• Embed Doc
  • Copy Link
  • Readcast
  • Collections
  • CommentGo Back
Download
 
Deactivation of the Limbic System During AcutePsychosocial Stress: Evidence from Positron EmissionTomography and Functional Magnetic ResonanceImagingStudies
Jens C. Pruessner, Katarina Dedovic, Najmeh Khalili-Mahani, Veronika Engert, Marita Pruessner,Claudia Buss, Robert Renwick, Alain Dagher, Michael J. Meaney, and Sonia Lupien
Background:
Stress-inducedmetabolicchangescanhavedetrimentalhealtheffects.Newlydevelopedparadigmstoinvestigatestressinneuroimaging environments allow the assessment of brain activation changes in association with the perception of and the metabolicresponse to stress.
Methods:
We exposed human subjects to a psychosocial stressor in one positron emission tomography (
n
10) and one functionalmagnetic resonance imaging (fMRI;
n
40) experiment.
Results:
Weobservedaprofounddeactivationoflimbicsystemcomponentsincludinghippocampus,hypothalamus,medio-orbitofrontalcortexandanteriorcingulatecortexinsubjectswhoreactedtothestressorwithasignificantincreaseoftheendocrinestressmarkercortisol.Further,inthefMRIstudy,thedegreeofdeactivationinthehippocampuswascorrelatedwiththereleaseofcortisolinresponsetothestresstask.
Conclusions:
Theobserveddeactivationoflimbicsystemstructuressuggestselevatedactivationatrestandduringnonstressfulsituations.A model is proposed where the observed reduction in limbic system activity is essential for the initiation of the stress response.
KeyWords:
Cortisol, default brain function, functional magneticresonanceimaging,hippocampus,positronemissiontomography,stress
 A
cuteandchronicstressorslikeemotionalorphysicaltrauma,orprolongedpsychosocialstressareofteninves-tigatedfortheirpotentialimplicationinhealthanddis-ease.Itisgenerallyacceptedthatstress-inducedactivityofthehypothalamic-pituitary-adrenal(HPA)axisandthereleaseof glucocorticoids(primarilycorticosteroneinanimalsandcortisolinhumans)fromtheadrenalcortexaccountatleastpartiallyfordetrimental health effects of chronic stress (1,2). While the effects ofstressonmemoryandcognitionarefrequentlyinvestigated(3–5), neural activation patterns associated with the initiation of thestressresponseremainpoorlyunderstood.ExperimentalinductionofstressandactivationoftheHPAaxishavebeensuccessfullyachievedbypreviouslaboratorytaskscontainingfree speech and mental arithmetic (6 – 8). Converting these tasks toaneuroimagingenvironmentischallengingandmustconsiderthesupineposition,headimmobilization,andisolationinthescanner room. Recently, Wang
et al.
(9)used a mental arithmetic taskinafunctionalfMRIenvironment,whichwasidenticaltothemental arithmetic part of the original Trier Social Stress Test (6).  Althoughtheheadmovementinducedbytheverbalresponsesfromthesubjectduringthescanningprocedurecouldbecon-sideredproblematic,theyreportactivationsoftheventralrightprefrontalcortexandtheleftinsula/putamen,arguingthatstresscausesactivationsofpathwaysassociatedwithnegativeemo-tions,andthattherightprefrontalcortexplaysakeyroleinthe central stress response. A study by Tillfors
et al.
(10)avoidedtheproblemofmovementartifactsbyscanningsubjectsemploying
15
O-waterinapositronemissiontomog-raphystudyjustprior(anticipationstress)versusjustafter(controlcondition)publicspeakingandfoundincreasedactivationinthehippocampus(HC)/amygdalaareaduringanticipation(althoughitcouldalsobearguedthatthiswasadecreaseinactivationduringthepoststressperiod).Finally, wehaverecentlydescribedaparadigmwherealaboratory stressorbasedonmentalarithmeticwithouttheneedfor verbal responses (7) has been converted to work in functionalneuroimaging environments (11). This task has been success- fully employed in combination with 11c-raclopride PET (12).  Animalstudiesprovideevidencethatthelimbicsystemplaysanimportantroleintheregulationofstressresponses.Com-monly,suchstudiesrevealthat30to60minaftertheonsetofastressor,molecularmarkersofneuralactivity(e.g.cFos,nerve-growth-factor-inducedproteinA)inmultiplelimbicstructuresareincreased,suggestingastimulatingeffectofstressonneuralactivity (13,14). Investigating stressors like social defeat in ani-mals (15), the HC has emerged as a limbic system structure that canhaveaninhibitoryeffectonHPAaxisactivity.ThisinhibitionisusuallydiscussedinthecontextofglucocorticoidfeedbackarrivinginthecentralnervoussystemintheaftermathofstresstosuppressfurtherHPAaxisactivity(‘negativefeedbackloop’),althoughitisacknowledgedthattheHCcanexertatonicinhibitory influence on HPA axis regulation (16,17). Althoughstudiesusingmolecularmarkerspresentacleardescriptionoftheconsequencesofstressonneuralsystems,they lackthetemporalresolutiontoshowwhichbrainactivationchangesareassociatedwiththeinitiationofthehormonalstressresponse.Here,functionalimagingstudiespossessanadvan-tage,sincetheyhavetheabilitytoassessbrainactivationswhilethestressfuleventisunfolding.
From the Center for Studies on Human Stress, Douglas Hospital ResearchCenter and Montreal Neurological Institute, Departments of Psychiatry(JCP, VE, MP, CB, MJM, SL), and Neurology and Neurosurgery (AD, KD,NK-M, RR), McGill University, Montreal, Quebec, Canada.Address reprint requests to Jens C. Pruessner, Ph.D., FBC 2115.2, DouglasMental Health University Institute, 6875 LaSalle Boulevard, Montreal,QC, H4H 1R3, Canada; E-mail: jens.pruessner@mcgill.ca.Received October 5, 2006; revised March 29, 2007; accepted April 18, 2007.
BIOLPSYCHIATRY2008;63:234–2400006-3223/08/$34.00doi:10.1016/j.biopsych.2007.04.041©2008SocietyofBiologicalPsychiatry 
 
To investigate neuronal activation changes during the expo-sure to psychosocial stress in humans, we conducted twofunctional brain imaging studies, using
15
OH
2
0 positron emissiontomography (PET), as well as functional magnetic resonanceimaging (fMRI). Considering previous findings from animal andhuman studies about the regulation of the stress response (16),  we hypothesized that stress exposure would cause significantbrain activation changes in limbic system structures.
Methods and Materials
Subjects
 A total of 50 healthy young subjects were recruited from thelocal population of students at McGill University by posting flyersand through advertisements in local newspapers. Informed con-sent was obtained before participation in accordance with therequirements of the Research Ethics Board of the MontrealNeurological Institute. Ten healthy men (mean age: 25.4
2.5 years) participated in the PET study, and 40 healthy subjects (20men, mean age 23.35
3.17 years; 20 women, mean age 22.75
3.11 years) participated in the fMRI study.
Behavioral Task 
Psychological stress was induced using the Montreal ImagingStress Task (MIST), where subjects are exposed to challengingmental arithmetic presented on a computer screen, to which they have to respond by choosing a one-digit number from a rotary dial using a three-button mouse. In the experimental condition,the difficulty of the tasks adapts to the user performance to yielda 45–50% performance range. A mock performance indicatorsuggests poor performance on the part of the subject in compar-ison to the average user. This is combined with negative feed-back by both the program (after each task) and the investigator(between runs). In the control condition, the mental arithmetic ispresented without negative feedback components, and in therest condition, the user interface is displayed without mentalarithmetic tasks being shown. The task and its effect on physio-logical parameters are explained in detail elsewhere (11).
Endocrine Measurements and Analysis
Cortisol was assessed from saliva samples, which were col-lected during the scanning in the PET study and the fMRI study.In addition, in the fMRI study, saliva for cortisol analysis was alsosampled on a separate day at least two days apart from thetesting, at the time of awakening, and 15, 30, 45, and 60 min afterawakening.In the PET experiment, saliva samples were assessed beforeand after the rest, control arithmetic without stress, and experi-mental arithmetic with stress conditions, for a total of sixsamples. Time distance between measurements was approxi-mately 30 min. In the fMRI study, nine saliva samples wereacquired, starting 50 min before onset of the MIST until 50 minafter the MIST. Cortisol was measured from saliva using atime-resolved fluorescence immunoassay (18). Intra- and inter- assay variability have been shown to be routinely less than 10%and 12%, respectively.
Functional Imaging Data Acquisition and Analysis
PET scans were performed with a CTI/Siemens (HR 
scanner; Siemens, Erlangen, Germany) operated in 3D acquisitionmode, yielding images with an approximate resolution of 4.6mmfull width at half-maximum. The H
215
O water bolus technique(19)was used to measure regional cerebral blood flow (rCBF). Each subject underwent nine one-min PET scans, three scans ineach experimental condition (resting, control arithmetic task,stress arithmetic task). The order of experimental, control, andrest scans was counterbalanced between subjects. If the stresscondition was not the last condition to be performed, it wasfollowed by an additional 20 min rest in order to avoid elevatedcortisol levels having a carry-over effect on subsequent testing.In the fMRI study, subjects were scanned on a 1.5 T SiemensMagnetom Vision Scanner, employing a block design with tworuns. Each run consisted of rest, control, and experimentalblocks, completed twice over the course of the nine min. Whilethe beginning condition (rest, stress or control) was counterbal-anced between subjects, the sequence of the three conditions was kept constant between subjects: control was following restand preceding stress.Statistical analysis of all functional data was performed withthe software packages fmristat and multistat (20,21). In the PET study,task-specificregionalcerebralbloodflow(rCBF)wasdeterminedbycontrastingrCBFdifferencesbetweeneitherexperimentalandcontrol,experimentalandrest,andcontrolandrestconditions.Usinga14mmFWHMblurringkernel,peakactivationswereconsideredsignificantiftheyexceededathresh-oldof 
t 
4.5(
 p 
.05,corrected).InthefMRIstudy,a6mmblurringkernelwasemployedandyieldeda
t 
-statthresholdof 
t 
3.66asborderofstatisticalsignificance(
 p 
.05,corrected)(21). (See Supplement 1 for more details on inclusion andexclusioncriteria,dataacquisition,andstatisticalanalysis.)
Results
CortisolStressResponsesandBaselineMeasures
ToestablishwhethertheexperimentalconditioncausedasignificantincreaseinsalivacortisollevelsinthePETstudy,weperformedatwofactormixeddesign(condition
time) ANOVA.Theresultswerehighlysignificant(
F 
12.46,
 p 
.001;Figure 1 A), and a Newman Keuls posthoc test confirmed that this wasduetochangesinthepre-postexperimentalcondition,
Figure1.
Cortisol levels in the different experiments and subgroups.
(A)
Post–pre cortisol levels of the three testing conditions rest, control, andexperimentalinthepositronemissiontomography(PET)study(
n
10).
(B)
Cortisollevels(wholegroup
n
40)duringthefMRIexperiment.
(C)
Cortisollevels in the responder (
n
21) and nonresponder (
n
19) subgroupsduring the fMRI experiment.
(D)
Cortisol levels during the first hour afterawakening on a separate day from the fMRI experiment. Error bars shownare SEM. fMRI, functional magnetic resonance imaging.
 J.C.Pruessner
etal.
BIOLPSYCHIATRY2008;63:234–240
235
www.sobp.org/journal
 
suggesting HPA axis activation. No significant cortisol increasescould be observed in the control, or the rest condition.Cortisol levels also increased significantly after stress in thefMRI study, with a significant main effect (Figure 1B, wholegroup;
F 
9.4,
 p 
.001; confirmed with Newman Keulsposthoc). Due to significant heterogeneity in the individualcortisol responses, we split the whole group into subjects withcortisol increase or decrease over the course of the experiment.This resulted in 21 responders, and 19 nonresponders (Figure1C). A two factor (group by time) mixed design ANOVA indi-cated a significant group by time interaction (
F 
6.71,
 p 
.02), with a Newman Keuls posthoc test confirming that responderssignificantly increased their cortisol levels after stress. This resultallowed us to investigate brain activation changes dependingupon the endocrine stress response of the task. There was nodifference in number of correct responses between the groups(
t 
1,
 p 
.20; responder group: 48.5%; nonresponder group:45%). The ratio of men to women was not significantly differentbetween the responders (10:11), and the nonresponders (10:9;c
2
1.16,
 p 
.20). Also, cortisol levels after awakening (22)from a separate day were available in this study, thus we testedthe difference between responders and nonresponders for thismeasure. A two factor (group by time) mixed design ANOVArevealed no significant group effect (
F 
1.54,
 p 
.20), and nosignificant (group by time) interaction effect (
F 
1,
 p 
.20;Figure 1D). Finally, we compared self-esteem scores betweenresponders and nonresponders. There was a marginal difference(
t 
1.8,
 p 
.08) with responders having lower self-esteemscores. Also, the right HC volume was significantly lower inresponders compared to nonresponders (
t 
2.6;
 p 
.05), witha trend difference in the left hemisphere (
t 
1.9,
 p 
.09).Hippocampal volume was also positively related to levels of self-esteem (
r 
.39,
 p 
.05).
Brain Activation as a Result of Mental Arithmetic
To identify brain activations in response to stressful andnonstressful mental arithmetic, we subtracted the rest from thecontrol and experimental conditions in all studies. This contrastallowed us to visualize the effects of performing stressful andnonstressful mental arithmetic on brain activation. In the (control
rest) contrast, in both studies, and independent of cortisolresponders or nonresponders in the fMRI study, we observedactivations in the occipital cortex in the area of the visualassociation cortex (Brodman Area 18 and 19), and the angulargyrus. We also observed signs of activation in the motor cortex,and the cerebellum. In the experimental
rest condition in bothstudies, we observed additional activations in the motor andpremotor areas. All these activations are in line with performinga visual task, using your arms and hands, and solving mentalarithmetic (23). No significant activations could be observed inthese contrasts in limbic system structures like HC or amygdala,hypothalamus, or anterior cingulate.
Brain Activation as a Result of Stress
One of the main goals of the two studies was to investigate theeffects of the psychosocial stress components on brain activations, while controlling for activation changes induced by the mentalarithmetic. For this purpose, we subtracted the control from theexperimental condition, thus looking at the main difference be-tween stressful mental arithmetic and nonstressful mental arith-metic. In the PET study, we failed to observe significant activationsin this contrast. In the fMRI study, we observed activations in the leftpremotor area (
t 
6.67,
 p 
.05), the medial left prefrontal cortex(
t 
4.35,
 p 
.05)andbilaterallyintheareaofthecingulum/whitematter(
t 
5.5,
 p 
.05).Inaddition,wecouldobservesignificantactivation in the area of the occipital cortex (
t 
4.38;
 p 
.05; Figure2). No other activations could be observed as a consequence of perceivingstimuliconsideredtorepresentasocialevaluativethreat.InvestigatingthesubgroupsofrespondersandnonrespondersseparatelyforbrainactivationsinthepresenceofstressinthefMRIstudydidnotrevealdifferentialactivations.
ObservedDeactivationsinthePresenceofPsychosocialStress
Finally,weinvestigatedtheactivationmapsforsignsofnegativepeaksandclusters.Itwasatthislevelofanalysisthatwefoundthemostprofoundeffectsofstressonbrainactivationchanges.Ingeneral,respondingwithacortisolincreasetothestressfultaskwasassociatedwithadecreaseofactivityinlimbicsystemcomponents, whencontrastedagainstthecontrolcondition.InthePETstudy,weobservedreducedcerebralbloodflowafterstressinanetworkof structureslinkedtothelimbicsystem,i.e.HCandamygdala(
t 
8.2;
 p 
.05),ventralstriatum(
t 
10.6,
 p 
.05),anddorsal(
t 
9,
 p 
.05)andventral(
t 
15,
 p 
.05)medialprefrontalcortex including the anterior cingulate cortex (Figure 3 A). We alsoobservedsignificantdeactivationsbilaterallyinthesuperiortempo-ralgyrus(
t 
7.9,
 p 
.05),andintherightsuperiorinsula(
t 
7.7,
 p 
.05).InthefMRIstudy,thispicturewasmoredifferenti-ated.Wholegroupanalysisforsignsofdeactivationsgenerally replicatedthefindingsofthePETstudy,withlimbicsystemstruc-turesshowingsignsofreducedbrainactivityunderstress,including ventralstriatum(
t 
4.8,
 p 
.05),hypothalamus(
t 
5.3,
 p 
.05),amygdala(
t 
4.5,
 p 
.05),andHC(
t 
5.7,
 p 
.05).However,asystematicdifferenceappearedbetweentheactivationmapsoftherespondersandnonrespondersinthisstudy.Inthenonrespondergroup,whilethedeactivationwassignificantintheareaofthemedio-orbitofrontalcortex(
t 
5.9,
 p 
.05)and
Figure2.
Significant activations in the (experimental minus control) con-trastinthefMRIstudy.x,y,z
sagittal,coronalandhorizontalviewinworldcoordinates. L, left; R, right; PMA, premotor area; OCC, occipital cortex;MPFC,medialprefrontralcortex;Cing,cingulum;fMRI,functionalmagneticresonance imaging.
236
BIOLPSYCHIATRY2008;63:234–240
 J.C.Pruessner
etal.
www.sobp.org/journal
of 00

Leave a Comment

You must be to leave a comment.
Submit
Characters: ...
You must be to leave a comment.
Submit
Characters: ...