Clinical Subpopulations in a Sample of North American Children Diagnosed With Acute Flaccid Myelitis, 2012-2016

Key Points

Question

Does the epidemiologic case definition of acute flaccid myelitis (AFM) encompass subgroups with distinct clinical presentations and possible alternative diagnoses?

Findings

In a case series of 45 patients meeting the US Centers for Disease Control and Prevention case definition of AFM, 24% had a definable alternative neurologic diagnosis. We identified clinical characteristics that distinguish between a homogenous subgroup of patients with AFM and a separate subgroup of children with definable alternative diagnoses that fulfill the broader epidemiologic AFM case definition.

Meaning

The epidemiologic case definition of AFM likely includes multiple distinct neurologic diagnoses that may present with overlapping clinical symptoms; the distinguishing clinical features identified here may be useful for defining a more homogeneous research population to enhance the power of studies on etiology and treatment.

Abstract

Importance

Acute flaccid myelitis (AFM) is an emerging poliolike illness of children whose clinical spectrum and associated pathogens are only partially described. The case definition is intentionally encompassing for epidemiologic surveillance to capture all potential AFM cases. Defining a restrictive, homogenous subpopulation may aid our understanding of this emerging disease.

Objective

To evaluate the extent to which the US Centers for Disease Control and Prevention (CDC) case definition of AFM incorporates possible alternative diagnoses and to assess the plausibility of a case definition that enriches the biological homogeneity of AFM for inclusion in research studies.

Design, Setting, and Participants

Retrospective case analysis of children younger than 18 years diagnosed as having AFM between 2012 and 2016 using the CDC case definition. Group 1 included patients recruited from the United States and Canada based on the CDC case definition of AFM. Group 2 included patients referred to the Johns Hopkins Transverse Myelitis Center for evaluation of suspected AFM. Patients’ records and imaging data were critically reviewed by 3 neurologists to identify those cases with definable alternative diagnoses, and the remaining patients were categorized as having restrictively defined AFM (rAFM). Clinical characteristics were compared between patients with rAFM (cases) and those with alternative diagnoses, and a case description distinguishing these AFM groups was identified. Interrater reliability of this description was confirmed for a subset of cases by a fourth neurologist. Data were analyzed between May 2017 and November 2018.

Main Outcomes and Measures

Proportion of patients with possible alternative diagnosis.

Results

Of the 45 patients who met the CDC AFM case definition and were included, the mean age was 6.1 years; 27 were boys (60%); and 37 were white (82%), 3 were Asian (7%), 1 was Hispanic (2%), and 4 were mixed race/ethnicity (9%). Of the included patients, 34 were classified as having rAFM, and 11 had alternate diagnoses (including transverse myelitis, other demyelinating syndromes, spinal cord stroke, Guillain-Barre syndrome, Chiari I myelopathy, and meningitis). Factors differing between groups were primarily asymmetry of weakness, lower motor neuron signs, preceding viral syndrome, symptoms evolving over hours to days, absence of sensory deficits, and magnetic resonance imaging findings. A case description was able to reliably define the rAFM group.

Conclusions and Relevance

We present an approach for defining a homogeneous research population that may more accurately reflect the pathogenesis of the prototypical poliomyelitis-like subgroup of AFM. The definition of rAFM forms a blueprint for inclusion criteria in future research efforts, but more work is required for refinement and external validation.

This study evaluates the extent to which the US Centers for Disease Control and Prevention case definition of acute flaccid myelitis incorporates possible alternative diagnoses.

Introduction

Pseudo-polio has been described sporadically over the last several decades. An increase in cases in the United States was first suspected in California in 2012 and was followed by clear outbreaks in the late summer and fall of 2014, 2016, and 2018.^{1,2,3,4,5,6,7} After the 2014 epidemic in the United States, the US Centers for Disease Control and Prevention (CDC), along with the Council of State and Territorial Epidemiologists, adopted a standardized case definition of acute flaccid myelitis (AFM) for purposes of epidemiologic surveillance of this disorder. A confirmed case of AFM was defined as acute onset of focal limb weakness and a magnetic resonance image (MRI) showing a spinal cord lesion largely restricted to gray matter and spanning 1 or more spinal segments. A probable case of AFM was defined as acute onset of focal limb weakness and cerebrospinal fluid (CSF) with pleocytosis.⁸ Epidemiologic evidence suggested that the outbreaks of AFM in the United States may be associated primarily with enterovirus D68,⁹ although other implicated viruses include enterovirus-A71, Epstein-Barr virus, and adenovirus.¹⁰

Scientists have since adopted this case definition to define research cohorts,^{4,7,8,9,10,11,12,13,14,15,16} and clinicians have used it to guide the diagnosis and treatment of children presenting with limb weakness. While this definition has aided comprehensive ascertainment of cases for epidemiologic surveillance of AFM, our experience suggests that the CDC case definition may also capture children with acute weakness owing to other definable illnesses. Among these are transverse myelitis, Guillain-Barre syndrome, ischemic myelopathy, acute disseminated encephalomyelitis, anti–myelin oligodendrocyte glycoprotein myelitis, and other myelopathies.^14,17

Acute flaccid myelitis is a poorly understood syndrome, and much remains to be learned regarding etiology, pathophysiology, biomarkers, prognosis, and treatment. Research is urgently needed to address these knowledge gaps. However, the power of such studies would suffer if subjected to the inadvertent inclusion of patients with neurologic disorders other than the prototypical infectious poliomyelitis-like illness that is intended to be captured within the epidemiologic case definition of AFM. In the absence of a biomarker or criterion standard for diagnosis of AFM, leveraging the most distinctive clinical features of AFM as viewed through the lens of classical poliomyelitis (ie, pure lower motor neuron impairment with spinal gray matter lesion) offers a rational approach to defining a subpopulation. The aims of this study are 2-fold: first, to determine the extent to which the CDC case definition may include patients with alternative diagnoses, and second, to identify clinical characteristics that differentiate between restrictively defined AFM (rAFM) and the alternative diagnoses that also meet the CDC case definition of AFM.

Here, we present a case series of patients drawn from 24 states/provinces across the United States and Canada with suspected AFM at the time of clinical presentation and who met the CDC case definition. Within this AFM population, we define a subgroup of patients with rAFM based on the absence of a definable alternative diagnosis and generate a case description that reliably identifies this rAFM population. This restrictive definition generates a more homogeneous population that likely reflects a shared pathophysiology, although external validation will be required.

Methods

Study Population

Inclusion criteria for this study were the CDC case definition for confirmed or probable AFM: an illness with onset of acute flaccid paralysis in at least 1 limb, plus supportive evidence of myelitis including MRI showing a spinal cord lesion largely restricted to gray matter and spanning 1 or more spinal segments and/or cerebrospinal fluid (CSF) with pleocytosis (white blood cell count >5 cells/μL; to convert to ×10⁹/L, multiply by 0.001),⁸ and diagnosis occurring in the United States or Canada after January 2012.

Patient information was collected from 2 partially overlapping patient groups. Group 1 included patients self-referred or referred by physicians with approval of the Johns Hopkins institutional review board for a study of host genetic susceptibility to AFM. Patients with medical records describing their acute clinical course were included. Group 2 consisted of patients presenting consecutively between 2014 and 2017 to the Johns Hopkins Transverse Myelitis Center for consultation regarding suspected AFM. A subset of group 2 patients was reported in 2018.¹⁸ Written consent was obtained for patients in group 1. Group 2 was granted a waiver of consent by the institutional review board.

Review of Clinical Data

Medical records and primary source MRI (when available) were reviewed by a neurologist (M.J.E.) for group 1 and by 2 neurologists (E.G.-L. and C.A.P.) for group 2. Patients who met a well-defined alternative diagnosis in addition to the AFM case definition were assigned to the “AFM with possible alternative diagnosis” (AFM-ad) category. The remaining patients, who met only the CDC case definition for AFM and no other alternative diagnosis, were classified as having rAFM.

To identify characteristics that differed between patients with rAFM and AFM-ad, each clinical variable was compared using either χ² or t tests. P values less than .05 were considered significant, and all P values were 2-sided. We generated a description of rAFM based on characteristics that were present in 100% of patients classified as rAFM and the absence of features that were present only in patients with AFM-ad or commonly seen in disorders in the differential diagnosis for AFM, including transverse myelitis, spinal cord ischemia, Guillain-Barre syndrome, and acute demyelinating encephalomyelitis, but not present in patients with rAFM cases nor reported in the AFM literature. A final provision was made for the consideration and exclusion of alternative diagnoses.

To validate interrater reliability of this definition, a subset of patients from group 1 was randomly selected for independent neurologist review (J.R.N.). Group 2 was not used for evaluation of interrater reliability because it was not predicted to contain a sufficient number of patients with AFM-ad. The reviewer was blinded to the prior assignment of diagnoses by study neurologists, was not previously involved in the development of the case description of rAFM, and had not been involved in the care of any of these patients. The reviewer was provided only the written case description, with no verbal discussion on its application. Following review of the records of 10 randomly selected patients, the description was further revised owing to divergent classifications for 1 patient. One additional patient was also excluded from the study owing to insufficient data available; specifically, a description of the presentation and neurologic examination in the acute phase was not provided in the medical record. An additional 10 randomly selected patients from group 1 were reviewed by J.R.N. using the revised description, with complete agreement on classification.

Results

Forty-five patients were included in total, all of whom met the CDC case definition for AFM. The mean age was 6.1 years; 27 were boys (60%); and 37 were white (82%), 3 were Asian (7%), 1 was Hispanic (2%), and 4 were mixed race/ethnicity (9%). Group 1 included 32 patients, of whom 26 had MRI available in addition to written medical records. Group 2 included 19 patients evaluated in person, including review of records and imaging where available. Six patients belonged to both groups and thus were included only once in this study. On record and imaging review, a total of 34 patients were classified as having rAFM: 17 from group 1, 12 from group 2, and 5 who were part of both groups. Eleven patients (24.4%) were classified as having AFM-ad: 9 from group 1, 1 from group 2, and 1 who was in both groups. The most common alternative diagnoses were transverse myelitis and spinal cord ischemia, among others (Table 1).

Table 1. Alternative Diagnoses for Patients Meeting the CDC Case Definition for AFM After Record Review.

Diagnosis	Cases, No.
Transverse myelitis	4
Spinal cord ischemia	3
Polyradiculoneuropathy	1
Clinically isolated syndrome, demyelinating	1
Meningitis	1
Myelopathy owing to severe Chiari I malformation	1

Abbreviations: AFM, acute flaccid myelitis; CDC, US Centers for Disease Control and Prevention.

Clinical characteristics, including demographics, symptoms, and examination at presentation (Table 2) and MRI features, electrophysiology, and CSF studies (Table 3), are outlined for all patients. Among patients with rAFM, the data were notable for a wide range of severity, from a single affected limb to quadriparesis with or without respiratory failure. Of note, quadriparesis and monoparesis were the 2 most common presentations, respectively, although cases representing the full spectrum of severity between these extremes were also present. Features that were typical of rAFM included a preceding fever or viral syndrome, asymmetric onset of weakness, hypotonia, and hyporeflexia. Patients with rAFM had longitudinally extensive MRI lesions ranging from gray matter predominant to exclusive involvement of the anterior horns, with or without involvement of the dorsal pons, dorsal medulla, and dentate nuclei of the cerebellum, and with or without ventral nerve root enhancement. Cerebrospinal fluid findings often showed lymphocytic pleocytosis with mild or no protein elevation, although CSF cell counts were normal in 1 case. Electromyography and nerve conduction studies were performed in only 9 patients with rAFM (26%) and largely on long-term follow-up. In all patients with rAFM, electromyography and nerve conduction studies showed an isolated motor neuropathic disorder, also with electromyography features of denervation in 8 of 9 (88%).

Table 2. Clinical Characteristics of 45 Paralyzed Children With CDC-Defined AFM.

Characteristic	No. (%)		P Valuea
	Restrictively Defined AFM (n = 34)	AFM With Possible Alternative Diagnosis (n = 11)
Male	22 (64.7)	5 (45.4)	.26
Age at onset, mean (range), y	5.48 (<1-15.0)	8.02 (<1-17)	.08
Age at onset, median, y	5.0	10.0	NA
Any weakness	34 (100.0)	10 (90.9)	.07
1 Limb only	9 (26.5)	1 (9.1)	.23
2 Limbs only	8 (23.5)	6 (54.5)	.05b
3 Limbs only	5 (14.7)	0	.18
4 Limbs	11 (32.3)	3 (27.3)	.75
Bilateral lower extremity only	4 (11.8)	5 (45.5)	.02b
Facial	7 (20.6)	0	.10
Bulbar	8 (23.5)	1 (9.1)	.30
Neck	12 (35.3)	1 (9.1)	.10
Asymmetric onset	33 (97.1)	4 (36.4)	<.001b
Decreased tone	32 (97.0)	7 (77.8)	.05b
Decreased reflexes	26 (96.3)	9 (90.0)	.45
Absent reflexes	18 (66.7)	6 (60.0)	.71
Increased tone	0	2 (22.2)	.006b
Increased reflexes or clonus	0	5 (45.5)	<.001b
Respiratory compromise	8 (23.5)	1 (9.1)	.30
Bowel/bladder compromise	2 (5.9)	6 (54.5)	<.001b
Sensory deficits	0	7 (63.6)	<.001b
Extremity pain or paresthesia	9 (27.3)	4 (36.4)	.57
Autonomic deficits	1 (2.9)	0	.56
Encephalopathy	0	0	NA
Onset to peak, h
<1	0	3 (30.0)	.001b
<6	7 (21.2)	4 (40.0)	.43
7-24	8 (24.2)	0 (0)	.60
Preceding illness	34 (100.0)	7 (63.6)	<.001b
Viral syndrome	31 (91.2)	3 (27.3)	<.001b
Fever	31 (91.2)	3 (30.0)	<.001b
Congestion or rhinorrhea	22 (66.7)	2 (20.0)	.009b
Cough	22 (66.7)	2 (20.0)	.009b
Difficulty breathing	8 (24.2)	1 (9.1)	.28
Myalgia	2 (6.1)	0	.40
Vomiting/diarrhea	8 (24.2)	1 (9.1)	.28
Neck pain or stiffness	11 (33.3)	2 (18.2)	.34

Abbreviations: AFM, acute flaccid myelitis; CDC, US Centers for Disease Control and Prevention; NA, not applicable.

^aP values were determined from χ² for categorical variables and from t test for continuous variables.

^bSignificant P value.

Table 3. Diagnostic Testing Results of 45 Paralyzed Children With CDC-Defined AFM.

Result	No. (%)		P Valuea
	Restrictively Defined AFM (n = 34)	AFM With Possible Alternative Diagnosis (n = 11)
MRI: T2-hyperintense spinal cord signal	34 (100.0)	11 (100.0)	NA
Gray matter predominant	31 (96.9)	8 (72.7)	.02b
Fully limited to gray matter	18 (56.2)	2 (18.2)	.03b
Anterior horns predominant	12 (37.5)	5 (45.4)	.64
Completely limited to anterior horns	1 (3.12)	0	.55
MRI: nerve root enhancement
Any including cauda equina	12 (37.5)	3 (30.0)	.67
Anterior root predominant	6 (18.7)	0	.14
Cauda equina enhancement	6 (18.7)	2 (20.0)	.93
Parenchymal enhancement (brain or spinal cord)	0	1 (11.1)	.13
MRI: T2-hyperintense signal
Dorsal pons	16 (61.5)	2 (22.2)	.04b
Supratentorial white matter (any size)	0	3 (33.3)	.002b
Supratentorial white matter (>1 cm)	0	2 (25.0)	.009b
CSF
White blood cells, mean (SD), cells/μL	211 (40.5-102.5)	50 (9.0-91.0)	.40
Protein, mean (SD), g/dL	54.3 (32.8-65.0)	144 (50.1-119.0)	.02b
Pleocytosis >5 cells/mm3	30 (96.8)	6 (60.0)	.002b
Elevated protein (by reported normal range of the laboratory)	10 (45.4)	7 (70.0)	.20
Patients with EMG/NCS data, No.	9	1
NCS
Decreased motor amplitudes	9 (100.0)	0	.002b
Decreased motor velocities	0	0
Decreased sensory amplitudes	0	0
Decreased sensory velocities	0	0
EMG
Denervation changes	8 (88.9)	1 (100.0)	.72
No voluntary motor units	3 (42.9)	0	.41
Myopathic changes	0	0	NA

Abbreviations: AFM, acute flaccid myelitis; CDC, US Centers for Disease Control and Prevention; CSF, cerebrospinal fluid; EMG, electromyography; MRI, magnetic resonance imaging; NA, not applicable; NCS, nerve conduction studies.

SI conversion factor: To convert protein to grams per liter, multiply by 10; white blood cell count to ×10⁹/L, multiply by 0.001.

^aP values were determined from χ² for categorical variables and from t test for continuous variables.

^bSignificant P value.

Several features differed between rAFM and AFM-ad cases (Tables 2 and 3). The pattern of limb involvement proved to be a major discriminating factor, with nearly all patients with rAFM showing asymmetric onset of symptoms. By contrast, bilateral lower extremity onset was more common in AFM-ad, reflecting the pattern of symptoms often seen in other causes of myelopathy such as transverse myelitis and ischemic injury. Decreased muscle tone and reflexes at the time of presentation were common in both groups, although later evolution to increased tone or hyperreflexia was seen only in AFM-ad. All hyperacute presentations (less than 1 hour) had additional features supportive of a diagnosis of spinal cord ischemia, whereas weakness in rAFM was often acutely recognized but progressed to nadir over hours to several days. Symptoms of impaired bowel or bladder function were more common in AFM-ad. Patients with rAFM were more likely to have MRI lesions predominantly or completely restricted to spinal cord gray matter or to have involvement of the dorsal pons, and did not have supratentorial brain lesions. Cerebrospinal fluid studies were largely a poor discriminator between the groups, with the exception that patients with rAFM had lower CSF protein values than those with alternative diagnoses.

An infectious prodrome was present in 100% of patients with rAFM but only 7 patients with AFM-ad (63.6%), further highlighting plausible differences in pathogenesis. However, identification of a specific pathogen occurred only in a minority of patients with AFM, and the proportion of patients with a pathogen identified did not differ between groups. In the rAFM group, specific pathogens were identified in 13 patients (38.2%) systemically (serum, nasal, sputum, or stool) including 5 with enterovirus D68, 2 with unspecified enterovirus, 2 with rhinovirus, 2 with adenovirus (all of the aforementioned by polymerase chain reaction [PCR]), and 2 with mycoplasma (by acute and convalescent titers). In 1 patient, Epstein-Barr virus was identified in the CSF by PCR, although with negative IgM titer and positive IgG possibly reflecting reactivation rather than primary infection, and no pathogen was identified systemically. Among patients with AFM-ad, 4 (36.3%) also had a pathogen present. Three such cases had pathogens identified systemically: 1 with untyped rhinovirus/enterovirus (by PCR) and mycoplasma (by IgM titer only), 1 with rhinovirus B by PCR, and 1 with enterovirus D68 by PCR (this latter patient had a superimposed pneumonia and presumed bacterial meningitis complicated by spinal cord infarction). The fourth patient had untyped rhinovirus/enterovirus identified in CSF by PCR.

Reproducible and distinctive features of rAFM cases were identified, and these characteristics were verified for interrater reliability as described in previous paragraphs. This case description is presented in the Box.

Box. Description of Restrictively Defined AFM .

1. Prodromal fever or viral syndrome

2. Weakness in a lower motor neuron pattern involving 1 or more limbs, neck, face, and/or bulbar muscles

   1. Limb weakness should be accompanied by decreased tone and decreased or absent tendon reflexes

3. Supportive evidence including at least 1 of the following:

   1. Gray matter predominant T2-hyperintense lesion on MRI of the spinal cord, spanning multiple levels, with or without ventral nerve root enhancement

   2. Electromyography and nerve conduction study evidence of a motor neuronopathy with intact sensory nerve conductions

   3. CSF studies showing pleocytosis (white blood cell count >5 cells/μL)

4. Absence of:

   1. Objective sensory deficits on neurologic examination

   2. Supratentorial white matter or cortical lesions greater than 1 cm

   3. Encephalopathy that cannot be explained by fever, illness, respiratory distress, or metabolic abnormalities

   4. Elevation of CSF protein greater than 2 times the upper limit of normal in the absence of CSF pleocytosis

   5. Presence of a definable alternative diagnosis

Discussion

We report clinical data on 45 children from the United States and Canada presenting between 2012 and 2016 who met the CDC case definition for AFM. Among these patients, the clinical presentation of up to 1 in 4 was consistent with an alternative diagnosis, most of these being transverse myelitis and spinal cord ischemia. These results highlight that the CDC case definition, while appropriately sensitive for epidemiologic ascertainment of possible AFM cases, also encompasses other neurologic diseases that can cause acute weakness. Previously published case series have used either the CDC case definition alone^7,11,12,13 or with varying degrees of modification,^{4,8,9,10,14,15,16} indicating variable approaches to defining research populations. This raises the possibility that the populations reported in these studies may represent a heterogeneous collection of neurologic illnesses or that AFM may be co-occurring with another defined neurologic disease. More specific criteria are needed to define AFM for future rigorously defined research studies to determine etiology, biomarkers, pathogenesis, and treatment strategies for this emerging disorder.

The clinical characteristics and case description of rAFM represent 1 possible strategy to define a homogeneous research cohort that is likely to represent a common pathophysiology underlying the prototypical form of AFM. The description of rAFM is sufficiently discrete to permit accurate assignment of diagnoses by a blinded evaluator without verbal instruction as to its application. This description is internally valid within our cohort but still warrants external validation. We propose that the definition of rAFM presented here be used as a starting point for developing inclusion and exclusion criteria for future research studies of AFM.

The results of our study may also inform the clinical diagnosis of AFM. Specifically, the presence of the characteristics differentiating between rAFM and AFM-ad (Tables 2 and 3) should raise the clinical suspicion of AFM as opposed to alternative diagnoses. The clinical data described here that best discriminate rAFM from other causes of paresis are all readily available at or shortly following presentation to the hospital. Early diagnosis of AFM would facilitate initiation of respiratory monitoring, initiation of potentially effective treatments such as intravenous immunoglobulin, and avoid the use of potentially harmful approaches, such as immunosuppression, that are often used for neuroinflammatory conditions in the differential diagnosis but might exacerbate AFM.¹⁹ Timely exclusion of AFM in other patients would enable appropriate treatments to be administered that might otherwise have been avoided or delayed. Prompt recognition may also encourage more vigorous investigation for a specific pathogen, whose presence is often transient and requires testing from nonneurologic sites that might otherwise not be considered.²⁰ Early diagnosis may also provide prognostic information to help guide treatment and rehabilitation decisions for patients and their families.

However, we emphasize that the rAFM case description presented here is intentionally designed with high specificity as a goal and acknowledge that sensitivity will be less than that of the CDC case definition. The description of rAFM will therefore be most useful for the strict definition of AFM in a research population where homogeneity of cases may aid our understanding of disease process. In the clinical setting, stringent application of exclusion criteria may prove problematic in a small number of cases. For example, some reports of AFM during the 2018 outbreak associated with enterovirus A71 have presented with overlapping features of meningoencephalitis.²¹ We suggest that the presence of putative exclusion criteria in a patient should prompt careful evaluation of alternative diagnoses rather than absolute disqualification of a diagnosis of AFM. We also acknowledge that AFM is still not well understood and this definition may change as we better understand the spectrum of disease from the growing number of children diagnosed with AFM.

Limitations

A major limitation of our study is the inability to independently validate the use of the proposed description of rAFM. Such validation would require a separate and prospectively defined cohort of children presenting with acute weakness. We therefore present this description as provisional and expect that an iterative process of refinements can better serve the research needs of the AFM community. The ongoing 2018 outbreak of AFM provides the opportunity to validate and refine these criteria, which will be the focus of future work.

Conclusions

Acute flaccid myelitis now appears to have established a biennial pattern of recurrence with seasonality, and further outbreaks should be anticipated in 2020 and beyond. Clinicians should be prepared to recognize cases presenting in the late summer and fall, often following a viral respiratory infection, and presenting with asymmetric flaccid weakness that progresses over hours to days, associated with characteristic MRI and CSF findings. Much work is needed to better diagnose, treat, and prevent future cases of AFM. We have demonstrated here the need for stringent inclusion criteria to support future research efforts and propose a provisional case description to meet this need.