This study was conducted as a retrospective review of medical records. Children with epilepsy evaluated at Seoul National University Bundang Hospital between January 2020 and December 2024 were screened for eligibility. The inclusion criteria were (1) a clinical diagnosis of epilepsy; (2) a family history of epilepsy or suspected genetic epilepsy; and (3) completion of testing with an ‘early-onset epileptic encephalopathy gene panel’ during the study period. Suspected genetic epilepsy was determined clinically by a pediatric neurologist. The exclusion criteria were (1) age at seizure onset greater than 18 years and (2) the presence of major dysmorphisms, defined as ‘structural abnormality of morphogenesis that has significant medical, surgical, or cosmetic consequences and/or results in functional impairment, morbidity, or mortality, that occurs in <1% of the general population.’ All eligible patients were included in the final analysis.

The multigene test panel consisted of 29 genes frequently associated with early-onset and genetic epilepsies, including genes implicated in developmental and epileptic encephalopathies and familial epilepsy syndromes (ALDH7A1, ARHGEF9, ARX, ATP1A2, CDKL5, CHD2, GABRA1, GABRB3, GABRG2, HCN1, IQSEC2, KCNA2, KCNQ2, KCNQ3, KCNT1, MECP2, MEF2C, PCDH19, PNPO, PRRT2, SCN1A, SCN1B, SCN2A, SCN8A, SLC25A22, SLC2A1, SPTAN1, STXBP1, and SYNGAP1).

Genomic DNA was extracted from peripheral blood samples, enriched using a targeted capture-based sequencing approach, and sequenced on the NextSeq 550Dx (Illumina Inc., San Diego, CA, USA) platform with Local Run Manager software version 2.4.0.2385 (Illumina Inc.). Sequencing yielded a mean depth of >150×, with 99.0% of targeted bases covered at ≥10×.

Sequencing data were processed using a standardized bioinformatics pipeline for alignment, variant calling, and annotation. Exonic variants causing nonsynonymous changes and intronic variants within ±10 base pairs of exon-intron boundaries were included in the analysis.

Identified variants were filtered according to allele frequency in population databases, including the Human Gene Mutation Database (http://www.hmgd.org) and the Genome Aggregation Database (https://gnomad.broadinstitute.org). Variants with a minor allele frequency >0.1% in any population were excluded. In silico pathogenicity was assessed using multiple algorithms, including SIFT, MutationTaster, and Align-GVGD. All retained variants were classified according to the American College of Medical Genetics and Genomics (ACMG) guidelines [13]. Pathogenic variants, likely pathogenic variants, and variants of uncertain significance (VUS) were confirmed by Sanger sequencing when applicable. The gene panel configuration remained unchanged throughout the study period.

The diagnostic rate of gene panel testing was calculated. For patients with negative epilepsy gene panel results, CMA was subsequently performed to evaluate copy number variations. When CMA results were negative, trio-based WES was conducted (Fig. 1). Variant filtering, annotation, and classification for WES followed the same analytical framework used for the targeted gene panel, with a focus on genes previously linked to epilepsy.

For each eligible patient, data on sex, age at seizure onset, clustered seizures at onset (two or more seizures within 24 hours), and a history of status epilepticus were collected. Developmental status at seizure onset and at the last follow-up was determined clinically. Electroencephalography findings were extracted and categorized as normal, focal interictal epileptiform discharges, or generalized epileptiform discharges. Brain magnetic resonance imaging (MRI) findings were classified as positive if a potentially epileptogenic lesion was identified and as negative otherwise. A family history of epilepsy was also recorded. To identify clinical factors associated with gene panel positivity, associations between these clinical variables and gene panel positivity were analyzed.

To assess the clinical implications of a positive genetic diagnosis, changes in clinical management immediately after diagnosis were evaluated for each patient, with overlapping categories permitted. Specifically, selection or avoidance of ASMs, family screening, and reproductive counseling were assessed.

For descriptive statistics, medians and interquartile ranges were used for continuous variables. For inferential analyses, the chi-square test or Fisher exact test was used for categorical variables, and the Mann–Whitney U test was used for continuous variables. A P<0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics version 31 (IBM Co., Armonk, NY, USA). GraphPad Prism version 10.2.3 (GraphPad Software Inc., San Diego, CA, USA) was used to generate the figures.

This study was approved by the Institutional Review Board of Seoul National Bundang University Hospital (IRB No. B-2503-958-102). The requirement for informed consent was waived because of the retrospective study design.

A total of 79 patients with suspected genetic epilepsy or a familial epilepsy background were included in the analysis. Forty-five patients (57.0%) were male, and the median age at seizure onset was 11.9 months. Developmental delay was present in 16 patients (20.3%) at seizure onset and increased to 38 patients (48.1%) at the last follow-up. Clustered seizures at onset were reported in 25 patients (31.6%), and seven patients (8.9%) had status epilepticus. Structural abnormalities on brain MRI were identified in 11 cases (13.9%), and 10 patients (12.7%) had a first-degree family history of epilepsy (Table 1).

Through multigene panel testing, seven pathogenic variants, 10 likely pathogenic variants, and 16 VUS were identified. Among the 16 VUS, four were reclassified as likely pathogenic according to ACMG guidelines, yielding a final diagnostic rate of 26.6% (21/79). Among the 58 panel-negative patients, 16 underwent CMA, and two received an additional genetic diagnosis (Table 2 and Fig. 1). WES was performed in five panel-negative patients and yielded three additional genetic diagnoses (DLG4, FMR1, SHANK3) (Table 2). Overall, this stepwise approach resulted in a total diagnostic rate of 32.9%.

Variants identified through multigene panel testing involved well-established epilepsy-associated genes, including PRRT2 (n=5), SCN1A (n=5), SCN2A (n=3), PCDH19 (n=3), STXBP1 (n=2), CDKL5 (n=1), KCNQ3 (n=1), and SYNGAP1 (n=1) (Table 2). The distribution of seizure onset ages showed gene-specific patterns. PRRT2 showed a consistent pattern of onset in early infancy, typically between 4 and 6 months. PCDH19 cases presented from early infancy through the first year of life. In contrast, SCN1A cases showed a broader onset range extending from early infancy through later toddlerhood (Fig. 2A).

To identify clinical factors associated with gene panel positivity, we compared baseline variables at the time of evaluation, including age at seizure onset, sex, clustered seizures at onset, history of status epilepticus, abnormal brain MRI findings, developmental delay, and first-degree family history of epilepsy. Among these factors, age at seizure onset was the only variable significantly associated with gene panel positivity, whereas the remaining clinical variables did not differ significantly between the positive and negative groups (Table 1 and Fig. 2B).

Consistent with this finding, the distribution of seizure onset age showed enrichment of early-onset presentations among gene panel-positive cases. Seventeen of 21 panel-positive patients had seizure onset within the first 12 months, 19 of 21 within 24 months, and all 21 within 36 months. Seizure onset before 12 months was associated with significantly higher panel positivity (41.5% vs. 10.5%, P=0.002). In contrast, the panel-negative group showed a broader onset distribution, with a larger proportion of later-onset cases beyond early childhood (Fig. 2C).

Among the 26 patients with a genetic diagnosis, 12 (46.2%) experienced clinically meaningful management changes attributable to the genetic findings. These implications were summarized as discrete action items, with overlapping categories permitted. Specifically, the most frequent consequence was selection or avoidance of ASMs (n=7). In addition, family screening was undertaken in four patients, and family planning-related genetic counseling was performed in two. Notably, 11 of the 12 patients with management changes after genetic diagnosis were identified through gene panel testing, highlighting the practical utility of this approach.

In this study, we showed that targeted epilepsy gene panel testing provided substantial diagnostic value in a real-world pediatric epilepsy cohort. The diagnostic yield of 26.6% is consistent with previously reported yields for targeted multigene panel-based approaches, which range from 18 to 28% [3,14,15]. Seizure onset before 12 months of age was significantly associated with gene panel positivity, consistent with previous reports [7].

Our cohort reflects routine clinical practice rather than a highly selected research population. Patients were referred for genetic testing based on broadly applicable clinical indicators, such as early seizure onset or a family history of epilepsy, rather than predefined electroclinical syndromes or severe developmental impairment. The relatively high diagnostic yield observed in our cohort suggests that clinically meaningful genetic diagnoses can be obtained even when classic ‘red flags’ for genetic epilepsy are not uniformly present [2].

Despite the continued expansion in the number of genes associated with epilepsy, a limited subset accounts for most identified variants [14]. In our cohort, variants were identified in eight of the 29 curated genes, and the top five genes (SCN1A, PRRT2, PCDH19, SCN2A, and STXBP1) accounted for 69.2% of genetic diagnoses. Previous studies have shown that 20 to 30 genes account for more than 80% of genetic diagnoses in epilepsy, supporting the use of multigene panel testing as a first-tier diagnostic approach for pediatric epilepsy [12,14]. A meta-analysis showed that gene panel testing was the most cost-effective diagnostic modality [16].

In our cohort, CMA was positive in two of 16 patients, yielding a diagnostic rate of 12.5%. The reported diagnostic yield of copy number variants is approximately 10%, which is consistent with our findings [3,14]. Because copy number variants may not be detected by gene panel testing or short-read WES, CMA can be considered in panel-negative cases before conducting WES [12].

In the rapidly evolving landscape of genetic testing technology, recent studies suggest the utility of WES or whole-genome sequencing (WGS) as first-tier testing in epilepsy [17]. More recently, WGS has emerged as a robust tool for genetic evaluation in neonatal and infantile epilepsy in research settings [18]. As a first-tier test, WGS not only improved diagnostic performance but also showed potential cost savings in 94.05% of simulations [19]. However, because access to WES and WGS remains variable, expert agreement on first-tier testing in epilepsy remains below 50%, according to an online survey of 45 centers across 23 European countries [20]. In Korea, lack of insurance coverage remains a major barrier to the use of WES or WGS as a first-tier test. Therefore, targeted epilepsy gene panels offer a favorable balance between diagnostic yield and feasibility in real-world clinical practice at the time of writing.

Our study shows that genetic diagnosis in pediatric epilepsy can lead to meaningful clinical changes beyond etiologic classification. Among patients with a confirmed genetic diagnosis, approximately half underwent direct modifications in clinical management, consistent with previous reports [7,12]. One of the most important effects of genetic diagnosis was on ASM management. In patients with PRRT2-related epilepsy, identification of a pathogenic variant supported the diagnosis of self-limited familial infantile epilepsy and facilitated rapid tapering of ASMs in accordance with the typically favorable natural course [21]. In these cases, the genetic diagnosis increased clinician confidence in avoiding overtreatment and in counseling families regarding expected seizure remission.

In addition, genetic diagnoses informed medication avoidance and expectation management in other genetic epilepsies. In patients with SCN1A variants, confirmation of a genetic etiology supported avoidance of sodium channel-blocking ASMs (e.g., oxcarbazepine), which are known to exacerbate seizures in this population [8,22]. Similarly, in patients with SCN2A variants, decisions regarding sodium channel-targeting agents were optimized, either supporting their use in early-onset cases or prompting reconsideration and treatment modification in others [23].

Identification of PCDH19 variants helped explain seizure clustering patterns and variable treatment responsiveness, thereby reducing uncertainty when seizure control remained incomplete despite multiple medication trials [24]. In these cases, the genetic diagnosis shifted the focus of management from aggressive escalation of ASMs to a more expectant approach, thereby minimizing adverse effects. These examples illustrate that the clinical impact of genetic diagnosis is not limited to the availability of gene-specific therapies.

Genetic diagnoses also had important implications for family screening. In one patient with an SCN1A variant, a family history of febrile seizures in the paternal lineage prompted genetic testing of a second-degree relative, which identified the same variant and enabled a diagnosis of genetic epilepsy with febrile seizures plus. In another patient with an SCN1A variant, both parents had a history of early-life seizures and underwent genetic testing with negative results, supporting a diagnosis of Dravet syndrome in the patient. In one patient with a PRRT2 variant, genetic testing was performed in a sibling with febrile seizures; the result was negative, providing reassurance regarding the sibling’s prognosis.

Genetic diagnoses also had important implications for reproductive counseling. In one patient with a PCDH19 variant, the patient’s mother had intellectual disability and a history of clustered seizures. Identification of the same pathogenic variant clarified the mother’s diagnosis and enabled appropriate reproductive counseling. In another patient with a PCDH19 variant, a male sibling was screened to determine whether he carried the same variant. The result was negative, providing reassurance to the family.

Several limitations of this study should be acknowledged. First, this was a retrospective, single-center study with a relatively small sample size, which may limit the generalizability of our findings and reduce statistical power to detect associations with less prevalent clinical factors. As a result, some potentially relevant predictors of genetic diagnosis may not have reached statistical significance. Second, patient selection was based on clinical suspicion of genetic epilepsy, including early seizure onset or a family history of epilepsy, introducing inherent referral and selection bias. Consequently, the diagnostic yield observed in this cohort may not be directly applicable to unselected pediatric epilepsy populations. Third, genetic testing strategies were not uniform across all patients. Although all patients underwent targeted epilepsy gene panel testing, subsequent CMA or trio-based WES was performed only in a subset of panel-negative patients, depending on clinical judgment and parental consent. This heterogeneity in sequential testing may have led to underestimation of the true overall genetic diagnostic yield. Fourth, the inherent technical limitations of multigene panel testing should be acknowledged, as such panels cannot detect repeat expansions, deep intronic variants, and methylation defects. Rapid technological advances in WES and WGS, together with increasing accessibility, should also be considered. Finally, although we documented clinically meaningful management changes following genetic diagnosis, the study was not designed to systematically evaluate long-term outcomes, such as sustained seizure control, neurodevelopmental trajectories, or quality of life. Future prospective studies with standardized follow-up are warranted to assess the longitudinal impact of genetic diagnosis on long-term clinical outcomes in pediatric epilepsy.

In this real-world pediatric epilepsy cohort, targeted epilepsy gene panel testing yielded a substantial diagnostic rate and provided clinically meaningful information that influenced patient management and family counseling. Earlier seizure onset was the key predictor of gene panel positivity. These findings support the integration of genetic testing into routine epilepsy care, particularly for patients with early-onset seizures.