43-6 Propagation Emanates from Localized Sources to Remaining Atria During Human AF
A, Left atrial AF rotor. B, Directionality shows propagation emanating from rotor to remaining atria. C, Left atrial AF focal source. D, Directionality shows propagation emanating from focal source to remaining atria. The arrows indicate the activation direction ⁵³ between isochrones (color bar).

Figure 43-7 Acute Termination of AF to Sinus Rhythm by Focal Impulse and Rotor Modulation (FIRM) Ablation
A, Left atrial rotor with counterclockwise activation (red to blue) and disorganized right atrium in AF. B, FIRM ablation at left atrium rotor terminated AF to sinus rhythm in <1 min. C, Right atrium rotor (clockwise) and simultaneous left atrium focal source during persistent AF. D, FIRM ablation at right atrium rotor terminated AF to sinus rhythm in 5.5 min. E, Left atrium focal source in AF. F, FIRM ablation at left atrium focal source terminated AF to sinus rhythm. All patients are AF-free on implanted cardiac monitors at >1 y. Scalebar = 1 s. CS, coronary sinus electrogram. Atrial orientations as in Figure 43-3. (Adapted from Narayan et al,¹⁶ with permission.)

Targeted Ablation Limited to Rotors and Focal Sources Acutely Terminates AF

Recent computational advances enabled us to perform FIRM mapping for human AF in near-real time at electrophysiological study. This provided the opportunity to target these regions for ablation prospectively,¹⁶ as illustrated in a recent video case report.⁵¹

Radiofrequency energy was targeted directly at identified rotors and focal sources before PV isolation. Radiofrequency energy was delivered using a 3.5-mm-tip irrigated catheter (Thermocool, Biosense-Webster, Diamond Bar, California) at 25 to 35 W, or, in heart failure subjects, an 8-mm-tip nonirrigated catheter (Blazer, Boston Scientific, Natick, Massachusetts) at 40 to 50 W, target 52° C. Other centers have subsequently employed a wide variety of energy sources for FIRM ablation cases, including cryoenergy. In each case, the catheter is maneuvered near basket electrodes overlying each source using fluoroscopy, then ablation is applied for ≈30 s at each site. The catheter is moved within the area representing the center of rotation or focal origin, for the acute end point of AF termination or ≤10 minutes, whichever comes first. AF termination is followed by attempted non-reinducibility of AF. Only if AF is subsequently non-reinducible do we consider the end point of AF termination achieved, unlike prior reports in which AF reinduction is rarely attempted. If AF did not terminate, we assessed for abruptly increased AF cycle length >10%. In our validation studies, an AF cycle length increase >10% indicates elimination of a secondary AF source, as suggested in computer simulations by others (using a 3% to 4% cutpoint).54,55 Our acute end point was thus the composite of AF termination or ≥10% slowing.

The acute effects of targeted ablation limited to rotor or focal sources were reported in the CONFIRM trial,¹⁶ in patients aged 63 ± 9 years, of whom 81% had persistent AF, and in an initial report from first cases at other centers,¹⁵ in patients aged 58 ± 12 years, of whom 92% had persistent AF.

FIRM ablation achieved the acute end point in 86% (31/36) of patients receiving FIRM-guided ablation in the CONFIRM trial ¹⁶ and in 100% (12/12) of patients in the first external series.¹⁵ Figure 43-7 shows AF termination by FIRM ablation at (A) a posterior left atrial rotor, (B) a right atrial rotor, and (C) a left atrial focal beat. Preliminary findings from a larger multicenter FIRM-guided ablation experience show similar results (manuscript in preparation).

The AF termination end point alone was observed in 56%¹⁶ and 67%¹⁵ of patients before PV isolation after a median of <5 min FIRM ablation at the primary source. In the CONFIRM trial, this was achieved with a median of 2.5 min ablation at the primary source (interquartile range [IQR], 1 to 3 min).¹⁶ Moreover, AF termination by FIRM was predominantly to sinus rhythm (23/28 terminations in pooled studies; 82%^15,16), unlike conventional (nonguided) ablation, in which AF terminates typically to an atrial tachycardia (<13% to sinus rhythm⁵⁵). Notably, AF sources were stable over time in each patient. AF sources were persistent over the time scale required for mapping and ablation (mean, 115 ± 57 minutes). AF sources were conserved for months in a subset of subjects who failed conventional ablation in the control limb of CONFIRM, then re-presented for FIRM-guided ablation.¹⁶

Discussion

These data demonstrate that human AF is sustained by localized rotors and focal sources in patients with paroxysmal, persistent, and long-standing persistent AF. Notably, localized sources were observed in patients undergoing first AF ablation and in those with recurrent AF after prior ablation. The observation that brief targeted ablation (FIRM) limited to rotors or focal sites was able to terminate AF and render it noninducible, before anatomically based ablation of PV triggers, provides strong evidence for the mechanistic role of patient-specific AF sources in maintaining the arrhythmia.

Our finding that human AF is sustained by a small number (1 to 3) of rotors and focal sources that are spatially stable over time ¹⁰ was initially surprising and differed from previous studies. A detailed recent epicardial mapping study was interpreted to show that AF is maintained by numerous complex nonrepeating activation fronts,³⁶ yet that study used limited field-of-view (<20% of dilated atrial surfaces) that may miss AF sources outside the mapped field, and did not prove that the disordered waves were causal rather than bystanders from unmapped and unidentified mechanisms.

Our results extend elegant prior studies suggesting rotors in human AF. Atienza et al reported regions of high spectral dominant frequency near the pulmonary veins in patients with paroxysmal AF whose frequency increased after adenosine administration,²⁶ consistent with a reentrant mechanism. Using noninvasive body surface mapping (ECG imaging [ECGI]), Cuculich et al¹³ detected rotational activity in AF patients, but in a minority of patients, typically of only 1 rotation, and with very short cycle length (<75 ms).¹³ These differences, which may reflect the respective mapping approaches, require further definition.

Clinical Implications

The presence of a small number of stable and temporally conserved rotors or focal sources in any given patient with AF is consistent with many mapping studies, and suggests a novel approach to classifying AF and tracking the natural history of AF—by noting an increase or a decrease in the number of sources. The precise pathophysiology that underlies the formation of AF sources requires further study, which may lead to future diagnostic or therapeutic applications. However, the most immediate therapeutic application is the direct elimination of AF sources. Accordingly, we hypothesized that targeted FIRM ablation at patient-specific AF rotors and focal sources would improve long-term freedom from AF compared with conventional ablation.

Ablation of Localized Sources—the CONFIRM Trial

The CONFIRM trial (CONventional ablation with or without Focal Impulse and Rotor Modulation) was a prospective case cohort study ¹⁶ that enrolled pateints at 107 AF ablation procedures who had failed at least one antiarrhythmic medication with standard indications. Of this population, 36 underwent FIRM-guided ablation then conventional ablation, and 71 underwent conventional ablation alone (1 : 2 allocation). The only exclusion criterion was an unwillingness to consent. Patients had paroxysmal AF (episodes that self-terminate in <7 days), persistent AF (episodes that last >7 days and require electrical cardioversion), and long-standing persistent AF (persistent AF with >1 year of continuous AF).² Overall, three-quarters of patients had persistent or long-standing persistent AF (conventional, 66%; FIRM-guided, 81%), and subjects had a wide range of ages (20 to 81 years), left ventricular ejection fractions (20% to 75%), and comorbidities.

FIRM was performed in all patients with spontaneous or induced AF (n = 101, including all FIRM-guided patients). As has been noted, maps of spontaneous and induced AF converge by 5 to 10 min. FIRM-guided ablation was performed as described for the acute end point of AF termination or AF cycle length increase >10% by FIRM ablation of ≤10 min per source. When AF terminated, we attempted to reinitiate AF; if successful, FIRM was repeated for ≤3 sources (≤30 min), but only 1 FIRM map was generated in the CONFIRM trial. In more recent series, repeat mapping is possible.

Conventional ablation was then performed: in FIRM-blinded subjects, immediately after basket data were acquired, and in FIRM-guided subjects, after FIRM-guided ablation was completed. Conventional ablation ² comprised wide area circumferential ablation to isolate the left and right pulmonary veins in pairs, with verification of pulmonary vein isolation using a circular mapping catheter.

Patients were evaluated after the single index procedure in clinic quarterly for up to 2 years. Repeat ablation was not permitted at any time, even in the first 3 months (“blanking” period), and antiarrhythmic medications were discontinued at 3 months. Recurrent arrhythmias were detected using implanted subcutaneous ECG monitors whenever possible—Reveal XT™ (Medtronic, Minneapolis, Minnesota), or clinically indicated pacemakers/defibrillators with AF detection algorithms. Remaining subjects received intermittent ambulatory ECG monitoring at each visit.

CONFIRM Trial Results

The single-procedure freedom from AF was 82.4% in FIRM-guided patients compared with 44.9% in FIRM-blinded patients (P < .001) after 273 days (median IQR, 132 to 681). Notably, no FIRM-guided case recurred after ≈7 months. FIRM-guided therapy maintained its benefit over FIRM-blinded therapy for first-time ablation cases (P < .001; Figure 43-8 for patients off antiarrhythmic medications) and for all prespecified subgroups.

Figure 43-8 Cumulative Freedom from Atrial Fibrillation in Patients Off Antiarrhythmic Medications
For all cases (bold lines), and for those at first ablation (dashed lines). Intention-to-treat analysis and P values reflect the complete follow-up period. (Adapted from Narayan et al,¹⁶ with permission.)

FIRM could not be completed in a small number of patients because of esophageal heating, phrenic nerve capture, or other factors; these patients were included as failures in the intention-to-treat results. Moreover, FIRM-guided patients had a greater number of comorbidities and more rigorous follow-up than FIRM-blinded patients (subcutaneous ECG monitors in 88.2% vs. 26.1%; P < .001), suggesting that the results may actually understate the relative benefits of FIRM-guided ablation. Patient safety was equivalent, with no difference in adverse events between groups. Finally, the acute and chronic end point was achieved in patients with all forms of AF (paroxysmal, persistent, and long-standing persistent AF). However, complete FIRM mapping was not possible in patients with atria larger than the largest basket diameter (55 to 60 mm), who were less likely to achieve the acute end point.

Discussion of CONFIRM Trial

The CONFIRM trial showed for the first time that brief ablation (FIRM) limited to patient-specific AF-maintaining sources was able to acutely terminate persistent or paroxysmal AF before conventional ablation in a majority of patients, and, when followed by conventional ablation, it substantially increased long-term AF elimination after a single procedure compared with conventional AF ablation alone.

The CONFIRM trial is notable for many reasons. First, it reports among the highest single procedure success rate of any AF ablation trial to date, verified using implanted ECG monitors that are far more sensitive than symptoms or the intermittent ECG monitoring used in most prior ablation trials. Second, the CONFIRM trial is one of the few AF trials to date 9,56,57 to compare a novel ablation strategy with a control strategy of conventional ablation⁵⁸ (not to antiarrhythmic medications). Third, CONFIRM is the first trial to identify and directly target demonstrated AF mechanisms, proven by achievement of the acute end point in a patient-tailored fashion. Fourth, FIRM-guided ablation was more effective than conventional ablation in all patient subgroups, including those undergoing their first procedure and those with AF despite prior conventional ablation (see Figure 43-8).

Results of the CONFIRM trial may explain many observations from conventional AF ablation. First, most sources lay in the left atrium. Nevertheless, the presence of right atrial sources in one-quarter of patients may explain the 70% to 80% success ceiling ² of conventional (mostly left atrial) ablation for AF, and the need for right atrial ablation in some patients.³² Second, diverse source locations are consistent with reports that widespread ablation may be required.⁸ Third, the higher number of sources in persistent than in paroxysmal AF is consistent with lower success rates and more difficult procedures in persistent AF patients. Fourth, how widespread conventional ablation can miss localized sources may be explained by AF source locations often remote from typical lesion sets, and the fact that conventional ablation may dwell briefly at each location (≈30-60 s) before the catheter is moved, but FIRM targets sources for >≈2.5 min of ablation (similar to ablation for AV node reentry, atrial tachycardias, or other “simple” arrhythmias^17–19).

CONFIRM has several limitations. The first is its nonrandomized design, although subjects were enrolled consecutively and were treated prospectively for prespecified end points. In fact, FIRM-guided subjects had a higher prevalence of persistent AF, a greater number of comorbidities, and more intense monitoring than FIRM-blinded subjects, thus actually biasing the study against and likely underestimating the benefit of FIRM-guided ablation. Second, the maximal current basket size (55 to 60 mm) places an upper limit on patients in whom complete atrial mapping is achievable. Although the basket has suboptimal resolution, the theoretical design considerations above and the size of single-ablation lesions (≈5 to 7-mm diameter) may limit the need for substantially higher resolution. Third, a randomized controlled trial with a greater representation of female subjects is required and is already under way. Fourth, future studies are required to define the benefit of mechanistically targeted FIRM ablation alone, without ablation at AF trigger sites; such studies are also under way.