Periictal cerebral tissue hypoxemia: A potential marker of SUDEP risk


Address correspondence to Brian D. Moseley, 710 Westwood Plaza, Los Angeles, CA 90095, U.S.A. E-mail:


Cerebral oximetry has not been explored in patients experiencing seizures in the epilepsy monitoring unit (EMU). The purpose of our study was to evaluate the feasibility of periictal measurement of cerebral oxygenation using noninvasive cerebral tissue oximetry and to determine whether there was evidence of cerebral hypoxemia during generalized seizures. Cerebral oxygen saturation findings were subsequently correlated with sudden unexpected death in epilepsy (SUDEP) risk factors. We prospectively evaluated six patients admitted to our EMU with histories of generalized tonic–clonic seizures (GTCS) with prolonged scalp electroencephalography (EEG) and two regional cerebral oxygen saturation (rSO2) sensors. Minimum rSO2 values were recorded in the 5 min preceding seizure onset, during the seizure, and in the 5 min following seizure offset. SUDEP risk was assessed using the SUDEP-7 Inventory. Cerebral oximetry was well tolerated, with a mean duration of rSO2 monitoring of 81.1 h. Cerebral oxygen saturation data were available from at least one sensor in 9 (90%) of 10 seizures; only 6 (60%) of 10 seizures had useable periictal digital pulse oximetry data. GTCS were associated with significantly lower minimum ictal (p = 0.003) and postictal (p = 0.004) %rSO2 values than the minimum preictal value. Patients with at least one seizure with a %rSO2 decrease of ≥20% tended to have higher SUDEP-7 Inventory scores (mean SUDEP-7 Inventory score 7 ± 2.8) versus patients without recorded desaturations (4.3 ± 0.5, p = 0.08). Larger studies are needed to determine the value of cerebral oximetry in the identification of patients at risk of SUDEP.


Severe hypoxemia as measured by digital pulse oximetry is known to occur during and after seizures, particularly generalized tonic–clonic seizures (GTCS) (Bateman et al., 2008; Moseley et al., 2011). It has been hypothesized that periictal hypoxemia and depression of cerebral activity are important mechanisms contributing to sudden unexpected death in epilepsy (SUDEP) (So, 2008). However, digital pulse oximetry can only be considered an approximation of cerebral oxygenation. Currently, no feasible direct measure of cerebral oxygenation is routinely available in epilepsy monitoring units (EMUs). The technology of cerebral tissue oximetry has been validated in patients undergoing carotid endarterectomy (Carlin et al., 1998), but its use during seizures has not been explored. The purpose of our study was to evaluate the feasibility of periictal measurement of cerebral tissue oxygenation by using a novel noninvasive method of cerebral tissue oximetry. We also evaluated whether greater periictal cerebral hypoxemia was associated with SUDEP risk factors.


Patients admitted to our adult EMU with past histories suggesting GTCS were prospectively recruited between December 2011 and May 2012. All subjects were evaluated with continuous 30-channel scalp electroencephalography (EEG) using the International 10–20 system for electrode placement. Digital pulse oximetry, which displays oximetric data every second on the EEG, was also utilized. Two near infrared spectroscopy (NIRS) regional cerebral oxygen saturation (rSO2) sensors were placed on each side of the forehead. Data from the rSO2 sensors were recorded every 4 s by a Nonin EQUANOX Model 7600 Regional Oximeter (Nonin, Plymouth, MN, U.S.A.). Minimum rSO2 values in the 5 min preceding EEG seizure onset, during the seizure, and in the 5 min following EEG seizure offset were recorded. When both sensors yielded useable data, the minimum value recorded by either sensor was used. Data from pulse oximeters and cerebral oximeters were considered useable only when values were recorded during the entire periictal period. Each patient’s epilepsy was classified using the International League Against Epilepsy (ILAE) Commission on Classification and Terminology 2005–2009 report (Berg et al., 2010). All available clinical data were abstracted from the electronic medical record to assess SUDEP risk using the SUDEP-7 Inventory (DeGiorgio et al., 2010).

Data entry and statistical analysis were performed using IBM SPSS Statistics Version 19 (IBM, Armonk, NY, U.S.A.). We utilized paired-sample t-tests and Mann-Whitney U tests depending on the data analyzed. p-Values < 0.05 were considered statistically significant.

This study was approved by the Mayo Clinic, Rochester Institutional Review Board.


Patient demographics and seizure characteristics

Six patients were recruited. Demographic data are listed in Table 1. Ten seizures were recorded; these included three focal onset seizures with impairment of awareness but no secondary generalization, four focal onset seizures with secondary generalization, and three primary GTCS.

Table 1.   Patient demographics (n = 6)
  1. AEDs, antiepileptic drugs.

Gender1/6 (16.7%) male, 5/6 (83.3%) female
Age at time of admission (years)34.8 ± 12.2 (21–47)
Age at seizure onset (years)20.5 ± 12.8 (8–42)
Epilepsy mode of onset3/6 (50%) focal, 3/6 (50%) generalized
Epilepsy etiology2/6 (33.3%) structural/metabolic, 4/6 (66.7%) unknown
Epilepsy syndrome6/6 (100%) no syndrome
Developmental delay/intellectual disability0/6 (0%)
Neurologic examination abnormalities0/6 (0%)
MRI head abnormalities1/6 (16.7%) mesial temporal sclerosis, 1/6 (16.7%) previous tumor resection/surrounding encephalomalacia, 4/6 (66.7%) normal
Number of previously tried AEDs4.3 ± 2.2 (2–7)
Number of current AEDs on admission2.3 ± 0.5 (2–3)
Baseline seizure frequency at admission1/6 (16.7%) ≥ daily, 2/6 (33.3%) daily to ≥ weekly, 1/6 (16.7%) weekly to ≥ monthly, 2/6 (33.3%) monthly to ≥ yearly

Feasibility of cerebral oximetry

The mean duration of rSO2 monitoring was 81.1 h (range 18.7–186.7 h). Nine (90%) of the 10 recorded seizures yielded useable periictal rSO2 data from at least one sensor; six yielded useable periictal data from both sensors. There were no significant differences between the minimum periictal %rSO2 values recorded (p = 0.55) or the absolute change from minimum preictal %rSO2 to minimum ictal/postictal %rSO2 (p = 0.19) when comparing the two pads in these six seizures. Only 6 (60%) of 10 had useable periictal digital pulse oximetry data (see Fig. S1).

Cerebral oximetric changes during seizures

Of the nine seizures with useable cerebral oximetry data, 8 (88.9%) were characterized by minimum ictal %rSO2 values that were lower than the minimum preictal %rSO2 values. The one seizure not marked by a decrease in the minimum ictal %rSO2 value was characterized by a Sp02 desaturation <90%. Lower ictal minimum %rSO2 values persisted into the postictal period in 7 (77.8%) of 9 seizures (see Table 2). There was no significant difference between the minimum ictal and postictal %rSO2 values versus minimum preictal measurements in the three nongeneralized seizures. However, primary/secondarily GTCS were marked by significantly lower minimum ictal (54.2 ± 12.8% vs. 70.83 ± 8.3%, paired-sample t-test p = 0.003) and postictal %rSO2 values (50.8 ± 14.5% vs. 70.8 ± 8.3%, paired-sample t-test p = 0.004) versus minimum preictal measurements. All six GTCS were characterized by periictal rSO2 decreases of at least ≥10% compared to baseline, and two were characterized by decreases of ≥20%. The three seizures not marked by %rSO2 decreases of ≥10% were characterized by Sp02 desaturations <90%.

Table 2.   Cerebral oximeter (rSO2) and digital pulse oximetry measurements
Patient IDSz IDSz typeOnsetMinimum preictal rSO2Minimum ictal rSO2Minimum postictal rSO2Periictal desaturation ≥10%Periictal desaturation ≥20%Periictal digital oximetry desaturation <90%PGESSUDEP-7 Inventory score
  1. 2G, secondarily generalized tonic–clonic; FO, focal onset without secondary generalization; Gen, generalized; L, left; N, no; N/A, not applicable/unusable data; PG, primary generalized tonic clonic; PGES, postictal generalized EEG suppression; R, right; Sz, seizure; Tem, temporal; Y, yes.

C1FOL Tem616661NNYN4
E1FOL Tem646163NNYN4
22GL Tem614647YNN/AY
3FOR Tem615866NNYN
F12GR Tem816663YNN/AN5
22GR Tem745962YNNN

Cerebral oximetric changes and SUDEP risk

The mean SUDEP-7 Inventory score for all six patients was 5.2 ± 1.9 (range 4–9). There was a trend for higher SUDEP-7 Inventory scores in patients with at least one recorded seizure with %rSO2 decreases of ≥20% (7 ± 2.8 vs. 4.3 ± 0.5, Mann-Whitney U test p = 0.08). No statistical difference in SUDEP 7 scores were found in association with the presence of at least one seizure characterized by postictal generalized EEG suppression (5.5 ± 2.4 vs. 4.5 ± 0.7, Mann-Whitney U test p = 0.8) or periictal digital pulse oximetry desaturation of <90% (5.7 ± 2.9 vs. 4.5 ± 0.7, Mann-Whitney U test p = 1).


Our results demonstrate that noninvasive cerebral tissue oximetry is feasible in patients undergoing prolonged video-EEG monitoring. Cerebral oximetry was well tolerated, with a mean monitoring duration of >3 days. Cerebral oximetry also provided more consistently useable data than digital pulse oximetry, which is currently the standard for detecting hypoxemia in EMUs. This was despite the occurrence of generalized tonic–clonic convulsive movements, which often rendered our digital pulse oximetry readings unreliable. The occurrence of Sp02 desaturations during three seizures that were not marked by %rSO2 decreases of ≥10% was unexpected. However, it may be explained by the different normative values for cerebral and digital pulse oximetry. Unlike digital pulse oximetry, where values ≥90% are considered normal, the mean %rSO2 value in normal subjects on room air is 76.6 ± 4.9% (MacLeod, 2012). Therefore, we may have underestimated significant rSO2 changes when we chose the ≥10% cutoff. In addition, it is possible the Sp02 desaturations recorded during those three seizures were secondary to phenomenon (e.g., peripheral vasoconstriction) that was not occurring in the central nervous system (CNS). In such a circumstance, cerebral oximetry would yield more reliable results regarding cerebral hypoxemia.

Our finding that GTCS are associated with significant changes in %rSO2 deserves further study. If replicated in larger studies, such changes might provide a glimpse into cerebral tissue oxygenation during seizures and allow evaluation of this parameter while performing periictal SUDEP physiologic research. For years, epileptologists have recognized that periictal systemic hypoxemia is common, occurring in 25–33% of monitored seizures (Bateman et al., 2008; Moseley et al., 2010, 2011). Given that cases of witnessed SUDEP have been characterized by respiratory difficulty and/or hypoventilation (Langan et al., 2000; Bateman et al., 2010), the role of such respiratory changes in sudden death deserves further study. The trend toward higher SUDEP-7 Inventory scores in our patients with ≥20% reduction in %rSO2 during the periictal period suggests that such changes may be detrimental, particularly when profound and/or prolonged. It is possible that when cerebral hypoxemia exceeds a certain threshold or duration, it may contribute to periictal suppression of brainstem respiratory centers. If hypoxemia and brainstem suppression are parts of a vicious cycle of events (including reduced activity of pulmonary stretch receptors, increased carotid chemoreceptor sensitivity, and asystole), sudden death could theoretically result (So, 2008). If true, detection of profound cerebral hypoxemia could one day be utilized to stratify SUDEP risk and identify patients at need for interventions to reduce such risk.

Because the primary aim of our study was to demonstrate the feasibility of rSO2 monitoring in EMU patients, the sample size of our study was small. Future studies with larger sample sizes are needed to determine if the cerebral oximetric changes demonstrated in our patients can be replicated. It is possible that the addition of a second digital pulse oximeter would have increased the reliability of this technology to a similar level as that with two cerebral oximeter pads. Given that we did not save interictal rSO2 values for later analysis, we cannot comment on false positive desaturations. This limits our ability to comment on the usage of cerebral oximeters to detect seizures, particularly GTCS. Despite these limitations, the findings of our study prove rSO2 monitoring is feasible in epilepsy patients with focal onset and GTCS. Larger studies are now needed to determine the value of cerebral oximetry in the identification of patients at risk of SUDEP.


This study was sponsored by discretionary funding made available by the Mayo Clinic Department of Neurology Research Committee. The regional cerebral oximeter was provided at no cost by Nonin Medical, Inc. The authors would like to thank our clinical neurophysiology technologists Susan Senjem, Eric Marshall, Jean Varner, Randy Berge, Kim Le, and Judith Johnson for their valuable assistance with the conduction of this study.


Dr. So serves on the editorial boards for Epilepsia, Epilepsy Research, and the Journal of Clinical Neurophysiology and is an unpaid volunteer for UCB’s patient advocacy project. Drs. Moseley, Britton, and Lee and Cindy Nelson have no conflicts of interest to disclose. 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.