Sudden unexplained death in epilepsy (SUDEP) is the most common cause of death in patients with epilepsy, with an annual incidence of 2.2–10 per 1,000 patients (Téllez-Zenteno et al., 2005). Cardiac arrhythmias, central apnea, neurogenic pulmonary edema, and asphyxia are postulated as mechanisms for SUDEP. These assumptions are supported by clinical evidence of ictal and postictal apneas and cardiac arrhythmias (Dasheiff & Dickinson, 1986; Nashef et al., 1996; So et al., 2000). Four cases of SUDEP and two cases of near-SUDEP with electroencephalography (EEG) monitoring have been discussed in a recent review (Tomson et al., 2008). A fifth case of SUDEP with EEG monitoring was published as an abstract (Purves et al., 1992). We report two cases of SUDEP occurring in patients undergoing video-EEG telemetry (VET) at two epilepsy centers. Permission for reporting these cases was obtained from appropriate authorities at each institution. No other cases of SUDEP have occurred at either epilepsy monitoring unit (EMU) in the last two decades.
Sudden unexplained death in epilepsy (SUDEP) is a common cause of death in patients with epilepsy, with cardiorespiratory dysfunction and a primary cessation of cerebral function proposed as causes. We report two cases of SUDEP in patients with intractable temporal lobe epilepsy undergoing video-EEG (electroencephalography) telemetry at two centers. Both had secondarily generalized convulsions. EEG, electrocardiography (ECG), and respiratory changes in these two patients are reported herein. Ictal/postictal hypoventilation may contribute to SUDEP with the resulting hypoxemia and acidosis leading to failure of recovery of cortical function and eventual cardiac failure.
A 42-year-old woman had a 7-year history of seizures, with symptoms of déjà vu, depersonalization, and visual disturbance followed by staring, unresponsiveness, postictal confusion, and dysphasia and occasional generalized tonic–clonic seizures. She failed to respond to five anticonvulsants. On admission she was taking phenytoin and lamotrigine.
Anticonvulsants were tapered and then discontinued on the third day of admission to provoke seizures. On the second day of admission, she had two simple partial or nonepileptic seizures without EEG changes. On day 4 she had one secondarily generalized tonic–clonic seizure of left temporal onset, and later that day, while unattended, she had a seizure that preceded her death.
She was asleep on her right side and then opened her eyes and had repetitive blinking. She lifted her head, turning partly to the right. She rolled over into a prone position with her head midline and legs extended. Clonic movements occurred 56 s after eye opening and ceased 32 s later. Postictally, she was prone with her arms under her chest and head partially on the pillow. She was in stage 2 sleep when the seizure began, with rhythmic 2-Hz sharp waves in the left anterior to midtemporal region, increasing in frequency, spreading to the left central and right temporal regions after 25 s, and generalizing after 43 s. The seizure terminated after 102 s. Postictally there was diffuse muscle artifact for 3 min. The EEG remained diffusely attenuated. Respiratory movements or movement artifacts were seen on video or EEG for 12 min after the seizure ended. These became increasingly infrequent and irregular before ceasing. Preictally there was a sinus rhythm of 90 beats per min (bpm). There was ictal tachycardia up to 160 bpm with return to baseline following seizure termination. Occasional premature atrial and ventricular contractions occurred in the early postictal period. Over the next 18 min, the sinus rhythm became slower and more irregular, with ST elevations and peaked T waves and without any escape beats. Several 3–12 s asystoles were seen and electrocardiography (ECG) activity terminated 18 min after the seizure ended (Fig. 1). She was subsequently found to be pulseless and unresponsive. Resuscitation was unsuccessful.
A 62-year-old man had a 56-year history of seizures consisting of a strange feeling and a “need to get away,” nervousness, and dysphasia with or without loss of consciousness followed by postictal dysphasia, disorientation and fatigue, and rare generalized tonic–clonic seizures. He had failed 10 anticonvulsant agents. Interictal EEG showed left temporal epileptiform discharges.
Anticonvulsant medications were tapered to provoke seizures. Over the first 2 days of admission, he experienced six typical simple partial seizures associated with brief rhythmic 3–4 Hz activity over bilateral frontotemporal regions. On the fourth day, he had two typical seizures, one secondarily generalized, with bilateral frontotemporal 3–4 Hz rhythmic waveforms followed by left temporal rhythmic 5–7 Hz waveforms before generalization. There was ictal tachycardia of 140 bpm. The next morning, while unattended, asleep, and prone in bed, he had another secondarily generalized seizure lasting approximately 2 min with electrographic onset in the left temporal region. Postictally the EEG was diffusely attenuated (Fig. 2). Respiratory movements and audible breath sounds were present. The respiratory rate was 15 breaths per min prior to the seizure onset. Following seizure termination the respiratory rate gradually decreased, and when measured at 15 s intervals was 10, 10, 8, 8, 7, 5, 4, 2, and 1 breath(s) per min. There was no ictal tachycardia with this event, but a few premature beats and peaked T waves were present postictally. ECG and respirations ceased approximately 2 min after the seizure. Resuscitation efforts were unsuccessful.
Both patients had intractable partial epilepsy and both experienced a terminal convulsive seizure after which they were prone in bed. The EEG remained diffusely suppressed following the seizures. Respiratory movements and ECG activity were still visible for several minutes postictally, but it could not be determined whether the patients were adequately ventilating. ST segment or T-wave changes were present postictally. In Patient 1, there was progressive sinus bradycardia without escape rhythm. These findings suggest possible contributions from hypoxemia and respiratory and/or metabolic acidosis to cardiac failure.
Five cases of SUDEP and two cases of near-SUDEP with EEG monitoring have been reported in the literature (Purves et al., 1992; Tomson et al., 2008). A postictal primary cessation of cerebral function, or cerebral electrical shutdown (CES), was postulated as the primary mechanism of patient death in three of these cases (Tomson et al., 2008). Despite the lack of respiratory data, hypoxemia was felt not to be contributory because of the absence of the typical EEG changes that accompany acute hypoxia (McLean & Wimalaratna, 2007). The absence of such EEG findings in our patients does not exclude hypoxia as a potential contributor to death. The EEG background in both SUDEP cases was diffusely suppressed postictally. The typically described acute hypoxia pattern of progressive slowing and attenuation of the EEG ultimately leading to an isoelectric background would not be expected to be seen in an already diffusely suppressed EEG. Rather, the EEG may simply remain attenuated without postictal recovery in response to persistent hypoxia.
Ictal-associated hypoxemia, hypercarbia, and respiratory and/or metabolic acidosis persisting in the postictal period may have contributed to death in our cases. There were ST-segment elevations and peaked T waves on the post-ictal ECGs and a lack of an escape rhythm despite profound sinus bradycardia in Patient 1. Similar changes were described in a canine model of anoxia and asphyxia, where in the presence of hypercarbia and acidosis, asystole was the cause of death (Kristoffersen et al., 1967). Acidosis reduces heart rate without affecting intraventricular action potential propagation (Aberra et al., 2001). Tall T waves occur with hyperkalemia and also in metabolic acidosis without hyperkalemia (Dreyfuss et al., 1989). The mechanisms underlying persistent postictal hypoxemia remain to be clarified.
Subsequent to these two cases of SUDEP occurring, we demonstrated that ictal hypoxemia occurs in one-third of partial-onset seizures in patients undergoing VET (Bateman et al., 2008). Given these data, we postulate that ictal hypoventilation may be a predisposing factor for SUDEP. This is supported by a sheep model of epileptic sudden death where hypoventilation, hypercarbia, and acidosis contributed to death (Johnston et al., 1997). Suffocation while in the prone position in the postictal period may have contributed to death in our two cases. Nevertheless, we have shown that the incidence and severity of ictal hypoxemia and hypercapnia with partial and secondary generalized seizures is unrelated to patient position (Bateman et al., 2008).
Central or mixed ictal apneas were reported in one case series with accompanying bradyarrhythmias in 4 of 10 patients, suggesting potentiation of cardiorespiratory reflexes by apnea and hypoxia (Nashef et al., 1996). In a case of near-SUDEP, there was persistent postictal apnea with progressive bradycardia and cardiac arrest following a complex partial seizure (So et al., 2000). Obstructive respiratory failure may also contribute to SUDEP (Tomson et al., 2008).
Changes have been instituted at both our EMUs as a result of these deaths and the insights gained from our study (Bateman et al., 2008). Continuous pulse oximetry with alarm is now in use for all patients at both sites. At one site, VET data, ECG, and oxygen saturation are continuously monitored by EEG technologists. At the other EMU, nursing staff are automatically alerted by pager when the patient seizure event button is pressed, the oxygen saturation drops below 85%, or the heart rate falls outside preset parameters, and patients are continuously monitored by cardiac telemetry technologists trained to recognize seizures. Given the findings in these cases of SUDEP and the incidence and severity of ictal hypoxemia in patients with partial epilepsy, monitoring of respiratory parameters should be considered by other epilepsy centers.
We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
None of the authors has any conflict of interest to disclose.