Dynamic Electrophysiological Behavior of Human Atria During Paroxysmal Atrial Fibrillation
Background The aims of our study were to investigate the meaning of local atrial activation and its behavior during paroxysmal atrial fibrillation and to study the effect of overdrive pacing on local atrial activity.
Methods and Results Twenty-five patients with lone paroxysmal atrial fibrillation underwent electrophysiological study. Functional and effective atrial refractoriness was determined. Mean and fifth percentile values of 100 consecutive atrial fibrillation intervals (FF) were evaluated at three atrial sites either at arrhythmia onset or at self-termination (or at minute 5). A high-voltage burst pacing was performed after 6 minutes of stable atrial fibrillation in 10 patients. Mean FF intervals were evaluated 5 seconds before and after atrial pacing. Forty-nine atrial fibrillation episodes were induced: 39 self-terminating within 5 minutes and 10 long-lasting. A significant correlation was found between mean FF and atrial functional refractory period (r=.73, P<.001) and between fifth percentile FF and atrial effective refractory period (r=.57, P<.005). Atrial fibrillation self-termination was associated with significant mean FF prolongation, whereas long-lasting fibrillation behaved the opposite. In 10 patients, burst pacing resulted in significant shortening of the mean FF at the stimulation site; no changes were observed in the two distant recording sites.
Conclusions The analysis of the FF intervals demonstrates a strict correlation with atrial functional refractoriness. The self-termination of atrial fibrillation is related to a prolongation of the functional refractoriness (mean FF), whereas a shortening of both functional and effective refractoriness (fifth percentile) is associated with atrial fibrillation persistence. The provoked shortening of the mean FF at the stimulation site is consistent with the presence of a gap of excitability during atrial fibrillation in the human atria.
The electrophysiological mechanisms of atrial fibrillation (AF) onset in humans have been referred to as dispersion of refractoriness or conduction within the atria,1 2 3 4 5 6 7 8 9 10 11 12 13 14 evaluated primarily during atrial stimulation by the use of single or multiple delivered atrial extra beats. The results of these studies, however, give us the opportunity to evaluate only the behavior of the aforementioned parameters at a chosen basal heart rate and during a regular rhythm, which is a far different condition than an irregular and high-rate rhythm such as AF. In other words, those studies do not give information on the dynamic behavior of atrial conduction and refractoriness that may occur at high atrial rates during AF. A relation between local atrial activation during fibrillation and atrial refractoriness was hypothesized, and clues were subsequently proposed in experimental and clinical studies.9 15 Moreover, because it was demonstrated that a transient linking of atrial excitation during AF is at least a local phenomenon,16 17 it is conceivable that we can obtain local capture of the atrium by pacing during the most regular phases of atrial activation. Given that evidence in humans, the existence of an excitable gap during AF could eventually be inferred, as demonstrated in the experimental setting.16 17
The aims of our study were to evaluate the electrophysiological meaning of the local atrial activation during AF, together with its dynamic behavior, and to evaluate the effect of atrial overdrive pacing on atrial activation during AF.
The study population comprised 25 consecutive patients (20 men, 5 women) ranging in age from 12 to 69 years (mean, 41±12 years) with frequent self-terminating episodes of AF and without any demonstrable organic heart disease. AF was previously documented on multiple Holter recordings and on hospital admissions; the patients were referred to our outpatient clinic during 1989 through 1994 for weekly recurrent episodes of AF. A careful clinical history and repeated Holter recordings of the clinical arrhythmia did not help to recognize any peculiar pattern of AF onset in any patient. Each patient underwent clinical evaluation, two-dimensional echocardiography, stress test, stress 201Tl tomography, and thyroid hormone plasma level evaluation. All these tests had to provide normal results for the patients to be enrolled in the study. AF was clinically well tolerated in all patients (ie, causing significant limitations only at a very high ventricular rate or during effort) primarily because of the fairly young age of and the absence of organic heart disease in the study population.
We specifically studied patients with lone paroxysmal AF (LPAF) because the arrhythmia could be induced without significant hemodynamic impairment and because it was usually not long-lasting, which allowed us to avoid electric cardioversion. Furthermore, because of the possibility of easy inducibility and self-termination, it was possible to follow the atrial activation pattern in all the main phases of each AF episode (Fig 1A⇓ and 1B⇓).
Every drug that could eventually interfere with atrial electrophysiological properties was discontinued at least five half-times before the beginning of the study. No patient was evaluated while under or after amiodarone therapy.
A group of 10 healthy patients undergoing electrophysiological study for reasons other than AF who never experienced a spontaneous episode of AF (9 men, 1 woman), ranging in age from 27 to 63 years (mean, 45±10 years), served as the control population.
Patients underwent an electrophysiological study in the nonsedated state. We used 6F USCI electrode catheters with a 5-mm interelectrode distance. Two quadripolar catheters (Josephson type) were placed in the high right atrium (HRA) and in the coronary sinus (CS) or the left atrium through a patent foramen ovale to record the atrial activity. The His bundle electrogram and low right atrium (LRA) recording were obtained by a tripolar catheter placed just across the tricuspid valve. A fourth quadripolar catheter was placed in right ventricular apex for assistance in drug response evaluation at the end of the study. Endocavitary atrial signals were recorded primarily as bipolar, but in 3 cases unipolar signals were recorded during AF to avoid misinterpretation of the local atrial activity because of fragmented waves that made measuring FF intervals difficult.18
Endocavitary and ECG leads I, aVF, V2, and V5 were recorded simultaneously on a seven-channel Siemens Mingograph at a standard paper speed of 100 mm/s. Programmed stimulation was performed by use of a square wave at 2.5 diastolic threshold and 2-ms duration. The stimulation protocol included the single extrastimulus method at basic cycle length and at incremental cycle lengths of 600, 500, and 400 ms. HRA was tested as the first stimulation site; then CS and LRA were tested. This sequence could have influenced our percentages of inducibility at the different sites as reported below.
Data Acquisition and Signal Processing
Two body surface ECG leads were recorded simultaneously with the three atrial electrograms on a seven-channel FM tape recorder (TEAC MR-30) and later digitized (1-kHz sampling frequency) for interval measurement. A computer program was developed to measure beat-to-beat atrial intervals (FF) from each recording site. The computer algorithm was divided into two main steps: atrial wave detection and local activation time measurement. For atrial wave detection, an adaptive threshold on the filtered intra-atrial signal was used.19 To avoid baseline drift, the algorithm first performed an incremental difference; then an energy collection was applied to enhance atrial waves. Atrial activation wave was detected when the filtered signal exceeded a threshold. To follow morphological changes of the atrial signal, threshold level and the other algorithm parameters were updated beat-to-beat. For each detected atrial activation wave, the local activation time was determined on the original signal, with 1-ms resolution, by specific algorithms. For bipolar recordings, the algorithm searched for the greatest amplitude of the atrial wave deflection; for unipolar electrograms, local activation time was measured on the steepest intrinsic deflection. To prevent double counting, detection was discontinued for a sufficient lag time after detection of a local activation. The electrograms were displayed on a computer graphic screen, together with the local activation times, to check for errors or missing activation marks. The operator might correct the computer output by deleting or adding activation times. In the few cases in which the automatic procedure erred too many times, atrial intervals were measured manually on the computer screen by positioning a cursor on the local atrial activation waves.
The RR interval sequence was also automatically measured by a similar procedure.19 Blood pressure was monitored with a femoral arterial sheath in all patients.
To obtain a relatively stable situation, we evaluated only fibrillatory episodes lasting longer than 1 minute. Self-termination was defined as spontaneous resumption of sinus rhythm within 5 minutes. An AF episode was defined as long-lasting if it lasted >5 minutes. At every recording site, we measured 100 consecutive FF intervals either at the onset or at the termination (or at minute 5 in the long-lasting cases) of each AF episode.
Mean FF (MFF) and the SD of FF values were calculated in all the episodes on the basis of either all 100 intervals or a subgroup of 20 FF intervals. To represent statistically the shortest FF intervals in the AF interval distribution, the fifth percentile (P5) value was used.
Rate-corrected sinus node recovery time (cSNRT) was calculated from the difference between sinus node recovery time and basic sinus cycle length.
Atrial vulnerability was assessed by programmed electric stimulation with 1 extrastimulus during sinus rhythm and atrial pacing at the three aforementioned cycle lengths.
The extrastimulus was delivered after 8 paced beats late in diastole, and the coupling interval was shortened by steps of 5 ms and finally of 1 ms until the effective atrial refractory period was reached. The atrial effective refractory period (ERP) was defined as the longest attainable S1S2 interval that did not evoke an atrial depolarization. The atrial functional refractory period (FRP) was measured with the proximal electrode of the stimulating catheter and was defined as the shortest attainable A1A2 interval. For each cycle length, refractory periods were compared between AF patients and control subjects only when obtained at the same atrial site. The AV node ERP was defined as the longest attainable A1A2 interval that did not evoke His depolarization. AV node FRP was the shortest attainable H1H2 interval evaluated in HBE.
ΔA1A2 was defined the maximum difference between the lengths of A1A2 at the stimulation site and at each distant atrial recording site. The A1A2 delay zone was defined as the range of S1S2 in which A1A2 showed a lengthening compared with S1S2. ΔA2 was defined as the maximum increase of A2 duration with respect to A1. The A2 delay zone was defined as the range of S1S2 in which A2 showed an increase compared with A1.
To represent atrial vulnerability, the vulnerable parameter adopted by Attuel et al13 was evaluated. This is defined as refractoriness times interelectrode distance divided by A2 duration, where the ratio of interelectrode distance to A2 duration gives an estimate of atrial conduction velocity. In our setting, latent atrial vulnerability index (LVI) was defined as LVI=ERP×0.5/A2 cm.
Diagnosis of AF was based on the surface ECG with the following criteria: no discrete P waves in any surface lead and F waves that were irregular in timing and morphology at a rate >320 beats per minute (bpm). These criteria were validated by the endocardial recordings that detected irregular atrial activation not separated by an isoelectric line (except for Wells’ type I AF)20 and with an SD of the FF interval >10 ms. Moreover, RR intervals had to be irregular, and no periodic pattern of the FF intervals could be present.21 22
We calculated atrial effective and functional refractoriness at the three aforementioned cycle lengths; we measured atrial FF intervals during AF either at the onset of the arrhythmia or before self-termination. If the AF episodes were long-lasting, measurement of the FF intervals was performed after 5 minutes. In 10 patients, a burst of atrial stimulation of 2-ms duration to 10 V was delivered twice at a rate of 750 to 1250 bpm invariably after minute 6 of the long-lasting AF episodes. The FF intervals covering the 5 seconds immediately preceding and those immediately following burst stimulation were measured at the stimulation site and at the distant recording sites.
Atrial and AV node refractory periods and the electrophysiological parameters at AF induction were compared with those measured in the control group at the calculation of the effective refractoriness by one-way ANOVA. When significant differences were detected, data were analyzed by Bonferroni’s t tests.
A correlation test was performed for both mean and P5 FF interval (evaluated at the induction site for either 100 or 20 consecutive intervals) and atrial FRP. The same correlation was tested with atrial ERP. Data on FF intervals were compared with refractoriness evaluated at the same recording site. Because of the easy inducibility of the AF episodes, an effective refractoriness was measured in 22 of 25 patients.
The spontaneous behaviors of MFF and the corresponding mean RR interval were investigated by ANOVA in both self-terminating (onset versus termination) and long-lasting (onset versus 5 minutes) AF episodes; when significant differences were found, data were analyzed by Bonferroni’s t tests. P5 behavior in the site of induction was also evaluated by the same statistical method. The effects of burst stimulation on MFF were evaluated by the same statistical method in the three recording sites.
Blood pressure recordings immediately before and at AF induction were compared by a paired t test.
Seventy-six AF episodes were induced; 49 fulfilled the criteria for sustained episodes (lasting >1 minute) and 27 did not; hence, these latter were discarded from subsequent analysis.
A total of 49 sustained atrial fibrillation episodes were recorded among the 25 AF patients; 35 (71%) were induced in HRA and 14 (29%) were induced in CS. Fifteen (31%) were induced at the 600-ms cycle length, 12 (25%) at the 500-ms cycle length, and 22 (44%) at the 400-ms cycle length. Thirty-nine were spontaneously terminating within 5 minutes; 10 were long-lasting. At least 1 AF episode was elicited by a single extrastimulus in all 25 patients. The average duration of self-terminating AF episodes was 3 minutes, 31 seconds±39 seconds (range, 2 minutes, 24 seconds to 4 minutes, 49 seconds); it was 36 minutes, 10 seconds±433 seconds (range, 28 minutes, 4 seconds to 48 minutes, 32 seconds; P<.001 versus self-terminating) for the long-lasting episodes.
AF was never induced in control subjects.
Basic Electrophysiological Parameters
No conduction disturbances were observed in terms of P-wave duration and PA, AH, and HV intervals; sinus cycle length and cSNRT were within normal limits in both groups (Table 1⇓).
No significant differences were found for AV node FRP (319±40 versus 345±37 ms at 600 ms, 318±44 versus 340±56 ms at 500 ms, and 320±40 versus 301±115 ms at 400 ms) and AV node ERP (268±24 versus 277±35 ms at 600 ms, 265±7 versus 278±52 ms at 500 ms, and 270±31 versus 244±93 ms at 400 ms) between fibrillating patients and the control group.
ARPs were shorter in AF patients than in control subjects at each cycle length: 240±30 versus 281±35 ms at 600 ms (P<.005), 227±30 versus 267±42 ms at 500 ms (P<.05), and 210±59 versus 266±37 ms at 400 ms (P<.01) for the atrial FRP and 193±23 versus 231±36 ms at 600 ms (P<.001), 190±24 versus 231±32 ms at 500 ms (P<.05), and 178±48 versus 234±34 ms at 400 ms (P<.001) for the atrial ERP.
There were significantly longer atrial conduction times in the fibrillating atria compared with controls as evidenced by greater ΔA1A2 (Table 2⇓). In particular, ΔA1A2 was prolonged in the fibrillating group: 38±11 versus 7±8 ms (P<.001).
The LVI was significantly shorter at all three cycle lengths in AF patients compared with control subjects (range, 1.75 to 2.6 cm at 600 ms, 1.78 to 2.58 cm at 500 ms, and 1.72 to 2.53 cm at 400 ms for AF patients and 3.53 to 5.8 cm at 600 ms, 3.38 to 6.57 cm at 500 ms, and 3.1 to 5.2 cm at 400 ms for control patients; Table 2⇑).
Relation Between FF Intervals and Atrial Refractoriness
A significant correlation was found between the mean of 100 FF intervals calculated at AF onset and the atrial FRP evaluated at the same recording site (r=.73, P<.001; Fig 2A⇓). This degree of correlation was lower when only 20 FF intervals were considered (r=.46, P<.01) or when the mean of 100 FF intervals was correlated with effective refractoriness (r=.54, P<.05). A lower, although still statistically significant, level of correlation was also found between the atrial ERP and the P5 of 100 consecutive FF intervals (r=.57, P<.005; Fig 2B⇓). There was no significant correlation when only the first 20 FF intervals were evaluated (Table 3⇓).
Dynamic Behavior of FF Intervals During AF
The dynamic behavior of AF intervals was analyzed with only the mean of 100 consecutive intervals because it was demonstrated that only this parameter fitted with atrial functional refractoriness at a high degree of correlation; the mean of 20 FF intervals showed a poorer correlation.
The mean of 100 FF values prolonged shortly before the arrhythmia terminated spontaneously in all three atrial recording sites in the self-terminating episodes (Table 4⇓, Fig 3⇓). It shortened at minute 5 in the long-lasting cases (Table 4⇓, Fig 4⇓). No major dispersion of the MFF values contemporarily evaluated in the three atrial recording sites was observed. The corresponding mean RR interval showed no significant variation from AF onset to its self-termination or at minute 5 in long-lasting episodes (Table 4⇓, Fig 5⇓).
The P5 FF value (calculated at the induction site), on the other hand, was not modified in the self-terminating episodes (157±22 versus 157±21 ms), whereas it shortened significantly at minute 5 in the long-lasting episodes (132±17 versus 148±18 ms, P<.05; Fig 6⇓). Before the spontaneous arrhythmia termination, there was a trend for the fibrillatory wave to regularize compared with the arrhythmia onset as indicated by a decrease of the SD of the FF intervals (Table 4⇑). This behavior was not observed in the long-lasting episodes (Table 4⇑).
Blood Pressure Variations
Systolic pressure slightly decreased at AF onset (117±11 versus 123±13 mm Hg; range, 102 to 140 mm Hg at onset and 100 to 140 mm Hg before AF; P<.01), and diastolic pressure increased (77±5 versus 73±6 mm Hg; range, 60 to 88 mm Hg at onset and 60 to 84 mm Hg before AF; P<.01) at AF onset compared with baseline, whereas no change was observed for mean blood pressure (89±7 versus 89±8 mm Hg; range, 78 to 101 mm Hg at onset and 76 to 101 mm Hg before AF; P=NS) in self-terminating episodes.
The same pattern was observed in the long-lasting episodes: 114±9 versus 120±12 mm Hg (range, 100 to 130 mm Hg at onset and 104 to 138 mm Hg before AF; P<.005) for systolic pressure, 77±5 versus 74±5 mm Hg (range, 70 to 84 mm Hg at onset and 64 to 82 mm Hg before AF; P=.05) for diastolic pressure, and 88±6 versus 89±6 mm Hg (range, 81 to 98 mm Hg at onset and 79 to 100 mm Hg before AF; P=NS) for mean blood pressure.
Effect of Overdrive Pacing on FF Intervals During AF
In 10 patients, high-rate atrial pacing was performed twice with the above-mentioned protocol after 6 minutes of persistent AF; in 7 patients, it was performed in HRA; it was performed in CS in 3 patients. MFF was calculated at 5 seconds immediately before and after burst pacing in each atrial recording site. Atrial activation was significantly modified in the stimulating site as shown by shortening of MFF (158±8 ms after burst versus 173±10 ms before burst, P<.05), whereas no change occurred in the other distant recording sites (Table 5⇓, Fig 7⇓).
Localized Flutter Activity in AF
In only 1 case did the analysis of FF intervals allow us to demonstrate the presence of rapid atrial flutter in the left atrium, while the right atrium was still fibrillating (Fig 8⇓) and the ECG recorded AF. As Fig 8⇓ shows, the electric activity pattern is different between the right and left atria. In the right atrium, an irregular FF interval with a mean value of 144 ms and an SD of 15.9 ms, characteristic of fibrillation, was measured; meanwhile, in the left atrium the MFF was 130 ms with an SD of 4.7 ms. Moreover, as previously demonstrated,21 22 the sequence of FF intervals in the left atrium displayed regular oscillations in the atrial cycle length correlated to the ventricular activity. The typical pattern of rapid atrial flutter,22 showing a decrease of the atrial interval after the ventricular beat and a increase in the subsequent atrial intervals, is clearly discernible in Fig 8⇓.
Our observations confirm that both shorter refractoriness and dispersion of intra-atrial conduction are significant characteristics of atria prone to fibrillation compared with controls.
Dispersion of Atrial Conduction and AF Inducibility
In our cases, as in the 12 patients in the study of Cosio et al,3 conduction of nonpremature beats did not differ in the patients with AF compared with control subjects. There was, however, a remarkable enhancement of atrial conduction delay by premature stimuli without an apparent close relation to cardiac cycle (Table 2⇑). In this setting, the way that a single extrastimulus conducts intra-atrially and interatrially may be an important index of atrial vulnerability.6 7 12 This observation is confirmed by the fact that a dispersion of intra-atrial conduction >25 ms among the three atrial sites was almost invariably present in the fibrillating atria and not in control subjects. This was evident already at the 600-ms cycle length when a significantly shorter atrial ERP compared with that of control subjects also resulted. This combination of short atrial ERP and dispersion of conduction even at a 600-ms cycle length could explain how the inducibility of AF was so easily obtained even with one extrastimulus at a relatively slow heart rate.
It is well known that the measurement of atrial wavelength provides a critical parameter for the assessment of atrial vulnerability. For this purpose, the measure of conduction velocity is a critical issue in clinical electrophysiology because of the well-known limitations of atrial mapping. Therefore, we used the same parameter adopted by Attuel and coworkers13 as an index of atrial vulnerability.
Our finding of a shorter left atrial vulnerability index at all cycle lengths in AF compared with control patients allows us to overcome limitations in calculating conduction velocity and makes our observations of a very high vulnerability to fibrillation of this patient population meaningful.
It has to be noticed that our data are quite similar to those reported in the literature13 (2.27±0.6 cm for AF patients and 4.5±1.7 cm for control subjects), thus providing further validation of the left atrial vulnerability index as a vulnerable parameter in AF patients. In particular, the difference of left atrial vulnerability index values observed in our AF (1.72-2.6 cm) and control (3.1-6.57 cm) subjects sets left atrial vulnerability index as a “vulnerable parameter” that is very useful in characterizing the arrhythmogenic substrate in the clinical setting.
Our observations on AF induction are in general agreement with previous reports1 2 3 4 5 6 7 8 9 10 11 12 13 14 and establish the background for the high degree of vulnerability to AF in our selected study population.
Meaning of FF Intervals
LPAF is a condition in humans that offers the possibility of observing the spontaneous behavior of fibrillatory waves during the entire arrhythmic episodes, often from onset until its termination.
Previous studies9 23 sought a relation between average FF interval and atrial refractoriness. One study dealt with patients undergoing heart surgery9 ; in this study, the AF interval was defined as the mean of the histogram of local activation at the recording site. It was assumed a realistic representation of atrial refractoriness during fibrillation on the basis of previous investigation in the experimental setting15 in which the average interval between local activations during AF was used as an index for local refractoriness. It was hypothesized that the average or shortest FF interval reflects local refractoriness because, during fibrillation, cells can be reexcited as soon as their refractory period ends.9 15
The results of our study underline the existence of a significant and close relation between functional refractoriness and the mean of 100 consecutive FF intervals evaluated at AF onset and between the P5 FF interval and effective atrial refractoriness. This result is of practical importance because it allowed us to follow the electrophysiological behavior of the atrium during fibrillation, with possible meaningful implications in the pharmacological management of arrhythmia.
In the study of Asano et al,23 the MFF interval, defined as the average value of 10 FF intervals sampled from the HRA electrogram, was considered related to atrial refractoriness, although no data were provided to substantiate that claim. In our study, we demonstrated that even the average of 20 consecutive intervals is poorly related to atrial refractoriness. On the contrary, 100 consecutive FF intervals showed a good correlation. Because we are dealing with an arrhythmia characterized by a very irregular atrial activation pattern, it is probably mandatory to consider a certain number of FF intervals to obtain a representative picture of atrial activation behavior. Therefore, we think that the results of studies dealing with the calculation of few FF intervals can be misleading. Modulation of atrial electrophysiological properties may occur as part of reflex adjustments at AF onset, and this could explain the low degree of correlation with the first 20 FF intervals. These adjustments, however, are thought to be minor owing to the supine position of patients during electrophysiological study as demonstrated by the relatively stable mean blood pressures recorded in our patients.
In their study, Asano et al23 demonstrated that MFF was significantly prolonged in HRA before termination in 20 patients with self-terminating episodes, and this trend was present after intravenous administration of disopyramide in 4 of 10 patients with long-lasting episodes. Unfortunately, they did not evaluate the behavior of the MFF before drug administration, which might have been useful in understanding the mechanisms of AF persistence. Furthermore, in our study, the spontaneous lengthening of the MFF in self-terminating episodes and its shortening in the long-lasting episodes occurred in all three atrial sites, thus providing evidence for a peculiar pattern of arrhythmia behavior.
Electrophysiological Mechanism Basis of AF Self-Termination
It is of interest to note the different behaviors of atrial activation between sustained and self-terminating AF episodes.
The mean of 100 FF intervals (related to atrial FRP) was prolonged before self-termination but was shortened in the long-lasting AF episodes. At the same time, the P5 FF interval (related to atrial ERP) was significantly shortened in the long-lasting episodes but was unchanged in the self-terminating cases. Thus, self-termination of AF is probably related more to prolongation of atrial conduction (atrial FRP) than atrial effective refractoriness, which probably mirrors a decrease of the number of atrial wavelets below a critical value.18 Whenever effective refractoriness is progressively shortened during AF (long-lasting episodes), the number of AF wavelets constantly stays above the critical value, and the arrhythmia persists.
In the same way, the results of Boahene et al24 demonstrated that a lengthened FF interval is the hallmark of successful termination of AF by procainamide or propafenone in Wolff-Parkinson-White patients. In the study by Asano et al,23 the spontaneous trend toward an increase in FF interval before self-termination was mimicked by disopyramide, although to a greater extent.
Sih et al25 recently reported on the mechanisms of AF termination. By observing AF spontaneous termination in 7 patients, they did not find an increase in atrial cycle in 3 patients, whereas all 8 procainamide-induced AF terminations showed a significant increase in atrial cycle. This latter finding is in agreement with previously published observations24 25 of the effect of class 1A and 1C drugs on the atrial cycle during fibrillation. The heterogeneous behavior of atrial cycle in the 7 spontaneous terminations is not clear; as the authors suggested, critical changes in circulating wavelets within the atria, although undetected, may have occurred. Given that 12 of 14 patients underwent electrophysiological study for reasons other than AF, the possibility of observing electrophysiological patterns different from patients with clinical AF episodes must also be considered.
The progressive shortening of the atrial ERP found in our study in the long-lasting episodes seems to be the hallmark of a more extensive electrophysiological atrial derangement that makes the possibility of spontaneous reversion to sinus rhythm more unlikely and is, in itself, promoting AF perpetuation.
Previous observations emphasized the importance of dispersion of atrial refractoriness as the basis of AF induction7 12 13 ; however, these studies had considerable limitations in technical aspects such as obtaining stable and reproducible contact of the catheter tips in different recording sites with consequent fluctuations in stimulation threshold values and hence in refractory period measurement.1 2 Nonetheless, in our study, the MFF (closely related to the atrial FRP) did not show any difference during simultaneous recording at the three different atrial sites (several centimeters apart from each other).
Effect of Overdrive Pacing on Human AF
In an experimental setting, Allessie and colleagues16 and Kirchhof and colleagues17 recently demonstrated regional entrainment of AF in a dog model. Kirchhof et al17 observed that pacing during fibrillation at a cycle slightly shorter than the median fibrillation interval led to penetration of the paced wave fronts into an excitable gap between the wandering fibrillation waves. Prolongation of the pacing cycle allowed the fibrillation waves to approach the pacing site and resulted in loss of capture. In contrast, shortening of the pacing cycle led to acceleration of AF, which in turn led to loss of entrainment. Kirchhof and colleagues17 observed that this pacing-induced acceleration of AF is due to the induction of small leading circle reentrant circuits near the site of pacing. These small reentrant circuits changed in both size and position (wandering leading circle), and they always terminated spontaneously, which invariably led to the resumption of the original fibrillation rate. Thus, AF acceleration by overdrive pacing seems to be dependent of the induction of temporary leading circle reentry; rate-dependent shortening of the refractory period, caused by rapid pacing, was also considered a facilitating mechanism.
Because of the different scenario of clinical electrophysiology, our study is far from such an elegant demonstration of local entrainment of AF; nonetheless, the observations by Kirchhof et al17 clearly show that our data are consistent with the existence of an excitable gap during AF in humans.
In fact, atrial overdrive by burst pacing resulted in MFF acceleration at the stimulation site, which was short-lived and followed by the resumption of the original AF rate. There was no chance to influence atrial activation at distant sites. Our data are consistent with the statistical possibility of influencing local atrial activation owing to the very high rate of delivered impulses, which might fall at different degrees of atrial cell recovery.
The correlation observed between MFF and functional refractoriness in our patients suggests that local entrainment may depend not only on prolongation of refractoriness, as suggested by the Sicilian Gambit document,26 but also on the slowing of conduction, which can extend an already existent, small excitable gap.
In a minority of cases, such as that reported in Fig 8⇑, it is possible to observe how an atrial flutter at a particularly high atrial rate could be a possible reentrant circuit basis of AF, and the remnant part of the atria could be unable to follow because its own functional refractoriness was longer than the reentry cycle.
Criticism of the Study
One could argue that some or many of our AF episodes resemble atrial flutter or that they may start as AF and then change to atrial flutter.
An analysis of atrial flutter was performed by Wells et al27 and more recently by Lammers et al21 and Ravelli et al.22 These authors outlined how the SD of the FF intervals in atrial flutter is very small, on the order of 4 to 5 ms. Moreover, in atrial flutter, a typical pattern of phasic FF interval variation correlated to ventricular contraction21 22 was shown, such as in the case described in Fig 8⇑. These features were not observed in our cases in which the SD of FF intervals was well above 10 ms even during the most regular phases that occurred before self-termination (Table 2⇑).
Only in a one of our cases was a pattern of rapid atrial flutter observed in the left atrium, thus indicating the presence of a regular reentrant circuit in one part of the atria as a possible cause of an irregular AF in the remnant part. By then, the peripheral ECG was recording a typical pattern of AF. This diagnosis was confirmed by the presence of regular oscillations of the FF, re-corded from the left atrium and correlated to the ventricular activity21 22 (Fig 8⇑).
Furthermore, our observations are not applicable to AF in general but can be accepted only for a specific AF pattern represented by LPAF patients with relatively frequent and self-terminating AF episodes. This is, however, a unique setting for the investigation of purely electrophysiological mechanisms of AF.
Although an exhaustive understanding of the electrophysiological mechanisms during AF requires evaluation of the wavelength, the difficulty of obtaining a reliable estimate of conduction velocity during fibrillation is well recognized. Nonetheless, the correlation between MFF and atrial refractoriness offers a valuable clinical index closely linked to atrial wavelength.
Given the peculiar setting of electrophysiology, heart rate changes cannot be sufficient to investigate the possible autonomic reactions implicated in arrhythmia onset; for this reason, blood pressure was also evaluated. Some limitations, however, must be taken into account by this approach because gross blood pressure fluctuations may not be observed in the supine position (as required by electrophysiological study) in patients without organic heart disease, such as study population.
Slight and not clinically significant changes were observed for systolic and diastolic pressures at AF induction; in fact, mean blood pressure did not change at all in both groups of AF episodes. The absence of major hemodynamic fluctuations can be considered a marker of substantially well-preserved autonomic balance and rules out significant autonomic impairment at AF onset. These minor adaptive changes are consistent with the expected good preservation of hemodynamics in our population.
No significant changes were detected during AF in both self-terminating and long-lasting episodes for either blood pressure or RR interval. Although blood pressure and heart rate measurements may not completely depict both autonomic nervous system limb activity, these observations make significant autonomic changes during AF unlikely.
Analysis of the FF intervals demonstrates a close correlation with atrial refractoriness; thus, recording the pattern of atrial activation gives us a chance to better understand the electrophysiological mechanisms of AF maintenance or self-termination.
In humans and in the experimental setting, a local gap of excitability seems to be present during AF and probably represents a useful parameter for pharmacological purposes. In rare cases, a localized high-rate flutter may be detected while AF is recorded at the surface ECG.
We are grateful to Prof Michiel J. Janse, MD, FACC, FESC, University of Amsterdam, for his critical review of the manuscript.
- Received January 13, 1995.
- Revision received March 6, 1995.
- Accepted March 10, 1995.
- Copyright © 1995 by American Heart Association
Simpson RJ. Electrophysiology: normal and diseased atria. In: Falk RH, Podrid PJ, eds. Atrial Fibrillation: Mechanisms and Management. New York, NY: Raven Press Publishers; 1992:321-332.
Janse MJ, Allessie MA. Experimental observations in atrial fibrillation. In: Falk RH, Podrid PJ, eds. Atrial Fibrillation: Mechanisms and Management. New York, NY: Raven Press Publishers; 1992:41-57.
Ohe T, Matsuhisa M, Kamakura S, Yamada I, Sato I, Nakajima K, Shimomura K. Relation between the widening of the fragmented atrial activity zone and atrial fibrillation. Am J Cardiol. 1983;53:1219-1222.
Michelucci A, Lagi A, Caneschi A, Giovannini T, Mezzani A, Salvi S, Padeletti L. Importance of left atrial variables in defining atrial arrhythmogenesis. New Trends Arrhythmias. 1990;6:795-800.
Attuel P, Pellerin D, Gaston J, Seing S, Quatre JM, Mugica J, Coumel P. Latent atrial vulnerability: new means of electrophysiologic investigations in paroxysmal atrial fibrillation. In: Attuel P, Coumel P, Janse MJ, eds. The Atrium in Health and Disease. Mount Kisko, NY: Futura Publishing Co; 1989:159-200.
Lammers WJEP, Allessie MA, Rensma PL, Schalij MJ. The use of fibrillation cycle length to determine spatial dispersion in electrophysiological properties and to characterize the underlying mechanism of fibrillation. New Trends Arrhythmias. 1986;2:109-112.
Allessie M, Kirchhof C, Scheffer GJ, Chorro F, Brugada J. Regional control of atrial fibrillation by rapid pacing in conscious dogs. Circulation. 1991;84:1689-1697.
Kirchhof C, Chorro F, Scheffer GJ, Brugada J, Konings K, Zetelaki Z, Allessie M. Regional entrainment of atrial fibrillation studied by high-resolution mapping in open-chest dogs. Circulation. 1993;88:736-749.
Allessie MA, Lammers WJEP, Bonke FIM, Hollen J. Experimental evaluation of Moe’s multiple wavelet hypothesis of atrial fibrillation. In: Zipes DP, Jalife J, eds. Cardiac Electrophysiology and Arrhythmias. San Diego, Calif: Grune & Stratton; 1985:265-275.
Nollo G, Del Greco M, Ravelli F, Speranza G, Antolini R, Disertori M. ECG-oesophageal recording for beat-to-beat analysis of A-V conduction time variability. In: Computers in Cardiology. Los Alamitos, Calif: IEEE-Computer Society Press; 1992:433-436.
Lammers WJEP, Ravelli F, Disertori M, Antolini R, Furlanello F, Allessie MA. Variations in human flutter cycle length induced by ventricular beats: evidence of a reentrant circuit with a partially excitable gap. J Cardiovasc Electrophysiol. 1991;2:375-387.
Ravelli F, Disertori M, Cozzi F, Antolini R, Allessie MA. Ventricular beats induce variations in cycle length of rapid (type II) atrial flutter in man: evidence of leading circle reentry. Circulation. 1994;89:2107-2116.
Task Force of the Working Group on Arrhythmias of the European Society of Cardiology. The Sicilian Gambit: a new approach to the classification of the antiarrhythmic drugs based on their actions on antiarrhythmogenic mechanisms. Circulation. 1991;84:1831-1851.
Wells JL, MacLean WAH, James TN, Waldo AL. Characterization of atrial flutter: studies in man after open surgery using fixed atrial electrodes. Circulation. 1979;60:665-673.