Neuroimaging findings in primary insomnia

Etude de l’insomnie primaire par imagerie cerebrale

J.N. O’Byrne a,b, M. Berman Rosa c, J.-P. Gouin c, T.T. Dang-Vu a,*,b,d

a Department of Exercise Science, Concordia University, 7141 Sherbrooke St W, Montreal, Quebec, H4B 1R6 Canadab Center for Studies in Behavioral Neurobiology, Concordia University, 7141 Sherbrooke St W, Montreal, Quebec, H4B 1R6 Canadac Department of Psychology, Concordia University, 7141 Sherbrooke St W, Montreal, Quebec, H4B 1R6 Canadad Institut Universitaire de Geriatrie de Montreal, Universite de Montreal, 4565, chemin Queen-Mary, Montreal, Quebec, H3W 1W5 Canada

Pathologie Biologie 62 (2014) 262–269

A R T I C L E I N F O

Article history:Received 30 October 2013Accepted 13 May 2014Available online 14 August 2014

Keywords:InsomniaNeuroimagingSleepSleep disorderHyperarousalPositron emission tomographySingle-photon emission computedtomographyMagnetic resonance imagingMagnetic resonance spectroscopy

Mots cles :InsomnieNeuroimagerieSommeilTroubles du sommeilHyperactivationTomographie par emission de positronsTomographie d’emission monophotoniqueImagerie par resonance magnetiqueSpectroscopie en resonance magnetiquenucleaire

A B S T R A C T

State-of-the-art neuroimaging techniques have accelerated progress in the study and understanding ofsleep in humans. Neuroimaging studies in primary insomnia remain relatively few, considering theimportant prevalence of this disorder in the general population. This review examines the contributionof functional and structural neuroimaging to our current understanding of primary insomnia. Functionalstudies during sleep provided support for the hyperarousal theory of insomnia. Functional neuroimagingalso revealed abnormalities in cognitive and emotional processing in primary insomnia. Results fromstructural studies suggest neuroanatomical alterations in primary insomnia, mostly in the hippocampus,anterior cingulate cortex and orbitofrontal cortex. However, these results are not well replicated acrossstudies. A few magnetic resonance spectroscopy studies revealed abnormalities in neurotransmitterconcentrations and bioenergetics in primary insomnia. The inconsistencies among neuroimagingfindings on insomnia are likely due to clinical heterogeneity, differences in imaging and overall diversityof techniques and designs employed. Larger samples, replication, as well as innovative methodologiesare necessary for the progression of this perplexing, yet promising area of research.

! 2014 Elsevier Masson SAS. All rights reserved.

R E S U M E

Les techniques d’imagerie cerebrale ont permis des avancees considerables dans l’etude du sommeil chezl’humain. Cependant, les etudes par imagerie cerebrale dans l’insomnie primaire demeurent peunombreuses, particulierement en regard de la prevalence importante de ce trouble du sommeil dans lapopulation generale. Cette revue examine la contribution des etudes d’imagerie cerebrale fonctionnelle etstructurelle a la comprehension de l’insomnie primaire. Les etudes d’imagerie fonctionnelle au cours dusommeil appuient la theorie de l’hyperactivation dans l’insomnie. D’autres etudes fonctionnelles ont reveledes alterations dans le traitement cerebral des processus cognitifs et emotionnels dans l’insomnie primaire.Les resultats des etudes structurelles suggerent des modifications neuroanatomiques, particulierementdans l’hippocampe, le cortex cingulaire anterieur et le cortex orbitofrontal. Cependant, ces resultats ne sontpas concordants d’une etude a l’autre. Quelques etudes spectroscopiques ont revele des alterations dans lesniveaux de neurotransmetteurs, ainsi que des changements bioenergetiques dans l’insomnie primaire. Lemanque de concordance entre les resultats d’imagerie cerebrale en insomnie pourrait etre lie al’heterogeneite des differentes populations cliniques etudiees, ainsi qu’a la diversite des techniquesd’imagerie et d’analyse employees. La neuroimagerie constitue une voie d’exploration prometteuse del’insomnie, mais la poursuite des avancees dans ce domaine necessite de reunir de plus grands echantillons,de reproduire et confirmer les resultats existants, tout en developpant l’utilisation de nouvelles modalites.

! 2014 Elsevier Masson SAS. Tous droits reserves.

* Corresponding author.E-mail address: tt.dangvu@concordia.ca (T.T. Dang-Vu).

Available online at

ScienceDirectwww.sciencedirect.com

http://dx.doi.org/10.1016/j.patbio.2014.05.0130369-8114/! 2014 Elsevier Masson SAS. All rights reserved.

1. Introduction

Insomnia is a remarkably prevalent disorder. Depending on thedefinition used, it affects 6–20% of the general population [1–5]. Asa result, sleep dissatisfaction counts among the most commonhealth complaints in primary care [6] and the associatedhealthcare expenditures, in addition to the costs of sleep aidsand absenteeism at work, contribute to a considerable economicburden [7,8]. Symptoms of insomnia include difficulties fallingasleep and staying asleep, and feelings of non-restorative sleep [4].Daytime fatigue, mood disruption and cognitive impairmentsassociated with insomnia negatively affect productivity andquality of life [9–11]. While insomnia symptoms can be a transientresponse to stress or changes in sleep-wake schedule, 70% ofindividuals with insomnia display persistent symptoms for morethan three months (i.e., chronic insomnia) [12].

Relatively few neuroimaging studies have examined thephysiology of this common sleep disorder [13]. Neuroimagingtechniques can be useful in identifying the cerebral mechanisms ofinsomnia pathogenesis, and the neural correlates of insomniasymptoms. In this paper, we review the findings of thesepioneering studies, which examined insomnia through the lensesof single-photon emission computed tomography (SPECT), posi-tron emission tomography (PET), magnetic resonance imaging(MRI), functional MRI (fMRI) and magnetic resonance spectroscopy(MRS). PET, SPECT and fMRI, are functional modalities thatexamine changes in brain metabolism, blood flow or bloodoxygenation. Structural modalities, such as MRI and MRS, mapout subtle changes in brain anatomy and content. In synthesizingthe strengths and limitations of these studies, we propose futuredirections in this expanding area of research. The scope of thisreview will be limited to primary insomnia (PI), which is defined bysleep disturbances occurring in the absence of comorbid medical orpsychological conditions [14].

2. Functional neuroimaging

2.1. PET and SPECT

The first neuroimaging studies to examine PI used PET andSPECT functional imaging techniques. PET and SPECT bothinvolve the injection of a radiolabeled isotope (the tracer) intothe bloodstream. Depending on the tracer employed, the scanscan offer indices of cerebral blood flow, cerebral metabolic rateof glucose (CMRglu) or neurotransmission. Smith et al. [15]employed SPECT with technetium-99m-hexamethylpropylena-mine oxime (99mTc-HMPAO), a gamma-emitting radionuclideimaging agent, in order to observe regional cerebral blood flowduring non-rapid-eye-movement (NREM) sleep in 5 PI patientsand 4 good sleepers, all 9 of them female. Compared to controls,PI patients displayed cerebral hypoperfusion during NREM sleepin eight pre-selected regions of interest. The most pronouncedhypoperfusions were observed in the basal ganglia (Fig. 1), andto a lesser extent in the frontal medial, occipital and parietalcortices. In a later study, the same group re-scanned 4 of the 5 PIpatients after 8 weeks of behavioral therapy for insomnia. Theyfound that a 43% reduction in sleep onset latency after treatmentwas accompanied by a 24% restoration of regional cerebral bloodflow, especially in the basal ganglia [16]. These changes werethought to represent normalization of sleep processes. Theauthors further speculated that increased sleep debt from partialsleep deprivation in PI may accentuate the normal cerebraldeactivation during sleep, as a homeostatic compensatorymechanism.

In contrast, the next functional study by Nofzinger et al.provided support for the hyperarousal theory of insomnia [17]. The

hyperarousal theory explains PI as a fundamental imbalance in thesleep-promoting and arousal systems, resulting in a state of globalcortical and physiological arousal across the sleep-wake cycle[20,21]. In the study by Nofzinger et al., 7 men and women with PIwere compared to 20 age- and gender-matched healthy controlsduring wakefulness and NREM sleep, using 18F-fludeoxyglucose(18F-FDG) PET in order to measure regional cerebral metabolism,indexed by CMRglu. In line with hyperarousal theory, PI patientsrelative to controls were found to have a smaller reduction inrelative metabolism from wakefulness to NREM sleep in theascending reticular activating system, hypothalamus, thalamus,hippocampus, anterior cingulate cortex (ACC), medial prefrontaland insular cortices (Fig. 1). In addition, PI patients had lowerwaking metabolism than healthy controls in cortical (bilateralfrontal, left superior temporal, parietal and occipital cortices) andsubcortical regions (thalamus, hypothalamus and brainstemreticular formation).

Nofzinger et al.’s results lend support to Espie’s integratedpsychobiological inhibition model [22], according to whichheightened arousal in PI is attributable to the inhibition of normalcortical deactivation during the transition from waking to NREMsleep. This model at once explains two major symptoms ofinsomnia:

! difficulty falling asleep because of restricted sleep onset-relatedcortical inhibition and;! difficulty staying asleep because of the same disinhibition

occurring following arousals over the course of the night. Thesearousals would otherwise go unnoticed because of rapid corticaldeactivation in normal sleep [22].

Fig. 1. Primary insomia: functional studies. Regional cerebral metabolism duringwake and NREM sleep in PI. Smith et al. [15,16] found reduced regional cerebralblood flow (SPECT) in the basal ganglia in insomniacs. Nofzinger et al. [17] foundsmaller reductions in regional metabolism (18F-FDG PET) during the transition tofrom wake to NREM sleep in patients with PI. Nofzinger et al. [18] found acorrelation between WASO and metabolism in thalamocortical pathways and thepontine tegmentum. Altena et al. [19] and Nofzinger et al. [17] found evidence forprefrontal deactivation during wake. Adapted from Desseilles et al. [13], and fromillustrations by Patrick J. Lynch and C. Carl Jaffe. http://creativecommons.org/licenses/by/2.5/.

J.N. O’Byrne et al. / Pathologie Biologie 62 (2014) 262–269 263

Nofzinger’s research group extended their findings in anotherPET study using 18F-FDG to examine CMRglu in relation to wakeafter sleep onset (WASO) in a sample of 15 patients with PI [18].They found positive correlations between WASO, and increasedglucose metabolism during NREM sleep in the thalamus, ACC,temporal and frontal cortices, as well as in the pontine tegmentum(Fig. 1). Conspicuously, these regions largely overlap with thoseregions from their previous study showing less reduction in thetransition from waking to NREM sleep in PI.

2.2. Functional MRI studies

Functional imaging can be obtained from blood-oxygen-leveldependent (BOLD) contrast fMRI. The BOLD signal is sensitive tothe relative decrease in deoxyhemoglobin concentration thatfollows the local increase in cerebral blood flow in an activatedbrain area. Functional MRI does not require the injection of a tracer,but requires that the subject lies in the scanner in real-time duringthe period of interest. A first fMRI study on insomnia by Altenaet al. [19] examined BOLD response during the completion of averbal fluency task. Previous research alludes to cognitivedysfunction in insomnia [23], but behavioral studies have yieldedinconsistent results [24,25]. In order to detect cerebral alterationsduring cognitive performance, the researchers asked 21 olderadults with PI (17 females, mean age " s.d.: 61 " 6.2 years) and12 age-, sex- and education-matched controls to complete letter andcategory fluency tasks, with a counting-backwards task provided as abaseline. During the cognitive tasks, PI patients showed hypoactiva-tion in the left prefrontal cortex and left inferior frontal gyrus, incomparison to good sleepers (Fig. 1). As part of this same fMRI study,PI patients underwent 6 weeks of multimodal non-pharmacologicaltherapy, including cognitive behavioral therapy (CBT) and lightexposure therapy. Post-therapy, sleep efficacy and sleep onset latencyimproved significantly in PI patients. Activation in PI patients waspartially restored in the medial prefrontal cortex during the categoryfluency task, and in the inferior frontal gyrus during the letter fluencytask. The authors concluded that individuals with PI are cognitivelycompromised, as shown by altered brain responses during taskperformance [19]. Furthermore, these effects on brain responsesappear reversible with treatment. It should be noted that thegeneralization of this study is limited to older populations, given thesexagenarian mean age of the participants.

To further investigate cognitive impairments in PI, Drummondet al. scanned a sample of 25 young adults PI patients (12 females,mean age " s.d.: 32.3 " 7.2 years) and 25 controls matched for sex,age and education, while they completed an N-back working memorytask [26]. During the N-back task, individuals with PI showed lessactivation than good sleepers in the thalamus, fronto-parietal cortexand cerebellum, brain regions normally associated with workingmemory and motor and visual processing. No relationship wasobserved between task difficulty and brain activation for PI patients,whereas good sleepers showed increasing activation in these regionsin direct proportion to task difficulty. This indicates that insomniacsfailed to recruit brain areas typically engaged for performance of thistask. Furthermore, good sleepers showed deactivation of the middlefrontal gyrus, posterior cingulate and orbital frontal gyrus, brainregions involved in the ‘default mode’ network, with increasing taskdifficulty, whereas PI patients showed no change in these regions. Thedefault mode network is composed of brain regions active when thebrain is not otherwise engaged in goal-oriented behavior [27]. Lack ofdeactivation of the default mode network may indicate an inability todeactivate task-irrelevant brain regions during performance. The datasupport a PI cognitive task-performance profile characterized byfailure to engage task-appropriate processes, while failing todisengage task-irrelevant processes. Actual performance in the taskwas however, unchanged. These data may explain the frequent

subjective PI complaints of reduced cognitive performance, in theabsence of actual deficits [26].

Insomnia is often comorbid with emotional disorders, andelevated emotional reactivity is thought to represent an importantfactor in the etiology of insomnia [28–30]. Accordingly, Huanget al. investigated abnormalities in PI emotional processing usingresting-state fMRI connectivity analysis [31]. Resting-state fMRIhas emerged as a useful tool for identifying broadly connectedfunctional networks. The researchers compared 10 medication-naıve PI participants with 10 age- and sex-matched healthycontrols during wakefulness. They found several abnormalities inPI emotional network connectivity compared to controls. Speci-fically, functional connectivity between the amygdala and broadlydistributed cortical and subcortical areas was altered in PIcompared to controls. Furthermore, an observed increase inamygdala to premotor cortex, sensorimotor cortex connectivitywas correlated with total Pittsburgh Sleep Quality Index (PSQI)score, a subjective measure of sleep quality. Greater connectivitybetween the amygdala, a fear and threat processing centre, and thepremotor cortex, which prepares motor action in response tothreat perception, may reflect a hyper-reactivity to perceivedthreat in PI patients [32,33]. An elevated threat response isconsistent with the hyperarousal model of PI [31].

3. Structural neuroimaging

The present section examines the findings that structuralneuroimaging techniques (with MRI and MRS) have revealedconcerning anatomical and molecular brain changes associatedwith PI.

3.1. Volumetric differences in the hippocampus

Cognitive deficits have been observed in individuals with PI,including impairment in hippocampus-dependent memory con-solidation [23,34]. Together with evidence of suppressed hippo-campal neurogenesis in sleep-deprived rats [35,36], these findingsimply possible structural changes in the brains of individuals withPI, particularly in the hippocampus. Riemann et al. [37] wereamong the first to employ structural neuroimaging to investigateneuroanatomical differences between good sleepers and indivi-duals with PI. Using MRI (1.5 Tesla), they measured dorsolateralprefrontal cortex (dlPFC), hippocampus, orbitofrontal cortex (OFC),ACC and amygdala volumes in individuals with PI (n = 8, meanage " s.d.: 48.4 " 16.3 years) compared to controls. Of these regions,the hippocampus was found to be significantly reduced in the PIgroup (Fig. 2). This finding, however, was rendered non-significantupon familywise correction for multiple comparisons.

Winkelman et al. [38] also scanned for differences inhippocampal volume between normal sleepers and individualssuffering from PI. Their study included 20 PI (mean age " s.d.:39.3 " 8.7 years) and 15 controls and employed a 3.0 Tesla MRIscanner. Contrary to Riemann et al. [37], no differences inhippocampal volume between the PI group and healthy controlswere observed. However, in the PI group, there was a correlationbetween reduced volumes in the bilateral hippocampus andactigraphy-derived poor sleep efficiency and increased WASO. Thesethree findings were later corroborated in a retrospective study by thesame group [39], examining two independent samples from previousstudies [38,44] totalling 41 PI patients and 35 controls.

A recent study by Noh et al., [40] obtained MRI (1.5 Tesla) datafrom 20 physician-referred PI subjects (18 females; meanage " s.d.: 50.8 " 10.8 years) and 20 healthy sleepers. Similar tothe results of Winkelman et al. [38], the volumes of the left and righthippocampi in individuals with PI did not significantly differ from

J.N. O’Byrne et al. / Pathologie Biologie 62 (2014) 262–269264

that of healthy controls. However, it was found that smaller volumesof either the left or right hippocampus inversely correlated withinsomnia duration and polysomnography- (PSG-) defined arousalindex. Participants also completed a battery of neuropsychologicaltests. In accordance with previous findings [45], individuals with PIhad poorer performance in tests for attention, working memory,verbal and visual memory compared to controls. Lower hippocampalvolumes were associated with decreased cognitive performances.

Spiegelhalder et al. [41] investigated neuroanatomical changesin a clinically-referred sample of patients with PI using MRI. Theirsample consisted of 28 patients (18 females, mean age " s.d.:43.7 " 14.2 years) diagnosed with PI and 38 good sleeper controls.Unlike Riemann et al. [37], their analyses revealed no statisticallysignificant between-group differences in hippocampal volumes.Furthermore, no significant correlations were found between self-reported measure of insomnia severity or total sleep time and left orright hippocampal volumes.

Some differences among studies regarding the anatomicaldelineation of the hippocampus should be noted. WhereasRiemann et al. [37] included the alveus, fimbria and hippocam-pal-amygdala transition area (HATA) in calculating hippocampalvolume, Noh et al. [40] excluded the fimbria and HATA, andWinkelman et al. [38,39] excluded all three areas. Such metho-dological inconsistencies may in part be responsible for thediscrepant findings about hippocampal volume in PI.

3.2. Volumetric differences in the ACC

Nofzinger et al. [17,21] identified the ACC as an area of interestin the neurobiology of insomnia. In their PET study, the ACC wasamong the regions that showed smaller reductions in activationfrom wakefulness to sleep in PI patients than in controls. Thisfinding was corroborated by an observed reduction in GABA in thisregion in PI subjects relative to controls [46]. Winkelman et al.[39] retrospectively analyzed MRI data collected from twoindependent studies (i.e., [38] and [46]) with comparable designsand sample characteristics. The primary area of interest inves-tigated was the bilateral rostral ACC (rACC). Morphometric

analysis revealed that, compared to healthy controls, bothindependent PI samples presented with significantly larger rACCvolumes (Fig. 2). Furthermore, this increase in volume in the rACCwas positively correlated with sleep onset latency and WASO.That is, a larger volume in this region was associated with worsesleep [39]. In addition, in the course of the original MRS study [44](described below), the PI patients in the second sample werefound to have significantly lower levels of inhibitory neurotrans-mitters in the ACC than controls. The ACC is believed to beinvolved in cognitive and emotional processing [47], andalterations in rACC volume have been associated with majordepressive disorder [48]. The increases in rACC volume andaltered neurotransmission observed may then relate to emotionaldysregulation in PI [28], as well as to high comorbidity of PI withdepression [30].

3.3. Fronto-parietal volumetric differences

With the use of voxel-based morphometry (VBM), Altena et al.[42] examined whether volumetric differences in white and graymatter concentrations existed between a PI group composed of24 participants (17 females, mean age " s.d.: 60.3 " 6.0 years) and acontrol group comprising 13 good sleepers. Individuals withinsomnia had smaller volumes of gray matter in three areas: theleft OFC, the bilateral anterior precuneus of the parietal cortex and thebilateral posterior precuneus in the occipitoparietal cortex, comparedto controls (Fig. 2). Notably, the reduction in the OFC was stillsignificant after familywise correction for multiple comparisons. Anegative correlation between insomnia severity, as measured by theSleep Disorder Questionnaire, and left OFC volume was observed. Thereduction in hippocampal volume observed by Riemann et al. [37]was not replicated.

Joo et al. [43] also used VBM to analyze differences in gray andwhite matter volumes between a PI cohort (n = 27, 25 females,mean age " s.d.: 52.3 " 7.8 years) and a healthy control group. Incontrast to earlier VBM studies, the researchers employed SPM8-based VBM, which introduces a registration method termedDiffeomorphic Anatomical Registration Through ExponentiatedLie algebra (DARTEL) to the morphometric analysis. DARTELincreases the sensitivity and accuracy of VBM in the detection ofdifferences in gray and white matter volumetric composition. Inorder to assess the cognitive features associated with PI and theirpossible relationship with neuroanatomical changes, participantswere asked to complete several neuropsychological tests assessingattention, working memory, executive function and verbal function.Imaging analysis revealed a significant reduction of gray matterconcentration in the dlPFC of PI subjects compared to controls(Fig. 2). This included the bilateral superior, middle and inferiorfrontal gyri. The gray matter decrements also extended to the OFC,in line with the findings reported by Altena et al. [42]. In addition,individuals with PI scored significantly lower than controls on testsof attention, and nonverbal memory. Worse cognitive performanceon nonverbal memory tasks was correlated with PSG-derivedshorter total sleep time and poorer sleep efficiency, indicating a linkbetween poor sleep and memory dysfunction. Furthermore, anegative correlation was found between gray matter concentrationsin the left middle frontal gyrus and the Insomnia Severity Index.There was however no correlation between cognitive performancesand grey matter concentrations. It should be noted that none of thegroup differences in gray matter concentrations remained sig-nificant after correction for multiple comparisons.

Spiegelhalder et al. [41] investigated fronto-parietal volumetricdifferences in a large sample (PI: 28; controls: 38). Images wereacquired using a 3.0-Tesla MRI scanner and were subsequentlyanalyzed via VBM using DARTEL registration for differences in grayand white matter concentrations. In contrast with findings by Joo

Fig. 2. Primary insomnia: structural studies. Volumetric differences and differencesin gray matter volume in PI compared to good sleeper controls. Riemann et al. [37]found that the hippocampus was significantly smaller in PI patients, but this findingwas not replicated [38–43]. Winkelman et al. [39] found the ACC to be enlarged inPI. Altena et al. [42] and Joo et al. [43] found reductions in gray matter densities indifferent neocortical areas. Adapted from Desseilles et al. [13], and fromillustrations by Patrick J. Lynch and C. Carl Jaffe. http://creativecommons.org/licenses/by/2.5/.

J.N. O’Byrne et al. / Pathologie Biologie 62 (2014) 262–269 265

et al. [43] and Altena et al. [42], VBM analysis revealed nosignificant between-group differences in gray and white matterconcentrations.

There is little agreement in the white and gray matterconcentration data, with the exception that both Altena et al.[42] and Joo et al. [43] found reduced gray matter concentra-tions in the OFC of individuals with PI, compared to controls.Spiegelhalder et al. [41] found no such difference in a largersample, however. Still, the observed prefrontal neuroanatomicalchanges may hold significance with regard to PI cognition.This hypothesis is tempered by the lack of any correlationbetween gray matter deficits and neuropsychological scores[43], so it is equally likely that these deficits relate insteadto other processes associated with PI, such as emotionaldysregulation [28].

3.4. MRS studies

In vivo MRS is a non-invasive method allowing theestimation of relative concentrations of specific molecules inthe brain. Building on recent neuroimaging advances, Winkel-man et al. utilized single-voxel proton MRS (1H-MRS) tocompare daytime in vivo levels of gamma-aminobutyric acid(GABA) neurotransmission in 16 PI patients (8 females, meanage " s.d.: 37.3 " 8.1 years) and 16 healthy controls matched forage and sex [49]. Winkelman et al. determined that global GABAlevels were 30% lower in PI patients than in healthy controls. Lowerglobal GABA levels were also associated with higher PSG-quantified WASO within the PI group. As GABA is the primaryinhibitory neurotransmitter in the human brain, deficits in GABAare likely to result in difficulty regulating cortical arousal at night,in accord with hyperarousal theory. It is important to note that thisfirst study averaged the entire brain GABA concentration into oneglobal index lacking spatial resolution.

A subsequent study from the same group utilized 1H-MRS toestimate differences in GABA levels among PI patients with greaterspatial specificity [44,50]. In a new sample of 20 PI patients and 20age- and sex-matched controls, Plante et al. obtained resultsconsistent with their first study. Among PI patients, GABA levelswere significantly lower in the occipital cortex by 33% and in theACC by 21%, but were unchanged in the thalamus, compared togood sleepers [44]. In contrast, Morgan et al. [50] observed thatGABA levels were 12% higher in the occipital cortex of PI patientscompared to good sleepers. Morgan et al. [50] also detected anegative correlation between global GABA levels and PSG-measured WASO across both groups. Plante et al. [46] notedabout this apparent discrepancy that there was a limited overlapbetween the voxels each study used to measure GABA levels in theoccipital cortex.

One last functional neuroimaging study used MRS withphosphorous (31P-MRS) to investigate differences in brainenergetic compounds such as phosphocreatine among PIpatients and controls [51]. Harper et al. found that 16 PIpatients showed reductions in gray matter phosphocreatine,compared to 16 good sleepers. This might indicate thatinsomniacs experience a greater cortical energy demand thannormal sleepers, which is consistent with a continual state ofhyperarousal in insomnia.

4. Conclusions

In spite of a recent increase in neuroimaging research into PI, wehave yet to glean a consistent story about its neuropathology,especially with regard to structural studies of brain alterations.Functional studies are too few and diverse in methodology to yieldany general conclusions, whereas results of structural studies are

either contradictory or require replication. Tables 1 (functional)and 2 (structural) provide summaries of the main results of eachstudy reviewed.

The data tend toward an agreement with the hyperarousalconcept of insomnia, whereby PI is the result of a chronic state ofcentral nervous system (CNS) arousal that prevents and disturbssleep. Reduced deactivation in the transition from wakefulness tosleep [17], heightened connectivity of the emotional and threatresponse systems [31], diminished modulation of irrelevantcognitive processes [26], depletion of inhibitory neurotransmis-sion [44,49,50] and increased brain energetic demands [51] all areconsistent with an inability to appropriately decrease CNS arousalduring sleep onset and maintenance.

Structurally, PI reduction in hippocampal volume was found inone study [37] and was not successfully replicated [38,40–42].Lower gray matter concentrations were observed in the prefrontalcortex, specifically in the OFC [42,43] and in the dlPFC [43], butagain, they were not replicated by a later study [41]. Rostral ACCvolumes were increased in two independent PI samples [39], andinterestingly, GABA levels in the ACC were significantly decreasedin one of these samples [44]. Globally, GABA levels in the PI brainseem significantly diminished [49]; however, findings concerningGABA levels in the occipital cortex conflict [44,50]. Lastly, lowerphosphocreatine concentrations in the gray matter of individualswith PI seem to indicate altered cortical energetic demands [51].Overall, these promising structural findings are in need ofreplication.

Variability in sampling and methodology across studies mayexplain the variability in findings (Tables 1 and 2). Samplecharacteristics such as mean age and sex-ratio varied extensivelyacross studies. Differences in diagnostic assessment of PI andpatient history of pharmacological treatment may also havecontributed to cross-study variability. Furthermore, differences inPI severity and duration may also account for the discrepanciesobserved in the literature. Individuals who experienced insomnia-related sleep deprivation over many years may differ from thosewith more recent insomnia. The diagnostic criteria for PIencompass a heterogeneous group of individuals. The DSM-IVinsomnia subtypes have not received strong empirical support [52]and have been removed from the DSM-V. Nonetheless, there isconsiderable variability in the extent to which the subjective sleepcomplaints are associated with PSG-derived objective sleepdisturbances. Indeed, some studies found correlations betweenbrain alterations and PSG-defined sleep disturbances[38,39,43,50], while others found correlations with subjectiveparameters of severity [39,43]. The difficulty in identifyingempirically-supported subtypes of insomnia may also explainthe variable results obtained with neuroimaging. In structuralstudies, the norms for anatomically delineating brain regions ofinterest and the subsequent methods of morphometric analysisalso varied [37,38,40]. Future studies should pay careful attentionto previous methodologies when attempting to replicate findingsand ensure comparable sample characteristics and methods ofanalysis. Greater sample sizes are also needed in order to improvestatistical power.

Inconsistencies among the neuroimaging findings to datemay seem daunting to further research. To the contrary, suchdisagreement in the data should spur a still more structured andsystematic program of neuroimaging research on insomnia,using balanced samples, consistent diagnostic criteria andmethodologies, adequate subjective and objective sleep mea-sures, larger sample sizes and importantly, replication ofexisting findings. In addition, new imaging modalities, such asconnectivity analyses and cortical thickness measurements, mayshed light on the question of brain modifications in primaryinsomnia.

J.N. O’Byrne et al. / Pathologie Biologie 62 (2014) 262–269266

Table 1Functional neuroimaging studies of insomnia.

Study Neuroimagingtechnique

Sample size(number offemales)

Mean agein years (s.d.)

PI diagnosis andassessment

PI duration History of pharmacologicaltreatment

Main findings in PI compared to GS(significance level)

PI GS PI GS

Smith et al., 2002 [15] 99mTC-HMPAOSPECT

5 (5) 4 (4) 37.8 (12.1) 34.5 (11.9) ICSD-2, PSG >6 mo Off sleep aids for>4 weeks,off SSRIs for>1 yr

Hypoperfusion of basal ganglia andother regions during NREM sleep(P#0.05, uncorr.)

Smith et al., 2005 [16] 99mTC-HMPAOSPECT

4 (4) None 34.5 (12) None ICSD-2, PSG >6 mo Off sleep aids for>4 weeks,off SSRIs for>1 yr

Partial reestablishment of activation inbasal ganglia after BT (P#0.05, uncorr.)

Nofzinger et al., 2004 [17] 18F-FDG PET 7 (4) 20 (13) 34.2 (8.9) 32.6 (8.4) DSM-IV, PSG $1 mo PI using med. were excluded Smaller reduction in glucosemetabolism during transition to sleep;prefrontal hypoactivation during wake(P#0.05, corr.)

Nofzinger et al., 2006 [21] 18F-FDG PET 15 (7) None 36.9 (10.5) None DSM-IV, PSG $1 mo PI using med. were excluded Correlation between WASO andthalamocortical activation, includingpontine tegmentum (P<0.05, corr.)

Altena et al., 2008 [22] fMRI (1.5 T) 21 (17) 12 (9) 61 (6.2) 60 (8.2) RDC, PSG $2.5 yrs Off med. for$2 mo Prefrontal hypoactivation during verbalfluency task, partially restored afterCBT (P<0.05, uncorr.)

Huang et al., 2012 [31] fMRI (3.0 T) 10 (5) 10 (5) 37.5 (12.4) 35.5 (8.7) DSM-IV, PSG n.r. Medication-naive Altered connectivity in amygdalarpathways, particularly to the premotorcortex; amygdala-premotorconnectivity was correlated to PSQI(P<0.05, corr.)

Drummond et al., 2013 [26] fMRI (3.0 T) 25 (12) 25 (12) 32.3 (7.2) 32.4 (7.1) DSISD,actigraphy,PSG

$3 mo PI using med. were excluded During cognitive task, reducedactivation in task-relevant areas andreduced deactivation of default moderegions (P<0.05, uncorr.)

corr: corrected for multiple comparisons; uncorr.: uncorrected for multiple comparisons; n.r.: not reported; s.d.: standard deviation; PI: primary insomnia; GS: good sleeper controls; NREM: non-rapid-eye-movement; BT: behaviortherapy; CBT: cognitive behavioral therapy; PSQI: Pittsburgh Sleep Quality Index; PSG: polysomnography; WASO: wake after sleep onset; ACC: anterior cingulate cortex; med.: medication; SSRI: selective serotonin reuptakeinhibitor; CNS: central nervous system; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, edition IV; ICSD-2: International Classification of Sleep Disorders-2; RDC: Research Diagnostic Criteria for insomnia; DSISD:Duke Structured Interview for Sleep Disorders; mo.: months; yrs: years.

J.N.

O’B

yrne

et

al.

/

Pa

tho

logie

Bio

logie

62

(20

14

)

26

2–

26

9

26

7

Table 2Structural neuroimaging studies of insomnia.

Study Neuroimagingtechnique

Sample size(number offemales)

Mean age in years(s.d.)

PI diagnosis andassessment

Mean PIdurationin yrs (s.d.)

History ofpharmacologicaltreatment

Main findings in PI comparedto GS (significance level)

PI GS PI GS

Riemann et al., 2007 [37] MRI (1.5 T) 8 (5) 8 (5) 48.4 (16.3) 46.3 (14.3) DSM-IV, PSG 11.6 (8.9) Off med. for$2 weeks Reduction in HCV (P#0.05, uncorr.)Altena et al., 2010 [47] VBM-MRI (1.5 T) 24 (17) 13 (9) 60.3 (6.0) 60.2 (8.4) DSM-IV 17.7 (15.8) Off hypnotic med.

for$2 moReduction in GMCs in OFC, (P<0.05 corr.)

Winkelmanet al., 2010 [38]

MRI (3.0 T) 20 (10) 15 (6) 39.3 (8.7) 38.8 (5.3) DSM-IV,actigraphy, PSG

>6 mo No regular (>1/week)treatment with CNS-active med. for$3 mo

No diff. in HCV. SE and WASO negativelycorrelated with HCV (P#0.05)

Noh et al., 2012 [41] MRI (1.5 T) 20 (18) 20 (18) 50.8 (10.8) 50.4 (11.7) ICSD, PSG 7.6 (6.1) No hypnotic med.for$1 mo

No diff. in HCV. HCV negatively correlatedwith higher arousal index (P<0.05) andlonger insomnia duration (P<0.001)

Joo et al., 2013 [48] VBM-MRI (1.5 T)with DARTEL

27 (25) 27 (23) 52.3 (7.8) 51.7 (5.4) ICSD-2, PSG 7.6 (6.1) Medication-naive Reduction in GMCs in dlPFC and OFC.Negative correlations between: left middlefrontal gyrus GMCs and ISI; rightpostcentral gyrus GMCs and SOL; rightprecentral gyrus GMCs and WASO(P#0.001 uncorr.)

Spiegelhalderet al., 2013 [43]

VBM-MRI (3.0 T)with DARTEL

28 (18) 38 (21) 43.7 (14.2) 39.6 (8.9) DSM-IV, PSG 12.1 (11.0) Off psychoactive med.for$2 weeks

No diff. in HCV, no diff. in GMCs and WMCs.No correlation between ISI and HCV, norbetween HCV and total sleep time(P<0.05, corr. and P<0.001, uncorr.)

Winkelman et al., 2013(Study 1) [39]

MRI (3.0 T) 20 (10) 15 (6) 39.3 (8.7) 38.8 (5.3) DSM-IV, PSG $6 mo Off CNS-active med.for$2 weeks

Increased rACC volume. rACC volumecorrelated positively with SOL and WASO,negatively with SE (P#0.05, uncorr.)

Winkelman et al., 2013(Study 2) [39]

MRI (3.0 T) 21 (14) 20 (12) 35.8 (9.5) 34.1 (9.9) DSM-IV, PSG $6 mo Off CNS-active med.for$2 weeks

Increased rACC volume. Right ACC volumecorrelated with SOL (p#0.05, uncorr.)

Winkelmanet al., 2008 [49]

1H-MRS (4.0 T) 16 (8) 16 (7) 37.3 (8.1) 37.6 (4.5) DSM-IV,actigraphy, PSG

$6 mo No regular (>1/week)treatment with CNS-active med. for$3 mo

30% lower global GABA levels (P = 0.039)

Morgan et al., 2012 [50] 1H-MRS (4.0 T) 16 (10) 17 (9) 39 (9) 36 (9) DSM-IV, PSG $1 yr Off CNS-active med.for>3 mo

12% higher levels of GABA in the occipitalcortex (P<0.05)

Plante et al., 2012 [44] 1H-MRS (4.0 T) 20 (12) 20 (12) 34.3 (8.3) 34.1 (9.9) DSM-IV,actigraphy, PSG

$6 mo Off CNS-active med.for>2 weeks

33% lower global GABA levels in occipitalcortex, 21% lower in ACC (P<0.05)

Harper et al., 2013 [51] 31P-MRS (4.0 T) 16 (8) 16 (7) 37.2 (8.4) 37.6 (4.7) DSM-IV,actigraphy, PSG

>6 mo Off CNS-active med.for>3 mo

Lower phosphocreatine in gray matter,(P<0.05, corr.)

corr.: corrected for multiple comparisons; uncorr.: uncorrected for multiple comparisons; s.d.: standard deviation; diff.: difference; NA: not applicable; PI: primary insomnia; GS: good sleeper controls; HCV: hippocampal volume;GMC: gray matter concentrations; WMC: white matter concentration; rACC: rostral anterior cingulate cortex; PC: parietal cortex; dlPFC: dorsolateral prefrontal cortex; ISI: Insomnia Severity Scale; PSG: polysomnography; WASO:wake after sleep onset; SOL: sleep onset latency; SE: sleep efficiency; med.: medication; CNS: central nervous system; GABA: gamma-aminobutyric acid; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders, edition IV;ICSD-2: International Classification of Sleep Disorders-2; mo.: months; yrs: years.

J.N.

O’B

yrne

et

al.

/

Pa

tho

logie

Bio

logie

62

(20

14

)

26

2–

26

92

68

Disclosure of interest

The authors declare that they have no conflicts of interestconcerning this article.

Acknowledgements

Dr. Dang-Vu receives research support from the CanadianInstitutes of Health Research (CIHR), the Natural Sciences andEngineering Research Council of Canada (NSERC), the Fonds deRecherche du Quebec–Sante (FRQS), the Sleep Research SocietyFoundation (SRSF), and the Petro-Canada Young InnovatorsAwards Program. Dr. Gouin receives research support from theCanada Research Chair program, the Canada Foundation forInnovation (CFI), the CIHR, and the Social Sciences and HumanitiesResearch Council (SSHRC).

References

[1] Ohayon MM. Epidemiology of insomnia: what we know and what we still needto learn. Sleep Med Rev 2002;6(2):97–111.

[2] Fernandez-Mendoza J, Vgontzas AN, Bixler EO, Singareddy R, Shaffer ML,Calhoun SL, et al. Clinical and polysomnographic predictors of the naturalhistory of poor sleep in the general population. Sleep 2012;35(5):689–97.

[3] Roth T, Coulouvrat C, Hajak G, Lakoma MD, Sampson NA, Shahly V, et al.Prevalence and perceived health associated with insomnia based on DSM-IV-TR; International Statistical Classification of Diseases and Related HealthProblems, Tenth Revision; and Research Diagnostic Criteria/InternationalClassification of Sleep Disorders, Second Edition criteria: results from theAmerica Insomnia Survey. Biol Psychiatry 2011;69(6):592–600.

[4] Morin CM, LeBlanc M, Daley M, Gregoire JP, Merette C. Epidemiology ofinsomnia: prevalence, self-help treatments, consultations, and determinantsof help-seeking behaviors. Sleep Med 2006;7(2):123–30.

[5] Sivertsen B, Krokstad S, Overland S, Mykletun A. The epidemiology of insom-nia: associations with physical and mental health. The HUNT-2 study. JPsychosom Res 2009;67(2):109–16.

[6] Aikens JE, Rouse ME. Help-seeking for insomnia among adult patients inprimary care. J Am Board Fam Pract 2005;18(4):257–61.

[7] Hillman DR, Murphy AS, Pezzullo L. The economic cost of sleep disorders. Sleep2006;29(3):299–305.

[8] Stoller MK. Economic effects of insomnia. Clin Ther 1994;16(5):873–97 [dis-cussion 54].

[9] Zammit GK, Weiner J, Damato N, Sillup GP, McMillan CA. Quality of life inpeople with insomnia. Sleep 1999;22(Suppl. 2):S379–85.

[10] Van Dongen HP, Maislin G, Mullington JM, Dinges DF. The cumulative cost ofadditional wakefulness: dose-response effects on neurobehavioral functionsand sleep physiology from chronic sleep restriction and total sleep depriva-tion. Sleep 2003;26(2):117–26.

[11] Durmer JS, Dinges DF. Neurocognitive consequences of sleep deprivation.Semin Neurol 2005;25(1):117–29.

[12] Morin CM, Belanger L, LeBlanc M, Ivers H, Savard J, Espie CA, et al. The naturalhistory of insomnia: a population-based 3-year longitudinal study. Arch InternMed 2009;169(5):447–53.

[13] Desseilles M, Dang-Vu T, Schabus M, Sterpenich V, Maquet P, Schwartz S.Neuroimaging insights into the pathophysiology of sleep disorders. Sleep2008;31(6):777–94.

[14] American Academy of Sleep Medicine. International classification of sleepdisorders, 2nd ed.: Diagnostic and coding manual, Westchester, Illinois:American Academy of Sleep Medicine; 2005.

[15] Smith MT, Perlis ML, Chengazi VU, Pennington J, Soeffing J, Ryan JM, et al.Neuroimaging of NREM sleep in primary insomnia: a Tc-99-HMPAO singlephoton emission computed tomography study. Sleep 2002;25(3):325–35.

[16] Smith MT, Perlis ML, Chengazi VU, Soeffing J, McCann U. NREM sleep cerebralblood flow before and after behavior therapy for chronic primary insomnia:preliminary single photon emission computed tomography (SPECT) data.Sleep Med 2005;6(1):93–4.

[17] Nofzinger EA, Buysse DJ, Germain A, Price JC, Miewald JM, Kupfer DJ. Func-tional neuroimaging evidence for hyperarousal in insomnia. Am J Psychiatry2004;161(11):2126–8.

[18] Nofzinger EA, Nissen C, Germain A, Moul D, Hall M, Price JC, et al. Regionalcerebral metabolic correlates of WASO during NREM sleep in insomnia. J ClinSleep Med 2006;2(3):316–22.

[19] Altena E, Van Der Werf YD, Sanz-Arigita EJ, Voorn TA, Rombouts SA, Kuijer JP,et al. Prefrontal hypoactivation and recovery in insomnia. Sleep 2008;31(9):1271–6.

[20] Bonnet MH, Arand DL. 24-Hour metabolic rate in insomniacs and matchednormal sleepers. Sleep 1995;18(7):581–8.

[21] Perlis ML, Smith MT, Andrews PJ, Orff H, Giles DE. Beta/Gamma EEG activity inpatients with primary and secondary insomnia and good sleeper controls.Sleep 2001;24(1):110–7.

[22] Espie CA. Insomnia: conceptual issues in the development, persistence, andtreatment of sleep disorder in adults. Annu Rev Psychol 2002;53:215–43.

[23] Fulda S, Schulz H. Cognitive dysfunction in sleep disorders. Sleep Med Rev2001;5(6):423–45.

[24] Vincent NK, Walker JR. Perfectionism and chronic insomnia. J Psychosom Res2000;49(5):349–54.

[25] Drummond SP, Smith MT, Orff HJ, Chengazi V, Perlis ML. Functional imaging ofthe sleeping brain: review of findings and implications for the study ofinsomnia. Sleep Med Rev 2004;8(3):227–42.

[26] Drummond SP, Walker M, Almklov E, Campos M, Anderson DE, Straus LD.Neural correlates of working memory performance in primary insomnia. Sleep2013;36(9):1307–16.

[27] Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. Adefault mode of brain function. Proc Natl Acad Sci U S A 2001;98(2):676–82.

[28] Baglioni C, Spiegelhalder K, Lombardo C, Riemann D. Sleep and emotions: afocus on insomnia. Sleep Med Rev 2010;14(4):227–38.

[29] Riemann D, Spiegelhalder K, Feige B, Voderholzer U, Berger M, Perlis M, et al.The hyperarousal model of insomnia: a review of the concept and its evidence.Sleep Med Rev 2010;14(1):19–31.

[30] Tsuno N, Besset A, Ritchie K. Sleep and depression. J Clin Psychiatry2005;66(10):1254–69.

[31] Huang Z, Liang P, Jia X, Zhan S, Li N, Ding Y, et al. Abnormal amygdalaconnectivity in patients with primary insomnia: evidence from resting statefMRI. Eur J Radiol 2012;81(6):1288–95.

[32] Avendano C, Price JL, Amaral DG. Evidence for an amygdaloid projection topremotor cortex but not to motor cortex in the monkey. Brain Res1983;264(1):111–7.

[33] Hoshi E, Tanji J. Functional specialization in dorsal and ventral premotor areas.Prog Brain Res 2004;143:507–11.

[34] Backhaus J, Junghanns K, Born J, Hohaus K, Faasch F, Hohagen F. Impaireddeclarative memory consolidation during sleep in patients with primaryinsomnia: Influence of sleep architecture and nocturnal cortisol release. BiolPsychiatry 2006;60(12):1324–30.

[35] Guzman-Marin R, Suntsova N, Methippara M, Greiffenstein R, Szymusiak R,McGinty D. Sleep deprivation suppresses neurogenesis in the adult hippo-campus of rats. Eur J Neurosci 2005;22(8):2111–6.

[36] Hairston IS, Little MT, Scanlon MD, Barakat MT, Palmer TD, Sapolsky RM, et al.Sleep restriction suppresses neurogenesis induced by hippocampus-depen-dent learning. J Neurophysiol 2005;94(6):4224–33.

[37] Riemann D, Voderholzer U, Spiegelhalder K, Hornyak M, Buysse DJ, Nissen C,et al. Chronic insomnia and MRI-measured hippocampal volumes: a pilotstudy. Sleep 2007;30(8):955–8.

[38] Winkelman JW, Benson KL, Buxton OM, Lyoo IK, Yoon S, O’Connor S, et al. Lackof hippocampal volume differences in primary insomnia and good sleepercontrols: an MRI volumetric study at 3 Tesla. Sleep Med 2010;11(6):576–82.

[39] Winkelman JW, Plante DT, Schoerning L, Benson K, Buxton OM, O’Connor SP,et al. Increased rostral anterior cingulate cortex volume in chronic primaryinsomnia. Sleep 2013;36(7):991–8.

[40] Noh HJ, Joo EY, Kim ST, Yoon SM, Koo DL, Kim D, et al. The relationship betweenhippocampal volume and cognition in patients with chronic primary insom-nia. J Clin Neurol 2012;8(2):130–8.

[41] Spiegelhalder K, Regen W, Baglioni C, Kloppel S, Abdulkadir A, Hennig J, et al.Insomnia does not appear to be associated with substantial structural brainchanges. Sleep 2013;36(5):731–7.

[42] Altena E, Vrenken H, Van Der Werf YD, van den Heuvel OA, Van Someren EJ.Reduced orbitofrontal and parietal gray matter in chronic insomnia: a voxel-based morphometric study. Biol Psychiatry 2010;67(2):182–5.

[43] Joo EY, Noh HJ, Kim JS, Koo DL, Kim D, Hwang KJ, et al. Brain gray matter deficitsin patients with chronic primary insomnia. Sleep 2013;36(7):999–1007.

[44] Plante DT, Jensen JE, Schoerning L, Winkelman JW. Reduced gamma-amino-butyric acid in occipital and anterior cingulate cortices in primary insomnia: alink to major depressive disorder? Neuropsychopharmacology2012;37(6):1548–57.

[45] Shekleton JA, Rogers NL, Rajaratnam SM. Searching for the daytime impair-ments of primary insomnia. Sleep Med Rev 2010;14(1):47–60.

[46] Plante DT, Jensen JE, Winkelman JW. The role of GABA in primary insomnia.Sleep 2012;35(6):741–2.

[47] Bush G, Luu P, Posner MI. Cognitive and emotional influences in anteriorcingulate cortex. Trends Cogn Sci 2000;4(6):215–22.

[48] Koolschijn PC, van Haren NE, Lensvelt-Mulders GJ, Hulshoff Pol HE, Kahn RS.Brain volume abnormalities in major depressive disorder: a meta-analysisof magnetic resonance imaging studies. Hum Brain Mapp 2009;30(11):3719–35.

[49] Winkelman JW, Buxton OM, Jensen JE, Benson KL, O’Connor SP, Wang W, et al.Reduced brain GABA in primary insomnia: preliminary data from 4 T protonmagnetic resonance spectroscopy (1H-MRS). Sleep 2008;31(11):1499–506.

[50] Morgan PT, Pace-Schott EF, Mason GF, Forselius E, Fasula M, Valentine GW,et al. Cortical GABA levels in primary insomnia. Sleep 2012;35(6):807–14.

[51] Harper DG, Plante DT, Jensen JE, Ravichandran C, Buxton OM, Benson KL, et al.Energetic and cell membrane metabolic products in patients with primaryinsomnia: a 31-phosphorus magnetic resonance spectroscopy study at 4 Tesla.Sleep 2013;36(4):493–500.

[52] Edinger JD, Wyatt JK, Stepanski EJ, Olsen MK, Stechuchak KM, Carney CE, et al.Testing the reliability and validity of DSM-IV-TR and ICSD-2 insomnia diag-noses. Results of a multitrait-multimethod analysis. Arch Gen Psychiatry2011;68(10):992–1002.

J.N. O’Byrne et al. / Pathologie Biologie 62 (2014) 262–269 269