ÌÇÐÄvlog

Key Points

QuestionÌý What is the incidence and natural history of arterial ischemic stroke due to large vessel occlusion in the pediatric population?

FindingsÌý In this cohort study that included 166 admissions for pediatric arterial ischemic stroke, 39 (23.5%) had a large vessel occlusion. The incidence of pediatric large vessel occlusion stroke was 0.24 per 100 000 children per year, and patients with large vessel occlusion who were treated conservatively had poor neurological outcomes, with 19 (73.1%) experiencing moderate to severe disability or death at 3 months.

MeaningÌý In this study, the natural history of large vessel occlusion stroke in children was poor when treated conservatively, contrary to historic beliefs.

ImportanceÌý The incidence and natural history of large vessel occlusion (LVO) stroke in children is largely unknown. These knowledge gaps limit the uptake of reperfusion therapies and reduce the efficiency of pediatric acute stroke pathways.

ObjectiveÌý To determine the incidence and natural history of pediatric LVO stroke.

Design, Setting, and ParticipantsÌý This retrospective population-based cohort study was conducted between January 2010 and December 2019, with a mean (SD) follow-up of 37.0 (28.8) months. Admissions from all pediatric hospitals in the state of New South Wales, Australia, with a final diagnosis of arterial ischemic stroke (AIS) in patients 1 month to younger than 17 years were included. A total of 85 of 251 identified cases were excluded based on selection criteria. Data were analyzed from July 2020 to June 2021.

ExposuresÌý One-third of patients with LVO received mechanical thrombectomy with or without intravenous thrombolysis while the remainder were treated conservatively.

Main Outcomes and MeasuresÌý The primary outcome was the pediatric modified Rankin Scale (ped-mRS) score 3 months after stroke. Ordinal logistic regression was used to compare non-LVO, LVO without thrombectomy, and LVO with thrombectomy groups.

ResultsÌý Of 161 included patients, 56 (34.8%) were female, and the mean (SD) age was 6.1 (5.4) years. A total of 166 AIS admissions were studied, and clinical follow-up was available for 164 of 166 admissions. LVO was present in 39 admissions (23.5%). The incidence of LVO stroke was 0.24 per 100 000 patients per year (95% CI, 0.13-0.35). Patients with LVO who did not receive thrombectomy (n = 26) had poor neurological outcomes, with 19 (73.1%) experiencing moderate to severe disability or death (ped-mRS score of 3 to 6) at 3 months (6 of 12 patients receiving thrombectomy [50.0%]; 25 of 38 patients with LVO [65.8%]). Patients with LVO without thrombectomy had significantly worse clinical outcomes than patients with non-LVO at 3 months (odds ratio, 3.64; 95% CI, 1.68-7.87; P = .001). Most patients with LVO presented within time windows suitable for thrombectomy (27 of 39 [69.2%] within 6 hours; 35 of 39 [89.7%] within 24 hours).

Conclusions and RelevanceÌý In this population-based cohort study, the natural history of pediatric patients with LVO stroke treated conservatively was poor, with most experiencing lifelong disability or death. Nearly 90% of pediatric patients with LVO presented within time windows suitable for thrombectomy.

Large vessel occlusion (LVO) causes approximately one-quarter of arterial ischemic stroke (AIS) presentations in adults¹ and has worse long-term outcomes than non-LVO stroke.² Five randomized clinical trials showed mechanical thrombectomy within 6 to 8 hours since last seen well improved outcome compared with standard medical therapy in those with anterior circulation LVO.^3-9 The time window of benefit was extended to 16 and then 24 hours by 2 additional trials in selected adult patients.^10,11

Children and adolescents were not included in those trials, and to our knowledge, there are no randomized clinical trials for mechanical thrombectomy or any acute therapies for pediatric AIS. Evidence for thrombectomy in children is predominantly from studies that are retrospective and single arm and have small sample sizes.¹² A recent larger retrospective multicenter cohort study assessing outcomes following thrombectomy in 73 children reported favorable long-term outcomes, with a median modified Rankin Scale (mRS) score of 1.0.¹³ Current guidelines for treatment of LVO stroke in children make only cautious recommendations on its use,^14,15 although recent literature suggests increasing uptake of thrombectomy.^12,13

Absence of population data regarding the natural history of LVO stroke in children has hindered confident interpretation of the existing pediatric data. Randomizing pediatric patients to noninterventional treatment would be unethical given the benefit in adults (number needed to treat of 2.6).⁸ In addition, such a trial would likely not be feasible, as observed in attempts to undertake a trial for intravenous (IV) thrombolysis in children.¹⁶ Further limitations on the uptake of reperfusion therapies include absence of LVO stroke incidence data in children,¹² slow triage and imaging pathways,¹⁷ and historic beliefs that children with LVO present too late for reperfusion and have better outcomes than adults when treated conservatively.^18,19 Addressing these evidence gaps may support resource allocation and facilitate formation of stronger, evidence-based guidelines for pediatric stroke.

In this study, we aim to address the following gaps in knowledge: (1) the annual incidence of pediatric LVO stroke; (2) the proportion of pediatric AIS cases caused by LVO; (3) the proportion of pediatric LVO cases that would fulfill adult selection criteria for mechanical thrombectomy (other than age); and (4) the long-term clinical outcomes of pediatric LVO stroke without thrombectomy.

Ethics approval was granted by the Sydney Children’s Hospital Network Human Research Ethics Committee, including a waiver for consent. The design was a multicenter retrospective cohort study of all pediatric patients (1 month to younger than 17 years) with acute AIS across the state of New South Wales (NSW), Australia, covering the 10 years from January 1, 2010, to December 31, 2019. Four centers were included: all 3 pediatric hospitals in NSW and an adult comprehensive stroke center that provides thrombectomy to older adolescents. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology () reporting guideline.

Three centers had existing prospective databases for AIS, while the fourth center required creation of a retrospective database using diagnostic coding from which AIS cases were identified. Clinical and imaging records were then assessed for the inclusion and exclusion criteria to confirm the case. The presence of AIS was based on the final clinical diagnosis made by the treating pediatric neurologist.

Inclusion criteria were age 1 month to younger than 17 years on the date of initial assessment, admission, or follow-up at one of the 4 involved centers between January 2010 and December 2019, and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) diagnostic codes of I63 (cerebral infarction), I65 (occlusion and stenosis of precerebral arteries, not resulting in cerebral infarction), or I66 (occlusion and stenosis of cerebral arteries, not resulting in cerebral infarction).

The following patient groups were excluded because they have markedly different etiologies, presentations, treatments, and outcomes: those with hemorrhagic stroke, neonatal stroke (younger than 29 days) (n = 35); those with cerebral venous sinus thrombosis (n = 9); those with group B Streptococcal meningitis–related stroke (n = 7); those with herniation syndromes from traumatic brain injury resulting in stroke (n = 8); and those with global hypoxic-ischemic injury (eg, cardiac arrest, drowning) (n = 26). Clinical and imaging records of patients identified with ICD-10 diagnostic codes I65 and I66 were assessed for evidence of infarction; in the absence of infarction or reperfusion therapy, these patients were excluded. Patients 17 years and older were excluded because a substantial proportion were treated at adult hospitals.

For this study, LVO was defined as acute occlusion of the intracranial segments of the internal carotid artery (ICA), M1 (first) or M2 (second) segments of the middle cerebral artery, A1 or A2 segments of the anterior cerebral artery (ACA), intracranial segment of the vertebral artery, basilar artery, and P1 or P2 segments of the posterior cerebral artery (PCA), as demonstrated on computed tomography angiography (CTA), magnetic resonance angiography (MRA), or digital subtraction angiography (DSA). Acute vs chronic occlusion was differentiated based on presence or absence of well-developed lenticulostriate or transdural collaterals.

This definition as it applies to the anterior circulation is derived from the first and largest positive randomized clinical trial for mechanical thrombectomy in AIS (Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands [MR CLEAN] trial: ICA, M1, M2, A1, A2),³ supported by Class I (ICA, M1) and Class IIb (ACA, M2) recommendations for mechanical thrombectomy in adults from the American Heart Association (AHA).^3,9 For the posterior circulation, it is based on AHA Class IIb recommendations in adults.⁹

The population of persons younger than 17 years in NSW as estimated in June of each year from 2010 to 2019 by the Australian Bureau of Statistics was used to determine the state pediatric population.²⁰ The annual and mean incidence rates measured in cases per 100 000 persons younger than 17 per year were calculated for pediatric patients with AIS and LVO.

The primary outcome was the functional clinical outcome 3 months after stroke onset measured using the pediatric mRS (ped-mRS) score. The primary analysis compared patients with LVO who did not undergo mechanical thrombectomy and/or IV thrombolysis with patients with non-LVO. Ped-mRS score was assessed retrospectively by a pediatric neurologist or pediatric interventional neuroradiologist based on prior assessments undertaken by a pediatric neurologist, neurosurgeon, or rehabilitation physician 2 to 4 months after AIS. Ped-mRS scores were able to be derived from the clinical records in 164 of 166 admissions (98.8%). The primary safety outcome was the rate of symptomatic intracranial hemorrhage.

Selection of patients with LVO to receive reperfusion therapy was not randomized or based on standardized selection criteria but rather was determined by the treating pediatric neurologist in the context of the clinical presentation and resource availability. The study period extended before and after mechanical thrombectomy became widely used in adults (2015 onwards), affecting patient selection. An a priori decision was thus made to make the primary analysis a comparison of the LVO without thrombectomy and non-LVO groups.

Secondary outcomes were final ped-mRS score available (including timing), stroke etiology, method of initial angiographic imaging (CTA, MRA, or DSA), timing of imaging (hours since last seen well), timing of clinical presentation (less than 6, 6 to 24, 24 to 48, or more than 48 hours since last seen well), IV thrombolysis status, mechanical thrombectomy status, and selection criteria fulfillment status (other than age) for the MR CLEAN trial,³ DWI or CTP Assessment with Clinical Mismatch in the Triage of Wake-Up and Late Presenting Strokes Undergoing Neurointervention with Trevo (DAWN) trial,¹¹ and AHA Class I-IIb recommendations⁹ for use of thrombectomy in adults. Note that the MR CLEAN trial³ included only patients with anterior circulation stroke presenting within 6 hours since last seen well, and for the DAWN trial,¹¹ within 6 to 24 hours since last seen well. These criteria are detailed in eMethods 1 and 2 in the Supplement. Secondary analyses included comparison of ped-mRS scores between patients with LVO with and without thrombectomy, and dichotomous assessment of 3-month and final ped-mRS scores classified into favorable outcome (ped-mRS score of 0 to 2) or poor outcome (ped-mRS score of 3 to 6).

All available CTA, MRA, and DSA studies for patients with AIS were analyzed independently by a pediatric interventional neuroradiologist (K.D.B.) for the presence of LVO. Imaging for all LVO cases was further independently assessed by 2 pediatric neuroradiologists at separate centers, blinded to the clinical presentation and outcome (R.G. and P.M.). Assessors recorded the presence, side, and location of the primary (most proximal) intracranial occlusion.

The primary analysis was assessed by ordinal logistic regression using MiniTab version 19.2020.2.0 (MiniTab LLC) to determine the odds ratio (OR) and P value for a worse outcome on the ped-mRS score (ie, higher score on the scale from 0 to 6) comparing patients with non-LVO, LVO without thrombectomy, and LVO with thrombectomy. For secondary analyses, ped-mRS score data were compared using ordinal logistic regression, dichotomous variables using the χ² test, and continuous variables using the unpaired t test. Confounders contributing to selection for thrombectomy were analyzed using binary logistic regression. Complete-case analysis was used to account for missing data. All P values were 2-tailed, and significance was set at P &±ô³Ù; .05.

From January 2010 to December 2019, 166 AIS admissions in 161 pediatric patients were identified in NSW. Of 161 included patients, 56 (34.8%) were female, and the mean (SD) age was 6.1 (5.4) years. A total of 166 AIS admissions were studied, and clinical follow-up was available for 164 of 166 admissions. Demographic characteristics, etiology, and timing of presentation are reported in Table 1.

LVO was present in 39 of 166 admissions (23.5%). Initial angiographic imaging was available in 36 of 39 patients with LVO, with 100% agreement of occlusion location between blinded assessors. Follow-up imaging was available for the remaining 3 cases, all demonstrating territorial infarcts consistent with previous LVO. Imaging results are reported in Table 2.

A total of 13 patients (33.3%) underwent mechanical thrombectomy (10 underwent thrombectomy alone; 3 underwent thrombectomy and IV thrombolysis), and 26 patients (66.7%) did not (ie, conservative treatment). Patients with LVO were older than patients with non-LVO (mean [SD] age, 8.23 [5.45] years vs 5.48 [5.23] years; P = .007). Patients with LVO undergoing thrombectomy were older than conservatively treated patients (mean [SD; range] age, 12.41 [4.41; 4.0-16.9] years vs 6.23 [4.73; 0.4-14.5] years; P &±ô³Ù; .001).

Across the 10-year period, the mean (SD) annual incidence of pediatric (ie, younger than 17 years) AIS was 1.02 (0.18) cases per 100 000 persons per year (95% CI, 0.89-1.16). The mean (SD) annual incidence of acute pediatric LVO was 0.24 (0.15) cases per 100 000 persons per year (95% CI, 0.13-0.35). Annual estimated incidence data are reported in the eTable in the Supplement.

At 3 months following stroke, ped-mRS score was 3 to 6 in 19 of 26 patients with LVO without thrombectomy (73.1%), 6 of 12 with LVO with thrombectomy (50.0%), and 52 of 126 with non-LVO (41.2%) (Table 3). A total of 24 of 26 cases involved the anterior circulation, of which 18 patients (75%) had a poor outcome at 3 months. Patients with LVO without thrombectomy had significantly worse outcomes than patients with non-LVO on the ped-mRS 3 months after stroke using ordinal logistic regression analysis (OR, 3.64; 95% CI, 1.68-7.87; P = .001) (Table 3).

Patients with LVO without thrombectomy also had significantly worse outcomes than patients with non-LVO on final available ped-mRS assessment (mean [SD] time from stroke, 37.0 [28.8] months; OR, 3.23; 95% CI, 1.50-6.95; P = .003) (Table 4). Dichotomous analysis results comparing ped-mRS scores of 0 to 2 vs 3 to 6 between groups were also significantly worse for patients with LVO without thrombectomy than for patients with non-LVO at 3 months (χ²₂ = 8.803; P = .01) and final assessment (χ²₂ = 7.695; P = .02).

Patients with LVO without thrombectomy had significantly worse outcomes than those treated with thrombectomy at (3 months: OR, 3.75; 95% CI, 1.05-13.45; P = .04; final assessment: OR, 6.07; 95% CI, 1.55-23.73; P = .01). These worse outcomes were more pronounced in the anterior circulation subgroup (Table 5).

Among 13 patients undergoing thrombectomy, there were no occurrences of symptomatic intracranial hemorrhage. One patient had a retroperitoneal hemorrhage from the access site (treated with a covered stent) (Table 5). Two patients with dissection and 1 with focal cerebral arteriopathy were treated with thrombectomy, with no procedural complications.

Patients with LVO presented more quickly than patients with non-LVO (Table 1). Most with LVO (27 of 39 [69.2%]) presented to the hospital within 6 hours since last seen well and 35 of 39 (90%) presented within 24 hours. Selection criteria other than age for the MR CLEAN trial³ were fulfilled in 20 of 39 patients (51.3%) and the DAWN trial¹¹ in an additional 7 patients (17.9%). If posterior circulation occlusions are included (AHA Class IIb recommendations⁹), then 32 of 39 patients (82.1%) were potentially eligible. Binary logistic regression analysis of potential confounders (age, sex, timing of imaging, year, site of occlusion) only identified age (OR, 1.31; 95% CI, 1.06-1.61; P = .01) as a significant predictor that patients with LVO underwent thrombectomy.

In our study, outcomes for pediatric LVO stroke treated conservatively were poor. Moderate to severe disability or death (ped-mRS score of 3 to 6) was present in 19 of 26 (73.1%) at 3 months (75% of anterior circulation LVO cases). The HERMES collaboration,⁸ which pooled data from 5 randomized trials on thrombectomy for anterior circulation LVO in adults, found 73.5% of control patients had a poor outcome (mRS score of 3 to 6) at 3 months—remarkably similar to the rate we identified in children (75%). In their interventional group, 54.0% had a poor outcome, and the OR of a favorable outcome following thrombectomy was 2.35.⁸ These results indicate children with LVO stroke treated conservatively have similarly poor outcomes to adults with LVO and worse outcomes than adult patients undergoing thrombectomy.

With only 5 posterior circulation strokes due to LVO, we have insufficient data to comment on this cohort. A recent literature review of pediatric basilar artery occlusion, however, reported poor long-term neurological outcomes in patients treated conservatively (n = 111), with moderate to severe disability or death in 58%.²¹

Our natural history cohort consists of patients not treated with thrombectomy or IV thrombolysis. This group may have had a worse baseline prognosis, and their outcomes may be affected by selection bias. However, 73% of patients with LVO without thrombectomy fulfilled adult selection criteria for thrombectomy, patients were included from before (2010 to 2014) and after (2015 to 2019) the publication of positive adult trials, and uptake of thrombectomy in children was variable until recently.

Previous studies assessing outcomes from pediatric AIS did not analyze patients by LVO status.^15,22,23 Pooling of patients with LVO among all stroke cases allowed for a potential underestimate of LVO stroke severity. Pediatric patients with LVO (Table 1) have differing stroke etiology to patients with non-LVO, with higher proportions of cardioembolic, dissection, and focal cerebral arteriopathy cases. However, demonstration of worse outcomes for adult patients with LVO^1,2 was a major impetus for the thrombectomy trials.^3,8 Demonstration of worse outcomes in children with LVO is important to distinguish an identifiable subgroup with poor outcomes when treated conservatively.

In our study, patients with LVO treated conservatively had significantly worse functional outcomes than patients undergoing thrombectomy (final assessment: OR, 6.07; 95% CI, 1.55-23.73; P = .01). However, the sample size is small, and these patients were older than those in the conservatively treated group. Therefore, these results should be interpreted with caution.

A retrospective single-arm cohort study of pediatric patients who underwent thrombectomy (n = 73) demonstrated favorable neurological outcomes, with 87% having an mRS score of 0 to 2 at 6 months.¹³ Symptomatic intracranial hemorrhage occurred in 1 patient.¹³ A meta-analysis of pediatric patients undergoing thrombectomy (n = 113) demonstrated favorable long-term outcomes in 90.6%.¹² These results were superior to adult trials (mRS score of 0 to 2 in 46% at 3 months⁸) and are likely affected by selection bias.¹³ However, they indicate mechanical thrombectomy in children is feasible and effective.^12,13 In our study, 50% of patients undergoing thrombectomy had a ped-mRS score of 0 to 2 at 3 months and 67% at final assessment.

Our results can now provide a natural history reference group for this population. Given the poor natural history and feasibility of thrombectomy in children, the risk-benefit balance for intervention is likely altered, and pediatric guidelines should be updated. Prospective multinational registries should be encouraged and funded.

In our cohort, 69% of children with LVO presented within 6 hours since last seen well, and 90% presented within 24 hours. Four of 5 children with LVO (82.1%) fulfilled adult selection criteria for consideration of thrombectomy other than age.^3,9,11 These high rates of early presentation likely reflect the severity of the neurological syndrome in LVO. The focus of pediatric stroke pathways should therefore be on reducing triage and imaging times with appropriate anesthetic support.^15,18

While we have used adult stroke criteria for reference, time-based limits from adult practice may be too restrictive for children. Children have strong pial collaterals, providing greater compensation in acute ischemic stroke.^24,25 Low Pediatric National Institutes of Health Stroke Scale score or presentation more than 24 hours since last seen well should not be absolute contraindications for thrombectomy in children. Rather, imaging assessment of perfusion or infarct volume may be valuable.^23,24,26

LVO was present in nearly one-quarter of pediatric AIS admissions (23.5%)—similar to the rate in adults (18% to 24%).^1,27,28 Even if ACA and PCA occlusions are excluded, the rate of LVO in our study was 22.3% (37 of 166). Our estimate of pediatric AIS incidence (1.02 cases per 100 000 persons per year) is concordant with that of the Canadian Pediatric Ischemic Stroke Registry when neonatal strokes are excluded (1.2).²² Our cohort was predominantly male (65.2%), and this sex imbalance is well established for pediatric AIS.^15,22,29 In the state of NSW, with a pediatric population of approximately 1.7 million persons²⁰ and a mean annual LVO incidence of 0.24 cases per 100 000 persons per year (95% CI, 0.13-0.35), we could anticipate 2 to 6 pediatric LVO cases per year. This data allows for greater planning of pediatric stroke services.

There are several limitations of this study. Patients from border communities may have been transferred to interstate centers. However, we expect a similar number of transfers into our state. Pediatric patients with no deficit may have been treated in community hospitals. We deviated from our original protocol by excluding patients aged 17 years after determining that a substantial proportion were treated in adult hospitals. We excluded neonatal stroke because of differing presentation, etiology, and treatment.¹⁵

The absence of randomization or standardized selection criteria for thrombectomy is a major limitation owing to selection bias. Comparison of patients with LVO treated with thrombectomy with conservatively treated patients is also weakened by the small sample size and age difference. Patients with AIS are more likely to have underlying long-term health conditions that may also affect long-term outcomes.^15,22,28

Our study used retrospective assessment of ped-mRS scores, which is more prone to bias. Reviewers of the ped-mRS score were not blinded to thrombectomy status. The mRS score was originally adapted for children by Bigi et al³⁰ but has not been validated in prospective pediatric stroke studies. Dichotomous classification of outcome into favorable (ped-mRS score of 0 to 2) and poor (ped-mRS score of 3 to 6) is not validated in children, and thus, the primary analysis was undertaken using ordinal logistic regression.

In this population-based cohort study, LVO was present in nearly one-quarter of pediatric patients with AIS, and most fulfilled adult selection criteria for mechanical thrombectomy. Pediatric patients with LVO treated conservatively had poor long-term outcomes that were significantly worse than those with non-LVO and those treated with thrombectomy. Future pediatric stroke guidelines may take account of this poor natural history.

Accepted for Publication: January 28, 2022.

Published Online: March 28, 2022. doi:10.1001/jamaneurol.2022.0323

Corresponding Author: Kartik D. Bhatia, MBBS, PhD, Department of Medical Imaging, Children’s Hospital at Westmead, Cnr Hainsworth Street and Hawkesbury Road, Westmead, NSW 2145, Australia (kartik.bhatia@health.nsw.gov.au).

Author Contributions: Messrs Bhatia and Andrews had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Bhatia, Muthusami, Worthington, Andrews.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Bhatia, Goetti, Ang.

Critical revision of the manuscript for important intellectual content: Bhatia, Briest, Webster, Troedson, Dale, Muthusami, C. Miteff, F. Miteff, Worthington, Tastula, Andrews.

Statistical analysis: Bhatia, Briest.

Administrative, technical, or material support: Bhatia, Briest, Goetti, Webster, Dale, C. Miteff, Tastula, Andrews.

Study supervision: Ang, Andrews.

Conflict of Interest Disclosures: None reported.