Two cases of sudden unexpected death in epilepsy in a GEFS+ family with an SCN1A mutation


To the Editors:

Individuals with epilepsy may die suddenly and unexpectedly without a clear cause, known as sudden unexpected (or unexplained) death in epilepsy (SUDEP). This is defined as sudden, unexpected, witnessed or unwitnessed, nontraumatic, and nondrowning death in patients with epilepsy, with or without evidence for a seizure and excluding documented status epilepticus, where postmortem examination does not reveal a cause (Nashef, 1996). Without postmortem, sudden death occurring in benign circumstances with no known competing cause for death is classified as probable SUDEP (Annegers & Coan, 1999). SUDEP rates vary with the cohort studied from 0.35/1,000 person-years (Ficker et al., 1998) in a population-based study to 3–9/1,000 person-years in highly selected cohorts (Tomson et al., 2005). The background risk of sudden death in the general population is 0.05–0.1/1,000 person-years for those under 45 years and 3/1,000 if older (Annegers & Coan, 1999).

Different risk factors and mechanisms may operate with a final common pathway of cardiorespiratory compromise. Some individuals may be more at risk because of social factors, lifestyle, suboptimal management and lack of adherence to treatment. Young age, early onset of seizures, generalized tonic–clonic seizures (GTCS), lack of supervision as well as some medication-related factors are among risk factors identified in case–control studies. Potential mechanisms include respiratory arrest and cardiac arrhythmia or asystole (Hitiris et al., 2007).

Generalized epilepsy with febrile seizures plus (GEFS+) is an autosomal dominant epilepsy first reported in 1997 (Scheffer and Berkovic, 1997). GEFS+ has a spectrum of phenotypes including FS+, defined as “FS extending beyond 6 years with or without afebrile generalized tonic–clonic seizures.” GEFS+ now encompasses many seizure types and syndromes including myoclonic astatic epilepsy (MAE), severe myoclonic epilepsy of infancy (SMEI), absence syndromes, and temporal lobe epilepsy with or without hippocampal sclerosis. Many families with GEFS+ have mutations in the Na+ channel gene SCN1A, and less frequently in SCN1B, SCN2A, or the GABAA receptor gene, GABRG2. The affected gene is only known in an estimated 20% of GEFS+ families (Scheffer et al., 2005).

SCN1A is also implicated in other disorders, mainly sporadic SMEI and closely related phenotypes such as borderline SMEI (SMEB). Over 100 SCN1A missense, nonsense and splice-site mutations have now been described. SCN1A is also implicated in autosomal dominant febrile seizures (FEB3) (Mantegazza et al., 2005), intractable childhood epilepsy with generalized tonic–clonic seizures (Fujiwara et al., 2003) and familial hemiplegic migraine (Dichgans et al., 2005; Vanmolkot et al., 2007).

There are no published reports of familial cases of sudden unexpected death in epilepsy. We report an otherwise typical GEFS+ family with a novel SCN1A mutation and two cases of SUDEP.

Subjects and Methods

The study protocol was approved by Multi-Regional Ethics Committee of Northern & Yorkshire 04/MRE03/56. The family was self-referred. A detailed family pedigree was ascertained and power for linkage studies assessed using SLINK. Blood or cheek swabs were taken for DNA. Clinical information was gained from semistructured interviews and hospital records. Information sought included onset, details of types of seizures throughout lifetime, frequency, and response to medication. Electroencephalograms (EEGs) and imaging results were sought if available. Clinical information for SUDEP cases was gained from next of kin. Linkage analysis using MLINK was carried out on microsatellite data obtained from an 8cM genome-wide scan performed by DeCode Genetics, Iceland.

Person-years were totaled to allow for an estimate of SUDEP incidence and calculated from onset of febrile/afebrile seizures to either (1) death; (2) the present, if active epilepsy; or (3) age of remission. Remission was defined as seizure-free or no antiepileptic treatment for 2 years. Person-years were also calculated for all those known to be carrying the mutation irrespective of the epilepsy as well as the three affected individuals who died suddenly or drowned, presumed but not proven to carry the mutation, from date of birth until death or the present. Confidence intervals were calculated using Poisson distribution tables.

Screening for inherited long QT (LQT) gene mutations (KCNQ1, KCNH2, SCN5A, KCNE1, and KCNE2) was performed in Pavia, Italy (Molecular Cardiology Laboratory, Policlinico S. Matteo).


Clinical features

Table 1 contains clinical information for each affected individual in the pedigree (see Fig. 1). Of 13 clinically affected individuals, 8 had FS. The youngest age of onset was 4½ months (mean 10.7 months). All had several episodes of FS, two 5–10 episodes and the remaining six > 10 episodes. Five affected individuals without reported FS were in one branch of the family. It was believed by some family members that FS occurred beyond the age of 6 years in the eldest generation (V.1, V.3, and V.4), but this could not be confirmed. Of the other five, four continued to have FS beyond the age of 6, although age of offset was not clear as afebrile seizures commenced prior to the cessation of FS. The remaining individual is currently under 6. FS were reported as simple seizures in all but VII.2 and VII.3, whose FS included focal features.

Table 1.  Clinical information
Pedigree referenceAge assessedFebrile seizuresAfebrile seizuresEpilepsy classification
Age of onset/remissionType of seizures/ numberAge of onset/ remissionType of seizures/ number
  1. FoS, focal onset seizures; CPS, complex partial seizures; Ab, absences; MAE, myoclonic astatic epilepsy; sev., several; cont., continuing

V.1/proband67 years5 months/?>6 yearsGTCS/many? <6 years/cont.GTCS/many
GTCS from childhood; focal seizures clinically in early adulthood
V.360 yearssev. months/?>6 yearGTCS/sev<6 years/15 yearsGTCS/sev.FS+
V.464 yearssev. months?>6 yearGTCS/many<6 years/13 yearsGTCS/manyFS+
V.560 yearsNoNo13 years 6 months/33 yearsGTCS/manyEpilepsy: GTCS only
V.8Deceased (SUDEP 23 years)NoNoTeens/throughout lifeGTCS/manyEpilepsy: GTCS only
V.9Deceased (drowned 41 years)NoNo7 years/throughout lifeGTCS/manyEpilepsy: GTCS only
VI.236 years1 years 6 months/teensGTCS/manyChildhood/31 yearsGTCS/many ?FoS/manyFocal and generalized features; difficult to classify
(SUDEP 19 years)
14 months/throughout lifeGTCS/many
Ab/many? FoS
2 years/throughtout lifeGTCS/manyFS+ with other seizure types
VI.1033 yearsNoNo17 years/28 yearsGTCS/manyEpilepsy: GTCS only
VI.1128 yearsNoNo13 years/20 yearsGTCS/6Epilepsy: GTCS only
VII.110 years 4 months8 months/cont.GTCS/many3 years 2 months/cont.Ab/many
Drop attacks/many
VII.27 years 3 months12 months/cont.Complex FS/2
Simple FS
4 years 5 months/5 yearsDrop attacks/5FS+
VII.33 years 10 months4.5 months/cont.Complex FS/3NoNoFS
Simple FS/8 
Total FS/11 
Figure 1.

Pedigree of family showing typical features of GEFS+. Four marker haploypes are shown below in most individuals, with boxes to indicate the haplotype segregating with affected status. Markers are given below the pedigree, also showing the relative position of SCN1A. The proband is indicated by the arrow and the two SUDEP cases by #.

All but one individual, aged 3 years 10 months, experienced afebrile seizures. Age of onset of afebrile seizures was variable and ranged between 3 and 17 years. Age of offset was similarly variable. The youngest to remit was 13 years old. Two other individuals remitted by age 20 and three in their late 20s/early 30s. Three individuals continue to experience afebrile seizures; the proband (V.1) and her grandniece (VII.2) have infrequent seizures while her grandnephew (VII.1) has regular seizures despite treatment.

Three affected individuals with uncontrolled epilepsy died. No DNA was available. One (V.9) drowned during a seizure aged 41 and two (V.8 and VI.4) died suddenly and unexpectedly (one probable and one definite SUDEP) at the age of 23 and 19, respectively. V.8 was not reported as having FS, but developed afebrile seizures in her teens. Cognitive development was normal. She suffered from frequent generalized seizures. She was found dead in bed but did not have a postmortem. VI.4 commenced with FS at 14 months. She started experiencing afebrile seizures soon after and started antiepileptic medication around the age of 2. She had associated mild learning difficulties. Seizures were not controlled around the time of death. She was found dead having fallen out of bed and was thought to have had a convulsion. Postmortem examination was negative.

A total of 288 person-years of febrile seizures and/or active epilepsy were observed (269 person-years of epilepsy alone). Two cases of SUDEP give a rate of about 7/1,000 person-years (95% CI: 1–25). If we considered the total number of person-years (514) for all those with the mutation, irrespective of a history of seizures, the incidence of sudden death observed remains raised at 4/1,000 person-years.

Linkage study

Linkage analysis showed linkage to 2q24, where Na+ channel genes SCN1A, SCN2A, and SCN3A are located. Haplotypes from this region are shown in the pedigree (Fig. 1). Mutational analysis of SCN1A, already linked to GEFS+, identified a novel mutation 5600T>C with an amino acid change I1867T in 12 individuals (Fig. 2). All 10 affected living individuals carried the SCN1A mutation. Two additional nonpenetrant individuals, one an obligate carrier, carried the mutation. DNA was not available from the two SUDEP cases. Two individuals (VI.1 and VI.7) with inconsistent histories and unassigned clinical status were negative for the mutation.

Figure 2.

SCN1A mutation in affected individuals (A) Sequencing trace of affected individual, heterozygous for 5600T>C mutation, resulting in Ile 1867 being replaced by Thr (I1867T). (B) Position of I1867T near C-terminal end of Na+ channel.

The missense mutation changed a nonpolar to a polar amino acid near the C-terminus of the SCN1A protein as shown in Figure 2. LQT gene mutations were excluded in two members of this family who, with consent and ethical approval, acted as surrogates for the SUDEP cases, in one case the father (V.3) for (VI.4) and the other a sibling (V.5) for (V.8).


Here we describe a family with a history compatible with GEFS+ and a novel SCN1A mutation. This mutation segregates with the disease and results in a change from a nonpolar to polar amino acid, likely to have a functional effect on the Na+ channel. An unusual observation in this family is that there are two cases of SUDEP (one definite, one probable) in different affected branches. Family history of SUDEP is not usually reported. The incidence of SUDEP in population-based cohorts is less than 1/1,000 person-years of follow-up. The incidence in this family is much higher at 7/1,000, although the 95% confidence intervals (1–25) do not exclude chance. The incidence remains high at 4/1,000 even if total years of life are considered. Although another genetic defect may result in a susceptibility to sudden death irrespective of seizures, the two SUDEP cases were associated with uncontrolled epilepsy and with no other family history of sudden premature death.

As no DNA was available from the two SUDEP cases, we are unable to prove categorically that they carry the mutation. The pedigree and reported phenotypes, however, are consistent with this. In an autosomal dominant pedigree, the probability of inheriting the mutation is 50%. The probability of a phenocopy is substantially less, given that the risk of an idiopathic epilepsy in the general population by age 20 years is of the order of 1%.

Long QT syndrome (LQTS) is an important cause of sudden cardiac death in the population. Mutations in eight genes that cause LQTS alter ion channel function, resulting in a defect of cardiac electrical repolarization (Schwartz & Priori, 2004), producing a propensity toward ventricular fibrillation, syncope, and unexpected sudden death, usually in children or young adults. One of these genes (SCN5A) encodes a Na+ channel and the others are ion channel or related genes. None of these genes have been implicated in epilepsy. The coding exons of genes KCNQ1, KCNH2, SCN5A, KCNE1, and KCNE2 (five out of eight known LQT genes) were excluded in a first-degree relative of each of the SUDEP cases in two branches of this family. It is therefore unlikely that a separately inherited mutation in an LQT gene is the cause of death in the SUDEP patients. Disease-causing mutations are, however, identified in only 70% of patients with definite LQTS.

The deaths observed may have reflected a more severe epilepsy without the need to invoke a separate predisposition. Nevertheless, it is worth considering a coexisting genetic susceptibility to sudden death. By screening for known LQT mutations in siblings, we attempted to exclude an additional unrelated biological susceptibility. The alternative hypothesis is that the same mutation, presumed present in those who died, may confer increased susceptibility to both epilepsy and sudden death. A similar situation occurs with MECP2 mutations in Rett syndrome, the phenotype of which includes seizures, in which mortality is increased. Studies demonstrated labile breathing patterns and reduction in cardiac vagal tone indicating brainstem immaturity (Julu et al., 2001).

Mortality rates in SMEI (also caused by SCN1A mutations) are raised and families are counseled accordingly. Anecdotally, the incidence of SUDEP appears higher in SMEI cohorts than other childhood epilepsies of similar severity. In Dravet's series of 63 patients with a mean age at follow-up of 11 years 4 months, 10 died; 2 of these were sudden and 1 unknown. In the Tokyo series, 7 of 39 patients died, 1 from an unknown cause (Dravet et al., 2005).

An increased susceptibility to sudden death may be through cardiac mechanisms, reflecting underlying processes common to both neurological and cardiac functions. While SCN1A is primarily a neuronal gene, several studies have shown that Nav1.1 (SCN1A gene product) is present in various regions of the heart in rat and mouse (Rogart et al., 1989; Dhar et al., 2001; Marionneau et al., 2005), in rabbit neonate (Baruscotti et al., 1997), and in dog (Haufe et al., 2005).

There is good evidence for a role for Nav1.1 in pacemaker function of the sino-atrial (SA) node. In mice, Nav1.1 (but not Nav1.5, SCN5A gene product) was detected in the SA node, and moreover, when brain-type Na+ channels were selectively blocked, significantly reduced spontaneous heart rate and greater heart rate variability were observed (Maier et al., 2003). A role for Nav1.1 in pacemaker activity in the mouse SA node was confirmed in a similar but independent study (Lei et al., 2004). Further support comes from a study in rats, where Nav1.1 was also found in the SA node. Heart failure, induced by volume overload, resulted in SA node dysfunction and downregulation of Nav1.1 expression (Du et al., 2007).

In contrast, evidence for a role for Nav1.1 in ventricular function is contradictory. Maier et al. (2002) showed that the majority of voltage-gated Na+ channels in mouse ventricular myocytes contain Nav1.5 subunits, but they also detected brain-type Na+ channel α-subunits (including Nav1.1). When brain-type Na+ channels were blocked, ventricular function was reduced, suggesting a role in excitation–contraction coupling (Maier et al., 2002). However, no reduction in ventricular function was observed in a similar study in rat ventricular myocytes (Brette & Orchard, 2006).

The above discussion focuses on whether the known mutation in this “neuronal” Na+ channel has potential functional effects on cardiac function. Other possible mechanisms include an effect of the mutation on brainstem control of respiration or autonomic function. As described earlier, SCN1A mutations are associated with neuronal dysfunction other than epilepsy, including familial hemiplegic migraine. Studies of migraine pathophysiology implicate the brainstem, with imaging studies showing reproducible changes (Goadsby, 2007). A study describing a missense mutation in SMEI also suggested brainstem dysfunction (Kimura et al., 2005). Two brothers with SMEI and their father, with two simple FS aged <4 years, were found to have a missense mutation. Sleep–wake cycle was deranged after late infancy in both siblings suggesting dysfunction of brainstem aminergic neurons. As these neurons have crucial roles in brain maturation, their early involvement may explain the associated decline in cognitive function. Studies in rat models have also shown good expression of SCN1A in the brainstem. Developmentally in the rat, expression of these brainstem channels increases dramatically during the third week postnatally, peaking at the end of the first month and then decreasing to adult levels at 50% of peak value (Gong et al., 1999).

In our recent review (Nashef et al., 2007), we noted that there is no clinical bridging evidence between susceptibility to sudden cardiac death and idiopathic epilepsy and that further genetic and epidemiological studies are needed. Systematic assessment of cardiac, respiratory, and autonomic function among those with SCN1A mutations may be informative. This pedigree raises the possibility of a shared genetic predisposition to epilepsy and sudden death, and thus in a subgroup with epilepsy.


We are indebted to BrainWave and all the family members who have participated in this research. This research was funded by Guy's & St. Thomas' Charitable Foundation, Sir Jules Thorn Charitable Trust, Epilepsy Bereaved, and Fund for Epilepsy.