SEARCH

SEARCH BY CITATION

Keywords:

  • Epilepsy;
  • Hypoxemia;
  • Hypercapnia;
  • Sudden unexpected death in epilepsy;
  • EEG suppression

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References

Purpose:  The relationship of postictal generalized electroencephalography (EEG) suppression (PGES) with sudden unexpected death in epilepsy (SUDEP) is controversial. It has been suggested that PGES is associated with respiratory inhibition leading to SUDEP, but the relationship between PGES and respiratory depression is unknown. Respiratory rate and amplitude of airflow increase following seizures but there is persistent hypercapnia and hypoxemia.

To determine whether seizures with PGES result in respiratory dysfunction, we analyzed respiratory parameters recorded during video-EEG telemetry in patients with localization-related epilepsy.

Methods:  Secondarily generalized convulsive seizures (GC) with PGES on scalp EEG or bilateral postictal attenuation (BA) on intracranial recordings were compared to GC without PGES/BA. Oxygen desaturation nadir and duration, end-tidal CO2 (ETCO2), apnea duration, and duration of the seizure and of the convulsive component were compared in GC with or without PGES/BA.

Key Findings:  There was no significant difference between GC with (n = 30) or without PGES/BA (n = 72) for total seizure duration or duration of the convulsion. GC with PGES/BA had a mean oxygen desaturation nadir of 68.8 ± 11.8% (71.5, 43–88) (mean ± standard deviation [median, range]) that was lower (p = 0.002) than seizures without PGES/BA (76.31 ± 10.17% [79, 42–93]). The duration of desaturation was significantly longer and peak ETCO2 higher in GC with PGES/BA. There was no difference in apnea duration. Apnea did not start during PGES/BA and did not typically extend into the postictal period in GC with or without PGES/BA.

Significance:  PGES is not associated with postictal central apnea but is more likely related to the severity of seizure-associated intrinsic pulmonary dysfunction.

The risk of sudden unexpected death in epilepsy (SUDEP) is more than 20 times that reported in the general population (Shorvon & Tomson, 2011). In patients with refractory epilepsy the risk of SUDEP is 6.3–9.5 per 1,000 patient years (Tomson et al., 2008). Cardiac arrhythmias and/or respiratory depression are considered to be likely mechanisms (Surges et al., 2009). A recent study suggested that postictal generalized electroencephalography (EEG) suppression (PGES) >20 s following generalized convulsions (GC) is a risk factor for SUDEP (Lhatoo et al., 2010). No respiratory information was available in this retrospective study; however, it was hypothesized that PGES represented a profound cortical neuronal inhibitory mechanism of abrupt onset, possibly with associated inhibition of brainstem respiratory centers and resultant postictal central apnea (Lhatoo et al., 2010). We have previously demonstrated that, rather than the occurrence of postictal central apnea, both the respiratory rate and amplitude of airflow are increased following seizures (Seyal et al., 2010). Despite these respiratory changes, there is persistent postictal hypercapnia and hypoxemia that cannot be accounted for by relatively brief ictal apneas (Seyal et al., 2010). In another study of 17 matched SUDEP and control patients, PGES was shown to be significantly associated with secondarily generalized convulsions, but not to be an independent risk factor for SUDEP (Surges et al., 2011), suggesting that PGES alone may not be sufficient to induce seizure-related death (Surges et al., 2011).The relationship between PGES and SUDEP remains uncertain and it is undetermined whether PGES is associated with respiratory dysfunction.

To determine whether GC with PGES are associated with more pronounced respiratory dysfunction relative to GC without PGES, we analyzed video-EEG telemetry (VET) data from our epilepsy monitoring unit (EMU) that incorporates recording of respiratory parameters concurrently with EEG.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References

Details of respiratory parameter monitoring during inpatient VET in our EMU have been published previously (Bateman et al., 2008; Seyal et al., 2010) and included recording of oxygen saturation with digital pulse oximetry, nasal airflow, abdominothoracic excursions, and end-tidal CO2 (ETCO2). Prior approval for the study was obtained from the local institutional review board.

Periods of PGES were determined using previously published criteria, that is, a generalized absence of electroencephalographic activity >10 μV in amplitude, allowing for muscle, movement, breathing, and electrode artifacts (Lhatoo et al., 2010; Surges et al., 2011). Periods of PGES ≥2 s were analyzed. In our analysis, bilateral EEG attenuation (BA) present in recordings from intracranial electrodes was included along with scalp EEG recordings, with the assumption that BA recorded intracranially was equivalent to PGES seen in scalp recordings. A previous study in patients with intractable localization-related epilepsy recorded with bilateral intracranial electrodes demonstrated that apnea during the ictal period is related to bilateral spread of seizures (Seyal & Bateman, 2009). It was, therefore, felt that combining data from scalp and intracranial studies was justifiable. Data from scalp-recorded VET alone were also analyzed separately.

Consecutive GC with interpretable data from digital pulse oximetry and/or nasal airflow and abdominothoracic excursions were analyzed during the periictal period. We compared oxygen desaturation nadir, duration of oxygen desaturation below 90%, peak ETCO2 and ETCO2 change from preictal baseline, total seizure duration, duration of the convulsive component of the seizure, and ictal apnea duration in GC with PGES/BA and without PGES/BA. We noted whether ictal apnea extended into the postictal period and whether any apnea started in the postictal period.

The primary end point for determination of sample size was a 10% difference in oxygen desaturation nadir between GC with PGES/BA >20 s and GC without PGES/BA. We determined that with a two-sided two-sample t-test with assumption of equal variances at a significance level of 0.05 and power of 0.8, a sample size of 24 seizures would be required in both groups. The oxygen saturation data was not normally distributed (Kolmogorov-Smirnov test). Therefore, the Mann-Whitney rank sum test was used to analyze respiratory and seizure variables between the two groups (GC with PGES/BA with those without PGES/BA). A two-sided p-value < 0.05 was considered statistically significant. Data are presented as mean ± standard deviation (median, range).

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References

Summary statistics

Consecutive GC recorded in patients with localization-related epilepsy in the EMU were reviewed. Data collection was stopped when 24 consecutive GC with PGES/BA >20 s were collected. At this point, data from a total of 102 GC in 37 patients (17 female) had been collected, including 30 GC (14 patients) with PGES/BA of any duration and 72 GC (29 patients) without PGES/BA. Six patients had GC both with and without PGES. Mean age was 35.4 ± 12.9 years (34.5, 18–66). We excluded seizures that did not secondarily generalize from analysis, since GC associated with PGES >20 s were identified as the subgroup of seizures shown to be significantly associated with an increased risk of SUDEP (Lhatoo et al., 2010). Seven patients were studied with both scalp and intracranial electrodes as part of their presurgical evaluations. For scalp-recorded seizures only, 85 GC (32 patients) had available oxygen saturation data. PGES was present in 16 of these seizures (11 patients) and in 12 seizures (8 patients) had a duration of >20 s. Sixty-seven scalp-recorded GC (27 patients) had no PGES.

Of the GC with PGES/BA, 10 were of right temporal onset, 7 were of left temporal onset, 2 were of near simultaneous bitemporal onset, 8 were of frontal onset, and the onset could not be determined in the remaining 3 seizures. Of the GC without PGES/BA, 32 were of right temporal onset, 23 were of left temporal onset, 1 of near simultaneous bitemporal onset, 7 were of frontal onset, and seizure onset could not be determined for the remaining 9 seizures.

Analysis of combined scalp-recorded and invasive EEG data

Seizure characteristics for all GC seizures and PGES/BA

The mean seizure duration for all GC with PGES/BA was 133.4 ± 98.3 s (105, 62–516). The mean duration of PGES/BA was 41.55 ± 25.29 s (36, 2–117). For GC without PGES/BA the mean seizure duration was 147.76 ± 85.9 s (131.5, 38–455). The seizure duration was not significantly different between the two groups (p = 0.136). The mean duration of the convulsive component of GC with PGES/BA was 70.67 ± 16.99 s (67, 48–116). The mean duration of the convulsive component of GC seizures without PGES/BA was 80.86 ± 48.71 s (69, 16–324). There was no significant difference in duration of the convulsive component of the seizures between the two groups (p = 0.826) (Table 1, Fig. 1).

Table 1.   Seizure and respiratory characteristics in GC with and without PGES/BA
Combined scalp and invasive recordingsNo PGES/BA (n = 72)All PGES/BA (n = 30)aPGES/BA >20 s (n = 24)bMann-Whitney p-value (comparison with No PGES/BA)
  1. ‘–’ insufficient data.

  2. *Statistically significant result.

  3. ap-value for all PGES/BA.

  4. bp-value for PGES/BA >20 s.

Mean total seizure duration (s)147.76133.4137.460.136a 0.230b
Mean duration of convulsion (s)80.8670.6773.250.826a 0.777b
Mean O2 saturation nadir (%)76.3168.8*67.92*0.002*a 0.003*b
Duration of O2 desaturation <90% (s)81.58140.69*138.520.015*a 0.096b
Mean ictal apnea duration (s)67.4185.150.246a
Preictal to postictal respiratory rate change (breaths/min)12.88.80.196a
Scalp recordings only(n = 67)(n = 16)(n = 12) 
Mean total seizure duration (s)149.72159.72181.00.936a 0.508b
Mean duration of convulsion80.5571.1176.50.876a 0. 589b
Mean O2 saturation nadir (%)75.6968.065.80.002*a 0.004*b
Duration of O2 desaturation <90% (s)86.11174.13185.180.004*a 0.044*b
image

Figure 1.   For all GC with PGES/BA, the mean duration of the entire seizure (yellow plus red bars) and the convulsive component of the seizure (red bars) are depicted in the traces on top. The mean duration of PGES/BA is shown in black. The light blue bar shows the mean onset and termination of apneas. Mean seizure duration and apnea duration for GC without PGES/BA are shown below. The mean time of onset and termination of oxygen desaturation below 90% (relative to seizure onset) is shown by the base of the dark blue triangle. The height of the triangle is proportional to the mean nadir of oxygen saturation (76% for GC without PGES/BA [bottom] and 69% for GC with PGES/BA [top]). The thin black horizontal line represents 60 s. Note that the timing of an oxygen desaturation event includes a circulation delay of approximately 16 s from the pulmonary alveoli to the finger oximeter (Zubieta-Calleja et al., 2005) and an additional oximeter delay.

Download figure to PowerPoint

Seizure characteristics for GC seizures and PGES/BA >20 s

The mean seizure duration for GC with PGES/BA >20 s was 137.46 ± 105.36 s (105, 67–516). This was not significantly different from GC without PGES/BA (p = 0.230). The mean duration of the convulsive component of GC with PGES/BA >20 s was 73.25 ± 16.88 s (70, 50–116) and was not significantly different from GC without PGES/BA (p = 0.777) (Table 1).

Respiratory changes with all GC seizures and PGES/BA

For GC with PGES/BA, the mean time for oxygen saturation to fall below 90% was 67.63 ± 43.41 s (58, −12 to 208) after seizure onset, reaching a nadir at 124.44 ± 71.84 s (104, 45–417) after seizure onset, and returning to above 90% at 210.0 ± 137.26 s (174.5, 76–765) after seizure onset. For GC without PGES/BA, the mean time for oxygen saturation to fall below 90% was 91.68 ± 69.68 s (72.5, 0–404) after seizure onset, reaching a nadir at 124.62 ± 66.82 s (110.5, 45–418) after seizure onset and returning to above 90% at 169.16 ± 69.93 s (162, 52–454) after seizure onset (Fig. 1).

Adequate ictal airflow and abdominothoracic excursion information was available for 35 GC, with and without PGES/BA. Mean apnea onset time was 11.1 ± 19.3 s (14, −27 to 36) after seizure onset in seizures with PGES/BA and 52.3 ± 88.9 s (39, −36 to 322) after seizure onset without PGES/BA. This difference was not statistically significant (p = 0.189). In 32 of the 35 GC, apnea ended before the end of the seizure. In the remaining 3 GC, an apnea outlasted one seizure without PGES/BA by 4 s and two seizures with PGES/BA by 8 and 13 s, respectively. With the latter two seizures, PGES/BA persisted beyond the apnea for 20 and 56 s, respectively. No apneas began in the postictal period in GC with or without PGES/BA. The mean ictal apnea duration was 85.15 ± 48.8 s (89, 11–173) in seizures with PGES/BA, compared to 67.41 ± 23.7 s (68.5, 23–108) in seizures without PGES/BA (p = 0.246) (Fig. 1). Preictal and postictal respiratory rates (RRs) could be determined in 40 seizures, 12 with PGES/BA and 38 without PGES/BA. RRs increased in both groups postictally. In seizures associated with PGES/BA, mean RR change was 8.8 ± 3.5 per min (9, 3–15), whereas in seizures without PGES/BA, mean RR change was 12.8 ± 9.4 per min (12, −6 to 39). There was no significant difference between the two groups (p = 0.196).

The mean oxygen desaturation nadir for all GC with PGES/BA was 68.8 ± 11.8% (71.5, 43–88) and for GC without PGES/BA the mean oxygen desaturation nadir was 76.31 ± 10.17% (79, 42–93). The oxygen desaturation nadirs were significantly different (p = 0.002) (Table 1, Fig. 1). GC with PGES/BA >20 s had a mean oxygen desaturation nadir of 67.92 ± 12.81% (70, 43–88), and this nadir was significantly different from seizures without PGES/BA (p = 0.003) (Table 1). Linear regression analysis showed a significant positive association between SaO2 nadir and duration of PGES (p = 0.028).

The mean duration of oxygen desaturation below 90% for GC with PGES/BA was 140.69 ± 137.14 s (111, 8–712) and for GC without PGES/BA the mean duration of oxygen desaturation was 81.58 ± 52.44 s (70, 10–225). The duration of desaturation was significantly different in the two groups (p = 0.015) (Table 1, Fig. 1). Linear regression analysis showed a significant positive association between duration of PGES and duration of oxygen desaturation (p = 0.02). GC with PGES/BA >20 s had a mean oxygen desaturation duration of 138.52 ± 151.70 s (107, 8–712) and this was not significantly different from GC without PGES/BA (p = 0.096) (Table 1).

ETCO2 data were available for 10 GC with PGES/BA and 28 GC without PGES/BA. The mean peak ETCO2 in GC with PGES/BA was 61.3 ± 10.22 mm Hg (62.5, 44–74). The mean peak ETCO2 in GC without PGES/BA was 53.68 ± 12.27 mm Hg (53.5, 38–94). There was a significant difference between the two groups (p = 0.031). The mean ETCO2 change from baseline was 28.2 ± 11.89 mm Hg (28, 9–46) in GC with PGES/BA, and in GC without PGES/BA the mean ETCO2 change was 15 ± 11.56 mm Hg (11, 2–50). There was a significant difference between the two groups (p = 0.005). A representative VET recording showing PGES with accompanying respiratory data is shown in Fig. 2.

image

Figure 2.   Data from a 28-year-old man with partial onset seizure with a secondarily generalized convulsion. Four frames (top to bottom) depicting onset of respirations following the end of a convulsive seizure. In each frame, the top 25 tracings are EEG channels (calibration bar 400 msec, 100 μV). The next two channels depict nasal airflow and thoracoabdominal excursions). The three electrocardiography channels are leads II, V1, and V5. The next channel is digital display of SaO2 values (% saturation) and the bottom channel shows ETCO2 values (mm Hg). A central apnea is present until the end of the seizure. The narrow arrow indicates oscillations in the nasal airflow and thoracoabdominal channels related to cardiac pulsations. The wide arrow indicates the postictal onset of respirations during EEG suppression. SaO2 drops to 68% (second frame). ETCO2 peaks at 86 mm Hg (third frame) and remains elevated in the fourth frame. Low amplitude delta frequency EEG activity starts in the fourth frame.

Download figure to PowerPoint

Analysis of the subset of scalp-recorded VET data

Seizure characteristics

In GC with PGES, the mean total seizure duration was 159.72 ± 119.85 s (129, 62–516). The mean duration of PGES was 35.74 ± 26.7 s (32, 2–117). In GC without PGES the mean seizure duration was 149.72 ± 88.46 s (132, 38–455). The mean seizure duration was not significantly different (p = 0.936). For GC with PGES >20 s, the mean seizure duration was 181 ± 136.49 s (140.5, 67–516), and this was not significantly different from GC without PGES (p = 0.508). The mean duration of the convulsive component of the seizures with PGES was 71.11 ± 19.79 s (70.5, 48–116). The mean duration of the convulsive component of GC without PGES was 80.55 ± 49.98 s (69, 16–324). The duration of the convulsive component of GC was not significantly different in the two groups (p = 0.876). The mean duration of the convulsive component of GC with PGES >20 s was 76.5 ± 20.45 s (74, 50–116), and this duration was not significantly different from GC without PGES (p = 0.589) (Table 1).

Respiratory characteristics

GC not followed by PGES had a mean oxygen desaturation nadir of 75.69 ± 10.26% (76, 42–93). For GC with PGES the mean oxygen desaturation nadir was 68.0 ± 10.52% (70.5, 43–79). There was a significant difference between the two groups (p = 0.002). For GC with PGES >20 s, the mean oxygen desaturation nadir was 65.83 ± 11.85% (70, 43–79) and this nadir was significantly different from GC without PGES (p = 0.004). In GC without PGES, the mean duration of oxygen desaturation below 90% was 86.11 ± 52.12 s (79, 12–225). In GC with PGES the mean duration of oxygen desaturation was 174.13 ± 162.23 s (122, 30–712). The duration of oxygen desaturation was significantly different in the two groups (p = 0.004). In GC with PGES >20 s, the mean duration of oxygen desaturation was 185.18 ± 195.21 s (115, 30–712) and was significantly different from GC without PGES (p = 0.044) (Table 1).

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References

PGES/BA is not related to either the total duration of the preceding secondarily generalized seizure or to the duration of the convulsive component of the preceding seizure. The presence of PGES/BA is significantly associated with a lower nadir of seizure-related hypoxemia and a higher rise in ETCO2 relative to GC without PGES/BA. The duration of oxygen desaturation is significantly longer in seizures with PGES/BA than in those without. Oxygen desaturation starts and reaches its nadir before the end of the seizure and onset of PGES (Fig. 1). Apnea is not initiated coincident with PGES in the postictal period and ictal apneic events typically do not progress into the postictal period. We have shown previously that the rate and amplitude of airflow increase in the postictal period (Seyal et al., 2010) and the present study confirms that in the setting of PGES the respiratory rate is increased postictally. These data do not support the contention that PGES is associated with inhibition of brainstem respiratory centers (Lhatoo et al., 2010). Rather, our data suggest that PGES more likely is related to the severity of seizure-related intrinsic pulmonary dysfunction.

There is evidence both from animal and human data that seizures can cause pulmonary pathophysiologic changes (Terrence et al., 1981; Johnston et al., 1996; Szabo et al., 2009). In a sheep model of “neurogenic” pulmonary edema with bicuculline-induced status epilepticus, a primary role of pulmonary vascular pressure increases in the production of periictal neurogenic pulmonary edema was demonstrated (Johnston et al., 1996). No direct neural effect on the pulmonary microvasculature could be demonstrated during seizures when the hemodynamic effects were eliminated (Johnston et al., 1996). Loss of capillary integrity can occur after brief increases in vascular pressure, which may explain observations of continued increased capillary permeability in the absence of sustained pulmonary microvascular pressure increase (Johnston et al., 1996). In a study of epileptic baboons, it was shown that almost all animals that died suddenly without apparent cause had pulmonary congestion or edema without evidence of trauma, systemic illness, or heart disease. In contrast, pulmonary edema occurred in only 12% of control baboons, most of whom had evidence of other concurrent nonpulmonary disease (Szabo et al., 2009). In eight cases of young patients with epilepsy who died of SUDEP, it was shown that there was moderate to severe pulmonary edema as well as alveolar hemorrhage. There was no evidence of myocardial disease in these patients (Terrence et al., 1981). Increased lung weights and pulmonary edema are frequently found in autopsy studies of SUDEP cases (Leestma et al., 1989).

Minute ventilation was not directly evaluated and, therefore, our study does not exclude the possibility that, despite the increase in respiratory rate postictally seizures, with PGES impair maximal ventilatory effort. Therefore, postictal hypoventilation rather than seizure-related pulmonary dysfunction resulting in hypoxemia and hypercapnia remains a viable explanation. We have shown that apnea does not occur postictally. However, the present study does not negate the possibility that, at least in some cases of SUDEP, postictal apnea may be the relevant underlying pathophysiologic mechanism.

The mechanisms underlying PGES remain undetermined. PGES may represent the direct effect of seizure-induced hypoxia on the brain. There is evidence that anoxia may result in transient suppression of EEG without prior slowing of background rhythms. Suppression of EEG activity without preceding or following slowing of background frequency has been reported in association with malignant ventricular arrhythmias (Aminoff et al., 1988). Brief anoxia depresses excitatory and inhibitory postsynaptic potentials in the rat. Hypoxia results in rapid and reversible changes in cellular excitability and synaptic mechanisms that combine to achieve cortical circuit function (Fano et al., 2007). Hypoxia rapidly suppresses oscillatory activity in the rat hippocampus (Fano et al., 2007).

Periictal hypercapnia is a potential contributor to PGES. Hypercapnia depresses neuronal activity in tissue slices, rats, and nonhuman primates (Balestrino & Somjen, 1988; Zappe et al., 2008). Mild hypercapnia suppresses cortical afterdischarges and terminates electrographic seizures in animal models and patients with epilepsy (Tolner et al., 2011). Diffuse EEG suppression has been reported in association with acute normoxic hypercapnia in cats and humans (Schindler & Betz, 1976; Mises et al., 1982).

Seizure-related cortical spreading depression (CSD) may also be related to PGES and exacerbated by hypoxia. Leao suggested that CSD and propagating focal seizures were related phenomena, generated by the same cellular elements (Leao, 1944). Since then, the co-occurrence of CSD and seizure activity has been observed in animal studies (Van Harreveld & Stamm, 1953; Koroleva & Bures, 1983) and in human neocortical tissue slices (Avoli et al., 1991; Gorji & Speckmann, 2004). Cortical spreading depression is associated with tissue hypoxia, with loss of transmembrane ion gradients occurring within seconds. During CSD there is a drop in pO2, reflecting a period during which O2 consumption transiently exceeds vascular delivery of O2 (Takano et al., 2007). CSD duration is a direct function of oxygen concentration in inspired air and increasing oxygen availability shortens the duration of CSD (Takano et al., 2007). Seizure-related hypoxia may precipitate CSD, as it has been shown that hypoxia lowers the threshold for elicitation of CSD (Takano et al., 2007).

Our data demonstrate that the severity of ictally triggered hypoxemia and hypercapnia are associated with PGES. Seizure-associated apnea does not typically persist into the postictal period, and there is no evidence of apnea beginning postictally. The presence of PGES following secondarily generalized convulsive seizures may be the consequence of intrinsic seizure-related pulmonary dysfunction, possibly transient pulmonary edema. Our studies do not support the hypothesis that PGES is associated with central apnea in the postictal period.

Acknowledgment

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References

All coauthors have been substantively involved in the study and/or the preparation of the manuscript; no undisclosed groups or persons have had a writers; and all coauthors have seen and approved the submitted version of the paper and accept responsibility for its content.

Disclosure

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References

The authors report no conflicts of interest. 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.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Disclosure
  8. References