Risk Factors in Sudden Death in Epilepsy (SUDEP): The Quest for Mechanisms


Address correspondence and reprint requests to Dr. L. Nashef at Neurology Department, Kings College Hospital, Denmark Hill, London SE5 9RS, U.K. E-mail: Lina.Nashef@kingsch.nhs.uk;


Summary:  People with epilepsy may die suddenly and unexpectedly without a structural pathological cause. Most SUDEP cases are likely to be related to seizures. SUDEP incidence varies and is <1:1,000 person-years among prevalent cases in the community and ∼1:250 person years in specialist centres. Case–control studies identified certain risk factors, some potentially amenable to manipulation, including uncontrolled convulsive seizures and factors relating to treatment and supervision. Both respiratory and cardiac mechanisms are important. The apparent protective effect of lay supervision supports an important role for respiratory factors, in part amenable to intervention by simple measures. Whereas malignant tachyarrhythmias are rare during seizures, sinus bradycardia/arrest, although infrequent, is well documented. Both types of arrhythmias can have a genetic basis. This article reviews SUDEP and explores the potential of coexisting liability to cardiac arrhythmias as a contributory factor, while acknowledging that at present, bridging evidence between cardiac inherited gene determinants and SUDEP is lacking.

People with epilepsy may die unexpectedly without a clear structural or pathologic cause for death. Sudden unexpected death in epilepsy, or SUDEP, is a convenient category in which to classify such deaths. It accounts for a significant proportion of deaths in epilepsy (Forsgren et al., 2005; Lhatoo and Sander, 2005). Different risk factors and mechanisms may operate with a final common pathway of cardiorespiratory compromise. Understanding mechanisms is necessary to formulate prevention strategies.

We briefly review incidence data and then focus on risk factors and potential mechanisms. Evidence supports respiratory mechanisms, with a protective effect through lay supervision in the community, where respiratory compromise may be amenable to simple measures. We are interested here in exploring the potential contribution of cardiac mechanisms, in particular the hypothesis that a coexisting genetic susceptibility to epilepsy and sudden cardiac death may exist.


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, in which postmortem examination does not reveal a structural or toxicologic cause for death (Nashef, 1997). Where autopsy is not performed, and for the purpose of epidemiologic studies, sudden death occurring in benign circumstances with no known competing cause for death is classified as probable SUDEP (Annegers, 1997)


Reported SUDEP rates vary with the cohort studied (Stollberger and Finsterer, 2004; Tomson et al., 2005) and range from ∼1:100 person-years in intractable epilepsy surgery cohorts to 0.35/1,000 person-years in a population-based study (Jay and Leestma, 1981; Dasheiff, 1991; MRC, 1991; O'Donoghue and Sander, 1997; Ficker et al., 1998; Walczak et al., 2001; Nashef, 2004; Ryvlin et al., 2005; Tomson et al., 2005). Most publications reflect risk for younger adults, and extrapolation to other age groups is not appropriate. SUDEP is thought to be relatively rare in children (Sillanpaa et al., 1998; Donner et al., 2001; Camfield et al., 2002; Camfield and Camfield, 2005; Weber et al., 2005), but studies are limited, often with small numbers.


Although some authors make a distinction between the two (Sillanpaa et al., 1998), it is not possible to separate with certainty those related to a “terminal” epileptic seizure and those occurring independently. The presence of fresh tongue biting is suggestive (Ulrich and Maxeiner, 2003), but its absence does not exclude a seizure. A number of studies suggest that SUDEP, as defined earlier, is frequently a periictal event. As the majority are unwitnessed, the evidence is largely indirect. The proportion of reported witnessed cases varies from 7 to 38%. Of those witnessed, between one third and all were reportedly related to convulsions (Langan et al., 2000). A series of witnessed SUDEP cases, among 135 ascertained, reported convulsions in 12 of 15 (Langan et al., 2000). Signs of preceding seizures were reported in 67% in a case control study of 42 SUDEP cases from Norway (Kloster and Engelskjon, 1999). Case–control studies report a higher relative risk in people with more-frequent seizures and in those with a history of generalized convulsive seizures (Birnbach et al., 1991; Nilsson et al., 1999; Walczak et al., 2001; Langan et al., 2005). A SUDEP case recorded in a video-telemetry unit occurred during a secondarily generalized epileptic seizure (Bird et al., 1997). Although studies suggest that prevention of seizures through successful epilepsy surgery reduces excess, it is still possible that those who do not respond constitute a different subgroup (Persson et al., 2005; Ryvlin et al., 2005).


The association with epileptic seizures, especially convulsive, accounts for some of the evidence only. Some individuals may be more at risk because of social factors, lifestyle, suboptimal management, and lack of adherence to treatment. Others may have additional biologic susceptibility. This could be related, but not necessarily, to the type of epilepsy.

Risk factors from descriptive cohorts, considered variables for further study, include youth, male sex, remote symptomatic epilepsy, structural findings on neuropathology, severe epilepsy, unwitnessed seizures, alcohol abuse, abnormal EEGs with epileptiform changes and greater variations, mental handicap, psychotropic medication, African-American ethnicity, lack of adherence to treatment, abrupt medication changes, and low AED levels (Nashef, 2004). Risk factors emerging from case–control studies with varying methods include convulsive seizures (Birnbach et al., 1991), prone body position (Kloster and Engelskjon, 1999), tonic–clonic seizure in the past year, >50 seizures per month, full-scale IQ <70, and more than two AEDs (odds ratios adjusted for seizure frequency, number of generalized tonic–clonic seizures (GTCSs), and number of AEDs when indicated) (Walczak et al., 2001), more-frequent seizures, epilepsy type, younger age at onset of epilepsy, polytherapy, frequent medication changes and carbamazepine (CBZ) above the usual quoted range (conditional multiple logistic regression analysis performed with relative-risk estimates adjusted for seizure frequency and other variables if applicable) (Nilsson et al., 1999; 2001), history of GTCSs, history of GTCSs in the last 3 months, bedroom shared by someone capable of giving assistance (protective), use of listening devices (protective), a history of asthma (protective), more than four AEDs ever taken, having never taken AEDs, and current use of CBZ (backward stepwise conditional logistic regression analysis was performed) (Langan et al., 2005). Some of these risk factors are discussed further later.


Although SUDEP was observed well before modern AED therapy, putative risk factors relating to AED treatment, such as drug levels, adherence to treatment, abrupt withdrawal, polytherapy, and choice of AED are potentially important, as they are amenable to manipulation in routine management.

The Stockholm study, which included only SUDEP cases of treated epilepsy, identified polytherapy and frequent medication changes as independent risk factors for SUDEP (Nilsson et al., 1999). Although these may still be surrogate markers for epilepsy severity, theoretically, both could increase risk, for example, by affecting postictal state or causing autonomic instability. A large U.K.-based study, although not confirming polytherapy as a risk factor, found that total number of AEDs ever taken was associated with increased risk, reflecting intractability of the epilepsy. The same study also found that untreated epilepsy was a significant risk factor (Langan et al., 2005).

Studies on adherence to treatment are contradictory (Bowerman et al., 1978; Lund and Gormsen, 1985; George and Davis, 1998; Opeskin et al., 1999). This is difficult to study, as accurate information about medication omitted shortly before death is not usually available, and postmortem AED blood levels can be unreliable (Tomson et al., 1998). A recent study from Cardiff of AED levels in sequential 1-cm hair-segment samples showed greater coefficient of variation in SUDEP cases than in other cohorts studied (Williams et al., 2006). The difference was not attributed to prescribed medication changes and may be due to poorer adherence to treatment.

The question has been raised as to whether certain AEDs are associated with a higher risk of SUDEP. Older descriptive series simply reflected the then-current prescribing practices. The study of Walczak et al. (2001) did not identify any specific association with any particular AED. An association with CBZ use, or higher than usual levels, has been observed in a small number of studies (Timmings, 1998; Nilsson et al., 2001; Langan et al., 2005). Rare cases of heart block secondary to CBZ occur in predisposed individuals, and CBZ has an effect on heart-rate variability (Kenneback et al., 1997; Tomson et al., 1998; Hennessy et al., 2001; Persson et al., 2003). The relevance of these findings, however, to the majority of cases with epilepsy remains uncertain. It is probable that patients with more-severe epilepsy/seizures were more likely to be prescribed higher doses of CBZ, a mainstay of epilepsy treatment. Furthermore, AED treatment must be appropriate to the epilepsy syndrome, and CBZ is not broad spectrum. In one study, some SUDEP patients taking CBZ had uncontrolled idiopathic generalized epilepsy (Nashef et al., 1998). Overall, although surgical or medical treatment interventions have the potential to worsen epilepsy, the evidence suggests that the risk/benefit ratio favours AED treatment, with an increased risk of SUDEP among those never treated (Langan et al., 2005).


Most unwitnessed SUDEP deaths occur in bed, presumably during sleep. Nocturnal seizures may be different pathophysiologically; this observation may be also due to lack of assistance, in the event of respiratory compromise. Some evidence supports this. The Norwegian case–control study found a significant difference in the position of the body compared with that expected, with 71% prone and only 4% supine (Kloster and Engelskjon, 1999). In another study of circumstances of SUDEP, in 11 of 26 cases, the body was found in a position that could have resulted in respiratory obstruction (Nashef et al., 1998). The U.K.-based case–control study of Langan et al. investigated supervision as a possible factor and showed decreased risk associated with the use of listening devices or sharing a bedroom with someone capable of giving assistance (Langan et al., 2005).



Respiratory changes frequently occur in seizures. Central and obstructive apnea, excess bronchial and oral secretions, pulmonary edema, and hypoxia are all well documented (Nashef et al., 1996; Blum et al., 2000; So et al., 2000; Swallow et al., 2002; O'Regan and Brown, 2005). Pulmonary edema is found at autopsy in the large majority of SUDEP cases. Central apnea can occur secondary to the ictal discharge, but a role has also been suggested for secondary endogenous opioid release. During postictal coma, hypercapnea and hypoxia may be less-potent respiratory stimuli. In a sheep model of ictal sudden death, animals that died had a greater increase in pulmonary vascular pressure and hypoventilation. When airway obstruction was excluded by tracheostomy, central apnea and hypoventilation were observed in all, causing or contributing to death in two, whereas a third animal developed heart failure with significant pathologic cardiac ischemic changes (Johnston et al., 1997). The apparent protective effect of supervision favours an important primary role for respiratory factors (Langan et al., 2005), as these can be influenced by relatively unskilled intervention, such as airway protection, repositioning, or stimulation. It is unknown what proportion of SUDEP cases may be prevented by such intervention.


Primary or secondary cardiac mechanisms are also likely to be important (Stollberger and Finsterer, 2004). Changes in heart-rate variability are well documented in epilepsy and can be related both to the pathophysiologic basis of the epilepsy and AEDs. Studies of routine ECG changes in people with epilepsy have had a low diagnostic yield. A variety of changes, of rhythm or repolarization, are documented during seizures (Nei et al., 2000; Opherk et al., 2002; Stollberger and Finsterer, 2004). Sinus tachycardia is very frequently observed in seizures. Malignant tachyarrhythmias are relatively rarely recorded. Sinus bradycardia/arrest is better documented, although one study observed ictal cardiac asystole in only five of 1,244 inpatients undergoing prolonged video-EEG recording (Rocamora et al., 2003). A study of long-term cardiac rhythm monitoring in a highly selected group with poorly controlled epilepsy with an implantable device eventually revealed episodes of sinus arrest, for which pacing was carried out in four of 19 patients (Rugg-Gunn et al., 2004). The wider significance of these findings remains to be established. A small study found an increase in QT interval, largely within normal limits, during interictal epileptiform discharges in 11 SUDEP cases (Tavernor et al., 1996). Routine autopsies in SUDEP cases, by definition, do not reveal a cardiac cause for death. Specialized cardiac pathology studies have been carried out, with two studies showing myocardial fibrosis (Natelson et al., 1998; P-Codrea Tigaran et al., 2005). In another study, morphologic abnormalities of the cardiac conduction system were observed in four of 10 SUDEP cases, a similar finding to the six of 10 of the control group. The significance of these changes is uncertain (Opeskin et al., 2000).

In relation to possible cardiac mechanisms in SUDEP, we can postulate a number of hypotheses, which are not mutually exclusive:

  • 1Malignant cardiac tachy/bradyarrhythmias or cardiac failure may occur secondary to the ictal discharge (Stollberger and Finsterer, 2004) and/or apnoea/hypoxia.
  • 2Long-term cardiac changes secondary to uncontrolled epilepsy may predispose to excess sudden death. Limited evidence supports this (Natelson et al., 1998, P-Codrea Tigaran et al., 2005).
  • 3Some individuals with epilepsy may have a coexisting “mild” susceptibility to sudden cardiac death, which can manifest in the presence of uncontrolled seizures. This hypothesis could apply equally to acquired or genetic disorders. Coexisting acquired disorders as a risk factor for SUDEP have generally been underinvestigated and are more likely in older age groups. Strict SUDEP definitions currently in use, as well as study methods focusing on younger adults, exclude those with competing acquired cardiac diseases from the “pure” SUDEP category (Davis and McGwin, 2004), even though these may increase the likelihood of dying in a seizure. In younger people, a genetic predisposition would be more likely than acquired disorders.


We wish to explore further the possibility of a coexisting genetic susceptibility to sudden cardiac death. Such a genetic susceptibility may be unrelated to the epilepsy. Alternatively, some variants may confer increased susceptibility both to epilepsy and to sudden cardiac death, reflecting underlying processes common to cardiac and neurologic functions. A number of arguments might be put forward against this hypothesis: (a) An absence of epidemiologic data exists indicating a higher incidence of epilepsy among relatives of patients with inherited susceptibility to arrhythmia; (b) family history of early sudden cardiac death is not reported in SUDEP series; (c) malignant tachyarrhythmias are relatively uncommonly recorded in seizures, with bradyarrhythmias more often reported; and (d) respiratory changes occur more frequently than serious arrhythmias during or after seizures.

The epidemiological data relating to the first argument are not available, and a lack of association cannot be assumed. Scope exists for potentially useful epidemiologic studies looking at the incidence of epilepsy in families of sudden cardiac death victims; furthermore, it may be helpful to study the incidence of syncope in patients with idiopathic epilepsy and their relatives.

With regard to the second argument, although largely unaddressed, two studies, failed to elicit a family history of early sudden cardiac death in SUDEP cases (Nashef et al., 1998; Langan et al., 2005). However, we are not postulating a mendelian trait but a genetic susceptibility, requiring variants in several genes or environmental triggers, which these studies would not have identified. In an autopsy study of sudden death in those aged 5–35 years, sudden cardiac death was reported in a first-degree relative in only 4.5% of decedents, although 56.4% were deemed cardiac, with 29% arrhythmia related (Puranik et al., 2005).

Third, whereas inherited susceptibility to long-QT syndrome and tachyarrhythmias is widely recognized, it is perhaps less widely appreciated that genetic susceptibility is also very well documented in sinus node dysfunction and bradyarrhythmias.

Last, whereas respiratory compromise is clearly a very important factor, arrhythmias in a susceptible individual may be more likely in the presence of apnea/hypoxia (Nashef et al., 1996).

We review the literature for gene mutations implicated in cardiac paroxysmal disorders causing both tachy- and bradyarrhythmias, including sinus node dysfunction (Table 1), with respect to related genes known to cause epilepsy (Table 2) and refer to emerging evidence in support of susceptibility to arrhythmia inherited as a complex trait.

Table 1. Gene mutations involved in paroxysmal cardiac disorders.
KCNQ1 (or KVLQT1)Romano-Ward, Jervell-Lange Nielsen: LQT1 (Geelen et al., 1998)KV7.1 (K channel α subunit)Heart, ear, kidney, intestine, lung, thymus, pancreasHearing loss1. Cardiac events more likely during exercise and increased symptomatic activity (Schwartz et al., 2001)
Short QT syndrome 2 (Bellocq et al., 2004) 2. Not expressed in brain but in same family as KCNQ2/3 (mutations cause BFNC)
Familial atrial fibrillation (Chen et al., 2003) 
KCNH2 (or HERG)Romano-Ward: LQT2KV11.1 (K channel α subunit)Heart, brain, eye, muscle, lung, thymus, adrenal Cardiac events associated with arousal (Schwartz et al., 2001).
Short-QT syndrome 1 (Brugada et al., 2004) 
SCN5ARomano-Ward: LQT3Brugada syndrome (Chen et al., 1998)NaV1.5 (Na channel α subunit)Heart, brain 1. Cardiac events associated with bradycardia (Schwartz et al., 2001)
Sick sinus syndrome 1 (Benson et al., 2003) 2. Mutations in brain-expressed genes SCN1A, SCN2A, and SCN1B cause GEFS+
Progressive cardiac conduction defect (Schott et al., 1999) 
ANK2Romano-Ward: LQT4Ankyrin-2 (Ion channel/transporter-binding protein)Heart, brain Patients often have severe sinus bradycardia and paroxysmal AF (Priori and Napolitano, 2004).
Sick-sinus syndrome with bradycardia (Mohler et al., 2004) 
KCNE1Romano-Ward, Jervell-Lange Nielsen: LQT5 (Geelen et al., 1998)MinK (K channel α subunit)Heart, ear, kidney, intestine, lung, thymus, pancreasHearing lossNot expressed in the brain (α subunit for KCNQ1)
KCNE2Romano-Ward: LQT6MiRP1 (K channel α subunit)Heart, brain α Subunit for KCNH2 and HCN family
KCNJ2 (or HHIRK1)Romano-Ward: LQT7Anderson (Plaster et al., 2001)Kir2.1 (K channel α subunit)Heart, brain, muscle, lung, kidney, placentaDysmorphic features 
Short-QT syndrome 3 (Priori et al., 2005) 
HCN2Heart, brain −/− Mice have spontaneous absence seizures and sinoatrial dysfunction
HCN4Sinus node dysfunction (Schulze-Bahr et al., 2003) Heart, brain −/− Mice die young because of poor cardiac pacemaker function
CACNA1CTimothy syndrome: LQT8 (Splawski et al., 2005, Splawski et al., 2004)CaV1.2 (L-type Ca channel α subunit)Heart, brainWebbing of digits, congenital heart disease, immune deficiency, intermittent hypoglycemia, cognitive abnormalities, autism−/− Mice die young
CACNA1D CaV1.3 (L-type Ca channel α subunit)Heart, brain −/− Mice have slowed pacemaker activity and arrhythmia
RYR2Catecholaminergic polymorphic ventricular tachycardia (Priori et al., 2001)Ryanodine receptor IIHeart, brain 
Arrhythmogenic right ventricular cardiomyopathy type 2 (Tiso et al., 2001) 
Table 2. Some genes known to cause monogenic idiopathic epilepsy
Epilepsy syndromeLocationGeneReference
Benign familial neonatal convulsions20q13.3KCNQ2Singh et al., 1998; Biervert et al., 1998
8q24KCNQ3Charlier et al., 1998; Hirose et al., 1999
Benign familial neonatal infantile convulsions2q24SCN2AHeron et al., 2002; Berkovic et al., 2004
Benign familial infantile convulsions2q24SCN2AStriano et al., 2006
Generalized epilepsy with febrile seizures plus19q13.1SCN1BWallace et al., 1998
2q24SCN1AEscayg et al., 2000
2q24SCN2ASugawara et al., 2001
5q31GABRG2Harkin et al., 2002
Absence epilepsy and febrile seizures5q31GABRG2Baulac et al., 2001; Wallace et al., 2001
Severe myoclonic epilepsy of infancy2q24SCN1AClaes et al., 2001
Autosomal dominant nocturnal frontal lobe epilepsy20q13CHRNA4Steinlein et al., 1995
1q21CHRNB2Fusco et al., 2000
Autosomal dominant juvenile myoclonic epilepsy5q34GABRA1Cossette et al., 2002
Juvenile myoclonic epilepsy6p21EFHC1Suzuki et al., 2004
Autosomal dominant idiopathic generalized epilepsy3q26CLCN2Haug et al., 2003
Autosomal dominant partial epilepsy with auditory features10q24LGI1Kalachikov et al., 2002


These predispose to inherited susceptibility to tachyarrhythmias and bradyarrhythmias. The best known are genes causing long-QT syndrome (LQTS), with an estimated frequency of 1 in 5,000 people (Kass and Moss, 2003). LQTS was first described by Jervell and Lange-Nielson (1957) in an autosomal recessive syndrome, associated with sensorineural hearing loss. LQTS is also transmitted as an autosomal dominant trait, as in Romano-Ward syndrome. LQTS can also be acquired, usually as a result of pharmacologic therapy. The altered ion-channel function results in a defect of cardiac electrical repolarisation (Wang et al., 1995) and produces prolongation of the action potential and propensity to torsades de pointes ventricular tachycardia, syncope, and unexpected sudden death, usually in children or young adults. Diagnosis may be difficult, as the QT interval is normal for ∼10% of the time and is only borderline prolonged for a further 30% (Vincent, 1998). Diagnostic features include prolongation of the heart rate–corrected QT interval (>440 ms) (Geelen et al., 1998) and characteristic T-wave abnormality, most commonly a bifid T-wave.

Lengthening of the QT interval can be caused by increasing the sodium current to prolong the initial phase of the cardiac action potential or by reducing the potassium current to delay the repolarisation phase. The cardiac delayed-rectifier current IK is a major determinant of phase 3 of the cardiac action potential. This comprises two independent components: one rapid IKr and one slow catecholamine-sensitive component, IKs. Not surprisingly, sodium channel (1) and potassium channel (5) genes account for six of the eight LQTS genes with known autosomal dominant mutations. The remaining known long-QT genes are CACNA1C, which encodes the calcium channel that provides the main cardiac calcium current (Gargus, 2006), and ANK2, coding for ankyrin B, a membrane protein that affects the localization of ion channels and transporters (Mohler and Bennett, 2005). Importantly, some of these also cause sinus node dysfunction (Table 1). In three of the LQTS genes, other rarer mutations have been described that shorten the QT interval and also predispose to cardiac arrhythmias and sudden death, especially in the very young (Brugada et al., 2005). Six of the eight long-QT genes are also expressed in the brain: KCNE2, KCNJ2, HERG/KCNH2, SCN5A, CACNA1C, and ANK2. These and other genes predisposing to arrhythmias, and the interactions between them, are discussed later.

Potassium channel genes

Potassium channels drive the repolarizing phase in action potentials.

LQT2 (HERG or KCNH2): LQT2, results from mutations in HERG, a human ether-a-go-go–related gene, which encodes a voltage-gated K+ channel α subunit (Gong et al., 2004) and is abundant in the heart, brain, and other tissues (Wymore et al., 1997). It encodes the α subunit of the cardiac rapidly activating, delayed rectifier (IKr) channel. In the heart, the HERG potassium channel is modulated by MiRP1, a β subunit encoded by KCNE2 (LQT6) (Abbott et al., 1999).

LQT7 (KCNJ2): This gene, which is expressed in the brain and heart (Raab-Graham et al., 1994), encodes the inward rectifier K+ channel Kir2.1. Loss-of-function mutations cause LQTS, including Anderson syndrome, a rare inherited disorder characterized by periodic paralysis, LQT with ventricular arrhythmias, and skeletal developmental abnormalities (Tristani-Firouzi et al., 2002).

LQT6 (KCNE2) and LQT5 (KCNE1): KCNE genes encode a family of single transmembrane domain proteins called MinK-related peptides (MiRPs) that function as ancillary or β subunits of Kv channels. When coexpressed in heterologous systems, they alter Kv channel conductance (McCrossan and Abbott, 2004). KCNE2 (LQT6) encodes the MiRP1 β subunit, which was first described as a modulator of the HERG (LQT2) potassium channel in the heart (Abbott et al., 1999), where it is highly expressed in the sinoatrial node (Moosmang et al., 2001). It also has been shown to be present in the brain, in areas that also express KCNQ2 and KCNQ3 (see later). When expressed in COS cells, MiRP1 associates with the KCNQ2/KCNQ3 complex and accelerates the deactivation kinetics of the potassium channel (Tinel et al., 2000). It has been demonstrated that MiRP1 also acts as an α subunit for the HCN family (see later) (Yu et al., 2001). The related gene KCNE1 (LQT5) codes for MinK, which appears to be the β subunit for KCNQ1 (implicated in LQT1, see later), neither of which appear to be expressed in the brain.

LQT1 (KCNQ1 or KVLQT1): The KCNQ subfamily (KCNQ1-KCNQ5) are all α subunits (Robbins, 2001). All but KCNQ5 are associated with inherited diseases. KCNQ1, the gene mutated in LQT1, encodes the α subunit of the potassium channel conducting the IKs component of the cardiac delayed-rectifier current described earlier. Although not expressed in the CNS, it has significant homology with the brain-expressed KCNQ2 and KCNQ3 (Steinlein et al., 1999), in both of which mutations have been found to be a cause of benign familial neonatal convulsions (Biervert et al., 1998; Charlier et al., 1998; Singh et al., 1998). These genes encode subunits of neuronal M-type K+ channels, key regulators of brain excitability. They are expressed in an overlapping distribution in the brain, with high levels in the hippocampus, neocortex, and thalamus (Rogawski, 2000).

Sodium channel genes

Diseases due to inherited Na+-channel mutations include a diverse group of dominantly inherited paroxysmal disorders from skeletal muscle hyperexcitability to epilepsy and cardiac arrhythmias. Mutations in three sodium channel genes (two α subunits: SCN1A and SCN2A; and one β subunit: SCN1B) are causative in many families with generalized epilepsy with febrile seizures plus (GEFS+). Affected individuals manifest a variety of epilepsy phenotypes including febrile and nonfebrile seizures. SCN1A mutations, in a gain-of-function abnormality, result in persistent inward Na+ current and cause prolonged membrane depolarisation (Lossin et al., 2002). De novo mutations in SCN1A have also been shown to cause severe myoclonic epilepsy of infancy (SMEI) and related conditions (Claes et al., 2001), including intractable childhood epilepsy with generalized tonic–clonic seizures (Rhodes et al., 2005). SMEI has a particularly bad prognosis. Dravet and colleagues reported on a series of 63 patients, 10 of whom had died by the age of 27 of a variety of causes including two sudden deaths (Dravet et al., 1992). It may be significant that SCN1A is also expressed in the heart (Maier et al., 2004). Besides GEFS+, other mutations in the sodium-channel gene SCN2A are implicated in benign familial neonatal–infantile seizures (Berkovic et al., 2004) and benign familial infantile seizures (Striano et al., 2006).

SCN5A (LQT3) encodes the cardiac Na+-channel α subunit responsible for initiating the cardiac action potential (Wang et al., 1995). LQT3 mutations cause a gain of function, delaying ventricular repolarisation and causing QT-interval prolongation. Mutations in SCN5A also cause sick-sinus syndrome. SCN5A has been detected in the human brain (Donahue et al., 2000), and the homologous protein (Nav1.5) was observed in mouse brain cerebral cortex, thalamus, hypothalamus, basal ganglia, cerebellum, and brainstem, clustering at high density in neuronal processes, mainly axons (Wu et al., 2002).

Other LQT genes

ANK2 (LQT4): This encodes ankyrin B, a membrane protein that interacts with many proteins including ion channels and transporters. Like SCN5A, ANK2 mutations cause sick sinus syndrome as well as LQTS. LQT4 families with loss-of-function ANK2 mutations are reported to have bradycardia, sinus arrhythmia, idiopathic ventricular fibrillation, catecholaminergic polymorphic ventricular tachycardia, and risk of sudden death (Mohler et al., 2004). ANK2 is expressed in the brain (Otto et al., 1991).

CACNA1C (LQT8): This gene is discussed later with the other calcium signaling genes.

H channels

Hyperpolarization-activated cation (HCN or h) channels play an important role in pacemaker activity in both cardiac and neuronal cells, including oscillations in thalamocortical relay neurons (Seifert et al., 1999). Four HCN channels are known, all of which are permeable to both K+ and Na+ ions, and have the opposite dependence on membrane potential to voltage-gated K+ channels, opening in response to membrane hyperpolarization and closing with depolarization. Of particular interest are HCN2 and HCN4, which are expressed in mouse thalamic relay neurons (Moosmang et al., 1999). HCN2-deficient mice exhibited spontaneous absence seizures from a pronounced hyperpolarizing shift of the resting membrane potential. They also displayed cardiac sinus dysrhythmia, a reduction of the sinoatrial HCN current (Ludwig et al., 2003). HCN4 is a pacemaker channel that plays an important role in sinus node automaticity in the heart. It is found not only in the cells of sinus nodes but also in other cells of the conduction system (Ueda et al., 2004) and the brain, with the thalamus being the predominant area of HCN4 expression (Seifert et al., 1999; Stieber et al., 2003). Two mutations in HCN4 have been described in patients with sinus node dysfunction (Schulze-Bahr et al., 2003; Ueda et al., 2004).

As we have already seen, MiRP1 (encoded by KCNE2) is a β subunit for the α subunit potassium channels encoded by HERG, KCNQ2, and KCNQ3. It is also a β subunit for the HCN family. In heterologous expression systems, coexpression of KCNE2 with HCN1, 2, or 4 enhances the current and alters the kinetics of activation and deactivation (Yu et al., 2001; Decher et al., 2003; Qu et al., 2004). MirP1 is therefore able to modulate both outward IKr and inward If pacemaker currents and may play a similar role in both cardiac and thalamocortical pacemaker function.

Calcium signaling genes

Three key molecules concerned with cardiac muscle excitation–contraction coupling are the two L-type voltage-gated calcium channel α subunits Cav1.2 and Cav1.3, encoded by genes CACNA1C and CACNA1D (Lipscombe et al., 2004), and the ryanodine receptor II (RyR2), encoded by RYR2 (Lehnart et al., 2004). Both Cav1.2 and Cav1.3 are expressed in the sinoatrial node, where Cav1.3 plays a critical role in pacemaker activity (Mangoni et al., 2003). Human mutations have been described for CACNA1C in Timothy Syndrome (TS, recently referred to as LQT8) (Splawski et al., 2004; Splawski et al., 2005; Gargus, 2006). As originally described, TS is characterized by LQT, webbing of digits (syndactyly), congenital heart disease, immune deficiency, intermittent hypoglycemia, cognitive abnormalities, and autism. In each of several unrelated cases, the same mutation (G406R in exon 8A of CACNA1C) was found, which is likely to cause intracellular Ca2+ overload in multiple cell types (Splawski et al., 2004) and reflects the importance of Cav1.2 in a wide range of functions. In a recent study, two other cases were described with a more severe cardiac phenotype (TS2), including an especially prolonged QT interval, but no syndactyly (Splawski et al., 2005). One of these had a G406R mutation in the alternative exon 8, rather than exon 8A as in TS, whereas the other TS2 case had a different mutation (G402S) also in exon 8. The different expression profiles of the two alternative splice variants of CACNA1C presumably explain the phenotypic differences between TS and TS2. Missense mutations in RyR2 cause ventricular arrhythmias and sudden cardiac death in RyR2 mutation carriers in familial polymorphic ventricular tachycardia (Priori et al., 2001; Lehnart et al., 2004). A recent study also showed that in 49 cases of sudden unexplained deaths, seven were due to mutations in RYR2, causing arrhythmogenic right ventricular dysplasia type 2 (ARVD2) (Tester et al., 2004).

These three calcium signaling genes are also highly expressed in the brain (Hell et al., 1993; Ludwig et al., 1997). Electrophysiologic, immunocytochemical, and in situ hybridization studies have shown that thalamic neurons express different calcium-channel subtypes including Cav1.2 and Cav1.3 (Pape et al., 2004). Importantly, heart and brain share key calcium signaling functions, including the well-characterized thalamocortical relay neurons (Kano et al., 1995; Budde et al., 2000; 2002). Elsewhere in the CNS, such as the cerebellum, Cav1.2 and Cav1.3 have also been implicated in oscillatory currents (Liljelund et al., 2000). Cellular Ca2+ signaling is an important factor in the control of neuronal metabolism and electrical activity. Although the role of these Ca2+ channels is well established for muscle, less is known about their expression and roles in the CNS, especially in the human brain. In the brain, RyRs are localized primarily to endoplasmic reticulum, where they regulate intracellular Ca2+ concentration. Widespread expression of RyR isoforms in the human hippocampus and cerebellum suggests that ryanodine-receptor proteins may have a central role in Ca2+ signaling and Ca2+ homeostasis in the human CNS (Martin et al., 1998).

T-type voltage-gated calcium channel α subunits may also play a role in pacemaker function in heart and brain. Of these genes, CACNA1H, encoding Cav3.2, is expressed in many tissues, including brain and heart (Cribbs et al., 1998). Mice lacking this gene have several cardiac defects, whereas in humans, 12 different CACNA1H mutations have been identified in 118 Han Chinese patients with childhood absence epilepsy (CAE) (Chen et al., 2003; Vitko et al., 2005). CAE patients, however, are considered at low risk of SUDEP.


The realization of the importance of respiratory mechanisms at a vulnerable stage in development of the infant led to dramatic success in reducing the incidence of SIDS, simply by recommending the supine rather than the prone position (Dwyer et al., 1991). Meanwhile, recognising that SIDS was likely to be multifactorial, Schwartz and colleagues prospectively tested their hypothesis that QT-interval prolongation may contribute to SIDS by increasing the risk of life-threatening ventricular arrhythmias. They recorded electrocardiograms on the third or fourth day of life in 34,442 newborns and monitored them prospectively for 1 year. Twenty-four of 34 deaths followed-up were SIDS and had a longer corrected QT interval (QTc) than survivors [mean (±SD), 435 ± 45 vs. 400 ± 20 ms; p < 0.01) or infants dying of other causes (393 ± 24 ms; p < 0.05). Twelve of 24 SIDS victims, but none of the other infants, had a prolonged QTc >440 ms. The odds ratio for SIDS in infants with a prolonged QTc was 41.3 (95% CI, 17.3–98.4) (Schwartz et al., 1998). Later, this group reported a sporadic SCN5A mutation in a near-miss SIDS case when a collapsed infant, living very close to a hospital, was successfully resuscitated from a ventricular fibrillation (Schwartz et al., 2000). Other examples have been found since then. A de novo mutation in KCNQ1 (LQT1) was identified by the same group in a child who died of SIDS (Schwartz et al., 2001). A HERG (LQT2) mutation also was found as a novel missense mutation in a case with SIDS associated with torsades de pointes tachycardia (Christiansen et al., 2005). A study found SCN5A (LQT3) mutations in two of 93 SIDS cases (Ackerman et al., 2001), and recently Plant et al. (2006) reported an increased risk of SIDS in infants homozygous for another SCN5A variant, as discussed later. The exact contribution to SIDS of genetic susceptibility to cardiac arrhythmia is still to be established, but direct causation has clearly been proved in some cases of SIDS, as well as in stillbirths (Schwartz, 2004). In the same way, it may be necessary for examination of candidate genes from many SUDEP victims before a role for cardiac genes can be established or excluded.


A detailed examination of this field is beyond the scope of this review. However, as in epilepsy, in considering inherited liability to arrhythmias, we must consider more-complex gene interactions beyond single-gene effects. Not surprisingly, mutations were first identified in mendelian pedigrees, but evidence in favour of combined influence with other genetic factors, acquired diseases, other medication, or lifestyle is gradually accumulating (Viswanathan et al., 2003; Ravn et al., 2005).

A variant allele (S1103Y) of the cardiac sodium-channel gene SCN5A, carried by 13% of African Americans, was found in a study to have a subtle effect on risk, manifesting in the presence of additional acquired risk factors, with most carriers never having an arrhythmia (Splawski et al., 2002). The same variant in a white family (where it is very rare) was associated with considerable risk of syncope, ventricular fibrillation, and sudden death (Chen et al., 2002). It was recently reported that an excess of homozygotes carrying both copies of the minor allele was present in 133 African-American SIDS cases, suggesting a 24-fold increased risk of SIDS for this rare genotype (Plant et al., 2006). DNA variants also predispose to drug-associated “acquired” long-QT syndrome. Another study screened coding regions of KvLQT1, HERG, and SCN5A in an “acquired” LQTS cohort and a control group and found mutations in a small proportion (Yang et al., 2002). A mutation in SCN5A was found in an elderly Japanese woman with documented QT prolongation and torsades de pointes during treatment with cisapride (Makita et al., 2002). Of particular interest is a study looking at single nucleotide polymorphisms (SNPs) in five long-QT syndrome genes: KCNQ1 (LQT1), HERG (LQT2), SCN5A (LQT3), KCNE1 (LQT5), and KCNE2 (LQT6) in a cohort of 141 white subjects. Four SNPs were associated with QTc interval, one in KCNE1, one in KCNE2, and two in SCN5A (Aydin et al., 2005). Two of the genes are expressed in the brain (Ravn et al., 2005).


Research in this field must be guided by the potential for the prevention of SUDEP. Seizure prevention, prediction, detection, and response to seizures are all important, as is the effect of therapy (mono- or polytherapy) on ictal and postictal states in relation to depressed consciousness and respiratory and cardiac parameters. Further studies of long-term cardiac rhythm recordings using implantable devices in less-selected cohorts, larger studies of ECG changes and QT interval in people with epilepsy, epidemiologic clinical studies of the association between syncope and inherited arrhythmias and epilepsy all must be considered. Although, at present, bridging evidence between cardiac inherited gene determinants and SUDEP is lacking, the possibility of a coexisting “mild” susceptibility to sudden cardiac death, be it independent of or related to the epilepsy, which becomes symptomatic in the presence of uncontrolled seizures, must be explored.


Acknowledgment:  We thank the Fund for Epilepsy and Guy's and St Thomas's Charitable Trust for funding Dr. Neeti Hindocha.