Correlation of Individual Differences in Schizotypal Personality Traits With Amphetamine-Induced Dopamine Release in Striatal and Extrastriatal Brain Regions

Sixty-three participants drawn from two studies of individual differences in d-amphetamine-induced dopamine release were included in this investigation. Study procedures were identical for the two studies, except that one group was given identical-appearing capsules containing placebo or amphetamine on PET scanning days (placebo-controlled cohort; N=48), whereas the other group was not blind to amphetamine administration (open-label cohort; N=15). Fourteen participants in the open-label cohort were included in a previous report (21). Table 1 summarizes the characteristics of the total sample and of each cohort. All participants received a physical and neurological examination that included ECG, blood chemistries, urine analysis, urine drug screen, and T₁, T₂, and T₂ flair MRI scans. Exclusion criteria included a history of neurological or psychiatric disorder; severe past or concomitant medical illness; borderline elevated blood pressure; abnormal results on ECG, comprehensive medical panel, CBC, or urine analysis; positive finding on 10-panel urine drug screen; brain abnormalities revealed on MRI scanning; use of psychotropic medication during the preceding 6 months; history of substance abuse or dependence; lifetime use of cocaine or amphetamines; any illicit drug use in the previous 2 months; and pregnancy or lactation. Axis I psychopathology was ruled out using the Structured Clinical Interview for DSM-IV Axis I Disorders (22). Axis II psychopathology was not formally assessed. The study was approved by the Vanderbilt University Institutional Review Board, and written informed consent was obtained from each participant.

Study and image acquisition procedures have been described previously (21). Prior to scanning, participants completed the Schizotypal Personality Questionnaire (2). The focus of this investigation is on the total score; however, factor-analytic studies indicate that the Schizotypal Personality Questionnaire can be separated into cognitive-perceptual, paranoid, negative, and disorganized factors (3). Exploratory analyses of the factor scores were undertaken to determine whether dopamine release is related to a specific dimension of schizotypy.

Participants underwent two PET scans with [¹⁸F]fallypride. The first was a baseline scan; the second occurred on a different day, 3.5 hours after administration of oral d-amphetamine (0.43 mg/kg) or placebo. As noted, 15 participants were not blind to d-amphetamine administration, and 48 participants received identical-appearing capsules on scan days that contained either placebo or d-amphetamine. Physiological measures (blood pressure, heart rate, temperature, and respirations) were monitored on scan days, and participants completed a brief screen of possible side effects. Participants also completed a brief neurological screen at baseline and at the end of the scan protocol. Blood samples for CBC and a comprehensive medical panel were also obtained at baseline and at completion of the scan protocol.

PET imaging was performed at Vanderbilt University Medical Center on either a GE Discovery LS scanner (N=30) or, after the center's scanner was upgraded, a GE Discovery STE system (N=33). All participants received their baseline and d-amphetamine scans on the same scanner. To assess the validity of combining data across scanners, we compared dopamine release in each of the anatomical regions of interest (described below) between scanners. No differences were observed in any region of interest. Moreover, voxel-wise analysis comparing dopamine release between the two scanners did not identify any clusters after whole brain correction at t=2.5 (lowest cluster-level p value >0.90).

Three-dimensional emission acquisitions and transmission attenuation correction scans were performed following a 5.0 mCi slow bolus injection of [¹⁸F]fallypride (specific activity >3,000 Ci/mmol). Serial scans started simultaneously with the bolus injection of [¹⁸F]fallypride and were obtained for approximately 3.5 hours. The extended scanning time allowed for stable kinetic model fits in both striatal and extrastriatal brain regions. The initial scan sequence coincided with the start of the [¹⁸F]fallypride injection and included the following frames: eight for 15 seconds, six for 30 seconds, five for 1 minute, two for 2.5 minutes, three for 5 minutes, and three for 10 minutes. After the initial scan sequence, a 10-minute transmission scan was obtained, and then the participant was allowed a break. Approximately 85–90 minutes after the injection, a second scan sequence of two frames of 25 minutes each followed by a second transmission scan was obtained. The participant was then allowed a second break, and at approximately 165–170 minutes, a 40-minute emission scan followed by a third transmission scan was obtained. Serial PET scans were coregistered using a mutual-information rigid-body algorithm to minimize potential modeling errors due to head motion within and between scans (23). Consistent with our prior studies with [¹⁸F]fallypride (for example, reference 24), parametric binding potential (BP_ND) images of dopamine D₂/D₃ receptor density were calculated using the full (four-parameter) reference region model (25) with the cerebellum serving as the reference region. Previous studies in our lab (26) have shown that this method produces BP_ND estimates that closely agree with those derived from Logan plots (27) using a metabolite-corrected plasma input function.

A high-resolution T₁-weighted MRI scan was also obtained for each participant, and PET and MRI scans were coregistered to one another (23). After coregistration, each participant's BP_ND image was warped to a canonical brain that had been normalized to the MNI152 template brain and resampled to 2 mm³. Parametric images of dopamine release, in percent, were created by subtracting each participant's d-amphetamine scan from his or her baseline scan and dividing the difference by the baseline scan using the ImCalc function in SPM2 (http://www.fil.ion.ucl.ac.uk/spm). In addition, dopamine release values for several anatomically defined regions of interest were extracted from the parametric images of dopamine release by calculating the mean of the voxels within each region of interest. The regions of interest included the left and right striatum, thalamus, amygdala, and hippocampus. Dopamine release values were averaged across hemispheres to produce one value for each region of interest. The striatum regions of interest were taken from the Laboratory of Neuro Imaging, UCLA (LONI) Probabilistic Brain Atlas (28) and partitioned into limbic, associative, and sensorimotor functional subdivisions using previously described criteria (29, 30). Briefly, the striatum atlas was divided into five regions of interest: ventral striatum, dorsal caudate rostral to the anterior commissure (AC), dorsal putamen rostral to the AC, postcommissural caudate, and postcommissural putamen. The limbic subdivision comprised the ventral striatum, the associative striatum was the weighted average of the pre- and postcommissural dorsal caudate and precommissural putamen, and the sensorimotor subdivision consisted of the postcommissural putamen. Dopamine release in the entire striatum, weighted by the size of each subdivision, was also calculated. The thalamus region of interest was derived from the International Consortium for Brain Mapping (ICBM) Deep Nuclei Probabilistic Atlas (http://www.loni.ucla.edu/Atlases), thresholded at 80% to avoid partial volume effects. The hippocampus and amygdala regions of interest were created using the WFU (Wake Forest University) PickAtlas and manually edited using criteria previously described by our group to avoid partial volume effects from adjacent structures (24).

The relationship between Schizotypal Personality Questionnaire score and dopamine release was examined with region-of-interest and voxel-wise analyses. First, Schizotypal Personality Questionnaire score was correlated with dopamine release in the regions of interest. The correlations for the striatum and striatum subdivisions were thresholded at a p value of 0.05, given our a priori hypothesis that dopamine release in the striatum is correlated with overall schizotypal traits. The significance threshold was set to p=0.016 for extrastriatal regions to correct for the number of structures examined. For the voxel-wise analysis, multiple regression analysis was used, with Schizotypal Personality Questionnaire score entered as a predictor of dopamine release at each voxel. Given our hypothesis, we first examined the extent to which dopamine release in the striatum was related to schizotypal traits by restricting the voxel-wise analysis to the LONI Probabilistic Brain Atlas striatum map using the small-volume-correction tool in SPM2. Only clusters within the LONI Probabilistic Brain Atlas striatum mask that exceeded the cluster-wise corrected threshold at a voxel-wise p value of 0.05 are reported. Next, we examined positive correlations throughout the brain. Only clusters exceeding the whole brain cluster-wise corrected alpha of 0.05 for voxel-wise t=2.5 are reported (31). The voxel-wise analysis was masked to exclude voxels with mean BP_ND values below 0.40 on the amphetamine scan. Significant clusters were converted to Talairach coordinates using ICBM_SPM2Tal (32). The estimated smoothness of the statistical parametric map generated for the voxel-wise regression analysis in the x, y, and z planes, respectively, was 6.6 mm, 7.5 mm, and 6.4 mm. Age, sex, and cohort (open-label or placebo-controlled) were included as nuisance covariates in both the region-of-interest and voxel-wise analyses. Scanner was not included as a covariate for two reasons. First, all participants in the open-label cohort were scanned on the Discovery LS scanner, so inclusion of scanner as a covariate would have been redundant given that cohort and scanner were not independent. Second, as noted above, neither voxel-wise nor region-of-interest analyses revealed any significant differences in dopamine release between scanners.

Correlations between dopamine release and Schizotypal Personality Questionnaire scores in the total sample and in the placebo-controlled subgroup are presented in Table 2. For the striatum, overall schizotypal traits were correlated with dopamine release in the whole striatum and associative subdivision. No correlations reached the corrected alpha level (p=0.016) in the extrastriatal regions, although dopamine release in the amygdala was correlated with the Schizotypal Personality Questionnaire score at the uncorrected alpha level. The results were virtually identical when the analysis was restricted to the placebo-controlled cohort. The mean peak plasma d-amphetamine level was 72 ng/ml (SD=19). Consistent with our previous report on the open-label cohort (21), peak plasma d-amphetamine level was unrelated to striatal dopamine release. Baseline BP_ND values and dopamine release for each region of interest are presented in Table S1 in the data supplement that accompanies the online edition of this article.

The results of the voxel-wise analysis are presented in Table 3 and Figures 1 and 2. Small-volume correction within the striatum revealed positive correlations bilaterally in the striatum. The clusters were centered in the head of the caudate but extended into the ventral striatum (see Figure 1). The corresponding correlations between mean dopamine release extracted from each cluster and Schizotypal Personality Questionnaire score, after controlling for age, gender, and cohort, were r=0.41 (p=0.001) for the left striatum and r=0.40 (p=0.002) for the right striatum. The findings were unchanged when the analysis was restricted to the placebo-controlled cohort (left striatum: r=0.45, p=0.002; right striatum: r=0.36, p=0.014).

Whole brain analysis identified two additional clusters (see Table 3 and Figure 2). They included a region within the left middle frontal gyrus corresponding to Brodmann's area 9/10 and the left supramarginal gyrus within the inferior parietal lobule. The corresponding correlations between mean dopamine release extracted from each cluster and Schizotypal Personality Questionnaire score, after controlling for age, gender, and cohort, were r=0.44 (p=0.0004) and r=0.46 (p=0.0002) for the left middle frontal and supramarginal gyrus clusters, respectively. The findings were unchanged when the analysis was restricted to the placebo-controlled cohort (left middle frontal gyrus: r=0.48, p=0.001; left supramarginal gyrus: r=0.55, p=0.0001).

No inverse correlations between dopamine release and Schizotypal Personality Questionnaire score were identified in the striatum or whole brain.

We examined the relationship between Schizotypal Personality Questionnaire factor scores and dopamine release in the regions of interest to determine whether a particular facet of schizotypy was related to dopamine release (see Table 2). No statistical correction was applied given the exploratory nature of this analysis. Robust correlations were observed between disorganized schizotypal traits and dopamine release in the whole striatum and associated subdivisions, in the amygdala, and in the thalamus. Similar results were observed in the placebo-controlled cohort. The other factor scores were not correlated with dopamine release in either the entire sample or the placebo-controlled cohort.

Given the widespread correlations observed between dopamine release and disorganized traits in the region-of-interest analysis, we performed an exploratory voxel-wise multiple regression analysis regressing disorganized factor scores on dopamine release with age, sex, and cohort as nuisance covariates. The results were thresholded at a whole brain cluster-wise corrected alpha of 0.05 for voxel-wise t=2.5. Dopamine release was correlated with disorganized traits in several subcortical and cortical regions (see Table 3; also see Figure S1 in the online data supplement), including the left and right striatum; the right thalamus and pregenual cingulate/medial prefrontal cortex; the left and right temporal cortex; the superior frontal gyrus; and the left and right insula. All of the clusters remained significant when the analysis was restricted to the placebo-controlled cohort (all cluster p values <0.003).