Dravet syndrome (DS) is a severe epilepsy presenting in the first year of life with clonic seizures triggered by fever, often unilateral and long lasting in a child with previously normal development. Other seizure types may develop between one and four years of age, including myoclonic jerks, focal seizures and atypical absences. Electroencephalogram (EEG) is usually normal at onset, and then shows slowing of basal activity with asymmetrical spikes or polyspikes-waves 1 2 . In addition to pharmacoresistant seizures and frequent episodes of status epilepticus (SE) often triggered by fever, severity of the condition results from the frequent occurrence of progressive worsening of psychomotor delay and sudden death. Mutations in SCN1A have been identified in 75% of DS patients 3 , and mutations in PCDH19 were found in 10% of the SCN1A-negative population 4 .

Developmental delay in DS is usually considered as a result of severe epilepsy, and the term “epileptic encephalopathy” therefore often applied 5 6 7 . Occurrence of SE and early spikes on EEG proved to be predictors of a worse developmental outcome in a large DS cohort of SCN1A mutated patients 8 . However preliminary data question this relationship between cognitive outcome and the course of epilepsy. Two patients with SCN1A truncation and severe cognitive decline presented with divergent epilepsy severity 9 . Cognitive scores were not related to the frequency of convulsive seizures in a series including 16% of SCN1A-negative patients 10 . DS patients with PCDH19 mutations seem to develop better than those with SCN1A mutations despite the occurrence of explosive convulsive clusters 4 . The first attempt to detect any difference in cognitive outcome according to the presence or not of SCN1A mutation was unsuccessful 11 .

Whether Nav.1 dysfunction plays a role in the cognitive decline independently from the epilepsy has not been proved so far. While epilepsy features in children affected by DS have been described in detail for a long time 1 12 13 , neurodevelopmental features were only recently addressed and available data mostly consist of small and/or retrospective series 7 8 9 10 11 14 15 , sometimes without formal neuropsychological testing 8 .

The aim of this study was to analyze the neuropsychological and behavioral features prospectively in a large cohort of patients with DS and its relation with SCN1A mutation and the characteristics of epilepsy.

All children and adolescents followed at the Neuropediatric Department of Necker Hospital, with clinical and EEG features of typical DS, based on the ILAE classification 16 , were prospectively enrolled in this study from January 2006. Inclusion criteria were the following: i) normal infant with normal EEG and without preexisting cerebral lesion, developing normally until the first seizure occurring before one year of age, ii) refractory clonic or tonic-clonic seizures affecting one or both sides simultaneously or alternatively, with eventually additional seizure types during follow-up, iii) exclusion of any other identified epilepsy syndrome. All SCN1A negative patients were tested for PCDH19 and those with PCDH19 mutation were excluded.

Clinical data were assessed prospectively from 2006 to 2010 by the same three pediatric neurologists (RN, OD, CC). Those obtained before 2006 were retrieved from the medical files. Demographic data included familial (history of febrile seizures/epilepsy) and personal (pre- and peri-natal) antecedents. Epilepsy data (used as variables) included first seizure (age, febrile/afebrile, status epilepticus [SE]/non SE [the term SE designated seizures lasting more that 30 minutes], unilateral/generalized), seizures during course (presence or absence of tonic-clonic, clonic, tonic, absences, myoclonic or focal seizures and sensitivity to fever or not), SE (presence or absence of SE, and if any: age at first SE, number of SE, sensitivity to fever or not), photosensitivity (yes/no on EEG), and treatment (therapy used, age at stiripentol introduction if any, use or not of any antiepileptic drugs considered inappropriate for DS and duration if any). Neurological data included gait and neurological examination.

Mutation analysis of the SCN1A was performed by direct sequencing followed, in negative patients, by multiplex ligation probe amplification (MLPA) as reported previously 4 . Patients were considered mutated when they presented any abnormality on SCN1A gene (mutation, rearrangement, deletion) and “non-mutated” when they had no abnormality detectable on SCN1A.

Neuropsychological evaluations were prospectively performed by the same neuropsychologist from 2006. Raw data of evaluations performed before 2006 were reanalyzed and used for this study. Neurodevelopmental features and adaptation were assessed according to age and level of collaboration using Wechsler Scales (WPPSI, WISC-IV) (IQ, intellectual quotient) 17 18 , Brunet-Lezine Developmental Scale (DQ, developmental quotient) with sub-scores to be completed up to 6 years (the superior age at which the Brunet-Lezine scale is applicable) 19 . Behavior troubles were screened by clinical observation, parent’s interview, Achenbach Child Behavior Checklist and Conners Rating Scale 20 21 . In addition, psychomotor development was assessed by the pediatric neurologist with the neuropsychologist during free play using a semi-quantitative psychomotor scale (SQPS) (no/moderate/severe delay) for global, motor and language development.

For the present analysis, we used all the DQ/IQ and DQ sub-scores obtained during the whole follow-up, whereas Achenbach, Conners and SQPS were collected from the last evaluation.

We studied the relationship between psychomotor/cognitive/behavioral profile, age, epilepsy, and genetic background. First we studied the relationship between DQ/IQ scores, DQ sub-scores, Achenbach, and Conners, on one hand, and age on the other hand. A special analysis of the changes with age in DQ/IQ and DQ sub-scores was performed in the sub-population of patients longitudinally assessed. Secondly we studied the relationships between DQ/IQ and epilepsy variables. Thirdly we studied the relationships between the presence or absence of SCN1A abnormality on one hand, and neuro-developmental scores at last evaluation (DQ/IQ, DQ sub-scores, and semi-quantitative psychomotor assessment) and epilepsy variables on the other hand. Statistical analyses were performed using bivariate and univariate tests: correlation for continuous variables, chi-square and if necessary Fisher exact for categorical variables, and t-test and if necessary signed rank test to compare means.

We performed a total of 81 neuropsychological evaluations with Wechsler or Brunet-Lezine Scales in this series (67 patients). In 10 patients (15%) such evaluation was not possible because of opposing (7 cases) or autistic (3 cases) behavior, and we could score 85% of our series. We also obtained responses for Conners scale in 58 patients and for Achenbach in 59.

Based on a large and homogeneous population of children with typical DS mostly prospectively and partly longitudinally assessed for cognitive development, this study shows that although retardation worsens with age, there is no loss of abilities and patients make slow but regular progress along the first decade. As suspected, cognitive outcome is related to epilepsy course and seizure characteristics since myoclonia and focal seizures are associated with a lower QD/IQ level after 3 years of age. However, epilepsy does not account for the whole cognitive picture since mutated patients tend to exhibit worse psychomotor course than non-mutated ones although epilepsy tends to be less severe. SCN1A mutation is therefore a key factor of cognitive delay, in addition to epilepsy. We also observe a dissociated cognitive profile, particularly for patients with SCN1A mutation: speech develops better than visuo-motor function, from the first years of life and before cognitive slowing that occurs mainly after age three.

DS: Dravet syndrome; EEG: Electroencephalogram; SE: Status epilepticus; DHPLC: Denaturing high performance liquid chromatography; MLPA: Multiplex ligation probe amplification; WISC: Wechsler intelligence scale for children; WPPSI: Wechsler preschool and primary scale of intelligence; IQ: Intellectual quotient; DQ: Developmental quotient; SQPS: Semi-quantitative psychomotor scale; VPA: Valproate; STP: Stiripentol; CLB: Clobazam; SD: Standard deviation.

The authors declare that they have no competing interest.

RN conceived the study design, followed the patients, made the interpretation of the results, and wrote the manuscript. NC followed patients and analyzed the results. MC followed patients and participated in the results analysis. GB participated in the analysis of the results. CB, CD and DL performed the neuropsychological evaluations of the patients. IJ participated to conceive the study and made the supervision of the neuropsychological part of the paper. GD performed statistical analyses. OD followed and referred patients and extensively reviewed the manuscript. CC participated to conceive the study, followed and referred patients, and drafted the manuscript. All authors read and approved the final manuscript.

RN (MD, PhD) is the coordinator of the Reference Centre of Necker-Enfants Malades Hospital (Paris, France) and the head of the Group Phenotype/Genotype in the Research Unit (U663) Paediatric Epilepsies and Brain Plasticity. NC, MC, GB are paediatric neurologists and CB, CD, DL are neuropsychologists at the Reference Centre of Rare Epilepsies (NC and GB also are PhD students). IJ (PhD) is head of the Group Neuropsychology in the Research Unit. OD (MD) was head of the Neuropediatric department of Necker Hospital. GD (PhD) is researcher in epidemiology and statistics. CC (MD, PhD) is the head of the Research Unit Paediatric Epilepsies and Brain Plasticity.