Sleep,rest-activity fragmentation and structural brain changes related to the ageing process

Download (0)

Full text

(1)

Sleep,

rest-activity

fragmentation

and

structural

brain

changes

related

to

the

ageing

process

Marion

Baillet

1

and

Christina

Schmidt

1,2

Increasingevidencesuggestsanassociationbetweentypical age-relatedchangesinsleepandbrainstructure.Herewe reviewstudiesexploringtheassociationbetweenhuman histo-pathologicalandinvivoneuroimagingmarkersofbrain structureandsleep-wakeparametersinhealthyolderadults. Evidencefrombothlarge-scaleepidemiologicalstudiesand in-labquantificationofspecificsleepsignaturesarereviewedand advantagesandpitfallshighlighted.Overall,theresultspointto anassociationbetweensleep-wakedisruptionandbothlocal anddiffusechangesinbrainstructure.Theassociativestrength largelyvariesbetweenstudiesandseemstopartiallydepend onthesleeptraitunderinvestigation.Theroleofspecific sleep-wakeregulatingmechanismsonhumancognitiveandbrain fitnessandmoreparticularlytheircausalrelationshipremains tobedisentangled.

Addresses

1GIGA-Institute,CyclotronResearchCenter/InVivoImaging,Sleepand ChronobiologyLab,UniversityofLie`ge,Lie`ge,Belgium

2PsychologyandNeuroscienceofCognitionResearchUnit,Facultyof PsychologyandEducationalSciences,UniversityofLie`ge,Lie`ge, Belgium

Correspondingauthor:Schmidt,Christina(christina.schmidt@uliege.be)

CurrentOpinioninBehavioralSciences2019,33:8–16

ThisreviewcomesfromathemedissueonCognitionandperception –*sleepandcognition*

EditedbyMichaelCheeandPhilippePeigneux

https://doi.org/10.1016/j.cobeha.2019.11.003

2352-1546/ã2019ElsevierLtd.Allrightsreserved.

Introduction

Overthelastfew years,growinginterest isobserved in

understanding theprotective effects of sleep onhealth

andqualityoflife,especiallyinthecontextoftheageing

population.Inhumans,thegrossframeworktosleepand

wakefulness alternation is provided by the interaction

betweenasleep-dependenthomeostaticprocessandthe

circadiantimingsystemresponsibleforendogenous

mod-ulationsof sleep and wakepropensity over the24-hour

day [1]. Those regulating processes act as important

modulators for brain and behavior (e.g. [2]) and their

potential implication in cognitive and brain ageing is

increasinglyconsidered([3–5];seealso[6]for areview).

Here we review studies exploring the association

betweenhuman post-mortemand in vivomagnetic

reso-nance imaging (MRI) markers of brain structure and

sleep-wakeparameters in healthyolder adults.As

sum-marized in Table 1, brain structural measures include

quantificationofgreymatteraswellaswhitematterfrom

cross-sectional and longitudinal designs, using

whole-brainorregionsofinterestapproaches.Wewill

differen-tiate studies investigating sleep-wake parameters

collected through (1) self-reported questionnaires, (2)

polysomnography(sleep architectureand

electrophysio-logicalsignatures),and(3)actimetryrecordings(integrity

of the temporal distribution of rest-activity cycles and

estimatedsleepfragmentationin real-lifesettings).

Association

between

brain

structure

and

self-reported

sleep

Sleep complaints are increasing while getting older

including altered sleep perception, changes in sleep

architecture or increased variability in the timing and

duration of sleep [7]. Sleep questionnaires have been

developedtoprovideageneraloverviewofthesubjective

quality of sleep. While they are often used as a first

diagnostictoolinclinicalpractice,qualitativeassessments

of sleep perception donot inform about specific

sleep-regulating mechanisms and are the consequence of an

amalgamofvariousbiologicalandenvironmentalfactors,

varyingover time.However,they remain costeffective

andeasytouseallowingthereforethegenerationoflarge

andrepresentativesamplesbyadoptingepidemiological

approaches.

Cross-sectionalstudiesdetectedan associationbetween

poorself-reportedsleepqualityandreducedgreymatter

volumeincludingorbitofrontalregions[8],insula[9],and

age-related thalamic and hippocampal atrophy [10]. A

longitudinalapproach reportedthatgreymatteratrophy

of frontal, parietal and temporal, but not hippocampal

regionswaspredictivefor poorsleepqualityat3.5years

follow up [11]. Compared to sleep quality, reports on

sleepdurationhaveproducedmore mixedresults. Most

cross-sectionalstudiesfailedtoobserveasignificant

asso-ciation between self-reported sleep duration and grey

mattervolume [9,12,13]. A longitudinal study reported

that both short (<7hours) and long (>7hours) sleep

durations were associated with higher rates of cortical

(2)

Table1

Crosssectionalandlongitudinalstudiesofsleep-wakestatesandstructuralbrainimaging

Paper Design n,meanageSD Project

Brainmeasures Mainfindings

Sleepquestionnairestudies Stoffers[8] Crosssectional n=65,40.59.7

Netherlandsstudyof depressionandanxiety

GMvolume–VBM8,voxel wiseanalysis

EMAwasassociatedwithlow volumewithintheleftOFC.No associationwithDIS,DMS Branger[9] Crosssectional n=51,64.110.6

Multimodalimagingofearly stageAlzheimer’sdisease

GMvolume–VBM5,voxel wiseanalysis

Highnumberofnocturnal awakeningswasrelatedtolow insularvolume.Noassociationwith sleeplatency,durationandquality Liu[10] Crosssectional n=54young,21.61.6

n=94older,66.25.1 Independentsample

GM,WM,hippocampaland thalamicvolumes–VBM8

Poorsleepqualitywasassociated withage-relatedGM,hippocampal andthalamicatrophy

Sexton[11] Longitudinal3.5years n=147,53.915.5 Cognitionandplasticity throughthelifespan

Corticalvolume–Freesurfer, vertexwiseanalysis Hippocampalvolumes– Freesurfer

Lowvolumewithintherightsuperior frontalcortex(crosssectional)and increasedrateofatrophywithin frontal,temporal,andparietal regions(longitudinal)waspredictive forpoorsleepquality.No associationwithhippocampal volumeoratrophy

Lo[12] Longitudinal2years n=66,67.45.8

Singapore-longitudinalaging brainstudy

GM,WM,hippocampal, ventricles,inferiorand superiorfrontalgyrivolumes –Freesurfer

Shortsleepdurationwasrelatedto increasedventricularexpansion.No associationwithotherbrainvolumes andsleepquality

Lutsey[13] Crosssectional n=312,61.75 Theatherosclerosisriskin communitiesstudy

Frontal,temporal,parietal andoccipitalcortical volume.Hippocampal,and deepgreymattervolume– Freesurfer

WMHvolumeandbrain infarcts

Noassociationbetweensleep durationandMRImeasures

Spira[14] Longitudinal8years n=122,66.68

Baltimorelongitudinalstudy ofaging

Corticalthickness– Freesurfer,vertexwise analysis

Longsleepdurationwasassociated withlowcorticalthicknessinthe inferioroccipitalgyrusandsulcusof thelefthemisphere(crosssectional). Longandshortsleepdurationwere associatedwithincreasedrateof corticalthinninginfronto-temporal area(longitudinal)

Carvalho[15] Crosssectional n=1374,>50

Mayoclinicstudyofaging

Frontal,parietal,temporal andoccipitalcortical thicknessandhippocampal volume–Freesurfer

Excessivedaytimesleepinesswas associatedwithlowcortical thicknessinfrontal,parietal, temporalandoccipitalregionsand hippocampalvolume

DelBrutto[16] Crosssectional n=237,708 Atahualpaproject

WMHvolume,lacunar infarctsanddeep microbleeds

Poorsleepqualitywasassociated withWMHpresenceandseverity. Noassociationwithsilentlacunar infarctordeepmicrobleeds Kocevska[17] Longitudinal5years n=2,529,55.86.0

Rotterdamstudy

FA,MD–tractography Sleepcomplaintswererelatedto decreasedFAinthemiddle cerebellarpeduncleandmedial lemniscusovertime.Shortsleep durationwasrelatedtoincreased WMHburdenovertimeandhigh sleepefficiencywithhighWMH volumeatfollowup.Noassociation betweenWMmeasuresatbaseline andchangesinsleepcomplaints Sexton[19] Crosssectional n=448,69.25.1

WhitehallIIimaging substudy

FA,AD,RD-FSL,TBSS, voxelwiseanalysis WMHvolume

Poorsleepqualitywasrelatedtolow FA,highADandRDvaluesinfrontal lobe.Noassociationwithsleep duration,sleepefficiency,WMH volumeandpersistenceofpoor sleep16yearspriorbrainimaging

(3)

Table1 (Continued)

Paper Design n,meanageSD Project

Brainmeasures Mainfindings

Yaffe[21] Crosssectional n=613,45.4 Coronaryarteryrisk developmentinyoungadults

Frontal,parietal,temporal andoccipitalFA,MD WMHvolume

Shortsleepdurationwasassociated withhighMDvaluesinfrontal, parietalandtemporalregionsand highparietalWMHvolume Ramos[22] Crosssectional n=1,244,70.09.0

NorthernManhattanstudy

WMHvolume Longsleepdurationwasassociated withhighWMHvolumeonlyinolder adultssufferingfromdiabetes Polysomnographicstudies

Gelber[28] Crosssectional n=167,83.7(79–95) Honolulu-Asiaagingstudy

Brainautopsy GreaterSWSdurationwas associatedwithlessbrainatrophy Dube´ [29] Crosssectional n=30young,23.492.79

n=33older,60.355.71 Independentsample

Corticalthickness–CIVET, voxelwiseanalysis

HigherSWdensitywasassociated withage-relatedinsular,superior temporal,parietalandmiddlefrontal thickness.HigherSWamplitude wasassociatedwithage-related middlefrontal,medialprefrontal, andmedialposteriorthinning Mander[30] Crosssectional n=18young,20.42.1

n=18older,72.46.1 Independentsample

mPFCvolume–VBM8 ReducedSWAwasassociatedwith age-relatedmPFCatrophy

Varga[31] Crosssectional n=18young,18–23 n=13older,51–85 Independentsample

mPFCvolume–Freesurfer ReducedfrontalSWAwasassociatd withage-relatedmPFCatrophy

Helfrich[32] Crosssectional n=20young,20.42.0 n=32older,73.75.3 Independentsample

mPFC,dlPFC,lateral orbitofrontalcortex,thalamic andhippocampalvolumes– VBM8

Impairedslowwave-spindle couplingwasrelatedtoage-related mPFCatrophy.Noassociationwith hippocampus,thalamus,lateral OFCanddlPFC

Fogel[33] Crosssectional n=13young,24.13.5 n=15older,62.23.8 Independentsample

GMvolume–FSL,voxelwise analysis

Sleepspindleswererelatedtogrey mattervolumewithinthe

hippocampus,cerebellum, cingulateandparietalcortexin youngbutnotolderadults Mander[34] Crosssectional n=20young,20.42.0

n=31older,73.55.2 Independentsample

MD-FSL,TBSS,voxelwise analysis

Decreasedinfastspindledensity wasassociatedwithage-related increasedMDvaluesin commissuralandprojectingfiber tracts

Gaudreault[35] Crosssectional n=30young,22.92.7 n=31older,59.85.4 Independentsample

FA,MD,AD,RD-FSL,TBSS, voxelwiseanalysis

SleepspindleswererelatedtoWM modificationsoverfrontalregionsin youngbutnotolderadults

Actigraphicstudies Lim[37] Crosssectional n=45,33healthyolderand

12AD,89.4(85.2–93.5) Rushmemoryandaging project

Brainautopsy Highfragmentationofsleepand activestateswasrelatedtolow galanin-immunoreactiveneuronsin theintermediatenucleus.No associationwithself-reportedsleep duration,DIS,DMS

Lim[38] Crosssectional n=315older,90.46.1 Rushmemoryandaging project

Brainautopsy Highsleepfragmentationwas associatedwithsevere

arteriolosclerosisandsubcortical macroscopicinfarcts

Wang[39] Crosssectional n=17,10healthyolderand 7AD,90.45.5

Rushmemoryandaging project

Brainautopsy Lowrest-activityamplitudewas associatedwithlow VIP-immunoreactiveneuronsinthe suprachiasmaticnucleus

(4)

while an association between short sleep duration and

ventricular expansion over two years was detected in

anotherstudy[12].Besidesleep,Carvalhoetal.reported

that excessive daytime sleepiness was associated with

reducedhippocampalvolumeandcorticalthickness,

par-ticularly in temporal regions, suggesting that not only

alteredsleepbutalsoresultingdaytimesleepinessmaybe

related tobrainstructure [15].

Regarding white matter, poor sleep quality has been

associatedwiththepresenceofwhitematter

hyperinten-sities(WMH)incross-sectional andlongitudinalstudies

[16,17].WMHarecommonlyobservedduringageingand

reflectwhitemattertissue injurysuchasdemyelination,

axonallossandgliosis[18].Importantly,WMHhavebeen

related to cognitive decline, especially executive

dys-function, and incident dementia [18]. Another study

Table1 (Continued)

Paper Design n,meanageSD Project

Brainmeasures Mainfindings

Lim[40] Crosssectional n=141,82.9(median,IQR 77.4–87.4)

Rushmemoryandaging project

Volume,surfaceandareaof Desikan-Killianyatlas regions–Freesurfer

Highsleepfragmentationwas relatedtolowvolumeswithinthe lateralorbitofrontalandinferior frontalparsorbitalisandthickness withintherightinferiorfrontalpars orbitalis.Noassociationwith corticalarea,subcorticalandWM volume,self-reportedsleepduration andnocturnalawakenings Andre´ [41] Crosssectional n=30cognitively

unimpaired,73.37 n=36SCDand/orMCI, 71.58.2 IMPAP+study GMvolume–SPM12,voxel wiseanalysis

Highvariabilityofsleep

fragmentationduringthefirstpartof thenightwasrelatedtolowvolume withinthethalamusincognitively unimpairedolderadults.No associationwithsleep fragmentationintensity vanSomeren[42] Crosssectional n=138,69.18.5

Independentsample

Medialtemporallobeand globalcorticalatrophy– visualrattingscales

Highrest-activityfragmentationwas associatedwithmedialtemporal lobeatrophy

Baillet[43] Crosssectional n=58,76.10.5(SEM) AMImagestudy

GMvolume–VBM8,voxel wiseanalysis

FA,MD,RD,AD-FSL,TBSS, voxelwiseanalysis WMHvolume

Lowrest-activityamplitudeandhigh sleepfragmentationwererelatedto lowFA,highMDandRDvaluesin thewholeWM.Thisassociationwas partlyassociatedwithWMH.No associationwithrest-activity fragmentation,stabilityofactivity rhythms,sleepdurationandGM volume

Kocevska[44] Longitudinal5.8years n=1,201,59.37.9 Rotterdamstudy

FA,MD–tractography Shortsleepduration,highWASO andlowsleepefficiencywere relatedtoWMmodificationsin severalWMtractsuptosevenyears afterbrainimaging.Noassociation withsleeplatencyandwithWM modificationchangesovertime Oosterman[45] Crosssectional n=162,69.18.6

Independentsample

WMHvolume–visualrating scale

Lowrest-activityamplitudeand unstableactivityrhythmswere associatedwithfrontaldeepWMH. Noassociationwithperiventricular WMHandrest-activity

fragmentation Zuurbier[47] Crosssectional n=970,59.27.5

Rotterdamstudy

WMHvolume,lacunar infarctsandcerebral microbleeds

Rest-activityfragmentationwas associatedwithhighWMHvolume andcerebralmicrobleeds.No associationwithstabilityofactivity rhythms,sleepduration,WASO, self-reportedsleepqualityand lacunarinfarcts

Abbreviations:GM=greymatter,WM=whitematter,OFC=orbitofrontalcortex,mPFC=medialprefrontalcortex,dlPFC=dorsolateralprefrontal cortex,VIP=vasoactiveintestinalpolypeptideneurons,WMH=whitematterhyperintensities,VBM=voxelbasedmorphometry,FA=fractional anisotropy,MD=meandiffusivity,RD=radialdiffusivity,AD=axialdiffusivity,TBSS=tractbasedspatialstatistics;EMA=earlymorning awaken-ings,DIS=difficultyinitiatingsleep,DMS=difficultymaintainingsleep,SWS=slow-wavesleep,SWA=slow-waveactivity,WASO=wakeafter sleeponset.

(5)

observedthatpoorsleepquality,andespeciallylongsleep

latency, was associated with modifications of Diffusion

Tensor Imaging (DTI) parameters, particularly radial

diffusivity(RD),butnotwithWMHvolume.Persistence

ofpoorsleepqualitysixteenyearspriorMRIwashowever

notassociatedwithwhitemattermeasures[19].Ofnote,

ageingisrelatedtoreducedfractionalanisotropy(FA)and

increased diffusivity parameters (mean diffusivity,MD

andRD;lessprominentforaxialdiffusivity,AD)

reflect-ingimpairedfiberintegrityduetoalossofcoherenceona

preferred direction and an increase of diffusion rate.

Changes in RD are assumed to reflect myelin damage

whilechangesinAD presumablyreflectaxonaldamage

[20]. Kocevskaet al. revealed no significant association

betweenwhitemattermeasuresatbaselineandchanges

in sleep complaints over time, but observed that sleep

complaintsatbaselinewereassociatedwithdecreasedFA

values in brainstem tracts over a five-year longitudinal

design; thereby suggesting that sleep complaints may

predict future age-related white matter modifications

[17]. With respect to sleep duration, shorter durations

(<6hours)havebeenassociatedwithhighWMHvolume

in parietalareas [21] and increased WMHburden over

time [17]. Short self-reported sleepers presented also

higherMDvaluesmainlyinfrontal,parietalandtemporal

whitematterregionsand,higherprevalenceof vascular

risk factors compared to moderate sleepers [21]. Other

studiesdidnotobserveasignificantassociationbetween

self-reported sleep duration and WMH in older adults

[13,22]; note that Ramos et al. reported an association

betweenlongsleep duration(>9hours)andWMH

vol-umeonlyin olderadultssuffering fromdiabetes[22].

Association

between

brain

structure

and

electrophysiological

signatures

of

sleep

Sleeparchitectureischaracterizedbytheultradian

alter-nation between rapid eye movement (REM) and

non-REM (NREM) sleep as assessed by polysomnographic

(PSG) recordings. Standard PSG-assessed variables and

moreparticularlyfrequency-bandspecificspectralpower

exhibits stable and trait-like-interindividual differences

(e.g. [23]) which make them interesting candidates for

brain-behaviorcorrelations.Sleepspindleshavefor

exam-ple been proposed to reflect featuresof trait and

time-varyingpropertiesofneuroplasticity[24]andhavebeen

associatedwithgeneralintelligence(e.g.[25]).

Compro-mised generation and propagation of sleep oscillations

(spindlesandslowwaves)havebeenlinkedtoage-related

corticalthinning[26].

Post-mortem assessments from theHonolulu-Asia Aging

Study [27] revealed that short slow-wave sleep (SWS)

duration was associated with more global grey matter

atrophy at death [28]. In vivo MRI studies similarly

revealed that during adulthood,changes in slow waves

densityandamplitudewereassociated withage-related

corticalthinninginfrontal,temporal,parietalandinsular

regions[29].Importantly,age-relatedgreymatteratrophy

within the medial prefrontal cortex (mPFC) has been

relatedtodecreasedslow-waveactivity[30,31]andslow

wave—sleepspindlecoupling[32],bothassociatedwith

overnight sleep-dependent memory impairment. Note

howeverthatanotherstudyreported significant

associa-tionsbetweensleepspindlecharacteristicsandgrey

mat-tervolume in younger adultsonly [33].Withrespectto

white matter, Mander et al. observed that age-related

increase of MD values in several white matter tracts,

suchasthecorpuscallosum,wasassociatedwithalossof

frontalfastsleepspindles.Thiswhitemattermodification

moderated the impairment of overnight sleep

spindle-dependent motor memory consolidation [34]. Another

study failed to observe such association in a younger

agerange[35].

Association

between

brain

structure

and

rest-activity

cycles

The application of in-lab approaches allow to uncover

processesunderlyingdifferentsleepphenotypes.At

con-trast,fieldactigraphyenablesmonitoringof masked

24-hourrest-activitycyclesoverdays,weeksorevenmonths.

Difficulty however remains in the inference of

physio-logicallymeaningful sleep phenotypes.Strength of this

approach is that, unlike single-night polysomnographic

recordings, it allows a rather unique assessment of the

temporalprofilesofrest-activitypatternsandtherebythe

inferencesabout putative temporaldisruption of

sleep-wakestates.Assuch,notonlysleep,butalsothetemporal

organization of rest-activity cycles has been associated

withcognitivefunctioning(e.g. [3–5]).

Knowledgeonhistopathological changesrelatedto

rest-activity cycles in ageing mainly arises from actigraphic

studiesoftheRushMemory andAging Project[36].In

the latter, Lim et al. observed that higher rest-activity

fragmentationwasrelatedtoalowernumberof

galanin-immunoreactiveneuronswithintheintermediatenucleus

ofthehypothalamus(homologueofthesleep-promoting

ventrolateralpreopticnucleusinrodents;[37]).

Further-more, higher sleep fragmentation was associated with

more severe arteriolosclerosis and subcortical

macro-scopic infarcts at autopsy (27% and 31% higher odds

foreach1standarddeviationincreaseof sleep

fragmen-tation;[38]).Finally,Wangetal.reportedthatalower

24-hour cycle amplitude (i.e. less differentiation between

day-time activity and night-time rest) was related to a

lowernumber of vasoactiveintestinal polypeptide

neu-rons of the suprachiasmatic nucleus implicated in the

maintainance of a high-amplitude circadian output and

photicresetting[39].

Fewstudiesevaluatedtheassociationbetween

actigra-phy-derivedrest-activityparametersandgreymatterin

older adults. Highsleep fragmentation has been

(6)

and inferior frontal regions [40], and its variability

during the first part of the night with low thalamic

volume[41].Recently,vanSomerenetal.reportedthat

rest-activity fragmentation accounted for 19% of the

variance in medial temporal lobe atrophy, evaluated

withavisualratingscale[42].Incontrast,anotherstudy

did not observe any significant association between

actigraphic measures of rest-activity cycles and grey

matter volume[43].

Regarding white matter, a population-based study

reportedthatactigraphic derivedpoorsleepquality and

short sleep duration wererelated to whitematter

mod-ifications(highFAandlowMD)uptosevenyearslater

butnotwiththeevolutionofDTIparametersovertime.

Howeverasactigraphicrecordingswerenotperformedat

baseline, causality between sleep and brain structure

remains difficulttoinferatthis stage[44].Threecross

sectionalactigraphicstudiesprovidedfirstevidencethat

rest-activity disturbances observed in older adults are

related to cerebral small vessel disease manifestations.

Oosterman et al. observed that a lower 24-hour

rest-activityamplitude,especiallydecreaseindaytime

activ-ity,andunstableactivityrhythmswererelatedtohigher

deepWMHvolumesinfrontalregions[45].Inmorethan

900participantsoftheRotterdamStudy[46],rest-activity

fragmentation has been associated with WMH volume

and cerebral microbleeds but not with the number of

lacunarinfarcts[47].IntheonlystudyusingDTI,Baillet

et al. observed that low 24-hour rest-activity amplitude

andhigh sleepfragmentationwerebothassociatedwith

diffusewhitemattermodifications(lowFA,highMDand

RD) and WMH volume. Interestingly, rest-activity

amplituderemained associatedwith whitematter

prop-ertieswhenaccountingforsleepfragmentation,

suggest-ing aspecific effectof the temporaldistributionofrest

andactivityover the24-hourday[43].

Discussion

Ageing is associated with modifications of brain

struc-ture,sleep-wakeregulationandcognitionbuthowthese

factors are related to each other remains to be

deter-mined.Atthisstage,itmightappearprematuretodraw

firm conclusions regarding region-specific implications

(see also Figure 1a for a conceptual representation).

Regions of interest based approaches mainly targeted

on frontal, temporal, hippocampal, and thalamic grey

mattervolume,consideringthepredominantimplication

of those regions in brain ageing trajectory but also in

sleep-wakeregulationormorespecifically,inthe

gener-ation and/or coupling of slow oscillations and sleep

spindles.OfthirteenMRIstudiesincludingsubcortical

structures[8–13,15,32,33,40–42,43],onlyfourreported

apositiveassociationbetween sleep-wakepatternsand

hippocampal [10,15,42]orthalamicvolumes[10,41]. In

contrast, all polysomnographic studies included inthis

Figure1

(a)

(b) Self-reported sleep Actigraphy Polysomnography

Brain

ageing

Trait-like sleep phenotyping

Effect size

Sample size

Representativity

Duration Quality Night-time rest Rest-activity cycles SWS REM Spindles Cortical thinning White matter changes

CSVD manifestations Sleep perception

Sleep macro- and microstructure Rest-activity cycles

(e.g. mPFC, hypothalamus)

Sleep-wake states Diffuse process

Local process

Current Opinion in Behavioral Sciences

Conceptualrepresentationof(a)resultsfromstudiesreportinganassociationbetweensleep-wakestatesandMRI-derivedbrainstructureand(b) methodologicalconsiderations.

(7)

reviewsuggestthatadecreaseinslowwavesparameters

isassociatedwithcorticalatrophy,includingthemedial

prefrontal cortex [29–31,32] suggesting a possible

neuroanatomicalsubstratetoexplain sleep-related

cog-nitive impairment. Rather consistent results were

observed regarding white matter, and especially with

actigraphicrest-activitypatternsandcerebralsmall

ves-sel disease manifestations as WMH, cerebral

micro-bleeds and related cerebrovascular pathologies

[38,43,45,47]. Actingas adiffuse process,these

man-ifestationsdisruptnotonlythenormalappearingwhite

matterbutalso connected corticalregionsthrough

sec-ondaryneurodegeneration disrupting efficient

informa-tiontransferinbrainnetworks[48].Cerebralsmallvessel

diseaseisunderlinedbyreducedcerebralbloodflowand

blood-braindysfunctionandassociatedwithvascularrisk

factors [48], suggesting that rest-activity disturbances

mayberelatedtovascularabnormalities.

Mostofthestudiesexploringtheassociationbetweensleep

andstructuralbrainchangesatolder agereliedon

ques-tionnaire-basedsubjectiveassessments ofsleep duration

andquality.Thismightatleastpartiallyexplainthe

vari-abilityinobservedresultswithrespecttobrainstructural

changesovertheageingprocess.Itmightbeassumedthat

replicabilityofstructuralbrain-behaviorassociations

criti-callyreliesuponthedefinitionandprecisionofthesleep

traitunderinvestigation(e.g.[49]).Itisthusimportantto

completepopulation-basedsampleswithin-lab

measure-mentsofmoreunmaskedsleepphenotypeswhile

control-lingforpriorsleep-wakehistory.Withinthiscontext,

trait-likeand thusreplicable characteristics havebeen attributed

forexampletoelectrophysiological-derivedsleep

charac-teristics(e.g.[50]).Thosestudies,testingforanassociation

betweenelectroencephalographic-derivedspectralpower

measures(mainlyslow-waveactivity,spindledensityand

theircoupling)andbrainstructuremostlyrevealedstrong

effectsizes([29–31,34]butsee[33,35]).Addingcomplexity

tosuchphenotypes,asforinstancetakingintoaccountthe

dynamic interaction between REM and NREM sleep

whentestingfor putativeassociations will completethe

pictureabouttheimportanceofsleeponbrainageing[51].

It wasrecently observed forexamplethatNREMand REM

sleepstagesdifferentlyaffectcognitiveoutputslinkedto

theageing process(e.g.[52])therebyre-emphasizingon

REMsleepasanimportantcontributortobrainfunction.

Finally,theroleofthecircadianclockonshapinghuman

brain ageing has been widely underestimated. This

appearssurprisingbecauseofitswidespreadimplication

and regulation of cognitive brain function [2]. Even

thoughnotassessingtheinternalclock,actigraphyallows

toinfer abouttemporalprofiles of rest-activitypatterns.

Note that most studies have focused on sleep and its

disturbanceswithoutconsideringitasapartofa24-hour

cycle.Ofseven[40,41,42,43,44,45,47]studiesusingMRI

andactigraphy,onlyfourdescribedrest-activitypatterns

[42,43,45,47]andmostof thememphasizedan

associa-tion with white matter properties. Furthermore,

post-mortem studies provide evidence that actigraphic

rest-activity patterns are associated with degeneration of

hypothalamic neurons implicated in circadian timing

and sleep-wake regulation, regionsnot easilyreachable

withconventionalMRI[37,39].Thesestudieshinttothe

importance of the temporal framework given to sleep

regulation for brain fitness at older age. Older adults

present an advanced phase of circadian rhythms (e.g.

classicalendocrinemarkerssuchasmelatoninsecretion),

leadingtoearliersleeptimesandareducedamplitudein

bothcircadian sleepand wake consolidation [53],

puta-tivelyleadingtomorefragmentedactiveandsleepstates

(i.e. intrusion of sleep into daytime activities and

frag-mentationofnight-timesleepbywakeperiodsor

noctur-nalarousals;e.g. [54]).Note that atool regrouping

cur-rentlyavailableanalysismethodswasrecentlydeveloped

inourlabinordertocatchsuchtemporalspecificitiesin

actigraphy-derivedrest-activitydata,makingthistypeof

analysisopenlyavailableforfuturestudies(Pyactigraphy,

doi:https://doi.org/10.5281/zenodo.2537921).

Assummarizedin Figure1b,bothlarge-scale

epidemio-logicalstudiesbutalsocontrolledin-labquantificationof

specific sleep phenotypes present strengths and pitfalls

suchthattheirrespectiveoutputsshouldnotbetakenas

equalbutratherusedinacomplementarymanner.Inthe

samevein,inferencesweremostlybasedoncross-sectional

correlationsbetweenbrainandsleepvariables.

Longitudi-naldesignsaredecisivetodifferentiatebetween

intraindi-vidualandinter-individualvariabilityandtherebyofferthe

possibilitytodeterminewhethersleep-wakedisturbances

haveapredictivevalueoncognitiveandbrainchanges.In

subjectivecognitivedeclinepopulation,definedas

preclin-icalstageofAD,poorsleepqualityhasforexamplebeen

reportedtoprecedemedialtemporallobeatrophy[55].

Notethatourreviewfocusesonpossiblecontributionsof

sleep-wake states on non-symptomatic age-related

changesin brainstructure. Insights frompositron

emis-sion tomography imaging studies, not included in this

review, revealed for example that poor sleep and

frag-mented rest-activity cycles are related to Ab and Tau

burden(e.g. [56];seealso [57]for areview). The latter

might partially account for the modification in brain

structure,openingnewavenuestoexploretheassociation

betweensleepandage-relatedchangesinbrainintegrity.

Conflict

of

interest

statement

Nothingdeclared.

Acknowledgements

TheresearchersarefundedbyanERCStartingGrant(GA757763)andby theFondsdelaRechercheScientifiquedeBelgique(F.R.S.-FNRS).

(8)

References

and

recommended

reading

Papersofparticularinterest,publishedwithintheperiodofreview, havebeenhighlightedas:

 ofspecialinterest

1. DaanS,BeersmaDG,Borbe´lyAA:Timingofhumansleep: recoveryprocessgatedbyacircadianpacemaker.AmJ Physiol1984,246:R161-183.

2. MutoV,JasparM,MeyerC,Kusse´ C,ChellappaSL,DegueldreC, BalteauE,Shaffii-LeBourdiecA,LuxenA,MiddletonBetal.:Local modulationofhumanbrainresponsesbycircadian

rhythmicityandsleepdebt.Science2016,353:687-690.

3. LuikAI,ZuurbierLA,HofmanA,VanSomerenEJW,IkramMA, TiemeierH:Associationsofthe24-hactivityrhythmandsleep withcognition:apopulation-basedstudyofmiddle-agedand elderlypersons.SleepMed2015,16:850-855.

4. OostermanJM,vanSomerenEJW,VogelsRLC,VanHartenB, ScherderEJA:Fragmentationoftherest-activityrhythm correlateswithage-relatedcognitivedeficits.JSleepRes 2009,18:129-135.

5. TranahGJ,BlackwellT,StoneKL,Ancoli-IsraelS,PaudelML, EnsrudKE,CauleyJA,RedlineS,HillierTA,CummingsSRetal.: Circadianactivityrhythmsandriskofincidentdementiaand mildcognitiveimpairmentinolderwomen.AnnNeurol2011, 70:722-732.

6. ScullinMK,BliwiseDL:Sleep,cognition,andnormalaging: integratingahalfcenturyofmultidisciplinaryresearch. PerspectPsycholSci2015,10:97-137.

7. Ancoli-IsraelS:Sleepanditsdisordersinagingpopulations. SleepMed2009,10(Suppl.1):S7-S11.

8. StoffersD,MoensS,BenjaminsJ,vanTolM-J,PenninxBWJH, VeltmanDJ,VanderWeeNJA,VanSomerenEJW:Orbitofrontal graymatterrelatestoearlymorningawakening:aneural correlateofinsomniacomplaints? FrontNeurol2012,3:105.

9. BrangerP,Arenaza-UrquijoEM,TomadessoC,Me´zengeF, Andre´ C,FloresR,de,MutluJ,SayetteVdeL,EustacheF, Che´telatGetal.:Relationshipsbetweensleepqualityandbrain volume,metabolism,andamyloiddepositioninlate adulthood.NeurobiolAging2016,41:107-114.

10. LiuY-R,FanD-Q,GuiW-J,LongZ-L,LeiX,YuJ:Sleep-related brainatrophyanddisruptedfunctionalconnectivityinolder adults.BehavBrainRes2018,347:292-299.

11. SextonCE,StorsveAB,WalhovdKB,Johansen-BergH,FjellAM: Poorsleepqualityisassociatedwithincreasedcortical atrophyincommunity-dwellingadults.Neurology2014, 83:967-973.

12. LoJC,LohKK,ZhengH,SimSKY,CheeMWL:Sleepduration andage-relatedchangesinbrainstructureandcognitive performance.Sleep2014,37:1171-1178.

13. LutseyPL,NorbyFL,GottesmanRF,MosleyT,MacLehoseRF, PunjabiNM,ShaharE,JackCR,AlonsoA:Sleepapnea,sleep durationandbrainMRImarkersofcerebralvasculardisease andAlzheimer’sdisease:theatherosclerosisriskin communitiesstudy(ARIC).PLoSOne2016,11:e0158758.

14. SpiraAP,GonzalezCE,VenkatramanVK,WuMN,PachecoJ, SimonsickEM,FerrucciL,ResnickSM:Sleepdurationand subsequentcorticalthinningincognitivelynormalolder adults.Sleep2016,39:1121-1128.

15. CarvalhoDZ,StLouisEK,BoeveBF,MielkeMM,PrzybelskiSA, KnopmanDS,MachuldaMM,RobertsRO,GedaYE,PetersenRC etal.:Excessivedaytimesleepinessandfatiguemayindicate acceleratedbrainagingincognitivelynormallatemiddle-aged andolderadults.SleepMed2017,32:236-243.

16. DelBruttoOH,MeraRM,ZambranoM,LamaJ,DelBruttoVJ, CastilloPR:Poorsleepqualityandsilentmarkersofcerebral smallvesseldisease:apopulation-basedstudyin community-dwellingolderadults(TheAtahualpaProject).SleepMed2015, 16:428-431.

17. KocevskaD,CremersLGM,LysenTS,LuikAI,IkramMA, VernooijMW,TiemeierH:Sleepcomplaintsandcerebralwhite matter:aprospectivebidirectionalstudy.JPsychiatrRes2019, 112:77-82http://dx.doi.org/10.1016/j.jpsychires.2019.02.002.

18. PrinsND,ScheltensP:Whitematterhyperintensities,cognitive impairmentanddementia:anupdate.NatRevNeurol2015, 11:157-165.

19. SextonCE,ZsoldosE,FilippiniN,GriffantiL,WinklerA, MahmoodA,AllanCL,TopiwalaA,KyleSD,SpiegelhalderKetal.: Associationsbetweenself-reportedsleepqualityandwhite matterincommunity-dwellingolderadults:aprospective cohortstudy.HumBrainMapp2017,38:5465-5473http://dx.doi. org/10.1002/hbm.23739.

20. SalthouseTA:Neuroanatomicalsubstratesofage-related cognitivedecline.PsycholBull2011,137:753-784.

21. YaffeK,NasrallahI,HoangTD,LauderdaleDS,KnutsonKL, CarnethonMR,LaunerLJ,LewisCE,SidneyS:Sleepduration andwhitematterqualityinmiddle-agedadults.Sleep2016, 39:1743-1747.

22. RamosAR,DongC,RundekT,ElkindMSV,Boden-AlbalaB, SaccoRL,WrightCB:Sleepdurationisassociatedwithwhite matterhyperintensityvolumeinolderadults:theNorthern Manhattanstudy.JSleepRes2014,23:524-530.

23. TuckerAM,DingesDF,VanDongenHPA:Traitinterindividual differencesinthesleepphysiologyofhealthyyoungadults.J SleepRes2007,16:170-180.

24. KooPC,MarshallL:Neuroscience:asleeprhythmwithmultiple facets.CurrBiol2016,26:R813-815.

25. Bo´dizsR,KisT,La´za´rAS,Havra´nL,Rigo´ P,ClemensZ,Hala´szP: Predictionofgeneralmentalabilitybasedonneuraloscillation measuresofsleep.JSleepRes2005,14:285-292.

26. ZhongH-H,YuB,LuoD,YangL-Y,ZhangJ,JiangS-S,HuS-J, LuoY-Y,YangM-W,HongF-Fetal.:Rolesofaginginsleep. NeurosciBiobehavRev2019,98:177-184.

27. GelberRP,LaunerLJ,WhiteLR:TheHonolulu-Asiaagingstudy: epidemiologicandneuropathologicresearchoncognitive impairment.CurrAlzheimerRes2012,9:664-672.

28. GelberRP,RedlineS,RossGW,PetrovitchH,SonnenJA, ZarowC,Uyehara-LockJH,MasakiKH,LaunerLJ,WhiteLR: Associationsofbrainlesionsatautopsywith

polysomnographyfeaturesbeforedeath.Neurology2015, 84:296-303.

29. Dube´ J,LafortuneM,BedettiC,BouchardM,GagnonJF,DoyonJ, EvansAC,LinaJ-M,CarrierJ:Corticalthinningexplains changesinsleepslowwavesduringadulthood.JNeurosci 2015,35:7795-7807.

30. ManderBA,RaoV,LuB,SaletinJM,LindquistJR,Ancoli-IsraelS, JagustW,WalkerMP:Prefrontalatrophy,disruptedNREMslow waves,andimpairedhippocampal-dependentmemoryin aging.NatNeurosci2013,16:357-364.

31. VargaAW,DuccaEL,KishiA,FischerE,ParekhA,KoushykV, YauPL,GumbT,LeibertDP,WohlleberMEetal.:Effectsofaging onslow-wavesleepdynamicsandhumanspatialnavigational memoryconsolidation.NeurobiolAging2016,42:142-149.

32.

 HelfrichbrainscomeRF,ManderuncoupledBA,Jagustinsleep:WJ,slowKnightwave-spindleRT,WalkerMP:Old synchrony,brainatrophy,andforgetting.Neuron2018, 97:221-230.e4.

Recent studyreporting an association between slow wave — sleep spindlecouplingandspecificage-relatedgreymatteratrophyoverthe medialprefrontalcortex.

33. FogelS,VienC,KarniA,BenaliH,CarrierJ,DoyonJ:Sleep spindles:aphysiologicalmarkerofage-relatedchangesin graymatterinbrainregionssupportingmotorskillmemory consolidation.NeurobiolAging2017,49:154-164.

34. ManderBA,ZhuAH,LindquistJR,VilleneuveS,RaoV,LuB, SaletinJM,Ancoli-IsraelS,JagustWJ,WalkerMP:Whitematter structureinolderadultsmoderatesthebenefitofsleep

(9)

spindlesonmotormemoryconsolidation.JNeurosci2017, 37:11675-11687.

35. GaudreaultP-O,GosselinN,LafortuneM,Deslauriers-GauthierS, MartinN,BouchardM,Dube´ J,LinaJ-M,DoyonJ,CarrierJ:The associationbetweenwhitematterandsleepspindlesdiffersin youngandolderindividuals.Sleep2018,41.

36. BennettDA,SchneiderJA,BuchmanAS,MendesdeLeonC, BieniasJL,WilsonRS:Therushmemoryandagingproject: studydesignandbaselinecharacteristicsofthestudycohort. Neuroepidemiology2005,25:163-175.

37. LimASP,EllisonBA,WangJL,YuL,SchneiderJA,BuchmanAS, BennettDA,SaperCB:Sleepisrelatedtoneuronnumbersinthe ventrolateralpreoptic/intermediatenucleusinolderadultswith andwithoutAlzheimer’sdisease.Brain2014,137:2847-2861.

38.

 fragmentation,LimASP,YuL,Schneidercerebralarteriolosclerosis,JA,BennettDA,BuchmanandbrainAS:infarctSleep pathologyincommunity-dwellingolderpeople.Stroke2016, 47:516-518.

Post-mortemstudyreporting anassociationbetween actigraphy-esti-matedsleepfragmentationandcerebrovascularpathologies.

39.

 WangBuchmanJL,LimAS,AS,BennettChiangDA,W-Y,HuK,HsiehSaperW-H,CB:LoSuprachiasmaticM-T,SchneiderJA, neuronnumbersandrest-activitycircadianrhythmsinolder humans.AnnNeurol2015,78:317-322.

Post-mortemstudyreporting anassociationbetween actigraphy-esti-matedrest-activitypatternsandnumberofVIP-expressingneuronsinthe suprachiasmaticnucleus.

40. LimASP,FleischmanDA,DaweRJ,YuL,ArfanakisK,BuchmanAS, BennettDA:Regionalneocorticalgraymatterstructureand sleepfragmentationinolderadults.Sleep2016,39:227-235.

41. Andre´ C,TomadessoC,deFloresR,BrangerP,RehelS, Me´zengeF,LandeauB,SayetteV,dela,EustacheF,Che´telatG etal.:Brainandcognitivecorrelatesofsleepfragmentationin elderlysubjectswithandwithoutcognitivedeficits.Alzheimers Dement(Amst)2019,11:142-150.

42. VanSomerenEJW,OostermanJM,VanHartenB,VogelsRL, GouwAA,WeinsteinHC,PoggesiA,ScheltensP,ScherderEJA: Medialtemporallobeatrophyrelatesmorestronglyto sleep-wakerhythmfragmentationthantoageoranyotherknown risk.NeurobiolLearnMem2018,160:132-138http://dx.doi.org/ 10.1016/j.nlm.2018.05.017.

43.

 BailletActivity/restM,DilharreguycycleandB,Pe´re`sdisturbancesK,DartiguesofstructuralJ-F,MayobackboneW,CathelineofG: cerebralnetworksinaging.Neuroimage2017,146:814-820.

Diffusiontensor imagingstudyreportinganassociationbetween acti-graphy-estimatedrest-activitypatternsandwhitemattermodifications. 44.

 KocevskaSomerenEJW,D,TiemeierIkramMA,H,LysenVernooijTS,MW,deGrootLuikAI:M,TheMuetzelprospectiveRL,Van associationofobjectivelymeasuredsleepandcerebralwhite mattermicrostructureinmiddle-agedandolderpersons. Sleep2019,42 http://dx.doi.org/10.1093/sleep/zsz140.

Longitudinal studyreporting an association between actigraphy-esti-matedsleepqualityandwhitemattermodificationsuptosevenyears later,butnowithdiffusionparameterschangesovertime.

45. OostermanJ,vanHartenB,VogelsR,GouwA,WeinsteinH, ScheltensP,ScherderE:Distortionsinrest-activityrhythmin agingrelatetowhitematterhyperintensities.NeurobiolAging 2008,29:1265-1271.

46. HofmanA,BretelerMMB,vanDuijnCM,KrestinGP,PolsHA, StrickerBHCH,TiemeierH,UitterlindenAG,VingerlingJR, WittemanJCM:Therotterdamstudy:objectivesanddesign update.EurJEpidemiol2007,22:819-829.

47. ZuurbierLA,IkramMA,LuikAI,HofmanA,VanSomerenEJW, VernooijMW,TiemeierH:Cerebralsmallvesseldiseaseis relatedtodisturbed24-hactivityrhythms:apopulation-based study.EurJNeurol2015,22:1482-1487.

48. WardlawJM,SmithC,DichgansM:Smallvesseldisease: mechanismsandclinicalimplications.LancetNeurol2019, 18:684-696.

49. KharabianMasoulehS,EickhoffSB,HoffstaedterF,GenonS: Alzheimer’sdiseaseneuroimaginginitiative:empirical examinationofthereplicabilityofassociationsbetweenbrain structureandpsychologicalvariables.eLife2019,8.

50. FinelliLA,AchermannP,Borbe´lyAA:Individual“fingerprints”in humansleepEEGtopography.Neuropsychopharmacology 2001,25:S57-S62.

51. ScullinMK,GaoC:Dynamiccontributionsofslowwavesleep andREMsleeptocognitivelongevity.CurrSleepMedRep 2018,4:284-293.

52. dellaMonicaC,JohnsenS,AtzoriG,GroegerJA,DijkD-J:Rapid eyemovementsleep,sleepcontinuityandslowwavesleepas predictorsofcognition,mood,andsubjectivesleepqualityin healthymenandwomen,aged20–84years.FrontPsychiatry 2018,9.

53. SchmidtC,PeigneuxP,CajochenC:Age-relatedchangesin sleepandcircadianrhythms:impactoncognitive

performanceandunderlyingneuroanatomicalnetworks.Front Neurol2012,3.

54. HuangY-L,LiuR-Y,WangQ-S,VanSomerenEJW,XuH,Zhou J-N:Age-associateddifferenceincircadiansleep-wakeand rest-activityrhythms.PhysiolBehav2002,76:597-603.

55. LauriolaM,EspositoR,PizziSD,deZambottiM,LondrilloF, KramerJH,RabinoviciGD,TartaroA:Sleepchangeswithout medialtemporallobeorbraincorticalchangesin community-dwellingindividualswithsubjectivecognitivedecline. Alzheimer’sDement2017,13:783-791.

56. LuceyBP,McCulloughA,LandsnessEC,ToedebuschCD, McLelandJS,ZazaAM,FaganAM,McCueL,XiongC,MorrisJC etal.:Reducednon–rapideyemovementsleepisassociated withtaupathologyinearlyAlzheimer’sdisease.SciTranslMed 2019,11:eaau6550.

57. VanEgrooM,NarbutasJ,ChylinskiD,VillarGonza´lezP,MaquetP, SalmonE,BastinC,ColletteF,VandewalleG:Sleep-wake regulationandthehallmarksofthepathogenesisof Alzheimer’sdisease.Sleep2019,42.

Figure

Updating...

References