Febrile seizures (FS) occur in about 2–3% of children from 3 months to 5 years of age with a phenotype outcome that varies from simple febrile seizures to Dravet-like syndrome and epileptic encephalopathy (Sharkov, 2020). Atypical features of febrile seizures such as focal semiology, prolonged episodes (>15 min), or multiple seizure types within the same fever episode are of increased risk of developing epilepsy. After the third febrile seizure, the risk of additional episodes is approaching 50% and the risk of epilepsy formation is up to 15.8% (Bertelsen et al., 2016). Recent studies showed the contribution of genetic causes in the development of febrile-associated epilepsy including inborn errors of metabolism (mitochondrial, peroxisomal, lysosomal diseases, organic acidurias, aminoacidopathies, glycosylation, and urea cycle disorders), monogenic early-onset epileptic encephalopathies, and epilepsy phenotype caused by copy number variations (CNVs) such as the Cornelia de Lange, Seckel, and Rubinstein–Taybi syndromes (Dadali et al., 2016; Wang et al., 2017; Deng et al., 2018).

Febrile onset of epilepsy or seizure exacerbation during febrile/afebrile illness is observed among patients with pathogenic variants in genes that are associated with epileptic and developmental encephalopathies (DEEs) including PRRT2, STX1B, and PCDH19, and genes encoding voltage-gated sodium channels such as SCN1A, SCN2A, SCN3A, and SCN8A are known as the cause for fever-associated seizures or epilepsy (Deng et al., 2018; Yokoi et al., 2018; Shibata et al., 2021).

However, the SCN1A gene is the most common cause of febrile-associated epilepsy among early-onset DEE (Verbeek et al., 2013; Symonds et al., 2019). SCN1A-related epilepsy is often represented by Dravet syndrome (DS) which typically presents in the first year of life in a normal child with prolonged, afebrile and febrile (triggered by mild fever or hot bath), focal (usually hemiclonic), and generalized tonic–clonic seizures. The first seizure occurs before 12 months of age in over 90% of cases, usually between 4 and 8 months (Gataullina and Dulac, 2017). Yet in rare cases, a later onset of seizures was described in patients up to 18–20 months of age (Wirrell et al., 2017).

The onset of myoclonic seizures is typically by the age of two in most cases. Non-convulsive status, focal seizures with impaired awareness, and atypical absence seizures generally occur after 2 years. Between 1 and 5 years of age, in the “worsening stage” of DS, motor seizures become more frequent but shorter, although their severity is still linked to mild hyperthermia.

Patients with DS have normal development prior to seizure onset, and neurological abnormalities typically become evident in 3–4 years of age and include crouched gait, hypotonia, incoordination, and impaired dexterity. Seizures are typically exacerbated with the use of sodium channel blocking drugs such as carbamazepine, oxcarbazepine, phenytoin, and lamotrigine (Gataullina and Dulac, 2017; Wirrell et al., 2017).

Here, we report a patient with epileptic and developmental symptoms similar to those of SCN1A-related encephalopathy but with an unusually good response to sodium channel blockers and his diagnostic odyssey.

Patients with KS have a pathogenic or likely pathogenic variant in KMT2D or KDM6A genes and distinctive facial features. The most commonly occurring features of KS are long palpebral fissures, eyelid ectropion, and fingertip pads. A large majority of people with KS are mildly or moderately intellectually disabled from birth, although some patients may have severe intellectual disability (ID). In a minority of patients the intelligence quotient (IQ) may be within the normal range as reported in (Matsumoto and Niikawa, 2003).

Boniel et al. demonstrated that patients with KS generally have low IQ levels and upon testing with the CARS, ADOS, and ADI-R scales (used in autism spectrum disorders), they tend to fulfill the criteria for autism. The prevalence of epilepsy in patients with KS has varied between 5 and 16–36% according to different sources. Seizure types most commonly include focal seizures, bilateral tonic–clonic seizures, and myoclonus. There are no reports of patients with KS and epileptic encephalopathies (Boniel et al., 2021).

Although, the segregation analysis confirmed the de novo status of the c.2831dupA variant in the KDM6A gene and the variant is predicted to lead to LoF which is in agreement with the molecular mechanism of KS we believe that the diagnosis could not be confirmed based solely on genetic data. The absence of typical clinical signs of KS and signs of epileptic encephalopathy in the proband raise doubts concerning its causality. The patient had only mild dysmorphic facial features (Figure 1E), had normal motor and near-normal developmental milestones until seizures onset, and had none of the supportive symptoms according to the international consensus diagnostic criteria of KS (Adam et al., 2019).

This incomplete penetrance in our patient could be explained by the almost complete skewed XCI, which is in agreement with previously reported clinically unaffected female carriers of a pathogenic variant in KDM6A (Lederer et al., 2014; Khodaeian et al., 2021). Several studies reported that even though the KDM6A gene escapes inactivation the expression level from the inactivated X-chromosome is significantly lower (Xu et al., 2008; Lederer et al., 2012). In this scenario, in our patient, there are near-normal levels of WT KDM6A protein which can rescue the phenotype. However, we could not fully exclude that the frame shifting variant in KDM6A, even with such skewed XCI may affect and modify the epilepsy phenotype in our patient.

The second reported variant p. His127Asp in the SCN1A gene was previously described in patients with DS and generalized epilepsy with the febrile seizures plus (GEFS+) phenotype (Zuberi et al., 2011; Cetica et al., 2017; Lindy et al., 2018; Brunklaus et al., 2020; Minardi et al., 2020). Nowadays, DS can be diagnosed among patients who met at least four of the five inclusion criteria: 1) normal or near-normal cognitive and motor development before seizure onset; 2) ≥2 febrile or afebrile seizures before 1 year of age; 3) seizure semiology consisting of myoclonic, hemiclonic, or generalized tonic-clonic seizures; 4) ≥2 seizures lasting longer than 10 min; and 5) failure to respond to first-line antiepileptic drug therapy with continued seizures after 2 years of age (Wu et al., 2015).

An additional feature of SCN1A-DS is an exacerbation of seizures during the use of sodium channel blocking drugs. Our patient had an unusually good response to carbamazepine. Also, we did not observe a non-convulsive status and atypical absence which generally occur after 2 years. Moreover, the proband had severe muscle hypotonia with increased tendon reflexes which is also rarely observed in SCN1A-related epilepsy.

Based on the fact that the variant in the SCN1A gene was observed 20 times in a heterozygous state in the gnomAD database, the fact that the proband father was an unaffected carrier and that the clinical picture was unusual for SCN1A-related epilepsy a different genetic etiology was suspected.

The favorable response to sodium channel inhibitors was suggestive for the SCN2A gene where in the previous analysis the intronic variant was missed. SCN2A encodes the pore-forming protein type 2 alpha subunit Nav1.2 of neuronal voltage-gated sodium channels. NaV1.2 is widely expressed throughout the human central nervous system, predominately in excitatory, glutamatergic neurons. It is located in the initial segment of the axons and is involved in the initiation and propagation of action potentials in a range of neuron classes (Boiko et al., 2003; Sanders et al., 2018).

SCN2A pathogenic variants have been identified as a prominent cause of a wide range of conditions, from benign neonatal or infantile seizures to neurodevelopmental disorders, including ASD, ID, and infantile-onset seizures of varying severity (Syrbe et al., 2016).

Also, few patients with unusual courses of diseases such as schizophrenia (Carroll et al., 2016) and recurrent ataxia (Leach et al., 2016) were reported.

SCN2A-epilepsy was divided into two groups depending on seizure onset: early infantile epilepsies (<3 months) and later onset (≥3 months), and described favorable effects to sodium channel blockers. Even in the late-onset epilepsy group, such medications were less effective and in some cases, induced seizure worsening (Wolff et al., 2017). Some patients with SCN2A-epilepsy have seizure exacerbation with intercurrent febrile or afebrile illnesses, including Dravet-like syndrome phenotype (Shi et al., 2009).

Reanalysis of the NGS data by neurogeneticists revealed a novel heterozygous intronic variant in the SCN2A gene: c.1035–7A>G. Following analysis of the variant has shown his potentially causative role and explained the course of the disease. Overall, based on the atypical clinical picture, lack of cosegregation with the previously reported variant in the SCN1A, and the result of the functional analysis which are concordant with the molecular pathogenesis of SCN2A-related DEE we believe that the intronic variant is causative in the reported patient.

Reanalysis of NGS data is an important tool in cases with no causative variants (Liu et al., 2019). An additional investigation by a novel specialist could dramatically increase the diagnostic yield, not only by finding causative variants in novel, previously undescribed genes, but also by finding variants that were filtered out during the first interpretation. In such cases, deep phenotyping and detailed clinical examination are of great importance.

A major challenge in DEE is establishing the pathogenicity of novel nucleotide variants and performing robust genotype-phenotype correlations. Considering the genetic heterogeneity of DEE, functional analysis is an important tool that helps in understanding the molecular consequences of the genetic variant and is crucial for proper genetic counseling. Unlike low-throughput, time-consuming approaches for investigation of the molecular defect on the protein level, for some variants, functional analysis on the RNA level can be used as an effective alternative (Wai et al., 2020). In such cases, the major limitation is the expression level of the gene of interest in an available biological sample. However, even for genes with tissue-specific expression patterns such as SCN2A for a subset of variants that are predicted to affect splicing functional analysis using the minigene assay can overcome this limitation.

Splicing variants are more and more recognized as a major cause of Mendelian disorders. Some estimate that up to 50–60% of all pathogenic variants in monogenic diseases could in fact be splicing variants (Wai et al., 2020). Abnormal splicing is more often associated with LoF, due to frameshift and nonsense-mediated decay activation. Yet, the exact molecular mechanism of splicing variants is very challenging to predict, given the complexity of the splicing machinery. In such cases, functional analysis using different approaches such as the minigene assay is crucial for a proper understanding of the molecular pathogenesis. In our patient, the splicing change leads to the insertion of two amino acids and does not alter the reading frame. This observation fits with the molecular mechanism of SCN2A-related DEE, where LoF variants are often found in patients with intellectual disability and GoF variants are linked to epileptic phenotype (Wolff et al., 2019).

Here, we report a diagnostic odyssey of a patient with an atypical Dravet-like phenotype, but with a good response to sodium channel inhibitors. Reanalysis of genetic data and functional analysis confirmed the causative role of the undescribed intronic variant in the SCN2A gene c.1035–7A>G. This report highlights the importance of a phenotype-driven strategy in DEE diagnosis and broadens the mutational spectrum of SCN2A epileptic disorders.

Subjects: The proband was clinically examined in the Veltischev Research and Clinical Institute for Pediatrics of the Pirogov Russian National Research Medical University, Russia. Genetic analysis was performed in Genomed Ltd., Russia. Functional analysis was performed in the Research Centre for Medical Genetics., Russia. We reviewed medical files, EEG tracing, video EEG recordings, magnetic resonance imaging (MRI), and seizure course during the follow-up ranging from 2 to 3.5 years.