Late diagnoses of Dravet syndrome: How many individuals are we missing?

Abstract We report new genetic diagnoses of Dravet syndrome in a group of adults with complex epilepsy of unknown cause, under follow‐up at a tertiary epilepsy center. Individuals with epilepsy and other features of unknown cause from our unit underwent whole‐genome sequencing through the 100 000 Genomes Project. Virtual gene panels were applied to frequency‐filtered variants based on phenotype summary. Of 1078 individuals recruited, 8 (0.74%) were identified to have a pathogenic or likely pathogenic variant in SCN1A. Variant types were as follows: nonsense (stopgain) in five (62.5%) and missense in three (37.5%). Detailed review of childhood history confirmed a phenotype compatible with Dravet syndrome. Median age at genetic diagnosis was 44.5 years (range 28‐52 years). Tonic‐clonic seizures were ongoing in all despite polytherapy including valproate. All had a history of fever sensitivity and myoclonic seizures, which were ongoing in two (25%) and three (37.5%) individuals, respectively. Salient features of Dravet syndrome may be less apparent in adulthood, making clinical diagnosis difficult. Regardless of age, benefits of a genetic diagnosis include access to syndrome‐specific treatment options, avoidance of harmful drugs, and monitoring for common complications.


SILVENNOINEN Et aL.
Dravet syndrome is typically caused by loss of function variants in the gene SCN1A, particularly affecting inhibitory interneurons. 4 The majority of pathogenic variants arise de novo. 5 DS is now widely recognized by pediatricians and neurologists, and SCN1A molecular testing is available in many countries. However, older patients especially may remain undiagnosed 6 ; the prevalence in age epochs across adulthood is unknown. We describe a series of individuals diagnosed with DS in adulthood based on whole-genome sequencing.

This study was approved by the Camden & Kings Cross
Research Ethics Committee (reference 11/LO/2016). The participants did not have capacity to provide informed consent; written assent for participation was obtained from a personal consultee for each individual following the approved protocol.
Participants fulfilling criteria for the "epilepsy plus other features" category (epilepsy with structural abnormality of the brain or other organs, cognitive impairment, autism or consanguinity), 7 with no known genetic diagnosis, were recruited to the UK 100 000 Genomes Project and underwent wholegenome sequencing. 8 Reads were aligned to build GRCh38 of the human genome. Virtual gene panels 9 were chosen based on the phenotype summary entered at time of recruitment and applied to frequency-filtered variants (Table S1). Results were reviewed in a multidisciplinary meeting with epileptology, clinical, and molecular genetics input, and classified according to the Association for Clinical Genomic Science guidelines. 10 Further clinical data were obtained from medical records and epilepsy genomics clinic reviews.
Prior to this analysis, one individual was identified through screening within the Genomics England Research Environment for stopgain variants in the SCN1A gene region (chr2:165989160-166128013). 6 The finding was confirmed in the present analysis.
Descriptive statistics were calculated using Microsoft Excel version 16.38. Due to small sample size, central tendency was expressed using medians.

| RESULTS
A total of 1078 individuals were recruited from our unit. Eight individuals (six females and two male) were found to have heterozygous pathogenic variants in SCN1A (Table 1). The median age at genetic diagnosis was 44.5 years (range 28-52; Table 2). In one of the individuals (12.5%), a diagnosis of Dravet syndrome had been previously suspected by the treating physician. In three others (37.5%), electronic patient records were, in retrospect, sufficient for suspecting the diagnosis. In the remaining four (50%), sufficient details to make a clinical diagnosis of DS were not present in available electronic patient records, but subsequent review of historical (paper) notes highlighted that their phenotype was indeed compatible with DS ( Table 2). All variants were absent from The Genome Aggregation Database (gnomAD). 11 Four of the variants had been previously reported in individuals with DS, with additional functional evidence for two of these variants (Table S2). Due to the age of our patients and inability to obtain parental samples in many cases, parental testing was possible only in one individual, with confirmation of de novo status of the SCN1A variant.
None of these individuals had any additional filtered variants felt to be contributing to their phenotype.
The median age of seizure onset was 6 months (2.5-10). In seven (87.5%), the first seizure occurred in the context of pyrexia. Two individuals had received a vaccination in the preceding 24 hours.
Median age of onset of developmental delay was 2.5 years (range 1.25-4). All patients had a history of bilateral tonicclonic seizures (TCS) and myoclonic jerks. Six individuals had a history of focal-onset nonmotor seizures with impaired awareness (FIAS); EEGs were not available to confirm atypical absences. Other seizure types ever included focal-onset motor seizures (two individuals), unclassified drops/episodes of head nodding (two individuals), and tonic seizures (one individual). All patients had ongoing TCS. Myoclonic seizures were ongoing in three. FIAS continued in one. One patient had unclassified episodes of eyelid fluttering. Fever or intercurrent illness was elicited as an ongoing seizure precipitant in two.
Seven (87.5%) patients had data on previous and current motor and language skills. All seven had deterioration in mobility compared to their best-attained level; however, all continued to be able to walk for at least short distances. Language skills ranged from no verbal communication to ability to have a basic conversation using sentences. Four of seven (57.1%) had deterioration in language skills compared to their best level.
All patients were taking valproate at current presentation. The median number of current anti-seizure treatments (including ketogenic diet) was 3 (range 2-4). The median number of previously tried anti-seizure treatments (excluding rescue medications) was 11 (range 5-15). All patients had a history of sodium channel blocker (SCB) treatment. Five individuals (62.5%) had documented deterioration in seizure frequency and/or severity while on lamotrigine or carbamazepine.

| DISCUSSION
Dravet syndrome is among the most common monogenic epileptic encephalopathies, with an estimated population-based incidence of about 1/15500 live births. 12 Although some individuals succumb in childhood, recent estimates suggest over 80% will require care in adult services. 13 We conclude, therefore, that a number of adult patients are currently undiagnosed and have unmet health needs. Our experience highlights the need to consider a genetic diagnosis among older individuals with treatment-resistant epilepsy. Dravet syndrome is now, and has been historically, typically diagnosed in childhood; therefore, the commonly appreciated key clinical features reflect the childhood presentation. It is recognized that TCS persist in adulthood in the majority of individuals with DS, while seizure types characteristic in childhood, including myoclonic seizures and atypical absence seizures, continue to occur only in a minority. 2,3,14 In our series also, TCS were ongoing in all, while half of patients had no other definite seizures. In previous series of adult with DS, gait impairment of variable severity, including crouch gait, and significant language impairment were reported in the majority 2,3 ; swallowing difficulties are also a recognized late feature in some. 3 In keeping with the previous literature, all our patients had at least one of these three features. While nonspecific, these features might alert to a possible diagnosis of DS in adults with refractory epilepsy.
In our group of adults with epilepsy and other features, a new genetic diagnosis of DS could be made in 0.74%, a relatively high proportion for a single syndrome. In our view, all adults with refractory epilepsy and intellectual disability of unknown cause should be suspected of having a possible genetic cause, including DS, and be offered genetic testing. In those with seizure onset before age of 1 year, fever sensitivity, and history of myoclonic seizures, testing for SCN1A variants might be undertaken directly. While reviewing the childhood notes of all adults with refractory epilepsy for features of syndromic diagnoses would seem prudent, in reality, such notes may not be available and such review would be a sizeable task in large busy clinics. Panels incorporating a number of genes associated with epilepsy, such as those used in this study (Table S1), provide a cost-effective way to screen for variants in multiple genes, including other genes associated with a Dravet-like phenotype.
Among the widely available anti-seizure medications (ASMs), established treatments for DS include valproate, clobazam, and topiramate. 15 Despite lack of a syndromic diagnosis, all our patients had arrived at polytherapy incorporating valproate and half also took regular clobazam. Despite these treatments, all continue to have TCS. Emerging or licensed treatments for DS include stiripentol, cannabidiol, and fenfluramine. 15 Establishing a diagnosis of DS may help fulfill local criteria necessary for access to these drugs or future treatments on a research basis or through early access programs.
One of the diagnostic clues for DS is exacerbation of seizures by SCBs, 15 and avoiding these presents one of the earliest genetics-driven treatment approaches. 16 All of our patients had a history of SCB use; in five, this was associated with clearlydocumented exacerbation of seizures. One of these patients remains on oxcarbazepine. Withdrawal of SCBs has been associated with benefit also in older individuals 3 and will be considered in this patient.
A multidisciplinary approach is helpful to address the common complications of DS that include dysphagia and progressive gait problems. 3 Making the diagnosis allows for appropriate monitoring and therapy input as necessary. People with DS are at high risk of sudden unexpected death in epilepsy (SUDEP), 17 providing further motivation to optimize seizure control. Arriving at a genetic diagnosis may provide an end to a decades-long diagnostic odyssey for families. The diagnosis may also have implications in terms of genetic counseling for the wider family, as some causal variants may be inherited.
Estimation of rare disease prevalence is a step forward in promoting disease-specific treatments, as prevalence influences funding priorities and is helpful for planning of clinical trials. Estimation of the prevalence of DS in adulthood currently relies on incidence at birth. 12, 13 We suggest our cross-sectional study highlights the need for widespread access to genetic testing among adults with treatment-resistant epilepsies, as there are clearly undiagnosed adults.