caudate glucose metabolic rate changes with both drug and bt for ocd (baxter, 1992)

Caudate Glucose Metabolic Rate ChangesWith Both Drug and Behavior Therapy

for Obsessive-Compulsive DisorderLewis R. Baxter, Jr, MD; Jeffrey M. Schwartz, MD; Kenneth S. Bergman, MD; Martin P. Szuba, MD;

Barry H. Guze, MD; John C. Mazziotta, MD, PhD; Adina Alazraki; Carl E. Selin, MS;Huan-Kwang Ferng, MD; Paul Munford, PhD; Michael E. Phelps, PhD

\s=b\We used positron emission tomography to investigate lo-cal cerebral metabolic rates for glucose (LCMRG1c) in pa-tients with obsessive-compulsive disorder before and aftertreatment with either fluoxetine hydrochloride or behaviortherapy. After treatment, LCMRG1c in the head of the rightcaudate nucleus, divided by that in the ipsilateral hemi-sphere (Cd/hem), was decreased significantly comparedwith pretreatment values in responders to both drug andbehavior therapy. These decreases in responders were alsosignificantly greater than right Cd/hem changes in nonre-sponders and normal controls, in both of whom values didnot change from baseline. Percentage change in obsessive-compulsive disorder symptom ratings correlated signifi-cantly with the percent of right Cd/hem change with drugtherapy and there was a trend to significance for this samecorrelation with behavior therapy. By lumping all respond-ers to either treatment, right orbital cortex/hem was signif-icantly correlated with ipsilateral Cd/hem and thalamus/hem before treatment but not after, and the differencesbefore and after treatment were significant. A similar pat-tern was noted in the left hemisphere. A brain circuitinvolving these brain regions may mediate obsessive-compulsive disorder symptoms.

(Arch Gen Psychiatry. 1992;49:681-689)

Obsessive-compulsive disorder1 (OCD) is characterizedby recurrent, unwanted thoughts (obsessions) and

conscious, ritualized acts (compulsions) usually at¬tributed to attempts to deal with anxiety generated by theobsessions. There is now ample evidence that both medi-

cations that are strong serotonin re-uptake inhibitors2 andspecific behavioral therapies that employ the princi¬ples of exposure and response-prevention3 are highlyeffective in reducing the symptoms of OCD, although thelatter may be more effective for compulsions than for ob¬sessions. Based on a wide variety of evidence, manyinvestigators have postulated a role for the basal ganglia,along with limbic, thalamic, and cortical brain regions, inthe mediation of OCD symptoms.411 We postulated pre¬viously that the head of the caudate nucleus plays a cen¬tral role in OCD symptom mediation and that successfultreatment of OCD by either medication or behavior ther¬apy would be accompanied by a change in caudate nucleusfunction that might be detected with positron emissiontomography (PET) and the 18-F-fluorodeoxyglucose(fludeoxyglucose F 18 [FDG]) method.5,8 Cerebral glucosemetabolism is a sensitive indicator of brain function.12

We decided, therefore, to study OCD patients withFDG-PET before and after either drug or behavior therapy.We chose fluoxetine hydrochloride as our treatmentdrug,1315 knowing a similar study was in progress usingclomipramine hydrochloride.16

PATIENTS, SUBJECTS, AND METHODSThis study was conducted in accordance with guidelines

established by the UCLA Human Subjects' Protection Committee.

PatientsThere were initially 10 patients with OCD in each of the two

treatment groups, but a computer failure led to the loss of datafor the initial scan of one subject in each group. The nine subjectsremaining in each treatment group are included in the analysesreported herein.

All patients were clinical outpatients and /or inpatients in theUCLA Neuropsychiatrie Hospital Mood and Anxiety DisordersTreatment Program who were asked to undergo PET scanningbefore and after the clinical treatment they elected (after a carefuldiscussion of all available treatment options), which was eitherfluoxetine hydrochloride or behavior therapy without drugs. Allhad a current primary diagnosis of DSM-III-R OCD which hadbeen present for at least 1 year. Diagnoses for each patient weremade based on both a nonstructured psychiatric interview andthe Schedule for Affective Disorders and Schizophrenia-Lifetime

Accepted for publication February 8, 1992.From the Departments of Psychiatry (Drs Baxter, Schwartz, Bergman,

Szuba, Guze, Ferng, and Munford), Radiological Sciences (Drs Baxter,Guze, Mazziotta, and Phelps, Ms Alazraki, and Mr Selin), and Neurol-ogy (Dr Mazziotta and Ms Alazraki) and the Laboratory of NuclearMedicine (Drs Baxter, Guze, Mazziotta, and Phelps, Ms Alazraki, andMr Selin), UCLA School of Medicine. Dr Bergman is now at the Univer-sity of Washington School of Medicine, Seattle; Ms Alazraki is now atEmory University School of Medicine, Atlanta, Ga; and Dr Ferng is nowat National Defense Medical Center, Taipei, Republic of China.

Reprint requests to the Department of Psychiatry, UCLA/NPI, 760Westwood Plaza, Los Angeles, CA 90024 (Dr Baxter).

Downloaded From: http://archpsyc.jamanetwork.com/ on 07/01/2012

PopulationsDrug

TreatmentBehavior

Treatment

No.No. drug freeAge, ySex, M/FNo. R-handedY-BOCS

TotalObsessive (items 1-5)Compulsive (items

6-10)HAM-DHAM-AGAS

99

31.2±12.93/6

24.9±3.612.7+2.3

12.2±1.610.4±5.426.6±9.757.8 + 9.7

99

34.7±6.04/5

23.9+5.512.0±3.8

11.9+2.97.8±5.7

22.8±7.459.1 ±8.6

"OCD indicates obsessive-compulsive disorder; Y-BOCS, Yale-Brown Obsessive-Compulsive Scale; HAM-D, Hamilton DepressionRating Scale; HAM- , Hamilton Anxiety Rating Scale; and GAS, GlobalAssessment Scale. Values are numbers or means±SDs.

version (SADS-L).17 All patients were drug free for at least 2 weeksbefore their initial PET scan.

In the drug treatment group, there were concomitant diagno¬sis of cyclothymic disorder, panic disorder, Tourette's disorder(mild), and social phobia in one subject each. In the behaviortherapy group, there were concomitant diagnoses of cyclothymicdisorder, panic disorder, and acrophobia in one subject each.Many patients in each group had had major depression in thepast, but all were euthymic now, as determined by both clinicalassessment and the SADS-L. All patients denied substance abuse;we did not do urine drug screens. Patient demographics and rat¬ing scale scores are given in Table 1.

Normal Control SubjectsSubsequent to analyses of patient PET data, a small group of

normal control subjects were collected for comparison. Thesewere two men and two women with a mean (±SD) age of 29 ±8years who were judged to be healthy by history and recent phys¬ical examination. None had any personal or family history of anDSM-HI Axis I disorder. All were drug free by history for at least1 month before initial PET scanning. They were rescanned 10±2weeks after initial scanning.

Symptom Severity RatingsSubject behaviors were rated with the Yale-Brown Obsessive-

Compulsive Scale1819 (Y-BOCS), the 17-item Hamilton DepressionRating Scale20 (HAM-D), the Hamilton Anxiety Rating Scale21(HAM- ), and the Global Assessment Scale17 (GAS) at the time ofeach PET scan. Responders to treatment were defined a priori asthose who were rated as either "much improved" or "very muchimproved" on item 18 of the Y-BOCS (which is taken from theClinical Global Impression Scale) at the time of the second PETscan. Nonresponders were those who scored "no change" or

"minimally improved" on the same item.

TreatmentsDrug Treatment.—Patients in this group were treated with oral

fluoxetine hydrochloride, started at 20 mg/d and titrated within2 weeks to 60 to 80 mg/ d, as tolerated, before the second PET scanwas obtained. This is our standard clinical treatment for OCDwith this drug. All patients enrolled in drug treatment were ableto attain this dosage range with minimal side effects and stay onit for the duration of the study. No other drugs were used, and

patients were not in behavior therapy or formal psychotherapy.If they asked about how they should respond to obsessions andcompulsive urges, these patients were advised to see if they couldrefrain from these behaviors, as comfort permitted, if they wished.All did receive supportive advice from their treaters (L.R.B.,M.P.S., and J.M.S.), with whom they met once or twice a week.

Behavior Therapy.— These patients elected to have behaviortherapy consisting of exposure and response-prevention, whichwas individualized for the patient. Exposure and response-prevention exercises were facilitated by cognitive techniques.22Patients met once or twice a week with their therapist ([.M.S.,K.S.B., or L.R.B.) for approximately 1 hour to review assignmentsfor exposure and response-prevention, which they did as home¬work and self-monitored with diaries and/or graphs. Homeworkwas reviewed and evaluated with therapists at the next meeting.Some had occasional therapist-aided exposure and response-prevention sessions. Six also attended a cognitive-behavioraltherapy group for patients with OCD run by one of us (J.M.S.).None took any psychoactive medications during the study. Allreceived supportive advice from their individual treaters.

All patients had 10±2 weeks of the therapy they elected beforeundergoing a second PET scan. Those treated with behaviortherapy abstained from medications from the first through thesecond scan, while those having drug treatment were all on flu¬oxetine at the time the second scan, with the last dose given theday before scanning.

PET MethodsAll subjects were injected with FDG while in the supine position

in a room with no conversation, low ambient light, and environ¬mental noise (mostly from the scanner gantry), as previously de¬scribed.23 The subjects' ears and eyes were open, and they wereinstructed to look at the diffusely lit white ceiling above the to¬mograph. Each subject received 185 to 370 mBq (5 to 10 mCi) ofFDG, prepared as previously described.24 "Arterialized" venousblood was obtained by having the subject's hand in a handwarmer. Blood sampling and determinations of plasma glucoseand FDG concentrations have been described in detail previ¬ously.25·26

Scanning was performed with a PET tomograph (831 NeuroECAT III, Seimens-CTI, Knoxville, Tenn). Fifteen transverse sec¬tions of the brain, spaced 6.75 mm apart, were acquired simulta¬neously at an angle parallel to the canthomeatal plane. Each sub¬ject's head was held in a special head holder27 during scanning toallow accurate positioning using the low-power laser marker ofthe tomograph.

Before injection of the tracer, a 2-minute transmission scan withradiation from a germanium Ge 68 (68Ge) ring source was

performed to obtain similar planes for intersubject comparisonsand to allow accurate repositioning for follow-up scans done af¬ter treatment. Next, data for measured attenuation correctionwere obtained by means of a 20-minute transmission scan usingthe ring source. Emission scanning commenced 60 minutes afterinjection of FDG. Total imaging time was 40 minutes, and eachimage was reconstructed from 2 million to 3 million counts. 18-F-Fluorodeoxyglucose was injected between 10:00 and 15:00. Im¬ages were reconstructed by filtered back-projection using of a

Shepp filter with a cutoff of 0.6 of the Nyqvist frequency (0.95cycles/cm). The in-plane resolution used in these studies was 6X6mm, and the axial resolution was 6.75 mm.28

Images were displayed on a video display (SuperMac Technol¬ogy, Sunnyvale, Calif) in a 256X256 pixel display, with a black-on-white format. Neuroanatomic regions of interest (ROIs) wereidentified in all tomographic planes in which they occurred, andthe glucose metabolic rates were determined as described previ¬ously (rate constants used for gray-matter metabolic rate deter¬minations were k,=0.102, k2=0.13, k3=0.062, 1 ,=0.0068,26 and thelumped constant was 0.5229·30).25'26·31 To accomplish this, each scanwas compared with template sets obtained from normal anatomicand PET studies with the same tomograph. The size and site ofthe ROI were then copied from these templates in a standardizedfashion appropriate to the individual brain at hand, as previously


Treatment Response by GrouptDrug Treatment Behavior Treatment

PretreatmentScore

PosttreatmentScore

PretreatmentScore

PosttreatmentScore

Y-BOCSTotalObsessive (items 1-5)Compulsive (items 6-10)

HAM-DHAM-A

GAS

25.8±3.713.3±2.212.5±1.710.3+3.224.2 ±4.257.5±7.7

Responders

13.0±2.96.6 + 1.56.4±1.75.6±2.2

20.1 ±4.3

68.2±6.9

.02

.02

.02

.04NSNS

22.3±4.111.3±3.611.0+2.45.5±3.3

20.2±4.360.0±9.6

13.5+4.07.0±2.36.5±2.75.2 + 3.1

18.3±4.471.8±4.8

.04

.22

.04

NSNS

NS

Y-BOCSTotalObsessive (items 1-5)Compulsive (items 6-10)

HAM-DHAM-A

GAS

24.0±2.012.0±2.012.0±0.010.0±9.032.5±16.556.0±13.0

Nonresponders19.5±0.58.5±0.5

11.0+1.05.5 + 2.2

19.5±3.559.0±11.0

NSNSNSNSNSNS

27.0±5.713.3±3.113.7+2.612.3±5.828.0±8.257.3±3.3

22.0±4.911.3±2.610.7±2.65.3±4.1

21.7±4.658.7±6.3

NSNSNSNSNSNS

"OCD indicates obessive-compulsive disorder.tin the drug treatment group, seven of nine subjects were responders and two of nine were nonresponders; in the behavior treatment group, six

of nine subjects were responders and three of nine were nonresponders. Values are means±SDs. See Table 1 for expansion of abbreviations. NSindicates not significant.described.32 For patients, this was done by one of us (A.A.), whowas blind to subject identity, diagnosis, scan sequence, and thehypothesis being tested. Patient scans were scattered among 85from a variety of studies being processed at the same time.Supratentorial hemispheric values were obtained in a similarfashion and included ventricles. The scans of control subjects an¬

alyzed subsequently had their ROIs defined by a technician whowas also blind to scan sequence and the hypothesis being tested.For each subject, an average local glucose metabolic rate value(LCMRGlc) was determined for each structure by weighing thatstructure's planar metabolic value by its cross-sectional area us¬

ing the following formula:

LCMRG1c=2(MR)íAí/2a¡i = l i=l

where indicates the number of planes that include the structure,A] indicates the cross-sectional area of the structure in plane i, andMR¡ indicates the metabolic rate of the structure in plane i. TheLCMRGlc is expressed in milligrams of glucose used per 100 gof brain tissue per minute. LCMRGlc for each ROI was "normal¬ized" by dividing the LCMRGlc of that region by that of the ip¬silateral cerebral hemisphere (ROI/hem), which yielded a dimen-sionless ratio. Using this metabolic ratio, rather than absolutevalues, reduces the variance in the resultant data set and gives ameasure of activity relative to that in the rest of the hemisphere.

Statistical AnalysesAll results are presented as the mean±SD. We had made a clear

a priori prediction that metabolic rates in the caudate nuclei, nor¬malized to the ipsilateral hemisphere (Cd/hem), would differ sig¬nificantly before and after treatment for responders to both drugand behavior therapy. In addition, the degree of change in re¬

sponders would show significantly more change in Cd/hem thanin nonresponders or controls, in whom change would not occurafter a similar time interval. This was our primary hypothesis.

Statistical comparisons for parametric measures (eg, ROI/hemvalues) were made with the paired t test for data from the sameindividual at the two times of scanning. Student's f test was used

for comparisons between groups. However, because of low subjectnumbers and concerns about normal distributions, we also per¬formed the nonparametric Wilcoxon test on critical results thatwere significant by Student's f test, and the Wilcoxon Signed-Ranktest for similar intrasubject evaluations. The report by Benkelfat etal16 and doubts about our previous report5 concerning direction ofcaudate metabolic nucleus change after an unconventional drugtreatment of depressed OCD patients (see below) led us to use two-tailed statistical tests to identify a difference, regardless of direc¬tion, in Cd/hem values before and after successful treatment.

We also examined the orbital gyrus/hem, putamen/hem, an¬terior cingulate gyrus/hem, and thalamus/hem, given the find¬ings and theories of others,6'7·16'33,34 as secondary hypotheses. Nocorrections were made for multiple tests. Other brain ROIs (an-terolateral prefrontal cortex, sensory-motor region, parietal cor¬tex, lateral temporal lobes, hippocampus-parahippocampalgyrus complex, amygdaloid nuclei complex, and cerebellar hemi¬spheres), available for analysis as part of the ROI protocol forscans from the various studies undergoing processing at the time,were examined in an a posteriori survey, with no correctionsmade for multiple tests.

For comparisons among more than two groups, the nonpara¬metric Kruskal-Wallis analysis of variance was used, since normaldistributions and equal variances did not obtain. Post hocWilcoxon tests were used to distinguish between relevant pair¬ings of OCD treatment responders and normal controls.

The nonparametric Kendall's , with a correction for tied val¬ues, was used for rank-order correlations between percentagechange in the nonparametric behavioral rating scale scores andpercentage change in normalized regional metabolic rates. (TheKendall is valid with n£8.35) The was also used for parametricdata (metabolic rates), because normal distributions, as deter¬mined by visual inspection of frequency histograms, did not ob¬tain for many brain regions.

RESULTSDrug-treated patients underwent their second PET scanning

after 10.5±1.3 weeks of treatment; patients undergoing behaviortherapy did so after 10.8± 1.1 weeks. Normal control subjects wererescanned after 9.8±1.3 weeks.


Drug Treatment (df=6) Behavior Treatment (df=5) -1 -1

Site Pretreatment Posttreatment f Pretreatment Posttreatment t

R head of caudatenucleus 1.23±0.08 1.17±0.10 5.81 .001 § 1.28±0.12 1.18±0.15 4.12 .009||

L head of caudatenucleus 1.10±0.09 1.08±0.09 1.07 .33 1.14±0.13 1.10±0.13 1.41 .22

R putamen 1.38±0.08 1.38±0.10 0.06 .96 1.44±0.12 1.42±0.16 0.56 .60L putamen 1.33±0.09 1.37±0.10 1.21 .27 1.46+0.15 1.40±0.14 2.24 .08R anterior

cingulate gyrus 1.20±0.05 1.14±0.06 2.89 .03 1.18±0.07 1.18±0.05 0.01 .99L anterior

cingulate gyrus 1.26±0.04 1.28±0.07 0.84 .43 1.25±0.08 1.24±0.08 0.75 .49R orbital gyri 1.17+0.09 1.17±0.11 0.12 .91 1.15±0.04 1.12±0.08 0.93 .40L orbital gyri 1.18±0.07 1.23+0.10 1.12 .30 1.17+0.07 1.16±0.09 0.31 .77Rthalamus 1.17+0.11 1.10±0.09 1.71 .14 1.15±0.14 1.12+0.13 1.07 .33L thalamus 1.23±0.09 1.16±0.09 2.75 .03 1.19±0.13 1.20±0.13 0.29 .78

'Fluoxetine hydrochloride-treated responders (n=7) and behavior therapy-treated responders (n=6).tRatio of glucose metabolic rate in brain region of interest, divided by that in the ipsilateral hemisphere.iHead of caudate nuclei predicted a priori to show significant changes.§Wilcoxon z=2.27, P=.02.¡Wilcoxon z=2.04, f>=.04.

eS>? 00

Drug Therapy Behavior Therapy

,c.e<s NormalControlSubjects

Fig 1.—Percent change after drug or behavior therapy {[(posttreatmentvalue-pretreatment valueVpretreatment valuejx 100} in glucose meta¬bolic rate in the right head of the caudate nucleus, divided by that of theipsilateral hemisphere (Cd/hem) for responders and nonresponders totreatment, as well as normal controls scanned twice with a between-scaninterval similar to that of the patients. Differences between respondersand nonresponders to treatment are significant (P<.05) for both treat¬ment groups, as are differences between treatment responders and nor¬mal controls. Arrows indicate data for subjects illustrated in Fig 2.

Seven of the drug-treated patients and six of the behavior ther¬apy patients were judged to be responders by the preestablishedcriterion; the rest were nonresponders. Rating scale scores beforeand after treatment for these groups are presented in Table 2. Al¬though not the determinant of response here, there was no over-

lap between responders and nonresponders in percentage changein the total Y-BOCS score before and after treatment: respondersdecreased scores by 30% or more and nonresponders by less than30%, comparable with the differences between responders andnonresponders on the Y-BOCS in other studies.36

Table 3 presents pretreatment and posttreatment ROI/hemvalues obtained in those patients judged treatment responders forthe brain regions postulated on a primary and secondary a prioribasis. Whole-hemisphere LCMRGlc values did not change sig¬nificantly from before to after treatment in either group ofresponders (percent change in drug treatment group: left hemi¬sphere, 9.2%±63.8% [f=.32, df=6, P=not significant]; right hemi¬sphere, 12.1%±69.4% [f=.39, df=6, P=not significant]; percentchange in behavior therapy group: left hemisphere, 11.6% ± 16.9%[{=1.54, df=5, P=not significant]; right hemisphere: 14.1%±18.3%[f=1.73, df=5, P=not significant]).

Although right anterior cingulate gyrus/hem and leftthalamus/hem showed significant changes with successful drugtreatment, only the head of the right caudate nucleus showed a

significant change with both successful drug and behaviortherapies—values decreased. While responders changed rightcaudate nucleus ratios —5.2% ±2.3% for drug treatment and-8.0% ±4.8% for behavior therapy, nonresponders changed only0.3%±1.0% (P=not significant) and 2.6%±3.2% (P= not signifi¬cant), respectively. Normal controls changed only 0.4% ±2.0% (P=not significant). Individual subject values for percentage changein right Cd/hem are represented in Fig 1. The Kruskal-Wallisanalysis of variance, applied to right Cd/hem among all groups,was significant (Kruskal-Wallis test statistic, 14.17; P=.007). Dif¬ferences between responders and nonresponders were significantfor both drug treatment (f=2.84, df=7, P=.025; Wilcoxon z=1.90,P=.06) and behavior therapy (f=3.15, df=7, P=.02; Wilcoxon z=2.19,P=.03). Right Cd/hem percentage changes for normal controlswere less than those for drug treatment (z=2.17, df=9, P=.03) andbehavior therapy (z=2.24, df=8, P=.025) responders. For illustra¬tion, Fig 2 presents FDG-PET scans of representative OCD treat¬ment responders before and after each treatment.

Contrary to our expectations of significant Cd/hem results forboth sides of the brain, the left Cd/hem did not show a statisti¬cally significant change with either treatment. Therefore, separateanalyses of results in relationship to handedness were conducted.Results were unchanged when left and right caudate nucleus


Fig 2.— 18-F-Fluorodeoxyglucose (fludeoxyglucose F 18) positron emis¬sion tomographic scans of representative patients in a horizontal planeat a middle level of the head of the caudate nuclei before and after suc¬cessful drug treatment or behavior therapy (Behav Tx) for obsessive-compulsive disorder (OCD). Scans were processed to reflect the ratio ofglucose metabolic rate registered by each pixel element, divided by thatof whole brain; color bar reflects this with linear scaling. Arrowheads in¬dicate right head ofcaudate nucleus. (Display follows radiologie and an¬atomic convention ofdisplaying the right side on the viewer's left.) Theseparticular examples were chosen for illustration because ofexactness ofscan repositioning and because they demonstrate various degrees of vis¬ible left-right asymmetry of caudate nucleus change from before to aftertreatment.

values for the one left-handed patient in each treatment group(both responders) were switched, and when they were excludedfrom the analyses.

There was a significant, positive rank-order correlation be¬tween percentage change in total Y-BOCS score before and aftertreatment and the percentage change in right Cd/hem whenconsidering all nine subjects having drug therapy ( =.48, P=.04)and a trend for those patients undergoing behavior therapy( =37, P=.09).

None of the other brain regions surveyed that are not listed inTable 3 showed significant changes in pretreatment to posttreat¬ment results in the treatment responders (all P>.10).

Since there were significant changes with both successful drugand behavior treatment only for right Cd/ hem, and we have alwaysobserved high correlations between normalized left and right headof caudate nucleus values in normal controls (J.C.M. and L.R.B., un¬

published data, 1982 to 1992), we wondered what happened to sucha correlation in the OCD treatment responders before and aftertreatment. To have adequate numbers for valid correlation coeffi¬cients, we had to combine responders to both treatments.

As shown in Table 4, before treatment there was not a significantcorrelation of left and right Cd/hem in responders. After success¬ful treatment there was a significant positive correlation betweenleft and right Cd / hem. At this point, since we and others have pos¬tulated pathology in brain circuits involving the caudate, thala¬mus, orbital, and cingulate gyrus brain regions in OCD,511 we ex¬amined correlations among these regions, too (Table 4).

Right orbital cortex/hem was significantly correlated with ip¬silateral Cd/hem and thalamus/hem before treatment but notafter, and the differences before and after treatment were clearlysignificant (P<.05). There was a similar pattern of correlationamong these structures on the left side. Also of interest, there wasno significant correlation between left cingulate/hem and left

ValueBefore

Treatment

ValueAfter

Treatment

L to R caudateL to R orbitL to R thalamusL to R cingulateL caudate to L orbitR caudate to R orbitL orbit to L thalamusR orbit to R thalamusL caudate to L thalamusR caudate to R thalamusL caudate to L cingulateR caudate to R cingulateL cingulate to L orbitR cingulate to R orbitL cingulate to L thalamusR cingulate to R thalamus

.28

.54+

.67+

.17

.49*

.44*

.33

.41*

.28

.21-.10-.04.00.12

-.05-.01

.77*

.59+

.69*

.03

.00-.03-.21-.21

.33

.46+

.41 +

-.03.33.03.00.05

*P<.001 (two-tailed).+P<.01 (two-tailed).+P<.05 (two-tailed).

Cd/hem before treatment, but a clearly significant change to a

significant positive correlation obtained after successful treat¬ment. No other correlations between ROIs were calculated.Numbers of nonresponders and controls were too small for validcorrelation coefficients.

COMMENTAlthough subject numbers are small, and our findings

clearly in need of replication, we were able to provide evi¬dence that glucose metabolic rates in the right head of thecaudate nucleus change when OCD is treated successfullywith either fluoxetine or behavior therapy. Normal subjectsscanned twice under similar conditions hours to days aparthave not shown such changes in other laboratories,37,38 nor

did our small group of normal subjects with a time intervalbetween scans similar to that of our patients with OCD.

We also found evidence for significant correlations oforbital cortex activity with both the caudate nucleus andthe thalamus before treatment in treatment responders.These correlations disappeared with successful treatment.Left cingulate glucose metabolism was not significantlycorrelated with left caudate nucleus activity before treat¬ment but was significantly correlated after treatment.

Although half of our primary a priori hypothesis con¬

cerning the caudate nuclei was fulfilled, we did not provideevidence that left Cd/hem changes with effective OCDtreatment. We had expected similar findings for both leftand right Cd/hem. This seems all the more confusing inthat Benkelfat et al16 observed a statistically significant de¬crease in the normalized left caudate nucleus of OCD re¬

sponders vs nonresponders, but did not do so for the rightcaudate nucleus in a group of patients with OCD under¬going FDG-PET before and after clomipramine hydrochlo¬ride treatment. They did, however, find a similar mean per¬centage decrease in right caudate nucleus glucose meta-


bolic rates in responders (—7.4% ± 18.6%) vs nonresponders(0.07% ±8.7%), as we did, but the variance in their data forthe right caudate nucleus was much greater than that ob¬tained for the left caudate nucleus, and, thus, results were

not significant for the right side. Clomipramine has signif¬icant direct interactions with the dopamine system16-39 inaddition to its better-known serotonin reuptake blockadeproperties.2,40 Both dopamine agonists and antagonistshave been shown to have strong effects on caudate nucleusglucose metabolic rates, and Benkelfat et al acknowledgedthat such effects, unrelated to OCD symptom response,could have confounded their caudate nucleus findings. Flu¬oxetine itself, however, has indirect effects on the dopaminesystem, as should all agents affecting serotonin.41

Likewise, in our data, despite the fact that some casesthat showed a clear response-related change in the rightCd/hem had a visible lack of similar effects on the left (Fig2), it should be emphasized that we did observe an overallmean decrease in left Cd/hem with both OCD treatments,even though this change was not statistically significant.Also, our correlational analysis implicates the left caudatenucleus almost as strongly as the right.

Our study findings should not be taken as evidenceagainst the findings of Benkelfat et al.16 Both studies had a

high probability of Type II error, and taken together they are

complementary in showing a similar change in caudate nu¬cleus glucose metabolism after treatment with two chemi¬cally different drugs and after behavior therapy. It is note¬worthy, however, that Hollander et al42 found increasedright compared with left brain neurologic soft signs in theirpopulation of patients with OCD, perhaps implying thatright brain function is more disordered than left in OCD.

Type II error is also possible in the evaluation of the otherbrain regions we surveyed. Surrounded by white matterand ventricular space, the head of the caudate nucleus isone of the easiest brain regions to determine visually withhigh-resolution FDG-PET scanning. The boundaries ofmost of the other brain regions we examined are not asdistinct. Both Benkelfat et al16 and Swedo et al34 report sig¬nificant changes in normalized orbital cortex after OCDtreatment, yet we note that only subregions of orbital cor¬tex showed this change; we did not attempt such sub¬divisions. We did find significant changes in correlationsbetween orbital gyrus and caudate nucleus and thalamusbefore and after successful treatment, however. We alsonote that Swedo et al did not find caudate nucleus changeswith OCD treatment. Differences in treatment durationmight account for these seeming disparities among thethree studies of OCD before and after treatment. We havesuggested that orbital brain function changes with OCDtreatment would occur some time after caudate nucleuschanges.8 Further, work in our laboratory43 has demon¬strated that as an individual learns to perform a motor taskmore efficiently, the critical brain structures mediating thebehavior show a reduction in both spatial extent and mag¬nitude of activation on PET scanning while performing thetask than when the task was new. With time, might the cau¬

date nucleus become "more efficient" in controlling OCDsymptoms and its change(s) in critical functions no longerbe detectable with present PET methods? Swedo et al res-tudied their subjects after at least 1 year (mean, 20 months),Benkelfat et al did so after a mean of 16 weeks, and we didso after only 10 weeks. It is interesting that the study of in¬termediate length found both caudate nucleus and orbitalchanges, while the shorter and longer ones found only cau-

date nucleus and orbital changes, respectively. Whetherthose who respond to OCD treatment show various re¬

gional brain changes before, at the time of, or after clinicalresponse could be investigated with FDG-PET.

We also found decreases in right anterior cingulate gyrus/hem and left thalamus/hem with successful drug treatmentbut not behavior therapy. These findings may be specific toOCD treatment with fluoxetine vs exposure and response-prevention, but could also be related to other factors in thosepatients who chose drug treatment over behavior therapy.Other studies are needed to resolve this issue.

In our first report of FDG-PET findings in OCD, we re¬

ported increases in both left and right Cd/hem ratios aftertreatment with trazodone hydrochloride and tranyl-cypromine sulfate.5 However, most of those subjects hadmajor depression as well as OCD. We had previously re¬

ported a similar increase in Cd/hem with treatment ofunipolar major depression44 and observed that the changein depression scores in the depressed patients with OCDgave a significant negative correlation with the change inCd/hem, while the change in the OCD scale used did not.5Nevertheless, we associated this increase in Cd/hem withimprovement in OCD symptoms, not just depression.5 Wenow believe that that conclusion was in error and that boththe increases in Cd/hem and the decreased OCD symp¬toms observed in that study population were secondary toan improvement in depression, rather than a primary im¬provement in OCD per se. Obsessive-compulsive disorderoften worsens when major depression is superimposed.4546The findings of Benkelfat et al16 in a group of nondepressedpatients with OCD, and our own failure to show significantimprovement in a placebo and doxepin hydrochloridecontrolled treatment study of trazodone hydrochloride innondepressed patients with OCD (L.R.B., J. M. Thompson,MD, and J. M. Schwartz, MD, unpublished data, 1990) ledus to suspect this error. Consequently, we used a two-tailed hypothesis and statistics in the present study.

It should be pointed out, however, that a decrease inCd/hem does not imply a "decrease" in some critical cau¬date nucleus "function." There are so many interacting ex¬

citatory and inhibitory circuits in the caudate nucleus47 thatall one can say is that there is a change in function; thecritical element(s) in the behavioral mediation may be ei¬ther increasing or decreasing.

Known functions of the caudate nucleus, however, doseem to fit the symptomatology of OCD. Rapoport andcolleagues11 have pointed out the similarities between thebehaviors commonly seen in compulsive rituals and in¬nate, species-specific behavioral routines that may bereleased when the caudate nucleus is dysfunctional. (Also,see Villabianca and Olmstead48 for a detailed review ofrelevant behaviors in cats with caudate nucleus lesions.)Another basal ganglia function, "gating," by which certainmotor, sensory, and perhaps cognitive impulses are eitherallowed to proceed through to perception and behavior orare held back ("filtered") and dissipated, seems to speak tothe psychodynamic concept of disordered "repression" inOCD.8 In this regard, it seems appropriate to note that in¬dividuals likely to develop Huntington's disease, whohave Cd/hem values that are significantly below those ofnormals, have increased expressions of "anger and hostil¬ity" on the Profile of Mood States compared with siblingswith normal Cd/hem values who are less likely to developthis neurologic disorder.49 Obsessive-compulsive disordersymptoms have often been viewed as the result of attempts


to overcontrol these same emotions. Further, specifically inregard to the localization of this brain change after behav¬ior therapy for OCD, there is now a wealth of data impli¬cating the caudate nucleus in "procedural" or "processlearning," including the ability to acquire new habits andskills necessary for the successful initiation of approach oravoidance behaviors.5053

Although our findings are consistent with the idea thatthe head of the caudate nucleus is involved in the media¬tion of OCD's symptomatic expression,8 and there is an at¬tractive theory as to how caudate nucleus dysfunction maydevelop in OCD and related disorders,54 our data do notprove that caudate nucleus dysfunction is the "cause" ofOCD. That may be further up the afferent stream. Indeed,brain glucose metabolic rates largely reflect the work ofneuronal firing at synaptic nerve terminals, and not themetabolic demands of cell bodies in and processes efferentfrom a structure.55 The orbital region of the brain has beendemonstrated to be abnormal in OCD with FDG-PET5'33'56,57and sends extensive efferent projections to both the headof the caudate nucleus and the thalamus.7 Our correla¬tional analysis of brain region activity suggests that insymptomatic, treatment-responsive OCD activity in theorbital region is closely coupled with that in the caudatenucleus and thalamus, while with symptom improvementthese relationships are broken, perhaps through a changein caudate nucleus function.

Although it seems hazardous to tally pluses and minusesamong the multiple excitatory and inhibitory elements inorbital-basal ganglia-thalamic circuits to predict a final out¬come, we will now invoke our data to support proposed the¬ories6,7,910 of OCD symptom mediation by elements in thesesame circuits6"11 (Fig 358). According to our present view, be¬fore effective treatment a deficit in caudate nucleus functionleads to inadequate "filtering" (or "repression") of orbital"worry" inputs, and thus allows them to drive the relevantfraction of inhibitory caudate nucleus output to the globuspallidus. This results in decreased pallidal inhibitory outputto relevant parts of the thalamus. Consequently, orbitalworry inputs may also come to have an undue, rigid influ¬ence on thalamic outputs to other brain regions mediatingOCD symptoms that are not localized to the orbit. Further,excitatory thalamic input to the orbit makes this a potentiallyself-reinforcing loop that is difficult to "break." With effec¬tive treatment, however, adequate filtering activity in thecaudate nucleus damps out this self-driving circuitry, andthereby allows the individual to better limit OCD symptomsonce an obsessive worry is "launched" from the orbital re¬

gion. These orbital-basal ganglia-thalamic circuits may haveevolved to allow significant threats involving OCD themes(violence, contamination, etc) to capture and direct attentionfor needed action—and to rivet behavior to those concernsuntil the danger is judged passed. In treatment-responsiveOCD the threshold for system "capture" may be too low.Caution is warranted about the generalizability of theseideas, however, since we do not have adequate subject num¬bers to examine drug treatment and behavior therapy sep¬arately or to comment on treatment nonresponders. In ad¬dition to further human PET work, PET studies and simul¬taneous electrophysiologic recordings from these samebrain regions in animal models of OCD, before and after ef¬fective drug treatment,59 are warranted.

Based on reported effects of cingulotomy in OCD60 andtheory put forward by others,6 we expected significantcingulate-caudate correlations before but not after effective

Relevant Outputto Other Brain

Regions

Thalamus (+)

Io Block FromTreatment

Relevant Outputto Other Brain

2° Block-

Fig 3.—Although known connections among brain systems proposed tobe involved in obsessive-compulsive disorder (OCD) symptom media-f/on6'7·10 (top) are complex,™ a subsystem (top) maybe the locus ofpatho¬logic dysfunction. Pretreatment correlations suggest that "worry" outputsfrom the orbital (Orb) region of the brain may be driving OCD-relevantcircuits in the caudate nucleus (Cd), and thus increase inhibitory outputto relevant regions of the globus pallidus (GP). This would reduce inhi¬bition of the thalamus, making it vulnerable to being driven by the orbitas well. The excitatory connections between thalamus and orbit makethis a potentially self-sustaining circuit, and thus difficult to break. Witheffective treatment (bottom), increases in filtering functions in the caudatenucleus may reduce its inhibitory output to the globus pallidus, which inturn would increase inhibitory output to the thalamus. This would resultin an uncoupling of this fixed "worry circuit" and allow the patient tomore easily terminate OCD behaviors. With time, orbital "worry activity"may also decrease. Arrows indicate effect ofstimulating one brain regionon the rate of neuronal firing in the other; plus signs, excitatory output;minus signs, inhibitory output; and broken lines, reduced effect.

treatment. Instead, we observed the opposite. This mayhave been a chance occurrence, although we have someideas as to how cingulate-mediated "will" and/or "sup¬pression" functions might be operating.22 Such discourse isbeyond space limitations here.

A significant input to the caudate nucleus comes from se¬rotonin neuron bodies in the dorsal raphe.61 Serotonin ishighly implicated in the pathophysiology of OCD.40 We be¬lieve it unlikely, however, that changes in the activity ofthese serotonin terminals themselves are what is beingmeasured directly when Cd/hem changes with successfultreatment of OCD. These terminals are but a tiny fractionof those in the caudate nucleus,61 and changes in glucoseutilization in serotonin terminals alone are thus not likelyto be directly responsible for as large a percentage change


in Cd/hem as measured in our study. A more specific PETneurotracer than FDG may be needed to detect changes inactivity at serotonin synapses. (A comparable situation ex¬ists with dopamine terminals in Parkinson's disease whereFDG-PET does not show consistent striatal abnormalitiesbut 18F-fluoro-DOPA-PET does.62) It seems more likely thatthe changes in Cd/hem we observed are a reflection ofchanges in the activity of one or more of the many neuro-

chemically diverse populations of small interneurons in thecaudate nucleus.61 Our correlational analysis suggests thepossibility, however, that loss of a key modulating processin the caudate nucleus may allow inputs from the orbitalregion to drive interneuron activity in the caudate nucleusand that with successful treatment adequate regulatorycontrol is (re)established. Serotonin agonists have generallybeen observed to decrease brain glucose metabolic rates inanimals, but effects in the caudate nucleus are complex, pre¬sumably dependent on the classes of serotonin receptorsthat may be most affected by any given agent, and are noteasily interpreted.63 Critical interneuron function changeswith OCD treatment could be effected by serotonin affér¬ents to the caudate nucleus acting on other, larger popu¬lations of neurons, but there are many possible candidatesfor the title of "critical element" in the neurochemical mo¬saic of the caudate nucleus.

The "serotonin hypothesis" of OCD rests most firmly onevidence that chemically diverse drugs that are strong se¬rotonin re-uptake inhibitors are effective in the treatment ofOCD, while similar agents that affect other neurotransmit-ters are ineffective.40,64,65 Some may wonder how behaviortherapy could produce brain function changes similar toneurochemically specific drugs. Even in lower animals,such as the sea slug, Aplysia, however, it is changes at syn¬apses that use serotonin that seem to mediate learnedchanges in stimulus-response behavior (see Kandel66 for re¬

view). Further, direct applications of serotonin at these syn¬apses can produce the same lasting changes in synapticfunction and behavior as seen with behavior modificationin Aplysia.66,67 Thus, the possibility of both a serotonin re¬

uptake inhibitor and behavior modification treatmentshaving the same neural effects is not as farfetched as itmight seem to some at first glance.

Unipolar major depression6870 and the panic/agorophobia disorder complex71"75 can also be treated witheither drugs or behavior therapy, although, as in OCD, cer¬tain symptoms may respond better to one intervention thanto another. Positron emission tomography has providedsignificant, replicated findings in mood disorders (see Bax¬ter76 for review) and anxiety disorders.5,33,56,57,7679 Positronemission tomographic studies that compare and contrastthe effects of these overtly different treatment interventionsmay provide a powerful method for investigating the brainmediation of symptoms in these disorders, too.

This investigation was supported in part by contract AM03-76-SF00012 from the US Dept of Energy, grant MH37916, Research Sci¬entist Development Award KO1-MH00752 (Dr Baxter) and AcademicCareer Development Award KO7-MH00912 (Dr Guze) all from theNational Institute of Mental Health, Bethesda, Md; US Public HealthService-National Institute of Mental Health training grantT32MH19140 (Dr Szuba); funds from the National Defense MedicalCenter, Republic of China (Dr Ferng); National Alliance for Researchon Schizophrenia and Depression Young Investigator (Drs Guze andSzuba) and Established Investigator (Dr Baxter) awards; donationsfrom the Jennifer Jones Simon Foundation, and the Judson BraunChair in Psychiatry, UCLA.

We are indebted to Eva Allen and Juan Barrio for clerical assistance;

. Satyamurthy, PhD, and his cyclotron staff, and Jorge Barrio, PhD,and his chemistry staff for synthesis of FDG; Ron Sumida, LawrencePang, Francine Aguilar, Gloria Stocks, Anne Irwin, Judy Edwards,Der-Jen Liu, Mike Raghian, and Mark Hulgan for technical assistancewith scanning; Goran Rosenqvist, Ann Echelmeyer, and Ron Sumidafor other technical assistance; and Lee Griswold and Wendy Wilsonfor help with the preparation of illustrations. We would also like tothank Monte Buchsbaum, MD, and Lynn Fairbanks, PhD, for sugges¬tions for further data analyses.

References1. Jenike MA, Baer L, MinichielloW WE, eds. Obsessive-Compulsive Disor-

ders: Theory andManagement. 2nd ed. St Louis, Mo: Mosby-Year Book; 1990.2. DeVeaugh-Geiss J. Pharmacologic treatment of obsessive-compulsive

disorder. In: Zohar J, Insel T, Rasmussen S, eds. The Psychobiology ofObsessive-Compulsive Disorder. New York, NY: Springer Publishing Co Inc;1991:187-207.

3. Baer L, Minichiello WE. Behavior therapy for obsessive-compulsivedisorder. In: Jenike MA, Baer L, Minichiello WE, eds. Obsessive-CompulsiveDisorders: Theory and Management. 2nd ed. St Louis, Mo: Mosby-YearBook; 1990:203-232.

4. Cummings JL, Frankel M. Gilles de la Tourette's syndrome and theneurological basis of obsessions and compulsions. Biol Psychiatry. 1985;20:1117-1126.

5. Baxter LR, Phelps ME, Mazziotta JC, Guze GH, Schwartz JM, Selin CE.Local cerebral glucose metabolic rates in obsessive-compulsive disorder: a

comparison with rates in unipolar depression and normal controls. Arch GenPsychiatry. 1987;44:211-218.

6. Rapoport JL, Wise SP. Obsessive-compulsive disorder: is it a basal gan-glia dysfunction? Psychopharmacol Bull. 1988;24:380-384.

7. Modell JG, Mountx JM, Curtis GC, Greden JF. Neurophysiologic dys-function in basal ganglia/limbic striatal and thalamocortical circuits as apathogenetic mechanism of obsessive-compulsive disorder. J Neuropsychi-atry. 1989;1:27-36.

8. Baxter LR, Schwartz JM, Guze BH, Bergman K, Szuba MP. Neuroimagingin obsessive-compulsive disorder: seeking the mediating neuroanatomy. In:Jenike MA, Baer L, Minichiello WE, eds. Obsessive-Compulsive Disorders: The-ory andManagement. 2nd ed. St Louis, Mo: Mosby-Year Book; 1990:167-188.

9. Insel TR. Obsessive-compulsive disorder: a neuroethological perspec-tive. Psychopharmacol Bull. 1988;24:365-369.

10. Insel TR, Winslow JT. Neurobiology of obsessive-compulsive disorder.In: Jenike MA, Baer L, Minichiello WE, eds. Obsessive-Compulsive Disorders:Theory and Management. 2nd ed. St Louis, Mo: Mosby-Year Book; 1990:116\x=req-\131.

11. Rapoport JL. Recent advances in obsessive-compulsive disorder.Neuropsychopharmacology. 1991;5:1-10.

12. Phelps ME, Mazziotta JC. Positron emission tomography: human brainfunction and biochemistry. Science. 1985;228:799-809.

13. Pigott TA, Pato MT, Bernstein SE, Grover GN, Hill JL, Tolliver TJ, MurphyDL. Controlled comparisons of clomipramine and fluoxetine in the treatmentof obsessive-compulsive disorder. Arch Gen Psychiatry. 1990;47:926-932.

14. Turner SM, Jacob RG, Beidel DC, Himmelhoch J. Fluoxetine treatmentof obsessive-compulsive disorder. J Clin Psychopharmacol. 1985;5:207-212.

15. Jenike MA, Buttolph L, Baer L, Ricciardi J, Holland A. Open trial of flu-oxetine in obsessive-compulsive disorder. Am JPsychiatry. 1989;146:909-911.

16. Benkelfat C, Nordahl TE, Semple WE, King C, Murphy DL, Cohen RM.Local cerebral glucose metabolic rates in obsessive-compulsive disorder:patients treated with clomipramine. Arch Gen Psychiatry. 1990;47:840-848.

17. Spitzer RL, Endicott J. Schedule for Affective Disorders and Schizo-phrenia. New York, NY: New York State Psychiatric Institute; 1978.

18. Goodman WK, Price LH, Rasmussen SA, Mazure C, Fleischmann RL,Hill CL, Heninger GR, Charney DS. The Yale-Brown Obsessive CompulsiveScale, I: development, use, and reliability. Arch Gen Psychiatry. 1989;46:1006-1011.

19. Goodman WK, Price LH, Rasmussen SA, Mazure C, Delgado P,Heninger GR, Charney DS. The Yale-Brown Obsessive Compulsive Scale, II:validity. Arch Gen Psychiatry. 1989;46:1012-1016.

20. Hamilton M. Development of a rating scale for primary depressive ill-ness. BrJ Soc Psychol. 1967;6:278-296.

21. Hamilton M. Diagnosis and rating of anxiety. Br J Psychiatry. 1969;3(special issue):76-79.

22. Schwartz JM, Martin K, Baxter LR. Neuroimaging and cognitive-behavioral self-treatment in obsessive-compulsive disorder: practical andphilosophical considerations. In: Hand I, Goodman W, eds. Obsessive-Compulsive Disorder: Recent Research. New York, NY: Springer-Verlag NYInc. In press.

23. Mazziotta JC, Phelps ME, Miller J. Tomographic mapping of humancerebral metabolism: normal unstimulated state. Neurology. 1981;31:503\x=req-\516.

24. Barrio JR, MacDonald NS, Robinson GD, Najafi A, Cook JS, Kuhl DE.Remote, semiautomated production of F-18-labeled 2-deoxy-2-fluoro-D-glucose. J Nucl Med. 1981;22:372-375.

25. Phelps ME, Huang SC, Hoffman EJ, Selin C, Sokoloff L, Kuhl DE. To-mographic measurement of local cerebral glucose metabolic rate in humans


with (F-18) 2-fluoro-2-deoxyglucose: validation of method. Ann Neurol.1979;6:371-388.

26. Huang SC, Phelps ME, Hoffman SC, Sideris K, Selin CE, Kuhl DE.Noninvasive determination of local cerebral metabolic rate of glucose inman. Am J Physiol. 1980;238:E69-E82.

27. Mazziotta JC, Phelps ME, Meadors AK, Ricci A, Winter J, Bentson JR. An-atomical localization schemes for use in positron computed tomography usinga specially designed headholder. J Comput Assist Tomogr. 1982;6:848-853.

28. Cherry SR, Dahlbom M, Hoffman EJ. 3D PET using a conventional mul-tislice tomograph without septa. J Comput Assist Tomogr. 1991;15:655-668.

29. Reivich M, Alavi A, Wolf AP, Fowler J, Russell J, Arnett C, Mac\x=req-\Gregor RR, Shiue CY, Atkins H, Anaud A, Dahn R, Greenberg JH.Glucose metabolic rate kinetic model parameter determination in man:the lumped constants and rate constants for [18F]fluorodeoxyglucose and[11C]deoxyglucose. J Cereb Blood Flow Metab. 1985;5:179-192.

30. Brooks RA, Hatazawa J, Di Chiro G, Larson SM, Fishbein DS. Humancerebral glucose metabolism determined by positron emission tomography:a revisit. J Cereb Blood Flow Metab. 1987;7:427-432.

31. Reivich M, Kuhl D, Wolf A, Greenberg JH, Phelps ME, Ido T, Cassel laV, Fowler J, Hoffman EJ, Alvani A, Sokoloff L. The [18F]fluorodeoxy-glucosemethod for the measurement of local cerebral glucose utilization in man. CircRes. 1979;44:127-137.

32. Mazziotta JC, Phellps ME, Plummer D. Optimization and standard-ization of anatomical data in neurobehavioral investigations using positronCT. J Cereb Blood Flow Metab. 1983;3(suppl):S266-S267.

33. Swedo SE, Schapiro MB, Grady CL, Cheslow DL, Leonard HL, KumarA, Friedland RP, Rapoport SI, Rapoport JL. Cerebral glucose metabolism inchildhood-onset obsessive-compulsive disorder. Arch Gen Psychiatry. 1989;46:518-523.

34. Swedo SE, Pietrini P, Leonard HL, Schapiro MB, Rettew DC, Gold-berger EL, Rapoport SI, Rapoport JL, Grady CL. Cerebral glucose metabolismin childhood-onset obsessive-compulsive disorder: revisualization duringpharmacotherapy. Arch Gen Psychiatry. In press.

35. Hayes WL. Statistics. New York, NY: Holt Rinehart & Winston; 1963.36. The Clomipramine Collaborative Study Group. Clomipramine in the

treatment of patients with obsessive-compulsive disorder. Arch Gen Psychi-atry. 1991;48:730-738.

37. Bartlett EJ, Brodie JD, Wolf AP, Christman DR, Laska E, Meissner M.Reproducibility of cerebral glucose metabolic measurements in resting hu-man subjects. J Cereb Blood Flow Metab. 1988;8:502-512.

38. Bartlett EJ, Barouche F, Brodie JD, Wolkin A, Angrist B, Rotrosen J,Wolf AP. The stability of resting deoxyglucose metabolic values in PET stud-ies of schizophrenia. Psychiatry Res. 1991;40:11-20.

39. Austin LS, Lydiard RB, Ballenger JC, Cohen BM, Laraia MT, Zealberg JJ,Fossey MD, Ellinwood EH. Dopamine blocking activity of clomipramine in pa-tients with obsessive-compulsive disorder. Biol Psychiatry. 1991;30:225-232.

40. Zohar J, Insel TR. Obsessive-compulsive disorder: psychobiologicalapproaches to diagnosis, treatment, and pathophysiology. Biol Psychiatry.1987;22:667-687.

41. Lipinski JF, Mallya G, Zimmerman P, Pope HG. Fluoxetine-inducedakathisia: clinical and theoretical implications. J Clin Psychiatry. 1989;50:339-342.

42. Hollander E, Schiffman E, Cohen B, Rivera-Stein MA, Rosen W, GormanJM, Fyer AJ, Papp L, Liebowitz MR. Signs of central nervous system dysfunctionin obsessive-compulsive disorder. Arch Gen Psychiatry. 1990;47:27-32.

43. Mazziotta JC, Grafton ST, Woods RC. The human motor system stud-ied with PET measurements of cerebral blood flow: topography and motorlearning. In: Lassen NA, Ingvar DH, Raichle ME, Frieberg L, eds. Brain Workand Mental Activity: Quantitative Studies With Radioactive Tracers: Pro-ceedings of the 31st Alfred Benzon Symposium. Copenhagen, Denmark:Munksgaard; 1991:280-293.

44. Baxter LR, Phelps ME, Mazziotta JC, Schwartz JM, Gerner RH, SelinCE, Sumida RM. Cerebral metabolic rates for glucose in mood disorders:atudies with positron emission tomography and 18F-fluorodeoxyglucose.Arch Gen Psychiatry. 1985;42:441-447.

45. Mavissakalian M, Michelson L. Tricyclic antidepressants in obsessive\x=req-\compulsive disorder: antiobsessional or antidepressant agents? J Nerv MentDis. 1983;171:301-306.

46. Fogelson DL, Bystritsky S. Imipramine in the treatment of OCD withand without major depression. Ann Clin Psychiatry. 1991;3:233-237.

47. Penney JB, Young AB. Speculations on the functional anatomy of basalganglia disorders. Annu Rev Neurosci. 1983;6:73-94.

48. Villablanca JR, Olmstead CE. The striatum: a finer tuner of the brain.Acta Neurobiol Exp (Warsz). 1982;42:227-299.

49. Baxter LR, Mazziotta JC, Pahl JJ, Grafton ST. St George-Hyslop P,Haines JL, Gusella J, Szuba MP, Selin CE, Guze BH, Phelps ME. Psychiatric,genetic and PET evaluation of persons at-risk for Huntington's disease. ArchGen Psychiatry. 1992;49:148-154.

50. Phillips AG, Carr GD. Cognition and the basal ganglia: a possiblesubstrate for procedural knowledge. Can J Neurol Sci. 1987;14:381-385.

51. Packard MG, Hirsh R, White NM. Differential effects of fornix andcaudate nucleus lesions on two radial maze tasks: evidence for multiplememory systems. J Neurosci. 1989;9:1465-1472.

52. Mittleman G, Whishaw IQ, Jones GH, Koch M, Robbins TW. Cortical,hippocampal and striatal mediation of schedule-induced behaviors. Behav

Neurosci. 1990;104:399-409.53. Packard MG, White NM. Lesions of the caudate nucleus selectively

impair 'reference memory' acquisions in the radial maze. Behav Neural Biol.1990;53:39-50.

54. Swedo SE, Rapoport JL, Cheslow DL, Leonard HL, Ayoub EM, HosierDM, Wald ER. High prevalence of obsessive-compulsive symptoms inpatients with Sydenham's chorea. Am J Psychiatry. 1989;146:246-249.

55. Mazziotta JC, Phelps ME. Positron emission tomography studies of thebrain. In: Phelps ME, Mazziotta JC, Schelbert HR, eds. Positron Emission To-mography and Autoradiography. New York, NY: Raven Press; 1986.

56. Baxter LR, Schwartz JM, Mazziotta JC, Phelps ME, Pahl JJ, Guze BH,Fairbanks L. Cerebral glucose metabolic rates in nondepressed patients withobsessive-compulsive disorder. Am J Psychiatry. 1988;145:1560-1563.

57. Nordhal TE, Benkelfat C, Semple WE, Gross M, King AC, Cohen RM.Cerebral glucose metabolic rates in obsessive compulsive disorder. Neu-ropsychopharmacology. 1989;2:23-28.

58. Albin RL, Young AB, Penney JB. The functional anatomy of basal gan-glia disorders. Trends Neurosci. 1989;12:366-375.

59. Rapoport JL, Ryland DH, Kriete M. Drug treatmentof canine acral lick:an animal model of obsessive-compulsive disorder. Arch Gen Psychiatry. Inpress.

60. Jenike MA, Bear L, Ballantine HT, Martuza RL, Tynes S, Giriunas I,Buttolph L, Cassem NH. Cingulotomy for refractory obsessive-compulsivedisorder: a long-term follow-up of 33 patients. Arch Gen Psychiatry. 1991;48:548-555.

61. Greybiel AM, Ragsdale CW. Biochemical anatomy of the striatum. In:Emson PC. Chemical Neuroanatomy. New York, NY: Raven Press; 1983:427-504.

62. Leenders KL, Salmon EP, Tyrrell P, Perani D, Brooks DJ, Sager H, JonesT, Marsden CD, Frackowiak RSJ. The nigrostriatal dopaminergic system as-sessed in vivo by positron emission tomography in healthy volunteer subjectsand patients with Parkinson's disease. Arch Neurol. 1990;47:1290-1298.

63. Grome JJ, Harper AM. Local cerebral glucose utilization followingindoleamine- and piperazine-containing 5-hydroxytryptamine agonists. JNeurochem. 1986;46:117-124.

64. Leonard HL, Swedo SE, Koby EV, Rapoport JL, Lengne MC, CheslowDL, Hamburger SD. Treatment of obsessive-compulsive disorder with clomi-pramine and desmethylimipramine in children and adolescents: a double-blind crossover comparison. Arch Gen Psychiatry. 1989;46:1088-1092.

65. Goodman WK, Price LH, Delgado PL, Palumbo J, Krystal JH, Nagy LM,Rasmussen SA, Heninger GR, Charney DS. Specificity of serotonin reuptakeinhibitors in the treatment of obsessive-compulsive disorder: comparison offluvoxamine and desipramine. Arch Gen Psychiatry. 1990;47:577-585.

66. Kandel ER. Genes, nerve cells, and the remembrance of things past.J Neuropsychiatry. 1989;1:103-125.

67. Montarolo PG, Goelet P, Castellucci VF, Morgan J, Kandel ER,Schacher S. A critical period for macromolecular synthesis in long-term heterosynaptic facilitation in Aplysia. Science. 1986;234:1249-1254.

68. Elkin I, Shea MT, Watkins JT, Imber SD, Sotsky SM, Collins JF, GlassDR, Pilkonis PA, Leber WR, Docherty JP, Fiester SJ, Parloff MB. National In-stitute of Mental Health treatment of depression collaborative research pro-gram: general effectiveness of treatments. Arch Gen Psychiatry. 1989;46:971-982.

69. Frank E, Kupfer DJ, Perel JM, Cornes C, Jarrett DB, Mal linger AG, ThaseME, McEachran AB, Grochocinsky VJ. Three-year outcomes for maintenancetherapies in recurrent depression. Arch Gen Psychiatry. 1990;47:1093-1099.

70. Klerman GL. Treatment of recurrent unipolar major depressive disorder:commentary on the Pittsburgh study. Arch Gen Psychiatry. 1990;47:1158\x=req-\1162.

71. Mavissakalian M, Michelson L. Agoraphobia: relative and combinedeffectiveness of therapist-assisted in vivo exposure and imipramine. J ClinPsychiatry. 1986;47:117-122.

72. Mavissakalian M. Imipramine in agoraphobia. Compr Psychiatry.1986;27:401-406.

73. Ballenger JC, Burrows GD, DuPont RL Jr, Lesser IM, Noyes R Jr, Pec-knold JC, Rifkin A, Swinson RP. Alprazolam in panic disorder and agorapho-bia: results from a multicenter trial, I: efficacy in short-term treatment. ArchGen Psychiatry. 1988;45:413-422.

74. Marks IM, Gray S, Cohen D, Hill R, Mawson D, Ramm E, Stern RS.Imipramine and brief therapist-aided exposure in agoraphobics having self\x=req-\exposure homework. Arch Gen Psychiatry. 1983;40:153-162.

75. Salkovskis PM, Jones DRO, Clark DM. Respiratory control in thetreatment of panic attacks: replications. Br J Psychiatry. 1986;148:526-532.

76. Baxter LR. PET studies of cerebral function in major depression andobsessive-compulsive disorder: the emerging prefrontal cortex consensus.Ann Clin Psychiatry. 1991;3:103-109.

77. Reiman EM, Raichel ME, Butler FK, Herscovitch P, Robins E. A focalbrain abnormality in panic disorder, a severe form of anxiety. Nature. 1984;310:683-685.

78. Reiman EM, Raichle ME, Robins E, Minton MA, Fusselman MJ, Fox PT,Price JL, Hackman KA. Neuroanatomical correlates of a lactate-inducedanxiety attack. Arch Gen Psychiatry. 1989;46:493-500.

79. Nordahl TE, Semple WE, Gross M, Mellman TA, Stein MB, Goyer P,King AC, Uhde TW, Cohen RM. Cerebral glucose metabolic differences inpatients with panic disorder. Neuropsychopharmacology. 1990;3:261-272.


caudate glucose metabolic rate changes with both drug and bt for ocd (baxter, 1992)

Documents