Medicine 65 (2020) 142e146

Contents lists available at ScienceDirect

Sleep Medicine

journal homepage: www.elsevier.com/locate/sleep

Original Article Reduced sleep spindle activity in children with primary

* Pablo E. Brockmann a, b, , Oliviero Bruni c, Leila Kheirandish-Gozal d, David Gozal d a Department of Pediatric Cardiology and Pulmonology, Division of Pediatrics, School of Medicine, Pontificia Universidad Catolica de Chile, Santiago, Chile b Pediatric Sleep Center, School of Medicine, Pontificia Universidad Catolica de Chile, Santiago, Chile c Department of Developmental and Social Psychology, Sapienza University, Rome, Italy d Department of Child Health and Child Health Research Institute, University of Missouri School of Medicine, Columbia, MO, 65201, USA article info abstract

Article history: Background: Habitually snoring children are at risk of manifesting disease-related problems even if their Received 10 July 2019 sleep studies are overall within normal limits. Received in revised form Study objectives: To compare sleep spindle activity in children with primary snoring and healthy controls. 28 September 2019 Methods: Sleep spindle activity including analysis of fast and slow spindles (ie, >13 Hz and <13 Hz, Accepted 1 October 2019 respectively) was evaluated in polysomnographic (PSG) recordings of 20 randomly selected children with Available online 11 October 2019 primary snoring (PS; normal PSG recordings except for objective presence of snoring; 12 boys, mean age 6.5 ± 2.1 years), and 20 age- and gender-matched PSG-confirmed non-snoring controls. Keywords: fi Sleep spindles Results: PS children showed signi cantly lower spindle indices in all non-rapid eye movement (NREM) < ¼ Snoring sleep stages (p 0.05). In contrast, fast spindles were found in 40% (n 8) children with PS and in 25% Sleep microstructure (n ¼ 5) controls. Sleep spindle activity was particularly higher in NREM sleep stage 2 in controls compared PS (76% versus 43% of all marked sleep spindles events in NREM sleep stage 2, p < 0.001). Pediatric Conclusions: Children with PS exhibit significantly reduced spindle activity when compared to matched controls. Reduced sleep spindle activity may be an indicator of sleep disruption and, therefore, could be involved in the development of disease-related consequences in snoring children. © 2019 Elsevier B.V. All rights reserved.

1. Introduction while fast spindles have been correlated with other cognitive processes, such as global intelligence [13]. Sleep spindles are bursts of 10e16 Hz electroencephalographic A large number of conditions may be associated with reduced (EEG) activity, typically between 0.5 and 2 s in duration [1]. These spindle activity in childhood, and this is particularly the case cortical electrical activity elements are characteristic of N2 stage of among children with intellectual disabilities [14] in whom virtually non-REM (NREM) sleep, but are also present, albeit to a much lesser no sleep spindles are found. On the other hand, children with extent during the other NREM sleep stages [1]. Spindle activity normal IQ and dyslexia showed an increase in spindle activity changes with age, and such changes are believed to reflect the strictly related to the extent of improvement after rehabilitation maturation of thalamocortical regulatory pathways [2]. Sleep [15]. In contrast to the relatively higher number of studies spindles have been classified according to their frequency into fast regarding spindle activity and neurocognition, few studies have and slow spindles (ie, >13 Hz and <13 Hz, respectively) [3e5]. shown a relation between sleep-disordered-breathing (SDB) and a Sleep spindle activity has been associated with memory decrease of sleep spindle activity in adults [16] and in children [17]. consolidation [3], development of social skills [6], intelligence Moreover, a recent study showed that spindle activity seems to be quotient [7], and overall learning capability [5,7e11]. In adults, slow significantly different between children with obstructive sleep spindles have been associated with visual perceptual learning [12], apnea (OSA) and controls [17]. Previous studies have failed to demonstrate differences in standard sleep architecture and arousals between children with primary snoring and children who have never snored [18]. Notwithstanding, snoring children show significant increases in the * Corresponding author. Department of Pediatric Cardiology and Pulmonology, e Pediatric Sleep Center, School of Medicine, Pontificia Universidad Catolica de Chile, risk of manifesting disease-related consequences [19 22], even in Lira 85, 5to piso, 8330074, Santiago de Chile, Chile. the absence of apneas, desaturations or classic EEG arousals. We E-mail address: [email protected] (P.E. Brockmann). https://doi.org/10.1016/j.sleep.2019.10.001 1389-9457/© 2019 Elsevier B.V. All rights reserved. P.E. Brockmann et al. / 65 (2020) 142e146 143 hypothesize that spindle activity is different in children with pri- with a duration >0.5 s, usually maximal in amplitude using central mary snoring than in controls. Therefore, in this study, we aimed to derivations. In this study, all spindles included in the count had an investigate whether sleep spindle activity differs in children with amplitude exceeding 10 mV and had typical waxing-waning PS compared with matched non-snoring controls. morphology. Sleep spindle activity was detected via HypnoLab® software 2. Methods developed for automatic detection of sleep spindles and other micro-sleep phenomena (SWS Soft, Italy). Central (C3eC4) elec- 2.1. Subjects trodes for each epoch of N2 and N3 sleep were used for spindle detection. The program identified spindle events with a frequency (PSG) recordings were randomly selected of 11e16 Hz, 14 mV or higher in amplitude, and 0.5e3 s in duration from an existing database of community-dwelling children aged based on standard spindle criteria. five to nine years who were recruited between 2006 and 2014 Spindles were then visually checked by the investigator (PEB) from the Louisville and Chicago metropolitan areas. This age for consistency after the automated analysis. Spindle indices in group was selected for convenience, as snoring is prevalent in stages N1, N2 and N3 (slow wave sleep) were calculated as the ratio preschoolers. Children from Louisville were recruited through of the number of sleep spindles to the number of minutes in stage. collaboration with public schools, and children from Chicago After an initial sleep spindle analysis, a second differentiated were recruited through community announcements and distri- analysis between fast and slow spindles (ie, >13 Hz and <13 Hz, bution of materials at the University of Chicago Medical Center. respectively) was conducted using the above-mentioned software. This study was approved by University of Louisville (protocol Again, spindles were then visually checked for consistency after the #474.99), and the University of Chicago (protocol 09-115-B), second automated analysis. informed consent was obtained from each parent and assent was solicited from each participant. From the database, otherwise healthy children who had never been clinically referred for sus- 2.5. Statistical analysis pected SDB were identified as having primary snoring (PS) based on their sleep questionnaire (snoring >3 nights/week) and Descriptive statistics were used to analyze children's de- corroborated by an overnight PSG, and then an age-, gender-, mographic and PSG characteristics (ie, numbers, percentages, location-, parental higher education-matched child was identi- median, minimum, maximum for non-normal distributed data; fied from non-snoring controls with normal PSG. and mean and standard deviation for those with normal distri- bution). The comparison between spindles indices obtained in 2.2. Procedures children with primary snoring and controls was performed by means of the ManneWhitney test for independent datasets. Based All included subjects completed a full single night sleep lab- on a previous pilot sample [17], a sample size analysis was per- based PSG and were accompanied by one of their parents formed. Considering 0.05 type-I-error and 0.2 type-II-error to throughout the night. A blinded investigator interpreted each PSG. detect a significant difference in mean sleep spindles indices, EEG based signal analysis was performed in order to extract data for n ¼ 10 individuals were needed at least in each group. Differences spindle activity analysis. This analysis was performed using Hyp- were considered statistically significant at p < 0.05. The nolab® software. commercially available software SPSS® version 20, was used for these statistical tests. 2.3. Polysomnography

PSG was conducted using a commercially available data acqui- 3. Results sition system (Polysmith; Nihon Kohden America, Inc., Irvine, CA), and interpreted according to the current international standards PSG studies from 20 randomly selected children with PS and 20 published by the American Academy of Sleep Medicine [23]. The age-, gender- and parental education-matched controls were study montage comprised the following channels: 8-lead electro- included for the purposes of this study. Demographic data of encephalography, 2-lead electrooculography, 3-lead submentalis included subjects are shown in Table 1. In addition, none of the electromyography, chest and abdominal wall movements, nasal children were receiving any chronic medication and per parental pressure transducer, nasal-oral airflow measured with a thermistor, reports, all children were performing normally in school. snoring, pulse oximetry-derived arterial hemoglobin oxygen satu- Sleep macrostructure results are provided in Table 2. Children ration and pulse waveform, electrocardiogram and instantaneous with PS had a significantly reduced total sleep time beat-to-beat heart rate, digital audio and video. A minimum of 6 h (436.8 ± 57.6 min versus 485.3 ± 68.7 min, p ¼ 0.029). Sleep effi- of sleep time was required for interpretation. Respiratory events ciency was similar in both groups (90.4 ± 4.5% versus 85.6 ± 9.2%, and sleep architecture were analyzed according to the above- p ¼ 0.123). Sleep macrostructure results were similar between both mentioned criteria [23]. For this study, only children with a groups, NREM and REM sleep distribution were almost identical normal obstructive apnea hypopnea index (AHI) were selected (ie, (see Table 2). AHI < 1/hour total sleep time). A sleep medicine specialist, who was Table 3 shows the findings of sleep spindle analyses for both blinded to all subject's data, performed all PSG-based spindles groups. Fast spindles (ie, >13 Hz) were found only in eight children analysis. with primary snoring and in five controls. Children with primary snoring showed significantly lower spindle indices in each NREM 2.4. Sleep spindle activity sleep stages (p < 0.05 for all sleep stages). Sleep spindle distribution was significantly different in children After a first software-based identification of EEG events in with PS compared with controls: sleep spindle activity was NREM sleep, sleep spindles were automatically identified in the significantly more concentrated in each NREM sleep stage 2 in central channel (C3 or C4) in all epochs scored as stage N2 or N3 for controls than in children with primary snoring (76% versus 43% of the entire night as trains of distinct waves with frequency 10e16 Hz all marked sleep spindles events in NREM sleep stage 2, p ¼ 0.001). 144 P.E. Brockmann et al. / Sleep Medicine 65 (2020) 142e146

Table 1 Demographic characteristics of the sample.

Total PS Controls P value N ¼ 40 N ¼ 20 N ¼ 20

Demographic characteristics Age (y) 6.5 ± 2.1 6.4 ± 2.4 6.5 ± 2.5 >0.05 Males, n (%) 24 (60) 12 (60) 12 (60) >0.05 Parental high educational level, n (%) 30 (75) 15 (75) 15 (75) >0.05 Race (black, %) 16 (40) 8 (40) 8 (40) >0.05 Asthma (%) 4 (10) 3 (15) 1 (5) >0.05 BMI z score (% obese) 1.28 ± 0.23 (25) 1.37 ± 0.32 (30) 1.14 ± 0.21 (20) >0.05 AHI 0.72 ± 0.21 0.83 ± 0.24 0.67 ± 0.20 >0.05

If not otherwise noted, results show mean and SE. Abbreviations: AHI e obstructive apnea-hypopnea index; high educational level reflects completion of a Bachelor degree or equivalent.

Table 2 robust correlate, and fail to predict behavioral or cognitive conse- Sleep macrostructural characteristics in children with primary snoring and controls. quences in children with SDB. In a study of 1010 community-based Primary Snoring Controls p children, our group showed that only children with a relatively high N ¼ 20 N ¼ 20 AHI (ie, >5 events/hour of sleep) exhibited a significantly measur- fi TIB-min 509.1 24.0 535.6 64.6 0.029 able risk of signi cant OSA-related neurocognitive consequences SPT-min 475.3 24.7 505.1 59.7 0.011 [16]. However, even milder forms of OSA or SDB seem to manifest TST-min 436.8 57.6 485.3 68.7 0.029 increased risk for disease-related consequences that sometimes are SE% 90.4 4.5 85.6 9.2 0.123 similar in magnitude to those of children with frank OSA N1% 3.2 1.4 1.7 1.3 0.529 e fi N2% 59.5 7.3 62.9 8.5 0.280 [20,24 27]. Hence, the identi cation of sleep disruption that are N3% 23.5 9.1 23.0 6.2 0.739 predicated on PSG analyses and predict the presence of disease REM % 13.8 7.7 12.4 6.1 0.900 related consequences in these milder forms of SDB seems to be an All results are given as mean and SD. Abbreviations: REM, rapid eye movement important research task. sleep; SE, sleep efficiency; SPT, sleep period time; SWS, slow wave sleep; TST, total In a previous study, we already showed that spindle activity sleep time. differs among children with and without OSA [17]. In that study, sleep spindle activity was correlated with several domains of neurocognitive performance, especially concerning memory [17]. Table 3 Comparison of spindle indices between children with primary snoring and controls. In the present study, we aimed to go a step further by exploring potential changes in sleep spindle characteristics among children Primary Controls p with an even milder form of sleep-disordered-breathing (ie, pri- ¼ snoring N 20 fi N ¼ 20 mary snoring). The ndings support substantial vulnerability of sleep spindle activity, and therefore provide a potential highly Mean SD Mean SD sensitive marker of sleep disruption. We cannot infer form the Spindle Index N2 14.8 9.0 132.1 65.7 0.001 present study as to the mechanisms underlying the unique sus- Spindle Index SWS 16.9 5.8 41.3 21.7 0.029 ceptibility of spindles to snoring. It is possible that the reduced Spindle Index NREM 14.1 7.7 98.3 52.8 0.001 sleep spindle activity in children with PS is either (i) a marker of Spindle index: ratio of the number of sleep spindles to the number of minutes in sleep disruption due to snoring and increased upper airway resis- each stage of NREM (ie, N2, SWS ¼ N3, total NREM sleep). tance and increased respiratory effort, or (ii) alternatively it may Abbreviations: NREM, not rapid eye movement; SD, standard deviation; and SWS, slow-wave sleep. indicate diffuse thalamo-cortical dysfunction during sleep in chil- dren with PS. The first hypothesis concerning sleep disruption due to the presence of this distinctive sleep spindle pattern may indi- 4. Discussion cate early sleep microstructural modifications in children with PS. These findings are in line with previous studies showing subtle This study shows that PS children display reduced spindle ac- forms of sleep disruption in children with PS [28,29]. tivity compared with healthy controls. Sleep spindle activity was The hypothesis that thalamo-cortical dysfunction occurs in SDB significantly lower in each NREM sleep stages in children with PS. has already been suggested in adults with OSA [30]. A study of 21 € Thus, even in the absence of apneic events and oxyhemoglobin adults with OSA performed by Schonwald et al., showed reduced desaturations, children with PS exhibit a different sleep micro- sleep spindle activity when compared with seven healthy adults. In architecture throughout NREM sleep stages. This finding is poten- that study, sleep frontal sleep spindle activity in patients with OSA tially of great relevance since it reveals for the first time the extent was significantly reduced, compared with more posterior brain of sleep disruption among children with the mildest presentation regions that appeared to maintain spindle frequency during sleep of SDB. [30]. However, the exact mechanisms that account for this specific The identification of specific PSG features that may reliably sleep spindle pattern in children with PS remain to be established. correlate with the presence of behavioral and neurocognitive In line with previous pediatric studies, fast spindles were found problems among children even with only PS is critically needed if only in a proportion of children in our cohort. Hoedlmoser et al. [3], we wish to better phenotype at-risk children and identify those described the predominant presence of slow spindles in healthy who may benefit from therapeutic intervention even when stan- children, and the hypothesis that fast spindles were not present in dard classic disease criteria do not apply, such as in PS. Indeed, all children was postulated by these authors. In the present study, classical respiratory-based indices like the AHI do not provide a we used a wide range of frequencies (ie, 11e16 Hz), in order to P.E. Brockmann et al. / Sleep Medicine 65 (2020) 142e146 145 identify the highest possible quantity of spindles. However, types of The ICMJE Uniform Disclosure Form for Potential Conflicts of spindle and even inter-spindle frequency are noteworthy. Slow Interest associated with this article can be viewed by clicking on the spindles seem to be more prevalent in frontal regions, whereas fast following link: https://doi.org/10.1016/j.sleep.2019.10.001. spindles occur more often in parietal locations [31]. Furthermore, “ ” inter-spindle frequency modulation or chirping has emerged as a References possible marker of differential cortical response to apnea in adults [16]. In contrast with the existing evidence in adults with OSA, [1] Purcell SM, Manoach DS, Demanuele C, et al. Characterizing sleep spindles in sleep spindle activity has been less frequently evaluated in children 11,630 individuals from the national sleep research resource. Nat Commun with OSA. In our previous study, we already reported that controls 2017;8:15930. [2] Clawson BC, Durkin J, Aton S. Form and function of sleep spindles across the had a higher spindle index compared with children with OSA, lifespan. Neural Plast 2016;2016:6936381. especially during NREM sleep stage N2 [17]. Current findings are in [3] Hoedlmoser K, Heib DP, Roell J, et al. Slow sleep spindle activity, declarative e line with that previous study. Thus, there seems to be an apparent memory, and general cognitive abilities in children. Sleep 2014;37:1501 12. [4] Tamaki M, Matsuoka T, Nittono H, et al. Fast sleep spindle (13-15 hz) activity reduction in sleep spindle activity in children with SDB, even in the correlates with sleep-dependent improvement in visuomotor performance. mildest form (ie, PS). To which extent this specific sleep micro- Sleep 2008;31:204e11. structure characteristic may be considered as a clinically relevant [5] Chatburn A, Coussens S, Lushington K, et al. Sleep spindle activity and cognitive performance in healthy children. Sleep 2013;36:237e43. dysfunction remains unknown. It is possible that sleep spindle [6] Wilhelm I, Groch S, Preiss A, et al. Widespread reduction in sleep spindle density reductions may reflect compensatory mechanisms related activity in socially anxious children and adolescents. J Psychiatr Res 2017;88: to sleep homeostasis in the context of SDB. However, such 47e55. fi [7] Gruber R, Wise MS, Frenette S, et al. The association between sleep spindles compensatory mechanism may lead to signi cant consequences, and IQ in healthy school-age children. Int J Psychophysiol 2013;89:229e40. such as deterioration of several synaptic processes as reflected by [8] Bergmann TO, Molle M, Diedrichs J, et al. Sleep spindle-related reactivation of and learning [3,5,7,8,10,17]. category-specific cortical regions after learning face-scene associations. Neu- fi roimage 2012;59:2733e42. In the present study, sleep spindle activity was signi cantly [9] Eschenko O, Molle M, Born J, et al. Elevated sleep spindle density after more concentrated in NREM sleep stage 2 in controls than in chil- learning or after retrieval in rats. J Neurosci 2006;26:12914e20. dren with primary snoring. To our knowledge, this finding is novel [10] Schabus M, Hodlmoser K, Gruber G, et al. Sleep spindle-related activity in the in children. Reasons for this specific sleep stage distribution of human EEG and its relation to general cognitive and learning abilities. Eur J Neurosci 2006;23:1738e46. spindle activity are unknown. Sleep spindles present changes in [11] Schabus M, Hoedlmoser K, Pecherstorfer T, et al. Interindividual sleep spindle their topographic distribution during life [7]: Spindles become differences and their relation to learning-related enhancements. Brain Res e identifiable after the sixth week of life [32]. After that, spindles may 2008;1191:127 35. [12] Bang JW, Khalilzadeh O, Hamalainen M, et al. Location specific sleep spindle appear asymmetric (ie, only in one cerebral hemisphere). After the activity in the early visual areas and perceptual learning. Vis Res 2014;99: first year of life they become more synchronous and are easier to 162e71. identify [33]. To which extent, snoring may affect the normal [13] Bodizs R, Kis T, Lazar AS, et al. Prediction of generalmental ability based on e fi measures of sleep. J Sleep Res 2005;14:285 92. development of sleep spindle activity and lead to speci c sleep [14] Gruber R, Wise MS. Sleep spindle characteristics in children with neuro- stage differences, is unclear. developmental disorders and their relation to cognition. Neural Plast The present study has some limitations to acknowledge. First, 2016;2016:4724792. [15] Bruni O, Ferri R, Novelli L, et al. Sleep spindle activity is correlated with this is relatively small sample size, considering a larger sample may reading abilities in developmental dyslexia. Sleep 2009;32:1333e40. add information on the differences between primary snorers and [16] Carvalho DZ, Gerhardt GJ, Dellagustin G, et al. Loss of sleep spindle frequency controls. We also did not perform analyses regarding cognitive or deceleration in . Clin Neurophysiol 2014;125:306e12. [17] Brockmann PE, Damiani F, Pincheira E, et al. Sleep spindle activity in children behavioral outcomes, which would be interesting to address in with obstructive sleep apnea as a marker of neurocognitive performance: a future studies. Finally, PSG reflects one night in children's sleep, and pilot study. Eur J Paediatr Neurol 2018;22:434e9. the lack of information regarding sleep in the nights prior to the [18] Brockmann PE, Urschitz MS, Noehren A, et al. Risk factors and consequences e PSG or an adjustment night may have also influenced partially our of excessive autonomic activation during sleep in children. Sleep Breath Schlaf Atmung 2011;15:409e16. results. Notwithstanding these shortcomings, current findings [19] Brockmann PE, Bertrand P, Pardo T, et al. Prevalence of habitual snoring and point to the need for further understanding of the impact of snoring associated neurocognitive consequences among Chilean school aged children. e and other forms of sleep-disordered breathing on sleep micro- Int J Pediatr Otorhinolaryngol 2012;76:1327 31. [20] O'Brien LM, Mervis CB, Holbrook CR, et al. Neurobehavioral implications of structure and spindle activity. habitual snoring in children. Pediatrics 2004;114:44e9. [21] Smith DL, Gozal D, Hunter SJ, et al. Parent-reported behavioral and psychiatric problems mediate the relationship between sleep-disordered breathing and 5. Conclusions cognitive deficits in school-aged children. Front Neurol 2017;8:410. [22] Smith DL, Gozal D, Hunter SJ, et al. Frequency of snoring, rather than apnea- Children with primary snoring exhibit significantly lower spin- hypopnea index, predicts both cognitive and behavioral problems in young children. Sleep Med 2017;34:170e8. dle activity when compared to carefully matched controls. The [23] Berry RB, Budhiraja R, Gottlieb DJ, et al. Rules for scoring respiratory events in identification of reduced sleep spindle activity may be an indicator sleep: update of the 2007 AASM manual for the scoring of sleep and associ- of early sleep disruption in PS. The impact of this specific sleep ated events. Deliberations of the sleep apnea definitions task force of the American Academy of sleep medicine. J Clin Sleep Med JCSM e Offic Publ Am spindle pattern on disease-related consequences has yet to be Acad Sleep Med 2012;8:597e619. established. [24] Brockmann PE, Urschitz MS, Schlaud M, et al. Primary snoring in school children: prevalence and neurocognitive impairments. Sleep Breath e Schlaf Atmung 2012;16:23e9. Funding [25] Ersu R, Arman AR, Save D, et al. Prevalence of snoring and symptoms of sleep- disordered breathing in primary school children in istanbul. Chest 2004;126: 19e24. FONDECYT project (#1180397) supported PEB. LKG and DG are [26] Marcus CL, Hamer A, Loughlin GM. Natural history of primary snoring in supported in part by National Institutes of Health grants HL130984 children. Pediatr Pulmonol 1998;26:6e11. and HL140548. [27] Sahin U, Ozturk O, Ozturk M, et al. Habitual snoring in primary school chil- dren: prevalence and association with sleep-related disorders and school performance. Med Princ Pract 2009;18:458e65. Conflict of interest [28] Lopes MC, Guilleminault C. Chronic snoring and sleep in children: a demon- stration of sleep disruption. Pediatrics 2006;118:e741e6. [29] Tauman R, O'Brien LM, Holbrook CR, et al. Sleep pressure score: a new index The authors have no conflict of interest to declare. of sleep disruption in snoring children. Sleep 2004;27:274e8. 146 P.E. Brockmann et al. / Sleep Medicine 65 (2020) 142e146

[30] Schonwald€ SV, Carvalho DZ, de Santa-Helena EL, et al. Topography-specific [32] Metcalf DR. The effect of extrauterine experience on the ontogenesis of EEG spindle frequency changes in obstructive sleep apnea. BMC Neurosci 2012;13: sleep spindles. Psychosom Med 1969;31:393e9. 89. [33] Kellaway P. An orderly approach to visual analysis: characteristics of the [31] Jobert MPE, Jahnig€ P, Schulz H, et al. Topographical analysis of sleep spindle normal EEG of adults and children. Curr Pract Clin Electroencephalogr 1990;2: activity. Neuropsychobiology 1992;26:210e7. 139e99.