Understanding Enterovirus 71 Neuropathogenesis and Its Impact on Other Neurotropic Enteroviruses

Abstract

Enterovirus A71 (EV‐A71) belongs to the species group A in the E nterovirus genus within the P icornaviridae family. EV‐A71 usually causes self‐limiting hand, foot and mouth disease or herpangina but rarely causes severe neurological complications such as acute flaccid paralysis and encephalomyelitis. The pathology and neuropathogenesis of these neurological syndromes is beginning to be understood. EV‐A71 neurotropism for motor neurons in the spinal cord and brainstem, and other neurons, is mainly responsible for central nervous system damage. This review on the general aspects, recent developments and advances of EV‐A71 infection will focus on neuropathogenesis and its implications on other neurotropic enteroviruses, such as poliovirus and the newly emergent Enterovirus D68. With the imminent eradication of poliovirus, EV‐A71 is likely to replace it as an important neurotropic enterovirus of worldwide importance.

Introduction

Enterovirus 71 (EV‐A71) belongs to a large group of enteroviruses that traditionally includes poliovirus, coxsackievirus A and B, echovirus and other numbered enterovirus (enteroviruses discovered later and given unique consecutive numbers) 86. More recent classifications of enteroviruses have utilized viral genomic sequence to further refine the nomenclature 44. Currently, the Enterovirus genus in the Picornaviridae family includes nine species groups (species A to H and species J), each consisting of a varying number of enterovirus serotypes. Species group A (25 serotypes) comprises about half the original coxsackieviruses A, including coxsackievirus A16 (CV‐A16), and an important numbered enterovirus, EV‐A71. Species group B, the largest group (63 serotypes), comprises all coxsackieviruses B, echoviruses and numbered enteroviruses. Species group C (23 serotypes) comprises the three poliovirus serotypes, the rest of the coxsackieviruses A and numbered enteroviruses, and species group D (five serotypes) includes enterovirus D68 (EV‐D68). Human rhinovirus species groups have recently been included to the Enterovirus genus 44.

EV‐A71 and CV‐A16 are major causes of hand‐food‐mouth disease (HFMD) and herpangina in infants and young children. In addition, EV‐A71 infection has proven to be complicated by aseptic meningitis, acute flaccid paralysis (AFP) and encephalomyelitis 100. In fact, EV‐A71 was first isolated in 1969 from the feces of a 9 month old infant with fatal encephalitis in the USA 94. Since then, many large HFMD outbreaks with significant neurological complications and mortality have been reported worldwide in Europe, Australia, USA and Asia 39, 67, 68, 112. With the expected eradication of poliovirus, EV‐A71 could well emerge as the most prevalent neurotropic enterovirus. This review supplements previous reviews 67, 68, 85, 100 on the general aspects of EV‐A71 infection and on more recent developments and advances. How the pathology and neuropathogenesis of AFP and encephalomyelitis and impacts on our understanding of other neurotropic enteroviruses, including poliovirus and the newly emergent Enterovirus D68 (EV‐D68), will be discussed.

Virology

EV71 shares a common morphology and structure with other members of the Picornaviridae family. The non‐enveloped, symmetrically icosahedral virion of about 30 nm in diameter, is one of the smallest viruses 90. The virus capsid or protein shell is made up of 60 repeating units called protomers, each composed of four viral structural proteins (VP1 to VP4). The crystal structure of EV‐A71 has been recently determined at 3.8 Å resolution 88, 89. Enteroviruses are relatively stable viruses, insensitive to organic solvents and resistant to acid pH, thus their ability to survive conditions in the gastrointestinal tract 86.

The EV‐A71 genome is a single‐stranded, positive (+) sense RNA molecule of approximately 7.5 kb. It encodes for four structural proteins and several non‐structural proteins. The entire genome of some virus strains has been sequenced. Phylogenetic analysis of EV‐A71 strains using variations in the VP1 region, has been used to study the molecular epidemiology of outbreaks 68, 100. Based on these genomic variations, EV‐A71 was divided into three genogroups: A (prototype virus); B (B1‐B5) and C (C1‐C5). This has enabled epidemiologists and virologists to study the source and spread of the viruses across countries and continents. It was found that B1 and B2 genogroup viruses were predominant in Japan, Europe and the USA in the 70s and 80s. More recently, B3 to B5 genogroup viruses circulated in Southeast Asia, while genogroup C1 to C5 viruses were also detected in Southeast Asia, the Far East and Australia 68, 100.

Several putative virus receptors on host plasma membranes that enable EV‐A71 attachment and viral entry have been found, including the human scavenger receptor class B2 (SCARB2), P‐selectin glycoprotein ligand‐1 (PSGL‐1), annexin II, sialic acid, dendritic cell‐specific ICAM‐3 grabbing non‐integrin, nucleolin and heparin sulphate 54, 77, 101, 102, 119, 121, 122. Perhaps the best studied virus receptor is SCARB2, a ubiquitously expressed major transmembrane lysosomal protein. In the human CNS, SCARB2 was found mainly on neurons and glial cells. A human SCARB2 transgenic mouse model that demonstrated susceptibility to EV‐A71 infection and convincing similarity to HFMD complicated by encephalomyelitis was successfully developed recently, confirming SCARB2 as an important functional receptor 28. PSGL‐1, dendritic cell‐specific ICAM3 grabbing non‐integrin, and sialic acid are limited to leukocytes, dendritic cells, epithelial, and endothelial cells, respectively 54, 77, 121, 122. The ubiquitous heparin sulphate and nucleolin apparently could facilitate attachment but not viral entry 101, 102.

Viral Transmission

Like poliovirus, humans are the only natural hosts for EV‐A71. Person‐to‐person transmission is usually through fecal‐oral or oral‐oral routes, and sometimes via droplets 86. Viral isolation from throat swabs appears to be significantly higher than from rectal swabs or feces, suggesting that the oral‐oral route may be more important than the fecal‐oral route 82, 83. Moreover, detection of viral antigens and RNA in tonsillar crypt squamous epithelium but not in other parts of the gastrointestinal tract, suggests that the palatine tonsil and oral mucosa is the major source of viral shedding into the oral cavity, and very likely in the feces as well 34. As virus can be cultured from skin vesicles 82, 83, it is conceivable that cutaneous‐oral transmission could also occur but its relative significance is unknown. Figure 1 summarizes a hypothesis for the possible routes of viral entry into the body and the CNS, viral replication sites and shedding in Enterovirus 71 encephalomyelitis, based on available data from human and animal studies.

Figure 1.

A hypothesis for the possible routes of viral entry into the body and CNS, viral replication sites and viral shedding in Enterovirus 71 encephalomyelitis, based on available data from human and animal studies. While person‐to‐person transmission is mainly via fecal‐oral and oral‐oral routes, the portal/s for viral entry are unknown but likely to be via the palatine tonsil and oropharyngeal mucosa after viral replication in these tissues. Viral entry through the gastrointestinal tract could possibly occur as well. Viremia spreads infection to the skin and perhaps other organs where there is further viral replication. Neuroinvasion occurs through retrograde axonal transport up cranial and spinal motor nerves to infect motor nuclei and anterior horn cells in the brainstem and spinal cord, respectively. Within the CNS, motor and other neural pathways spread virus further. However, hematogenous spread into the CNS still cannot be excluded. As no gastrointestinal tract replication sites have been identified, the source of fecal viral shedding may be mainly the tonsil/oropharynx.

Clinical Manifestations

The incubation period after EV‐A71 infection is about 3 to 6 days 64. Uncomplicated EV‐A71 infection manifests as HFMD or the more localized herpangina 85. Typically, HFMD patients have a self‐limiting, acute febrile illness presenting with oral ulcers on the tongue and buccal mucosa, and vesicular or small erythematous maculopapular rashes on the hands, feet, knees or buttocks. Oral ulcerations in herpangina are similar to HFMD, except that they are more confined to the posterior oral cavity involving the buccal mucosa, tonsillar pillars, soft palate or uvula. HFMD and herpangina are considered to represent both ends of the clinical spectrum of mucocutaneous manifestations of EV‐A71 85.

The neurological complications are well known but rare in EV71 infection 37, 85. Aseptic meningitis is characterized by mononuclear infiltration in the meninges without involving the brain parenchyma. Meningitis may occur by itself or accompany HFMD/herpangina. AFP is characterized by acute limb weakness plus decreased reflexes without disturbance of limb sensation, similar to classical poliomyelitis. It is due to isolated infection of groups of anterior horn cells in the spinal cord. Encephalomyelitis is an inflammation involving both the brain and spinal cord. It often occurs with manifestations of HFMD/herpangina, but aseptic meningitis is invariably found. The typical encephalomyelitis patient presents with a few days of non‐specific fever, lethargy and very acute onset of neurological symptoms and signs, and often dies within hours of hospital admission 35, 37, 61. Encephalomyelitis is characterized by hyporeflexic flaccid muscle weakness, myoclonic jerks, ataxia, nystagmus, oculomotor and bulbar palsies and coma, in various combinations 112. Brainstem encephalitis or rhombencephalitis has been used to describe a similar syndrome without muscle weakness. In the terminal stages, there is autonomic dysregulation (cold sweating, mottled skin, tachycardia, hypertension, hyperglycemia, tachypnea), fulminant pulmonary edema and myocardial dysfunction 37. The four clinical stages of infection have been defined as: stage 1, HFMD/herpangina; stage 2, CNS involvement; stage 3, cardiopulmonary collapse; stage 4, convalescence/sequelae 35, 37. Most infections are stage 1, with some cases progressing to stage 2, and a few advancing to the most severe stage 3. The exact prevalence of severe neurological complications is unknown but is probably much less than 1%. In stage 4, patients may recover or develop significant neurological sequelae including cerebellar dysfunction, severe respiratory and motor impairment, delayed neurodevelopment and reduced cognitive function 14, 38.

Typically in encephalomyelitis, the T2 weighted MRI shows mainly hyperintense lesions in the spinal cord and brainstem 19, 21, 96. In the spinal cord, there was involvement of the anterior horns bilaterally and at all levels, continuing into the posterior areas of the medulla, pons and midbrain. More rarely, lesions may be found in the thalamus, frontal and parietal lobe white matter and other parts of the brain 96, 126. In AFP patients, lesions were confined to the anterior horn regions of the cord and ventral nerve roots 17, 18, 19.

Laboratory Diagnosis

As a biosafety level 2 virus, EV‐A71 isolation and identification in clinical specimens, eg, throat swabs and cerebrospinal fluid (CSF), can be done in most laboratories and is the gold‐standard for diagnosis. For this purpose, human rhadomyosarcoma and Vero cells (African green monkey kidney cells) remain the most commonly used, and the cytopathic effects are characteristic. Isolated viruses can be identified by conventional neutralization tests, immunofluorescence staining using EV‐A71‐specific monoclonal antibodies or by specific reverse transcriptase‐polymerase chain reaction (RT‐PCR) 85, 112. Overall, the rate of virus isolation from CSF is very low (<5%) compared to throat, vesicle and rectal swabs 82, 112. Recently, several new molecular diagnostic methods for simple, rapid identification of EV‐A71 in clinical specimens have been described, including nested RT‐PCR, real‐time RT‐PCR, reverse transcription loop‐mediated isothermal amplification and droplet digital PCR 29, 59, 97, 127.However, these new methods require specific primers (targeted to the VP1 gene) that have to be constantly revised accordingly to the prevalent EV‐A71 genogroups/variants in a new outbreak area.

Pathology of Fatal EV‐A71 Encephalomyelitis

Based on autopsy cases, the pathology of EV‐A71 encephalomyelitis is beginning to be better understood 9, 10, 34, 36, 61, 98, 114, 120, 123, 125, 129. EV‐A71 encephalomyelitis showed a stereotyped distribution of inflammation in which the most intense inflammation was found in the spinal cord, brainstem (Figure 2A), hypothalamus and cerebellar dentate nucleus (Figure 2B). The inflammation was typical of viral encephalitis consisting of perivascular cuffing, edema, necrosis, microglial nodules and neuronophagia (Figure 2B–D). At all levels of the cord, the anterior horn cells showed direct evidence of neuronotropism with demonstrable viral antigens/RNA in neuronal bodies and processes (Figure 2E,F). Similarly, neuronal viral antigens/RNA may also be demonstrated in other inflamed areas in the medulla, pontine tegmentum/floor of the fourth ventricle (Figure 2A) and the midbrain posterior to the corticospinal tracts (crus cerebri), hypothalamus and subthalamic nucleus. However, viral antigens/RNA may be very focal even in severely inflamed areas. The cerebral cortex, shows much milder and less consistent inflammation mostly in the motor area. Inflammation was absent in the globus pallidus, putamen, caudate nucleus, thalamus, hippocampus, anterior pons and cerebellar cortex. The whole of the medulla, including the inferior olivary nucleus and cranial nerve nuclei, reticular formation and nucleus ambiguus area showed inflammation (Figure 2D). No viral inclusions were found. Some inflammatory cells around infected neurons may show phagocytosed viral antigens but generally, meninges, glial cells, blood vessels and choroid plexus were all negative for viral antigens/RNA. Mild meningitis was found throughout the CNS. Inflammatory cells were predominantly CD68+ macrophages and CD15+ neutrophils 114, 123. CD3+, CD8+ and CD20+lymphocytes, if present, were found sparsely in perivascular spaces or in the parenchyma 114, 123.

Figure 2.

A. Horizontal section of the whole cerebellum at the mid‐pons level from an autopsy of EV‐A71 encephalomyelitis in which the distribution of moderate (blue dots) and severe inflammation (black dots) was shown. The inflammation was confined to the tegmentum/floor of 4th ventricle (C) and dentate nucleus (B), sparing the anterior pons (C, arrow) and cerebellar hemisphere. Intense parenchymal inflammation, necrosis and perivascular cuffing was seen in the medulla (D). In inflamed areas, viral antigens (E) and RNA (F) were localized to infected neuronal bodies and processes with or without evidence of neuronophagia. In an experimental mouse model of encephalomyelitis infected by intramuscular injection into jaw/facial muscles, viral antigens (G, arrow) and RNA (H, arrow) were demonstrated in cranial nerves. Permissions to reproduce images were obtained for figure C 115 and for figure G and H 103.

Although lungs revealed marked pulmonary edema with multifocal hemorrhage with or without inflammation, viral antigens/RNA were not detectable. Myocarditis is generally absent or if present, is very mild 9, 120. Viral antigens/RNA were absent in the myocardium. The mesenteric lymph nodes, Peyer's patches, spleen and thymus showed congestion and/or reactive hyperplasia 34, 123, 129. In the palatine tonsil, viral antigens/RNA were demonstrated in squamous epithelium lining the crypts but not in the squamous epithelium covering the external surface nor in lymphoid cells or blood vessels 34. Viral antigens/RNA were not detected in the gastrointestinal mucosa, liver, pancreas, kidney and skeletal muscle fibers. EV‐A71 has been isolated from CNS tissues including cerebrum, cerebellum, pons, medulla, and spinal cord 37, 61, 120, and from non‐CNS tissues, including tonsils, intestines, pancreas, myocardium and blood. However, viral cultures from lungs, liver and kidney showed negative results 9, 22.

Neuropathogenesis of EV‐A71 Encephalomyelitis

Neuronotropism was first demonstrated in autopsy cases of EV‐A71 encephalomyelitis from a HFMD outbreak in Malaysia in 1999 10, 61, 113. Viral antigens/RNA were detected mostly in neurons that were undergoing neuronophagia or degeneration, suggesting that viral cytolysis is an important mechanism for tissue injury. Several recent animal models of EV‐A71 infection, including mouse and monkey models, have since confirmed viral neuronotropism and cytolysis 28, 75, 76, 80, 128. However, other mechanisms of tissue injury, eg, immune mediated or bystander effects, cannot be excluded because neuronal viral antigens/RNA were often more focal than expected from the extent and severity of inflammation 114. There is a possibility that following EV‐A71 infection, apoptosis or autophagy may occur as demonstrated in vitro and in animal studies 25, 40, 48, 99, 111, 118. However, we have been unable to demonstrate apoptosis in our series of encephalomyelitis cases (unpublished observation).

Based on the stereotyped distribution of inflammation and viral antigens/RNA in human autopsies supported by findings in a mouse model, it was postulated that viral entry into the CNS may be via peripheral motor nerves 80, 114. As peripheral motor nerves originate in cord anterior horn cells, these neurons will be the first to be infected. It was reported in a mouse model that efferent motor axons adjacent to infected anterior horn cells in the spinal cord could be infected 80. Furthermore, cranial motor nerves (Figure 2G,H) could similarly facilitate viral retrograde axonal transport to reach the murine brainstem motor nuclei and reticular formation after intramuscular injection of virus into the jaw muscles 103. Being the first neurons to be infected, it is not surprising that motor and adjacent neurons in the spinal cord, medulla, pons and midbrain showed the most severe inflammation. The mild inflammation in the motor cortex is consistent with a delay in viral spread from cord/brainstem motor nuclei to the motor cortex via motor pathways. Other neural pathways may also be involved with viral transmission as the severely inflamed dentate nucleus is not directly connected to motor pathways. Interestingly, in contrast to the severely inflamed inferior olivary nucleus with which it has direct neural connections, the cerebellar cortex was never found to be inflamed. Like poliomyelitis, it is assumed that AFP is due to a localized, focal infection of a group of motor neurons in the spinal cord. Based on data from mouse models, which demonstrated severe skeletal muscle infection before CNS involvement, it is tempting to speculate that virus spreads from skeletal muscle to motoneuron junctions, up peripheral motor nerves and thence into the motor nuclei in the CNS 81, 103. However, human skeletal muscle infection has not be demonstrated.

Hematogenous spread of virus through the blood–brain barrier (BBB) has not been well studied and cannot be excluded. In in vitro experiments, EV‐A71 was able to infect, activate and/or induce apoptosis in human endothelial cells suggesting that viral penetration into the CNS via the BBB may be possible 53, 63. Intraperitoneal or intravenous inoculation in mouse and monkey models was followed rapidly by CNS infection presumably as a result of viremia but it is difficult to demonstrate if neuroinvasion was due to a breach in the BBB or due to virus entering skeletal muscle/motor nerves 28, 76, 80. Assuming that virus is able to cross the BBB, it may preferentially infect certain groups or types of neurons, eg, motor neurons, resulting in the stereotyped distribution of inflammation and viral antigens/RNA in the CNS.

Pulmonary edema and cardiopulmonary collapse, both terminal events in encephalomyelitis, may be related to neurogenic pulmonary edema 100, a concept that has been described in sudden deaths in bulbar poliomyelitis and other CNS diseases 4. As reviewed recently, there may a complex interplay of various pathogenetic mechanisms involving a cytokine storm of inflammatory mediators arising from strong systemic and CNS inflammatory responses and an excessive sympathetic hyperactivity/catecholamine release as a result of brainstem encephalitis 100. The end result is cardiotoxicity/dysfunction and acute pulmonary edema. Proinflammatory cytokines such as TNF‐α, IL‐1β, IL‐6, IL‐10, IL‐13 and IFN‐γ, and chemokines such as IL‐8, IFN‐γ induced protein, monocyte chemoattractant protein, monokine induced by IFN‐γ, were found to be increased in the cerebrospinal fluid and/or plasma in patients with brainstem encephalitis with or without pulmonary edema 109. Most of these inflammatory mediators that form part of the cytokine storm, except TNF‐α, IL‐1β, IL‐8 and IFN‐γ, could modulate changes in pulmonary vascular permeability to cause pulmonary edema. Pulmonary edema apparently could not be explained by hemodynamic factors because patients initially had normal cardiac function, pulmonary artery pressure and vascular resistance. Depletion of CD4 and CD8 lymphocytes, natural killer cells and lowered EV‐A71‐specific cellular response may also contribute to pulmonary edema 13, 106. Interestingly, neurogenic pulmonary edema in bulbar poliomyelitis was reported to best correlate with damage to the dorsal nuclei of the vagus and medial reticular nuclei (vasomotor) 4. We speculate that other contributory factors to sudden collapse may be damage to medullary respiratory control center and respiratory muscle paralysis following severe cord motor neuron injury. Unlike humans, pulmonary edema has not been reported in animal models including rodents and non‐human primates 28, 55, 75, 76, 80, 111.

So far, there is no evidence that mutations in the viral genome of EV‐A71 confer neurovirulence, unlike poliovirus in which mutations in the 5′ UTR can determine virulence 45. So far, analysis of whole EV‐A71genomic sequences derived from fatal and non‐fatal cases has not revealed any mutations responsible for neurovirulence. Nevertheless, several studies have demonstrated that mutations in structural and non‐structural genes may confer or attenuate neurovirulence in mice and monkey models 2, 3. There is still every possibility that neurovirulent viral strains could arise from specific viral mutations in future outbreaks of HFMD as recombinant events are very common among EV‐A71 strains 49, 58, 69. Other viral factors like the infective viral dose, host factors like immune status caused by diminishing protective maternal antibodies 62, 131 or immunosuppression 1, 41, genetic predisposition [2′‐5′‐oligoadenylate synthetase1, HLA A33, chemokine (c‐c motif) ligand 2] 8, 15, 32, and gender may be important and should be investigated further.

Treatment and Vaccines

There are currently no approved antiviral drugs for treatment of EV‐A71 infections but numerous candidates have been tested including type 1 interferons, ribavirin, suramin, pleconaril, inhibitors of EV‐A71 non‐structural proteins (eg, 3D polymerase) and cellular receptor inhibitors. Detailed results from these studies have been recently reviewed 95. Suramin, which binds to the viral capsid, prevents viral attachment to cells 92, 110. As it has demonstrated exceptional post‐exposure and therapeutic efficacy in mouse and rhesus monkey models, it appears to be a promising drug 92. Type 1 interferon and ribavirin have also shown efficacy in mouse models 52, 56. Milrinone, a cyclic nucleotide phosphodiesterase inhibitor which is used as a treatment for congestive cardiac failure may reduce mortality in patients 105, 107.

Normal, polyclonal or hyperimmune (anti‐EV‐A71) intravenous immune globulin (IVIG) has been used as a therapeutic modality for EV‐A71 infection. Some studies reported improvements in severe EV‐A71 infection after IVIG administration 12, 84. Chemokine and cytokine levels before and after treatment apparently showed reduction 108. In Taiwan, IVIG is now routinely used for the treatment of CNS complications 12, 105. In addition to IVIG, passive immunization using specific anti‐EV‐A71 monoclonal antibodies (human or humanized) is being actively developed as better therapies based on results in animal experiments 11, 24, 43, 51. Interestingly, passive immunization was found to be useful to ameliorate experimental brainstem encephalitis 103.

The success of both the formalin‐inactivated and live attenuated vaccines in controlling poliovirus epidemics underscores the potential use of mass vaccination as an effective means to control EV‐A71 epidemics. Numerous studies have demonstrated that inactivated EV‐A71 virus vaccine can protect against lethal EV‐A71 infection in mouse models, either by transplacental transfer of maternal antibodies, passive immunization using antibodies from immunized mice or active immunization 81, 116, 124. A concerted effort to develop vaccines suitable for human populations reached a milestone when inactivated vaccines were tested in three independent phase 3 clinical trials involving more than 10 000 children each 50, 130, 132. All three vaccines could elicit protective immune responses over the course of two epidemic seasons with vaccine efficacies of > 90%, and only mild adverse effects, such as fever. Unfortunately, these vaccines did not protect against CV‐A16 infection 50, 130, 132. Other vaccine formats, including DNA vaccines, viral‐like particles, synthetic peptides, and recombinant VP1 proteins derived from milk of transgenic mice expressing VP1have been developed and found to be immunogenic 20, 23, 27, 104.

Poliovirus, Coxsackievirus A16, Enterovirus D68 and Other Neurotropic Enteroviruses

The best known neurotropic enterovirus is undoubtedly poliovirus, which was discovered more than a century ago 91. The spectrum of neurological diseases caused by poliovirus includes, aseptic meningitis, paralytic poliomyelitis (spinal poliomyelitis, bulbar poliomyelitis and bulbospinal poliomyelitis), and the rare encephalitis 86. These clinical syndromes generally resemble those of EV‐A71, although poliovirus uses a different virus receptor, CD155 70. Poliovirus neuronotropism in the CNS has been convincingly demonstrated in mouse and monkey models by immunohistochemistry and by ultrastructural studies 33, 74, 76, and viral RNA has been localized in the human spinal cord motor neurons 26. As far as stereotypic distribution of inflammation is concerned, perhaps the closest to EV‐A71 encephalomyelitis is bulbospinal poliomyelitis 6. EV‐A71 associated AFP also appears similar to spinal poliomyelitis as evidenced by clinical and MRI features 17, 37. It is widely believed that poliovirus utilizes both hematogenous as well as retrograde axonal transport based on evidence extensively reviewed elsewhere 86, 91. The transgenic mouse model that expresses the human CD155 poliovirus receptor has provided additional support for fast retrograde axonal transport in motor nerves 46, 78, 93 similar to findings in the EV‐A71 mouse model 16, 80, 103. Furthermore, it is postulated that the poliovirus could enter peripheral nerve via the motoneuron junction after viral endocytosis and endosomal movement along microtubules to reach and infect cord motor neurons 45, 73, 79. The probable use of the motoneuron junction by EV‐A71 for entry into peripheral nerves remains to be investigated, although skeletal muscle infection in the mouse model is consistent with this notion. There is evidence that in poliovirus and EV‐A71 infections, the palatine tonsil may play an important role as a site for viral replication and as a portal for viral entry. Poliovirus has been cultured from tonsils and thought to be a portal for viral entry in experimentally infected chimpanzees 5. EV‐A71 has been isolated from human tonsils 22 and viral antigens/RNA localized to tonsillar crypt squamous epithelium 34.

Although similar in many ways to EV‐A71, neurological complications in CV‐A16 infection are probably very rare 30. Most of the evidence of CV‐A16 neurovirulence in anecdotal reports has been indirect because virus was not isolated from CSF or CNS tissues, but from extra CNS sites, such as throat or rectum. In large epidemics of HFMD where neurological complications are more likely to occur, it is not uncommon to have EV‐A71, CV‐A16 and other enteroviruses circulating together, thus confirmation of the causative neurotropic virus is difficult, unless virus was isolated from clinically relevant sites. With the advent of MRI and molecular diagnosis, perhaps better evidence for CV‐A16 neurotropism will be forth coming in future. Nonetheless, in at least one CV‐A16 associated rhombencephalitis (virus cultured from rectal swab), brain MRI showed hyperintense lesions that were strikingly similar to EV‐A71 30. We believe that CV‐A16 neurotropism has not been unequivocally demonstrated in human infected tissues or animal models 7, 57, 66.

EV‐D68 as a species group D serotype is unique among enteroviruses in that it has properties of both enteroviruses and rhinoviruses, which are predominant causes of the common cold 42. Until recently, EV‐D68 usually causes a rare self‐limited respiratory illness. From 2008, localized outbreaks were reported, culminating in more widespread outbreaks of >1000 cases of severe lower respiratory disease in 2014. Reports of AFP (later called acute flaccid myelitis) associated with EV‐D68 surfaced from 2005 onwards in the USA, France and Norway. Not surprisingly, most virus isolations from these outbreaks were obtained from patients' respiratory secretions; only rarely from blood and cerebrospinal fluid 31, 71, 87. The reasons for the emergence of neurotropic strains of EV‐D68 are still unknown. So far, no viral genomic variations have been linked to neurovirulence 31, 117. In Denmark, recent isolates had >98% genome homology with viruses associated with AFP but no AFP was reported 72. MRI showed hyperintensive, non‐enhancing lesions in bilateral anterior horns and/or central gray areas, and ventral roots, over extensive areas of the spinal cord, most prominently in the cervical cord. In the brainstem, lesions were found in the pontine tegmentum, medulla, midbrain, anterior pons and dentate nucleus 65, 71. No supratentorial lesions were found. By and large, the distribution of these lesions corresponded very well with MRI and autopsy findings in EV‐A71 encephalomyelitis 19, 114, 115, 126. The only post‐mortem case diagnosed as EV‐D68 infection by RT‐PCR and sequencing, showed extensive meningoencephalitis characterized by inflammation in the cerebellum, midbrain, pons, medulla and cervical cord, with prominent neuronophagia of motor nuclei 47. Unfortunately, further details of topographic distribution of inflammation or detection of viral antigens/RNA were not available. Based on available data, we speculate that like EV‐A71, EV‐D68 may be able to enter the CNS via retrograde viral transmission up peripheral and cranial motor nerves to infect motor neurons in the cord and brainstem, respectively. Moreover, it is likely that EV‐D68 is neuronotropic, with a predilection for motor neurons.

Many more enterovirus serotypes from various species groups have been associated with neurological syndromes 42. However, good evidence for some of these associations may be lacking because viral isolation from a clinically relevant site is often unavailable because of poor collection/storage techniques, obtaining specimens outside the optimal virus replication/shedding period and lack of proper lab facilities. Even in the recent EV‐D68 outbreaks, although epidemiological, clinical and other data suggested this, it remains unproven that EV‐D68 is the causative virus of the neurological complications. Further studies are needed to resolve these issues. In anecdotal reports, eg, echovirus 7 (B species group) encephalitis, a positive viral culture from the CSF, and good supporting MRI data, even in the absence of post‐mortem findings, is probably good enough evidence that echovirus 7 is neurotropic. Interestingly, the MRI findings in echovirus 7 encephalitis were similar to EVA‐71 encephalomyelitis 60. Thus, overall, neurotropic enteroviruses may share common neuropathogenesis and pathways to invade the CNS.

Conclusion

With the imminent eradication of poliovirus as the most prevalent neurotropic enterovirus the world has known, other non‐polio, neurotropic enteroviruses, notably EV‐A71, could replace it as a cause of widespread morbidity and mortality 68, 100. Much work remains to be done to more fully understand the neuropathogenesis of these viruses. Wider availability of newer molecular techniques, CNS diagnostic imaging and greater awareness of neurotropic potential of enteroviruses could help with the diagnosis and identification of future outbreaks of emerging or re‐emerging complicated enteroviral infections.

Conflict of Interest

The authors declare no conflict of interest.