Clinical manifestations

Presentation and course

A landmark sudden unexpected death in epilepsy study is the Mortality in Epilepsy Monitoring Units Study (MORTEMUS), a multicenter, retrospective analysis of deaths in epilepsy monitoring units (103). In this study, 29 deaths in patients undergoing live monitoring were identified. There were 16 sudden unexpected death in epilepsy cases, 9 near-sudden unexpected death in epilepsy cases (2 of them subsequently fatal), and four deaths of other etiologies. There was only one pediatric case in a 10-year-old child (near-sudden unexpected death in epilepsy). The remainder were adults, with a mean age of 41.3 years (19 to 62). All monitored sudden unexplained deaths in epilepsy (11 of 16) and all near-sudden unexplained deaths in epilepsy had a seizure immediately prior to cardiorespiratory arrest. In all monitored sudden unexpected death in epilepsy cases and in 7 of 9 of near-sudden unexpected death in epilepsy cases, including both fatal near-sudden unexpected death in epilepsy cases, the terminal seizure was a generalized convulsive seizure. For the remaining 2 of 9 near-sudden unexplained deaths in epilepsy, the seizure was classified as “complex partial seizure”. In all but one sudden unexpected death in epilepsy case and in both fatal near-sudden unexplained deaths in epilepsy, cardiorespiratory arrest occurred at night. Nine of 11 sudden unexpected death in epilepsy cases were found prone.

Prognosis and complications

Because sudden unexpected death in epilepsy is a terminal event, this section does not apply.

Clinical vignette

MORTEMUS summarizes the most common phenotype described in previous sudden unexpected death in epilepsy case reports (116; 12). Sudden unexpected death in epilepsy tends to affect young and middle-aged adults, who are commonly found overnight in prone position with signs of having suffered a generalized convulsive seizure (120). However, sudden unexpected death in epilepsy does not exclusively happen after generalized convulsive seizures and sudden unexpected death in epilepsy and near-sudden unexpected death in epilepsy cases have been described to occur also without a preceding seizure (72; 08).

A 23-year-old right-handed man first presented to the epilepsy clinic with intractable epilepsy, characterized by a vague, nonspecific cephalic aura followed by generalized tonic-clonic seizures. He had begun to have seizures at the age of 13 years, and had at least one seizure a month despite repeated medication changes for intractability. He had no epilepsy risk factors or significant family history. He had normal routine bloodwork, no routine EEG abnormalities, and was MRI lesion negative. An epilepsy monitoring unit assessment showed rare 5 to 10 second bursts of generalized 3 to 4 Hertz spike wave discharges. A diagnosis of a genetic generalized epilepsy was made and his medication changed from carbamazepine and levetiracetam combination to 1 of lamotrigine (200 mg twice daily) and valproic acid (1500 mg twice daily). His seizure frequency decreased to one seizure every 3 to 4 months. A year after his in-patient EEG monitoring discharge, he was found dead in bed one morning by his mother. Autopsy revealed petechial conjunctival hemorrhages and mild pulmonary edema. No anatomical or toxicological cause of death was found, and his death was deemed natural. A diagnosis of definite sudden unexpected death in epilepsy was made. Although the clinical vignette described above is the prototypical scenario for sudden unexpected death in epilepsy, sudden unexpected death in epilepsy and near-sudden unexpected death in epilepsy have been described in patients without evidence of a terminal seizure (72; 08). In a case series of two definite sudden unexpected death in epilepsy and one probable sudden unexpected death in epilepsy adult patients, and in a case report of a near-sudden unexpected death in epilepsy in a pediatric patient, the terminal event occurred without a prior generalized convulsive seizure or typical clinical seizure (72).

Biological basis

Etiology and pathogenesis

Despite increasing awareness and multiple studies of seizure patho-phenomenology, the ultimate pathophysiology of sudden unexpected death in epilepsy is still unknown (30). Evidences for both cardiovascular and respiratory dysfunction have been reported and are reviewed below.

Breathing dysfunction. Breathing dysfunction has been reported before, during, and after epileptic seizures (141; 11; 63; 64; 99; 138).

Ictal central apnea occurs in 36.5% to 36.9% of seizures and is a semiological feature of focal onset epileptic seizures, predominantly in temporal lobe epilepsy (65; 64; 138). Ictal central apnea can start before, at, or after EEG seizure onset and is usually a self-limiting phenomenon (64). However, prolonged ictal central apnea can occur, is associated with severe hypoxemia, and may pose increased risk of death (64; 138). Postconvulsive central apnea can occur after 18% to 22.1% of generalized convulsive seizures, with or without ongoing EEG seizure discharges (139). Apart from periictal central apnea (ictal central apnea/postconvulsive central apnea), ictal and postictal obstructive apnea, such as with laryngospasm, have also been described in humans and in animal models of sudden unexpected death in epilepsy, although their incidence is lower than that of periictal central apnea (06; 86; 141; 127; 83; 118; 64).

Availability of cardiorespiratory data in 10 of 16 sudden unexpected death in epilepsy cases in MORTEMUS allowed identification of a relatively consistent pattern leading to death. This comprised a postconvulsive, immediate increase in breathing rate followed by bradycardia and transient periods of apnea and asystole, along with postictal generalized EEG suppression. Respiratory collapse was terminal in a third of patients and although transient restoration was found in the remainder, respiratory deterioration followed. Terminal asystole was always preceded by terminal apnea, suggesting apnea as the final mechanism of death, in line with previous animal models of sudden unexpected death in epilepsy (54; 103).

Despite the consistent observation of breathing arrest preceding sudden unexpected death in epilepsy, its physiopathological basis is still a topic under investigation. One of the potential explanations for postictal apnea and hypoventilation is an excessive activation of oxygen conserving reflexes (14; 15). Another plausible explanation is a failure of mechanisms of autoresuscitation and blunted hypercapnic response to CO2 increase in the aftermath of a generalized convulsive seizures (105; 45; 35). Most of the available models of breathing dysfunction after epileptic seizures are animal based. Further information is provided under the “molecular mechanisms and animal models section.”

Cardiovascular dysfunction. Seizure-induced arrhythmias are frequent and usually self-limited (89; 136). The most frequently noted periictal rhythm change is sinus tachycardia, noted in up to 80% of epileptic seizures (136). This observation is concordant with signs of increased sympathetic tone during epileptic seizures, such as increased levels of circulating catecholamines, decrease in heart rate variability, and elevated electrodermal activity (131; 87; 88; 113; 39; 114). Less frequent dysrhythmias are bradycardia and its extreme manifestation, asystole (136). AV conduction blocks, atrial flutter/fibrillation, and ventricular tachycardia/fibrillation have also been reported (136). Bradycardia, defined as a heart rate under the first centile of normal heart rate frequency, accounted for 25 of 162 (15.4%) of total seizure-related arrhythmias in a metaanalysis (136). In the same metaanalysis, prevalence of ictal asystole, defined as an R-R interval of more than 3 or 4 seconds (criteria differed among studies), was found to be 0.177% (95% CI 0.177-0.178%) (136), although the rate of ictal asystole in studies with long-term EKG recordings has been reported to be 1.6% to 3.2%. Both ictal bradycardia and asystole were only found in seizures of focal onset (136). Postictal asystole usually occurred in the context of focal seizures evolving to generalized convulsive seizures (136). Ictal asystole has been proposed as benign and self-limiting, even as a mechanism of seizure termination because asystole-induced reduction in cerebral blood flow potentially aborts perpetuation of the seizure discharge (108). Accordingly, ictal asystole has not been reported in sudden unexpected death in epilepsy cases and can recur in up to 40% of cases, suggesting a benign semiological phenomenon rather than a causative sudden unexpected death in epilepsy mechanism (136; 44; 128). In contrast, postictal asystole has been described in sudden unexpected death in epilepsy and near-sudden unexpected death in epilepsy cases, suggesting a different physiopathology and potentially lethal implications (136; 139).

A prospective analysis of cardiac long-term recordings and the incidence of potentially deleterious cardiac arrhythmias in patients with epilepsy have shown conflicting results. Although in one study no clinically relevant cardiac arrhythmias were reported in patients with epilepsy, two other studies found cardiac arrhythmias that required additional interventions in 1.6% and 9.7% patients, respectively (135; 110; 115). These discrepancies are likely due to methodological differences because one of the studies excluded patients with suspected ictal asystole (135).

A deeper state of sleep five minutes prior to the seizure has been associated with a slower nadir in postictal heart rate (107). In addition, patients with sudden unexpected death in epilepsy have been found to have an absence of significant changes in heart rate during and after hyperventilation, suggesting an abnormal cardiac autonomic response to the sympathetic stimulation triggered by hyperventilation (123). In generalized convulsive seizures, the presence of potentially high-risk cardiac arrhythmias has been associated with duration of periictal hypoxemia, although no association was found with the magnitude of oxygen saturation drop (94). Moreover, postictal asystole has been demonstrated concurrently with or preceded by alveolar hypoventilation including human cases and animal models of sudden unexpected death in epilepsy (86; 54; 94). However, the potential causality between cardiac and respiratory dysfunction is still unclear. Changes in blood pressure have been described during and after seizures (51; 16). In a cohort study of focal seizures, ictal hypertension was found in 15 of 57 (26.3%) seizures, whereas ictal hypotension was found in 5 of 57 (8.8%) seizures (52). The physiopathology underlying blood pressure changes during epileptic seizures may involve the activation of cortical and subcortical structures responsible for blood pressure control, such as the medial prefrontal cortex, insular cortex, and amygdala, structures that may be damaged in patients with epilepsy (07; 66; 61; 60). Two different patterns in blood pressure changes have been described during and after generalized convulsive seizures, although data are scarce due to methodological difficulties inherent to blood pressure data acquisition in the seizure setting (87). One pattern consists of simultaneous increase of blood pressure and heart rate, whereas the second one shows an initial raise in blood pressure followed by a quick drop of blood pressure to baseline or even below preictal values, although with sustained tachycardia (43). The first pattern is another manifestation of the aforementioned increase in sympathetic tone during epileptic seizures (136; 87; 88). Several explanations have been proposed for the second pattern. First, a spread to Brodman area 25 may be responsible for a decrease in blood pressure because stimulation of this area elicits a significant decrease in blood pressure (61). Secondly, muscular hyperemia after generalized convulsive seizures may lead to a decrease in peripheral vascular resistance, resulting in a decrease in systemic blood pressure (87). Another potential explanation is primary cardiac dysfunction, in the setting of a seizure-induced Takotsubo cardiomyopathy, although such a scenario transpires over a period of many hours to days (119; 26).

The occurrence of changes in heart and coronary vasculature due to repeated surges in catecholamines and hypoxemia in patients with epilepsy, leading to electrical and mechanical dysfunction, has been labelled as the “epileptic heart.”

Postictal generalized electroencephalographic suppression and cerebral shutdown. Postictal generalized electroencephalographic suppression is defined as the immediate postictal (within 30 seconds) generalized absence of electroencephalographic activity higher than 10 µV in amplitude (70). Postictal generalized electroencephalographic suppression follows generalized convulsive seizures in 27% to 74% of cases and is a feature of the generalized convulsive seizures in those patients (109; 67; 139). Prolongation of postictal generalized electroencephalographic suppression is a marker of seizure severity and has been suggested as a premortem risk identifier and was noted in all cases of monitored sudden unexpected death in epilepsy in the MORTEMUS study (103). In a case control study, postictal generalized electroencephalographic suppression duration above 50 seconds conferred an almost 9-fold increase in sudden unexpected death in epilepsy risk (OR 8.9; CI 95% 1.47-53.46) (70). However, other studies have not found significance (55), possibly due to methodological issues. What seems apparent, however, is that postictal generalized electroencephalographic suppression is frequently found associated with other markers of seizure severity.

Postictal generalized electroencephalographic suppression mechanisms are poorly understood (112). It is more commonly seen in seizures arising from sleep (69; 92). It is also associated with longer postictal immobility periods and generalized convulsive seizures-induced respiratory dysfunction, defined by lower oxygen desaturation nadir value, longer hypoxemia (SpO₂ < 90%) duration, and a higher rise in end tidal CO₂ (112; 59; 99). Conversely, early administration of oxygen therapy after generalized convulsive seizures has been associated with a lower incidence of postictal generalized electroencephalographic suppression and shorter hypoxemia and postictal generalized electroencephalographic suppression durations (111; 02; 99; 140). Postictal generalized electroencephalographic suppression is not related to total generalized convulsive seizures duration (112; 02) but is associated with longer tonic phase duration and with the presence of bilateral tonic symmetric extensor posturing (decerebration) (126; 02; 92). Postictal generalized electroencephalographic suppression duration is also longer in the presence of decerebration compared to other tonic phase semiology types (140). Postictal generalized electroencephalographic suppression is also linked to postictal tonic electromyographic activity, with direct correlation between postictal generalized electroencephalographic suppression duration and tonic electromyographic duration (93; 139). This suggests that postictal tonic electromyographic activity may be a good surrogate for postictal generalized electroencephalographic suppression (11). Lastly, postictal generalized electroencephalographic suppression duration has also been associated with higher electrodermal activity and sympathetic overactivation (97).

Neuroimaging. Structural and functional abnormalities have been reported in sudden unexpected death in epilepsy and patients with generalized convulsive seizures. Morphometric and cortical thickness studies have shown decreased volume in the posterior thalamus bilaterally, in the medial and orbital frontal cortex, temporal cortex, and brainstem (50; 143; 91). Conversely, cortical thickening has been reported in postcentral gyri and limbic structures such as the insula, amygdala, and cingulate cortex (91). Alterations in functional connectivity between cortical and subcortical studies have also been described (03). Peri-ictal hypoxia accompanying generalized convulsive seizures has been associated to the extent of regional brain volume loss in key structures for cardiorespiratory homeostasis (05). Further studies are needed to elucidate the relationship between these neuroimaging findings and sudden unexpected death in epilepsy risk (04). Premortem use of MRI to determine sudden unexpected death in epilepsy risk is currently not recommended or utilized.

Pathology. The diagnosis of sudden unexpected death in epilepsy is based on the absence of positive findings in autopsy studies that conclusively explain mechanism of death (85). However, postmortem examinations of sudden unexpected death in epilepsy patients often show abnormalities that could be potential contributors to death or could be the cause or consequence of seizures and epilepsy or other comorbidities either related or unrelated to epilepsy (84).

Pulmonary pathology. Pulmonary edema is the most frequent pulmonary finding in sudden unexpected death in epilepsy patients (71%), followed by pulmonary hemorrhage (3%) and lung hyperinflation (0.3%) (84). Pulmonary edema has also been described after seizures in living epilepsy patients based on bilateral edema on chest x-ray following generalized convulsive seizures (56). Moreover, the presence of postictal pulmonary x-ray abnormalities is associated with seizure duration (56). The etiology of pulmonary edema is unknown and both neurogenic and negative pressure (ie, due to laryngospasm) are plausible (124; 13). However, the extent of the edema observed does not account for death.

Cardiac pathology. The most common cardiac finding in postmortem examination of sudden unexpected death in epilepsy patients is interstitial fibrosis, followed by myocyte hypertrophy and vacuolization (84). However, evidence of a higher incidence of cardiac fibrosis in sudden unexpected death in epilepsy patients compared to controls was not found in a study (28). Coronary lesions have been found to be higher among sudden unexpected death in epilepsy patients compared to controls (147).

Brain pathology. Histopathologic studies of sudden unexpected death in epilepsy patients have shown focal gliosis within amygdaloid subnuclei as well as an increased right parahippocampal volume, in concordance with neuroimaging studies (129; 143; 117). Moreover, a decrease in adenosine A2A receptors in the temporal lobe and an increase of amygdalar adenosine A1 receptor expression has been noted in high sudden unexpected death in epilepsy risk patients (95). Neuropathological studies have also demonstrated a reduction in somatosatin neurons, neurokinin 1 receptor in the ventrolateral medulla, as well as a reduction of galanin and tryptophan hydroxylase in the ventrolateral medulla and raphe (96). These findings may represent the pathological basis for cardiorespiratory dysfunction in sudden unexpected death in epilepsy patients (80; 96; 95). No markers of inflammation or acute neuronal injury have been found to be different in sudden unexpected death in epilepsy patients from nonsudden unexpected death in epilepsy epileptic patients (79).

Genetics. There is no known sudden unexpected death in epilepsy gene, although certain genetic mutations may confer increased risk (58). Evidence comes from genetic analyses of ion channel mutations expressed in both cardiac and brain tissue, resulting in syndromes that have increased risk of sudden death, such as long QT and Dravet syndromes. However, genetic studies in sudden unexpected death in epilepsy are still limited by availability of DNA samples of sudden unexpected death in epilepsy patients extracted either during life or postmortem evaluation (24).

Long QT syndrome. Long QT syndrome is a genetically heterogeneous cardiac channelopathy that produces abnormally prolonged repolarization of cardiomyocyte action potentials (142). This translates to a prolonged QT interval that predisposes to ventricular tachyarrhythmias such as torsade de pointes, which can degenerate into ventricular fibrillation and death (142). There are 17 long QT syndrome subtypes with monogenic mutations of 15 autosomal genes with variable penetrance (142). Long QT syndrome is autosomal dominant except long QT syndrome associated with sensorineural deafness (Jervell and Lange-Nielsen syndrome), which is autosomal recessive. De novo pathogenic mutations are infrequent. Almost 50% of long QT syndrome cases are caused by a mutation in the KCNQ1 gene (long QT syndrome type 1). This gene encodes the alfa subunit of the voltage gated potassium channel Kv7.1, the mutation of which is responsible for the abnormally slow repolarization current in cardiomyocytes. A mutation in KCNH2, which encodes for Kv1.1, has been associated with long QT events triggered by auditory stimuli (long QT syndrome type 2) and a gain-of-function mutation in SCN5A, and encoding Nav 1.5 has been associated with gastrointestinal symptoms (long QT syndrome type 3). Of note, SCN5A loss-of-function mutation are also associated with Brugada syndrome (146). Mutations in KCNQ1, KCNH2, and SCN5A comprise the vast majority of long QT syndrome cases (53). Kv7.1 has been found in brain tissue and mice bearing the human mutation in this channel show seizures and malignant cardiac arrhythmias (42). In addition, presence of Kv7.1 neurons within mouse vagal brainstem suggests that a mutation could disrupt not only cardiac but also respiratory physiology (42). Both KCNH2 and SCN5A are also expressed in neuronal tissue in animals (31; 33). Patients with KCNH2 mutations (long QT syndrome type 2) have a seizure phenotype more common (47%) than long QT syndrome types 1 (22%) and 3 (25%) (53). Overall, 27% of patients with long QT syndrome have an epilepsy phenotype defined as a personal or close family member history of epilepsy (53).

Dravet syndrome (severe myoclonic epilepsy of infancy) and other genetic developmental encephalopathies. Dravet syndrome is an intractable epileptic encephalopathy characterized by recurrent febrile seizures during the first year of age, multiple seizure types, developmental delay, and severe neurologic deficits, with an increased risk for sudden unexpected death in epilepsy (77; 25). A mutation in the SCN1A gene, which encodes for the alpha subunit of the voltage-gated sodium channel, has been found in 70% to 80% of the cases, mainly occurring de novo, although mutations in other genes have been described (23; 77). Expression of SCN1A has been found in heart and brain tissues (76). Patients with Dravet syndrome have decreased heart rate variability and periictal respiratory dysfunction (57). In a mouse model of Dravet syndrome and sudden unexpected death in epilepsy, blockade of peripheral muscarinic receptors did not prevent postictal apnea, bradycardia, and death, whereas selective central blockade of medullary muscarinic receptors did, suggesting a primary respiratory failure as the cause of cardiac arrest and death (57). It is assumed that death is due to fatal arrhythmias, although there is no direct evidence of this. In one study of sudden unexpected death in epilepsy rates among patients with genetic developmental and epileptic encephalopathies in which 12 genes were studied, 16 out of 20 sudden unexpected death in epilepsy cases occurred in patients with SCN1A mutations (12 of these with Dravet syndrome).The remaining 4 out of 20 cases were seen in patients with pathogenic variants of SCN2A, SCN8A, and STXBP1 (32).

Other genes. Sudden unexpected death in epilepsy accounts for at least 43% of deaths in patients with isodicentric chromosome 15 duplication syndrome (dup15q). Other genes related to epilepsy, sudden death, and cardiac disease include but are not limited to: FBN1, HCN1, SCN4A, EFHC1, CACNA1A, SCN11A, SCN10A, and DEPDC5 (24).

Molecular mechanisms and animal models.

Serotonergic and purinergic mechanisms. Animal models of sudden unexpected death in epilepsy and sudden unexpected death in epilepsy prevention suggest serotoninergic and adenosinergic pathways may play a significant role in sudden unexpected death in epilepsy and sudden infant death pathophysiology (35). Serotoninergic neurons in the central nervous system are located in the brainstem, primarily within the midline raphe nuclei of the midbrain, pons, and medulla (101). These neurons project throughout the central nervous system, including cortex, hippocampus, and other brainstem nuclei, such as the major respiratory control centers (101). Serotoninergic neurons in midbrain act as chemoreceptors activated in response to decreases in pH induced by hypercapnia. This in turn activates thalamocortical pathways responsible for arousal (19). Additionally, they project to the preBozinger complex and nucleus of the solitary tract to enhance their breathing pacemaker activity (19). Therefore, serotonin plays a crucial protective role against hypercapnia. In adult Lmx1bf/f/p mice, which lack more than 99% of serotoninergic neurons in the central nervous system, induction of seizures results in lower seizure threshold and increased seizure-induced mortality (18). Respiratory failure is followed by terminal asystole (18). Mortality in both Lmx1bf/f/p mice and controls (Lmx1bf/f) decreases with administration of mechanical ventilation during seizures or serotonin receptor (5-HT2a) agonist pretreatment (18). Administration of citalopram, a selective serotonin reuptake inhibitor, does not reduce mortality in Lmx1bf/f/p mice but it does in controls (18).

DBA/1 and DBA/2 mice are serotonin deficiency models and susceptible to seizure induced respiratory arrest (133; 37; 134). In these, acoustic stimulation triggers generalized seizures and post seizure, apneic sudden death if resuscitation is not rapidly instituted (37; 134). Cardiac rhythm is consistently detectable at onset of respiratory arrest and arrhythmia does not occur until several seconds after apnea onset (37). Death is prevented by prompt respiratory support with mechanical ventilation, oxygen administration, or prior administration of serotoninergic agents or adenosine antagonists (37; 38; 36). Similarly, in Scn8a mutant mice, apnea initiated during the tonic phase was noted in all the cases of seizure induced death preceding asystole; although it was also noted in nonfatal seizures, suggesting that apnea was necessary but not sufficient to produce death (144). These findings suggest primacy of breathing failure in sudden unexpected death in epilepsy rather than a cardiac trigger (37). Conversely, adenosine promotes seizure termination and suppresses cardiorespiratory function by activation of its A1 receptors whereas blockage of A1 receptors or activation of A2a receptors has the opposite effect (34; 47; 17; 95).

Spreading depolarization. Spreading depolarization has been noted in Kv1.1 KO and SCN1A mice, producing cardiorespiratory arrest preceded by EEG suppression and apnea (01). It is postulated that some high-sudden unexpected death in epilepsy risk individuals undergo such a phenomenon in the postseizure state, shutting down cardio-respiratory brainstem nuclei that would otherwise affect homeostatic return of cardiac and breathing function to the preseizure state. Homozygous Cacna1a mice with missense mutation S218L in the alfa subunit of the P/Q voltage-dependent calcium channel show nonfatal and fatal seizures during adulthood (74). Seizure-induced brainstem spreading depolarization has been demonstrated in this model too, suggesting spreading depression is a viable hypothesis (74).

Epidemiology

Incidence of sudden unexpected death in epilepsy varies considerably depending on the population studied. According to a metaanalysis, overall sudden unexpected death in epilepsy incidence is 0.5 per 1000 patient-years (95%CI 0.31-1.08) (46). During childhood, sudden unexpected death in epilepsy incidence is 0.22 per 1000 patient-years (95% CI 0.16-0.31) and it increases to 1.2 per 1000 patient-years (95% CI 0.64-2.32) in adulthood, with the highest incidence among epilepsy surgery candidates, in whom it ranges from 6.3 to 9.3 per 1000 patients-year (130).

Multiple studies have focused on identifying risk factors for sudden unexpected death in epilepsy. To date, the major risk factor for sudden unexpected death in epilepsy is the presence and frequency of generalized convulsive seizures and ongoing seizures despite adequate treatment (48; 46; 121). According to a combined analysis of case control studies, patients with generalized convulsive seizures have a 10-fold increased risk of sudden unexpected death in epilepsy compared to those without generalized convulsive seizures (48; 46). In patients with three or more generalized convulsive seizures per year, the risk rises 15-fold (48; 46). The same combined case-control study pointed out a potential deleterious effect of polytherapy in patients with generalized convulsive seizures (48). However, after adjusting individual antiepileptic drugs and number of antiepileptic drugs for generalized convulsive seizures frequency, they did not find increased risk, reinforcing current evidence that generalized convulsive seizures frequency is the strongest risk factor for sudden unexpected death in epilepsy (49; 106; 145; 121). In fact, when adjusting by generalized convulsive seizure frequency in one study, polytherapy with three or more antiepileptic drugs has been associated with a reduction of sudden unexpected death in epilepsy risk by two thirds, whereas a poor adherence has been associated with increased sudden unexpected death in epilepsy risk (122). No antiepileptics have been associated with an increase in sudden unexpected death in epilepsy risk. Although a safety warning was issued on lamotrigine regarding a potential proarrhythmogenic effect based on in vitro results, (41), further studies have demonstrated a lack of association between sudden unexplained death risk and lamotrigine (20; 90). Levetiracetam is the sole medication in monotherapy that has been associated with a decreased sudden unexpected death in epilepsy risk even after adjusting for generalized convulsive seizure frequency (122). An increased risk of sudden unexpected death in epilepsy has also been reported in patients with early onset (< 16 years) and longstanding (> 15 years) epilepsy (48). However, these associations were not corroborated in a population-based case-control study when age at onset and epilepsy duration were adjusted by generalized convulsive seizures frequency (121). Sudden unexpected death in epilepsy risk was reported to be higher in males compared to females in one study (48), but gender differences have not been replicated in more recent studies (121).

Death is frequently nocturnal and risk of sudden unexpected death in epilepsy is 2.6-fold (CI 95% 1.3-5.0) increased in patients with nocturnal seizures compared to those without (68). In a case control study, nocturnal generalized convulsive seizures were associated with a 9-fold risk for sudden unexpected death in epilepsy. However, there was no association between sudden unexpected death in epilepsy and nocturnal nongeneralized convulsive seizures (121).

An additional proposed risk factor for sudden unexpected death in epilepsy is the prone position (73). However, this is the final position sudden unexpected death in epilepsy cases are frequently found in, and not necessarily the patient’s natural sleep position. Thus, proning may be the consequence of forced ictal body version rather than sleeping prone (71).

Other clinical risk factors associated with sudden unexpected death in epilepsy after adjustment for generalized convulsive seizures frequency are a history of substance abuse or alcohol dependence. No increased risk for sudden unexpected death in epilepsy was found with psychiatric comorbidities once adjustment for generalized convulsive seizures frequency was applied (121). Finally, living conditions have an influence in sudden unexpected death in epilepsy risk. Compared with sharing a bedroom, living alone or sharing a household but not a bedroom are associated with an increased risk for sudden unexpected death in epilepsy, even after adjustment for generalized convulsive seizures frequency (121). A low socioeconomic status has been associated with an increased risk of all-cause mortality in patients with epilepsy and an increased risk of sudden unexpected death in epilepsy as well. These disparities may be explained by lower access to specialty care and medications as well as poorer understanding of their disease in those patients from lowest-income communities compared to those from the highest income (22).

In spite of the aforementioned risk factors, sudden unexpected death in epilepsy can also occur in a minority with relatively benign phenotypes, such as benign focal epilepsies of childhood, or without apparent intractability where a first ever or second ever seizure is fatal (137). Therefore, further research is needed in order to understand sudden unexpected death in epilepsy physiopathology and identify those patients at higher risk.

Prevention

Because the major sudden unexpected death in epilepsy risk factor is frequent generalized convulsive seizures (48; 46) and no targeted preventive strategies exist, the main sudden unexpected death in epilepsy prevention approach currently is achievement of seizure control as far as possible (46). Active management of epilepsy likely results in sudden unexpected death in epilepsy risk reduction; this includes aggressive pharmacological treatment, epilepsy surgery, vagal nerve or deep brain stimulation, and responsive neurostimulation (46; 27; 104; 122). In fact, a declining pattern of adherence to antiseizure medications has been associated with increased risk of sudden unexpected death in epilepsy (125). Patients with nocturnal seizures are at a higher risk and physical nocturnal supervision or supervision through remote listening devices is recommended, although the evidence for this is weak (68; 75). Advances in wearable and bed-based devices may facilitate sudden unexpected death in epilepsy risk management through multiple pathways (78; 102). First, improvement of seizure frequency counts may guide treatment optimization. Second, caregiver notification during a seizure allows immediate intervention, likely decreasing risk. Lastly, biosignals recorded interictally and during seizures may help identify patients at higher risk (102). Considering the potential role of serotonin and adenosine in sudden unexpected death in epilepsy physiopathology, specific pharmacological strategies have been suggested targeting their respective pathways (100; 62; 82). Administration of fluoxetine has been shown to prevent seizure-induced respiratory arrest in a mouse model of sudden unexpected death in epilepsy (133; 38). Moreover, selective serotonin reuptake inhibitors in humans are associated with reduced likelihood of ictal oxygen desaturation in patients with partial seizures and a decreased incidence of ictal central apnea (10; 62). Therefore, selective serotonin reuptake inhibitors have been proposed as potential prophylactic treatment for sudden unexpected death in epilepsy, although prospective clinical trials are needed (62). Similarly, adenosine antagonists such as caffeine and other methylxanthines have been suggested for sudden unexpected death in epilepsy prophylaxis (36). However, methylxanthines may prolong seizure duration and increase seizure frequency and further clarifications are necessary (21).

Early oxygen administration is associated with shorter hypoxemia duration and lower hypoxemia severity after generalized convulsive seizures (99). Moreover, it is associated with lower incidence and duration of postictal generalized EEG suppression (02). Also, because most sudden unexpected death in epilepsy patients are found prone, repositioning patients early in order to ensure airway protection may conceivably reduce sudden unexpected death in epilepsy risk (73). Cardiac pacemaker implantation for sudden unexpected death in epilepsy prevention is controversial and has no clear role. For ictal asystole and bradycardia, pacemakers may reduce incidence of syncopal loss of consciousness and tone, protecting against fall related injuries (09). However, their effectiveness in sudden unexpected death in epilepsy prevention is questionable because primary respiratory failure may be the key event in sudden unexpected death in epilepsy (103). On the other hand, measures aimed at restoring breathing after a seizure, including diaphragmatic pacing, have been able to prevent seizure-related deaths in mice (98). Further studies are needed to assess the feasibility and efficacy of such intervention to prevent sudden unexplained death in humans.