Customize your JAMA Network experience by selecting one or more topics from the list below.
Borlot F, Regan BM, Bassett AS, Stavropoulos DJ, Andrade DM. Prevalence of Pathogenic Copy Number Variation in Adults With Pediatric-Onset Epilepsy and Intellectual Disability. JAMA Neurol. 2017;74(11):1301–1311. doi:10.1001/jamaneurol.2017.1775
What is the role of copy number variation investigation in adults with unexplained epilepsy and intellectual disability?
In this cross-sectional study, pathogenic and/or likely pathogenic copy number variations were found in 16.1% of the probands investigated using microarray analysis; 8 nonrecurrent rare copy number variations were identified. Lennox-Gastaut syndrome was associated with an ABAT duplication and Jeavons syndrome with a KIAA2022 deletion.
This study highlights the importance of investigating adults with childhood-onset epilepsy and intellectual disability without an etiologic diagnosis.
Copy number variation (CNV) is an important cause of neuropsychiatric disorders. Little is known about the role of CNV in adults with epilepsy and intellectual disability.
To evaluate the prevalence of pathogenic CNVs and identify possible candidate CNVs and genes in patients with epilepsy and intellectual disability.
Design, Setting, and Participants
In this cross-sectional study, genome-wide microarray was used to evaluate a cohort of 143 adults with unexplained childhood-onset epilepsy and intellectual disability who were recruited from the Toronto Western Hospital epilepsy outpatient clinic from January 1, 2012, through December 31, 2014. The inclusion criteria were (1) pediatric seizure onset with ongoing seizure activity in adulthood, (2) intellectual disability of any degree, and (3) no structural brain abnormalities or metabolic conditions that could explain the seizures.
Main Outcomes and Measures
DNA screening was performed using genome-wide microarray platforms. Pathogenicity of CNVs was assessed based on the American College of Medical Genetics guidelines. The Residual Variation Intolerance Score was used to evaluate genes within the identified CNVs that could play a role in each patient’s phenotype.
Of the 2335 patients, 143 probands were investigated (mean [SD] age, 24.6 [10.8] years; 69 male and 74 female). Twenty-three probands (16.1%) and 4 affected relatives (2.8%) (mean [SD] age, 24.1 [6.1] years; 11 male and 16 female) presented with pathogenic or likely pathogenic CNVs (0.08-18.9 Mb). Five of the 23 probands with positive results (21.7%) had more than 1 CNV reported. Parental testing revealed de novo CNVs in 11 (47.8%), with CNVs inherited from a parent in 4 probands (17.4%). Sixteen of 23 probands (69.6%) presented with previously cataloged human genetic disorders and/or defined CNV hot spots in epilepsy. Eight nonrecurrent rare CNVs that overlapped 1 or more genes associated with intellectual disability, autism, and/or epilepsy were identified: 2p16.1-p15 duplication, 6p25.3-p25.1 duplication, 8p23.3p23.1 deletion, 9p24.3-p23 deletion, 10q11.22-q11.23 duplication, 12p13.33-13.2 duplication, 13q34 deletion, and 16p13.2 duplication. Five genes are of particular interest given their potential pathogenicity in the corresponding phenotypes and least tolerability to variation: ABAT, KIAA2022, COL4A1, CACNA1C, and SMARCA2. ABAT duplication was associated with Lennox-Gastaut syndrome and KIAA2022 deletion with Jeavons syndrome.
Conclusions and Relevance
The high prevalence of pathogenic CNVs in this study highlights the importance of microarray analysis in adults with unexplained childhood-onset epilepsy and intellectual disability. Additional studies and comparison with similar cases are required to evaluate the effects of deletions and duplications that overlap specific genes.
Copy number variations (CNVs) are thought to be an important mechanism of genomic diversity and evolutionary changes in humans.1,2 Often, CNVs are seen in healthy control individuals, and determination of the pathogenicity of newly identified CNVs can be challenging. Since 2004, genome-wide identification of CNVs has become efficient through array comparative genomic hybridization, allowing the detection of causative submicroscopic CNVs that were not previously identified by karyotype,1,3 especially in neuropsychiatric disorders.4
The importance of rare CNVs has been well recognized in patients with epilepsy with partial and generalized seizures,5,6 epileptic encephalopathies,7 genetic generalized epilepsy with intellectual disability (ID),8,9 atypical Rolandic epilepsy,10 fever-associated syndromes,11 and absence epilepsy.12 Rare CNVs have also been identified in individuals with developmental disabilities and/or congenital anomalies.13,14 Of note, these comorbidities are seen in a high proportion of adult patients who did not outgrow their childhood-onset epilepsies.15
In an adult epilepsy clinic, it can be difficult to diagnose specific childhood-onset epilepsy syndromes for several reasons. The clinical features might have evolved since childhood. Key diagnostic features, such as characteristic electroencephalograms (EEGs), or certain seizure types may no longer be present; even if phenotyping were performed during childhood, without those previous notes, the current features may no longer be enough for a clinical diagnosis. In addition, parents often have difficulties remembering the characteristics of certain seizures that have long been replaced by other seizure types. Finally, for some patients, despite phenotyping, the clinical manifestations are not diagnostic of any specific epileptic syndrome. Therefore, genetic testing in adults, interpreted in light of clinical features, may help clarify the etiologic diagnosis.16,17
We sought to determine the prevalence of pathogenic and likely pathogenic CNVs in adults with pediatric-onset epilepsy and ID of unknown origin. We also aimed to identify novel genes within these CNVs that may be related to seizures and could then be studied further and compared with similar cases.
Patients were recruited from the Toronto Western Hospital epilepsy outpatient clinic from January 1, 2012, through December 31, 2014. Inclusion criteria for this cross-sectional study were (1) onset of seizures between birth and adolescence and ongoing seizure activity throughout adulthood, (2) neither obvious causal structural abnormalities in their neuroimaging studies nor evident metabolic conditions that could explain their symptoms, and (3) ID of any degree, diagnosed through a formal neuropsychological evaluation and classified according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision18 (when IQ test results were available) or DSM-519 (when patients could not be tested). The exclusion criteria were (1) patients presenting with a classic phenotype of chromosomal abnormalities (eg, Down syndrome) and (2) patients previously diagnosed with well-known single-gene mutations that cause seizure phenotypes (eg, sodium channel, neuronal type I, α subunit [SCN1A] [OMIM 182389], and cyclin-dependent kinase-like 5 [CDKL5] [OMIM 300203]). Ethical approval for the study was granted by the University Health Network Research Ethics Board, and verbal informed consent was obtained from patients’ parents or caregivers.
DNA of all patients was screened for CNVs using clinical genome-wide microarray platforms; labeling and hybridization were performed following standard protocols using platform 4 × 180K Oligonucleotide Array (Agilent Technologies) and CytoSure interpret (Oxford Gene Technologies) analysis software. Some samples were studied with CytoScan HD SNP Array (Affymetrix) genomic platform and ChAS (Affymetrix) analysis software.
Patients who harbored a CNV of interest were offered segregation testing. Available relatives were tested with fluorescence in situ hybridization (FISH) analysis with probes obtained from The Centre for Applied Genomics or with microarray testing using the same technology that was used to identify the CNV in their relatives. For the latter cases, only the region of interest was subject to analysis.
Pathogenicity of CNVs was predicted based on their size and gene content according to the American College of Medical Genetics guidelines for interpretation of CNV results.20 Whenever known epilepsy genes were within the deleted or duplicated interval and there was a correlation with the patient’s phenotype, the CNV was considered to be pathogenic.
For each relevant Online Mendelian Inheritance in Man (OMIM) gene that could play a role in the patient’s phenotype, we applied the Residual Variation Intolerance Score (RVIS) and the percentile of most intolerant genes (MIG) to guide the interpretation of potential clinical significance.21 The RVIS is based on allele frequency as represented in whole-exome sequence data from the National Heart, Lung, and Blood Institute Exome Sequencing Project 6500 data set. A gene with a positive score has more common functional variation, and a gene with a negative score has less variation and is referred to as intolerant. The more intolerant a gene is, the more likely it is associated with disease if mutated or at abnormal levels.
Of the 2335 adult patients, a subset of 143 probands (6.1%) was investigated using clinical microarray (mean [SD] age, 24.6 [10.8] years; 69 male and 74 female). There were 23 probands (16.1% of the total sample investigated) and 4 affected relatives (2.8%) with 1 or more pathogenic or likely pathogenic CNV (Table 1). The investigated patients' age ranged from 17 to 41 years (median, 22 years), and 16 were female.
Among the 27 individuals with pathogenic and likely pathogenic CNVs, deletions were found in 15 patients and duplications in 10. These CNVs varied from 0.08 to 18.9 Mb (median, 4.84 Mb). Five probands had more than 1 CNV detected, including 2 patients with 2 pathogenic CNVs (patients 4 and 12; both had unbalanced translocations confirmed by FISH or G-banding), and 3 patients with a second CNV deemed to be a variant of unknown significance (patients 3, 18, and 21).
Parental testing revealed de novo CNVs in 11 of 23 probands (47.8%) and CNVs inherited from parents in 4 (17.4%), the latter representing unbalanced segregation of parental balanced rearrangements in 3 probands (patients 3, 4, and 6) and maternal X-linked inheritance in a male proband (patient 26). One mother (patient 6) presented with mild dysmorphic features similar to her offspring and borderline intellectual functioning.
Eight adult patients (34.8%) had only 1 or none of the parents available for testing. However, the CNVs in 7 of them (patients 12, 15, 16, 17, 18, 19, and 20) were pathogenic according to the American College of Medical Genetics guidelines.20 The remaining patient (patient 21) had 1 likely pathogenic CNV and 1 large (>500 kb) variant of unknown significance.
Table 1 contains the seizure phenotype, cognitive profile,22,23 systemic manifestations, and microarray results for 23 unrelated patients and 4 affected siblings, each of whom shared the respective proband’s clinically relevant CNV(s). Sixteen unrelated patients (69.6%) presented with genetic disorders associated with seizures, including OMIM cataloged human genes and genetic disorders and defined hot spots for CNVs in epilepsy.
Among the CNVs previously associated with seizures, we found a phenotype caused by KIAA2022 (OMIM 300524) deletion (patient 27; de novo Xq13.3 deletion) of particular interest: a 23-year-old woman with Jeavons syndrome (JS) and cognitive difficulties since elementary school. Her seizures were characterized by absence with eyelid myoclonia and rare bilateral convulsive seizures. Her EEGs showed polyspikes and generalized spike waves induced by eye closure, disappearing with eye opening. Her seizures were well controlled with topiramate monotherapy. Results of investigation for glucose transporter 1 deficiency were negative (solute carrier family 2 [facilitated glucose transporter], member 1 [SLC2A1] [OMIM 138140] sequencing was normal), as were results for another 476 genes tested in the same epilepsy genetic panel.
Eight nonrecurrent, rare pathogenic or likely pathogenic CNVs that contained 1 or more genes associated with ID, autism, and seizures were identified (Table 2)24-48: 2p16.1-p15 duplication, 6p25.3-p25.1 duplication, 8p23.3p23.1 deletion, 9p24.3-p23 deletion, 10q11.22-q11.23 duplication, 12p13.33-13.2 duplication, 13q34 deletion, and 16p13.2 duplication. Four genes found in these CNVs may have had a role in patients’ neurodevelopment phenotype and had low RVISs as follows: 4-aminobutyrate aminotransferase (ABAT) (OMIM 137150), –0.33 (30.7% MIG); collagen, type IV, α-1 (COL4A1) (OMIM 120130), –2.82 (0.6% MIG); calcium channel, voltage-dependent, L type, α-1C subunit (CACNA1C) (OMIM 114205), –2.09 (1.5% MIG); and SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily a, member 2 (SMARCA2) (OMIM 600014), –1.97 (1.82% MIG).
ABAT duplication (patient 21; 16p13.2 duplication) was identified in a 23-year-old man with developmental delay and a history of multiple seizure types since the age of 6 months. He died shortly after genetic testing was completed. Over time, his seizure phenotype evolved and became consistent with Lennox-Gastaut syndrome (LGS) with tonic, atonic, and atypical absence seizures and corroborant EEGs. He required partial callosotomy at the age of 4 years, ketogenic diet from the ages of 8 to 9 years, vagus nerve stimulator insertion when he was 10 years of age, and completion of callosotomy at 11 years of age. At the time of testing, he was nonverbal and incontinent and presented with a wide-based gait and bilateral hand dystonic posture. He had no dysmorphic features. The case was reviewed in the context of the CNV finding, and thus, cerebrospinal fluid γ-aminobutyric acid (GABA) testing could have provided support for GABA-transaminase overexpression as a consequence of his ABAT duplication. However, this analysis was not performed because his family decided to not pursue an invasive test.
COL4A1 deletion and CACNA1C duplication (patient 12; 13q34 deletion and 12p13.33-13.2 duplication) were seen in a 24-year-old woman with severe ID and dysmorphisms characterized by a prominent forehead, low hair implantation, and bilateral auricular appendix. She was born in Ethiopia to nonconsanguineous parents. There was no family history of seizures or ID. Seizure phenotype was characterized by bilateral convulsive seizures starting when the patient was 17 years of age; despite her severe ID, her EEG showed a normal background and spike and slow wave discharges over the left frontal and central regions. Seizures were controlled with levetiracetam and topiramate.
SMARCA2 deletion (patients 4 and 5; 9p24.3-p23 deletion and 6p25.3-p25.1 duplication) was present in a pair of siblings who developed focal-onset seizures in adolescence (15 and 17 years of age); their seizures have been controlled with monotherapy with levetiracetam. In addition to ID and autistic features, mood oscillation and intermittent aggression outbursts not aggravated (or ameliorated) by the antiepileptic therapy have been their main problems. The female sibling was born with cleft palate, and the male has developed sensorineural deafness over time.
This study of adults with epilepsy and ID revealed that a high proportion of patients had clinically relevant CNVs. Twenty-three probands (16.1%) had pathogenic or likely pathogenic CNVs, including 8 nonrecurrent rare CNVs. These CNVs encompassed genes that could explain the epilepsy and should be further studied. Of importance, this study underscores the role of reinvestigating adults with epilepsy and ID who may have been investigated as children, when the current technology was not available to clarify their diagnosis.
The role of CNV has been well studied in children with seizures.6,7,9 The largest study,6 which analyzed 805 pediatric patients with epilepsy, reported a CNV as an explanation for the patient’s phenotype in 5% of patients. In addition, in a systematic assessment7 of epileptic encephalopathies, rare CNVs were found in 7.9% of 315 patients. In 4.1%, the CNVs were pathogenic.7 Only 1 study49 has examined the presence of CNVs in adults with epilepsy and ID, and rare CNVs were present in 9.3% of 279 patients; however, only 29% of the patients studied had ID, and patients with severe ID, major neuropsychiatric issues, and epileptic encephalopathy were not included. In our cohort, all patients had some degree of ID, 47.8% of probands had autistic features, and 30.4% presented with other neuropsychiatric symptoms.
The current study indicates a higher prevalence of CNVs (16.1%) most likely because of our broader inclusion criteria, including patients with pharmacosensitive and pharmacoresistant seizure disorders associated with ID of variable severity. Another explanation for higher yield in our sample compared with the pediatric cohorts could be that studies in children may include patients in whom the phenotype was not entirely developed (eg, ID is not yet apparent or later onset of seizures). Although one could argue that the CNVs found in our study might be the cause of ID and not seizures, all CNVs (except 8p23.3p23.1 duplication and 10q11.22-q1.23 duplication) contain at least 1 gene in which heterozygous mutations have already been associated with epilepsy or that has a mechanism of action that could explain the occurrence of seizures (Table 1 and Table 2).
Of interest, for 69.5% of our cohort, the CNVs or genes present in those CNVs identified had been previously associated with well-known conditions, such as Angelman syndrome and 22q11.2 deletion syndrome. Although these genetic syndromes could have been diagnosed in childhood, (1) genetic tests were not available and/or easily performed when some of these adults were children, and karyotype was not capable of detecting the smaller CNVs currently detected by microarray; (2) mild or atypical phenotypes associated with classic genetic syndromes are sometimes difficult to recognize50,51; and (3) some of these conditions are relatively newly described compared with the median age of our cohort (eg, protocadherin 19 [PCDH19] [OMIM 300460] causing epilepsy restricted to females with intellectual disability or methyl-CpG-binding protein 2 [MECP2] [OMIM 300005] duplication syndrome in males).
The function of most genes identified in the 8 rare nonrecurrent CNVs in the present cohort was previously known, but the effect of hemizygous deletions or duplications was not always clear. As indicated in Table 2, selected genes are expressed in the brain and, overall, present a lower threshold to variation according to their RVISs. Five notable candidate genes were found: ABAT in a patient with LGS; KIAA2022 in a patient with JS; and COL4A1, CACNA1C, and SMARCA2 in 2 patients with seizures and ID not characteristic of any known epileptic syndrome.
Four γ-aminobutyrate aminotransferase gene (ABAT) recessive mutations have been described; however, no duplication of this gene has been previously identified to our knowledge. Homozygous or compound heterozygous mutations lead to GABA-transaminase deficiency, causing a hyper-GABAergic state. Such a hyper-GABAergic state underlies severe neurologic conditions characterized by psychomotor retardation and refractory seizures.44-48 The EEGs in some of these patients have demonstrated hypsarrhythmia,46 a condition that may evolve into LGS (which patient 21 with ABAT duplication also developed). It is not clear whether the patient’s duplication led to higher amounts of GABA-transaminase and therefore a hypo-GABAergic state. Of interest, hypo-GABAergic states can also lead to epilepsy with a wide spectrum of seizure severity but not yet including LGS.52 The definitive diagnosis usually requires measuring levels of GABA in the cerebrospinal fluid or its activity in liver or lymphocytes, but GABA-transaminase activity analysis was not available. However, one could hypothesize that ABAT duplication could be causative of the LGS phenotype in patient 21.
Patient 27 is a female with JS and an Xq13.3 de novo deletion overlapping the KIAA2022 gene. Male patients with KIAA2022 mutations have been described as having mild to severe ID and several types of seizures, including bilateral convulsive seizures, epileptic spasms, and LGS.53,54 Female patients with heterozygous KIAA2022 mutations may present with a milder phenotype compared with males, likely because of random X chromosome inactivation. A female patient with 100% inactivation of the normal X chromosome had a severe phenotype similar to that of males.55 de Lange and colleagues56 recently described 14 female patients with heterozygous de novo mutations of the KIAA2022 gene. Thirteen mutations resulted in a frameshift or premature stop codon that could elicit nonsense-mediated decay, predicting a complete loss of the protein function. Most of these female patients presented with intractable epilepsy with predominant myoclonic seizures and/or absences, with onset in infancy or early childhood, behavioral problems, and mild to severe ID.56 Likewise, the phenotype in patient 27 is fairly similar to that in previously described patients except that eye closure–induced absence seizures (typical of JS) were not previously described. Thus, it appears that the deletion that contained the KIAA2022 gene identified in patient 27 may be associated with JS.
With respect to the other genes less tolerant to variation found in our cohort, COL4A1, CACNA1C, and SMARCA2 had the most negative scores. COL4A1 was found in the 13q34 deletion and CACNA1C in the 12p13.33-13.2 duplication, both in patient 12. This patient did not have clinical and neuroimaging abnormalities previously associated with COL4A1 mutations,57,58 such as porencephaly, which is usually accompanied by perinatal intraventricular hemorrhage; brain small vessel disease with or without ocular anomalies; or hereditary angiopathy with nephropathy, aneurysms, and muscle cramps. Of importance, this patient also presented with CACNA1C duplication. Heterozygous mutation in the CACNA1C gene causes Timothy syndrome, characterized by autosomal dominant inheritance, long QT syndrome with syndactyly, dysmorphic features, and neuropsychiatric involvement. This L-type calcium channel gene has not been associated with an epilepsy syndrome, but a patient who harbored 2 variants of unknown significance (12p13.33 duplication and 15q11.2 deletion) that overlapped the CACNA1C gene has been described.42 The seizure phenotype was described as unspecified seizure disorder, with the EEG showing focal epileptiform activity and magnetic resonance imaging showing gray matter heterotopia.42 At present, it is not possible to determine the effects of CACNA1C duplication in our patient, even though some of her craniofacial and neuropsychiatric findings are similar to Timothy syndrome despite absence of cardiac involvement.
Finally, heterozygous mutations in the SMARCA2 gene cause Nicolaides-Baraitser syndrome, characterized by severe ID, short stature, microcephaly, sparse hair, brachydactyly, prominent interphalangeal joints, behavioral problems, and seizures.38 Seizures have been reported in two-thirds of patients; however, the seizure phenotypes of such patients have not been detailed except by a single case report of a patient with myoclonic-astatic epilepsy.59 In that report, Tang and colleagues59 described an electrographic correlation of negative axial myoclonus, normal magnetic resonance imaging results, and good response to sodium valproate therapy in a patient with SMARCA2 missense mutation (c.3721C>G; p.Gln1241Glu). The patients in our study (patients 4 and 5) who had deletions, including SMARCA2, had some nonspecific features of Nicolaides-Baraitser syndrome, such as ID and behavioral problems. However, their seizures began in adolescence and had a focal onset. They never developed myoclonic-astatic seizures and did not have the typical dysmorphic features described in Nicolaides-Baraitser syndrome.
Identification of the origin of epilepsy and ID in adult patients finally ends their prolonged diagnostic process and the associated emotional60 and financial costs. In addition, diagnosis can help in making treatment decisions; for instance, adults with 22q11.2 deletion (patients 23 and 24) can develop Parkinson disease, and 50% of these patients have psychiatric problems.61 Clozapine is a commonly used antipsychotic because of its lower propensity to induce extrapyramidal adverse effects; however, patients with 22q11.2 deletion syndrome have a high chance of developing clozapine-induced seizures.22,23 Therefore, the diagnosis in these cases is directly linked to treatment recommendations. Another example of clinical management based on genetic results was seen in patient 26: at the time of diagnosis of MECP2 duplication (maternally inherited) in the patient, the patient’s sister was 6 months pregnant and did not want to know the child’s sex. However, after learning that she could also be a carrier of this duplication and, if carrying a boy, there would be a 50% chance of passing on the MECP2 duplication, she agreed to be tested for her carrier status. She was not a carrier of this duplication, and she gave birth to a healthy child.
Finally, after genetic diagnosis, patients’ families often appreciated being part of support groups, such as the PCDH19 Alliance (http://pcdh19info.org) (patient 25), the Dup 15q Alliance (http://www.dup15q.org (patients 13 and17), and the Rare Disease Foundation.
Although our study has several aforementioned strengths, there were also some limitations. Given that our study was a clinical investigation, no gene function analysis was performed. The candidate CNVs and genes identified in our study need to have further functional evaluation and comparison with similar cases. Of note, 16 of the 23 probands underwent further genetic testing through gene panels and whole exome and whole genome sequencing, and no additional pathogenic variants were identified.
This study highlights the importance of reinvestigating adults with childhood-onset epilepsy and ID without a clear etiologic diagnosis. Although most of these patients were thoroughly investigated when their symptoms first appeared, the current genetic technology to allow such diagnosis was not available at that time. In this study, 8 nonrecurrent, rare pathogenic and likely pathogenic CNVs were identified. Two epilepsy syndromes, LGS and JS, were associated with an ABAT duplication and a KIAA2022 deletion, respectively. Further studies and comparison with similar cases are needed to evaluate the functional effects of the CNVs on specific genes. However, interpretation of genetic findings might be complex at times; therefore, phenotyping and continuous collaboration between neurologists and geneticists are important. Alternatively, epilepsy genetics programs in which neurologists seeing adult patients have genetic training could help with etiologic diagnosis and treatment decisions based on genetic findings.
Corresponding Author: Danielle M. Andrade, MD, MSc, FRCPC, Division of Neurology, Krembil Neuroscience Centre, Toronto Western Hospital, University of Toronto, 399 Bathurst St, West Wing, Room 5-445, Toronto, ON M5T 2S8, Canada (firstname.lastname@example.org).
Accepted for Publication: May 30, 2017.
Published Online: August 28, 2017. doi:10.1001/jamaneurol.2017.1775
Author Contributions: Dr Andrade had full access to all the data and takes full responsibility for the integrity of the data and the accuracy of data analysis.
Study concept and design: Borlot, Bassett, Andrade.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Borlot, Regan.
Critical revision of the manuscript for important intellectual content: Bassett, Stavropoulos, Andrade.
Administrative, technical, or material support: Borlot, Regan.
Study supervision: Stavropoulos, Andrade.
Conflict of Interest Disclosures: Dr Bassett reported holding the Dalglish Chair in 22q11.2 Deletion Syndrome at the Toronto General Hospital. No other disclosures were reported.
Funding/Support: Dr Andrade received research support from EpLink–The Epilepsy Research Program of the Ontario Brain Institute. The Ontario Brain Institute is an independent nonprofit corporation, funded partially by the Ontario government.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.
Additional Contributions: Cory Tam, MSc, and Kelly Mo, BSc, helped with data collection. Both were compensated for their work. We are grateful to the families who participated in our research.
Create a personal account or sign in to: