Presynaptic Dopaminergic Dysfunction in Schizophrenia:  A Positron Emission Tomographic [¹⁸F]Fluorodopa Study FREE

The original dopamine (DA) hyperactivity hypothesis of schizophrenia,¹ although undergoing revisions and reformulations,^2- 3 remains a pivotal neurochemical hypothesisof the illness that is yet to be fully confirmed or refuted. Multiple componentsof dopaminergic neurotransmission may cause dopaminergic overactivity, includingincreased DA synthesis, release, receptor number and/or affinity, and DA-mediatedpostsynaptic effector mechanisms. Certain of these processes can be quantifiedin vivo in patients, allowing a more comprehensive test of the DA overactivityhypothesis than was previously available.

To date, most imaging studies have examined DA receptor changes in theillness, but positron emission tomography (PET) and single-photon emissioncomputed tomographic studies of DA D₁ and D₂ receptorshave produced equivocal results. Recently, prefrontal D₁ receptorshave been reported to be elevated, unchanged, or lowered in the illness,^4- 6 while 2 meta-analysesof striatal D₂ receptor changes^7- 8 suggestthat increases of D₂ receptors, if present in the illness, areof much smaller magnitude than the reported postmortem changes, where theconfounding effect of neuroleptic exposure is more difficult to control. Directin vivo evidence of enhanced DA release has been more consistently obtainedin the recent studies by Laruelle et al^9- 11 andBreier et al.¹² These authors used the single-photonemission computed tomographic tracer [¹²³I]iodobenzamide and thePET radiotracer [¹¹C]raclopride to index amphetamine-induced DArelease in the striatum of patients with schizophrenia. Binding of these radiotracersto striatal DA D₂ receptors is sensitive to endogenous levels ofDA (see Laruelle¹⁰ for comprehensive review)such that increases of synaptic DA decrease the radiotracer's specific bindingand decreases of DA increase specific binding. Patients with schizophreniashow greater reductions of striatal [¹²³I]iodobenzamide bindingor [¹¹C]raclopride after amphetamine challenge compared with healthyvolunteers,^11- 12 a finding compatiblewith enhanced DA release. Recently, these findings have been replicated ina second cohort of patients with schizophrenia.¹³ Furthermore,worsening of positive symptoms in patients correlated with the degree of DArelease, estimated by striatal [¹²³I]iodobenzamide displacement.Finally, after pretreatment with α-methyl-para-tyrosine (AMPT), whichlowers synaptic DA levels, with the radiotracer[¹²³I]iodobenzamide,an increased "DA D₂ availability" has been observed in patientscompared with control subjects,¹⁴ implyingthat synaptic levels of "baseline" DA may be high in the illness.^15- 16

A complementary method of imaging presynaptic dopaminergic functionunder baseline conditions measures the formation and storage of DA in presynapticterminals with [¹⁸F]fluorodopa, a radioactive analogue of L-dopa, the precursor of DA. [¹⁸F]fluorodopa is taken upby presynaptic monoaminergic neurons and is metabolized by aromatic acid decarboxylase(AADC) to [¹⁸F]fluorodopamine ([¹⁸F]DA), which is trappedand stored within vesicles in the nerve terminals. [¹⁸F]fluorodopauptake, quantified as the influx constant K_i, measures AADC activity and vesicular storage capacity.¹⁷ Highvalues for [¹⁸F]fluorodopa K_i areobserved in areas of dense DA nerve terminal innervation (eg, the striatum),and [¹⁸F]fluorodopa uptake correlates with surviving nigrostriatalcell numbers in both monkey and human studies.^18- 19 [¹⁸F]fluorodopa has been extensively used to probe the structural andfunctional integrity of striatal dopaminergic neurons, particularly in Parkinsondisease and other movement disorders.^20- 23

To date, 6 studies using [¹⁸F]fluorodopa and 1 using dopalabeled with carbon 11 ([¹¹C]dopa) have been reported in patientswith schizophrenia. Five of the 7 studies describe elevated dopa metabolismin the striatum,^24- 28 whereasno difference was detected by Dao-Castellana and colleagues²⁹ and1 study reported reduced [¹⁸F]fluorodopa striatal uptake.³⁰ Elevations of [¹⁸F]fluorodopa and [¹¹C]dopa in the striatum were observed in both medication-naive andmedication-free patients. The total numbers of schizophrenic patients studiedwith [¹⁸F]fluorodopa and [¹¹C]dopa to date have beenlimited, however (n = approximately 50), with 5 to 12 patients per study.Furthermore, quantification of cortical [¹⁸F]fluorodopa uptakehad not been attempted or achieved, in most studies probably because of thesmall signal magnitude in cortical areas and/or sensitivity of the previousgeneration of PET cameras.

Thus, the available in vivo imaging evidence points to alterations ofpresynaptic striatal dopaminergic function in schizophrenia. However, it isnot clear how presynaptic striatal dopaminergic dysfunction is related, ifat all, to mesocortical dopaminergic function in schizophrenia or to activityin cortical areas such as the prefrontal cortex. An inverse reciprocal relationshiphas been postulated between mesocortical (particularly prefrontal) DA systemsand striatal DA systems,^31- 33 withprefrontal DA lesions giving rise to enhanced subcortical DA activity in someanimal studies. Also, prefrontal lesions modulate striatal DA function insome but not all studies (for full discussion, see Grace³⁴ andCarlsson et al³⁵). Pertinent to these issues,the most recent [¹⁸F]fluorodopa study demonstrated a negative correlationbetween elevated striatal [¹⁸F]fluorodopa K_i and prefrontal activation in patients with schizophrenia.²⁸ Although small (n = 6), this study provided one ofthe first in vivo demonstration of the hypothesized link between prefrontaldysfunction and altered subcortical DA function in the illness.

In the present study, we extend previous studies of [¹⁸F]fluorodopain schizophrenia by studying a larger sample of patients (n = 16), with ahighly sensitive PET camera with an optimized [¹⁸F]fluorodopa scanningprotocol. We determined whether elevated striatal [¹⁸F]fluorodopauptake could still be detected in medicated patients with schizophrenia, aspreviously reported in a number of studies of unmedicated patients. Second,we tested the hypothesis that prefrontal [¹⁸F]fluorodopa uptakewould be reduced in the illness. Third, we explored whether regional [¹⁸F]fluorodopa uptake in striatal and cortical areas correlated withpositive or negative symptoms of schizophrenia; we hypothesized that positivesymptoms would correlate with striatal [¹⁸F]fluorodopa uptake whilenegative symptoms would correlate inversely with [¹⁸F]fluorodopauptake in prefrontal regions. Finally, we examined the relationships, if any,between regional [¹⁸F]fluorodopa uptake and performance on neuropsychologicaltests sensitive to the cognitive impairment in schizophrenia.³⁶

The control group comprised 12 right-handed healthy volunteers (agerange, 29-49 years; mean age, 38.3 years; SD, 7.1 years) with normal resultsof neurologic examination. Twenty patients (age range, 25-65 years; mean age,39.9 years; SD, 11.3 years) who met DSM-IV criteriafor schizophrenia were recruited to the patient group from routine hospitaloutpatient clinics. All participants were assessed by a trained psychiatrist(S.M.), and current and past psychiatric morbidity was excluded in healthyvolunteers by routine psychiatric interview (S.M.) and the General HealthQuestionnaire³⁷ with a cutoff of 5 points orless. All patients were taking neuroleptic medication. Left-handed participantswere not excluded from the patient group, as this may reflect the disorder.Exclusion criteria for both groups were a history of neurologic illness, seriousphysical illness, or substance dependence.

All subjects gave written informed consent after a full explanationof the procedure. Permission to undertake the study was granted by the ethicscommittees of participating hospitals. Approval to administer radiotracerswas obtained from the Administration of Radioactive Substances Advisory Committee,United Kingdom.

A 3-dimensional PET scanner (HR++/966 EXACT; CTI PET Systems, Knoxville,Tenn) was used, which reconstructs high-resolution images of the whole brainfrom 95 planes with a slice spacing of 2.425 mm. In brief, the physical characteristicsof the scanner, which have been described in full elsewhere,³⁸ includea spatial resolution of 4.8 ± 0.2 mm (full-width half-maximum–transaxial)and a sensitivity of 2.6 × 10¹² cps/Ci per milliliter (69cps/Bq per milliliter), which compares favorably with PET cameras used inprevious studies (spatial resolution, 5-7 mm; sensitivity, 1.0 × 10¹¹ to 5.3 × 10¹¹ cps/Ci per milliliter [2.8-14.4 cps/Bqper milliliter^24- 30]).To correct for attenuation, a 5-minute transmission scan was carried out beforeemission scanning, with the use of a 4.05-mCi (150-MBq) cesium 137 rotatingpoint source.

To further enhance specific signal detection, 1 hour before the startof each scan, 150 mg of carbidopa, a peripheral AADC inhibitor, and 400 mgof entacapone, a peripheral catechol-O-methyltransferase inhibitor, were givenorally. These compounds minimize metabolism of [¹⁸F]fluorodopaby peripheral AADC and reduce the formation of radioactively labeled metabolitesof [¹⁸F]fluorodopa by peripheral catechol-O-methyltransferase,which may cross the blood-brain barrier and increase the background cerebralsignal.^39- 40

The radiotracer [¹⁸F]fluorodopa, 2.97 mCi (110 MBq) (range,2.76-3.65 mCi [102-135 MBq]), was administered intravenously over 30 secondswith a scan protocol of 26 time frames starting with a background frame of30 seconds, followed by injection of [¹⁸F]fluorodopa at the beginningof four 60-second frames, three 120-second frames, three 180-second frames,and, finally, fifteen 300-second frames. Participants were positioned in thescanner with the orbitomeatal line parallel to the transaxial plane of thetomograph. Head position was monitored via laser crosshairs and video camera.

For each participant, a structural magnetic resonance (MR) T1 imagewas obtained for coregistration and comparison of any volumetric differencesbetween groups.

Four patients were excluded: 1 patient had bilateral perisylvian atrophyon MR imaging; in 1 patient there were technical problems with the PET scan;and 2 patients had excessive head movement during the PET scan. Data are thusreported for 16 patients and 12 controls.

Premorbid IQ was assessed with the revised version of the National AdultReading Test.⁴¹ Participants completed theverbal fluency (FAS) task,⁴² Symbol Digit ModalitiesTest (SDMT),⁴³ and the Stroop Color-Word Test.^44- 45 The FAS was scored as the total numberof correct words produced in 180 seconds. A second, more sensitive clusterscore was also calculated, providing an index of the production of words withinsemantic subcategories.⁴⁶ The SDMT was scoredas the number of correctly completed targets in 90 seconds. For the Strooptest, the performance measure was an interference score, calculated as thecolor − color-word difference score.⁴⁴ Oneparticipant did not complete the National Adult Reading Test or the FAS becauseEnglish was not his first language. Symptoms during the preceding month wererated on the scan day by a trained psychiatrist (S.M.) using the Present Statesection of the Comprehensive Assessment of Symptoms and History.⁴⁷ Themean sum of positive symptoms was 4.2 (range, 0-14), and the mean sum of negativesymptoms was 6.3 (range, 0-16), yielding overall mean total symptom scoresfor patients of 10.6 (range, 3-28). By means of the subcategories of symptomsin the Comprehensive Assessment of Symptoms and History, patients could beclassified as follows: predominantly positive (n = 1), predominantly negative(n = 1), mixed high (n = 1), and mixed low (n = 13). In addition, the presenceof any abnormal movements was assessed with the Abnormal Involuntary MovementScale, with a mean score of 0.1 in patients (range, 0-2).³⁷

Two methods of image analysis of [¹⁸F]fluorodopa PET scanswere used: (1) statistical parametric mapping (SPM99; Wellcome Departmentof Cognitive Neurology, London, England) and (2) an automated region-of-interest(ROI) approach. One advantage of using more than one image analysis methodis that results, if physiologically valid, should be independent of the imageanalysis performed. All data analysis was performed by one worker (S.M.) ona workstation (Sun Sparc; Sun Microsystems, Silicon Valley, Calif) using imageanalysis software (AnalyzeAVW 3.0 Biomedical Imaging Resource; Mayo Foundation,Rochester, Minn) and in-house software written in Matlab (version 5; The MathWorks,Inc, Natick, Mass), which, by a multiple time graphical approach,⁴⁸ generates parametric images of [¹⁸F]fluorodopainflux rate constants (K_i) or values of K_i for defined ROIs. In both methods, the occipitalcortex was used as a reference region to generate the input function for thePatlak analysis, the region being drawn on MR images coregistered to the PETscans.

A template of [¹⁸F]fluorodopa uptake was created for usewithin the SPM system by using combined images of [¹⁸F]fluorodopaand T1 MR images (details available on request). The SPM results were thresholdedusing a value of P<.05 corrected, with the useof small volume corrections for the striatal and prefrontal regions. The anatomicmask defining striatal areas was taken from Mawlawi et al⁴⁹ andthe prefrontal mask derived from the atlas of Hammers et al,⁵⁰ withan in-house modification to include regions of the prefrontal cortex.

The data were analyzed by means of standardized ROIs drawn with AnalyzeAVWimage analysis software on the representative single participant T1 MR imageavailable in SPM. This T1 image originated from the Montreal NeurologicalInstitute brain database. For the striatal ROI, the volume was subdividedas follows: all planes containing striatal structures below the anterior commissure–posteriorcommissure plane were operationally defined as the ventral striatum ROI, andall planes above the anterior commissure–posterior commissure planewith striatal structures formed the dorsal striatum ROI. For the anteriorcingulate (ACC), in the sagittal orientation, the genu of the corpus callosumwas used as an anatomic landmark to divide the anterior portion of the gyrusinto dorsal and ventral ROIs bilaterally. In the same sagittal orientation,an ROI was defined anterior to the cingulate ROIs for the medial prefrontalcortex. Finally, a unified cerebellar ROI was defined on both hemispheres,extending over 3 planes. In this manner, these ROIs are defined in a knownorientation and a standardized space, which is important for generating ROIswhere arbitrary, rather than anatomically visible, criteria are being used.The [¹⁸F]fluorodopa template used in the SPM analysis was in thesame standardized space as the single-subject T1 MR image. By means of SPM,the [¹⁸F]fluorodopa template was normalized to each individual[¹⁸F]fluorodopa summation image. The individual specific mathematicaltransformation parameters to accomplish this process were then applied tothe ROIs defined on the single-subject T1 MR image, thereby normalizing ROIsto each individual PET scan. With this method, observer bias in defining ROIsfor each individual scan is avoided. These normalized ROIs were then usedto generate values of [¹⁸F]fluorodopa K_i for each individual.

As an a priori hypothesis existed that patients would show increasesin striatal [¹⁸F]fluorodopa K_i anddecreases in prefrontal cortical K_i, resultsof the SPM analysis were thresholded at P<.05corrected for the volumes (striatum and prefrontal cortex) analyzed. For theautomated ROI analysis of K_i values, the2 groups were compared by means of the 2-way t testcomparison for 2 populations for each of the brain regions. P values were set at P<.01 to correct forthe 5 principal ROIs examined (ventral striatum, dorsal striatum, ventralanterior cingulate, dorsal anterior cingulate, and medial prefrontal cortex).For the neuropsychological scores, all data were normally distributed andanalyzed by independent-sample t tests. Pearson product-momentcorrelation coefficients were calculated between the cognitive test scoresand [¹⁸F]fluorodopa K_i valuesin the ROIs.

The SPM analysis showed increases in [¹⁸F]fluorodopa uptake K_i in the striatum of patients compared withcontrols (P<.05 corrected; df = 1,26) (Figure 1, Table 1). The increase in [¹⁸F]fluorodopauptake was predominantly located in the ventral striatum. No other significantdifferences were seen between patients and controls.

With the use of standardized ROIs, only K_i values in the whole striatum and the ventral striatum were increasedin patients compared with controls (striatum: P =.001, t₂₆ = −3.92; ventral striatum: P = .001, t₂₆ = −3.79;dorsal stria tum: P = .09, t₂₆ = −1.77), corroborating the SPM analysis. No significant differenceswere detected in any of the other ROIs between the 2 groups (Figure 2 and Figure 3).

There was no significant difference in age between the control groupand the patients (P = .86, t₂₆ = 0.18) (Table 2). Therewas no significant difference in IQ between controls and patients, althoughIQ was generally lower in patients at a trend level (P =.07, t₂₅ = 1.93). Only one of the patientswas left-handed. Of the 16 patients, 10 were taking atypical antipsychoticmedication and 4 were taking typical neuroleptics. Two patients were receivingboth typical and atypical antipsychotic drugs. Medication dose (expressedas chlorpromazine equivalents) did not correlate with [¹⁸F]fluorodopa K_i in any of the ROIs (striatum: r = 0.06, P = .83, n = 15; ventral striatum: r = 0.2, P = .47, n = 15; dorsalstriatum, r = −0.07, P =.8, n = 15; 1 outlier 4 SDs from the average neuroleptic dose was excludedin this analysis). [¹⁸F]fluorodopa K_i did not correlate with age among both controls and patients, and inthe patient group [¹⁸F]fluorodopa uptake K_i did not correlate with duration of illness or age at onset of illness(data not shown).

Striatal volumes did not differ between patients and controls whetherassessed as whole striatum or as dorsal and ventral components (Table 3) (independent-samples t test: striatum: P = .80, t₂₆ = 0.26;ventral striatum: P = .11, t₂₆ = 1.65; dorsal striatum: P = .23, t₂₆ = −1.22).

Most patients had symptoms of relatively low severity, a reflectionthat most were stable clinically and had been recruited from hospital outpatientclinics. [¹⁸F]fluorodopa uptake K_i in the ROIs was not significantly correlated with either positiveor negative symptom scores, summed from the Comprehensive Assessment of Symptomsand History (positive: striatum: r = −0.03, P = .91, n = 16; ventralstriatum: r = 0.33, P =.21, n = 16; dorsal striatum: r = −0.25, P = .35, n = 16; negative: striatum: r = 0.45, P = .08, n = 16; ventralstriatum: r = 0.22, P =.42, n = 16; dorsal striatum: r = 0.25, P = .35, n = 16). Likewise, [¹⁸F]fluorodopa uptake K_i in prefrontal cortex ROIs did not correlatewith symptom scores (data not shown).

Schizophrenic patients were impaired on both the SDMT and FAS (totaland cluster scores) but showed the same degree of Stroop interference as didcontrols (Table 4). There weresignificant negative correlations between Stroop interference scores and dorsalACC K_i values in both groups (P = .01 for patients and trend-level P = .08for controls), with greater ACC K_i valuesreflecting reduced interference (Table 5). The magnitude of this correlation did not differ between groups(Fisher r to Z transformation, Z = 0.35, P = .72 [2-tailed]).The 2 groups did diverge, however, in the pattern of correlations betweencognitive scores and K_i values in theventral striatum. Thus, FAS cluster scores showed a significant positive correlationwith ventral striatal K_i values in controls,but not in patients, and the 2 correlation coefficients significantly differed(Z = 2.94, P = .004). Forthe SDMT, there was a significant negative correlation with ventral striatal K_i values in patients, but not in controls.Again, these 2 correlation coefficients were significantly different (Z = 2.04, P = .04).

The SPM analysis comparing [¹⁸F]fluorodopa uptake K_i values between patients and controls on a voxel-by-voxelbasis confirmed our hypothesis of striatal K_i increases among patients with schizophrenia. These changes were confinedprimarily to the ventral striatum. The subsequent ROI analyses using automatedmeasures confirmed the SPM findings for the ventral striatum. Thus, 2 independentmethods of image analysis demonstrated a similar finding. The elevation instriatal [¹⁸F]fluorodopa uptake K_i replicates previous findings in medication-naive and medication-freepatients with schizophrenia.^24- 26,28 Inthese previous studies, no distinction was made between ventral and dorsalstriatum, although even in our study with a high-performance camera, partialvolume effects would dictate that up to 30% of a ventral [¹⁸F]fluorodopasignal could be contributed by the dorsal striatum. However, if the increasedventral striatal [¹⁸F]fluorodopa signal observed in this studyhad arisen from dorsal striatal changes alone, proportionately larger localizedchanges in dorsal striatal [¹⁸F]fluorodopa signal would have beenobserved, which was not the case in the SPM and ROI analyses.

Although the exact relationship between striatal [¹⁸F]fluorodopauptake K_i and synaptic DA release is notknown, in Parkinson disease reduced [¹⁸F]fluorodopa uptake K_icorrelates with blunted [¹¹C]raclopridedisplacement with amphetamine,⁵¹ suggestingthat under certain conditions these 2 measures can be functionally related.Whether this relationship pertains in schizophrenia is unknown, although theresults of this study appear compatible with greater DA release with amphetaminechallenge in patients with schizophrenia.¹² Speculatively,enhanced DA release could be a consequence of increased DA synthesis, as suggestedby our results.

Using a high-sensitivity camera and a PET protocol designed to maximize[¹⁸F]fluorodopa signals, we were able to obtain measures of [¹⁸F]fluorodopa K_i in cortical areasin addition to the striatum. In both groups, a detectable signal was foundin the anterior cingulate cortex on ROI analysis. However, no differencesin cortical [¹⁸F]fluorodopa uptake K_i in patients compared with controls could be detected in the ROI analyses.Cortical K_i values were about 25% of striatalvalues but had similar estimates of variability. Thus, although striatal K_i values and variability allowed for a detectionof a 10% increase of signal with 16 participants (power, 80%; α P<.05, 2-tailed test), the corresponding number of participantsfor detecting a similar change in cortical areas was prohibitively large (eg,n = 144 for dorsal anterior cingulate). Thus, without a 30% or greater changein signal, it would not have been possible to detect cortical changes in thisstudy by ROI approaches despite an optimized scanning protocol. Despite suchlimitations, however, Stroop performance correlated with dorsal anterior cingulate[¹⁸F]fluorodopa K_i in bothgroups independently, suggesting that the cortical signal, although of smallmagnitude and highly variable, has physiological relevance for tasks thatare known to invoke anterior cingulate activity.⁵²

Striatal K_i represents a measureof the uptake of [¹⁸F]fluorodopa into presynaptic dopaminergicterminals in the basal ganglia and its conversion by AADC to [¹⁸F]DA,which enters the vesicular compartment. As all the patients in this studywere being treated with antipsychotic medication, the effect of such drugson AADC activity and hence K_i is an importantconsideration in the interpretation of our findings. Although AADC is notthe rate-limiting step in the synthetic pathway for DA, it has been suggestedthat AADC activity may influence the rate of DA synthesis.^53- 54 Inrats, increases in AADC activity in vitro and in vivo have been reported afteracute treatment with DA antagonists,^55- 58 andin anesthetized pigs (n = 3), the estimated dopa decarboxylation rate wasincreased by short-term haloperidol infusion.⁵⁴ Conversely,short-term treatment with the DA agonist apomorphine decreases [¹¹C]dopainflux in monkeys.⁵⁹ Evidence of such effectsin humans, however, is extremely limited. Thus, in the only comprehensivestudy to date, Grunder et al⁶⁰ recently reporteda decrease in [¹⁸F]fluorodopa K₃ in 9 patients with schizophrenia after 5 weeks' treatment with haloperidol,⁶⁰ suggesting that long-term neuroleptic administrationwill tend to decrease AADC activity and hence DA synthesis. Given that ourmedicated patients, like previous unmedicated cohorts, showed elevations of[¹⁸F]fluorodopa K_i, our findingssuggest that neuroleptic treatment has not fully normalized the elevated [¹⁸F]fluorodopa K_i in the illness.The absolute magnitude of the [¹⁸F]fluorodopa K_i increase in our medicated patient sample (striatum K_i, 0.0144 in patients vs 0.0128 in controls)is very similar to that reported in the study closest in design to our owninvolving medication-naive patients (K_i,0.0149 in medication-naive patients vs 0.0129 in controls²⁵).Finally, in the other studies reporting [¹⁸F]fluorodopa K_i increases, all patients were not taking medication andthe majority were drug naive. Therefore, neuroleptic medication or previousexposure does not appear to readily explain the observed increase in [¹⁸F]fluorodopa K_i seen in patientswith schizophrenia. However, further longitudinal studies of patients, onand off neuroleptic medication, may be necessary to fully address this issue.

Another potentially confounding variable in interpreting the elevationof [¹⁸F]fluorodopa K_i is theinfluence of smoking on PET measures of [¹⁸F]fluorodopa uptake.One recent study in healthy volunteers demonstrated increased [¹⁸F]fluorodopa K_i in smokers compared with nonsmokers.⁶¹ Psychiatric patients tend to smoke more than healthyvolunteers,⁶¹ and this was observed in ourdata (25% of controls and 50% of patients). However, [¹⁸F]fluorodopa K_i for smokers and nonsmokers did not differin our study (data available on request).

A further issue that should be considered when our findings are interpretedconcerns possible striatal volumetric differences between patients and controls.Long-term treatment with neuroleptics has been reported to increase striatalvolume in patients with schizophrenia.⁶² Potentially,this might alter the [¹⁸F]fluorodopa K_i signal in patients via partial volume effects. However, when striatalROI volumes drawn on the T1 MR images by an investigator blind to diagnosiswere used to compare patients against controls, no difference in striatalvolume was found (Table 3) (patientvs control striatum: P = .23, t₂₆ = −1.22).

Elevated [¹⁸F]fluorodopa K_i values did not correlate with symptom ratings in our study. However,all the patients were relatively stable in terms of their psychopathologyand the severity of their symptoms was mild for most of the group; thus, correlationsmay have been underpowered because of the low variance of symptom severity,perhaps induced by neuroleptic treatment. However, Hietala et al²⁶ demonstratedcorrelations of [¹⁸F]fluorodopa only with depressive symptoms intheir drug-naive cohort, suggesting that neuroleptic treatment is not a significantconfound in the lack of correlation with symptomatology. Interestingly, Laruelleand Abi-Dargham⁶³ observed that the increasedDA release demonstrated by [¹²³I]iodobenzamide displacement isonly present in patients who are currently relapsing, suggesting that differingmeasures of presynaptic DA function may be dissociable at the level of symptomatologyor overall clinical state (stable vs relapsing).

A number of investigators have proposed a role for dysfunction in DAsystems in the core cognitive deficits thought to underlie schizophrenic symptomatology,^64- 66 although theoriesplace different emphases on the importance of cortical vs subcortical DA alterationsin relation to cognitive impairment. In this study, we found a relationshipbetween [¹⁸F]fluorodopa K_i inthe ACC and Stroop performance in both patients and controls. However, incontrast to some studies,⁴⁴ schizophrenic patientsshowed Stroop interference effects of the same magnitude as those in healthycontrols. The [¹⁸F]fluorodopa K_i cingulate data extend previous findings showing a correlation betweenStroop performance and both blood flow⁶⁷ andmetabolism⁶⁸ in the ACC in schizophrenic patients.In both of those studies, greater ACC activation was seen in patients producingmore errors, which has been taken to support the influential view that theACC is involved in detecting response conflict during task performance thatmight be associated with errors.⁶⁹ In our study,increased K_i values in the ACC were associatedwith reduced interference, which would support the idea that increased DAneurotransmission, by increasing neural signal-to-noise ratios, leads to increasedprocessing efficiency.⁶⁵

In contrast, the 2 groups appeared to show qualitative differences inthe relationship between cognitive performance and ventral striatal [¹⁸F]fluorodopa K_i values. Schizophrenicpatients performed significantly less well on the SDMT and FAS tests. Furthermore,the relationship between [¹⁸F]fluorodopa K_i and verbal fluency and SDMT performance in patients was significantlydifferent from that in controls, with controls showing a positive correlationbetween K_i values and performance, butpatients showing a negative relationship. Speculatively, our results are compatiblewith the hypothesized inverted U-shaped curve of dopaminergic tone (actingvia D1 receptor stimulation) and cognitive performance described by a numberof authors.^70- 71 Thus, in a hyperdopaminergicstate (evidenced as abnormally increased [¹⁸F]fluorodopa K_i), ventral striatum DA function may lead toimpaired performance while increases of [¹⁸F]fluorodopa K_i, within normal levels, would be predicted to improvecognitive performance. A number of investigators have suggested a prominentrole for ventral striatal DA dysfunction in the cognitive impairment seenin schizophrenia.^64,72- 73 Inparticular, Gray et al⁶⁴ argued that the corecognitive deficit in schizophrenia lies in the context-dependent use of storedregularities in guiding current action. Although the precise nature of thisdeficit in biological terms is unclear, the current data provide evidenceof a role for ventral striatal DA alterations in the cognitive impairmentof schizophrenia. Both SDMT (which requires the use of regular symbol-digitmappings to guide action) and verbal fluency (which relies on the use of storedsemantic regularities to guide output) would seem to require the kind of cognitiveprocesses described by Gray and colleagues in their neuropsychological modelof schizophrenia.⁶⁴

Altogether our results add to an accumulating body of in vivo imagingdata that indicate abnormal presynaptic striatal DA function in treated aswell as untreated patients with schizophrenia. Whether abnormal striatal DAfunction is primary or, perhaps more likely, secondary to pathophysiologicchanges in other neurotransmitter systems (eg, prefrontal glutamatergic or γ-aminobutyricacid–ergic systems) remains to be fully determined.

Corresponding author and reprints: Paul Grasby, MD, MRCPsych, CyclotronUnit, MRC Clinical Sciences Centre, Hammersmith Hospital, Imperial College,Du Cane Road, London, W12 0NN England (e-mail: paul.grasby@csc.mrc.ac.uk).

Submitted for publication January 24, 2003; final revision receivedJune 18, 2003; accepted July 10, 2003.

This study was supported by the Medical Research Council, London, England.