Cortical grey matter changes, behavior and cognition in children with sleep disordered breathing

This paper investigated cortical thickness and volumetric changes in children to better understand the impact of obstructive sleep disordered breathing (SDB) on the neurodevelopment of specific regions of the brain. We also aimed to investigate how these changes were related to the behavioral and cognitive deficits observed in the condition. Neuroimaging, behavioral, and sleep data were obtained from 30 children (15 non‐snoring controls, 15 referred for assessment of SDB) aged 7 to 17 years. Gyral‐based regions of interest were identified using the Desikan‐Killiany atlas. Student's t‐tests were used to compare regions of interest between the controls and SDB groups. We found that the cortical thickness was significantly greater in the right caudal anterior cingulate and right cuneus regions and there were volumetric increases in the left caudal middle frontal, bilateral rostral anterior cingulate, left, right, and bilateral caudate brain regions in children with SDB compared with controls. Neither cortical thickness nor volumetric changes were associated with behavioral or cognitive measures. The findings of this study indicate disruptions to neural developmental processes occurring in structural regions of the brain; however, these changes appear unrelated to behavioural or cognitive outcomes.


| INTRODUCTION
Sleep disordered breathing (SDB) in children is very common (Bixler et al., 2009) and describes a spectrum of respiratory-related sleep disturbances, characterised by intermittent obstruction of the upper airway during sleep, despite diaphragmatic effort.The severity of the condition ranges from primary snoring to obstructive sleep apnea (OSA) (Kheirandish-Gozal & Gozal, 2012).Severe OSA results in recurrent hypoxic events, elevated CO 2 levels and/or sleep fragmentation (Kheirandish-Gozal & Gozal, 2012).Primary snoring is not associated with gas exchange abnormalities or sleep disruption.It is well established that children with SDB are at increased risk of cognitive and behavioural difficulties (Cardoso et al., 2018).However, these have been shown to be unrelated to disease severity (Bourke et al., 2011a(Bourke et al., , 2011b)).As behavioural and cognitive disturbances have the potential to impact school performance and functioning, there is an increased need to understand how these difficulties are associated with sleep disordered breathing.
Brain injury as a result of intermittent hypoxic events is one proposed mechanism to explain the cognitive and behavioural difficulties in paediatric SDB (Yang et al., 2022).Volumetric analysis is a simple but effective means of measuring whole-and regional brain volumes in patients who are thought to have experienced volume loss due to brain injury.Several neuroimaging studies have utilised volumetric analyses to identify regional alterations occurring in paediatric SDB (Chan et al., 2014;Macey et al., 2018;Musso et al., 2020;Na et al., 2021;Philby et al., 2017).These studies have presented a widespread pattern of cortical injury with reports of reduced grey matter in the frontal (Chan et al., 2014;Musso et al., 2020;Philby et al., 2017), temporal, (Chan et al., 2014;Na et al., 2021;Philby et al., 2017), occipital (Musso et al., 2020), and parietal cortices (Philby et al., 2017) as well as subcortical regions such as the brain stem (Philby et al., 2017), thalamus, ventral posterior nucleus, and medial dorsal nucleus (Lv et al., 2017).Additionally, individuals with moderate-severe OSA are particularly susceptible to neuropathological changes (Chan et al., 2014).
However, volumetric analysis is a gross measure that can fail to detect subtle changes occurring on the surface of the brain.The advent of high resolution T1-weighted MRI and diffusion tensor imaging (DTI) allows for more specific investigations into how the brain is impacted by sleep disordered breathing.DTI measures the movement of water molecules in white matter tracts and provides metrics of white matter integrity (Basser & Pierpaoli, 2011) which can provide more granular information about neuropathology than volumetric analyses.A previous study by our group utilised DTI measures and detected mean diffusivity was reduced (indicating acute changes) in some specific regions of the brain of children with SDB and was increased (indicating chronic damage) in others (Horne et al., 2018).
Although these findings indicate potential white matter integrity abnormalities occurring as a result of hypoxic or ischaemic mechanisms, further investigations are required to elucidate cortical abnormalities in the condition to better understand the impact of sleep disordered breathing on neuropathology and to provide biomarkers for neuropathological progression.Measuring the thickness of the cortex is one method that is sensitive to changes in cortical regions, and has been utilised in a range of neurodevelopmental and neurodegenerative disorders (Blanken et al., 2015;Lehmann et al., 2011;Pereira et al., 2012;Tan et al., 2022).Throughout a lifetime, the human brain experiences fluctuations in cortical thickening and thinning (Fuhrmann et al., 2022).These occur as a result of normal developmental processes such as neural development, synaptic pruning, reduced blood flow etc.Until recently, few studies have assessed cortical thickness changes in paediatric subjects with SDB.Those which have, report significant cortical thinning in numerous brain regions, including the rostral middle frontal, caudal middle frontal, lateral occipital gyri (Musso et al., 2020), right superior parietal (Lee et al., 2022), superior frontal, and ventral medial prefrontal (Macey et al., 2018).Significant cortical thickening has also been observed in the precentral and central gyrus, medial prefrontal cortex, the mid and anterior insular, the posterior and subgenu of the anterior cingulate cortex as well as the medial temporal lobe areas (Macey et al., 2018).
Whilst these studies have been informative to our understanding of brain pathology in SDB, to date no studies have investigated the relationship between cortical thickness and behaviour and cognition in paediatric SDB.Such knowledge is essential to provide clinicians with a better understanding of the impact of sleep disordered breathing on the brain, and how these are related to the clinical symptoms associated with it.
This study aimed to provide more detailed evidence to the growing body of literature about the impact of sleep disordered breathing on regional cortical thickness, volume and behaviour, and estimates of IQ in children.Based on the existing literature, we hypothesised that regional cortical thickness and volumetric abnormalities would be present in many cortical areas, particularly those relating to behaviour and cognition.We also hypothesised that these changes would be related to the severity of disease, and would be correlated with measures of behaviour and cognition.

| Overnight polysomnography
Overnight polysomnography (PSG) was conducted using standard paediatric criteria (Berry et al., 2012).A minimum 4 h of sleep during the PSG was required to diagnose SDB severity.Prior to the PSG, children were weighed and measured and body mass index (BMI) z-scores were calculated (Ogden et al., 2002).Electrophysiological signals were recorded with a commercially available PSG system (E-series, Compumedics, Melbourne, Australia).In brief, electroencephalogram (EEG), Respiratory parameters included SpO 2 nadir (i.e., the lowest oxygen saturation value overnight), average SpO 2 drop, % (the average percentage drop in oxygen saturation associated with respiratory events); SpO 2 < 90%, events/h (number of times a participant's oxygen saturation dropped to below 90% per hour of TST), SpO 2 ≥ 4% drop, events/h (the number of times a participant's oxygen saturation dropped by 4% or more per hour of TST).
The obstructive apnea-hypopnea (OAHI) was defined as the total number of obstructive apneas, mixed apneas, and obstructive hypopneas per hour of TST.The OSA group was defined as participants with OAHI >1 event/h and non-snoring controls all had an OAHI ≤1 event/h and no history of snoring or snoring on the overnight PSG.The respiratory disturbance index (total number of apneas, hypopneas, and respiratory effort-related arousals, divided by the sleep time in hours) and central respiratory disturbance index (CRDI) were also calculated.

| Cognitive and behavioural assessment
Within 2 weeks of the PSG study, the children completed the following cognitive and behavioural assessments.

| Stanford-Binet intelligence scale -Fifth edition
The 10 core subtests from the Stanford-Binet Intelligence Scale -Fifth Edition were used to provide an estimate of nonverbal reasoning (performance IQ, PIQ), and vocabulary (verbal IQ, VIQ).Raw scores were converted to age-scaled standardised scores (M = 10, SD = 3) and were summed and converted to provide the standardised PIQ, VIQ and full-scale IQ (FSIQ, M = 100, SD = 15).Lower scores indicate worse performance on these measures.

| Child behaviour checklist (CBCL) -6-18, parent report
The CBCL is a subjective, standardised measure to assess behavioural and emotional problems in children and adolescents (Achenbach, 2001;Hensley, 1988;Nolan et al., 1996) occurring within the past 6 months.
Parents answer 118 questions rated on a Likert scale as either 0 (not true), 1 (somewhat or sometimes true), or 2 (very true or often true).The raw scores were calculated by totalling the items under each of eight subscales and then converting to normal-referenced T-scores ≥70, which are equivalent to scores above the 98th percentile, are considered to be of clinical concern (Marino et al., 2019).

| Behaviour rating inventory of executive function (BRIEF) -Parent report
The BRIEF parent report is a subjective, standardised measure of executive functioning in children and adolescents (Gioia et al., 2000).
It is an 86-item parent-reported questionnaire where parents respond to questions such as "when given three things to do, {your child} remembers only the first or last", on a series of Likert scales.The CBCL and BRIEF have been used widely and are well validated in a range of clinical and research settings (Biggs et al., 2015;Bourke et al., 2011a;Heubeck, 2000;Horne et al., 2018;Strauss et al., 2006;Toplak et al., 2008).It should be noted, however, that these measures can have significant bias due to the subjective nature of the parental report (Beebe, 2006;Biggs et al., 2011).

| MRI processing
T1-weighted images were processed to reconstruct cortical surfaces and to quantify measures of cortical and subcortical anatomy using FreeSurfer (version 7.3.1)(https://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferAnalysisPipelineOverview) (Dale et al., 1999;Fischl et al., 1999).The surface-based pipeline comprised several stages including volume registration, B1 bias field estimation, skull stripping, white matter segmentation, tessellation of the grey/white-matter boundary, automated topology correction and subcortical segmentation.Processed images were visually inspected for pial and grey-white matter boundaries, as well as any skull stripping errors.Edits were made to correct any errors manually.Cortical brain regions were provided by the Desikan-Killiany brain atlas, an automated, gyral-based labelling system consisting of 68 gyral-based regions (two hemispheres Â 34 maps) (Desikan et al., 2006).Subcortical brain regions were provided by the automated ASEG atlas using probabilistic information on the location of structures (Fischl et al., 2002).The cortical volume was calculated as the volume of all voxels within each region of interest.Cortical thickness was calculated as the distance between the inside (white/grey) and outside (grey/pial) boundaries of the cortex.

| Statistical analysis
Statistical analyses were completed using IBM SPSS V.24.0 for windows.Tests of normality and Levene's test for equality of variances were completed.Comparisons of demographics, sleep and respiratory characteristics, and demographics were compared between the SDB group and controls using independent samples Student's t-tests.Comparisons of cortical thickness and volume for each region of interest (ROI), estimates of IQ and behaviour were compared between the SDB group and controls using a series of ANCOVAs with age, sex, and BMI Z-Score as covariates.Bonferroni corrections were used to adjust for multiple comparisons.Initially, Pearson's correlations, adjusted for age and sex, were performed between cortical thickness and volume and behavioural (selected CBCL and BRIEF subscales) and cognitive (Stanford Binet index scores) measures.As we were interested in the impact of hypoxia and sleep fragmentation, we also performed Pearson's correlations between cortical measures and TST, sleep efficiency, SpO 2 nadir, average SpO 2 Drop %, <90% events/h, ≥4% drop events/h, total arousal index, WASO, respiratory arousal index, NREM1%, and OAHI.To reduce the number of analyses, correlations were only performed using cortical measures in those regions of interest that showed significant differences between control and SDB groups.To investigate the relationship between SD, the brain and cognition or behaviour, a mediation regression analysis was conducted with PSG measures of sleep fragmentation and hypoxaemia (i.e., SpO 2 nadir, Avg SpO 2 drop, SpO 2 < 90%h, SpO 2 > 4% drop, arousals I/h, respiratory arousals, total sleep time, sleep efficiency, wake after sleep onset (min), wake after sleep onset %, and OAHI as independent variables, with cognitive and behavioural measures as dependent variables, and cortical measures as mediating independent variables).
The results are presented as means and standard deviations (SD) for parametric data and median and interquartile range (IQR) for non-parametric data.A p value of <0.05 was considered statistically significant.

| Demographic and polysomnographic measures
There were no statistically significant differences in demographic characteristics between the two groups.There were no differences in sleep characteristics between the groups (see Table 1).By design the SDB group had a significantly higher OAHI and RDI, as well as a higher respiratory arousal percentage (see Table 1).

| Cognitive measures
Only 14 control and 12 children with SDB completed the cognitive assessment.There were no significant differences between the groups on any of the cognitive measures (PIQ, VIQ, FSIQ), although the SDB group tended to perform worse on all measures compared with controls (see Table 2).

| Behavioural measures
There were no significant differences between groups on any of the behavioural subtests from the CBCL and BRIEF, although parents of the SDB group tended to endorse higher levels of difficulty on all measures of behaviour (Table 2).

| Cortical measures and polysomnographic measures
There were no significant correlations between cortical thickness or volume with any sleep or respiratory measures related to sleep disruption or hypoxia (SpO 2 nadir, average SpO 2 drop %, <90% events/h, ≥4% drop events/h, arousal index, WASO, respiratory arousal index, OAHI).

| Cortical thickness (Table 3)
After controlling for age, sex and BMI Z-score, the SDB group demonstrated significantly greater cortical thickness in the right caudal anterior cingulate (F (1, 25) = 8.17, p = 0.008) (Figure 1).No statistically significant differences in cortical thickness were detected in any other regions of interest.As mentioned above, those with moderate-severe OSA are particularly susceptible to grey matter deficits (Chan et al., 2014).Therefore, we excluded the eight participants with OAHI <5 events/h and re-ran the above analyses, which elicited similar findings (data not presented).

| Mediation analysis
The results showed there was no significant mediating effect of structural brain change (i.e., cortical thickness, volume) on the relationship between any of the PSG measures and cognitive or behavioural measures.

| DISCUSSION
This study provides evidence of significant cortical thickness and volumetric changes occurring in children with SDB.These abnormalities were not correlated with polysomnographic measures, nor were they (Continues) correlated with cognitive or behavioural performance, suggesting that cortical surface changes in these regions are not the explanation for the clinical presentation seen in sleep disordered breathing.
We observed significant cortical thickening in the right caudal anterior cingulate equating to a 0.09 mm 2 ($3%) increase.The observed differences in cortical thickness in this study are consistent with current understandings of selective thickening occurring in specific cortical regions in children with OSA, although our results were notably in different regions to previous investigations (Macey et al., 2018).Cortical thickening may indicate abnormal neural development occurring in these brain regions.For instance, synaptic pruning is a process that occurs between early childhood and adulthood and is necessary for efficient brain function (Stiles & Jernigan, 2010;Toga et al., 2006).Previous studies have noted that cortical thinning of between 0.15 mm and 0.30 mm per year occurs as a result of synaptic pruning in children aged 5-11 years old (Stiles & Jernigan, 2010;Toga et al., 2006).Thus, the increases to cortical thickness we noted may indicate a disruption to synaptic pruning in these regions.It has been noted that the trajectory of cortical thickness in childhood is one of linear decline between the ages of 4.9 to 22.3 years (Ducharme et al., 2016).As such an observation of cortical thicken- explain some of the behavioural and cognitive deficits commonly observed in children with SDB (Beebe et al., 2004;Bourke et al., 2011aBourke et al., , 2011b;;Cardoso et al., 2018;Engleman & Joffe, 1999;Owens et al., 2000), however, the present study did not report a relationship between cortical thickening and behaviour and estimates of IQ.This was somewhat surprising, particularly given that the caudal anterior cingulate is anatomically and functionally responsible for executive functioning, attention, and emotional processing (Bush et al., 2000;Devinsky et al., 1995;Stevens et al., 2011), which are cognitive areas most impacted by OSA, particularly in children (Beebe et al., 2004;Landau et al., 2012;Owens et al., 2000).F I G U R E 1 Mean cortical thickness (mm 2 ) in the right hemisphere caudal anterior cingulate demonstrating significant differences between controls and SDB (Lee et al., 2022).*Significant at p < 0.05.
Similarly, volumetric analyses in this study revealed significant increases in the left caudal middle frontal, bilateral rostral anterior cingulate, left, right and bilateral caudate brain regions.Regional grey matter volume has previously been reported to be reduced in the frontal and cingulate regions in children with OSA compared with controls (Chan et al., 2014;Philby et al., 2017).However, our results of increased volume in these regions, as well as the bilateral caudate are novel.When comparing the general characteristics of our children with those from the above studies, our children were older, with a greater difference in BMI Z-score between controls and SDB group, and observed severity of OAHI were less severe than both (Chan et al., 2014;Philby et al., 2017).In the same cohort of children with SDB as the present study, Horne et al. (2018) reported widespread abnormalities to white matter integrity in numerous cortical regions.
When compared with the findings of this study, their results suggest white matter integrity is more vulnerable to sleep disordered breathing than the cortical surface.As with cortical thickening (described above), one explanation for increased volume could be a disruption in the synaptic pruning process related to OSA.However, another involves hypoxia induced neural inflammation (Kaur et al., 2013).This theory would posit that in paediatric SDB, injury or abnormal development affecting regional brain areas could be a result of intermittent hypoxia causing increased formation of reactive oxygen species and oxidative stress (Chandrakantan & Adler, 2019;Lavie, 2003).These could play a role in neuronal injury and cognitive dysfunction.However, if this were the case, one might expect a significant difference in in hypoxic exposure, measured bySpO 2 < 90% or SpO 2 ≥ 4% drop events/h, which we did not observe.Further studies are required to better elucidate this theory.
There were no significant correlations between cortical thickness or volume and the chosen behavioural and cognitive measures.This finding was unexpected as the regions in which we observed significant differences are thought to play a role in behavioural functioning (i.e., right caudal anterior cingulate, rostral anterior cingulate, and bilateral caudate), executive functioning (caudal anterior cingulate).
However, we note that in this sample we did not observe significant differences in CBCL and BRIEF subtest scores or cognitive index scales, despite the children with SDB having higher mean scores on behavioural subtests and lower scores on cognitive indices.Given the high degree of variance in these measures (see  likely that a sample size greater than 30 is required to detect a relationship between behavioural outcomes and subtle changes to the cortex.As mentioned above, we note that the severity of SDB was relatively low (mean OAHI, 3.9 event/h), ranging from mild to moderate (0.7-8.2 event/h) in our cohort, and we did not find a significant difference in SpO 2 between the groups.Therefore, it is possible that the children in our study did not meet a threshold of hypoxaemia or sleep disturbance to cause significant cognitive and behavioural impairment compared with controls.Additionally, we and others have previously shown that all severities of SDB affect behaviour in neurocognition (Bourke et al., 2011a(Bourke et al., , 2011b)).Alternatively, the cognitive and behavioural outcomes observed in SDB may be more susceptible to changes in white matter integrity, as noted previously (Horne et al., 2018), and further investigations into this area are warranted.
We also were unable to find significant correlations between cortical measures and polysomnographic measures (listed above).This finding is in contrast to the results found by Chan et al. (2014) in which grey matter volume deficits were observed in moderate-severe OSA (i.e., OAHI >5 events/h) only (Chan et al., 2014).In our cohort, the median OAHI was 1.8 events/h in the SDB group, thus providing a potential explanation for our inability to detect a relationship between SDB severity and cortical abnormalities.Even after excluding eight participants with an OAHI <5 events/h we were unable to detect any significant differences between groups.Previous studies have noted a significant relationship between the severity of SDB and white matter changes (Horne et al., 2018;Mei et al., 2021) and as mentioned above, it is possible that white matter integrity is more vulnerable to injury caused by sleep disordered breathing, with cortical changes occurring later.
This study found significant positive correlations between SDB severity and behavioural outcomes.The findings were such that increased severity of OAHI was associated with worse behavioural difficulties.It is unclear whether these results replicate the findings of several studies that note the severity of disease is correlated with behavioural impairment (Biggs et al., 2017;Bourke et al., 2011b;Tripuraneni et al., 2013).
We acknowledge certain limitations of this study.Firstly, the sample size may have been insufficient for detecting subtle changes occurring in grey matter regions.Due to many constraints (e.g., cost, time, obtaining parent consent), sample sizes in neuroimaging practical studies tend to be smaller compared with those in other fields (Szucs & Ioannidis, 2020).This can result in false-negative outcomes, and is a problem compounded by practical difficulties associated with research in paediatric populations.A power calculation indicated that our study may have lacked the statistical power required to detect significant changes in this instance.Nevertheless, our study sample size is comparable to many other neuroimaging studies (Szucs & Ioannidis, 2020), including a previous investigation by this group which found significant white matter integrity changes in paediatric SDB (Horne et al., 2018).Future investigations into this area may consider larger samples, or accessing existing datasets to provide greater statistical power.We also note that BMI-Z score was not able to be controlled for in our statistical analysis comparing mean volume and cortical thickness between the groups.There is evidence to suggest that cortical thickness and volume are impacted by BMI in children (Laurent et al., 2020;Solis-Urra et al., 2019), although the literature is not consistent in this regard (Sharkey et al., 2015).However, BMI Zscore was not statistically different between the groups assessed in our study, and therefore Student's t-tests were deemed appropriate for the purpose of this study.Furthermore, measures of neurobehavioural deficits have been criticised for a lack of sensitivity and specificity to PSG-defined SDB (Beebe, 2006;Goodwin et al., 2005).
Although the CBCL and BRIEF show strong technical adequacy (Hendrickson & McCrimmon, 2019;Nolan et al., 1996) and are more sensitive than single-item behavioural scales, it is still possible that variability in how a parent rates their child's behavioural difficulties limits their ability to detect statistical differences in group comparisons.This study examined the relationship between estimates of IQ, cortical brain measures, and SDB.However, we did not examine whether cortical brain measures and SDB were related to specific measures of cognitive domains likely to be impacted by hypoxaemia and sleep disruption such as attention, executive functioning, verbal and visual learning and memory due to the small sample size and increased risk of Type 1 error if we were to do so.Future studies would benefit from such analysis to elucidate the relationship between SDB and specific areas of cognition in a larger sample.
Although the present paper has shed further light on cortical abnormalities in children with SDB, this study and many others remain cross-sectional in design and provide no information about the time course of cortical alterations.Thus, longitudinal neuroimaging studies are required to track cortical changes and better understand neuropathological processes occurring in SDB.

| CONCLUSIONS
This study aimed to further existing knowledge about regional grey matter changes occurring in sleep disordered breathing, and their relationship to behavioural symptoms common in the disorder.These findings suggest that regional changes to grey matter (thickness and volume) may be a result of later-stage cortical changes which are secondary to white matter integrity changes.Furthermore, cognitive and behavioural symptoms observed in SDB may be closer related to white matter changes than grey matter alterations, however, certain limitations in this study (addressed above) do need to be taken into account when interpreting these results.Prospective longitudinal analyses are required to better understand the neuropathology of sleep disordered breathing.

2. 1
| Participants The Monash Health and Monash University Human Research Ethics Committees granted ethical approval (14024B).Written informed consent was obtained from parents and verbal assent from children before study onset.No monetary incentive was provided.Clinical participants were recruited from children attending the Melbourne Children's Sleep Centre for assessment of sleep disordered breathing.Non-snoring age-matched controls were recruited from the community via advertisements in newsletters.Exclusion criteria included conditions or medications known to affect sleep, breathing, or blood pressure, as well as neurodevelopmental conditions known to impact cognition and behaviour such as developmental delay, autism spectrum disorder, and attention deficit hyperactivity disorder.All children were well at the time of the study.Of the original 80 children recruited for a larger study, 38 participants (18 SDB, 20 controls) agreed to undergo MRI brain scans and the results of the DTI data have been published previously (Horne et al., 2018).Thirty had sufficient T1-weighted data for the analysis using FreeSurfer (see below).The final sample comprised 15 clinical SDB (7 males, 8 females; Age m = 12.4,SD = 2.4; BMI Z-score m = 1.1, SD = 1.2), and 15 non-snoring controls (8 males, 7 females; Age m = 12.4,SD = 2.5; BMI Z-score m = 0.3, SD = 0.7) aged between 7 and 17 years old.
left and right electrooculogram (EOG), submental electromyogram (EMG), and left and right anterior tibialis muscle EMG and electrocardiogram (ECG), thoracic and abdominal movement measurements, transcutaneous carbon dioxide (TcCO 2 ), nasal pressure, oronasal airflow, and oxygen saturation (SpO 2 ) were recorded.Paediatric sleep technologists sleep-staged and scored the PSG studies manually in 30 s epochs according to clinical practice (Berry et al., 2012).The sleep parameters recorded and calculated included: time in bed (TIB; the time from lights out until the end of the study), sleep period time (the amount of time from sleep onset until morning awakening), total sleep time (TST; the sleep period excluding any period of wake), sleep latency (the period of from lights out until sleep onset), REM latency (the period from sleep onset until the first period of REM sleep), sleep efficiency (the ratio of TST to TIB), % wake after sleep onset (WASO, wake after sleep onset as a percentage of sleep period time), percentage of TST in N1, N2, N3, and REM sleep, as well as sleep fragmentation (arousal index, NREM1%, and respiratory arousals).
The BRIEF yields eight subscales, each of which reflect a specific aspect of executive functioning: (i) inhibit; (ii) shift; (iii) emotional control; (iv) initiate; (v) working memory; (vi) plan/organise; (vii) organisation of materials; (viii) monitor.These subscales comprise the behaviour regulation index, metacognition index and global executive composite which were the three BRIEF measures used in this study.Higher scores on these scales indicate worse behavioural difficulties.
ing would indicate a disruption to normal brain development, however, it is uncertain to what degree the 0.1-0.2mm increases in cortical thickness we observed in sleep disordered breathing is clinically relevant.Others have posited that cortical thickening could

F
I G U R E 2 Volume (mm 3 ) in regions of interest demonstrating significant differences between controls and SDB.*Significant at p < 0.05.(a) left hemisphere caudate; (b) right hemisphere caudate; (c) bilateral caudate; (d) left hemisphere caudal middle frontal gyrus; (e) bilateral rostral anterior cingulate gyrus.
Sleep and respiratory statistics of controls and children with SDB presented as median and interquartile ranges T A B L E 1Abbreviations: CRDI, central respiratory disturbance index; NREM, non-REM; OAHI, obstructive apnea-hypopnea index; REM, rapid eye movement; Resp Ar, respiratory arousal index as a percentage of the total arousal index; SDB, sleep disordered breathing; SpO 2 , oxygen saturation; TST, total sleep time; WASO, wake after sleep onset.*Significant at p < 0.05 compared with controls.**Significantat p = 0.01 compared with controls.
Unadjusted means and standard deviations of cortical thickness (mm 2 ) and volume (mm 3 ) for each group by region of interest T A B L E 2 Mean and standard deviations of behavioural subscale and cognitive index scores by groupT A B L E 3 T A B L E 3 (Continued)

Table 2 )
, this result may be due to the small sample size in the study.Additionally, beha- (Kheirandish-Gozal & Gozal, 2012)red breathing can manifest in numerous behavioural phenotypes(Kheirandish-Gozal & Gozal, 2012)and is also likely subject to individual vulnerability and other factors such as co-morbid sleep disorders.With such prevalent heterogeneity in how sleep disordered breathing can present behaviourally, it is