This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
2
3Q2
4
56
7
8910
11121314151617
34
3536
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
NeuroImage xxx (2014) xxx–xxx
YNIMG-11586; No. of pages: 8; 4C: 4, 5, 6
Contents lists available at ScienceDirect
NeuroImage
j ourna l homepage: www.e lsev ie r .com/ locate /yn img
Reduced mediodorsal thalamic volume and prefrontal cortical spindleactivity in schizophrenia
OFAndreas Buchmann a,1, Daniela Dentico a,1, Michael J. Peterson a, Brady A. Riedner a, Simone Sarasso a,b,
Marcello Massimini a,b, Giulio Tononi a, Fabio Ferrarelli a,b,⁎a Department of Psychiatry, University of WI–Madison, USAb Department of Clinical Sciences, University of Milan, Italy
⁎ Corresponding author at: University of Wisconsin–MaE-mail address: [email protected] (F. Ferrarelli)
1 These authors contributed equally to the manuscript.
Background:We recently found marked deficits in sleep spindles, non-rapid eye movement (NREM) sleep oscil-lations that are generated within the thalamus and then amplified and sustained in the cortex, in patients withschizophrenia compared to both healthy and psychiatric controls. Here,we investigated the thalamic and corticalcontributions to these sleep spindle deficits.Methods: Anatomical volume of interest analysis (i.e., thalamic volumes) and electroencephalogram (EEG)source modeling (i.e., spindle-related cortical currents) were performed in patients with schizophrenia andhealthy comparison subjects.Findings: Schizophrenia patients had reduced mediodorsal (MD) thalamic volumes, especially on the left side,compared to healthy controls, whereas whole thalami and lateral geniculate nuclei did not differ betweengroups. Furthermore, left MD volumes were strongly correlated with the number of scalp-recorded spindles in
28
29
30
31
32
33
CTan anterior frontal region, and cortical currents underlying these anterior frontal spindles were localized in the
prefrontal cortex, in Brodmann area (BA) 10. Finally, prefrontal currents at the peak of spindle activity were sig-nificantly reduced in schizophrenia patients and correlated with their performance in an abstraction/workingmemory task.Conclusion: Altogether, these findings point to deficits in a specific thalamo-cortical circuitry in schizophrenia,which is associated with some cognitive deficits commonly reported in those patients.
Sleep spindles are waxing/waning, fast (11–16 Hz) EEG oscillationswhich occur during NREM sleep. While the functional role of spindlesis still not fully established, they are thought to be implicated inmemoryconsolidation and plasticity (Molle and Born, 2011), and are considereda proxymeasure for the individual's learning potential (Fogel and Smith,2011). This assumption is supported by animal studies, showing that thetiming and number of spindle sequences regulate plasticity-relatedchanges in cortical pyramidal neurons (Rosanova and Ulrich, 2005). Itis also consistent with findings in healthy humans, which have shownthat higher spindle activity is associatedwith better performance in ver-bal memory (Schabus et al., 2008), visuo-spatial memory (Clemenset al., 2006), as well as declarative learning tasks (Gais et al., 2002).
In two recent studies we found amarked reduction in several wholenight sleep spindle parameters, including duration, amplitude, andnumber in patients with schizophrenia compared to both healthy and
psychiatric controls (Ferrarelli et al., 2007, 2010). These sleep spindleimpairments were established in baseline conditions, given that pa-tients did not perform any cognitive task before sleep, therefore sug-gesting an intrinsic fundamental deficit of the underlying neuronalcircuitry. Deficits in cognitive tasks involving working memory andother learning paradigms have been consistently found in patientswith schizophrenia (Kalkstein et al., 2010). Furthermore, accumulatingevidence has revealed sleep spindle deficits associated to cognitive per-formance in schizophrenia patients. Two studies employing a finger-tapping motor sequence task (MST) before and after a night of sleepfound that patients with schizophrenia had reduced spindle density(Manoach et al., 2010) and coherence (Manoach et al., 2010; Wamsleyet al., 2012) compared to healthy controls, and in one of these studieslower spindle incidence predicted less overnight improvement in theMST (Wamsley et al., 2012). A reduction in spindle density was alsoestablished in schizophrenia patients compared to healthy subjects aswell as mood disordered patients during a 40 minute nap, and schizo-phrenia patients showed no after nap improvement in a procedurallearning task (Seeck-Hirschner et al., 2010).
Sleep spindles are generated within the thalamus, and are thentransferred to the cortex, where these oscillations are amplified and
mic volume and prefrontal cortical spindle activity in schizophrenia,
2 A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
sustained. To further characterize the thalamic and cortical contributionto sleep spindle deficits in schizophrenia, in this studywe combined an-atomical (i.e., thalamic volumes) and functional (i.e., cortical sourcesunderlying EEG sleep spindles)measures.We also investigatedwhethersome of those biological measures could predict performance in an ab-straction/working memory task in schizophrenia patients.
T
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
ORREC
Methods
Participants
Twenty-onepatientswith schizophrenia (mean±S.D.=36± 10.2;female = 8) and twenty-one age-matched healthy controls (mean ±S.D. = 36.2 ± 8.5; female = 4) participated in this study (Table 1). Inthese subjects, we performed whole night high-density (hd)-EEG re-cordings. Furthermore, in all schizophrenics as well as in a subset(N = 11) of healthy controls anatomical magnetic resonance imag-ing (MRI) scans were collected to perform region of interest analyses(ROIs, see below). This subset of healthy controls was also age-matched with the schizophrenia group (mean ± S.D. was 36.8 ±8.4 in controls, 36.0 ± 10.2 in patients with schizophrenia). The re-maining control subjects had completed their MRI scans prior to anupgrade of the scanner, and had used different acquisition parame-ters (i.e., number and thickness of individual slices) for the anatom-ical scans, and could not be used in comparison to the other subjects.Exclusion criteria for enrollment in the study were substance abuse/dependence within the last six months, neurological disorders, andsleep disorders. Among neurological conditions, individuals with ahistory of recurrent seizures or epilepsy, medical conditions thatcould increase the chance of having a seizure (e.g. — stroke, aneu-rysm, brain surgery, brain tumor or other structural lesion), as wellas a history of head trauma that caused neurological injury wereexcluded. Patients with current or past diagnoses of sleep disorders,including sleep apnea and restless leg syndrome, were also excluded.An additional exclusion criterion for healthy subjects was thepresence of first — degree relatives with psychiatric diagnoses.
A psychiatrist (MJP) interviewed all participants and administeredthe structured clinical interview for the diagnostic and statistical manu-al ofmental disorders (DSM-IV) to assess psychiatric diagnoses. Patientswith schizophrenia were diagnosed as paranoid (N= 11), undifferenti-ated (N=5), disorganized (N=1), or residual subtype (N=4). All butone was receiving second-generation antipsychotics. Patients withschizophrenia were outpatients with a mean duration of illness of13 years (SD = 7). After a complete description of the study, each par-ticipant gave written informed consent. The study was approved by theUniversity of Wisconsin Institutional Review Board.
UNC
Table 1Demographics and clinical ratings of subject groups.
PANSS: Positive and Negative Syndrome Scale.a Medication doses are expressed as Chlorpromazine equivalent.
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
ED P
RO
OF
Volume of interest analysis
Anatomical T1-weighted magnetic resonance (MR) images were ac-quired in a 3 T GE scanner. T1-weighted images used a high resolutionthree-dimensional gradient echo sequencewith 512 × 512 × 248 voxelsin x–y–z direction, resulting in a resolution of 0.5 × 0.5 × 0.8mm. TRwas7.448 ms, TE 1.496 ms and flip angle 10°. Additionally, we collected T2-weighted imageswith 39 sliceswith 256 × 256 voxels in x–y–z direction,resulting in a resolution of 0.94 × 0.94 mm, slice thickness 3 mm, TR was8000 ms, TE 1.68 ms and flip angle 90°. All participants included in thestudy had their MRI performed in the same scanner, using the same ac-quisition parameters. Scans were collected within a week of the wholenight sleep EEG recordings. MR images were bias-corrected with SPM8,and volumes of interest analysis (VOIs) were measured manually usingthe software MRIcroN. We selected the whole thalamus (WT),mediodorsal (MD) nucleus, and lateral geniculate nucleus (LGN) bilater-ally. The MD was chosen because it is the most recognizable nucleus ofthe dorsal thalamus, which is heavily interconnected with thalamic retic-ular nucleus (TRN), the spindle pacemaker. Anatomical and electrophys-iological evidence of this privileged interaction between dorsal thalamusand TRN is presented in two comprehensive review articles (Fuentealbaand Steriade, 2005; Pinault, 2004).Moreover, theMD is an anterior nucle-us of the thalamus, and in recent work we found that schizophrenia pa-tients showed the strongest reduction in sleep spindles in anterior(frontal–prefrontal), rather than posterior (parieto-occipital) regions(Ferrarelli et al., 2007, 2010). The LGNwas chosen as a control volume be-cause it is a posterior nucleus, and it is not highly inter-connected withthe TRN. VOIs were measured three times by two raters (A.B., F.F.).Intra-rater correlations ranged between 0.84 and 0.89, inter-rater correla-tions ranged from 0.70 to 0.76. The MD thalami were identified as thedarker medial structures starting anterior to the nucleus ruber and ex-tending posterior to the medial geniculate. The LGN was identified as aknee-shaped formation localized alongside the middle hippocampus.For all VOIs, the average values obtained across all measurements wereutilized for statistical analyses.
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
Scalp EEG recordings and analysis
Whole-night sleep recordings were performed in schizophrenia pa-tients and healthy controls with 256 electrode high-density (hd)-EEGsystems (EGI, Eugene, Oregon). There was no adaptation night specifi-cally for this study. However, all participants in this study had been en-rolled in previous sleep recordings and had spent at least a night in thelaboratory prior to those sleep hd-EEG recordings. Sleep stages werescored on C3A2 and C4A1 derivations based on the American Academyof Sleep Medicine criteria (AASM) (Silber et al., 2007). The AASM rec-ommends the use of F4, C4 and O2 for staging sleep; however the pres-ent study focused on sleep spindle activity, and therefore we utilizedthe C4 as themain derivation for sleep scoring. Furthermore, we scoredthe data with 20 second resolution, rather than 30 s, to minimize theamount of data rejected throughout the analysis. Artifacts were exclud-ed visually during sleep staging. Additional artifacts were removed au-tomatically, by rejecting epochs that exceeded thresholds based on themean power for each channel in the 0.75 to 4.5 Hz and 20 to 30 Hzbands. After removing noisy electrodes, whichwere excluded automat-ically based on high impedance value (N150 kΩ) aswell as by visual in-spection, EEG signals were re-referenced to the average of all retainedchannels, and sleep power spectra as well as spindle detection analysiswere performed. These analyses are described in detail elsewhere(Ferrarelli et al., 2007, 2010). Results from this dataset have been par-tially presented in a previous publication (Ferrarelli et al., 2010). Herewe focused on the correlation between thalamic VOIs and sleep spin-dles aswell as on sourcemodeling scalp EEG spindle activity. All spindleparameters (number, activity) were calculated per minute of NREMsleep.
mic volume and prefrontal cortical spindle activity in schizophrenia,
3A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
UNCO
RREC
Source modeling analysis
EEG data were band-pass filtered in the spindle frequency(11–16Hz) and Independent Component Analysis (ICA)was used to re-move ocular, muscle, and electrocardiographic artifacts. Specifically, ICAcomponents with activity patterns and topographic maps characteristicof artifactual activity were removed. ICAwas performed on NREM sleepdata, includingN2 andN3 epochs, duringwhich sleep spindles occurredin the frontal cluster showing the strongest correlation with MD tha-lamic volumes in both groups, as shown in Fig. 2. Spindleswere then de-tected on EEG scalp data with an in house automatic algorithm. Briefly,for each channel thresholds for detection relative to the mean ampli-tude of that channel were used, whereas for each spindle the amplitudewas the maximum above an upper threshold, while the beginning andend were the points preceding or following this maximum when theamplitude dropped below a lower threshold (Ferrarelli et al., 2010).For source modeling, the length of the analysis window matched theduration of each scalp detected spindle. Sleep spindles were localizedindividually, and solutions were then averaged across all detections.Those spindles occurred during sleep epochs that were artifact rejectedboth visually and automatically, and therefore there was no EEG base-line noise. Source localization was performed using a 4-shell headmodel derived from a template MRI. An MRI-coregistered set of elec-trodes was used to construct the forward model. The inverse matrixwas calculated using the minimum norm least-squares (L2) method,subject to depth weighting, Tikhonov regularization at 10−1, and LowResolution Electromagnetic Tomography (LORETA) constraint. Thevalue of each source was then standardized (sLORETA). The sourcespace was restricted to 2447 cortical voxels (7 mm3), each assigned toa gyrus based on the Montreal Neurological Institute probabilisticatlas. Source analyses were performed using GeoSource software (EGI,Eugene, Oregon).
Cognitive task
A subset of schizophrenia patients (N = 12) performed the PENNCNP, a computerized battery that includes several cognitive tests. How-ever, the analysis was focused on the Penn Abstraction, Inhibition andWorking Memory (AIM) task. The AIM, like theWisconsin Card SortingTest (WCST), is a measure of cognitive executive functions, which hasbeen found to be impaired in schizophrenia patients compared tohealthy comparisons (Glahn et al., 2000). More importantly, both ana-tomical and functional neuro-imaging studies have found that thosetypes of tasks implicate the prefrontal cortex (Berman et al., 1995;Szulc et al., 2012; Yuan and Raz, 2014), which is where we found thepeak of spindle activity in both groups. Here we investigated if AIM per-formance correlated with cortical spindle currents in schizophrenia pa-tients. We also assessed whether this spindle activity could predict theRaven's scores, a non-verbal measure of IQ, in schizophrenia patients.
Statistics
Differences in VOIs and sleep spindles between schizophrenia pa-tients and healthy controls were assessedwith unpaired t tests. Becausesix VOIs (WT, MD, and LGN bilaterally) were measured, threshold forsignificance was set at p (0.05/6) = 0.0083. Correlation analyses be-tween left MD thalamic volumes and the topography of sleep spindledensity were performed. Left MD nuclei were the most markedly andconsistently reduced VOIs in schizophrenia patients compared tohealthy controls. Furthermore, we chose spindle density, rather thanother spindle parameters, since sleep spindles are generated duringNREM sleep within the thalamus, and therefore spindle density, whichmeasures the number of spindles occurring per minute of NREM sleep,was the parameter most likely to reflect a thalamic defect in generatingsleep spindles.
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
ED P
RO
OF
We also performed unpaired t tests between healthy controls andpatients with schizophrenia for the strongest cortical currents. Specifi-cally, the dipoles corresponding to the top 1% of all cortical currentswere organized by Brodmann areas and the most intense currents atthe beginning, at the peak, and at the end of spindle activity. Becausethree measures were computed, threshold for significance was set at p(0.05/3) = 0.016. Finally, correlation analyses were performed be-tween medication doses and MD thalamic volumes, EEG frontal spin-dles, and spindle-related cortical currents.
Results
Schizophrenics had MD volume reduction compared to healthy controls
Schizophrenia patients showed a significant reduction in the volumeof both the left and right MD nuclei (Fig. 1). The left MD volume de-crease was significant even after Bonferroni correction for multiplecomparisons (Table 2). By contrast, the volume of both left and rightLGNwas not different between the two groups (Fig. 1, Table 2). Further-more, no difference in either left or rightwhole thalamus (WT) volumeswas found between groups (Table 2, Supp. Fig. 1). To assess whethergroup differences were affected by variability in VOI measurements,we calculated the margin of error (MOE) between raters as the meandifference in VOI measures and found that the largest inter-ratersMOE was 0.05 for the left WT and 0.07 for the right WT, whereasgroup differences for the same VOIs were 0.15 and 0.28 respectively.
MD volume reduction correlated with frontal EEG spindle deficitsin schizophrenia
Correlation analysis between left MD volumes and the EEG topogra-phy of spindle density in the pooled sample (schizophrenia patients andhealthy controls) established that spindles in posterior, parietal–occip-ital regions showed the lowest correlation values (Fig. 2, right). In con-trast, frontal–prefrontal areas had higher correlation coefficients, and aleft anterior frontal cluster of electrodes (N=23) showed the strongestcorrelations with the left MD volumes (r ≥ 0.5, p ≤ 0.04). To rule outthat the correlation between MD thalamic volume and sleep spindledensity was exclusively driven by the group differences on these pa-rameters, we performed a linear regression analysis in which spindledensity was the dependent variable whereas MD volumewas the inde-pendent, predictive variable with group as a categorical independentcovariate. We found that MD thalamic volumes could still significantlypredict variability in spindle number, even after accounting for thegroup effect (MD t= 2.381, p= 0.024). Furthermore,when comparingthe incidence of sleep EEG spindles in this left frontal cluster betweenthe two groups it was found that schizophrenics had a marked reduc-tion in spindle density (0.3 ± 0.25) compared to healthy controls(0.9 ± 0.3, p b 0.001, Fig. 2 left).
Sourcemodeling of spindle activity occurring in this left frontal clusterestablished that most of the cortical currents were localized in frontal/prefrontal areas. Moreover, in both patients with schizophrenia andhealthy subjects the strongest spindle cortical sources were localizedin the prefrontal cortex, in a region overlapping the boundaries be-tween BA 10 and BA 11, with the center of mass in BA 10 (Fig. 3 left).BA 10 showed the largest currents at the beginning, at the peak, andat the end of spindle activity in both groups. Furthermore, while bothgroups had similar current intensities at the beginning and end ofsleep spindles, cortical currents at the peak of spindle activity were re-duced in schizophrenia patients (0.4 ± 0.02) compared to healthy con-trols (0.53 ± 0.03, p = 0.01, Fig. 3 right). Finally, those spindle-relatedprefrontal cortical currents were significantly correlated with perfor-mance in the AIM task (Fig. 4); by contrast, spindle prefrontal currents
mic volume and prefrontal cortical spindle activity in schizophrenia,
Fig. 1. ROI's analysis of themediodorsal (MD) and lateral geniculate nucleus (LGN) in healthy subjects (blue) and patients with schizophrenia (red). Schizophrenia patients showed a vol-ume reduction in both left and right MD, but not in the LGN, compared to control subjects. Left MD reduction was significant (p = 0.001) after correction for multiple comparison (pthreshold = 0.0083, or 0.05/6).
t2:1Q1
t2:2
t2:3
t2:4
t2:5
t2:6
t2:7
t2:8
t2:9
t2:10
t2:11
4 A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
NCO
RREC
could not predict theRaven's scores in schizophrenia patients (r= 0.39;p = 0.2). Furthermore, we found no correlation between antipsychoticmedication doses, expressed as Chlorpromazine equivalents, and corti-cal currents, EEG spindle number, as well as MD thalamic volumes(r≥ −0.23, p ≥ 0.3).
Discussion
Variations in thalamic graymatter volumes have been previously in-vestigated in schizophrenia. Some studies found WT reductions inschizophrenia patients (Ettinger et al., 2007; van Haren et al., 2007),whereas other studies reported no WT differences between schizo-phrenics and healthy subjects (Yoshihara et al., 2008). In another recentwork LGN volumes of schizophrenics, health subjects and depressed pa-tients were comparable in size (Dorph-Petersen et al., 2009). We foundno difference in WT as well as LGN volumes between schizophrenicsand control subjects. By contrast, schizophrenia patients showed a vol-ume reduction inMD nuclei, more prominently on the left side. A stron-ger left-sided reduction in MD volumes in schizophrenia has beenreported in previous structural MRI (Byne et al., 2001) aswell as neuro-pathological, post-mortem (Danos et al., 2003) studies. Furthermore, inthe few studies measuring both WT and MD volumes it was found thatWT was reduced only in one study and on one side (Shimizu et al.,2008),whereas schizophrenics had consistently smaller leftMDnucleusvolumes compared to healthy controls (Kemether et al., 2003; Shimizu
U 357
358
359
360
361
362
363
364
365
366
367
368
369
370
Table 2
Volumes of interest (VOIs) Mean volume (±SD) p values(unpaired t test)
Healthycontrols
Schizophrenics
Left whole thalamus (WT) 6.00 ± 0.91 5.85 ± 1.00 p = n.s.Right whole thalamus (WT) 5.84 ± 1.06 5.56 ± 0.77 p = n.s.Left mediodorsal (MD) 0.74 ± 0.05 0.64 ± 0.10 p = 0.001a
Right mediodorsal (MD) 0.72 ± 0.05 0.62 ± 0.15 p = 0.01Left lateral geniculatenucleus (LGN)
0.72 ± 0.12 0.66 ± 0.15 p = 0.2
Right lateral geniculatenucleus (LGN)
0.62 ± 0.09 0.57 ± 0.11 p = 0.2
a Significant after Bonferroni correction for multiple comparisons.
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
ED
et al., 2008). The MD nucleus is a major higher order (HO) thalamicrelay that receives driver inputs from cortical layer V neurons, whereasfirst order (FO) thalamic nuclei, such as the lateral geniculate nuclei, re-ceivemost of their driver inputs from subcortical sources (Sherman andGuillery, 2002). HO thalamic nuclei are thought to be implicated intransferring information from one cortical area to another, and also re-ceive a copy of the motor message that is sent from the cortex tolower, mostly motor centers in the central nervous system (Guilleryand Sherman, 2002). Thus, it has been proposed that HO thalamic nucleimay play a role in internal motor monitoring mechanisms, also knownas efference copy or corollary discharge mechanisms, and a recentstudy found that such mechanisms are impaired in schizophrenia(Spering et al., 2013). The reduction in volume of theMD nuclei report-ed here may be implicated in the efference copy failure found in schizo-phrenia patients.
Here left MD volumes were correlated with left frontal EEG spindlesin both healthy and schizophrenic subjects. Our previous findings ofspindle deficits in schizophrenia (Ferrarelli et al., 2007, 2010) point toa defect in the thalamus, and particularly in the TRN, the spindle pace-maker. Becausemeasuring the volume of the TRNwith current anatom-ical neuro-imaging techniques is a practical impossibility, we exploitedthe location of the TRN, which covers most of the dorsal thalamus, aswell as its connectivity with dorsal thalamic nuclei, of which the MD isthe most recognizable structure. Smaller MD volumes reflect reducedneurons and connections between those neurons and the TRN, whichare likely to affect the incidence of sleep spindles. Consistent with thisprediction, we found a strong correlation between MD volumes andspindle density in scalp-recorded spindles, in a cortical region knownto be anatomically connected to the MD nucleus.
This finding suggests that the MD nucleus contributes to the inci-dence of frontal spindles, and is consistent with electrophysiologicalstudies in animals demonstrating that spindles are initiated within thethalamus by the interplay of the TRNwith (mostly) dorsal thalamic nu-clei (Fuentealba and Steriade, 2005), as well as with electro-corticographic recordings in humans showing an implication of MDand TRN in spindle generation (Nakamura et al., 2003). Furthermore,electrophysiological recordings in awake and attentive primates havedemonstrated that MD nucleus and the pulvinar, both HO thalamic nu-clei, have much greater rebound burst firing activity compared to FO
mic volume and prefrontal cortical spindle activity in schizophrenia,
Correlation between Left MD volume and spindle density
0.5 0.7 0.9
0.4
0.8
1.2
Slee
pSp
indle
dens
ity
Left MD volume
Fig. 2. Left MD volume reduction in schizophrenia patients significantly correlated with sleep spindle density in a frontal cluster of electrodes. This cluster (black-encircled red spot) wassignificantly reduced in schizophrenics compared to healthy controls (blue and red box plots, p b 0.01). On the bottom left, single subject correlation between left MD volume and spindlenumber for schizophrenia patient (red) and healthy controls (blue).
5A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
C
thalamic nuclei, including the LGN (Ramcharan et al., 2005), and thishigher burst propensity could be related to greater expression ofvoltage-dependent transient (T-type) calcium channels (Wei et al.,2011). Intriguingly, a gene encoding a T-type calcium channel(CACNA1I, which encodes CaV3.3), has been recently implicated inschizophrenia by two large genetic studies (2012; 2014), whereas an-other study has shown that CaV3.3 calcium channel, which is highlyexpressed in the TRN, is the major sleep spindle pacemaker in the thal-amus (Astori et al., 2011).
We also established that EEG frontal spindles showing a significantcorrelation with MD volumes originated from the anterior prefrontal
UNCO
RRE
Spindle Start
Spindle Peak
Spindle End
Healthy Subjects Schizop
BA 10 BA 10
BA 10 BA 10
BA 10 BA 10
A
Fig. 3. Source modeling (sLORETA) of sleep spindles in healthy subjects (blue) and schizophrenfrontal cortex (BA 10), and these prefrontal currents were significantly weaker in schizophren
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
ED
cortex (BA 10). Anatomical and immuno-histological studies in pri-mates have demonstrated that MD thalamus receives topographicallyorganized fibers from the TRN, the spindle pacemaker, and MDthalamo-cortical neurons are heavily interconnected with prefrontalcortical neurons (Barbas et al., 2011). Moreover, recent f-MRI experi-ments during sleep spindles found a strong blood-oxygen level depen-dent (BOLD) activation in the MD thalamus, which correlated with theactivity in prefrontal cortex (Schabus et al., 2007).
In this study we found that spindle-associated cortical prefrontalcurrents were decreased in schizophrenic compared to healthy individ-uals, and this reduction predicted impaired performance in an
p=0.01
hrenics
C u
r r
e n
t
i n
t e
n s
i t
y
a t
B
A 1
0
0
0.2
0.3
0.4
0.5
0.1
*
*
B
0
0.2
0.3
0.4
0.5
0.1
0
0.2
0.3
0.4
0.5
0.1
ia patients (red). In both groups the strongest cortical currents were localized in the pre-ia compared to healthy controls at the spindle activity peak.
mic volume and prefrontal cortical spindle activity in schizophrenia,
Fig. 4. Cortical prefrontal currents (BA 10) correlated with performance (left) as well as reaction time (right) in an executive function task (AIM from PENN CNP).
6 A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
UNCO
RREC
abstraction/workingmemory task in schizophrenia patients. It has beenshown that higher spindle activity predicts better performances in ver-bal memory, visuo-spatialmemory, aswell as declarativememory tasks(Fogel and Smith, 2011). Furthermore, recent studies have reported thatschizophrenia patients had reduced sleep spindle density compared tohealthy controls, and this reduction predicted worse post-sleep perfor-mance in procedural learning tasks (Seeck-Hirschner et al., 2010;Wamsley et al., 2012). Our findings not only confirmprevious data indi-cating executive function deficits in those patients (Glahn et al., 2000),but also point to defects in a TRN–MDnucleus–prefrontal cortex circuit-ry in schizophrenia. Intriguingly, we established that prefrontal spindlecurrents correlated with performance in the AIM task, but not with theRaven's scores. While this finding does not rule out that other tests maybe significantly associated to spindle deficits in schizophrenia(i.e., verbal IQ tasks), it suggests that the reduction in prefrontal spindleactivity established here is specifically related to the AIM performance.
We found no associations between cognitive performance and MDthalamic volumes. However, MD volumes predicted a reduced spindleincidence in anterior frontal regions. We think that this is due to thefact that a defect within the thalamus, which includes the MD and theTRN, will result in a reduced incidence of sleep spindles in anterior fron-tal regions. At the same time, when spindles do occur, the overall levelof activation of the prefrontal cortex, which is critical in performingthe AIM task correctly, is reduced in patients with schizophrenia com-pared to healthy controls, and this reduction is associated with an im-paired performance in this cognitive task.
At this stage, we can only speculate about the molecular mecha-nisms underlying this defective thalamo-cortical circuitry. One mecha-nism may involve reduced binding or expression of thalamo-corticalN-methyl-D-aspartate (NMDA) glutamate receptors. Postmortem stud-ies found reduced NMDA glutamate receptors in bothMD thalamus andprefrontal cortex in schizophrenia patients (Pakkenberg et al., 2009).Pharmacological manipulations with NMDA antagonists, including ke-tamine and phencyclidine (PCP), produce schizophrenia-like psychosisin humans, and animal studies have shown that Asenapine and Cloza-pine, two second-generation antipsychotics, could revert a PCP-induced hypoactivity of NMDA receptors in both MD thalamus(Santana et al., 2011) and prefrontal cortex (Jardemark et al., 2010). In-jections of NMDA antagonists in the TRN rat brain trigger delta-rangerhythmic bursting, thus suggesting that NMDA hypofunction underlieTRN-generated delta band EEG oscillations, a waking thalamo-corticaldys-rhythmia established in schizophrenia (Zhang et al., 2009). More-over, 2-deoxyglucose imaging data in mice characterizing the acute ef-fects of ketamine on brain functional connectivity found ketamine-induced impairments in a circuitry involving TRN, MD thalamus, andprefrontal cortex (Dawson et al., 2011).
Another molecular mechanism implicates deficits in the gamma-aminobutyric acid (GABA) thalamo-cortical system. The TRN consistsof GABA-ergic neurons, and electrophysiological experiments have
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
ED P
RO
Oshown that TRN GABA release plays a critical role in regulating sensorygating (Krause et al., 2003), which has been found to be defective inschizophrenia (Light and Braff, 1999). The presence of GABA impair-ments in schizophrenia is also suggested by data from postmortemstudies, which found a reduction in glutamate decarboxylase 67, an en-zyme involved in GABA synthesis, and in GABA membrane transporterdensity in cortical interneurons in schizophrenia patients (Lewis et al.,2005). Additionally, treatment studies have shown that Clozapine, oneof the most effective antipsychotics, is associated with enhancedthalamo-cortical GABA activity in schizophrenia patients, and the bene-ficial effects of Electroconvulsive Therapy (ECT) and Transcranial Mag-netic Stimulation (TMS) are related to increased GABA-mediatedinhibitory neurotransmission on excitatory cortical neurons(Daskalakis et al., 2008).
Schizophrenia patients were medicated when data were collected.They were all taking second generation antipsychotics, with the excep-tion of one patientwhowas on Fluphenazine. Ten patients were on Clo-zapine, five on Aripiprazole, five on Risperidone, three on Quetiapine,oneonOlanzapine and one on Ziprasidone. Six patientswere on two an-tipsychotic medications. However, medications are unlikely to accountfor these findings. Neuro-imaging anatomical studies on the effects ofatypical antipsychotics on the thalamus have reported a bilateral in-crease in thalamic volumes in both first episode psychosis (Dazzanet al., 2005) and chronic patients with schizophrenia, whereas otherstudies found a slight increase in the right WT and no changes in theleft WT (Deng et al., 2009; Tomelleri et al., 2009). Here we reportedno difference in bilateral WT, whereas we found a marked decrease inthe leftMDvolumes of schizophrenia patients compared to healthy con-trols. Moreover, in a recent study we found that non-schizophrenia pa-tients taking antipsychotic medications had no spindle deficitscompared to healthy controls (Ferrarelli et al., 2010) and in this studywe found that medication doses did not correlate with any of the mea-sures defective in schizophrenia patients, including left MD volumes,EEG frontal spindles, and spindle-related prefrontal cortical currents.
By collecting both anatomical and functional measures in the samegroupof patients,we identified deficits in a specific thalamo-cortical cir-cuitry underlying sleep spindle activity in schizophrenia. Future studiesare needed to fully characterize the involvement of this circuitry in theneurobiology of this disorder. Replicating thesefindings in larger groupsof patients will help to clarify the contribution of different clusters ofsymptoms to these deficits. Here we found that patients with paranoidschizophrenia had smallerMD thalamic volumes compared to the otherschizophrenia patients, but this reduction was not significant. Further-more, simultaneous fMRI/hd-EEG recordings will allow assessing moredirectly the reduction in activation of the thalamus, and especiallyTRN/MD thalamic nuclei, and prefrontal cortex during sleep spindlesin schizophrenia. As a first step in this direction, in a recent fMRI/TMSstudy schizophrenics showed reduced thalamic and anterior frontal ac-tivation following direct perturbation of the frontal cortex, and
mic volume and prefrontal cortical spindle activity in schizophrenia,
the cortex. Philos. Trans. R. Soc. Lond. B Biol. Sci. 357, 1695–1708.
7A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
UNCO
RREC
connectivity analyses revealed weaker thalamus–superior frontal cor-tex in schizophrenia patients compared to healthy controls (Gulleret al., 2012). Future studies could also contribute to assess the implica-tion of this thalamo-cortical circuit in the symptoms of schizophrenia,as recently suggested (Lisman, 2011). Finally, improvements in somemeasures of this thalamo-cortical circuitry may be utilized to establishthe effectiveness of both pharmacological and non-pharmacological in-terventions in schizophrenia. For example, it was recently found thatEszopiclone significantly increased sleep spindles in schizophrenia pa-tients, and this increase correlated with overnight motor sequencetask improvement (Wamsley et al., 2013), whereas a recent fMRIstudy showing a significant improvement in reality monitoring inschizophrenics after 80 h of computerized training, which correlatedwith increased activity in the prefrontal cortex (Subramaniam et al.,2012).
Conclusion
In this study we collected both structural (i.e., MRI-based VOI analy-sis of thalamic volumes) and functional (i.e., sleep spindle-related corti-cal currents) measures in patients with schizophrenia and healthycomparison subjects. We found that schizophrenia patients had re-duced mediodorsal (MD) thalamic volumes, that MD volumes werestrongly correlated with the number of scalp-recorded anterior frontalspindles, and that cortical currents underlying those frontal spindleswere localized in the prefrontal cortex (BA 10). Finally, we establishedthat prefrontal currents at BA 10 were significantly reduced in schizo-phrenia patients and predicted their performance in an abstraction/working memory task. Altogether, these findings point to deficits in aspecific thalamo-cortical circuitry in schizophrenia, whichmay underliesome cognitive impairments commonly reported in those patients.
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.neuroimage.2014.08.017.
Uncited references
Anon., 2012Anon., 2014
References
Anon., 2012. Genome-wide association study implicates HLA-C*01:02 as a risk factor atthe major histocompatibility complex locus in schizophrenia. Biol. Psychiatry 72,620–628.
Astori, S., Wimmer, R.D., Prosser, H.M., Corti, C., Corsi, M., Liaudet, N., Volterra, A., Franken, P.,Adelman, J.P., Luthi, A., 2011. The Ca(V)3.3 calcium channel is the major sleep spindlepacemaker in thalamus. Proc. Natl. Acad. Sci. U. S. A. 108, 13823–13828.
Berman, K.F., Ostrem, J.L., Randolph, C., Gold, J., Goldberg, T.E., Coppola, R., Carson, R.E.,Herscovitch, P., Weinberger, D.R., 1995. Physiological activation of a cortical networkduring performance of the Wisconsin Card Sorting Test: a positron emission tomog-raphy study. Neuropsychologia 33, 1027–1046.
Byne, W., Buchsbaum, M.S., Kemether, E., Hazlett, E.A., Shinwari, A., Mitropoulou, V.,Siever, L.J., 2001. Magnetic resonance imaging of the thalamic mediodorsal nucleusand pulvinar in schizophrenia and schizotypal personality disorder. Arch. Gen. Psy-chiatry 58, 133–140.
Clemens, Z., Fabo, D., Halasz, P., 2006. Twenty-four hours retention of visuospatial mem-ory correlates with the number of parietal sleep spindles. Neurosci. Lett. 403, 52–56.
Danos, P., Baumann, B., Kramer, A., Bernstein, H.G., Stauch, R., Krell, D., Falkai, P., Bogerts, B.,2003. Volumes of association thalamic nuclei in schizophrenia: a postmortem study.Schizophr. Res. 60, 141–155.
Daskalakis, Z.J., Christensen, B.K., Fitzgerald, P.B., Moller, B., Fountain, S.I., Chen, R., 2008.Increased cortical inhibition in persons with schizophrenia treated with clozapine.J. Psychopharmacol. 22, 203–209.
Dazzan, P., Morgan, K.D., Orr, K., Hutchinson, G., Chitnis, X., Suckling, J., Fearon, P.,McGuire, P.K., Mallett, R.M., Jones, P.B., Leff, J., Murray, R.M., 2005. Different effectsof typical and atypical antipsychotics on grey matter in first episode psychosis: theAESOP study. Neuropsychopharmacology 30, 765–774.
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
ED P
RO
OF
Deng, M.Y., McAlonan, G.M., Cheung, C., Chiu, C.P., Law, C.W., Cheung, V., Sham, P.C., Chen,E.Y., Chua, S.E., 2009. A naturalistic study of grey matter volume increase after earlytreatment in anti-psychotic naive, newly diagnosed schizophrenia. Psychopharma-cology (Berlin) 206, 437–446.
Dorph-Petersen, K.A., Caric, D., Saghafi, R., Zhang, W., Sampson, A.R., Lewis, D.A., 2009.Volume and neuron number of the lateral geniculate nucleus in schizophrenia andmood disorders. Acta Neuropathol. 117, 369–384.
Ettinger, U., Picchioni, M., Landau, S., Matsumoto, K., van Haren, N.E., Marshall, N.,Hall, M.H., Schulze, K., Toulopoulou, T., Davies, N., Ribchester, T., McGuire, P.K.,Murray, R.M., 2007. Magnetic resonance imaging of the thalamus and adhesiointerthalamica in twins with schizophrenia. Arch. Gen. Psychiatry 64, 401–409.
Ferrarelli, F., Huber, R., Peterson, M.J., Massimini, M., Murphy, M., Riedner, B.A., Watson, A.,Bria, P., Tononi, G., 2007. Reduced sleep spindle activity in schizophrenia patients. Am.J. Psychiatry 164, 483–492.
Ferrarelli, F., Peterson, M.J., Sarasso, S., Riedner, B.A., Murphy, M.J., Benca, R.M., Bria, P.,Kalin, N.H., Tononi, G., 2010. Thalamic dysfunction in schizophrenia suggested bywhole-night deficits in slow and fast spindles. Am. J. Psychiatry 167, 1339–1348.
Fogel, S.M., Smith, C.T., 2011. The function of the sleep spindle: a physiological index ofintelligence and a mechanism for sleep-dependent memory consolidation. Neurosci.Biobehav. Rev. 35, 1154–1165.
Fuentealba, P., Steriade, M., 2005. The reticular nucleus revisited: intrinsic and networkproperties of a thalamic pacemaker. Prog. Neurobiol. 75, 125–141.
Gais, S., Molle, M., Helms, K., Born, J., 2002. Learning-dependent increases in sleep spindledensity. J. Neurosci. 22, 6830–6834.
Guillery, R.W., Sherman, S.M., 2002. Thalamic relay functions and their role incorticocortical communication: generalizations from the visual system. Neuron 33,163–175.
Guller, Y., Ferrarelli, F., Shackman, A.J., Sarasso, S., Peterson, M.J., Langheim, F.J.,Meyerand, M.E., Tononi, G., Postle, B.R., 2012. Probing thalamic integrity inschizophrenia using concurrent transcranial magnetic stimulation and function-al magnetic resonance imaging. Arch. Gen. Psychiatry 69, 662–671.
Jardemark, K., Marcus, M.M., Shahid, M., Svensson, T.H., 2010. Effects of asenapine on pre-frontal N-methyl-D-aspartate receptor-mediated transmission: involvement of dopa-mine D1 receptors. Synapse 64, 870–874.
Kalkstein, S., Hurford, I., Gur, R.C., 2010. Neurocognition in schizophrenia. Curr. Top.Behav. Neurosci. 4, 373–390.
Kemether, E.M., Buchsbaum, M.S., Byne, W., Hazlett, E.A., Haznedar, M., Brickman, A.M.,Platholi, J., Bloom, R., 2003. Magnetic resonance imaging of mediodorsal, pulvinar,and centromedian nuclei of the thalamus in patients with schizophrenia. Arch. Gen.Psychiatry 60, 983–991.
Krause, M., Hoffmann, W.E., Hajos, M., 2003. Auditory sensory gating in hippocam-pus and reticular thalamic neurons in anesthetized rats. Biol. Psychiatry 53,244–253.
Light, G.A., Braff, D.L., 1999. Human and animal studies of schizophrenia-related gatingdeficits. Curr. Psychiatry Rep. 1, 31–40.
Lisman, J., 2011. Excitation, inhibition, local oscillations, or large-scale loops: what causesthe symptoms of schizophrenia? Curr. Opin. Neurobiol.
Manoach, D.S., Thakkar, K.N., Stroynowski, E., Ely, A., McKinley, S.K., Wamsley, E.,Djonlagic, I., Vangel, M.G., Goff, D.C., Stickgold, R., 2010. Reduced overnight consolida-tion of procedural learning in chronic medicated schizophrenia is related to specificsleep stages. J. Psychiatr. Res. 44, 112–120.
Molle, M., Born, J., 2011. Slow oscillations orchestrating fast oscillations andmemory con-solidation. Prog. Brain Res. 193, 93–110.
Nakamura, M., Uchida, S., Maehara, T., Kawai, K., Hirai, N., Nakabayashi, T., Arakaki, H.,Okubo, Y., Nishikawa, T., Shimizu, H., 2003. Sleep spindles in human prefrontal cor-tex: an electrocorticographic study. Neurosci. Res. 45, 419–427.
Pakkenberg, B., Scheel-Kruger, J., Kristiansen, L.V., 2009. Schizophrenia; from structure tofunction with special focus on the mediodorsal thalamic prefrontal loop. ActaPsychiatr. Scand. 120, 345–354.
Pinault, D., 2004. The thalamic reticular nucleus: structure, function and concept. BrainRes. Brain Res. Rev. 46, 1–31.
Ramcharan, E.J., Gnadt, J.W., Sherman, S.M., 2005. Higher-order thalamic relays burstmore than first-order relays. Proc. Natl. Acad. Sci. U. S. A. 102, 12236–12241.
Rosanova, M., Ulrich, D., 2005. Pattern-specific associative long-term potentiation in-duced by a sleep spindle-related spike train. J. Neurosci. 25, 9398–9405.
Santana, N., Troyano-Rodriguez, E., Mengod, G., Celada, P., Artigas, F., 2011. Activation ofthalamocortical networks by the N-methyl-D-aspartate receptor antagonist phency-clidine: reversal by clozapine. Biol. Psychiatry 69, 918–927.
Schabus, M., Dang-Vu, T.T., Albouy, G., Balteau, E., Boly, M., Carrier, J., Darsaud, A.,Degueldre, C., Desseilles, M., Gais, S., Phillips, C., Rauchs, G., Schnakers, C.,Sterpenich, V., Vandewalle, G., Luxen, A., Maquet, P., 2007. Hemodynamic cerebralcorrelates of sleep spindles during human non-rapid eye movement sleep. Proc.Natl. Acad. Sci. U. S. A. 104, 13164–13169.
Schabus, M., Hoedlmoser, K., Pecherstorfer, T., Anderer, P., Gruber, G., Parapatics, S.,Sauter, C., Kloesch, G., Klimesch, W., Saletu, B., Zeitlhofer, J., 2008. Interindividualsleep spindle differences and their relation to learning-related enhancements. BrainRes. 1191, 127–135.
Seeck-Hirschner, M., Baier, P.C., Sever, S., Buschbacher, A., Aldenhoff, J.B., Goder, R., 2010.
Effects of daytime naps on procedural and declarative memory in patients withschizophrenia. J. Psychiatr. Res. 44, 42–47.
Sherman, S.M., Guillery, R.W., 2002. The role of the thalamus in the flow of information to
mic volume and prefrontal cortical spindle activity in schizophrenia,
8 A. Buchmann et al. / NeuroImage xxx (2014) xxx–xxx
Shimizu, M., Fujiwara, H., Hirao, K., Namiki, C., Fukuyama, H., Hayashi, T., Murai, T., 2008.Structural abnormalities of the adhesio interthalamica and mediodorsal nuclei of thethalamus in schizophrenia. Schizophr. Res. 101, 331–338.
Silber, M.H., Ancoli-Israel, S., Bonnet, M.H., Chokroverty, S., Grigg-Damberger, M.M.,Hirshkowitz, M., Kapen, S., Keenan, S.A., Kryger, M.H., Penzel, T., Pressman, M.R.,Iber, C., 2007. The visual scoring of sleep in adults. J. Clin. Sleep Med. 3, 121–131.
Spering, M., Dias, E.C., Sanchez, J.L., Schutz, A.C., Javitt, D.C., 2013. Efference copy failureduring smooth pursuit eye movements in schizophrenia. J. Neurosci. 33,11779–11787.
Subramaniam, K., Luks, T.L., Fisher, M., Simpson, G.V., Nagarajan, S., Vinogradov, S., 2012.Computerized cognitive training restores neural activity within the reality monitor-ing network in schizophrenia. Neuron 73, 842–853.
Szulc, A., Galinska-Skok, B., Tarasow, E., Konarzewska, B., Waszkiewicz, N., Hykiel, R.,Walecki, J., 2012. Clinical and cognitive correlates of the proton magnetic resonancespectroscopy measures in chronic schizophrenia. Med. Sci. Monit. 18, CR390–CR398.
Tomelleri, L., Jogia, J., Perlini, C., Bellani, M., Ferro, A., Rambaldelli, G., Tansella, M.,Frangou, S., Brambilla, P., 2009. Brain structural changes associated with chronicityand antipsychotic treatment in schizophrenia. Eur. Neuropsychopharmacol. 19,835–840.
van Haren, N.E., Hulshoff Pol, H.E., Schnack, H.G., Cahn, W., Mandl, R.C., Collins, D.L.,Evans, A.C., Kahn, R.S., 2007. Focal gray matter changes in schizophrenia acrossthe course of the illness: a 5-year follow-up study. Neuropsychopharmacology32, 2057–2066.
UNCO
RRECT
Please cite this article as: Buchmann, A., et al., Reduced mediodorsal thalaNeuroImage (2014), http://dx.doi.org/10.1016/j.neuroimage.2014.08.017
Wamsley, E.J., Shinn, A.K., Tucker, M.A., Ono, K.E., McKinley, S.K., Ely, A.V., Goff, D.C.,Stickgold, R., Manoach, D.S., 2013. The effects of eszopiclone on sleep spindles andmemory consolidation in schizophrenia: a randomized placebo-controlled trial.Sleep 36, 1369–1376.
Wei, H., Bonjean,M., Petry, H.M., Sejnowski, T.J., Bickford, M.E., 2011. Thalamic burst firingpropensity: a comparison of the dorsal lateral geniculate and pulvinar nuclei in thetree shrew. J. Neurosci. 31, 17287–17299.
Yoshihara, Y., Sugihara, G., Matsumoto, H., Suckling, J., Nishimura, K., Toyoda, T.,Isoda, H., Tsuchiya, K.J., Takebayashi, K., Suzuki, K., Sakahara, H., Nakamura, K.,Mori, N., Takei, N., 2008. Voxel-based structural magnetic resonance imaging(MRI) study of patients with early onset schizophrenia. Ann Gen Psychiatry 7, 25.
Yuan, P., Raz, N., 2014. Prefrontal cortex and executive functions in healthy adults: ameta-analysis of structural neuroimaging studies. Neurosci. Biobehav. Rev. 42C,180–192.
Zhang, Y., Llinas, R.R., Lisman, J.E., 2009. Inhibition of NMDARs in the nucleus reticularis ofthe thalamus produces delta frequency bursting. Front. Neural Circ. 3, 20.
O
ED P
RO
mic volume and prefrontal cortical spindle activity in schizophrenia,