|
Attuel and co-workers were the first to observe an intriguing
abnormality in patients with paroxysmal atrial fibrillation1. In these patients the atrial refractory period was
much shorter than in patients without atrial fibrillation, and did not prolong upon slowing of the heart rate.
Le Heuzey and colleagues2 found that in isolated atrial preparations from patients in sinus rhythm, the majority
of transmembrane action potentials had a marked plateau, whereas in preparations from patients with atrial
fibrillation, 97% of the action potentials had a triangular shape and a short duration. Also in this study, failure
of rate adaptation of action potential duration was found in specimens from patients with atrial fibrillation.
In addition to short refractory periods, and failure of rate adaptation, increased dispersion in refractoriness
was also demonstrated. Our group recorded simultaneous atrial electrograms from up to 40 atrial sites during
brief periods of atrial fibrillation induced by premature stimulation in patients undergoing open heart surgery3.
Reasons for operation were the “corridor” operation in patients with drug refractory atrial fibrillation and surgical
ablation of an accessory pathway in patients with atrial fibrillation, and surgical treatment of post-myocardial
infarction ventricular tachycardia (the control patients). The average interval between local activations, the
so-called atrial fibrillation interval, was used as an index of local refractoriness3. The advantage of using this
index is that simultaneous recordings at 40 sites could be made in 4 seconds, whereas classical determination
of the refractory period by the extrastimulus method at 40 sites would take an inordinate amount of time, and
would be impossible during an operation. The average atrial fibrillation interval in patients with atrial fibrillation
was shorter compared to control patients (152±3 msec at a total of 247 sites versus 176±8.1 msec at a total
of 118 sites; p<0.05). Dispersion in atrial fibrillation intervals was three times larger in the atrial fibrillation group than in the control group (Fig. 1).
For some time, it was thought that these changes were the cause of atrial fibrillation, until recent studies
provided evidence that they might be the result of atrial fibrillation.
In chronically instrumented conscious goats, repetitive induction of atrial fibrillation by a fibrillation pacemaker
led to a slowly progressing shortening of the atrial refractory period during the first 48 hours of atrial fibrillation,
from an average of 146 msec to 81 msec. At the same time, atrial fibrillation became more stable and the
duration of paroxysms of atrial fibrillation increased progressively4. Similar findings were reported for dogs and
humans5,6. This led to the slogan: atrial fibrillation begets atrial fibrillation, and the process whereby prolonged
periods of rapid atrial activity lead to long lasting changes in electrophysiological properties became known as
“electrical remodeling”.
Importantly, the long term shortening of the atrial refractory period persists for a long time after restoration of
sinus rhythm, and this means that after succesful cardioversion of atrial fibrillation, conditions remain favorable
for recurrence of atrial fibrillation because the refractroy period, and therefore the wavelength, remain short.
This very probably accounts for the frequent early reccurrence of atrial fibrillation following cardioversion7.
Several studies were undertaken to unravel the underlying ionic mechanisms of electrical remodeling. Action
potential shortening can in principle be caused by an upregulation of outward currents carried by potassium
currents, or by downregulation of inward currents carried by sodium or calcium currents. Surprisingly, the
transient outward current Ito, and the sustained potassium current Iksus were actually reduced in in patients
with atrial fibrillation8, and in a dog model of rapid pacing, voltage clamp studies found no change in a number
of outward currents (Ik1, Ikr, Iks, Ikur)9. In the latter study, the transient outward current Ito was reduced,
but this was considered to have no effect on action potential duration. The reason for the action potential
shortening in the dog model was the marked reduction in the density of the L-type calcium channel9. The
action potential of atrial myocytes isolated after 42 days of rapid pacing had a triangular shape, and their
duration had shortened from values around 160 msec to 85 msec. Identical action potentials could be obtained
in normal myocytes subjected to the calcium antagonist nifedipine, while in remodeled myocytes the calcium
agonist BAY K 8644 largely restored the action potential plateau9. In a more recent study on atrial myocytes
isolated from fibrillating human atria, the L-type calcium current and the transient outward current Ito were
both reduced by to about 70%, and in addition, the inward rectifier Ik1 and the acetylcholine-activated
potassium current IkAch were increased10.
Whether or not electrical remodeling also involves changes in conduction velocity is less clear. In the chronic
goat model, conduction velocity does not change4, but in dogs subjected to rapid atrial pacing, density of
the sodium current decreases parallel to a decrease in conduction velocity11. Contradictory data are available
about changes in gap junction dictribution, which also could influence conduction: in dogs with atrial fibrillation
an increase in connexin 43 has been found12, but in goats with atrial fibrillation no change in connexin 43
was detected whereas there was a change in the distribution of connexin 4013.
The reasons why atrial myocytes shorten their action potential, primarily by a downregulation of the expression
of the L-type Ca++ current, are not clear, but it may be an attempt to prevent cellular calcium overload14.
In the study of Wijffels et al4, the refractory period reached a steady state within a few days following repetitive
induction of atrial fibrillation, whereas it took a few additional weeks for atrial fibrillation to become chronic.
Therefore, other factors besides electrical remodeling must play a role in the development of chronic atrial
fibrillation.
Already 30 years ago, Bailey and colleagues observed that long standing atrial fibrillation tended to perpetuate
itselc:\www “atrial fibrillation leads to a diffuse atrophy of muscle, and atrial fibrillation becomes irreversible”15.
Recently, it was shown that sustained atrial fibrillation in goats leads to structural changes in the atrium that
resemble those in ventricular myocytes during chronic hibernation: loss of myofibrils, accumulation of glycogen,
fragmentation of the sarcoplasmatic reticulum16. The electrophysiological consequences are not clear, but the
loss of atrial contractility is likely to contribute to atrial dilatation.
In 1961, Van Dam and Durrer showed that atrial electrograms from left atrial appendages from patients with
mitral stenosis and atrial fibrillation were fractionated, and they attributed this “functional dissociation between
various elements in these muscle strips” to “the development of connective tissue between the muscle fibers”,
and speculated that “such an increased functional dissociation may give rise to a reentry mechanism”17. Boyden
and co-workers18 found in dogs with “spontaneous” mitral valve fibrosis and atrial arrhythmias, including atrial
fibrillation, massive interstitial fibrosis, cellular hypertrophy and loss of myocytes, and surprisingly little changes
in the characteristics of atrial transmembrane potentials. They concluded that “dramatic abnormalities of cell
electrophysiology may not be involved in the genesis of arrhythmias in the enlarged canine atrium, and the
altered morphological features of the atrium in these dogs may be important in the genesis of persistent atrial
arrhythmias”18. They reasoned that the increased atrial size would permit the co-existence of many reentrant circuits, and that
the increase in connective tissue would alter anisotropic properties and could lead to slow, inhomogeneous
conduction, unidirectional block and reentry. Spach and colleagues19 demonstrated how in an atrial preparation
from a 62-year old patient with an enlarged and hypertrophied atrium, where virtually all atrial fibers were
surrounded by collagenous septae, atrial electrograms were fragmented, and microrentry occurred. This was
due to electrical uncoupling of side-to-side connections of parallel-oriented fibers, leading to “zig-zag”
conduction in the transverse direction of atrial fibers, and to fractionated electrograms20.
Fibrosis may result not only in slow, inhomogeneous conduction, but also in an increase in dispersion of
refractoriness3, and both factors increase the propensity for reentrant excitation. In well coupled cells,
current flow between cells during repolarization will tend to decrease dispersion in repolarization by prolonging
action potentials with an intrinsic short duration and by shortening action potentials with an intrinsic long
duration. Cellular uncoupling will unmask intrinsic differences in action potential duration.
In summary, there are a number of processes involved in atrial remodeling. leading to persistent atrial
fibrillation. The altered gene expression of ionic channels resulting in shortening of the action potential
and loss of rate adaptation of the refractory period is a process of several days. The development of
hibernation, resulting in loss of contractility and atrial dilatation, occurs in a matter of weeks. Finally, fibrosis
completes the structural abnormalities, and this may take several months.
Fig. 1: Atrial fibrillation intervals (ordinate, in msec) recorded simultaneously at 32 atrial sites
(electrodes) during atrial fibrillation induced by premature stimulation during open heart surgery in a
control patient and in a patient with paroxysmal atrial fibrillation. Reproduced by permission from
Ramdat Misier et al3
|
