OBJECTIVES:

We assessed the adherence to neuroimaging guidelines and the diagnostically relevant yield of neuroimaging in newly presenting early life epilepsy (ELE).

METHODS:

There were 775 children with a new diagnosis of epilepsy (<3 years old at onset) who were recruited through the ELE study at 17 US pediatric epilepsy centers (2012–2015) and managed prospectively for 1 year. The data were analyzed to assess the proportion of children who underwent neuroimaging, the type of neuroimaging, and abnormalities.

RESULTS:

Of 725 children (93.5%) with neuroimaging, 714 had an MRI (87% with seizure protocols) and 11 had computed tomography or ultrasound only. Etiologically relevant abnormalities were present in 290 individuals (40%) and included: an acquired injury in 97 (13.4%), malformations of cortical development in 56 (7.7%), and other diffuse disorders of brain development in 51 (7.0%). Neuroimaging was abnormal in 160 of 262 (61%) children with abnormal development at diagnosis versus 113 of 463 (24%) children with typical development. Neuroimaging abnormalities were most common in association with focal seizure semiology (40%), spasms (47%), or unclear semiology (42%). In children without spasms or focal semiology with typical development, 29 of 185 (16%) had imaging abnormalities. Pathogenic genetic variants were identified in 53 of 121 (44%) children with abnormal neuroimaging in whom genetic testing was performed.

CONCLUSIONS:

Structural abnormalities occur commonly in ELE, and adherence to neuroimaging guidelines is high at US pediatric epilepsy centers. These data support the universal adoption of imaging guidelines because the yield is substantially high, even in the lowest risk group.

What’s Known on This Subject:

MRI abnormalities in epilepsy have been reported across a broad range of ages, but limited information exists for early-onset epilepsy. Consensus guidelines recommend MRI, although there have been no data for evaluating if clinical practice reflects these recommendations.

What This Study Adds:

This prospective study reveals a 40% rate of etiologically related abnormalities, which is higher than previous best estimates. Structural abnormalities are high regardless of the presence of developmental delay, seizure type, or pathogenic genetic variants.

Neuroimaging is a core investigation in the evaluation of early childhood epilepsy, which occurs in 1 to 2 out of 1000 children <3 years old.1,2 Epilepsy has various causes, many of which manifest as structural brain abnormalities.3,4 Identification of structural abnormalities may inform additional testing, including genetic or metabolic investigations.5 Certain structural abnormalities portend a pharmacoresistant course and may accelerate plans for surgical intervention.6,8 Because neuroimaging serves an important role in parsing etiology and guiding management, clinical guidelines delineate the use of MRI in children newly presenting with unprovoked seizures.9,12 Recent international guidelines recommend MRI specifically in early life epilepsy (ELE) for any child with seizure onset before the age of 2.10 It is unknown how well these recommendations are implemented into current clinical care.

Previous prospective studies in which the authors addressed use of MRI4,13 occurred in the 1980s–1990s, before the widespread availability of seizure protocol MRI and the publication of neuroimaging guidelines. The authors of these studies reported neuroimaging rates of 53%3 and 80%4 overall, with only 14%13 and 63%4 specifically receiving MRI. This likely reflects that MRI was then a relatively new technology and did not have a firmly established place in the diagnosis and evaluation of children with epilepsy. A recent retrospective single-center study from Finland revealed use of MRI in 85% of patients for infantile epilepsy.14 

There are few contemporary studies in which the yield of neuroimaging for children with seizures in the first few years of life is assessed. In a small population-based study that was focused on incident cases during infancy, the authors identified MRI abnormalities in 26 of 51 cases (51%).15 In a study of children <2 years old in the emergency department with new-onset seizures (not necessarily epilepsy), the authors described MRI abnormalities in 50% of cases although MRI was completed in just over half of cases, potentially representing a test selection bias.16 An additional study of new-onset afebrile childhood seizures evaluated in the emergency department revealed relevant computed tomography (CT) and MRI findings in 11.8% and 27.7% of cases, respectively.17 The prevalence of structural abnormalities in prospective community cohorts ranged from 16% to 21% in a broad pediatric age range that was unselected for seizure type4,13; this increased to 26% when it was limited to children <2 years old.18 Since the 1990s, improvements in magnet strength, protocols, and experience in interpreting scans have occurred; MRI is likely more sensitive than it was when it first came into use ∼30 years ago. In our study, we identified the yield and findings of neuroimaging in incident cases of epilepsy in a high-risk age group among a contemporary cohort with high rates of neuroimaging.

We enrolled 775 children with newly diagnosed ELE through 17 US pediatric epilepsy centers (2012–2015). All centers obtained institutional review board approval. Parents and guardians gave informed written consent according to the requirements at each center. Eligible children had their first seizure before their third birthday and an initial diagnosis of epilepsy established at the participating center before the age of 42 months. Epilepsy was defined as the occurrence of ≥2 unprovoked seizures on separate days or ≥1 seizures on a single day with risk factors indicating a substantial risk of recurrence in children who received treatment.19 Pediatric epileptologists at each center identified eligible children. This was an observational study; no additional clinical investigations were performed as part of the study procedures. Trained research coordinators obtained study data from medical records and reviewed this information with site investigators before entry into a Research Electronic Data Capture20 database housed at Northwestern University Feinberg School of Medicine. All data were reviewed by a lead study coordinator and the study principal investigator (A.T.B.).

Subjects were managed prospectively for 1 year. All neuroimaging studies performed before or within the first year after the initial epilepsy diagnosis were reviewed by the site’s pediatric neuroradiologist (exception of 26 subjects from 2 centers whose images were reviewed by a neuroradiologist) for comparison with the final MRI report. For any subjects without neuroimaging, site investigators performed a chart review or discussed with the neurologists treating the subjects to ascertain the reason for no imaging. Additional data included previous seizure history, type of epilepsy or syndrome, and treatment selection and response. Pharmacoresistance was defined as the failure of ≥2 antiseizure medications within the 1-year study period.21 Neuroimaging categorization was assigned on the basis of cumulative information from all imaging studies. This categorization was independent of other etiologic factors and diagnoses (eg, tuberous sclerosis complex [TSC] or metabolic disorder).

Each center had a specific epilepsy protocol for brain MRI. This was not standardized, but shared features included the following: axial and coronal T2-weighted sequence; axial and/or coronal fluid-attenuation inversion recovery sequence; high resolution oblique coronal T2-weighted imaging of the hippocampus; anatomic, thin-slice volumetric T1-weighted gradient echo sequence; and maximal slice thickness not exceeding 5 mm (between 1 mm for T1 three-dimensional images to 5 mm for some T2-acquired sequences). The magnet strength used was 1.5 or 3.0 T but was not ascertained individually. Variances in protocols between centers are described in Supplemental Table 6.

Neuroimaging studies were obtained on a clinical basis as ordered by the treating clinician. Imaging reports were reviewed centrally by 2 investigators (J.C., A.T.B.) to establish the etiologic categorization. If questions arose, site investigators directly reviewed images to confirm the etiologic categorization.

Imaging findings were classified as “incidental” if the finding was abnormal but unlikely to be related to the cause of epilepsy (eg, Chiari 1 malformation, grade 1 intraventricular hemorrhage). Incidental findings were grouped with normal neuroimaging studies.

Imaging findings were grouped into the following categories: normal, acquired (eg, intraventricular hemorrhage, stroke, or hypoxic-ischemic encephalopathy), malformations of cortical development (MCDs) (eg, schizencephaly, polymicrogyria, or focal cortical dysplasia), other diffuse brain disorders (likely developmental abnormalities involving multiple structures, cell lineages, or spanning a broad spatial distribution [eg, combined infratentorial and supratentorial lesions or cortical and subcortical abnormalities]), tuberous sclerosis, metabolic-degenerative, microcephaly, tumor, vascular abnormality, hippocampal abnormality, undetermined, and other. Isolated microcephaly was a combined clinical and radiographic category on the basis of head circumference without evidence of other radiographic MCDs, although impaired neurogenesis is presumed.22,23 A radiographic classification of “metabolic-degenerative” was used when recognized MRI patterns of a metabolic-degenerative disorder were present (eg, bilateral basal ganglia hyperintensity with Leigh syndrome or progressive atrophy on serial MRI24,25; suspected vigabatrin-related changes not included) independent of serum or cerebrospinal fluid laboratory findings. Conversely, radiologic findings that were not strongly suspicious for a metabolic-degenerative disorder were characterized on the basis of whatever imaging features that were present even if children were diagnosed with metabolic disorders through clinical, genetic, or metabolic testing (eg, Zellweger syndrome placed in diffuse brain disorder on the basis of MR findings). Findings that were clearly abnormal but with insufficient information to classify the lesion (eg, focal cortical dysplasia versus low grade tumor; hypoxic injury versus metabolic) were deemed “undetermined.” “Other” was applied to findings associated with conditions such as neurofibromatosis and scattered diffusion restriction in a dystroglycanopathy.

Genetic testing included chromosomal microarray, karyotype, epilepsy-specific or other targeted gene panel, or whole exome sequencing. Additional details regarding the evaluation of genetic information and identification of pathogenic variants were previously published.26 

Descriptive statistics were used to define population characteristics. Bivariate analyses were performed by using appropriate methods for the form of the data. χ2 tests were used for dichotomous or multichotomous variables. t tests and Wilcoxon rank tests were used to test for differences in continuous variables. Strength of association was quantified with risk ratios and 95% confidence intervals. Minimal criteria for statistical significance required P < .05. Analysis was performed in SAS 9.4 (SAS Institute, Inc, Cary, NC).

A total of 775 subjects (367 girls, 47.4%) were enrolled and managed for 1 year. The mean age at epilepsy onset was 11.1 months (65.7% had onset before 1 year of age). The mean age at diagnosis was 12.7 months.

There were 725 subjects (93.5%) who received neuroimaging (Fig 1), and 714 subjects had MRIs (620 epilepsy protocol and 94 other protocol). Only 11 individuals had CT or ultrasound as their sole imaging modality. There were 625 children (80.6%) who had an MRI completed before or during their baseline evaluation for epilepsy (534 with designated epilepsy protocols). The remaining 89 children had an MRI completed within the first year of follow-up. Children were slightly less likely to receive an MRI if their presenting seizures were of generalized or mixed or unclear semiology (206 of 234 [88.0%]) versus focal semiology or spasms (467 of 492 [94.9%]) (P = .003).

FIGURE 1

Use of neuroimaging in a cohort with ELE from initial neurologic evaluation to 1 year after epilepsy diagnosis.

FIGURE 1

Use of neuroimaging in a cohort with ELE from initial neurologic evaluation to 1 year after epilepsy diagnosis.

At baseline evaluation, 589 subjects (76%) had completed an MRI (526 seizure protocol, 63 other protocol). An additional 59 subjects had an MRI ordered but not yet performed; 16 had CT or ultrasound only. Between baseline and 1-year evaluations, 126 additional subjects completed their initial MRI.

There were 61 children who did not receive an MRI (Table 1). Although overall neuroimaging uptake was high, there was a trend for decreasing use with increasing age at seizure onset (<12 months: 485 of 509 [95%]; 1–2 years: 139 of 151 [92%]; and 2–3 years: 102 of 115 [89%]; P = .006). Neither insurance type nor distance from the families’ homes to epilepsy center was associated with MRI use.

TABLE 1

Reasons Why MRI Was Not Completed (N = 61)

ReasonsTotal, n
Lost to follow-up 14 
Primary genetic diagnosisa 14 
Unknown 12 
Not indicatedb 
Genetic generalized epilepsy 
Recommended, not donec 
CT or ultrasound sufficient to confirm abnormality (eg, PVL/IVH or trauma) 
ReasonsTotal, n
Lost to follow-up 14 
Primary genetic diagnosisa 14 
Unknown 12 
Not indicatedb 
Genetic generalized epilepsy 
Recommended, not donec 
CT or ultrasound sufficient to confirm abnormality (eg, PVL/IVH or trauma) 
a

Trisomy 21 (6), SCN1A (3), PRRT2, ARX (prenatal diagnosis), DiGeorge syndrome, Wolf-Hirschhorn syndrome, tuberous sclerosis.

b

One with CT later diagnosed with Zellweger syndrome, 1 with trisomy 21 but seizure free, 1 with benign familial neonatal seizures; others doing well clinically with normal neurodevelopment and no focal EEG findings.

c

MRI deferred because of an increased risk of sedation with the patient’s comorbidities.25 

All children with neonatal seizures (n = 31) or acute provoked seizures (n = 28) before study entry had imaging; however, 8 of 85 children with previous febrile seizures did not receive any neuroimaging even after epilepsy was diagnosed. Of those without an MRI by 1 year after epilepsy diagnosis, two-thirds did not have an identified epilepsy etiology (n = 41 of 61). In children who underwent CT or ultrasound only, 7 of 11 had an identifiable etiology: periventricular leukomalacia and intraventricular hemorrhage (PVL/IVH; 2), SCN1A (2), Zellweger syndrome (1), Down syndrome (1), and trauma (1).

Anecdotally, some physicians did not perform imaging on typically developing children if they lacked focal EEG abnormalities and their seizures responded to treatment.27 

Etiologically relevant findings were present in 40% of children (290 of 725) (Tables 2 and 3). An additional 15% (n = 107) had incidental findings.

TABLE 2

Neuroimaging Results (N = 725)

ResultsSubjects, n (%)
Normala 435 (60) 
Acquired 97 (13.4) 
MCD 56 (7.7) 
Diffuse brain disorder 51 (7.0) 
Tuberous sclerosis 19 (2.6) 
Metabolic degenerative 18 (2.5) 
Microcephaly 16 (2.2) 
Tumor 10 (1.4) 
Vascular 9 (1.4) 
Other 8 (1.4) 
Hippocampal abnormality 3 (0.4) 
Undetermined 3 (0.4) 
ResultsSubjects, n (%)
Normala 435 (60) 
Acquired 97 (13.4) 
MCD 56 (7.7) 
Diffuse brain disorder 51 (7.0) 
Tuberous sclerosis 19 (2.6) 
Metabolic degenerative 18 (2.5) 
Microcephaly 16 (2.2) 
Tumor 10 (1.4) 
Vascular 9 (1.4) 
Other 8 (1.4) 
Hippocampal abnormality 3 (0.4) 
Undetermined 3 (0.4) 
a

Includes incidental MRI findings in 15% of subjects.

TABLE 3

Types of MRI Abnormalities Within the Composite Groups

Types of MRI AbnormalitiesTotal, n
Diffuse brain disorders (N = 51)  
 WM predominant 11 
 WM and abnormal cortex 
 Hypoplastic brainstem 
  Dysplastic cortex 
  WM abnormality 
  Trisomy 21 
 Hydrocephalus 
  Chiari II 
 Other posterior fossa abnormalities 
  Dandy Walker 
  Joubert (polymicrogyria, abnormal vermis) 
  Cerebellar hypoplasia or abnormal WM 
 Septo-optic dysplasia 
 Holoprosencephaly 
 Aicardi 
 Walker-Warburg 
MCD (N = 56)  
 Focal cortical dysplasia 21 
 Lissencephaly and band heterotopia 14 
 Polymicrogyriaa 13 
  Extensive 
  Perisylvian 
  Unilateral 
  Not specified 
 Schizencephaly 
 Periventricular 
 Nodular heterotopia 
 Hemimegalencephaly 
Acquired (N = 97)  
 PVL/IVH 34 
 Anoxic injury 22 
  Perinatal HIE 20 
  Postnatal 
 Stroke 21 
  Ischemic 20 
  Hemorrhagic 
 Trauma 
 Infections 
  Prenatal 
  Postnatal 
 Genetic and acquired 
  Microdeletion and IVH 
  STXBP1+ 
  Neurofibromatosis 
Types of MRI AbnormalitiesTotal, n
Diffuse brain disorders (N = 51)  
 WM predominant 11 
 WM and abnormal cortex 
 Hypoplastic brainstem 
  Dysplastic cortex 
  WM abnormality 
  Trisomy 21 
 Hydrocephalus 
  Chiari II 
 Other posterior fossa abnormalities 
  Dandy Walker 
  Joubert (polymicrogyria, abnormal vermis) 
  Cerebellar hypoplasia or abnormal WM 
 Septo-optic dysplasia 
 Holoprosencephaly 
 Aicardi 
 Walker-Warburg 
MCD (N = 56)  
 Focal cortical dysplasia 21 
 Lissencephaly and band heterotopia 14 
 Polymicrogyriaa 13 
  Extensive 
  Perisylvian 
  Unilateral 
  Not specified 
 Schizencephaly 
 Periventricular 
 Nodular heterotopia 
 Hemimegalencephaly 
Acquired (N = 97)  
 PVL/IVH 34 
 Anoxic injury 22 
  Perinatal HIE 20 
  Postnatal 
 Stroke 21 
  Ischemic 20 
  Hemorrhagic 
 Trauma 
 Infections 
  Prenatal 
  Postnatal 
 Genetic and acquired 
  Microdeletion and IVH 
  STXBP1+ 
  Neurofibromatosis 

HIE, hypoxic-ischemic encephalopathy; WM, white matter.

a

Three additional cases of polymicrogyria are listed with diffuse brain disorders (associated with septo-optic dysplasia).

Age at seizure onset was lower in children with abnormal imaging (8.9 ± 8.1 months) versus normal (12.0 ± 9.8 months; P < .001) imaging findings. This difference was noted for both acquired injuries (8.0 ± 6.7 months; P < .001) and MCDs (7.0 ± 7.3 months; P = .004), but not for the subset of focal cortical dysplasias (12.1 ± 10.4 months; P = .97). Abnormal neuroimaging was also more frequent among children with abnormal development at the baseline evaluation (160 of 262 [61%]) versus normal development (113 of 463 [24%]) (P < .001).

Diagnostic radiographic abnormalities were present in 38% of children (153 of 404) with nonsyndromic forms of epilepsy, in 50% (132 of 264) of children with West syndrome or spasms, and in 11% (6 of 57) of children with other electroclinical syndromes (Table 4).

TABLE 4

Association Between Identifiable Electroclinical Syndrome, Frequency of Neuroimaging, and Abnormalities

Electroclinical SyndromeTotal, NNeuroimaging, NNeuroimaging Abnormalities, N (%)
Nonsyndromic 440 404 153 (38) 
West syndrome or infantile spasms 272 264 132 (50)a; 50 acquired, 25 MCD, 13 TSC, and 44 other 
Dravet 11 11 0 (0) 
Myoclonic epilepsy of infancy 0 (0) 
Myoclonic-astatic (Doose) 0 (0) 
Febrile seizures plusb 1 (20) 
Ohtaharac 1 (25) 
Benign familial infantile epilepsy 0 (0) 
Benign epilepsy of infancy 0 (0) 
Childhood absence epilepsy 0 (0) 
Benign familial neonatal epilepsy 0 (0) 
EMEd 1 (33) 
Epilepsy with myoclonic absences 0 (0) 
LGSe 2 (100) 
Migrating focal seizures of infancy 0 (0) 
BCECTS 0 (0) 
Gelastic seizures with hamartomaf 1 (100) 
Myoclonic epilepsy, nonprogressive 0 (0) 
Electroclinical SyndromeTotal, NNeuroimaging, NNeuroimaging Abnormalities, N (%)
Nonsyndromic 440 404 153 (38) 
West syndrome or infantile spasms 272 264 132 (50)a; 50 acquired, 25 MCD, 13 TSC, and 44 other 
Dravet 11 11 0 (0) 
Myoclonic epilepsy of infancy 0 (0) 
Myoclonic-astatic (Doose) 0 (0) 
Febrile seizures plusb 1 (20) 
Ohtaharac 1 (25) 
Benign familial infantile epilepsy 0 (0) 
Benign epilepsy of infancy 0 (0) 
Childhood absence epilepsy 0 (0) 
Benign familial neonatal epilepsy 0 (0) 
EMEd 1 (33) 
Epilepsy with myoclonic absences 0 (0) 
LGSe 2 (100) 
Migrating focal seizures of infancy 0 (0) 
BCECTS 0 (0) 
Gelastic seizures with hamartomaf 1 (100) 
Myoclonic epilepsy, nonprogressive 0 (0) 

BCECTS, benign childhood epilepsy with centrotemporal spikes; EME, early myoclonic encephalopathy; LGS, Lennox-Gastaut Syndrome.

a

Imaging was omitted in 8 children with spasms, including 6 children with Down syndrome, 1 child with TSC2, and 1 child with a prenatally diagnosed disorder.

b

One case of GABRG2 with T2 hyperintense hippocampus.

c

One case with absence of corpus callosum.

d

One case of nonketotic hyperglycinemia with diffusion restriction of internal capsule, ventrolateral thalamus, and corona radiate (Fig 1).

e

One case with polymicrogyria; 1 case with PVL/IVH.

f

One case with hypothalamic hamartoma.

In the nonspasms groups, neuroimaging abnormalities were more common in children with seizures of focal (100 of 251 [40%]) or unclear semiology (32 of 76 [42%]) compared with children with generalized seizures (29 of 132 [22%]) (P < .001).

In children with the lowest risk profile (typical neurodevelopment, absence of focal semiology, or no spasms), 29 of 185 (16%) had abnormal neuroimaging.

Appropriate trials of ≥2 antiseizure medications failed to control seizures in 257 (34%) of 756 children for whom data were available. Compared with children with normal imaging, most groups of children with radiographic abnormalities had a higher likelihood of pharmacoresistant seizures (Table 5). The exceptions were tuberous sclerosis and tumors.

TABLE 5

Pharmacoresistance in Comparison With Children With Normal Neuroimaging

MRI GroupRisk RatioPharmacoresistant, N (%)χ2, P
Acquired 1.35 (1.01–1.81) 37 of 95 (38.9) .054 
MCD 1.73 (1.28–2.35) 27 of 54 (50.0) .0016 
DBD 1.84 (1.35–2.49) 23 of 49 (46.9) .0005 
Metabolic-degenerative 2.85 (2.19–3.72) 14 of17 (82.4) <.0001 
TSC 0.98 (0.73–1.32) 5 of 18 (27.8) .920 
Tumor 1.01 (0.67–1.53) 3 of 10 (30.0) .9381 
No imaging 0.30 (0.12–0.78) 4 of 46 (8.7) .0034 
Normal imaging — 123 of 426 (28.9) Reference 
MRI GroupRisk RatioPharmacoresistant, N (%)χ2, P
Acquired 1.35 (1.01–1.81) 37 of 95 (38.9) .054 
MCD 1.73 (1.28–2.35) 27 of 54 (50.0) .0016 
DBD 1.84 (1.35–2.49) 23 of 49 (46.9) .0005 
Metabolic-degenerative 2.85 (2.19–3.72) 14 of17 (82.4) <.0001 
TSC 0.98 (0.73–1.32) 5 of 18 (27.8) .920 
Tumor 1.01 (0.67–1.53) 3 of 10 (30.0) .9381 
No imaging 0.30 (0.12–0.78) 4 of 46 (8.7) .0034 
Normal imaging — 123 of 426 (28.9) Reference 

DBD, diffuse brain disorder; —, not applicable.

Overall, 331 children had both neuroimaging and genetic testing. Use of genetic testing was similar in those with normal (210 of 452 [46.2%]) and abnormal neuroimaging (121 of 273 [44.3%]). Both groups were more likely to receive genetic studies than those without any neuroimaging (11 of 50 [22%]; P = .004). Genetic testing was mostly used for those with metabolic-degenerative imaging abnormalities (16 of 18 [89%]), microcephaly (14 of 16 [88%]), and diffuse brain disorders (31 of 49 [63.3%]). Among 331 children who had ≥1 genetic test results available, 125 had a pathogenic variant, including 53 of 121 (44%) of those with abnormal neuroimaging and 70 of 210 (33.3%) of those with normal imaging (P = .06).

Our data reveal high levels of adherence with current imaging guidelines for infants and toddlers at US pediatric epilepsy centers.9,12 A brain MRI that is obtained up to 1 year after ELE diagnosis provides a high diagnostic yield regardless of various clinical factors, including seizure type or development.

As per current guidelines, the vast majority of children with neuroimaging studies received a brain MRI (98.5%), and most of these were epilepsy-protocol MRIs. The largest subgroup of children who did not receive an epilepsy-protocol MRI had acquired etiologies frequently obtained before study enrollment. Approximately one-third of subjects who did not receive neuroimaging were either lost to follow-up or did not complete testing despite physician recommendation, which suggests that the intent to image exceeded 94%.

There may be instances in which it is appropriate not to pursue imaging, such as EEG results consistent with a presumed genetic generalized epilepsy or other well-characterized self-limited electroclinical syndromes (eg, benign familial neonatal seizures).

The reasons for high use of MRI in children with newly diagnosed ELE likely include the following: specific guidelines9,12; data revealing that there may be immediately actionable treatments on the basis of imaging findings16; or improved availability, accessibility, and increased physician familiarity with MRI’s role in epilepsy. We found high adherence to pediatric neuroimaging guidelines in the United States, and we show that MRI use is increased as compared with earlier large prospective pediatric cohorts of epilepsy4 or first unprovoked seizure.13 

To our knowledge, this study is only the second class I level-of-evidence study28 for the determination of structural abnormalities in ELE; a smaller prospective study of infantile epilepsy in the United Kingdom revealed MRI abnormalities in 51% of subjects.15 The overall prevalence of abnormalities in this study (40%) is higher than the previous best estimates (16%–26%), which vary with respect to age range and whether it was a first seizure or incident epilepsy.4,13,18 Past studies included a wider age-of-onset range (up to 15 or 16 years) and a different mix of epilepsy syndromes, limiting comparison. Previous research in young age groups has been limited by examining of neuroimaging at first-time seizure rather than at diagnosis of epilepsy16 or by selection of only a subset of epilepsy with focal seizures.3 Additionally, previous studies with lower use rates may be limited by selection bias, with neuroimaging ordered in cases that were either more severe or clearly focal. Our high rate of neuroimaging use in the current study minimizes such bias.

Guidelines recommend an MRI for the evaluation of any child <2 years old presenting with new-onset seizures regardless of neurologic examination or seizure semiology (except for simple febrile seizures).10 In our cohort, 1 in 6 children presenting with typical development and without either focal seizures or spasms still had abnormal neuroimaging. This supports the continued adherence to guidelines.

Traditionally, structural and genetic classifications have been treated as a dichotomy, with etiologies being either 1 or the other. Consequently, clinicians may stop diagnostic testing when either a primary genetic or radiographic etiology is identified. However, MCDs are often of genetic origin, and structural abnormalities may be present in a number of genetic syndromes5,23 (Supplemental Table 7). In this study, nearly half of children with an abnormal MRI who had any genetic testing were found to have a pathogenic variant.

Structural abnormalities on an MRI do not preclude a genetic diagnosis.26 A specific genetic diagnosis can complement the radiographic diagnosis and inform specific treatment options, provide prognostic information, facilitate access to research protocols and disease-specific support groups, and carry reproductive counseling implications. Previous examples of malformations or other structural abnormalities in children with clear genetic etiologies, such as KCNQ2 or GRIN2B variants, have been published,29,30 including additional examples from this cohort.26 Understanding the constitutional nature of the disorder may give pause to expedited surgical evaluations given the increased possibility for diffuse or multifocal cortical involvement. The benefit of joint genetic testing and neuroimaging can be bidirectional. For example, identifying a pathogenic TSC variant typically warrants neuroimaging for dysplastic lesions and/or subependymal giant cell astrocytoma. Conversely, particular radiographic patterns may inform the highest yield genetic tests for an individual patient (eg, lissencephaly or band heterotopia gene panel, Leigh-associated genes). Both structural and genetic aspects of etiology may provide important treatment-related insights.

Potential limitations of the current study include the use of tertiary centers to identify patients and the potential selection of children who were more severely affected than those in the general population. This study was not population-based, and it is unknown how many patients with ELE were not referred to a tertiary-care center; however, centers strived to identify all incident cases at study centers within the study period and achieved high enrollment (97.5%). There was no common imaging protocol for performing or interpreting imaging studies across centers. Although there was no centralized review of images, the principal investigator clarified unclear neuroimaging reports with site investigators who reviewed the images, and all images were reviewed by neuroradiologists. This was particularly important in distinguishing incidental findings from those that were deemed etiologically relevant. There may also be inherent limitations in the way we categorized images. The radiographic groupings are themselves heterogeneous, and there remains a lack of granularity because of the number of distinct etiologies that occur in small numbers.

Pediatric epilepsy centers may be more likely to order imaging with a dedicated epilepsy protocol and have pediatric neuroradiologists more familiar with the interpretation of infant brain images than other sites. Any of these concerns could limit generalizability of our results. Nonetheless, our study reflects a diverse geographic range of US centers and suggests that there is broad acceptance of MRI as the standard in the early evaluation of ELE.

MRI guidelines should be adopted universally for ELE, including outside of a tertiary-care setting. Next steps should also include standardization of ELE MRI protocols. Although there has been widespread adoption of epilepsy protocols, these have been geared toward identification of hippocampal abnormalities in older children and adults rather than toward the young, myelinating brain. To our knowledge, there are limited data to reveal the relative diagnostic advantage with (ultra) high resolution T2 imaging,31 double inversion recovery sequences,32 diffusion tensor imaging,33 or, in select cases, magnetic resonance spectroscopy in this age group.34 In future work, researchers must also determine the frequency with which early imaging impacts care decisions, both acutely and chronically.

Neuroimaging, and MRI in particular, frequently identifies an etiology for ELE, enhancing the ability of neurologists to provide a precise diagnosis, offer anticipatory guidance, and consider the full array of available therapies.

     
  • CT

    computed tomography

  •  
  • ELE

    early life epilepsy

  •  
  • MCD

    malformation of cortical development

  •  
  • PVL/IVH

    periventricular leukomalacia and intraventricular hemorrhage

  •  
  • TSC

    tuberous sclerosis complex

Dr Berg developed the database, obtained funding for the study, oversaw the collection of data, performed statistical analyses, interpreted results, and critically revised the manuscript for content; Dr Coryell collected data, interpreted the results, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Gaillard, Shellhaas, Grinspan, Wirrell, Knupp, Wusthoff, Keator, Sullivan, Loddenkemper, Patel, Chu, Massey, Novotny, and Saneto collected data, provided critical review, and suggested revisions for the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Funded by the Pediatric Epilepsy Research Foundation in Dallas, Texas.

Ignacio Valencia, MD (St Christopher’s Hospital for Children, Drexel University College of Medicine), and Eric Kossoff, MD (Johns Hopkins Hospital), provided additional support for this research as site investigators.

1
Camfield
CS
,
Camfield
PR
,
Gordon
K
,
Wirrell
E
,
Dooley
JM
.
Incidence of epilepsy in childhood and adolescence: a population-based study in Nova Scotia from 1977 to 1985.
Epilepsia
.
1996
;
37
(
1
):
19
23
[PubMed]
2
Aaberg
KM
,
Gunnes
N
,
Bakken
IJ
, et al
.
Incidence and prevalence of childhood epilepsy: a nationwide cohort study.
Pediatrics
.
2017
;
139
(
5
):
e20163908
[PubMed]
3
Wang
PJ
,
Liu
HM
,
Fan
PC
, et al
.
Magnetic resonance imaging in symptomatic/cryptogenic partial epilepsies of infants and children.
Zhonghua Min Guo Xiao Er Ke Yi Xue Hui Za Zhi
.
1997
;
38
(
2
):
127
136
4
Berg
AT
,
Testa
FM
,
Levy
SR
,
Shinnar
S
.
Neuroimaging in children with newly diagnosed epilepsy: a community-based study.
Pediatrics
.
2000
;
106
(
3
):
527
532
[PubMed]
5
Parrini
E
,
Conti
V
,
Dobyns
WB
,
Guerrini
R
.
Genetic basis of brain malformations.
Mol Syndromol
.
2016
;
7
(
4
):
220
233
[PubMed]
6
Wyllie
E
.
Surgical treatment of epilepsy in children.
Pediatr Neurol
.
1998
;
19
(
3
):
179
188
[PubMed]
7
Sugano
H
,
Arai
H
.
Epilepsy surgery for pediatric epilepsy: optimal timing of surgical intervention.
Neurol Med Chir (Tokyo)
.
2015
;
55
(
5
):
399
406
[PubMed]
8
Wirrell
EC
,
Wong-Kisiel
LC
,
Nickels
KC
.
Seizure outcome after AED failure in pediatric focal epilepsy: impact of underlying etiology.
Epilepsy Behav
.
2014
;
34
:
20
24
[PubMed]
9
Hirtz
D
,
Ashwal
S
,
Berg
A
, et al
.
Practice parameter: evaluating a first nonfebrile seizure in children: report of the quality standards subcommittee of the American Academy of Neurology, The Child Neurology Society, and The American Epilepsy Society.
Neurology
.
2000
;
55
(
5
):
616
623
[PubMed]
10
Gaillard
WD
,
Chiron
C
,
Cross
JH
, et al;
ILAE Committee for Neuroimaging, Subcommittee for Pediatric
.
Guidelines for imaging infants and children with recent-onset epilepsy.
Epilepsia
.
2009
;
50
(
9
):
2147
2153
[PubMed]
11
Nunes
VD
,
Sawyer
L
,
Neilson
J
,
Sarri
G
,
Cross
JH
.
Diagnosis and management of the epilepsies in adults and children: summary of updated NICE guidance.
BMJ
.
2012
;
344
:
e281
[PubMed]
12
Wilmshurst
JM
,
Gaillard
WD
,
Vinayan
KP
, et al
.
Summary of recommendations for the management of infantile seizures: task force report for the ILAE Commission of Pediatrics.
Epilepsia
.
2015
;
56
(
8
):
1185
1197
[PubMed]
13
Shinnar
S
,
O’Dell
C
,
Mitnick
R
,
Berg
AT
,
Moshe
SL
.
Neuroimaging abnormalities in children with an apparent first unprovoked seizure.
Epilepsy Res
.
2001
;
43
(
3
):
261
269
[PubMed]
14
Gaily
E
,
Lommi
M
,
Lapatto
R
,
Lehesjoki
AE
.
Incidence and outcome of epilepsy syndromes with onset in the first year of life: a retrospective population-based study.
Epilepsia
.
2016
;
57
(
10
):
1594
1601
[PubMed]
15
Eltze
CM
,
Chong
WK
,
Cox
T
, et al
.
A population-based study of newly diagnosed epilepsy in infants.
Epilepsia
.
2013
;
54
(
3
):
437
445
[PubMed]
16
Hsieh
DT
,
Chang
T
,
Tsuchida
TN
, et al
.
New-onset afebrile seizures in infants: role of neuroimaging.
Neurology
.
2010
;
74
(
2
):
150
156
[PubMed]
17
Chang
T
,
Acosta
MT
,
Rosser
T
, et al
.
Neuroimaging in children during the acute evaluation of new onset seizures [abstract].
Ann Neurol
.
2002
;
52
(
suppl 3
):
S134
18
Berg
AT
,
Mathern
GW
,
Bronen
RA
, et al
.
Frequency, prognosis and surgical treatment of structural abnormalities seen with magnetic resonance imaging in childhood epilepsy.
Brain
.
2009
;
132
(
pt 10
):
2785
2797
[PubMed]
19
Fisher
RS
,
Acevedo
C
,
Arzimanoglou
A
, et al
.
ILAE official report: a practical clinical definition of epilepsy.
Epilepsia
.
2014
;
55
(
4
):
475
482
[PubMed]
20
Harris
PA
,
Taylor
R
,
Thielke
R
,
Payne
J
,
Gonzalez
N
,
Conde
JG
.
Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support.
J Biomed Inform
.
2009
;
42
(
2
):
377
381
[PubMed]
21
Kwan
P
,
Sperling
MR
.
Refractory seizures: try additional antiepileptic drugs (after two have failed) or go directly to early surgery evaluation?
Epilepsia
.
2009
;
50
(
suppl 8
):
57
62
[PubMed]
22
Alcantara
D
,
O’Driscoll
M
.
Congenital microcephaly.
Am J Med Genet C Semin Med Genet
.
2014
;
166C
(
2
):
124
139
[PubMed]
23
Barkovich
AJ
,
Dobyns
WB
,
Guerrini
R
.
Malformations of cortical development and epilepsy.
Cold Spring Harb Perspect Med
.
2015
;
5
(
5
):
a022392
[PubMed]
24
Patay
Z
,
Blaser
SI
,
Poretti
A
,
Huisman
TA
.
Neurometabolic diseases of childhood.
Pediatr Radiol
.
2015
;
45
(
suppl 3
):
S473
S484
[PubMed]
25
Barkovich
AJ
.
Pediatric Neuroimaging
, 4th ed.
Philadelphia, PA
:
Lippincott Williams & Wilkins
;
2005
26
Berg
AT
,
Coryell
J
,
Saneto
RP
, et al
.
Early-life epilepsies and the emerging role of genetic testing.
JAMA Pediatr
.
2017
;
171
(
9
):
863
871
[PubMed]
27
Andropoulos
DB
,
Greene
MF
.
Anesthesia and developing brains - implications of the FDA warning.
N Engl J Med
.
2017
;
376
(
10
):
905
907
[PubMed]
28
Gross
RA
,
Johnston
KC
.
Levels of evidence: taking neurology to the next level.
Neurology
.
2009
;
72
(
1
):
8
10
[PubMed]
29
Weckhuysen
S
,
Ivanovic
V
,
Hendrickx
R
, et al;
KCNQ2 Study Group
.
Extending the KCNQ2 encephalopathy spectrum: clinical and neuroimaging findings in 17 patients.
Neurology
.
2013
;
81
(
19
):
1697
1703
[PubMed]
30
Platzer
K
,
Yuan
H
,
Schütz
H
, et al
.
GRIN2B encephalopathy: novel findings on phenotype, variant clustering, functional consequences and treatment aspects.
J Med Genet
.
2017
;
54
(
7
):
460
470
[PubMed]
31
De Ciantis
A
,
Barba
C
,
Tassi
L
, et al
.
7T MRI in focal epilepsy with unrevealing conventional field strength imaging.
Epilepsia
.
2016
;
57
(
3
):
445
454
[PubMed]
32
Wong-Kisiel
LC
,
Britton
JW
,
Witte
RJ
, et al
.
Double inversion recovery magnetic resonance imaging in identifying focal cortical dysplasia.
Pediatr Neurol
.
2016
;
61
:
87
93
[PubMed]
33
Szmuda
M
,
Szmuda
T
,
Springer
J
, et al
.
Diffusion tensor tractography imaging in pediatric epilepsy - a systematic review.
Neurol Neurochir Pol
.
2016
;
50
(
1
):
1
6
[PubMed]
34
Rincon
SP
,
Blitstein
MB
,
Caruso
PA
,
González
RG
,
Thibert
RL
,
Ratai
EM
.
The use of magnetic resonance spectroscopy in the evaluation of pediatric patients with seizures.
Pediatr Neurol
.
2016
;
58
:
57
66
[PubMed]

Competing Interests

POTENTIAL CONFLICT OF INTEREST: Dr Wirrell has received a consulting fee from Biomarin; Dr Knupp has served as a consultant for Zogenix and has an additional grant funded by the Pediatric Epilepsy Research Foundation; Dr Sullivan has served as an expert witness in various cases that involve seizures in children in which he was not subpoenaed. Subjects who were enrolled in this study by Dr Sullivan were seen as part of routine clinical care; Dr Patel discloses consulting with Greenwich Biosciences, UCB Pharma, Supernus, and LivaNova. He has pending or active grants with the Pediatric Epilepsy Research Foundation, Upsher-Smith, Greenwich Biosciences, and LivaNova; Dr Loddenkemper discloses board membership service on the council (and as president) of the American Clinical Neurophysiology Society and the American Board of Clinical Neurophysiology. He serves as committee chair at the American Epilepsy Society (Special Interest Group and Investigator Workshop committees) and as founder and consortium principal investigator of the pediatric Status Epilepticus Research Group. He received grant support from federal epilepsy-related groups as well as from pharmaceutical companies, including research support from the National Institutes of Health, the Patient-Centered Outcomes Research Institute, the Epilepsy Research Fund, the American Epilepsy Society, the Epilepsy Foundation of America, the Epilepsy Therapy Project, the Pediatric Epilepsy Research Foundation, and Citizens United for Research in Epilepsy, and he received research grants from Lundbeck, Eisai, Upsher-Smith, Mallinckrodt, Sage, and Pfizer. Dr Loddenkemper is part of pending patent applications to detect and predict seizures and to diagnose epilepsy; the other authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data