Psychoneuroendocrinology (2014) 50, 95—105

Available online at www.sciencedirect.com

ScienceDirect

j ourna l h om epa ge : www.elsev ier .com/ locate /psyneuen

Neurophysiological correlates of metabolicsyndrome and cognitive impairment inschizophrenia: A structural equationmodeling approach

Laurent Boyera,∗,1, Julien Testartb,1, Pierre Michela,Raphaëlle Richierib, Catherine Faget-Agiusb, Violette Vanoyeb,Pascal Auquiera, Christophe Lancona,b, Eric Guedjc,d,e

a Aix-Marseille Univ, EA 3279 — Public Health, Chronic Diseases and Quality of Life — Research Unit, 13005Marseille, Franceb Department of Psychiatry, Sainte-Marguerite University Hospital, 13009 Marseille, Francec APHM, Hôpital de la Timone, Service Central de Biophysique et Médecine Nucléaire, 13005 Marseille,Franced Aix-Marseille Univ, CERIMED 13005, Marseille, Francee Aix-Marseille Univ, INT, 13005 Marseille, France

Received 18 March 2014; received in revised form 18 July 2014; accepted 18 July 2014

KEYWORDSSchizophrenia;Metabolic syndrome;Cognition;Single-photonemission computedtomography (SPECT);Cerebral blood flow;

SummaryObjective: We aimed to investigate the brain functional substrate underlying relationshipsbetween metabolic syndrome (MetS) and cognitive impairment in schizophrenia.Methods: In this cross-sectional study, we collected socio-demographic, clinical, anthropomet-ric, blood, and cognition data and performed brain 99mTc-ECD-SPECT imaging of cerebral bloodflow in patients with schizophrenia. Patients were grouped according to the absence or presenceof MetS. Whole-brain perfusion SPECTs were compared at voxel level between these two groups,and voxel-wise interregional correlation was performed to compare functional connectivity

(voxel level significance of p < 0.005, uncorrected; p < 0.05 for the cluster, uncorrected; using SPM;

Structural equationmodeling

SPM8). A structural equation model (SEM) was applied to examine the relationships betweenbrain perfusion, connectivity between brain areas, and cognition.Results: Of the 55 patients, 17 had MetS. They performed significantly worse than patientswithout MetS on tests of executive functions (processing speed p = 0.005 for TMT-A; and reactiveflexibility p = 0.014 for TMT-B), attention (D2 attention task p = 0.007), and memory (California

∗ Corresponding author at: EA 3279 — Self-Perceived Health Assessment Research Unit, School of Medicine, La Timone University, 13005Marseille, France. Tel.: +33 686936276; fax: +33 491433516.

E-mail address: laurent.boyer@ap-hm.fr (L. Boyer).1 These authors equally participate as first author.

http://dx.doi.org/10.1016/j.psyneuen.2014.07.0190306-4530/© 2014 Elsevier Ltd. All rights reserved.

96 L. Boyer et al.

Verbal Learning Test p = 0.039). In comparison to patients without MetS, those with MetS exhibitedsignificant hypoperfusion within the left orbital prefrontal cortex and greater functional connec-tivity from this left frontal cluster within the left insula and middle/superior frontal gyrus. SEMconfirmed the effect on executive functions of brain hypoperfusion and of increased connectivity,suggesting possible compensatory networks in patients with MetS.Conclusion: Our study identifies the brain functional impact of MetS on cognition, with orbitalprefrontal impairment and possible compensatory networks.© 2014 Elsevier Ltd. All rights reserved.

1

Cfnes(fibsi(

t(siMl2eiipnsaretnk(2rlhi

cb(reuaaum

vartprc2

2

2

ThFpciaIldIWaCa(eS

hNC(acric

2

The following data were collected:

. Introduction

ognitive impairments in memory, attention, and executiveunctions (Nuechterlein et al., 2004) are critical determi-ants of functional disability in schizophrenia (Dickinsont al., 2004; Heinrichs and Zakzanis, 1998), even afteruccessful treatment and reduction of psychotic symptomsMcGurk et al., 2004; Nuechterlein et al., 2012). Becauserst and second generation antipsychotics have limitedenefits on these impairments (Keefe et al., 2007), under-tanding the neurobiological mechanisms affecting cognitionn schizophrenia may serve to develop new approachesGold, 2004).

On the other hand, metabolic syndrome (MetS) andhe micro- and macro-cerebrovascular alterations it causesYaffe et al., 2004) are increasingly recognized as respon-ible for cognitive dysfunction. This may offer interestingnterventional prospects. In non-psychiatric populations,etS has a significant influence on cognition among ado-

escent (Yau et al., 2012), middle-aged (Hassenstab et al.,010), and older individuals (Dik et al., 2007; van den Bergt al., 2009; Yaffe et al., 2004). These findings are of utmostmportance in psychiatry, given that the prevalence of MetSs higher in patients with schizophrenia than in the generalopulation. Data on the association between MetS and cog-ition are, however, scarce and contradictory in patientsuffering from schizophrenia. The Clinical Antipsychotic Tri-ls of Intervention Effectiveness (CATIE) study found noelationship between MetS and cognitive impairment (Meyert al., 2005). In contrast, five recent studies have foundhat schizophrenia patients with MetS had more severe cog-itive impairments than patients without MetS in threeey domains: memory, attention, and executive functionsBoyer et al., 2013; Dickinson et al., 2008; Friedman et al.,010; Lancon et al., 2012; Lindenmayer et al., 2012). Neu-oimaging studies may provide important insights on theinks between MetS and cognition in schizophrenia. To date,owever, no functional neuroimaging study has specificallynvestigated this relationship.

99m Tc-ethyl cysteinate dimer single-photon emissionomputed tomography (99mTc-ECD-SPECT) is a valuablerain imaging tool to study regional cerebral blood flowrCBF), and functional connectivity (Cilia et al., 2011), in aange of psychiatric disorders such as schizophrenia (Boyert al., 2012b,c; Faget-Agius et al., 2012). rCBF, in partic-lar, is altered by the micro- and macro-cerebrovascular

lterations caused by MetS (Yaffe et al., 2004). Weimed to investigate here the brain functional substratenderlying relationships between MetS and cognitive impair-ent in schizophrenia, using whole-brain 99mTc-ECD-SPECT

1

2

oxel-based statistical analysis, i.e., without any regional priori hypothesis. To explore the exact nature of theelationships between MetS and cognition, we used struc-ural equation modeling (SEM), which is a useful statisticalrocedure to test a theory involving non-straightforwardelationships that is well suited for the management ofross-sectional data for inferential purposes (Falissard,005).

. Methods

.1. Study participants

he study evaluated all prospective patients attending ourospital in the daytime from January 2011 to June 2011.ifty-five outpatients with schizophrenia were included. Allatients provided written informed consent. The inclusionriteria were the following: being over 18 years of age, hav-ng a diagnosis of schizophrenia according to the Diagnosticnd Statistical Manual of Mental Disorders, 4th ed. (DSM-V-TR) criteria (APA, 2000), and speaking French as nativeanguage. The exclusion criteria included the following:iagnosis other than schizophrenia on Axis I of the DSM-V, decompensated organic disease, and mental retardation.e also included 26 healthy controls similar to patients in

ge and gender (mean age 36.1 [SD 9.2] years, 20 men).ontrols had to be free of neurologic/psychiatric diseasend of MetS, and had to have a normal brain MRI scanNCT00484523). They were in particular screened for anxi-ty and depression using the Anxiety and Depression Hospitalcale (subscores < 8) (Zigmond and Snaith, 1983).

The local ethics committee approved investigations forealthy subjects (registration number of clinical trial:CT00484523). The data collection was approved by theommission Nationale de l’Informatique et des LibertésCNIL number: 1223715). Our research was conducted inccordance with the Declaration of Helsinki and French goodlinical practices. In particular, both patients and controlseceived an explanation of the study and gave written,nformed consent after a standardized and structured clini-al interview.

.2. Data collection

. Socio-demographic information: age, gender, and educa-tional level.

. Clinical characteristics: duration of disease; psychoticsymptoms based on the Positive and Negative Syndrome

gnit

Tf(frwT(

ci(eerrsa

2

Dd

ttw(gmwwlpttciS

rvnS

damtda(dB

3

Neurophysiological correlates of metabolic syndrome and co

Scale (PANSS), which comprises three subscales, i.e.,positive, negative, and general psychopathology (Kayet al., 1986); depression based on the Calgary DepressionScale for Schizophrenia (CDSS) (Addington et al., 1993);smoking status assessed by the Fagerstrom Test for Nico-tine Dependence (FTND) (Heatherton et al., 1991); andalcohol dependence or abuse determined by the DSM-IV-TR (during the past year) (APA, 2000).

3. Drug information: antipsychotic medications (first-generation antipsychotics, second-generation antipsy-chotics). Antipsychotic medications were stable for aminimum of 3 months.

4. Neurocognitive assessment: based on previous research,several measures were selected to test memory, atten-tion, and executive functions (Baumstarck et al., 2013;Boyer et al., 2012a; Szoke et al., 2008; Godefroy, 2008).Memory was tested using the California Verbal LearningTest (CVLT), attention was tested using the D2 atten-tion task, and executive functions were tested using theStroop color-word test for inhibition capacity, the verbalfluency test (category domains) for spontaneous flexibil-ity, the Trail Making Test A and B for processing speedand reactive flexibility, and the Wechsler Adult Intelli-gence Scale — Third Edition (WAIS-III symbol coding) forupdating.

5. Metabolic syndrome: diagnosis of MetS was definedaccording to the National Cholesterol Education Pro-gram’s ATP-III criteria (Grundy et al., 2005), including3 or more of the following criteria: waist circumference>88 cm in women and >102 cm in men, fasting serum tri-glycerides ≥150 mg/dL; serum HDL <50 mg/dL in womenand <40 mg/dL in men, blood pressure ≥130/85 mmHg;and fasting blood glucose levels ≥110 mg/dL. Waistcircumference was measured at the midpoint betweenthe lower rib margins and the iliac crest. Arterial bloodpressure was recorded using a standard mercury sphyg-momanometer. Glucose and lipoprotein concentrationswere analyzed in fasting venous blood samples usingstandard enzymatic techniques. All the measures wereperformed at the same time.

2.3. SPECT protocol

All subjects had a brain SPECT, with the same camera andunder the same conditions as previously reported (Boyeret al., 2012b,c; Faget-Agius et al., 2012). A voxel-by-voxelgroup study was then performed using SPM8 (WellcomeDepartment of Cognitive Neurology, University College, Lon-don, UK), running on Matlab (Mathworks Inc., Sherborn, MA).

Data were standardized with the Montreal Neurolog-ical Institute (MNI) atlas, using a 12-parameter affinetransformation, followed by non-linear transformationsand trilinear interpolation. Dimensions of resulting voxelswere 2 mm × 2 mm × 2 mm. Standardized data were thensmoothed with a Gaussian filter (FWHM = 8 mm) to blurindividual variations in gyral anatomy and to increase signal-to-noise ratio. The following variables were considered as

confounding factors based on a threshold p-value ≤ 0.20from the univariate analysis comparing the two groupsof patients (absence or presence of MetS): age, diseaseduration and presence/absence clozapine or olanzapine.

3

T7

ive impairment in schizophrenia 97

he ‘‘proportional scaling’’ routine was used to controlor individual variation in global brain perfusion. The SPMT) maps were obtained at a height threshold of p < 0.005or the voxel and at p < 0.05 for the cluster (uncor-ected). Normalized perfusion values of significant clustersere extracted, and MNI coordinates were converted into

alairach coordinates using the Talairach Daemon databasehttp://ric.uthscsa.edu/projects/talairachdaemon.html).

Moreover, functional connectivity from the significantlusters previously identified in the between-group compar-son was studied using voxel-wise interregional correlationIRCA), as previously described (Cilia et al., 2011; Leet al., 2008; Morbelli et al., 2012). In brief, mean values ofxtracted perfusion clusters were used as covariates to findegions showing significant voxel-wise positive/negative cor-elations across subjects and between groups, and using theame nuisance variables and the same statistical thresholds in our SPM analysis.

.4. Statistical analysis

ata were expressed in proportion or mean and standardeviation.

Chi-squared tests or t-tests were conducted to comparehe characteristics of the study sample in terms of MetS sta-us (absence or presence). Functional brain abnormalitiesere determined by comparing the two groups of patients

absence or presence of MetS) after adjustment on age,ender, educational level, disease duration, patients’ treat-ent, smoking, and alcohol usage. Brain SPECT perfusionsere compared between the 2 patient groups (with andithout MetS) and the healthy subjects. Correlations were

ooked for between cluster(s) in each of the two groups ofatients, using Pearson’s coefficients. Fisher’s Z transforma-ion was then used to compare the correlations between thewo groups. All the tests were two-tailed and p-values wereonsidered significant when p < 0.05 (except for neuroimag-ng SPM analysis: see above). These data were analyzed usingPSS, version 20.0, software package.

According to the clinical/imaging results and literatureeview, we constructed hypothetical relationships amongariables (i.e., cognitive functions, brain perfusion, and con-ectivity values) and applied SEM for each group of patients.EM was conducted with IBM SPSS AMOS® (see Appendix).

We conducted a complementary exploratory approach toetermine which component of the MetS shows a greaterssociation with brain abnormalities and cognitive impair-ent. Multiple linear regressions were performed to assess

hese associations. The final models incorporated the stan-ardized ˇ coefficients for all the variables, which represent

change in the standard deviation of the dependent variableMetS components) resulting from a change of one standardeviation in the various independent variables (TMT-A, TMT-

and perfusion within the left orbital prefrontal cortex.

. Results

.1. Clinical characteristics

he mean age of the patients was 35.6 years (±12.0), and6.4% of them (n = 42) were male. The patients showed

9

m6agn(c

wstat

Pf1(—ta

p

8

oderate severity of symptoms, with a total PANSS score of8.1 (±17.9) and sub-scores of 13.6 (±4.5), 19.04 (±7.1),nd 35.45 (±9.6), respectively, for positive, negative, andeneral psychopathology factors. The mean duration of ill-ess was 12.0 years (±8.3). Of the total number of patientsn = 55), 90.9% had been taking second-generation antipsy-hotics (n = 50), and 30.9% (n = 17) met the criteria for MetS.

Compared to patients without MetS (n = 38), thoseith MetS (n = 17) did not differ statistically regarding

ocio-demographic data, clinical values, and antipsychoticreatment (Table 1). All the patients were taking atypicalntipsychotics. The distribution of antipsychotic medica-ions was comparable between the two groups (Table 2).

tTpp

Table 1 Sample characteristics and metabolic syndrome status (

Non-MetS(

Socio-demographic characteristicsSex ratio (%): male 78.9

Age in years: mean ± SD 34.0 ± 10.7Education level (%): ≥12 years 39.5

Clinical and paraclinical characteristicsDisease duration in years: mean ± SD 10.3 ± 7.0

PANSS total score: mean ± SD 66.7 ± 17.7Positive factor 13.2 ± 4.1

Negative factor 18.5 ± 7.0

General psychopathology factor 35.0 ± 9.8

Glycaemia: mean ± SD 4.5 ± 0.8

HDL cholesterol: mean ± SD 1.3 ± 0.5

Triglycerides: mean ± SD 1.2 ± 0.5

Blood pressure: mean ± SD 120 ± 15/7Waist circumference: mean ± SD 93.3 ± 12.2CRP (�g/mL): mean ± SD 3.4 ± 4.8

Antipsychotics (AP)Atypical AP (%) 100

Associated with typical AP (%) 7.9

Clozapine or Olanzapine 44.7

AP doses (equivalent chlorpromazine) 729.6 ± 55

Fagerstrom Test: mean ± SD 4.3 ± 2.9

Alcohol consumption (%) 27.0

Metabolic syndrome criteriaHyperglycaemia (%) 1 (2.6)

Low HDL cholesterol (%) 7 (18.4)

High triglycerides (%) 2 (5.3)

Hypertension (%) 8 (21.1)

Abdominal obesity (%) 11 (28.9)

Cognitive assessment: mean ± SDMemory

CVLT List A — 1—5 Memory 45.5 ± 11.6Attention

D2 attention task 138.3 ± 43Executive function

Processing speed TMT-A (s) 42.2 ± 32.3Cognitive flexibility TMT-B (s) 97.2 ± 49.1Stroop interference 41.4 ± 14.6Category fluency 25.8 ± 8.1

WAIS III — Digit Symbol-Coding 6.1 ± 3.0

Figures in bold indicate statistical significance.

L. Boyer et al.

atients with MetS had significantly lower cognitive per-ormance in the D2 attention task (p = 0.007), CVLT-List A—5 Memory test (p = 0.039), and executive function testsprocessing speed — TMT-A, p = 0.005; cognitive flexibility

TMT-B, p = 0.014). The largest difference between thewo groups concerned cognitive flexibility (delta = 52.8) andttention (delta = 42.8).

The duration of the illness was associated with aoorer cognition for the following tests: D2 attention

ask (r = −0.51; p < 0.001), TMT-A (r = 0.42; p = 0.002),MT-B (r = 0.35; p = 0.011), Stroop interference (r = −0.42;

= 0.004) and CVLT List A — 1—5 Memory (r = −0.30; = 0.029). There was no association with Category fluency

N = 55).

n = 38) MetS (n = 17) P-Value

70.6 .511 39.2 ± 14.1 .199

47.1 .510

15.7 ± 9.9 .065 71.3 ± 18.4 .353

14.4± 5.5 .55920.3 ± 7.3 .55336.6 ± 9.3 .5725.9 ± 1.6 <.0011.1 ± 0.5 .2012.6 ± 1.6 <.001

1 ± 11 135 ± 12/84 ± 8 <.001 108.8 ± 12.1 <.001

8.8 ± 7.4 .003

100 1.00011.8 .64770.6 .089

1.1 983.8 ± 671.4 .218

4.6 ± 2.1 .87431.2 .751

9 (52.9) <.000113 (76.5) <.000112 (70.6) <.000113 (76.5) <.000114 (82.4) <.0001

37.3 ± 12.1 .039

.7 94.5 ± 49.2 .007

52.8 ± 18.7 .005150.0 ± 106.1 .014

37.0 ± 10.7 .36322.6 ± 5.4 .2994.7 ± 2.3 .091

Neurophysiological correlates of metabolic syndrome and cognit

Table 2 Antipsychotic medication distribution of patientswith and without metabolic syndrome.

Metabolic syndrome Total

Absent(n = 38)

Present(n = 17)

n (%) n (%)

Typicals 0 (0%) 0 (0%) 0 (0%)

AtypicalsAmisulpride 2 (5.3%) 0 (0%) 2 (3.6%)Aripiprazole 8 (21.1%) 2 (11.8%) 10 (18.2%)Clozapine 8 (21.1%) 5 (29.4%) 13 (23.6%)Olanzapine 7 (18.4%) 3 (17.6%) 10 (18.2%)Risperidone 3 (7.9%) 1 (5.9%) 4 (7.3%)Risperidonelong-acting

2 (5.3%) 1 (5.9%) 3 (3.6%)

Atypi-cals + typicals

1 (2.6%) 0 (0%) 1 (1.8%)

More than oneatypical

5 (13.2%) 1 (5.9%) 6 (10.9%)

Clozapine + 2 (5.3%) 4 (23.5%) 6 (10.9%)

pap

piwTgoftfMpc(lc(fp

3c

Tbwflacfl2ZcleZm

imtabclTiec

pmfrontal cortex and executive function.

atypical/typical

(r = −0.14; p = 0.321) and WAIS III — Digit Symbol-Coding(r = −0.13; p = 0.334).

3.2. Brain SPECT findings

Brain SPECT findings are presented in Table 3 andFigs. 1 and 2. Bilateral fronto-temporal hypoperfusions werefound in patients, in comparison to healthy subjects, withwider abnormalities in patients with MetS, than those with-out. In comparison to patients without MetS, those with MetSexhibited significant hypoperfusion within the left orbitalprefrontal cortex (BA11, BA47). The severity of the hypoper-fusion was associated to the severity of MetS, as assessedusing the number of MetS criteria (r = −0.81; p < 0.001).In addition, perfusion within the left orbital prefrontalcortex differed between patients with and without MetSand healthy controls. Post hoc comparisons using the Bon-ferroni method revealed significant differences for eachpaired-comparison, confirming that patients with MetS hadlower orbital perfusion than either patients without MetS orhealthy subjects. No significant hyperperfusion was found inpatients with MetS, in comparison to those without MetS.Functional connectivity from this left frontal cluster wassignificantly increased in patients with MetS within theleft insula (BA13) and left middle/superior frontal gyrus(BA11), in comparison to patients without MetS. No sig-nificant decrease in functional connectivity from this leftfrontal cluster was found in patients with MeTS, in compar-ison to those without MetS.

Based on perfusion values of extracted clusters, Pearson’scorrelations were then looked for between left orbital pre-

frontal cortex and either left insula or left middle/superiorfrontal gyrus, to confirm the difference in functional con-nectivity.

an

ive impairment in schizophrenia 99

In the group with MetS, SPECT perfusion of left orbitalrefrontal cortex was significantly correlated with left insuland left middle/superior frontal gyrus (respectively, r = 0.92,

< 0.001 and r = 0.77, p < 0.001).In the group without MetS, SPECT perfusion of left orbital

refrontal cortex was only moderately correlated with leftnsula (r = 0.46; p = 0.004), and not significantly correlatedith left middle/superior frontal gyrus (r = 0.05; p = 0.768).hese correlations significantly differed between the tworoups (Z = −3.03, p = 0.002 for the correlations between leftrbital prefrontal cortex and left insula; Z = −3.37, p < 0.001or the correlations between left orbital prefrontal cor-ex and left middle/superior frontal gyrus), confirming thatunctional frontal connectivity is higher in patients withetS than in those without MetS. As a confirmation, SPECTerfusion of left orbital prefrontal cortex in the healthyontrol group was moderately correlated with left insular = 0.53; p = 0.005), and not significantly correlated witheft middle/superior frontal gyrus (r = 0.17; p = 0.393). Theseorrelations did not differed from the group without MetSrespectively p = 0.653 and p = 0.726), but significantly dif-ered from the group with MetS (respectively p = 0.015 and

= 0.003), with stronger correlations in patients.

.3. Structural equation model of clinicalharacteristics and SPECT findings

o apply SEM, we constructed hypothetical relationshipsetween the most impaired cognitive functions in patientsith MetS (executive functions, and in particular cognitiveexibility) and regions of interest identified in the SPECTnalysis. We hypothesized that the left orbital prefrontalortex perfusion primarily explains the level of cognitiveexibility in schizophrenia patients with MetS (Konishi et al.,002; Nakamura et al., 2008; Szatkowska et al., 2007;akzanis et al., 2005). Moreover, we also hypothesized thatoncomitantly increased connectivity within left insula andeft middle/superior frontal gyrus, known to be involved inxecutive functions (Dichter et al., 2010; Pauly et al., 2010;akzanis et al., 2005), may possibly reflect compensatoryechanisms.The SEM fitted to assess the hypothesized model is

llustrated in Fig. 3. In patients with MetS, the testingodel showed good fit based on the chi-squared statis-

ic (normed �2 = 2.53) and had RMSEA < 0.001, CFI = 1.00,nd SRMR = 0.02. The SEM revealed significant associationsetween decreased perfusion within left orbital prefrontalortex and increased connectivity within left insula andeft middle/superior frontal gyrus (path coefficients >0.83).hese brain SPECT characteristics (decreased perfusion and

ncreased connectivity) explained lower performances forxecutive functions (path coefficient = −1.00), mainly forognitive flexibility — TMT-B (path coefficient = 0.71).

In contrast, the testing model did not show good fit foratients without MetS (normed �2 > 3, RMSEA > 0.05). In thisodel, only a trend was observed between left orbital pre-

The same models were tested for attention and memorynd other executive functions and, as expected, results didot differ significantly for patients with and without MetS.

100 L. Boyer et al.

Table 3 Talairach coordinates of significant SPECT findings in patients (voxel level significance of p < 0.005, uncorrected; p < 0.05for the cluster, uncorrected).

Contrast Cluster equiv k p(unc) Peak T Peak Talairach coordinates Localization

p(unc) x y z

Hypoperfusion in patients withmetabolic syndrome, in comparisonto those without

681 0.012 3.51 <0.001 −12 63 −20 Left superior frontalgyrus (BA11)

3.41 0.001 −46 40 −14 Left inferior frontalgyrus (BA47)

Increased functional connectivityfrom left frontal cortex in patientswith metabolic syndrome, incomparison to those without

432 0.026 4.00 <0.001 −38 10 0 Left insula (BA13)377 0.036 3.90 <0.001 −24 55 19 Left middle frontal

gyrus (BA10)3.24 0.001 −20 60 −11 Left superior frontal

gyrus (BA11)

alueinate

3a

Mnttar

4

Tn2YiOlneIat

osm(puIicsLfiotis

(ehmrRtihtceescsbmsm

tfhrt(pcfgpenpttt

Results from SPM8 are listed in decreasing order of peak T-score vvoxels in the particular cluster. x, y and z are the Talairach coord

.4. MetS components associated with brainbnormalities and cognitive impairment

ultivariate analyses results are reported in Table 4. Sig-ificant associations were found between perfusion withinhe left orbital prefrontal cortex and hyperglycaemia, highriglycerides and abdominal obesity; between TMT-A andbdominal obesity; and between TMT-B and high triglyce-ides and abdominal obesity.

. Discussion

he impact of MetS on cognition has been largely recog-ized in patients with non-psychiatric disorders (Dik et al.,007; Hassenstab et al., 2010; van den Berg et al., 2009;affe et al., 2004; Yau et al., 2012), but the studies explor-ng this issue in schizophrenia are scarce and discordant.ne important finding of our study is the confirmation of the

ink between cognitive impairments and MetS in schizophre-ia (Boyer et al., 2013; Dickinson et al., 2008; Lancont al., 2012; Lindenmayer et al., 2012; Nasrallah, 2010).t allows us to confirm that results obtained for nearly

decade in somatic medicine are applicable to psychia-ry.

Beyond this finding, our data also support the notionf selective cognitive impairment associated with MetS inchizophrenia. In our study, cognitive impairments wereore pronounced on attention and executive functions

i.e., cognitive flexibility), contrary to studies with non-sychiatric disorders, which reported mild to moderateniform cognitive impairments (van den Berg et al., 2009).n agreement with our observation, two previous stud-es have also found that attention, working memory, andognitive flexibility were predominantly deteriorated whenchizophrenia is associated to MetS (Boyer et al., 2013;indenmayer et al., 2012). Of importance, our SPECTndings revealed significant hypoperfusion within the left

rbital prefrontal cortex in patients with MetS. The role ofhis brain area, and in particular from the left-side as foundn our study, has been described in the attentional controlystem (Baddeley, 1986), especially in shifting of attention

e

on

within each cluster. k value represents the number of significants (mm).

Konishi et al., 2002; Nakamura et al., 2008; Szatkowskat al., 2007). Interestingly, the orbital prefrontal cortexas been also involved in regulating impulse control, self-onitoring, and goal-directed behaviors, all of which play a

ole in eating behavior and food choice (Cohen et al., 2011;olls, 2007; Schoenbaum et al., 2006). We might proposehat the more severe dysfunction of orbital prefrontal cortexn patients with schizophrenia, through its relationship withunger and satiety (Del Parigi et al., 2002), may then worsenhe metabolic syndrome; the heterogeneity of prefrontalognitive impairments in schizophrenia (Rodriguez-Sanchezt al., 2005), and of this orbital region, may thus partlyxplain the occurrence or not of MetS. If this hypothe-is is confirmed in future studies, it may serve to developognitive remediation approaches in patients with MetS,pecifically based on cognitive functions involved in eatingehavior (Tchanturia and Lock, 2011). Cognitive remediationay improve eating behavior and then lower the metabolic

yndrome, which may in turn improve cognitive impair-ents.We found that connectivity in patients MetS was concomi-

antly increased within left insula and left middle/superiorrontal gyrus, in comparison to patients without MetS or toealthy controls. Interestingly, these brain areas have beeneported to participate with the left orbital prefrontal cor-ex in a large neural network involved in cognitive shiftingZakzanis et al., 2005). SEM confirmed that the decreasederfusion within left orbital prefrontal cortex and increasedonnectivity within left insula and left middle/superiorrontal gyrus were associated with executive function, sug-esting possible compensatory networks. Indeed, orbitalrefrontal cortex functions are relatively preserved in firstpisode psychosis, with progressive deterioration with ill-ess progression (Rodriguez-Sanchez et al., 2005). Becauseatients were relatively young in our study, we hypothesizehat the increased connectivity may reflect the solicita-ion of additional neural resources still relatively preservedhrough the neural reserve, in order to improve or maintain

xecutive functions.

Finally, we found that among the different componentsf MetS, hyperglycaemia, hypertriglycerides and abdomi-al obesity, were the most important features associated

Neurophysiological correlates of metabolic syndrome and cognitive impairment in schizophrenia 101

Figure 1 Anatomical localization of significant SPECT findings in comparison to healthy subjects (voxel level significance ofp < 0.005, uncorrected; p < 0.05 for the cluster, uncorrected). Bilateral fronto-temporal hypoperfusions were found in patients, incomparison to healthy subjects, with wider abnormalities in patients with metabolic syndrome, than those without. (For interpre-tation of the references to color in this figure legend, the reader is referred to the web version of the article.)

Table 4 MetS components associated with brain abnormalities and cognitive impairment: multivariate analyses.

Perfusion within the leftorbital prefrontal cortexˇ#

Processing speed TMT-A(s)ˇ#

Cognitive flexibilityTMT-B (s)ˇ#

Hyperglycemia −0.40* 0.22 0.18Low HDL cholesterol −0.13 0.05 −0.34High triglycerides −0.36* −0.01 0.43*

Hypertension −0.07 −0.27 −0.03Abdominal obesity −0.27* 0.30* 0.33*

#: standardized beta coefficient ( represents the change of the standard deviation in score resulting from a change of one standarddeviation in the independent variable).

* Statistically significant in bold: p < 0.01.

102 L. Boyer et al.

Figure 2 Anatomical localization of significant SPECT findings between patients (voxel level significance of p < 0.005, uncorrected;p < 0.05 for the cluster, uncorrected). (A) Hypoperfusions in patients with metabolic syndrome, in comparison to those without,on SPM surface rendered. (B) Increase of functional connectivity with left frontal region in patients with metabolic syndrome,in comparison to those without, on SPM surface rendered. (A) In comparison to patients without metabolic syndrome, those withmetabolic syndrome exhibited significant hypoperfusion within the left orbital prefrontal cortex (inferior and superior frontal gyrus,BA11 and BA47). (B) Functional connectivity from this left frontal cluster was significantly increased in patients with metabolicsyndrome within the left insula (BA13) and left middle/superior frontal gyrus (BA11), in comparison to patients without metabolics is figa

wf(lcasfmtvcei

neiacait(

yndrome. (For interpretation of the references to color in thrticle.)

ith brain perfusion and cognitive impairment. These threeactors are well-established risk factors for atherosclerosisCheng et al., 2002), suggesting that atherosclerosis may beinked to cognitive impairment in schizophrenia. Metabolichanges may have caused micro- and macro-cerebrovascularlterations which may underlie impaired neural transmis-ion and lead to cognitive impairments such as executiveunctions (Lindenmayer et al., 2012). In addition, the MetSight contribute to accelerate the atherosclerosis process

hat is associated with an inflammatory response (i.e., ele-

ated CRP in our study) and both phenomena could thenontribute to cognitive decline (Yaffe et al., 2004). Sev-ral authors have also proposed non-vascular mechanismsnvolving actions of adiposity on neuronal tissue through

4

Tp

ure legend, the reader is referred to the web version of the

eurochemical mediators, such as leptin (Bray, 2004; Holdent al., 2009; Li et al., 2002; Oomura et al., 2006). Interest-ngly, abdominal obesity was in our study the only factorssociated with perfusion within the left orbital prefrontalortex, TMT-A and TMT-B. This finding seems consistent with

recent study which found that an elevated body massndex may be a risk factor for decreased prefrontal cor-ex function and potentially impaired executive functionWilleumier et al., 2011).

.1. Limitations and prospects

here are several limitations to this study. First, the sam-le size was relatively small, especially for the MetS group

Neurophysiological correlates of metabolic syndrome and cognitive impairment in schizophrenia 103

ith

tTctmfeFvaM

5

To

Figure 3 Structural equation model: hypothesized model wsyndrome.

(n = 17). Our sample may be not representative of patientswith MetS. Moreover, patients were mostly middle-agedmales with more than 5 years of illness duration. Thus,we need confirmation through more diverse and largergroups of patients with schizophrenia. Second, this study islimited because it is cross-sectional rather than prospectivein design. No causal inference can be formally advanced.Future longitudinal studies are needed to confirm that ourhypotheses are temporally verified. In the same line, theSEM analysis cannot totally exclude the involvement of otherbrain areas underlying cognitive impairments and MetS.Future studies should explore this specific issue. Third,several important data were not available such as dura-tion of treatment. Further studies should integrate theseparameters to examine their role. Finally, an importantmethodological problem remains in the definition of MetS.Currently, several definitions of MetS exist (Kassi et al.,2011). Although we have chosen one of the most widely used

definitions in the scientific literature, ATP-III criteria seemsmore suitable for American than European populations andthe International Diabetes Federation criteria might havebeen more suitable in our study. Another limitation is

oniM

fitted coefficients for schizophrenia patients with metabolic

hat our findings derived from at-resting perfusion imaging.he relationship between cognitive impairment and MetSould be better understood by specifically testing cogni-ion in patients with and without MetS during functionalagnetic resonance imaging tasks, even if investigation of

unctional connectivity in the resting state has recentlymerged as a powerful approach (Viviani et al., 2011).inally, future studies will need to address whether pre-enting or lowering MetS will improve cognitive impairment,nd also whether improving cognitive impairment will loweretS.

. Conclusion

his is the first demonstration of brain functional impactf MetS on executive functions in schizophrenia, reflecting

rbital prefrontal impairment and possible compensatoryetworks. This may have important clinical implications inmproving the cognition of patients with schizophrenia andetS.

1

R

N

C

N

A

StrbpcmsasnTht0oethafacrpuccim

R

A

A

B

B

B

B

B

B

B

C

C

C

C

D

D

D

D

D

F

F

F

G

G

04

ole of the funding source

one.

onflict of interest

one declared.

ppendix. Structural equation modeling (SEM)

EM enables the simultaneous performance of several mul-iple linear regressions. After designing a path diagramepresenting the investigators’ hypotheses concerning linksetween variables, SEM was used to verify the statisticallausibility of these links. We evaluated model fit using thehi-squared statistic, root mean square error of approxi-ation (RMSEA), the comparative fit index (CFI), and the

tandardized root mean square residuals (SRMRs) (Jaccardnd Wan, 1996). A small (less than 3), non-significant chi-quared value indicates that the observed correlations areot significantly different from the expected correlations.he RMSEA indicates how well the model would fit theypothetical population covariance matrix. A value of lesshan 0.05 is indicative of a close-fitting model, between.05 and 0.08 is indicative of a reasonable fit, and 0.10r greater indicates a poor model. The CFI indicates thextent to which the hypothesized model provides a bet-er fit than the null model. The comparative fit index (CFI)as a range of 0—1.0; a value greater than 0.90 suggests

reasonably good fit. Finally, SRMR (i.e., the average dif-erence between the correlations predicted by the modelnd the observed correlations) values of less than 0.10 indi-ate good fit. The standardized path coefficients, which canange from −1 to +1, indicate the strength and sign of theaths. The significance of the path coefficient is assessedsing the standard errors and the p-values for each coeffi-ient. In addition to the statistical significance of the pathoefficients, the strength of the relationship plays a rolen determining whether the relationships are weak (<0.2),oderate (0.2—0.5), or strong (>0.5) (Cohen, 1982).

eferences

ddington, D., Addington, J., Maticka-Tyndale, E., 1993. Assessingdepression in schizophrenia: the Calgary Depression Scale. Br. J.Psychiatry Suppl., 39—44.

PA, 2000. DSM-IV. Diagnostic and Statistical Manual of MentalDisorders, 4th ed., text rev. American Psychiatric Association,Washington, DC.

addeley, A.D., 1986. Working Memory. Oxford University Press,Oxford.

aumstarck, K., Boyer, L., Boucekine, M., Aghababian, V., Parola,N., Lancon, C., Auquier, P., 2013. Self-reported quality of lifemeasure is reliable and valid in adult patients suffering fromschizophrenia with executive impairment. Schizophr. Res. 147(1), 58—67.

oyer, L., Cermolacce, M., Dassa, D., Fernandez, J., Boucekine,

M., Richieri, R., Vaillant, F., Dumas, R., Auquier, P., Lancon, C.,2012a. Neurocognition, insight and medication nonadherence inschizophrenia: a structural equation modeling approach. PLoSONE 7, e47655.

L. Boyer et al.

oyer, L., Guedj, E., Dassa, D., Lancon, C., 2012b. Leber’s hered-itary optic neuropathy associated with schizophrenia. J. Clin.Psychiatry 73, e1158—e1159.

oyer, L., Richieri, R., Dassa, D., Boucekine, M., Fernandez,J., Vaillant, F., Padovani, R., Auquier, P., Lancon, C., 2013.Association of metabolic syndrome and inflammation with neu-rocognition in patients with schizophrenia. Psychiatry Res.,http://dx.doi.org/10.1016/j.psychres.2013.06.020.

oyer, L., Richieri, R., Faget, C., Padovani, R., Vaillant, F., Mundler,O., Lancon, C., Auquier, P., Guedj, E., 2012c. Functional involve-ment of superior temporal sulcus in quality of life of patientswith schizophrenia. Psychiatry Res. 202, 155—160.

ray, G.A., 2004. Obesity is a chronic, relapsing neurochemical dis-ease. Int. J. Obes. Relat. Metab. Disord. 28, 34—38.

heng, A., Braunstein, J.B., Dennison, C., Nass, C., Blumenthal,R.S., 2002. Reducing global risk for cardiovascular disease:using lifestyle changes and pharmacotherapy. Clin. Cardiol. 25,205—212.

ilia, R., Cho, S.S., van Eimeren, T., Marotta, G., Siri, C., Ko,J.H., Pellecchia, G., Pezzoli, G., Antonini, A., Strafella, A.P.,2011. Pathological gambling in patients with Parkinson’s diseaseis associated with fronto-striatal disconnection: a path modelinganalysis. Mov. Disord. 26, 225—233.

ohen, J., 1982. Statistical Power Analysis for Behavioural Sciences.Lawrence Erlbaum Associates, Hillsdale, NJ.

ohen, J.I., Yates, K.F., Duong, M., Convit, A., 2011. Obesity,orbitofrontal structure and function are associated with foodchoice: a cross-sectional study. BMJ Open 1, e000175.

el Parigi, A., Gautier, J.F., Chen, K., Salbe, A.D., Ravussin, E.,Reiman, E., Tataranni, P.A., 2002. Neuroimaging and obesity:mapping the brain responses to hunger and satiation in humansusing positron emission tomography. Ann. N. Y. Acad. Sci. 967,389—397.

ichter, G.S., Bellion, C., Casp, M., Belger, A., 2010. Impaired mod-ulation of attention and emotion in schizophrenia. Schizophr.Bull. 36, 595—606.

ickinson, D., Gold, J.M., Dickerson, F.B., Medoff, D., Dixon,L.B., 2008. Evidence of exacerbated cognitive deficits inschizophrenia patients with comorbid diabetes. Psychosomatics49, 123—131.

ickinson, D., Iannone, V.N., Wilk, C.M., Gold, J.M., 2004. Generaland specific cognitive deficits in schizophrenia. Biol. Psychiatry55, 826—833.

ik, M.G., Jonker, C., Comijs, H.C., Deeg, D.J., Kok, A., Yaffe,K., Penninx, B.W., 2007. Contribution of metabolic syndromecomponents to cognition in older individuals. Diabetes Care 30,2655—2660.

aget-Agius, C., Boyer, L., Padovani, R., Richieri, R., Mundler,O., Lancon, C., Guedj, E., 2012. Schizophrenia with preservedinsight is associated with increased perfusion of the precuneus.J. Psychiatry Neurosci. 37, 110125.

alissard, B., 2005. Modèles structuraux. In: Falissard, B. (Ed.),Comprendre et utiliser les statistiques dans les sciences de lavie (Understanding and Using Statistics in Life-related Sciences).Masson, Paris, pp. 203—215.

riedman, J.I., Wallenstein, S., Moshier, E., Parrella, M., White,L., Bowler, S., Gottlieb, S., Harvey, P.D., McGinn, T.G., Flana-gan, L., Davis, K.L., 2010. The effects of hypertension and bodymass index on cognition in schizophrenia. Am. J. Psychiatry 167,1232—1239.

old, J.M., 2004. Cognitive deficits as treatment targets inschizophrenia. Schizophr. Res. 72, 21—28.

rundy, S.M., Cleeman, J.I., Daniels, S.R., Donato, K.A., Eckel,R.H., Franklin, B.A., Gordon, D.J., Krauss, R.M., Savage, P.J.,Smith Jr., S.C., Spertus, J.A., Costa, F., 2005. Diagnosis and

management of the metabolic syndrome: an American HeartAssociation/National Heart, Lung, and Blood Institute ScientificStatement. Circulation 112, 2735—2752.

gnit

N

N

N

G

O

P

R

R

S

S

S

T

v

V

W

Y

Y

Neurophysiological correlates of metabolic syndrome and co

Hassenstab, J.J., Sweat, V., Bruehl, H., Convit, A., 2010. Metabolicsyndrome is associated with learning and recall impairment inmiddle age. Dement. Geriatr. Cogn. Disord. 29, 356—362.

Heatherton, T.F., Kozlowski, L.T., Frecker, R.C., Fagerstrom, K.O.,1991. The Fagerstrom Test for Nicotine Dependence: a revisionof the Fagerstrom Tolerance Questionnaire. Br. J. Addict. 86,1119—1127.

Heinrichs, R.W., Zakzanis, K.K., 1998. Neurocognitive deficit inschizophrenia: a quantitative review of the evidence. Neuropsy-chology 12, 426—445.

Holden, K.F., Lindquist, K., Tylavsky, F.A., Rosano, C., Harris,T.B., Yaffe, K., 2009. Serum leptin level and cognition in theelderly: findings from the Health ABC Study. Neurobiol. Aging30, 1483—1489.

Jaccard, J., Wan, C., 1996. LISREL Approaches to Interaction Effectsin Multiple Regression. Sage Publications, Thousand Oaks.

Kassi, E., Pervanidou, P., Kaltsas, G., Chrousos, G., 2011. Metabolicsyndrome: definitions and controversies. BMC Med. 9, 48.

Kay, S.R., Opler, L.A., Fiszbein, A., 1986. Significance of positive andnegative syndromes in chronic schizophrenia. Br. J. Psychiatry149, 439—448.

Keefe, R.S., Bilder, R.M., Davis, S.M., Harvey, P.D., Palmer, B.W.,Gold, J.M., Meltzer, H.Y., Green, M.F., Capuano, G., Stroup,T.S., McEvoy, J.P., Swartz, M.S., Rosenheck, R.A., Perkins, D.O.,Davis, C.E., Hsiao, J.K., Lieberman, J.A., 2007. Neurocognitiveeffects of antipsychotic medications in patients with chronicschizophrenia in the CATIE Trial. Arch. Gen. Psychiatry 64,633—647.

Konishi, S., Hayashi, T., Uchida, I., Kikyo, H., Takahashi, E.,Miyashita, Y., 2002. Hemispheric asymmetry in human lateralprefrontal cortex during cognitive set shifting. Proc. Natl. Acad.Sci. U. S. A. 99, 7803—7808.

Lancon, C., Dassa, D., Fernandez, J., Richieri, R., Padovani,R., Boyer, L., 2012. Are cardiovascular risk factors associatedwith verbal learning and memory impairment in patients withschizophrenia? A cross-sectional study. Cardiovasc. PsychiatryNeurol., 204043.

Lee, D.S., Kang, H., Kim, H., Park, H., Oh, J.S., Lee, J.S., Lee, M.C.,2008. Metabolic connectivity by interregional correlation analy-sis using statistical parametric mapping (SPM) and FDG brain PET;methodological development and patterns of metabolic connec-tivity in adults. Eur. J. Nucl. Med. Mol. Imaging 35, 1681—1691.

Li, X.L., Aou, S., Oomura, Y., Hori, N., Fukunaga, K., Hori, T., 2002.Impairment of long-term potentiation and spatial memory inleptin receptor-deficient rodents. Neuroscience 113, 607—615.

Lindenmayer, J.P., Khan, A., Kaushik, S., Thanju, A., Praveen, R.,Hoffman, L., Cherath, L., Valdez, G., Wance, D., 2012. Relation-ship between metabolic syndrome and cognition in patients withschizophrenia. Schizophr. Res. 142, 171—176.

McGurk, S.R., Mueser, K.T., Walling, D., Harvey, P.D., Meltzer, H.Y.,2004. Cognitive functioning predicts outpatient service utiliza-tion in schizophrenia. Ment. Health Serv. Res. 6, 185—188.

Meyer, J.M., Nasrallah, H.A., McEvoy, J.P., Goff, D.C., Davis, S.M.,Chakos, M., Patel, J.K., Keefe, R.S., Stroup, T.S., Lieberman,J.A., 2005. The Clinical Antipsychotic Trials of InterventionEffectiveness (CATIE) Schizophrenia Trial: clinical comparison ofsubgroups with and without the metabolic syndrome. Schizophr.Res. 80, 9—18.

Morbelli, S., Drzezga, A., Perneczky, R., Frisoni, G.B., Caroli, A., vanBerckel, B.N., Ossenkoppele, R., Guedj, E., Didic, M., Brugnolo,A., Sambuceti, G., Pagani, M., Salmon, E., Nobili, F., 2012. Res-ting metabolic connectivity in prodromal Alzheimer’s disease. AEuropean Alzheimer Disease Consortium (EADC) project. Neuro-biol. Aging 33, 2533—2550.

Nakamura, M., Nestor, P.G., Levitt, J.J., Cohen, A.S., Kawashima,T., Shenton, M.E., McCarley, R.W., 2008. Orbitofrontal vol-ume deficit in schizophrenia and thought disorder. Brain 131,180—195.

Z

Z

ive impairment in schizophrenia 105

asrallah, H.A., 2010. Linkage of cognitive impairments withmetabolic disorders in schizophrenia. Am. J. Psychiatry 167,1155—1157.

uechterlein, K.H., Barch, D.M., Gold, J.M., Goldberg, T.E., Green,M.F., Heaton, R.K., 2004. Identification of separable cognitivefactors in schizophrenia. Schizophr. Res. 72, 29—39.

uechterlein, K.H., Subotnik, K.L., Ventura, J., Green, M.F.,Gretchen-Doorly, D., Asarnow, R.F., 2012. The puzzle ofschizophrenia: tracking the core role of cognitive deficits. Dev.Psychopathol. 24, 529—536.

odefroy, O., GREFEX éditeur, 2008. Fonctions executives etpathologies neurologiques et psychiatriques Solal (Ed.), Eval-uation en pratique clinique, Collection neuropsychologie,Marseille, France.

omura, Y., Hori, N., Shiraishi, T., Fukunaga, K., Takeda, H., Tsuji,M., Matsumiya, T., Ishibashi, M., Aou, S., Li, X.L., Kohno, D.,Uramura, K., Sougawa, H., Yada, T., Wayner, M.J., Sasaki, K.,2006. Leptin facilitates learning and memory performance andenhances hippocampal CA1 long-term potentiation and CaMK IIphosphorylation in rats. Peptides 27, 2738—2749.

auly, K., Seiferth, N.Y., Kellermann, T., Ruhrmann, S., Daumann,B., Backes, V., Klosterkotter, J., Shah, N.J., Schneider, F.,Kircher, T.T., Habel, U., 2010. The interaction of working mem-ory and emotion in persons clinically at risk for psychosis: anfMRI pilot study. Schizophr. Res. 120, 167—176.

odriguez-Sanchez, J.M., Crespo-Facorro, B., Perez-Iglesias, R.,Gonzalez-Blanch, C., Alvarez-Jimenez, M., Llorca, J., Vazquez-Barquero, J.L., 2005. Prefrontal cognitive functions in stabilizedfirst-episode patients with schizophrenia spectrum disorders: adissociation between dorsolateral and orbitofrontal functioning.Schizophr. Res. 77, 279—288.

olls, E.T., 2007. Understanding the mechanisms of food intake andobesity. Obes. Rev. 8 (Suppl. 1), 67—72.

choenbaum, G., Roesch, M.R., Stalnaker, T.A., 2006. Orbitofrontalcortex, decision-making and drug addiction. Trends Neurosci. 29,116—124.

zatkowska, I., Szymanska, O., Bojarski, P., Grabowska, A., 2007.Cognitive inhibition in patients with medial orbitofrontal dam-age. Exp. Brain Res. 181, 109—115.

zoke, A., Trandafir, A., Dupont, M.E., Meary, A., Schurhoff,F., Leboyer, M., 2008. Longitudinal studies of cognitionin schizophrenia: meta-analysis. Br. J. Psychiatry 192,248—257.

chanturia, K., Lock, J., 2011. Cognitive remediation therapy foreating disorders: development, refinement and future direc-tions. Curr. Top. Behav. Neurosci. 6, 269—287.

an den Berg, E., Kloppenborg, R.P., Kessels, R.P., Kappelle, L.J.,Biessels, G.J., 2009. Type 2 diabetes mellitus, hypertension, dys-lipidemia and obesity: a systematic comparison of their impacton cognition. Biochim. Biophys. Acta 1792, 470—481.

iviani, R., Messina, I., Walter, M., 2011. Resting state functionalconnectivity in perfusion imaging: correlation maps with BOLDconnectivity and resting state perfusion. PLoS ONE 6, e27050.

illeumier, K.C., Taylor, D.V., Amen, D.G., 2011. Elevated BMI isassociated with decreased blood flow in the prefrontal cortexusing SPECT imaging in healthy adults. Obesity (Silver Spring)19, 1095—1097.

affe, K., Kanaya, A., Lindquist, K., Simonsick, E.M., Harris, T.,Shorr, R.I., Tylavsky, F.A., Newman, A.B., 2004. The metabolicsyndrome, inflammation, and risk of cognitive decline. JAMA292, 2237—2242.

au, P.L., Castro, M.G., Tagani, A., Tsui, W.H., Convit, A., 2012.Obesity and metabolic syndrome and functional and structuralbrain impairments in adolescence. Pediatrics 130, e856—e864.

akzanis, K.K., Mraz, R., Graham, S.J., 2005. An fMRI study of theTrail Making Test. Neuropsychologia 43, 1878—1886.

igmond, A.S., Snaith, R.P., 1983. The hospital anxiety and depres-sion scale. Acta Psychiatr. Scand. 67, 361—370.