Clinical spectrum of early onset epileptic encephalopathies caused by KCNQ2 mutation




KCNQ2 mutations have been found in patients with benign familial neonatal seizures, myokymia, or early onset epileptic encephalopathy (EOEE). In this study, we aimed to delineate the clinical spectrum of EOEE associated with KCNQ2 mutation.


A total of 239 patients with EOEE, including 51 cases with Ohtahara syndrome and 104 cases with West syndrome, were analyzed by high-resolution melting (HRM) analysis or whole-exome sequencing. Detailed clinical information including electroencephalography (EEG) and brain magnetic resonance imaging (MRI) were collected from patients with KCNQ2 mutation.

Key Findings

A total of nine de novo and one inherited mutations were identified (two mutations occurred recurrently). The initial seizures, which were mainly tonic seizures, occurred in the early neonatal period in all 12 patients. A suppression-burst pattern on EEG was found in most. Only three patients showed hypsarrhythmia on EEG; eight patients became seizure free when treated with carbamazepine, zonisamide, phenytoin, topiramate, or valproic acid. Although the seizures were relatively well controlled, moderate-to-profound intellectual disability was found in all except one patient who died at 3 months.


De novo KCNQ2 mutations are involved in EOEE, most of which cases were diagnosed as Ohtahara syndrome. These cases showed distinct features with early neonatal onset, tonic seizures, a suppression-burst EEG pattern, infrequent evolution to West syndrome, and good response to sodium channel blockers, but poor developmental prognosis. Genetic testing for KCNQ2 should be considered for patients with EOEE.

Early onset epileptic encephalopathies (EOEEs) are characterized by developmental impairment and disastrous seizures starting from early infancy. Several genes have been demonstrated to be involved in the pathogenesis of EOEE: ARX in Ohtahara syndrome (OS) and West syndrome (WS), CDKL5 in WS, STXBP1 in OS, SLC25A22 in early myoclonic encephalopathy (EME), and SCN1A in Dravet syndrome (Claes et al., 2001; Stromme et al., 2002; Kalscheuer et al., 2003; Molinari et al., 2005; Kato et al., 2007; Saitsu et al., 2008). There are many infants who do not strictly fit the electroclinical parameters of these known syndromes (Holland & Hallinan, 2010). Identification of causative mutations associated with EOEE and its particular phenotypes is useful for genetic counseling and potentially for patient management.

KCNQ2 encodes the potassium channel subunit Kv7.2, and its mutations have been shown to cause benign familial neonatal seizures (BFNS) with a favorable prognosis (Biervert et al., 1998; Singh et al., 1998). In addition, rare sporadic and familial cases have been reported with neonatal-onset seizures and poor outcomes associated with KCNQ2 mutations (Borgatti et al., 2004; Steinlein et al., 2007). Recently, de novo KCNQ2 mutations have been found in a substantial proportion of patients with neonatal epileptic encephalopathy (Weckhuysen et al., 2012), confirming the association of KCNQ2 mutations with intractable seizures with poor outcomes. Some cases showed a suppression-burst pattern on EEG, tonic seizures, and profound intellectual disability, resembling OS. We also found three de novo missense mutations in KCNQ2 in 3 of 12 patients with OS by whole-exome sequencing (WES) (Saitsu et al., 2012a). Therefore, it is likely that de novo KCNQ2 mutations are one of the common causes of EOEE, including OS.

In this study, to delineate the clinical spectrum of EOEE caused by KCNQ2 mutations, we screened for KCNQ2 mutations in 239 patients with EOEE.

Patients and Methods


A total of 239 patients with EOEEs (51 patients with OS, 104 with WS, 4 with EME, and 80 with unclassified epileptic encephalopathy with an age of onset <1 year) were analyzed for KCNQ2 mutations by high-resolution melting (HRM) analysis or WES (30 of 50 patients with OS) (Saitsu et al., 2012a,b). The diagnosis was made based on clinical features and characteristic patterns on EEG. Mutations of STXBP1 or ARX had been excluded in all or male patients in advance, respectively. KCNQ3 mutations were excluded only in 30 patients with OS by WES. We obtained detailed clinical information on all the 12 patients with a KCNQ2 mutation, an EEG from 10 patients, and brain MRI or computed tomography (CT) images from 9 of them.

Mutation analysis

Genomic DNA was obtained from peripheral blood leukocytes by standard methods, amplified by GenomiPhi version 2 (GE Healthcare, Little Chalfont, United Kingdom), and used for mutation screening. Exons 1–17, covering the entire KCNQ2 coding region (transcript variant 1, NM_172107.2), were examined by HRM analysis or direct sequencing. Samples showing an aberrant melting curve pattern by HRM analysis were sequenced. Polymerase chain reaction (PCR) primers and conditions are available upon request. All novel mutations were verified in the original genomic DNA sample. For the families showing de novo mutations, parentage was confirmed by microsatellite analysis, as previously described (Saitsu et al., 2008). Appropriate biologic parentage was confirmed if three or more informative markers were compatible and the other markers showed no discrepancies.

Whole-exome sequencing

DNAs were captured using the SureSelectXT Human All Exon v4 Kit (Agilent Technologies, Santa Clara, CA, U.S.A.) and sequenced with four samples per lane on an Illumina HiSeq2000 (Illumina, San Diego, CA, U.S.A.) with 101-bp paired-end reads. Image analysis and base calling were performed by sequence control software real-time analysis and CASAVA software v1.8 (Illumina). Reads were aligned to GRCh37 with Novoalign (Novocraft Technologies, Selangor, Malaysia); duplicate reads were marked using Picard ( and excluded from downstream analysis. Local realignments around small insertions or deletions and base quality score recalibration were performed using the Genome Analysis Toolkit (DePristo et al., 2011). Single-nucleotide variants and small insertions or deletions were identified using the Genome Analysis Toolkit and were annotated using ANNOVAR ( (Wang et al., 2010).

TA cloning

For measurement of the ratio of wild-type and mutant alleles of patient 272's mother, PCR products using maternal DNA as a template was subcloned into pCR4-TOPO vector (Invitrogen, Carlsbad, CA, U.S.A.). Cloned fragments were amplified with PCR mixture containing 1× ExTaq buffer, 0.2 mm each dNTP, 0.5 μm each primer, and 0.375 U Ex TaqHS polymerase (Takara Bio, Ohtsu, Japan). M13 forward (5′-TAAAACGACGGCCAGTGAAT-3′) and M13 reverse (5′-CAGGAAACAGCTATGACCATGA-3′) primers were used for amplification, and M13 forward primer was used for sequencing.

Standard protocol approvals, registrations, and patient consents

The experimental protocols were approved by the institutional review boards for ethical issues of Yokohama City University School of Medicine and Yamagata University Faculty of Medicine. Informed consent was obtained from the families of all patients.


Identification of KCNQ2 mutations

The 10 mutations found in 12 patients are summarized in Table S1. A total of 10 mutations in 12 patients were identified (two mutations occurred recurrently). All the mutations are missense changes. In 11 patients the mutation arose de novo, and in one patient the mutation (c.854C>A) was transmitted from the mother, who had epilepsy and possessed the mutation as a somatic mosaic (Fig. S1). All the mutations were absent from our in-house 212 control exomes. One mutation (p.Arg333Trp) was identical to the de novo mutation found in a patient with neonatal tonic seizures and poor developmental outcome (Schmitt et al., 2005). Five mutations (p.Ala265Val, p.Gly290Ser, p.Ala294Val, p.Arg553Trp, p.Arg553Leu) caused alteration of amino acid residues at which different missense changes (p.Ala265Pro, p.Gly290Asp, p.Ala294Gly, and p.Arg553Glu) have been reported in patients with neonatal seizures (Steinlein et al., 2007; Weckhuysen et al., 2012). The mutations were distributed from the transmembrane S4 domain to the C-terminal cytoplasmic domain of the encoded protein (UniProtKB O43526).

Clinical features of patients with KCNQ2 mutation

KCNQ2 mutations were assigned to two patients with unclassified EOEEs and 10 patients with OS, including one patient showing the transition from OS to WS, in whom tonic seizures started 2–3 days after birth but the first medical examination was at 3 months. Detailed clinical information was obtained from all 12 patients (Table S1). The mothers of two patients had a history of epilepsy. Nine patients showed initial symptoms such as seizures or poor feeding within a few days, and also demonstrated initial epileptic attacks within a week. Tonic seizures were initially seen in 11 patients on an hourly or daily basis. Initial EEG studies showed a pattern of suppression-burst in 10 patients with a diagnosis of OS, 4 of whom had an asymmetric pattern and 4 of whom had a brief period of electrodecremental response (Fig. 1). Epileptic spasms arose in four patients, and hypsarrhythmia on EEG was noted in three patients. The most frequently used antiepileptic drug was phenobarbital. Eight patients became seizure-free with a treatment of a particular antiepileptic drug (patients 205, 14, 272, 232, and 168) or combinations of them (patients 297 and 17), or intramuscular injections of adrenocorticotropic hormone (patient 304). The most recent EEG displayed focal or multifocal paroxysmal discharges in eight patients, whereas diffuse paroxysms were seen in only two patients and hypsarrhythmia in one patient. One patient died at 3 months after palliative care. All patients showed intellectual disability from moderate to profound developmental delay. Two patients were able to walk without support, and seven patients were bedridden. Normal brain images were confirmed in three of nine patients for whom MRI findings were studied; the other six patients showed abnormal hyperintensities in the globus pallidus, particularly on T1-weighted images in the neonatal period and on T2-weighted images in early infancy. Mild atrophy of the frontal lobe and a thin corpus callosum were seen in three patients at later stages (Fig. S2).

Figure 1.

EEG of patients with a KCNQ2 mutation. EEG of patient 205 shows an asymmetric pattern of suppression-burst at both 1 day (d) and 44 days of age. EEGs of patients 272 and 136 show patterns of symmetric and asymmetric suppression-burst at 11 days and 2 months (m) of age, respectively. EEG of patient 304 shows hypsarrhythmia for the most part and also electrodecremental discharge that is identical to suppression-burst to some extent. The calibration bar in the lower right-hand corner indicates 1 s (horizontal line) and 100 μV (vertical line).


We identified 10 KCNQ2 mutations in 12 patients with EOEEs (10 cases with OS, two cases with unclassified EOEEs), suggesting that KCNQ2 mutations are one of major causes for OS in contrast to KCNQ3 mutations, which are causative for BFNS as well, but not for EOEEs in our cohort and the previous report (Weckhuysen et al., 2012). Consistent with the results of previous reports, all KCNQ2 mutations found in EOEE patients in this study were missense changes (Saitsu et al., 2012a; Weckhuysen et al., 2012). In contrast, frameshift or nonsense mutations in KCNQ2 are more frequently found among patients with BFNS (Singh et al., 2003; Steinlein et al., 2007); the electrophysiologic properties of mutant cells suggest that haploinsufficiency of the ion channel is the main mechanism of BFNS (Rogawski, 2000; Volkers et al., 2009). There are few experimental data on the function of mutant KCNQ2 in EOEE patients, but a missense mutation p.Ser247Trp within the fifth transmembrane region, found in a mother with BFNS and her son with “early epileptic encephalopathy due to undetermined etiology” has been studied (Dedek et al., 2003). This mutation results in a dominant negative effect on the current amplitude of homomeric wild-type and mutant KCNQ2 constructs. There may be common pathomechanisms between BFNS and EOEE, although further study is needed.

Six of 10 mutations occurred at amino acid residues in which the same or different missense changes have been previously reported: three amino acids (Arg333, Ala265, and Gly290) are mutated in patients with neonatal tonic seizures and poor developmental outcome (Schmitt et al., 2005; Weckhuysen et al., 2012), and two (Ala294, Arg553) are mutated in patients with BFNS. However, the mutations in patients with EOEEs did not cluster in any particular domain of the encoded protein, making it difficult to identify genotype–phenotype correlations. In mice, a combination of a subclinical mutation of Kcnq2 and a mild mutation of Scn2a exacerbates the phenotype (Kearney et al., 2006). It is possible that genetic modifiers contribute to the severity of phenotypes caused by de novo KCNQ2 mutations as well as a dominant negative effect.

The initial clinical symptoms of the patients with EOEEs and KCNQ2 mutation in this study are generally similar to the findings for BFNS. No significant prenatal or perinatal history except seizures can be found. The onset of seizures is within a week of birth, and tonic seizure is the most frequently observed initial type. In our cohorts, apparent autonomic symptoms, such as facial flushing, pale face, apnea, or poor feeding were noticed prior to evident seizures, as is seen in patients with BFNS (Ronen et al., 1993). Although some patients with BFNS experience tens of seizures per day, they seem to be even more frequent in patients with EOEEs and KCNQ2 mutation.

The most critical finding for discrimination between EOEE and BFNS caused by KCNQ2 mutation is the suppression-burst pattern on EEG that we observed in 18 (78%) of 23 patients (Saitsu et al., 2012a; Weckhuysen et al., 2012). Another case with KCNQ2 mutation (p.Ser247Trp), reported by Dedek et al. (2003), also showed a suppression-burst pattern and severe developmental delay, leading to a diagnosis of OS because of the combination of tonic seizures and suppression-burst on EEG. The asymmetrical pattern or brief span of suppression found in our cohort might be characteristic EEG features in patients with EOEE caused by KCNQ2 mutation.

In contrast to the typical clinical course of OS caused by other etiologies, such as brain malformation or mutations of ARX or STXBP1 (Kato et al., 2007; Saitsu et al., 2008), other features can be observed in patients with OS caused by KCNQ2 mutation. First, an evolution to WS characterized by epileptic spasms and hypsarrhythmia on EEG is infrequent. Second, medical control of seizures is relatively good. Third, effective antiepileptic drugs for EOEE with KCNQ2 mutation are unique (Ozawa et al., 2002). Of interest, the primary action of most of the antiepileptic drugs that were useful in our patients—carbamazepine, phenytoin, sodium valproate, and topiramate—is the blockade of voltage-gated sodium channels. Retigabine, which selectively opens the Kv7 potassium channel, might be more effective (Maljevic et al., 2011), possibly improving the neurologic prognosis of EOEE with KCNQ2 mutation.

Faint but unusual signal changes in the globus pallidus were seen on brain MRI in half of the patients. The abnormal signals disappeared in accordance with developmental age, as reported previously (Weckhuysen et al., 2012), but no associations were found between the hyperintensities and seizure prognosis or involuntary movement. Although the pathologic mechanism of the lesion is uncertain, it might be useful clinically to discriminate the EOEE patients with KCNQ2 mutation from others.

In summary, our data clearly demonstrate that de novo KCNQ2 mutations are involved in EOEEs, most of which cases are diagnosed as OS. The seizures responded to some anticonvulsants; however, the neurologic prognosis was very poor. More effective treatments are needed.


We would like to thank the patients and their families for their participation in this study. We thank Aya Narita for her technical assistance. This study was supported by the Ministry of Health, Labour and Welfare of Japan (24133701, 11103577, 11103340, 10103235), a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science (24591500), a Grant-in-Aid for Young Scientists from the Japan Society for the Promotion of Science (10013428, 12020465), the Takeda Science Foundation, the Japan Science and Technology Agency, the Strategic Research Program for Brain Sciences (11105137), and a Grant-in-Aid for Scientific Research on Innovative Areas (Transcription Cycle) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (12024421).


None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.