Decreased brain serotonin 5-HT_1A receptor availability in medication-naive patients with major depressive disorder: an in-vivo imaging study using PET and [carbonyl-¹¹C]WAY-100635

Abstract

Serotonin (5-HT) is involved in the pathophysiology of major depressive disorder (MDD). Among the numerous serotonergic receptors, the 5-HT_1A receptor subtype is of interest because of its involvement in cognition, hippocampal neurogenesis, and mechanism of action of antidepressant drugs. Previous imaging studies have suggested altered availability of 5-HT_1A receptors in MDD but prior antidepressant medication and chronicity of the illness may confound the interpretation. We examined 21 drug-naive primary-care patients with MDD using positron emission tomography (PET) imaging with [carbonyl-¹¹C]WAY-100635, a radioligand for 5-HT_1A receptors, along with 15 healthy control subjects. Binding to receptors was assessed both regionally and at voxel level with the binding potential (BP) that was estimated using arterial blood input. Compared with healthy controls, the BP of [carbonyl-¹¹C]WAY-100635 was reduced in patients with MDD in most brain regions, ranging from −9% to −25%. Voxel-level analysis confirmed this finding by showing a widespread reduction of [carbonyl-¹¹C]WAY-100635 BP. No statistically significant associations were observed between BP and total HAMD scores in the patients, but lower BP was associated with higher likelihood of insomnia. This study demonstrated a widespread reduction in the availability of serotonin 5-HT_1A receptors in a relatively large sample of drug-naive primary-care patients with MDD, suggesting the involvement of this receptor subtype in the pathophysiology of the illness. Lack of correlation with overall severity of the illness may relate to a largely trait-like nature of this abnormality in depressive disorders.

Depression, positron emission tomography, serotonin (5-HT), 5-HT_1A receptor

Introduction

The monoamine hypothesis of major depressive disorder (MDD) was primarily based on indirect pharmacological evidence (Carlsson et al., 1968; Kuhn, 1958; Schildkraut, 1965). Specifically, it suggests that a deficit in the functioning of monoamine neurotransmitters (such as serotonin; 5-HT) is central to the pathophysiology of the illness. Conversely, the use of drugs that increase the synaptic concentration of 5-HT by means of 5-HT transporter blockade [such as tricyclic antidepressants (TCA) and selective serotonin reuptake inhibitors (SSRIs)], is the first line of pharmacological treatment for MDD (Duval et al., 2005). Further support for this hypothesis is provided by studies of post-mortem samples of the human brain that have demonstrated abnormalities in the densities of various 5-HT receptor subtypes. However, a clear picture has not emerged from these studies, and it is often difficult to distinguish between disease-specific influences and confounding factors, such as psychiatric comorbidity and antidepressant medication (Stockmeier, 2003).

In-vivo imaging studies of the brain serotonergic system using positron emission tomography (PET) and single-photon emission computed tomography (SPECT) appear to be the best methods of revealing disease-specific abnormalities in the brains of living patients. Previous studies have examined serotonin 5-HT_1A receptor binding (Bhagwagar et al., 2004; Drevets et al., 1999; Meltzer et al., 2004; Parsey et al., 2006a; Sargent et al., 2000), serotonin 5-HT₂ receptor binding (Attar-Lévy et al., 1999; Bhagwagar et al., 2006; Biver et al., 1997; Messa et al., 2003; Meyer et al., 1999a, 2003; Mintun et al., 2004; Schins et al., 2005; Sheline et al., 2004; Yatham et al., 2000), and 5-HT transporter availability (Catafau et al., 2006; Herold et al., 2006; Ichimiya et al., 2002; Lehto et al., 2006; Malison et al., 1998; Meyer et al., 2001, 2004; Newberg et al., 2005; Parsey et al., 2006b; Reivich et al., 2004; Staley et al., 2006) in patients with major MDD relative to healthy comparison subjects. However, results from these studies have been mixed; factors likely to contribute to inconsistent findings may include clinical heterogeneity across different samples, the choice of radiotracer as well as the quantification methods applied, and whether or not patients have been on antidepressant drug treatment prior to imaging studies (for review see Stockmeier, 2003). A recent PET study demonstrated significantly increased levels of monoamine oxidase A (MAOA), an enzyme that breaks down 5-HT, norepidephrine, and dopamine, in the brains of patients with major depression (Meyer et al., 2006). The latter finding argues for a more general defect in monoamine function in major depression.

Of the 5-HT receptors the 5-HT_1A subtype is of particular interest (Albert and Lemonde, 2004). The 5-HT_1A subtype is expressed as two distinct populations in the brain: as somatodendritic presynaptic autoreceptors in the raphe nuclei, from where all serotonergic neurons ascend, and post-synaptic receptors in limbic structures and in the neocortex (Aznavour and Zimmer, 2007). Somatodendritic autoreceptors in the raphe nuclei regulate the firing of serotonergic neurons and are thus in a central position to control 5-HT release at projection terminals (Fisher et al., 2006). Desensitization of these receptors is thought to explain the delayed antidepressant response to SSRI treatment (Blier and Ward, 2003). Post-synaptic 5-HT_1A receptors influence cortical pyramidal cell functioning and thereby affect cognition and emotion (Díaz-Mataix et al., 2005; Yuen et al., 2005), and they also contribute to hippocampal neurogenesis, a neurobiological substrate for the antidepressant effect (Berton and Nestler, 2006; Manji et al., 2001). PET with the 5-HT_1A antagonist radioligand [carbonyl-¹¹C]WAY-100635 provides high-contrast images of 5-HT_1A receptor distribution in the living human brain (Farde et al., 1998; Gunn et al., 1998; Hirvonen et al., 2007; Parsey et al., 2000; Pike et al., 1996). This paradigm has been previously used to demonstrate altered in-vivo 5-HT_1A availability in MDD (Bhagwagar et al., 2004; Drevets et al., 1999; Meltzer et al., 2004; Parsey et al., 2006a; Sargent et al., 2000). Importantly, previous antidepressant drug treatment may contribute to the in-vivo availability of [carbonyl-¹¹C]WAY-100635 (Parsey et al., 2006a), thus emphasizing the need for studying MDD patients who have not been previously exposed to SSRI treatment.

The purpose of this study was to assess the in-vivo 5-HT_1A availability using PET and [carbonyl-¹¹C]WAY-100635 in a sample of medication-naive MDD patients as well as in healthy control subjects. Arterial blood samples were acquired from all subjects to allow for the accurate discrimination of specific and non-specific binding using kinetic compartmental modelling.

Method

This study was approved by the Joint Ethics Committee of the University of Turku and Turku University Central Hospital, and was conducted according to the Declaration of Helsinki. All subjects gave ethics committee-approved written informed consent.

Subjects and clinical evaluations

The study was carried out at Turku PET Centre and performed in collaboration with the psychiatric clinics of the Helsinki and Turku universities and the Research Centre of the Social Insurance Institution (Turku, Finland) during 2000–2004. This study was part of a larger project comparing psychotherapy and SSRI drug treatment in treating MDD. The patients with symptoms of depression were recruited from five occupational health service (OHS) units providing primary health care.

At the Research Centre, all subjects were interviewed by a psychiatrist, using the Structured Clinical Interview for DSM-IV Axis I disorders (First et al., 1997). The severity of the depression was assessed by the psychiatrist using the 17-item Hamilton Depression Rating Scale (HAMD; Hamilton, 1967). The subjects were also asked to complete the Beck Depression Inventory (BDI; Beck et al., 1961). The inclusion criteria for the study were: (1) mild or moderate episode of major depressive disorder, (2) HAMD score of ⩾13, (3) age 20–60 yr, (4) no psychotherapeutic or psychopharmacological treatment during the preceding 4 months, (5) no DSM-IV Axis I or Axis II comorbidity, (6) no severe somatic illnesses. Twenty-one patients diagnosed with MDD entered the PET study, and were compared against 15 healthy control subjects. Demographic and clinical characteristics of the study sample are given in Table 1. The patients with MDD were on average moderately ill: total HAMD scores ranged from 13 to 23 (median 18), and total BDI scores were from 10 to 40 (median 23) (data missing from one patient, n=20). Seventeen patients (81%) were experiencing their first depressive episode, while four (19%) had recurrent depression: of those, three (14.3%) had had one previous depressive episode, and one (4.8%) had suffered two previous episodes. Education was reported using three categorical levels based on years of education (‘1’, <13 yr; ‘2’, 13–15 years; ‘3’, ⩾16 yr). Five patients (24%) were using oral benzodiazepines (midazolam, oxazepam, temazepam) or non-benzodiazepine hypnotics (zopiclone, zolpidem).

Table 1

Demographic and clinical characteristics and radiochemical measurements of the study sample

BDI, Beck Depression Inventory; HAMD, Hamilton Depression Rating Scale; MDD, major depressive disorder; s.d., standard deviation.

a

t=−2.60, p=0.014. ^b χ²=0.86, p=0.769. ^c Data missing from one control (n=14), t=−0.49, p=0.625. ^d Data missing from one patient (n=20). ^e All p>0.3.

PET procedures

The radioligand [carbonyl-¹¹C]WAY-100635 was prepared according to published procedures (Hirvonen et al., 2007). PET experiments were carried out using a whole-body 3D PET scanner (GE Advance; GE, Milwaukee, WI, USA) with 35 slices of 4.25-mm thickness covering the whole brain, running in 3D mode, essentially as previously described (Hirvonen et al., 2007). Briefly, an intravenous bolus of [carbonyl-¹¹C]WAY-100635 was given in the right antecubital vein and rapidly washed with saline. The average (±s.d.) injected dose and mass were 232.6±52.7 MBq and 1.18±0.80 µg, respectively. Radiochemical purity was at least 96%. There were no statistically significant differences in any of the radiochemical parameters between patients with MDD and healthy controls (Table 1). The uptake of [carbonyl-¹¹C]WAY-100635 was measured for 57 min using 14 time-frames of increasing duration. To obtain arterial input function for modelling, the left radial artery was cannulated for obtaining arterial samples. Samples were collected for the measurement of plasma radioactivity and metabolites according to a previously published protocol (Hirvonen et al., 2007). Finally, an arterial plasma input curve for [carbonyl-¹¹C]WAY-100635 corrected for metabolites was constructed.

Automated region of interest (ROI) analysis

PET images were analysed using an automated ROI analysis based on spatial normalization into standard space of dynamic PET images and a ROI template (Hirvonen et al., 2006). First, a ligand-specific template for [carbonyl-¹¹C]WAY-100635 was constructed using data from 12 healthy volunteers and an MRI-assisted procedure (Meyer et al., 1999b). Transformation parameters were then estimated from individual integral (summed) [carbonyl-¹¹C]WAY-100635 PET images using SPM2 (Friston et al., 1995). These transformation parameters were then applied to dynamic [carbonyl-¹¹C]WAY-100635 PET images, frame by frame, using in-house software based on SPM2 routines. Finally, a set of ROIs were created in the standard space, and applied onto each individual spatially normalized dynamic [carbonyl-¹¹C]WAY-100635 PET image to yield regional tissue time–activity curves, using Imadeus software (version 1.2, Forima Inc., Turku, Finland). Appropriate ROI placement was ensured by visual inspection of every slice from all subjects. Similar strategies employing spatial normalization and a standard ROI set have been previously used to quantify [carbonyl-¹¹C]WAY-100635 binding (Rabiner et al., 2002), and at our laboratory, there is good convergence between values from the automated and traditional manual ROI analyses (J. Hirvonen and J. Hietala, unpublished observations). ROIs covered the following brain regions: amygdala, anterior cingulate cortex, dorsal raphe nuclei, dorsolateral prefrontal cortex, angular gyrus, inferior, middle, and superior temporal gyri, medial prefrontal cortex, orbitofrontal cortex, hippocampus, insular cortex, supramarginal gyrus, ventrolateral prefrontal cortex, and posterior cingulate cortex. The ROI for dorsal raphe nuclei was drawn directly on the [carbonyl-¹¹C]WAY-100635 template, since this structure is not readily visible in MR images. Cerebellar white matter was used as the reference region (Hirvonen et al., 2007; Parsey et al., 2005). Prior to modelling, contributions of total blood radioactivity to regional tissue time–activity curves was eliminated by assuming 5% blood volume in the ROI and subtracting it directly from regional radioactivity.

Quantification of [carbonyl-¹¹C]WAY-100635 availability

A linearized method based on non-negative least squares optimization was used to calculate regional total distribution volumes (V_T), according to the standard two-tissue compartmental model (2TM) (Hirvonen et al., 2007). Binding potential (BP) values were indirectly estimated from V_T values of ROIs and the reference region (Hirvonen et al., 2007). Two commonly used estimates of BP were considered: BP_P=V_T[ROI]−V_T[ref] and BP_ND=V_T[ROI]/V_T[ref]−1 (BP_P/V_T[ref]). BP_P is proportional to f_PB_avail/K_D (where f_P is the fraction of free or non-protein bound radiotracer in the plasma, B_avail is the total concentration for receptors, and K_D is the apparent equilibrium dissociation constant), whereas BP_ND is proportional to f_NDB_avail/K_D, where f_ND is the fraction of radioactivity originating from free radiotracer in the non-displaceable tissue compartment (Innis et al., 2007). The free fraction f_P was not measured in the present study.

Voxel-based analyses

To facilitate detailed visualization of the results, a confirmatory voxel-based analysis was performed. First, individual parametric BP_P and BP_ND maps were calculated by estimating V_T values voxel-wise using a linearized method based on non-negative least squares optimization and then performing the BP calculations voxel-wise, using the cerebellar white-matter V_T from the automated ROI analysis as an estimate of V_T[ref]. Preprocessing and statistical analysis was performed using SPM2 (Friston et al., 1995) running on Matlab 6.5 for Windows (MathWorks, Natick, MA, USA). Parametric BP maps were then spatially normalized into standard space using normalization parameters estimated for the automated ROI analysis (see above). Spatially normalized parametric BP maps were then smoothed using a 12-mm Gaussian kernel. Group differences were analysed using a two-sample t test option in SPM2. As BP values are quantitative in nature, no scaling of voxel-wise BP values was performed. Images were masked using the image global value as threshold, and zeros were ignored. Height threshold was set at p<0.05 (uncorrected voxel level), and extent threshold at 100 voxels. A cluster-level corrected p value of 0.05 was considered the criterion for statistical significance.

Statistical analyses of ROI-based data

Statistical analyses were performed on the natural logarithm of BP_P and BP_ND values, to reduce non-normality of distribution of BP values, as well as to facilitate direct comparison with a recent report on MDD subjects (Parsey et al., 2006a). Statistical analyses were performed using SPSS 13.0 for Windows (Release 13.0.1, SPSS Inc., Chicago, IL, USA). Data were analysed by means of repeated-measures analysis of variance (rmANOVA) with region (ROI) and hemisphere as within-subject factors and age as the between-subject predictor of BP. Group×region interaction was modelled to investigate regionally specific group differences in BP values. Sex entered the model as a covariate (Parsey et al., 2002). Age was not included in the model: it does not appear to influence the BP of [carbonyl-¹¹C]WAY-100635 in large samples of human subjects (Parsey et al., 2002; Rabiner et al., 2002). Consistent with published reports, we did not observe any effects of age on BP₁ in a sample of 36 healthy volunteers (aged 18–48 yr) scanned with [carbonyl-¹¹C]WAY-100635 at our laboratory (R²=0.00–0.07) (unpublished observations). We also tested for the effects of age, sex, body mass index, and years of education on the mean post-synaptic BP_P of [carbonyl-¹¹C]WAY-100635 by applying a linear regression model in the study groups separately, but none of the predictors were statistically significant (all p>0.12). To control for positive bias due to violation of the sphericity assumption associated with interaction terms including within-subject factor with three or more levels, Mauchly's tests of sphericity was performed and degrees of freedom in the averaged tests of significance were adjusted using the Greenhouse–Geisser method. A separate linear regression model (with group status and sex as regressors) was built for dorsal raphe nuclei, which is a midline structure. As a methodological validation, cerebellar white-matter V_T values were compared between the study groups to investigate possible group differences in V_T[ref]. Effect sizes for group differences were calculated as the absolute difference between MDD subjects and healthy controls divided by the standard deviation in the healthy control group. The associations between regional [carbonyl-¹¹C]WAY-100635 BP and total BDI and HAMD scores (being scale variables) was analysed using the parametric Pearson correlation coefficients, and the association between BP and individual HAMD subscores (being ordinate variables) was tested using the non-parametric Spearman correlation coefficient. Data are presented as mean±standard deviation unless otherwise specified. p values of <0.05 were considered criteria for statistical significance.

Results

Group differences in 5-HT_1A binding potential

MDD patients tended to have lower V_T[ref] values compared with controls (0.40±0.12 vs. 0.47±0.08, p=0.055, equal variances assumed). Thus, BP_P was selected as the outcome measure of choice, given the relative independence of this measure to changes in V_T[ref] as compared with BP_ND and the fact that f_P is not different between MDD patients and healthy controls (Meltzer et al., 2004; Parsey et al., 2006a).

The rmANOVA with group status and sex as the between-subject predictors and region and hemisphere as within-subject predictors of BP_P showed a significant main effect of group (F=5.00, p=0.032) but no effects of sex (F=1.25, p=0.271). Group×region and group×hemisphere interactions were not statistically significant, thus implying that the group difference was similar in magnitude across brain regions and in both hemispheres. BP_P in dorsal raphe nuclei was tested separately: the difference did not reach statistical significance in this region (t=−1.36, p=0.184). In all brain regions, MDD subjects had lower BP_P values compared with healthy controls (Figure 1). Standardized effect sizes for group differences were on average −0.69 (range −0.86 to −0.33) and percentage decrements were on average −18.7% (range −24.9% to −8.5%). Similar results were obtained by using cerebellar grey matter as reference region (−18.8%, average standardized effect size −0.67, main effect of group in the rmANOVA: F=3.94, p=0.056).

Figure 1

Mean [carbonyl-¹¹C]WAY-100635 BP₁ values across brain regions in healthy control subjects (□) and subjects with MDD (■) from the automated ROI analysis. Error bars represent standard deviation. AC, Anterior cingulated cortex; AMY, amygdala; ANG, angular gyrus; DLP, dorsolateral prefrontal cortex; ITG, inferior temporal gyrus; MTG, middle temporal gyrus; STG, superior temporal gyrus; HIP, hippocampus; INS, insular cortex; MFC, medial prefrontal cortex; ORB, orbitofrontal cortex; PC, posterior cingulated cortex; DRN, dorsal raphe nuclei; SG, supramarginal gyrus; VLP, ventrolateral prefrontal cortex.

The widespread decrease in BP_P values in MDD subjects compared with healthy controls was confirmed with the voxel-based analysis showing a single, very large cluster encompassing many regions of the brain known to express 5-HT_1A receptors (Figure 2).

Figure 2

Voxel-based analysis of parametric BP₁ maps revealed a single large cluster representing lower BP₁ in major depressive disorder subjects compared with healthy comparison subjects (k_E=123 746 voxels, cluster-level corrected p=0.013). The results are visualized on transaxial slices of the T1-weighted MRI template, and the colour bar represents the t value at the voxel level.

No statistically significant differences in BP_ND values were observed between the study groups: in the rmANOVA, the main effect of group was not statistically significant (F=0.62, p=0.436), neither was the group×region interaction (F=0.818, p=0.658). This phenomenon is consistent with MDD subjects having lower cerebellar white-matter V_T values, reflecting a change in free and non-specifically bound radiotracer in the brain, and making BP_ND an insensitive outcome measure to detect differences in specific binding to 5-HT_1A receptors.

Correlations of 5-HT_1A binding potential with clinical variables

The BP_P of [carbonyl-¹¹C]WAY-100635 did not show significant correlations either with the BDI total scores (R=−0.15 to 0.20, p=0.399–0.996) or the HAMD total scores (R=−0.23 to −0.88, p=0.326–0.704). In a post-hoc analysis of HAMD subscores, scoring on early and late insomnia were negatively correlated with BP_P in many brain regions (Table 2), suggesting that the lower the BP_P, the higher was the likelihood of suffering from insomnia. Duration of illness was not associated with BP_P (R=−0.02–0.20, p=0.377–0.925), nor were there any significant differences in BP_P between recurrent (n=4) and first-episode patients (n=17) (t=−0.58 to 0.26, p=0.570–0.988).

Table 2

Associations between [carbonyl-¹¹C]WAY-100635 BP₁ values and scoring on HAMD subscores representing early and late insomnia

Statistically significant associations are in bold.

BP, Binding potential; HAMD, Hamilton Depression Rating Scale; p, p value associated with the Spearman's correlation coefficient; ρ, Spearman's correlation coefficient.

Discussion

Patients with MDD manifest with a widespread reduction of serotonin 5-HT_1A receptor availability in the brain, as assessed using PET and [carbonyl-¹¹C]WAY-100635. To minimize the confounding contributions of previous antidepressant or psychotherapeutic treatments, all but one patient had been never treated before. Consequently, this represents the largest sample of medication-naive MDD subjects assessed with [carbonyl-¹¹C]WAY-100635 published to date. The patient sample described herein represents mild to moderate MDD with no comorbidities.

5-HT_1A receptors in patients with major depression vs. healthy comparison subjects

There have been previous reports on [carbonyl-¹¹C]WAY-100635 measurements with PET in subjects with MDD (Bhagwagar et al., 2004; Drevets et al., 1999; Meltzer et al., 2004; Parsey et al., 2006a; Sargent et al., 2000). These studies are summarized in Table 3. Drevets et al. (1999) studied 12 depressed patients (of whom eight were previously medicated patients with familial MDD; others were suffering from bipolar illness) and found decreased BP_ND values compared with controls. Sargent et al. (2000) found decreased BP_ND values in medication-naive (n=8), medication-free (n=8), and currently treated (n=20) patients compared with controls. Meltzer et al. (2004) studied 17 elderly patients with late-life depression, both medication-naive and previously treated, and found decreased BP_P values in the dorsal raphe nuclei. Bhagwagar et al. (2004) studied 14 medication-free male subjects recovered from MDD and found persistently reduced BP_ND values. In contrast to previous studies, Parsey et al. (2006a) recently compared 13 medication-naive subjects with MDD to 43 healthy controls and found increased BP_F and BP_P, but not BP_ND, values throughout the brain. As a group, patients with MDD (those who had received medication and those who had not) did not differ from controls, thus implying a possible modulatory role of antidepressant drug treatment on 5-HT_1A availability in MDD (Parsey et al., 2006a). While methodological issues, such as whether arterial input function was used in modelling, could theoretically explain the discrepancy between earlier studies and that by Parsey and colleagues (2006a), they are less likely to account for the contrasting findings in BP_P values between the present study and that of Parsey et al. (2006a). Clinical heterogeneity regarding illness severity and duration of current depressive episode is another issue that can be held accountable for the discrepancy (Table 3).

Table 3

A summary of PET imaging studies using [carbonyl-¹¹C]WAY-100635 in patients with MDD experiencing a depressive episode vs. healthy comparison subjects. Numbers denote mean±s.d., n, or n/n

BP, binding potential; DRN, dorsal raphe nuclei; HAMD, 17-item Hamilton Depression Rating Scale; n.a.., not available; ↓, lower 5-HT_1A binding in MDD vs. healthy controls. ↑, higher 5-HT_1A binding in MDD vs. healthy controls.

a

With total symptom scores. ^b Number of untreated first-episode patients. ^c In drug-naive MDD patients vs. controls.

Interestingly, it was recently demonstrated that adult monkeys characterized with depressive behaviour show a widespread reduction of 5-HT_1A receptor availability in the brain compared to normal monkeys, as assessed using PET and [¹⁸F]MPPF, another radiotracer for 5-HT_1A receptors (Shively et al., 2006). Although the similarities in the pathophysiology of depressive behaviour between monkeys and humans could be questioned, this finding lends support for a common neural substrate for liability to develop MDD that might have phylogenetic underpinnings.

5-HT_1A receptors and clinical characteristics of major depression

Generally [carbonyl-¹¹C]WAY-100635 availability did not correlate with BDI or HAMD total scores in any brain region in patients with MDD, although it should be emphasized that the overall severity of illness in our study ranged from mild to moderate: we cannot exclude correlations that might have occurred by including patients with more severe forms of MDD. There might also be a low threshold in the severity of MDD after which 5-HT_1A receptors down-regulate. Lack of correlations suggests that reduced 5-HT_1A receptor availability may represent a trait variable, that is associated with the vulnerability to develop MDD rather than with the clinical phenotype. Consistent with this view, Bhagwagar et al. (2004) found persistently reduced 5-HT_1A receptor availability in euthymic males recovered from MDD, although this interpretation is limited by the fact that these patients had previously undergone SSRI drug treatment. This view is also supported by the fact that previous PET studies have generally not found correlations with symptom severity and 5-HT_1A receptor availability (Table 3). However, we did find inverse correlations between [carbonyl-¹¹C]WAY-100635 BP and scoring on HAMD items representing early and late insomnia: thus, lower BP was associated with a greater likelihood of suffering from insomnia. Moreover, stronger associations were found with early (trouble getting to sleep) rather than late (waking up early) insomnia (Table 2). This is congruent with the proposed role of 5-HT function in the regulation of sleep (Adrien, 2002; Sharpley and Cowen, 1995; Thase, 2006; Tsuno et al., 2005; Ursin, 2002). Indeed, 5-HT has been suggested to modulate the sleep–wake cycle through reciprocal connections between the raphe nuclei and multiple brain regions involved in the homeostatic regulation of sleep. These regions include the thalamic reticular nuclei, the preoptic/basal forebrain area, the suprachiasmatic nucleus, and hypothalamus (Ursin, 2002). Generally, it is assumed that increased 5-HT function, as indicated by increased firing of raphe neurones and increased synaptic 5-HT content, promotes wakefulness and decreased 5-HT function promotes sleep (Adrien, 2002; Ursin, 2002), and there is some direct human evidence from neuroimaging studies to support this (Derry et al., 2006). However, this concept is too simplistic in the face of the complexity of serotonergic modulation (Adrien, 2002; Ursin, 2002). As to specific receptor subtypes, stimulation of thalamic 5-HT₂ receptors appears to promote wakefulness, whereas the functional role of 5-HT_1A appears to depend on the regional and ultrastructural localization of the receptors (Adrien, 2002; Boutrel et al., 2002; Muraki et al., 2004; Ursin, 2002). It has been hypothesized that insomnia in depression represents a compensatory attempt to reverse deficient 5-HT function, since sleep deprivation increases 5-HT function (Adrien, 2002). Our findings of reduced post-synaptic 5-HT_1A receptor availability, possibly reflecting deficient 5-HT function, associated with the probability of insomnia is compatible with this view. Regarding state vs. trait contributions, it may be that some aspects of the clinical phenotype of MDD are state-dependently modulated by 5-HT_1A function and the particular facets of the wide array of symptoms under the influence of 5-HT_1A may vary from one patient group to another. As an example of this, psychic and somatic anxiety was found to be differentially associated with 5-HT_1A availability in a recent study of patients with MDD (Sullivan et al., 2005), whereas no such associations were detected in the present study (although patients in the present study did not have many anxiety symptoms, let alone have comorbid anxiety disorder). Taken together, however, post-hoc analyses on individual HAMD items carry a risk of type I errors through multiple comparisons and thus, these correlations must be interpreted with caution.

Lower 5-HT_1A receptor availability may not be specific to MDD, but may rather represent a common trait for vulnerability towards disorders in the anxiety-depression spectrum. In support of this hypothesis, recent studies have demonstrated reduced 5-HT_1A receptor availability in panic disorder (Neumeister et al., 2004) and social anxiety disorder (Lanzenberger et al., 2007). Interestingly, the study by Neumeister et al. (2004) did not find any differences between those with and without concurrent comorbid depressive episode, suggesting that the depressive phenotype does not per se additionally affect 5-HT_1A levels. The exact neurobiological phenomena responsible for 5-HT_1A down-regulation in anxiety-depression disorders remain to be revealed.

Methodological issues

The outcome measure BP_P reflects the product of total number of receptors (B_avail), the apparent affinity for the radioligand (K_D), and the fraction of radioactivity originating from free (non-protein bound) radioligand in arterial plasma (f_P, ‘free fraction’). Patients with MDD had lower distribution volumes for the free and non-specifically bound radiotracer in the brain. Previously, Meltzer et al. (2004) also found lower V_T[ref] in depressed patients, whereas Parsey et al. (2006a) noticed slightly higher V_T[ref] in never-medicated depressed patients, as compared with control subjects. The explanation for altered V_T[ref] in patients with MDD is currently unknown. At equilibrium, free radiotracer in the plasma equals the free radiotracer in the non-displaceable tissue compartment and thus, V_T[ref] equals f_P/f_ND. Altered plasma free fraction f_P could theoretically contribute to this phenomenon, but previous studies have not found differences in f_P between MDD subjects and healthy controls (Meltzer et al., 2004; Parsey et al., 2006a). Higher tissue free fraction f_ND in MDD patients remains a possibility, but cannot be directly estimated without f_P measurements. These observations suggest that BP_ND is not an appropriate outcome measure for estimating specific binding of [carbonyl-¹¹C]WAY-100635 to 5-HT_1A in patients with MDD, and thus, BP_P was preferred in the present study. To evaluate whether lower V_T[ROI] and V_T[ref] in MDD patients is secondary to faster peripheral clearance of the unchanged tracer, we calculated total peripheral clearance for each subject by dividing the injected dose with the area under the metabolite-corrected plasma curve (Hirvonen et al. 2007); however, no significant differences were noted between patients and controls (91 vs. 101 l/h respectively, p=0.198). It has been reported that the availability of [carbonyl-¹¹C]WAY-100635 is unchanged by acute or chronic manipulations of endogenous synaptic 5-HT concentration in human (Moses-Kolko et al., 2007; Rabiner et al., 2002; Sargent et al., 2000), which suggests that the results may reflect receptor density, or conformational changes in the receptor structure leading to altered affinity to bind [carbonyl-¹¹C]WAY-100635.

Conclusions

In conclusion, a widespread reduction of brain 5-HT_1A receptor availability was found in patients with MDD in the present study, thus confirming previous observations in a relatively large sample of medication-naive patients. Lower 5-HT_1A receptor availability in MDD was not related to overall symptom severity in this sample of mild to moderate depression. It is possible that it represents a trait-dependent marker of vulnerability for depression and possibly anxiety disorders rather than directly affecting the clinical phenotype.

Acknowledgements

The staff of Turku PET Centre and the MRI Unit of Turku University Central Hospital are acknowledged for skilful assistance in performing PET and MRI scanning. This study was financially supported by Signe and Ane Gyllenberg Foundation, The Social Insurance Institution of Finland, and the Turku University Central Hospital (grant P3848).

Statement of Interest

Dr Hirvonen has received lecture fees from AstraZeneca, Bristol–Myers Squibb, Janssen-Cilag, Lundbeck, and Novartis, congress travel grants from AstraZeneca and Lundbeck, and research funding from Orion Pharma and Lundbeck. Dr Karlsson has received lecture fees from AstraZeneca, Eli Lilly, GlaxoSmithKline, Janssen-Cilag, Lundbeck, and Wyeth. Dr Markkula has received lecture fees from Eli Lilly, GlaxoSmithKline, and Janssen-Cilag. Dr Hietala has received lecture fees from AstraZeneca, Bristol–Myers Squibb, Eli Lilly, Janssen-Cilag, and Lundbeck, congress travel grants from AstraZeneca, Bristol–Myers Squibb, and Eli Lilly, and has acted as a consultant for Orion Pharma.