Authors: John A. Mason (1Department of Neurology, University of New Mexico, Albuquerque, NM, USA), Elizabeth Juarez-Colunga (2Department of Biostatistics and Informatics, University of Colorado Anschutz Medical Campus, Aurora, CO, USA), Kelly G. Knupp (3Department of Pediatrics and Neurology, University of Colorado, Anschutz Medical Campus, Aurora, CO, USA)
Categories: Article, Infantile spasms, scalp electroencephalography, nonepileptic paroxysmal disorder, West syndrome
Source: Epilepsia
Doi: 10.1111/epi.17967
Authors: John A. Mason, Elizabeth Juarez-Colunga, Kelly G. Knupp
Infantile Epileptic Spasms Syndrome (IESS) is a common and urgent diagnosis with seizure and nonseizure mimics. Evaluation with prolonged video electroencephalogram (EEG) can be time-consuming and costly. This study investigated the use of EEG review of a single sleep-wake cycle to exclude IESS.
We retrospectively reviewed video EEGs to rule out IESS in children between the ages of 2 months and 2 years in the period from January 2019 through June 2020. EEGs were reviewed from the start of the recording through the first sleep-wake cycle and scored as “normal,” “consistent with IESS,” or “abnormal but not diagnostic of IESS.” Scores were compared to the clinical report created by analysis of the entire video EEG.
Inclusion criteria were met in 238 EEG studies. The mean patient age was 7.6 months. The median duration of the full study was 908 minutes compared to 107.5 minutes for the first sleep-wake cycle only. The median difference in recording time was 801 minutes, *p-*value < .01. Scored outcomes were similar. Sixty eight percent were scored as “normal” on first sleep-wake cycle review as compared to 63% (p=.34) on full study review, 13% scored as “consistent with IESS” compared to 16% (p=.51), and 19% scored as “abnormal but not diagnostic of IESS” compared to 21% (p=.65). Sensitivity and specificity of the first sleep-wake cycle review for studies “consistent with IESS” was 84% and 100% respectively. No cases of IESS were scored as normal on first sleep-wake cycle review.
A single sleep-wake cycle captured on EEG can triage studies when IESS is suspected. A normal first sleep-wake cycle did not miss cases of IESS and could result in reduced EEG recording time. Since most of these cases presented to an emergency department, a normal first sleep-wake cycle may help providers determine the acuity, or necessity, of further testing.
Infantile Epilepsy Spasms Syndrome (IESS) is a common and serious epileptic encephalopathy of young children.(1) This epilepsy syndrome has an incidence of between 2.9 and 4.5 per 10,000 live births in the United States and Europe,(2,3) making it the most common epilepsy among infants.(4) IESS carries a high risk of developmental delay and evolution into an intractable epilepsy.(5) Over one third of these children have evolution of their epilepsy and later meet diagnostic criteria for Lennox Gastaut Syndrome.(6)
When the diagnosis of IESS is considered, prompt diagnosis and treatment initiation is required as developmental injury can accrue with ongoing disease.(7) Moreover, response to medication diminishes with time to treatment.(8) IESS is a high-stakes diagnosis that must be made in an accurate and timely manner.
The evaluation of children with suspected IESS can be challenging in young patients who are unable to provide a subjective experience of events and who manifest a large number of mimics, many of which are benign. Behavioral arrest, colic, sleep myoclonus, arousals from sleep, self-gratification, and shuddering are frequently encountered seizure mimics which are evaluated on EEG.(9–11) Nonepileptic paroxysmal events can be difficult to distinguish from seizures, with provider sensitivity as low as 50% for neonates and 75% adults.(12,13) Moreover, evaluation of paroxysmal events on video EEG can be prolonged, with one study reporting under 50% of patients receiving a diagnosis on the first day.(14)
IESS is associated with a severely abnormal background, termed hypsarrhythmia, but this finding is not invariably present.(15) IESS without hypsarrhythmia is reported, comprising 1–33% of cases.(16,17) Hypsarrhythmia may develop after the start of IESS and thus depend on latency to presentation.(18) A normal background, however, is rarely reported as consistent with untreated IESS.(19) The phenomenon of hypsarrhythmia negative IESS necessitates caution in investigating for epileptic spasms if the episode of interest has not been captured on video EEG and electrographically characterized. The presence or absence of hypsarrhythmia has not been shown to affect prognosis or response to treatment.(15)
In light of these challenges, the use of video EEG to evaluate for IESS is a frequent occurrence. Episode capture on video EEG is the gold standard test to evaluate paroxysmal events. The institutional costs of inpatient video EEG monitoring can be one to two thousand US dollars per day and physician interpretation and technologist fees add to that cost.(20) In addition, there is burden to the family in the form of lost work, need for childcare while one child is in the hospital, and emotional toll of an unplanned hospital admission. Given the high volume and long length of EEGs done to rule out IESS, reducing the time needed to perform this investigatory process would save resources.
The current study investigated the use of a single sleep-wake cycle to triage the acuity or need for prolonged video EEG for episode capture. This study compares video EEG of the first sleep-wake cycle alone to video EEG with episode capture. We investigated whether cases of IESS would be missed if a normal EEG was stopped after a first sleep-wave cycle.
This study was reviewed and approved by the Colorado Multiple Institutional Review Board. Using the Children’s Hospital Colorado EEG archive, we identified 694 bedside video EEGs performed on patients between the ages of 2 months and 2 years during the period from January 2019 through June 2020. These EEGs were further screened for inclusion criteria of indication for EEG (which included spell capture or concern for infantile spasms) and exclusion criteria of electronic medical record or technologist’s report indicating a preexisting diagnosis of IESS. Studies meeting these criteria were included in the study for first sleep-wake cycle review and data entry from full prolonged video EEG.
Identified video EEGs were retrieved from the Children’s Hospital Colorado archive for review. EEGs were analyzed from the beginning through the end of the first sleep-wake cycle by author JAM. First sleep-wake cycle review was performed on files which had been clipped for long-term storage by registered EEG technologists trained to include representative periods of wakefulness, sleep, background findings, and events. These studies were scored as “normal” for maturational state, “consistent with IESS,” or “abnormal but not diagnostic of IESS.” The length of time from the beginning of the video EEG until the end of the first sleep-wake cycle was recorded, as were details of the background. These details included the presence, if applicable, of background abnormalities such as generalized or focal slowing, generalized, focal, or multifocal spike discharges, and if seizures or spasms were present. A BASED score, as discussed below, was assigned.
The first sleep-wake cycle was defined as the beginning of the record, including wakefulness through the first sleep cycle (N1, N2, N3, REM). If the record started in N2 sleep, this was considered sufficient if 10 minutes of N2 sleep was obtained even if N1 was not obtained. If REM was not obtained, this was also considered sufficient. This procedure recognizes the relative importance of N2 and N3 sleep in detecting hypsarrhythmia when compared to REM sleep or the waking state.(18) Recording the post-arousal period would increase sensitivity for epileptic spasms; this was not specifically required, however, as the first-wake cycle review was intended to assess background rather than capture events.
Studies were scored as consistent with IESS in two ways. The first was the presence of electroclinical epileptic spasms on the recording. The literature has recognized IESS without background hypsarrhythmia, therefore electroclinical epileptic spasms were sufficient to score a study as consistent with IESS.(21) Second, because the first sleep-wake cycle review is designed to evaluate the electroencephalographic background and not characterize events, the first sleep-wake cycle review could also be scored as consistent with IESS if hypsarrhythmia was present as defined by score of 4 or 5 using the BASED score criteria, 2021 version, developed by Mytinger et al.(22,23)
First sleep-wake cycle scoring was performed independently of chart review and blinded to the interpretation of the complete video EEG. The full video EEG was scored similarly to the first sleep-wake cycle review with information extracted from chart review and the clinical report generated by analysis of the complete video EEG recording. Data were collected in three batched tasks done in the order of participant identification, then first sleep-wake cycle analysis, and then full chart review for clinical status and full EEG details. Batching and order of operations were done to reduce risk of chart review influencing analysis of the first sleep-wake cycle.
Study data were collected and managed using REDCap electronic data capture tools hosted at the Colorado Clinical and Translational Sciences Institute.(24) The following patient demographic and historical data were recorded patient age at time of EEG; patient sex; time since start of concerning episodes; presence or absence of developmental delay; the presence or absence of an existing epilepsy diagnosis; if patient had a diagnosis or features consistent with Tuberous Sclerosis Complex; the presence or absence of abnormal imaging; if antiseizure medications were taken at time of EEG. Patient electroencephalographic data from full video EEG and first sleep-wake cycle review were similarly collected in the REDCap electronic database.
Demographics and baseline characteristics were described overall and for the group with IESS, and odds ratios were calculated for studies consistent with IESS across the exposure variable (e.g. abnormal imaging). Unpaired continuous dependent variables were compared using a two-tailed Mann-Whitney U test. Paired nominal dependent variables were compared using the McNemar-Bowker test. Paired continuous dependent variables compared between the prolonged video EEG and first sleep-wake cycle review were analyzed using a Wilcoxon signed-rank test. Comparison of “IESS without hypsarrhythmia” to “abnormal but not IESS” was conducted using the Mann-Whitney U test and Fisher exact test as appropriate. Sensitivity, specificity, and odds ratio were calculated using a standard 2 by 2 contingency table. The Wilson score interval was used to calculate the confidence intervals for sensitivity and specificity. A significance level of 0.05 was used for all methods of comparison. Analyses were performed using IBM SPSS Statistics, version 28.0.00.
A total of 238 studies met all inclusion criteria. Patient demographics and baseline characteristics are summarized in Table 1 with a significant difference noted in developmental delay, taking antiseizure medication, and abnormal imaging between patients with IESS and those without. Fourteen percent of patients with studies consistent with IESS had Tuberous Sclerosis Complex compared to 3% of total studies (p=.01).
The first sleep-wake cycle review was significantly shorter than the full video EEG recording. The mean and median duration of the full recordings were 960 and 908 minutes respectively (range 55–4569). This is compared to 163 and 108 minutes (range 12–709) for the first sleep-wake cycle review (p = <0.01) (Table 2). The difference between the medians of the two durations was 801 minutes, or 13.3 hours.
There was a statistically significant difference in measured outcomes between the first sleep-wake cycle review and full study review. This was due to eleven studies scored as normal on first sleep-wave cycle review which were noted to have discharges or focal seizures on full EEG review and six studies scored as IESS on full EEG. Importantly, no confirmed cases of IESS were scored as normal on the first sleep-wake cycle review.
Test performance was measured by sensitivity and specificity for IESS comparing first sleep-wake cycle review to the full record as the gold standard. Sensitivity was 84% (95% CI 67–93%), which is a reflection of the presence of cases of IESS without hypsarrhythmia. Specificity was 100% (95% CI 98–100%); there were no cases of IESS on the first sleep-wake cycle review that were scored otherwise in the full report. This gave a positive predictive value of 100% (95% CI 86–100%) and a negative predictive value of 97% (95% CI 94–99%).
Looking at cases of IESS only, there was no statistically significant difference in scoring of background using the first sleep-wake cycle (Table 2). Six patients (16%) with IESS did not have hypsarrhythmia. Of these patients, 1 (16%) had a BASED score of 2 and 5 (83%) had a BASED score of 3. There was no difference in the number of studies scored as consistent with hypsarrhythmia (BASED score of 4 or 5) on the full review and first sleep-wake cycle review.
Comparing cases of IESS without hypsarrhythmia to other abnormal EEGs, there were no baseline patient characteristics that differentiated hypsarrhythmia negative IESS from other abnormal EEGs. The study did not have the power to detect small differences in patient characteristics between groups. Results are presented in Table 3.
Hypsarrhythmia was present in 84% cases of IESS. In cases without hypsarrhythmia, epileptiform discharges were present in the interictal background of all studies. By BASED criteria, 83% (5/6) of these patients were scored as 3, with multifocal epileptiform discharges and without abnormal high voltages over 200 microvolts in more than one channel on a longitudinal bipolar montage and 16% (1/6) scored as 2, with epileptiform discharges in under 3 foci and no channels with abnormal high voltage over 200 microvolts.(23) No case of IESS without hypsarrhythmia had a BASED score under 2.
The first sleep-wake cycle on EEG is an effective screening tool for IESS. It can be recorded in a short duration and reasonably exclude the diagnosis of IESS when normal. This occurred in a majority of infants (63%) referred for episode capture in our cohort. These patients could have potentially been spared prolonged video EEG monitoring. As we have shown, no case of IESS was scored as normal on the first sleep-wake cycle review. This can give confidence to clinicians that further investigation for IESS is unneeded in cases where the first sleep-wake cycle is normal.
The single sleep-wake cycle review was adequate to evaluate interictal background to exclude IESS, but this was not necessarily true of other seizure types. There was a high degree of agreement in BASED score and other interictal findings between first sleep-wake cycle review and full prolonged video EEG. In any case where the differential was limited to IESS and a nonepileptic mimic, a normal interictal background would be sufficient for reassurance.(25) Importantly, it would not rule out epilepsy. Other seizure mimics could suggest either IESS or other epilepsy syndromes. In many epilepsy syndromes a normal interictal background does not preclude the diagnosis of epilepsy; there are cases which would require further monitoring for diagnosis regardless of whether IESS had been excluded.(26–28) In such a case, a prolonged video EEG may be necessary to capture and electroencephalographically characterize an event.
Our study identified hypsarrhythmia in 84% of IESS cases using the BASED score. This is consistent with and within the range reported in the literature; a large multicenter study including 447 patients with IESS reported 82% with hypsarrhythmia.(15) No cases in that report had a normal background. In Watanabe’s 82 patient cohort, only 1 patient did not have hypsarrhythmia at presentation and did develop the pattern later in disease course.(18) In our review, the IESS cases without hypsarrhythmia had epileptiform abnormalities; no patients with IESS had an interictal background without epileptiform discharges. Only one EEG in our review had fewer than three spike foci, meaning that 97% had multifocal epileptiform discharges. There were no EEG studies with slowing as the only manifestation of encephalopathy associated with IESS.
Our study showed a low yield of video electroencephalogram for IESS overall (16%), suggesting that we screen a large number of children for IESS. This likely underestimates the pretest probability of studies done to rule out IESS as our inclusion criteria were intentionally broad to ensure that positive studies were not missed; the yield of studies done where providers specifically suspected IESS would likely be higher. Additionally, there was a low overall yield of abnormal studies. The proportion of normal studies was 63%. Neurologists today are tasked with evaluating a large number of children who are presenting with nonepileptic spells. Even assuming some of these spells have a neurologic basis other than epilepsy, many of these nonepileptic spells were nonpathologic variations of normal behavior.
Our study had several limitations. First, although amply powered for key measures, our study was not powered to detect small differences in the scoring of first sleep-wake cycle review compared to full prolonged video EEG. Manual rereview of a first sleep-wake cycle of sufficient studies for this purpose would have been heroic. This was similarly true of differences between patients with hypsarrhythmia negative IESS and other patients with an abnormal video EEG. Second, first sleep-wake cycle review for this project was altered by institutional curation of completed video EEGs. EEG files are clipped for long-term storage at our institution. Inability to view complete files is an acknowledged limitation of the study. A related limitation is that clipping could limit determination of length of first sleep cycle; when the end of the first sleep-wake cycle was clipped out of a file, the length reported for this study was recorded conservatively as the point at which clipped files definitively showed that the first sleep-wake cycle had passed. If the clip ended midway through a sleep cycle, provided that 10 minutes were recorded as outlined above in methods, the end of the sleep cycle was recorded as the point at which the next clip began. We reported the duration of the first sleep-wake cycle as the start and stop time inclusive of clipped epochs. This tended to overestimate the length of our reported first sleep-wake cycle review times and make our estimate of time savings more conservative. Third, a single investigator performed EEG analysis of the first sleep-wake cycle and chart review for clinical and full EEG details. Steps were taken to reduce the possibility of bias and are described in the methods section. Blinding the reviewer to patient name during the first sleep-wake cycle review was not feasible and is an acknowledged limitation of the study. Fourth, prior annotations on the EEG study from initial analysis for clinical interpretation were still present for retrospective review of the first sleep-wake cycle. Although this introduced the possibility of bias in favor of the interpretation of the original clinical analyst, it was not feasible to have an independent party strip these before retrospective review.
This study demonstrated that a normal sleep-wake cycle on video electroencephalography is inconsistent with a diagnosis of IESS. A normal result obviates the need for further diagnostic investigation for IESS. In the case of an abnormal first sleep-wake cycle, further testing may be required based on clinical concern. IESS can be present in the absence of hypsarrhythmia and thus, even if hypsarrhythmia is not clearly seen, a concern for IESS may persist. Reassuringly, although hypsarrhythmia can be absent, other background abnormalities should prompt a more thorough investigation. The negative predictive value of a normal single sleep-wake cycle approaches 100%. Based on our findings, providers can use the first sleep-wake cycle to triage studies into cases that do, and do not, need prolonged video EEG with episode capture. This could lead to a shortened time and decreased cost of evaluation for a subset of infants who are being evaluated for this life-altering diagnosis.