Epilepsy of Infancy With Migrating Focal Seizures (EIMFS): Expansion of Clinical Phenotypic And Genotypic Spectra


 EIMFS is a rare early infantile epileptic encephalopathy with unknown etiology and poor prognosis. This study included 36 patients who were diagnosed with EIMFS. 17/36 cases had causative variants across 11 genes, including 6 novel EIMFS genes: PCDH19, ALDH7A1, DOCK6, PRRT2, ALG1 and ATP7A. 13/36 patients had ineffective seizure control, 14/36 patients had severe retardation and 6/36 patients died. Of them, the genes for ineffective seizure control, severe retardation or death include KCNT1, SCN2A, SCN1A, ALG1, ATP7A and WWOX. 17 patients had abnormal MRI, of which 8 had ineffective seizure control, 7 had severe retardation and 4 died. 13 patients had hypsarrhythmia, of which 6 had ineffective seizure control, 6 had severe retardation and 2 died. Also, 7 patients had burst suppression, of which 1 had ineffective seizure control, 3 had severe retardation and 3 died. This study is the first to report that ALDH7A1, ATP7A, DOCK6, PRRT2, ALG1, and PCDH19 mutations cause the phenotypic spectrum of EIMFS to expand the genotypic spectrum. The genes KCNT1, SCN2A, SCN1A, ALG1, ATP7A and WWOX may be associated with poor prognosis. The patients presenting with MRI abnormalities, hypsarrhythmia and burst suppression in EEG may be associated with poor prognosis.

tonic-clonic, autonomic seizures and others (8/36, 22.2%). Hypsarrhythmia (classic or modi ed) was seen in 13 (13/36, 36.1%) patients, and a burstsuppression pattern was observed in 7 (7/36, 19.4%) patients. All patients underwent brain MRI and/or CT examinations, 17 of whom had abnormal brain imaging ndings. Diffuse cerebral atrophy (with an increase in extra axial uid) was observed in 3 children. Four cases demonstrated forehead dysplasia or widened brain space, 3 presented with hippocampal sclerosis, 3 presented with encephalomalacia foci, 1 presented with dysplasia of the corpus callosum, 5 presented with subependymal cyst, 1 presented with pachygyria and 1 presented with meningeal reinforcement. Speci cally, one of the 3 children with cerebral atrophy was accompanied with hippocampal sclerosis. One child had encephalomalacia foci, hippocampal sclerosis, subependymal cyst and widened brain space.   Table 3. We highlight the phenotypes of the ve most frequently implicated genes for EIMFS here. Severe retardation (n); Mild-moderate retardation (n); Normal (n);

KCNT1 EIMFS patients (n=3)
Three patients (3 males) with KCNT1 variants aged 1 year and 4 months to 2 years were studied (median 1 year 4months). The median seizure onset was 1 months (range 15 days to 4 months). All had focal seizures, 1 had tonic seizures, 2 had spasms seizures, and 2 had autonomic seizures. All had normal brain imaging ndings. EEG showed seizure migration in all patients, hypsarrhythmia in 2/3, and burst suppression in 1/3. Two patients had severe developmental retardation, and one died. The age of death was 2 years.

SCN2A EIMFS (n=6)
There were 6 patients with SCN2A EIMFS studied at 1 month to 6 years of age (median 1 years 5 months) with median seizure onset at 2 days (range 1 day to 9 months). All had focal seizures, 4 had tonic seizures,2 had spasms seizures, and 1 had an autonomic seizure. A total of 2/6 had normal brain imaging ndings, and 4/6 had abnormal imaging ndings. EEG data showed seizure migration in all patients, hypsarrhythmia in 3/6, and burst suppression in 2/6. The effective drugs for 6 patients with seizure control were oxcarbazepine, carbamazepine, vigabatrin and levetiracetam. Three patients had severe developmental retardation, three patients had mild-moderate development retardation, and no patients died.
PNPO and ALDH7A1 EIMFS (n=2) One patient (female) with PNPO variants was studied at 4 years and 4 months of age. Seizure onset was 2 days after birth. She had focal seizures and abnormal brain imaging ndings. One patient (male) with ALDH7A1 variants was studied at 3 years and 10 months of age. Seizure onset was 1 month. He had focal and clonic seizures and normal brain imaging. There was no hypsarrhythmia or burst suppression in the EEG of these two patients. Vitamin B6 was effective in both patients. One patient had only mild development retardation, and one patient had normal development.

PRRT2 EIMFS (n=1)
One patient (female) with PRRT2 variants was studied at 10.5 months of age. Seizure onset was 3 months. She had migarating focal seizures and normal brain imaging ndings. Migarating focal seizures and no hypsarrhythmia or burst suppression was evident in the EEG data of the patient (Figure 1).
Oxcarbazepine was effective in controlling seizures in this patient, who had normal development.

Treatment
Antiepileptic treatment effects on seizures were analyzed in 36 patients with EIMFS (Table 3). In addition to classical antiepileptic drugs (AEDs), vitamin B6 and adrenocorticotropic hormones (ACTH) were used to treat EIMFS patients. Vitamin B6 allowed patients with ALDH7A1 and PNPO mutations to achieve seizure-free status, oxcarbazepine was effective for patients with SCN2A, ATP7A, WWOX, and PRRT2 mutations, and ACTH was partly effective for DOCK6 mutation patients with spasms and hypsarrhythmia.

Outcome
We obtained data regarding seizure control and motor development for 36 patients: 14 Table 4. shown in Table 5.

Discussion
EIMFS is characterized by nearly continuous seizures involving multiple independent areas of both hemispheres with arrested psychomotor development. It is an age-dependent, often overlooked syndrome among the epileptic encephalopathies that can occur within the rst 6 months of life [7]. It may therefore be included among epileptic encephalopathies, together with others beginning in the same period, such as early infantile epileptic encephalopathy, early myoclonic encephalopathy, and infantile spasms [8]. The clinical manifestations of children with EIMFS are characterized by migrating and focal seizures, with an onset peak from 40 days to 3 months. Symptoms can manifest on EEG readings as diffuse slow waves in the background, multifocal discharge with or without spastic between episodes, typical or atypical hypsarrhythmia and never a suppression burst pattern [3]. EIMFS has no speci c neuroimaging changes, and the neuroimaging abnormalities reported in the current literature include delayed myelination, abnormal signals in the basal ganglia, dysplasia in the corpus callosum, diffuse brain atrophy, microcephaly, multiple cortical developmental malformations with multiple cerebellar gyri, focal cortical dysplasia, and hippocampal sclerosis [2,8,9]. We described the phenotype spectrum of 36 patients with EIMFS in this study. All patients had clinical seizure migration associated with a signi cant impact on development. We identi ed several key phenotypic features that have only been rarely reported in the EIMFS We describe the genotypic spectrum of 17 patients with EIMFS. We highlight the extensive genetic heterogeneity of EIMFS, which is similar to that in other epilepsy syndromes, such as infantile spasms (West syndrome) and Lennox-Gastaut syndrome, but with a considerably higher yield on current testing. We identi ed the etiology in 47.2% of our cohort, including 6 novel EIMFS genes. The most commonly involved genes were KCNT1 (8.3%) and SCN2A (16.7%), together explaining 25.0% of EIMFS cases. Among the 3 patients with KCNT1 gene mutations, the mutation sites of 1 patients were classi ed as a variant of unknown signi cance (VUS) by the American College of Medical Genetics (ACMG), and among the 6 patients with SCN2A gene mutations, the mutation site of 1 patient was classi ed as a VUS by the ACMG. However, according to the analysis of the clinical phenotypes and responses to drug therapy of the patients, we still believed that KCNT1 and SCN2A were the pathogenic genes of patients. This pattern differs from that reported in patients with neonatal-onset developmental and epileptic encephalopathy [10], including Ohtahara syndrome [11], for which KCNQ2 and SCN2A are most common. These studies highlight the importance of classifying a patient's epilepsy syndrome, which in uences genetic testing and interpretation [12]. Epileptic spasms have only been rarely reported in EIMFS patients [2]. In this series, 8/36 patients (22.2%) had epileptic spasms, including those with KCNT1, SCN2A, SCN1A, DOCK6 and PCDH19 variants. Epileptic spasms are a hallmark of CDKL5 encephalopathy, but EIMFS had not been previously described in this disease. Whether other genes predispose patients with EIMFS to epileptic spasms will require studies with large cohorts to enable phenotype-genotype correlation.
The PNPO gene encodes pyridoxine 5-prime-phosphate oxidase, and the maintenance of optimal pyridoxal 5-prime-phosphate levels in the brain is important in many neurologic disorders in which neurotransmitter metabolism is disturbed, which is associated with autosomal recessive pyridoxamine 5'-phosphate oxidase de ciency [13]. The PNPO gene was EIMFS-related genes reported in previous literature. In our study cohort, although the mutation sites of the gene were classi ed as VUS by the ACMG, taking into account the analysis of the clinical phenotypes and responses to drug therapy of the patient, we still believed that PNPO was the pathogenic genes of patients.
We discovered 6 novel EIMFS genes in our cohort -de novo PCDH19; paternal ALDH7A1, DOCK6, PRRT2 and ALG1; and maternal ALDH7A1, ATP7A, DOCK6 and ALG1encoding a wide range of proteins. These genes have not been described in patients with EIMFS, but they have been associated with other neurological diseases.
Considering the PCDH19 gene rst, most of the PCDH19 gene variants resulted in protein termination and nonsense-mediated decay and affected PCDH19 through impaired calcium binding, which is associated with X-linked early infantile epileptic encephalopathy-9 (EIEE-9) in females [14]. Considering that one of our patients was a female and had a de novo mutation of the PCDH19 gene, which is consistent with the X-linked genetic classi cation and the common clinical phenotype of spasms seizure and delayed motor development, the software predicted the likely pathogenicity of the mutation; thus, these ndings suggest that the PCDH19 gene could be associated with EIMFS.
The ALDH7A1 gene encodes an aldehyde dehydrogenase that is an alpha-aminoadipic semialdehyde dehydrogenase in the pipecolic acid pathway of lysine catabolism, which is associated with autosomal recessive pyridoxine-dependent epilepsy [15]. Considering that some of our patients had paternal and maternal mutations of the ALDH7A1 gene, which is consistent with the autosomal recessive genetic classi cation and the common clinical phenotype of seizures, delayed motor development and effective treatment with vitamin B6, the software predicted the likely pathogenicity of the mutation; thus, these ndings suggest that the ALDH7A1 gene could be associated with EIMFS.
DOCK6 belongs to subfamily C of the DOCK family and has a role in remodeling the actin cytoskeleton by functioning as a GEF for both CDC42 and RAC1 [16], which is associated with autosomal recessive Adams-Oliver syndrome-2. Adams-Oliver syndrome is a multiple congenital anomaly syndrome that is characterized by aplasia cutis congenita (ACC) as well as terminal transverse limb defects (TTLD) in addition to variable involvement of the brain, eyes, and cardiovascular system [17]. Although one of our patients had maternal and paternal mutations of the DOCK6 gene, which is consistent with the autosomal recessive genetic classi cation, and the software predicted the likely pathogenicity of the mutation, the patient lacked the typical clinical phenotype of Adams-Oliver syndrome-2. Therefore, we could not determine whether the DOCK6 gene was the causative gene of EIMFS, and further data are needed for veri cation.
The PRRT2 gene encodes proline transmembrane protein 2 and is involved in signal transduction between neurons. PRRT2-associated paroxysmal movement disorders (PRRT2-PxMD) include autosomal dominant genetic paroxysmal kinesigenic dyskinesia (PKD), benign familial infantile epilepsy (BFIE), paroxysmal kinesigenic dyskinesia with infantile convulsions (PKD/IC), and hemiplegic migraine (HM). The characteristics of BFIE include early onset, cluster seizures that could be self-healing or that respond well to the treatment of sodium channel blockers, and normal development. In addition, PRRT2 pathogenic variants have been identi ed in other childhood-onset movement disorders and different types of seizures, suggesting that the understanding of the spectrum of PRRT2-PxMD is still evolving [18]. The patient with this mutation in our cohort was very interesting. He presented with very frequent cluster seizures when he was 3 months old, video-EEG demonstrated migrating focal seizures and typical background characteristics of EIMFS, and he experienced obvious cognitive retardation after the onset of epilepsy. He achieved seizure-free status after the administration of oxcarbazepine, and follow-up showed normal cognitive development. The patient had a paternal mutation of the PRRT2 gene-his father had symptoms of dystonia in his youth-and the software predicted the likely pathogenicity of the mutation. We considered that the PRRT2 gene was the causative gene of EIMFS, and further data were needed for veri cation.
In the 10th patient, two pathogenic genes were detected simultaneously, namely, WWOX and ATP7A. The clinical phenotypes reported in the literature were Xlinked Menkes disease, occipital horn syndrome, distal spinal muscular atrophy 3, epileptic encephalopathy, early infantile 28, esophageal squamous cell carcinoma, somatic and spinocerebellar ataxia, and autosomal recessive 12, which theoretically can cause epileptic encephalopathy [19][20][21][22][23]. This patient also had a signi cant decrease in blue copper protein, had hair and skin color changes and could have been diagnosed with Menkes disease. WWOX conformed to the genetic classi cation. One of the two mutation sites was a pathogenic mutation, and the other was a VUS. We could not con rm whether it was a copathogenic gene.
The ALG1 gene encodes mannosyltransferase I (MT I). The biosynthesis of lipid-linked oligosaccharides is highly conserved among eukaryotes and is catalyzed by 14 glycosyltransferases in an ordered stepwise manner. MT I catalyzes the rst mannosylation step in this process [24]. The gene variants can result in autosomal recessive congenital disorder of glycosylation type Ik [25,26]. The characteristics of congenital disorder of glycosylation type Ik includes feeding problems and diarrhea, profound hypoproteinemia with massive ascites, muscular hypertonia, seizures refractory to treatment, recurrent episodes of apnea, cardiac and hepatic involvement and coagulation anomalies [27]. The clinical characteristics of this patient were consistent with the above core symptoms, the genetic results were consistent with the law of autosomal recessive inheritance, and the software predicted the likely pathogenicity of the mutation. We considered the ALG1 gene to be a newly pathogenic gene of EIMFS.
Most patients with EIMFS are refractory to AEDs, but some have shown good progression or a near satisfactory response to treatment. The AEDs used alone or in combination with one another that may achieve seizure control or reduction are potassium bromide, levetiracetam, ACTH, stiripentol, clonazepam, and ru namide [28]. Mikati et al. reported that quinidine is effective in the treatment of patients with KCNT1 gene mutations in EIMFS [29]. At present, there are few literature reports on the use of corresponding effective drug treatments for patients with different EIMFS gene mutations. In our cohort, we determined that vitamin B6 could allow patients with ALDH7A1 and PNPO mutations to achieve seizure-free status. Oxcarbazepine was effective for patients with SCN2A, ATP7A+WWOX, and PRRT2 mutations. One of the patients with a maternal SCN2A heterozygous mutation was treated with oxcarbazepine; subsequently, the patient's convulsions were controlled. The patient's mother had a normal heterozygote phenotype, which was consistent with the pathogenesis of autosomal dominant inheritance (incomplete penetrance). ACTH was partly effective for patients with DOCK6 mutations who had spasms and hypsarrhythmia.
While seizure outcomes and developmental prognoses are generally poor in EIMFS, there are rare reports of mildly affected patients [30]. In our study cohort, the incidence of poor prognosis was also relatively high; 6/36 (16.7%) patients died, and the related pathogenic genes were KCNT1, SCN1A, ALG1. 14/36 (38.9%) patients had severe retardation, and the genes for ineffective seizure control and severe retardation included KCNT1, SCN2A, WWOX and ATP7A. The results indicated that the related pathogenic genes KCNT1, SCN1A, ALG1, SCN2A, WWOX and ATP7A may be associated with ineffective seizure control and poor prognoses. While all patients experienced refractory epilepsy early in the course of the disease, 3/36 (8.3%) patients had normal mental and motor development. Genes associated with seizure-free, mild-moderate retardation or normal of mental and motor development included PRRT2, SCN2A, ALDH7A1, PCDH19 and PNPO.
In addition, we compared the association of MRI abnormalities, hypsarrhythmia and burst suppression in EEG with poor prognosis. The results found that patients with EIMFS characterized by abnormal MRI, hypsarrhythmia and burst suppression in EEG have a higher incidence of ineffective seizure control, severe retardation and a higher mortality rate. The results suggest that EIMFS patients who present with abnormal MRI, hypsarrhythmia and burst suppression in EEG may be associated with ineffective seizure control and poor prognosis. We need to further expand the sample to analyze and con rm these correlations.

Conclusions
In conclusion, our study expands the EIMFS clinical phenotype and genotype spectra. EIMFS patients who present with abnormal MRI ndings, hypsarrhythmia and burst suppression in EEG may be associated with ineffective seizure control and poor prognosis. Etiological analysis showed that in addition to genetic mutations, structural brain injury such as HIE could also be secondary to EIMFS, and structural and infectious etiology may be associated with poor prognosis. This study is the rst to report that ALDH7A1, ATP7A, DOCK6, PRRT2, ALG1, and PCDH19 mutations cause the phenotypic spectrum of EIMFS. The genes KCNT1, SCN1A, ALG1, SCN2A, WWOX and ATP7A may be associated with ineffective seizure control and poor prognosis. Through early diagnosis with genetic tests and the administration of the corresponding precise treatment, the outcomes of EIMFS can be notably improved.

Participants and phenotyping
Our EIMFS cohort comprised 36 patients from Hunan Children's Hospital and Qilu Hospital of Shandong University. A multicenter retrospective case study was performed over a 10-year period (January 2010 to January 2020). The parents of the patients provided written, informed consent. This study was approved by the Medical Ethics Committee of Hunan Children's Hospital and Qilu Hospital of Shangdong University.
According to the clinical characteristics of EIMFS described by Coppola [1], the inclusion criteria are as follows: (1) onset within 6 months after birth; (2) migrating focal seizures at onset; (3) multifocal seizures intractable to conventional antiepileptic drugs; (4) the EEG during the onset period is characterized by multifocal discharges, migrating in one hemisphere or between the two hemispheres, involving multiple parts, and the clinical onset is closely related to the time and location of the EEG discharge; and (5) delayed developmental progress or signs of psychomotor regression associated with seizure onset. The phenotypic information of all patients was collated on clinical presentation, disease course, EEG, neuroimaging, treatment strategies and the results of neurometabolic and diagnostic genetic investigations. All patients were followed up every 1-6 months by telephone or outpatient department.
It should be noted in particular that one of our patients met the inclusion criteria except for the onset age of 9 months. Considering that EIMFS with onset age of 9 months has been reported in literatures, we still included this patient in our study cohort.

Genetic Tests
Genetic testing was carried out using chromosome karyotype analysis, CNV analysis, mitochondrial genome sequencing, epilepsy gene panels and WES. CNV analysis was performed with the Illumina HumanOmniZhonghua-8 Bead Chip; Mitochondrial genome sequencing was subsequently performed on an Illumina HiSeq 2000 platform (Illumina, San Diego, CA, USA); details were provided by our team previously [31]. The epilepsy gene panel contained 265 epilepsyassociated genes was performed on methods previously reported by Lemke et al [32]. WES was also performed based on methods previously reported by our team [31] by using Illumina HiSeq X Ten (Illumina, San Diego, CA, USA) with 150-bp paired-end reads.