Abstract
Anti–N-methyl-
However, rates of clinical sleep symptoms with NMDA receptor encephalitis have not been previously reported in children. In one small (n = 32) single-center study, sleep disturbances were reported in 25% NMDA receptor encephalitis pediatric patients, but sleep information was not available for the rest of the patients. 10 Additionally, use of sleep medications and sedation medications in pediatric NMDA receptor encephalitis have not been reported. Sleep severity can be assessed by patient or caregiver report, or reflected by which sedation medications are used in children with NMDA receptor encephalitis. For example, propofol is only used in severe situations in children unable to be sedated with other medications because of the risk of propofol infusion syndrome. 11 Thus, propofol administration could be a clinical indicator of severe sleep dysfunction in pediatric NMDA receptor encephalitis. Understanding sleep dysfunction in NMDA receptor encephalitis is important as sleep disturbance could be used to help monitor clinical status, be used as a symptom to reflect underlying disease pathogenesis, and may affect clinical recovery. 12 One common metric for assessing outcomes in NMDA receptor encephalitis is the modified Rankin Score (mRS), a functional outcome scale in neurology. The mRS scale spans from 0 to 6, with 0 indicating no symptoms and 6 indicating death. 13
Our primary aim is to study the rates of sleep dysfunction at disease onset and the use of sleep and/or sedation medications in pediatric NMDA receptor encephalitis. Our secondary aim is to assess whether poor sleep at onset or persistent poor sleep at 1 year is associated with 1-year outcomes (good mRS = 0-2, poor mRS = 3-6) among children with NMDA receptor encephalitis.
Patients and Methods
Setting, Participants, and Study Size
Participant data in this pediatric NMDA receptor encephalitis retrospective observation cohort was collected from a single free-standing pediatric hospital in the southeastern United States between January 1, 2010, and June 1, 2022.14–16 Patients under the age of 21 years were included. Patients were identified by reviewing all patients with any diagnostic code for anti–NMDA receptor autoimmune encephalitis (G04.81). We then included only patients with confirmed NMDA receptor encephalitis, defined by positive anti–NMDA receptor antibody testing in the cerebrospinal fluid and only if discussion about sleep occurred at time of presentation in the medical record from manual chart review. The final sample size was 41 patients, which was obtained by including all eligible participants. One-year outcomes were assessed in 32 patients who had modified Rankin Scores available at 1 year (see Figure 1).
Flowchart of selection of patients with anti–N-methyl-
Data Sources and Variables
Data were obtained from chart review of the electronic medical record. Clinical information including demographics, symptoms, laboratory results, EEG, magnetic resonance imaging (MRI), and treatments, such as immunotherapy and sleep-related medications, were included in the study. Time to treatment was defined as the time from symptom onset to the time of receiving first immune treatment (steroids or intravenous immunoglobulin) administered in the hospital. Time to improvement from treatment was defined as the time of receiving first immune treatment to the time when the patient started to clinically improve, as discussed by the medical providers. Length of stay including hospital length of stay and inpatient rehabilitation length of stay were included. Sleep problems at onset was defined as documented sleep complaints, including insomnia or hypersomnia, noted at the time of presentation. Sleep problems at 1 year was defined as the presence or absence of documented sleep concerns at 1 year from onset. We also examine the use of sleep medications and/or sedation medications used in our cohort during their hospitalization and at 1 year. We also compared the sleep problems in our pediatric NMDA receptor encephalitis cohort to previously published rates of sleep problems in adults with autoimmune encephalitis 17 and healthy children and adolescents. 18
Bias, Missing Data, and Data Cleaning
Missing data are reported in Tables 1 to 3 and the Results section. Multiple imputation was not performed because of a small sample size and with the degree of missing data. Data on sleep and sedation medications were cleaned to reflect if medications were used for sleep/sedation and not for any other causes. To avoid bias, we only included patients in whom sleep problems were clearly documented and discussed by the patient or caregivers, and not inferred from the chart or other medical observers. This left 7 patients for whom we did not know the sleep status and who thus were excluded from the initial analysis. We also performed a sensitivity analysis in which the 7 patients were initially excluded due to unknown sleep problems as they were not discussed at onset by categorizing them as not having sleep problems.
Demographic and Clinical Characteristics of Pediatric Anti–N-Methyl-
Abbreviations: CSF, cerebrospinal fluid; F, female; ICU, intensive care unit; LOS, length of stay; M, male; MRI, magnetic resonance imaging; mRS, modified Rankin score; SD, standard deviation; WBC, white blood cell.
Boldface indicates significance (P < .05). Numbers in square brackets indicate the number of missing data points.
First-line treatment included steroids, intravenous immunoglobulin, and plasmapheresis.
Second-line treatments included rituximab and cyclophosphamide.
Time to first-line treatments defined as number of days from symptom onset to initial date for first-line treatments.
Time to second-line treatments defined as number of days from symptom onset to initial date for second-line treatments.
Time to improvement defined as numbers of days from the initial date of first treatment administered to date that patient began to improve.
Defined as improved after 4 weeks of treatment onset.
Demographic and Clinical Characteristics of Pediatric Anti–N-methyl-
Abbreviations: CSF, cerebrospinal fluid; F, female; ICU, intensive care unit; LOS, length of stay; M, male; MRI, magnetic resonance imaging; mRS, modified Rankin score; SD, standard deviation; WBC, white blood cell.
Sleep and sedation medications are also included. Boldface indicates significance (P < .05). Numbers in square brackets indicate the number of missing data points.
First-line treatment included steroids, intravenous immunoglobulin, and plasmapheresis.
Second-line treatments included rituximab and cyclophosphamide.
Time to first-line treatments defined as number of days from symptom onset to initial date for first-line treatments.
Time to second-line treatments defined as number of days from symptom onset to initial date for second-line treatments.
Time to improvement defined as numbers of days from the initial date of first treatment administered to date that patient began to improve.
Defined as improved after 4 weeks of treatment onset.
Results From Adjusted Models Showing Effect of Needing Propofol for Sedation on 1-Year Outcomes by Modified Rankin Score (Good = 0-2 and Poor ≥ 3).
Abbreviations: aOR, adjusted odds ratio; 95% CI, 95% confidence interval; OR, odds ratio.
Adjusted for time to first treatment, intensive care unit (ICU) admission, time to treatment response, abnormal electroencephalography (EEG), did not improve in 4 weeks, and abnormal magnetic resonance imaging (MRI).
Adjusted for time to first treatment, intensive care unit (ICU) admission, time to treatment response, abnormal electroencephalography (EEG), did not improve in 4 weeks, and abnormal magnetic resonance imaging (MRI).
Quantitative Variables
One-year outcomes, defined as 1 year from the onset of symptoms, were assessed from the clinical chart with the pediatric mRS from 0 to 6, with scores of 0 to 2 indicating good outcomes and scores of 3 or greater indicating poor outcomes.13,19 The mRS is a measure of functional motor status and has been used as a measure of disability after neurologic symptoms, including stroke. 13
Statistical Analysis
Statistical analyses were performed using SAS v.9.4 (Cary, NC). Descriptive statistics were performed to discuss rates of sleep problems and other clinical characteristics in our cohort. For continuous variables, means and standard deviations were reported for relatively normally distributed populations, and medians and interquartile ranges were reported for skewed data. Percentages and frequencies were reported for proportions. Differences in characteristics between patients with good and poor outcomes at 1 year were assessed using the Student t test for means for approximately normal distributions, Wilcoxon rank-sum test for skewed data, and the chi-square test for proportions.
A multivariate logistic regression model assessed 3 independent predictors: (1) poor sleep at onset, (2) use of propofol, or (3) poor sleep at 1 year to predict the dependent variable of poor outcomes at 1 year. Propofol was selected as a sign of severe sleep dysfunction as propofol is usually reserved for children who are unable to be sedated with other medications. 11 Other covariates were included in the model: time to first treatment, intensive care unit (ICU) admission, time to treatment response, abnormal electroencephalography (EEG), and abnormal MRI. These covariates were included because of their association with worse outcomes.20-22 Complete cases were included in the logistic regression. We then performed linear regression with mRS at 1 year as the outcome variable and then included a backwards stepwise regression analysis to determine the optimal model for poor outcomes at 1 year.
Results
Fifty-six patients were identified with a diagnosis of anti-NMDA receptor encephalitis in the medical record. One patient did not have a positive cerebrospinal fluid antibody, and cerebrospinal fluid anti–NMDA receptor antibody results were not available for 7 patients; the remaining 48 had confirmed NMDA receptor encephalitis with a positive cerebrospinal fluid anti–NMDA receptor antibody and neurologic symptoms. Of 48 patients, 41 reported sleep symptoms during the initial neurology consult note and were included in the study (Figure 1). Sleep problems were reported in 95% (39/41) of patients at the time of admission, with a majority (38/39) endorsing insomnia and only 1 patient with hypersomnia. If those excluded because of unknown sleep problems at onset were presumed to have no sleep problems, the proportion of sleep problems in our cohort would be 81% (39/48). Because the majority of NMDA receptor encephalitis patients in this study have reported sleep problems, with only 2 reporting no sleep problems, we then examined exposure to propofol as a marker for sleep severity in NMDA receptor encephalitis. When comparing characteristics in those who received propofol (n = 21) versus those who did not (n = 20), no differences were observed (Table 1). We also compared ICU admission and catatonia, that is, which of the patients who received propofol were in the ICU, and no differences in catatonia was observed between those who received propofol and those who did not.
As for 1-year outcomes, 32 patients (78%) had mRS scores available at 1 year. We then compared characteristics in those with poor (mRS 3-6) versus those with good (mRS 0-2) status to predict 1-year outcomes (Table 2). The characteristics that were different between those with good versus poor 1-year outcomes included use of second-line treatment, inpatient rehabilitation length of stay, ICU admission, no improvement after 4 weeks of treatment, and zolpidem (Table 2). As for persistent sleep problems, 34% (11/32) reported persistent sleep difficulty at 1 year after disease onset, which was insomnia for all 11 patients. ICU admission and catatonia did not increase the risk for sleep complaints at 1 year (P = .1958 and P = .7423, respectively). We also examined those who were still on sleep mediations at 1 year and found that 11 of 32 (34%) were on sleep medications at 1 year. Persistent sleep complaints at 1 year associated with use of sleep medications at 1 year (P = .001); sleep medications were still used at 1 year in 3 of 22 (14%) of those without sleep complaints, as compared to 8 of 10 (80%) of those with sleep complaints.
We then performed logistical regression to assess for whether the use of propofol, sleep problems at onset, or sleep problems at 1 year were associated with poor outcomes at 1 year. Propofol and sleep problems at 1 year did not predict poor outcomes, even after adjusting for time to first treatment, ICU admission, time to treatment response, abnormal EEG, abnormal MRI, and no clinical improvement in 4 weeks. We then performed sensitivity analysis by including those with missing data on sleep problems at onset; that is, because sleep was not discussed at the time of admission, we presumed that they had no sleep problems at onset. However, this sensitivity analysis did not change the results because those patients also did not have mRS available at 1 year. When we used linear regression with backward variable selection to predict mRS at 1 year, no improvement at 4 weeks was the only significant variable (OR 18.0, 95% CI 1.7, 191.5, P = .0166). However, when using the Spearman rank correlation coefficient, sleep problems at 1 year was associated with mRS scores at 1 year (P = .011). We also compared proportions of sleep problems in pediatric NMDA receptor encephalitis compared to healthy pediatric patients 18 and found higher rates of sleep problems in our pediatric NMDA receptor encephalitis cohort as compared to 22.6% of 855 children aged 4-9 years and 20% of 1047 adolescents aged 10-17 years15 (P < .00001). Sleep problems in pediatric NMDA receptor encephalitis were also increased as compared to adult autoimmune encephalitis patients, in which 19 of 26 (76%) reported sleep difficulties 17 (P < .01).
Discussion
This, to our knowledge, is the first assessment of sleep dysfunction in children with NMDA receptor encephalitis. We found that a high proportion of children with NMDA receptor encephalitis had sleep problems at the time of presentation with their disease. Even by presuming that those in whom sleep was not discussed had no sleep dysfunction, 81% of children with NMDA receptor encephalitis had sleep disturbances at onset of disease. Persistent sleep problems were also observed at 1 year from onset and may be associated with mRS scores. We also included ICU admission because it can affect sleep, including circadian rhythm sleep-wake disturbance, and although ICU admission did correlate with poor outcomes as has been shown in other studies,3,20,22-24 ICU admission was not associated with sleep complaints at 1 year.
Sleep disturbance (both insomnia and hypersomnia) is a common feature in the clinical course of NMDA receptor encephalitis, although not part of the core diagnostic criteria. Adults with NMDA receptor encephalitis report having refractory insomnia with difficulty falling asleep, maintaining sleep, and decreased need for sleep as some of the more prominent features. 9 In our cohort, all but 1 had insomnia, with 1 patient having hypersomnia. Another metric for sleep dysfunction can be measured through polysomnography (PSG). PSG includes electroencephalography (EEG) to monitor the electrophysiological characterizations of sleep stages. Alterations in sleep patterns have been observed on electroencephalography (EEG) in adults with NMDA receptor encephalitis. 25 Further studies could examine the association between sleep complaints and alterations in sleep stages. Insufficient data was available on whether sleep stages were captured on EEG in our cohort, and in a subset of our cohort (n = 11), we have previously demonstrated that loss of normal sleep architecture is associated with poor outcomes as defined by the modified Rankin Score (mRS ≥ 3) at 1 year. 21 These EEG findings included that the sleep stages of non–rapid eye movement (NREM) is absent in 4 and rapid eye movement (REM) is absent in 7 children with NMDA receptor encephalitis on the first 24-hour EEG at time of presentation, 21 suggesting that sleep is also electrographically affected in many pediatric NMDA receptor encephalitis patients. Additionally, additional studies could determine whether the return of electrographic sleep stages marks disease improvement and therefore serve as a biomarker of recovery.
The pathophysiology behind sleep disturbances in NMDA receptor encephalitis is likely complex and involving different pathways. Different neuroanatomic areas of the brain are involved in sleep along with different neurotransmitters. For example, anterior hypothalamic lesions can result in insomnia because of impaired GABAergic pathways, or lesions in the brainstem, basal forebrain (acetylcholine), and orexin-producing neurons in the posterior and lateral hypothalamus can impair arousal. 26 Moreover, glutamatergic signaling is associated with the parabrachial nucleus for wakefulness26,27; and with NMDA receptors being glutamatergic, 28 further studies are required to identify the pathways involved in sleep disturbances in NMDA receptor encephalitis.
Persistent sleep problems at 1 year were associated with increasing mRS scores and poor 1-year outcomes. We also included ICU admission as that can affect sleep, including circadian rhythm sleep-wake disturbance, and although ICU admission did correlate with poor outcomes as has been shown in other studies,3,20,22-24 ICU admission was not associated with sleep complaints at 1 year.
Recovery in NMDA receptor encephalitis includes resolution of symptoms, and sleep disturbances are part of the natural history of NMDA receptor encephalitis. Sleep is important for recovery after brain injury, such as in concussion, and ongoing sleep dysfunction may contribute to poor recovery. 29 Likewise, for recovery in NMDA receptor encephalitis, poor sleep may be contributing to delayed recovery and therefore poor outcomes. Whether persistent sleep problems reflect ongoing disease severity versus contributing to poor recovery is unknown. Either way, we recommend striving for sufficient sleep to patients as additional intervention to help with recovery. Moreover, quality of life issues may be overlooked in the management of children with NMDA receptor encephalitis, and thus treating sleep problems, such as modafinil for hypersomnia or melatonin or hypocretin receptor antagonists for insomnia, may improve quality of life or outcomes. Surprisingly, obtaining sufficient sleep is often undervalued and overlooked, as sleep deprivation affects as high as 68% of healthy children in high school. 30
Limitations of our study include that this is a retrospective single-center study. Moreover, the information about the reasons for sedation medications are limited because of the retrospective nature of the study and difficulty obtaining this information from chart review, including the use of propofol for agitation out of sleep versus agitation during wakefulness, the latter of which could be due to non-sleep disturbances. We also have limited information on any preexisting history of a circadian rhythm disorder, prior to the onset of NMDA receptor encephalitis. We also have limited information on whether circadian rhythm sleep-wake disturbance occurred in these patients, although we used ICU admission as a surrogate marker. Another limitation is the sample size; nevertheless, this is a large pediatric cohort for a single institution. Other limitations include missing data such as unknown sleep problems at onset, but we also included analyses presuming that those patients did not have sleep problems. We also did not include which sleep medications were most effective in these patients, as improvements in sleep could be confounded by other treatments, including concomitant immunotherapy. We did find that most patients were no longer on sleep medications at 1 year, and the majority on sleep medications at 1 year still had sleep complaints, which suggests that sleep improvements are independent of treatment but may be confounded by immunotherapies. Future prospective studies including multiple centers and sleep questionnaires could help further expand on sleep disturbances in NMDA receptor encephalitis. Questionnaires such as the Child Sleep Habits Questionnaire 31 could be administered from the onset of NMDA receptor encephalitis and during follow-up to monitor sleep status.
Conclusion
Sleep dysfunction also occurs in children with NMDA receptor encephalitis just as in adults with NMDA receptor encephalitis. Moreover, sleep problems can persist at 1 year from onset in a subset of patients, which may correlate with 1-year outcomes. Further investigation including multicenter prospective studies could help understand aberrations of sleep in NMDA receptor encephalitis to advance understanding of normal sleep physiology, and whether persistent sleep problems are a sign of ongoing disease severity versus contributing to recovery from NMDA receptor encephalitis.
Footnotes
Author Contributions
GG: conception or design of the work; and acquisition, analysis, or interpretation of data for the work; Drafting the work or revising it critically for important intellectual content; Final approval of the version to be published; Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. MM, LL, LB, RH: acquisition, analysis, or interpretation of data for the work; Drafting the work or revising it critically for important intellectual content; Final approval of the version to be published; Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Declaration of Conflicting Interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: GG receives salary support from the Centers for Disease Control and Prevention for surveillance for acute flaccid myelitis and from Novartis as a site for a pediatric multiple sclerosis clinical trial. The other authors have no conflicts of interest to disclose.
Ethical Approval
Institutional review board approval was obtained from Children's Healthcare of Atlanta IRB (STUDY00003330) and consent was waived for this retrospective observational cohort study of pediatric NMDA receptor encephalitis patients.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported in part by the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number UL1TR002378 and KL2TR002381.
