Panoramic Endocardial Optical Mapping Demonstrates Serial Rotors Acceleration and Increasing Complexity of Activity During Onset of Cholinergic Atrial Fibrillation.

Background Activation during onset of atrial fibrillation is poorly understood. We aimed at developing a panoramic optical mapping system for the atria and test the hypothesis that sequential rotors underlie acceleration of atrial fibrillation during onset. Methods and Results Five sheep hearts were Langendorff perfused in the presence of 0.25 µmol/L carbachol. Novel optical system recorded activations simultaneously from the entire left and right atrial endocardial surfaces. Twenty sustained (>40 s) atrial fibrillation episodes were induced by a train and premature stimuli protocol. Movies obtained immediately (Initiation stage) and 30 s (Early Stabilization stage) after premature stimulus were analyzed. Serial rotor formation was observed in all sustained inductions and none in nonsustained inductions. In sustained episodes maximal dominant frequency increased from (mean±SD) 11.5±1.74 Hz during Initiation to 14.79±1.30 Hz at Early Stabilization (P<0.0001) and stabilized thereafter. At rotor sites, mean cycle length (CL) during 10 prerotor activations increased every cycle by 0.53% (P=0.0303) during Initiation and 0.34% (P=0.0003) during Early Stabilization. In contrast, CLs at rotor sites showed abrupt decreases after the rotors appearances by a mean of 9.65% (P<0.0001) during both stages. At Initiation, atria-wide accelerations and decelerations during rotors showed a net acceleration result whereby post-rotors atria-wide minimal CL (CLmin) were 95.5±6.8% of the prerotor CLmin (P=0.0042). In contrast, during Early Stabilization, there was no net acceleration in CLmin during accelerating rotors (prerotor=84.9±11.0% versus postrotor=85.8±10.8% of Initiation, P=0.4029). Levels of rotor drift distance and velocity correlated with atria-wide acceleration. Nonrotor phase singularity points did not accelerate atria-wide activation but multiplied during Initiation until Early Stabilization. Increasing number of singularity points, indicating increased complexity, correlated with atria-wide CLmin reduction (P<0.0001). Conclusions Novel panoramic optical mapping of the atria demonstrates shortening CL at rotor sites during cholinergic atrial fibrillation onset. Atrial fibrillation acceleration toward Early Stabilization correlates with the net result of atria-wide accelerations during drifting rotors activity.

action potential duration

breakthrough pattern of impulse propagation

cycle length

5% minimal CL

dominant frequency

maximal DF

phase singularity point

Clinical Perspective

What Is New?

We developed a new panoramic endoscopic system enabling optical mapping of activation patterns simultaneously across intact endocardial surfaces of right and left atria in the isolated sheep heart to study how initial triggered activity turns into subsequent high frequency and complex excitation waves in a self‐sustained cholinergic atrial fibrillation model.

Ectopies or wavebreaks following a premature stimulus can occur in the other atrium and subsequent fibrillation is sustained only when the cycle intervals gradually shorten and rotors appear.

Rotors appear serially and temporarily accelerate activation in their vicinity, but the rotors’ acceleration is associated with net atria‐wide acceleration only initially and is followed by an equilibration between atria‐wide accelerations and decelerations; wavebreaks that do not develop into reentries do not associate with atrial acceleration at any stage, but their number, and hence fibrillation complexity, is increasing with time until stabilization.

What Are the Clinical Implications?

The appearance of initial ectopic waves, wavebreaks, and rotor patterns across the 2 atria during the onset of atrial fibrillation is consistent with the suggestion that discharges and driving activity can appear across the entire atria and in locations remote to the triggering event, which warrant panoramic mapping.

Only a subset of drifting rotors patterns associate with atria‐wide acceleration, sustenance and increased complexity, and specifically targeting those patterns could potentially improve ablative and nonablative therapies.

Atrial fibrillation (AF), the most common sustained clinical arrhythmia, is a challenging disease. It would be advantageous to better understand the initiation of AF to avoid its stabilization and prevent its progression. It is commonly accepted that triggered activity in patients can initiate AF episodes and indeed, AF episodes often follow the appearance of spontaneous repetitive focal activity of variable origin. , However the manner by which the discrete initial triggered activity turns into subsequent high frequency and complex excitation waves and self‐sustained AF remains elusive. ,

Optical mapping of the posterior left atrium (LA) in a sheep model of cholinergic AF initiation revealed that programmed pacing at the pulmonary veins generated the first wavebreaks and self‐sustained reentries at the septopulmonary bundle. There is also evidence from panoramic electrical mapping in humans that spontaneous or induced AF was initiated in the presence of repetitive focal, but mostly reentrant, patterns. Nevertheless, the dynamics and the role of observed reentrant activity across the 2 atria during initiation and early stabilization of AF have not been fully established. In particular, the mechanisms giving rise to the dramatic increase in activation rate and complexity across the entire atria, characteristic of the transition from organized rhythm to disorganized fibrillation, have not been studied.

Thus, this study has 2 objectives. We first aim at developing a wide‐view optical mapping approach to elucidate with high reliability the patterns of excitation across the entire and intact endocardial surfaces of the right atrium (RA) and LA. And second, we aim to use the new panoramic mapping capabilities to test the hypothesis that during the onset of cholinergic AF reentrant activity accelerates with an increased complexity of the fibrillation. Using the panoramic endocardial optical mapping system in isolated sheep hearts we find reentrant‐associated acceleration of activity across both the RA and LA. Our results support the notion that series of reentrant activity patterns, but not wavebreaks, underlie the acceleration of AF during onset.

Methods

This research was performed in its entirely at the University of Michigan in accordance with its Unit of Laboratory Animal Medicine Policies, Guidelines and Standard Operating Procedures. The study data, methods, and materials information will be made available to other researchers for purposes of reproducing the results or procedures upon a reasonable request.

Langendorff Perfused Heart Preparations

Five healthy male sheep weighing 30 to 40 kg were used. The animal protocol was approved by the University Committee on Use and Care of Animals of the University of Michigan and conforms to the Guide for Care and Use of Laboratory Animals by the US National Institutes of Health. Sheep hearts were isolated and Langendorff perfused as described previously. , , See Data S1 for more details.

AF Induction and Classification

All episodes of atrial activity were studied in the presence of 0.25 μmol/L carbachol mimicking increased vagal tone inducive of AF. , AF was induced with a stimuli train (S1) and premature stimulus (S2) programmed pacing protocol at the LA epicardial free wall (see Data S1 for details). Subsequent episodes were classified as nonsustained if lasting <40 s and self‐sustained if lasting ≥40 s. All nonsustained and all sustained AF episodes were included in the analysis. We focused on 2 periods of the AF onset: the first 7 s post‐S2 period (termed Initiation period), and 30 to 40 s post‐S2 (termed Early Stabilization period).

Wide‐View Endoscopic Optical Mapping

The novel wide‐view, panoramic, endoscopic mapping approach is described in Data S1 and Figure S1. Briefly, lesions were cut at left and right ventricular apexes to enable insertion of 2 dual‐channel solid borescopes equipped with wide‐view objective units directed toward the left and right atria. The borescopes were attached to cameras to record di‐4‐ANEPPS fluorescence from the endocardial surfaces in 2 10‐s duration movies (80 × 80 pixels, 600 frames/s) during the pacing and Initiation and during the Early Stabilization periods. Sample recordings are provided as Supplemental Videos.

Optical Data Analysis

Parameters analyzed in space and time included phases and phase singularity points (SPs), dominant frequencies (DFs), cycle lengths (CLs) and action potential durations (APDs). Details of data processing and definitions of SPs of rotors and non‐rotors (wavebreaks), maximal DF (DFmax), atria‐wide minimal CL (CLmin) as well as rotors drift distance and velocity are presented as Data S1.

Statistical Analysis

Analyzed values are presented as mean±SD, and with 95% CIs where noted. For time course of measures, repeated values in time were analyzed with linear regressions and ANOVA. Distributions of data values within groups were tested for normality using the Shapiro–Wilk test before all comparisons. For comparisons between normally distributed groups of data, paired or unpaired Student’s t test methods were used as appropriate. Unpaired t tests were applied under nonequal variances condition. The nonnormal distributions were compared with Wilcoxon signed‐rank test as noted. Statistical analyses were performed in Matlab (MathWorks, Natick, MA). Probability (P) value was considered significant if less than alpha=0.05, except when adjusted for multiple comparisons as noted.

Results

Time Course of AF Onset and Activation Patterns

In 5 hearts, 16 nonsustained AF episodes and 20 induced ≥40 s long AF episodes were analyzed. A representative time course of pacing induction, Initiation, and Early Stabilization of sustained AF is shown in Figure 1. A demonstrates a single pixel optical signal showing transition from a stimuli train S1 and premature stimulus S2 to self‐sustained more complex AF activity. During LA pacing the electrical activity was devoid of wavebreaks across either atrium, and APD70 for all RA regions was longer than the for all LA regions (112.6±19.3 ms versus 93.9±9.9 ms respectively, P<0.0001; see Video S1 and Figure S2). In episodes of self‐sustained AF the stimuli train‐S2 interval was longer than in nonsustained AF (127±18 ms, n=20, versus 109±15 ms, n=15, respectively; P=0.017). B and C demonstrate the increased complexity from the Initiation period, immediately following the S2 stimulus, to the Early Stabilization period 30 s later (see Videos S2 and S4 for Initiation and Videos S3 and S5 for Early Stabilization). The sample phase snapshots in panels B and C show an increase in the number of SPs and the time‐space plots show the temporal proliferation of patterns characteristic of reentrant activity, from a single site in the RA (cyan arrow, B) to multiple sites in both atria (C).

Figure 1

Induction and stages of AF onset.

A, A sample pixel fluorescence trace showing the stage of regular pacing (S1…S1) followed by a premature stimulus (S2) inducing self‐sustained AF. The AF onset was analyzed during its Initiation (0–3 s) and Early Stabilization (30–33 s) stages. Cyan vertical dash lines indicate times of phase maps in (B) and (C). B, Phase maps (top) and time‐space plots (bottom; along the white dash line in phase maps) during Initiation stage. Arrowhead circle in phase map points to the first rotor activity in the AF (Cyan arrow in the time‐space plot). C, Phase maps (top) and time‐space plots (bottom; along the white dash line in phase maps) during Early Stabilization stage. Arrowhead circles in phase map point to the multiple rotor or wavebreak activity. AF indicates atrial fibrillation; LA, left atrium; RA, right atrium; S1, stimuli train; and S2, premature stimulus.

Among the 20 sustained AF inductions, the first observed post‐S2 waves were 16 breakthrough (BT) and 4 rotor patterns (see Data S1 and Figure S3); the initial BTs appeared in sequences characterized by gradual decrease in activation intervals; rotors were observed in each of the 20 inductions leading to sustained AF (see Figure S4), but none in the inductions that failed to result in sustained AF (see Figure S5). During Initiation there was a trend for rotors to be more abundant in the RA (55.8±18.9%) than in the LA (44.2±18.9%, P=0.0621; see Figure S6). Thereafter the transition to Early Stabilization was characterized by a decrease in the RA free wall (from 35.5±18.7% to 25.6±16%, P=0.0121) and an increase in the LA roof (from 5.7±6.1% to 10.7±10.5%, P=0.0187) rotors presence and a trend to LA dominance over the RA (55.7±23.9% versus 44.3±23.9%, P=0.0670).

Acceleration of AF

Acceleration of activation rate during onset of sustained AF episodes was quantified by generating DF maps in 3‐s segments during the Initiation stage immediately after induction and during the Early Stabilization stage, 30 s later. Figure 2A shows representative DF maps from a sample heart in the course of AF onset. In this example, the DF map immediately following induction (0–3 s) shows a maximal DF (DFmax) domain of 9 Hz in the RA and most of the remaining atria activate at a DF of 8 Hz. Subsequent DF maps show an increase in DFmax to 11.5 Hz (4–7 s) and to 14 Hz (30–33 s) with reduced domain sizes indicating increased complexity of activity. Time course of DFmax values in 20 sustained AF inductions (Figure 2B) demonstrated a progressive averaged increase from an initial value of 11.71±1.82 Hz at 0 to 3 s to 13.45±1.88 Hz at 4 to 7 s (P=0.0006) and to 15.49±1.62 Hz at 30 to 33 s (P<0.0001) after induction. Thereafter, at 37 to 40 s post induction the DFmax values stabilized at 15.48±1.46 Hz (P=0.9553).

Figure 2

Time course of dominant frequencies (DFs) during AF onset.

A, Anatomy image and DF maps in a sample AF induction panoramic mapping of the RA and LA endocardial surfaces. From top to bottom panels: (1) The anatomy image of the panoramically mapped RA and LA. (2) DF maps in the respective atrium and at the indicated time periods (in s). Numbers in panels are sample DF values in Hz. Maximal DF (DFmax) site is seen to shift from the RA at 0 to 3 s to the LA afterwards. B, DFmax progression over time in all inductions (N=5 hearts, n=20 inductions). Colors indicate inductions in different hearts (1, 2, 4, 6, and 7 inductions/heart; Shapiro–Wilk test P=0.8584). Red markers indicate mean (horizontal bar), 95% CI of the mean (box) and SD (error bars) values. t test was performed in all the comparisons, except 30 to 33 versus 37 to 40 s, which was compared with Wilcoxon signed‐rank test. AF indicates atrial fibrillation; LA, left atrium; PVs, pulmonary veins; RA, right atrium; and SVC, superior vena cava.

Tracking the distribution of DFmax in the various RA and LA regions during the 20 AF inductions reveals that on average the posterior LA is hosting the atrial DFmax during all 4 periods of time analyzed, albeit not always significantly higher than the other regions (Figure S7). Interestingly, despite the posterior LA presenting the highest average DFmax in each period of time, in some inductions, other LA regions as the ridge, the free wall, and the roof presented similar or higher DFmax values. The LA highest DFmax values is also consistent with the lowest paced APD70 there (Figure S2) and with previous studies.

Local Acceleration at Rotors Sites

Analysis of time course of CLs in Figure S8 demonstrated gradual AF acceleration and stabilization consistent with DFmax time course in Figure 2. Thus we investigated the relationship between appearance of rotors and acceleration of the AF activation rate, as indicated by the decrease of CLmin anywhere in the atria during AF Initiation. Figure 3A shows representative phase snapshots before, during, and after the appearance of a rotor along with single pixel recordings. The phase maps show the appearance of a SP at time zero. The SP meanders and drifts for some time, as shown at 208 ms, until its disappearance shortly afterwards, before the time of the 438 ms map. Representative single‐pixel fluorescence recordings from 2 different locations near the meandering SP (sites a and b in inserts) show sequential CLs with a beat‐by‐beat variability. To characterize the time‐course variation in local CL relative to rotor appearance, the traces in Figure 3A were synchronized with their zero times coinciding with the appearance of the SP (solid red line). The traces demonstrate that the local CL near the SP increases from 105 ms during the first cycle of the rotor (red number, top trace), to 113 ms during the second cycle (red number, bottom trace).

Figure 3

Time ‐course of cycle length (CL) of rotors.

A, Representative example of CLs localized to a rotor site. Top: Phase maps showing a prerotor, rotor appearance, rotor drift, and postrotor snapshots at −117, 0, 208, and 438 ms (respectively; time is relative to the moment of rotor appearance). Inserts are magnifications of areas with rotors indicated by the arrowhead circles. Bottom: Sample fluorescence signals from 2 different pixels marked a and b near the SPs of the meandering rotor in the inserts. Numbers are beat‐by‐beat CLs in ms. Red numbers indicate the CLs of the rotor in the first (top trace) and second (bottom trace) in ms. Vertical red lines indicate the time of the phase snapshots. B, Time course of CLs of activity at the rotors sites during the Initiation (left, and Early Stabilization [right] stages of the AF onset [N=5, n=20 in each graph]). X axis is the cycle number of the activity relative to the moment of individual rotors’ appearances (cycle number=0; vertical red line). All CL values are presented as percentage of the prerotors averaged CLs (blue horizontal lines) at Initiation. Prerotor CLs were determined at the site of the rotor SP’s appearance. Thereafter the CLs were determined at different sites tracking the meandering SPs. Numbers indicate the numbers of CLs analyzed and shown as mean and SD. Dots represent rotors too infrequent for statistical comparisons. Red arrows: highlighting shortening of CLs following rotor appearance. Wilcoxon signed‐rank test was performed for all comparisons. AF indicates atrial fibrillation; and SP, phase singularity point. *P<10−4 versus prerotors averaged CLs.

The CLs at rotors sites during cycles before and during the rotors identified in 20 inductions of AF were collected using the tracking method shown in A of Figure 3. A total of 217 and 357 rotors detected during Initialization and Early Stabilization stages, respectively, appeared in series at rates of 2.0±1.37 versus 3.29±2.09 rotors/s during the 2 respective stages (P<0.0001; see Data S1 and Figure S9). Their number decreased with number of cycles lifespan before disappearance, as in previous studies. The rotors completed ≤6 cycles during Initiation and ≤10 cycles during Early Stabilization before dissipating, but the number of their cycles lifespan was not different (1.75±1.03 versus 1.76±1.2 rotations per rotor; P=0.9048).

Figure 3B shows the time course of the normalized local CLs for each cycle during Initiation (left) and Early Stabilization (right) as a function of cycle number, indexed with their zero as the last cycle before the rotor SP appearance. The average collective value of the local CLs in sites of SPs, during 10 cycles before the appearance of the SPs showed a reduction from 108.6±29.2 ms at Initiation to 91.8±31.1 ms at Early Stabilization (15.4% reduction of means, P<0.0001). Regression analyses of cycle‐by‐cycle 10 prerotor local CLs demonstrated a slow increase: 0.75 ms/cycle (P=0.0014) during Initiation and 0.38 ms/cycle (P=0.0333) during Early Stabilization. In contrast, the local CL following rotor appearance showed an abrupt acceleration with a reduction in the first rotation cycle to 97.7±20.3 ms (9.9%, P<0.0001) during Initiation and to 82.4±16.4 ms (10.2%, P<0.0001) during Early Stabilization. The average local CL of the first 2 cycles of rotors during Initiation (P<0.0001) and 3 cycles of rotors during Early Stabilization (P<0.0001) is 9.65% shorter than that during the past 10 cycles before the appearance of the SP (red arrows in Figure 3B). The local CL of succeeding cycles of the rotors either increased back to the prerotor values, or were too infrequent for statistical comparison.

Atria‐Wide CL Acceleration During Rotors Presence

The changes in CLs across the entire atria and AF onset during the specific times concurrent with the rotors are analyzed in Figure 4. A shows a synchronized sequence of panoramic phase (top) and CL (bottom) snapshots from the LA along with a single pixel recording. The phase maps include SPs at the pivoting center of either rotors or nonrotor (<1 rotation) patterns of activation. The CL maps show the spatial distribution of the CLs at a particular time and enable determination of minimal CL (CLmin) values across the atria pre‐ and postrotor presence. In this example, atrial CL before the rotor appearance is 110 ms and it decreases to 70 and 67 ms during the leftward drifting rotor presence. Once the rotor SP disappears a minimal atrial CL of 67 ms is observed.

Figure 4

The effect of rotors on the atrial CLs.

A, Fluorescence signal from a sample LA rotor site with corresponding phase (top row) and CL (bottom row) maps. Numbers superimposed on the fluorescence signal and in the CL maps indicate corresponding CLs in ms. Asterisks in CL maps indicate the site of the fluorescence signal near the rotor. The minimal atrial CL is seen in this sample drifting rotor to shorten from a value of 110 ms before the rotor formation to 67 ms it disappears (see CL maps at −100 and 330 ms respectively). Prerotor and postrotor black arrow markers in the fluorescence trace indicate times of determining atria‐wide CLmin in analysis. Solid arrowhead circles: location of a sample clockwise rotor in phase maps superimposed in same location also on CL maps. Dashed line circles: locations of sample nonrotor SPs. B, Alterations in the atrial CLmin of the AF during the presence of rotors (left bar graph) and nonrotors SPs (<1 rotation; right bar graph). CLmin values were averaged for all pre‐ and postrotors and nonrotor (SPs) activities in each individual 20 AF inductions in 5 hearts and are presented as mean±SD percentages of the preactivities CLs during the Initiation stage of AF onset of the corresponding induction. Number of rotors: 10.5±3.4 rotors/induction at Initiation and 18.1±6.8 rotors/induction at Early Stabilization. Nonrotors: 15.7±1.9 SPs/induction at Initiation and 17.4±1.2 SPs/induction at Early Stabilization. Red arrow: 4.5±6.8% (P=0.0042) shortening of CLmin of the AF from prerotor to postrotor values during Initiation. Green arrows: Shortening of CLmin of the AF from Initiation to Early Stabilization average values for both rotors (12.4±8.8%, P<0.0001) and non‐rotors (15.3±7.2%, P<0.0001) activity. t test performed for all comparisons. AF indicates atrial fibrillation; CL, cycle length; CLmin, 5% minimal cycle length; and SP, phase singularity point.

We collected atria‐wide CLmin values immediately before and after multiple SPs in 20 AF inductions to investigate whether SPs of rotor and nonrotor patterns were associated with acceleration of activation rate. Figure 4B shows cumulative analysis of the atrial CLmin values immediately before and after 572 rotor SPs and 661 nonrotor SPs for all inductions (28.6±9.3 and 33±2.6 SPs per induction, respectively). To account for interinduction variability, CLmin data was normalized by the average of the values measured immediately before appearance of rotors and nonrotors SPs during Initiation at each induction separately. During Initiation, the average atrial postrotor CLmin was 95.5±6.8% of the atrial prerotor CLmin (P=0.0042, red arrow) and the atrial nonrotor post‐SP was 99.7±1.5% of the atrial pre‐SP CLmin (P=0.1737). During transition from Initiation to Early Stabilization, atrial CLmin of both rotor and nonrotor SPs accelerated on average for both rotors (12.4±8.8%, P<0.0001) and nonrotors (15.3±7.2%, P<0.0001; green arrows). However, in contrast with the Initiation, during Early Stabilization there was no reduction in atrial CLmin between the appearance and disappearance of the SPs (rotors: 84.9±11.0% to 85.8±10.8%, P=0.4029; nonrotors: 84.5±7.3% to 84.6±7.4%, P=0.4921, respectively).

Variability in Atria‐Wide CL During Rotors Presence

The difference between pre‐ and postrotors CL values across the atria varied greatly among inductions and stages of AF. Although the example in Figure 4A shows atrial postrotor CL reduction (acceleration), another example shown in Figure S10 is demonstrating an atrial postrotor CL increase (deceleration) during the presence of a rotor that drifts less than the one in Figure 4A. In Figure S11, atria‐wide CLmin before specific rotors were compared with CLmin subsequent to the same rotors. Analysis of all prerotor and postrotor atrial CLmin values revealed that 53.8% of rotors during Initiation associated with CLmin reduction and the rest with CLmin increase. The entire group of rotors during Initiation of AF associated with a small but significant atrial CLmin net decrease (acceleration) from 90.2±32.3 to 87.3±33 ms (P=0.0439. See Discussion section). On the other hand, a similar analysis of all prerotor and postrotor atrial CLmin values during Early Stabilization showed no net difference (prerotor: 77.6±26 versus postrotor: 78.7±26.9 ms, P=0.1671). Overall, the presence of rotors that accelerate locally was distinctively associated with a net atria‐wide acceleration of AF at Initiation versus a stable AF rate at Early Stabilization (P=0.0112, see Figure S11). In contrast to the rotors, the nonrotor SPs did not associate with cumulative reduced CLmin either during Initiation or during Early Stabilization (Figure S12).

The Dynamics of Rotors Drift and Atria‐Wide CLs

Phase maps in Figure 4A show atria‐wide CL acceleration concurrent with a rotor that drifts more than in atria‐wide CL deceleration (Figure S10). Thus in Figure 5 we quantify the relationship between rotors drift distance and velocity with the atria‐wide CLmin acceleration and deceleration. A shows a general abbreviation of CLs in the LA peripheral sites that are activated by waves emanating from a drifting rotor (see Video S6). The distance and velocity of the rotor drift were measured along a vector connecting the beginning and ending locations of the SP trajectory (see Data S1). The drift data were further correlated with the atria‐wide CLmin acceleration or deceleration during the Initiation and Early Stabilization stages in Panel B. The graphs show that when considering all the rotors together, their drift distances decreased from 21.5±15.8 mm at Initiation to 17.9±12.2 mm at Early Stabilization (P=0.0015), but their drift velocities did not alter during that transition and remained at 0.10±0.07 mm/ms; (P=0.4442). Comparing between atria‐wide acceleration and deceleration during Initiation revealed a trend for larger distance (23.1±15.7 versus 19.5±15.8 mm, P=0.0523) and significantly larger velocity (0.11±0.07 versus 0.09±0.06 mm/ms, P=0.0103) of drift in rotors with atria‐wide acceleration versus deceleration (red arrows). In contrast, during Early Stabilization there were no observed differences in drift distance and velocity in the atria‐wide acceleration versus deceleration groups.

Figure 5

Rotor drift and atria‐wide CL alterations.

A, Top: sequential phase maps showing LA activation patterns during the presence of a rotor drifting from start (site begins at 1776 ms after S2) to finish (site ends at 2000 ms; black SP trajectory is superimposed on all maps). Begin‐to‐end drift vector is shown in leftmost map with distance (vector magnitude) and velocity along the vector (see Data S1). Arrowhead circles indicate location of SP. Bottom: single pixel fluorescence traces from sites near the SP trajectory (a) and the periphery (b–d) with CLs in ms. CLs in sites b and c are abbreviated following activating waves emanating from the rotor and SP trajectory area (red arrows). B, Left: Rotor drift distance (begin‐to‐end vector magnitude) for all rotors (All; n=210, n=362), rotors with atria‐wide CLmin acceleration (Accel; n=113, n=156) and rotors with atria‐wide CLmin deceleration (Decel; n=97, n=206) for Initiation and Early Stabilization (n respectively). Symbols are mean±SD. 95% CI for each distance group is ≤3.1 mm. Right: Rotors drift velocity for same conditions, n numbers and symbols similar as for the left graph. 95% CI for each velocity group is ≤0.0127 mm/ms. Red arrows, indicate trend (left graph) and significant (right graph) decrease differences. CL, cycle length; CLmin, 5% minimal cycle length; LA left atrium; and SP, phase singularity point.

Spatial Distribution of Rotors With Atria‐Wide Acceleration

In Figure 3 we demonstrate that on average all rotors abbreviate the CL near their core, but Figure 4 and Figure S11 show that only portions of the rotors abbreviate the atrial CLmin and accelerate the AF, thus raising the question of whether only certain regions are dominating the acceleration. In Figure 6 we demonstrate the regional distribution of those rotors that exclusively associate with AF acceleration in the 20 AF inductions. It is shown that all regions harbor such rotors but with varying proportions. On average, at Initiation the RA tended to have more rotors than in the LA (33.8±17% versus 23±15.2% respectively, P=0.0429) and their numbers became nonsignificantly different at Early Stabilization (22.8±12.8% versus 24±9.7% respectively, P=0.6948). The main regions responsible for the RA‐LA equalization in the number of AF acceleration‐associated rotors in Early Stabilization were the RA free wall, showing a reduction from 22.4±16.7% to 13.9±10.3% (P=0.0252), and the LA roof, showing an increase from 2.3±3.4% to 5.3±6% (P=0.0084).

Figure 6

Regional distribution of rotors accelerating the atria.

Stacked bars show the amount of rotors in each region in cases where an atrial CLmin at their termination was less than the atrial CLmin at their appearance (red cases in Figure S11). Distributions are shown for Initiation and Early Stabilization. Regional rotors amounts are expressed as mean±SD percentage of 20 AF inductions in 5 hearts from the total number of atrial rotors analyzed during those Initiation and Early Stabilization stages (210 and 253 respectively). Comparisons were made by t test for normal distributions (Shapiro–Wilk test) and by Wilcoxon signed‐rank test for nonnormal distributions. Statistical significance level is alpha=0.025 for dual comparisons. AF indicates atrial fibrillation; CLmin, 5% minimal cycle length; LA, left atrium; PLA, posterior left atrium; RA, right atrium; and SVC, superior vena cava.

Time Course of Complexity During AF Onset

The CL acceleration at the rotors sites and across the atria during arrhythmias onset found in our experiments should be linked to an increase in wave complexity. To demonstrate such conjecture, we counted the number of SPs per frame during 5‐s periods of Initiation and Early Stabilization. Figure 7A shows the number of SPs counted in bins of 100 ms during Initiation (left) and Early Stabilization (right) stages of a sample AF induction. In this example, the number of SPs gradually increased during Initiation at an average rate of 3.93e‐3 SP/ms (P<0.0001) and then their number stabilized with an average of 17.39 SPs for every 100 ms (P=0.795). In B we summarize the rate of change in SP numbers over 20 AF inductions. On average, during both Initiation and Early Stabilization of AF in our model the number of SPs increases with time, but the rate of increase is reduced from 2.76e‐3 SP/ms during Initiation to 1.30e‐3 SPs/ms during Early Stabilization (P=0.0056). The complexity of the AF, as quantified by the number of SPs, was further compared to the minimal CL in the atria at the moment SPs were detected throughout Initiation and Early Stabilization. The logarithmic scatter plot in Figure 7C shows the relationship between the number of SPs in 100 ms bins versus the atrial CLmin in that bin for each heart separately. Although each heart shows a different trend, all showed a negative log‐log slope, indicating a similar behavior of an increase in the number of SPs for shorter CLs. A linear regression model for log‐log data on 1875 SPs and their instantaneous CLs in the 20 inductions in 5 hearts (not shown) had a slope of −4.7667±0.1531 and an intercept of 5.7594±0.0358 (P<0.0001).

Figure 7

Time course of number of SPs during AF onset.

A, Representative example of number of SPs counted in bins of 100 ms during the Initiation (left) and Early Stabilization (right) stages of AF induction. Symbols show means and SDs of SPs per frame. Linear best fit formulas, lines and 95% CIs are superimposed on graphs. B, Summary of rate of change in SP numbers in 20 AF inductions in 5 hearts. Red markers indicate mean (horizontal bar), 95% CI of the mean (box) and SD (error bars) values. C, A logarithmic scatter plot between the number of SPs in 100 ms bins (n=1875) versus the CLmin in that bin for all 20 inductions in 5 separated hearts. SPmin=minimal number of SPs in a heart. Min(CLmin)=the minimal CLmin among all AF inductions in a heart. AF indicates atrial fibrillation; APD, action potential duration; BT, breakthrough pattern of impulse propagation; CL, cycle length; CLmin, 5% minimal cycle length; DF, dominant frequency; RA, right atrium; and SP, phase singularity point.

Discussion

Main Results

We developed a new panoramic optical method enabling simultaneous mapping of activation patterns across intact atria‐wide endocardial surfaces as never before. Using the new system to study the onset of cholinergic AF in the isolated sheep heart, our main observations were that, first, the BTs or wavebreaks following premature stimuli can occur remote from pacing in either atrium, and their succeeding CLs are gradually shortening only when rotors appear and the subsequent AF is sustained. Second, serial short‐lasting rotors temporarily accelerate activation in their vicinity, but that acceleration is associated with a net atrial‐wide acceleration only during the Initiation stage of the AF and is followed by a net equilibration between acceleration and deceleration maintained during the Early Stabilization stage. Third, distances and velocities of rotors’ drifts during atria‐wide CL acceleration are greater than during deceleration. And finally, wavebreaks that do not develop into reentries do not associate with net atrial acceleration at any stage, but their number, and hence AF complexity, is increasing with time and atrial acceleration until the Early Stabilization stage.

Wide‐View Endoscopic Optical Mapping of AF

The optical signals generated by fluorescence of voltage sensitive dyes in cardiac cells can map their activity with high reliability and spatiotemporal resolution. Although areas covered by optical mapping can typically capture propagation features of fibrillation, they are restricted to visible regions on the atrial epicardium or required a dissection of the atria to map the endocardium, with consequential artificial boundaries and possible ischemia, limiting investigation of epicardial‐endocardial activation differences. To avoid the atrial dissection we previously developed an endoscopic approach to map the posterior LA, which is of a major clinical importance. , As clinical mapping of extracellular potentials have demonstrated possible reentrant and focal drivers of AF distributed over the entire atrial endocardium, , , , it is desirable to use also panoramic optical mapping for a more comprehensive investigation of transmembrane impulse propagation during AF. , , Thus, we developed here a unique objective lens assembly attached via a borescope to a charge‐coupled device camera for a panoramic optical mapping of the entire endocardial surfaces viewed above the valves rims of the intact 2 atria. The panoramic view enables tracking the complete trajectory and life span of maximal number of endocardial rotors from their formation to termination.

Activation Patterns and Rotors in AF

Experimental and clinical studies have proposed that rotor activity plays a driving role in the maintenance of AF. , In a cholinergic and persistent sheep models of AF , , rotors have been found to be localized to areas at or adjacent to the highest DF domains during self‐sustained and stable AF, but it is not known what would be the dynamics of those rotors during the onset of AF. Here we tested the hypothesis that the rotors play a role in the acceleration of the activation rate during the transition from the organized and slow rhythm toward the faster and more complex fibrillation. Our panoramic mapping study is demonstrating a dynamic process during AF onset whereby rotors appear serially, are in general short living and discontinuous, and their CLs is shortening by an average of 9.65% in their first 2 to 3 cycles relative to the average CLs before their formation (Figure 3). Despite equal rotor lifespan and local acceleration (Figure 3), the rotors were associated with different atria‐wide acceleration dynamics between the Initiation and the Early Stabilization stages of the AF (Figure 4 and Figure S11); greater rotor drift distances and velocities were associated with AF acceleration during Initiation (Figure 5). In contrast, the activity following the formation of SPs that do not complete at least 1 rotation (ie, wavebreaks) did not show a cumulative CLs shortening at any stage of the AF, and their multiplication is suggested to be secondary to the rotor‐induced acceleration toward the Early Stabilization of the arrhythmia (Figures 4, 7, and Figure S12).

Our findings are consistent with other studies in patients and simulations showing evidence for rotational activity linked to transient highest activation rate and dominant frequency. , , However our study further finds for the first time that the acceleration at the rotors sites, although brief, associates with the progression of the AF. It was found that following an initial brief breakthroughs period, AF accelerations occurred only when rotors were observed and were otherwise absent during the 10 prerotor cycles (Figure 3B) and during the nonrotors wavebreaks (Figure 4B and Figure S12).

The acceleration of the atria surrounding the accelerating rotors is not universal. Inspection of each rotor separately (Figure S11) finds that in some cases the postrotor activity at surrounding tissue accelerates relative to the prerotor rate, and in other cases the activity decelerates. Data presented in Figure S11 show that each rotor during the Initiation stage reduces the atrial CLmin by a small, but statistically significant, average amount of 90.2 − 87.3=2.9 ms. This small net average acceleration during a rotor presence, if confirmed, may accumulate in time and explain the large atria‐wide reduction of CLmin (Figure S8) and increase of DFmax (Figure 2B) during the Initiation stage. The balance between the postrotor accelerating and decelerating atrial activity was found to shift from a small dominating acceleration during Initiation of the AF, to an equilibration during the Early Stabilization stage at about 30 s into the AF. For comparison, in the cases of nonrotors SPs, the atria‐wide acceleration is balanced with the deceleration at all stages of the AF onset (Figure 4B and Figure S12).

Mechanisms of Rotor‐Induced Acceleration and Deceleration

During the AF onset, CL accelerations occurred when rotors were observed, but not all accelerations colocalized with the rotors. The CL abbreviation at the rotor sites has been proposed to be mediated by electrotonic effects abbreviating the APD near the core of the pivoting waves reinforced in our cholinergic AF model by the enhanced inward rectifier K+ currents. Uninterrupted waves emanating from the rotor will abbreviate CL in its periphery (Figure 5A). In case of rotors drift as shown in Figure 5, the atria‐wide activity may be further abbreviated via the Doppler effect and advancement of local cycle phase of the rotating waves. However, the acceleration‐deceleration fate of the postrotor atria‐wide rate of activity may be the outcome of additional mechanisms related to rotors activations themselves as follows: On one hand, APDs and refractory periods shorten as membrane ionic currents adapt to fast rotor‐induced activations and could remain abbreviated after the rotor termination by a memory effect. , On the other hand, computational studies on dynamics of rotors in the ionic heterogeneous atrial tissue have demonstrated that rotors slow down their rotation rate as they drift toward areas of lower excitability, which is also strongly affected by the spatial distribution of the inward rectifier current. Noticeably, mechanisms underlying the opposing atrial acceleration and deceleration during and by rotors coexist and their net effect on the AF acceleration may not require a continuous presence of rotors. The panoramic optical mapping approach employed here, in combination with studies on ionic and structural properties, , , could be used to better elucidate on the acceleration rate across the atria during AF onset.

Clinical Implications

In clinical electrophysiology AF is defined when the arrhythmia lasts ≥30 s, but 30‐s long episodes have been consistently found and anticoagulation is typically initiated once AF episodes of 30 s or longer are present, highlighting the risk and the need to better understand, and possibly prevent, brief AF episodes as studied here. Our study can particularly shed light on the etiology of vagal‐mediated paroxysmal AF, which is observed in the absence of detectable heart disease in young male adults. , , In general AF in the clinic is often seen to initiate following focal activity from a varying origin. , , , , , The pulmonary veins have been described as the most common area of AF triggers, but triggered activity has been observed elsewhere as well and pulmonary vein isolation, the most common clinical technique for AF ablation, is suboptimal in terminating AF, and its efficacy is further reduced as AF progresses to persistent and permanent forms. , We find initial BTs and rotors across the 2 atria immediately following the premature pacing (Figures S3 and S4), consistent with the suggestion that discharges and driving activity can appear across the entire atria and in locations remote to the triggering event, which warrant a panoramic mapping. Our panoramic mapping further demonstrates that AF occurs following BTs and series of short lasting rotors, but only some of which associate with atria‐wide acceleration. The possible presence of initial BTs and specific subsets of serial rotors contributing to the net result of atria‐wide acceleration could potentially be a more specific target for improved ablative and nonablative therapies.

Limitations

The study investigates rotor dynamics during onset of AF in healthy isolated sheep hearts in the presence of 0.25 µmol/L carbachol. AF reentrant activity has been observed in other animal models and patients, under conditions such as increased intra‐atrial pressure and presence of remodeling, and we cannot exclude different dynamics of acceleration in onsets of noncholinergic AF. The new wide‐view lens used allows unprecedented panoramic mapping of endocardial surfaces. However, the optical mapping approach does not inform on activation patterns inside the thick myocardium and we also did not map the epicardial surfaces. As such, rotor activity and wavebreaks localized deeper than the subendocardial surfaces as well as on the epicardium and the distal portions of the appendages and veins could not be detected and analyzed. Furthermore, because of the inability of our optical mapping approach to detect mid‐myocardium and epicardial activity we cannot exclude the possibility that detected BTs represent the exit site of intramural reentries or wavebreaks. Finally, the differences found in the numbers of rotors with atria‐wide acceleration versus deceleration actions, as well as in the net results of accelerations, are small and would benefit from additional confirmation studies.

Conclusions

A new system for panoramic and simultaneous optical mapping of the endocardial surfaces of the intact RA and LA provides the most comprehensive perspective on the dynamics of AF initiation to date. The new mapping system demonstrates that the DF increase during the onset of cholinergic AF in the sheep model correlates with a net result of opposing atria‐wide acceleration‐deceleration actions occurring concomitantly with rotors. The results suggest that the cumulative effect of drifting rotors, but not of multiple wavebreaks, could underlie the AF acceleration during its onset.

Sources of Funding

This work was supported in part by the National Institutes of Health National Heart, Lung, and Blood Institute grants R01‐HL118304 and R21‐HL153694 (Berenfeld), and R01‐HL122352 (Jalife); the Michigan TRAC Life Sciences program (Berenfeld); the Leducq Foundation (Jalife, Berenfeld, and Pandit); the University of Michigan Health System and Peking University Health Sciences Center Joint Institute for Translational and Clinical Research (Jalife); Ministerio de Economía y Competitividad and Fondo Europeo de Desarrollo Regional (FEDER; Jalife); the JHRS fellowship program from Medtronic Japan, Uehara Memorial Foundation (Takemoto); American Heart Association postdoctoral fellowship (Takemoto); research grants from Gilead Sciences Inc (Jalife and Pandit); fellowship by Fundación Martín Escudero (Salvador‐Montañés).

Disclosures

Dr Jalife served as a consultant for Topera‐Abbott Laboratories and was a cofounder of Cartox, LLC. Dr Berenfeld was cofounder and Scientific Officer of Rhythm Solutions, Inc., Research and Development Director for S.A.S. Volta Medical, consultant to Acutus Medical and is a cofounder of Cor‐Dx LLC. Research grants were provided from Abbott, Medtronic Inc (Jalife and Berenfeld) and CoreMap Inc. (Berenfeld). None participated in this study. The remaining authors have no disclosures to report.

Data S1

Figures S1–S12

Videos S1–S6

References 46, 47, 48, 49

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.