Abstract
Background
Sudden Unexpected Death in Epilepsy (SUDEP) is the leading category of epilepsy-related mortality. Generalized, particularly nocturnal, convulsive seizures, longstanding epilepsy, and solitary living are retrospectively identified risk factors. No definite electroclinical biomarkers have been prospectively ascertained. This study aimed to identify SUDEP risk markers using multimodality data with long-term follow-up.
Methods
A prospective, multicenter observational cohort study enrolled 2632 children and adults with epilepsy undergoing prolonged videoelectroencephalographic monitoring from 2011 to 2021. Demographic, electroclinical, and cardiorespiratory data were collected at baseline. Patients were followed up long-term for mortality. Primary study endpoint was time to SUDEP. Cox proportional hazards models assessed significant risk factors.
Findings
Incident SUDEP mortality rate was 0.85 (95% confidence interval [CI]: 0.6—1.16) cases per 1000 person-years, from a cohort of 44,803 person-years. Among 2468 patients, 38 succumbed to SUDEP (12 definite SUDEP, 18 probable, 8 possible) and two experienced near-SUDEP. Living alone (hazard ratio [HR] 7.62 [95% CI: 3.94—14.71), experiencing >3 generalized convulsive seizures in the prior year (HR 3.1 [95% CI: 1.64—5.87]), longer ictal central apnea (HR 1.11 [95% CI: 1.05—1.18]), and longer postictal central apnea (HR 1.32 [95% CI: 1.14—1.54]) were significant predictors of increased SUDEP risk. In sub-analysis excluding possible and near-SUDEP cases, longer ictal central apnea was no longer significant.
Interpretation
This study demonstrates an association between premortem peri-ictal apnea and increased SUDEP risk. Cardiorespiratory monitoring during seizures may benefit epilepsy mortality risk assessments. Together with solitary living and convulsive seizure frequency, peri-ictal apnea may inform the development of a validatable SUDEP risk index.
BACKGROUND.
Sudden Unexpected Death in Epilepsy (SUDEP), a catastrophic outcome, is the commonest direct epilepsy-related category of death.1,2 Incidence ranges from 0.09–2.65/1000 person-years in community-based cohorts to 9.3/1000 person-years in epilepsy surgery candidates.1,3 Long term, 9% of persons with active epilepsy succumb to SUDEP.4 Risk factors identified in retrospective case-control studies include a generalized convulsive seizure history (increased risk with greater number of seizures in life or previous year)5,6, nocturnal seizures,6,7 longstanding epilepsy,5,8,9 and solitary living6. Postictal centrally mediated breathing failure and concurrent or subsequent cardiac rhythm disturbances are central components to the agonal cascade leading to SUDEP.10 Studies support potential biomarkers of peri-ictal breathing cessation11,12 and ictal and postictal brainstem-type posturing.12–14 However, since none have been prospectively ascertained,6,8,9,15 our rationale was to address this lack of definite electroclinical biomarkers, and enable development of preventive or interventional measures. Prospective risk factor confirmation could enhance development of premortem risk stratification indices. In our large prospective, multi-center study our objective was to identify risk factors and electroclinical biomarkers predicting SUDEP risk in epilepsy patients, using detailed phenotypic, multimodal electroencephalographic, and cardio-respiratory data acquired in Epilepsy Monitoring Units, with long-term follow-up.
METHODS.
Study Design and Conduct.
We conducted a prospective, multicentric observational cohort study of SUDEP in Epilepsy Monitoring Unit epilepsy patients. Under an approved IRB protocol, children older than two months and adults undergoing routine evaluations were enrolled after informed consent at nine centers from September 2011 until December 2021 (see Supplementary Methods). Participants were informed that they were consenting to participate in a study to specifically identify risk markers of SUDEP. They were informed about SUDEP risk and seizure reduction/mitigation as a means to reduce SUDEP risk as part of clinical care, but not as part of the study design, since this was a non-interventional, observational study. The first admission was considered baseline. Yearly follow-up was conducted through routine clinic visits, review of electronic health records, and, where necessary, telephone interviews to collect seizure frequency, medication status, and mortality information. For patients who died, information on circumstances of death was collected from next of kin, electronic health records, death certificates, and autopsy reports. For survivors, follow-up duration was based on last recorded contact. Two study investigators (S.D.L. and N.L.), experienced in epilepsy death adjudication, determined SUDEP classification using Nashef et al. 2012 criteria.16
Data Collection.
Demographic, clinical, and seizure data were collected on admission, including peri-ictal electroclinical and cardiorespiratory variables,14 detailed in Supplementary Methods. These baseline data were used for the primary, secondary and sub-analyses. Recorded seizures were classified as generalized convulsive seizures (generalized tonic-clonic seizures, generalized clonic seizures, focal-to-bilateral tonic-clonic seizures) and non-convulsive seizures (focal seizures without evolution to bilateral tonic-clonic, atonic, or absence).17 The term generalized convulsive seizures was used because 417 (37%) generalized convulsive seizures did not exhibit a tonic phase. Status epilepticus, brief non-convulsive seizures (<10 seconds), clinical seizures without electroencephalographic correlate, subclinical electroencephalographic seizures, seizures without video, and non-epileptic events were excluded from analysis. Patients who did not complete any follow-up visits were excluded from analyses.
Video-Electroencephalographic and Cardiorespiratory Monitoring.
All patients underwent prolonged video-electroencephalographic monitoring with surface 10–20 international electrode system or invasive electroencephalography (depth and/or subdural grids). Electroencephalogram and standard 1–3 channel electrocardiogram were acquired using clinical platforms. Peripheral capillary oxygen saturation was recorded with digital pulse oximetry, and chest wall and abdominal breathing excursions were recorded with inductance plethysmographs (See Supplementary Methods).
Study Assessments.
Postictal posturing14, peri-ictal breathing,10 peri-ictal arrhythmia18, hypoxemia13, ictal central apnea,19 and postictal central apnea12 assessment methods are detailed in Supplementary Methods. Interictal short-term heart rate variability was assessed, (normalized units of low-frequency power and high-frequency power) since these have been associated with SUDEP.20 Early nursing intervention (oxygen administration or airway suction during the seizure or within five seconds of seizure termination), was also assessed for generalized convulsive seizures.12
Electroencephalographic Assessment:
Postictal generalized electroencephalographic suppression21 was determined by a validated electroencephalographic suppression detection tool22 supplemented visually by two epilepsy neurophysiologists (S.D.L. and L.V.) when tool results were indeterminate (see Supplementary Methods).
Statistical Analysis.
Primary analysis:
Descriptive statistics for continuous variables were reported as median (first quartile, third quartile). For categorical variables, numbers and percentages were provided. SUDEP incidence rates were estimated using person-years. An aggregation method for repeated predictors23 assessed risk factors separately to identify most severe features of each patients’ seizures (maximum of continuous features or presence in any categorical features). The primary endpoint of this study was time to SUDEP or censoring due to other causes. Unadjusted Cox proportional hazards models assessed each risk factor’s hazard ratios. Two-sided p-value < 0.05 after Bonferroni correction assessed risk factors for significance. Risk factors tested included age at epilepsy onset, epilepsy duration, sex, living status, number of generalized convulsive seizures in the prior year, anti-seizure medication, drug resistance status, cardio-respiratory comorbidities, intellectual disability, any arrhythmia during seizures, apnea duration, postictal EEG suppression duration, total hypoxemia duration, postictal posturing, low/high frequency heart rate variability powers during awake and sleep. The proportional hazard assumption was checked by Schoenfeld residuals. The linearity assumption was assessed by fitting Cox proportional hazards models with continuous risk factors as spline terms. All analyses were performed using R (version 4.4.1).24
Secondary analysis:
Adjusted hazard ratios were assessed by entering significant risk factors from the primary analysis into a multivariable Cox proportional hazards model. The Kaplan-Meier method estimated SUDEP probabilities by patient groups stratified by statistically significant risk factors. Median cutoffs were used to measure Kaplan-Meier statistical significance of continuous variables. Subgroup analyses assessed patients who had seizures during admission, had generalized convulsive seizures and non-convulsive seizures only, were drug-resistant, and experienced near/possible SUDEP events. No inferences were planned for the secondary analysis (i.e., no p-value correction or p-values reported), and confidence intervals reported for these non-primary analyses should not be interpreted as hypothesis testing. The generalized linear mixed effects model was used to estimate if early nursing interventions had effects on post-convulsive electroclinical risk factors.
Sensitivity analysis and missing data:
Sensitivity analysis in Cox proportional hazards model analysis included adjustment from study center strata and clinical factors (e.g., epilepsy onset age, epilepsy duration, sex) and analysis from 100 imputed data using multivariate imputation by chained equations.25 Sensitivity analyses were also performed using Cox’s proportional hazards model, adjusting for significant electroclinical variables by seizure count during admission, and using median non-zero seizure values instead of most severe.
FINDINGS
Incidence of SUDEP and description of a prospective cohort
We enrolled 2632 epilepsy patients; 164 were lost to follow-up (Figure 1). Individuals lost to follow-up were similar in characteristics and electroclinical features to included patients (Tables S1 and S2). Among 2468 patients, 96 individuals died of whom 38 succumbed to SUDEP (12 definite SUDEP, 18 probable, and 8 possible; four were SUDEP-plus), , and 58 died from other causes (median follow-up, 35 months [IQR: 18—54]). Two experienced near-SUDEP. There were four SUDEP plus cases (11%): 2 definite and 2 probable, who were also included in the analysis. The incident SUDEP mortality rate for this prospective cohort is 0.85 (95% confidence interval [CI]: 0.6—1.16) cases per 1000 person-years, from a cohort of 44,803 person-years. The incident SUDEP mortality rate was 0.83 (0.38—1.57) for 728 patients treated with monotherapy and 0.89 (0.60—1.27) for 1648 patients treated with polytherapy. Table S3 details SUDEP and near-SUDEP cases. Two near-SUDEP cases, occurring during video-electroencephalographic monitoring after a generalized convulsive seizure in both, were reported previously.12
Figure 1.
Cohort enrollment, follow-up, and analysis. EEG denotes electroencephalographic; EMU, epilepsy monitoring unit; GCS, generalized convulsive seizures; PWE, persons with epilepsy; SUDEP, sudden unexpected death in epilepsy.
Table 1 summarizes demographic and clinical features at first epilepsy monitoring unit admission. During admission, 1660 included patients had 3352 seizures (Figure 1); 1432 patients had 1091 analyzable generalized convulsive seizures and 2117 non-convulsive seizures (electroclinical characteristics in Table 2 and Figure S1).
Table 1. Clinical characteristics of patients with epilepsy in the cohort.
| Characteristic | Overall N = 2,468 |
Non-SUDEP N = 2,428 |
SUDEP/near-SUDEP N = 40 |
|---|---|---|---|
| Age at epilepsy onset—yr, median (IQR) | 15 (7 – 27) | 15 (7 – 27) | 14 (8 – 33) |
| Adult | 18 (9 – 30) | 18 (9 – 30) | 17 (9 – 40) |
| Pediatric | 3 (0.7 – 8) | 3 (0.6 – 7) | 4 (0.1 – 8) |
| Unknown | 115 | 115 | 0 |
| Epilepsy duration at baseline—yr, median (IQR) | 12 (4 – 22) | 12 (4 – 22) | 12 (5 – 30) |
| Unknown | 123 | 123 | 0 |
| Follow-up time—month, median (IQR) | 35 (18 – 54) | 35 (18 – 54) | 33 (18 – 60) |
| Sex—no. (%) | |||
| Male | 1,086 (44) | 1,064 (44) | 22 (55) |
| Female | 1,382 (56) | 1,364 (56) | 18 (45) |
| Age at first EMU admission —yr, median (min-max) | 32 (0.17 – 86) | 32 (0.17 – 86) | 43 (2 – 72) |
| Pediatric | 11 (0.17 – 17) | 11 (0.17 – 17) | 6 (2 – 16) |
| Adult | 35 (18 – 86) | 35 (18 – 86) | 48 (19 – 72) |
| Age group at first EMU admission—no. (%) | |||
| Pediatric | 309 (13) | 302 (13) | 7 (18) |
| Adult | 2,159 (87) | 2,126 (87) | 33 (83) |
| Living status—no. (%) | |||
| Alone | 138 (8.2) | 124 (7.5) | 14 (36) |
| With someone | 1,549 (92) | 1,524 (92) | 25 (64) |
| Unknown | 781 | 780 | 1 |
| Number of GCS in the prior year—no. (%) | |||
| Less than 3 | 1,510 (67) | 1,494 (67) | 16 (41) |
| 3 or more | 745 (33) | 722 (33) | 23 (59) |
| Unknown | 213 | 212 | 1 |
| ASM Treatment—no. (%) | |||
| No medication | 90 (3.6) | 90 (3.7) | 0 (0) |
| Monotherapy | 728 (30) | 719 (30) | 9 (22) |
| Polytherapy | 1,650 (67) | 1,619 (67) | 31 (78) |
| Drug-resistant epilepsy—no. (%) | |||
| Yes | 1,592 (65) | 1,565 (64) | 27 (68) |
| No | 875 (35) | 862 (36) | 13 (33) |
| Unknown | 1 | 1 | 0 |
| Epilepsy type—no. (%) | |||
| Focal | 1,937 (79) | 1,902 (78) | 35 (88) |
| Combined | 86 (3.5) | 86 (3.5) | 0 (0) |
| Generalized | 442 (18) | 437 (18) | 5 (13) |
| Unknown | 3 | 3 | 0 |
| Cardiac comorbidities—no. (%) | |||
| Yes | 163 (6.6) | 157 (6.5) | 6 (15) |
| No | 2,297 (93) | 2,264 (94) | 33 (85) |
| Unknown | 8 | 7 | 1 |
| Respiratory comorbidities—no. (%) | |||
| Yes | 253 (10) | 246 (10) | 7 (18) |
| No | 2,208 (90) | 2,176 (90) | 32 (82) |
| Unknown | 7 | 6 | 1 |
| Intellectual disability—no. (%) | |||
| Yes | 406 (17) | 396 (17) | 10 (26) |
| No | 2,017 (83) | 1,988 (83) | 29 (74) |
| Unknown | 45 | 44 | 1 |
Percentages may not total 100% due to rounding.
ASM denotes antiseizure medication, EMU epilepsy monitoring unit, GCS generalized convulsive seizures, and SUDEP sudden unexpected death in epilepsy.
Table 2. Electroclinical features for each of the patients who experienced seizures during epilepsy monitoring unit admission.
| Characteristic | Overall N = 1,432 |
Non-SUDEP N = 1,401 |
SUDEP/near-SUDEP N = 31 |
|---|---|---|---|
| Number of seizures during EMU admission† —no. (%) | |||
| GCS | 1091 (34) | 1026 (35) | 65 (24) |
| Non-GCS | 2117 (66) | 1909 (65) | 208 (76) |
| Any arrhythmia‡? —no. (%) | |||
| Yes | 761 (54) | 734 (53) | 27 (87) |
| No | 652 (46) | 648 (47) | 4 (13) |
| Unknown | 19 | 19 | 0 |
| Any PGESθ‡? —no. (%) | |||
| Yes | 416 (74) | 400 (74) | 16 (80) |
| Duration—s, median (IQR) | 38 (27 – 47) | 37 (27 – 46) | 49 (44 – 64) |
| No | 144 (26) | 140 (26) | 4 (20) |
| Unknown | 872 | 861 | 11 |
| Any PCA‡? —no. (%) | |||
| Yes | 211 (21) | 198 (20) | 13 (43) |
| Duration—s, median (IQR) | 14 (9 – 25) | 13 (9 – 24) | 21 (14 – 50) |
| No | 820 (79) | 803 (80) | 17 (57) |
| Unknown | 401 | 400 | 1 |
| Any ICA‡? —no. (%) | |||
| Yes | 353 (43) | 344 (43) | 9 (36) |
| Duration—s, median (IQR) | 17 (10 – 28) | 17 (10 – 27) | 44 (28 – 49) |
| No | 471 (57) | 455 (57) | 16 (64) |
| Unknown | 608 | 602 | 6 |
| Any Hypoxemia‡? —no. (%) | |||
| Yes | 367 (98) | 354 (98) | 13 (100) |
| Duration—s, median (IQR) | 62 (35 – 103) | 62 (35 – 101) | 69 (40 – 159) |
| No | 8 (2.1) | 8 (2.2) | 0 (0) |
| Unknown | 1,057 | 1,039 | 18 |
| Hypoxemia severity‡ —no. (%) | |||
| SpO2 signal available but no desaturation | 205 (14) | 202 (14) | 1 (10) |
| SpO2 75–89% | 187 (33) | 180 (32) | 7 (44) |
| SpO2 <75% | 181 (32) | 175 (31) | 6 (38) |
| Unknown | 859 | 844 | 15 |
| Postictal posturing‡ —no. (%) | |||
| Yes | 59 (11) | 53 (9.9) | 6 (30) |
| No | 495 (89) | 481 (90) | 14 (70) |
| Unknown | 878 | 867 | 11 |
| LFP_awake—nu. (IQR) | 51 (33 – 71) | 51 (33 – 71) | 55 (28 – 78) |
| Unknown | 264 | 260 | 4 |
| HFP awake—nu. (IQR) | 45 (28 – 61) | 45 (28 – 61) | 41 (25 – 61) |
| Unknown | 276 | 272 | 4 |
| LFP_sleep—nu. (IQR) | 46 (27 – 65) | 46 (27 – 64) | 48 (24 – 75) |
| Unknown | 288 | 284 | 4 |
| HFP sleep—nu. (IQR) | 52 (35 – 67) | 52 (35 – 67) | 50 (27 – 61) |
| Unknown | 298 | 294 | 4 |
Percentages may not total 100% due to rounding. Postictal generalized electroencephalographic suppression were only available in generalized convulsive seizures. EMU denotes epilepsy monitoring unit, GCS generalized convulsive seizures, HFP high-frequency power, ICA ictal central apnea, LFP low-frequency power, nu normalized units, PCA postictal central apnea, PGES postictal generalized electroencephalographic suppression, SpO2 peripheral oxygen saturation, and SUDEP sudden unexpected death in epilepsy.
1117 patients, including 1091 (78%) non-SUDEP and 26 (84%) SUDEP/near-SUDEP patients, had more than one seizure during EMU admissions. The number of seizures may total over the number of patients.
Most severe features of each patient were reported.
Feature only encountered in generalized convulsive seizures.
Compared to survivors and non-SUDEP deceased, SUDEP (38) and near-SUDEP (2) patients had similar features (i.e., age of epilepsy onset, epilepsy duration at first admission, sex, anti-seizure medication number, drug resistance, cardiorespiratory comorbidities, ictal and postictal hypoxemia, central apnea, postictal generalized electroencephalographic suppression, and follow-up duration) (Table 1 and Table 2). SUDEP/near-SUDEP patients experienced more non-convulsive seizures ( median 3 vs 2) and convulsive seizures (median 2 vs 1) compared to non-SUDEP patients.. Compared to survivors and non-SUDEP deceased, SUDEP and near-SUDEP patients were likelier to live alone (36% vs 7.5%), have >3 generalized convulsive seizures in the prior year (59% of SUDEP/near-SUDEP cases vs 33%), have peri-ictal arrhythmias (87% of SUDEP/near-SUDEP cases vs 53%), have postictal central apnea (43% of SUDEP/near-SUDEP cases vs 20%), and have postictal brainstem posturing (30% of SUDEP/near-SUDEP cases vs 9.9%). Median duration of postictal central apnea was 21 seconds among SUDEP/near-SUDEP cases versus 13 seconds among survivors and deceased non-SUDEP patients (Table 2). SUDEP/near-SUDEP patients had longer duration ictal central apnea and postictal generalized electroencephalographic suppression compared to others (Table 2)
Patients with or without recorded seizures had similar age of onset, sex, and living status. However, patients with recorded seizures were more likely to have SUDEP when compared to patients who did not have seizures available for analysis (2.2% vs 0.8%), longer epilepsy duration (median: 14 vs 9 years), shorter follow-up times (32 vs 41 months), >3 generalized convulsive seizures in the prior year (37% vs 27%), and were treated with >2 antiseizure medications (75% vs 55%) (Table S4).
Primary analysis
Increased SUDEP risk was significantly associated with living alone (HR 7.62 [95%CI 3.94. 14.71], p<0.001), having >3 generalized convulsive seizures in the prior year (HR 3.10 [95%CI 1.64, 5.87], p<0.001)) , longer postictal central apnea duration (HR 1.32 [95%CI 1.14, 1.54] p<0.001), and longer ictal central apnea duration (HR 1.11 [95%CI 1.05, 1.18] p<0.001). (Figure 2 and Table 3). These findings were unchanged after correcting for participating centers as strata and corrections for demographic covariates (e.g., age onset of epilepsy, epilepsy duration, and sex) (Table S5A). Higher SUDEP risk was associated with longer postictal generalized electroencephalographic suppression, peri-ictal arrhythmia, and postictal posturing (Table 3). However, these associations were not significant after Bonferroni correction. The Schoenfeld residuals showed that Cox proportional hazards assumptions were satisfied in all models.
Figure 2.
Estimated cumulative hazard of SUDEP/near-SUDEP by the Kaplan-Meier method, for (A) the full cohort and subgroups by (B) living conditions, (C) number of GCS in the prior year, (D) PCA duration longer or no longer than 14s, and (E) ICA duration longer or no longer than 17s. The median durations of 14s and 17s for PCA and ICA (Table 2) were used as cutoffs, respectively. 95% confidence intervals were shown as shaded areas. GCS, denotes generalized convulsive seizures; ICA, ictal central apnea; PCA, postictal central apnea; s, seconds
Table 3. SUDEP hazard ratios for each risk factor.
Hazard ratios (HR) were estimated by individual Cox proportional hazards models, one for each risk factor. Uncorrected p-values are listed, and significant p-values (<0.05) after Bonferroni correction were marked by *. No estimates were available for non-ASM-treated patients due to no SUDEP cases observed in this subgroup.
| Characteristic | HR | 95% CI | P-value |
|---|---|---|---|
| Age at epilepsy onset, per 1 year increment | 1.01 | (0.99, 1.02) | 0.52 |
| Epilepsy duration at baseline, per 1 year increment | 1.01 | (0.99, 1.04) | 0.21 |
| Sex, male vs female (ref) | 1.68 | (0.90, 3.13) | 0.10 |
| Living status, living alone vs with someone (ref) | 7.62 | (3.94, 14.71) | <0.001* |
| Number of GCS in the prior year, 3 or more vs less than 3 (ref) | 3.10 | (1.64, 5.87) | <0.001* |
| ASM Treatment, Polytherapy vs Monotherapy (ref) | 1.53 | (0.73, 3.21) | 0.26 |
| Drug-resistant, Yes vs No (ref) | 1.33 | (0.68, 2.58) | 0.40 |
| Cardiac Comorbidities, Yes vs No (ref) | 2.34 | (0.98, 5.59) | 0.055 |
| Respiratory comorbidities, Yes vs No (ref) | 1.80 | (0.8, 4.09) | 0.16 |
| Intellectual disability, Yes vs No (ref) | 1.41 | (0.68, 2.89) | 0.35 |
| Any arrhythmia, yes vs no (ref) | 4.96 | (1.74, 14.2) | 0.0028 |
| PCA duration, per 10 second increment | 1.32 | (1.14, 1.54) | <0.001* |
| ICA duration, per 10 second increment | 1.11 | (1.05, 1.18) | <0.001* |
| PGES duration, per 10 second increment | 1.12 | (1.03, 1.22) | 0.0071 |
| Total hypoxemia duration, per 10 second increment | 1.03 | (0.95, 1.12) | 0.45 |
| Postictal posturing, Yes vs No (ref) | 3.36 | (1.29, 8.77) | 0.013 |
| LFP awake | 1.00 | (0.99, 1.02) | 0.76 |
| HFP awake | 1.00 | (0.98, 1.01) | 0.74 |
| LFP sleep | 1.01 | (0.99, 1.02) | 0.45 |
| HFP sleep | 0.99 | (0.98, 1.01) | 0.46 |
ASM denotes antiseizure medication, GCS generalized convulsive seizures, HFP high-frequency power, ICA ictal central apnea, LFP low-frequency power, PCA postictal central apnea, PGES postictal generalized electroencephalographic suppression, and SUDEP sudden unexpected death in epilepsy.
For the full cohort, the Kaplan-Meier estimated five-year probability of SUDEP was 2.2% (95% CI: 1.3% — 3.0%) (Figure 2A). The 5-year SUDEP risk was significantly higher among those living alone (14.9%, 4.1% — 24.4%) vs living with someone (2.1%, 1.1% — 3.1%) (Figure 2B); with >3 generalized convulsive seizures in the prior year (4.3%, 2.2% — 6.5%) vs <3 (1.4%, 0.5% — 2.3%) (Figure 2C); with >14 seconds (median value) postictal central apnea duration (13.4%, 0% — 24.9%) vs <14 seconds (3.6%, 1.8% — 5.3%) (Figure 2D), with > 17 seconds (median value) ictal central apnea duration (7.4%, 1.3% — 13.1%) vs <17 seconds (3.7%, 1.3% — 6.1%) (Figure 2E; see Figure S2).
Secondary analysis
Adjusted multi-variate Cox models hazard ratios indicating higher SUDEP risk remained significant among patients who lived alone (8.25, 95% CI: 3.39 – 20.1), had >3 generalized convulsive seizures in the prior year (2.75, 95% CI: 1.10 – 6.88), longer ictal central apnea duration (1.09 per 10-second increment, 95% CI: 1.01 – 1.17). Ictal central apnea lost significance after adjusting for clinical covariates, and postictal central apnea did not reach statistical significance in the adjusted multi-variate Cox model (Table 4). Findings for solitary living and >3 generalized convulsive seizures in the prior year were unchanged after adjusting covariates (Table 4). When comparing one or more generalized convulsive seizures in the prior year versus none, results were similar to those obtained when using >3 generalized convulsive seizures as a cutoff, and the confidence intervals remained robust (Table S5A). Results remained consistent after adjusting by seizure count, though ictal central apnea lost statistical significance after full adjustment (Table S5B). The generalized linear mixed effects model indicated that early nursing interventions, during or within five seconds after seizure termination, had no effects on presence of arrhythmia, postictal central apnea, or postictal generalized electroencephalographic suppression (Table S6, Table S7, and Table S8).
Table 4. Adjusted hazards ratios (AHR) by the multi-variate Cox’s proportional hazards models with the predictors for sudden unexpected death in epilepsy (SUDEP).
SUDEP predictors: living condition, number of generalized convulsive seizures in the prior year, postictal central apnea, and ictal central apnea duration. The results are presented without adjustment, with adjustment for centers as strata and with adjustment for the covariates (age of onset, epilepsy duration, sex) and center strata. Raw confidence intervals were reported.
| Characteristic | No covariate adjustment |
Adjusted for center |
Adjusted for center, age of onset, epilepsy duration, sex |
|||
|---|---|---|---|---|---|---|
| AHR | 95% CI | AHR | 95% CI | AHR | 95% CI | |
| Living condition | ||||||
| With someone | — | — | — | — | — | — |
| Alone | 8.25 | (3.39, 20.1) | 7.71 | (3.08, 19.3) | 9.07 | (3.17, 25.9) |
| Number of GCS in the prior year | ||||||
| Less than 3 | — | — | — | — | — | — |
| 3 or more | 2.75 | (1.10, 6.88) | 2.76 | (1.10, 6.94) | 2.84 | (1.12, 7.23) |
| PCA duration, per 10 second increment | 1.04 | (0.82, 1.33) | 1.03 | (0.80, 1.32) | 1.05 | (0.81, 1.36) |
| ICA duration, per 10 second increment | 1.09 | (1.01, 1.17) | 1.10 | (1.01, 1.19) | 1.07 | (0.98, 1.16) |
Abbreviations: GCS, generalized convulsive seizures; ICA ictal central apnea; PCA, postictal central apnea.
Subgroup analysis: Generalized convulsive seizures
Patients with generalized convulsive seizures during admissions were similar in most characteristics to those with only non-convulsive seizures recorded, except patients with convulsive seizures recorded had higher SUDEP rates (3.6% vs 1.2%), more with >3 generalized convulsive seizures in the prior year (61% vs 21%), more generalized epilepsy (14% vs 3.8%), peri-ictal arrhythmias (75% vs 40%), and severe hypoxemia (50% vs 14%) (Tables S9 and S10). Individual Cox proportional hazards models on patients with generalized convulsive seizures showed that seizure characteristics had similar trends to our primary analysis in associations with SUDEP under various covariate adjustment considerations, but did not achieve significance (Table S11A). A post-hoc analysis of electroclinical seizure characteristics in non-generalized convulsive seizures only indicated similar findings, except for higher risk of SUDEP in patients with peri-ictal arrhythmia (Table S12A).
Subgroup analysis: Drug-resistant versus initially drug-responsive patients
Percentages of SUDEP cases were similar between drug-resistant (25 SUDEP + 2 near- SUDEP/1592 patients) and initially drug-responsive groups (13/875 patients). At time of SUDEP, all patients were drug-resistant, despite resective epilepsy surgery in eight. In the drug-resistant group, increased SUDEP risk was associated with male sex (drug-resistant hazard ratio = 2.10, [95%CI: 0.96—4.59] vs drug-responsive (hazard ratio =1.05, [0.34—3.23]), three or more generalized convulsive seizures in the prior year (drug-resistant hazard ratio =4.32, [95% CI: 1.81—10.27] vs drug-responsive hazard ratio =1.71, [0.52 — 5.54]), and presence of peri-ictal arrhythmia (drug-resistant hazard ratio =17.72, [95% CI: 2.39 — 131.08] vs drug-responsive hazard ratio =0.74, [0.15 — 3.69]) (Table S13). Living alone increased SUDEP risk in both subgroups and was higher in drug-resistant patients (drug-resistant HR 8.11, [95% CI: 3.67—17.96] vs drug-responsive (HR 6.63, [95%CI 1.99—22.09]) (Table S13).
Subgroup analysis: Definite, probable, possible, and near-SUDEP categories
To sufficiently power analysis, we combined definite and probable groups since the probable SUDEP patients fulfilled all criteria for SUDEP, barring complete autopsy. Demographic, clinical (Table S14), and electroclinical characteristics (Table S15) were similar in all groups. Our primary findings of living status (HR 6.96, [95%CI: 3.24—14.93]), prior year GCS frequency (HR 3.54, [95%CI: 1.67—7.49]), and peri-ictal arrhythmia (HR 4.93, [95%CI: 1.46—16.59]) were unchanged after excluding near-SUDEP and possible SUDEP cases (Table S16). Postictal central apnea (HR 1.25, [95%CI: 1.0—1.56]), postictal generalized EEG suppression (HR 1.12, [95%CI: 1.0—1.25]), and ictal central apnea (HR 1.06, [95%CI: 0.82—1.37]) showed only borderline significance.
Sensitivity analysis.
Analysis of multiple imputed data was similar to primary analysis (Table S17). Using median non-zero seizure values instead of the most severe yielded similar results (Table S18). No non-linear effects of continuous variables were significant (Figure S3).
INTERPRETATION
This multicenter, prospective study of 2632 epilepsy patients (~45,000 person-years of follow-up) who underwent video-electroencephalography and cardiorespiratory monitoring identified 38 sudden unexpected death in epilepsy (SUDEP) outcomes; incident SUDEP mortality rate was 0.85 per 1000 person-years. Living alone, experiencing >3 generalized convulsive seizures in the prior year, longer ictal central apnea, and longer postictal central apnea were significant predictors of increased SUDEP risk in the primary analysis. Peri-ictal arrhythmias, prolonged postictal generalized electroencephalographic suppression, and postictal posturing were more prevalent among SUDEP and near-SUDEP cases than survivors. In addition to prospective confirmation that solitary living and frequent generalized convulsive seizures increase SUDEP risk, we also prospectively observed, for the first time, that prolonged periictal central apneas may be putative biomarkers for individualized SUDEP risk stratification and intervention trials.
Living alone was retrospectively associated with five-fold increased risk of SUDEP, and solitary sleeping with generalized convulsive seizures had 67-fold odds in a case-control study.6 Generalized convulsive seizures in the year preceding death conferred a 27-fold increased risk compared to those without convulsive seizures.6 In our study, SUDEP probability was increased >7-fold in those living alone and >3-fold in those with >3 generalized convulsive seizures in the previous year, even after correcting for age of onset, epilepsy duration, and sex. In our prospectively gathered data, hazard rates are lower than in retrospective study,6 possibly because of smaller SUDEP sample size, and cohort characteristics of patients with less intractable epilepsy.
In our cohort, five-year SUDEP risk was 2.2%, similar to retrospective studies in mixed drug-responsive and drug-resistant epilepsy cohorts.1,2 Living alone (15%) and prolonged postictal central apnea conferred the highest five-year risk (13%), followed by prolonged ictal central apnea (7%), peri-ictal arrhythmias, and >3 generalized convulsive seizures in the prior year (4%). In a large retrospective SUDEP study during video-electroencephalography monitoring, terminal postictal central apnea preceded terminal asystole.10 However, it was unknown if these features in premortem seizures predict death. Our results support that peri-ictal cardiorespiratory features associate with SUDEP risk and may help identify high-risk individuals, even when drug-responsive or with infrequent convulsions; for example, SUDEP can occur shortly after epilepsy diagnosis, when patients are not usually considered high-risk.26,27
Postictal central apnea occurs in 18% of generalized convulsive seizures, and has been shown to occur before terminal asystole in SUDEP occurring under monitoring in the MORTEMUS study.10–12 Our primary analysis supports that such apnea increases SUDEP risk independent of other factors. Prolonged (≥15 seconds) postictal central apnea may be a reliable risk biomarker and may potentially identify high-risk patients. The association was not significant in secondary analyses and should be interpreted in the context of these limitations. Postictal central apnea is hypothesized to reflect brainstem depression via spreading depolarization,28 seizure propagation29 or impaired CO2 detection, leading to ventilatory and arousal failure.30–32 Imaging33 and neuropathological studies34 of SUDEP individuals suggest damage in cardiorespiratory brainstem nuclei, including depletion of neuronal populations in the ventrolateral medulla and raphe nuclei.34 These are sites involved in respiratory rhythm generation and CO2 chemosensing respectively, and dysfunction may lead to failure of postictal homeostatic recovery and thus respiratory compromise.12 However, whether the presence of these neuropathological findings directly predispose or relate to premortem postictal central apnea is currently unknown.
Prolonged ictal central apnea (>17 seconds) was associated with SUDEP risk. However, no significance was observed across secondary analyses, highlighting the need for further prospective validation. Based on brain electrical stimulation studies35,36, ictal central apnea likely arises from seizure discharges invading brain structures that modulate respiration, such as the amygdala, hippocampus, and mesial temporal pole. Brief ictal central apnea is commonly seen in temporal lobe seizures (in ~37% of seizures), but prolonged ictal central apnea is less common.11,37 It is associated with postictal central apnea,11 suggesting that persons with ictal central apnea are more apt to develop postictal breathing cessation. It can cause near SUDEP phenomena,12 and possibly SUDEP. These observations strongly support cardiorespiratory monitoring usage during video-electroencephalography studies, especially in patients with generalized convulsive seizures.
SUDEP risk was also associated with longer postictal generalized electroencephalographic suppression duration, peri-ictal arrhythmia, and postictal posturing, but these findings did not retain significance after Bonferroni correction. Our SUDEP sample was relatively small despite prospective recruitment of many patients; larger cohorts may be confirmatory. However, prospective ascertainment of SUDEP, given its relatively low incidence, requires recruitment of several thousand persons with epilepsy, a hitherto cost-prohibitive constraint. We observed tendency for increased SUDEP risk with peri-ictal arrhythmias, which deserves further study. Concurrent postictal bradyarrhythmia was common in recorded SUDEP,10 and thus analysis of the seizure record for peri-ictal arrhythmias may be important for SUDEP risk assessments. Postictal brainstem posturing has previously been posited as a biomarker.12 Here prospectively, we found its presence tended to associate with increased risk. As a positive motor phenomenon in the absence of cortical seizure discharge, such posturing may reflect brainstem seizure effects.14 Attention to peri-ictal seizure semiology, an easily identifiable clinical phenomenon, may thus be important in SUDEP risk assessments.
Our study had limitations. First, although 38 SUDEP and two near-SUDEP cases represent the largest prospective SUDEP cohort, numbers are still small, limit statistical power and hence cautious interpretation of results is necessary. To increase power, we combined definite, probable, possible SUDEP, and near-SUDEP cases, as near-SUDEP cases would have likely succumbed without resuscitation (‘interrupted SUDEP’). Excluding possible and near-SUDEP resulted in longer ictal central apnea duration losing significance, and both postictal central apnea and postictal generalized EEG suppression became marginally significant, possibly reflecting the relatively small number of SUDEP cases. Most SUDEP occurs during sleep 1,38; our sleep analysis will be reported separately. Monitoring unit seizures may be more severe because of medication withdrawal, leading to higher rates of seizures than at home, and possibly more profound cardiorespiratory seizure effects, thus influencing the results seen. Absence of peri-ictal hemodynamic monitoring limits our understanding of other immediate cardiovascular mechanisms, such as the role of blood pressure, and hence identification of additional risk markers of interest may have been missed. Our definition of peri-ictal arrhythmia was broad since we hypothesized that one or two postictal bradycardic beats may have significant biomarker value.39
This study reveals prospective evidence that premortem detection of peri-ictal breathing dysfunction – namely prolonged ictal central apnea and prolonged postictal central apnea – can potentially identify patients at increased SUDEP risk. Together with active convulsive seizures, solitary living, peri-ictal arrhythmia, postictal generalized electroencephalographic suppression, and postictal posturing, these variables are easily studied in the epilepsy monitoring unit and may provide a SUDEP risk index for prospective validation in a large cohort of patients, a project currently under way. If successful, such an index has the potential to precisely identify individuals at high risk. Such identified persons with epilepsy are likely to benefit from heightened observation and monitoring for post seizure cardiorespiratory dysfunction (such as with wearable technologies), as well as with proactive treatment changes that mitigate seizure frequency (medication changes, epilepsy surgery, neuromodulation). Additionally, the identification of peri-ictal breathing dysfunction paves the way to the investigation of therapies that target breathing rescue. These may include pharmacotherapies in the form of respiratory stimulants, or mechanical interventions such as diaphragmatic pacing. Importantly, this study suggests that there may be significant value in the addition of respiratory assessments in the epilepsy monitoring unit as standard of care, to EEG and ECG, for the identification of breathing dysfunction and SUDEP risk.
Supplementary Material
Research in context.
Evidence before this study
Sudden unexpected death in epilepsy (SUDEP) occurs primarily in persons with drug resistant epilepsy who suffer generalized convulsive seizures. Active epilepsy with generalized convulsive seizures, frequent such seizures, and living alone status, have been identified as important risk factors in retrospective studies. Seizure related electroclinical phenomena, namely postconvulsive breathing cessation, bradyarrhythmias and postictal general electroencephalographic suppression, have been observed in a retrospective case series of epilepsy patients who died while being monitored in epilepsy monitoring units. Such cardiorespiratory and electroencephalographic phenomena can occur premortem in habitual seizures. Peri-ictal central apnea can manifest as ictal central apnea and/or postictal central apnea, both of which have been implicated in the pathogenesis of SUDEP. Since SUDEP is a relatively rare phenomenon with an annual incidence of approximately one per thousand in the epilepsy population, sufficiently powered prospective SUDEP electroclinical risk marker studies have hitherto been cost prohibitive, and have not been undertaken. This has been a major barrier to the precise identification of high-risk individuals with epilepsy, the development of targeted SUDEP interventions based on its pathogenesis, and randomized clinical trials for SUDEP prevention.
Added value of the study
This study was a prospective multicentric observational cohort study of persons with epilepsy in the epilepsy monitoring unit undergoing multimodality assessments of their seizures, including video-electroencephalography and cardiorespiratory measurements, and with long term follow up. The unique value of the study lies in its large scale prospective recruitment of persons with epilepsy, its use of cardiorespiratory monitoring (including oximetry, breathing rate, and multi-channel ECG), and forty cases of ascertained SUDEP and near-SUDEP. It is therefore the first study to prospectively observe SUDEP outcomes in individuals found to have cardiorespiratory dysfunction (peri-ictal central apnea, peri-ictal cardiac arrhythmia), certain seizure semiological features such as postictal posturing, and postictal generalized EEG suppression. Of these, prolonged ictal central apnea and prolonged postictal central apnea were particularly associated with SUDEP. The study also prospectively confirmed previous retrospective observations of the risk posed by frequent, active generalized convulsive seizures, and living alone status.
Implications of all the available evidence
In addition to the presence and frequency of generalized convulsive seizures and living alone status as SUDEP risk factors, this study identified specific, clinically identifiable respiratory risk markers contained in seizures in the form of ictal and postictal central apnea. Additionally, before Bonferroni correction peri-ictal arrhythmia, postictal posturing and prolonged postictal generalized EEG suppression were also associated with SUDEP. An important contribution of this study is the evidence for development of a validatable SUDEP risk index for precise identification of those most at risk of death using these risk markers, so that heightened observation and monitoring, and proactive treatment changes to reduce seizure frequency are made possible. Further, peri-ictal central apneas, as more frequent phenomena than SUDEP itself, have the potential to serve as surrogate markers of SUDEP in preventive clinical trials, thereby reducing cost and increasing feasibility. Lastly, this study paves the way to potential future targeted interventions that harness breathing rescue, whether pharmacological or mechanical, to prevent SUDEP.
Role of the funding source
The study was funded by the National Institutes of Health through grants P20NS076965, U01NS090405, U01NS090406, U01NS090407.
Funding
National Institutes of Health: P20NS076965, U01NS090405, U01NS090406, U01NS090407
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of interests
The authors do not have any direct or indirect conflicts of interest to declare that are pertinent to this work. Detailed declarations are attached.
Contributor Information
Manuela Ochoa-Urrea, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Xi Luo, Department of Biostatistics and Data Science, School of Public Health, The University of Texas Health Science Center at Houston, Houston, TX, USA.
Laura Vilella, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Nuria Lacuey, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Shirin Jamal Omidi, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Norma J Hupp, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Blanca Talavera, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Johnson P Hampson, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA.
Sandhya Rani M R, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Shiqiang Tao, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Xiaojin Li, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Christina Y Miyake, Division of Cardiology, Department of Pediatrics, Texas Children’s Hospital, and Department of Molecular Physiology and Biophysics, Houston, TX, USA; Division of Cardiology, Department of Pediatrics, Texas Children's Hospital, and Department of Molecular Physiology and Biophysics, Houston, TX, USA.
Licong Cui, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Jaison S Hampson, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Ganne Chaitanya, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA.
Yash Shashank Vakilna, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA.
Rup K Sainju, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Internal Medicine. Carver College of Medicine University of lowa, Iowa City, LA, USA; Department of Internal Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA, USA.
Daniel Friedman, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurology, NYU Grossman school of Medicine, New York, NY, USA; Department of Neurology, NYU Grossman School of Medicine, New York, NY, USA.
Maromi Nei, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurology, Sidney Kimmel Medical College, Thomas Jefferson University, Philadelphia, PA, USA; Department of Neurology, Sidney Kimmel Medical College, Thomas Jefferson University, Philadelphia, PA, USA.
Luke Allen, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; University College London Queen Square Institute of Neurology University college London, London, UK; University College London Queen Square Institute of Neurology, University College London, London, UK.
Catherine A Scott, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; University College London Queen Square Institute of Neurology University college London, London, UK; University College London Queen Square Institute of Neurology, University College London, London, UK.
Joana Oliveira, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; University College London Queen Square Institute of Neurology University college London, London, UK; University College London Queen Square Institute of Neurology, University College London, London, UK.
Brian Gehlbach, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Internal Medicine. Carver College of Medicine University of lowa, Iowa City, LA, USA; Department of Internal Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA, USA.
Stephan U Schuele, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Feinberg School of Medicine, Northwestern University, Chicago, IL, USA; Feinberg School of Medicine, Northwestern University, Chicago, IL, USA.
Jennifer A Ogren, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurobiology and the Brain Research Institute, University of California, Los Angeles, Los Angeles, CA, USA; Department of Neurobiology and the Brain Research Institute, University of California, Los Angeles, Los Angeles, CA, USA.
Ronald M Harper, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurobiology and the Brain Research Institute, University of California, Los Angeles, Los Angeles, CA, USA; Department of Neurobiology and the Brain Research Institute, University of California, Los Angeles, Los Angeles, CA, USA.
Beate Diehl, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; University College London Queen Square Institute of Neurology University college London, London, UK; University College London Queen Square Institute of Neurology, University College London, London, UK.
Lisa M Bateman, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurology, Cedars Sinai Medical Center, lot Angeles. CA, USA; Department of Neurology, Columbia University, New York, NY, USA; Department of Neurology, Columbia University, New York, NY, USA; Department of Neurology, Cedars-Sinai Medical Center, Los Angeles, CA, USA.
George B Richerson, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Internal Medicine. Carver College of Medicine University of lowa, Iowa City, LA, USA; Department of Internal Medicine, Carver College of Medicine, University of Iowa, Iowa City, IA, USA.
Jose-Miguel Yamal, Department of Biostatistics and Data Science, School of Public Health, The University of Texas Health Science Center at Houston, Houston, TX, USA.
Guo-Qiang Zhang, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Orrin Devinsky, National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA; Department of Neurology, NYU Grossman school of Medicine, New York, NY, USA; Department of Neurology, NYU Grossman School of Medicine, New York, NY, USA.
Samden D Lhatoo, Department of Neurology, McGovern Medical School, The University of Texas Health Science Center at Houston, Houston, TX, USA; National Institute for Neurological Disorders and Stroke Center for SUDEP Research, Houston, TX, USA.
Data sharing
Complete datasets of all study participants will be shared with bona fide researchers as a National Institutes of Health (NIH) Controlled Access Data Repositories (CADR). Deidentified individual participant data and detailed data dictionary will be made available as soon as requests are approved by the NIH designated Data Advisory Committee. Requests for data will be vetted by the Data Advisory Committee, and approved investigators will be asked to sign data use agreements.
REFERENCES
- 1.Sudden Devinsky O., unexpected death in epilepsy. N Engl J Med 2011;365(19):1801–1811. doi: 10.1056/NEJMra1010481 [DOI] [PubMed] [Google Scholar]
- 2.Devinsky O, Hesdorffer DC, Thurman DJ, Lhatoo S, Richerson G. Sudden unexpected death in epilepsy: epidemiology, mechanisms, and prevention. Lancet Neurol 2016;15(10):1075–1088. doi: 10.1016/S1474-4422(16)30158-2 [DOI] [PubMed] [Google Scholar]
- 3.Tomson T, Nashef L, Ryvlin P. Sudden unexpected death in epilepsy: current knowledge and future directions. Lancet Neurol. 2008;7(11):1021–1031. doi: 10.1016/S1474-4422(08)70202-3 [DOI] [PubMed] [Google Scholar]
- 4.Sillanpaa M, Shinnar S. Long-term mortality in childhood-onset epilepsy. N Engl J Med 2010;363(26):2522–2529. doi: 10.1056/NEJMoa0911610 [DOI] [PubMed] [Google Scholar]
- 5.Hesdorffer DC, Tomson T, Benn E, et al. Combined analysis of risk factors for SUDEP. Epilepsia 2011;52(6):1150–1159. doi: 10.1111/j.1528-1167.2010.02952.x [DOI] [PubMed] [Google Scholar]
- 6.Sveinsson O, Andersson T, Mattsson P, Carlsson S, Tomson T. Clinical risk factors in SUDEP: A nationwide population-based case-control study. Neurology. 2020;94(4):e419–e429. doi: 10.1212/WNL.0000000000008741 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sveinsson O, Andersson T, Carlsson S, Tomson T. Circumstances of SUDEP: A nationwide population-based case series. Epilepsia. 2018;59(5):1074–1082. doi: 10.1111/epi.14079 [DOI] [PubMed] [Google Scholar]
- 8.Walczak TS, Leppik IE, D’Amelio M, et al. Incidence and risk factors in sudden unexpected death in epilepsy: a prospective cohort study. Neurology. 2001;56(4):519–525. doi: 10.1212/wnl.56.4.519 [DOI] [PubMed] [Google Scholar]
- 9.Nilsson L, Farahmand BY, Persson PG, Thiblin I, Tomson T. Risk factors for sudden unexpected death in epilepsy: a case-control study. Lancet. 1999;353(9156):888–893. doi: 10.1016/s0140-6736(98)05114-9 [DOI] [PubMed] [Google Scholar]
- 10.Ryvlin P, Nashef L, Lhatoo SD, et al. Incidence and mechanisms of cardiorespiratory arrests in epilepsy monitoring units (MORTEMUS): a retrospective study. Lancet Neurol 2013;12(10):966–977. doi: 10.1016/S1474-4422(13)70214-X [DOI] [PubMed] [Google Scholar]
- 11.Vilella L, Lacuey N, Hampson JP, et al. Incidence, Recurrence, and Risk Factors for Peri-ictal Central Apnea and Sudden Unexpected Death in Epilepsy. Front Neurol 2019;10:166. doi: 10.3389/fneur.2019.00166 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Vilella L, Lacuey N, Hampson JP, et al. Postconvulsive central apnea as a biomarker for sudden unexpected death in epilepsy (SUDEP). Neurology. 2019;92(3):e171–e182. doi: 10.1212/WNL.0000000000006785 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rheims S, Alvarez BM, Alexandre V, et al. Hypoxemia following generalized convulsive seizures: Risk factors and effect of oxygen therapy. Neurology. 2019;92(3):e183–e193. doi: 10.1212/WNL.0000000000006777 [DOI] [PubMed] [Google Scholar]
- 14.Vilella L, Lacuey N, Hampson JP, et al. Association of Peri-ictal Brainstem Posturing With Seizure Severity and Breathing Compromise in Patients With Generalized Convulsive Seizures. Neurology. 2021;96(3):e352–e365. doi: 10.1212/WNL.0000000000011274 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Langan Y, Nashef L, Sander JW. Case-control study of SUDEP. Neurology. 2005;64(7):1131–1133. doi: 10.1212/01.WNL.0000156352.61328.CB [DOI] [PubMed] [Google Scholar]
- 16.Nashef L, So EL, Ryvlin P, Tomson T. Unifying the definitions of sudden unexpected death in epilepsy. Epilepsia. 2012;53(2):227–233. doi: 10.1111/j.1528-1167.2011.03358.x [DOI] [PubMed] [Google Scholar]
- 17.Beniczky S, Trinka E, Wirrell E, et al. Updated classification of epileptic seizures: Position paper of the International League Against Epilepsy. Epilepsia. 2025;66(6):1804–1823. doi: 10.1111/epi.18338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Park KJ, Sharma G, Kennedy JD, Seyal M. Potentially high-risk cardiac arrhythmias with focal to bilateral tonic–clonic seizures and generalized tonic–clonic seizures are associated with the duration of periictal hypoxemia. Epilepsia. 2017;58(12):2164–2171. doi: 10.1111/epi.13934 [DOI] [PubMed] [Google Scholar]
- 19.Lacuey N, Talavera B, Magana-Tellez O, et al. Ictal Central Apnea Is Predictive of Mesial Temporal Seizure Onset: An Intracranial Investigation. Ann Neurol 2024;95(5):998–1008. doi: 10.1002/ana.26888 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sivathamboo S, Friedman D, Laze J, et al. Association of Short-term Heart Rate Variability and Sudden Unexpected Death in Epilepsy. Neurology. 2021;97(24):e2357–e2367. doi: 10.1212/WNL.0000000000012946 [DOI] [PubMed] [Google Scholar]
- 21.Lhatoo SD, Faulkner HJ, Dembny K, Trippick K, Johnson C, Bird JM. An electroclinical case-control study of sudden unexpected death in epilepsy. Ann Neurol 2010;68(6):787–796. doi: 10.1002/ana.22101 [DOI] [PubMed] [Google Scholar]
- 22.Theeranaew W, McDonald J, Zonjy B, et al. Automated Detection of Postictal Generalized EEG Suppression. IEEE Trans Biomed Eng 2018;65(2):371–377. doi: 10.1109/TBME.2017.2771468 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Goldstein BA, Pomann GM, Winkelmayer WC, Pencina MJ. A comparison of risk prediction methods using repeated observations: an application to electronic health records for hemodialysis. Stat Med 2017;36(17):2750–2763. doi: 10.1002/sim.7308 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Team CR. R: A Language and Environment for Statistical Computing. [software] [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2017. Available from https://www.r-project.org/. R Foundation for Statistical Computing. 2022 https://www.r-project.org/ [Google Scholar]
- 25.van Buuren S, Groothuis-Oudshoorn K mice: Multivariate Imputation by Chained Equations in R. J Stat Softw 2011;45(3). doi: 10.18637/jss.v045.i03 [DOI] [Google Scholar]
- 26.Ross J, Tyrlikova I, Giermek A, Ochoa-Urrea M, Lhatoo S, Klein P. Sudden Unexpected Death in Epilepsy During Ambulatory Video-EEG Monitoring at Home. Neurology. 2024;102(2):e208041. doi: 10.1212/WNL.0000000000208041 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Leestma JE, Walczak T, Hughes JR, Kalelkar MB, Teas SS. A prospective study on sudden unexpected death in epilepsy. Ann Neurol 1989;26(2):195–203. doi: 10.1002/ana.410260203 [DOI] [PubMed] [Google Scholar]
- 28.Aiba I, Noebels JL. Spreading depolarization in the brainstem mediates sudden cardiorespiratory arrest in mouse SUDEP models. Sci Transl Med 2015;7(282):282ra46. doi: 10.1126/scitranslmed.aaa4050 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lertwittayanon W, Devinsky O, Carlen PL. Cardiorespiratory depression from brainstem seizure activity in freely moving rats. Neurobiol Dis 2020;134:104628. doi: 10.1016/j.nbd.2019.104628 [DOI] [PubMed] [Google Scholar]
- 30.Sainju RK, Dragon DN, Winnike HB, et al. Hypercapnic ventilatory response in epilepsy patients treated with VNS: A case-control study. Epilepsia. 2021;62(9):e140–e146. doi: 10.1111/epi.16997 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Buchanan GF. Impaired CO2-Induced Arousal in SIDS and SUDEP. Trends Neurosci 2019;42(4):242–250. doi: 10.1016/j.tins.2019.02.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Teran FA, Sainju RK, Bravo E, et al. Seizures Cause Prolonged Impairment of Ventilation, CO2 Chemoreception and Thermoregulation. J Neurosci 2023;43(27):4959–4971. doi: 10.1523/JNEUROSCI.0450-23.2023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Wandschneider B, Koepp M, Scott C, et al. Structural imaging biomarkers of sudden unexpected death in epilepsy. Brain. 2015;138(Pt 10):2907–2919. doi: 10.1093/brain/awv233 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Patodia S, Somani A, O’Hare M, et al. The ventrolateral medulla and medullary raphe in sudden unexpected death in epilepsy. Brain. 2018;141(6):1719–1733. doi: 10.1093/brain/awy078 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lacuey N, Hampson JP, Harper RM, Miller JP, Lhatoo S. Limbic and paralimbic structures driving ictal central apnea. Neurology. 2019;92(7):e655–e669. doi: 10.1212/WNL.0000000000006920 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Nobis WP, González Otárula KA, Templer JW, et al. The effect of seizure spread to the amygdala on respiration and onset of ictal central apnea. J Neurosurg 2020;132(5):1313–1323. doi: 10.3171/2019.1.JNS183157 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Lacuey N, Zonjy B, Hampson JP, et al. The incidence and significance of periictal apnea in epileptic seizures. Epilepsia. 2018;59(3):573–582. doi: 10.1111/epi.14006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Lamberts RJ, Thijs RD, Laffan A, Langan Y, Sander JW. Sudden unexpected death in epilepsy: people with nocturnal seizures may be at highest risk. Epilepsia. 2012;53(2):253–257. doi: 10.1111/j.1528-1167.2011.03360.x [DOI] [PubMed] [Google Scholar]
- 39.Vilella L, Miyake CY, Chaitanya G, et al. Incidence and Types of Cardiac Arrhythmias in the Peri-Ictal Period in Patients Having a Generalized Convulsive Seizure. Neurology. 2024;103(1):e209501. doi: 10.1212/WNL.0000000000209501 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Complete datasets of all study participants will be shared with bona fide researchers as a National Institutes of Health (NIH) Controlled Access Data Repositories (CADR). Deidentified individual participant data and detailed data dictionary will be made available as soon as requests are approved by the NIH designated Data Advisory Committee. Requests for data will be vetted by the Data Advisory Committee, and approved investigators will be asked to sign data use agreements.