Experimenting on the Past: The Enigma of von Economo's Encephalitis Lethargica

Abstract

Encephalitis lethargica (EL) was a complex and mysterious disease that appeared around the same time as the great influenza pandemic of 1918. The contemporaneous relationship of the 2 diseases led to speculation that they were causally related. Contemporary and subsequent observers conjectured that the influenza virus, directly responsible for the deaths of more than 20 million people, might also have been the cause of EL. A review of the extensive literature by observers of the EL epidemic suggests that most contemporary clinicians, epidemiologists, and pathologists rejected the theory that the 1918 influenza virus was directly responsible for EL. Disappearance of the acute form of EL during the 1920s has precluded direct study of this entity. However, modern molecular biology techniques have made it possible to examine archival tissue samples from victims of the 1918 pandemic in order to detect and study the genetic structure of the killer virus. Similarly, tissue samples from EL victims can now be examined for evidence of infection by the 1918 influenza virus.

Introduction

In April 1917, a paper by neurologist Constantin von Economo appeared in the Wiener Klinische Wochenschrift, describing a series of unusual cases that had occurred during the winter of 1916–1917 (1). Admitted under a variety of diagnoses from meningitis to delirium, the patients exhibited a recognizable and consistent pattern of symptoms that led von Economo to propose the emergence of a novel disease entity that he named encephalitis lethargica (EL).

After the 1916–1917 outbreak described by von Economo, the disease disappeared in Vienna, but reports of similar cases surfaced in France and England during 1918 (2–5). By late 1918, the disease had been reported in the United States (6, 7) and became widespread in Europe and North America in the early 1920s. The disease peaked in 1923 when it caused nearly 2,000 deaths in the United States. Incidence then waned rapidly and few new acute cases were reported after 1925 (8). Although occasional reports of sporadic cases have subsequently been reported, the clusters of cases characterizing the years 1919–1923 have never recurred. A substantial literature concerning the disease accumulated during the 1920s, including many detailed and thoughtful descriptions of the clinical, epidemiological, and pathological characteristics of EL.

Clinical Characteristics

Encephalitis lethargica presented with so many different symptoms that had the cases occurred in isolation, the disease might never have been recognized. Von Economo's description of the 1916–1917 outbreak in Vienna was the first to delineate a constellation of signs and symptoms suggestive of a distinct disease entity. He identified 3 distinct clinical patterns of EL (8) while emphasizing that symptoms often overlapped and that even in the same patient, characteristics of each of the different patterns might prevail at different stages of the disease.

The somnolent-opthalmoplegic form of EL predominated during the 1916–1917 outbreak in Vienna. It was characterized by a brief, nonspecific prodrome consisting of malaise, headache, and mild fever. Von Economo emphasized that the sudden high fever, myalgias, and upper respiratory symptoms characteristic of influenza were absent (1). The patient became increasingly somnolent until he fell into a deep sleep from which he could easily be roused, but into which he immediately relapsed when stimulation ceased. Somnolence usually lasted 1 or 2 weeks, after which it either worsened, resulting in coma leading to death, or gradually diminished until full recovery. The appearance of palsies during the first few days of the illness, especially of the external and internal eye muscles with resultant ptosis and diplopia, was a consistent characteristic. The extremities often became profoundly weak or hypotonic, but occasionally muscle groups developed rigidity. An expressionless, mask-like facies was frequently described. The descriptive epithet “sleepy sickness” was often attributed to this form of the disease. The mortality associated with this type of EL was approximately 50%, but chronic debility and neurologic sequelae were uncommon in this form of EL.

The hyperkinetic form of EL was characterized by restlessness as the dominant symptom, which assumed the form of motor disturbances including twitching or jerking of muscle groups or an anxious, even frenzied mental state. Onset was more sudden than in the somnolent form and often included severe back and neck pain, prostration, and weakness, followed by increasing mental and motor unrest. Insomnia or inversion of sleep patterns were common symptoms. Although this form of EL bore little resemblance to the somnolent variety, both included a profound disturbance of sleep patterns, reflecting localized damage in related functional areas of the brain. While diplopia and ptosis were uncommon in the hyperkinetic form, disturbances of the intra-ocular muscles were almost always present. As the disease progressed, involuntary movement increasingly interfered with voluntary actions, making the disease difficult to distinguish from chorea. prognosis of the hyperkinetic form varied. Mortality in the initial stages was higher than in the somnolent-opthalmoplegic form, with sudden death occurring at any time. The hyperkinetic form was frequently succeeded by the somnolent form and occasionally by the amyostatic-akinetic form described below. Generally, the later stages of the hyperkinetic form were associated with a lower mortality than the other forms, leading to an average death rate of approximately 40%, although complete recovery was less common than in the somnolent form.

Although the amyostatic-akinetic form of EL was the least common manifestation of acute EL, it was associated with a high proportion of chronic sequelae. In its chronic form, von Economo designated the disease as parkinsonism. The weakness and rigidity occasionally seen in somnolent EL became prominent symptoms in amyostatic-akinetic EL. The condition bore a striking resemblance to the bradykinetic component of the disease described by parkinson in 1817 as paralysis agitans (11), although concomitant tremor was unusual. These patients also exhibited sleep inversion and ophthalmoplegic disturbances, including diplopia, ptosis, and general weakness of the eye-muscles. Mortality in the acute stage of this form of EL was somewhat less than in the other forms, although progression to a chronic form was much more likely. Evolution of this form of EL into chronic parkinsonism was a recognized clinical entity as early as 1921 (10).

Postencephalitic parkinsonism

As the acute epidemic waned in the mid-1920s, numerous cases of delayed parkinsonism began to accumulate. Over a few months, patients became physically and mentally exhausted, with generalized rigidity of the limbs, bent posture, and unsteady gait. This syndrome was called postencephalitic parkinsonism (PEP), as the clinical constellation was similar to that of classical or idiopathic parkinson disease (IP). Although differentiation between pEP and IP was sometimes difficult, the diseases were not clinically identical (10, 12, 13). The main difference was in age of onset, as IP characteristically occurred after the age of 50, while pEP began at any age, including childhood. There were other clinical differences, including the absence in pEP of the characteristic “pill-rolling” tremor of IP. pEP generally progressed more rapidly and in spurts, in contrast to the slow and steady progression of IP. Von Economo noted the similarity between the 2 conditions and argued that they represented 2 different entities “which, in spite of the likeness of their symptoms, appear only accidentally to affect similar nerve-centres” (8). This judgment is reinforced by the dissimilar histopathology of the 2 conditions, as discussed below.

Epidemiology

Von Economo's initial cases occurred during the winter of 1916–1917 (1). He later found evidence for the existence of similar cases a few years earlier in Bulgaria (14). He and several other observers also found descriptions of disease outbreaks apparently resembling EL in medical histories. In particular, there were many descriptions of nonspecific nervous complications, weakness, mental disorders and somnolence related to past influenza epidemics (2, 8). The reports lacked sufficient detail to identify these historical episodes as EL with certainty, but many observers, including von Economo, accepted some of these descriptions as evidence that EL had occurred in the past and was not a new disease. A European influenza epidemic in 1580 was referred to by some as Schlafkrankheit (sleeping sickness) (8). Sydenham, writing in 1726, described an epidemic in London of febris comatosa (sleeping fever) in 1673–1675. Several reviews mention another epidemic referred to as Schlafkrankheit in Tübingen in 1712. Reviewers differed as to which historical description best represented the probable predecessor of EL.

The extensive number of descriptions of historical outbreaks suggestive of EL led many to conclude that it must have occurred in the past. Of particular interest was a disease called nona that was said to have been prevalent in Italy in the early 1890s. Von Economo, while admitting that there were “no scientific records…. of this malady,” nevertheless concluded from the descriptions that it represented an antecedent epidemic of EL (8). A review by Longuet in 1892 (15) stated that while nona had been extensively reported in the press, subsequent investigation revealed that “la nona n'existait pas” (nona never existed).

A British physician, A. J. Hall, who evaluated many cases of EL himself and reviewed the course of the epidemic in Britain, reached the following conclusion:

“Thus, even if we include, rightly or wrongly, the Tübingen outbreak, together with the epidemics of electric chorea and nona, the smaller outbreaks of Comby and Gordon, and the isolated scattered cases outlined above, the fact is emphasized that nothing approaching the present epidemic, either in extent or in severity, has been known to us during the past century. When epidemic encephalitis appeared in 1916–17, it was literally new to us” (2).

Hall wrote these lines in 1924, prior to the emergence of thousands of cases of postencephalitic parkinsonism (PEP), which were later considered a unique consequence of EL. In all of the historical accounts of cases resembling acute EL, there is no mention of such chronic sequelae, reinforcing Hall's conclusion that EL probably had never before occurred on a large scale.

Extent of the Epidemic 1917–1924

Beginning in 1918, EL occurred in numbers sufficient to trace the course of the epidemic, although we are aware of official epidemiological statistics on the extent of the epidemic only for the United States and the United Kingdom.

A small number of EL cases appeared in England in March 1918 (2, 4) and the disease became reportable at the end of 1918. In his review of the epidemic, Hall estimates that there were about 250 cases in England prior to January 1919. The disease recurred annually, usually between January and March, reaching a peak of 1,470 cases in 1921. The number of cases dropped to 454 in 1922, before increasing to more than 1,000 in both 1923 and 1924. The disease then practically disappeared in its epidemic form, although cases of delayed-onset parkinsonism continued to emerge for several years thereafter. There were approximately 5,500 cases of EL reported in England from 1918 to 1924, with an estimated mortality rate of 40%, suggesting a total of 2,200 deaths from acute EL. Establishing the number of victims of pEP is difficult because statistics on this entity apparently were not collected in England and reports of acute EL deaths ceased in 1924.

The first cases of EL in the United States were reported in New York City in September and October 1918. By the early spring of 1919, the disease had spread as far as Illinois, Louisiana, and Texas. A 1921 study investigated 255 EL cases that had been reported between September 1918 and May 1919 (7). This study was unique among epidemiological studies of EL in that reported cases were individually reconfirmed. The difficulty of diagnosing EL accurately, especially early in an epidemic, is reflected by the fact that over 20% of the individually reviewed cases initially reported as EL were subsequently reclassified as other disease entities. From 1919 on, EL was included as a separate cause of death in US mortality statistics. As in England, the number of cases rose sharply from 1919 to 1921 before dropping in 1922 and peaking in 1923.

EL continued to be listed as a separate cause of death in the United States throughout the 1930s, accounting for 500 to 1,000 deaths per year (16). Since widespread epidemics ceased after 1924, it seems likely that the deaths attributed to EL after 1925 reflected pEP and not acute EL. From 1919 through 1925, there were 8,542 EL deaths. Assuming a mortality rate of 40% in acute EL cases (2, 7, 8), this figure suggests a total of 21,000 acute EL cases during the epidemic phase of the disease. Another 14,400 deaths were attributed to EL from 1925–1938 reflecting almost twice as many deaths from pEP as from acute EL.

Many patients who recovered from acute EL later developed pEP. The majority of pEP cases emerged during the 1920s, usually within 6 months to 1 year following an acute episode of EL. Several studies suggest that the number of patients with parkinsonian symptoms tripled during the 1920s. At least 50% of pEP patients at that time had clinical histories of acute EL. pEP cases continued to emerge in the 1930s, but the distinction between IP and pEP became blurred. pEP symptoms became even more similar to those of IP, and the patients, although older, were still younger than typical IP patients and less likely to report a preceding bout of acute EL. Collectively, pEP accounted for approximately half of the cases of parkinsonism in the 1930s (12).

Because there were so many pEP cases in the 1920s and 1930s, the idea that all cases of parkinsonism were ultimately caused by EL became widespread. It was even predicted that parkinsonism would disappear as the generation that had been alive during the acute EL epidemic died out (17). As the EL cohort reached the age at which IP normally developed, it is probable that many IP cases were diagnosed as pEP. In an article on the etiology of parkinson disease, Duvoisin wrote in 1982:

“It is clear from reviewing medical charts of the 1930s and 1940s that to many physicians the terms ‘chronic encephalitis’ and ‘parkinsonism’ became synonymous, and that the nosological distinctions between parkinson's disease and postencephalitic parkinsonism had become confused. … In practice, a mere history of the flu sufficed to gain otherwise typical cases of parkinson's disease, the diagnosis of postencephalitic parkinsonism” (18).

In total, US mortality statistics ascribed 22,942 deaths to EL (16), of which one third comprised acute EL deaths occurring between 1919 and 1925. The other two thirds probably reflected pEP deaths occurring between 1925 and 1938, although an uncertain number of these may have been deaths from IP misdiagnosed as EL. A reasonable estimate of total EL cases in the United States ranges from 25,000–40,000.

The extent of the epidemic outside the United States and United Kingdom is particularly difficult to assess due to lack of official statistics. Numerous references cite outbreaks in France, Italy, Germany, Switzerland and elsewhere (2, 8, 19, 20). Von Economo reported that after the initial 1916 outbreak in Vienna, EL did not return until early 1920 in the form of an extensive epidemic of the hyperkinetic type. There was another widespread outbreak in spring 1921, but only sporadic or very small clusters of cases subsequently. Von Economo reports that outbreaks occurred throughout Germany, France, and Italy at slightly differing times during the winter and spring of 1920 and 1921, but unfortunately there are no official statistics associated with any of these reports. A French physician, Netter, is quoted by many observers as estimating that there were 10,000 cases in France (2, 8), but due to the absence of official statistics, it is difficult to gauge the reliability of this figure. Anecdotal reports from Europe suggest that it may be justifiable to extrapolate the rate of mortality reported in England and the United States to the rest of Europe. Assuming a population of approximately 300,000,000 in Western Europe in 1920, a total of 80,000–120,000 cases of EL during the epidemic period is a reasonable estimate.

Assessment of the extent of the EL epidemic worldwide is again hampered by the lack of statistics. Mention is made in some reports of outbreaks outside Europe, but clinical descriptions are lacking. For example, an extensive encephalitis outbreak took place in Japan during the summer of 1924, but the symptoms and seasonality suggest that this was not EL but was caused by Japanese encephalitis virus (8, 19). Regional medical literature might hold evidence of more extensive worldwide spread, but reviews written at the time suggest that the disease was concentrated in North America and Western Europe.

Pathology

The pathological findings of acute EL were described in detail by several investigators during the outbreak utilizing autopsy material. Microscopic pathology was first reported in von Economo's 1917 paper in reference to the brains of 2 of the 6 patients who died. These results were subsequently confirmed and expanded (2, 8, 21).

All observers reported similar findings, but the changes were nonspecific and compatible with those seen in other acute encephalitides, such as Japanese encephalitis, as noted by von Economo (8). Grossly, the changes were usually slight, including congestion of superficial vessels with occasional meningeal hemorrhages. On cross section, the brain and spinal cord might show congestion and minute hemorrhages, with brainstem involvement being much more marked than in the cerebellum and spinal cord. The pronounced involvement of the brainstem, however, including the substantia nigra and locus ceruleus represents a more specific pattern of disease.

The most prominent microscopic feature was a lymphoplasmacytic infiltrate in the perivascular Virchow-Robin spaces with rare neutrophils. This inflammatory infiltrate was most pronounced around vessels in the basal ganglia, mid-brain and pons, other brainstem nuclei, the aqueduct of Sylvius, and the floor of the fourth ventricle. Hemorrhages, when present, were usually microscopic. Neuronal changes were less marked than in poliomyelitis and neuronophagia with glial nodules, when present, was not pronounced (22). The same alterations were sometimes found in the cerebral cortex, however the greater part of the cortex was free of severe inflammatory lesions, as were the cerebral white matter and cerebellum.

In his 1929 monograph, von Economo wrote: “Encephalitis lethargica is therefore, in a few words, a non-purulent and properly speaking, non-hemorrhagic acute inflammation of the gray matter with approximately negative macroscopic findings. …These findings are not pathognomonic, that is to say, they are characterized not by features peculiar to encephalitis lethargica to the exclusion of other nervous diseases, but by the great constancy of the anatomical picture” (8).

The pathologic distinctions between IP and pEP have been reviewed in detail (23, 24). In IP, unilateral or bilateral neuronal loss is observed in the substantia nigra, locus ceruleus, and in other locations. Neuronal, intracytoplasmic Lewy bodies are usually present in dopaminergic and other subcortical nuclei, spinal cord, sympathetic ganglia and, frequently, the cerebral cortex.

In contrast, pEP showed extensive and severe bilateral diffuse degeneration and gliosis of substantia nigra and locus ceruleus in the absence of Lewy bodies, with damage to other parts of the brainstem and widespread presence of globose neurofibrillary tangles. In 1935, Hallervorden described neurofibrillary tangles in the substantia nigra and other brainstem nuclei in pEP cases (25). The presence of neurofibrillary tangles in the absence of Lewy bodies in the appropriate clinical background is characteristic of pEP. Although neurofibrillary tangles might also be present in cortical and subcortical areas of pEP brains, unlike Alzheimer disease, senile plaques were few or absent.

Etiology

There appears to be general agreement that while EL had characteristics of an infectious process, it was not acutely contagious. Reviews of large numbers of cases documented little evidence of transmission from patient to patient (2, 8, 26). The symptoms and pathology of the disease resembled a number of conditions, including poliomyelitis, post-measles encephalitis, and encephalitides following other acute illnesses, botulism, lead poisoning, cerebrospinal meningitis, syphilis, myasthenia gravis, and IP. Thus, on the basis of symptoms, there was no reason to exclude bacteria, viruses, toxins, secondary effects of infectious agents, or other degenerative conditions as the etiologic factor in EL. However, the geographic, seasonal, and age distribution seem most consistent with the spread either of an infectious agent that caused no symptoms in the vast majority of carriers or that infected only a minority of the population.

Separate disease outbreaks were similar, characterized by a number of cases scattered across a geographic area over the course of a few months. Epidemics usually peaked in late winter, from February through March, whereas summer incidence was very low. All ages were affected, with the average age in most series of cases being between 20 and 40 years. In both symptoms and sporadic distribution, EL most closely resembled poliomyelitis, with which it was initially confused (2, 8, 26, 27). As the epidemic expanded, it became clear that the seasonal and age incidence differed from polio and that symptoms of EL, while similar to those of polio, were distinguishable. The clinical similarities, however, led many to postulate an infectious neurotropic agent with an affinity for specific parts of the brain as the causative agent. Contemporary efforts to isolate the causative agent from the brain were unsuccessful (8, 26, 28).

Direct damage by a neurotropic infectious agent remains a likely cause of EL. A toxic agent, either environmental in origin or, perhaps, a component of a contemporary medical treatment may account for the symptoms and the limited duration of the epidemic, although a candidate toxin or medication with the requisite geographic, seasonal, and chronological distribution has not been described. Another possibility is that a familiar organism, either by itself or in combination with other agents, developed unusual characteristics that account for the EL epidemic, only to have reverted or disappeared after a critical mass of the world's population became immune. A candidate organism and scenario for its modification have never been firmly established. The 1918 influenza pandemic virus, however, has remained a prime suspect for 80 years.

EL and the 1918 Flu

It is curious that, although contemporary observers of the EL epidemic concluded that EL was not directly connected to the 1918 influenza pandemic, the influenza virus subsequently became widely accepted as the likely cause of EL. A number of factors contribute to the continued popularity of this thesis, the most important of which is the relatively contemporaneous appearance and disappearance of the 2 illnesses. As the years went by and neither the deadly influenza nor EL recurred, the speculation that they were causally related became entrenched.

Further support for this line of reasoning was provided in the late 1930s by the development of an influenza strain capable of replicating outside the respiratory tract. After multiple passages in chicken eggs, ferrets, and mice, the first influenza virus isolated from humans, A/WS/33, mutated into a new strain that caused systemic disease in mice. Most significantly, the new strain, A/WSN/33 (29, 30), proved to be neurotropic. As A/WS/33 was thought to be a descendent of the 1918 pandemic strain, the neurotropism might reflect a reversion to a trait first exhibited by its ancestor (31).

Contemporary observers discounted the link between EL and influenza for several reasons. The earliest EL cases occurred during the winter of 1916–1917, more than a year before the spring wave of the 1918 pandemic and nearly 2 years before the lethal fall wave. Many EL cases were seen in France and England in early 1918, months prior to the arrival of pandemic influenza in the late spring and early summer. It is possible to argue that the influenza strain responsible for the pandemic began circulating at a low level before the spring of 1918. Analysis of the genetic structure of the 1918 pandemic virus in fact suggests that the virus may have been adapting in a mammalian host for a few years prior to emerging as a pandemic virus (32–34). However, it is difficult to understand how a novel influenza virus could have spread throughout Europe causing dozens to hundreds of cases of EL in the absence of any outbreak of the typical respiratory symptoms of influenza. Von Economo explicitly states that there was no prevalence of influenza-like illness during the initial EL outbreak in Vienna. The early outbreaks in England and France also preceded the arrival of the 1918 flu.

As A. J. Hall wrote in his monograph on EL:

“When the first series of cases came under our observation in the Sheffield district in March 1918, many possible explanations were considered and discussed; but amongst them influenza was never thought of or suggested by any who saw this series of extraordinary cases. If there is any connection between epidemic encephalitis and influenza, it is unusually well concealed from the clinician” (35).

Physicians noted that the symptoms of the 2 diseases overlapped only very slightly. Secondary encephalitic complications associated with influenza were known at the time, but clinicians found them clearly distinguishable from EL (8, 36, 37). Such postinfectious complications still occur (38, 39) and are now more properly considered as acute disseminated encephalopathies in which vascular congestion and edema of the brain are thought to result from the peri-infectious increase in inflammatory cytokines (38). In spite of the fact that nearly one third of the world's population suffered an acute influenza attack in the fall of 1918, there are very few case reports of EL emerging during or soon after a bout of acute influenza. H. F. Smith studied 255 cases of EL occurring in the United States in the spring of 1919 (37). These cases represented the first wave of EL cases in the United States and were the cases closest in time to the worst influenza epidemics of October and November of 1918. In 54% of these EL cases, no preceding attack of influenza was reported.

The 2 diseases also lacked a consistent epidemiological relationship. Although influenza and EL both occurred in epidemic form between 1918 and 1923, the timing and extent of the outbreaks were dissimilar. Influenza spread around the world in a matter of weeks and peaked everywhere between September and November 1918, whereas EL required months to spread through Europe and then several more months to cross the North American continent (19, 37). While influenza recurred in ever-smaller epidemics in the winters of 1919 through 1922, the numbers of EL cases rose and fell during these years, peaking in most places in 1923 (16).

The relationship between the 2 diseases also varied with geographic location. In Chicago the number of influenza cases ranged from 605 in 1924 to 30,480 in 1920, reflecting a 50-fold variation, while EL cases ranged from only 48–90 cases per year during the same period (19). In New York City, influenza and EL deaths peak together in February 1920, but in 1921 EL peaked 1 month before influenza deaths, and in 1922 peaked 2 months later. Weekly case numbers from 6 English cities in 1924 also showed no clear temporal relationship (19).

A more recent study by Ravenholt and Foege (40), which compared influenza deaths in Seattle with EL cases by month of onset, describes a pattern whereby the onset of EL cases lagged a year behind influenza deaths. Without a much larger data set, it is difficult to know whether the pattern detected in Seattle reflects anything more than the fact that both EL and influenza tended to peak in late winter. It seems clear that on a global scale there was no such pattern; influenza deaths certainly peaked in late 1918 throughout the world at levels at least 10-fold greater than in subsequent years, while EL incidence began to increase in early 1919 and did not peak until 1923 (19).

The infectivity of the 2 diseases was also dramatically different. While influenza is one of the most contagious infectious diseases known, EL was apparently not contagious and evidence of transmission between individuals was extremely rare (26). The sheer number of cases was also vastly different. It is likely that influenza infected 500 million people between September of 1918 and March of 1919 (41), killing at least 20 million (42). By contrast, it is unlikely that there were more than 150,000 cases of EL worldwide over the course of 6 years. If the 1918 influenza virus caused EL, it did so in only a tiny percentage of influenza victims.

Evaluation of the pathology and epidemiology of the 1918 influenza pandemic suggests that it differed more in degree than in kind from other influenza pandemics (43). The 1918 virus caused more severe illness in a higher proportion of victims than other influenza strains, but it did not cause a different illness. Even the primary viral pneumonia that killed some victims in a matter of days, as frequently seen in 1918, can be caused by other influenza strains. There is no clinical or pathological evidence that the 1918 virus replicated anywhere outside its traditional environment, the respiratory tract (44–47).

EL demonstrated a consistent and distinct pathology, with CNS inflammation confined predominantly to the midbrain and other brainstem nuclei (2, 8, 21). In order for the influenza virus to cause EL directly, it must be capable of reaching the midbrain and brainstem and selectively infecting specific CNS cells. Since EL victims displayed none of the ordinary respiratory symptoms or pathology associated with influenza, this hypothesis requires that the influenza virus acquired a completely different tropism in a very small number of victims. A mechanism whereby an influenza virus can so dramatically alter its host cell specificity has not been proposed.

Descriptions of EL-like illnesses contemporaneous with past influenza epidemics suggested a causal connection between the 1918 influenza and EL that was appealing to many observers. Many historical sources describe ‘nervous complications’ of pandemic influenza. A closer look at papers describing the 1891 influenza pandemic reveals that the authors meant generalized symptoms such as headache, fever, muscle aches, and weakness (48–50), which are consistent with acute influenza but do not fulfill the diagnostic criteria of EL. As von Economo pointed out, these descriptions did not include distinctive CNS symptoms such as diplopia, ptosis, paresis, and other extrapyramidal signs. Moreover, apparently none of the prior outbreaks resulted in the complication of chronic parkinsonism, an important characteristic of the EL epidemic. In the absence of descriptions of CNS symptoms characteristic of EL, reports of various neurological complications of pandemic influenza before 1918 must be interpreted with caution.

Certainly, no syndrome resembling EL accompanied the 2 influenza pandemics occurring since 1918. The mortality levels in the 1957 and 1968 influenza pandemics were markedly lower than in 1918, but morbidity was extremely high. Despite hundreds of millions of influenza cases, there was no outbreak of EL following these pandemics (51–54). In the companion paper in this issue (pages 696–704), influenza RNA was not detected by the sensitive method of RT-PCR in archival samples of EL and pEP.

In summary, there is no clear connection between EL and pandemic influenza. Clinically, epidemiologically, and morphologically they represent distinct entities, and EL has disappeared. The historical reports of similar outbreaks cannot be confirmed and it is possible that the epidemic of 1916–1925 was the only time in history that this disease emerged on a large scale. While it seems reasonable to rule out influenza virus as the direct caused of EL, it will prove difficult to demonstrate another etiologic agent consistent with all that is known about EL.

Identifying the Cause of EL

Since 1918, many viruses that can cause encephalitis have been identified. None of them, however, consistently produce the typical range of acute symptoms and subsequent parkinsonian sequelae characterizing EL (55). Bacterial and environmental toxins and secondary reactions to infectious agents may also produce nonspecific encephalitic symptoms. pathological damage similar to that seen in EL can be caused by a variety of viruses and toxins (26, 56). However, in order for a specific agent to be proposed as the cause of EL, it must meet several criteria. First, it must be capable of causing not only general encephalopathic symptoms but also the characteristic spectrum of symptoms typical of EL. Furthermore, the disease must meet certain epidemiologic criteria: occurring in small outbreaks over the course of a few months in the late winter or early spring; showing almost no evidence of direct contagion; infecting people of all ages; causing death in 20%–50% of cases; and resulting frequently in chronic parkinsonism. Although sporadic cases resembling EL are occasionally described, it is evident that the disease has disappeared in epidemic form. Any proposed cause, therefore, should also include some explanation as to why the agent disappeared or changed after the 1920s.

The arboviruses, included in the families Togaviridae and Flaviviridae, among others, are small arthropod-borne RNA viruses that meet many of these criteria. Although none of the known arboviruses cause symptoms exactly like EL, the general disease course is similar (55, 56). Several arboviruses cause epidemics of encephalitis but, since infection occurs through an arthropod vector including mosquitoes and, less commonly, ticks, there is no evidence of direct contagion. All ages are affected and neurological sequelae sometimes result. However, encephalitis epidemics due to arboviruses have a distinct seasonality that is virtually the opposite of that associated with EL, tending to peak in late summer or early autumn. It is possible that an arbovirus carried by a different host such as fleas or mites may have caused EL, although it is difficult to explain how such a vector, which must have been widespread in Europe and North America from 1918–1925, could completely disappear.

EL was first confused with poliomyelitis, but it was soon recognized that the symptoms, seasonality, and, especially, the age distribution of cases of polio were inconsistent with EL. polio, however, provides a example of an infectious agent that appears to spread via many asymptomatic carriers and that produces its symptoms by destruction of specific target areas in the CNS. It is possible that an agent similar to poliovirus may have been responsible for EL (2, 7, 8). The Enterovirus group (Picornaviridae), to which poliovirus belongs, also includes the coxsackieviruses and echoviruses, some of which can cause encephalitis. None of them, however, are known to cause the full spectrum of symptoms comprising EL and all of them, like polio, tend to be diseases predominantly of childhood.

In addition to enteroviruses and arboviruses, there are many other viral agents that can cause encephalitis. A comprehensive review of the possible causes of postencephalitic parkinsonism by Casals and colleagues (55) includes a detailed discussion of the disease course and sequelae of many viral diseases. The authors conclude that, although the pathology and epidemiology of pEP are consistent with a viral etiology, none of these viruses result in chronic parkinsonian sequelae resembling pEP.

Although the pathology and epidemiology of EL are consistent with a viral etiology, other causes cannot be ruled out. A bacterial strain producing a neurotoxin, an environmental toxin, or a drug with a toxic side effect on the CNS represent alternative causes. Although conceivable, it is difficult to test the hypothesis that a common bacterial strain could acquire and then lose the capability to produce a neurotoxin, thereby explaining the disappearance of the disease. Many attempts were made to culture bacteria from acute EL cases and no organism was consistently found. The existence of a widespread environmental toxin or medical treatment has also not been demonstrated. An alternative explanation postulates that an autoimmune reaction to a virus by the mechanism of antigenic mimicry may have caused the symptoms of EL.

The cause of EL remains a mystery. No known agents are capable of producing the syndrome, and it is troubling to conclude that an agent capable of causing such a devastating illness could emerge from an unknown source and then disappear. Recently a hitherto unknown herpes virus (HHV-8) was identified as the causative agent in Kaposi's sarcoma using the molecular technique of representational difference analysis (57). perhaps progress toward the application of such techniques to paraffin-embedded tissue will someday make it possible to identify the agent responsible for the enigmatic entity of encephalitis lethargica.

Acknowledgment

The authors thank Dr. David Morens for helpful discussion. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense. This is US government work; there are no restrictions on its use.

References

Author notes