KCNQ2 encodes a channel subunit carrying the neuronal Im current whose inherited mutations were first described in autosomal dominant benign familial neonatal epilepsy (BFNE, OMIM#121200) 1 2 3 . Patients affected by a BFNE displayed stormy phase of motor seizures during the neonatal period, lasting 2 to 6 weeks in average. Interictal EEG was normal or slightly modified 4 . Subsequently, seizure frequency quickly decreased and the vast majority of patients became seizure free before the age of three months 5 . Motor and cognitive outcome were usually normal. Recently, de novo mutations of KCNQ2 have been described in early onset epileptic encephalopathies (EOEEs; OMIM#613720) 6 7 8 . EOEEs are a group of devastating epilepsies beginning before three months of age, with frequent seizures and abnormal interictal EEG leading to a rapid deterioration of motor, cognitive and sensori-neuronal functions. Patients carrying de novo KCNQ2 mutations displayed abnormal interictal EEG that could reveal multifocal spikes or a suppression-burst pattern, and all had poor neurological outcome 7 8 . This dramatic form of KCNQ2-related epilepsy, with very poor neurological outcome, was unexpected. In order to assess the importance of KCNQ2 screening for the molecular diagnosis of early onset epilepsies, and mostly to describe the outcome of the sporadically mutated patients, we have analyzed a cohort of 71 patients with an early onset, severe epilepsy, without any familial history of epilepsy.

This study was approved by CPP Sud Méditerannée (Comité de protection des personnes). Seventy one patients were included in a cohort of subjects who displayed an early onset epileptic encephalopathy. All the patients or their parents gave their informed consent to join the cohort. Inclusion in the cohort was decided according to the following criteria; (1) epilepsy onset within the first 3 months of age; (2) abnormal interictal EEG (3) brain MRI without obvious cortical malformation or hypoxic lesion; (4) normal metabolic screening (exclusion of nonketotic hyperglycinemia, hyperammonemia, urea cycle defect, organic aciduria, hyperlactacidemia, pyridoxine-dependent and pyridoxal-dependent seizures); (5) No mutation of STXBP1, a major gene involved in early onset epileptic encephalopathy with or without suppression-burst 9 ; (6) No mutation of ARX 10 in male patients (n=35); (7) patients must be regularly followed till now. All the girls that displayed early onset epileptic spasms and/or tonic seizures without any suppression-bursts were tested for CDKL5 (n=36). The epilepsy began during the neonatal period for 47/71 patients, the EEG showed a suppression-burst or discontinuous traces in 33 of them (Groupe A), and multifocal spikes in the remaining 14 (Groupe B). Epilepsy began between 1 and 3 months for the 24 patients of groupe C. The 18 coding exons (including alternative exons) of KCNQ2 were sequenced. Primer sequences are available upon request. The identified mutations were numbered according to the KCNQ2 reference sequence NM_172107.2.

We found heterozygous mutations in KCNQ2 in 16/71 patients (Table 1). All of them have occurred de novo. Typically, the first seizure was observed before the 5^th day of life (n=14/16), taking the form of clonic and/or tonic seizures resembling those observed in BFNE (12/16). These seizures were very frequent, rapidly leading to obvious neurological impairment before the end of the first week (10/16). Eight patients carrying a KCNQ2 mutation were initially diagnosed with an Ohtahara syndrome, with a typical suppression-burst pattern on the EEG (Table 1, Figure 1). The first EEG did not show any suppression-burst pattern, but discontinuous traces in the remaining patients (Table 1, Figure 1). In three cases, EEGs evolved into a hypsarythmic pattern, but the majority quickly developed into a continuous pattern with multifocal asynchronous spikes and/or slowing of the traces (13/16). The outcome of epilepsy was highly variable: 9/16 patients became seizure free during the follow-up, 6 of them before the end of the first year of life, while 7/16 patients were still epileptic, three of them had only myoclonic jerks, two of them had recurrent generalized tonic clonic seizures and two had focal seizures (Table 1). Fifteen patients had obvious developmental delay: 4/15 could walk but 3/4 had no language and 1/3 had autistic features; 11/15 were profoundly impaired with poor or absent head control and eye contact (8/15) or global/axial hypotonia with poor or absent hand use (3/15). One patient had a good evolution with normal neurological evaluation at age 6. The initial brain MRI was normal or showed very slight and transitory brain signal abnormalities in 12/16 patients, while in 3 patients, abnormal signal intensity were found, as previously described 7 (Table 2). Only one patient had extra-neurological features: congenital left hip luxation and cleft palate (patient 2). Fifteen patients had a mutated KCNQ2 in the group A (45%, n=33), one patient had a mutation of KCNQ2 in the group B (7%, n=14) and none of the patient had a KCNQ2 mutation in the group C. Thus, KCNQ2 was mutated in half of patients with a neonatal onset epileptic encephalopathy and an EEG showing either discontinuous or suppression-burst pattern. Here, we confirm that, besides well described entity BFNE, KCNQ2 mutations can also be associated with severe epileptic and cognitive phenotypes defining early onset epileptic encephalopathies 7 . Hence, it should be considered in the diagnosis workup of neonatal onset epilepsies especially those beginning during the first week of life with stormy clonic and/or tonic seizures, whatever the presence or absence of a familial history. If cognitive outcome is relatively reliable and good in familial cases of BFNE 5 , it is not the case in sporadic ones, where neurological outcomes range from dramatic to normal. We did not find any relationship between the initial history of the epilepsy and the severity of outcome. For example, patients 1 and 12 had relatively similar features at the beginning and displayed very different outcomes (Tables 1, 2 and Figure 1). Since KCNQ2 is now implicated in various forms of epilepsies, from the most benign to the most dramatic, additionnal data on phenotype/genotype correlations would be particularly relevant. Interestingly, none of the mutation reported in neonatal epileptic encephalopathies had previously been reported in BFNE, and the severe mutations that have been found in several patients (p.G290A 7 , p.T274M and p.A294V(present study)) lead to relatively similar features in terms of initial EEG and development, however different in terms of evolution of the epilepsy. The different epileptic features in patients carrying the same mutation of KCNQ2 may be due to genetic modifiers or non genetic factors. Overall, this cohort of patients highlights the heterogeneous evolution of the neurological phenotypes associated with de novo heterozygous mutations in KCNQ2. This heterogeneity could be at least partially related to the impact of the mutations on the Im current. Analysis of the functional consequences of “benign” versus “severe” mutations in KCNQ2 should be of paramount importance to better understand the molecular and cellular mechanisms involved in the emergence of an epileptic encephalopathy. This has recently been tested with two mutations of KCNQ2 affecting the same residue in the S4 domain of the protein KV7-2 but associated with either a benign phenotype, or with a neonatal epileptic encephalopathy with severe drug-resistant seizures and neurocognitive delay, suppression-burst pattern at EEG, and distinct neuroradiological features 11 . The authors showed that, while both mutations destabilized the open state of the channel causing a reduction of the voltage sensitivity, the functional changes were more pronounced in the “severe” mutation than in the benign one. In both cases, the functional impairment could be fully restored by the neuronal Kv7 activator retigabine. This study suggested that the clinical disease severity may be related to the extent of the mutation-induced functional K+ channel impairment and set the preclinical basis for the potential use of Kv7 openers as a targeted anticonvulsant therapy to improve developmental outcome in neonates. However, since two patients carrying the same KCNQ2 mutation do not have the same epileptic outcome, correlations between Im impairment and the severity of the encephalopathy should be made with caution. Other unknown factors may be involved in the occurrence of the epileptic encephalopathy. Moreover, the vast majority of the mutations we described here were localised on segment S6 and should have different consequences on Im current that those which have been previously described, affecting segment S4 11 . These consequences still have to be studied. Overall, it is not known whether ongoing brain dysfunction that is observed in several patients is due to the Kv7-2 channelopathy or if it is a sequel of neonatal epilepsy. This question would be of paramount interest.

HC head circumference, m months, y year.

GW gestational week, HC Head circumference, m month, Y year, PHB Phenobarbital, PHT phenytoine, VGB vigabatrin, TPM topiramate, CLN clonazepam, VPA Sodium Valproate, CBZ carbamazepine, CLB clobazam.

KCNQ2 is frequently found mutated de novo in early onset epileptic encephalopathies, especially if the epilepsy begins within the first week of life. Despite relatively stereotyped initial phenotype, the neurological and epileptic outcomes were highly variable, overall severe.

The authors declare that they have no competing interests.

MM, SA, CM, NV, AR, BH, MC, AK, CA, GB, DV, MAB, DH, DG, AA, ND, TBV, JP, BI, NG, LV, IA, AJ, PC BC, GL and LV designed the study and interpreted the data. NBK, JSS and CL made the experiments. MM, CM, GL and LV drafted and revised the MS. All authors read and approved the final manuscript.