Brain Serotonin 5-HT1A Receptor Binding in Schizophrenia Measured by Positron Emission Tomography and [11C]WAY-100635 FREE

THERE IS CONSIDERABLE evidence for a role of the neurotransmitter serotonin(5-hydroxytryptamine [5-HT]) in the pathophysiologic characteristics of schizophrenia.^1- 3 Recent interest in 5-HT has been fueled by the fact that novel antipsychotic drugs such as clozapine, olanzapine, quetiapine, risperidone, sertindole, and ziprasidone hydrochloride are potent 5-HT_2A receptor antagonists and relatively weaker dopamine D₂ antagonists.⁴ In addition, 5-HT_1A and 5-HT_2C receptors seem to contribute to the clinical effects of some novel antipsychotic drugs.⁵

Results of most postmortem studies^6- 14 show a pronounced elevation of 20% to 79% in cortical 5-HT_1A receptor density in schizophrenia using [³H]-8-OH-DPAT or [³H]WAY-100635 as ligands. Human autoradiographic findings¹³ revealed the highest density of 5-HT_1A receptors in the temporolimbic cortex, followed by brainstem raphe nuclei, the frontal cortex, and other neocortical regions, with very low or undetectable levels in the cerebellum. The brainstem receptors are somatodendritic autoreceptors, whereas the cortical receptors are mainly postsynaptic. Cortical 5-HT_1A receptors exert inhibitory control over striatal glutamate release, and 5-HT_1Aantagonists increase glutamate release in the striatum via corticostriatal efferents.¹⁵ In addition, 5-HT_1Aagonists increase the outflow of dopamine in the prefrontal cortex, without a similar change in striatal dopamine release.¹⁶ Stimulation of 5-HT_1A receptors seems to produce many of the same effects as antagonism of 5-HT_2A receptors.⁵

Earlier efforts to quantitatively analyze 5-HT_1A receptors using the agonist ligand 8-OH-DPAT were hampered by the fact that it labels5-HT_1A receptors only in their high-affinity state. This problem has recently been overcome by the discovery of WAY-100635, a selective high-affinity(K_d<1nM) 5-HT_1A antagonist, which labels both low- and high-affinity receptors.¹⁷ WAY-100635 has been labeled at the [carbonyl-¹¹C] position¹⁸ and can be used for the quantitative analysis of binding to 5-HT_1Areceptors in humans.¹⁹ Using [carbonyl-¹¹C]-N-[2-[4-(2-methoxyphenyl)-1-piperazinyl]ethyl]-

N-(2-pyridinyl)cyclohexane carboxamide ([¹¹C]WAY-100635) and positron emission tomography(PET), cortical 5-HT_1A binding can be quantitatively analyzed using the cerebellum as an input function of a simplified reference tissue model(SRTM).^20- 21 Using this method, an age-dependent decline in cortical 5-HT_1A receptor binding potential²² (BP) in healthy volunteers was recently demonstrated,²³ consistent with findings from postmortem studies,^24- 26 which showed a decline in 5-HT_1A receptor numbers with age.

We present a PET study in 14 neuroleptic drug–naive patients with a DSM-IV diagnosis of schizophrenia who experienced a first psychotic episode. On the basis of human postmortem studies, we hypothesized that in vivo 5-HT_1A receptor BP as measured with [carbonyl-¹¹C]WAY-100635 and PET is higher in the frontal and temporal cortices of patients with schizophrenia compared with an age-matched control group.

Fourteen right-handed patients (6 women and 8 men; mean age, 26 years; age range, 22-37 years) with a DSM-IV diagnosis of schizophrenia were included in the study. Each patient experienced a first psychotic episode, had not received any psychotropic medication except for benzodiazepines within 1 month of the PET scan, and had never been treated with an antipsychotic agent. Patients were recruited from the Schizophrenia and Continuing Care Program of the Centre for Addiction and Mental Health, where they had been evaluated as either inpatients or outpatients. The diagnosis of schizophrenia was ascertained using a Structured Clinical Interview for DSM-IV,²⁷ which was performed by an experienced psychiatrist (J.T., N.P.L.G.V., Z.J.D., or S.T.-W.).

Fourteen age-matched individuals (8 women and 6 men; mean age, 28 years; age range, 19-36 years) comprised the control group. These 14 controls were recruited from the community by advertisements and were part of a bigger pool of healthy individuals described in an earlier [¹¹C]WAY-100635 PET study.²³

Patients were excluded from the study if they experienced a serious medical or neurologic illness, had a significant head injury, or were pregnant. Furthermore, exclusion criteria for control subjects were any Axis I psychiatric diagnosis as confirmed by the Structured Clinical Interview for DSM-IV, nonpatient edition,²⁸ or treatment with psychotropic medications within 3 months of the study.

All patients gave written consent after the procedure had been fully explained. The study and recruitment procedures were approved by the research ethics board of the Centre for Addiction and Mental Health and the Department of Psychiatry, University of Toronto.

The selective 5-HT_1A receptor antagonist [¹¹C]WAY-100635 was synthesized according to modifications of the McCarron method²⁹ using a short fluorocarbon resin tube loosely packed with polypropylene wool as a substitute for the narrow polypropylene tubing originally used.³⁰ This procedure yielded syntheses with high purity (>95%) and average specific activity of 47 GBq/µM (1270 mCi/µM) at the time of injection.

Positron emission tomographic images were obtained during 60 minutes using a GEMS PC2048-15B camera (General Electric Medical Systems, Milwaukee, Wis) in 15 one-minute frames followed by another 9 five-minute frames after bolus injection of a mean ± SD of 9.8 ± 0.6 mCi (363 ±22 MBq) of [¹¹C]WAY-100635. The images were corrected for attenuation with a ⁶⁸Ge transmission scan and were reconstructed using filtered back projection (Hanning filter, 5 mm full-width at half maximum), and 15 axial slices, each 6.5-mm thick, were obtained.

For the quantification of 5-HT_1A receptor binding in human brain, 2 approaches were used: one based on predefined regions of interest(ROIs) and the other a voxel-wise analysis.

Each participant underwent magnetic resonance imaging (MRI) (GE Signa1.5-T scanner; spin-echo sequence T1- and proton density–weighted images; and x, y, and z voxel dimensions 0.78, 0.78, and 3.00 mm, respectively). The MRIs were coregistered to each PET image by using RView8/mpr software.³¹ For ROI analysis, brain regions were delineated on the coregistered MRI using previously defined landmarks.³² Anatomic ROIs were drawn bilaterally in the dorsolateral prefrontal (DLPFC), anterior cingulate (ACC), medial temporal (MTC), lateral temporal (LTC), and parietal cortices and in the cerebellum by an operator masked to the condition. The gray matter of the cerebellum was delineated on consecutive slices where the middle cerebellar peduncle was clearly visible. The DLPFC was delineated on axial MRI slices in which the caudate, putamen, and globus pallidus were all clearly visualized. The ACC was delineated on the sameslices and identified as a gray matter structure on both sides of the interhemispheric fissure extending posterior to the anterior genu of the corpus callosum. The MTC was delineated as a mediotemporal gray matter region corresponding to the hippocampus, amygdaloid nucleus, and parahippocampal gyrus. The LTC included temporal gray matter located laterally starting with the same slice, where the MTC was delineated, and extending superior to slices in which the caudate, putamen, and globus pallidus were clearly visualized. The parietal cortex was delineated on 3 slices corresponding to the inferior parietal lobule extending anterior to the postcentral sulcus.

Decay-corrected time-activity curves (TACs) were obtained for each ROI using the 60 minutes of the data acquisition period because 60-minute TACs in the SRTM yielded test-retest agreement comparable to 90-minute TACs.^23,33 Because we were scanning acutely psychotic patients, we tried to keep the scanning time as brief as possible.

Regional BP²² values were calculated as an estimate of 5-HT_1A receptor number in each ROI using the kinetic modeling tool of PMOD Medical Imaging Software (version 2.20).³⁴ To obtain BP values, the cerebellum was used as the reference region for an SRTM.²⁰

For the voxel-wise analysis, parametric 5-HT_1A receptor BP images were generated using the SRTM with PMOD. Parametric images were then spatially normalized within the standard Montreal Neurologic Institute brain space using Statistical Parametric Mapping version 99 (SPM99)³⁵ and a ligand-specific template.³⁶

Statistical analyses of the ROI data were performed using SPSS for Windows10.0.0 (SPSS Inc, Chicago, Ill). Parametric statistical analyses were applied after it had been assured that skewness, kurtosis, outliers, and homogeneity of variance of our data met the criteria for a normal distribution.³⁷ In a first step, all regions were pooled together to estimate the mean cortical 5-HT_1A receptor BP of each group. To test the hypothesis that 5-HT_1A receptor BP is elevated in the frontal and temporal cortices of patients with schizophrenia, regional BP values of patients and controls were compared using multivariate analysis of covariance, with all regional BP values as dependent variables, group (patients vs controls) as a fixed factor, and age as a covariate. Multiplying the area of each ROI by the number of slices and their respective thickness of 6.5 mm provided an approximation for the actual volume of interest (VOI). Additional separate multivariate analyses of variance were performed to compare the VOI and R₁ values between groups. R₁ is the ratio of tracer delivery to the tissue of interest (K) relative to the reference tissue (K′), and it can be described by the following equation: R₁ = K/K′. To compare the uptake in the cerebellum between patients and controls, an analysis of variance was performed, with the area under the curve of the cerebellar TAC as the dependent variable. For the ROI analysis, an α level of P<.05 was considered significant, and all tests were2-tailed. For the post hoc tests, a Bonferroni correction for multiple comparisons was performed using 9 as the denominator because we compared [¹¹C]WAY-100635 binding indices between patients and controls in 9 different cortical regions.

Potential correlations between age and the measured activity in the cerebellar ROI, between age and ROI size or regional BP values, and between each ROI BP value and its respective area were examined using 2-tailed Pearson product moment correlation coefficients. In addition, Pearson product moment correlation coefficients were calculated to reveal possible correlations between5-HT_1A receptor BP in patients and duration of untreated psychosis or severity of illness as measured using the Positive and Negative Syndrome Scale for schizophrenia.³⁸

For the voxel-by-voxel analysis, statistical parametric maps were generated using SPM99. To test whether the measured 5-HT_1A receptor BP differed between patients and controls in any given voxel, 2-tailed t tests were applied. Results were displayed as statistical parametric maps using an uncorrected height threshold of P<.01(t test, >2.48). We applied 2 contrasts (patient5-HT_1A BP higher than that of controls and vice versa) to a search volume of 93,209 voxels, each with a size of 2 × 2 × 2 mm.

The mean ± SD age of patients and controls was 26 ± 5 years and 28 ± 5 years, respectively. There was no significant difference in age between groups (t₂₆ = 1.071; P = .29).

The ROI-based analysis revealed a mean ± SD 5-HT_1Areceptor BP increase of 7.1% ± 6.4% in patients with schizophrenia across all 9 cortical ROIs (Figure 1and Figure 2). A multivariate analysis of covariance revealed a significant group effect (patients vs controls) on the BP values measured in 9 cortical ROIs (F_9,17 = 2.975; P = .02), but neither a significant effect of age (F_9,17 = 0.814; P = .61) nor a significant age × group interaction (F_18,36 = 1.716; P = .08) was found. Post hoc tests of between-subject effects revealed that 5-HT_1A receptor BP values of patients were significantly higher in the left MTC (F₁ = 9.339; P = .005) and the right MTC (F₁ = 4.453; P = .045)(Table 1). There were no significant differences in BP values in any other ROI. Only the result in the left MTC survived a Bonferroni correction for multiple comparisons (corrected P = .045).

A multivariate analysis of variance revealed no significant group differences in regional tracer delivery (R₁) between patients and controls(F_9,18 = 1.427; P = .25). Mean VOI values were 23 mL for the cerebellum, 21 mL for the left PFC, 22 mL for the right PFC, 15 mL for the ACC, 12 mL for the left temporal cortex, 12 mL for the right temporal cortex, 4 mL for the left MTC, 4 mL for the right MTC, 26 mL for the left parietal cortex, and 27 mL for the right parietal cortex. There was no significant difference in VOIs between patients and controls (F_10,17 = 0.882; P = .57). Furthermore, there was no significant difference in the area under the curve of the cerebellar TAC between groups (F₁ = 0.194; P = .66)(Figure 3).

Pearson correlation coefficients (r) did not reveal any significant correlation between age and the area under the curve of the cerebellum or BP in any cortical ROI, between age and VOI, or between regional BP values and their respective VOIs. Furthermore, there was no significant correlation between MTC 5-HT_1A receptor BP and duration of untreated psychosis in patients (left MTC: r = 0.06; P = .84; right MTC: r = −0.18; P = .57). Binding potential values were not significantly correlated to severity of illness as measured using the Positive and Negative Syndrome Scale for schizophrenia sum, positive, negative, and general scores, with r = 0.05 to 0.37 and the corresponding P = .41 to .92.

The voxel-wise analysis produced no robust evidence for a pronounced elevation in [¹¹C]WAY-100635 uptake in the prefrontal or temporal cortex in schizophrenia. Statistical parametric mapping showed only one activity cluster denoting higher [¹¹C]WAY-100635 uptake in patients with schizophrenia, consisting of 86 voxels extending laterally from the Montreal Neurologic Institute brain space coordinates –16, –6, and –40 mm (x, y, and z, respectively), with z = 2.99 and P₂₆ = .001. Plotting this activity cluster on an averaged composite MRI of all patients, which had been normalized to the Montreal Neurologic Institute brain space, revealed a region in the left MTC similar to the one in which we found the most pronounced difference between patients and controls with the ROI-based approach (Figure 4).

The main finding of this in vivo PET study was an increase in cortical5-HT_1A receptor BP in schizophrenia. The most pronounced difference between patients and controls was a 20% increase in [¹¹C]WAY-100635 binding in the left MTC of patients with schizophrenia found by ROI-based PET analysis, which was confirmed by an additional voxel-wise analysis using SPM99.

This in vivo PET study of 5-HT_1A receptors in schizophrenia did not detect a significant elevation in [¹¹C]WAY-100635 binding in the prefrontal cortex of patients, which is in contrast to results of postmortem studies^6- 13 showing a 20% to 79% elevation in 5-HT_1A receptor number in that region.

Postmortem studies have several limitations. All patients in the postmortem studies had received several years of antipsychotic drug treatment, and all patients except 5 in the study by Hashimoto et al⁶ were receiving antipsychotic agents at the time of death. In contrast, all patients in our study were antipsychotic drug naive. Hence, the more pronounced increase in postmortem 5-HT_1A receptors could be an effect of long-term drug treatment.

In contrast to the postmortem studies with an average illness duration of approximately 20 years, our sample consisted of first-episode patients with a mean ± SD duration of untreated psychosis of 21 ± 18 months (range, 7 months to 5 years). Owing to this relatively restricted range, we were not able to systematically investigate the effects of disease progression.

For tracer kinetic modeling, we used the cerebellum as the reference region for an SRTM because the cerebellum is relatively devoid of 5-HT_1A receptors.¹³ Furthermore, the SRTM proved to be more reliable than kinetic modeling using arterial data²¹ and provided excellent test-retest reproducibility with [¹¹C]WAY 100635.^21,23 A test-retest study in 6 control subjects provided strong evidence that we would be able to correctly identify a group difference of 20% to 79% with acceptable sensitivity.²³ Sixty- and 90-minute TACs gave comparable results. With 60-minute TACs, the magnitude of the mean error between 2 [carbonyl-¹¹C]WAY-100635 PET scans of the same individual ranged from 2% to 7% in cortical ROIs.²³

Most postmortem studies selected limited and often arbitrary brain regions to study. Only 2 groups^6- 7,12 studied samples from all cortical regions. All others investigated only the prefrontal cortex^9- 11,13 or the prefrontal and temporal cortices.⁸ Although all of the postmortem studies report "prefrontal" increases, the data came from Brodmann areas (BAs) 9⁸; 10⁶;11 and 12⁹; 24, 9a, and 44¹²;or 46.^10,13 To avoid an unjustified restriction to frontal brain areas, we analyzed 5-HT_1A receptor BP in cortical ROIs drawn in the DLPFC, ACC, MTC, LTC, and parietal cortex. However, we are aware of several limitations inherent to the ROI approach. The intrinsic spatial resolution of our PET camera is 4.5 to 5.5 mm in the transaxial plane. Therefore, we chose to delineate rather large areas comprising several BAs. In the case of the DLPFC, this ROI roughly corresponds to portions of BAs 9, 10, and 46, whereas in case of the ACC, we tried to confine this ROI to BA 32 and frontal parts of BA 24.

The process of coregistering introduces another source of error, which also contributes to the anatomic inaccuracy of the ROI approach. Moreover, if any given pathologic condition afflicts only parts of an ROI, a possible BP increase or decrease will be diluted, and, therefore, the chance to miss a group difference is high. On these grounds we chose to corroborate the results of the ROI analysis by applying an additional voxel-wise SPM99 analysis. Because of the restricted field of view of our PET camera (15 slices of 6.5 mm each, which translates into 9.75 cm with regard to the z-axis), even the voxel-wise approach cannot be considered to cover the entire brain. Nevertheless, SPM99 revealed an elevated 5-HT_1A receptor BP in the left MTC of patients, confirming the strongest result of the ROI-based approach.

We matched our control group for age but not for sex because it had been shown post mortem and in vivo using PET and [¹¹C]WAY-100635 that 5-HT_1A density declines with age, whereas in both studies sex did not significantly affect 5-HT_1A receptor BP.^23- 24

Based on the results of postmortem studies, we expected at least a 25% BP elevation in the frontal cortices of patients in vivo. Using values from controls (mean ± SD frontal BP, 3.2 ± 0.4), we estimated that our study had a power of 0.98 (1 − β) to detect a 25% elevation in frontal 5-HT_1A receptor BP in patients at α<.05. However, this study did not reveal such an elevation in frontal brain regions. On the other hand, to decide whether the 4% to 5% "nonsignificant" difference reported in the DLPFC is truly within chance or due to a type II error, we would have to increase our sample size 10-fold, which is not feasible because of practical limitations.

There is preclinical evidence^39- 40 to support a role for 5-HT_1A agonism in the antipsychotic action and extrapyramidal adverse effects of drugs. The 5-HT_1A agonist8-OH-DPAT enhanced the antipsychotic-like effect of the D₂/D₃ antagonists raclopride⁴¹ and haloperidol⁴² and antagonized the catalepsy induced by the D₁ agonist SCH23390 in rats.⁴³ Several atypical antipsychotic drugs are partial agonists at the 5-HT_1Areceptor, including clozapine, ziprasidone, quetiapine, and tiospirone hydrochloride. Clinical studies of adding 5-HT_1A partial agonists may help to clarify the possible importance of 5-HT_1A agonism in the treatment of schizophrenia.

The biological significance of elevated 5-HT_1A receptor numbers in schizophrenia as indicated by several postmortem studies and this in vivo PET study remains unclear. It has been suggested that given the location of most of the 5-HT_1A receptors on pyramidal cells, it may reflect an abnormal glutamatergic network.⁴⁴ Although we did not confirm a pronounced 5-HT_1A receptor elevation in frontal cortices of patients, the ROI-basedapproach demonstrated a 20% higher 5-HT_1A receptor BP in the left MTC and a 13% elevation in the right MTC. Underlining the robustness of the result in the left MTC was the fact that it survived a correction for multiple comparisons in 9 cortical ROIs and was confirmed using a voxel-wise analysis with SPM99. The left temporal cortex is an anatomic region known to be afflicted in schizophrenia,⁴⁵ but the significance and functional relevance of our finding of locally elevated[¹¹C]WAY-100635 uptake in patients with schizophrenia remains unclear and warrants replication in the future.

Submitted for publication February 26, 2001; final revision received August 16, 2001; accepted September 11, 2001.

This research was supported by the EJLB Foundation (Montreal, Quebec) and the Austrian Research Fund (Vienna).

We thank all patients and healthy volunteers for their participation; Corey Jones, BSc, Kevin Cheung, Alex Kecojevic, HBSc, Li Jin, and Armando Garcia for technical assistance; and Barb Brownlee, MSc, for proofreading.

Corresponding author and reprints: Johannes Tauscher, MD, Department of General Psychiatry, University of Vienna, Währinger Gürtel 18-20, A-1090 Vienna, Austria (e-mail: johannes.tauscher@akh-wien.ac.at).