Biphasic sleep is a feature of sleep disturbance in SYNGAP1-RD

Abstract

Study Objectives:

Sleep problems are common in children with SYNGAP1-Related Disorder (SYNGAP1-RD). The use of devices that objectively estimate sleep are complicated by co-occurring sensory disorders in this population. We examined the feasibility and validity of wrist actigraphy to examine sleep and rest-activity rhythm (RAR).

Methods:

Data from five children with SYNGAP1-RD and 42 typically developing children were analyzed. All children were asked to wear the Actiwatch-2 for 14 continuous days and caregivers were asked to complete a sleep diary and the Children Sleep Health Questionnaire (CSHQ). Parametric (alpha, beta, acrophase, amplitude, up/down mesor, mesor), nonparametric (intradaily variability, interdaily stability, M10/L5 and relative amplitude), and sleep metrics—total sleep time (TST), wake after sleep onset (WASO), fragmentation index (restlessness throughout the rest period), the time of sleep bout onset, the number, and duration of sleep bouts, and the number and duration of wake bouts—were analyzed.

Results:

In children with SYNGAP1-RD, sleep bouts were distributed across the day and night and were consistent with a higher alpha. Fewer sleep bouts occurred in the early morning aligning with parent report of early morning arousal. Although there was no difference in average TST, children with SYNGAP1-RD, had fewer sleep and wake bouts that were longer in duration. Additionally, they had greater WASO and higher fragmentation index, the latter was a strong predictor of subjects with SYNGAP1-RD.

Conclusion:

Actigraphy provides an objective measure of sleep that aligns with RAR, is consistent with parent report, and provides an ethological approach to monitoring sleep in children with SYNGAP1-RD.

Introduction

SYNGAP1-related disorder (SYNGAP1-RD) is characterized by intellectual disability (ID), developmental delay, altered sensitivity to sensory stimuli, epilepsy, sleep difficulties, autism spectrum disorder (ASD), and challenging behaviors^1–5. It is causally linked to heterozygous pathogenic and likely-pathogenic variants in the SYNGAP1 gene that encodes a synaptic Ras-GTPase activating protein^6,7. This protein is a negative regulator of Ras signaling and is essential for synaptic plasticity, learning, and memory^6,7. In addition to forebrain regions, SynGAP1 is expressed in the murine suprachiasmatic nucleus and is implicated in the regulation of circadian-gated locomotor activity and light entrainment in rodents⁸.

Sleep is a biological process that is critical for physiologic functioning^9–11 and is regulated by homeostatic sleep pressure and circadian rhythms^12,13. Poor sleep can negatively impact an individual’s quality of life and that of their caregivers^9,14,15. Sleep challenges are common in typically developing (TD) children with an estimated prevalence of 25–40%¹⁶, and are 2–3 times more prevalent and severe in children with neurodevelopmental disorders, such as autism spectrum disorder (ASD) and epilepsy^15,17,18. Although 60% of children with SYNGAP1-RD have a coexisting ASD diagnosis, sleep difficulties in this disorder appear to be independent of ASD status¹⁹. Factors contributing to sleep disturbance in SYNGAP1-RD are not well understood; however, sleep anxiety, parasomnias, and difficulty winding down are known predictors of poor sleep quality in SYNGAP1-RD¹⁹.

Polysomnography (PSG) is the gold standard for sleep assessment²⁰; however, a notable constraint of PSG in children is its reliance on a specialized sleep laboratory and multiple body-worn sensors²¹. Children with neurodevelopmental disorders are often unable to tolerate the demands of in-lab PSG²², thus increasing the need for more accessible tools to characterize sleep in this population. Wrist-worn accelerometers (i.e., “wrist-actigraphy”) have emerged as an accessible alternative to evaluate sleep/wake cycles in children with neurodevelopmental disorders^23,24. Actigraphy is a non-invasive, objective method that captures movement-based measures of sleep and rest activity rhythms (RAR) in the natural environment²⁵, and has been validated against polysomnography to estimate sleep/wake patterns^23,26–29. Using the Actiwatch-2 (Philips Respironics, Bend, OR) and a community control sample of typically developed peers³⁰ we previously showed that participants with SYNGAP1-RD had shorter latency to sleep onset, shorter duration of wakefulness after sleep onset, more wake bouts and shorter wake bout duration compared to SYNGAP1-RD¹⁹. The finding of shorter latency diverged from parent-reported outcomes on the CSHQ, as sleep onset delay was more common in SYNGAP1-RD than TD controls⁴. Interestingly, participants with SYNGAP1-RD exhibited a higher number of wake bouts with shorter duration of wakes bouts, while this is consistent with increased nighttime awakenings on the CSHQ it also suggests that sleep in SYNGAP1-RD might be more fragmented. This hypothesis prompted a more granular analysis of the actigraphy data, which is the focus of this manuscript. Specifically, we examined the validity of parametric and nonparametric approaches to assess sleep quality, RAR and circadian and ultradian rhythmicity in SYNGAP1-RD compared to control subjects.

Methods

Recruitment

The study protocols were approved by the Institutional Review Board of the Johns Hopkins University School of Medicine. Adult caregivers of TD children and children with pathogenic or likely-pathogenic variants in the SYNGAP1 gene provided informed consent to participate in this study. Participants were recruited between April 2019 and August 2023 from the Center for Synaptic Disorders (CSD) at the Kennedy Krieger Institute, Baltimore, Maryland, USA. TD children were recruited by the Center for Neurodevelopmental and Imaging Research (CNIR) at the Kennedy Krieger Institute in Baltimore Maryland, USA, as part of an ongoing research study of sleep and circadian rhythms in children with and without ASD. Children between the ages of 3–13 years old were included in the study and excluded if they had a variant of uncertain significance in the SYNGAP1 gene or lacked sufficient actigraphy data.

Questionnaires

Caregivers provided demographic information, clinical diagnoses and the names of prescribed medications and supplements used by their child. Caregivers also completed the Children’s Sleep Habit Questionnaire (CSHQ), which contains 45 items rated on a three-point Likert scale³¹. Responses inform eight subscales: Bedtime Resistance, Sleep Onset Delay, Sleep Duration, Sleep Anxiety, Night Awakenings, Parasomnias, Sleep Disordered Breathing, and Daytime Sleepiness. The subscale scores are combined and produce a Total Sleep Disturbance score. Larger scores indicate greater sleep disturbance and a score ≥41 is a marker of clinically significant sleep difficulties³¹.

Actigraphy data collection and analysis of sleep metrics and Rest Activity Rhythms

The Actiwatch-2 (Philips Respironics, Bend, Oregon) is a solid-state piezo-electric accelerometer that detects motion at a 32Hz sampling rate from a range of 0.5–2g³². Participants were instructed to wear the Actiwatch-2 device on their non-dominant wrist for 14 consecutive days. The data were summed into 30 second epochs in the CNIR dataset and 60 second epochs in the CSD dataset. All 30-second-epoch data was standardized to 60 second epochs by activity summation.

We developed an R script that automated the processing pipeline of minute-level data for actigraphy wear-time validation. Valid days were defined as days with a minimum of 20 hours of wear time, consistent with the National Sleep Research Resource data processing standards,³³ and were synchronized to start at midnight (00:00) of the first valid day. Participants with < 3 valid days were excluded from analysis, and periods of non-wear were defined as 90 or more consecutive minutes of zero activity³⁴. During the automated removal process, periods of non-wear were excluded by employing a Run Length Encoding function to identify contiguous bouts of zero activity. To ensure adequate data quality, days with <1200 minutes were excluded from parametric analyses and days with less than 1440 minutes were removed from nonparametric analyses. The automated removal process was complemented with visual inspection of the actograms generated by Actiware software (Version 6.3.0, Philips Respironics) for secondary validation. Caregivers or typically developing peers provided a sleep log to document non-wear, time in/out of bed, time of lights on/off, and the start and end of naps. In cases of prolonged inactivity, sleep logs were reviewed by AD to determine if the prolonged epoch was associated with non-wear.

Sleep metrics (e.g., total sleep time—total duration of sleep in the rest period, sleep efficiency¸—the percentage of the rest period spent asleep, WASO—duration of wakefulness after the onset of sleep, fragmentation index— an index of restlessness throughout the rest period, #sleep bouts—the number of sleep episodes during the rest period, sleep bout duration—duration of sleep bouts during the rest period, #wake bouts—the number of awakenings within the rest period, and average wake bout duration—duration of awakenings within the rest period) from a defined sleep period were determined using Actiware. To evaluate sleep across a 24-hour day we used a method agnostic to sleep period, we defined sleep bouts as epochs of zero activity lasting ≥ 4 minutes (after assigning non-wear bouts). This approach, termed IS4 (“inactive state 4 minutes) successfully captures relative hypersomnia in adults with epilepsy with or without intellectual disability^37,38 We identified the time of day that a sleep bout began and calculated the total sleep time, and the number and duration of sleep bouts.

We performed parametric and nonparametric analyses of RAR using two open-source R-studio packages (https://github.com/JessLGraves/RAR and https://github.com/junruidi/ActCR). Parametric measures of RAR were extracted from a sigmoidal-transformed extended cosinor model to account for the more square-like features of human RAR³⁵. Alpha (α) determines the relative width of the rest to active phase where larger α values indicate a longer rest period (wider trough) while smaller α values indicate a longer active period (wider peak). The beta parameter is the sigmoid steepness or rate of change when transitioning from rest to activity, such that larger beta values indicate a faster transition between rest and peak activity. The amplitude is the difference between the largest and smallest values of the RAR; higher values indicate more activity during the day and less activity at night. Acrophase is defined as the time of day when peak activity occurs, later time of peak activity may indicate delayed RAR. The midline estimating statistic of rhythm (MESOR) parameter is a rhythm-adjusted average activity level; higher values indicate a more robust average activity. The up-MESOR parameter is the clock time of the transition from sedentary to active where activity level surges from below MESOR to above MESOR while the down-MESOR parameter is the clock time where activity levels decline from above MESOR to below MESOR. The F-statistic represents the robustness of the RAR, with larger values indicating greater rhythmicity (Figure 1)^35–39.

Figure 1.

Parameters derived from the sigmoidally transformed extended cosine model. Alpha (relative width of rest to active phase) and beta are specific shape-defining parameters. Acrophase, up-MESOR, and down-MESOR are reported as clock times. Amplitude and MESOR are reported as activity counts. Created using Biorender.com

Nonparametric analyses includes measurements of interdaily stability (IS; day-to-day rhythm consistency), intradaily variability (IV; rhythm fragmentation, frequency of transitions between rest and activity states within a 24-hour period), relative amplitude (RA; rhythm strength), most active consecutive 10h period (M10), and the least active consecutive 5-h period (L5)⁴⁰. IV values vary between 0 and 2, where 0 reflects a perfect sine wave and 2 indicates Gaussian noise (high fragmentation), IV values closer to 0 indicate less RAR fragmentation. IS values closer to 1 indicate strong alignment with light/dark, while 0 indicates Gaussian noise (random pattern)⁴¹.

Spectral decomposition of the cleaned minute-level actigraphy signal was performed using a Lomb–Scargle periodogram implemented in MATLAB (R2024a)³⁷. For each participant, the validated activity vector was converted to a numeric array and analyzed at a sampling frequency of 1/60 Hz (one sample per minute). A predefined frequency grid spanning ~2–30-hour periods was applied resulting in the normalized spectral density representing the contribution of each period length to overall activity variance. The spectral output for each participant was exported to Graphad Prism 10.0 (Graphpad, La Jolla, California), where individual spectral profiles were compiled and averaged for group visualization.

Statistical analysis

Graphical representation and inferential analyses were performed in Graphad Prism 10.0 and RStudio (Posit Software, Boston, Massachusetts). Between group comparison of CSHQ total and subscale scores, parametric and non-parametric measures were performed using the Mann-Whitney U (MWU) test. Chi-square (Fisher exact) was used to compare demographics and rates of clinically significant CSHQ scores. To visualize the average RAR for each population, activity counts were segmented into 1m epochs and averaged for all days of valid wear time. Curves were smoothed by a second-order polynomial and averaging 200 neighboring points. Cumulative and relative frequencies of sleep bout distributions were compared by Kolmogorov-Smirnov test. Statistically significance was defined as p value of <0.05 for all analyses.

Results

Characterization of SYNGAP1-RD and Controls

Actigraphy and CSHQ data were collected from 53 participants. Five children between the ages of 3–10 years with pathogenic or likely-pathogenic variants in SYNGAP1 were included in the analysis and 42 TD children between the ages of 6–12 years were included as controls. One child with a variant of uncertain significance for SYNGAP1 was excluded, and five controls were excluded due to low-quality actigraphy data. There was no significant difference in mean ages between groups (p = 0.339). Males comprised 69% of the control group and 80% of SYNGAP1-RD group (Table 1). All children with SYNGAP1-RD used medications to assist with sleep (melatonin, clonidine, or valerian root), 60% had epilepsy and reported using antiseizure medications (cannabidiol and/or lamotrigine), and 40% used amphetamine salts or sertraline to regulate mood/behavior. The control participants reported “no” to using prescription or over the counter medication.

Table 1.

Demographic characteristics of participants with SYNGAP1-RD and controls.

	SYNGAP1-RD	Controls	p-value
Sample Characteristics	n = 5	n = 42
Age, years. mean (SD)	8.00y (1.542)	9.86y (0.259)	0.339
Range (3y-12y)
Biological Sex (%, n)
Male	80% (4)	69.1% (29)	>0.999
Female	20% (1)	29.5% (13)
Sleep Aid Use	100% (5)	0% (0)	<0.0001
Epilepsy	60%(3)	0%(0)	0.0006

p-value Chi-square (Fisher exact)

We then compared CSHQ scores and note that clinically significant CSHQ scores (total score >41) were observed in all individuals with SYNGAP1-RD compared to 30.95% of control subjects, p = 0.006. Total scores were significantly elevated in SYNGAP1-RD compared to controls p< 0.001, and subscale scores were significantly elevated p< 0.01 in all domains except bedtime resistance and daytime sleepiness (Table 2).

Table 2.

CSHQ total and subscale scores.

	SYNGAP1-RD	Controls	p-value
CSHQ Variables	n = 5
Clinically Significant (≥ 41)	100% (n=5)	30.95% (n=13)	0.006
Total Score M(SEM)	56 (6.173)	38.29 (0.695)	<0.001
Bedtime Resistance	8.8 (1.655)	6.857 (0.235)	0.135
Sleep Onset Delay	2.4(0. 4)	1.286 (0.098)	0.005
Sleep Duration	6.0 (0.949)	3.357 (0.127)	<0.001
Sleep Anxiety	7.2 (0.583)	4.690 (0.203)	<0.001
Night Awakenings	6 (1.000)	3.310 (0.093)	0.002
Parasomnias	12.40 (1.749)	7.738 (0.160)	<0.001
Sleep Disordered Breathing	4.0 (0.447)	3.024 (0.063)	0.004
Daytime Sleepiness	12.60 (1.720)	10.38 (0.378)	0.092

CSHQ scores were clinically significant and significantly elevated for children with SYNGAP1-RD in almost all subscales. p-values (Mann Whitney test)

RAR Characteristics of SYNGAP1-RD and Controls

The mean wear time of the Actiwatch-2 did not differ between the SYNGAP1-RD and controls, with average durations of 12.20 and 12.17 days, respectively. Parametric analysis revealed that children with SYNGAP1-RD exhibited greater alpha, implying a longer relative rest period (p = 0.006), and a lower amplitude, suggesting less activity during the day and more at night (p = 0.045). Acrophase and up-MESOR in SYNGAP1-RD were delayed by 23 and 52 minutes respectively (Figure 2a), but neither were significantly different from controls (Table 3). We next employed nonparametric analyses to examine stability and rhythmicity of the RAR. SYNGAP1-RD showed lower within-day fragmentation of the rhythm (IV) and greater across-day similarity (IS), but neither were statistically significant. The average activity during the most active 10 hours (M10) and least active 5 hours (L5) were not significantly different between cohorts, neither were the relative amplitudes (Table 4). Spectral decomposition analysis identified a dominant circadian peak for both groups at 24-hours and ultradian peaks at 6, 8 and 12-hours. A slight leftward shift was noted at 8,12 and 24-hours in SYNGAP1-RD (Figure 2b).

Figure 2.

a). Average actogram from patients with SYNGAP1-RD and Controls, showing delayed up Mesor and lower amplitude. b). Spectral decomposition of raw actigraphy data demonstrating similar major peak (circadian peak ~24 hours) and ultradian peaks at ~ 6, 8 and 12 hours (inset).

Table 3.

Parametric rest-activity-rhythm metrics for SYNGAP1-RD and controls.

	SYNGAP1-RD	Controls	p-value
Parametric Variables	Mean (±SEM)	Mean (±SEM)
Alpha	−0.259 (0.022)	−0.345 (0.011)	0.006
Beta	16.500 (5.166)	11.60 (0.758)	0.429
Acrophase Clock Time (h:min)	15:10 (0:35)	14:47 (0:08)	0.449
Amplitude	106.2 (16.45)	158.3 (9.392)	0.045
MESOR	23.52 (2.113)	22.98 (0.897)	0.802
Up MESOR Clock Time	8:17 (0:35)	7:25 (0:08)	0.147
Down MESOR Clock Time	22:18 (0:36)	22:08 (0:10)	0.752
F-Statistic	6256 (1134)	7783 (466.2)	0.271

p values (Mann-Whitney test).

Table 4.

Nonparametric rest-activity-rhythm metrics for SYNGAP1-RD and controls.

	SYNGAP1-RD	Controls	p-value
Nonparametric Variables	Mean (±SEM)	Mean (±SEM)
IV	0.638 (0.087)	0.820 (0.026)	0.081
IS	0.619 (0.028)	0.568 (0.016)	0.242
M10	542.3 (59.43)	587.5 (21.51)	0.470
L5	16.02 (2.168)	14.13 (1.173)	0.287
RA	0.9403 (0.010)	0.95 (0.004)	0.271

IV (intradaily variability) values closer to 0 (perfect sine wave) than 2 (Gaussian noise) indicate less RAR fragmentation while IS (interdaily stability) values closer to 1 (perfect stability) than 0 (Gaussian noise) indicate greater RAR light/dark stability. M10 (most active consecutive 10h period), L5 (least active consecutive 5h period), RA (relative amplitude. Mann-Whitney test p-values.

Sleep Characterization

We first analyzed sleep/wake epochs of the nighttime sleep period using Actiware and found no significant differences between controls and SYNGAP1-RD in average total sleep time (460 ± 6.0 minutes vs. 483.5 ± 10.21 minutes; p = 0.20) or sleep efficiency (84.72% ± 0.8 vs. 85.84% ± 1.25; p = 0.85) (Figure 3). However, compared to controls, children with SYNGAP1-RD experienced shorter sleep onset latency (2.73 ± 1.67 vs. 14.54 ± 2.5 minutes; p = 0.002), longer sleep bout duration (20.98 ± 2.08 vs. 16.32 ± 1.98 minutes; p = 0.01), but fewer sleep bouts (26.72 ± 2.44 vs. 37.72 ± 1.33, p = 0.005), longer duration of wake bouts (2.47±0.10 vs 1.29 ± 0.05, p<0.001) and fewer wake bouts (25.90 ± 2.44 vs. 37.12 ± 1.34, p = 0.005). Interestingly, WASO and fragmentation index were higher in SYNGAP1-RD compared to controls p = 0.007 and p <0.001 respectively.

Figure 3.

Sleep metrics analyzed using Actiware (v.6.3.0 Philips Respironics). *p< .05, **p< .01, ***p<.001

Since sleep in children with NDD have been reported to occur during the daytime we used an approach that is agnostic to the use of sleep diaries to denote sleep periods. We putatively defined sleep bouts as continuous epochs of zero activity lasting ≥ 4 minutes (IS4) to determine bout duration, timing, and total daily sleep time³⁷. We first examined the validity of this approach by measuring total sleep time from the 8:00PM to 8:00AM using IS4 and Actiware for both controls and SYNGAP1-RD cohorts. Using simple linear regression analyses we then examined the relationship between the total sleep time from IS4 and Actiware for each cohort and determined that these was no significant difference in the slopes F=0.12. DFn =1, DFd = 43, p = 0.73 (Figure 4a). Additionally, although the average total sleep time was not significantly different in SYNGAP1-RD compared to controls, p = 0.2; there are fewer sleep bouts in SYNGAP1-RD (p =0.005) and the sleep bout duration was longer (p = 0.01) (Figure 4b–d). This finding was consistently seen across Actiware and the IS4 method (Figure 4 e–g).

Figure 4.

Sleep estimates. a). Simple linear regression analysis of IS4 vs Actiware. b-d). Sleep estimates from Actiware analysis. e-g). Sleep estimates from IS4 analysis. h-j). Frequency distribution of sleep bout onset and sleep bout duration assessed across 24-hour period. TD-typically developed controls, SG-SYNGAP1-RD. * p< .05, **p< .01 (Mann Whitney Test)

We next used the IS4 approach to query the time of sleep bout onset and duration over the 24-hours and showed that participants with SYNGAP1-RD had a relative decline in sleep bouts with onset between 04:00 and 08:00, pointing to early morning arousals, a small increase in sleep bouts with onset between 11:00 and 16:00 hours, and a more substantial increase and earlier onset of sleep bouts starting at 18:00 hours (Figure 4h). The controls had a well-consolidated sleep pattern with sleep onset and offset at approximately 20:00 hours and 09:00 hours, respectively (Figure 4h). Approximately 73% of sleep bouts occurred between midnight to 0800 for controls while 55% of sleep occurs during the same period for SYNGAP1-RD (Figure 4i). Sleep bouts are generally less than 20 mins in length (Figure 4c), with 75% of sleep bouts being less than 17 minutes in SYNGAP1-RD and less than 12 minutes in controls (Figure 4j). These findings were generally consistent with the duration of sleep bouts measured by Actiware. The frequency distributions for sleep bout onset and sleep bout duration were significantly different (p< 0.001) between SYNGAP1-RD and TD controls.

Lastly, we explored the relationship of SYNGAP1-RD with RAR and sleep metrics. We dichotomously assigned each participant to one of two groups, SYNGAP1-RD positive or negative, and performed two logistic regression models using sex, age, acrophase, beta, alpha, L5, M10, IS, IV, F or sex, age, fragmentation index, duration of sleep bouts, number of sleep bouts, total sleep time, WASO and sleep onset latency with SYNGAP1-RD positive status as the outcome. We show in figure 5A that individuals with higher beta, delayed acrophase, wider alpha or lower M10 were more likely to have SYNGAP1-RD, and in figure 5B that individuals with higher fragmentation index, fewer sleep bouts or longer sleep time were more likely to have SYNGAP1-RD. Participants with a higher fragmentation index, i.e. more restlessness during the rest period have a 1.7x higher odds of having SYNGAP1-RD, while those with a high beta have 2x higher odds of having SYNGAP1-RD.

Figure 5.

Odds ratio and 95% confidence intervals (logistic regression) that determines actigraphic rhythms (a) and sleep variables (b) that predict genotype. *p< .05, **p< .01, ***p< .001

Discussion

In-lab PSG, the gold standard for sleep assessments, provides insight into the physiologic characteristics of sleep over 1–2 nights. It uses an electroencephalogram to determine sleep stages (N1, N2, N3, REM) and oxygen and airflow to detect sleep related breathing disorders. It is limited by its inability to capture longitudinal data on sleep-wake patterns and due to the many sensors, it is not well tolerated by children with neurodevelopmental disorders. Unlike PSG, actigraphy captures movement that is used to estimate sleep timing and duration. It is non-invasive and is used in the natural environment to assess sleep-wake patterns over days to months and can be used to distinguish insomnia from circadian rhythms disorders. For these reasons this study utilized wrist actigraphy to characterize sleep and RAR in children with SYNGAP1-RD, a disorder with a high prevalence of subjective and objective sleep disturbance.

SYNGAP1-RD and control cohorts were comparable in age and sex, however the use of medications to aid with sleep were endorsed by all children with SYNGAP1- RD and by none of the typically developed controls. Consistent with current literature, our CSHQ analysis showed that all children with SYNGAP1- RD had clinically significant sleep scores, compared to 30% of controls⁴ and subscale scores were significantly elevated in all domains except bedtime resistance and daytime sleepiness, which although elevated did not meet statistical significance. Additionally, there was no between group difference in mean wear time of the Actiwatch2, indicating that the device is well tolerated in children with altered sensory sensitivities. This finding is consistent with prior reports of Actiwatch use in children with autism and other neurodevelopmental disorders^23,24,30,42.

The results of the current study are largely consistent with our prior study, in which no between-group differences were observed in average total sleep time, sleep efficiency, or sleep latency. In contrast, the present analysis reveals fewer wake bouts and longer wake-bout duration relative to controls¹⁹. This difference likely reflects the use of an alternative comparator group.

Analysis of nighttime sleep in children with SYNGAP1-RD revealed shorter latency to sleep onset but longer sleep bout duration, fewer sleep bouts, longer duration of wake bouts and fewer wake bouts. Taken together in the context of the relative frequency plots these data suggests that there are stable boundaries between the sleep and wake states across the night. Interestingly, WASO and fragmentation index -- restlessness during the overnight period allocated for sleep -- were higher in SYNGAP1-RD compared to controls. This finding of increased restlessness is consistent with the findings of lower amplitude in RAR (less activity during the day, more activity at nights). Additionally, SYNGAP1-RD had onset of sleep bouts that were distributed across the day and night, a finding that is consistent with larger alpha, i.e., a longer relative rest phase. These findings agree with the wider rest phase seen in adult with epilepsy and co-occurring psychiatric diagnoses³⁷. The relative frequency plots also reveal a decline in sleep bouts in the early morning which is consistent with the anecdotal report of early morning arousal by parents of children with SYNGAP1-RD and supports the model that higher fragmentation index is a predictor of SYNGAP1-RD.

Melatonin and antiseizure agents are known to modify sleep behavior; however, the extent to which these agents modify sleep and circadian patterns in SYNGAP1-RD is not known. The small sample size of the SYNGAP1-RD cohort is a limitation of this study, however this is the first study to our knowledge, to utilize parametric and non-parametric analyses of rest activity rhythms in SYNGAP1-RD.

In conclusion, our study shows that actigraphic analysis of sleep is consistent with parent report of early morning arousal and restless sleep. It aligns with RAR metrics in SYNGAP1-RD and shows that sleep in participants with SYNGAP1-RD is biphasic with a more prominent secondary sleep period. Together, these measures may provide objective endpoints that complement survey-based assessments of sleep, behavior, and quality of life.

Highlights.

• Actigraphy is a feasible and valid tool to assess sleep/wake cycles in children with SYNGAP1-RD.

• Sleep is biphasic in children with SYNGAP1-RD with a more prominent secondary sleep period

• Reduced early morning sleep bouts and higher fragmentation index align with parent reporting and provides an objective measure of sleep disturbance in this disorder.