Unraveling SUDEP: Mechanisms of Seizure-Induced Cardiac and Respiratory Impairment

Introduction

People with epilepsy (PWE) have a 2- to 3-fold higher mortality rate than the general population, and sudden unexpected death in epilepsy (SUDEP) accounts for roughly half of that mortality. Indeed, SUDEP is the cause of 100 to 120 deaths per 100,000 person-years for PWE, compared to estimates of sudden unexplained death in the general population being 1 to 3 deaths per 100,000 person-years. ¹ Among the neurological disorders, SUDEP comes second only to stroke for the number of life-years lost, ² underscoring the significant impact it has on PWE and their families. Because SUDEP is rarely witnessed, it is one of the most challenging and enigmatic outcomes in epilepsy to be studied.

A priori logic suggests that the mechanism of SUDEP may include cardiac and/or respiratory dysfunction in addition to seizures, as failure of these systems results in sudden death. Clinical studies do not yet paint a clear picture on this front; some evidence points to primary cardiac dysfunction as a precipitating event, ³ and other studies suggest that seizure-induced respiratory failure frequently precedes cardiac arrest. ⁴ Furthermore, while seizure frequency is a major predictor of SUDEP risk, so are living alone and psychiatric comorbidities.^1,5 It is increasingly clear that SUDEP is multifactorial on 2 levels: first, as a heterogeneous diagnosis encompassing distinct pathophysiological pathways across cases, and second, as a sequence of converging events—such as nocturnal seizures, seizure-induced respiratory and/or cardiac dysfunction, prolonged postictal unresponsiveness, and disrupted autonomic regulation—that together culminate in death.

The goal of the 2025 AES SUDEP Special Interest Group was to examine how seizures and their underlying epilepsy etiology, as well as epilepsy-linked gene variants that are expressed in the brain and heart, may cause cardiac and respiratory impairment, which could lead to SUDEP. This includes peri-ictal and interictal physiological mechanisms of altered respiratory, cardiac, and autonomic function due to underlying epilepsy and seizures. Evidence from both humans and animal models that sheds light on these mechanisms and highlight pathways that may underlie SUDEP is discussed.

Seizure-Induced Apnea as a Mechanism of SUDEP

Most witnessed SUDEP cases are preceded by a generalized convulsive seizure, and evidence from both epilepsy monitoring units and animal models indicates that death often results from postictal respiratory arrest. ⁶ In the MORTEMUS study, 9 patients who died in epilepsy monitoring units with sufficient cardiorespiratory monitoring were analyzed. ⁴ In all cases, a convulsive seizure produced a terminal apnea that reliably occurred minutes before terminal asystole. While this data set is limited, it suggests that in many cases, SUDEP may be due to apnea secondary to seizures. This may not be too surprising, as apnea has been reported in a large percentage of PWE during and after convulsive and nonconvulsive seizures.^7–9 Thus, a major focus of SUDEP research is to identify the mechanisms of seizure-induced apnea.

A Causal Role for the Amygdala in Seizure-Induced Apnea

The Dlouhy laboratory has sought to define the causal neural substrate of seizure-induced central apnea through invasive intracranial recordings and direct electrical stimulation in patients with medically refractory epilepsy. Initial studies used intracranial electroencephalogram (EEG) monitoring, combined with direct electrophysiologic and continuous respiratory monitoring to localize seizure foci. In these patients, seizures that propagated to the amygdala produced rapid and profound central apnea accompanied by oxygen desaturation, while seizures restricted to other cortical and subcortical regions did not alter respiration. ¹⁰ Direct electrical stimulation of the amygdala reproduced these effects, causing apnea and desaturation, whereas stimulation of adjacent regions such as the hippocampus, insula, or cingulate cortex had no effect on breathing. Remarkably, during stimulation-induced apnea, patients were unaware they had stopped breathing, suggesting that the amygdala not only halts breathing but also suppresses the conscious perception of air hunger.

Subsequent studies extended these observations across many adult and pediatric patients, including children as young as 3 years old. The consistent finding across patients was that stimulation-induced apnea is specific to the amygdala. Using high-resolution electrode localization and machine learning approaches, a discrete subregion within the amygdala where stimulation most reliably induced apnea, which was named the amygdala inhibition of respiration (AIR) site. This site is localized to the medial subregion of the basal nuclei, cortical and medial nuclei, and amygdala transition areas. ¹¹ This work revealed that not all amygdala nuclei affect breathing equally; rather, a focal subregion exerts potent inhibitory control over the brainstem respiratory network.

Mapping the Circuitry of Long-Lasting Postictal Apnea

Work in the Dlouhy lab further investigated postictal mechanisms underlying SUDEP, by examining how seizures originating from or propagating to the amygdala alter breathing after seizure termination. In some patients, amygdala seizures produced both ictal and persistent postictal apnea, lasting up to minutes beyond seizure cessation. ¹² Electrical stimulation mapping, combined with across-subject analyses and machine learning approaches, localized this persistent effect to a discrete subregion within the AIR site. This region, designated the persistent AIR (pAIR) site, appears to drive sustained respiratory suppression that continues beyond seizure termination in some patients. To uncover the broader neural circuitry engaged by the pAIR site, electrical stimulation concurrent with functional magnetic resonance imaging (fMRI) was applied, allowing identification of causal, rather than correlative, connections to the AIR site. Stimulation of the amygdala during fMRI produced decreased BOLD activity in the pons and medulla—key brainstem respiratory centers—while increasing activity in the insula, a region implicated in interoception and breathing perception.

Collectively, these studies identify a focal, circumscribed human amygdala region (pAIR site) that exerts control over respiratory networks, producing ictal and postictal apnea without awareness of respiratory distress. This discovery provides a mechanistic framework linking seizure spread to the amygdala with respiratory arrest and impaired alarm and arousal in SUDEP. The findings also offer a translational foundation for targeted therapeutic strategies. Ongoing work continues to map the connectivity and physiological function of this site and explores neurosurgical and neuromodulatory approaches to mitigate its lethal downstream effects. These convergent human studies represent a significant step toward resolving the mystery of SUDEP and open a path toward evidence-based, mechanistic prevention.

SUDEP Genetics and Long QT Syndrome

While SUDEP occurs in all epilepsy etiologies,^13–16 SUDEP risk in genetic epilepsies varies in a gene-specific manner. PWE with variants in ion channel genes expressed in both the brain and the heart have the highest SUDEP risk.^17–19 For instance, Bagnall et al ¹⁷ performed whole exome sequencing of 61 SUDEP cases. Interestingly, 15% (N = 9) of cases had variants in genes associated with cardiac arrhythmias. Specifically, 7% (N = 4) of the SUDEP cases had variants in genes linked to the classically studied inherited cardiac disease, long QT syndrome (LQTS). Interestingly, when SUDEP is compared to sudden cardiac death secondary to LQTS, especially to LQT3, which is linked to variants in the voltage-gated sodium channel (VGSC) gene, SCN5A, there are parallels in the circumstances of death. ¹⁸ Sudden cardiac death occurs ∼3 times more frequently in PWE compared to the general population^20,21 and a large number of PWE, including individuals from SUDEP cohorts, have pathogenic variants in cardiac genes.^20,22–25

Similarly, seizures and epilepsy are reported in Brugada syndrome (BrS), ²⁶ catecholaminergic polymorphic ventricular tachycardia (RyR2 variants), ²⁷ HCN2 loss-of-function (LOF), ²⁸ and LQTS.^17,29–36 There is a 3-fold higher prevalence of seizures/epilepsy in patients whose genotype is positive for LQTS, compared to their genotype-negative family members. ³⁰ Patients with LQT2 have the highest prevalence of seizures/epilepsy (18%), and LQT2 is an independent risk factor for seizures/epilepsy. ³⁰ In a cohort of LQT2 patients that underwent EEG evaluation, 15% exhibited epileptiform discharges. ²⁹ Additionally, there is an increased prevalence of abnormal EEG findings and epileptogenic activity in people with LQTS. ³² People with LQT2 exhibit the highest prevalence of abnormal EEG findings (40%) and epileptogenic activity (10%). ³²

SUDEP in a Rabbit Model of LQT2

The Auerbach laboratory developed the first genetic rabbit model of LQT2 syndrome. ³⁷ Using CRISPR-Cas9 technology, a knock-in variant in the endogenous Kcnh2 gene was introduced, which encodes the K_v11.1 protein, a delayed rectifier potassium channel and current (I_Kr). Kcnh2 was initially cloned from a human brain cDNA library, ³⁸ and is highly expressed in the brain and heart. ³⁷ I_Kr is important in cardiac repolarization. In the brain, it helps stabilize the resting membrane potential and suppresses repetitive neuronal action potential (AP) firing.^4,39–42 The LQT2 rabbits exhibit reduced Kcnh2/K_v11.1 expression and cardiac QT_c prolongation. ²² Similar to people with LQT2, there is a significant increase in the prevalence of spontaneous epileptiform activity, seizures, and sudden death. ²² Epileptic seizures and cardiac arrhythmias are detected prior to sudden death.

Simultaneous EEG, electrocardiogram (ECG), and capnography recordings were used, as previously described, ⁴³ to capture neuro-cardiac-respiratory abnormalities and facilitated the detailed assessments of the temporal evolution and concordance between these multisystem pathologies surrounding seizures and leading up to sudden death (unpublished data). Cerebral abnormalities included spontaneous epileptiform discharges, generalized tonic-clonic seizures, and postictal generalized suppression. During the postictal period, there were episodes of bradycardia, atrial bigeminy, second- and third-degree atrioventricular block, and ventricular fibrillation. During the postictal period, there were periods of apnea.

In summary, these mutant rabbits provide a clinically relevant animal model of LQT2, which facilitates future studies investigating the mechanisms underlying seizure-mediated cardiorespiratory changes and sudden death (eg, SUDEP). Similar to the MORTEMUS study, ⁴ during the postictal period LQT2 rabbits develop cardiorespiratory abnormalities that ultimately result in sudden death. Detailed assessments of simultaneous multisystem recordings yield a deep understanding of the temporal evolution of pathophysiological changes leading up to SUDEP, as well as the mechanistic substrates and triggers of these events.

Channelopathies With a High Incidence of SUDEP

To gain insight into SUDEP mechanisms, the Isom laboratory focuses on genetic channelopathies with a high incidence of SUDEP. One such disorder is Dravet syndrome (DS), a devastating form of developmental and epileptic encephalopathy (DEE) characterized by multiple pharmaco-resistant seizure types and a high risk of SUDEP.^44–46 In addition to severe seizures, DS patients have profound intellectual disability, developmental delays, movement and balance issues, language and speech disturbances, growth defects, sleep abnormalities, disruptions of the autonomic nervous system, and mood disorders. ⁴⁷ While all PWE are at risk for SUDEP, DS patients have the highest risk: up to 20%. ¹³ The mean age of SUDEP in DS patients is 4.6 years. ¹⁶ In ∼90% of cases, DS is caused by de novo monoallelic LOF variants in SCN1A, encoding the VGSC Nav1.1 α subunit, resulting in haploinsufficiency.^48,49 A smaller cohort of DS patients have inherited biallelic LOF variants in SCN1B, encoding the VGSC nonpore-forming β1 subunits.^50,51 These patients present clinically with DS or the more severe early infantile DEE. A related DEE, DEE13, is linked to de novo monoallelic gain-of-function (GOF) variants in SCN8A, encoding the VGSC α subunit Nav1.6. ⁵² VGSCs are responsible for the generation of the rising phase and propagation of APs in neurons ⁵³ and cardiac myocytes (CMs).^53,54 SCN1A, SCN8A, and SCN1B are expressed in both the heart and brain of humans and mice.^55–57 Thus, the Isom lab has proposed that cardiac arrhythmias contribute to the mechanism of SUDEP in DEE.

Cardiac Arrhythmias Due to Ion Channel Variants

A body of work from the Isom laboratory demonstrated evidence for cardiac arrhythmias in 4 different mouse models of DEE: (1) Scn1a^R1507X/+ DS, (2) Scn8a^N1768D/+ DEE13, (3) Scn1b null DEE52, and (4) Scn1b^C89/C89 DEE52: acutely isolated Scn1a DS mouse ventricular CMs have increased transient (I_Na) and late (I_Na,L) sodium current, AP prolongation, and increased incidence of early afterdepolarizations. ⁵⁸ ECGs in freely moving mice showed prolonged QT intervals. Scn8a DEE13 mouse ventricular CMs have AP prolongation, prolonged calcium (Ca²⁺) transients, and aberrant Ca²⁺ release. AP prolongation and aberrant Ca²⁺ release in Scn8a DEE13 CMs are tetrodotoxin-sensitive (TTX-S) and sensitive to the Na⁺–Ca²⁺-exchanger inhibitor SN-6, implicating increased I_Na,L via Nav1.6 channel GOF and subsequent activation of reverse Na⁺/Ca²⁺ exchange as the mechanism of arrhythmogenesis. ⁵⁵ ECGs in these mice are normal at rest, but arrhythmias can be induced with epinephrine and caffeine to mimic an ictal sympathetic surge. Consistent with this work, patients with SCN8A-linked DEE13 variants were reported to experience cardiac arrhythmias and ictal asystole. ⁵⁹

SCN1B variants are linked to human cardiac disease in addition to DEE, including BrS5 and atrial fibrillation familial 13. However, other work suggests that SCN1B may not be a monogenic cause of BrS38. Scn1b null mouse ventricular CMs have increased I_Na and I_Na,L, AP prolongation, prolonged Ca²⁺ transients, and increased incidence of delayed afterdepolarizations.^60,61 Scn5a/Nav1.5 and Scn3a expression, as well as ³H-saxitoxin binding, which measures levels of TTX-S VGSC expression, are increased in Scn1b null hearts. Like Scn8a DEE13 mice, AP prolongation and aberrant Ca²⁺ release in Scn1b null mice are TTX-S. Scn1b null mouse ECGs show prolonged QT intervals. Scn1b deletion also results in altered expression of genes associated with atrial dysfunction. ⁶² Scn1b null hearts have a significant accumulation of atrial collagen, increased susceptibility to pacing-induced atrial fibrillation (AF), sinoatrial node dysfunction, and increased numbers of cholinergic neurons in ganglia that innervate the sinoatrial node. Atropine reduced the incidence of AF in null animals, suggesting that autonomic changes contributed to AF. AP duration was prolonged in null atrial myocytes, with increased I_Na,L density, and reduced L-type Ca²⁺ current density. Finally, mice with biallelic expression of the DEE52 variant Scn1b-c.265C>T, predicting p.R89C (Scn1b^C89/C89), have spontaneous and hyperthermia-induced generalized seizures and SUDEP. ⁶³ Scn1b^C89/C89 mouse CMs have increased transient outward potassium current (I_to) density and heart sections show ventricular fibrosis. ⁶⁴ Scn1b^C89/C89 mice are susceptible to pacing-induced cardiac arrhythmias. ⁶⁴ Taken together, this body of work demonstrates the critical role played by Scn1b in atrial and ventricular physiology during early postnatal mouse development, suggesting that SCN1B LOF variants that result in DEE52 may also significantly impact the developing pediatric heart.

Stress, Seizures, the Hypothalamic–Pituitary–Adrenal Axis, and Psychiatric Comorbidities in Epilepsy

There appears to be a bidirectional relationship between stress and epilepsy, where stress is a common seizure trigger, and seizures also increase the levels of stress hormones. ⁶⁵ Clinical studies link stressful life events with increased seizure frequency.^66,67 The physiological stress response is mediated by the hypothalamic–pituitary–adrenal (HPA) axis, involving release of corticotropin-releasing hormone (CRH) from the hypothalamus, which induces release of adrenocorticotropic hormone from the pituitary, leading to release of cortisol from the adrenal gland in humans (corticosterone in rodents). PWE exhibit elevated basal cortisol levels compared to healthy controls,^68,69 which are further increased following a seizure.^69,70 Data from the Maguire laboratory demonstrates a positive correlation between seizure frequency and cortisol levels in both PWE ⁶⁸ and in rodent models of temporal lobe epilepsy,^71,72 and that stress and neuroendocrine mediators of stress exert proconvulsant actions across animal models of seizure susceptibility and epilepsy.^73,74

Psychiatric comorbidities are highly prevalent in PWE, affecting approximately 75%, with depression (55%) and anxiety (25%-50%) being most common.^75–77 Similar behavioral deficits are also demonstrated in chronically epileptic mice.^78–81 A cardinal feature of depression, the psychiatric disorder most commonly diagnosed in PWE, is HPA axis hyperactivity. ⁸² Given that stress and HPA axis hyperactivity are known to worsen epilepsy outcomes, ⁶⁵ HPA axis dysfunction may represent a pathophysiological mechanism linking the comorbidity between psychiatric illnesses and epilepsy. In fact, chronically epileptic mice with HPA axis dysfunction exhibit an increased vulnerability to behavioral deficits ⁷³ ; whereas, mice with HPA axis hypofunction exhibit decreased behavioral comorbidities associated with epilepsy. ⁸⁰ Thus, stress and HPA axis activity may play a role in the comorbidity between psychiatric illnesses and epilepsy.^83,84

Stress, Seizures, HPA Axis Dysfunction, and Psychiatric Comorbidities in SUDEP Risk

The HPA axis also mediates critical homeostatic functions, including regulation of the metabolic, immune, and cardiovascular systems. Interestingly, chronically epileptic mice with HPA axis dysfunction exhibit increased mortality due to SUDEP. ⁷³ The increased SUDEP in this model was directly mediated by HPA axis dysfunction, since chemogenetic or pharmacological suppression of HPA signaling prevented the vulnerability to SUDEP. ⁷³ The translational relevance of these findings was validated by demonstrating alterations in neuroendocrine mediators in blood samples from PWE who died of suspected SUDEP compared to non-PWE or PWE (without suspected SUDEP). ⁷³ Thus, these data demonstrate that HPA axis dysfunction may be a novel mechanism contributing to SUDEP and is the first potential link to an environmental insult implicated in SUDEP.

An important, yet often overlooked and underexplored fact is that psychiatric comorbidities in epilepsy are associated with increased mortality. ⁸⁵ In fact, the incidence of SUDEP was found to be increased 5-fold in female patients with psychiatric comorbidities. ⁵ There is a well-established mortality gap in individuals with psychiatric illnesses compared to the general population, with life expectancy estimated to be 10 to 20 years shorter in patients with psychiatric conditions. ⁸⁶ Numerous mechanisms are proposed as contributing factors to increased mortality in individuals with psychiatric illnesses,^86–88 with cardiac dysfunction suggested to be a leading cause of death.^89,90 Potentially informative is evidence that HPA axis dysfunction is associated with the increased risk of cardiovascular disease, coronary heart disease, and cardiac death associated with depression. ⁹¹ Related to the mechanistic link between psychiatric comorbidities, epilepsy, and SUDEP, both serotonin ⁹² and HPA axis dysfunction ⁷³ are implicated, which necessitates further exploration.

Potential Mechanisms Mediating HPA Axis Dysfunction and SUDEP Risk

There is ample evidence that SUDEP is related to a sudden cardiac and/or respiratory arrest ⁹² mediated by central brainstem circuits regulating cardiorespiratory homeostasis. ⁹³ While the majority of studies investigating the role of brainstem circuits in SUDEP risk focus on spreading depolarization,^94,95 potential seizure-related trigger(s) to this global electrophysiological event is unknown. The HPA axis directly influences cardiorespiratory centers, which are likely responsible for the impact of stress mediators, including glucocorticoids, on overall respiration and cardiovascular function (for a review, see Kc and Dick ⁹⁶ ). The resulting cardiorespiratory dysfunction is then linked to increased mortality rates associated with epilepsy (for a review, see Tao et al ⁸⁵ , Yuen et al ⁹⁷ ) specifically SUDEP. ⁹⁸ In addition to their projections to the pituitary, hypothalamic CRH neurons send direct projections to cardiovascular and respiratory brainstem control centers,^99–101 including the nucleus of the solitary tract (NTS). NTS receives a multitude of excitatory glutamatergic interoceptive afferents that influence cardiorespiratory function, including those that trigger cardiac and respiratory depression. Activation of hypothalamic CRH neuronal projections enhances interoceptive afferent reflex responses through NTS. ⁹⁹ The Maguire and Boychuk laboratories made the exciting observation that chronically epileptic mice with HPA axis dysfunction and elevated SUDEP risk exhibit overactive reflexes that result in cardiac and respiratory arrest. ¹⁰² This establishes for the first time a potential trigger for SUDEP and identifies HPA axis dysfunction as a novel pathophysiological mechanism contributing to SUDEP risk.

Conclusions

SUDEP remains a devastating and incompletely understood cause of premature mortality in PWE. Human intracranial recording studies demonstrate that seizure propagation to a discrete amygdala subregion can induce profound ictal and postictal apnea, suppress awareness of respiratory failure, and causally inhibit brainstem breathing centers. Parallel work in genetic epilepsies underscores the importance of cardiac substrates. Animal models of ion channelopathies reveal intrinsic electrical dysfunction in both neurons and cardiomyocytes, producing arrhythmogenic conditions that can be unmasked or exacerbated by seizures. These models mirror clinical observations that gene variants traditionally associated with cardiac arrhythmias, such as LQTS genes, are enriched in SUDEP cases and that individuals with cardiac ion channel gene variants exhibit elevated seizure susceptibility. In addition, stress physiology and psychiatric comorbidities emerged as important, underrecognized contributors. HPA axis hyperactivity increases seizure susceptibility, worsens epilepsy outcomes, and heightens mortality in experimental epilepsy models.

Collectively, these multidisciplinary findings support the hypothesis that SUDEP results from the intersection of respiratory suppression, cardiac vulnerability, autonomic dysregulation, and stress-responsive physiology. Continued integration of human, genetic, and mechanistic studies will be crucial for developing targeted prevention strategies.