Clinical Significance of Arrhythmogenic Foci in Atrial Fibrillation

Atrial fibrillation (AF) is initiated by pulmonary vein (PV) and non-PV foci, which could be associated with initiating and maintaining AF. The development of the re‐ modeling process and preexistent anatomical structures are likely to relate to the structural and electrophysiological changes in the PVs and non-PV area, which could promote the local conduction abnormalities and cause an increased PV/non-PV arrhythmogenicity. In this section, we assessed the features and relating factors of PV/non-PV arrhythmogenicity in patients with AF and evaluated its clinical im‐ plication. As a result, we realized the atrial anatomical features, such as the left at‐ rial roof shape, left lateral ridge, and Marshall vein provided us with an understanding of PV and non-PV arrhythmogenicity in patients with AF. In addi‐ tion, the presence of residual arrhythmogenic non-PV foci is associated with in‐ creased AF recurrence after catheter ablation; therefore, the information of arrhythmogenic foci (AMF) is also useful for determining the appropriate strategy of ablation for AF.


Introduction
Atrial fibrillation (AF) is caused by triggers from pulmonary veins (PVs) [1], and a rapid firing from the PVs could be responsible for initiating and maintaining arrhythmias in patients with AF. The enhanced automaticity or triggered activity mechanisms could be involved in the initiation of AF [2,3]. In addition, the PV's circumference is also most likely crucial for sustaining the reentry of maintaining AF [4], which can enhance a condition for persistent AF.
Non-PV foci can also arise from the crista terminalis, ostium of the coronary sinus, interatrial septum, superior vena cava, left atrial posterior free wall, and Marshall bundle [5,6] with the incidence ranging from 3.2 to 47 % [7,8,9]. The dominant triggering sites of non-PV have a slow diastolic depolarization, increasing the chance of the spontaneous depolarization [10], and the triggered activity from the non-PV sites could also be involved in the initiation and perpetuation of AF. Previous studies have reported that the increased delay after depolarizations has been documented from the superior vena cava [10], coronary sinus (CS) [11], Marshall bundle and the coronary sinus [12], atrial muscle that extends into the mitral valve [13], and working muscle [14]. Especially, the Marshall bundle may be a crucial structure to initiate catecholamine-sensitive AF.
The development of the remodeling process and preexistent anatomical structures seems to be related to the structural and electrophysiological remodeling in the PVs and atrium, which can increase the local abnormal conduction and develop an increased PV/non-PV arrhythmogenicity leading to AF persistency [15,16,17,10,11].
In this section, we assessed the features and relating factors of PV/non-PV arrhythmogenicity in patients with AF and evaluated their clinical implication during catheter ablation procedure.

The feature and clinical implication of arrhythmogenic foci of atrial fibrillation 2.1. The method of induction and detection of PV/non-PV arrhythmogenic foci
We used five multipolar catheters recording the electrograms to search for the location of the arrhythmogenic foci (AMF). A 20-pole catheter covered the SVC to the crista terminalis, CS, and the left PVs. A mapping catheter was located at the right superior PV (Figure 1). When the AMF have originated from a non-PV area uncovered by the catheters, we searched the location with a mapping catheter. The 12-lead ECG and intracardiac electrograms were filtered between 30 to 500 Hz (DUO EP Laboratory; Bard Electrophysiology, Lowell, MA, USA).
The occurrence of PV/non-PV foci could be influenced by the induction methods, and PV arrhythmogenicity may be enhanced by the stimulation with acetylcholine or isoproterenol (ISP) [2,18]. The relationship between the ISP dose and arrhythmogenicity remains unclear; however, the PV/non-PV foci are likely to be revealed with a high-dose isoproterenol up to 20 g/min or subsequent cardioversion of AF [19,20]. High-dose ISP can cause the vagally mediated nerve reflex bradycardia, which seems to increase arrhythmogenicity after autonomic nerve competition.
Both atrial spontaneous AMF were carefully searched before the PV isolation procedure under an intravenous infusion of isoproterenol (ISP) without sedation. During sinus rhythm, ISP was initially delivered at 1-2 g/min for 5 min, and then the dose was gradually increased up to 20 g/min with careful monitoring of the blood pressure. When the blood pressure dropped under 70 mm Hg, the dose of ISP was reversed to the basal level. When AF persisted and/or spontaneously occurred, direct cardioversion (DC) was attempted up to three times. The DC energy was delivered by using an external biphasic wave up to 270 J, and sinus rhythm was temporarily or successfully restored in all enrolled patients.
The ISP administration was maintained at basal level (1-2 g/min) during the ablation. At the end of the session, the increased dose of ISP was administered up to 20 g/min. AMF were confirmed as direct AF triggers and/or reproducible atrial premature beats with coupling intervals of <350 ms or frequent firings ( Figure 2).

The location of PV/non-PV arrhythmogenic foci
Two hundred fourteen consecutive patients with drug-refractory AF episodes were enrolled in this study (mean age of 61 years, male, 71 %; persistent AF, 21%). The clinical and electrophysiologic characteristics of the AMF were demonstrated in Figure 3. Five hundred AMF were observed. PV and/or non-PV foci were detected in 201 of 214 (93.9 %). Two hundred sixtythree foci (52.6 %) in 174 patients (81.3 %) were confirmed as the triggers directly shifted to AF, and 237 foci (47.4 %) in 150 patients (70.1 %) showed reproducible premature atrial beats with an interval of <350 ms or repetitive firings. PV foci were confirmed in 195 of 214 patients (91 %) and non-PV foci in 107 of 214 (50 %), accounting for one third of all the AMF.  Catheter locations for detecting AMF. Twenty-pole circular catheters were positioned at the left superior and inferior PV. A 10-or 20-pole catheter was located in the CS, and then terminal crest and the SVC were covered with a 20-pole catheter. The mapping catheter was located at the RSPV. When arrhythmogenic foci arise from the RIPV or a non-PV area, the suspected area was confirmed by a mapping catheter. The location of the PV/non-PV foci showing the earliest atrial focus was referred to the local electrogram or onset of the ectopic P wave. In addition, the direction of the earliest activated site of the PV/non-PV foci could be also determined by the sequence of the activation recorded from multipolar catheters allowing to detect the non-PV sites in both atria. Non-PV foci were induced 15 % of the time with no ISP, 30 % with 1-2 g/min, and 55 % with 2-20 g/min. PV foci were revealed 25 % of the time with no ISP, 43 % with 1-2 g/min, and 32 % with 2-20 g/min. The distribution of the inducibility according to the ISP dose significantly differed between PV foci and non-PV foci (p<0.001).
In half of the enrolled patients, non-PV foci were confirmed and accounted for one third of all AMF. High-dose ISP could improve the ratio of the detection of both PV and non-PV foci; however, the dose of the ISP was significantly higher for the non-PV foci than the PV foci. The predominant non-PV trigger sites seemed to be related to anatomical structures such as the terminal crest or ligament/vein of Marshall, known to be catecholamine-sensitive structures [5]. These evidences may explain why a high dose of ISP was needed to reveal non-PV foci.

The PV/non-PV arrhythmogenic foci between paroxysmal and persistent AF
. The incidence of PV foci and non-PV foci from the left atrium was not significantly different between paroxysmal and persistent AF patients. The incidence of non-PV foci, the sum number of foci, the number of non-PV foci, the incidence of right atrial foci, and the occurrence of multiple foci were significantly higher in the persistent than paroxysmal AF. In a multivariate analysis, multiple foci were one of the independent contributing factors to persistent AF as well as the left atrial dimension.
Furthermore, Figure 4 demonstrated that the number of foci was significantly higher in >24 h than <  . The location of induced AMF. AMF were determined as direct AF triggers or spontaneous reproductive premature atrial premature contraction (PAC) with the coupling interval less than 350 ms or repetitive firing in arrhythmogenic foci. Black circle represents foci which directly shifted to AF; white circle represents reproductive PAC and repetitive firing. To detect the location of foci, we simultaneously used five multipolar catheters to record the electrograms outside the PVs, coronary sinus, SVC, and crista terminalis to search for the AMF during isoproterenol administration.
Clinical Significance of Arrhythmogenic Foci in Atrial Fibrillation http://dx.doi.org/10.5772/60646 persistent AF. Therefore, these findings may imply that the presence of increased foci may possibly facilitate the development of a shift from paroxysmal to persistent AF, although that may gradually become less significant as long-lasting AF develops.  The multiple triggers may allow a greater chance of reinitiating AF even after the AF selftermination and may facilitate the progression to the persistency from a paroxysmal to persistent AF. In the meantime, the enhanced triggered activity of multiple sites as the cause of AF persistency could also beget AF perpetuation by making new wavelets and less likelihood of AF self-termination. Furthermore, the enhanced dispersion of the atrial refractoriness may also be a crucial factor for AF persistency. The presence of increased atrial dispersion might promote the progression from paroxysmal to persistent AF state [21]. These observations may provide a clue as to why multiple triggers are associated with the development of the fibrillatory process in AF persistency.

Mutual linkage of left PV AMF
PV myocardial sleeves with complex muscle bundle orientations or specific autonomic nervous system may have the same interactions between each PV. Thus, we determined the mutual linkage of AMF around PVs. AFC from the left superior PV were significantly associated with AFC of the left inferior PV (42 % vs. 23 %, p<0.05), left-sided left posterior wall (20 % vs. 5 %, p<0.05), and roof area (8 % vs. 2 %, p<0.05) ( Figure 5). In case of foci from LSPV, the occur-rence of AMF was 68 % in LIPV, 85 % in the left side LA posterior wall, and 75 % in the roof. Right PVs had no significant mutual association for AFC between each other ( Figure 6).

Left atrial roof shape and PV/non-PV foci
The remodeling process is associated with the structural and electrophysiological abnormality in the PVs and atrium, which could promote local conduction delay and lead to an increased PV/non-PV arrhythmogenicity developing to AF persistency [15,16,17,10,11]. These evidences might imply that the morphological features of the PVs and atrium can contain a crucial role to detect their preexisting arrhythmogenicity, although the evaluation of the morphological features is limited in a quantitative manner because of their variable and unique structure.
The left atrial (LA) roof consisting of the upper wall of the left atrium and upper PVs was demonstrated as the silhouette of LA roof and could simply be visualized by PV angiography or left atrial CT imaging. In addition, the dominancy of morphological PVs/LA and the features of the LA roof silhouette could be easily determined in patients with AF. Thus, the relation between PV/non-PV arrhythmogenicity and LA roof silhouette was examined in this study.
Based on the PVs and LA dominancy, The LA roof shape was classified into a deep V shape (possible PV dominancy), shallow V shape, and flat-coved shape (possible LA dominancy) by cine angiography (Figure 8]. Angiography was conducted by both contrast media from the long sheath locating at the right and left superior PVs. The LA roof shape was assessed by A-P projection and was determined by the angle of the LA roof silhouette between the right and left LA wall. The deep V shape was defined as <140°, shallow V shape was 140° to180°, and flatcoved shape was > 180°. AF is likely to originate from larger PVs [22], and the enlarged PVs may often be associated with the arrhythmogenic PVs [23]. Enlarged PV by the atrial stretch can enhance the PV's automaticity and triggered activity for AF initiation [24]. In addition, the atrial remodeling process may promote the increased triggered activity of non-PV lesions. The presence of multiple PV/non-PV foci could be related to longer AF duration, an older age, and larger atrial dimensions [25]. And also, LA enlargement could predispose LA posterior wall triggers [15].

Marshall bundle and arrhythmogenic foci
Marshall reported that a "vestigial fold of the pericardium" lies dorsal to the left atrial appendage in 1850. The small oblique vein of Marshall (VOM) is often connected to the vestigial folds going around the ostium of the left PVs. VOM drains into CS and separates the great cardiac vein and CS. The muscle sleeves of the VOM are also connected to CS musculature [26]. The vein or its ligament of Marshall is usually connected to the left PVs [27,28,29], and its distal ends are directly connected to the posterior wall of the left atrium [27,30].  AF can originate from VOM or its ligament because of its catecholamine-sensitive structure [31,5]. Preserved persistent left superior vena cava as a remnant of VOM can also include the similar electrical and anatomical features [32]. VOM or its ligament has connections to muscle bundles of the left atrium as well as of the surrounding coronary sinus (CS) in histological studies. The distal end often connects to the LA lateral area and to the left PVs [27,29]. Therefore, recognition of the VOM anatomy in AF patients would help access to non-PV foci around the left PVs, which would lead to favorable clinical procedural result.

Angiographic vein of Marshall and AMF
In 100 AF patients, we examined the anatomy of VOM with balloon-occluded venography of coronary sinus using a balloon wedge pressure catheter (Goodtec, Huntington Beach, CA). The landmark of VOM orifice was identified at the junction of the CS and great cardiac vein. The right anterior oblique, left anterior oblique, and anteroposterior views were obtained in enrolled AF patients ( Figure 9). To identify the anatomical association for VOM to both the superior and inferior left PVs, we performed selective superior and left inferior left PVs angiography by using injection of 5-10 ml of contrast medium from long sheaths. The grade of VOM development was measured from the AP view and classified into two grades (poor, reaching below superior left PV distributed in LA, and good, above superior left PVs). VOM was visualized by balloon-occluded CS venography in 73 AF patients (73 %). There were no significant differences in clinical characteristics of the two groups. VOM development was poor in 55 patients (75 %) and good in 18 patients. In the anteroposterior image, the VOM running behind the mitral isthmus line was confirmed. VOM going through the mitral isthmus area was observed in 51 patients (51 %). The branches originating from the end of VOM was observed in 49 patients (67 %).
The incidence of PV foci from the left superior PV was significantly higher in patients with VOM than in those without it (66 % vs. 42 %, P=0. 05). And also, the incidence of left superior PV foci as direct AF initiator was significantly higher in patients with VOM than in those without it (50 % vs. 30 %, P<0. 05). The incidences of e left superior PV foci were 41 % in none, 69 % in poor VOM, and 56 % in good VOM. The incidences of the left superior PV foci as direct AF initiator were 30 % in none, 56 % in poor VOM, and 33 % in good VOM.
Twelve patients had non-PV foci in the LA posterior wall, and nine (75 %) of these patients also had PV foci in the left superior PV around VOM structure even after the successful PV isolation procedure at PV ostium level. The correlation between angiographic VOM anatomy and surrounding non-PV foci is shown in Figure 4. After ablating the site at the branch of VOM connecting to the LA wall, we can often successfully terminate AF. Twenty-eight patients had 30 non-PV foci surrounding left superior PVs including LA posterior free wall, LA roof, and LA lateral wall, and 12 of 30 non-PV foci were directly shifted to AF.
The branches of the VOM were a good landmark to identify the location of non-PV foci around left PVs (Figure 10). We could successfully ablate the residual non-PV foci at the distal end of VOM in 11 patients (39 %) after PV ostial isolation (AF termination after RF delivery, 3; disappearance of reproductive atrial premature contractions, 8). Successful terminations of non-PV foci were observed in 5 in left LA posterior wall, 4 in LA lateral wall, and 2 in LA roof. Among 28 patients with non-PV foci surrounding left PVs, non-PV foci were successfully deleted in 17 patients, whereas 11 patients of them had AF recurrence. Seven of 11 (67 %) with successful non-PV foci termination were free from AF recurrence. The presence of VOM is associated with a higher incidence of AF triggers of the left superior PVs. The incidence of left superior PV foci was significantly higher in patients with visible VOM than in those without visible VOM. In angiographic findings, the distal braches of VOM are commonly distributed around both the left superior and inferior PVs, especially in patients with good visual VOM. The VOM and its ligament richly innervated by sympathetic nerves could be served as a cause of isoproterenol-sensitive automatic activity [5,33]. These evidences support arrhythmogenic foci from the VOM, and its ligament can be inducible by using highdose isoproterenol administration.
Left PVs foci and non-PV foci adjacent to the left PVs can have an influence on each other. Approximately 40 % of non-PV foci around the left superior PV were successfully ablated by targeting the distal end of VOM or its branches. These evidences demonstrate the angiographic data of VOM, and its branches may indicate the site of catheter ablation of non-PV foci. Radiofrequency energy applied to the areas of VOM distal ends occasionally delineated non-PV foci originating from the surrounding area of left PVs. Thus, we believe that understanding of the VOM anatomy can improve the clinical outcome of ablation in cases with catecholaminesensitive AF.

Conduction along the left lateral ridge and the arrhythmogenicity of the left pulmonary veins
The ligament and VOM containing the Marshall bundle (MB) with richly innervates the sympathetic and parasympathetic nerves is within the left lateral ridge (LLR) which is longitudinally running between the left atrial appendage and left pulmonary vein (LPVs), and they can serve as a source of triggers and the substrate of reentry of atrial fibrillation (AF) [1,2]. If the distinctive dominant conduction along the LLR is present, possibly due to the continuous and/or partial MB conduction, its conduction may be associated with the increased arrhythmogenicity of the LPVs. In this study, we examined the relationship between the preferential conduction properties of the MB and the arrhythmogenicity of the LPVs in 40 AF patients. After the RF delivery along the LLR, the PV potentials of the upper LPV completely disappeared in one patient and that of the lower LPV in two patients. The conduction time between the LPVs and CS stimulus site was significantly prolonged during the RF delivery (before vs. after; upper, 91±26 ms vs. 127±38 ms, p<0.001; lower, 86±21ms vs. 103±22ms, p<0.001). A remarkable prolongation of more than 30 ms was observed in 19 of 40 patients (48 %) (both LPVs, 6; only the upper LPVs, 12; and only the lower LPV, 1). The sites of the remarkable prolongation during the RF delivery were observed at the carina between the LPVs [4], anterior site of the upper LPV carina [10], anterior wall of the lower LPV [3], and bottom of the lower LPVs [2]. The conduction time from the CS to the earliest activated upper PV after the RF delivery was significantly longer in patients with AMF from the upper LPV than in those patients without (107±36 ms vs. 146±40 ms, p<0.01), and the conduction time from the CS pacing site to the earliest activation site of the upper LPV was significantly prolonged in the patients with AMF than in those without during the RF delivery (44±22ms vs. 17±11ms, p<0.01).
In this study, the earliest site of AMF from the LPVs was often determined to be around the carina region. These observations are likely to be consistent with the previous report [9]. In addition, the complex crossing of the muscular connections, bridges, neural inputs, and the adjoining muscle sleeves, possibly related to the MB conduction in the inter-PV carina, might promote electrical arrhythmogenicity including spontaneous ectopies of AF [10]. And also, the earliest activated site of the upper LPVs during CS pacing was highly observed around the  carina region, and also a remarkable prolongation jump during the RF delivery was highly observed around the carina and/or adjacent anterior area. A previous report suggested that the distal exit of the MB into the upper LPV is commonly located around the inter-PV junction, possibly bypassing the LPV junction to the left atrium [34]. These specific muscle orientations and the dominant MB conduction toward the carina region could promote the preferential conduction properties.
In addition, the prolongation of the conduction time between the CS and LPVs during the RF delivery was significantly more commonly observed in patients with upper LPV AMF than in those without. The preferential properties of the MB connecting to the LPVs might involve cross talk that promotes an increased LPV arrhythmogenicity [3,4,11]. A larger amount of preserved MB muscle as a remnant of the LSVC, which is related to the conduction properties of the LPVs, may be crucial for determining the increased arrhythmogenicity of the LPVs.

The efficacy of the sinus restoration strategy to detect arrhythmogenic foci for persistent AF
Catheter ablation (CA) of persistent AF may commonly be performed during ongoing AF, mainly targeting sites exhibiting complex atrial fractionated electrograms (CFAEs) and/or dominant frequencies (DFs) in addition to pulmonary vein (PV) isolation [35,36,37]. However, CA during ongoing AF may be limited especially in patients with a trigger dominant-type AF [20,38,19]. The rapid firing from the PVs and non-PV foci may beget enhanced automaticity, triggered activity, and localized micro-reentry as AF initiation and maintenance [37,3,10].
Our prior data suggested that an increased number of AMF are more highly observed during a vigorous sinus rhythm (SR) restoration strategy in persistent rather than paroxysmal AF [39], and the failure of the elimination of the AMF initiating an immediate recurrence of AF was significantly associated with the recurrence of persistent AF [40]. In this study, we performed CA based on a vigorous SR restoration strategy for persistent AF and evaluated the relationship between the electrophysiological features of the inducible AMF and recurrent AF episodes after the CA in 117 persistent AF patients.
The AF ablation strategy is summarized in Figure 12. We initially performed the PV isolation procedure by using a double circular mapping catheter technique. The DC energy was delivered with an external biphasic waveform of up to 270 J before the PV isolation. The electrical PV isolation was successfully accomplished with monitoring the circumferential electrical isolation at the antrum level: approximately 1-2 cm from of both the right and left PVs ostium.
After the PV isolation procedure, an additional RF energy application was primarily applied to create an LA roof line. When the AF was still persisted or inducible after LA roof line, additional mitral isthmus line or ablation of the area showing complex fractionated atrial electrograms (CFAEs) in left atrium was accomplished. When AF could not be terminated in these series of procedures, direct cardioversion was delivered to restore SR again in such cases. Then, we confirmed whether complete block lines were created at the roof and mitral isthmus. At the end of the CA, residual AMF were still found in 48 of 117 patients (41.0 %) (directly shifted to AF, 22; reproducible atrial premature beats, [26]. The locations included the left atrial posterior wall [6], superior vena cava [3], crista terminalis [4], left lateral area [1], interatrial septum [1], coronary sinus ostium [1], and unknown [32]. The number of AMF during the CA Figure 12. The summarized vigorous SR restoration strategy during the ablation according to the pacing-oriented AF inducibility. SR rhythm was restored by using external direct cardioversion before the PV isolation and line creation at the end of the ablation. AF was no longer inducible after only the PV isolation procedure in 24 of 117 patients (20.5 %). During the PV isolation, SR shifted to AF spontaneously and/or was triggered by the roving catheter in 68 of 117 patients  Figure 13 demonstrates the AF recurrence ratio in the patients with and without residual AMF. AF episodes after the CA were significantly greater in the patients with residual AMF than in those without (50 % vs. 26 %, p=0.002). The result of the study demonstrated that the residual AMF was a useful predicting parameter for the outcome of CA, and the clinical course was impressively favorable in patients without residual AMF (AF recurrence after initial session at two years was 26%). (58.1 %). At the end of the ablation, residual AF was still inducible in 37 of 117 patients (31.6 %). Figure 13. The AF recurrence ratio in the patients with and without residual arrhythmogenic foci during the follow-up. Residual foci were observed in 48 of 117 patients (41 %). The AF free ratio between both groups was compared by a log rank analysis, and AF episodes after the CA were significantly higher in the patients with residual foci than in those without (50 % vs. 26 %, p=0.002). The mean follow-up period was 937 days.
After only a PV isolation, AF was no longer inducible in approximately one fifth of the patients with a favorable outcome even though they underwent a less aggressive intervention. This information might allow us to reduce the number of unnecessary additional RF applications during CA. On the other hand, non-PV foci were also highly confirmed even in patients with persistent AF [39]. Several studies have also addressed the importance of modifying the non-PV foci to improve the outcome of CA for AF [20,23]. A vigorous SR restoration strategy might facilitate determining the non-PV arrhythmogenicity.
The data from this study also showed that non-PV AMF were closely associated with the AF pacing inducibility during CA. The development of the atrial remodeling process might enhance the triggered activity of the non-PV lesions, because the increased non-PV arrhythmogenicity may be associated with the atrial remodeling process including an enlarged LA [25,41]. A recent study demonstrated that the response to ISP after CA more accurately predicted AF recurrences in patients with paroxysmal AF [38]. Residual AMF could increase the chance of AF initiation, and the significance of those may be especially pronounced particularly in cases that develop atrial remodeling.
In conclusion, these data support the role of arrhythmic triggers in determining eventual recurrences in patients with persistent AF and point to AMF as a potentially valuable early index of long-term ablation success.

Conclusions
AMF could be involved in mechanism of the AF development. In addition, atrial anatomical structure such as left atrial roof shape, left lateral ridge, and Marshall vein provided us with an understanding of the arrhythmogenicity of the PVs and non-PVs in patients with AF.
Because the presence of residual AMF is associated with increased AF recurrence after ablation, the information of AMF is useful for determining the appropriate strategy of ablation for AF.