Acute Flaccid Myelitis in the United States: 2015–2017

From January 1, 2015, to June 30, 2017, a clinical case of AFM was defined as a patient with acute onset of focal limb weakness. In June 2017, the case definition was modified from focal to flaccid to more accurately describe the weakness associated with AFM. After July 1, 2017, a clinical case of AFM was defined as a patient with acute onset of flaccid limb weakness. Health departments and clinicians submitted to the CDC demographic, clinical, epidemiologic, laboratory, and radiographic information on patients meeting the clinical case criteria. An AFM expert panel, consisting of pediatricians and neurologists from CDC and US academic centers, classified patients as confirmed or probable on the basis of a thorough review of all available data for each patient. A confirmed case was a patient who met the clinical case criteria and had an MRI showing a spinal cord lesion largely restricted to gray matter and spanning ≥1 spinal segments. A probable case was a patient who met the clinical case criteria and had CSF pleocytosis (white blood cell [WBC] count >5 cells/mm³). The primary analysis included confirmed pediatric cases (patients ≤21 years of age); collection of data for adult and probable cases began in August 2015 and is reported in Supplemental Tables 5 and 6. Patients who met the clinical case criteria but could not be classified as confirmed or probable cases were excluded from the primary analysis (Supplemental Table 7).

The CDC requested sterile-site (eg, blood or serum and CSF) and nonsterile-site (eg, nasopharyngeal or oropharyngeal respiratory and stool) specimens from each patient. We tested for poliovirus using virus isolation and molecular methods. During 2015 and 2016, the CDC tested all specimens for EVs and rhinoviruses (EVs/RVs) using a qualitative real-time reverse transcription–polymerase chain reaction (RT-qPCR) pan-EV assay¹¹  and an EV typing assay (species A–J) by viral protein 1 reverse transcription-seminested PCR and Sanger sequencing.¹²  Specimens were also tested for parechoviruses (species A–B) by using RT-qPCR.¹³  Respiratory specimens were further tested by using an EV-D68–specific assay.¹⁴  Any respiratory specimens that tested positive for EV-D68 by either viral protein 1 seminested sequencing or by the EV-D68–specific assay were considered EV-D68 positive. Because of wide availability of the pan-EV/RV RT-qPCR assay in the United States and in an effort to preserve CSF specimen volume, the CDC modified its testing algorithm in 2017 to perform only typing of EV/RV-positive specimens submitted by external laboratories. We provide results for the earliest specimen submitted if >1 specimen was submitted per specimen site. Results submitted to the CDC from other laboratories are also summarized in this report; diagnostic testing protocols varied by laboratory.

Data were entered into a Microsoft Access database. Differences in categorical variables were assessed by using the binomial test and Mantel-Haenszel χ² tests. Annual and state-level incidence rates were calculated by dividing the total number of confirmed pediatric cases by the corresponding estimate of the population ≤21 years of age. Descriptive analyses were performed by using SAS version 9.4 (SAS Institute, Inc, Cary, NC). Data were collected as part of standardized public health surveillance and determined by the CDC not to be research involving human subjects.

From 2015 through 2017, 43 states reported 305 clinical cases. The CDC confirmed 193 pediatric AFM cases from 41 states, with most cases (143, 74%) reported in 2016 (Fig 1). The average annual incidence was 0.71 per million in patients ≤21 years of age (Supplemental Fig 2 A–C). The majority of patients confirmed with AFM were male (118 cases, 61%), white race (102 cases, 53%), and non-Hispanic ethnicity (67 cases, 35%). Median age was 6 years (range: 3 months to 21 years; interquartile range [IQR]: 3–10 years). Onset of limb weakness occurred from August through November for 118 (61%) patients; 104 (88%) occurred in 2016 during the same period. Overall, 161 (83%) patients had fever, cough, rhinorrhea, vomiting, and/or diarrhea for a median of 5 days (range: 0 to 28 days; IQR: 2–7.5 days) before limb weakness onset; 120 (62%) had fever, 127 (66%) had respiratory illness, and 45 (29%) had gastrointestinal illness. In total, 106 patients (55%) had only 1 or 2 limbs affected compared with 87 patients (45%) who had 3 or 4 limbs affected (P > .05); 152 (79%) patients had at least 1 upper limb affected. At the time of limb weakness onset, 63 (33%) patients also presented with cranial nerve findings, 69 (36%) had quadriplegia, 51 (28%) had altered mental status, and 59 (33%) required mechanical ventilation (Table 1). Twenty-eight (15%) had underlying medical conditions, of which 15 (8%) reported asthma (Supplemental Table 8). One death was reported in 2017 of a 21-year-old man with cerebral edema and meningoencephalomyelitis. Spinal cord histopathology demonstrated mononuclear inflammatory infiltrates and focal neuronal necrosis; no etiologic agent was identified by using molecular and immunohistochemical techniques.

MRI testing protocols varied by hospital, and not all patients meeting the clinical case criteria had total spine imaging performed. Lesions were most commonly identified in the cervical spine in 144 (80%) patients, of which 21 (15%) had all cervical levels affected (Table 2). Among these 144 patients with cervical spine lesions, 122 (85%) also had documented upper limb weakness. Brain MRIs were conducted in 178 (92%) patients, of which 65 (38%) had brainstem lesions, with abnormalities most frequently identified in the pons or medulla (Table 2).

A total of 183 (95%) patients had CSF analysis (WBC count, protein, glucose) results available. The median interval between onset of limb weakness and CSF collection was 2 days (range: −4 to 26; IQR: 1–3). CSF pleocytosis was present in 144 patients (81%), with most having a lymphocytic predominance. Median CSF WBC count was 75 cells per microliter (range: 0 to 3261; IQR: 13–158). Median CSF protein and glucose were 47 mg/dL (range: 13 to 915; IQR: 32–66) and 60 mg/dL (range: 4 to 125; IQR: 53–70), respectively (Supplemental Table 9). Overall, at both CDC and non-CDC laboratories, 91 (47%) of 193 patients had a pathogen detected from any site, 20 (10%) had a pathogen detected from a sterile site (CSF and sera), and 82 (42%) had a pathogen detected from a nonsterile site (upper respiratory and stool specimens). Time from limb weakness onset or respiratory or febrile illness onset to specimen collection is available in Supplemental Figs 3 and 4.

Only 1 of 81 patients with CSF tested at a CDC laboratory had a positive result (CV-A16 in a 2015 patient). Among 72 patients with serum specimens tested, 2 were positive for EVs: EV-D68 was detected in a patient with onset of limb weakness in 2016, and CV-A16 was detected in the same patient with CV-A16 in the CSF just described. Among 90 patients with upper respiratory specimens tested, 32 (36%) were positive for EV/RV and 2 were positive for parechovirus. Among 77 patients with stool specimens tested, 15 (19%) were positive for EV/RV and 1 was positive for parechovirus; none were positive for poliovirus (Table 3). Respiratory and stool specimens collected within the initial 5 days of respiratory or febrile illness onset were more likely to be RT-PCR positive than those collected after >5 days (67% vs 25%, respectively; P = .0002).

CSF was tested in 170 patients; 4 were positive for EV. Three of the 4 were sent to the CDC for confirmation and typing; only 1 was positive for EV and subsequently typed as CV-A16 (2015 patient). Adenovirus, Epstein-Barr virus, human herpesvirus 6, and mycoplasma were also detected in CSF from 6 patients. Of the 123 patients with sera tested, 9 were positive for EV, West Nile virus, mycoplasma, and CV-B. Among 151 patients with respiratory specimens tested, 61 (46%) were positive for EV/RV. Among 78 patients with stool tested, 22 (52%) were positive for EV/RV (Table 4).

In our summary of national AFM surveillance from 2015 to 2017, we demonstrate that cases were widely distributed across the United States, the majority of cases occurred in late summer or fall, children were predominantly affected, there is a spectrum of clinical severity, and no single pathogen was identified as the primary cause of AFM. We conclude that symptoms of a viral syndrome within the week before limb weakness, detection of viral pathogens from sterile and nonsterile sites from almost half of patients, and seasonality of AFM incidence, particularly during the 2016 peak year, strongly suggest a viral etiology, including EVs.

EVs, such as poliovirus and EV-A71, can cause a wide clinical spectrum ranging from asymptomatic infection to much rarer presentations, such as multisystem organ failure and neuroinvasive disease (eg, limb paralysis and meningitis).^15–17  In 2014, multiple reports of AFM cases and concurrent outbreaks of severe respiratory disease caused by EV-D68 nationwide supported a temporal association between EV-D68 and AFM. Similar to other neuroinvasive EVs, EV-D68 can readily induce flaccid paralysis in mouse models and has been detected in CSF specimens of patients with AFM, albeit rarely.^18–26  In 2014, EV-D68 was the most common virus detected, accounting for 20% of respiratory specimens tested. However, multiple testing strategies (including virus-specific PCR or RT-PCR, broad “family-specific” PCR or RT-PCR, and metagenomic sequencing) did not identify a convincing single etiologic signal in sterile-site specimens.⁸ 

EV testing from 2015 through 2017 demonstrated similar findings to the 2014 investigation. Diagnostic yield from CSF was low (CV-A16 was detected in one patient in 2015, and EV-D68 was detected in one patient in 2014), multiple pathogens were identified from both sterile and nonsterile sites in approximately half of patients, and poliovirus was not detected in any cases. EV-D68 positivity from respiratory specimens of AFM patients with onset in 2015 to 2017 (24%) was similar to 2014; however, EV-D68 was also detected in patients later classified as noncases in this study. After the 2014 investigation, testing for EV-D68 increased because of wider availability of an EV-D68–specific assay, which allowed for more-rapid detection at treating facilities but still limited broader characterization of other type-specific EV circulation that could also be contributing to respiratory and neuroinvasive disease within a community.^27–29  Subnational estimates of EV-D68 circulation in the same regions where AFM cases are occurring have not been well characterized, although studies to evaluate these trends are being implemented. Although a temporal association between EV-D68 and AFM has been reported, additional evidence is still needed to more clearly establish a causal association.

EV circulation is widespread, with a US seasonal peak in the late summer and early fall²¹  complicating interpretations correlating EV-positive specimens from nonsterile sites with rare clinical outcomes like paralysis. Seroprevalence estimates of neutralizing antibodies for EVs range from 33% to 99%, depending on serotype.³⁰  The large burden of EV-D68 respiratory disease detected in 2014 led to hypotheses that EV-D68 was a newly emerging EV also associated with AFM. In a 2012 study of 436 subjects aged 2 to 81 years from Kansas City, Missouri, with sera collected before the 2014 EV-D68 respiratory outbreak, researchers demonstrated detection of neutralizing antibodies against the major 2014 EV-D68 outbreak strain for all subjects.³¹  Serologic evidence of widespread infection with EV-D68, even before the first notable increase of AFM in 2014, suggests that if EV-D68 was the primary cause of AFM in 2014 and 2016, other factors must play a role in the development of this rare outcome.

Because of limited availability of spinal pathology specimens, it remains unknown whether paralysis associated with AFM results from direct pathogen invasion of the central nervous system or a postinfectious process, possibly immune mediated. AFM may be analogous to other diseases that present as AFP, such as transverse myelitis or GBS, which are linked to immune-mediated damage of the spinal cord or peripheral nerves, respectively.^32,33  Alternatively, it may be similar to other EV pathogenesis, like poliovirus, which causes paralysis by direct viral destruction of motor neurons in the anterior horn of the spinal cord. EV-A71 has also been shown to cause brainstem encephalitis and paralysis via direct mechanisms.²  The short interval between a preceding viral syndrome and onset of limb weakness in these AFM cases could support direct viral invasion as a mechanism. However, given that identification of a pathogen in the CSF is rare, further research into potential immunologic and genetic mechanisms leading to AFM, as well as more sophisticated pathogen evaluations, is warranted. Examinations of soluble and cell-associated markers of immune system activation, particularly in the central nervous system, are underway, as well as investigations of the potential role of other immune-mediated pathogeneses.

Globally, countries vary widely in their surveillance for acute flaccid limb weakness. Many countries conduct AFP surveillance to rule out poliovirus, a critical component of the poliovirus eradication initiative. Some countries, including France, Sweden, and the Netherlands, identify AFM cases through laboratory-based EV surveillance networks, which limits reporting of AFM cases caused by other potential pathogens.^34–38  In our data, viruses other than EVs were identified in 60% of patients with pathogens detected from a sterile site. Important differences in AFM case ascertainment make it challenging to compare the global epidemiology of AFM across countries.

There are several limitations to this analysis, including reliance on voluntary reporting from clinicians and state and local health departments. Although AFM is not a nationally notifiable condition, all states have either made AFM reportable or have reporting requirements for AFP without an alternate diagnosis or for new or emerging conditions of public health importance.. Variability in awareness can lead to underreporting to the health department and affect timely collection of appropriate specimens. Although the case definition was expanded in 2015 to include persons of all ages, adults may also have other reasons for weakness (eg, spinal cord stroke, GBS), making diagnosis of AFM more challenging in this population. AFM likely has a range of clinical severity, and only moderate or severe cases may present for clinical evaluation. Ongoing analysis of the spectrum of illness is necessary to inform surveillance methods and improve specificity and sensitivity of the current case definition. Lastly, EV surveillance to characterize type-specific annual trends and geographic variability is limited, challenging our ability to attribute one specific EV as the etiologic cause of AFM.

With this report, we provide the most comprehensive summary of clinical, laboratory, radiologic, and epidemiologic data of AFM in the United States to date. The clinical symptoms shortly before weakness onset in most patients and detection of pathogens in sterile and nonsterile sites suggest viruses play a key role in the pathogenesis of AFM. Although EVs should continue to be evaluated as an etiology of AFM, ubiquitous circulation of multiple types of EVs and the rarity of AFM, even during EV season, highlight important knowledge gaps. Additional studies are needed to assess risk factors, establish causality, and develop a more-comprehensive understanding of the mechanisms that lead to AFM.

We thank the Vaccine Preventable Diseases Surveillance Coordinators at the local and state health departments that have contributed to the launch and maintenance of the AFM surveillance program, the Council of State and Territorial Epidemiologists for their continued collaboration on AFM surveillance, Sandra Roush and Holly Vins from the National Center for Immunization and Respiratory Diseases Surveillance Office for their ongoing support to the CDC AFM team, and Drs Wun-Ju Shieh and Sarah Reagan-Steiner from the National Center for Emerging and Zoonotic Infectious Diseases Pathology Branch for their expertise and consultation.

AFM

acute flaccid myelitis

AFP

acute flaccid paralysis

CDC

Centers for Disease Control and Prevention

CSF

cerebrospinal fluid

CV

coxsackievirus

EV

enterovirus

EV/RV

enterovirus and rhinovirus

GBS

Guillain-Barré syndrome

IQR

interquartile range

RT-qPCR

qualitative real-time reverse transcription–polymerase chain reaction

WBC

white blood cell