ABSTRACT
Objective
Febrile seizures (FS) are currently classified as simple or complex primarily to describe seizure characteristics, with limited ability to identify the underlying etiology. We evaluated whether molecular diagnosis could complement conventional classification and improve the diagnosis and management of FS.
Methods
This nationwide, multi‐center, prospective cohort study, conducted from January 2019 to February 2023, enrolled patients who were clinically diagnosed with FS. The panel targeted 189 genes associated with fever‐induced seizures. Initially, children were categorized into simple febrile seizures (SFS) and complex febrile seizures (CFS) groups based on medical history and clinical evaluation. Variant classification followed the ACMG guidelines. We analyzed the cohort demographic characteristics, potential genetic etiologies across different clinical phenotypes, and clinical utility of genetic diagnosis.
Results
Our study included 3300 children with FS, and ultimately analyzed sequencing data from 2552 children, of whom 273 received a positive diagnosis. The molecular diagnostic yield was 11% (273 out of 2552 patients), with the top 10 genes accounting for 70.4% of all confirmed genetic diagnoses. No significant difference in molecular diagnostic yield was observed between the SFS (9.5%, 111/1163) and CFS (11.7%, 162/1389) groups (p = 0.085). Molecular diagnostics had immediate clinical utility, influencing management (76.2%) and informing prognosis (72.6%). The cost of a single panel sequencing for a three‐person family was $240, with a turnaround time of 14 days.
Interpretation
Molecular diagnosis may improve the diagnosis of FS by identifying genetic fever‐sensitive epilepsies that may be missed by phenotype‐based classification, with implications for risk stratification and individualized management.
Keywords: febrile seizures, gene screening, genetic etiology, genotype‐phenotype correlation, molecular diagnosis
Abbreviations
- ACMG
American College of Medical Genetics and Genomics
- CFS
complex febrile seizures
- DEE
developmental epileptic encephalopathy
- DS
Dravet Syndrome
- FS
febrile seizures
- GEFS+
Generalized Epilepsy with Febrile Seizures Plus
- MAF
minor allele frequency
- NGS
next‐generation sequencing
- P/LP
pathogenic/likely pathogenic
- SCBs
sodium channel blockers
- SFS
simple febrile seizures
- VUS
variants of uncertain significance
1. Introduction
Febrile seizures (FS) represent the most common type of childhood seizures, associated with a febrile illness but without evidence of intracranial infection or defined cause [1]. The global incidence of FS in children before the age of 5 is approximately 2%–14%. However, significant geographic variations are observed, with higher rates reported in Asian populations [2, 3]. To date, the evaluation of FS primarily consists of characterizing a patient's type of FS and no specific treatment for FS is indicated other than supportive care and evaluation for underlying fever causes [4].
It is important to recognize that clinical classification may not always be precise in pediatrics due to reliance on parental or caregiver descriptions of the child's symptoms, which can be subjective, and children's lower tolerance for certain diagnostic procedures. Furthermore, there is significant overlap between FS and certain fever‐related epilepsies regarding age at onset, seizure types, and fever‐induced characteristics, especially in genetic epileptic syndromes such as Generalized Epilepsy with Febrile Seizures Plus (GEFS+), Dravet syndrome, and PCDH19‐related epilepsy [5]. This phenotypic overlap is difficult to clearly distinguish based solely on medical history and clinical features, which can lead to misdiagnosis and treatment delays that may contribute to intractable epilepsy and adverse neurodevelopmental outcomes. Importantly, these fever‐sensitive epilepsies typically have a clear genetic cause and benefit from precision treatment strategies. An early molecular diagnosis can therefore transform the management approach from the conventional view that “FS require no special treatment” to targeted interventions based on genetic etiology. With the development of precision medicine, traditional clinical classification seems to be inadequate to meet the current needs of diagnosis and treatment.
Although the majority of FS cases appear to benign and self‐limiting [6], research has shown that approximately 7% of children with FS develop epilepsy or afebrile seizures subsequently [7]. Especially in children with complex FS, the prolonged duration of seizures, focal features, and high recurrence rate may indicate an underlying risk of neurological injury. Even with simple FS, approximately 1%–2% of children face a risk of developing secondary epilepsy [8]. Whether these adverse outcomes are directly caused by febrile seizures or are confounded with other epilepsy syndromes, an early and definitive etiological diagnosis can guide targeted treatment, potentially altering the incidence of unfavorable outcomes [9]. However, despite the high incidence of FS, recommendations for genetic testing remain limited.
Gene panels target curated sets of genes associated with specific clinical phenotypes, aiming to maximize coverage, sequencing depth, sensitivity, and specificity [10]. The diagnostic yield for the target genes surpasses that of exome or genome sequencing, while offering lower costs and a shorter turnaround time, making it suitable for large‐scale clinical screening [11]. In this study, we therefore designed a panel comprising 189 target genes focused on phenotypes associated with fever‐induced seizures. We initiated a nationwide, multi‐center, prospective genetic research project, which included 3300 patients with FS. Comprehensive analyses of the panel sequencing data included genetic heterogeneity, immediate clinical utility of molecular diagnosis and the exploration of potential modifying genes, to evaluate the role of molecular diagnosis in optimizing clinical decision‐making. This study aimed to determine whether molecular testing could address the limitations of conventional phenotype‐based classification and improve etiological diagnosis of febrile seizures.
2. Methods
2.1. Participants
We initiated a nationwide, multicenter, cohort study that encompassed 180 hospitals across 30 administrative divisions in China. Eligibility for the study was extended to all participants who had experienced at least one febrile seizure and consented to undergo multigene panel testing. The inclusion criteria were a diagnosis of FS based on the definition by the International League Against Epilepsy (ILAE) [5] and the occurrence of FS without any previous onset of afebrile seizures. Exclusions were applied to children diagnosed with epilepsy, central nervous system infections, known acquired or genetic causes, as well as those with incomplete data. Clinical diagnoses were made by trained pediatricians at each center, based on participants' anamnesis and clinical examinations. The case report form (CRF) documented sex, gestational age, FS details (age at onset, type of seizure, frequency of seizures, and EEG findings), family medical history, developmental assessments, and other neurological and non‐neurological features. The EEG results for all patients were either normal or showed nonspecific abnormalities. Participants were classified based on clinical manifestations into two groups:
2.1.1. Simple Febrile Seizures (SFS)
These are primarily generalized seizures lasting less than 15 min and do not recur within 24 h. They occur in children with normal psychomotor development and without the previous history of afebrile seizures [5].
2.1.2. Complex Febrile Seizures (CFS)
These are characterized by focal seizures or status epilepticus (lasting 15 min or longer) and may recur within 24 h (occurring three or more times). This category also includes febrile seizures in children with previous neurological deficits or seizures associated with postictal neurological abnormalities [5].
The program was registered with the Chinese Clinical Trial Registry (No. ChiCTR1900022164). The study received approval from the Ethics Committee of the Chinese PLA General Hospital and the ethics boards of the participating hospitals. Written informed consent was obtained from the parents or guardians of all participants. All clinical information of participants was collected by their physicians and entered into the database by a designated researcher.
2.2. Panel Design
To design the panel, we first conducted an initial screening using Mygeno Company's Inhouse database, which includes over 6000 epilepsy cases, to focus on genes associated with FS or FS‐related epilepsy. This screening identified 123 high‐prevalence genes specific to the Chinese population. We then expanded the gene set by searching the NCBI, OMIM, and HPO databases with keywords such as “febrile seizures” and “fever‐related seizure”. This process allowed us to collect FS‐related genes, epilepsy‐related genes with a history of FS, and FS‐related genes of scientific interest, resulting in a final panel of 189 genes, as summarized in Figure S1. The panel targeting these 189 genes was downloaded from the Human Gene Mutation Database (HGMD). The GenCap liquid‐phase targeting capture technology, developed by Mygeno Company, was utilized in the design of the panel. Double‐stranded DNA capture probes were designed to target the coding regions as well as 50 base pairs upstream and downstream. The average sequencing depth for each sample exceeded 500X, with the target region coverage reaching at least 99% at 10X and 98% at 20X.
2.3. The Establishment of a Network Cloud Platform
A multicenter data network cloud platform has been established based on the Mygeno cloud platform (https://cloud.mygeno.cn/). Each physician will have their own account to access and update clinical information about their participants. They can further share data with researchers within the cloud platform alliance through their accounts, facilitating better research and exchange in genetic medicine clinical practice in accordance with the Human Genetic Resources Administration of China (HGRAC) regulations. Project training is conducted for all participating hospitals or departments. Our multicenter communication group manages the project in three tiers: data collection, project promotion, and better genetic education, to ensure effective interaction. Basic information on 189 genes and medical genetic knowledge are shared with the physicians in the alliance. Participants' clinical information is collected by local doctors from the children's parents and entered into the database by a trained researcher, with a secondary review conducted by another experienced pediatric neurologist. After the genetic reports are completed, they undergo a third review by the local doctors, who also make corrections and add other clinical data. Genetic data includes the location, reference alleles, mutant alleles, allele frequency of all variants, exact allele count, allele frequency data for different subtypes of patients and ethnic groups, and allele frequency data referenced from another project: the ChinaMAP project (http://www.mBiobank.com). All variants can be searched by gene symbol, rs ID, genomic region, or location.
2.4. Targeted Next‐Generation Sequencing Panels and Bioinformatics Analysis
DNA samples were obtained from probands and their biological parents. Genomic DNA was extracted from whole blood using standard protocols. Illumina libraries were prepared using the DNA Sample Prep Reagent Set (MyGenostics, Beijing, China). The capture experiment and PCR product purification were conducted according to the manufacturer's protocol. Sequencing of the enrichment libraries was performed by Illumina HiSeq X ten sequencer with 150 bp paired‐end reads. After sequencing, the raw data were saved as a FASTQ format. Reads containing more than 30% low‐quality bases (quality score < 20) were removed using Cutadapt software (http://code.google.com/p/cutadapt/). Sequencing reads were aligned to the human genome build 37 (GRCh37/hg19) using the BWA algorithm of Sentieon software (https://www.sentieon.com/), and aggregated into a BAM file. PCR duplicates were removed using the “driver” function of Sentieon software, which was also used for base correction to ensure that the quality score of bases in the final output BAM file's reads more accurately reflected the real probability of mismatches with the reference genome. The mapped reads were then used for variant detection. Single nucleotide variants (SNVs) and insertions/deletions (InDels) were called using the Genome Analysis Toolkit (GATK), and the data were transformed into the Genomic Variant Call Format (gVCF). Each variant was annotated using ANNOVAR software (http://annovar.openbioinformatics.org/en/latest/) and associated with multiple databases, such as 1000 Genomes, ESP6500, dbSNP, ExAC, Inhouse (MyGenostics), HGMD, and predictions by SIFT, PolyPhen‐2, MutationTaster, GERP++.
2.5. Selection of Variants
All sequencing data underwent a rigorous quality control process, which included the following four steps: (i) Mutation reads must exceed 10, and the mutation ratio should be at least 25%; (ii) Mutations with a minor allele frequency (MAF) greater than 0.05 in the 1000 Genomes Project, ESP6500, ExAC, gnomAD and Inhouse database were excluded; (iii) Mutations present in the InNormal database (MyGenostics) were discarded; (iv) Synonymous mutations not listed in the HGMD were also excluded. Variant classification adhered to the criteria of ACMG [12]. Sanger sequencing was utilized to validate the variants and analyze their parental origins. Children with pathogenic (P) or likely pathogenic (LP) variants in genes consistent with their phenotypes and modes of inheritance were considered to have diagnostic genetic test reports. For the children with variants of uncertain significance (VUS) where the phenotype strongly suggested pathogenicity, parental analysis was conducted, allowing for the possibility of reclassification, in accordance with current recommendations [13].
2.6. Rare Variants Burden Analysis
We annotated variants in our final call set using the Ensembl Variant Effect Predictor (VEP) assigned properties such as gene name and consequence to each variant. The differences between rare variants with MAF < 0.05 on phenotypic traits were analyzed following two approaches using the software RVTESTS [14].
For gene‐based tests, we generated p value using the Sequence Kernel Association Test (SKAT). This method is based on a weighted kernel statistical test aimed at assessing the aggregated effect of multiple rare variants within a gene. First, genotype information is organized into a matrix where rows represent different samples and columns represent distinct variant sites within the gene. For each variant, genotype data is converted into numerical values and assigned appropriate weights based on factors such as variant frequency and effect size. Next, a nonparametric permutation test is performed to calculate the p value by randomizing the pairing between phenotype and genotype data, thereby constructing the distribution of the test statistic under the null hypothesis. Specifically, the SKAT statistic (which reflects the aggregated effect of rare variants within the target gene region) is computed using the actual data; then, the phenotype–genotype pairings are randomized 10,000 times to generate new pairings, with the SKAT statistic recalculated for each permutation to obtain a distribution of permutation test statistics. Finally, the observed SKAT statistic's position within the permutation distribution is determined, and the p value is calculated as the proportion of permutation statistics that are greater than or equal to the observed statistic.
For single variant‐based tests, we performed a Single Wald test to investigate the differences in the frequency of single variant between the SFS and CFS subtypes. A logistic regression model was used with genotype as the independent variable and phenotype as the dependent variable. The Wald test was employed to assess the significance of the observed regression coefficients, with the p value reflecting whether there is a statistically significant association between the genetic variants and the phenotype. We further performed statistical analysis of the p values for each gene variant and applied the Benjamini‐Hochberg method for multiple testing correction to control the false discovery rate (FDR). The corrected p values reflect the statistical significance of each gene variant between the two groups.
2.7. Analysis of Voltage‐Gated Sodium Channel Genes Variants
Voltage‐gated sodium channel gene accounted for the largest proportion of the genes detected in children with genetic diagnosis. We stratified the sodium channel genes variants according to their functionality into loss‐of‐function (LOF) or gain‐of‐function (GOF). LOF includes nonsense, frameshift, deletion, and splice site variants (including canonical ±1 splice donor site disruptions), while GOF refers to variants that result in enhanced protein function or abnormal activation. For the missense variants, we estimated the functional consequences using an in silico prediction tool (funNCion: http://funNCion.broadinstitute.org) [15]. This tool classifies variants as GoF or LoF depending on a large dataset of clinical disease phenotypes, agreement with the gold standard patch‐clamp recordings ranged from 50%–96% [16]. Then we integrated these missense variants into a free online webtool to facilitate genotype–phenotype associations interpretation (https://wlab.ethz.ch/protter/start/). We also performed the alignment of linear protein sequences using the “PER viewer” (pathogenic variant enriched regions) and the “SCN viewer” (sodium channel gene family) to aide variant analysis and prediction of phenotypes (http://per.broadinstitute.org; http://SCN‐viewer.broadinstitute.org) [16, 17].
2.8. Statistical Analysis
Socio‐demographic features and clinical characteristics were expressed as frequencies and percentages for categorical data, and as median values with interquartile ranges for continuous variables. Comparative analyses were conducted using two‐tailed Fisher's exact test, Mann–Whitney test (based on normality assessment using Kolmogorov–Smirnov and Shapiro–Wilk tests). Statistical analyses were performed in IBM SPSS 26 program. A value of p < 0.05 was considered statistically significant.
2.9. Role of the Funding Source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, writing of the report, or the decision to submit for publication.
3. Results
3.1. Characteristics of Participants
Between January 2019 and February 2023, we contacted 209 hospitals and included 3300 participants, out of which 2552 participants diagnosed with FS from 180 of these hospitals were confirmed as eligible for the study (Figure S2). The geographic distribution of these 2552 participants across 30 administrative divisions in China is depicted in Figure S3. The median age at the time of testing was 40 months (interquartile range [IQR] 25–63), with an age range of 3–166 months. Based on clinical manifestations, 1163 (45.5%) out of 2552 children were clinically diagnosed with SFS, and 1389 (54.4%) were clinically diagnosed with CFS. At enrollment, the EEG results for all patients were either normal or showed nonspecific abnormalities. Brain MRI revealed that 1650 children had no significant abnormalities, 145 children showed nonspecific findings such as widened extracerebral spaces or sinusitis, and 757 children did not undergo MRI. Regarding birth history, 2429 children had no notable complications, whereas 123 children experienced issues such as jaundice, pneumonia, or meconium aspiration during the neonatal period.
We compared the main demographic characteristics of the two clinical diagnostic subgroups to illustrate the representativeness of our sample. The results showed no significant difference in gender composition between the two groups (p = 0.104), with both groups predominantly comprising males. Children with CFS presented with a younger age at seizure onset compared to children with SFS (median = 16 months vs. 18 months, p < 0.0001). No significant statistical differences were observed in family history of seizures, ethnic groups, or referral source between the two groups (Table 1).
TABLE 1.
Participant demographics for SFS and CFS.
| SFS (n = 1163) | CFS (n = 1389) | p | |
|---|---|---|---|
| Sex | — | — | 0.104 |
| Male, n (%) | 736 (63.3) | 922 (66.4) | — |
| Female, n (%) | 427 (36.7) | 467 (33.6) | — |
| Age at seizure onset, median (IQR), months | 18 (12–26) | 16 (11–24) | < 0.0001 |
| Family history of seizures (first‐degree or second‐degree relative, n [%]) | 533 (45.8) | 649 (46.7) | 0.661 |
| Ethnic groups | — | — | 0.962 |
| Han, n (%) | 1074 (92.3) | 1282 (92.3) | — |
| Minorities, n (%) | 89 (7.7) | 107 (7.7) | — |
| Referral source | — | — | 0.052 |
| General Hospital, n (%) | 602 (51.8) | 675 (48.6) | — |
| Children's Hospital, n (%) | 457 (39.3) | 609 (43.8) | — |
| Maternity and Child Care Hospital, n (%) | 104 (8.9) | 105 (7.6) | — |
Note: Data are n (%) or median (IQR). p Value determined by two‐tailed Fisher's exact, or Mann–Whitney test. Bold values indicate statistically significant differences p < 0.05.
3.2. Screening Results for Genetic Causes
Overall, a positive result was reported in 273/2552 (11%) children in this cohort. Among the remaining 2279 undiagnosed children, 54 had no detected variants, while 2225 carried variants classified as VUS or heterozygous P/LP variants in genes associated with autosomal recessive conditions, based on ACMG guidelines. Children with genetic diagnosis had significantly earlier age at seizure onset (median = 12 months, IQR = 8–22) than children without genetic diagnosis (median = 18 months, IQR = 12–24; p < 0.0001). Children with a strong family history of seizures had a significantly higher genetic diagnostic yield (30/91, 33.0%) than those without a strong family history (243/2461, 9.9%; p < 0.0001). Notably, there was no significant difference in the diagnostic yield for genetic causes between children clinically diagnosed with SFS (9.5%, 111/1163) and those with CFS (11.7%, 162/1389; p = 0.085). In addition, no statistically significant differences were found between the groups with and without a positive genetic diagnosis in terms of gender, history of prematurity or history of birth abnormalities (Table 2).
TABLE 2.
Statistical comparison of genetic diagnosis and no genetic diagnosis.
| Category | Genetic diagnosis (n = 273) | No genetic diagnosis (n = 2279) | p |
|---|---|---|---|
| Sex | — | — | 0.097 |
| Male (n = 1658) | 165/1658 (10.0%) | 1493/1658 (90.0%) | — |
| Female (n = 894) | 108/894 (12.1%) | 786/894 (87.9%) | — |
| Premature infant (n = 147) | 12/147 (8.2%) | 135/147 (91.8%) | 0.306 |
| Age at seizure onset, months | 12 (8–22) | 18 (12–24) | < 0.0001 |
| Diagnostic yield | — | — | |
| Type of FS | — | — | 0.085 |
| SFS (n = 1163) | 111/1163 (9.5%) | 1052/1163 (90.5%) | — |
| CFS (n = 1389) | 162/1389 (11.7%) | 1227/1389 (88.3%) | — |
| Family history of seizures (first‐degree or second‐degree relative, n = 1043) | 105/1043 (10.1%) | 938/1043 (89.9%) | 0.392 |
| Strong family history of seizures (two or more first‐degree relatives; at least two generations of the family had similar conditions, n = 91) | 30/91 (33.0%) | 61/91 (67.0%) | < 0.0001 |
| History of birth abnormalities (n = 123) | 18/123 (14.6%) | 105/123 (85.4%) | 0.148 |
Note: Data are n (%) or median (IQR). p Value determined by two‐tailed Fisher's exact, or Mann–Whitney test. Bold values indicate statistically significant differences p < 0.05.
3.3. Genetic Heterogeneity
We identified 277 variants classified as pathogenic/likely pathogenic (P/LP) across 50 genes in 273 patients (including four cases with compound heterozygous mutations and one case with a homozygous mutation) within our cohort (Table S1). Out of the 50 genes with diagnostic variants, 35 (70%) were implicated more than once in these children receiving a genetic diagnosis (Figure 1A), while 15 genes were identified only once. The top 10 genetic variants (SCN1A, GABRG2, BSN, SCN1B, GABRB3, PCDH19, ATP1A2, CACNA1A, SCN9A, and KCNQ2) accounted for 70.4% of all genetic diagnoses. Of these, pathogenic variants in the SCN1A gene alone accounted for 41.5% (115/277) of the diagnostic yield. Out of the 277 diagnostic variants, 268 (96.8%) were single nucleotide variants or small insertions/deletions, and nine (3.2%) were copy number variants (Figure 1B). These diagnostic variants were mostly associated with an autosomal dominant inheritance pattern, while 21 variants had X‐linked inheritance and three variants had autosomal recessive inheritance. Furthermore, 134 of these diagnostic variants were novel, having not been previously reported in the literature.
FIGURE 1.

Distribution of genes with diagnostic variants. (A) 35 genes reoccurred at least two times in the children with the number of variants in each gene. (B) Types of diagnostic variants. (C) Genetic diagnoses arranged by age at seizure onset within the two clinical diagnostic groups. Blue boxes represent the group initially diagnosed with complex febrile seizures (CFS), and orange boxes represent the group initially diagnosed with simple febrile seizures (SFS). The frequency of each recurrent gene variant (n ≥ 2) is shown within the boxes. Color intensity reflects the frequency of variants. Darker shades indicate higher frequency, and lighter shades indicate lower frequency. A total of 27 overlapping genes identified across different age groups in the two clinical diagnostic groups are highlighted with black borders.
Among the 273 patients with genetic causes, 111 were clinically diagnosed with SFS and 162 with CFS. Genetic diagnoses arranged by age at seizure onset within the two groups are depicted in Figure 1C. Among them, 27 genes with diagnostic variants showed a significant overlap across different ages between the two clinical diagnostic subgroups. The remaining 23 genes exhibited differences, with 9 genes exclusively detected in the SFS group and 14 solely in the CFS group. Genes identified in the CFS group are associated with a broader range of biological pathways than those in the SFS group (Table S2).
3.4. Onset Age Distribution Characteristics of High Frequency Pathogenic Genes
We further analyzed the relationship between the top 10 genes (by diagnostic yield) and age at seizure onset, aiming to provide reference for prioritizing potential genetic causes based on onset age (Figure 2). Among these genes, SCN1A and SCN1B were involved across all age groups. The SCN1A variant group had the earliest onset, with ages primarily clustered in the lower infancy range (median = 6 months, IQR = 3–9). In contrast, the SCN1B variant group showed a significantly later onset (median = 18 months, IQR = 12–24), with broader individual variation. Additionally, the first seizure onset ages for GABRG2, GABRB3, PCDH19, and SCN9A variants were relatively similar, falling between SCN1A and SCN1B. Finally, BSN, ATP1A2, CACNA1A, and KCNQ2 variants were associated with later seizure onset, with median ages of 22 months (IQR = 18–28), 24 months (IQR = 20–28), 26 months (IQR = 22–32), and 20 months (IQR = 16–26), respectively.
FIGURE 2.

Age at seizure onset associated with the top 10 genes by diagnostic yield. Each box plot represents a pathogenic gene group, distinguished by different colors. The y‐axis indicates age at seizure onset (in months), and the x‐axis shows gene names. The plots display the median age and interquartile range (IQR) for each gene group, with outliers marked accordingly.
3.5. Rare Variants Burden Analysis
Previous genome‐wide association studies have identified several common genetic variants associated with FS [18, 19]. We conducted genotype–phenotype association analyses for all rare variants to preliminarily explore whether rare variants contribute to the phenotypic complexity of FS. A total of 698,714 variants was compiled from 2552 children who met the quality control criteria in gene panel sequencing data, excluding variants of inferior quality. Utilizing RVTESTS software with a sequencing depth of at least 30X and a minor allele frequency (MAF) threshold of less than 0.05, a subset of 48,084 rare variants was discerned. Gene‐level analysis via SKAT revealed that coding variants in five genes were significantly different between the FS subtypes (Table S3). We performed a Single Wald test to compare the variant frequencies of 189 genes between the SFS and CFS groups. The results showed that the differences between the two groups did not reach statistical significance, with Benjamini−Hochberg‐corrected p values close to 0.05 (Table S6).
We further analyzed the 1869 rare variants from Wald test with nominal significance (p < 0.05) to explore the potential contribution of single variant to phenotypic differences in clinical subgroups (Figure 3A). The RBFOX1 gene contained the highest frequency of variants (n = 176), and SLC12A5 exhibited the highest density of variants per kilobase (kb) (Figure 3B). Moreover, six variants are located in the coding region, predicted to have a high or moderate impact on gene function and/or expression (Figure 3C). Of these, two variants were more frequently found in the SFS subtype. Conversely, four variants showed a higher frequency in the CFS subtype (Table S4). Furthermore, within the five genes identified by SKAT, 47 variants showed significant frequency differences between subgroups, including three synonymous variants in SLC12A5 and 44 variants located in intronic, upstream, or downstream regions potentially affecting regulatory functions. These results further supported the findings of the SKAT analysis.
FIGURE 3.

Results of single‐variant Wald tests in rare variant analysis. (A) Classification of 1869 variants with nominal significance based on genomic function. (B) The top 20 genes ranked by the frequency of variants (left) and the density of variants per kb of gene sequence (right). (C) The forest plot depicts six variants located within coding regions that exhibit significant differences between the FS subtypes. Variants demonstrating protective effects on the FS phenotype are marked in green, while those indicating increased risks for the complex phenotype are highlighted in red. The dashed line represents a beta value of 0, serving as a reference for distinguishing between risk‐increasing and risk‐reducing variants.
3.6. Genetic Landscape of Variants in Voltage‐Gated Sodium Channel Genes
Among the patients with genetic diagnosis, patients with a sodium channel gene (SCN1A, SCN2A, SCN3A, SCN8A, and SCN9A) variant accounted for the largest proportion of 47.6% (130/273). To facilitate prediction of phenotype, we further investigate the functional type and location of all 130 variants within the sodium channel genes. Protein‐truncating variants are known to cause complete LOF [20] and relate to severe phenotypes. For the missense variants, the functional consequences were assessed using the tool “funNCion”. Consequently, 101 variants are classified as LOF (66 missense, 16 nonsense, 12 frameshift, four splicing, and three deletion) and 23 variants are classified as GOF (23 missense).
A 2D illustration visually showed the distribution of these missense variants across the SCN1A protein structure (Figure 4A). Additionally, six missense variants within SCN2A, SCN8A, and SCN9A were not shown due to their low frequency. Out of these 89 missense variants, 60 associated with more severe phenotypes and 29 with milder phenotypes. In the fast inactivation gate (D3‐4 linker) and conserved N‐terminus, all seven variants were associated with more severe phenotypes (Figure 4B). All five variants found in the cytoplasmic side (D1‐2 linker) were associated with milder phenotypes. Variants with more severe phenotypes (46.7%, 28/60) appeared more frequently across all four S5‐6 pore‐loop regions compared to variants with milder phenotypes (31.0%, 9/29). The region‐specific distribution of variants with more severe phenotypes in comparison to variants with milder phenotypes is statistically significant (two‐tailed Fisher's exact test, p = 0.008). However, when grouped according to initial clinical diagnosis (SFS and CFS), no differences were observed in this region‐specific distribution (p = 0.868).
FIGURE 4.

Domain distribution of sodium channel gene variants. (A) The alpha subunit consists of four homologous domains (D1–D4), each formed of six transmembrane segments (S1–S6). The location of SCN1A missense variants with GOF or LOF is displayed as different colored dots (Visualization link: https://3mw.cn/8m8e). (B) Percentage stacked histogram of severe versus mild and LOF versus GOF variants across different protein parts.
Collectively, these 89 missense variants primarily occurred in the homologous domains D1‐4 of the sodium channel proteins, with very few variants occurring in cytoplasmic regions (N‐terminus, D1‐2 linker). Variants distributed across all four S5‐6 pore‐loop regions included both LOF (42.4%, 28/66) and GOF variants (39.1%, 9/23). However, GOF variants appeared more frequently in the voltage‐sensing domains (including S4 and S4‐5) compared to LOF variants (39.1% [9/23] vs. 10.6% [7/66]; p = 0.004; Figure 4B).
3.7. Clinical Utility of Molecular Diagnostics
Children with positive genetic findings are referred to pediatric neurology (Figure 5). Clinical utility was demonstrated for all 277 variants identified with diagnostic findings by providing early molecular diagnoses, guiding clinical management, optimizing risk assessment, and helping to avoid treatment delays or the use of inappropriate medications (Table S5). The diagnostic findings from 211 (76.2%) of the 277 variants influenced therapeutic management strategies, predominantly optimizing follow‐up monitoring protocols and the choice of antiseizure medication. Moreover, the diagnostic findings from 201 (72.6%) of the 277 variants provided further prognostic implications beyond what was determined by the initial FS diagnosis. Importantly, genetic findings for 148 (53.4%) of the 273 children advised genetic testing for other family members.
FIGURE 5.

Workflow of genetic screening in children with febrile seizures. Pediatricians at participating centers conduct clinical evaluations of children meeting the inclusion criteria for febrile seizures and classify them as simple febrile seizures (SFS) or complex febrile seizures (CFS). Blood samples are collected and sent to a third‐party laboratory for genetic panel testing. The test results are shared via a cloud‐based data platform. After receiving the genetic report, each participant undergoes a follow‐up evaluation, during which trained physicians interpret the results. Children with identified genetic etiologies receive a corresponding molecular diagnosis and are referred to pediatric neurology. Individualized management plans and genetic counseling are provided based on clinical presentation. For variants of uncertain significance (VUS) with high clinical suspicion, family‐based segregation analysis and data reanalysis are performed to clarify their clinical relevance.
3.8. Efficiency
In this study, the cost of a single panel sequencing for a three‐person family was $240, averaging $80 per individual, with additional family validation or data reanalysis provided free of charge. The panel achieved an average sequencing depth of over 500X for target genes, with a coverage rate of over 98% at ≥ 20×, and a turnaround time of 14 days. We compared the diagnostic performance of the panel and WES for SCN1A, the most frequently detected hotspot gene in neurological disorders, to reflect the overall efficiency difference between the two methods. According to the Mygeno Inhouse database, WES achieves > 100× depth for SCN1A with 97% coverage at ≥ 20×, but the turnaround time typically exceeds 2 months and costs approximately $1000 per test.
4. Discussion
This nationwide, multicenter prospective cohort study is the first to systematically evaluate the potential genetic etiologies and distribution of pathogenic genes in Chinese children with FS, revealing the clinical utility of molecular diagnosis in precision medicine for this population. The study confirmed the feasibility of gene panel sequencing in large cohorts and underscored its significant advantages as a rapid clinical molecular diagnostic tool.
Just as the diverse etiologies for epileptic encephalopathies, it is likely that there are multiple etiologies of FS. For some children, the etiology of FS also contributes to subsequent afebrile seizures and/or psychiatric and associated morbidities [9]. After witnessing their child's seizure, the majority of parents express a keen interest in accessing early, rapid, and economical genetic testing to identify the underlying cause [21]. Our study revealed a molecular diagnostic yield of 11% using the panel encompassing 189 genes associated with fever‐induced seizures within the FS cohort. Since its design in 2019, our gene panel has not been updated, primarily due to the impact of the COVID‐19 pandemic on research progress and a decrease in pediatric visit rates, which delayed the planned sample recruitment and data analysis. Moreover, as a multi‐center prospective cohort study, the core objective in the early stage was to validate the detection performance of the existing panel. Previous research indicates that the diagnostic yield of the NGS panel appears to be determined by the number of commonly mutated genes included rather than the total number of genes in the panel [22]. This conclusion was confirmed by our findings: the top 10 genes accounted for 70.4% of all confirmed diagnoses, underscoring the dominant role of specific genes in this disease phenotype spectrum. For patients with isolated FS or without significant neurodevelopmental abnormalities, the current panel may be sufficient, as it has a short turnaround time—making it an accessible molecular diagnostic tool for outpatient settings. In contrast, for patients with coexisting neurodevelopmental disorders or other epilepsy‐related phenotypes, whole‐exome or whole‐genome sequencing may be chosen for a more comprehensive genetic evaluation.
Patients with variants in sodium channel genes accounted for the largest proportion of all genetically diagnosed cases. Variants occurring in the fast inactivation gate, pore‐loop regions, and conserved terminal regions are frequently linked to more severe phenotypes and earlier onset of seizures. These findings are similar to previous studies [16]. However, this genetic pattern was not observed in groupings of clinical subtypes, indicating that reliance solely on clinical presentations might not capture the genetic intricacy and individual variability of the condition. To take full advantage of genetic information for refined clinical classification of patients, we extended genotype–phenotype correlation to the overall rare genetic variability obtained by NGS analysis in an FS cohort. Rare variants were found to have larger effects on both the frequency and likelihood of pathogenicity than common variants [23]. Both gene‐level association studies and Single Wald tests revealed that the burden of rare variants with nominal significance in the SLC12A5 gene is associated with the FS phenotype, with SLC12A5 showing the highest variant density per kb, suggesting its role as a genetic susceptibility locus for FS. Furthermore, the rs2297201 polymorphism in the SLC12A5 gene has been identified as a predisposing factor for FS [24].
Patients with SCN1A variants accounted for the largest proportion of all genetically diagnosed cases. Although widely recognized as a hotspot gene for epilepsy, its detection also facilitated the early identification of Dravet syndrome and GEFS+, which are the most challenging to differentiate from FS [25, 26]. Particularly in patients with Dravet syndrome, the conditions may mimic CFS during their first year [26]. A retrospective study reported a significant diagnostic delay, with the mean time from seizure onset to an accurate diagnosis being 7.4 years [27]. GEFS+ is an epilepsy syndrome traditionally diagnosed at the family level [28]. However, recent studies have shown that GEFS+ can result from polygenic inheritance or de novo mutations [29]. The condition often exhibits incomplete penetrance and phenotypic variability, making it prone to being underdiagnosed in families with only a few affected individuals [30]. For patients from small families or without a clear family history, we prefer molecular diagnosis based on reported variant sites, gene function, and mutation type to guide more precise treatment. For patients with a family history but no definitive diagnosis, additional functional validation, family extension analysis, and regular reanalysis of genetic data may be considered.
In our cohort, 76.2% of diagnostic findings influenced clinical management, while 72.6% provided additional information for prognostic assessment. These findings have the potential to transform current diagnostic and treatment practices by shifting the “empiric therapy” model toward individualized management. For example, when an SCN1A variant indicates a high risk of Dravet syndrome, sodium channel blockers, such as carbamazepine, should be avoided because they may exacerbate seizures. Although not all children with FS will develop genetic epilepsy, for individuals identified by genetic testing as potentially high‐risk, dynamically monitoring clinical evolution will allow for prompt adjustments in management strategies upon the appearance of early epileptic features.
Notably, some children initially classified as SFS were found to carry pathogenic variants associated with a less favorable prognosis. In routine practice, children with SFS generally undergo no further evaluation once the cause of fever has been identified. In such cases, a molecular diagnosis may prompt neurological follow‐up and closer monitoring for evolving epileptic features that might otherwise be overlooked. In CFS group, the identification of pathogenic genes overcomes the limitations of traditional risk factor‐based assessments (such as seizure duration or focal features), enabling early identification of genetically high‐risk individuals and the initiation of personalized monitoring. In addition, genetic testing has significant implications for reproductive counseling, as it applies to both inherited variants, thereby providing a scientific basis for reproductive interventions.
5. Limitations
Firstly, accurately interpreting genetic test results remains challenging. Family segregation analysis and epigenetics further increase the difficulty of interpretation. Additionally, VUS may increase the psychological burden for families. Personalized genetic counseling, psychological support, and regular follow‐up are essential to address this. Secondly, only the short‐term utility of genetic testing is reported here. Longitudinal follow‐up is needed to evaluate the impact on clinical outcomes, quality of life, and economic efficiency. Understanding parental perspectives can support better decision‐making about NGS panel use. Lastly, all findings are based on a predominantly Chinese cohort, limiting generalizability to other ethnic groups due to differences in allele frequencies and disease incidence. Economic factors, insurance coverage, and differences in awareness of genetic testing also influence testing uptake. Future research should validate these findings in diverse populations.
6. Conclusions
Overall, based on this large‐scale, multi‐center cohort study, our findings demonstrate the potential of molecular diagnostics in advancing precision medicine for FS and highlight the prospects of more accurately classifying patients using genetic information. According to our findings, FS may not constitute an independent disease clinically but rather represent an early phenotype of multiple disorders. In an era of rapidly developing precision medicine, we advocate for gene panel as an accessible molecular diagnostic tool for outpatient settings for FS, a condition with a high incidence rate.
Author Contributions
Concept and design: Liping Zou, Jian Wu, Jia Wang, Mengna Zhang, Yangyang Wang, and Xiuyu Shi. Acquisition, analysis, or interpretation of data: Liping Zou, Jian Wu, Jia Wang, Mengna Zhang, Yangyang Wang, Xiuyu Shi, Baoguang Li, Xiuxia Wang, Dan Sun, Sanqing Xu, Jianbo Zhao, Lijun Du, Qiming Pang, Xiaoming Liu, Chunhui Tang, Xiaohua Li, Jiuwei Li, Yanli Ma, Jianxiang Liao, Gulixian Abudureyimu, Jianmin Liang, Yuan Wang, and De Wu. Drafting of the manuscript: Liping Zou, Jia Wang, Mengna Zhang, and Yangyang Wang. Critical revision of the manuscript for important intellectual content: Liping Zou, Jian Wu, Jia Wang, Mengna Zhang, Yangyang Wang, and Xiuyu Shi. Statistical analysis: Liping Zou, Jia Wang, Mengna Zhang, and Yangyang Wang. Administrative, technical, or material support: Liping Zou, Jian Wu, Baoguang Li, Xiuxia Wang, Dan Sun, Sanqing Xu, Jianbo Zhao, Lijun Du, Qiming Pang, Xiaoming Liu, Chunhui Tang, Xiaohua Li, Jiuwei Li, Yanli Ma, Jianxiang Liao, Gulixian Abudureyimu, Jianmin Liang, Yuan Wang, De Wu. Supervision: Jia Wang, Mengna Zhang, Yangyang Wang, Xiuyu Shi, Baoguang Li, Xiuxia Wang, Dan Sun, Sanqing Xu, Jianbo Zhao, Lijun Du, Qiming Pang, Xiaoming Liu, Chunhui Tang, Xiaohua Li, Jiuwei Li, Yanli Ma, Jianxiang Liao, Gulixian Abudureyimu, Jianmin Liang, Yuan Wang, De Wu, Jian Wu, and Liping Zou. All authors reviewed and edited the manuscript, as well as had full access to all the data in the study and accept responsibility to submit for publication.
Funding
This work was supported by the National Key Research and Development Program of China, 2023YFC2706304; Innovation Talent Fund of Senior Department of Pediatrics, The Seventh Medical Center of PLA General Hospital, QZX‐04‐EKLHJH‐2.
Ethics Statement
This study is performed according to the international rules of acceptable clinical trials, and informed consent is obtained from the parents of the patient. This study was approved by the Ethical Committee of Chinese PLA General Hospital (S2019‐028‐01).
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1: Genes included in the panel.
Figure S2: Flowchart of cohort development.
Figure S3: Geographic distribution of 2552 participants over 30 administrative divisions in China.
Table S1: Genetic diagnoses made by NGS panel.
Table S2: Twenty‐three genes exhibiting heterogeneity between SFS and CFS and their associated biological pathways.
Table S3: The association between variants and the FS subtypes by SKAT.
Table S4: Wald test in difference analysis of coding variants between SFS and CFS.
Table S5: Utility of genetic diagnoses.
Table S6: Results of the single‐variant association analysis between simple and complex febrile seizures.
Acknowledgments
We thank the patients and their families for their participation and cooperation in this study. We thank all the clinicians who completed case report forms for patients included in the study. No one was financially compensated for their contribution. The multi‐center data network cloud platform is supported by Mygeno Company. This study was supported by the National Key Research and Development Program of China (Grant number: 2023YFC2706304) and the Innovation Talent Fund of the Senior Department of Pediatrics, The Seventh Medical Center of PLA General Hospital (QZX‐04‐EKLHJH‐2).
Funding Statement
This work was funded by National Key Research and Development Program of China grant 2023YFC2706304; Innovation Talent Fund of Senior Department of Pediatrics, The Seventh Medical Center of PLA General Hospital grant QZX‐04‐EKLHJH‐2.
Contributor Information
Jian Wu, Email: jw2231@mygeno.cn.
Liping Zou, Email: zouliping21@hotmail.com.
Data Availability Statement
Deidentified data information included in this analysis and NGS results reported for the participants with genetic diagnosis are provided in Appendix S1. The study protocol is available after publication by request. Requests should be addressed to Liping Zou (zouliping21@hotmail.com).
References
- 1. Subcommittee on Febrile S, American Academy of P , “Neurodiagnostic Evaluation of the Child With a Simple Febrile Seizure,” Pediatrics 127, no. 2 (2011): 389–394, 10.1542/peds.2010-3318. [DOI] [PubMed] [Google Scholar]
- 2. Leung A. K. and Robson W. L., “Febrile Seizures,” Journal of Pediatric Health Care 21, no. 4 (2007): 250–255, 10.1016/j.pedhc.2006.10.006. [DOI] [PubMed] [Google Scholar]
- 3. Sillanpaa M., Camfield P., Camfield C., et al., “Incidence of Febrile Seizures in Finland: Prospective Population‐Based Study,” Pediatric Neurology 38, no. 6 (2008): 391–394, 10.1016/j.pediatrneurol.2008.02.006. [DOI] [PubMed] [Google Scholar]
- 4. Xixis K. L., Samanta D., Smith T., and Keenaghan M., Febrile Seizure (StatPearls, Ineligible Companies. Disclosure: Debopam Samanta Declares No Relevant Financial Relationships With Ineligible Companies. Disclosure: Travis Smith Declares No Relevant Financial Relationships With Ineligible Companies. Disclosure: Michael Keenaghan Declares No Relevant Financial Relationships With Ineligible Companies, 2024). [Google Scholar]
- 5. Capovilla G., Mastrangelo M., Romeo A., and Vigevano F., “Recommendations for the Management of “Febrile Seizures”: Ad Hoc Task Force of LICE Guidelines Commission,” Epilepsia 50, no. Suppl 1 (2009): 2–6, 10.1111/j.1528-1167.2008.01963.x. [DOI] [PubMed] [Google Scholar]
- 6. Cross J. H., “Fever and Fever‐Related Epilepsies,” Epilepsia 53, no. Suppl 4 (2012): 3–8, 10.1111/j.1528-1167.2012.03608.x. [DOI] [PubMed] [Google Scholar]
- 7. Annegers J. F., Hauser W. A., Shirts S. B., and Kurland L. T., “Factors Prognostic of Unprovoked Seizures After Febrile Convulsions,” New England Journal of Medicine 316, no. 9 (1987): 493–498, 10.1056/NEJM198702263160901. [DOI] [PubMed] [Google Scholar]
- 8. Lee S. H., Byeon J. H., Kim G. H., Eun B. L., and Eun S. H., “Epilepsy in Children With a History of Febrile Seizures,” Korean Journal of Pediatrics 59, no. 2 (2016): 74–79, 10.3345/kjp.2016.59.2.74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Mewasingh L. D., Chin R. F. M., and Scott R. C., “Current Understanding of Febrile Seizures and Their Long‐Term Outcomes,” Developmental Medicine and Child Neurology 62, no. 11 (2020): 1245–1249, 10.1111/dmcn.14642. [DOI] [PubMed] [Google Scholar]
- 10. Adams D. R. and Eng C. M., “Next‐Generation Sequencing to Diagnose Suspected Genetic Disorders,” New England Journal of Medicine 379, no. 14 (2018): 1353–1362, 10.1056/NEJMra1711801. [DOI] [PubMed] [Google Scholar]
- 11. Consugar M. B., Navarro‐Gomez D., Place E. M., et al., “Panel‐Based Genetic Diagnostic Testing for Inherited Eye Diseases Is Highly Accurate and Reproducible, and More Sensitive for Variant Detection, Than Exome Sequencing,” Genetics in Medicine 17, no. 4 (2015): 253–261, 10.1038/gim.2014.172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Richards S., Aziz N., Bale S., et al., “Standards and Guidelines for the Interpretation of Sequence Variants: A Joint Consensus Recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology,” Genetics in Medicine 17, no. 5 (2015): 405–424, 10.1038/gim.2015.30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Chen D., “A Chinese Interpretation for the “ACGS Best Practice Guidelines for Variant Classification in Rare Disease 2020”,” Zhonghua Yi Xue Yi Chuan Xue Za Zhi 40, no. 8 (2023): 915–921, 10.3760/cma.j.cn511374-20221017-00692. [DOI] [PubMed] [Google Scholar]
- 14. Zhan X., Hu Y., Li B., Abecasis G. R., and Liu D. J., “RVTESTS: An Efficient and Comprehensive Tool for Rare Variant Association Analysis Using Sequence Data,” Bioinformatics 32, no. 9 (2016): 1423–1426, 10.1093/bioinformatics/btw079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Heyne H. O., Baez‐Nieto D., Iqbal S., et al., “Predicting Functional Effects of Missense Variants in Voltage‐Gated Sodium and Calcium Channels,” Science Translational Medicine 12, no. 556 (2020): eaay6848, 10.1126/scitranslmed.aay6848. [DOI] [PubMed] [Google Scholar]
- 16. Brunklaus A., Feng T., Brunger T., et al., “Gene Variant Effects Across Sodium Channelopathies Predict Function and Guide Precision Therapy,” Brain 145, no. 12 (2022): 4275–4286, 10.1093/brain/awac006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Pérez‐Palma E., May P., Iqbal S., et al., “Identification of Pathogenic Variant Enriched Regions Across Genes and Gene Families,” Genome Research 30, no. 1 (2020): 62–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Skotte L., Fadista J., Bybjerg‐Grauholm J., et al., “Genome‐Wide Association Study of Febrile Seizures Implicates Fever Response and Neuronal Excitability Genes,” Brain 145, no. 2 (2022): 555–568, 10.1093/brain/awab260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Feenstra B., Pasternak B., Geller F., et al., “Common Variants Associated With General and MMR Vaccine‐Related Febrile Seizures,” Nature Genetics 46, no. 12 (2014): 1274–1282, 10.1038/ng.3129. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Rivas M. A., Pirinen M., Conrad D. F., et al., “Human Genomics. Effect of Predicted Protein‐Truncating Genetic Variants on the Human Transcriptome,” Science 348, no. 6235 (2015): 666–669, 10.1126/science.1261877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. D'Gama A. M., Mulhern S., Sheidley B. R., et al., “Evaluation of the Feasibility, Diagnostic Yield, and Clinical Utility of Rapid Genome Sequencing in Infantile Epilepsy (Gene‐STEPS): An International, Multicentre, Pilot Cohort Study,” Lancet Neurology 22, no. 9 (2023): 812–825, 10.1016/S1474-4422(23)00246-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Carvill G. L., Heavin S. B., Yendle S. C., et al., “Targeted Resequencing in Epileptic Encephalopathies Identifies De Novo Mutations in CHD2 and SYNGAP1,” Nature Genetics 45, no. 7 (2013): 825–830, 10.1038/ng.2646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Kryukov G. V., Pennacchio L. A., and Sunyaev S. R., “Most Rare Missense Alleles Are Deleterious in Humans: Implications for Complex Disease and Association Studies,” American Journal of Human Genetics 80, no. 4 (2007): 727–739, 10.1086/513473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Dimitrijevic S., Jekic B., Cvjeticanin S., et al., “KCC2 rs2297201 Gene Polymorphism Might Be a Predictive Genetic Marker of Febrile Seizures,” ASN Neuro 14 (2022): 17590914221093257, 10.1177/17590914221093257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Camfield P. and Camfield C., “Febrile Seizures and Genetic Epilepsy With Febrile Seizures Plus (GEFS+),” Epileptic Disorders 17, no. 2 (2015): 124–133, 10.1684/epd.2015.0737. [DOI] [PubMed] [Google Scholar]
- 26. Korff C., Laux L., Kelley K., Goldstein J., Koh S., and D. Nordli, Jr. , “Dravet Syndrome (Severe Myoclonic Epilepsy in Infancy): A Retrospective Study of 16 Patients,” Journal of Child Neurology 22, no. 2 (2007): 185–194, 10.1177/0883073807300294. [DOI] [PubMed] [Google Scholar]
- 27. Bremer A., Lossius M. I., and Nakken K. O., “Dravet Syndrome—Considerable Delay in Making the Diagnosis,” Acta Neurologica Scandinavica 125, no. 5 (2012): 359–362, 10.1111/j.1600-0404.2011.01609.x. [DOI] [PubMed] [Google Scholar]
- 28. Scheffer I. E. and Berkovic S. F., “Generalized Epilepsy With Febrile Seizures Plus. A Genetic Disorder With Heterogeneous Clinical Phenotypes,” Brain 120, no. Pt 3 (1997): 479–490. [DOI] [PubMed] [Google Scholar]
- 29. Zhang Y. H., Burgess R., Malone J. P., et al., “Genetic Epilepsy With Febrile Seizures Plus: Refining the Spectrum,” Neurology 89, no. 12 (2017): 1210–1219. [DOI] [PubMed] [Google Scholar]
- 30. J. Engel, Jr. and International League Against E , “A Proposed Diagnostic Scheme for People With Epileptic Seizures and With Epilepsy: Report of the ILAE Task Force on Classification and Terminology,” Epilepsia 42, no. 6 (2001): 796–803. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Genes included in the panel.
Figure S2: Flowchart of cohort development.
Figure S3: Geographic distribution of 2552 participants over 30 administrative divisions in China.
Table S1: Genetic diagnoses made by NGS panel.
Table S2: Twenty‐three genes exhibiting heterogeneity between SFS and CFS and their associated biological pathways.
Table S3: The association between variants and the FS subtypes by SKAT.
Table S4: Wald test in difference analysis of coding variants between SFS and CFS.
Table S5: Utility of genetic diagnoses.
Table S6: Results of the single‐variant association analysis between simple and complex febrile seizures.
Data Availability Statement
Deidentified data information included in this analysis and NGS results reported for the participants with genetic diagnosis are provided in Appendix S1. The study protocol is available after publication by request. Requests should be addressed to Liping Zou (zouliping21@hotmail.com).
