Childhood epilepsy is an exceptionally heterogeneous neurological condition with a high incidence of 5–8 in 1,000 in China (Trinka et al., 2019). Genetic diagnosis of childhood epilepsy provides the most accurate pathogenetic evidence and is crucial for providing disease-specific treatments. Monogenic forms of seizure disorders tend to manifest earlier in life, presenting as childhood epilepsy, with a broad clinical spectrum, ranging from benign, self-limited epilepsies to severe epileptic encephalopathies (Sands and Choi, 2017; Jiang et al., 2021). Hundreds of genes have been associated with epilepsy, and the list is increasing continuously (Wang et al., 2017). With the aforementioned genes, diagnosis of epilepsy based on clinical features and biochemical investigations is now moving to the genetic testing stage. Clinical application of next-generation sequencing (NGS) targeting these genes has contributed greatly to the identification of causative genetic mutations (Lindy et al., 2018; Hebbar and Mefford, 2020). Diagnostic application of NGS for cohorts of adult and children with epilepsy have gradually evolved from chromosomal microarray analysis (CMA) to target panel sequencing and subsequently expanded exome range, and the diagnostic yield ranged from 23 to 42% (Yang et al., 2019; Johannesen et al., 2020). Retrospective analysis of clinical and genetic diagnoses of childhood epilepsy in different cohorts and ethnic groups could lead to a better understanding of the genetic etiology of epilepsy and provide information on prognosis, treatment, and genetic counseling in the era of precision medicine.

Studies have revealed that genetic factors similarly play a regulatory role in acquired epilepsies resulting from trauma, stroke, neoplasm, infection, or congenital malformations (Thomas and Berkovic, 2014). Moreover, abundant genomic data of epilepsy patients have enabled research into common underlying genetic risks, and polygenic burden has been proven to be significantly enriched in multiple cohorts of patients with epilepsy (Leu et al., 2019; Moreau et al., 2020). This polygenic background information is not diagnostic; however, it suggests the possibility of rare exome variants and their combinations in epilepsy pathogenesis (Lindy et al., 2018). The contribution of rare exome-level variants that do not fit the criteria for pathogenetic diagnosis under the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) guidelines for Mendelian genetic disease still deserves exploration and study. In this study, we retrospectively analyzed the clinical and genetic characteristics of Chinese childhood epilepsy patients and examined possible exome-level rare variant accumulation as a polygenic risk factor for childhood epilepsy.

The major underlying etiology of epilepsy is genetic, especially in early-onset childhood epilepsy; thus, genetic testing has been widely accepted and applied in the clinical diagnosis of epilepsy (Costain et al., 2019; Rochtus et al., 2020). Although multiple options are available for the diagnostic investigation of the complex genetic architecture of epilepsy, high-throughput sequencing is currently accepted as the first-tier choice (Dunn et al., 2020). Targeted panel sequencing offers superior cost performance and reliable diagnosis, while whole-exome sequencing and whole-genome sequencing further increase the diagnostic yield for novel gene discovery. The advantages of high-throughput sequencing include deletion/duplication analysis, thus gradually removing chromosomal microarray analysis from clinical applications. The diagnostic yield of high-throughput sequencing in our childhood epilepsy cohort was 30.61% (90/294), including CNVs and SNVs. The mutation spectrum of our cohort consisted of 12 CNVs and SNVs in 45 genes. CNVs affecting the 16p11.2 genetic locus are associated with a range of neurodevelopmental disorders, including autism spectrum disorder, intellectual disability, and epilepsy, and were detected in three patients in our cohort (Rein and Yan, 2020). Animal models of 16p11.2 deletions or duplications recapitulate many core behavioral phenotypes, including social and cognitive deficits, and exhibit altered synaptic function across various brain areas. Therefore, treatment of cognitive deficits in these children should be emphasized when epilepsy is controlled.

Ion channel gene variants accounted for the majority of monogenic diagnoses for childhood epilepsy in our study. We categorized the monogenic diagnoses into the four classifications of epilepsy-associated genes proposed by Wang et al. (Wang et al., 2017). Pure epilepsy symptoms manifested in 43 patients with epilepsy genes. Seven patients had neurodevelopmental-associated genes. Twenty-one patients had epilepsy-related genes, and two had putatively associated epilepsy genes. Six patients were identified with variants in genes not currently categorized into the four classifications. Further investigation of these six cases revealed three possible explanations (Table 2). Patients with metabolic disorder-related genes (ALDH5A1 and MMUT) are known to develop neurological disorders; however, seizures are not always the first or the most remarkable symptom requiring medication. Two patients were diagnosed with genes associated with mitochondria-related disorders (RARS2 and SURF1). Mitochondrial diseases usually affect multiple organs and exhibit strong clinical heterogeneity (Rahman, 2019). Some patients may only have nervous system abnormalities or epilepsy as the main manifestation; however, patients previously reported with epilepsy bearing variants in RARS2 and SURF1 are rare. The last two genes were newly discovered neurodevelopmental syndrome genes (LARP7 and YY1), which require more cases to summarize clinical characteristics and to update the epilepsy-associated gene lists.

The genetic diagnostic yield in pediatric patients with epilepsy with different clinical features was further analyzed in our study. Patients with generalized onset had a higher diagnostic yield than patients with focal onset, which is consistent with previous research (Torabian et al., 2019). Diagnostic yield in the remaining subgroups based on clinical features was consistent with the current understanding of the genetic basis of epilepsy, as patients with early-onset seizures, clinically recognizable epilepsy syndromes, and abnormal brain MRI have a higher diagnostic yield. The diagnostic rate between patients in the effective and non-effective treatment groups did not differ significantly.

Anti-seizure medication administration usually starts immediately after the diagnosis of epilepsy; however, the turn-around time of gene testing is as long as 6–8 weeks in developing countries, such as China. Treatment targeting the pathogenesis of known epilepsy genes should improve the effectiveness of treatment, as the pathogenesis and relationship between genotypes and phenotypes are relatively clear for some epilepsy genes, especially the common ion channel genes (Møller et al., 2019). Patients with PRRT2 mutations responded to oxcarbazepine or levetiracetam. Ketogenic diet was very effective for patients, with the SLC2A1 variant, who were diagnosed with glucose transporter 1 deficiency. Most other patients with gene mutations, such as SCN1A, KCNT1, CDKL5, are more likely to be drug-resistant. Thus, adjusting anti-seizure medications according to the results of genetic testing will be the key to improving the diagnosis and treatment of epilepsy.

With the decrease in sequencing cost and the increased clinical relevance of genetics, genomic data are currently being adopted in the daily clinical practice of neurology. Clinicians should interpret and apply genetic data for the management of patients with epilepsy. The risk of generalized epilepsy in first-degree relatives of patients (siblings and children) is increased five- to ten-fold relative to the background population (Peljto et al., 2014). However, in our study of childhood epileptic patients with a family history, only 31.58% of those were monogenically diagnosed. A large proportion of the undiagnosed epilepsy population may still suffer from a genetic etiology that has not been fully revealed, leading to the assumption of a polygenic etiology.

Common variant risk has been proven to be significantly enriched in multiple cohorts of patients with epilepsy compared to controls using polygenic risk scores (Johannesen et al., 2020). Detection of common variants has a limited contribution to genetic diagnosis; however, studies have equally illustrated the possibility of measuring genetic risk by accumulating singleton loss-of-function variants in neurodevelopmental disorders (Chen et al., 2018). Thus, a comparison of the enrichment of rare but deleterious exonic variants between patients and control cohorts could provide insight into the multigenic burden of epilepsy pathogenesis.

Epilepsy-associated genes can be grouped into the following five broad functional categories: ion transport, cell growth and differentiation, regulation of synaptic processes, transport and metabolism of small molecules within and between cells, and regulation of gene transcription and translation (Symonds and McTague, 2020). Thus, we calculated the enrichment of rare but deleterious exonic variants (LGD and MIS30 variants) in the gene lists of patients with a monogenic diagnosis, those with negative results, and control cohorts for evaluation purposes. LGD variants were not significantly different between the groups in any of the five gene categories. These LGD variants are usually highly emphasized in the analysis of monogenic traits. When excluded from the WES analysis workflow, the LGD variants were less likely to be disease-related. A three-fold possibility of MIS30 variant occurrence in ion channel and synaptic genes in the undiagnosed group was identified, which may contribute to the multigenic risk of childhood epilepsy. In the process of analyzing the WES data of epilepsy patients, these MIS30 variants (most of them inherited from parents) are always classified as variants with unknown clinical significance, which cannot be used for monogenic diagnosis and cannot completely exclude the pathogenicity of the variant. Our findings suggest that the enrichment and accumulation of MIS30 variants in ion channel- and synaptic function-related genes may represent a multigenic burden affecting the severity of epilepsy or risk of epilepsy.

In conclusion, genetic testing based on exome sequencing has a satisfactory diagnostic rate in childhood epilepsy patients. However, multiple deleterious variants at the exome level supposedly contribute to the pathogenesis of epilepsy in patients without a monogenic diagnosis. Harmful missense mutations in genes related to ion channels and synapses are most likely to produce a multigenic burden in childhood epilepsy. Mutation and gene spectra associated with epilepsy should be updated frequently to improve genetic diagnosis in clinical practice. Nevertheless, further studies in large epilepsy cohorts should reveal more reliable evidence of multigenic causes of epilepsy for use in clinical diagnosis.