Ataxia of the Pulse, Delirium Cordis, and Pulsus Irregularis

In 1628, Harvey provided logically compelling arguments to show that the heartbeat caused the circulation of the blood rather than the converse, and he particularly noted that contraction of the left ventricle was the sole cause of arterial pulsation. Laennec [5], after his discovery of mediate auscultation in 1918, continued to deny Harvey's views because of discordance between the pulsation of the heart and pulsation of the arteries in some cases of "intermission of the pulsation of the heart" (interrupted pulsation or extra beat) [5]. Laennec's experience with irregular heartbeat led him to disparage reliance on palpation of the pulse, considering it an unreliable guide to the state of the circulation. In 1835, Bouilland [6] distinguished between cases of pulsus intermittens and "ataxia of the pulse," in which the heartbeats follow each other at unequal intervals but with equal or unequal force. He likened this state of anarchy of the heart to a delirium of the brain. The invention of the kymograph by Ludwig in 1846 and the provision of a graphical method of investigating the pulse by Vierordt in 1855 enabled Nothnagel to publish three arterial pulse curves typical of the irregular heartbeat [7]. Nothnagel picked up on this complete irregularity of the pulse as a defining feature of the arrhythmia that he called delirium cordis: "In this form of arrhythmia the heartbeats follow each other in complete irregularity. At the same time, the height and tension of the individual pulse waves are continuously changing." Another essential feature of delirium cordis is its persistence, noted by Hering, who believed it at the time to be permanent and named it "pulsus irregularis perpetuus" [8].

Atrial Contraction Causes Venous Pulsation

After making many direct observations in animal hearts, Harvey became convinced that systole, rather than diastole, was the active part of the cardiac cycle that begins first in the atria [9]. This valuable insight lay fallow for nearly 200 years [5], until John Hunter determined that the veins in the neck of the dog distend and pulsate because the blood in them stagnates while the contraction of the heart prevents entry [10]. Clinicians had written about venous pulsations in the neck as early as 1704, particularly in association with tricuspid regurgitation, but there was great dispute about whether the impulse was generated in the vein, in the adjacent artery, or in the heart [11]. Clarification began in 1863, when Bamberger and Marey [12, 13] provided the first graphical records of the venous pulse, although these lacked timing and were of poor technical quality. In 1882, Riegel [14] published arterial and venous pulse curves that had been produced using the new technique and suggested that the initial wave resulted from contraction of the right atrium. MacKenzie, who published the first of his polygraphic studies in 1894, realized the value of study of the venous pulse in understanding events on the right side of the heart [11]. He provided tracings and detailed arguments to show that the initial wave results from auricular contraction and that the second results from ventricular contraction, events that precede the carotid arterial impulse. In his monumental book, The Study of the Pulse, MacKenzie [15] provided the first evidence to show that, in cases of advanced mitral stenosis when the irregular pulse supervenes, no signs of atrial activity are visible in the venous recordings. In cases reported in 1904 and 1905, MacKenzie showed that the a waves disappeared at the onset of the continuously irregular pulse and returned when the pulse became regular again [16, 17]. Hering confirmed these findings [18], and thereafter it was generally accepted that the three essential features of the pulsus irregularis perpetuus were the absolute irregularity of the pulse, the persistence of the rhythm, and the absence of demonstrable activity of the atria manifested by the absence of venous a waves [19].

Direct and Experimental Observations on Fibrillation of the Heart

McMichael [5] has collected early instances of direct observations of fibrillating hearts. In 1628, Harvey observed an undulation of the blood of the right auricle in the dying animal heart long after all normal beating had ceased. In 1683, Pechlin observed the ventricles of a dying patient through an open chest wound and clearly described terminal fibrillation. When all regular muscular contraction had stopped, "their fibres are still agitated. They form oblique ripples and there is seen. an undulation of many varieties." In 1783, de Senac associated obstructive lesions of the mitral valve with dilated atria whose fibers became irritated and whose contractions became more lively. The resulting palpitations were recognized as something other than normal beating of the heart; this recognition presaged the association of the irregular pulse with mitral stenosis, which would become clear in later centuries. Erichsen [20], in 1842, briefly described terminal fibrillation of the animal ventricle when the coronary arteries were acutely ligated. But the clinical relevance of these early observations appears not to have been recognized.

Formal investigation of fibrillation of the heart is generally accepted as having begun with the work of Hoffa and Ludwig [21]. Interest in the effect of electricity (which had only become available in continuous form early in the 19th century) on animal and human tissue led to the discovery that a strong and continuous current (faradic current) applied directly to the ventricle of a dog heart induced fibrillation. While the auricles continued to beat normally, the ventricles began a rapid, uncoordinated twitching that led to the dilation of the ventricle, a decrease in blood pressure, and the death of the dog. The investigator's interpretation of this experiment was that "the individual anatomic elements detach from their mutual relationships and abandon the synchrony of their contractions. As a consequence, disarray is created with respect to rhythm and intensity." In 1874, Vulpian [22] reported the same phenomenon in the atrium; he called it "fremissement fibrillaire." Aubert and Dehn [23] showed that infusing potassium salts into the blood of a dog led to irregular movements and sudden death. This led the investigators to hypothesize that there was a coordinating center for the regular heartbeat that was altered by the potassium. Kronecker and Schmey [24] inserted a needle into the upper third of the ventricular septum, throwing the ventricles into fibrillation; they surmised that they had found and destroyed this center of coordination for the regular heartbeat. MacWilliam, a graduate of Aberdeen University who had trained in Ludwig's Leipzig laboratory in the early 1880s, did extensive experiments using induction shocks on the hearts of invertebrates and mammals [25, 26]. He showed that the "arhythmic (sic) fibrillar contraction" induced by faradization of the muscle was a phenomenon common to many animal hearts. He also showed that the occurrence of fibrillar contraction of the ventricles was independent of any nervous influence or coordinating center of the heart and that it occurred in muscle isolated by zigzag section or in an isolated apical section. In the auricles, he found a different response—a rapid coordinate fluttering that could be suspended or inhibited temporarily during vagus nerve stimulation.

Arrhythmia Perpetua: A Manifestation of Atrial Fibrillation

In his 1887 paper [26], MacWilliam refers to his heart experiments as if the fibrillar contraction was familiar to researchers at the time under such names as Herz-Delirium and delirium cordis. He failed to note that, clinically, there was a pulse of the same name. Cushny [27] clarified the terminology, saying that, in physiology, delirium cordis referred to fibrillary contractions that arrest the circulation and prove rapidly fatal, whereas clinically, the term referred to extreme irregularity of the heartbeat. Cushny was impressed by the work of Krehl and Romberg [28], which had led to the general recognition that the irregularity of the pulse was caused by disorders of the higher part of the heart, where the normal rhythm was known to originate. Interestingly, Cushny pointed to the exact resemblance between the pulse curves of the dog with induced auricular fibrillar contraction and those obtained from patients with delirium cordis, but he refrained from claiming that they were identical. Similarly, in his 1903 paper on pulsus irregularis perpetuus [8], Hering mentioned his hunch that the irregular pulse resulted from a process familiar to researchers that was known as atrial fibrillation. In 1907, Cushny and Edmunds [29], who were doing experiments on anesthetized dogs, had the opportunity to record a change from regular beating to auricular delirium and suggested formally that the appearance of its pulse curve was identical to that seen in a young woman with relapsing irregular pulse [29].

The first electrocardiographic recordings of atrial fibrillation came shortly after Einthoven invented the string galvanometer in 1901. In 1906 [30], he published a tracing from a case of pulsus irregularis that showed QRS complexes with normal appearance; these occurred irregularly but had too much background interference to permit the identification of atrial activity. Hering [31], reporting on the electrocardiograms of similar patients in 1908, stated that one could see no signs on the electrocardiogram of action of the auricles, although F waves are clearly evident in his beautiful recording (Figure 1). The presence of these fine waves between ventricular beats had drawn the attention of Wenckebach [32], who wondered, after the suggestion of Cushny and Edmunds, whether these might be the waves of atrial fibrillation and might be responsible for the irregular rhythms whose pulse curves had been published by MacKenzie. MacKenzie, too, had considered this interpretation but had dismissed the fine waves evident on his records as artifacts created by a compression thrill of the vein [33].

View larger version (69K): [in this window] [in a new window]   Figure 1. The first electrocardiographic recording of atrial fibrillation showing F waves. In the paper that accompanied this recording, Hering remarked that no atrial activity was evident, meaning that no P waves could be seen [31].

Definitive proof of the identity of atrial fibrillation and arrhythmia perpetua was provided by Rothberger and Winterberg [34] in June 1909. In comparing tracings from animals with experimentally induced atrial fibrillation with those from patients with arrhythmia perpetua, these investigators noted three points of resemblance: absolute ventricular arrhythmia, the absence of P waves, and the presence of irregular oscillations of the venous galvanometer string caused by the fibrillary waves themselves. Lewis confirmed these criteria, apparently independently, in November 1909, emphasizing that the irregular electrocardiographic waves seen in diastole occur only with fibrillation of the auricle [35].

Fibrillation of the Atria or of the Entire Heart?

In his 1874 paper describing experimentally induced atrial fibrillation, Vulpian [22] stated clearly that stimulated ventricular fibrillation did not extend to the auricles, which continue to contract rhythmically even after the ventricles lose all excitability. MacWilliam [26], summarizing his extensive studies, made the same point again in 1887. However, MacKenzie, who was concentrating on irregular pulses and did not yet know that they were the same as atrial fibrillation, took a different view. In 1902, he speculated that the auricles became paralyzed in cases of irregular pulse [15]. By 1904, he had expanded on this idea, suggesting that the rhythm of the heart in pulsus irregularis originated from the ventricles and not from the origin of the great veins in the auricles as it normally did [16]. He was even more specific in 1907, stating that the abnormal rhythm proceeded from the fibers joining auricle and ventricle [36]. Discovery of the atrioventricular node by Aschoff and Tawara in 1907 caused MacKenzie to specify this node as the origin of the abnormal rhythm [37]. He reasoned that the hypertrophy sometimes seen in the auricles at autopsy meant that the auricles must not, after all, be paralyzed in such instances. Therefore, from his viewpoint, the only other possible explanation was that the atria contracted simultaneously with the ventricles as the result of an impulse originating in the atrioventricular node; he thought that the atrial contraction was concealed by the ventricular contraction in the pulse tracings.

MacKenzie's influence over clinicians was pervasive, particularly in Europe, because the experimentalists had definitive proof by 1904 that atrial fibrillation was separate from ventricular fibrillation [38]. Fredericq [39] cut the bundle of His (repeating an earlier experiment that His himself had conducted on normal hearts, an experiment that had produced complete heart block) and found that the atria continued to fibrillate while the ventricles ceased to beat irregularly and resumed an independent, regular rhythm. Proof that this finding had clinical relevance had to wait until 1910, when Lewis noted from electrocardiographic studies that the R wave was relatively normal in cases of irregular pulse [40]. Lewis argued that ventricular contraction must therefore originate from its usual starting point. He took the fine oscillations between the R waves, already noted by Mackenzie and Wenckebach, to be evidence of atrial activity throughout the cardiac cycle. From detailed study of the chest leads, Lewis showed that these oscillations originated from the atria rather than from the atrioventricular node.

The Mechanism of Atrial Fibrillation

Initial Theories

With atrial fibrillation understood to be a uniquely atrial rhythm underlying arrhythmia perpetua, debate about and investigation into its nature quickened. The experimentalists had already initiated the debate after the experiments with electric current, infused potassium, and cutting of the heart into strips mentioned above. In 1895, Engelmann [41] proposed that each heart fiber becomes independently rhythmic and that each is a focus of its own impulse formation as a result of increased excitability. This theory, which became known as the multiple heterotopous centers theory, was taken up and developed by Winterberg in 1906 [42] and by Lewis [43], who reasoned that activity from one or more heterogeneous centers would account for single, premature beats; for regular tachycardias; and, ultimately, for the completely incoordinate activity seen in auricular fibrillation. Later, Rothberger and Winterberg [44] presented a contrasting theory when they proposed that extreme acceleration of the rate from a single focus, discharging at about 3000 per minute, accounted for atrial fibrillation and that discharges of about 500 per minute accounted for atrial flutter. Notably, they showed that such rapid contractions require a marked shortening of the refractory period; Rothberger and Winterberg [44] and Lewis and colleagues [45] showed in dog hearts that this was probably due to vagal influence.

However, these two theories were soon replaced by a new one called "circus movement," which came to prevail for the next 30 years. According to Lewis [46], this idea first came to formal light as a result of Mayer's observations of a ring of tissue excised from the sub-umbrella of the jellyfish and floated in seawater [47]. Chemical or mechanical stimulus of the quiescent ring resulted in a band of contraction that went around the circuit in one direction for hours to days until another stimulus stopped it. Mines [48], using rings of tortoise heart excised so that a strip of atrium formed a ring with a strip of ventricle, produced a similar circuit of contraction in conditions of slow conduction and a shortened refractory period. He suggested that sufficient conditions exist in the intact, fibrillating heart so that a closed circuit of muscle of considerably greater length than the wave of excitation could propagate the circuit round and round indefinitely. Garrey [49] carried the idea further. When he cut a fibrillating chamber into four pieces, each fragment continued to fibrillate; this could not have happened if the fragments were dependent on a single tachysystolic center. On the other hand, Garrey found that although portions of the isolated heart muscle were capable of fibrillating, a critical amount of muscle mass was necessary; this made the individual fiber or multiple heterotopous centers theory untenable. Finally, Garrey could get a single impulse to propagate across a narrow bridge of cardiac tissue, but he could not get fibrillation waves to do so. He resolved these facts with a hypothesis that concerned conduction block, and he reasoned that impulses can travel in any direction but that they become limited by localized blocks in the tissue mass. His experiments had shown that the blocks were transitory and shifting in nature but that they forced the contraction wave into other, more circuitous paths, making possible a series of ring-like circuits of shifting location and multiple complexity—the circus contractions essential to fibrillation [49].

Lewis took up this experimental work and resolved to show that it had clinical relevance. In 1920, with his colleagues [46], he published the results of studies in which direct leads on a dog auricle were used. He claimed that in atrial flutter, a constant, central wave passes through a natural ring or cylinder of muscle around the vena cava and through the taenia terminalis [46]. From calculations on the vector of the atrial flutter wave in a human patient, Lewis argued that the wave rotated through 360 degrees around an isoelectric center compatible with circus movement [50]. In two patients with chronic atrial fibrillation, he did detailed studies of the auricular axis and found that its plane of movement swung through 60 to 90 degrees, compatible with a single circulating wave that temporarily entered and traveled through a new channel for a few cycles, only to revert to its original channel [51]. Rothberger [52] was immediately critical of Lewis's first paper because it was based on one experiment with incomplete results. Furthermore, he argued that the F waves of atrial fibrillation were the result of the activity of the entire auricles and not of one fixed circuit [53]. He persisted in defending a tachysystolic origin of atrial fibrillation over the years (while still allowing for the possibility that small regional circuits might exist), pointing out that a distinct band of muscle around the orifices of the great veins could not be shown anatomically [54]. However, with the authority of Lewis (and his rigid commitment to his own opinion, a trait that led to a falling out with Mackenzie in 1921 [38]) behind the circus movement hypothesis, it prevailed as the dominant explanation of the underlying mechanism of atrial fibrillation, particularly in the English medical literature. The circus movement theory was developed further to explain the complexity of experimental findings; these developments included the concept of a single "mother" ring that propagated the arrhythmia and gave rise to "daughter" rings and, finally, to multiple independent rings. Even Garrey, one of the original proponents of this theory, was moved to distance himself from the overly rigid way in which this theory was being applied. He stated in a later review article [55] that the circuit, although present, could no longer be thought of as a simple circuit because of blocks to its conduction: "the impulse is diverted into different paths, weaving and inter-weaving through the tissue mass, crossing and recrossing old paths again to course over them or to stop short as it impinges on some barrier of refractory tissue ... ."

Modern Theories

Further experimental work on the mechanism of atrial fibrillation had to wait until an animal model was found that produced fibrillation lasting longer than the transient episodes resulting from electrical stimulation. Scherf [56] found such a model using aconitine, a crystalline alkaloid from the tuberous root of monkshood. When he injected a blister of aconitine into the epicardium at the sinoatrial node, he often produced auricular fibrillation. If he cooled the area, fibrillation stopped, only to resume promptly with rewarming. This resumption, Scherf argued, was not compatible with circus movement because the circuit would have been interrupted and stopped by the cooling. If he injected aconitine into the auricular appendices, the result was similar; this led him to re-postulate the theory of the single tachysystolic center [57]. He also recognized that stimulating the vagus nerve had the paradoxical effect of accelerating the fibrillation rate; this was only seen consistently when the injection was away from the sino-atrial node. Extending this finding to experiments with acetylcholine-stimulated fibrillation, he showed that cooling the injection site would not abolish the fibrillation unless the sinoatrial and atrioventricular nodes were also cooled. He postulated that other self-sustaining tachysystolic centers must be set up at these sites in atrial fibrillation and at innumerable centers in ventricular fibrillation, possibly in the Purkinje fibers. Prinzmetal and colleagues [58] used high-speed cinematography and multiple-channel electrocardiography to show that the aconitine-stimulated or electrically stimulated fibrillation wave proceeded out uniformly in all directions from the point of stimulation. No circus movement or daughter waves could be shown. Cutting or burning the route of the specific circus pathway proposed by Lewis had no effect on the propagation of the fibrillation wave [58]. The theory of circus movement as the mechanism of atrial fibrillation was apparently no longer tenable [59].

Central to our present understanding of the mechanism of atrial fibrillation is the block hypothesis, first proposed by Porter in 1894 [60]. Porter suggested the following: "Fibrillar contractions may be due to an alteration in the cement substance leading to an interruption of the contraction-wave. The contraction-wave would thus be prevented from running its usual course, and the normal coordinated action of the ventricular cells would give place to the confusion conspicuous in fibrillary contractions ... .A change sufficient to block the passage of the contraction-wave might not be recognizable with the present histological methods." The only experimental evidence for this hypothesis was MacWilliam's earlier demonstration that fibrillation is more apt to ensue in conditions that have slowed myocardial conduction [26]. Mayer's circuit in the jellyfish had required that the initial stimulus be blocked from conducting in one of the directions around the ring; this encouraged speculation about the role of block in abnormal cardiac rhythms. Closely associated with the idea of block was the idea that the area might be re-excited or reentered later, after excitability had recovered and before the impulse died out. The block had to be in place to avoid collision and extinction when two wave fronts met on the other side of the ring, but it had to be unidirectional to permit reentry of the impulse that arrived from around the ring. Mines [48, 61] formally stated the features necessary for establishing the presence of reentry: unidirectional block; re-circulation of the impulse to its point of origin; and elimination of the rhythm by cutting the pathway. In 1928, Schmitt and Erlanger [62] produced unidirectional block in a strip of heart muscle. They showed how Mines's criteria could be met by means of a classic diagram of a twig of the atrioventricular bundle dividing into terminal fibers, forming two sides of a triangle (Figure 2). The resulting circuit would allow reentry, causing a ventricular extrasystole and, thereby, tachycardia.

View larger version (15K): [in this window] [in a new window]   Figure 2. Schematic drawing of a way in which reentry could occur. A penultimate twig (D) of the atrioventricular bundle is shown; anastomoses with ventricular muscle are at B and C. Conduction at AB would be monodromic: impossible in the direction of the atrioventricular node but almost normal in the opposite direction. An impulse from D would be blocked at A but, by way of the other terminal branch, could travel in the BA direction, ultimately to reach and stimulate muscle at C [62].

For reentry to occur, the circuit must be long enough for excitability to be recovered. The dilatation of the atria seen with cases of irregular pulse complicating mitral stenosis, already noted by de Senac in 1783, by Bouilland in 1835, by Marey in 1863, and by Riegel in 1882, suggested that one way to accomplish the delay was to increase the physical length of the circuit [5, 38]. Factors that slow conduction had already been noted by MacWilliam and others to facilitate fibrillation. In 1915, Einthoven [63] showed that the duration of the electrical impulse in atrial fibrillation is not prolonged with slowing of the heart rate as it is in the normal heart. Rothberger and Lewis [44, 45], a few years later, showed that stimulation of the vagus nerve further shortened the refractory period in atrial fibrillation; this established the basis for a third means of early recovery of excitability [44, 45].

In 1959, Moe [64] developed a dog-heart model of atrial fibrillation that could be sustained indefinitely using a strong cholinergic stimulus. Using the idea of block, Moe proposed a new mechanism of atrial fibrillation as a self-sustaining arrhythmia independent of focal discharge, and he called it the multiple wavelet hypothesis. He thought that although the fibrillation may have originated from one focus, it did not depend on it to persist and propagate but rather depended on random fractionation of the wave front around islands of refractory tissue: "The grossly irregular wave front becomes fractionated as it divides about islets or strands of refractory tissue, and each of the daughter wavelets may now be considered an independent offspring ... .Fully developed fibrillation would then be a state in which many such randomly wandering wavelets coexist."

Modern electrophysiologic evidence for block with reentry began in 1977, when Allessie showed a reentrant circuit in an isolated rabbit atrium by using precisely timed premature impulses to produce a sustained tachycardia [65]. Of particular interest to an explanation of atrial fibrillation is the type called "leading circle," in which the initiation of reentry takes place because of nonuniform refractory periods in atrial fibers in close proximity to one another. The initiating impulse conducts in fibers with short refractory periods and is blocked in those with longer ones, allowing reentry to them before the impulse has died out. Impulses circulate around a central area that is kept refractory, and thereby blocked, by centripetal wavelets arriving from all sides. The size of the circle is determined by the recovery time of the tissue forming the circuit, because tissue on both sides is kept depolarized by the leading circle. It is said that what goes around, comes around; Lewis would have been charmed to hear about this last finding by Allessie! Later studies done on an isolated, perfused canine atrial preparation in which induced atrial fibrillation was sustained by acetylcholine and recorded using a multiplexing recording technique, produced findings consistent with multiple, random-type wavelets of the leading circle type as the basis for fibrillation in such an experimental model [66, 67]. A reentrant circuit mechanism has been shown in human atrial fibrillation by cardiac surface mapping at the time of open heart surgery [68, 69]. Subsequent work by Allessie and colleagues [70] on conscious dogs underscores the importance of slowed conduction and shortened refractoriness—a short wavelength, the distance traveled by the depolarization wave during the refractory period—in determining the onset of atrial fibrillation during rapid electrical stimulation. Allessie does not exclude the possibility that the independent mechanism is also sustained by one or more undetected sites of primary impulse generation. Capture of regional areas of the fibrillating atrium up to 3 cm in radius by rapid atrial pacing leaves this theoretical possibility open [71]. Evidence showing that the wavelets are not entirely random, but rather that the earlier ones influence the course of subsequent ones, has recently been provided in humans having cardiac catheterization [72].

Whether the random wavelet mechanism accounts for clinical atrial fibrillation in all its forms or whether some is generated and sustained from a unique site, as shown by Scherf, remains open for further investigation [73]. What is well accepted is that sustained atrial fibrillation requires that the depolarizing wave fronts must continuously encounter excitable tissue, a circumstance favored by the shortening of atrial refractoriness, the slowing of conduction, and increased atrial circumference [74]. In the historical development of the disease concept, an animal model of fibrillation was studied for 50 years before atrial fibrillation was associated with its clinical counterpart, arrhythmia perpetua, at the beginning of this century. Atrial fibrillation was the only clinically recognized form until electrocardiography enabled ventricular fibrillation to be distinguished from it. Attempts to understand the basis of atrial fibrillation were polarized between the idea of an abnormal focus acting independently and that of a self-sustaining circus mechanism. Modern ideas about the origin and nature of atrial fibrillation are rooted in historical developments of the concepts and technology of cardiology. Inquiry into the nature of atrial fibrillation continues because of the tantalizing possibility that the disease, electrical in nature, might have an electrical cure.