Skip to main content
Skip main navigationClose Drawer MenuOpen Drawer Menu

Central Illustration



The aim of this study was to demonstrate that a stochastic vector-based mapping approach could guide ablation of atrial fibrillation (AF) drivers as evidenced by ablation response and long-term follow-up outcomes.


The optimal method for mapping and ablation of AF drivers is yet to be defined.


Patients undergoing persistent AF ablation were recruited. Patients underwent pulmonary vein isolation (PVI) with further ablation guided by the stochastic trajectory analysis of ranked signals (STAR) mapping method. The proportion of the time an electrode’s activation was seen to precede its neighboring electrodes activation was used to determine early sites of activation (ESA). A positive ablation response at ESA was defined as AF termination or cycle length slowing of ≥30 ms. Clinical outcome was defined as recurrence of AF/atrial tachycardia (AT) during a follow-up of 12 months.


Thirty-five patients were included (AF duration of 14.4 ± 5.3 months). After PVI, an average of 2.6 ± 0.8 ESA were ablated per patient with study-defined ablation response achieved in all patients. Of the 86 STAR maps created post-PVI, the same ESA was identified on 73.8 ± 26.1% of maps. ESA that resulted in AF termination were more likely to be identified on both pre- and post-PVI maps than those associated with cycle length slowing (23 of 24 vs. 16 of 49; p < 0.001). During a follow-up of 18.5 ± 3.7 months, 28 (80%) patients were free from AF/AT.


The ablation response at ESA suggests they may be drivers of AF. Ablation guided by STAR mapping produced a favorable clinical outcome and warrants testing through a randomized controlled trial. (Identification, Electro-mechanical Characterisation and Ablation of Driver Regions in Persistent Atrial Fibrillation [STAR MAPPING]; NCT02950844)


The mechanisms of persistent atrial fibrillation (AF) remain unclear. Recent mapping studies have demonstrated high rates of AF termination when targeting localized sources (1–4). However, these successes have not always been replicated (1,5,6), suggesting that these techniques may be dependent on operator knowledge and skill. The optimal way to identify drivers for persistent AF remains unclear.

Recent data suggest that drivers may be spatially conserved but with temporal periodicity (2,4). Because of the chaotic nature of AF, it is not feasible to determine the earliest site of activation (ESA) in relation to a fixed reference point. Stochastic trajectory analysis of ranked signals (STAR) is a novel-mapping method for identifying localized sources that potentially play a mechanistic role in maintaining AF. The principle of STAR mapping is to use data on multiple individual wavefront trajectories to identify atrial regions that most often precede activation of neighboring areas with the aim of identifying intermittent drivers (7). The STAR mapping method has been validated in vivo using optical mapping of calcium transit in spontaneous fibrillating HL1 cells and in human studies mapping atrial tachycardias (ATs) (7).

The study’s aim was to prospectively target sites identified by STAR mapping as relevant to the maintenance of persistent AF. We sought to determine the temporal and anatomical stability of these sites identified as potential drivers across multiple maps. Clinical relevance of these sites was assessed by the immediate ablation response and the freedom from AF/AT during 12 months of follow-up.


Patients with persistent AF were included (AF duration <24 months and no previous AF ablation) (Online Appendix). All patients provided informed consent for their study participation. Ethical approval was granted by the United Kingdom National Research Ethics System (16/LO/1379). The study was prospectively registered on ( NCT02950844).

Electrophysiology study

Procedures were performed using CARTO mapping system (Biosense Webster, Diamond Bar, California) (Online Appendix). Right atrial (RA) and left atrial (LA) geometries were created with detailed bipolar voltage maps using a 2-6-2 mm PentaRay NAV catheter (Biosense Webster) with a color fill threshold of 5 mm. Basket catheters (Constellation, Boston Scientific, Natick, Massachusetts; and FIRMap, Abbott Vascular, Santa Clara, California) were used to record unipolar signals for 5 min pre- and post-pulmonary vein isolation (PVI). The basket catheter was sized and positioned as previously described (8). A minimum of 2 recordings of 5 min each was performed at each stage. If coverage was limited during these recordings, additional recordings were taken following catheter manipulation. Unipolar signals were recorded using the Bard Labsystem Pro Electrophysiology recording system (Boston Scientific). A locational point was taken on the CARTO system at the start of each recording segment and every 30 s for the duration of each 5-min recording to obtain locational coordinates for the basket catheter electrode poles and geometry coordinates. A decapolar catheter (Biosense Webster) positioned in the inferior vena cava was used as an internal indifferent electrode reference. Unipolar signals were filtered with a bandpass filter of 0.5 to 500 Hz.

The STAR method imported electrogram signals from Bard and corresponding chamber geometry (vertices and polygons coordinates) and catheter/electrode location data (xyz coordinates) from CARTO. These were processed by a MATLAB custom written script (MATLAB 2017b, MathWorks, Natick, Massachusetts). The STAR maps created then allow ESA to be identified on a replica CARTO atrial geometry. For a leading site to be classified as an ESA, it was required to be leading for at least 75% of the time.

Initial mapping and ablation was performed in the LA, and RA mapping was only performed if the ESA was considered likely to be within the RA (coronary sinus activation predominantly proximal to distal and LA ESA at the LA septum).

Ablation strategy

All patients had PVI performed by wide area circumferential ablation (WACA) using radiofrequency ablation applied with a SMARTTOUCH catheter (Biosense Webster) (see the Online Appendix). The post-PVI STAR maps were then used to guide further ablation at the identified ESA. A lesion was delivered at the center of the driver site with further ablation surrounding the initial lesion in a cluster, avoiding the creation of linear lesions so as not to affect any AF mechanisms in this way. Ablation at driver sites was delivered with a contact force of 5g to 40g, with a power of 30 to 40 W (30 W posteriorly and 40 W anteriorly). The CARTO VISITAG size used was 3 mm.

AF cycle length (CL) was measured over 30 beats from a 2-6-2 mm spacing PentaRay NAV catheter positioned in the LA appendage. AF CL pre- and post-ESA ablation was used to assess ablation response. We opted to use the electrograms within the LA appendage to monitor AF CL as these were in the majority of cases distinct and of high amplitude, making CL measurements easier, which is consistent with the procedure used by other studies (9,10). All ESA on a STAR map were targeted in order of ranked priority whereby sites that were leading 100% of the time being targeted first, followed by 90%, and so on. If multiple ESA with the same ranked priority were identified, what order to ablate the ESA was left to the operator’s discretion.

Ablation at a driver site was stopped in the following situations: a total of 5 min of ablation had been performed at an ESA including consolidating ablation lesions; no signal remained at the ablation site; or a study-defined ablation response had been achieved. A study-defined ablation response was either AF termination or CL slowing of ≥30 ms. Although small changes in CL (usually 5 to 6 ms) have been used previously to determine an ablation response (9,11), it was thought that ablation of a clear driver ought to have a more substantial effect, therefore a CL slowing of ≥30 ms was used.

If AF terminated but other ESA had not been ablated, these sites were not empirically targeted. Beyond targeting ESA, no other empirical ablation was allowed including the creation of lines. If AF organized into an AT, this was mapped and ablated during the procedure. Direct current cardioversion was performed at the end of the procedure if AF did not terminate following ablation of all identified ESA.

Ablation of ESA was dependent on the operator’s interpretation of the location of these sites on the LA geometry. A blinded observer reviewed all STAR maps to determine whether the maps identified the same anatomical location for the ESA ablated as was indicated by the operator.


All patients underwent clinical follow-up at 3, 6, and 12 months, with 48-h ambulatory Holter monitoring at 6 and 12 months. At the clinician’s discretion and patient’s preference, patients could be followed up beyond the 12-month period. A 3-month “blanking period” was observed, with all medication including antiarrhythmic drugs continued during this time. Clinical success was defined as freedom from AF/AT lasting >30 s off antiarrhythmic drugs as per consensus recommendations (12).

STAR mapping principals

The principle of STAR mapping is to use data from multiple individual wavefront trajectories to identify regions of the atrium that most often precede activation of neighboring areas. By gathering data from many hundreds of activations, a statistical model can be formed. This permits regions of the atrium to be ranked according to the amount of time that activations precede those of adjacent regions.

Electrode pairing and geodesic distance

In addition to statistical analysis of multiple activations, the geodesic distance between compared regions was limited (7). The intention was to minimize the chance of ascribing a relationship between 2 regions activated by unrelated wavefronts, and instead assign relationships only to true sequential activations. Unipolar activation timing was taken as the maximum negative deflection (peak negative dv/dt).

Filtering using refractory periods was performed with conservative values for a minimum atrial refractory period used (70 ms) (13,14). If an activation fell within this nominated refractory period it was not considered representative of a separate wavefront and was rejected by the mapping method. Electrode timing relationships that were implausible due to conduction velocity restraints were discarded (Online Figures 1A and 1B).

Creating and interpreting a STAR map

Color-coded electrode positions were projected on to patient’s LA geometry representation. Colors represented the proportion time spent leading in relation to the other paired electrodes. Further visual cues highlighting the importance of sites with high-leading proportions included size scaling of electrode sites and arrowhead lines highlighting wavefront direction.

Offline analysis

The unipolar signals recorded pre-PVI was processed offline and these STAR maps were compared with those created post-PVI to assess the stability of the ESA and the impact of PVI. Pre-PVI STAR maps were also reviewed to determine whether ESA were identified near the PV at sites of the WACA line. All the post-PVI STAR maps created in each patient were also compared to determine whether the ESA were consistent between maps.

The ESA identified by STAR mapping and ablation sites were co-registered offline. A custom written MATLAB script, using locational data for the ablation lesions and ESA, was used to ensure the ablation sites were within 1 cm of the identified ESA on the STAR map. The relationships between ESA where a study-defined ablation response was achieved and low voltage zones (LVZ) were assessed. LVZ were defined as areas with a bipolar voltage of <0.5 mV (15).

Duration of recordings and consistency of data over time

Post hoc offline analysis of shorter time windows was performed. The 5 min of unipolar electrogram recording collected post-PVI was split into 10 unipolar recordings of 30-s duration, and each unipolar recording was used to create separate consecutive STAR maps. ESA identified on longer recordings and targeted prospectively with ablation were also assessed on each individual 30-s segment to assess their consistency and to determine boundaries for recording time.

Statistical analysis

Statistical analyses were performed using SPSS (IBM SPSS Statistics, version 24, IBM Corp., Armonk, New York). Continuous variables are displayed as mean ± SD or median (interquartile range). Categorical variables are presented as a number and proportion. Chi-square test was used for the comparison of nominal variables. Student’s t-test, or when appropriate, its nonparametric equivalent, Mann-Whitney U test, was used for comparison of continuous variables. A p value of <0.05 was deemed significant.


Thirty-five patients were included in the study. Baseline characteristics are shown in Table 1. The mean age was 60.9 ± 9.4 years with 24 patients (68.6%) being male. The mean AF duration was 14.4 ± 5.3 months (n = 7 AF duration <12 months and n = 28 AF duration >12 months) and 23 of 35 patients (65.7%) were on an antiarrhythmic drug pre-ablation. All patients were in AF at the commencement of the procedure.

Table 1. Baseline Characteristics (N = 35)

Age, yrs60.9 ± 9.4
Male24 (68.6)
Diabetes mellitus0
Hypertension11 (31.4)
TIA/CVA1 (2.9)
Ischemic heart disease2 (5.7)
Cardiac surgery1 (2.9)
Left ventricular EF ≥55%31 (88.6)
LA size, cm2
20–3027 (77.1)
31–408 (22.9)
LA volume, ml56.8 ± 8.1
AF duration, months14.4 ± 5.3
CHA2DS2-VASc score1.1 ± 1.3
Previous AT ablation
Cavo-tricuspid isthmus-dependent flutter2 (5.7)
Current medical strategy
Beta-blockers including sotalol17 (48.6)
Amiodarone21 (60.0)
Flecainide1 (2.9)

Values are mean ± SD or n (%).

AF = atrial fibrillation; AT = atrial tachycardia; CHA2DS2-VASc = Congestive Heart Failure, Hypertension, Age ≥75 Years, Diabetes Mellitus, Prior Stroke or Transient Ischemic Attack or Thromboembolism, Vascular Disease, Age 65 to 74 Years, Sex; CVA = cerebrovascular accident; EF = ejection fraction; LA = left atrial; TIA = transient ischemic event.

Procedure-related data

Twenty-one procedures were performed under local anesthetic and conscious sedation (71.4%), with the remainder being under general anesthetic. The mean procedure time was 225.4 ± 65.6 min (Online Table 1). The total procedural radiofrequency time while in AF was 54.1 ± 7.2 min. The total radiofrequency time to achieve sinus rhythm with ablation (including PVI, ESA ablation + AT ablation) was 63.5 ± 10.2 min. The average radiofrequency time at ESA that resulted in AF termination was 2.6 ± 0.6 min. The mean fluoroscopy time was 1.9 ± 3.2 min. One patient experienced cardiac tamponade, which was noted at the end of the procedure, that required pericardiocentesis. No other complications were encountered. A total of 170 STAR maps were created pre-PVI (84 maps) and post-PVI (86 maps), with an average of 2.4 ± 0.6 maps per patient at each stage.

Procedural outcomes are summarized in the CONSORT (Identification, Electro-mechanical Characterisation And Ablation of Driver Regions in Persistent Atrial Fibrillation) flow diagram (Figure 1).

Figure 1.
Figure 1.

CONSORT Flow Chart

The diagram summarizes the CONSORT (Identification, Electro-mechanical Characterisation And Ablation of Driver Regions in Persistent Atrial Fibrillation) procedural outcomes. AF = atrial fibrillation; AT = atrial tachycardia; CL = cycle length; ESA = early sites of activation; Pts = patients; PVI = pulmonary vein isolation; SR = sinus rhythm.


Of the 35 patients, 3 patients terminated to sinus rhythm on PVI, and post-PVI ESA were not identified or targeted in these patients. A total of 92 ESA were identified in the remaining 32 patients post-PVI, 83 ESA (90.2%) were targeted with ablation. The mean surface area ablated per ESA was 1.5 ± 0.3 cm2. The total surface area ablated for all the ESA that were ablated on a per-patient basis was 3.4 ± 0.6 cm2. The 9 ESA that were not ablated were in patients who had ablation at a previous site that resulted in AF termination. An average of 2.6 ± 0.8 ESA were ablated per patient. An ablation response that met the study criteria was seen with 73 ESA (88.0%) with an average of 2.3 ± 0.6 sites per patient (Central Illustration, Figures 2, 3, and 4, Online Figure 2). For the 10 ESA where a study-defined ablation response was not demonstrated, the mean CL slowing was 20 ± 0.7 ms.

Central Illustration.
Central Illustration.

Patient #6

(A) A stochastic trajectory analysis of ranked signals (STAR) map in an anterior-posterior (AP) view highlights 2 early sites of activation (ESA) at the (1) low anterior wall and (2) high anterior wall bordering roof. Unipolar electrograms at each ESA demonstrate that the electrode at the ESA (star) is leading its neighboring electrodes (arrows) as assigned by the STAR mapping method. (B) Ablation for 3.0 min resulted in cycle length slowing of ≥30 ms with site 1 and ablation for 1.9 min resulted in AF termination to sinus rhythm with site 2 as highlighted on the (C) electrograms obtained from Bard. AP = anterior-posterior; LAA = left atrial appendage; LUPV = left upper pulmonary vein; MVA = mitral valve annulus; RUPV = right upper pulmonary vein; STAR = stochastic trajectory analysis of ranked signals.

Figure 2.
Figure 2.

Patient #15 STAR and CARTO Maps

(A) A stochastic trajectory analysis of ranked signals (STAR) map of the left atrium in a posterior-anterior (PA) view shows an early site of activation (ESA) on the posterior-inferior wall (1). Unipolar electrograms at the ESA demonstrate that the electrode at the ESA (star) is leading its neighboring electrodes (arrows) as assigned by the STAR mapping method. (B) CARTO left atrial geometry shows that 1.8 min of ablation resulted in (C) organization of atrial fibrillation to atrial tachycardia as shown on the electrograms obtained from Bard. This was mapped to a roof-dependent flutter and effectively ablated to sinus rhythm with a roof line. LUPV = left upper pulmonary vein; RUPV = right upper pulmonary vein.

Figure 3.
Figure 3.

Patient #21 STAR and CARTO Maps

(A) A global STAR left atrial map in an anterior-posterior (AP) view highlights an ESA along the anteroseptum (B) where ablation for 2.8 min as shown on the CARTO left atrial map in an AP view resulted in (C) termination of atrial fibrillation to sinus rhythm as shown on the Bard electrograms. LAA = left atrial appendage; MVA = mitral valve annulus; other abbreviations as in Figures 1 and 2.

Figure 4.
Figure 4.

Patient #27 STAR and CARTO maps

(A) A STAR map in an AP view highlights an ESA along the anteroseptum (B) where ablation for 1.8 min as shown on the CARTO left atrial map in an AP view resulted in (C) termination of atrial fibrillation to atrial tachycardia as shown on the Bard electrograms. Abbreviations as in Figures 1, 2, and 3.

The ablation response achieved on a per-patient basis is shown in Table 2. On a per-ESA basis, AF termination was achieved with 24 sites (18 organizations to AT and 6 terminations to sinus rhythm) and CL slowing of ≥30 ms was achieved with 49 sites (Online Table 2). The anatomical distribution of the ESA and their responses to ablation is summarized in Figure 5. All AT were effectively ablated to achieve sinus rhythm at the end of the procedure. The mechanism of 2 AT was focal/micro-re-entrant that was mapped to the ablation site of the ESA, which resulted in AF termination (Online Table 2). Eight patients who remained in AF underwent direct current cardioversion at the end of the procedure. Eight of the 32 patients met the study criteria to undergo RA mapping. In these 8 patients, 3 ESA were identified (2 septal RA and 1 lateral RA) in 3 patients. Ablation of all sites resulted in CL slowing of ≥30 ms (Online Table 2, Online Figure 2).

Table 2. Ablation Response Seen in the Cohort of Patients and the Mechanism of the AT Mapped

Ablation response32
Termination to sinus rhythm6
Organization to AT18
CL slowing ≥30 ms8
AT mapped18
Mitral isthmus-dependent flutter5
Roof-dependent flutter7
Cavo-tricuspid isthmus-dependent flutter3
LA anteroseptal1
Branch of the coronary sinus1
LA roof1

Values are n.

CL = cycle length; other abbreviations as in Table 1.

Figure 5.
Figure 5.

Anatomical Distribution of ESA Using the STAR Mapping

The AP (Ai) and PA (Aii) views of the anatomical distribution of ESA identified in the left atrium using STAR mapping. Red circles highlight sites that resulted in cycle length slowing (SCL) of ≥30 ms. Blue triangles highlight sites that resulted in AF termination into either atrial tachycardia or sinus rhythm. Orange rectangles highlight sites that resulted in no effect that met the study-defined ablation response criteria. Black circles indicate sites that were not targeted with ablation. Three ESA were identified in the right atrium. Abbreviations as in Figures 1, 2, and 3.

At least 1 ESA was identified in every patient, and at least 1 ESA was identified on 81 of the 86 STAR maps created post-PVI (94.2%). On a per-patient basis, these sites showed consistency whereby they were identified on average on 73.8 ± 26.1% of the STAR maps created in each patient. Of the 92 ESA identified, a majority of the ESA were leading 100% of the time (86 of 92, 93.5%). Out of the remaining 6 ESA, 1 was leading 90% of the time, 3 were leading 85% of the time, and 2 were leading 75% of the time. There was no difference in the AF termination rates with ESA that were ablated and were leading 100% of the time when compared with those that were leading <100% of the time (22 of 77 [28.6%] vs. 2 of 6 [33%]; p = 1.00).

The blinded observer agreed with the operator in regard to the anatomical location of 81 of the 83 ablated ESA (97.6%). With the 2 ESA for which there was disagreement, with 1 the operator identified it to the roof whereas the blinded observer identified it to the anterior wall, and with the other ESA, the operator identified it to the septum whereas the blinded observer identified it to the anteroseptum.

Of the 32 patients that underwent STAR mapping–guided ablation, 21 patients (65.6%) were on an antiarrhythmic drug. There was no significant difference in the ESA detection (2.9 ± 1.0 vs. 2.8 ± 0.7 ESA per patient; p = 0.9) or AF termination rates on ablation of ESA (17 of 21 vs. 7 of 11; p = 0.40) in those on an antiarrhythmic drug versus those that were not on an antiarrhythmic drug.

Offline analysis

In the 3 patients who had AF, termination was achieved during PVI; ESA were identified at the PV ostium on the site of the WACA lines; and these ESA were collocated at the sites where AF terminated during PVI. Of the 73 ESA where a study-defined ablation response was achieved, 39 were identified on the pre-PVI STAR maps (53.4%). The sites that were identified on both pre- and post-PVI STAR maps were more likely to be associated with AF termination than CL slowing (23 of 24 [95.8%] vs. 16 of 49 [32.7%]; p < 0.001).

All lesions on CARTO (Biosense Webster) were within 1 cm of targeted ESA (mean distance 0.45 ± 0.28 cm). Of 73 ESA where a study-defined ablation response was achieved, 52 (71.2%) colocated to sites of LVZ.

The 5-min STAR map recordings consisted in total of 860 segments of 30-s duration. Of the 92 ESA seen, 80 (87.0%) were consistently identified across the 30-s segments. All of the 24 ESA that resulted in AF termination were seen in all the segmented recordings.

Follow-up data

During an average follow-up of 18.5 ± 3.7 months (all patients reached ≥12 months of follow-up), 28 patients (80%) were free from AF/AT and off antiarrhythmic drugs. Of the 7 patients that had recurrent AF/AT during follow-up, 6 had documented AT of which all underwent AT ablation. The AT mapped and ablated in these patients included 2 roof-dependent flutters, 3 mitral isthmus–dependent flutters, 1 cavotricuspid isthmus–dependent flutter, 2 focal/micro-re-entrant AT (1 mapped to the WACA line and terminated during reisolation of the PV and the other was mapped to the anterior wall where previous ablation during the AF procedure had resulted in AF termination). Two of the 6 patients had 2 AT mapped and ablated during the same procedure. During an average follow-up of 11.5 ± 3.4 months after a second procedure, all patients remained free from AF/AT. One patient remained on an antiarrhythmic drug after a second procedure.


This study has demonstrated that ESA can be identified during persistent AF using a novel mapping method. Radiofrequency ablation at these ESA resulted in a high proportion of AF slowing or termination, suggesting that these sites may represent AF drivers. Furthermore, a high proportion of patients remained free from AT/AF during long-term follow-up. ESA showed temporal consistency, being seen on most STAR maps even if shorter 30-s segments were analyzed. Therefore shorter duration of unipolar recordings may suffice to allow the identification of ESA that are potentially mechanistically important.

The ESA showed reproducibility and were identified on a majority of the STAR maps created in each patient. AF drivers have been shown to not be temporally stable (2,4,16), which can have an impact on their reproducibility during electrical recordings (4,16). The ability of a majority of STAR maps to identify the same ESA is reassuring and its reproducibility rate is consistent using another mapping system (4,16).

Localized drivers playing a mechanistic role in maintaining AF have previously been proposed (17–19). On these grounds, several mapping systems have been created to enable the mapping of these drivers (1,2,4,16,20). The FIRMap mapping system has shown good success rates and freedom from AF both in the short term and during extended follow-up (1,21). However, the evidence of benefit with regard to clinical outcomes is conflicting (5,6), which may explain why it has, as yet, not gained widespread clinical acceptance. The ECGI mapping system (CardioInsight, NonInvasive 3D mapping system, Medtronic, Minneapolis, Minnesota) has also suggested localized drivers exist in persistent AF, again demonstrating high success rates of freedom from AT/AF (2). However, the recent AFACART (Noninvasive Mapping Before Ablation for Atrial Fibrillation) study reported high rates of AT during follow-up when using the ECGI system in a multicenter trial (22). Likewise mapping using the CARTOFINDER system (Biosense Webster) has demonstrated similar driver patterns with an immediate response to ablation, although long-term follow-up data are lacking (4,16). Thus far, an optimal technique to identify localized drivers in persistent AF has not been established. STAR mapping, presented here, identifies sites demonstrating earlier activation than neighboring sites within a specified geodesic distance. These are localized using a replica of the 3-dimensional (3D) geometry created using CARTO. This thereby aids operators when they navigate to ablation target sites. This is in contrast to the Topera mapping system (Abbott, Menlo Park, California) (1) that relies on a 2D animation of the geometry, or the ECGI mapping system (2) that uses a geometry obtained from computed tomography imaging data. CARTO has recently produced CARTOFINDER, which times electrograms relative to each other in a 250-ms window that then moves through a continuous recording to show wavefront movement over time. This has enjoyed some success demonstrating focal and rotational activations in AF, and an automated approach to their detection has been developed (4,16,23). The Topera and ECGI mapping systems rely on phase mapping to identify drivers. These technologies have also enjoyed some success, although there has also been concern that phase mapping could demonstrate rotors incorrectly at times (17). In contrast, the STAR mapping method utilizes minimal computation or data manipulation and simply compares activation times across electrode pairs to identify sites that are most often leading compared with neighboring sites. Furthermore, all currently available mapping systems require a degree of interpretation by the operator, such as for analysis of phase maps with the ECGI, or analysis of dynamic wavefront activation maps with CARTOFINDER. In contrast, the STAR mapping method is not dependent on the analysis of dynamic wavefront maps but works through highlighting the ablation target sites by a colored circle (circle color being representative of the time an electrode is leading relative to its pairs). The STAR mapping methodology therefore has some potential advantages for mapping in AF. In this study, ESA on the STAR maps correlated well to the ablation lesions on CARTO, emphasizing the good anatomical correlation of STAR maps that permits targeted ablation. The STAR mapping method identified 91 ESA of which 83 were ablated. The response of 73 of these strongly implies a mechanistic role of such sites. The high rate of AF termination, which is compatible to that reported in other studies using alternative mapping systems (2,22) further emphasized the mechanistic role of these sites. During a minimum follow-up of 12 months, 80% of patients were free from arrhythmia and off antiarrhythmic drugs. Furthermore, the patients had an average AF duration of more than 14 months with 31.4% having persistent AF and 68.6% having long-standing persistent AF (2). It is well established that patients with a longer AF duration have poorer outcomes (24), and this clinical result can be seen as highly promising for this cohort.

Localized drivers that demonstrate greater consistency may be of greater mechanistic importance (16). This was observed in the current study: the majority of the ESA, where ablation resulted in the study’s pre-defined electrophysiological endpoints, were consistently identified on the majority of STAR maps created in each patient. Furthermore, as shown in our previous work (4,16), ESA that were identified pre-PVI were also more likely to be associated with AF termination rather than CL slowing. This raises the question whether these sites could be targeted pre-PVI and result in the same clinical endpoints. However, it has been shown that PVI has the ability to eliminate wavebreaks and organize wavefronts to form re-entry (25), raising the question as to whether PVI contributes to the formation of these drivers. Furthermore, high rates of AT were observed in the AFACART study (22) whereby localized drivers were targeted pre-PVI with the aim of terminating AF. Targeting drivers post-PVI may therefore limit ablation outside the PV, although additional mapping prior to PVI may also be helpful to elicit the hierarchy in which localized drivers should be targeted. Furthermore, as we have previously shown, PVI does not affect driver characteristics proposing that the ablation response elicited at the ESA is independent of the PV. Nevertheless, we target ESA 20 min post-PVI so that no delayed effect of ablation would affect rhythm or CL during ESA ablation. In this study we avoided empirically targeting ESA. As this is the first study evaluating the STAR mapping method we believe further characterization of these ESA is required before proposing the empirical targeting of these sites. Having not performed empirical targeting of ESA this still allowed us to obtain a high freedom from AF/AT during a long follow-up period. With the conflicting evidence with regard to the mapping and ablation of localized drivers in AF and the lack of consensus of the optimal way to target localized drivers, we believe alternative mapping and ablation strategies need to be evaluated to allow the identification of an optimal mapping and ablation strategy for persistent AF.

In this study we effectively mapped ESA using whole-chamber basket catheters. Even though mapping with basket catheters can result in limited coverage and contact, our group has previously shown that these parameters can effectively be improved through accurate basket catheter sizing and catheter positioning (8). Basket catheters are limited with regard to the resolution achieved due to the electrode spacing. However, we aimed to overcome this through repositioning the basket catheter and performing a minimum of 2 recordings in different positions in each patient. This allowed unmapped areas to be effectively mapped. We have previously demonstrated the effective mapping of localized drivers using whole-chamber basket catheters (4,16,26,27). Our group has also demonstrated that localized drivers can be identified through sequential endocardial mapping using multipolar catheters. Applying the STAR mapping method in this setting is currently being evaluated.

About two-thirds of the patients in this study were on an antiarrhythmic drug pre-procedure. There was no difference in the number of ESA detected or achieving AF termination on ablation of ESA when comparing those that were on an antiarrhythmic drug versus those that were not. This potentially suggests that antiarrhythmic drugs did not seem to influence ESA detection or ablation response achieved. Furthermore, all patients had their antiarrhythmic drug stopped following the blanking period, emphasizing that the freedom of AF/AT achieved was independent of the use of an antiarrhythmic drug.

In this study the majority of ESA were mapped to sites of LVZ. Previous studies have shown that localized drivers in the form of rotational activity are more likely to colocate to sites of LVZ (4,16,26,28). Studies have also shown that drivers with rotational activity dominate over focal drivers in patients with persistent AF (1,2). The STAR mapping method does not have the ability to distinguish between rotational or focal drivers in AF but uses ESA as a surrogate for localized drivers. The STAR mapping method thereby makes no assumptions about AF mechanism but simply identifies the location from which atrial activation is emerging. Validation of the STAR mapping method has however demonstrated that the STAR mapping method can accurately identify macro- and micro-re-entrant circuits in AT. STAR maps in this setting do not identify an ESA leading 100% of time in a macro- or micro-re-entrant circuit as there are electrodes leading at the leading edge of the circuit. The proportion of cycles that an electrode is leading relative to its peers will be dependent on the number of electrodes it is paired with ahead of and behind the wavefront, which in turn is influenced by catheter position, orientation, and the limitations placed on geodesic electrode pairing distance. However, with a chaotic rhythm such as AF this pattern is less clear on the STAR maps in the context of rotational drivers. Both driver types appear to be of mechanistic importance in AF, so distinguishing the type of driver may not be particularly important. This is supported in this study because an ablation response was achieved with the majority of ESA, and high freedom from AF/AT during follow-up was also achieved when performing ablation as guided by the STAR mapping method.

Study limitations

This nonrandomized study recruited a relatively small number of patients. However, as this is using a novel mapping method that has not previously been applied in patients with persistent AF, the small cohort was thought reasonable as a proof-of-concept or feasibility study. The number of patients undergoing persistent AF ablation is similar to previously reported studies using an alternative mapping system (1,6). Holter monitoring was not performed at 3 months; however, with all patients undergoing follow-up for longer than 12 months, this study does still provide detailed follow-up.

All procedures were performed in a single center by a limited number of operators. Although strict pre-specified study protocols were followed, whether such outcomes can be reproduced in another center or environment will need to be tested, ideally against a blinded control.

As the STAR mapping method is currently applied using a custom written script and is not incorporated into a mapping system, ablation of ESA was dependent on the operator’s interpretation of the location of these sites on the LA geometry. However, as the STAR maps utilize the same geometry created in CARTO, it allows an accurate visual colocation of the ablation targets on the CARTO geometry. With further development of the mapping method it would be ideal to have the ability to coregister the maps live during ablation.


Mapping leading activation sites using this novel mapping method showed consistent results, and ablation at these sites terminated AF or slowed CL substantially in a majority of patients. Ablation at these sites in addition to PVI led to freedom from AF/AT in a high proportion of patients long term. Further testing of this method is warranted and ultimately a prospective randomized controlled trial is needed to determine its clinical utility.


COMPETENCY IN MEDICAL KNOWLEDGE: The success rates for catheter ablation of persistent AF are limited. Uncertainty remains as to the mechanisms sustaining persistent AF. Targeting of drivers has been reported albeit with variable success and this remains controversial. This novel STAR mapping method was used to identify ESA in patients with persistent AF. Targeting these ESA terminated AF in 75% of patients and slowed CL substantially in the remainder. Ablation guided by the STAR mapping method resulted in an 80% freedom from AF/AT.

TRANSLATIONAL OUTLOOK: This study demonstrated the clinical utility of the novel STAR mapping method in targeting ESA that were seen as representing localized AF. Randomized controlled studies with long-term follow-up are needed to determine whether targeting ESA affects outcomes.


Online Data

  • 1. Narayan S.M., Krummen D.E., Shivkumar K., Clopton P., Rappel W.J. and Miller J.M. : "Treatment of atrial fibrillation by the ablation of localized sources: CONFIRM (Conventional Ablation for Atrial Fibrillation With or Without Focal Impulse and Rotor Modulation) trial". J Am Coll Cardiol 2012; 60: 628.

    View ArticleGoogle Scholar
  • 2. Haissaguerre M., Hocini M., Denis al. : "Driver domains in persistent atrial fibrillation". Circulation 2014; 130: 530.

    CrossrefMedlineGoogle Scholar
  • 3. Miller J.M., Kalra V., Das al. : "Clinical benefit of ablating localized sources for human atrial fibrillation: the Indiana University FIRM Registry". J Am Coll Cardiol 2017; 69: 1247.

    View ArticleGoogle Scholar
  • 4. Honarbakhsh S., Schilling R.J., Dhillon al. : "A novel mapping system for panoramic mapping of the left atrium: application to detect and characterize localized sources maintaining atrial fibrillation". J Am Coll Cardiol EP 2018; 4: 124.

    Google Scholar
  • 5. Benharash P., Buch E., Frank al. : "Quantitative analysis of localized sources identified by focal impulse and rotor modulation mapping in atrial fibrillation". Circ Arrhythm Electrophysiol 2015; 8: 554.

    CrossrefMedlineGoogle Scholar
  • 6. Shivkumar K., Ellenbogen K.A., Hummel J.D., Miller J.M. and Steinberg J.S. : "Acute termination of human atrial fibrillation by identification and catheter ablation of localized rotors and sources: first multicenter experience of focal impulse and rotor modulation (FIRM) ablation". J Cardiovasc Electrophysiol 2012; 23: 1277.

    CrossrefMedlineGoogle Scholar
  • 7. Honarbakhsh S., Hunter R.J., Finlay al. : "Development, in vitro validation and human application of a novel method to identify arrhythmia mechanisms: the stochastic trajectory analysis of ranked signals mapping method". J Cardiovasc Electrophysiol 2019; 30: 691.

    CrossrefMedlineGoogle Scholar
  • 8. Allan V., Honarbakhsh S., Casas al. : "Are cardiovascular risk factors also associated with the incidence of atrial fibrillation? A systematic review and field synopsis of 23 factors in 32 population-based cohorts of 20 million participants". Thromb Haemost 2017; 117: 837.

    CrossrefMedlineGoogle Scholar
  • 9. Takahashi Y., O'Neill M.D., Hocini al. : "Characterization of electrograms associated with termination of chronic atrial fibrillation by catheter ablation". J Am Coll Cardiol 2008; 51: 1003.

    View ArticleGoogle Scholar
  • 10. Haissaguerre M., Lim K.T., Jacquemet al. : "Atrial fibrillatory cycle length: computer simulation and potential clinical importance". Europace 2007; 9 Suppl 6: vi64.

    MedlineGoogle Scholar
  • 11. Hunter R.J., Diab I., Tayebjee al. : "Characterization of fractionated atrial electrograms critical for maintenance of atrial fibrillation: a randomized, controlled trial of ablation strategies (the CFAE AF trial)". Circ Arrhythm Electrophysiol 2011; 4: 622.

    CrossrefMedlineGoogle Scholar
  • 12. Calkins H., Hindricks G., Cappato al. : "2017 HRS/EHRA/ECAS/APHRS/SOLAECE expert consensus statement on catheter and surgical ablation of atrial fibrillation". Heart Rhythm 2017; 14: e275.

    CrossrefMedlineGoogle Scholar
  • 13. Hunter R.J., Diab I., Thomas al. : "Validation of a classification system to grade fractionation in atrial fibrillation and correlation with automated detection systems". Europace 2009; 11: 1587.

    CrossrefMedlineGoogle Scholar
  • 14. Nademanee K., McKenzie J., Kosar al. : "A new approach for catheter ablation of atrial fibrillation: mapping of the electrophysiologic substrate". J Am Coll Cardiol 2004; 43: 2044.

    View ArticleGoogle Scholar
  • 15. Sanders P., Morton J.B., Davidson al. : "Electrical remodeling of the atria in congestive heart failure: electrophysiological and electroanatomic mapping in humans". Circulation 2003; 108: 1461.

    CrossrefMedlineGoogle Scholar
  • 16. Honarbakhsh S., Schilling R.J., Providencia al. : "Characterization of drivers maintaining atrial fibrillation: correlation with markers of rapidity and organization on spectral analysis". Heart Rhythm 2018; 15: 1296.

    CrossrefMedlineGoogle Scholar
  • 17. Lewis S. : The Mechanism and Graphic Registration of the Heart Beat . Chicago, IL: Chicago Medical Book Company1924.

    Google Scholar
  • 18. Jalife J., Berenfeld O. and Mansour M. : "Mother rotors and fibrillatory conduction: a mechanism of atrial fibrillation". Cardiovasc Res 2002; 54: 204.

    CrossrefMedlineGoogle Scholar
  • 19. Jais P., Haissaguerre M., Shah al. : "A focal source of atrial fibrillation treated by discrete radiofrequency ablation". Circulation 1997; 95: 572.

    CrossrefMedlineGoogle Scholar
  • 20. Grace A., Verma A. and Willems S. : "Dipole density mapping of atrial fibrillation". Eur Heart J 2017; 38: 5.

    CrossrefMedlineGoogle Scholar
  • 21. Narayan S.M., Baykaner T., Clopton al. : "Ablation of rotor and focal sources reduces late recurrence of atrial fibrillation compared with trigger ablation alone: extended follow-up of the CONFIRM trial (Conventional Ablation for Atrial Fibrillation With or Without Focal Impulse and Rotor Modulation)". J Am Coll Cardiol 2014; 63: 1761.

    View ArticleGoogle Scholar
  • 22. Knecht S., Sohal M., Deisenhofer al. : "Multicentre evaluation of non-invasive biatrial mapping for persistent atrial fibrillation ablation: the AFACART study". Europace 2017; 19: 1302.

    CrossrefMedlineGoogle Scholar
  • 23. Daoud E.G., Zeidan Z., Hummel al. : "Identification of repetitive activation patterns using novel computational analysis of multielectrode recordings during atrial fibrillation and flutter in humans". J Am Coll Cardiol EP 2017; 3: 207.

    Google Scholar
  • 24. Rostock T., Salukhe T.V., Steven al. : "Long-term single- and multiple-procedure outcome and predictors of success after catheter ablation for persistent atrial fibrillation". Heart Rhythm 2011; 8: 1391.

    CrossrefMedlineGoogle Scholar
  • 25. Kim Y.H., Garfinkel A., Ikeda al. : "Spatiotemporal complexity of ventricular fibrillation revealed by tissue mass reduction in isolated swine right ventricle: further evidence for the quasiperiodic route to chaos hypothesis". J Clin Invest 1997; 100: 2486.

    CrossrefMedlineGoogle Scholar
  • 26. Honarbakhsh S., Schilling R.J., Orini al. : "Structural remodeling and conduction velocity dynamics in the human left atrium: relationship with re-entrant mechanisms sustaining atrial fibrillation". Heart Rhythm 2019; 16: 18.

    CrossrefMedlineGoogle Scholar
  • 27. Honarbakhsh S., Schilling R.J., Providencia al. : "Automated detection of repetitive focal activations in persistent atrial fibrillation: validation of a novel detection algorithm and application through panoramic and sequential mapping". J Cardiovasc Electrophysiol 2019; 30: 58.

    CrossrefMedlineGoogle Scholar
  • 28. Haissaguerre M., Shah A.J., Cochet al. : "Intermittent drivers anchoring to structural heterogeneities as a major pathophysiological mechanism of human persistent atrial fibrillation". J Physiol 2016; 594: 2387.

    CrossrefMedlineGoogle Scholar

Abbreviations and Acronyms




atrial fibrillation


atrial tachycardia


cycle length


early site(s) of activation


left atrium/atrial


low voltage zone(s)


pulmonary vein isolation


right atrium/atrial


stochastic trajectory analysis of ranked signals


wide area circumferential ablation


This work was funded by the British Heart Foundation (grant PG/16/10/32016). Dr. Hunter has received travel grants from Medtronic. Prof. Schilling has received speaker and travel grants from Biosense Webster; research grants from Biosense Webster and Boston Scientific. Drs. Honarbakhsh, Hunter, and Finlay and Prof. Schilling are shareholders and co-inventors of the STAR mapping system. 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.