Pediatric spinal cord stroke: clinical presentation, MRI features, and suspected mechanisms

Abstract

Background and Objectives:

Spinal cord stroke (SCS) is an underrecognized cause of severe acute myelopathy in children that can be misdiagnosed as an infectious or inflammatory process. Limited characterization of clinical and neuroimaging features of pediatric SCS impede timely and accurate diagnoses. Our objectives were to (1) identify clinical and radiological features of SCS to prompt timely evaluation and accurate diagnosis and (2) gain insight into mechanisms of SCS.

Methods:

We conducted a retrospective case analysis of children (<18-years-old) diagnosed with SCS at a specialized referral institution between 2010 and 2025. Clinical records, imaging, and laboratory data were reviewed to identify key features of pediatric SCS.

Results:

Among the 56 included patients, 28 were male (50%), and the age distribution was bimodal, with peaks at one and 14 years. Pediatric SCS presented with hyperacute onset of motor, sensory, bladder/bowel, and pain symptoms. The lesions were predominantly anterocentral, longitudinally extensive, with the cervical spinal cord being most often affected. Over time from the onset of symptoms, MRI diffusion restriction decreased while gadolinium enhancement increased. CSF profile was non-inflammatory in 95% of cases. Suspected mechanisms were identified in 63% of cases, with vascular compression being the most commonly identified mechanism in 17 subjects (30%), while 38% were idiopathic. Chiari I malformations were identified as the main suspected cause of vascular compression, occurring in a subgroup of young patients with cervical SCS (n=8, 14%). Serum studies identified hypercoagulability in 38% of 40 subjects evaluated.

Discussion:

Recognition of specific clinical and neuroimaging profiles in pediatric patients with SCS may facilitate earlier identification and intervention. We propose a systematic evaluation of clinical history, with particular attention to the temporal course from symptom onset to nadir, early acquisition of spine MRIs with DWI sequences, serologic evaluations for hypercoagulability, and CSF studies to exclude inflammatory etiologies. This approach will provide a framework for continued efforts to better understand the mechanisms underlying pediatric SCS and develop prevention and treatment strategies.

Introduction

Spinal cord stroke (SCS) is a rare but consequential cause of acute myelopathy. Estimates suggest SCS may account for 6 to 15% of all myelopathies and 1.2% of all strokes in the whole population^1,2. In a specialized myelopathy referral center, approximately one-third of patients of all ages referred for “transverse myelitis” have noninflammatory etiologies of myelopathy, in which SCS accounts for nearly half of those cases³. Although no epidemiological data are available for the pediatric population, our data suggested that SCS may account for approximately 25% of myelopathies in children³. Thus, we suspect SCS may be underrecognized in children as with adults, leading to misdiagnosis and suboptimal management^1,4.

There are several challenges in diagnosing SCS in children. First, a detailed understanding of the natural history and pathogenesis of pediatric SCS is lacking, as prior reports have been limited to series of 13 or fewer patients^5–9. Second, the diagnosis of SCS often depends on recognizing lesion patterns or the acquisition of diffusion-weighted imaging (DWI) on spine magnetic resonance imaging (MRI)^1,10. However, MRI findings in children are poorly described, and DWI sequences are not standard in routine spine MRI protocols. Even when appropriate spine MRIs are obtained, resolution may be limited, and the likelihood of artifacts is high¹¹.

The mechanisms of SCS in children outside a periprocedural context are unclear². Traditional vascular risk factors for cerebral stroke, such as hypertension, hyperlipidemia, and diabetes are uncommon in children and identified only in a minority of SCS patients of any age^1,10. On the other hand, inherited hypercoagulability has been associated with increased risk for cerebral stroke in children¹². In a meta-analysis, the highest odds ratios were observed for protein C deficiency, antiphospholipid antibodies, lipoprotein(a), and genetic variants in Factor V, Factor II, and methylenetetrahydrofolate reductase (MTHFR)¹². Additionally, some causes of SCS may result from disturbances in the spinal cord’s unique vascular anatomy¹³. We sought to characterize the clinical, laboratory, and imaging features of SCS in a large cohort of pediatric cases to improve diagnostic accuracy and elucidate potential mechanisms.

Methods

Clinical Characterization

We conducted a retrospective chart review of 56 pediatric patients (<18-years-old) diagnosed with SCS, who presented to the Johns Hopkins Hospital or Kennedy Krieger Institute between 2010 and 2025. Cases were identified from a research registry for patients with myelopathies. Patients were evaluated at the time of acute presentation or had details of the initial presentation confirmed during a neurologist's interview at the clinical visit.

We included patients who met the following inclusion criteria: (1) presented with clinical myelopathy, defined as the onset of motor, sensory, bladder, bowel, or autonomic symptoms localizable to the spinal cord based on clinical history, physical, and neurological evaluation as previously outlined¹⁴; (2) hyperintense lesions on T2-weighted MRI sequences within a vascular territory or watershed area that explained the symptoms, per our review or per radiology report in cases where we did not have MRI images available for review; and (3) other etiologies excluded by the available clinical, imaging, and laboratory methods obtained for clinical care as previously outlined^3,14 or more consistent with alternative diagnoses per prior consensus diagnostic criteria^15–17. SCS diagnostic criteria¹ were applied, and certainty of SCS was considered “definite” if a vascular abnormality was demonstrated by conventional angiography; “probable” if there was diffusion restriction or an explanatory mechanism; or “possible” if none of these conditions were met (eTable 1).

The temporal profile from symptom onset to nadir was classified as hyperacute (<6 hours), acute (6–48 hours), subacute (>48 hours–21 days), or chronic (>21 days). Nadir was defined as the point of worst neurologic function, before improvement or plateau, based on history and neurologic examination. The degree of disability at nadir was measured using the modified Rankin Scale for Children (mRS). Gait, micturition, and defecation disturbance were measured with the modified Aminoff-Longe scale (eTable 2)¹⁸.

MRI Studies

We analyzed clinically acquired spine MRIs classified into four time intervals from symptom onset to image acquisition: <2 days, 2–10 days, 11–30 days, and >30 days, corresponding to the hyperacute, acute, subacute, and chronic phases of MRI findings in cerebral strokes¹⁹. If available, one MRI per period (the earliest MRI demonstrating a lesion if multiple were available) was included for each subject. All MRIs were reviewed by a physician with experience in SCS imaging (DA-G) and validated by a neuroradiologist (DL). T1-weighted, T2-weighted, T1-weighted with gadolinium (T1W-Gad), and DWI sequences were reviewed. Lesion location was recorded and categorized at each vertebral level in the earliest available MRI with a lesion on T2-weighted sequences.

Lesions were considered longitudinally extensive if they involved 3 or more adjacent vertebral levels. The axial distribution was recorded as predominantly involving one or more of the following compartments: anterior (ventral horns and white matter), posterior (dorsal horns or white matter), lateral (lateral white matter), or central (commissure and surrounding white matter). SCS was considered multifocal if there were two or more discrete, noncontiguous lesions. Imaging signs described in SCS were recorded as present or absent at any level. The presence of intra-axial parenchymal or nerve root enhancement on T1W-Gad sequences was recorded. Diffusion restriction was defined as hyperintensity on DWI with matching hypointensity on the apparent diffusion coefficient (ADC). For the topographical analysis, the earliest available MRI demonstrating the lesion on T2-weighted sequences was used.

Imaging signs classically described in SCS were characterized, including bilateral anterior gray matter hyperintensity (owl’s eyes sign), complete gray matter involvement (H sign or butterfly sign), anteromedial white matter hyperintensity, continuous hyperintensity of the anterior horns and central gray matter forming a U or V shape (anterior U/V lesion), longitudinal pencil-like signal hyperintensity change in the sagittal plane, and signal intensity change involving the complete transverse section^9,23. Regarding spinal cord volumetric appearance, tumefaction or atrophy (increased or decreased volume in the sagittal or axial plane, respectively) and cystic myelomalacia (defined as T2-weighted bright hyperintensity with T1-weighted hypointensity) were recorded. Vertebral bone infarction was defined as T2-weighted hyperintensity with associated T1-weighted heterogeneous hypointensity in a vertebral body, including or adjacent to the compromised vascular territory; when unclear, the presence of gadolinium enhancement and changes over time on imaging were used to support this finding.

Laboratory Evaluation

Hypercoagulability studies used as part of stroke risk factor work-up¹² were recorded. Cerebrospinal fluid (CSF) analysis performed within 10 days of symptom onset was recorded. Pleocytosis was defined as WBC >5 cells/mm³, protein elevation as protein >45 mg/dL, and positive oligoclonal bands as >2 bands in the CSF but not in the serum.

Explanatory SCS Mechanism Analysis

Using the clinical history and presentation, neuroimaging data, laboratory workup, and the neurologist’s initial evaluation and clinical impression, we classified the suspected mechanisms of SCS for each patient into the six following categories: vascular compression, hypercoagulability, hypoperfusion, fibrocartilaginous embolism (FCE), arterial dissection or disease, and unknown. Definitions and examples for each mechanistic category are supplied in eTable3. In cases where more than one mechanism was suspected, we classified based on the mechanism considered most likely to be the primary mechanism of SCS.

Data Analysis

Quantitative data are described with median and interquartile ranges, and qualitative data with frequencies. Where data were unavailable for all patients in our cohort, the denominator indicates the number available for analysis. All analyses were performed using R (v4.5.2; R Core Team) and Prism (v10.6.1, GraphPad Software).

Standard protocol approvals, registrations, and patient consents

The Johns Hopkins Medicine Institutional Review Board approved the study and waived patient consent (IRB00115274).

Data availability statement

Data will be made available by request to corresponding authors upon publication. Researchers must provide a methodologically sound proposal with approval from an independent or institutional review board and sign a data access agreement.

Results

Clinical presentations

Fifty-six pediatric patients (50% male) diagnosed with SCS were included. Clinical characteristics of the cohort, including demographics and symptoms at presentation, are outlined in Table 1. All patients were previously healthy. The median age at symptom onset was 12-years-old, but a bimodal distribution was observed, with peaks at 1-year and 14-years (eFigure 1). Five patients (9%) were categorized as definite SCS, 37 (66%) as probable SCS, and 14 (25%) as possible SCS. Practical constrains during the acute-phase diagnostic evaluation limited the extent of the initial assessment, which may have reduced diagnostic certainty and contributed to the inability to classify additional cases as definite or probable SCS. Only 14 patients (25%) underwent angiography, and while 55 patients (98%) had spine MRI performed within a month of symptom onset, only 27 patients (48%) had DWI sequences included in their early spine MRIs.

Table 1.

Demographic and clinical characteristics

	Subjects (n=56)
Male (%)	28 (50%)
Median age, year (Q1-Q3)	12 (6–14)
Race (%)
White	37 (66%)
Asian	7 (13%)
Black/African American	5 (9%)
Other	7 (13%)
Ethnicity (%)
Hispanic/Latino	5 (9%)
Spinal cord stroke diagnostic criteria (%)
Definite	5 (9%)
Probable	37 (66%)
Possible	14 (25%)
Angiogram obtained (%)	14 (25%)
Spine MRI obtained within 30 days of symptom onset (%)	55 (98%)
Spine MRI included DWI/ADC sequences	27 (48%)
Temporal profile of symptom onset to nadir (%)
Hyperacute (< 6h)	46 (82%)
Acute (6–48h)	9 (16%)
Unknown	1 (2%)
Presenting Symptoms / Total with available data (%)
Motor symptoms	56/56 (100%)
Sensory symptoms	47/49 (96%)
Bowel/Bladder symptoms	44/54 (81%)
Acute back, neck, or perineal pain	30/41 (73%)
Cranial nerve dysfunction	14/56 (25%)
Cognitive symptoms or altered mental status	5/52 (10%)
Visual symptoms	1/44 (2%)
All limbs affected	24/56 (43%)
Severity of symptoms at nadir, Median (Q1-Q3), Total with available data
Pediatric Modified Rankin Score	4 (4–5), n = 56
Gait disturbance	5 (4–5), n = 55
Micturition dysfunction	3 (2–3), n = 50
Defecatory dysfunction	2 (1–3), n = 41
Intensive care needs at presentation / Total with available data (%)
PICU admission	27/47 (57%)
Respiratory failure needing intubation	14/47 (30%)
Respiratory failure needing tracheostomy	9/14 (64%)

Most patients presented with a combination of motor, sensory, and bladder/bowel symptoms (Table 1). Acute back, neck, or perineal pain was a frequent complaint at presentation (n=30/41, 73%). Cranial nerve, cognitive, mental status, were less common, though observed in patients with simultaneous cerebral stroke (n=3), and patients with respiratory failure (n=14). Visual symptoms were described in one patient who complained of unilateral blurry vision without clear evidence of impairment in the visual pathway. The temporal profile from symptom onset to nadir was largely hyperacute (n=46/56, 82%). At symptom nadir, patients exhibited moderate to severe disability, with a median mRS score of 4, severe gait disturbance with inability to stand or walk, micturition dysfunction with total urinary incontinence or retention, and defecatory dysfunction with occasional incontinence or severe constipation. Approximately half of the patients for whom we have admission data (n=27/47, 57%) were initially admitted to the Pediatric Intensive Care Unit, and 30% experienced respiratory failure requiring intubation (n=14/47), with 9 of those ultimately receiving a tracheostomy.

MRI Characteristics

Spine MRI studies were performed for all patients. Image files were available for our review for 50 patients (89%). The remaining six cases had radiology reports consistent with SCS, but the image files were not available for our review (n=5) or had poor imaging quality (n=1), so they were excluded from the imaging analysis. The timing and number of spine MRIs per patient were highly variable. We analyzed up to 4 studies per patient at the indicated time intervals, for a total of 114 studies (38 at <2 days, 26 at 2–10 days, 14 at 11–30 days, and 36 at >30 days). Longitudinal lesion topology showed a bimodal distribution with the lower cervical and lower thoracic levels being more frequently affected (Figure 1A). Lesions were predominantly anterocentral, longitudinally extensive, and monofocal (Figure 1B–C, eFigure 2, eTable 4).

Figure 1. Topography of MRI lesions in pediatric spinal cord stroke.

A. Distribution of SCS T2-weighted lesions on sagittal spinal levels. B. Axial areas of SCS T2-weighted lesions. C. Sagittal length of SCS T2-weighted lesions at different timepoints from symptom onset. D. Axial T2-weighted lesion area at different timepoints from symptom onset. N=50 patients with SCS where MRI was available for review. Ant = Anterior, Cent = Central, Lat = Lateral, Post = Posterior.

Several patterns emerged from temporal MRI analysis (Figure 2). First, diffusion restriction was present in most MRI studies obtained within 2 days of symptoms and decreased over time (Figure 2B, eTable 4). Six of the 27 patients who underwent diffusion imaging had two MRI time points. Notably, one patient had a diffusion-negative lesion at day 1 after symptom onset, which became diffusion-positive at day 5. Second, gadolinium enhancement increased during the first 30 days after symptom onset, reaching 69% of studies (n=9/13), before subsequently declining (Figure 2C, eTable 4). Among the 36 patients who underwent contrast-enhanced MRI at two timepoints, 10 demonstrated non-enhancing lesions on the initial study obtained at <2 days of symptom onset that later became enhancing. Third, T2-weighted axial lesion extent peaked at 2–10 days, with 56% (n=14/25) of MRI studies showing lesions involving the entire cord in the axial plane, whereas the remainder involved only a partial cross-section extent of the cord (Figure 1D). Myelomalacia was present in most MRI studies obtained >30 days after symptom onset (Figure 2D, eTable 4). Finally, we assessed the frequency of MRI signs previously associated with SCS, including pencil-like lesions, owl’s eyes sign, H-sign, anteromedial white matter lesions, and vertebral body infarcts (eFigure 3, eTable 4). The most prevalent finding was owl’s eyes sign, which was observed in 50% (7/14) of MRI studies performed 11–30 days after symptom onset. Overall, the prevalence of each sign varied over time, and no single sign was present in the majority of spine MRIs.

Figure 2. Evolution of MRI findings in pediatric spinal cord stroke.

A. Representative images displaying the evolution of SCS lesions on MRI over time for an individual patient. B-D. MRI characteristics of SCS lesions at different timepoints from symptom onset for cohort of SCS patients where MRI was available for review, n=50. B. Proportion of MRIs with diffusing restricting lesions. C. Proportion of MRIs with GAD-enhancing lesions. D. Proportion of MRIs with myelomalacia. (B-D) Corresponding data can be found in eTable 4.

Laboratory studies

Forty patients (71%) underwent evaluation for hypercoagulability, of whom 15 (38%) had abnormal findings, including positive antiphospholipid antibodies, elevated lipoprotein(a), elevated homocysteine, reduced Protein C/S levels, elevated fibrinogen, and positive Factor V Leiden (Table 2). In eight cases, hypercoagulability was considered to be the primary explanatory mechanism for SCS. In five cases the laboratory abnormalities were not deemed to confer a clinically meaningful increase in vascular risk, including heterozygous status for MTHFR C677T, quantitative protein C or S deficiency with normal function, or fibrinogen only slightly above the normal range without other findings of a hypercoagulable state. In two cases, alternative mechanisms, specifically FCE and vascular compression, were considered the primary cause. CSF studies obtained within 10 days of symptom onset were available for 43 patients (77%) and were largely noninflammatory (Table 2). When present, pleocytosis or protein elevation were minimal, with the exception of one case with no pleocytosis but elevated protein of 159 mg/dL.

Table 2.

Laboratory Findings

Hypercoagulability Studies	Abnormal/Total Tested (%)	Difference from normal limit, Median (Range)
Any abnormal lab testing	15/40 (38%)
Antiphospholipid antibodies	5/34
Lipoprotein A	4/27	98.5 (79–493) nmol/L
Homocysteine	2/31	2.75 (1.6–3.9) μmol/L
Protein C or S	3/29	−20 (4–36) %
Fibrinogen	3/16	93 (15–106) mg/dL
Pathogenic Genetic Variants
Factor V Leiden	2/24
MTHFR C677T	7/12
PT G20210A	0/15
CSF Studies (Criteria for abnormal)	Abnormal/Total Tested (%)	Abnormal values, Range
Pleocytosis (WBC > 5 cells/mm3)	2/43 (5%)	8–14 cells/mm3
Protein > 45 mg/dL	7/42 (17%)	47–159 mg/dL
CSF specific oligoclonal bands (>2 bands)	2/35

SCS mechanisms

We identified suspected mechanisms of SCS in 35 patients (63%, Table 3). The most common mechanisms were vascular compression (n=17), hypercoagulability (n=8), and hypoperfusion (n=4). Together, these three etiologies accounted for approximately 83% of cases with a suspected mechanism. Among cases attributed to vascular compression, some appeared secondary to structural abnormalities, including Chiari malformations, basilar kyphosis, or disc herniation in the context of connective tissue disease, whereas others appeared to be dynamically functionally induced, likely precipitated by trauma or prolonged cervical hyperextension (eTable3). Different mechanisms were associated with distinct cord lesion topographies. Vascular compression, particularly in the setting of structural abnormalities, as well as arterial dissection or stenosis commonly involved the upper spinal cord, whereas hypercoagulability, hypoperfusion, and FCE commonly involved the lower spinal cord. Three patients also had concurrent cerebral infarctions: one attributed to arterial dissection, one to hypoperfusion, and one of unknown etiology.

Table 3.

Mechanism of Spinal Cord Stroke and Lesion Distribution

Primary Etiology	n (%) Total n=56	Age of Onset Median (Q1-Q3), yrs	Avail. MRIs, n	Longitudinal Distribution, n			Axial Distribution, n
				C	T	L	Ant	Cent	Lat	Post
Vascular compression	17 (30%)	11 (1–14)	15	11	6	0	12	11	5	6
Hyper-coagulability	8 (14%)	13 (11–14)	8	3	6	0	8	7	3	6
Hypoperfusion	4 (7%)	6 (2–12)	4	2	3	1	2	3	3	3
Fibrocartilaginous embolism	3 (5%)	14 (14–15)	3	0	3	2	2	3	2	3
Arterial Dissection or Stenosis	3 (5%)	13 (9–15)	3	2	1	0	2	2	0	1
Unknown	21 (38%)	12 (10–14)	18	9	14	4	15	12	8	7

C = Cervical, T = Thoracic, L = Lumbar

Ant = Anterior, Cent = Central, Lat = Lateral, Post = Posterior

Avail.= Available, MRI= Magnetic Resonance Imaging

SCS associated with Chiari I malformation

Notably, 8 patients (14%) in our cohort were found to have associated Chiari I malformation at the time of SCS presentation (Figure 3, eTable 5). This subgroup presented at a mean age of approximately 1-year-old and had greater symptom severity than the cohort overall (Figure 4A, eTable 5). All 8 patients presented with quadriparesis, and 6 required intensive care management. On MRI, all lesions affected the upper cervical cord, with no consistent axial predominance, although the posterior cord was frequently affected bilaterally at the superior aspect of the lesion (Figure 4B–C). The suspected SCS mechanism was vascular compression in 6/8 cases, based on imaging evidence of compression, and remained unknown in two cases. One patient underwent posterior fossa decompression because of severe compression of the upper cervical cord by the cerebellar tonsils and progressive clinical deterioration with neurogenic shock.

Figure 3. Illustrative cases of spinal cord strokes associated with Chiari I malformations.

Examples of individual patient MRIs with SCS associated with Chiari I malformations. A. Case of SCS with associated syringomyelia. MRI obtained 11 days after symptoms onset shows a T2-weighted hyperintensity involving mainly the posterior and central compartments (Aa) and GAD enhancement (Ab) in the anterior horns resembling the Owl’s eyes sign. B. MRI showing lesion extending from the cervicomedullary junction to the mid-thoracic cord (Ba, caudal end not shown), involving the posterior cord bilaterally in its upper portion. A follow-up study demonstrates a vessel coursing between the cord and the herniated cerebellar tonsils (Bb, arrowhead). C. Tumefactive lesion and herniated cerebellar tonsils abutting the cord at the foramen magnum (Ca). Due to clinical deterioration, foramen magnum decompression was performed. Follow-up MRI three days later shows a decrease in edema and lesion extension, with residual hyperintensity in the posterior compartment (Cb).

Figure 4. Spinal cord strokes associated with Chiari I malformations.

Comparison of subgroup with Chiari I malformation (blue) relative to the whole cohort (gray) A. Age of onset. B. Sagittal distribution of SCS T2-weighted lesions on spinal levels. C. Axial distribution of SCS T2-weighted lesions.

Discussion

We characterized 56 children diagnosed with SCS, the largest cohort to date. Our goals were to (1) identify clinical and radiological features of SCS to prompt timely evaluation and accurate diagnosis, and (2) gain insight into mechanisms of SCS.

Pediatric SCS presents as hyperacute onset of severe motor, sensory, bladder/bowel, and pain symptoms

Clinical presentation most commonly included a combination of motor, sensory, and bladder/bowel symptoms. Motor symptoms were present in all patients. Acute back, neck, or perineal pain was also a common presenting symptom. The frequency of pain was similar to that reported in adults, and it often followed a radicular distribution that may reflect ischemic injury of the sensory roots or dorsal root or disturbances in the sensory pathways integration system, dorsal horns and sensory afferents, in the segments of the spinal cord affected ganglia²⁰. Symptom onset was typically hyperacute, with a nadir reached within 6 hours, although symptoms were not always maximal at onset, as is often seen in cerebral stroke. This clinical pattern is consistent with the expected pathophysiology of SCS and with prior reports in both pediatric and adult populations^1,3,5,6,10. At nadir, most patients were unable to ambulate and were incontinent, requiring substantial care. Accordingly, half were initially admitted to intensive care units, with a considerable proportion experiencing respiratory compromise.

Timing of MRI matters: diffusion restriction decreases while GAD enhancement increases over days from symptom onset

Our findings support early acquisition of DWI sequences, as diffusion restriction appears to diminish over time, particularly beyond 10 days after symptom onset²¹. Because DWI is not routinely included in standard spine MRI protocols, early clinical recognition of SCS is essential to ensure appropriate imaging. Consistent with previous reports, we identified one patient with serial MRI studies who demonstrated conversion from absent to present diffusion restriction over several days^6,10. Accordingly, when clinical suspicion is high but initial imaging is diffusion negative, repeat DWI within a short interval may be warranted. In contrast, gadolinium enhancement of cord lesions increased with time from symptom onset, up to 30 days. Prior studies have likewise reported enhancement in a substantial proportion of SCS lesions^1,10, underscoring that enhancement alone is not sufficient to distinguish inflammatory from vascular etiologies.

In contrast, the utility of spinal angiograms for diagnosing SCS in children was less clear, providing a definitive SCS diagnosis in 5 of 14 cases where it was obtained. Practical considerations hinder our ability to consistently obtain angiograms in a timely manner, particularly for pediatric patients, as the test requires both a well-trained specialist and a sufficiently stable patient to perform. Delay in obtaining angiograms may obscure subtle signs of dissection, particularly in children, who heal quickly. Moreover, for cervical lesions, angiography visualizing only vertebral arteries may miss findings from the subclavian artery, typically the main blood supplier for the cervical cord²².

Lesions were mostly anterocentral, longitudinally extensive, and cervical

Topographically, lesions on T2-weighted imaging were more anterocentral, though they could span the entire cord axially, particularly when imaged 2–10 days after onset. Lesions were also largely longitudinally extensive during the first 30 days after symptom onset. Beyond 30 days, lesions became smaller and were frequently associated with myelomalacia. These findings are consistent with prior reports of SCS in both children and adults and likely reflect the time course of vasogenic edema associated with the infarct, which resolves over time^1,3,5,6,10.Interestingly, we found that the cervical spinal cord was most frequently affected, with a second peak in the thoracolumbar region. This differs from the findings of Bar et al.⁵, who reported the thoracolumbar region as the most commonly affected in their 7 pediatric SCS patients and 50 additional literature reports, although their cohort included fewer very young patients. By contrast, the cervical region was involved in 3 out of 6 pediatric SCS cases reported by Elgehiny et al.⁶, 8 out of 13 cases reported by Ben-Artzi Cohen et al.⁹, and other case reports^7,8. In our cohort, vascular compression, particularly with Chiari malformations in very young patients, comprised the majority of the group with cervical cord involvement and may account for some of the differences between cohorts. The second peak in the thoracolumbar region may reflect anatomical vulnerability as the artery of Adamkiewicz is the principal arterial supply to the anterior spinal artery in the lower thoracic and lumbar cord, potentially predisposing this region to SCS²³. Studies of SCS in adults likewise report predominantly thoracic cord involvement, although this is largely associated with aortic atherosclerosis, which is rare in children¹⁰. Furthermore, previously reported MRI signs associated with SCS in adults, including pencil-like lesions, owl’s eyes, H-sign, and anteromedial white matter lesions, were not pervasive in our cohort^5,10. Overall, these data highlight the importance of recognizing that the mechanisms and imaging patterns of SCS may differ substantially between children and adults across the lifespan.

Laboratory studies suggest hypercoagulability in serum and noninflammatory CSF

A hypercoagulability workup was available for a subset of our cohort and revealed abnormalities in 38% (n=15/40), of whom 8 were considered to have clinically significant findings. The most common abnormality were antiphospholipid antibodies and lipoprotein(a) elevation. Bar et al.⁵ reported abnormalities in 21% of their cases, whereas other studies have reported no abnormalities in their patients^6,9. Additional studies are therefore needed to elucidate the relevance of hypercoagulability as a risk factor for pediatric SCS. CSF profiles were largely noninflammatory, suggesting that the absence of CSF inflammation should prompt consideration of vascular etiologies such as SCS. In rare cases, we observed mild pleocytosis or increased protein in the CSF, at a rate consistent with prior reports in ischemic strokes²⁴.

Vascular compression was the most suspected mechanism

The most frequently suspected mechanism of SCS in our cohort was vascular compression (30%, n=17), commonly due to disc herniation indenting the anterior surface of the cord or occluding the intervertebral foramen, which involves the anterior spinal artery or its feeding radiculomedullary arteries, respectively¹³. Although frequently proposed as a common cause of SCS²⁵, we suspected fibrocartilaginous embolism in only 5% of our cases (n=3), consistent with prior estimates²⁶. Of note, we did not identify a mechanism for SCS in 38% of our cases.

Chiari I malformation was found in a subgroup of pediatric SCS patients.

Because vascular compression was the most frequently suspected mechanism, we further evaluated these cases and identified a subgroup of eight cases associated with Chiari I malformation. Interestingly, this subgroup corresponded to the first peak in our age distribution, which was characterized by a younger age of onset at or before one-year of age, with only one patient aged 14-years. These patients were all previously healthy and presented with a hyperacute onset of flaccid quadriplegia, resulting in severe presentations (median mRS=5). Although only a few cases have been reported in the literature, the clinical findings in these cases are remarkably consistent²⁷.

In all eight cases, the lesions involved the cervical spinal cord. In six patients, we suspected vascular compression of the spinal arteries by the herniated cerebellar tonsils at the level of the vertebral arteries as the underlying mechanism ²⁷. In contrast, the remaining two patients showed no neuroimaging evidence of cord compression. Anatomically, the posterior spinal arteries arise from descending branches of the distal vertebral arteries near the foramen magnum²⁸. Greater cervical spinal laxity at younger ages, coupled with vascular compression by ectopic cerebellar tonsils, may increase susceptibility to ischemic injury⁸. The predominance of posterior cord involvement in the upper portion of the lesions is consistent with the expected topography of ischemia resulting from compression of this vascular territory²⁹. Although the risk conferred by Chiari I malformation for pediatric SCS is difficult to estimate, the frequency of Chiari I malformation in our cohort (14%) is likely higher than the estimated prevalence in children (~4%)³⁰. Larger studies are therefore needed to better define both the associated risk and the potential utility of operative decompression in these cases.

Limitations

Our study has several limitations. First, this was a retrospective, single-institution study. Because patients presented either acutely to our emergency department or later to our outpatient clinics for follow-up after the initial event, the diagnostic workup and timing of investigations varied substantially across patients. As a result, we lacked sufficient data to meaningfully assess the contribution of cardiac evaluations, including electrocardiography and echocardiography, as well as the evaluation of genetic risk factors beyond the hypercoagulability studies described above. In addition, patients received heterogeneous treatments, including steroids, IVIg, plasma exchange, and aspirin, and rehabilitation intensity also varied. These differences limited our ability to interpret outcomes, as inconsistent follow-up timepoints precluded meaningful quantification of recovery trajectories or treatment effects. Anecdotally, however, we observed gradual improvement over time regardless of treatment, with substantial variability in both the pace and degree of recovery, consistent with prior reports ^5,6.

Conclusions

Given the rarity of pediatric SCS, prospective multicenter natural history studies are needed to support standardized acute and longitudinal evaluations. Based on our data, we recommend that clinicians caring for children presenting with acute myelopathy obtain a detailed history, with particular attention to the temporal course from symptom onset to nadir, as well as early spine MRI including diffusion-weighted imaging (DWI) sequences, serologic testing for hypercoagulability to identify potential underlying causes, and CSF studies to exclude inflammatory etiologies. These evaluations may provide greater insight into the mechanisms of pediatric SCS and inform strategies for prevention. In addition, rehabilitation outcome measures should be assessed at regular intervals after SCS to better characterize the recovery trajectory and improve counseling for patients and families.

Take-home points:

• Spinal cord stroke (SCS) is an underrecognized cause of severe acute myelopathy in children presenting with hyperacute (≤6 hours) motor, sensory, and bladder/bowel symptoms, often with pain and severe disability at nadir.

• MRI timing impacts sensitivity of findings: diffusion restriction is most reliable early and resolves by ~10 days, while gadolinium enhancement increases over the first 30 days. Spine MRI with DWI should be obtained early and repeated if initial DWI is negative but suspicion is high.

• Hypercoagulability workup is abnormal in ~38% of tested patients, supporting serologic thrombophilia testing as part of the evaluation for SCS. Conversely, CSF is typically noninflammatory in SCS, so a normal CSF profile in a child with acute myelopathy should raise suspicion for a vascular rather than inflammatory cause.

• Suspected mechanisms were identified in about 2/3 of cases, most often vascular compression. In a distinct subset of young children with cervical lesions, Chiari I malformation was thought to have caused vascular compression. No mechanism was identified in about 1/3 of cases.