In the past, atrial fibrillation (AF) was regarded as an arrhythmia caused by multiple wavelets rotating in a totally random fashion around both atria. Recent studies, however, show that a variable amount of organization is involved in the fibrillation process, most likely due to anisotropic electrical conduction properties of anatomic structures and streaming phenomena along electrical borders.1,2 In animal models of AF,3 it could be demonstrated that the beat to beat AF intervals (FF) and the degree of organization decreased due to structural and electrical changes in both atria when AF became chronic. Mapping of human AF was previously preferentially carried out using conventional catheters reflecting electrical activity only adjacent to the small number of exploring electrodes4,5 or by using epicardial patch electrodes during cardiac surgery. Multisite mapping data of human AF are limited and mostly derived from open heart studies using epicardial patch electrodes.6-8 Basket catheters were developed for endocardial multisite recordings of electrical activity of intact atria9 and ventricles10 in the electrophysiologic laboratory. We used a basket catheter composed of 64 electrodes (as recently described by our group11), which was deployed in the right atrium (RA) for mapping of AF. The aim of this study was to investigate the anatomical distribution of FF intervals and the local degree of organization in human persistent, induced sustained and induced non-sustained episodes of AF and to correlate these findings with the tendency of AF to persist. Patients. The study population consisted of 54 patients referred for electrophysiologic (EP) study to the German Heart Center in Munich. All patients had either AF at admission or AF episodes that were induced during EP study. The patients were divided in 3 groups with respect to the duration of their AF episodes (Table 1). Electrophysiologic study and catheter deployment. All patients underwent EP study in the fasting postabsorptive state after written consent was obtained. All antiarrhythmic drugs were discontinued at least 5 half-lives before the study. Transesophageal echocardiography was performed prior to the study in all patients with persistent AF who had not been treated with coumadin for at least 4 weeks with an INR > 2.0 to rule out clots or thrombus formation in the left atrium before cardioversion. In addition, transthoracic echocardiography was performed for estimation of the right and left atrial dimensions to select the appropriate basket catheter size. The Constellation basket catheter (Boston Scientific/Scimed, Inc., Maple Grove, Minnesota) was advanced through an 11 French (Fr) long vascular sheath inserted into the right or left femoral vein and positioned in the RA as recently described by our group (Figure 1).11 The basket catheter was composed of 8 flexible self-expanding nitinol splines; each spline carried 8 ring electrodes equally spaced at 4 or 5 mm apart with a deployed diameter of 48 or 60 mm, respectively. Each spline was identified by a letter (from A to H) and each electrode by a number (1-8, with 1 at the tip and 8 at the base of the basket catheter). During biplane fluoroscopy, spline A was identified by 1 radiopaque marker and spline B by 2 radioopaque markers; thus, the approximate site of each spline in the atrium could be assessed. From each basket spline, seven bipolar signals were derived by a combination of electrodes 1 and 2, 2 and 3, and so forth until 7 and 8 (Figure 2). In addition, intracardiac ultrasound was used to identify the exact location of each spline in the RA, particularly with respect to the crista terminalis. The intracardiac ultrasound system used in this study consisted of a 9 Fr ultrasound catheter (Boston Scientific/Scimed, Inc.) with a 9 MHz transducer, which was recently described.12 The catheter was advanced via a 10 Fr venous sheath into the RA. In all patients in groups I and II, a 7.5 Fr internal cardioversion catheter (Alert ® EP-Medical, Berlin, New York) was inserted in addition via a brachial or femoral vein using an 8 Fr vascular sheath. It was advanced through the RA and right ventricular outflow tract into the left pulmonary artery. The internal atrial defibrillation shocks were delivered between a distal set of electrodes (distal array: cathode, active surface 2.5 cm2) in the left pulmonary artery and a proximal set (anode, active surface 2.5 cm2) along the right atrial lateral wall. When the catheters were placed in the correct positions, anticoagulation was performed by administration of a 5,000 IU bolus of heparin followed by continuous infusion of 800-1,400 IU per hour to keep the activated clotting time between 200 and 350 seconds. Data acquisition and analysis. All intracardiac electrograms were simultaneously acquired (Bard, Lab-System). Intracardiac electrograms were amplified, digitized at 1,000 Hz, filtered (30-500 Hz) and recorded on an optical disk. Custom-made software was used for automated recognition of the atrial activation in each of the 56 bipolar basket recordings. We used the cross-correlation method as a detection algorithm for activation timing, as recently described in detail.13 This technique compares the actual bipolar signal with a mathematical sine curve. The software calculates the matching of the sine pattern with the bipolar signal curve as a mathematical function. When the maximum value of this function is reached and this maximum exceeds a user definable threshold, an activation is marked by the computer. The activation markers given by the software can be corrected manually for every channel. Atrial fibrillation episodes of 10-second duration were exported from the Bard Lab-System to a personal computer. In group I, we evaluated a 10-second period directly before the first cardioversion shock; in groups II and III, a 10-second period 1-2 minutes after induction of the AF episode for further evaluation. AF classification. We used a modified Wells14 classification of AF adjusted to the high number of bipolar recordings. The existence or absence of an isoelectrical baseline and the morphology of the recorded signals were considered main characteristics of AF. We distinguished distinct signals with stable morphology from signals with a rapid change of signal morphology and number of baseline crossings. To describe the degree of organization, we defined 4 different types of AF (Figure 2) with the highest degree of AF organization in AF type I and the lowest degree in AF type IV, as follows: Type I-AF: isoelectric baseline and AF signals with stable morphology; Type II-AF: isoelectric baseline as in type I, with unstable morphology of the AF signals; Type III-AF: no isoelectric baseline; and Type IV-AF: transitional states with changes between type I-III within 500 ms. The 2 basket splines located anterior of the crista terminalis were considered to represent the lateral RA, the 2 splines posterior of the crista terminalis the posterior RA, the following 2 splines the septal area and the remaining 2 splines were assigned to the anterior region of the RA. The signals obtained from the 2 anterior splines were discarded from further evaluation due to bad wall contact or mainly ventricular activation of the electrodes caused by the close proximity to the tricuspid annulus and right ventricle. The type of AF was evaluated for each bipole during the recorded 10-second episode of AF. Afterward, these data were added together for each region to obtain the percentage of each AF type during the recorded 10-second episode for each region. Activation sequence. We evaluated the activation sequence along each basket spline (Figure 3). A craniocaudal activation pattern was defined as a sequential activation of all electrodes of 1 spline in a superior-to-inferior direction; a caudocranial activation pattern was defined as an activation in the opposite direction. Activation was considered simultaneous when all bipoles in a spline were activated at the same time. The activation was called indeterminable when none of the former patterns were observed. We further compared the activation pattern of the 2 splines located in 1 anatomical region (lateral, posterior, septal RA) to get an estimate of the homogeneity of the electrical activity within this region (Figure 4). A synchronized activation was assumed when the 2 adjacent splines showed the same activation pattern simultaneously (Figure 3). Statistical analysis. Values are expressed as percentages, ranges or means ± standard deviations. Statistical comparisons were performed using the Student s unpaired 2-tailed t-test. For comparison of FF intervals in the 3 AF groups, a 1-way ANOVA test was performed. A p-value Feasibility and safety of basket catheter application. The basket catheter could be deployed in all patients without problems. Compared to a standard mapping catheter, a small amount of additional time (2-3 minutes) was needed for advancement of the basket catheter though a long vascular sheath and correct placement within the right atrium. No basket catheter-related problems occurred. Anatomical distribution of FF intervals (Table 2). Persistent AF exhibited the shortest mean FF intervals (162 ± 30 ms), while induced non-sustained AF exhibited the longest mean FF intervals (193 ± 29 ms). The mean FF interval of induced, sustained AF was intermediary (162 ± 32 ms), but was only slightly longer than the FF interval of persistent AF. The 1-way ANOVA test revealed a significant increase of FF intervals in the lateral and septal region between persistent, acute sustained and acute non-sustained AF (Table 2). The posterior RA showed a similar increase without reaching statistical significance. AF organization. Isoelectric baselines with sharp monomorphic signals were a consistent finding at all basket splines during sinus rhythm in all patients. Organized AF (types I and II) with an isoelectric baseline was frequently observed in all 3 patient groups along the lateral wall (86%, 76% and 87% of the time, respectively). The main finding, however, was a significant increase of organized AF along the posterior and septal region from persistent and induced sustained AF to induced non-sustained AF (Figure 5) (p Synchronized activation. Along the lateral RA, synchronized activation occurred more frequently compared to the posterior or septal RA in the 3 patient groups. Overall, the percentage of synchronized activation increased from sustained to non-sustained AF. This increase was most evident in the septal RA region (Figure 4) (p FF interval distribution. The FF interval depends mainly on 2 factors: 1) the existence of anatomical or electrical obstacles that influence the course of meandering wavefronts; and 2) the local effective refractory period that determines whether the atrial tissue can be reexcited by a new wave wavefront. During AF, the effective refractory period shortens over time with a consecutive shortening of FF intervals. This process, named electrical remodeling, was described by several groups studying animals and humans.3,15,16 We had therefore expected to find longer FF intervals in the group of patients with induced AF compared to the patients with persistent AF who were in AF for at least 48 hours at the time of the EP study. Surprisingly, the mean FF interval of persistent AF was only slightly shorter than the FF intervals of induced sustained AF. Spontaneously terminating AF, however, showed significantly longer FF intervals compared to induced sustained AF. These findings are in accordance with Capucci et al., who reported significantly shorter FF intervals in patients with acute sustained compared to acute non-sustained AF.17 Wijffels et al.3 described in a goat model of AF that the shortening of the FF interval during AF is an important factor for the sustenance of AF. Our data support this hypothesis in man, and also show that the FF interval at the beginning of an AF episode serves as a predictor for the tendency of an AF episode to sustain itself or to terminate spontaneously. AF organization and activation. In this study, organized AF with an isoelectric baseline (type I or II) was a very frequent finding along the lateral wall in the 3 groups of AF patients, as already described by other groups.2,4 The occurrence of an isoelectric baseline was considered important because it represents an electrically silent period between AF signals without detection of near or far-field electrical activity. Along the lateral wall, predominantly organized AF was found in the 3 patient groups. In contrast to the lateral region, the septal and posterior areas exhibited a significant increase in organized AF from sustained to non-sustained AF. Synchronized activation occurred more frequently during non-sustained compared to sustained AF in all 3 RA regions. The increase in FF interval, AF organization and local synchronization along the septum during non-sustained AF may reflect an overall increase of AF organization caused by a decrease of the number of circulating wavelets not only of the right but also of the left atrium that may serve as a necessary precondition for the spontaneous termination of AF. Recently, several groups could demonstrate that the canine interatrial septum is electrically disconnected between the 2 atria, with only 2 electrical breakthrough sites in the superior and inferior septum.18,19 From human studies, it is assumed that there may be at least 3 electrical transseptal breakthroughs: Bachmann s bundle, the fossa ovalis and the coronary sinus.20 The fact that electrical activation enters the septal area not only from the adjacent right atrial sites but also from the left atrium via the connections described may explain our finding that during sustained AF the septum and parts of the posterior wall showed shorter FF intervals and shorter episodes of organized AF compared to the lateral wall. Study limitations. This study has some limitations. First, the shape of the RA is complex, sometimes causing incomplete coverage of the right atrial appendage and subeustachian isthmus area by the basket catheter.11 Second, we used bipolar signals with an interelectrode distance of 3-4 mm. A closer spacing would have allowed a more precise evaluation of local activation. Third, we did not perform direct left atrial basket mapping. Conclusion. Longer FF intervals and a significantly higher degree of AF organization and synchronization, especially in the septal region, are the main differences found between persistent and non-sustained episodes of AF. Induced sustained AF ranked in between these 2 groups. However, its behavior resembles from its beginning much more closely persistent than non-sustained AF.