Sleep habits, academic performance, and the adolescent brain structure

1 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 www.na u e.com/scien i ic epo s Sleep habi s, academic pe o mance, and he adolescen b ain s uc u e Anna S. U ila1,2,3, E ic A iges1,4, Jessica Massico e1, Ruben Mi anda1, Hélène Vulse 1, Pauline Bézi in-F e e1, Winok Lapidai e1, He é Lemaî e1, Jani Pen ilä1,5, Pa icia J. Con od6,7, Hugh Ga a an8, Ma ie-Lau e Paillè e Ma ino 1,9, Jean-Luc Ma ino 1,10 & he IMAGEN conso ium† , He e we epo he i s and mos obus e idence abou how sleep habi s a e associa ed wi h egional b ain g ey ma e olumes and school g ade a e age in ea ly adolescence. Sho e ime in bed du ing weekdays, and la e weekend sleeping hou s co ela e wi h smalle b ain g ey ma e olumes in on al, an e io cingula e, and p ecuneus co ex egions. Poo school g ade a e age associa es wi h la e weekend bed ime and smalle g ey ma e olumes in medial b ain egions. The medial p e on al - an e io cingula e co ex appea s mos igh ly ela ed o he adolescen s’ a ia ions in sleep habi s, as i s olume co ela es in e sely wi h bo h weekend bed ime and wake up ime, and also wi h poo school pe o mance. These indings sugges ha sleep habi s, no ably du ing he weekends, ha e an ala ming link wi h bo h he s uc u e o he adolescen b ain and school pe o mance, and hus highligh he need o in o med in e en ions. Adolescen s oday sleep less and expe ience mo e day ime sleepiness symp oms as compa ed o p e ious gen- e a ions1,2. Du ing adolescence, sleep unde goes majo changes: sleep du a ion and dep h dec ease, and sleep shi s owa ds e ening hou s. A endency owa ds e eningness becomes e iden du ing he adolescen yea s as a esul o in e nal and ex e nal in luences on b ain mechanisms egula ing sleep and ci cadian hy hm3. S aying up la e combined wi h ea ly mo ning awakenings o school easily lead o insu icien sleep and accumula ion o sleep deb du ing he school week. Adolescen s ypically a emp o pay back hei sleep deb du ing weekends, especially by sleeping in on weekend mo nings4. Since he combina ion o delayed bed imes and ea ly school s a imes esul s in sleep deb o a la ge po ion o he adolescen popula ion, he e is an ongoing public deba e on how o a ange school s a ing imes ha would be sui able, applicable, and bene icial o adolescen s’ heal h5,6. E idence om bo h epidemiological and expe imen al sleep es ic ion s udies sugges s ha insu icien and/ o la e iming o sleep po en ially has a la ge spec um o nega i e e ec s on adolescen s’ academic success, heal h, and sa e y7. No su p isingly, bo h sho sleep and la e sleeping hou s ha e been shown o co ela e wi h poo school pe o mance, possibly ia a pa hway in ol ing educed a en ion and inc eased day ime somnolence8–10. Cogni i e p ocesses suppo ed by he ne wo ks associa ed wi h he p e on al co ex, such as a en ion and exec- u i e unc ions, appea especially sensi i e o sleep loss11–12. Despi e he de imen al e ec s o poo and inadequa e sleep on adolescen academic success and heal h, e i- dence on he e ec s o long- e m sleep habi s on he de eloping adolescen s’ b ain s uc u e is s ill lacking. The only published s udy o b ain g ey ma e olumes (GMVs) in ela ion o sleep habi s was pe o med in a mixed 1Ins i u Na ional de la San é e de la Reche che Médicale, INSERM Uni 1000 “Neu oimaging & Psychia y”, Uni e si y Pa is Sud – Uni e si y Pa is Saclay, Uni e si y Pa is Desca es, 97 Bd de Po Royal, 75014, Pa is, F ance. 2Na ional Ins i u e o Heal h and Wel a e, Depa men o Heal h, Uni o Men al Heal h, P.O. Box 30, 00271 Helsinki, Finland. 3Uni e si y o Helsinki and Helsinki Uni e si y Cen al Hospi al, Depa men o Psychia y/Adolescen Psychia y, P.O. Box 803, 00029 HUS, Helsinki, Finland. 4Depa men o Psychia y 91G16, 4 place du Géné al Lecle c, O say Hospi al, GH No d-Essonne, 91400 O say, F ance. 5Adolescen Psychia y Depa men , Medical school, FI-33014 Tampe e Uni e si y, Finland. 6Depa men o Psychia y, Uni e si é de Mon éal, CHU S e Jus ine Hospi al, 175 Chemin de la Cô e-Sain e-Ca he ine, Mon éal, QC H3T 1C4, Canada. 7Depa men o Psychological Medicine and Psychia y, Ins i u e o Psychia y, Psychology & Neu oscience, King’s College, 16 De C espigny Pa k, London SE5 8AF, Uni ed Kingdom. 8Depa men s o Psychia y and Psychology, 6436 UHC, Uni e si y o Ve mon . 1 Sou h P ospec S ee , Bu ling on, VT 05401, USA. 9APHP, Adolescen Psychopa hology and Medicine depa men , Maison de Solenn, Cochin Hospi al, 97 Bd de Po Royal, 75014 Pa is, F ance. 10CENIR a ICM ins i u e, Cen e de Neu oimage ie de Reche che Bâ imen ICM, 47-83 boule a d de l’Hôpi al, 75651 Pa is Cedex 13,ss F ance. †A comp ehensi e lis o conso ium membe s appea s a he end o he pape . Co espondence and eques s o ma e ials should be add essed o J.-L.M. (email: [email p o ec ed]) ecei ed: 29 Ap il 2016 Accep ed: 28 Decembe 2016 Published: 09 Feb ua y 2017 OPEN www.na u e.com/scien i ic epo s/ 2 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 sample o child en and adolescen s. I showed ha sleep du a ion du ing weekdays co ela ed wi h GMVs in bila - e al hippocampal and igh do sola e al p e on al co ex (DLPFC)13. Taking in o accoun he adical changes in sleep s uc u e, leng h, and iming du ing adolescen de elopmen , he need o s udying a mo e homogeneous age g oup o adolescen s appea s e iden . Fu he mo e, he e a e o da e no epo s on he associa ions be ween sleep iming and b ain s uc u e in his age g oup. We examined he ela ionship be ween adolescen s’ sleep habi s and b ain g ey ma e olume in a homoge- neous sample o 14–yea -old adolescen s (n = 177) using magne ic esonance imaging (MRI) and oxel-based mo phome y (VBM). The VBM app oach gi es a comp ehensi e assessmen o ana omical olume di e ences h oughou he b ain wi hou bias owa ds any speci ic egion14. Fi s , we hypo hesized ha sho e sleep, la e bed imes (indica ing a endency owa ds e eningness), and sleeping in on weekend mo nings (indica ing accu- mula ion o sleep deb du ing he school week and eco e y sleep du ing weekends) would be associa ed wi h smalle egional GMVs. Based on p e ious esea ch13,15, we expec ed ha he g ea es changes in GMVs would be obse ed in he medial p e on al and an e io cingula e co ex (ACC) as well as he hippocampus. Second, we hypo hesized ha sleep habi s would co ela e wi h school g ade a e age, and, inally, ha smalle GMVs co e- la ing wi h sleep habi s would be also associa ed wi h school g ade a e age. We ound ha , among adolescen s, la e weekend bed imes co ela ed wi h smalle b ain GMVs in on al, an e io cingula e, and p ecuneus co ex egions. La e weekend bed imes we e associa ed wi h poo e school g ade a e age, and bo h o hese we e u he associa ed wi h small GMV in he medial PFC egion. Sho e weekday ime in bed co ela ed wi h smalle GMVs in on al egions. These esul s highligh especially he pos- sible ad e se link o la e iming o weekend sleep wi h he ma u ing adolescen b ain and school pe o mance. Resul s De ails on he sample a e p esen ed in Table1. Sleep habi s and GMVs. GMVs co ela ed wi h sleep habi s in se e al co ical egions (Fig.1, Table2). La e wake up ime on weekends co ela ed wi h smalle GMV in a clus e comp ising he le on al medial o bi al co ex and he le ACC. La e bed ime on weekends co ela ed wi h smalle GMVs in h ee sepa a e clus- e s loca ed in (i) he igh p ecuneus and pa acen al lobule, (ii) he igh middle/supe io on al gy us, and (iii) he igh on al supe io medial co ex and le ACC. Longe ime in bed du ing weekdays co ela ed wi h la ge GMV in a on al clus e comp ising he le supe io and middle on al gy us. No o he co ela ions be ween sleep habi s and GMVs we e obse ed. Sleep habi s and school g ade a e age. The a e age bed imes, wake-up imes, and ime in bed on week- days and on weekends a e p esen ed in Table1. Boys had on a e age la e bed imes on weekends han gi ls (mean bed ime on weekends boys 23:40 ± 1:08 s. gi ls 23:18 ± 00:59; independen samples - es p = 0.020, = − 2.354). No o he s a is ically signi ican gende di e ences we e no ed in sleep habi s. In eg ession analyses con olling o gende (β = 0.570), pube al s age (β = − 0.211), and IQ (β = − 0.061), he e was a ela ionship be ween poo school g ade a e age and la e weekend bed imes (p = 0.00472; β = 0.211; = 2.863). School g ade a e age was no co ela ed wi h any o he sleep habi . Va iable Mean (o %)aSD Demog aphy Age 14.4 0.5 Gende 52.0% (n = 92) emale BMI 20.1 2.7 IQ 109.3 9.0 PDS 2.9 0.6 Sleep habi s Wake up ime WD 7:06 0:23 Wake up ime WE 9:45 1:09 Bed ime WD 22:20 0:44 Bed ime WE 23:29 1:05 Time in bed WD (h) 8.8 0.8 Time in bed WE (h) 10.3 1.2 School pe o mance G ade a e age 2.54 1.22 Global b ain olumes To al GMV 805.2 77.7 To al WMV 452.8 48.8 To al CSF 541.5 95.6 To al olume 1799.49 178.12 Table 1. Cha ac e is ics and b ain olumes in 177 communi y adolescen s. a% ins ead o mean is indica ed o nominal a iables when app op ia e. BMI = body mass index; IQ = in elligence quo ien ; PDS = Pube al De elopmen Scale sco e; WD = weekday; WE = weekend; DAWBA = De elopmen and Well-Being Assessmen ; GMV = g ey ma e olume; WMV = whi e ma e olume; CSF = ce eb ospinal luid. www.na u e.com/scien i ic epo s/ 3 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 School g ade a e age and GMV co ela es. Based on he six oxel-wise co ela ion analyses be ween sleep habi s and GMV, h ee egion-o in e es masks we e iden i ied o he ollowing a iables: (1) wake-up ime du ing weekends (including a eas o he le on al medial o bi al co ex and he le ACC); (2) bed ime du ing weekends (including a eas o he igh p ecuneus and pa acen al lobule, he igh middle and supe io on al gy us, he igh on al supe io medial co ex, and he le ACC); and (3) ime in bed du ing weekdays (including a eas o he le middle and supe io on al gy us). School g ade a e age co ela ed wi h GMV wi hin wo o he h ee applied egion-o -in e es masks in medial b ain egions: (i) wi hin he weekend wake-up ime egion-o -in e es mask, in a clus e o he bila e al an e io cingula e co ex ex ending o he bila e al on al supe io medial co ices, and (ii) wi hin he weekend bed ime egion-o -in e es mask, in h ee clus e s loca ed in he le pa acen al lobule and middle cingula e, he le an e- io cingula e, and he igh an e io cingula e co ex (Table3). No s a is ically signi ican co ela ions be ween school g ade a e age and GMV we e ound wi hin he weekday ime in bed egion-o -in e es mask. Causal media ion analyses showed ha weekend bed ime explained a signi ican ac ion o he ela ionship be ween GMV and school pe o mance (Table4). Weekend bed ime media ed he ela ionship be ween he GMVs wi hin wo o he h ee egions-o -in e es and school g ade a e age. Weekend bed ime accoun ed o 43.8% (p < 0.001) o he o al e ec be ween GMV in he on al supe io medial co ex and ACC, and school pe o mance. Also, i accoun ed o 33.7% (p = 0.03) o he o al e ec be ween GMV in he p ecuneus and pa - acen al lobule, and school pe o mance. No media ion e ec was ound in any o he o he media ion analyses pe o med. Discussion Sho e weekday sleep and la e weekend sleep habi s associa ed wi h smalle b ain egional GMV among 14-yea -old communi y adolescen s. The medial p e on al co ex (PFC) GMV co ela ed bo h wi h bed ime and wake-up imes du ing weekends, and i was also associa ed wi h school g ade a e age. These indings implica e ha he medial PFC, especially he an e io cingula e co ex, migh be he b ain egion ha is mos sensi i e o indi idual di e ences in sleep habi s du ing adolescence. Weekend bed imes, weekend wake-up imes, and weekday ime in bed, which likely ep esen h ee sepa a e bu pa ly o e lapping domains o sleep (as deno ed in Supplemen a yTableS1), had di e en ial co ela ions wi h b ain egional GMV. While weekday ime in bed desc ibes he habi ual sleep amoun , weekend bed ime e lec s a endency owa ds e eningness, and weekend wake-up ime possibly e lec s eco e y sleep due o accumula ion o sleep deb du ing he week3,4. He ein, wake-up ime and bed ime du ing weekdays and ime in bed du ing Figu e 1. B ain egions whe e g ey ma e olumes co ela ed wi h sleep habi s. Red colou and op igh panel illus a e he DLPFC egions whe e GMV was co ela ed wi h ime in bed on weekdays; ligh blue colou and bo om igh panel illus a e he medial p e on al and an e io cingula e egions whe e GMV co ela ed nega i ely wi h wake-up ime on weekends; da k blue colou and bo om le panel illus a e he DLPFC, an e io cingula e and p ecuneus egions whe e GMV was nega i ely co ela ed wi h bed ime on weekends. All images a e p esen ed wi h heigh h eshold p < 0.001, and FWE co ec ion o mul iple spa ial compa isons ac oss he whole b ain (1200 oxels). The sca e plo s a e p esen ed o he oxel o maximal s a is ical signi icance; he solid line ep esen s he linea eg ession line. GMV is exp essed in a bi a y uni s. BA = B odmann a ea. www.na u e.com/scien i ic epo s/ 4 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 weekends we e no co ela ed wi h b ain GMV, which is likely explained by he ac ha school s a ing imes dic a e sleeping imes, pa icula ly wake-up imes, du ing weekdays. Thus, in e -indi idual di e ences in sleep iming become mo e e iden on weekends wi h a ee schedule, while sleep cu ailmen occu s mos commonly du ing weekdays3,4. The mos signi ican esul s we e indeed ela ed o iming o sleep du ing weekends: he la e he adolescen s slep du ing weekends, he smalle we e hei GMVs in a la ge clus e cen e ed in he medial o bi o on al co ex (OFC) and he an e io cingula e co ex (ACC). These medial PFC egions ha e been p e iously implica ed in sleep pa hologies16–19. Mo eo e , hese egions ha e been obse ed in heal hy adul s o co ela e posi i ely wi h habi ual ‘sleep c edi ’ (sleeping in excess o sleep need)20 and nega i ely wi h subjec i e sleepiness21. Ou da a adds o con e ging e idence ha GMV o he ACC and medial OFC a e ela ed o sleep habi s. The PFC is among he la es ma u ing b ain egions, wi h changes seen un il he second decade o li e22, which po en ially makes i ulne able o he e ec s o sleep habi s du ing adolescence. A he age o 14 yea s, he peak GMV has al eady passed in on al a eas, and b ain ma u a ion is e lec ed in GMV educ ion23. The medial PFC B ain egion Clus e Peak oxel BA k p MNI coo dina es pa x y z Time in bed du ing weekdays Supe io on al gy us L 9 1222 0.049 − 28 36 37 3.41E-05 4.08 Middle on al gy us L 8− 39 30 46 5.04E-05 3.98 Wake up ime du ing weekends F on al medial o bi al L 11 5163 5.68E-05 − 2 34 − 14 1.35E-06 * 4.85 An e io cingula e L 32 − 2 50 4 1.84E-06 * 4.78 Bed ime du ing weekends P ecuneus R 7 3068 0.0014 10 − 36 46 7.86E-09 * 5.94 Pa acen al lobule L 5− 12 − 33 52 7.31E-06 4.46 Middle on al gy us R 9/10 2220 0.0062 33 27 33 5.23E-07 * 5.06 Supe io on al gy us R 8 38 30 49 1.15E-04 3.76 F on al supe io medial R 10 3219 0.0011 14 56 10 5.24E-07 * 5.06 An e io cingula e L 32/24 − 3 44 − 5 6.88E-05 3.90 Table 2. Regional co ela ions be ween sleep habi s and g ey ma e olumes in communi y adolescen s. BA = B odmann A ea; k = clus e size, exp essed in numbe o oxels; MNI = Mon eal neu ological Ins i u e coo dina es in millime e s; R = igh ; L = le . MNI coo dina es a e gi en o he oxel o maximal s a is ical signi icance. Co ela ions a e nega i e o wake up ime and bed ime du ing weekends, and posi i e o ime in beddu ing weekdays. aHeigh h eshold p < 0.001; clus e ex en h eshold p < 0.05 FWE co ec ed (k > 1200 oxels). *Heigh h eshold p < 0.05 Family Wise E o (FWE) co ec ed. B ain egion BA Clus e Peak oxel k p MNI coo dina es pa x y z egion-o -in e es mask: wake up ime weekend An e io cingula e L 32 739 0.0004 − 10 46 10 1.34E-05 4.32 F on al supe io medial L 32 − 3 27 36 1.41E-04 3.74 An e io cingula e L 24 − 2 33 0 2.03E-04 3.54 An e io cingula e R 32 4 39 − 2 2.29E-04 3.50 F on al supe io medial R 10 2 52 15 4.71E-04 3.27 egion-o -in e es mask: bed ime weekend Mid Cingula e L/Pa acen al lobule 31 183 0.011 − 12 − 30 51 4.18E-05 4.03 Mid Cingula e R 31 4 − 27 46 5.33E-05 3.97 An e io cingula e L 32 122 0.016 − 4 45 3 6.63E-05 3.91 An e io cingula e L 24 − 2 33 0 2.03E-04 3.62 An e io cingula e R 32 7 0.042 6 39 − 2 1.80E-04 3.62 Table 3. Regional nega i e co ela ions be ween school g ade a e age and g ey ma e olumes in communi y adolescen s. aHeigh h eshold p < 0.05 FWE co ec ed o mul iple compa isons, ex en h eshold p < 0.05 FWE co ec ed. BA = B odmann A ea; k = numbe o oxels; MNI = Mon eal neu ological Ins i u e coo dina es in millime e s; R = igh ; L = le . MNI coo dina es a e gi en o he oxel o maximal s a is ical signi icance. www.na u e.com/scien i ic epo s/ 5 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 is in ol ed in cogni i e and emo ional con ol, and selec i e a en ion24, and i s unc ion has been shown o be impai ed du ing sleep dep i a ion15. La e bed ime du ing weekends was addi ionally associa ed wi h widesp ead GMV educ ions in he igh middle, medial, and supe io on al co ex, and in he p ecuneus. Sleep slow wa e ac i i y, an elec ophysio- logical measu e o ‘deep’ sleep o sleep homeos asis, has been shown o be closely co ela ed wi h GM hickness in mos o hese b ain egions in child en and adolescen s25. He ein, he middle and supe io on al gy i GMV co ela ed nega i ely wi h bo h weekend bed ime and weekday ime in bed. They a e pa o he do sola e al p e- on al co ex (DLPFC)26, hus ex ending p e ious indings linking DLPFC s uc u e and unc ion wi h sleep13, sleep loss27, and sleep pa hologies19. The p ecuneus is an epicen e o he de aul mode ne wo k28. I s ac i i y, sim- ila ly o he ac i i y o he ACC, is highly a iable du ing he heal hy sleep-wake cycle: high du ing wake ulness and low du ing slow-wa e sleep29. A educ ion in p ecuneus GMV has been obse ed in insomnia pa ien s17, bu ou s udy is he i s o demons a e an associa ion be ween GMV o he p ecuneus and sleep habi s in heal hy indi iduals, using a whole b ain VBM app oach. Poo e school g ades associa ed wi h smalle GMV in he medial on al a eas. This is he i s epo o show ha bo h sleep habi s and school g ades a e linked o b ain GM mo phome y o he medial PFC in ado- lescen s. In e es ingly, hese egions a e epicen e s o unc ions such as mul i asking30 and sel - ela ed men al ep esen a ions28 ha migh be a s ake in adolescen s’ academic achie emen s. Al hough he di ec ion o he media ion analyses (i.e. ha weekend bed ime media ed he ela ion be ween GMV and school pe o mance) was no exac ly as we ini ially hypo hesized ( ha GMVs would media e he e ec o sleep habi s on school pe o - mance), he esul s o he media ion analyses al oge he sugges ha la e weekend bed ime migh d aw a dele e i- ous ela ion be ween GMV and school g ades. Based on ou da a i emains specula i e how much e.g. modi ying bed imes could a ec school pe o mance. Fu he in e en ion s udies a e equi ed o answe hese ques ions. Con a y o ou expec a ions and p e ious indings13, we did no see a co ela ion o sleep habi s wi h hip- pocampal GMV. This disc epancy may be a ibu ed o me hodological di e ences, no ably he egion-o -in e es app oach used in he s udy by Taki e al.13 as compa ed o he p esen whole-b ain s a is ical pa ame ic mapping app oach, and he ac ha ma u a ion o he hippocampus ollows a di e en ial ajec o y han he neoco ex. Fu he , con a y o he olume o he neoco ex, hippocampal olumes migh e en inc ease o e he cou se o adolescence31. The hippocampus egion migh he e o e be mo e ulne able o sleep habi s a a di e en phase o de elopmen han he neoco ex, which could pa ly explain he di e en ial indings be ween he p esen homo- geneous age g oup o adolescen s (age ange 13.4–15.5 yea s) and he ea lie s udy including a mixed sample o child en and adolescen s aged 5–18 yea s13. Finally, in he p esen VBM s udy we used a conse a i e ex en h eshold (1200 oxels), which may ha e p ecluded he de ec ion in small b ain s uc u es. Indeed, in supplemen- a y analyses wi h a mo e libe al h eshold (p < 0.001 unco ec ed; ex en h eshold a 50 oxels unco ec ed), we ound a co ela ion be ween smalle bila e al hippocampal olumes and la e weekend bed imes as well as sho e weekday ime in bed (see Supplemen a yFigu esS1 and S2). These co ela ions we e, howe e , s a is ically much weake han he neoco ical indings. The majo s eng h o he cu en s udy is he sample wi h a na ow age ange ec ui ed om he gene al pop- ula ion. Addi ionally, he a iabili y in pube al de elopmen al s age wi hin he sample was aken in o accoun by including he adolescen s’ PDS sco es as a con ounding co a ia e in he s a is ical models. Thus he e ec s o age/ pube al s a us on ou esul s we e minimized. Mo eo e , he bed imes and wake up imes in ou sample a e in line wi h hose p e iously epo ed in la ge -scale epidemiologic s udies on adolescen sleep habi s4. Al hough ou sample included also a small p opo ion o indi iduals wi h p obable psychia ic diso de s, he main esul s we e no accoun ed o by he p esence o psychia ic mo bidi ies among he pa icipan s (see Supplemen a yTableS2), implica ing an independen ole o sleep habi s. E ec ype Poin es ima e 95% CI p- alue P ecuneus/pa acen al clus e (BA 7/BA 5) Media ion e ec − 1.429 [− 2.948; − 0.163] 0.03 (GMV – bed ime – school pe o mance) Di ec e ec − 2.765 [− 5.825; 0.322] 0.08 (GMV – school pe o mance) To al e ec − 4.194 [− 7.052; − 1.340] < 0.001 P opo ion o o al e ec ia media ion 0.337 [0.030; 1.158] 0.04 F on al supe io medial and ACC clus e (BA10/BA 32) Media ion e ec − 2.339 [− 4.807; − 0.562] < 0.001 (GMV – bed ime – school pe o mance) Di ec e ec − 2.851 [− 7.764; 1.978] 0.28 (GMV – school pe o mance) To al e ec − 5.190 [− 9.821; − 0.884] 0.02 P opo ion o o al e ec ia media ion 0.438 [0.075; 2.191] 0.02 Table 4. Causal media ion analyses on he ela ionship be ween he GMVs and school g ade a e age wi h bed ime du ing weekends as media o . GMV = G ey ma e olume; BA = B odmann A ea; 95% CI = 95% con idence in e al o he poin es ima e. www.na u e.com/scien i ic epo s/ 6 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 Limi a ions include he lack o objec i e measu es o sleep and he c oss-sec ional design. Since no objec i e assessmen o sleep was pe o med, he design does no allow any in e ences on he mo e exac neu obiological mechanisms unde lying he ela ionship be ween sleep habi s and b ain g ey ma e olumes. C oss-sec ional da a canno es o causal ela ionships be ween a iables, i.e. whe he poo sleep habi s a e causing o caused by egional GMV loss. The c oss-sec ional design o he s udy also p ecludes in e ences ela ed o b ain ma u a ion. Longi udinal app oaches a e needed o add ess whe he he GMV changes a e in ac ha m- ul o help ul o he indi iduals in he long un. Du ing adolescen yea s he no ma i e cou se o de elopmen is owa ds a dec ease in bo h GMVs and sleep du a ion, and owa ds delayed sleep iming: he mo e ma u e he adolescen is, he less GMV(s) he is expec ed o ha e3,23. The sample we s udied was pa icula ly homogeneous in e ms o age, and, u he mo e, pube al de elopmen al s age was con olled in he analyses. On a cau ious no e, we would no expec ha he indings would e lec a posi i e ma u a ional change, bu a he a de ia ion in ma u a ion, i.e. a GMV dec ease beyond he a e age le els. The link be ween smalle GMV and poo e school pe o mance would also a o he idea o he changes o be a he maladap i e han de elopmen ally a o able. The li e a u e on sleep pa hologies in adul pa ien s would also suppo he second al e na i e. Finally, he de e minan s o adolescen s’ sleep habi s in ol e biological, beha iou al, psychosocial, and cul- u al ac o s ha a e di icul o disen angle, and may in e e e wi h b ain ma u a ion. Demons a ing he ole o hese ac o s was beyond ou objec i es. In conclusion, he p esen epo highligh s an ala ming associa ion be ween he a ia ions in sleep habi s, adolescen egional GMV, and school g ade a e age. I espec i e o he causal unde pinnings o his ela ionship, he indings suppo ha paying a en ion o sleep habi s du ing his pe iod o ma u a ion migh be a p ecau- iona y p inciple. We would encou age all pa en al, socie al, and educa ional suppo o adolescen s o ensu e main enance o a heal hy sleep-wake hy hm. Especially, a oiding la e bed imes du ing weekends would seem impo an in o de o make he mos ou o he b ain’s de elopmen al po en ial and o ensu e op imal academic success. Me hods Pa icipan s. A o al o 265 adolescen s we e ec ui ed based on hei age om he gene al popula ion om 9 h and 10 h g ade s uden s in schools nea Pa is, F ance, ia lye s and school isi s. Exclusion c i e ia included se e e medical soma ic condi ions, any his o y o head inju ies, and any con aindica ions o MRI. These ado- lescen s pa icipa ed in a la ge Eu opean mul i-si e longi udinal s udy o adolescen de elopmen 32. Only he F ench adolescen s we e assessed o hei sleep cha ac e is ics and we e hus eligible o he s udy. Fo he pu - pose o his s udy, we excluded all pa icipan s who d opped ou be o e he MRI scan, pa icipan s wi h missing sleep da a, o whose b ain images did no pass quali y con ol, o wi h ma ked alcohol consump ion (Alcohol Use Diso de s Iden i ica ion Tes (AUDIT) o al sco e > 7), esul ing in he exclusion o 88 adolescen s. He e, we hus p esen da a om he emaining 177 adolescen s. The s udy p o ocol was app o ed by he E hics Commi ee o Pa is CPP IDF-VII and he s udy was conduc ed in acco dance wi h he Decla a ion o Helsinki. W i en in o med assen and consen we e ob ained om all he pa icipan s and hei pa en s, espec i ely. The pa icipan s we e assessed o psychia ic symp oms wi h he De elopmen and Well-Being Assessmen (DAWBA) (), a sel -adminis e ed compu e -based diagnos ic ques ionnai e consis ing o open and closed ques- ions33. O he assessmen s included he Pube al De elopmen Scale (PDS) ques ionnai e34, and he Wechsle In elligence Scale o Child en-Fou h Edi ion (WISC-IV)35. Acco ding o DAWBA, he majo i y o he adoles- cen s we e ee o any clinically signi ican psychia ic symp oms, while n = 28 (15.8%) adolescen s in o al we e a ed as ha ing a p obable DSM-IV psychia ic diagnosis. Al hough some pa icipan s exp essed psychia ic symp oms, none o hem was using psycho opic medica ion o men al heal h consul a ion a he ime o he s udy, and all pa icipan s we e a ending school egula ly. Supplemen a y analyses on he egional co ela ions be ween sleep habi s and GMVs we e also pe o med in a subsample o 149 adolescen s wi hou any p obable psychia ic diagnosis (Supplemen a yTableS2). Assessmen o sleep. Sleep habi s we e assessed by asking he adolescen s hei habi ual bed imes and wake-up imes du ing weekdays and weekends sepa a ely. The exac ques ions we e “A e age ime o going o bed on weekdays”; “A e age ime o waking up du ing he week”; “A e age ime o going o bed du ing he weekend”; “A e age ime o waking up du ing he weekend” (o iginal ques ions in F ench: “Heu e moyenne de couche la semaine, Heu e moyenne de le e la semaine; Heu e moyenne de couche le week-end; Heu e moyenne de le e le week-end”). The ques ions did no conce n a speci ic e ospec i e ime ange bu he adolescen s we e asked o epo hei usual sleeping habi s. Time in bed was calcula ed based on he bed imes and wake-up imes epo ed by he pa icipan s sepa a ely o weekdays and weekends. The sleep assessmen s we e pe o med on he same day as he MRI. Assessmen o school g ade a e age. The pa icipan s epo ed hei school g ade a e age o he las e m as pa o he Eu opean School Su ey P ojec on Alcohol and D ugs (ESPAD)36 by answe ing he ques ion ”Which o he ollowing bes desc ibes you a e age g ade in he end o he las e m”? wi h an 8-poin scale ang- ing om A (93–100) o C- (70–72). (See Supplemen a yTableS3 o de ails). MRI da a acquisi ion and p ep ocessing. All adolescen s unde wen MRI examina ion wi h a Siemens T io 3 Tesla scanne a he Neu ospin Cen e (F ance). High- esolu ion s uc u al T1-weigh ed images we e ob ained using a s anda dized 3D T1-weigh ed magne iza ion p epa ed apid acquisi ion echo (MPRAGE) sequence based on he ADNI p o ocol (h p://adni.loni.usc.edu/me hods/m i-analysis/m i-acquisi ion/) wi h he www.na u e.com/scien i ic epo s/ 7 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 ollowing acquisi ion pa ame e s: epe i ion ime = 2300 ms: echo ime = 2.8 ms, lip angle = 8°; 256 × 256 × 170 ma ix, 1.1 × 1.1 × 1.1 mm oxel size). Images we e p ep ocessed wi h S a is ical Pa ame ic Mapping 8 so wa e (SPM8) using Voxel-Based Mo phome y (VBM)7. The T1-weigh ed images we e segmen ed and no malized using cus omized Tissue P obabili y Maps. The no malized, segmen ed and modula ed g ey ma e (GM) images we e smoo hed using a 10-mm ull-wid h a hal -maximum (FWHM) Gaussian ke nel. Head size was measu ed by he Volume Scaling Fac o (VSF), which is based on he a ine ans o ma ion pe o med du ing spa ial no - maliza ion (h ps://su e .nm .mgh.ha a d.edu/ swiki/eTIV). S a is ical analysis. The demog aphic, sleep and school pe o mance a iables we e analysed wi h he IBM SPSS S a is ics Ve sion 22. Reg ession analyses using he gene al linea model (GLM) me hod we e conduc ed o assess whe he sleep habi s we e associa ed wi h school g ade a e age. These analyses we e con olled o pube al s age (PDS) and gende . Bon e oni-co ec ed p- alues o < 0.0083 we e conside ed as s a is ically signi ican . The co ela ions be ween sleep habi s - bed ime, wake up ime, and ime in bed, sepa a ely on weekdays and weekends - and GMVs, we e examined in he whole b ain using mul iple eg ession models in SPM8. Fo each o he six co ela ion analyses, we used one sleep habi as co a ia e o in e es , and gende , PDS and VSF as con ounding co a ia es. Signi ican clus e s o GMVs we e iden i ied using a heigh h eshold o p < 0.001, wi h amily-wise e o (FWE) co ec ion o mul iple spa ial compa isons a he clus e le el ac oss he whole b ain (p < 0.05, spa ial ex en 1200 oxels) unde non-s a iona i y assump ion37. A e wa ds, analyses we e conduc ed in o de o de e mine whe he he egions whe e GMVs co ela ed wi h sleep habi s did also ela e wi h school g ade a e age. These analyses we e conduc ed wi hin 3 egion-o -in e es masks ob ained om signi ican co - ela ion analyses be ween sleep habi s and GMVs. Co ela ion analyses be ween he school g ade a e age and GMV wi hin he egion-o -in e es masks we e conduc ed wi h a heigh h eshold se a p < 0.05 FWE co ec ed o mul iple compa isons. Causal media ion analyses we e conduc ed o de e mine whe he , in he egions-o -in e es ha we e ound o co ela e wi h sleep habi s, (1) sleep habi s could media e he ela ion be ween he GMVs and school g ade a e - age, o (2) GMVs could media e he ela ion be ween he sleep habi s and school g ade a e age. Causal media ion analyses we e pe o med wi h algo i hms de ised by Imai e al.38. School g ade a e age was en e ed as dependen ac o and (1) GMVs and (2) sleep habi s (wake up ime and bed ime du ing weekends) as independen ac o . (1) Sleep habi s and (2) GMVs we e en e ed as media o a iable. Gende , PDS and VSF we e en e ed as con ound- ing a iables. The media o models we e i wi h he gene al linea model and ou pu objec s we e boo s apped 5000 imes wi h eplacemen using a pa ame ic media ional analysis. In causal media ion analyses, a signi ican media ing e ec was de ined as a 95% con idence in e al ha does no include ze o. Re e ences 1. Ma icciani, L., Olds, T. & Pe ko , J. In sea ch o los sleep: secula ends in he sleep ime o school-aged child en and adolescen s. Sleep Med. Re . 16, 203–211 (2012). 2. K onholm, E. e al. T ends in sel - epo ed sleep p oblems, i edness, and ela ed school pe o mance among Finnish adolescen s om 1984 o 2011. J. Sleep Res. 24, 3–10 (2015). 3. Hagenaue , M. H., Pe yman, J. I., Lee, T. M. & Ca skadon, M. A. Adolescen changes in he homeos a ic and ci cadian egula ion o sleep. De . Neu osci. 31, 276–284 (2009). 4. G adisa , M., Ga dne , G. & Dohn , H. Recen wo ldwide sleep pa e ns and p oblems du ing adolescence: a e iew and me a- analysis o age, egion, and sleep. Sleep Med. 12, 110–118 (2011). 5. Whea on, A. G., Fe o, G. A. & C o , J. B. School s a imes o middle school and high school s uden s - Uni ed S a es, 2011–12 school yea . MMWR. Mo b. Mo al Wkly. Rep. 64, 809–13 (2015). 6. Minges, K. E. & Redeke , N. S. Delayed school s a imes and adolescen sleep: A sys ema ic e iew o he expe imen al e idence. Sleep Med. Re . 28, 82–91 (2016). 7. Owens, J., Adolescen Sleep Wo king G oup. Commi ee on Adolescence. Insu icien sleep in adolescen s and young adul s: an upda e on causes and consequences. Pedia ics 134, e921–32 (2014). 8. Me ikan o, I., Lah i, T., Puusniekka, R. & Pa onen, T. La e bed imes weaken school pe o mance and p edispose adolescen s o heal h haza ds. Sleep Med. 14, 1105–11 (2013). 9. Pe ez-Llo e , S. e al. A mul i-s ep pa hway connec ing sho sleep du a ion o day ime somnolence, educed a en ion, and poo academic pe o mance: an explo a o y c oss-sec ional s udy in eenage s. J. Clin. Sleep Med. 9, 469–73 (2013). 10. Hysing, M., Ha ey, A. G., Lin on, S. J., Askeland, K. G. & Si e sen, B. Sleep and academic pe o mance in la e adolescence: esul s om a la ge popula ion-based s udy. J. Sleep Res., doi: 10.1111/js .12373 (2016). 11. Ho ne, J. A. Human sleep, sleep loss and beha iou . implica ions o he p e on al co ex and psychia ic diso de . B . J. Psychia y 162, 413–419 (1993). 12. Nilsson, J. P. e al. Less e ec i e execu i e unc ioning a e one nigh ’s sleep dep i a ion. J. Sleep. Res. 14, 1–6 (2005). 13. Taki, Y. e al. Sleep du a ion du ing weekdays a ec s hippocampal g ay ma e olume in heal hy child en. Neu oimage 60, 471–475 (2012). 14. Ashbu ne , J. & F is on, K. J. Voxel-based mo phome y– he me hods. Neu oimage 11, 805–821 (2000). 15. Ma, N., Dinges, D. F., Basne , M. & Rao, H. How acu e o al sleep loss a ec s he a ending b ain: A me a-analysis o neu oimaging s udies. Sleep 38, 233–240 (2015). 16. Joo, E. Y. e al. Reduced b ain g ay ma e concen a ion in pa ien s wi h obs uc i e sleep apnea synd ome. Sleep 33, 235–241 (2010). 17. Al ena, E., V enken, H., Van De We , Y. D., an den Heu el, O. A. & Van Some en, E. J. Reduced o bi o on al and pa ie al g ay ma e in ch onic insomnia: a oxel-based mo phome ic s udy. Biol. Psychia y 67, 182–185 (2010). 18. Sche le , C. Whi e and g ay ma e abno mali ies in na colepsy wi h ca aplexy. Sleep 35, 345–351 (2012). 19. Joo, E. Y. e al. B ain g ay ma e de ici s in pa ien s wi h ch onic p ima y insomnia. Sleep 36, 999–1007 (2013). 20. Webe , M. e al. Habi ual ‘sleep c edi ’ is associa ed wi h g ea e g ey ma e olume o he medial p e on al co ex, highe emo ional in elligence and be e men al heal h. J. Sleep Res. 22, 527–534 (2013). 21. Killgo e, W. D., Schwab, Z. J., Kipman, M., DelDonno, S. R. & Webe , M. Voxel-based mo phome ic g ay ma e co ela es o day ime sleepiness. Neu osci. Le . 518, 10–13 (2012). 22. Gog ay, N. e al. Dynamic mapping o human co ical de elopmen du ing childhood h ough ea ly adul hood. P oc. Na l. Acad. Sci. USA 101, 8174–8179 (2004). www.na u e.com/scien i ic epo s/ 8 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 23. Giedd, J. N. e al. Ana omical b ain magne ic esonance imaging o ypically de eloping child en and adolescen s. J. Am. Acad. Child Adolesc. Psychia y 48, 465–470 (2009). 24. Bush, G., Luu, P. & Posne , M. I. Cogni i e and emo ional in luences in an e io cingula e co ex. T ends Cogn. Sci. 4, 215–222 (2000). 25. Buchmann, A. e al. EEG sleep slow-wa e ac i i y as a mi o o co ical ma u a ion. Ce eb al Co ex 21, 607–615 (2011). 26. Cieslik, E. C. e al. Is he e “one” DLPFC in cogni i e ac ion con ol? E idence o he e ogenei y om co-ac i a ion-based pa cella ion. Ce eb al Co ex 23, 2677–2689 (2013). 27. Thomas, M. e al. Neu al basis o ale ness and cogni i e pe o mance impai men s du ing sleepiness. I. E ec s o 24 h o sleep dep i a ion on waking human egional b ain ac i i y. J. Sleep Res. 9, 335–352 (2000). 28. Ca anna, A. E. & T imble, M. R. The p ecuneus: a e iew o i s unc ional ana omy and beha iou al co ela es. B ain 129, 564–583 (2006). 29. Dang-Vu, T. T. e al. Func ional neu oimaging insigh s in o he physiology o human sleep. Sleep 33, 1589–1603 (2010). 30. Roca, M. e al. The ole o A ea 10 (BA10) in human mul i asking and social cogni ion. Neu opsychologia 49, 3525–3231 (2011). 31. Os by, Y. e al. He e ogenei y in subco ical b ain de elopmen : A s uc u al magne ic esonance imaging s udy o b ain ma u a ion om 8 o 30 yea s. J. Neu osci. 29, 11772–11782 (2009). 32. Schumann, G. e al. The IMAGEN s udy: ein o cemen - ela ed beha iou in no mal b ain unc ion and psychopa hology. Mol Psychia y 15, 1128–1139 (2010). 33. Goodman, R., Fo d, T., Richa ds, H., Ga wa d, R. & Mel ze , H. The De elopmen and Well-Being Assessmen : desc ip ion and ini ial alida ion o an in eg a ed assessmen o child and adolescen psychopa hology. J. Child Psychol. Psychia y 41, 645–655 (2000). 34. Pe e sen, A. C., C ocke , L., Richa ds, M. & Boxe , A. A sel - epo measu e o pube al s a us: Reliabili y, alidi y, and ini ial no ms. J. You h Adolesc. 17, 117–133 (1988). 35. Wechsle , D. WISC-IV. Technical and In e p e i e Manual. San An onio, TX: The Psychological Associa ion (2003). 36. Ande sson, B., Bja nason, T., Kokke i, A., Mo gan, M. & Na usk, A. The 1995 ESPAD epo : Alcohol and o he d ug use among s uden s in 26 Eu opean coun ies. Swedish Council o In o ma ion on Alcohol and O he D ugs; S ockholm. (1997). 37. Nichols, T. E. Combining oxel in ensi y and clus e ex en wi h pe mu a ion es amewo k. Neu oimage 23, 54–63 (2004). 38. Imai, K., Keele, L. & Tingley, D. A gene al app oach o causal media ion analysis. Psychol Me hods. 15, 309–334 (2010) Acknowledgemen s P o . Mau i Ma unen is hanked o his help and suppo and o p o iding he acili ies o conduc he esea ch a he Na ional Ins i u e o Heal h and Wel a e. This wo k ecei ed inancial suppo om: he Academy o Finland (g an numbe 276612 o ASU); he Emil Aal onen Founda ion; he Eu opean Union- unded FP6 In eg a ed P ojec IMAGEN (LSHM-CT- 2007–037286), he F ench unding agency ANR (ANR-12- SAMA-0004), The E ane – Neu on g an (p ojec AF12-NEUR0008-01 - WM2NA), he Fonda ion de F ance, he Fonda ion pou la Reche che Médicale, he Mission In e minis é ielle de Lu e-con e-les-D ogues-e -les- Condui es-Addic i es (MILDT), an IDEX Pa is Saclay PhD-g an 2012 o D . Hélène Vulse , and an INSERM - APHP 2010 in e ace g an o D . Ma ie-Lau e Paillè e Ma ino . The sponso s o he s udy had no ole in s udy design, da a collec ion, da a analysis, da a in e p e a ion, w i ing o he epo , o decision o submi he pape o publica ion. The co esponding au ho had ull access o all he da a in he s udy and had inal esponsibili y o he decision o submi o publica ion. Au ho Con ibu ions J.L.M., A.S.U., E.A., M.L.P.M., J.M. and he IMAGEN conso ium designed he s udy. J.L.M., E.A., J.M., R.M., J.P., H.V., P.B., P.C., H.G., M.L.P.M., pa icipa ed in da a collec ion (design o acquis ion) and quali y con ol. A.S.U., E.A., J.M., H.L., H.V., R.M., W.L., unde ook o ad ised on he da a analysis. A.S.U., J.L.M., E.A., M.L.P.M., w o e he manusc ip . All au ho s con ibu ed o he subsequen d a s and app o ed he inal e sion. Addi ional In o ma ion Supplemen a y in o ma ion accompanies his pape a h p://www.na u e.com/s ep Compe ing inancial in e es s: The au ho s decla e no compe ing inancial in e es s. How o ci e his a icle: U ila, A. S. e al. Sleep habi s, academic pe o mance, and he adolescen b ain s uc u e. Sci. Rep. 7, 41678; doi: 10.1038/s ep41678 (2017). Publishe 's no e: Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps and ins i u ional a ilia ions. This wo k is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons license, unless indica ed o he wise in he c edi line; i he ma e ial is no included unde he C ea i e Commons license, use s will need o ob ain pe mission om he license holde o ep oduce he ma e ial. To iew a copy o his license, isi h p://c ea i ecommons.o g/licenses/by/4.0/ © The Au ho (s) 2017 www.na u e.com/scien i ic epo s/ 9 Scien i ic RepoR s | 7:41678 | DOI: 10.1038/s ep41678 Conso ia The IMAGEN conso ium Tobias Banaschewski11, He a Flo 11, Mi a Fau h-Bühle 11, Louise Pou ska11, F auke Nees11, Y onne G imme 11, Ma en S u e11, Andeas Heinz12, And eas S öhle12, Viola Kappel12, Be eke Ma ia an Noo 12, Jean-Bap is e Poline13, Yanick Schwa z13, Benjamin Thy eau13, James I eland14, John Roge s14, Nadège Bo das1, Zuleima B icaud1, I ina Filippi1, And é Galinowski1, Fanny Gollie -B ian 1, Vincen Ména d1, Gun e Schumann7, Syl ane Des i iè es7, Anna Ca ell7, Robe Goodman7, A gy is S inga is7, Cha lo e Nymbe g7, Lau ence Reed7, Ga e h J Ba ke 15, Be nd I e mann16, Ruedige B ühl16, Michael Smolka17, Thomas Hübne 17, Ka h in Mülle 17, A un L. W. Bokde18, Ch is ian Büchel19, Uli B ombe g19, Ju gen Gallina 19, Tahmine Fadai19, Pennylai e Gowland20, C Law ence20, Tomas Paus21 11Cen al Ins i u e o Men al Heal h, Mannheim, Ge many 12Cha i é Hospi al, Be lin, Ge many 13Commissa ia à l’Ene gie A omique, Pa is, F ance 14Delosis, Twickenham, G ea e London, Uni ed Kingdom 15 The Depa men o Neu oimaging, Ins i u e o Psychia y, Psychology & Neu oscience, King’s College London, Uni ed Kingdom 16Physikalisch-Technische Bundesans al (PTB), Be lin; Ge many 17Uni e si y o D esden, Ge many 18T ini y College Dublin, I eland 19Uni e si y o Hambu g, Ge many 20Uni e si y o No ingham, Uni ed Kingdom 21Uni e si y o To on o, Canada