Does CognitiveBehavioral TherapyChange the Brain? ASystematic Review ofNeuroimaging in AnxietyDisordersPatricia Ribeiro Porto, M.S.Leticia Oliveira, Ph.D.Jair Mari, Ph.D.Eliane Volchan, Ph.D.Ivan Figueira, Ph.D.Paula Ventura, Ph.D.

This systematic review aims to investigate neuro-biological changes related to cognitive-behavioraltherapy (CBT) in anxiety disorders detectedthrough neuroimaging techniques and to identifypredictors of response to treatment. Cognitive-behavioral therapy modified the neural circuitsinvolved in the regulation of negative emotionsand fear extinction in judged treatment respond-ers. The only study on predictors of response totreatment was regarding obsessive-compulsivedisorder and showed higher pretreatment regionalmetabolic activity in the left orbitofrontal cortexassociated with a better response to behavioraltherapy. Despite methodological limitations, neu-roimaging studies revealed that CBT was able tochange dysfunctions of the nervous system.

(The Journal of Neuropsychiatry and ClinicalNeurosciences 2009; 21:114–125)

Neuroscience has developed several methods to an-alyze the cognitive function and potentiate the

understanding of the mental functioning of healthy andmentally disordered individuals. The recent advancesin neuroimaging techniques have helped to increase theunderstanding of the neuronal correlates of mental dis-orders.

Psychological interventions can promote changes inthe thoughts, feelings, and behaviors of patients. Canwe then say that the psychological treatment promotesbrain changes? Unfortunately, the biological mecha-nisms related to psychotherapy are little known. On theother hand, the arrival of neuroimaging techniquesmake it possible to investigate the neurobiological con-sequences of psychological treatment. Such investiga-

Received August 11, 2008; revised October 11, 2008; accepted October17, 2008. Ms. Porto is affiliated with the Universidade Federal do Riode Janeiro in Brazil; Dr. Oliveira is affiliated with the UniversidadeFederal Fluminense; Dr. Mari is affiliated with the Department ofPsychiatry and Medical Psychology of the Sao Paulo Medical School,Universidade Federal de Sao Paulo (EPM-UNIFESP), in Brazil; Dr.Volchan is affiliated with the Institute of Biophysics Carlos ChagasFilho, Universidade Federal do Rio de Janeiro (IBCCF-UFRJ), in Bra-zil; Dr. Figueira is affiliated with the Institute of Psychiatry, Univer-sidade Federal do Rio de Janeiro (IPUB-FRJ), in Brazil; Dr. Ventura isaffiliated with the Institute of Psychology, Universidade Federal doRio de Janeiro. Address correspondence to Patrıcia Ribeiro Porto,Federal University of Rio de Janeiro, Rio de Janeiro, Brazil;portopat@gmail.com (e-mail).

Copyright © 2009 American Psychiatric Publishing, Inc.

114114 http://neuro.psychiatryonline.org J Neuropsychiatry Clin Neurosci 21:2, Spring 2009

SPECIAL ARTICLES

tion is highly important, as a better understanding ofthe brain mechanisms underlying therapy can promoteimprovements in the therapeutic interventions as wellas increase our knowledge on the formation and main-tenance of symptoms.1,2 Elucidating the neural corre-lates associated with symptom reductions has been thesubject of research aimed at the identification of thebiological mechanisms of psychotherapy.1–5

Raedt6 points out in his work that behavioral therapyoffers an interesting perspective for integration with theneuroscience field, since any intervention is connectedto a support of experimental and empirical research.Cognitive behavioral therapy (CBT) proposes to treatvarious mental disorders. The literature has reportedthat CBT has treatment models with high efficacyrates.7,8

One of the basic assumptions of CBT is that feelingsand behaviors are largely influenced by the way thesituations are interpreted. It is believed that individualsrespond to the cognitive representations of the events,instead of responding to the events themselves. Conse-quently, they can process information in a way thatdoes not match their reality, characterizing the cogni-tive distortions. Thus, the ways in which the facts areconstrued play an important role in the formation andmaintenance of psychiatric disorders.7

This systematic review aimed to investigate CBT-re-lated neurobiological changes in anxiety disorders, de-tected through neuroimaging techniques, and to iden-tify predictors of response to treatment. By gatheringresults of studies on the new findings in the area, wecan eventually contribute with knowledge for the in-crease of therapy efficacy, either through the improve-ment of new therapeutic strategies or through the po-tentialization with drugs that facilitate the basicmechanisms of action. The understanding of the under-lying neurobiological mechanism can eventually aid inthe choice of the most indicated treatment for a partic-ular patient.

METHODS

An electronic search was carried out on January 26,2007, in the following databases: PubMed, PsychInfo,and Web of Science. The search strategies used areavailable upon request. Searches were also performedfrom the references of the systematic review articles onneurobiological changes and psychotherapy.

We included studies that evaluated neurobiologicalchanges due to CBT through neuroimaging techniquesand studies that involved adult patients with anxietydisorders. We excluded studies that used concomitanttreatment besides CBT in the same sample of patients.Moreover, articles that dealt with subclinical anxietywere excluded. The outcome measures studied werethe neurobiological changes resulting from CBT, as-sessed through neuroimaging techniques. Thirteenstudies were found: five on patients with obsessive-compulsive disorder (OCD), three on posttraumaticstress disorder (PTSD), two on specific phobia, two onpanic disorder, and one on social phobia. No studieswere found for generalized anxiety disorder. Threestudies were excluded because the patients weretreated concomitantly with psychotherapy and medica-tion (one from the OCD group and two from the post-traumatic stress disorder group). Thus, 10 articles metthe selection criteria of this systematic review.

RESULTS AND DISCUSSION

Changes Related To CBT in Anxiety DisordersNine of the 10 studies identified provided data as to thefirst objective of this work—studies assessing the neu-robiological changes of CBT detected through neuroim-aging techniques (Table 1).

Studies with Spider Phobia Cognitive behavioral ther-apy has been shown to be effective to reduce symptomsof specific phobia.9 The neuroimaging studies in CBTwere conducted by Paquette et al.10 and Straube et al.11

Both studies used the paradigm of symptom provoca-tion and functional magnetic resonance (fMRI).

In their 2003 study, Paquette et al.10 evaluated thesubjects with fMRI before and after CBT treatment. Theparticipants of the study comprised of 12 women withspider phobia and a comparison group of 13 womenwithout history of neurological or psychiatric diseaseand absence of anxiety response to spider exposure.The treatment with CBT consisted of the gradual expo-sure to spiders and cognitive restructuring. All subjectsresponded successfully to the therapy.

The neuroimaging findings showed that before thetreatment, phobic patients presented significantly acti-vated dorsolateral prefrontal cortex and parahippocam-pal gyrus. The findings after CBT treatment showedthat there was no significant activation of these struc-

PORTO et al.

115J Neuropsychiatry Clin Neurosci 21:2, Spring 2009 http://neuro.psychiatryonline.org 115

tures in the phobic subjects. The absence of activation ofthe dorsolateral prefrontal cortex and parahippocampalgyrus after CBT demonstrated for the authors strongsupport to the hypothesis that CBT reduces phobicavoidance through the extinction of the contextual fearlearned in the hippocampal/parahippocampal regionand reduces the dysfunctional and catastrophicthoughts in the prefrontal cortex. Therefore, the processof extinction would be able to prevent reactivation oftraumatic memories, allowing the phobic subject tomodify his or her perception of stimuli which evokedfear before the treatment. With the modification of stim-ulus perception, it ceases to be a threat, and this cogni-tive restructuring could inhibit the activation of brainregions previously associated with a phobic reaction.10

Straube et al.11 performed another study of symptomprovocation using fMRI. They also investigated theneurobiological effect of a successful therapeutic inter-vention with CBT. The study included healthy as well

as phobic individuals on a waiting list. The phobic sub-jects were scanned before and after CBT treatment.

Twenty-eight women with spider phobia and 14healthy women took part in the study. The subjectswith spider phobia were randomly assigned to the ther-apy group and comparison group on a waiting list. Thegroups did not differ in phobia severity, age, or educa-tional level. The CBT treatment consisted of gradualexposure to spiders and cognitive restructuring. All ofthe subjects in the therapy group responded success-fully to the treatment.

The neuroimaging findings before the treatmentshowed that only the phobic subjects displayed activa-tion of the insula and anterior cingulate cortex, whilethe activation of the amygdala was restricted to thehealthy control subjects. No activation of other areaswas found among the phobic groups.

The neuroimaging findings after the treatmentshowed significant differences between the waiting list

TABLE 1. Neurobiological Changes Associated with CBT

StudyMental

DisorderNeuroimaging

Technique Neuroimaging Findings

Paquette et al.(2003)

Spider phobia fMRI After CBT there was significant activation of the dorsolateral prefrontal cortexand parahippocampal gyrus regions.

Straube et al.(2006)

Spider phobia fMRI After treatment the CBT group displayed absence of activation of the anteriorventral insula and failed to show any difference from the healthy controlsubjects.

Furmark et al.(2002)

Social phobia PET After treatment with CBT and citalopram, there was reduction of activities atthe temporal lobe regions, mainly at the right hemisphere. Decreasedactivity at the right amygdala, hippocampus, rhinal activity, andperiamygdaloid.

Farrow et al.(2005)

PTSD fMRI In the social cognition of empathy there was significant activation of the leftposterior and anterior medial temporal gyrus and posterior cingulate gyrus.As for the forgiveness-related cognition, there was activation of the posteriorcingulate, medial frontal gyrus, posterior cingulate activation, medial frontalgyrus, and left posterior medial temporal gyrus.

Baxter et al.(1992)

OCD PET After treatment the CBT and medication groups showed decreased activationof the head of the right caudate nucleus.

Schwartz et al.(1996)

OCD PET After treatment the patients presented significant decrease of activation of theright caudate nucleus. When the subjects of this study were combined withthose of the previous study (Baxter et al., 1992), they showed as wellreduced activation of the left caudate nucleus.

Brody et al.(1998)

OCD PET The higher metabolism of the left frontal orbital cortex before treatment wasassociated with a better response to BT

Nakao et al.(2005)

OCD fMRI After treatment there was a decrease in the activation of the frontal orbitalcortex.

Prasko et al.(2004)

Panic PET The increase of activity in the left hemisphere was mainly in the prefrontal,temporoparietal and occipital regions. In the right hemisphere, in theposterior cingulate. The decrease was predominant at the left hemisphere inthe frontal region, and at the right hemisphere in the frontal, temporal, andparietal region.

Sakai et al.(2006)

Panic PET After treatment there was decrease in the metabolism at the righthippocampus, left ventral anterior cingulate cortex, uvula, and pyramid ofthe left cerebellum and pons. The increased metabolism was found in thebilateral medial prefrontal region.

fMRI�functional magnetic resonance imaging; PET�positron emission tomography; CBT�cognitive behavior therapy; PTSD�posttraumaticstress disorder; OCD�obsessive-compulsive disorder

NEUROIMAGING IN ANXIETY DISORDERS

116116 http://neuro.psychiatryonline.org J Neuropsychiatry Clin Neurosci 21:2, Spring 2009

group and the therapy group. The therapy groupshowed absence of activation during symptom elicita-tion in the anterior cingulate cortex and only a smallarea of activation of the anterior ventral insula, whilethe waiting group displayed marked responses bilater-ally in the insula and anterior cingulate cortex. Thetherapy group did not show any significant differencefrom the healthy comparison subjects, while the waitinglist patients showed more activation of the right insulaand anterior cingulate cortex.

Straube et al.11 showed that the processing of phobicthreat is associated with increased activation of the in-sula and anterior cingulate cortex in subjects with spe-cific phobia. It is important to highlight that successfulCBT led to reduction of hyperactivity in these regions.Consequently, the subjects in the therapy group, ascompared to those of the waiting list, showed attenua-tion of phobic symptoms and reduction of the activityof the insula and anterior cingulate cortex during thesecond neuroimaging scan.

The neuroanatomic functioning associated with thesymptoms of spider phobia is not clear. While Paquetteet al.10 pointed to the participation of the dorsolateralprefrontal cortex and parahippocampal gyrus in theprocessing of phobic fear, Straube et al.11 do not sup-port this hypothesis. This discrepancy between the find-ings merits further investigation. Both studies did notfind evidence of the involvement of the amygdala in theprocessing of fear stimulus of the subjects with phobia,but in the research by Straube et al.11 the healthy com-parison subjects presented activation of the amygdala.Although Paquette and Straube obtained distinct re-sults for the brain areas involved before the treatment,the CBT in both studies proved to be capable of reduc-ing the symptoms and modifying the possible dysfunc-tional neuronal circuits after the treatment.

Studies with Social Phobia The neurofunctional changesassociated with the reduction of social anxiety in pa-tients submitted to CBT treatment were investigated byFurmark et al.12 through PET. The only existing studyon social phobia used the paradigm of emotional acti-vation. The research also had the aim of exploringwhether the brain change was associated with the long-term results of the treatment.

Eighteen individuals who fulfilled the criteria for di-agnosis of social phobia according to DSM-IV partici-pated in the study. The participants were sorted accord-ing to symptom severity, age, and gender, and then

randomized for treatment with citalopram, CBT, andwaiting list; each group comprised six participants. TheCBT group used techniques of exposure, cognitive re-structuring, and homework.

The gravity of the symptoms of social phobia wassignificantly reduced after 9 weeks, both in the CBT andcitalopram groups, while the waiting-list group did notshow any improvement. There was no statistical differ-ence between the CBT and citalopram groups concern-ing treatment outcome.

The therapeutic effect on regional blood flow wasassessed by contrasting the task of speaking in publicbefore and after the treatment in each group. Thus, theimprovement of social anxiety was associated with sig-nificant reduction of the response of regional bloodflow bilaterally in the amygdala, hippocampus, andmedian and anterior temporal cortex, including the en-torhinal, perirhinal, parahippocampal, and periamyg-daloid areas both for the CBT and citalopram groups.There was no significant alteration in the regional bloodflow in the waiting-list group. The CBT and citalopramgroups differed only in the perfusion of the right thal-amus, which presented a greater increase in the citalo-pram group.

The study showed that the degree of reduction of thelimbic response with the treatment is associated withthe long-term clinical outcome. The decrease in the re-sponse of the brain blood flow in the amygdala, peri-aqueductal gray matter, and left thalamus can indicatewhich patients show greater improvement in an inter-val of 1 year. Thus, favorable results at 1-year follow-upwere associated with greater attenuation of the subcor-tical blood flow response while speaking in public.

In their discussion, Furmark et al.12 reported that theamygdala and the hippocampus are structures relatedto the conditioning of aversive stimuli in individualswith social phobia. These structures, together with therhinal, parahippocampal, and periamygdaloid wouldform an alarm system that can be activated by threat-ening stimuli. The reduction of the activity in the amyg-dala-hippocampal region and adjacent cortex can beimportant mechanisms through which both pharmaco-logical and psychotherapeutic treatments could exert ananxiolytic effect.

The study concluded that the neural sites of activa-tion for the treatment with citalopram and CBT in socialanxiety converge to the amygdala, hippocampus, andadjacent cortical areas, possible representing a commonway in the successful treatment of anxiety. The attenu-

PORTO et al.

117J Neuropsychiatry Clin Neurosci 21:2, Spring 2009 http://neuro.psychiatryonline.org 117

ation of the activity in the amygdalar and limbic regionwith the treatment was associated with a favorablelong-term result and can be a prerequisite for clinicalimprovement.

Studies with Posttraumatic Stress Disorder No neuroim-aging studies were found with a focus on CBT in iso-lation evaluating the correlation between improvementof posttraumatic stress disorder (PTSD) symptoms andthe brain areas involved in this disorder. Nevertheless,Farrow et al.13 investigated the effects of PTSD, unre-lated to combat, on the physiology of social cognition.According to them, PTSD symptoms could affect theprocessing of social and emotional cognition. Thesechanges could attenuate brain activation in the areasrelated to social cognition, specifically the ability to for-give and empathize, and for the authors CBT couldregularize this brain activation. In a previous study,Farrow et al.14 showed that the ability to forgive andempathize in healthy subjects was related to activationof the following areas: left medial prefrontal cortex, leftanterior medial temporal gyrus, left inferior frontal gy-rus, orbital frontal gyrus, and posterior cingulate gy-rus/precuneus.

In the 2005 study,13 Farrow et al. evaluated whetherCBT could regularize the activation of these areas in-volved with the physiology of social cognition (the abil-ity to forgive and empathy) in individuals with PTSD,as described in the 200114 report. The participants un-derwent fMRI before and after treatment with the par-adigm of evaluation for social cognition of empathy andforgiveness. Thirteen subjects who fulfilled the criteriaof the DSM-IV for PTSD participated in the trial. Thepatients presented significant reduction of PTSD symp-toms.

The main finding was that after the treatment, thePTSD patients experienced symptom improvement ac-companied by increased brain activity in areas whichwere previously related to social cognition.14 Specifi-cally, there was an increase in the activation of the leftmedial temporal gyrus in response to the paradigm ofempathy. The same process occurred with the posteriorcingulate gyrus, which had its activation increased inresponse to the condition of forgiveness after the treat-ment. From this study Farrow et al.13 concluded thatCBT can promote changes in the brain area.

Studies with Obsessive-Compulsive Disorder Three stud-ies were identified evaluating the neurobiological ef-

fects of CBT in patients with OCD. Baxter et al.15 andSchwartz et al.16 used PET, and Nakao et al.17 examinedthe patients through fMRI.

Baxter et al.15 investigated the changes in brain me-tabolism resulting from the treatment with behavioraltherapy and fluoxetine in OCD through FDG-PET.Eighteen OCD patients took part in the study, with eachtreatment group containing nine participants, and themethod of allocation was based on the preference ofsubjects. The group of healthy control subjects wascomposed of four participants. All patients werescanned at rest.

In the fluoxetine group there were comorbidities withcyclothymic disorder, panic disorder, Tourette’s disor-der, and social phobia. In the behavioral therapy groupthere were comorbidities with cyclothymic disorder,panic disorder, and agoraphobia.

The behavioral therapy utilized techniques of expo-sure with prediction of response which was individu-alized for each patient. The exposures and predictionsof response were facilitated by cognitive techniques.None of the patients in the therapy group took medi-cation during the study, but six patients participated ingroup therapy for OCD.

The participants presented improvement of symp-toms both in the fluoxetine and in the behavioral ther-apy group. The neuroimaging findings after treatmentshowed decrease of the right anterior cingulate and leftthalamus in the fluoxetine group who responded to thetreatment. The head of the right caudate nucleus pre-sented a significant decrease in both treatments.

Baxter et al.15 concluded that the glucose metabolismof the head of the right caudate nucleus was changed inthe patients treated successfully with both behavioraltherapy and fluoxetine. There was a significant correla-tion of activity of the orbital cortex with the caudatenucleus and the thalamus before the treatment in pa-tients who responded. This correlation disappeared af-ter the success of the treatment.

In a subsequent study, the same research group16

investigated through PET neurobiological changes inpatients with OCD before and after behavioral therapy.The aim of that study was to replicate the previousfindings with an independent sample and increase thesample of subjects whose results could be combinedwith those treated with behavioral therapy in the firststudy.15

The results of the first study15 corroborate the ideathat the pathological activity of the cortical-striate-tha-

NEUROIMAGING IN ANXIETY DISORDERS

118118 http://neuro.psychiatryonline.org J Neuropsychiatry Clin Neurosci 21:2, Spring 2009

lamic circuit could be responsible for the brain media-tion of both fixed, repetitive thoughts and the behaviorsobserved in OCD. Also, the effective treatment couldreestablish the functioning of the head of the caudatenucleus of selecting if a particular stimulus will activatethe psychopathology-related circuit.

The nine participants of the subsequent study wereanalyzed separately and with the nine subjects of thefirst study by Baxter et al.15 The patients did not havecomorbidities with other axis I pathologies, which dif-fered from the sample from the first study;15 however,the treatment with behavioral therapy was similar tothe first study. Six patients responded to the treatment.

Schwartz et al.16 replicated the previous findings ofsignificant decrease of the activity of the right caudatenucleus in those who responded to therapy. When thepatients of the subsequent study were combined with thepatients of the first study, including those with good orlittle response to therapy, the right caudate nucleus pre-sented a statistically significant decrease after the treat-ment. The information of the new patients combined withthose of the old patients shows that the left caudate nu-cleus also displayed a significant change after treatment inthose who responded to the therapy. It is important tohighlight that in the first study, with a smaller sample, theleft caudate nucleus did not show any significant differ-ence between responders and nonresponders.

Schwartz et al.16 concluded that the results of thisstudy replicated those of the first study,15 presenting asignificant change in the metabolic activity of the rightcaudate, which was normalized after effective behav-ioral therapy. This change was not observed in the pa-tients who did not respond to treatment. When theparticipant data from the first study were combinedwith the participant data from the subsequent study, itwas possible to demonstrate a statistically significantpretreatment correlation between the right orbital gy-rus, caudate nucleus, and the thalamus, which de-creases after effective treatment. In the previous study,similar results were found with a sample treated withbehavioral therapy or fluoxetine. The finding that theseeffects can be demonstrated following effective treat-ment only with behavioral therapy (without using med-ication) and that the correlation between regions is notobserved in healthy control subjects, suggests that theassociation of activity between elements of the cortical-striate-thalamic circuit may be related to expression ofOCD symptoms.

Nakao et al.,17 in order to understand the pathophys-

iology of OCD, also evaluated the regional brainchanges through fMRI before and after treatment withbehavioral therapy and medication, using the paradigmof symptom elicitation and cognitive tasks.

Ten patients with OCD participated in the study.Patients with comorbid axis I disorders, neurologicaldisease, head injury, serious medical condition, historyof drug or alcohol use, or IQ below 80 evaluated by theWAIS, were excluded from the study.

Ten patients were randomly assigned to receive flu-voxamine (n�4) or behavioral therapy (n�6). Aftertreatment the clinical symptoms in both groups weresignificantly reduced; two patients in the medicationgroup did not show any improvement.

Concerning the neuroimaging findings, the patientspresented activation of the left orbital frontal cortex,temporal cortex, and parietal cortex during the task ofsymptom provocation before the treatment. After thetreatment, the patients showed decrease of the activa-tion in the orbital frontal cortex.

The study concluded that the hyperactivation of thecircuits involved in the symptomatic expression ofOCD, namely, orbital frontal cortex, anterior cingulategyrus and basal ganglia, can decrease with symptomimprovement. However, it is important to underscore alimitation of the study concerning the analysis of brainchange. Due to the small number of participants, it wasnot possible to analyze separately the patterns of brainactivation resulting from the intervention with behav-ioral therapy and with fluvoxamine.

Studies with Panic Disorder Two studies were per-formed investigating the neurobiological substrates ofCBT in patients with panic disorder through PET.18,19

Prasko et al.18 submitted resting subjects to PET scansbefore and after treatments with CBT and medication.All of the patients were without medication for at least2 weeks before the first [18F]fluorodeoxyglucose (FDG)-PET procedure.

Twelve patients fulfilling the criterion for panic dis-order with or without agoraphobia according to theDSM-IV participated in the study. Ten of the 12 patientssuffered from agoraphobia. No other comorbidity wasdiagnosed. The patients were randomly distributed intothe two treatment groups. The patients of both groupsdid not show any significant difference in symptomseverity at the beginning of the study. The CBT andmedication groups were composed of six individuals.The exclusion criterion included score above 15 on the

PORTO et al.

119J Neuropsychiatry Clin Neurosci 21:2, Spring 2009 http://neuro.psychiatryonline.org 119

Hamilton Depression Rating Scale (HAM-D), preg-nancy, having used psychotropic medication for the last2 weeks, and suffering from serious physical disease orother psychiatric disorder other than panic and agora-phobia.

The CBT treatment consisted of psychoeducation,cognitive restructuring, training of diaphragmaticbreathing and relaxation, and interoceptive and in vivoexposure. The medication group was treated with 3months of antidepressants. All of the patients con-cluded the study. The participants in both groups pre-sented significant symptom reduction following thetreatment.

Prasko et al.18 concluded that both treatments wereeffective regarding panic symptoms. The changes ofbrain metabolism in the cortical regions were similar forboth treatments. The increased activity of the brain me-tabolism in the left hemisphere was mainly at the pre-frontal, temporoparietal, and occipital regions and pos-terior cingulate. The decrease was predominantly at theleft hemisphere of the frontal region, and at the righthemisphere of the frontal, temporal, and parietal re-gion. No changes were observed in the metabolic activ-ity of subcortical areas. The results of the study indicatethat both CBT and antidepressant treatments can acti-vate the temporal cortical processing.

Sakai et al.19 also used FDG-PET to investigate thechanges in the use of regional brain glucose associatedwith anxiety reduction after CBT treatment. The au-thors worked on the hypothesis that regions above theamygdala such as the medial prefrontal cortex, anteriorcingulate cortex, and hippocampus could be modulatedin the patients who responded to CBT. Also, the amyg-dala bilaterally, hippocampus, thalamus, midbrain,caudal pons, medulla, and cerebellum would presentan increase in glucose uptake at the baseline conditionbefore the treatment and would have a reduction of thisactivation after treatment. According to the authors,these regions would be part of the “neurocircuit ofpanic.”

Twelve patients who fulfilled the criteria for panicdisorder of the DSM-IV and who had not used fluox-etine and CBT prior to the study participated in thestudy. The participants had comorbidity with agora-phobia. Individuals with the following comorbiditieswere excluded from the study: major depression, bipo-lar disorder, schizophrenia, social phobia, obsessive-compulsive disorder, posttraumatic stress disorder,generalized anxiety, suicide risk, substance abuse, per-

sonality disorder, and physical disease (unspecified inthe study). The first PET scan was performed before thetreatment with CBT and the second was performedafter the treatment. The individuals remained at restduring the performance of the procedure.

CBT treatment consisted of psychoeducation, musclerelaxation training, breathing control, in vivo exposure,attention control technique, self-instruction, self-rein-forcement, thought stopping, and cognitive restructur-ing. Eleven of the 12 patients responded to the treat-ment. The patient who continued to suffer panic attacksafter CBT treatment was excluded from the second neu-roimaging procedure.

The neuroimaging findings after CBT showed de-creased metabolism in the right hippocampus, left ven-tral anterior cingulate cortex, left cerebellum, and pons.The increased regional brain glucose metabolism wasfound at the prefrontal medial region bilaterally.

The PET results following successful CBT treatmentevidenced that the level of glucose uptake in the righthippocampus, medial prefrontal cortex, and left ventralcingulate cortex was modulated by the treatment. Thefindings are consistent with the hypothesis that regionsabove the amygdala can be adaptively modulated inpatients who respond to CBT. Thus, Sakai et al.19 con-cluded that improvement in panic symptoms throughCBT can promote brain effects.

Prediction Factors of Treatment ResponseOnly one of the identified studies presented data con-cerning the second objective of this review—to identifyresponse predictors (Table 1).

Obsessive-Compulsive Disorder Brody et al.20 used FDG-PET to investigate if the metabolic activity of regionspreviously associated with OCD symptoms could pre-dict the response to behavioral therapy. Twenty-sevenpatients participated in the study. All of them fulfilledcriteria for OCD. All subjects were scanned at rest be-fore the behavioral therapy or fluoxetine treatment. Atthe time of the procedure no subjects were taking med-ication. The method of allocation was based on thepreference of the subjects.

The behavioral therapy group was composed of 18participants and the medication group had nine partic-ipants. In the medication group the participants hadcomorbidities such as cyclothymic disorder, panic dis-order, Tourette’s syndrome, and social phobia. In thebehavioral therapy group there was comorbidity with

NEUROIMAGING IN ANXIETY DISORDERS

120120 http://neuro.psychiatryonline.org J Neuropsychiatry Clin Neurosci 21:2, Spring 2009

cyclothymic disorder, panic disorder, and agoraphobia.In both groups the pretreatment metabolism of the leftfrontal orbital cortex was correlated with the percentageof decrease in the Yale-Brown Obsessions-CompulsionsScale (Y-BOCS). In the therapy group it was found thatthe higher metabolism at the pretreatment left frontalorbit was associated with a significant improvement inthe Y-BOCS score. In the medication group, on the con-trary, a lower pretreatment metabolism at the left fron-tal orbital cortex was associated with a significant im-provement in the Y-BOCS.

The study20 concluded that a higher pretreatmentmetabolism at the left frontal orbital cortex was associ-ated with a better response to treatment with behavioraltherapy. On the contrary, a lower metabolic activity atthe left frontal orbital cortex was associated with a bet-ter response to fluoxetine treatment. The results of bothgroups of behavioral therapy and medication suggestthat OCD patients with particular models of brain me-tabolism can respond preferentially to a particular typeof treatment.

In their discussion, Brody et al.20 point out that thefunctions ascribed to the frontal orbital cortex couldexplain why a higher metabolism at this region predictsa better response to behavioral therapy. Among thefunctions of the frontal orbital cortex they highlightwhich could be related to the findings of the study.First, the frontal orbital cortex is important to mediatebehavioral responses in situations in which the affectivevalue of the stimulus can change; and second, this areaseems to have an important role in the mediation ofextinction. In a successful treatment with behavioraltherapy, the patients experience a change in the affec-tive value that they used to ascribe to the stimulus andthus extinguish compulsions. Consequently, for Brodyet al.,20 the subjects with a higher pretreatment metab-olism at the frontal orbital cortex would be more capa-ble of changing the attribution of affective value to thestimulus and thus would be more capable of extin-guishing the compulsive responses. Therefore, theseabilities would allow a better response to behavioraltherapy.

CONCLUSION

This systematic review aimed to investigate CBT-re-lated neurobiological changes in anxiety disorders, de-tected through neuroimaging techniques, and to iden-

tify predictors of response to treatment. Although thenumber of studies considered in this review is small,they demonstrate that CBT is able to modify the dys-functional neural activity related to anxiety disorders inthe patients who responded to treatment. Such resultconfirms previous reviews of psychotherapy and neu-roimaging,1,2,4,5 but the present review differs from theprevious ones as it focus solely on CBT and anxietydisorders.

The studies by Straube et al.11 and Furmark et al.12

included in their methodology the randomization ofpatients for CBT and waiting-list group, thus evidenc-ing that the neurobiological changes in the therapygroup were a result of the CBT interventions ratherthan an effect of the passing of the time. The neuroim-aging findings in studies by Paquette et al.,10 Baxter etal.,15 and Schwartz et al.16 revealed that after treatmentthe patients presented activation similar to healthycomparison subjects.

As for the second aim of this review, the identifica-tion of treatment response predictors, Brody et al.20

reported important results. They showed that OCD pa-tients with particular patterns of brain metabolism canrespond preferentially to a particular type of treatment.This is because the patients who responded to behav-ioral therapy presented a higher metabolism at the leftfrontal orbital cortex before the treatment. On the otherhand, the lower metabolic activity at the left frontalorbital cortex was associated with better response tofluoxetine treatment. It should be highlighted that onlyone study was found concerning response prediction.The lack of studies about response prediction highlightsthe importance of future research in this area. The iden-tification of treatment response predictors has greatclinical importance, as knowledge of the pretreatmentbrain metabolism can eventually aid in the choice of themost indicated intervention for a given patient.

A particularly interesting aspect of the present re-view concerns the neuroimaging findings resultingfrom CBT treatment versus medication, revealing acommon way of brain modification. Therefore, it sug-gests that psychotherapy with CBT and drug therapymay act in similar brain circuits.12,15,18 Furmark et al.12

pointed that the neural regions related to treatmentwith citalopram and CBT in social phobia converge tothe amygdala, hippocampus, and adjacent cortical ar-eas, and possibly mean a common way in the successfultreatment of social anxiety. Baxter et al.15 detected in-creased glucose metabolism in the right caudate nu-

PORTO et al.

121J Neuropsychiatry Clin Neurosci 21:2, Spring 2009 http://neuro.psychiatryonline.org 121

cleus in patients with OCD. After treatment with fluox-etine and behavioral therapy, the neuroimaging studiesrevealed decreased activity at this region in both treat-ments. Finally, Prasko et al.,18 studying patients withpanic symptoms, concluded that treatment with CBTand with antidepressants can activate the cortical tem-poral processing.

The study by Farrow et al.13 stands out because of itsdistinct methodology. It proposed to identify the brainareas involved with social cognition in PTSD patients—specifically, the ability to forgive and experience empa-thy—and found attenuated activation of the related ar-eas with the referred cognitive processes. It may bespeculated that numbing symptoms related to thegroup could impair such abilities as empathy and for-giveness, as the individuals have difficulty feeling suchemotions as intimacy and tenderness and feel discon-nected from themselves. The study by Farrow et al.13

showed that CBT can help in the remission of PTSDsymptoms as well as promote the activation of brainareas related to social cognition of empathy and for-giveness. In their meta-analysis study, Etkin and Wa-ger21 underscore that PTSD is a more complex disorderthan the other anxiety disorders, specifically social andspecific phobia. This is because the patients with PTSDdisplayed a pattern of activation and hypoactivationthat differed from those of other pathologies. The re-sults revealed more frequent hyperactivity in the insulaand amygdala in patients with social and specific pho-bia than in PTSD patients. Also, only patients withPTSD displayed hypoactivation of the dorsal and ros-tral anterior cingulate cortex and medial prefrontal cor-tex. For the authors, PTSD may be related to a dysfunc-tion in the system of emotional regulation in which fearby itself is only an element of this system, while socialand specific phobia would be related to a stage of in-tense fear and therefore would present greater activityof the insula and amygdala.

It should be mentioned that the heterogeneity of thestudies presented here limits the possibility of directcomparison (Table 2). In many of them, such as Baxteret al.,15 Schwartz et al.,16 and Brody et al.,20 althoughthe therapeutic modality was referred to as behavioraltherapy, in the description of strategies there was ref-erence to cognitive techniques, suggesting that it wasCBT. The different number of sessions also made itdifficult to compare them. The neuroimaging methodsin the studies were also different, as four studies usedfMRI and six used PET. Other considerations must be

made as for the methods of the studies. Some studiesused control groups while others did not. For evalua-tion of neurobiological alterations, three types of exper-imental paradigms were used—symptom provocation,cognitive tasks performance, and testing at rest. Mostimportantly, all the reviewed studies are small in sam-ple size. This limitation restricts statistical power, en-hances false negative results, and therefore has limitedgeneralizability. To elucidate differences in brainchanges resulting from cognitive behavior therapy, alarger number of patients must be examined before andafter treatment. Further studies in this area need to beundertaken. Nakao et al.,17 for instance, carried out astudy with OCD patients, where the patterns of brainactivation resulting from the comparison between be-havioral therapy and fluvoxamine could not be statis-tically analyzed separately, due to the small number ofparticipants. However, Schwartz et al.16 combined re-sults from a previous study15 and were able to replicatethe original findings. Despite these acknowledged lim-itations, it can be concluded that CBT may indeed pro-mote neurobiological changes.

The investigation of changes in brain activity resultingfrom therapy is a new area of research which has majorimplications for better understanding the mechanisms offormation and maintenance of symptoms. Moreover, theycan help reveal the biological mechanisms associated withthe improvement of symptoms due to successful CBTtreatment. After an analysis of studies we could proposeplans for the performance of new clinical trials which cananswer questions on neurobiological changes and psycho-therapy. The methodology of studies should include ran-domized comparison groups for waiting list and a placebogroup. We would thus have more evidence that the brainchanges that occurred would be due to interventions withpsychotherapy.

While Etkin and Wager21 in their meta-analysis of struc-tures related to disorders of specific and social phobia andPTSD found a common route for anxiety that would bethe hyperactivation of the amygdala and insula, in thepresent systematic review we did not find such a model ofactivation. Possibly, this different result is related to thesmall number of studies that met the criteria for our re-view. Consequently, the inconsistency of results indicatesthe need for future research. However, the studies in-cluded in our systematic review pointed out in their neu-roimaging findings structures that participated both in thebrain circuits involved with extinction and in those in-volved with cognitive regulation of emotion. The results

NEUROIMAGING IN ANXIETY DISORDERS

122122 http://neuro.psychiatryonline.org J Neuropsychiatry Clin Neurosci 21:2, Spring 2009

TA

BL

E2.

Met

hod

olog

yof

Stu

die

s

Dis

ord

ers

Sp

ider

Ph

obia

Soc

ial

Ph

obia

PT

SD

OC

DP

anic

Stu

die

sP

aqu

ette

etal

.(2

003)

Str

aub

eet

al.

(200

6)Fu

rmar

ket

al.

(200

2)Fa

rrow

etal

.(2

005)

Bax

ter

etal

.(1

992)

Sch

war

tzet

al.

(199

6)B

rod

yet

al.

(199

8)N

akao

etal

.(2

005)

Pra

sko

etal

.(2

004)

Sak

aiet

al.

(200

6)

Hea

lthy

com

pari

son

grou

p13

subj

ects

14su

bjec

ts4

subj

ects

4su

bjec

ts

Com

pari

son

wai

ting

-lis

tgr

oup

12su

bjec

ts6

subj

ects

Med

icat

ion

grou

p6

subj

ects

9su

bjec

ts9

subj

ects

4su

bjec

ts6

subj

ects

The

rapy

grou

p12

subj

ects

13su

bjec

ts6

subj

ects

13su

bjec

ts9

subj

ects

9su

bjec

ts18

subj

ects

6su

bjec

ts6

subj

ects

12pa

tien

tsA

lloca

tion

met

hod

Ran

dom

Ran

dom

Subj

ect

pref

eren

ceSu

bjec

tpr

efer

ence

Ran

dom

Ran

dom

Tec

hniq

uefM

RI

fMR

IPE

TfM

RI

PET

PET

PET

fMR

IPE

TPE

TE

xper

imen

tSy

mpt

ompr

ovoc

atio

nSy

mpt

ompr

ovoc

atio

nSy

mpt

ompr

ovoc

atio

nSo

cial

cogn

itio

nac

tiva

tion

Res

tR

est

Res

tSy

mpt

ompr

ovoc

atio

nan

dco

gnit

ive

task

s

Res

tR

est

The

rape

utic

mod

alit

yC

BT

CB

TC

BT

CB

TB

TB

TB

TB

TC

BT

CB

T

Num

ber

ofse

ssio

ns4

sess

ions

2se

ssio

ns8

sess

ions

4-10

sess

ions

8-24

sess

ions

8-24

sess

ions

8-24

sess

ions

12se

ssio

ns20

sess

ions

10se

ssio

ns

fMR

I�fu

ncti

onal

mag

neti

cre

sona

nce

imag

ing;

PET

�po

sitr

onem

issi

onto

mog

raph

y;B

T�

beha

vior

alth

erap

y;C

BT

�co

gnit

ive

beha

vior

ther

apy;

PTSD

�po

sttr

aum

atic

stre

ssd

isor

der

;OC

D�

obse

ssiv

e-co

mpu

lsiv

ed

isor

der

PORTO et al.

123J Neuropsychiatry Clin Neurosci 21:2, Spring 2009 http://neuro.psychiatryonline.org 123

showed that CBT especially regulated the dysfunctionalneural circuits involved with the regulation of negativeemotions and fear extinction.

The literature shows that many mental disorders areinvolved with the inability to control fear22,23 and diffi-culty in regulating negative emotions.24,25 These data sug-gest that the conditioning of fear and the difficulty inregulating emotions play a major role in the formationand maintenance of anxiety disorders. Mocaiber et al.24

highlight the research on neural circuits of extinctionwhich has an important clinical implication. This is be-cause anxiety disorders are in part characterized by resis-tance to the extinction of emotional reactions to anxiogenicstimuli and by avoidance behaviors.

It is important to highlight that CBT treatment containsspecific techniques (exposure, distraction, and cognitiverestructuring) which allow both the extinction of condi-tioned fear and the cognitive regulation of emotions.

Cognitive behavior therapy has proved to be effectivein the treatment of various mental disorders, although

the neurobiological effects of its action are little known.CBT favors the restructuring of thought, modification offeelings and behaviors, and promotes new learning.Consequently it involves synaptic changes.26 This re-view had the aim of identifying the studies that haveproposed to understand the brain alterations resultingfrom CBT. The investigation of changes in brain activityresulting from successful CBT treatment allows us toclarify the neural substrates underlying psychotherapy.

Neuroimaging studies provide a means to observeand characterize changes in brain functioning related topsychological and pharmacological interventions. Con-sequently, to understand how individuals process astimulus can be an important piece of information fortherapeutic response.4

The neuroscientific findings associated with theneuroimaging studies can enhance our knowledge ofthe neurobiological foundations of psychotherapies,as well as improve interventions in order to increasetreatment efficacy.

References

1. Kumari V: Do psychotherapies produce neurobiological ef-fects? Acta Neuropsychiatrica 2006; 18:61–70

2. Linden D: How psychotherapy changes the brain: the contri-bution of function neuroimaging. Mol Psychiatry 2006; 11:528–538

3. Beauregard M: Mind does really matter: evidence from neu-roimaging studies of emotional self-regulation, psychother-apy, and placebo effect. Prog Neurobiol 2007; 81:218 –236

4. Etkin A, Phil M, Pittenger C, et al: Toward a neurobiology ofpsychotherapy: basic science and clinical applications. J Neu-ropsychiatry Clin Neuroscience 2005; 17:145–158

5. Roffman J, Marci C, Glick D, et al: Neuroimaging and func-tional neuroanatomy of psychotherapy. Psychological Medi-cine 2005; 35:1–14

6. Raedt R: Does neuroscience hold promise for the further de-velopment of behavior therapy? The case of emotional changeafter exposure in anxiety and depression. Scand J Psychol2006; 47:225–236

7. Beck A: The current state of cognitive therapy. Arch GenPsychiatry 2005; 62:953–959

8. Foa E: Psychosocial therapy for posttraumatic stress disorder.J Clin Psychiatr 2006; 67(suppl 2):40–45

9. Ost L-G: One-session group treatment of spider phobia. BehavRes Ther 1996; 34:707–715

10. Paquette V, Levesque J, Mensour B, et al: Change the mindand you change the brain: effects of cognitive behavior ther-apy on the neural correlates of spider phobia. Neuroimage2003; 18:401–409

11. Straube T, Glauer M, Dilger S, et al: Effects of cognitive be-havior therapy on brain activation in specific phobia. Neuro-image 2006; 29:125–135

12. Furmark T, Tillfors M, Marteinsdottir I, et al: Common

changes in cerebral blood flow in patients with social phobiatreated with citalopram or cognitive behavior therapy. ArchGen Psychiatry 2002; 59:425–433.

13. Farrow TFD, Hunter MD, Wilkinson ID, et al: Quantifiablechange in functional brain response to empathic and for-givability judgments with resolution of posttraumatic stressdisorder. Psychiatry Research-Neuroimaging 2005;140:45–53

14. Farrow TFD, Zheng Y, Wilkinson I, et al: Investigating thefunctional anatomy of empathy and forgiveness. Neuroreport2001; 12:2433–2438

15. Baxter LR, Schwartz JM, Bergman KS, et al: Caudate glucosemetabolic-rate changes with both drug and behavior-therapyfor obsessive-compulsive disorder. Arch Gen Psychiatry 1992;49:681–689

16. Schwartz JM, Stoessel PW, Baxter LR, et al: Systematicchanges in cerebral glucose metabolic rate after successfulbehavior modification treatment of obsessive-compulsive dis-order. Arch Gen Psychiatry 1996; 53:109–113

17. Nakao T, Nakagawa A, Yoshiura T, et al: Brain activation ofpatients with obsessive-compulsive disorder during neuro-psychological and symptom provocation task before and aftersymptom improvement: a functional MRI study. Biol Psychi-atry 2005; 57:901–910

18. Prasko J, Horacek J, Zalesky R, et al: The change of regionalbrain metabolism (18FDG PET) in panic disorder during thetreatment with cognitive behavior therapy or antidepres-sants. Neuro Endocrinol Lett 2004; 25:340 –348

19. Sakai Y, Kumano H, Nishikawa M, et al: Changes in cere-bral glucose utilization in patients with panic disordertreated with cognitive behavior therapy. Neuroimage 2006;33:218 –226

NEUROIMAGING IN ANXIETY DISORDERS

124124 http://neuro.psychiatryonline.org J Neuropsychiatry Clin Neurosci 21:2, Spring 2009

20. Brody AL, Saxena S, Schwartz JM, et al: FDG-PET predictorsof response to behavioral therapy and pharmacotherapy inobsessive compulsive disorder. Psychiatry Res 1998; 84:1–6

21. Etkin A, Wager T: Functional neuroimaging of anxiety: ameta-analysis of emotional processing in PTSD, social anxietydisorder, and specific phobia. Am J Psychiatry 2007; 164:1476–1488

22. Sotres-Bayon F, Cain C, LeDoux J: Brain mechanisms of fearextinction: historical perspectives on the contribution of pre-frontal cortex. Biol Psychiatry 2006; 60:329–336

23. Liggan DY, Kay J: Some neurobiological aspects of psycho-therapy. J Psychother Pract Res 1999; 8:103-114

24. Mocaiber I, Oliveira L, Garcia Pereira M, et al: Neurobiologyof emotion regulation: implications for cognitive behaviortherapy. Psicologia em Estudo, Maringa 2008; 3:513–520

25. Ochsner K, Gross J: The cognitive control of emotion. TrendsCogn Sci 2005; 9:242–249

26. Moraes K: The value of neuroscience strategies to accelerateprogress in psychological treatment research. Can J Psychiatry2006; 51:810–822

CME CALENDAR

CME CalendarTo announce educational events in the CME Calendar,please provide dates, location, sponsorship, number of CMEcredits, and contact information. Mail announcements forthe Fall 2009 issue by August 15 to Katie Duffy, StaffEditor, CME Calendar, The Journal of Neuropsychiatryand Clinical Neurosciences, American PsychiatricPublishing, Inc., 1000 Wilson Boulevard, Suite 1825,Arlington, VA 22209-3901.

November 7, 2009, 23rd Annual Epilepsy Update.Sponsored by the Mayo Clinic, Scottsdale, Arizona.The program is intended for neurologists; physicians

in internal medicine, family practice and generalpractice; physician assistants; and nurses who areseeking a review or update in current clinical practiceof seizures and epilepsy management. An interactiveand didactic format allows utilization of an audienceresponse system for all case studies andpresentations. Each session will conclude with anopportunity for participants to ask questions on thecases discussed or present their own cases to thefaculty panel. For information, contact Staci King,Mayo Clinic Scottsdale, 13400 E. Shea Boulevard,Scottsdale, AZ 85259; 480-301-4580.

PORTO et al.

125J Neuropsychiatry Clin Neurosci 21:2, Spring 2009 http://neuro.psychiatryonline.org 125