Rationale
Epilepsy is a chronic neurological condition characterized by an enduring predisposition for unprovoked and recurrent seizures. A seizure is a paroxysmal and transient event associated with the sequential progression of signs and symptoms generated as a consequence of hypersynchronous neuronal firing in the brain. Reference Fisher, Acevedo and Arzimanoglou1 The prevalence of epilepsy changes with age. In childhood, it has been estimated to affect 4–5 per 1000 children between birth and 15 years of age in population-based surveys. Reference Prasad, Sang, Corbett and Burneo2 While the precise contribution of genetic etiologies to epilepsy remains unknown, it is estimated that in about two-thirds of persons with epilepsy, there may be an inherited component. Reference Thomas and Berkovic3 Within the subgroup of epileptic encephalopathies undergoing diagnostic exome sequencing, a genetic basis has been confirmed in a proportion as high as 43.3%. Reference Helbig, Farwell Hagman and Shinde4
Identification of a genetic basis for childhood epilepsy is gaining importance. A number of recurrent chromosomal copy number variants (CNVs) have been associated with higher seizure susceptibility and variability within families, while other specific de novo CNVs have been found to be causal in patients with epilepsy, recognizable dysmorphisms, and developmental delay. Reference Allen5,Reference Myers, Johnstone and Dyment6 In terms of single-gene variants as direct causes of epilepsy and comorbid conditions, more than 1000 genes are presently identified, with a dozen frequent players. The list of inherited or sporadic gene defects associated with epilepsy is numerous with significant phenotypic overlap; these include but are not limited to structural causes, channel defects, neurotransmitter impairment, inborn errors of metabolism, and multisystem syndromes. Reference Helbig, Farwell Hagman and Shinde4,Reference Myers, Johnstone and Dyment6,Reference Epi7
There are obvious benefits to establishing a genetic basis. With a specific molecular diagnosis, the family can move on to learning about the disease, its comorbidities, and prognostic implications. Reference Trump, McTague and Brittain8 Diagnosis often informs management, even in the absence of curative or disease-specific therapy. Reference Costain, Cordeiro, Matviychuk and Mercimek-Andrews9 A timely etiological diagnosis allows for better management, such as influencing choice of anti-seizure medication (ASM), initiation of targeted metabolic and/or dietary treatment, improved surveillance for comorbidities, ability to provide accurate genetic counseling regarding recurrence risks in the family, provision of “closure”, and access to specific support groups for families. Reference Ottman, Hirose and Jain10
Genetic testing for epilepsy in Ontario is in transition. There is currently no next-generation sequencing (NGS)-based multigene panel available for epilepsy as a licensed clinical diagnostic test in Ontario, Canada. Testing is presently accessible through commercial US-based laboratories and paid for by the Ministry of Health and Long-Term Care of Ontario (MOHLTC). The Genetic Testing Advisory Committee was established in Ontario to review the clinical utility and validity of genetic tests and the provision of genetic testing in Ontario. As part of their mandate, the committee also developed recommendations and criteria for genetic testing in epilepsy. These include mandatory prerequisites such as an epileptologist/medical geneticist/clinical biochemical geneticist consultation, a list of diagnostic procedures to be undertaken before genetic testing, criteria for circumstances in which genetic testing is indicated and not indicated, and guidance for selection of genetic tests, including their limitations and considerations. Reference Jain, Andrade and Donner11 In 2018, an expert Working Group was formed by the Laboratories and Genetics Branch of the MOHLTC of Ontario which included medical geneticists, pediatric neurologists/epileptologists, biochemical geneticists, and clinical molecular geneticists from Ontario to develop a programmatic approach to implementing epilepsy panel testing as a provincial service. Reference Dyment, Prasad and Boycott12
The goal of this study is to collate objective evidence of the current state of the utility of genetic testing technologies in clinical practice at our center. We have completed a retrospective observational study of 105 children attending a tertiary care epilepsy program based in London, Ontario. Our findings represent an overview of the changing landscape of genetic testing and replicate the previous evidence of the utility of integrating NGS-based technologies into the diagnostic pathway of children with epilepsy in a representative population in Ontario. Reference Costain, Cordeiro, Matviychuk and Mercimek-Andrews9,Reference Demos, Guella and DeGuzman13 Our results further underline the continuing importance of detailed phenotyping and careful selection of genetic testing modality.
Methods
A retrospective chart review of infants and children from birth to 18 years of age, with a clinical diagnosis of epilepsy, seen in the Epilepsy Clinic of Children’s Hospital, London Health Science Centre, from January 1, 2008 to March 31, 2018, was performed to determine diagnostic yield of genetic testing technologies currently accessible in an academic clinical practice setting. Inclusion and exclusion criteria are listed below
Inclusion criteria:
-
1) A clinical diagnosis of epilepsy meeting current ILAE definition of epilepsy who have undergone any form of cytogenetic and molecular genetic testing
-
2) Genetic testing results (positive, negative, and equivocal) available.
Exclusion criteria:
-
2) A prior established diagnosis of a genetic syndrome associated with epileptic seizures as a major clinical presentation.
This project was reviewed and approved by Western University Health Sciences Research Ethics Board (approval number Project ID 111378).
The approach toward utilization of genetic tests in patients with epilepsy at our center has changed over this period. Pediatric neurologists in our division use chromosomal microarray (CMA) as a first-line test in patients selected for genetic testing in patients with epilepsy with or without developmental delay. If the microarray is negative, or a distinctive epilepsy phenotype is noted (e.g. Dravet syndrome, Generalized Epilepsy with Febrile Seizures Plus), then a single-gene targeted testing was often chosen in the initial years. Since gene panel testing was not universally accessible in our province, evaluation often included a genetics consultation. The clinical geneticist would in these cases decide to either use a targeted gene panel or a single gene depending on the epilepsy phenotype. Whole-exome sequencing (WES) was carried out in a very selected group of individuals who met the provincial criteria, in the last few years of the study period, when other tests carried out had not provided a diagnostic result.
We carried out a search through laboratory records in our cytogenetic and molecular laboratories based at the London Health Sciences Centre for relevant genetic tests performed in children with epilepsy. We also obtained a list of patients whose DNA had been sent out of country for molecular genetic testing to commercial laboratories (mostly based in the USA, as multigene panels have not yet been repatriated to our province). We simultaneously screened the clinic datasets maintained in the division of pediatric neurology, as well as the local database maintained in the Division of Genetics, for all children with a diagnosis of epilepsy attending outpatient clinics within the time frame described in the inclusion criteria. We then matched the above clinical datasets with the relevant genetic testing to generate a list of patients that met our selection criteria (Figure 1). Data extraction from the clinical and electronic patient chart were carried out to include different variables for each case identified: age, gender, age of onset of epilepsy, seizure type, family history of epilepsy, EEG findings, results of brain imaging studies (CT/MRI), genetic testing, treatments including number of ASMs used, and finally, impact of testing. Variants in genes and chromosomal CNVs were classified using ACMG criteria and variant aggregator datasets such as “ClinVar”. Reference Landrum, Lee and Benson14,Reference Richards, Aziz and Bale15 Diagnostic yield associated with CMA, single-gene testing, multigene panel testing, and WES was calculated as proportion and 95% confidence intervals (CIs) of likely pathogenic and pathogenic variants in the numerator and the number of tests requested (microarray, single-gene testing, targeted multigene panels, and WES) as the denominator. The impact of genetic testing on outcomes was also assessed in descriptive terms. While every effort was made to minimize missing data during data abstraction, data on some variables remained incomplete. Missing data elements for each variable were treated as missing at random during analysis.
Flow chart depicting scheme for patient selection for the study population.
We used an automated approach to fitting a logistic regression model so as to identify variables associated with a positive genetic diagnosis on genetic tests (CMA, single-gene panel, multigene manel, or WES). Specifically, we carried out a stepwise selection procedure with the alpha level set at 0.10 for both entry and removal of the candidate independent variables, which included age of onset, seizure type, epileptiform abnormality type, background, number of seizure medications, and presence of developmental delay. In addition, we repeated the stepwise selection procedure while setting the alpha level for both entry and removal of variables at 0.20. For the regression modeling, we used PROC LOGISTIC in SAS v9.4 (SAS Institute, Inc., Cary, North Carolina).
Results
There were a total of 2678 children seen in the epilepsy clinic for “seizures” over the time period of this study. Only 105 (3.92%) of them completed genetic testing, based on our inclusion criteria (Figure 1). Of the 105 children with epilepsy, there were 55 male (52.38%) and 50 female (47.61%), who met the inclusion criteria. Patient demographics and characteristics are summarized in Table 1. The patients were divided into three groups based on their age of onset of seizures as follows: <1 year, 1–5 years, and 6–15 years. Seizure type was identified based on clinical semiology described in the neurology consultation notes in the patient charts. The types of seizures were grouped as focal, generalized, mixed (focal and generalized or multiple seizure types), and non-epileptic (those that were considered as non-epileptic events in the clinic). The majority of seizures at onset were generalized (n = 63, 60.57% including two patients with infantile spasms). Thirty-three patients (31.73%) had focal onset seizures, four (3.85%) had mixed seizures, and three (2.88%) had non-epileptic seizures. In one patient, seizure semiology was poorly characterized (0.96%). Developmental delay was noted in 83 (79.04%) of our patients. Of these, motor delay was noted in 40 (48.19%), speech delay in 54 (65.06%), and learning disability in 47 (56.62%). Global delay was described in 29 (34.93%) and autism spectrum disorder in 11 (13.25%) of these patients. Nearly, half of the children, 54 (51.43%) in the study population were enrolled in an individualized educational program at school. The remaining were deemed developmentally normal or were in the preschool age group. Dysmorphic features were identified in 25 (24.04%) cases.
Patient demographics and characteristics
*Out of n = 105, unless otherwise specified.
IEP: individualized education plan.
Eight children (7.69%) received no ASMs, 28 (26.67%) were on monotherapy, 26 (24.76%) received two ASMs, while 43 (40.95%) received 3 or more than 3 (maximum 7) ASMs. In terms of seizure control over the 12-month period prior to the last clinic visit, 61 children (58.10%) were noted to have >90% seizure control in comparison to baseline seizure frequency, 11 (10.48%) had moderate control (>50%–90% compared to baseline), and 32 (30.48%) were noted to be poorly controlled (<50% compared to baseline). Only 34 children (32.38%) were hospitalized once, 25 children (23.81%) were hospitalized on at least 2 occasions (range 1–6), and the remaining 46 (43.81%) were never hospitalized.
EEG Findings and Imaging Studies
Ninety-four children had EEG reports available for review; in 81 children (77.88%), the EEG was interpreted as abnormal. Epileptiform abnormalities were documented in 73 (70.9%) of these 81 children, which included focal spikes or spike waves in 25 (34.25%), generalized epileptiform abnormalities in 40 (54.79%), and multifocal epileptiform discharges in 8 children (10.96%). Focal slowing was noted in 5 (6.17%) and 14 (17.28%) records documented generalized slowing of the background rhythms. Magnetic resonance imaging studies of the brain were performed and results were available in 88 children, the majority, and 55 scans (52.38%) were reported as normal, and in 12 records (13.63%) nonspecific changes (either as thinning of the cortical ribbon or the white matter due to ex vacuo change, or nonspecific signal abnormalities in T2-weighted sequences in the white matter) were reported. A specific abnormality was reported in 22 (20.95%). These abnormalities included periventricular nodular heterotopia, subcortical band heterotopia, flattening of the temporal gyri compatible with lissencephaly, agenesis of corpus callosum, multiple subependymal nodules, microencephaly, brain iron accumulation in substantia nigra, Chiari Type I malformation, solitary frontal subependymal heterotopia, multiloculated pineal cyst, and cerebellar atrophy.
Genetic Investigations and Diagnostic Yield
The results of the different genetic test modalities and the respective diagnostic yield are summarized in Table 2. We identified a significant number of novel variants in known epilepsy genes, which are all listed in Tables 3–6 and in Supplementary Table 1.
Genetic investigations and diagnostic yield
*Cases with multiple variants.
VUS: variant of uncertain clinical significance, WES: Whole-exome sequencing.
Results of microarray testing
CNV: copy number variants
Single-gene testing results
VUS: variant of uncertain clinical significance
Gene panel test results
Whole-exome sequencing results
Chromosomal Microarray (CMA)
CMA was ordered in 84 (80.77%) of 104 patients, the results were normal in 58, variants of uncertain significance (VUS) in 19, likely pathogenic CNV in 4, and pathogenic CNV in 3 patients (Table 3). The diagnostic yield for CMA was estimated to be 8.33% (95% CI 3.41, 16.41).
There were several previously established CNVs detected in our study, such as deletion 2p16.3 involving exons 3–6 of the NRXN1 (Neurexin-1) gene in a patient with severe intellectual disability and refractory epilepsy; deletion 15q26 involving exons 35–39 of the CHD2 (Chromodomain Helicase DNA Binding Protein 2) gene in a patient with generalized epilepsy, mild developmental delay, and dysmorphic features, who passed away suddenly at the age of 18 months; and deletion 9q34.11 that removed exons 13–20 of the STXBP1 (Syntaxin-Binding Protein 1) gene in a patient with global developmental delay, hypotonia, microcephaly, and focal seizures. A microdeletion at 2q23.1 of 0.176 Mb involving exons 1 and 2 of the MBD5 (Methyl-CpG Binding Domain Protein 5) gene was detected in a patient with global developmental delay, frontal lobe dysplasia, focal frontal lobe epilepsy, and failure to thrive and was reported as likely pathogenic. The microdeletion sizes of this region are variable, and the smallest previously reported microdeletion was approximately 0.038 Mb. Another deletion of 3.78 Mb at 9p24.3p24.2, which is at the extreme distal end of the 9p deletion syndrome region (OMIM# 158170), was also reported as likely pathogenic in a patient with mild developmental delay and febrile seizures (Table 3).
A novel CNV identified in our study was a large 14.82 Mb duplication at 8q21.13q22.1 interpreted as pathogenic due to its large size and gene content (73 genes, including 37 OMIM genes) in a patient with generalized seizures, speech delay, and subtle dysmorphic features including a prominent forehead and a single palmar crease. The duplication was the product of an unbalanced recombination of an insertion, transmitted from an asymptomatic mother who carried a balanced insertion of the 8q21.13q22.1 segment on her chromosome 14. A similar sized CNV has been reported in three patients with epilepsy, mild developmental delay and epilepsy, supporting the causality in our patient. Reference Rezazadeh, Borlot, Faghfoury and Andrade16 Cascade testing in this family, therefore, allowed for accurate genetic counseling for the significantly increased recurrence risk.
Single-Gene Testing
This was performed in 43 patients, the results were negative in 33, VUS in 2, likely pathogenic in 2, and pathogenic in 6 patients (Table 4). The diagnostic yield was estimated at 18.60% (95% CI 8.39, 33.40). Among the genes in which pathogenic or likely pathogenic variants were identified by single-gene testing were the SCN1A (Sodium Voltage-Gated Channel Alpha subunit 1), TSC1 (Tuberous sclerosis-1), TSC2 (Tuberous sclerosis-2), FLNA, (Filamin A) and PTRH2 (Peptidyl-TRNA Hydrolase 2).
Multigene Panels
These were completed in 26 patients (23.08%). Eleven patients had normal results. Seventeen VUS were identified in 10 patients with multiple variants, likely pathogenic variants in 2 patients, and pathogenic variants in 3 patients (Table 5). The diagnostic yield for a targeted epilepsy multigene panel was estimated at 19.23% (95% CI 6.55, 39.35). The number of genes in multigene panels performed in our patients ranged from 38 to 471.
Among genes in which pathogenic variants were identified by multigene panels were GABRA1 (Gamma-Aminobutyric Acid Type-A Receptor Subunit Alpha1) (in a 2-year-old patient with Generalized Epilepsy Febrile Seizure plus), IQSEC2 (IQ Motif and Sec7 Domain ArfGEF 2) (a 2-year-old male patient with X-linked intellectual disability and epileptic encephalopathy), and STXBP1 (4-year-old boy with Ohtahara syndrome who had infantile spasms since 3 months of age, developmental delay, and brain atrophy). The variants reported as likely pathogenic were a de novo variant in the DCX (Doublecortin) gene in a patient with extensive band heterotopia and flattening of the lateral temporal gyri compatible with lissencephaly and a de novo variant in the KCNQ2 (Potassium voltage-gated channel subfamily KQT member 2) gene in a patient with epileptic encephalopathy, severe developmental delay, and intractable epilepsy in early life.
Some of the above likely pathogenic and pathogenic variants identified by single-gene testing or multigene panels were novel and have not been previously reported in ClinVar (Tables 4 and 5), except for the IQSEC2 gene variant that has been previously reported as pathogenic. Reference Allen5,Reference Le, Prasad, Rupar, Debicki, Andrade and Prasad17
Whole-Exome Sequencing
Results of WES were available for 14 patients and were normal in 7, VUS reported in 2, likely pathogenic variant in 1, and pathogenic variants in 4 patients, with an estimated diagnostic yield of 35.71% (95% CI 12.76, 64.86) (Tables 2 and 6). The overall diagnostic yield utilizing combining the different genetic test modalities was 23.81% (95% CI 16.04, 33.11).
One of the likely pathogenic variants found by WES is in the SLC2A1 (Solute Carrier Family 2 Member 1), the gene causative of GLUT-1 deficiency. This diagnosis was further confirmed by detecting low glucose levels in the patient’s cerebrospinal fluid (CSF Glucose 2.1mmol/l). It also led to diagnostic closure for this family and a switch to the ketogenic diet in terms of management. Finding of a pathogenic variant in the ANKRD11 (Ankyrin Repeat Domain 11) gene resulted in the diagnosis of KBG syndrome (OMIM #148050) in an adolescent patient without a previous diagnosis. This patient had focal temporal lobe epilepsy, mild intellectual disability, short stature, and dysmorphic features. As expected in KBG syndrome, she grew out of her seizure disorder by the time she reached adolescence. This patient had a family history of Charcot–Marie–Tooth disease and tested positive for a mutation in the GJB1 (Gap Junction Protein Beta 1) gene, and she is currently asymptomatic but will be monitored by a neurologist for the development of symptoms. Finding of a de novo variant in the PACS1 (Phosphofurin Acidic Cluster Sorting Protein 1) gene in a patient with neonatal seizures (subsequently resolved), developmental delay, and mild dysmorphic features resulted in the diagnosis of Schuus-Hoeijmakers syndrome (OMIM #615009).
Another pathogenic variant in the GABRB3 (Gamma-Aminobutyric Acid Type A Receptor Subunit Beta3) gene was found in a 6-year-old girl with early infantile epileptic encephalopathy, microcephaly, ataxia, developmental delay, and history of developmental regression starting at the age of 14 months. Thus, the finding of variants that closely link with the patients’ phenotype and provide biological plausibility for the finding help further patient management by providing a diagnosis.
Impact of Genetic Testing on Patient Outcomes
Based on the results of genetic testing, a change in ASM was made in 4 patients, ketogenic diet was introduced as treatment in 1 patient, and screening for potential complications was implemented in 11 patients. All diagnosed patients received counseling on natural history of their disease, possible complications and recommended screening, management, recurrence risks, and possibilities of preimplantation or prenatal genetic diagnosis.
In 24 children, the genetic diagnosis ended a diagnostic odyssey for the parents and permitted diagnostic closure as well as a reduced need for further investigations. Thirty-one family members (majority were parents) underwent genetic testing following the identification of the pathogenic/likely pathogenic mutation. Overall, a positive impact on management was made possible in 17 patients (16.34%) based on genetic testing results. In the remaining eight (7.69%) patients, the genetic test served to confirm the clinical diagnosis that had already been disclosed to the patient (e.g. Tuberous sclerosis, structural brain malformation). The impact of genetic testing has been summarized in Table 7.
Outcomes and impact of genetic testing
Clinical Variables as Predictors of Outcome of Genetic Testing
At p < 0.05 level of statistical significance, no single variable emerged as a predictor for the likelihood of a positive genetic diagnosis. With p < 0.10 level of statistical significance, the presence of developmental delay was the single variable that was retained in the stepwise model with a point estimate of 0.276 (95% CI 0.077, 0.984). Further relaxation at a p < 0.20 level of statistical significance, the presence of developmental delay continued to be the only variable retained in the predictive model (Supplementary Table 2).
Discussion
As it stands, the type and timing of genetic testing in epilepsy in Ontario is largely determined by individual neurologists and/or geneticists caring for these patients, based on different variables under consideration for each physician. The use of genetic testing also depends on access to certain types of tests sent out of province, due to ministry restrictions. To help understand the practice patterns prior to the development and implementation of provincial genetic testing criteria in epilepsy, the goal of this study was to document experience with genetic testing in epilepsy in pediatric tertiary care hospital, based in London, Ontario between 2008 and 2018. Reference Jain, Andrade and Donner11,Reference Dyment, Prasad and Boycott12 Our study outlines the utility of all clinically available genetic testing during that time period, including CMA, single-gene testing, targeted multigene panels, and WES, in finding an underlying genetic cause in this population. Our overall rate of a genetic diagnosis of 23.81% is in keeping with published literature, particularly with studies that span several years. Reference Costain, Cordeiro, Matviychuk and Mercimek-Andrews9,Reference Hamdan, Myers and Cossette18
Chromosomal Microarray
The diagnostic yield of CMA of 8% is similar to that reported from other published data of approximately 10%, Reference Hamdan, Myers and Cossette18,Reference Mercimek-Mahmutoglu, Patel and Cordeiro19 and the majority (77%) of patients in whom CMA was requested had developmental delay in addition to epilepsy. Among the pathogenic CNVs identified in our study, there were well-known epilepsy hotspots, Reference Monlong, Girard and Meloche20 as well as previously unreported findings (such as the 8q21.13q22.1 duplication). In keeping with published literature, significant CNVs were detected among our patients with epilepsy accompanying a multisystem syndrome, most often in the setting of developmental delay and dysmorphic features. Reference Mercimek-Mahmutoglu, Patel and Cordeiro19,Reference Olson, Shen and Avallone21
There were some limitations to the interpretation of CNVs in our epilepsy patients. Firstly, due to the timing of their presentation and variable degrees of genetics follow-up, some of our patients with epilepsy were only offered CMA (and no molecular testing), and we were not able to exclude the contribution of single-gene variants in these patients. There were also instances where familial cascade testing was not possible, due to unavailability of parent/parents, limiting our ability to effectively assess the contribution of particular CNVs, especially the ones which are established to have variable presentations.
Molecular Testing: Single-Gene Testing, Multigene Panels, WES
Recent development of NGS technologies has identified several new genes responsible for monogenic epilepsy with high penetrance with the development of multigene epilepsy panels Reference Hamdan, Myers and Cossette18,Reference Butler, da Silva, Alexander, Hegde and Escayg22,Reference Heyne, Singh and Stamberger23 designed to maximize the yield. Early integration of WES and targeted multigene panels into the diagnostic pathway has been shown to increase not only diagnostic yield and clinical utility in such cases but also cost-effectiveness. Reference Trump, McTague and Brittain8,Reference Demos, Guella and DeGuzman13 Given that our study spanned 10 years prior to 2018, single-gene testing was performed in a considerable number of patients (n = 41, 39.4%), mostly before NGS panels became clinically available. Multigene panel testing was only performed in a minority of our cohort (n = 26, 25%) and even a smaller number of patients (n = 13, 12.5%) had access to WES. This is due to various factors, including the recent clinical availability of either test in Ontario (multigene panels for the latter 5 years and WES for the latter 3 years of this study), the provincial restrictions on out-of-country testing, as well as the change in practice culture among physicians, who likely became increasingly familiar with NGS only toward the end of the last decade. Reference Boycott, Hartley and Adam24 The advantage of testing multiple (mostly hundreds of) genes at the same time over single-gene sequencing is indisputable. However, the overall diagnostic yield of multigene panels (19.23%) is very similar in comparison to the yield of single-gene testing (18.61%) in our cohort. Although our numbers are small, this finding speaks to the importance of accurate phenotyping of seizure semiology, as well as clear delineation of clinical features, to make informed and cost-efficient choices when it comes to genetic testing. Further, genotype–phenotype correlations are improving particularly in the recognition of early-onset epileptic encephalopathies, where the diagnostic yield is decidedly higher. 25,Reference Rochtus, Olson and Smith26
As expected and shown in multiple studies worldwide, WES had the highest diagnostic yield of 35.71% (Tables 2 and 6) in our pediatric epilepsy cohort. Reference Helbig, Farwell Hagman and Shinde4,Reference Allen5,Reference Costain, Cordeiro, Matviychuk and Mercimek-Andrews9,Reference Demos, Guella and DeGuzman13,Reference Rochtus, Olson and Smith26 . In all 14 patients, WES was requested as the last-tier genetic investigation, after CMA, single-gene and/or multigene panel testing failed to reveal a genetic diagnosis. The yield would likely be even higher, or the clinical impact of the results would be stronger, if WES was available to more patients and earlier in the diagnostic odyssey in this population. While we do not have data into “time-to-diagnosis” in our cohort, multiple studies have now shown the benefit of first-tier WES on multiple occasions, especially in the pediatric intensive care setting. Reference Smith, Willig and Kingsmore27 - Reference Elliott, du Souich and Lehman30 WES is not currently performed in Ontario as a clinical diagnostic test and requires the Ministry of Health approval to be performed as an out-of-country test, while the technology and expertise already exists in the province. Our results confirm that patients with epilepsy and their caregivers in Ontario would certainly benefit from repatriation of multigene epilepsy panels and WES to our province.
Impact of Genetic Testing
The findings of our study endorse the multiple benefits of detecting an underlying genetic diagnosis in pediatric patients with epilepsy. Overall, 17 (16.34%) patients had a change in their epilepsy management, surveillance, or prognosis based on their genetic testing results and many families benefited from more specific counseling and increased options. While our numbers are small, the individual impact on each of the individuals and their families is significant and speaks to the benefit of the increasing implementation of genetic testing in epilepsy care in our institution.
Amongst the 2678 children evaluated for epilepsy during the period of the study, there likely were individuals who were not eventually diagnosed with epilepsy, or who had genetic testing performed elsewhere, or who did not meet the criteria for genetic testing in our province, the rate of systematic access to genetic testing was still lower than expected. We hope our findings of the increasing impact of genetic testing in epilepsy will provide further incentive for clinicians to consider these tests earlier in the diagnostic pathway.
Study Limitations
An observational study of this nature carries all the limitations associated with retrospective data collection and analysis. The descriptions of seizure semiology and EEG interpretations relied on the reports of individual physicians. The sample size being small, the resulting diagnostic yields have wide CIs; hence, the reader is advised caution in the interpretation of test results and its application on a wider population basis. It may also explain the limitations of the statistical model in identifying reliable clinical predictors of diagnostic test results.
Conclusions
This study had the advantage of an established clinical collaboration between epileptologists, geneticists, and laboratory professionals, which enabled deep phenotyping of both epilepsy semiology and non-neurological features, cascade familial testing whenever available to help resolve results, consistent variant interpretation, and comprehensive genetic counseling of patients and families. Being a single-center experience in Ontario was both a strength (real-life example of a particular time frame in Canada depicting temporal trends) and a limitation (small sample size and limited availability of NGS tests) of our study. Despite being able to offer some form of genetic testing to most of our patients with epilepsy, we were not able to show any predictive marker for positive genetic testing results, other than developmental delay, likely due to our sampling size. The revolution and changing trends we witnessed in genetic testing increased our yield and therapeutic success, but our study also highlighted the somewhat arbitrary selection of genetic testing based on physician experience, preference, and access. While some of our conclusions remain speculative, our results confirm the increased need of pediatric epilepsy patients in Ontario for a consistent approach to genetic testing and access to more genome-wide testing in a timely manner.
Acknowledgments
The authors wish to acknowledge that funding support for this study was provided by funds raised by the Marostica family in the memory of their daughter Melaina Marostica. Furthermore, the authors acknowledge the assistance received from Dr. I Karp, Department of Epidemiology, Western University for assistance with the statistical analysis.
Disclosures
The authors have no conflicts of interest to declare.
Statement of Authorship
SL participated in the collection and abstraction of clinical and laboratory data, creation of a dataset, and analysis of results.
NK participated in the intial development of research protocol, ethics approval, variant interpretation, and writing up of the results and discussion about the genetic testing and implications for outcome. She also mentored So Lee through the entire period for the collation of data.
EZA participated in the development of research protocol and ethics approval, created a dataset of epilepsy patients seen in the neurology clinics, and identified clinical variables that were to be included.
PY participated in the development of study protocol, faciliated the search of laboratory data on microarray studies in patients with epilepsy, and reviewed the draft of manuscript.
BS participated in the development of study protocol, faciliated the search of laboratory data on molecular genetic studies in patients with epilepsy, and reviewed the manuscript drafts.
TBB participated in the discussion of results of molecular genetic testing, variant interpretation, and writing up of the discussion in the final drafts of the manuscript.
ANP was the lead in the study design, ethics submission, supervision and mentoring during data collection, writing up of the rationale, methods, results and discussion, and final submission of the paper.
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/cjn.2020.167
