CPAP Treatment Partly Normalizes Sleep Spindle Features in Obstructive Sleep Apnea

Resea ch A icle CPAP T ea men Pa ly No malizes Sleep Spindle Fea u es in Obs uc i e Sleep Apnea Tiia Saunamäki,1Ee o Huupponen,2Juho Loponen,3and Sa i-Leena Himanen2,3 1Depa men o Neu ology and Rehabili a ion, Tampe e Uni e si y Hospi al, Tampe e, Finland 2Depa men o Clinical Neu ophysiology, Medical Imaging Cen e and Hospi al Pha macy, Pi kanmaa Hospi al Dis ic , Tampe e, Finland 3Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland Co espondence should be add essed o Sa i-Leena Himanen; sa i-leena.himanen@u a. i Recei ed 28 Oc obe 2016; Re ised 26 Decembe 2016; Accep ed 15 Janua y 2017; Published 2 Feb ua y 2017 Academic Edi o : Ma co Zucconi Copy igh © 2017 Tiia Saunam¨ aki e al. This is an open access a icle dis ibu ed unde he C ea i e Commons A ibu ion License, which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. Objec i e. Obs uc i e sleep apnea (OSA) dec eases sleep spindle densi y and equency. We e alua ed he e ec s o con inuous posi i e ai way p essu e (CPAP) ea men on di e en ea u es o sleep spindles. Me hods. Twen y OSA pa ien s unde wen wo nigh polysomnog aphies in a diagnos ic phase and one nigh polysomnog aphy a e 6 mon hs o CPAP ea men . The con ol g oup comp ised 20 heal hy con ols. Sleep spindles we e analyzed by a p e iously de eloped au oma ed me hod. Unila e al and bila e al spindles we e iden i ied in cen al and on opola b ain loca ions. Spindle densi y and equency we e de e mined o he i s and las hal o he NREM ime. Resul s. The densi y o bila e al cen al spindles, which did no change in he un ea ed OSA pa ien s, inc eased owa ds he mo ning hou s du ing CPAP ea men and in he con ols. Cen al spindles did no become as e wi h sleep in OSA pa ien s and he cen al spindles emained slow in he le hemisphe e e en wi h CPAP. Conclusion. CPAP ea men no malized spindle ea u es only pa ially. The changes may be associa ed wi h de ici s in halamoco ical spindle gene a ing loops. Signi icance. This s udy shows ha some sleep spindle changes pe sis a e CPAP ea men in OSA pa ien s. The associa ion o hese changes o day ime symp oms in OSA pa ien s needs o be u he e alua ed. 1. In oduc ion By s anda d de ini ion, sleep spindles a e composed o ansien elec oencephalog aphy (EEG) oscilla ions in he equency ange o 11 Hz o 16 Hz las ing a leas 0.5 s [1]. In heal hy subjec s, as e spindles o spindle ac i i y occu mo e o en in he pa ie al a ea, while slowe spindles a e mo e p e alen in he on al egions [2–8]. The in aspindle equency inc eases du ing he mo ning hou s [2, 9]. Pa ic- ula ly he cen al spindles occu in pe iodic sequences which a e ound o be mo e equen a he mo ning hou s [10]. Spindles a e abundan in ligh non-REM (NREM) sleep and a e supposed o p e en a ousing s imuli om eaching heco ex[11], husensu ingsolidsleep[2].Mo e ecen ly, spindles ha e been ound o play an impo an ole in many b ain unc ions such as in lea ning and memo y p ocesses [12]. These unc ions seem o be dependen on he long- e m po en ia ion (LTP) induced by he spindle bu s s [13]. In pa icula , he as spindles (usually >13 Hz) ha e been ound o ha e an impo an ole in connec ing he hippocampus and heneoco exandin ans e ing empo a yhippocampal memo ies o he neoco ex o s o age [14, 15]. In addi ion, local spindles ha e been p esen ed o a ise in co ical a eas ha we e ac i e in mo o o isuomo o asks du ing p eced- ing wake ulness [16, 17]. Obs uc i e sleep apnea (OSA) is cha ac e ized by epe - i i e episodes o uppe ai way obs uc ion du ing sleep and esul s in oxygen desa u a ion and a ousals om sleep. In OSA pa ien s, spindle densi y dec eases, he pe cen age o slow spindles inc eases, and he spindles emain slow h oughou he nigh [18–20]. Con inuous posi i e ai way p essu e (CPAP) ea men is a i s -line he apy in mod- e a e and se e e OSA. CPAP ea men educes obs uc i e b ea hing e en s and no malizes oxyhemoglobin sa u a ion and sleep agmen a ion. The esul o CPAP ea men is imp o ed sleep quali y and educed day ime symp oms. Hindawi Sleep Diso de s Volume 2017, A icle ID 2962479, 10 pages h ps://doi.o g/10.1155/2017/2962479 2Sleep Diso de s Howe e ,someday imesymp omssuchascogni i ede ici s seem o emain [21, 22]. The e ec o CPAP ea men on spindle ac i i y in OSA pa ien s is, howe e , an open ques ion. In he p esen s udy, i is assumed ha CPAP mos ly no malizes he dis u bed sleep spindle p ocess in OSA bu ha some impai men in he spindle p ocess may s ill emain. Since many spindle ea u es change dynamically du ing noc u nal sleep [9], spindle ea u es a e e alua ed in he i s and las pa o he noc u nal NREM sleep. 2. Ma e ials and Me hods Thes udycoho included20maleOSApa ien sand20age- ma chedheal hycon ols.Thepa ien shadbeen e e ed o he sleep labo a o y because o suspec ed OSA. All pa ien s me he diagnos ic c i e ia o OSA, and hei i s ea men choice was CPAP. The con ol subjec s we e heal hy olun- ee s ec ui ed h ough ad e isemen s. They we e no paid o hei pa icipa ion in he s udy. Bo h pa ien s and con ols we e i s in e iewed o ensu e hey me he ini ial eligibili y c i e ia: wo king-aged, igh -handed, no (o he ) sleep diso de s, no clinically sig- ni ican medical diso de (e.g., neu ological o psychia ic disease, hypo/hype hy oidism, o lung disease o he han asymp oma ic as hma), and no subs ance o alcohol abuse. Bo h s udy g oups unde wen a clinical in e iew and diagnos ic polysomnog aphy (PSG) in a sleep labo a o y. The OSA diagnosis was based on a clinical pic u e, subjec i e complain s and an apnea/hypopnea index (AHI) alue o >10 pe hou o sleep.Thecon olshad obeasymp oma icand ha e an AHI alue o <5 pe hou o sleep. The pa ien s and con ol subjec s who, acco ding o he esul s o he i s all- nigh diagnos ic PSG, me he eligibili y c i e ia unde wen a second all-nigh PSG, and he da a om his nigh we e used in he analysis. A e 6 mon hs o CPAP ea men , he pa ien s unde - wen a hi dPSG.Objec i eCPAPcompliancemeasu eswe e downloaded om he de ices and he minimum equi ed CPAP adhe ence was se a 4 hou s pe nigh and a leas 5 nigh s pe week. All pa ien s me he CPAP adhe ence c i e ia and we e included in o he analyses. The s udy was app o ed by he E hical Commi ee o he Pi kanmaa Hospi al Dis ic and all subjec s ga e hei w i en in o med consen . 2.1. Reco dings and Visual Analysis. Subjec s wen o sleep be ween 10 pm and 12 pm, acco ding o hei own habi ual bed imes. Six EEG de i a ions (Fp1-A2, Fp2-A1, C3-A2, C4- A1, O1-A2, and O2-A1), wo elec ooculog aphy channels, submen al muscle onus, elec oca diog am, ai low p essu e by nasal ansduce , he mis o , ho acoabdominal espi a- o y mo emen s, and blood oxygen sa u a ion we e eco ded. In addi ion, anscu aneous ca bon dioxide ension, leg mo emen s, body posi ion and body mo emen s we e also eco ded. Polysomnog aphies we e eco ded wi h an Embla N7000 de ice (Embla, Na us Medical Inc., USA). The EEG signalswe esampleda 200Hz(16bi s)wi habandwid ho 0.3 Hz o 90 Hz. As he s udy began be o e he e ision o he sleep s aging ules, he on opola EEG channels Fp1 and Fp2 we e used ins ead o he on al channels F3 and F4 ha a e ecommended nowadays. The sleep s aging was, howe e , pe o med in acco dance wi h he ules es ablished by he AASM in 2007 [23] wi h he excep ion o he EEG channels. Somnologicaso wa e (Medca e/Flaga, Iceland) was used o he isual analyses. The second diagnos ic nigh and he ea men nigh we e classi ied by wo independen sco e s in o s ages. The le el o ag eemen be ween he 2 sco e s was 86.4% (𝐾= 0.76). Based on he independen sco ings, he consensus sleep s aging was o med and used in he s a is ical analyses.TheAHI aluewascalcula edas henumbe o obs uc i e apneas and hypopneas (las ing ≥10 s) pe hou o sleep. Obs uc i e apneas we e de ined as a leas a 90% educ ion in he he mal signal ampli ude, whe eas hypop- neas we e de ined as diminu ion o a leas 30% o he nasal p essu e signal ha was associa ed wi h an a ousal o desa u- a ion o 3% [24]. Mic oa ousals we e sco ed acco ding o he c i e ia o he Ame ican Sleep Diso de s Associa ion [25]. In he diagnos ic nigh , 20 pa ien s had an AHI alue o ≥10/h (p e-CPAP g oup). Thei sleep pa ame e s and spindle ea u es we e compa ed wi h he pa ame e s o he age-ma ched heal hy olun ee s (con ol g oup). A e 6 mon hs o CPAP ea men , he pa ien s unde wen ano he polysomnog aphy and he alues o ha nigh (CPAP g oup) we e compa ed wi h he p e-CPAP alues and wi h he alues o he con ol g oup. 2.2. Spindle Analysis. Spindles we e analyzed by applying p e iously de eloped au oma ed spindle analysis me hods [3, 26]. The main componen o hese me hods is a sigma index ha is based on he as Fou ie ans o m (FFT). The equency band o 10.5 Hz o 16 Hz was used in compu ing he sigma index. A high sigma index alue indica es a high p obabili yo aspindleand ice e sa,and u he indica es how dominan he spindle peak is when compa ed wi h o he EEG ac i i ies. Sigma index alues o e 4.5 gene ally indica e spindles [26]. The sigma index has buil -in a i ac ejec ion unc ionali ies[3],andi canalsobecomplemen edwi h spindle ampli ude in o ma ion [26]. In he p esen wo k, he wo cen al (C3-A2, C4-A1) and he wo on opola (Fp1-A2, Fp2-A1) EEG channels we e analyzed sepa a ely. Thus, he p e ious me hod, desc ibed p e iously [3] needed some mino adjus men s. As a esul , wo EEG channels we e p ocessed a a ime ins ead o ou . The NREM sleep ime was included in he p esen analysis. Asa i s s ep, hep ocedu ede ailedin he o me wo k[26] was used o ob ain a eco ding-speci ic ampli ude h eshold o he b ain egion ( wo EEG channels) by es ima ing he lowe limi o he spindle ampli udes. Then, in o de o de ec a spindle ( om one EEG channel a a ime), he sigma index had o exceed 4.5 and also he ampli ude h eshold had o be exceeded a he same ime ins an (as in Sec ion 2.2.4 in [3]). The ini ial spindle de ec ion, ob ained a 0.33-s ime esolu ion, was hen p ocessed u he , and in his way indi idual unila e al (as in Sec ion 2.2.5 in [3] using m om one EEG channel a a ime) and bila e al (as in Sec ion 2.2.6 in [3] wi h 𝐷=1) spindles we e iden i ied. Sleep Diso de s 3 Table 1: Median and ange o he sleep pa ame e s in he s udy g oups and he 𝑝 alues o he compa isons. Con ols P e-CPAP CPAP Con ols e sus p e-CPAP1Con ols e sus CPAP1P e-CPAP e sus CPAP2 Median Range Median Range Median Range 𝑝 alue 𝑝 alue 𝑝 alue Age 48.5 34–60 50.0 37–65 0.170 BMI 24.9 20.1–30.2 29.6 24.7–41.4 <0.001 TST min 431.5 348–527 435.8 375–532 395 264–503 0.344 0.256 0.006 SEI% 90.6 76.7–97.1 93.2 80.0–97.5 92.0 58.2–97.0 0.417 0.946 0.322 N1% 5.4 2.2–16.3 5.0 1.2–11.2 4.8 1.5–16.1 0.285 0.617 0.490 N2% 60.0 45.4–71.6 73.0 59.5–85.3 68.0 46.8–79.0 <0.001 0.016 0.005 N3% 12.5 2.1–25.4 2.6 0.0–14.1 6.6 0.6–29.3 <0.001 0.029 0.002 REM% 20.0 13.1–27.8 16.8 8.6–27.3 17.6 11.4–26.4 0.117 0.285 0.563 SL min 8.3 3.5–32.5 6.3 1.0–33.0 4.5 2.0–22.0 0.163 0.014 0.162 AHI 𝑛/h 1.8 0.0–4.9 42.0 14.0–103.0 0.7 0.0–14.0 <0.001 0.080 <0.001 ARI 𝑛/h 12.8 4.3–22.8 30.0 8.0–99.6 16.0 4.6–27.0 <0.001 0.152 <0.001 ODI4 𝑛/h 0.5 0.0–6.0 34.0 0.0–88.0 0.8 0.0–3.0 <0.001 0.496 <0.001 Sa02min% 91.0 83.0–95.0 79.0 62.0–92.0 90.5 84.0–94.0 <0.001 0.713 <0.001 Sa02mean% 95.9 93.0–97.0 93.6 83.1–97.5 96.1 93.3–97.3 <0.001 0.551 <0.001 1Mann-Whi ney 𝑈 es ; 2Wilcoxon es . The Bon e oni co ec ed 𝑝 alues a e epo ed: Bon e oni co ec ion ac o = 3. BMI = body mass index; min = minu es; n/h = numbe /hou ; TST = o al sleep ime; SEI%= sleep e iciency index; N1%–N3%= ime pe cen age o sleep s ages N1–N3 e e ed o as TST; REM%= ime pe cen age o REM sleep e e ed o as TST; SL = sleep la ency; AHI = apnea/hypopnea index; ARI = a ousal index; ODI4 = oxygen desa u a ion index; Sa02min%= minimum oxygen sa u a ion pe cen age; Sa02mean%= mean oxygen sa u a ion pe cen age. Bold on = s a is ically signi ican 𝑝 alue. Spindle ea u es we e hen de i ed. These ea u es included spindle densi y (numbe o spindles pe hou o NREM sleep) and median o he in aspindle equency alues (10.5 Hz o 16 Hz, numbe o cycles pe second, in Hz; in bila e al spindles he equency alues we e calcula ed o bo h hemisphe es sepa a ely) du ing NREM sleep. In addi ion, he bila e ali y index (C bil index 1 = he numbe o bila e al cen al spindles di ided by he numbe o unila e al C4 spindles (bil C/C4), C bil index 2 = bil C/C3, Fp bil index 1 = bil Fp/Fp2, Fp bil index 2 = bil Fp/Fp1) o he spindles was calcula ed. I he bila e ali y index is <1, he spindles a e mo e o en unila e al han bila e al. Finally, he ime occupied by spindle sequences (as a pe cen age o NREM ime, a leas h ee consecu i e spindles, wi h in e spindle in e al ≤5s) was calcula ed. As spindle ea u es a e ound o change dynamically du ing henigh ,wecalcula edallspindle ea u essepa a ely o he i s and las pa o noc u nal NREM sleep. Fo his, he NREM ime was di ided in o wo. The di ision was made a he hal -way poin be ween he i s and las second o he NREM sleep. The ob ained ampli ude h eshold o he on opola spindles was 9.6 𝜇V in he con ol g oup (median, ange 6.5 o 13.3), 10.6 𝜇V(7.4 o15.2)in hep e-CPAPg oup,and10.0𝜇V (6.9 o 14.6) in he CPAP g oup. The espec i e h esholds o he cen al spindles we e 11.7 𝜇V (median, ange 8.5 o 16.6), 11.7 𝜇V (8.2 o 15.4), and 11.1 𝜇V (8.3 o 16.0). The indi idually de e mined ampli ude h esholds did no di e s a is ically be ween he g oups. 2.3. S a is ics. Since some o he pa ame e s we e no no - mally dis ibu ed, median and ange we e used as desc ip i e s a is icsandnonpa ame ic es swe echosenino de o compa e he s udy g oups. The Mann-Whi ney 𝑈 es was used o compa e he con ols wi h he pa ien g oups. The Wilcoxon es was used o compa e p e-CPAP pa ien s wi h heCPAPg oup.Tohandle hemul iplecompa isons,app o- p ia e Bon e oni co ec ion ac o s we e used ( he o iginal 𝑝 alues we e mul iplied wi h he numbe o compa isons). In his way, he signi icance le el was se a 0.05 o all analyses and he epo ed 𝑝 alues a e based on wo- ailed es s. 3. Resul s The demog aphic da a and sleep pa ame e s o he s udy g oups a e p esen ed in Table 1. The g oups did no di e acco ding o age. The OSA pa ien s we e mo e obese han he con ols. The con ols had mo e N3 and less N2 han he p e- CPAP g oup, and hey had ewe a ousals, desa u a ions and b ea hing e en s han he p e-CPAP g oup. CPAP ea men dec eased he o al sleep ime o he pa ien s and no malized henumbe o apneas,a ousals,obs uc i eb ea hinge en s and hypoxemic e en s. Howe e , pa ien s wi h CPAP con in- ued o show mo e N2 sleep and less N3 sleep han heal hy con ols, and hei sleep la ency was sho e . 3.1.SpindleFea u esin heFi s Pa o heNREMTime. In he i s pa o he nigh , nei he on opola (bil Fp, Fp1, andFp2)no cen al(bilC,C3,andC4)spindledensi ies di e ed be ween he g oups (Table 2). In all g oups, he Fp- and C- spindles we e mo e o en unila e al han bila e al, as e ealed by he alues o he bila e ali y indices (Table 2). The e we e no s a is ical di e ences in he bila e ali y indices be ween he g oups. Spindle sequences we e ew in he i s 4Sleep Diso de s Table 2: Median and ange o he spindle ea u es in he s udy g oups in he i s pa o he NREM ime. Con ols P e-CPAP CPAP Con ols e sus p e-CPAP1Con ols e sus CPAP1P e-CPAP e sus CPAP2 Median Range Median Range Median Range 𝑝 alue 𝑝 alue 𝑝 alue Densi y 𝑛/h Bil Fp dens 33.7 3.4–94.8 22.5 0.3–117.7 23.7 0.0–0.0 0.671 1.0 1.0 Fp2 dens 77.5 14.6–138.6 48.2 8.8–183.9 50.8 0.0–0.0 1.0 1.0 1.0 Fp1 dens 68.0 15.0–129.0 49.0 6.9–143.0 50.9 0.0–0.0 0.912 1.0 1.0 Bil C dens 55.9 11.8–161.2 29.1 1.0–215.7 27.8 1.4–202.8 0.250 0.361 1.0 C4 dens 104.7 36.4–194.2 58.3 13.7–194.3 79.6 16.5–181.5 0.369 1.0 0.741 C3 dens 116.7 50.2–219.6 56.3 17.3–216.3 85.3 29.0–221.5 0.085 0.224 0.574 F equency Hz Bil Fp2 eq 11.8 10.9–12.5 11.5 10.9–13.1 11.4 10.6–12.9 0.702 0.250 0.068 Bil Fp1 eq 11.8 10.9–12.9 11.5 10.7–13.1 11.5 10.8–12.8 0.874 0.516 1.0 Fp2 eq 11.8 10.9–12.3 11.5 10.8–13.1 11.5 10.8–12.8 1.0 1.0 0.013 Fp 1 eq 11.8 11.0–12.5 11.5 10.9–13.0 11.4 10.9–12.8 0.893 0.272 0.042 Bil C4 eq 12.9 11.2–14.4 13.1 11.0–14.6 12.6 10.7–14.0 1.0 0.703 0.006 Bil C3 eq 12.9 11.1–14.3 13.1 11.3–14.9 12.8 10.9–14.4 0.670 1.0 0.007 C4 eq 12.5 11.2–13.7 12.5 11.0–14.2 11.9 11.0–13.8 1.0 0.283 0.043 C3 eq 12.5 11.1–13.8 12.8 11.3–15.1 11.7 11.0–13.9 1.0 0.207 0.001 Sequences % Fp sequences 0.0 0.0–0.4 0.0 0.0–1.1 0.0 0.0–0.3 1.0 1.0 0.999 C sequences 0.3 0.0–2.4 0.1 0.0 4.4 0.0–3.1 0.720 0.176 0.224 Bila e ali y indices Fp bil index 1 0.48 0.21–0.98 0.36 0.02–0.81 0.44 0.07–0.68 0.185 0.388 1.0 Fp bil index 2 0.54 0.21–0.89 0.36 0.04–1.18 0.44 0.06–0.95 0.388 0.275 1.0 C bil index 1 0.64 0.20–1.31 0.40 0.03–1.16 0.38 0.08–1.12 0.297 0.085 1.0 C bil index 2 0.50 0.14–1.13 0.35 0.06–1.0 0.33 0.05–0.98 0.611 0.145 1.0 1Mann-Whi ney 𝑈 es ; 2Wilcoxon es . The Bon e oni co ec ed 𝑝 alues a e epo ed: Bon e oni co ec ion ac o = 3. Bil Fp/bil C = bila e al on opola /cen al spindle; bil C4, bil C3 = bila e al igh -side and le -side spindles, espec i ely; Fp2, Fp1, C4, and C3 =unila e al spindles; sequences = pe cen age o ime e e ed o as TST occupied by pe iodic on opola o cen al spindles; Fp bil index 1 = numbe o bil Fp spindles di ided by he numbe o unila Fp2 spindles (bil Fp/Fp2); Fp bil index 2 = numbe o bil Fp spindles di ided by he numbe o unila Fp1 spindles (bil Fp/Fp1); C bil index 1 = numbe o bil C spindles di ided by he numbe o unila C4 spindles (bil C/C4); C bil index 2 = numbe o bil C spindles di ided by he numbe o unila C3 spindles (bil C/C3). pa o he nigh in all g oups and no g oup-wise di e ences we e ound. Hemisphe ic compa isons e ealed ha heal hy con ols had mo e unila e al cen al spindles in he le han in he igh hemisphe e. The p e-CPAP g oup p esen ed mo e on opola spindles in he le han in he igh hemisphe e. O he hemisphe ic spindle densi y di e ences we e no p esen (Table 3). Rega ding he in aspindle equency, all cen al spindle ypes (bil C4, bil C3, C4, and C3) we e as e han he co esponding on opola spindles (bil Fp2, bil Fp1, Fp2, and Fp1, esp.) in all g oups (𝑝 alues in Table 4). The equency o he on opola o cen al spindles o he con ol g oup did no di e om he alues o he p e-CPAP o CPAP g oups (Table 2). The p e-CPAP g oup had as e Fp2, Fp1, bil C4, bil C3, C4, and C3 spindle equencies han he CPAP g oup. Heal hy con ols did no ha e hemisphe ic spindle equency di e ences (Table 3). Ins ead, bo h he p e-CPAP and CPAP g oups p esen ed a mino bu signi ican di e ence; unila e al on opola spindles we e as e in he igh hemisphe e han in he le . 3.2. Spindle Fea u es in he Las Pa o he NREM Time. In he las pa o he nigh , he con ol g oup’s bil C and C3 spindle densi ies we e highe han he densi ies in he p e- CPAP g oup (Table 5). The con ol g oup had mo e bil C spindles han heCPAPg oup.Thep e-CPAPg ouphad mo e Fp1 and C4 spindles han he CPAP g oup. The cen al spindles o he con ol g oup we e mo e o en bila e al han in he wo pa ien g oups, as assessed by he bila e ali y indices. No s a is ical di e ences be ween he g oups we e ound in he spindle sequence imes. Heal hy con ols did no p esen in e hemisphe ic spindle densi y di e ences, bu bo h pa ien g oupshadmo e on opola spindlesin he igh hemisphe e han in he le (Table 3). As in he i s pa o he NREM ime, all cen al spindle ypes we e as e han he espec i e on opola spindles Sleep Diso de s 5 Table 3: 𝑝 alues o he hemisphe ic igh -le compa isons o he bila e al and unila e al spindle equencies and unila e al densi ies in he di e en s udy g oups. Con ols P e-CPAP CPAP Fi s pa o heNREM ime Densi y Fp2 e sus Fp1 0.575 0.044 0.550 C4 e sus C3 0.030 0.737 0.156 F equency Bil Fp2 e sus bil Fp1 0.717 0.809 0.070 Fp2 e sus Fp1 0.126 0.035 0.014 Bil C4 e sus bil C3 0.588 0.235 0.064 C4 e sus C3 0.173 0.145 0.794 Las pa o he NREM ime Densi y Fp2 e sus Fp1 0.601 0.006 0.048 C4 e sus C3 0.469 0.478 0.502 F equency Bil Fp2 e sus bil Fp1 0.717 0.658 0.225 Fp2 e sus Fp1 0.444 0.225 0.313 Bil C4 e sus bil C3 0.444 0.097 0.896 C4 e sus C3 0.365 0.082 0.911 Numbe s e e o 𝑝 alues based on he Wilcoxon es . Table 4: 𝑝 alues o he compa isons be ween he cen al and on opola spindle equencies in he di e en s udy g oups. Con ols P e-CPAP CPAP Fi s pa o he NREM ime Bil C4 e sus bil Fp2 <0.001 <0.001 0.003 Bil C3 e sus bil Fp1 <0.001 <0.001 0.003 C4 e sus Fp2 <0.001 0.001 0.022 C3 e sus Fp1 <0.001 <0.001 0.002 Las pa o he NREM ime Bil C4 e sus bil Fp2 <0.001 0.002 <0.001 Bil C3 e sus bil Fp1 <0.001 <0.001 <0.001 C4 e sus Fp2 <0.001 0.001 0.001 C3 e sus Fp1 <0.001 <0.001 <0.001 Numbe s e e o 𝑝 alues based on he Wilcoxon es . (Table 4). The spindle equencies o he con ol g oup did no di e om he equencies o he p e-CPAP o CPAP g oups (Table 5). The CPAP g oup had slowe spindles han he p e- CPAP g oup wi h he excep ion o he equency o he bil Fp1 spindles. Hemisphe ic spindle equency di e ences we e no oundinanyo heg oups(Table3). 3.3. Compa ison o he Spindle Fea u es be ween he Fi s and he Las Pa o he NREM Time. None o he on opola spindle pa ame e s (densi y, bila e ali y index, sequences, and equency) showed any s a is ical di e ences be ween he i s and he las pa o he nigh in any o he s udy g oups. In he con ol g oup and he CPAP g oup, he densi y o he bila e al cen al spindles inc eased owa ds he end o he Bil C C4 C3 Bil C C4 C3 Bil C C4 C3 Fi s pa Las pa Con ols P e-CPAP CPAP p = 0.002 p = 0.028 0 20 40 60 80 100 120 Numbe /hou ∗ ∗ Figu e 1: Cen al spindle densi ies in he i s and he las pa o he NREM sleep in he s udy g oups. The s a is ically signi ican di e ences a e ma ked wi h an as e isk and deno ed wi h he 𝑝 alues. Con ols P e- CPAP CPAP Con ols P e- CPAP CPAP Fi s pa Las pa Bila e ali y index 1 Bila e ali y index 2 p = 0.001 p < 0.001 p = 0.007 0 0.2 0.4 0.6 0.8 1 ∗ ∗ ∗ Figu e 2: Bila e ali y indices o he cen al spindles in he i s and las pa o he NREM sleep. The s a is ically signi ican di e ences a e ma ked wi h an as e isk and deno ed wi h he 𝑝 alues. nigh (Figu e 1). Ins ead, unila e al cen al spindle densi ies did no show any ma ked changes du ing he nigh in any o he g oups. In he con ol g oup, he p opo ion o bila e al cen al spindles washighe in he las pa o he nigh han in he i s (Figu e 2). In he CPAP g oup, bila e ali y index 2 showed an inc ease in la e sleep. The p e-CPAP g oup did no p esen signi ican changes. In hecon ols, he imeoccupiedwi hcen alspindle sequences inc eased owa ds he end o he nigh , whe eas he sequence ime o he pa ien g oups did no change du ing he nigh (Figu e 3). In he con ols, all cen al spindle ypes we e as e in he las pa o he nigh han in he i s pa (Figu e 4). The spindle equencies o he p e-CPAP pa ien s did no change ma kedly du ing he nigh , while bo h bila e al and unila e al C4 spindles became as e in he CPAP g oup. 6Sleep Diso de s Table 5: Median and ange o he spindle ea u es in he s udy g oups a he las pa o he NREM ime. Con ols P e-CPAP CPAP Con ols e sus p e-CPAP1Con ols e sus CPAP1P e-CPAP e sus CPAP2 Median Range Median Range Median Range 𝑝 alue 𝑝 alue 𝑝 alue Densi y 𝑛/h Bil Fp dens 43.7 7.3–110.3 23.2 0.5–105.7 23.6 1.1–102.6 0.350 0.533 1.0 Fp2 dens 85.1 26.6–136.9 52.5 10.6–157-7 54.4 10.1–159.3 0.768 1.0 0.741 Fp1 dens 65.6 15.9–150.5 43.8 7.1–141.3 46.4 12.7–129.0 0.223 0.518 0.033 Bil C dens 76.5 18.3–215.7 32.6 0.5–282.5 35.4 4.4–282.0 0.010 0.018 0.436 C4 dens 111.5 63.8–183.5 78.2 21.6–196.8 90.1 18.6–191.8 0.223 0.626 0.027 C3 dens 117.1 46.4–229.7 64.3 20.2–231.6 85.4 26.9–217.7 0.033 0.093 0.238 F equency Hz Bil Fp2 eq 11.7 11.1–12.4 11.6 11.0–13.6 11.3 10.8–13.0 1.0 0.074 0.001 Bil Fp1 eq 11.8 11.1–12.4 11.6 10.8–13.3 11.4 10.7–13.0 1.0 0.243 0.131 Fp2 eq 11.7 11.2–12.4 11.7 10.9–13.0 11.5 11.0–12.8 1.0 0.395 0.022 Fp 1 eq 11.9 11.2–12.4 11.6 11.1–13.0 11.4 11.0–12.9 0.640 0.159 0.037 Bil C4 eq 13.1 11.4–14.3 13.3 10.8–14.9 13.0 10.9–14.6 1.0 1.0 0.007 Bil C3 eq 13.1 11.2–14.3 13.3 10.8–14.8 12.8 10.9–14.8 0.912 1.0 0.004 C4 eq 12.9 11.2–14.1 12.6 11.0–14.5 12.5 10.9–13.9 1.0 1.0 0.041 C3 eq 12.8 11.1–14.3 13.0 10.9–15.0 12.4 10.9–13.9 1.0 1.0 0.005 Sequences % Fp sequences 0.0 0.0–0.5 0.0 0.0–0.5 0.0 0.0–0.5 1.355 1.0 0.999 C sequences 0.6 0.0–4.0 0.1 0.0–7.3 0.0 0.0–7.1 0.086 0.097 1.0 Bila e ali y indices Fp bil index 1 0.47 0.15–1.22 0.36 0.04–0.73 0.43 0.11–0.75 0.168 0.388 1.0 Fp bil index 2 0.61 0.15–1.07 0.48 0.04–1.15 0.46 0.09–1.05 0.506 0.191 1.0 C bil index 1 0.79 0.17–1.55 0.36 0.03–1.44 0.42 0.14–1.47 0.006 0.008 1.0 C bil index 2 0.72 0.20–1.38 0.45 0.03–1.22 0.36 0.15–1.30 0.015 0.009 1.0 1Mann-Whi ney 𝑈 es ; 2Wilcoxon es . The Bon e oni co ec ed 𝑝 alues a e epo ed: Bon e oni co ec ion ac o = 3. Abb e ia ionsasinTable2. 4. Discussion The indings o ou s udy ha e alua ed sleep spindles a e six mon hs o CPAP ea men e ealed ha many spindle ea u es we e no malized by CPAP ea men , bu some de ici s emained. P e ious s udies ha e shown an inc ease in spindle densi y ac oss consecu i e NREM sleep episodes in heal hy subjec s [20, 27–29]. The p esen esul s sugges ha he densi y inc ease is mos ly induced by he cen al, as e spindlesbu no by on opola spindles.Thisisinlinewi h he indings o Sch¨ onwald and cowo ke s [18]. In con as o heal hy con ols, ou un ea ed OSA pa ien s did no p esen any signi ican inc ease in cen al spindle densi y du ing sleep, bu wi h CPAP ea men he inc ease was eco e ed. The inc ease o cen al spindles in heal hy con ols and CPAP- ea ed pa ien s was es ic ed o bila e al cen al spindles only. The densi y o local cen al spindles did no show any ma ked changes du ing sleep in any o he g oups. Acco ding o he ea lie s udies, local spindles a e equen and e en mo e equen han global spindles [30, 31]. Ou esul s a e in line wi h hese indings; bila e ali y indices show ha he numbe o unila e al on opola and cen al spindles is highe han bila e al spindles in all g oups. Howe e , spindles a e ound o become mo e global in la e sleep in a small g oup o neu osu gical pa ien s [30, 31]. Ou inding ha he p opo ion o bila e al cen al spindles inc eased in la e sleep in he heal hy con ols co esponds wi h his inding. In OSA pa ien s be o e CPAP ea men , howe e , no inc ease was ob ained. CPAP ea men pa ly no malized his phenomenon as he sha e o cen al bila e al spindles inc eased when compa ed wi h le -sided unila e al spindles, bu no changes we e ound in he igh hemisphe e. Spindles a e abundan in a mode a e le el o co ico ha- lamic hype pola iza ion, in ligh sleep [32, 33]. Wi h u he hype pola iza ion, in deep sleep, he e a e only a ew spindles [32, 34–36]. Co ically gene a ed slow oscilla ion seems o be he p incipal p ocess behind NREM sleep [32, 37–39] by sp eading o he halamus in synch onous olleys and d i ing spindles o he co ex h ough he halamoco ical ne wo k [40, 41]. The e o e, we wonde whe he he de ec in he inc easing bila e ali y o spindles in OSA would be a consequence o dis up ed co ico halamic spindle d i ing loops. As pe iodic spindles e lec he pe iodically ecu ing Sleep Diso de s 7 Con ols P e-CPAP CPAP Fi s pa Las pa p = 0.004 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 (%) ∗ Figu e 3: Pe cen age o ime occupied wi h cen al spindle sequences in he i s and las pa o he NREM sleep. The s a is ically signi ican di e ences a e ma ked wi h as e isk and deno ed wi h he 𝑝 alues. Bil C4 Bil C3 C4 C3 Bil C4 Bil C3 C4 C3 Bil C4 Bil C3 C4 C3 p = 0.001 p = 0.012 p = 0.022 p = 0.024 p = 0.033 p = 0.007 Fi s pa Las pa 0 2 4 6 8 10 12 14 (Hz) Con ols P e-CPAP CPAP ∗ ∗∗ ∗ ∗∗ Figu e 4: F equency o he cen al spindles in he i s and las pa o he NREM sleep in he s udy g oups. The s a is ically signi ican di e ences a e ma ked wi h an as e isk and deno ed wi h he 𝑝 alues. spindle d i ing slow oscilla ion du ing ligh sleep [39], ou inding ha spindle sequences we e no ound in OSA pa ien s suppo s his hough . CPAP ea men did no ha e an e ec on he occu ence o spindle sequences. This migh implica e a mo e pe manen de ec in he co ico halamic ne wo k ha dis u bs he spindle p ocess. On he o he hand, spindle densi y has been ound o inc easelocallyin hosea easo heb ain ha ha ebeen ained du ing wake ulness in a use-dependen manne [16, 17]. In he p esen s udy, he i s pa o he NREM ime did no e eal any spindle densi y di e ences be ween he g oups, bu local spindles in he le cen al a ea (unila e al C3) we e mo e abundan han local spindles in he igh cen al a ea (unila e al C4) in he con ols. Pa ien s wi h o wi hou CPAP did no p esen hemisphe ic di e ences in he unila e al cen al spindle densi ies. In he las pa o he NREM ime, he densi y o unila e al le -side cen al spindles was highe in he con ol g oup han in he OSA pa ien s be o e ea men . I is possible ha he le -side abundance in cen al spindles in heal hy con ols s ems om he inding ha he le hemisphe e needs mo e sleep han he igh hemisphe e, possibly due o di e ences in day ime ac i i y [42]. I unila e al spindle densi y is conside ed use- dependen , hen i can be specula ed ha ou con ol subjec s had used hei le cen al co ical a eas mo e han hei igh ones and ha he OSA pa ien s had used hei le cen al b ain a eas less ac i ely han he con ols. As expec ed, all on opola spindle ypes we e slowe han he co esponding cen al spindles. The equency o he on opola spindles did no change du ing he nigh in any o he g oups. Ins ead, all cen al spindle ypes became as e owa ds he mo ning hou s in he con ols, whe eas no equency changes we e obse ed in he newly diagnosed OSA pa ien s. CPAP ea men pa ly no malized he spindle equencies as bila e al and unila e al igh -side cen al spindles became as e . Howe e , he equencies o he le -sided cen al spindles did no change. The occu ence o spindles has been p o en o be depen- den on he le el o he hype pola iza ion in he halamo- co ical ne wo k, whe eas he equency o spindles has been ound o depend on he du a ion o he hype pola iza ion- ebound sequences o he halamoco ical cells [43]. I ebound sequences a e long, he spindle is slow. In heal hy subjec s, hese wo p ocesses in ol ed in de e mining spindle ea u es seem o co ela e because, in gene al, spindles a e abundan and as inligh sleepand ewandslowindeep sleep [9, 33]. Howe e , hese p ocesses a e p esen ed o be dissocia edinOSApa ien ssinceslowspindlesa e ound in ligh sleep [19]. Co esponding esul s we e ound in he p esen s udy as spindles o he p e-CPAP g oup did no become as e in la e sleep. In addi ion, he p esen indings sugges ha his dissocia ion is pa ly dissol ed wi h CPAP ea men because only le -sided cen al spindles did no become as e in la e sleep wi h CPAP ea men . A somewha su p ising inding was ha mos spindle equencies we e slowe wi h CPAP han wi hou ea men in OSA pa ien s in bo h he i s and he las pa o he nigh . Ano he in e es ing inding was ha spindle sequences we e no eco e ed wi h CPAP ea men . Howe e , OSA pa ien s do no usually ha e deep sleep, which means ha halamoco ical hype pola iza ion s ays a mode a e le els. Wi h CPAP, he pa ien s each slow wa e sleep wi h mo e neg- a i e memb ane po en ials o he halamoco ical ne wo k. This migh explain he eme gence o slow spindles [9] and a dec ease in he numbe o pe iodic spindles [10]. In his way, he inc ease in hype pola iza ion would explain bo h hese indings. In addi ion, i migh be ha 6 mon hs o CPAP ea men does no abolish he inc eased sleep p essu e o OSA pa ien s p ope ly and inc eased sleep p essu e inc eases he hype pola iza ion le el esul ing in slowe spindles wi h no pe iodici y. The inc ease in spindle densi y has been ound o be associa ed wi h memo y unc ion imp o emen , and i is only he as spindles ha ha e been associa ed wi h memo y and lea ning p ocessing [14, 44–48]. The e o e, we wonde whe he he changes in spindle dynamics migh be ela ed o he di e en symp oms o OSA pa ien s such as di e en cogni i e p oblems [49, 50]. CPAP ea men is known o, 8Sleep Diso de s a leas pa ially, educe di e en symp oms in OSA pa ien s [21, 51], and he ac ha many spindle ea u es no malized wi h CPAP is in line wi h ou hypo hesis. The numbe and equency o spindles a e dependen on many ac o s ( o e iew see [33]). Spindles and SWA ha e a ecip ocal ela ionship [52], and sleep dep i a ion dec eases spindling [53]. In addi ion, age, gende and ci cadian hy hm all a ec spindle ac i i y [27, 54–56]. In he p esen s udy, we s udied g oup-wise di e ences be ween heal hy con ols and OSA pa ien s wi h and wi hou CPAP ea men , bu mo e impo an ly we used a wi hin-subjec design o OSA pa ien s wi hou and wi h CPAP ea men o e alua e spindle ea u es.This,inpa , educed hee ec so heabo e men ioned ac o s ha a ec he spindle p ocess. A clea limi a ion o ou s udy is ha he assessmen o di e en cogni i e domains was no included and u u e s udy is needed o e alua e he e ec o CPAP ea men on bo h spindle ea u es and di e en cogni i e skills. To conclude, CPAP ea men only pa ially no malizes sleep spindle ea u es. Fu he s udies a e he e o e needed o e alua e whe he hese spindle p ocess changes a e ela ed o he emaining cogni i e p oblems o OSA pa ien s. Addi ional Poin s Highligh s.(i)Obs uc i esleepapneadis o ssleepspindle p ocess. (ii) CPAP ea men no malizes many spindle ea- u es bu some de ici s emain. (iii) The emaining spindle changes migh be ela ed o cogni i e diso de s o CPAP- ea ed pa ien s. Compe ing In e es s None o he au ho s ha e con lic o in e es s o be disclosed. Au ho s’ Con ibu ions Saunam¨ aki Tiia execu ed he p ojec , analyzed da a, in e - p e edda a,d a ed hemanusc ip ,andapp o ed he e sion o be published. Huupponen Ee o de eloped he sleep spindle de ec o , analyzed spindles in he p esen wo k, w o e spindle analysis me hods, and app o ed he e sion o be published. Loponen Juho pe o med p elimina y s a is ical analyses, analyzed da a, in e p e ed da a, d a ed he manusc ip , and app o ed he e sion o be published. Himanen Sa i- Leena analyzed polysomnog aphies, calcula ed s a is ics, in e p e ed da a, d a ed he manusc ip , and app o ed he e sion obepublished. Acknowledgmen s This s udy was inancially suppo ed by Compe i i e S a e Resea ch Financing o he Expe Responsibili y A ea o Tampe e Uni e si y Hospi al (G an s 9P013, 9R007, and 9S007). Re e ences [1] M.H.Silbe ,S.Ancoli-Is ael,M.H.Bonne e al.,“The isual sco ing o sleep in adul s,” Jou nal o Clinical Sleep Medicine, ol. 3, no. 2, pp. 121–131, 2007. [2] D. Aeschbach, D.-J. Dijk, and A. A. Bo b´ ely, “Dynamics o EEG spindle equency ac i i y du ing ex ended sleep in humans: ela ionship o slow-wa e ac i i y and ime o day,” B ain Resea ch, ol.748,no.1-2,pp.131–136,1997. [3] E. Huupponen, A. Kulkas, M. Tenhunen, A. Saas amoinen, J. Hasan, and S.-L. Himanen, “Di use sleep spindles show simila equency in cen al and on opola posi ions,” Jou nal o Neu oscience Me hods, ol.172,no.1,pp.54–59,2008. [4] E.We h,P.Ache mann,D.-J.Dijk,andA.A.Bo b ´ ely, “Spindle equency ac i i y in he sleep EEG: indi idual di e ences and opog aphic dis ibu ion,” Elec oencephalog aphy and Clinical Neu ophysiology, ol.103,no.5,pp.535–542,1997. [5] J. Zei lho e , G. G ube , P. Ande e , S. Asenbaum, P. Schimicek, and B. Sale u, “Topog aphic dis ibu ion o sleep spindles in young heal hy subjec s,” Jou nal o Sleep Resea ch, ol.6,no.3, pp.149–155,1997. [6] J. Zygie ewicz, K. J. Blinowska, P. J. Du ka, W. Szelenbe ge , S. Niemcewicz, and W. And osiuk, “High esolu ion s udy o sleep spindles,” Clinical Neu ophysiology, ol.110,no.12,pp.2136– 2147, 1999. [7] M. Jobe , E. Poiseau, P. J¨ ahnig, H. Schulz, and S. Kubicki, “Topog aphical analysis o sleep spindle ac i i y,” Neu opsy- chobiology, ol.26,no.4,pp.210–217,1992. [8] P.Ande e ,G.Kl ¨ osch,G.G ube e al.,“Low- esolu ionb ain elec omagne ic omog aphy e ealed simul aneously ac i e on al and pa ie al sleep spindle sou ces in he human co ex,” Neu oscience, ol.103,no.3,pp.581–592,2001. [9] S.-L. Himanen, J. Vi kkala, H. Huh ala, and J. Hasan, “Spindle equencies in sleep EEG show U-shape wi hin i s ou NREM sleep episodes,” Jou nal o Sleep Resea ch, ol.11,no.1,pp.35–42, 2002. [10] S.-L. Himanen, J. Vi kkala, E. Huupponen, J. Niemi, and J. Hasan, “Occu ence o pe iodic sleep spindles wi hin and ac oss non-REM sleep episodes,” Neu opsychobiology, ol.48,no.4, pp.209–216,2003. [11] M. S e iade, D. A. McCo mick, and T. J. Sejnowski, “Thalamo- co ical oscilla ions in he sleeping and a oused b ain,” Science, ol. 262, no. 5134, pp. 679–685, 1993. [12] J.Bo n,B.Rasch,andS.Gais,“Sleep o emembe ,”Neu oscien- is , ol. 12, no. 5, pp. 410–424, 2006. [13] Y. Dan and M.-M. Poo, “Spike iming-dependen plas ici y o neu al ci cui s,” Neu on, ol.44,no.1,pp.23–30,2004. [14] M. M¨ olle,T.O.Be gmann,L.Ma shall,andJ.Bo n,“Fas and slow spindles du ing he sleep slow oscilla ion: dispa a e coalescence and engagemen in memo y p ocessing,” Sleep, ol. 34, no. 10, pp. 1411–1421, 2011. [15]M.Schabus,T.T.Dang-Vu,G.Albouye al.,“Hemodynamic ce eb al co ela es o sleep spindles du ing human non- apid eye mo emen sleep,” P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o Ame ica, ol.104,no.32,pp. 13164–13169, 2007. [16] M. Nishida and M. P. Walke , “Day ime naps, mo o memo y consolida ion and egionally speci ic sleep spindles,” PLoS ONE, ol.2,no.4,a iclee341,2007. [17] L. A. Johnson, T. Blakely, D. He mes, S. Hakimian, N. F. Ramsey, and J. G. Ojemann, “Sleep spindles a e locally modula ed by Sleep Diso de s 9 aining on a b ain-compu e in e ace,” P oceedings o he Na ional Academy o Sciences o he Uni ed S a es o Ame ica, ol. 109, no. 45, pp. 18583–18588, 2012. [18] S. V. Sch¨ onwald, D. Z. Ca alho, E. L. de San a-Helena, N. Lemke, and G. J. L Ge ha d , “Topog aphy-speci ic spindle e- quency changes in obs uc i e sleep apnea,” BMC Neu oscience, ol. 13, no. 1, a icle 89, 2012. [19] S.-L. Himanen, J. Vi kkala, E. Huupponen, and J. Hasan, “Spin- dle equency emains slow in sleep apnea pa ien s h oughou he nigh ,” Sleep Medicine, ol.4,no.3,pp.229–234,2003. [20] B. Ondze, F. Espa, Y. Dau illie s, M. Billia d, and A. Besse , “Sleep a chi ec u e, slow wa e ac i i y and sleep spindles in mild sleep diso de ed b ea hing,” Clinical Neu ophysiology, ol.114, no. 5, pp. 867–874, 2003. [21] A. I. S´ anchez, P. Ma ´ ınez, E. Mi ´ o, W. A. Ba dwell, and G. Buela-Casal, “CPAP and beha io al he apies in pa ien s wi h obs uc i e sleep apnea: e ec s on day ime sleepiness, mood, and cogni i e unc ion,” Sleep Medicine Re iews, ol.13,no.3, pp. 223–233, 2009. [22] T. Saunam¨ aki,M.Jehkonen,E.Huupponen,O.Polo,andS.-L. Himanen, “Visual dys unc ion and compu a ional sleep dep h changes in obs uc i e sleep apnea synd ome,” Clinical EEG and Neu oscience, ol.40,no.3,pp.162–167,2009. [23]C.Ibe ,S.Ancoli-Is ael,A.L.Chesson,andS.F.Quan, The AASM Manual o he Sco ing o Sleep and Associa ed E en s: Rules, Te minology and Technical Speci ica ions,Ame - ican Academy o Sleep Medicine, Wes ches e , Ill, USA, 1s edi ion, 2007. [24] R. B. Be y, R. Budhi aja, D. J. Go lieb e al., “Rules o sco ing espi a o y e en s in sleep: upda e o he 2007 AASM manual o hesco ingo sleepandassocia ede en s,”Jou nal o Clinical Sleep Medicine, ol.8,no.5,pp.597–619,2012. [25] AASM, “EEG a ousals: sco ing ules and examples: a p elim- ina y epo om he Sleep Diso de s A las Task Fo ce o he Ame ican Sleep Diso de s Associa ion,” Sleep, ol.15,no.2,pp. 173–184, 1992. [26] E. Huupponen, G. G´ omez-He e o, A. Saas amoinen, A. V¨ a i, J. Hasan, and S.-L. Himanen, “De elopmen and compa ison o ou sleep spindle de ec ion me hods,” A i icial In elligence in Medicine, ol.40,no.3,pp.157–170,2007. [27] M. Guazzelli, I. Feinbe g, M. Amino , G. Fein, T. C. Floyd, and C. Maggini, “Sleep spindles in no mal elde ly: compa ison wi h young adul pa e ns and ela ion o noc u nal awakening, cogni i e unc ion and b ain a ophy,” Elec oencephalog aphy and Clinical Neu ophysiology, ol.63,no.6,pp.526–539,1986. [28] D.-J. Dijk, B. Hayes, and C. A. Czeisle , “Dynamics o elec oen- cephalog aphic sleep spindles and slow wa e ac i i y in men: e ec o sleep dep i a ion,” B ain Resea ch, ol.626,no.1-2,pp. 190–199, 1993. [29] L. De Genna o, M. Fe a a, and M. Be ini, “Topog aphical dis ibu ion o spindles: a ia ions be ween and wi hin NREM sleep cycles,” Sleep Resea ch Online, ol.3,no.4,pp.155–160, 2000. [30] Y. Ni , R. J. S aba, T. And illon e al., “Regional slow wa es and spindles in human sleep,” Neu on, ol.70,no.1,pp.153–169,2011. [31] T. And illon, Y. Ni , R. J. S aba e al., “Sleep spindles in humans: insigh s om in ac anial EEG and uni eco dings,” Jou nal o Neu oscience, ol. 31, no. 49, pp. 17821–17834, 2011. [32] F. Amzica and M. S e iade, “Elec ophysiological co ela es o sleep del a wa es,” Elec oencephalog aphy and Clinical Neu o- physiology, ol.107,no.2,pp.69–83,1998. [33] L. De Genna o and M. Fe a a, “Sleep spindles: an o e iew,” Sleep Medicine Re iews, ol.7,no.5,pp.423–440,2003. [34] D. A. McCo mick and H. C. Pape, “P ope ies o a hype pola iza ion-ac i a ed ca ion cu en and i s ole in hy hmic oscilla ion in halamic elay neu ones,” The Jou nal o Physiology, ol.431,no.1,pp.291–318,1990. [35] N. Le esche, S. Ligh owle , I. Sol esz, D. Jassik−Ge schen eld, and V. C unelli, “Low- equency oscilla o y ac i i ies in insic o a and ca halamoco ical cells,” The Jou nal o Physiology, ol. 441, no. 1, pp. 155–174, 1991. [36] M. S e iade, R. C. Dossi, and A. Nunez, “Ne wo k modula ion o a slow in insic oscilla ion o ca halamoco ical neu ons implica ed in sleep del a wa es: co ically induced synch o- niza ion and b ains em choline gic supp ession,” The Jou nal o Neu oscience, ol. 11, no. 10, pp. 3200–3217, 1991. [37] M. S e iade, A. Nunez, and F. Amzica, “A no el slow (<1Hz) oscilla ion o neoco ical neu ons in i o: depola izing and hype pola izing componen s,” Jou nal o Neu oscience, ol.13, no. 8, pp. 3252–3265, 1993. [38] P. Ache mann and A. A. Bo b´ ely, “Low- equency (<1Hz) oscilla ions in he human sleep elec oencephalog am,” Neu o- science, ol.81,no.1,pp.213–222,1997. [39] M. S e iade and F. Amzica, “Coalescence o sleep hy hms and hei ch onology in co ico halamic ne wo ks,” Sleep Resea ch Online, ol.1,no.1,pp.1–10,1998. [40] M. S e iade, P. Gloo , R. R. Llinas, F. H. Lopes de Sil a, and M. M. Mesulam, “Repo o IFCN commi ee on basic mechanisms. Basic mechanisms o ce eb al hy hmic ac i i ies,” Elec oencephalog aphy and Clinical Neu ophysiology, ol.76, no. 6, pp. 481–508, 1990. [41] D. Con e as and M. S e iade, “Spindle oscilla ion in ca s: he ole o co ico halamic eedback in a halamically gene a ed hy hm,” Jou nal o Physiology, ol.490,no.1,pp.159–179,1996. [42] P. Ache mann, L. A. Finelli, and A. A. Bo b´ ely, “Unihemisphe ic enhancemen o del a powe in human on al sleep EEG by p olonged wake ulness,” B ain Resea ch, ol.913,no.2,pp.220– 223, 2001. [43] M. S e iade and R. R. Llinas, “The unc ional s a es o he halamus and he associa ed neu onal in e play,” Physiological Re iews, ol. 68, no. 3, pp. 649–742, 1988. [44] M. Tamaki, T. Ma suoka, H. Ni ono, and T. Ho i, “Fas sleep spindle (13-15 Hz) ac i i y co ela es wi h sleep-dependen imp o emen in isuomo o pe o mance,” Sleep, ol.31,no.2, pp. 204–211, 2008. [45] M. Ba aka , J. Doyon, K. Debas e al., “Fas and slow spindle in ol emen in he consolida ion o a new mo o sequence,” Beha iou al B ain Resea ch, ol. 217, no. 1, pp. 117–121, 2011. [46] B. A. Mande , S. San hanam, J. M. Sale in, and M. P. Walke , “Wake de e io a ion and sleep es o a ion o human lea ning,” Cu en Biology, ol. 21, no. 5, pp. R183–R184, 2011. [47] K. C. And ade, V. I. Spoo make , M. D esle e al., “Sleep spindles and hippocampal unc ional connec i i y in human NREM sleep,” Jou nal o Neu oscience, ol.31,no.28,pp.10331– 10339, 2011. [48] A. G. Siapas and M. A. Wilson, “Coo dina ed in e ac ions be ween hippocampal ipples and co ical spindles du ing slow- wa e sleep,” Neu on, ol. 21, no. 5, pp. 1123–1128, 1998. [49] T. Saunam¨ aki and M. Jehkonen, “A e iew o execu i e unc- ions in obs uc i e sleep apnea synd ome,” Ac a Neu ologica Scandina ica, ol. 115, no. 1, pp. 1–11, 2007.