Main

The serotonin-1 (5-HT1) family of receptors is characterized by seven transmembrane loops linked to G-proteins and modulate adenylate cyclase activity (Zifa and Fillion 1992). The 5-HT1A receptors are distinguished from other 5-HT1 family members by their unique genetic sequence and ligand-binding profile (Zifa and Fillion). Until recently, there were no specific antagonist ligands for this receptor; therefore, most studies relied on selective agonists such as 8-hydroxy-2-di-n-propylaminotetraline (8-OH-DPAT) (Fletcher et al. 1993). Agonist studies are limited by the fact that they bind only to 5-HT1A receptors in the high-affinity state. Furthermore, 8-OH-DPAT has never been approved for human use. The discovery of selective and potent antagonists, such as WAY-100135 and WAY-100635, which both can be used in humans, has opened new possibilities for 5-HT1A research and therapeutics (Fletcher et al. 1993).

A human autoradiographic study showed the highest density of 5-HT1A receptors in the temporolimbic cortex, followed by brainstem raphe nuclei, frontal cortex, and other neocortical regions, with very low or undetectable levels in the cerebellum (Burnet et al. 1997). The brainstem receptors are somatodendritic autoreceptors, whereas the cortical receptors are mainly postsynaptic. The cortical 5-HT1A receptors are localized on axon hillocks of pyramidal cells, especially in the layers II, III, and V of the cortex (Burnet et al. 1997). These receptors are involved in the inhibitory modulation of cortico–cortical association and cortico–striatal efferent fibers (Bowen et al. 1993). Cortical 5-HT1A receptors exert inhibitory control over striatal glutamate release, and 5-HT1A antagonists increase glutamate release in the striatum via cortico–striatal efferents (Dijk et al. 1995). In addition, 5-HT1A agonists increase the outflow of dopamine in the prefrontal cortex without a similar change in striatal dopamine release (Wȩdzony et al. 1996).

Post-mortem studies show a decline in 5-HT1A receptor number with age (Dillon et al. 1991; Lowther et al. 1997; Matsubara et al. 1991). However, these studies have either used 8-OH-DPAT, an agonist ligand, which measures 5-HT1A receptors in the high affinity state, or spiperone, an antagonist, which lacks specificity. These limitations have recently been overcome by the discovery of WAY-100635, a selective high-affinity (Kd < 1 nM) 5-HT1A antagonist (Fletcher et al. 1993). WAY-100635 labels almost twice as many 5-HT1A receptors as 8-OH-DPAT, because it labels both low- and high-affinity receptors (Fletcher et al. 1993). To adapt WAY-100635 for human studies, it has been labeled at the [carbonyl-11C] position (Farde et al. 1997) and can be used for the quantitative analysis of binding to 5-HT1A receptors in humans (Farde et al. 1998).

Recently, positron emission tomography (PET) and [11C]WAY-100635 were used to show decreased 5-HT1A receptor binding potential (BP [Mintun et al. 1984]) in patients with major depression (Drevets et al. 1999; Sargent et al. 2000) and in clozapine-treated schizophrenic patients (Bantick et al. 2000), as compared to healthy control groups.

Because of the noted decrease in 5-HT1A receptor number with age in post-mortem studies, our hypothesis was that 5-HT1A receptor BP as measured with [11C]WAY-100635 and PET would decline with age in healthy subjects.

MATERIALS AND METHODS

Subjects

Nineteen healthy subjects (8 female/11 male; mean age 34 years; range: 22–53) were examined. Exclusion criteria were: (1) any Axis-I psychiatric diagnosis confirmed by the Structured Clinical Interview for DSM-IV, nonpatient edition (SCID-I/NP [First et al. 1997]); (2) presence of serious medical or neurological illness or of significant head injury; (3) a history of alcohol or substance dependence; (4) treatment with psychotropic medications within the last 3 months before the study; or (5) pregnancy. All subjects gave their written consent after the procedure had been fully explained. The study had been approved by the Human Subjects Review Committee of the University of Toronto.

To establish test–retest reliability, six of the 19 subjects were studied on a second occasion 1 to 2 weeks after the first scan.

[11C]WAY 100635 PET Scanning Protocol

Radiolabeling

The selective 5-HT1A receptor antagonist carbonyl-[11C]-N-[2-[4-(2-methoxyphenyl)-1-piperazinyl]ethyl]-N-(2-pyridinyl)cyclohexane carboxamide ([11C]WAY-100635) was synthesized by modifications of the McCarron method (McCarron et al. 1996) using a short Teflon tube loosely packed with polypropylene wool as a substitute for the narrow polypropylene tubing originally used (Houle et al. 2000). This procedure yielded syntheses with high purity and average specific activity of 41 GBq/μ mole at time of injection.

Data Acquisition

PET images were obtained with a GEMS PC2048-15B camera (General Electric Medical Systems, Milwaukee, WI, USA) in 15 1-minute frames followed by another 15 5-minute frames after bolus injection of 9.56 mCi [11C]WAY-100635. The images were corrected for attenuation with a 68Ge transmission scan and reconstructed by filtered back projection (Hanning filter, 5 mm full width at half maximum) and 15 6.5 mm-thick axial slices were obtained.

Image Analysis

For the quantification of 5-HT1A receptor binding in human brain, two approaches were used: region of interest (ROI) and voxelwise analysis.

Each subject had a magnetic resonance imaging (MRI) scan (GE signa 1.5 T scanner; spin echo sequence T1- and PD-weighted image; x,y,z voxel dimensions: 0.78, 0.78, and 3 mm, respectively). MRI scans were co-registered to each PET image by using RView8/mpr software (Studholme et al. 1997). For ROI analysis, nine distinct brain regions were delineated on the co-registered MRI using landmarks previously defined (Bremner et al. 1998). Anatomical ROI were drawn in dorsolateral prefrontal (DLPFC) and orbitofrontal (OFC), anterior cingulate (ACC), medial temporal (MTC), and lateral temporal (LTC), parietal and occipital cortex, and the cerebellum as a reference region. In addition, a raphe nuclei ROI was delineated in the following way: midbrain sections were identified on the co-registered MRI as those consecutive slices, where the interpeduncular cistern was clearly visible. A fixed-size circular ROI was then placed on a dorsal midbrain area with high tissue radioactivity in the corresponding two slices of a PET summation image yielding a constant volume of interest (VOI) of 0.6 cm3 for the raphe nuclei.

Decay-corrected time activity curves (TAC) were obtained for each ROI using the first 60 minutes of the data acquisition period. Additionally, for the six subjects scanned twice, TAC of 90 minutes data acquisition periods were generated. Regional binding potential (BP) values were calculated to estimate the 5-HT1A receptor number in each ROI (Mintun et al. 1984). To obtain BP values, the cerebellum was used as the input function of a simplified reference tissue method, because the cerebellum is relatively devoid of 5-HT1A receptors (Burnet et al. 1997) and this method proved to be superior to kinetic modeling using arterial data (Gunn et al. 1998).

For the voxelwise analysis, parametric 5-HT1A receptor BP images were generated using the simplified reference tissue model (Gunn et al. 1997). In a next step, parametric images were spatially normalized within the standard Montreal Neurologic Institute (MNI) brain space using Statistical Parametric Mapping version 99 (SPM99) (Friston et al. 1995) and a ligand-specific template (Meyer et al. 1999).

Statistical Analysis

Statistical analyses of the ROI data were performed using SPSS for Windows 10.0.0, SPSS Inc., 1999. The mean and standard deviation of BP values were calculated for each ROI. To test the hypothesis that 5-HT1A receptor BP declines with age, one-tailed Pearson product-moment correlation coefficients were calculated, using a threshold of p < .05 for significance. Potential correlations between age and the measured activity in the cerebellar ROI, between age and volume of interest (VOI) size, and between each ROI BP value and its respective volume were examined using two-tailed Pearson product-moment correlation coefficients.

The test–retest agreement of the estimates for BP in each ROI was assessed by calculating the difference between scan 1 and scan 2. Furthermore, we calculated average measure intraclass correlation coefficients (ICC) for each ROI and the repeatability coefficient (RC), which were calculated over the six subjects scanned twice. The RC is equal to twice the standard deviation of the difference between the BP values of scan 1 and 2. Thus, it can be expected that 95% of the differences are less than the RC (Bland and Altman 1986). In addition, to facilitate comparisons across regions, the RC was calculated as a percentage of the mean BP (RM):

Voxel-by-voxel analysis was carried out using SPM99. To test the hypothesis that age influenced the measured 5-HT1A receptor BP in any given voxel, analysis of covariance (ANCOVA) was applied using age as the covariate of interest with no global normalization of brain activity levels, no grand mean scaling, and a gray matter threshold of 80%. Results of the ANCOVA were displayed as parametric maps using a height threshold of p = .001 with no extent threshold and two contrasts to examine increases and decreases with age. Furthermore, SPM99 was used to test: (1) whether the assumed age-related 5-HT1A receptor BP decline occurred globally or in distinct brain areas; and (2) if gender influenced 5-HT1A receptor BP.

RESULTS

After injection of 9.56 mCi (± .77 [SD]) [11C]WAY-100635, ROI analysis of 60 minutes TAC by means of a simplified reference tissue method (Gunn et al. 1998) yielded 5-HT1A BP values (mean ± SD) ranging from 5.10 (± 0.69) in MTC to 2.10 (± 0.65) in the raphe nuclei (Table 1). The rank order for BP values was MTC > LTC > ACC > OFC > DLPFC > parietal cortex > occipital cortex > raphe nuclei. Pearson correlation coefficients revealed significant age-related 5-HT1A BP decreases in DLPFC (r = −0.56; p = .006 [Figure 1]), ACC (r = −0.44; p = .029), OFC (r = −0.42; p = .036), LTC (r = −0.40; p = .047), occipital (r = −0.43; p = .034) and parietal cortex (r = −0.65; p = .001 [Figure 2]). No significant correlation between 5-HT1A BP and age was found in MTC (r = −0.27; p = .132 [Figure 3]) and raphe nuclei (r = −0.16; p = .263).

Table 1 Mean Regional 5-HT1A Receptor BP Values and Their Respective Volume of Interest (in cm3) for 19 Healthy Subjects
Full size table
Figure 1
figure 1

Scatterplot of the serotonin 5-HT1A receptor BP decline in the dorsolateral prefrontal cortex with age in 19 healthy subjects

Full size image
Figure 2
figure 2

Scatterplot of the serotonin 5-HT1A receptor BP decline in the parietal cortex with age in 19 healthy subjects

Full size image
Figure 3
figure 3

Scatterplot of the serotonin 5-HT1A receptor BP in the MTC showing no significant decline with age in 19 healthy subjects

Full size image

Volume of interest (VOI) values ranged from 54.4 cm3 in parietel cortex to 0.6 cm3 for the raphe nuclei (Table 1). There was no correlation of age and VOI size in most ROI and in the cerebellum, with the exception of the occipital cortex (r = 0.52; p = .026) and OFC (r = −0.66; p = .002). However, VOI size and their respective BP values were not significantly correlated for any given ROI. Furthermore, there was no correlation of age and the measured activity in the reference region cerebellum (r = −0.09; p = .720).

A test–retest study in six healthy subjects scanned twice revealed excellent reproducibility for BP values of each ROI except for the raphe nuclei (Table 2). Sixty minute TAC gave slightly better results for test–retest reproducibility than 90 minute TAC, as described by the mean error between scan 1 and 2 in percentage, the ICC for BP scan 1 and 2, the RC and the RM. With 60 minute TAC, the size of the mean error between scan 1 and scan 2 ranged from 2 to 7% in cortical ROI, and was 19% in the raphe nuclei. Average measure ICC confirmed a very low test–retest error between repeated 60 minute scans ranging for cortical regions from 0.93 to 0.99. Again, the raphe showed the poorest result with an ICC of 0.58.

Table 2 Test-Retest Reproducibility in Six Healthy Subjects Described by the Mean Error in Absolute Percentage Between Scan 1 and Scan 2 for Each ROI, Their Respective 5-HT1A Receptor Binding Potentials Intraclass Correlation Coefficients (ICC), Repeatability Coefficients (RC), and Their Associated Percentage of the Mean Parameter Value (RM)
Full size table

Voxelwise analysis revealed a significant global decrease of 5-HT1A BP with age (set level < .001; 12 clusters). Four of these 12 clusters survived a correction for multiple comparisons over the entire volume (Table 3). Parametric mapping using SPM99 showed that the age-related decline in cortical 5-HT1A receptor BP was not limited to specific brain regions but occurred globally all over the cortex (Figure 4). The only exception was one cluster in the mediotemporal cortex of each hemisphere, which showed a significantly lesser decline in 5-HT1A receptor BP values with age (Figure 5).

Table 3 Voxelwise Analysis with SPM99
Full size table
Figure 4
figure 4

Parametric maps generated with SPM99 show a widespread and significant decrease of 5-HT1A receptor BP with age (for a summary of statistics, see Table 3). Red colored voxels denote brain regions with a significant age-related decline in 5-HT1A receptor BP. A grid shows the area of the brain which was actually scanned

Full size image
Figure 5
figure 5

Parametric maps using proportional scaling to depict areas with significantly lesser than average age-related 5-HT1A receptor BP decline. Note two clusters, occurring bilaterally one in each MTC, which survived correction for multiple comparisons (left MTC: pcorrected = .003, cluster size: 285 voxels; right MTC: pcorrected = .006, cluster size: 253 voxels)

Full size image

A multivariate analysis of variance with the eight regional BP values as the dependent variables (DV), gender as a fixed factor and age as a covariate revealed no significant interaction between gender and 5-HT1A BP in any of the ROI. Testing for a differential influence of gender on the demonstrated age-dependent 5-HT1A receptor BP decline, the interaction term of “gender*age” did not predict the DV for six of the eight tested ROI. However, there was a significant interaction between “gender*age” and 5-HT1A BP in the DLPFC (F = 5.548; p = .015) and in the parietal cortex (F = 6.534; p = .008), but none of these survived a correction for multiple comparisons. For the voxelwise analysis, we used SPM99 with “female” and “male” treated as two conditions, age as the covariate of interest and a corrected threshold level of p < .05 for statistical significance. There was no significant interaction between: (1) the covariate age and BP; and (2) the interaction term “age*gender” and BP for any given voxel. In summary, both voxelwise and ROI analysis did not reveal a significant influence of gender on 5-HT1A receptor BP values, and no differential effect of gender on the demonstrated age-dependent 5-HT1A receptor BP decline.

DISCUSSION

In vivo imaging of 5-HT1A receptors in humans with PET and [11C]WAY-100635 revealed a significant age-related decline of brain 5-HT1A receptors, which occurred globally all over the cortex, with the exception of a mediotemporal area in both hemispheres. Both approaches—ROI and voxel-wise analysis—yielded similar results: 5-HT1A receptor BP declined approximately by 10% per decade in our sample of healthy volunteers aged between 22 and 53 years.

Our findings, except for the apparent lack of age-dependent 5-HT1A receptor BP decline in the MTC and in the raphe nuclei, are in correspondence with post-mortem data. Quantitative autoradiographic analysis of 5-HT1A receptors with [3H]8-OH-DPAT as a ligand revealed age-related decreases in receptor density (Bmax) without apparent changes in affinity (Kd) in several cortical and hippocampal regions (Dillon et al. 1991). As in our study, gender had no effect on 5-HT1A receptors density. Another autoradiographic post-mortem examination with [3H]8-OH-DPAT found a significant negative correlation between age and the number of 5-HT1A receptors in the frontal cortex of suicide victims, but not in a control group (Lowther et al. 1997). In contrast, an earlier study had reported a significant negative correlation of age and 5-HT1A Bmax in frontal cortex of a control group but not in suicide victims (Matsubara et al. 1991). The interpretation of these post-mortem results is compromised by the fact that the agonist ligand 8-OH-DPAT measures 5-HT1A receptors only in the high-affinity state.

Previous in vivo studies in healthy volunteers with PET and [11C]WAY-100635 failed to show a significant decrease of 5-HT1A receptors with age. In two studies, it may be attributed to a lack of power because of a relatively small sample size of six (Farde et al. 1998; Gunn et al. 1998) as opposed to 19 in our study. In a recent study reporting a widespread reduction of 5-HT1A receptor binding with major depression, no correlation with age was found in 18 healthy controls (Sargent et al. 2000). The age range of the control group (27–56 years) was comparable to our sample (22–53 years). Unfortunately, the individual ages were not reported, but the distribution of age seemed slightly narrower in the Sargeant et al. (2000) study, as indicated by a standard deviation of 8.3, as compared to 10.2 in our study. Finally, another study reporting a reduced 5-HT1A receptor binding potential in the raphe and MTC of 12 depressive patients in comparison to eight healthy controls, reported a trend toward an inverse correlation with age (r = −0.6) in the controls, which did not reach significance at the p < .05 level (Drevets et al. 1999). Again, the failure to show a significant relationship might be attributed to the small sample size (n = 8).

One shortcoming of the present study might be that we studied healthy volunteers over a restricted age range of 22 to 53 years. Although it would have been possible to enroll slightly younger adults, we must point out that, for ethical reasons, we were not allowed to include volunteers younger than 18 years in this PET study comprising injection of a radioactive labeled compound. On the other hand, the inclusion of older subjects might have provided additional scientific insight. Further studies including younger and older subjects seem to be warranted, especially to address the question whether the reported age dependent 5-HT1A receptor decline is a linear phenomenon over the whole life span or not.

Previously, 60 and 90 minute TAC were used for tracer kinetic modeling of [11C]WAY-100635. We used 60 minute TAC for the calculation of the 5-HT1A receptor binding potential with a simplified reference tissue model. In our own test–retest study, 60 minute TAC gave comparable if somewhat better repeatability measures as 90 minutes (Table 2). In a kinetic modeling study Gunn et al. (1998) reported a better (RC) for the MTC with 60 minutes; whereas, 90 minutes were superior for the insula and cingulate cortex, and the raphe nuclei. In comparison, all RC for each single ROI and their associated percentage of the mean parameter value (RM) were far better with both 60 and 90 minute TAC in our study (Table 2). Because we intend to investigate 5-HT1A receptors with PET and [11C]WAY-100635 in psychotic patients in the future, the point can be made that scanning time should be kept as brief as possible to avoid possible behavioral problems and movement artifacts.

Because we used the cerebellum for a simplified reference tissue model, the age-related decline of 5-HT1A receptor BP could theoretically be attributed to a higher [11C]WAY-100635 uptake with age. However, in our sample, there was no correlation of cerebellar uptake as measured by the area under the cerebellar TAC and age.

In the age range of our sample; that is, between 22 and 53 years, a linear decline of gray matter cortical volume of approximately 6–7% per decade had been shown in healthy subjects (Zipursky et al. 1998; Pfefferbaum et al. 1994). To rule out cortical atrophy as a possible source of error, we examined a possible correlation of the actual VOI size and the measured BP. There was no correlation of VOI size and BP in any given VOI. Pearson correlation coefficients did not reveal any correlation between age and VOI size for the cerebellum, DLPFC, ACC, LTC, MTC, parietal cortex, and raphe nuclei. However, there was a positive correlation between age and VOI size in occipital cortex, and a negative correlation between age and the OFC VOI size. Because VOI size did not decrease with age, with the exception of the OFC, it is quite unlikely that the age-related decline of 5-HT1A receptor BP was attributable to partial volume effects. Furthermore, in the occipital cortex, a positive correlation between age and VOI might have lead to an underestimation of the age-related decline of 5-HT1A receptor BP.

On the other hand, partial volume effects might have contributed to a possible underestimation of 5-HT1A receptor BP values in the raphe nuclei. With 0.6 cm3, this was by far the smallest VOI and the mean BP value found in our study was considerably lower than those reported earlier by other groups (Farde et al. 1998; Gunn et al. 1998). Furthermore, it could not be delineated on the co-registered MRI, and showed the poorest test–retest reproducibility in our study (Table 2) and in another test–retest study (Gunn et al. 1998). These limitations must be taken into account when interpreting results of 5-HT1A receptor binding potential in raphe nuclei with PET and [11C]WAY-100635.

Great emphasis was laid on the accurate delineation of the ROI on the co-registered MRI. We used previously defined landmarks (Bremner et al. 1998) to delineate the ROI on the MRI and transferred the obtained template on a summation PET image where it was adjusted accordingly. The validity of our ROI approach is underlined by the fact that voxelwise analysis with SPM99 yielded similar results.

CONCLUSION

The main finding of this in vivo brain-imaging study in healthy volunteers was a significant decline of 5-HT1A receptor density in the range of approximately 10% per decade, except for the medial temporal cortex and the raphe nuclei, where we did not find such a decline. Hence, the effect of age must be considered in future studies using PET and [11C]WAY-100635 for the visualization of brain 5-HT1A receptors. Based on our data, careful age matching must be advised for group comparisons of psychiatric patients and healthy controls to avoid this source of error.