Enterovirus A71 (EV-A71) is a non-polio enterovirus that causes hand, foot, and mouth disease and neurological disease, particularly in young children.¹ In the Asia–Pacific region, EV-A71 has caused large-scale, cyclical epidemics (every 1–3 years) since the late 1990s. This virus has been associated with brainstem encephalitis, leading to fatal cardiopulmonary collapse, and severe cases of acute flaccid myelitis, leading to long-term paralysis.²

Outside of Asia, EV-A71 circulation is poorly understood and therefore difficult to predict; however, outbreaks have been described in Australia (in 2013),³ Spain (in 2016),⁴ France (in 2016–17 and 2019),⁵ and Germany (in 2015–16 and 2019).⁶ Although small-scale, sporadic, regional outbreaks have occurred in the USA (in 2003 and 2005),⁷ EV-A71 is not commonly reported, accounting for less than 1% of typed enteroviruses that have been reported to the US Centers for Disease Control and Prevention (CDC) National Enterovirus Surveillance System since 1970.^8-10 In our study, we aimed to describe the clinical characteristics of an outbreak of EV-A71 neurological disease amongst children in Colorado, USA in 2018.

Between March 10, and Nov 10, 2018, 74 children presenting to Children’s Hospital Colorado met our case definition for enterovirus neurological disease. EV-A71 was the most commonly identified enterovirus serotype, which was identified in 43 (58%) of 74 children (figure 1; figure 2).

The median age of the 43 children with EV-A71 neurological disease was 22·7 months (IQR 4·0–31·9); three (7%) children were aged 4 years or older at presentation (table 1). Most children had previously been healthy (39 [91%] children) and were male (34 [79%] children), white (33 [77%] children), and non-Hispanic (36 [84%] children). All children with EV-A71 neurological disease resided within the greater Denver metropolitan area, which has a population of approximately 3 million people. 20 (47%) children attended daycare, including a cluster of three (7%) children with acute flaccid myelitis from the same daycare classroom. 12 (28%) children had household contacts who were unwell, including one (2%) child with EV-A71-associated meningitis with a sibling from the same household who presented concurrently to a different hospital with untyped enteroviral meningitis.

Neurological symptoms were commonly preceded by fever (in 43 [100%] children; median duration 2·5 days [IQR 2·0–3·0]; median maximum temperature 39·1°C), irritability (in 37 [86%] children), vomiting (21 [49%] children), and respiratory symptoms (20 [47%] children; table 1). Prodromal symptom onset preceded neurological symptom onset by a median time of 1 day (IQR 1–3). 18 (42%) children with EV-A71 neurological disease had hand, foot, or mouth lesions (or a combination of the three) at or before neurological onset.

EV-A71 was most frequently detected in rectal specimens (in 32 [94%] of 34 specimens) and least frequently detected in cerebrospinal fluid (in eight [20%] of 40 specimens; table 2). Of the cases of EV-A71 with specimens from nasopharyngeal and oropharyngeal sites tested, 14 (54%) of 26 children had enterovirus detected at only one site. Amino-terminal VP1 sequences (297 nucleotides, including the B-C and D-E variable regions) of all EV-A71 strains from this outbreak belonged to clade C1 genotype, except for a single C2 genotype; however, all strains showed 100% consistency in amino acid (capsid) identity from translated nucleic acid sequences. VP1 nucleotide sequences from all enteroviruses identified in our study have been deposited in the GenBank sequence database (accession numbers MK836112–81 and MK491180).

40 (93%) children with EV-A71 neurological disease had findings suggestive of meningitis (table 1). Of 39 children with EV-A71 who had cerebrospinal fluid collected and analysed, 34 (87%) children had a pleocytosis (median white blood cell count, 135 cells per μL [IQR 50–292]) with a neutrophil predominance in 16 (47%) of 34 children and lymphocyte predominance in 13 (38%) of 34 children (table 2). Six (75%) of the eight children who had EV-A71 that was identified in cerebrospinal fluid had these samples collected within 2 days of symptom onset, six (75%) of eight samples were from infants younger than 2 months, and six (75%) of eight children presented with fever alone without evidence of encephalitis or myelitis. Eight (33%) of 24 children who were examined by MRI showed leptomeningeal enhancement.

31 (72%) children with EV-A71 neurological disease showed evidence of encephalitis, which included neurological findings of myoclonus in 26 (60%) children, ataxia in 25 (58%) children, and autonomic instability in 20 (47%) children, often in association with irritability (table 1). Myoclonus that worsened during sleep and truncal ataxia interfering with normal gait were commonly cited reasons for caregivers seeking medical attention (appendix; video). None of the six children with jerking movements captured while receiving EEG monitoring showed electrographic seizure activity. 12 (28%) children showed irregular breathing patterns, ten (23%) children showed persistent tachycardia (without evidence of myocarditis in any of the five children assessed by cardiology evaluation), six (14%) children showed mottling of the skin, and two (5%) children showed hypertension, all of which were features consistent with autonomic instability. MRI abnormalities consistent with encephalitis included dorsal brainstem lesions (in 16 [66%] of 24 children imaged) and lesions in the dentate nuclei of the cerebellum (in 12 [50%] of 24 children imaged); however, supratentorial lesions were uncommon (in three [13%] of 24 children imaged; table 2).

14 (33%) children with EV-A71 neurological disease were treated with intravenous immunoglobulin (1 g/kg per day for 2 days) to treat brainstem encephalitis with autonomic instability or acute flaccid myelitis (or both; table 3), and two children with brainstem encephalitis and acute flaccid myelitis refractory to intravenous immunoglobulin subsequently received intravenous methylprednisolone (30 mg/kg per day for 3–5 days). Children with EV-A71 neurological disease were treated in hospital for a median duration of 3 days (IQR 2–5) with ten (23%) children receiving intensive care unit support for autonomic instability (n=7), bulbar paralysis (n=1), altered mental status (n=1), or seizures (n=1). Three (7%) children subsequently required acute inpatient rehabilitation for acute flaccid myelitis. 20 (47%) children had fully recovered by discharge and 17 (41%) children received outpatient rehabilitation therapy. 39 (93%) of 42 children with EV-A71 neurological disease showed complete recovery by 1–2 months of follow-up, including nine (90%) of ten children with acute flaccid myelitis. Persistent deficits included residual, but improved, tremor and ataxia in two (5%) children at 1–2 months of follow-up, which subsequently resolved, and complete paralysis of one arm in a child with acute flaccid myelitis, which persisted for 1 year after disease onset. One child with residual tremor at discharge was lost to follow-up.

During the study period, an additional 31 cases of neurological disease associated with 11 enteroviruses other than EV-A71 were reported (tables 1-3). EV-A71 and other enteroviruses were most commonly identified in rectal specimens, but EV-A71 was less likely to be identified in cerebrospinal fluid than other enteroviruses (in eight [20%] of 40 specimens vs 20 [77%] of 26 specimens; p<0·0001). Fever was common in all children, but EV-A71 showed a longer median duration of fever (2·5 vs 1·0 days; p=0·0066), irritability (in 37 [86%] of 43 children vs nine [29%] of 31 children; p<0·0001), and presence of hand, foot, or mouth lesions (in 18 [42%] children vs six [19%] children; p=0·041) than other enteroviruses. Neurological signs (myoclonus, ataxia, and autonomic instability) and imaging findings, suggestive of encephalitis with brainstem and cerebellar involvement, best distinguished EV-A71 cases from other enterovirus-associated cases, and weakness was also more common in children with EV-A71. No children with EV-A71 developed non-cardiogenic pulmonary oedema or cardiopulmonary collapse, and there were no deaths of children with EV-A71 in our cohort; however, one neonate with echovirus 11 sepsis with multiorgan system involvement died at age 6 days.

Ten (23%) children with EV-A71 neurological disease met the Council of State and Territorial Epidemiologists case definition of acute flaccid myelitis (table 4). All were male and had cerebrospinal fluid pleocytosis suggestive of meningitis, and all but one had concomitant signs of encephalitis. Three of these ten children with EV-A71-associated acute flaccid myelitis were hypotonic at initial presentation, with most developing more generalised and symmetric hypotonia and weakness, including the trunk and extremities, later in the disease course. Urinary retention was observed in six of these children.

Longitudinal grey matter-predominant spinal cord lesions were present on MRI in 14 (74%) of 19 children with EV-A71 neurological disease who received imaging, including all children with acute flaccid myelitis, which affected the cervical (in 14 [74%] of 19 children imaged), thoracic (in eight [44%] of 18 children imaged), and lumbar regions (in seven [47%] of 15 children imaged), and these lesions were accompanied by nerve root enhancement in six (32%) of 19 children imaged. Notably, four (29%) of 14 children with abnormalities in the central grey matter of the cervical cord had no clinically documented weakness. An additional three children with EV-A71 neurological disease had weakness but did not receive imaging; therefore, these cases were not classified as acute flaccid myelitis.

A comparison of children with EV-A71-associated acute flaccid myelitis during the study period (n=10) and children with EV-D68-associated acute flaccid myelitis presenting to Children’s Hospital Colorado between 2013 and 2018 (n=8) is shown in table 4. Enteroviruses were detected more frequently in rectal and oropharyngeal specimens in children with EV-A71-associated acute flaccid myelitis (in our cohort) and in nasopharyngeal specimens in children with EV-D68-associated acute flaccid myelitis (in the children presenting to Children’s Hospital Colorado between 2013 and 2018). Both groups showed low detection of enterovirus in cerebrospinal fluid. Compared with children with EV-D68-associated acute flaccid myelitis, children with EV-A71-associated acute flaccid myelitis were younger (aged 19 vs 100 months; p=0·034) and had neurological onset earlier after prodromal symptom onset (median 1 vs 6 days; p=0·011). The presence of hand, foot, or mouth lesions (in six of ten children with EV-A71 disease vs zero of eight children with EV-D68 disease; p=0·013) and irritability distinguished EV-A71-associated acute flaccid myelitis; whereas, in children with EV-D68-associated acute flaccid myelitis, prodromal respiratory symptoms were more uniformly noted across the groups (in three of ten children with EV-A71 vs eight of eight children with EV-D68; p=0·04). Notably, patterns of brainstem and spinal cord involvement on MRI were comparable in both groups. Compared with children with EV-D68-associated acute flaccid myelitis, those with EV-A71-associated acute flaccid myelitis had milder weakness at onset and improved strength at 1–2 months of follow-up (figure 3). Those with EV-A71-associated acute flaccid myelitis also had a shorter duration of hospital stay (median 6 vs 32 days; p=0·023) and greater proportion of these children showed full neurological recovery (nine [90%] of ten children vs one [13%] of eight children; p=0·0029).

To our knowledge, this is the largest reported outbreak to date of EV-A71 neurological disease in the Americas. Children with EV-A71 in this outbreak were predominantly boys, with the clinically distinguishing findings of myoclonus, ataxia, weakness, and autonomic instability. Approximately half of this cohort had the clinical hallmark of hand, foot, or mouth lesions. Characteristic imaging findings included changes in the dorsal brainstem, dentate nuclei of the cerebellum, and grey matter of the spinal cord that were similar to previous outbreaks of EV-A71.¹⁷ We found no cases of non-cardiogenic pulmonary oedema, cardiopulmonary collapse, or death, as has been reported with previous EV-A71 outbreaks, and most children showed full recovery, despite most having evidence of encephalitis or myelitis.

Temporal and geographic clustering of young children presenting with a febrile illness associated with myoclonus, ataxia, and autonomic instability indicated a possible EV-A71 outbreak in our community. One factor provoking the initial outbreak investigation was an increase in enterovirus-associated neurological cases in the spring, before the expected late summer to early autumn circulation of enteroviruses in Colorado. The reasons for the geographically isolated outbreak and prolonged period of circulation, extending from spring into the late autumn, are unclear. Other notable epidemiological features of this outbreak included clustering of cases amongst children aged 4 years and younger in the Denver metropolitan area, with evidence of intraclassroom and intrahousehold spread. Outside of the paediatric cases included in this report, EV-A71 neurological disease was identified in two adults in the Denver metropolitan area during the outbreak period (unpublished data). Both cases occurred in adults who were immunocompromised, one of whom met Council of State and Territorial Epidemiologists criteria for confirmed acute flaccid myelitis and the other had probable acute flaccid myelitis; both had EV-A71 identified in cerebrospinal fluid.

Early recognition of the EV-A71 outbreak enabled educational outreach to help frontline providers to recognise the unique clinical features differentiating neurological disease due to EV-A71 from other clinical syndromes, to reinforce the value of timely specimen collection and enterovirus testing, and to standardise management strategies. These steps were important because some of the neurological symptoms due to EV-A71, such as myoclonus, ataxia, or autonomic dysfunction, could be easily misattributed to other diagnoses. For instance, myoclonus could prompt evaluation for seizures or be mistaken for benign infantile myoclonus; ataxia can be misdiagnosed as post-infectious cerebellitis, acute cerebellar ataxia, or suspected toxic ingestion; and autonomic dysfunction can present as irregular breathing patterns, which can be mistaken for Kussmaul respirations, prompting evaluation for diabetic ketoacidosis or toxic ingestion, as persistent tachycardia, prompting cardiology evaluation for myocarditis, or as skin mottling, prompting evaluation and therapy for presumed bacterial sepsis. When these symptoms present in the setting of a febrile illness, a heightened index of suspicion for neurological disease associated with EV-A71 is warranted to avoid delayed diagnosis.

The Colorado Department of Public Health and Environment conducted public health outreach via an immediate electronic public health notification network (Epi-X) to alert health departments in other states and three statewide Health Alert Network messages to inform Colorado health-care providers. The CDC did an investigation of enterovirus-related cases of neurological disease throughout the USA, finding no additional clusters of EV-A71 neurological disease (unpublished data).

Identification of children with EV-A71 in this outbreak relied upon viral detection from non-sterile site specimens, particularly rectal specimens, where EV-A71 can be detected for weeks to months after an infection.¹⁸ The low frequency of EV-A71 detection in the cerebrospinal fluid of children with neurological disease is similar to that of other outbreaks;^4,7 however, this outbreak additionally highlights that EV-A71 might be more likely to be found in cerebrospinal fluid early in the course of disease, as evidenced by increased detection in cerebrospinal fluid from infants undergoing lumbar puncture shortly after fever onset. Although identification of enterovirus in cerebrospinal fluid is specific, inclusion of non-sterile site specimen collection and testing is essential to the sensitivity of enterovirus surveillance and neurological case identification.¹⁹ Further, nasopharyngeal and oropharyngeal specimens are commonly considered interchangeable for enterovirus detection; however, we found a high discrepancy of enterovirus testing results between these two sites in EV-A71 cases. Optimal enterovirus detection in neurological cases should include enterovirus PCR testing of cerebrospinal fluid, the rectum or stool, skin and oral lesions (when present), and nasopharyngeal and oropharyngeal specimens. Although not done in our cohort, increasing evidence suggests that blood might also be a high-yield specimen for enterovirus detection, particularly in neonates.²⁰ Novel serological assays have detected enterovirus antibodies in cerebrospinal fluid from patients with acute flaccid myelitis, including during this outbreak, which provides additional supportive evidence and could become the diagnostic standard for enterovirus neurological infection in the absence of detected virus in cerebrospinal fluid.²¹

The high number of children showing full neurological recovery in our cohort is similar to that reported in EV-A71 outbreaks in Spain (in 2016)⁴ and China (in 2013),²² but differs from the long-term outcomes found in Australia (in 2013),³ Taiwan (in 1998–2003),²³ and Colorado (in 2003 and 2005),⁷ where a higher prevalence of long-term paralysis due to myelitis, neurodevelopmental delays due to encephalitis, and mortality due to cardiopulmonary collapse have been noted. More severe disease in some outbreaks of EV-A71 might reflect a combination of previously identified factors, including a genetic predisposition among certain ethnic groups,^24,25 differences in viral strains,²⁶ and different treatment protocols.^27,28 Children with brainstem encephalitis were selectively administered intravenous immunoglobulin if they showed signs of autonomic instability or acute flaccid myelitis; methylprednisolone was administered to those refractory to intravenous immunoglobulin and at least 2 days from symptom onset, similar to protocols used in previous outbreaks, with good outcomes.⁴ However, our uncontrolled observational cohort study does not allow systematic evaluation of response to therapy, and most children recovered regardless of treatment.

This outbreak demonstrates that EV-A71 can cause clusters of acute flaccid myelitis that are clinically distinguishable from EV-D68-associated acute flaccid myelitis.²⁹ Hand, foot, or mouth lesions were only found in children with EV-A71-associated acute flaccid myelitis and not those with EV-D68-associated acute flaccid myelitis, whereas preceding respiratory symptoms were more characteristic of EV-D68-associated acute flaccid myelitis. EV-A71-associated acute flaccid myelitis was initially difficult to diagnose because most children presented within a day of illness onset with myoclonus, ataxia, and irritability before developing weakness; whereas children with EV-D68-associated acute flaccid myelitis presented with limb weakness at the time of neurological onset without these associated findings. By contrast with EV-D68-associated acute flaccid myelitis, the weakness that ultimately developed with EV-A71 was more generalised and more symmetrical, often involving the trunk and proximal extremities, despite similar spinal cord imaging findings in patients in both groups.³⁰ The weakness in EV-A71-associated acute flaccid myelitis fully resolved in all but one patient: an older child with persistent asymmetrical upper limb weakness similar to cases of EV-D68-associated acute flaccid myelitis. The frequency of full neurological recovery in EV-A71-associated acute flaccid myelitis cases from this outbreak was much higher than in some previously described cases of EV-A71-associated acute flaccid myelitis³¹ and of EV-D68-associated acute flaccid myelitis.³² Notably, during this outbreak, we detected EV-A71 in some children with acute flaccid myelitis and EV-D68 in others, underscoring that several non-polio enteroviruses that likely cause acute flaccid myelitis can circulate concurrently but have differentiating clinical presentations and outcomes.

Limitations of our outbreak investigation include that no standardised comprehensive laboratory and imaging evaluation approaches were used; testing was based on the clinical discretion of treating providers, likely biasing results towards more severe presentations. Only children meeting the confirmed acute flaccid myelitis case definition were classified as having acute flaccid myelitis, which excluded potential cases of acute flaccid myelitis with weakness that we speculate would have had characteristic lesions had MRI been performed. Follow-up was restricted to documented neurological examinations at follow-up visits after 1–2 months without standardised neurocognitive or functional assessments; therefore, subtle longer term neurological sequelae cannot be ruled out. Cases were presumed to be caused by enterovirus infection if an enterovirus was identified by VP1 sequencing in any specimen; however, enteroviruses are ubiquitous and can be shed from non-sterile sites without being the cause of disease. Additionally, the case definition for EV-A71 neurological disease included all children who had EV-A71 identified in any specimen, allowing the potential for including co-infections in this group, as evidenced by one enterovirus coinfection (EV-A71 detected in an oropharyngeal specimen with EV-D68 detected in nasopharyngeal and rectal specimens). Because we did several statistical comparisons between findings in the different groups, p values should be interpreted accordingly.

In conclusion, despite low-level seasonal endemic EV-A71 circulation in the USA since the 1980s and previous small sporadic outbreaks, including in Colorado,⁷ EV-A71 has not previously been reported to cause an outbreak of this size or duration in the USA. However, since the 1990s in the Asia–Pacific region, EV-A71 has continued to cause large epidemics of hand, foot, and mouth disease, and neurological disease every 1–3 years, which has prompted the development of EV-A71 vaccines.³³ These cyclical patterns of enterovirus circulation are driven by serotype-specific immunity and the growth of susceptible birth cohorts,³⁴ with sudden shifts in dynamics due to changes in viral properties; however, why similar patterns have not been seen with EV-A71 to date in the USA is unclear. Enhanced clinical and laboratory surveillance for enterovirus neurological disease, including non-sterile site testing, is necessary to determine whether the 2018 Colorado EV-A71 outbreak is a sporadic, isolated incident or foreboding of impending cycles of neurological disease associated with EV-A71 in the USA.