INTRODUCTION

Positron emission tomography (PET) studies examining pindolol occupancy of the 5-HT1A receptor in healthy volunteers in vivo have found a higher occupancy at the autoreceptor as compared to the postsynaptic receptor (Martinez et al, 2001; Rabiner et al, 2000). Similarly, a preferential occupancy at the autoreceptor has been reported in a PET study in the rat (Hirani et al, 1999), as well as in some (Castro et al, 2000; Serrats et al, 2000), but not all (Raurich et al, 1999), in vitro human and rat studies. The significance of this finding is not clear, as the available evidence indicates that the 5-HT1A autoreceptor and postsynaptic receptor are structurally identical (Albert et al, 1990; Radja et al, 1992). On the other hand, considerable evidence exists for differences between the autoreceptors and postsynaptic receptors in vivo, such as differences in receptor reserve between the two sites (Meller et al, 1990).

An earlier study from our unit (Rabiner et al, 2001) reported the 5-HT1A autoreceptor occupancy by pindolol in depressed patients on selective serotonin re-uptake inhibitors (SSRIs). An incidental finding of that study, not reported in our original publication, but presented here, was the attenuation of the expected preferential occupancy of the 5-HT1A autoreceptor in this group. Activation of the 5-HT1A autoreceptors inhibits 5-HT neurone firing, and hence 5-HT release in terminal synapses (Adell and Artigas, 1991; Invernizzi et al, 1992; Sharp et al, 1989; VanderMaelen et al, 1986). Activation of the postsynaptic 5-HT1A receptors mediates some of the effects of 5-HT released from the nerve terminals (Blier et al, 1987; Chaput et al, 1991; Invernizzi et al, 1991; Sinton and Fallon, 1988), and may be critical to the antidepressant action of SSRIs. As preferential binding at the autoreceptor may be crucial for the proposed mechanism of action of pindolol in augmenting SSRIs (Artigas et al, 2001; Martinez et al, 2001; Rabiner et al, 2000), we further investigated pindolol 5-HT1A occupancy effects in a larger group of controls and depressed patients.

In this study we used [11C]WAY-100635 PET to examine a further group of seven depressed patients on SSRIs treatment before and after a single oral dose of 10 mg of pindolol, a dose known to produce consistent preferential occupancy in healthy volunteers (Rabiner et al, 2000). This group of seven depressed patients was compared to a group of seven healthy volunteers, who were also examined with [11C]WAY-100635 PET before and after a single oral dose of 10 mg of pindolol.

MATERIALS AND METHODS

Subjects

All studies were approved by the Imperial College School of Medicine Ethics Committee and the Administration of Radioactive Substances Advisory Committee. All subjects underwent a psychiatric interview with a qualified psychiatrist and a full physical examination, and gave written informed consent to the study. Depressed patients were treated with a variety of SSRIs (Paroxetine 20–40 mg, Fluoxetine 20–40 mg, Sertraline 50–100 mg), or Venlafaxine (75 or 150 mg). The 17-item Hamilton Depression Rating Scale and the 21-item Beck's Depression Questionaire were collected from all depressed subjects at the time of the pindolol scan.

Study 1

The methodology of this study has been described in our previous publication (Rabiner et al, 2001). Briefly, eight patients (seven M, one F, ages 24–61) meeting DSM-IV criteria for MDD, on therapeutic doses of antidepressant medication only (SSRI n=7, Venlafaxine n=1), and not fully recovered were examined. All subjects received two open label [11C]WAY-100635 PET scans. The first was a baseline scan (on antidepressant treatment only), while the second was conducted following a 7–14 days period of augmentation with pindolol 2.5 mg t.i.d. (n=4), or 5 mg t.i.d. (n=4). Although the autoreceptor occupancy by pindolol in this group was reported previously, preferential occupancy was not reported, and forms the focus of this paper.

Study 2

Seven patients on antidepressant treatment (SSRI n=5, Venlafaxine n=2) were included in this study. All subjects met the DSM-IV criteria for a major depressive episode. The control group consisted of seven medication-free healthy volunteers. Four of the healthy volunteers were recruited and examined previously (Rabiner et al, 2000), while the other three were newly recruited for this study. All subjects received two open label [11C]WAY-100635 PET scans (median interval 8 days, range 3–258 days). The first was a baseline scan, while the second was conducted 2 h following 10 mg p.o. of pindolol.

The demographic details of all subjects for both studies are summarized in Table 1.

Table 1 Subject Details
Full size table

PET Data Acquisition

PET scans were performed on an ECAT 953B PET camera (CTI/Siemens, Knoxville, TN, USA) (Spinks et al, 1992) in three-dimensional mode with dual window scatter correction (Grootoonk et al, 1996) and a measured attenuation correction. [11C]WAY-100635 was prepared at the Cyclotron Unit by 11C-carboxylation of a Grignard reagent (McCarron et al, 1996), and injected intravenously as a bolus (activity injected 334±69 MBq, mass of WAY 100635 injected 2.85±2.24 μg). In order to assess the influence of pindolol on the metabolism of [11C]WAY-100635, three venous samples were collected at 1000, 2000, and 3000 seconds and percentage of injected [11C]WAY-100635 estimated at each time point. In addition, venous blood samples were taken for estimation of plasma levels of pindolol at the time of the PET scan.

PET Data Analysis

The [11C]WAY-100635 PET scans were analyzed using a reference tissue compartmental model, with the cerebellum as a reference tissue as described previously (Gunn et al, 1997, 1998; Lammertsma and Hume, 1996; Rabiner et al, 2002a). The reference tissue model allows the estimation of binding potential (BP=f2 BAVAIL/KD, where f2 is the ‘free fraction’ of the radioligand in the tissue not specifically bound, BAVAIL is the concentration of available binding sites and KD is the equilibrium dissociation rate constant of the radioligand (Cunningham and Lammertsma, 1994)) and the ratio of radioligand delivery in the region of interest relative to the reference region (RI). Occupancy of the 5-HT1A receptor sites was inferred as a reduction of BP, and hence BAVAIL, under the assumption that f2 and KD remain constant for the two scans.

Brain regions were defined as described previously (Rabiner et al, 2002a). Briefly, cortical and limbic regions were defined via the warping of a brain region map drawn on an MR image in standard space, onto an individual subject image. The midbrain raphe nuclei (RN) region is a small region readily seen on a [11C]WAY 100635 PET image, but not apparent on an MR image. The raphe region was therefore manually defined as a fixed size region (900 mm3), on individual subject PET images summed over 20–90 min of the scan. Occupancy was calculated for the 5-HT1A autoreceptor (OCCAUTO, inferred from the occupancy of the midbrain RN) and the 5-HT1A postsynaptic receptor (OCCPOST inferred from the average occupancy of the cortical and limbic regions of interest). Preferential Occupancy (PREFocc)was defined as:

Plasma Data Analysis

Plasma pindolol was measured at the time of injection of the radioligand and at 15 min postradioligand injection. A mean pindolol plasma value was calculated for the first 15 min of the PET scan, and this value was correlated to the occupancy levels achieved.

Statistics

A difference in BP (ΔBP as per Eq. 3)) was calculated for the autoreceptor and postsynaptic regions, from a database of 15 healthy volunteers scanned on two occasions (Rabiner et al, 2002a).

OCCAUTO and OCCPOST for the pindolol groups were compared to the ΔBPAUTO and ΔBPPOST from the test–retest group using a one-way ANOVA, with post hoc testing by a Dunnet's t-test (two sided). PREFOCC was compared in the same way to the ΔBPauto–ΔBPpost of the test–retest group. In addition, a direct comparison of PREFOCC between the healthy volunteer and the depressed groups of study 2 was performed via a Student's t-test.

Changes in the RI were assessed in the same way as the changes in BP. The change in the percent of parent WAY-100635 following pindolol administration was assessed via a repeated measures ANOVA, with time being the within subject factor, and experimental group being the between subject factor.

[11C]WAY-100635 binding was quantified by the simplified reference tissue model, making it important to examine the reference region (here the cerebellum) for changes in radioligand kinetics that may lead to erroneous conclusions. Although changes in the nonspecific binding of the radioligand in the cerebellum cannot be quantified with a reference region model, a change in the shape of the cerebellar time–activity curves (Cb TACs) would be consistent with alterations in the kinetics of [11C]WAY 100635 nonspecific binding. The Cb TAC for each subjects scan1 and scan 2 were each normalized to their own peak, and the peak normalized TAC for scan 2 subtracted from the corresponding TAC for scan 1, generating a difference TACs (Diff TAC). The same procedure was performed on the 15 subjects of the test–retest group. The Diff TACs of the subjects who received pindolol were compared to the Diff TACs of the test–retest group using a repeated measures ANOVA (time being a within subject factor and group a between subject factor) to assess the effect of pindolol on the time course of the radioligand in the cerebellum.

All statistics were performed on SPSS version 10.1.

RESULTS

Study 1

The data from Study 1 are presented in Table 2. The mean injected dose of [11C]WAY 100635 in Study 1 was 333±65 MBq (mean mass of WAY 100635 injected 3.4±2.8 μg). There were no significant differences in either the injected dose of [11C]WAY 100635 or injected mass of WAY 100635 between the baseline and pindolol scans (injected dose mean difference 21±98 MBq, injected mass mean difference 0.07±2.35 μg). One-way ANOVAs compared pindolol occupancy at the autoreceptor (OCCAUTO) and postsynaptic (OCCPOST) receptor sites, as well as preferential occupancy (PREFOCC) in the following groups: 15 healthy volunteers Test–Retest (data from Rabiner et al (2002a)), four depressed patients following a 7–14-day course of pindolol augmentation (7.5 mg daily), and four depressed patients following a 7–14-day course of pindolol augmentation (15 mg daily). Pindolol had a significant effect on OCCPOST (F2,26=11.02, p=0.001) but not OCCAUTO (F2,26=2.02, p=0.159) or PREFOCC (F2,26=1.94, p=0.170). Post hoc Dunnett's t test (two sided) revealed a significant effect of the 15 mg (p<0.001) but not the 7.5 mg dose (p=0.054) on OCCPOST.

Table 2 5-HT1A Receptor Occupancy Following Repeated Doses of Pindolol (Study 1)
Full size table

An analogous analysis of the RI values demonstrated no significant effect of pindolol on OCCPOST (F2,26=0.62, p=0.550), OCCAUTO (F2,26=1.73, p=0.203), or PREFOCC (F2,26=2.77, p=0.087).

An examination of the differences in cerebellar time–activity curves before and after pindolol administration revealed no effect of pindolol on the binding of [11C]WAY-100635 in the reference region (main effect of group F2=1.89, p=0.177). Examination of the metabolism of [11C]WAY 100635 over the period of 1000–3000 s after injection revealed no main effect of time (F1.41=3.50, p=0.071), or group (F2=0.032, p=0.969), but a time by group interaction (F2.82=4.80, p=0.018).

Study 2

The data from Study 2 are presented in Table 3. The mean injected dose of [11C]WAY 100635 in study 2 was 335±72 (mean mass of WAY 100635 injected 2.5±1.8 μg). There were no significant differences in either the injected dose of [11C]WAY 100635 or injected mass of WAY 100635 between the baseline and pindolol scans (injected dose mean difference 13±85 MBq, injected mass mean difference −0.51±2.35 μg). One-way ANOVAs compared pindolol occupancy at the autoreceptor (OCCAUTO) and postsynaptic (OCCPOST) receptor sites, as well as preferential occupancy (PREFOCC) in the following groups: 15 healthy volunteers Test–Retest (data from Rabiner et al, 2002a), seven healthy volunteers following a single 10 mg dose of pindolol and seven depressed patients on SSRIs following a single 10 mg dose of pindolol. Pindolol had a significant effect on OCCPOST (F2,26=13.64, p<0.001), OCCAUTO (F2,26=11.68, p<0.001), and PREFOCC (F2,26=6.40, p=0.005). Post hoc Dunnett's t tests (two sided) revealed that pindolol 10 mg single dose had a significant effect in both the patients and the healthy volunteers on OCCPOST (p<0.001 and p=0.005 respectively) and the OCCAUTO (p<0.009 and <0.001 respectively). By contrast, PREFOCC was significantly different only for the healthy volunteers (p=0.005), but not for patients (p=0.994).

Table 3 5-HT1A Receptor Occupancy Following a Single Dose of Pindolol (Study 2)
Full size table

An analogous analysis of the RI values demonstrated no significant effect of pindolol on OCCPOST (F2,26=0.25, p=0.779), but both OCCAUTO (F2,26=4.80, p=0.017) and PREFOCC (F2,26=4.38, p=0.023) were significantly different. Post hoc Dunnett's t test (two sided) revealed no significant effects in either of the groups, though the patient group approached significance for both OCCAUTO (p=0.054) and PREFOCC (p=0.051).

An examination of the differences in cerebellar [11C]WAY-100635 time–activity curves before and after pindolol administration revealed no effect of pindolol on the time course of [11C]WAY-100635 in the reference region (main effect of group F2=1.358, p=0.703). Examination of the metabolism of [11C]WAY 100635 over the period of 1000–3000 s after injection revealed no main effect of time (F1.10=0.225, p=0.665), or a time by group interaction time (F2.20=0.661, p=0.544). There was however a main effect of group (F2=12.24, p=0.001). Post hoc tests revealed a significant increase in percent of parent [11C]WAY 100635 in the depressed patients on SSRIs following a single 10 mg dose of pindolol (p=0.011), but not in healthy volunteers.

Preferential Occupancy

Overall, these results demonstrate a preferential occupancy of the 5-HT1A autoreceptor compared to the postsynaptic receptor by pindolol, in healthy volunteers, but not in depressed patients on SSRIs (Figure 1). The loss of preferential occupancy, in depressed patients on SSRI treatment is confirmed by the comparison of PREFOCC in healthy volunteers (mean PREFOCC=22.6±7.7%) and depressed patients following a 10 mg single dose of pindolol (mean PREFOCC=2.9±10.8%, Student's t=3.94, df=12, p=0.002).

Figure 1
figure 1

Preferential occupancy of 5-HT1A autoreceptors by pindolol. Mean±SD of healthy volunteers (solid bars) and depressed patients (open bars). Striped bar represents healthy volunteer data from Martinez et al (2001).

Full size image

Pindolol Plasma Levels

Plasma pindolol levels were measured as previously detailed (Rabiner et al, 2000). There were no significant correlations between the plasma pindolol levels and either autoreceptor, postsynaptic receptor or preferential occupancy (Pearson's r; −0.13 (p=0.58), 0.18 (p=0.45) and −0.33 (p=0.16) respectively). In study 2, the mean plasma pindolol concentration was notably higher in the patient (29.7±22.8 ng/ml) as opposed to the healthy volunteer group (13.4±8.6 ng/ml), though this difference was not statistically significant (t=−1.74, df=7.87, p=0.12). The higher plasma pindolol levels did not lead to a significant difference in either the autoreceptor or postsynaptic receptor occupancy (t=1.83, df=12, p=0.09, and t=−0.86, df=12, p=0.41 respectively).

DISCUSSION

Pindolol is a mixed β-adrenergic/5-HT1A partial agonist (Clark et al, 1982; Clifford et al, 1998; Frishman et al, 1979; Hjorth and Carlsson, 1986; Meltzer and Maes, 1996; Newman-Tancredi et al, 1998; Sanchez et al, 1996; Sprouse et al, 2000) in several in vivo and in vitro assays, which has been proposed to accelerate and augment the antidepressant effects of SSRIs via the blockade of 5-HT1A autoreceptors (Artigas, 1993; Blier and de Montigny, 1994; Hjorth and Sharp, 1993), though an effect through its adrenergic activity cannot be excluded. The 5-HT1A receptor is a G-protein coupled receptor (GPCR) with seven trans-membrane domains, and exists as both a somatodendritic autoreceptor expressed on the 5-HT cell bodies and dendrites in the midbrain RN (Riad et al, 2000), and a postsynaptic heteroreceptor expressed in cortical and limbic areas (Hall et al, 1997; Pazos et al, 1987; Pike et al, 1996), on neuronal soma, especially perisynaptically. 5-HT1A receptors are also present on astroglia (Azmitia et al, 1996), though the density of non-neuronal 5-HT1A receptors may be considerably lower (Kia et al, 1996; Verge et al, 1986). The 5-HT1A autoreceptors control the release of 5-HT in the forebrain projection areas (Sharp et al, 1989; Sprouse and Aghajanian, 1987), and the therapeutic action of selective SSRIs has been hypothesized to depend on the desensitization of these autoreceptors (Artigas, 1993; Blier and de Montigny, 1994; Hjorth and Sharp, 1993).

Autoreceptor desensitization leads to a decrease in the inhibitory control of serotonergic neuronal activity by the 5-HT1A autoreceptors, and therefore an enhanced 5-HT neurotransmission (Artigas et al, 1996). The proposal to accelerate, and/or augment the antidepressant actions of SSRIs via a blockade of RN autoreceptors (Artigas, 1993; Hjorth, 1993; Hjorth and Sharp, 1993), depends therefore on significant pindolol binding to the 5-HT1A autoreceptor site, but has been criticized on the grounds that pindolol would also bind to the postsynaptic 5-HT1A site and therefore counteract any proposed beneficial effects of an increase in forebrain 5-HT (Sprouse et al, 2000). While the receptor subtype through which 5-HT exerts its antidepressant action is not certain, considerable evidence points to postsynaptic 5-HT1A receptors being important in this regard. Though the subsensitivity of postsynaptic 5-HT1A receptors in depressed patients has been debated (Cowen et al, 1994; Meltzer and Maes, 1995), 5-HT1A postsynaptic heteroreceptors are activated by long-term antidepressant treatment (Blier et al, 1987, 1990; de Montigny and Aghajanian, 1978; Haddjeri et al, 1998), while the antidepressant actions of SSRIs may depend on the enhanced levels of synaptic 5-HT acting on these receptors (Blier et al, 1987; De Vry, 1995). As a blockade of postsynaptic 5-HT1A receptors may block the beneficial effects of raised 5-HT produced by 5-HT1A autoreceptor blockade, compounds blocking the two receptor populations equally may not be effective augmenting agents. Preferential occupancy of the 5-HT1A autoreceptor would therefore make pindolol an attractive candidate for a role in accelerating/augmenting antidepressant effects of SSRIs.

The fact of preferential binding by pindolol seems well established, but the mechanism is uncertain, with regional differences in the affinity status of the 5-HT1A receptor being a plausible hypothesis (Martinez et al, 2001; Rabiner et al, 2000). The extended ternary model of ligand-receptor binding (Samama et al, 1993) posits the receptor to exist in several conformational states with agonists binding preferentially to, and stabilizing some states selectively, and promoting the formation of the ternary receptor–ligand–G-protein complex while antagonists have equal affinity for all conformational states. As pindolol is a partial agonist it may be expected to bind preferentially to the ‘high-affinity’ (R*), compared to the ‘low-affinity’ (R) receptor conformation. If the ratio of R* sites is higher for the autoreceptor compared to the postsynaptic receptor, then pindolol will appear to have a preferential occupancy of the autoreceptor compared to the postsynaptic receptor, when examined with an antagonist radioligand.

Considerable evidence exists for differences in the functional response of RN as opposed to the cortical and limbic 5-HT1A receptors to pharmacological challenges (Romero et al, 1996; Sinton and Fallon, 1988), although not all studies have come to this conclusion (Corradetti et al, 1998). The most common explanation given is that the RN 5-HT1A receptors possess greater ‘receptor reserve’ than the cortical and limbic areas (Jolas et al, 1995; Meller et al, 1990; Yocca et al, 1992). The mechanistic Whaley model of GPCR–Agonist–G-protein interaction (Whaley et al, 1994) is equivalent to the earlier empirical Furchgott model (1966) (see Clark et al, 1999 for review) and renders the concept of ‘receptor reserve’ as ‘both misleading and irrelevant’ (Clark et al, 1999). An increased ratio of ‘high affinity’ sites, would however, provide a higher concentration of receptors that can be acted upon effectively by agonists, therefore increasing the efficacy of a partial agonist (Clark et al, 1999). Such a mechanism would provide an explanation for compounds with 5-HT1A agonist effects, having greater efficacy in the RN compared to the postsynaptic areas such as the hippocampus (Andrade and Nicoll, 1987; Fabre et al, 1997; Greuel and Glaser, 1992; Hjorth and Sharp, 1990; Sprouse and Aghajanian, 1988).

Alterations in the functional status of the 5-HT1A receptor may be caused either by the antidepressant treatment or the pathophysiology of depression. Reductions in [11C]WAY-100635 binding to human 5-HT1A auto and heteroreceptors in depressed patients have been reported in several in vivo studies (Sargent et al, 2000; Drevets et al, 1999; Parsey et al, 2002) and have been interpreted as a modest reduction in 5-HT1A receptor numbers in depressed patients. The functional significance of a 10–15% reduction in receptor number in these patients is unclear, especially considering recent studies which demonstrated a 70% blockade of the 5-HT1A receptor in healthy volunteers without significant side effects (Rabiner et al, 2002b). On the other hand, the functional consequences of an alteration of receptor–G-protein coupling may be much more pronounced, but not apparent in an examination using an antagonist radioligand, [11C]WAY 100635, to determine receptor density.

Chronic agonist stimulation, such as that induced by long-term administration of SSRIs, causes a desensitization of the somatodendritic 5-HT1A receptors, but not the postsynaptic receptors (see Hensler, 2003 for a review). Mechanisms at several levels may explain receptor desensitization without a downregulation of receptor numbers. Receptor level mechanisms, such as regulation at the level of G-protein coupled receptor kinases (GRKs), arrestin binding, and endocytosis appear to dominate receptor desensitization. Downstream effects, such as agonist-induced phosphorylation of G-proteins, phospholipase C, and adenylate cyclase, may become important as the duration of stimulation increases. The reason for the differential effects of SSRIs on 5-HT1A desensitization in the RN compared to the postsynaptic receptors is unclear, but may relate to the differential coupling of these receptors to G-proteins, and effector mechanisms. For instance, while frontal cortex 5-HT1A receptors are coupled to both Go and Gi3, those in the RN are coupled to Gi3 only (Hensler, 2003).

If we hypothesize that preferential occupancy is a consequence of differing proportions of ‘high-affinity’ and ‘low-affinity’ sites in the RN compared to the cortical regions, then the loss of preferential occupancy in depressed patients on chronic SSRIs would imply that in these patients there is a change in the proportions of ‘high-affinity’ and ‘low-affinity’ sites in the autoreceptor and the postsynaptic receptor regions. In fact, the decrease in the autoreceptor occupancy, with the relative preservation of postsynaptic receptor occupancy seen in the patients in this study (see Table 3), implies that the loss of preferential occupancy is a consequence of the loss of ‘high-affinity’ sites in the RN. This loss of ‘high-affinity’ sites can be a consequence of either the SSRI treatment or that of the illness. From the discussion above, chronic SSRI treatment may be expected to lead to receptor desensitization because of chronic stimulation of the receptor by the endogenous agonist (5-HT), as discussed above. The effects of depression on the efficacy status of 5-HT1A receptors are less well documented. Chronic ultramild stress (CUMS), a plausible model of depression, causes a desensitization to the effects of the 5-HT1A partial agonist ipsapirone in mice (Lanfumey et al, 1999). In addition, other forms of stress as well as the application of corticosterone to brain slices have been shown to desensitize RN 5-HT1A autoreceptors (Laaris et al, 1995, 1999).

An unpredicted result of this study was a significant negative correlation (Spearman's ρ=−0.728, N=14, p=0.003) between the severity of depression (as judged by the HAM-D score) and the PREFOCC (Figure 2). The correlation between the Beck's Depression Inventory (BDI) and PREFOCC did not reach significance (Spearman's ρ=−0.439, N=15, p=0.102). The BDI consists of Factor 1 (cognitive and mood items) and Factor 2 (somatic items) (Schotte et al, 1997). PREFOCC correlates significantly with the somatic items of the BDI (Factor 2, Spearman's ρ=−0.539, N=15, p=0.038) but not the cognitive and mood items BDI (Factor 1, Spearman's ρ=−0.472, N=15, p=0.075). These findings support the view that depressive illness affects pindolol occupancy and, by implication, the functioning of the RN 5-HT1A autoreceptors.

Figure 2
figure 2

Correlation of preferential occupancy and severity of depression: (a) correlation of PREFOCC with HAM-D. (b) correlation of PREFOCC with BDI, Factor 2.

Full size image

If the desensitization of RN 5-HT1A receptors is considered to be a necessary effect of SSRI treatment, one would expect that patients who respond to treatment (and therefore have lower HAM-D scores) will have a lower degree of PREFOCC. Examination of Figure 2, on the other hand, reveals the opposite, meaning that patients with a lower HAM-D score have higher PREFOCC than those with more severe symptoms. This finding raises the possibility that the loss of PREFOCC may be a result of the illness, and as such is an index of the severity of depression, rather than an index of the effects of SSRI treatment. If the attenuation of PREFOCC is an index of depression severity, it could provide a valuable surrogate in the investigation of the underlying mechanisms of depression. However, the attenuation of PREFOCC may represent an interaction of two separate mechanisms, with both SSRI mediated effects and the direct effects of depression playing a role.

A recent review (Blier, 2003) concluded that the evidence for acceleration of antidepressant effects by pindolol is much more convincing than an augmentation effect. If the attenuation of PREFOCC is, at least in part, a consequence of the effect of chronic SSRI administration, and preferential occupancy is important for the mechanism of action of pindolol, in augmenting antidepressant actions of SSRIs, it may be expected that patients who receive pindolol at the start of their treatment (concurrent with the SSRIs) will have a better clinical response than those that receive pindolol after a prolonged period of antidepressant treatment. Similarly, if the severity of depression correlates negatively with PREFOCC, then the less severe patients will benefit from pindolol augmentation more than the more chronic, treatment resistant patients. These suppositions have some support from clinical practice.

Further studies are needed to elucidate the contributions of SSRIs and depression severity, in the attenuation of PREFOCC by pindolol; however, the discussion above indicates that while augmentation of antidepressant effects of SSRIs by pindolol may be an efficacious intervention, only a subset of patients (antidepressant free subjects with mild to moderate depression severity) may benefit. In addition, as we have discussed previously (Rabiner et al, 2001), the dose of pindolol commonly used in clinical trials, 2.5 mg t.i.d., is inadequate to produce a consistent occupancy of the 5-HT1A autoreceptor in depressed patients on SSRIs, a factor contributing to the variability of clinical response in these trials (Rabiner et al, 2001). This finding, supported by pindolol occupancy studies in healthy volunteers (Martinez et al, 2001; Rabiner et al, 2000), indicates that the utility of pindolol in the clinic may be limited, because an increase in dose will also lead to an increased incidence of β-adrenergic side effects. Novel compounds, selective for the 5-HT1A receptor, and having a similar weak partial-agonist profile as pindolol, may prove to be more successful. Nevertheless, pindolol will remain useful as a tool compound in the investigation of serotonergic neurotransmission in humans in vivo.

Limitations of This Study

We examined pindolol occupancy in a small group of depressed patients on SSRI treatment, which does not allow us to differentiate the effects of illness form the effects of treatment. Future studies will need to examine drug-free depressed patients, and healthy volunteers following several weeks of SSRI treatment, in order to disentangle these components. In addition, this work combines two different pindolol regimes, a single 10 mg dose, and repeat dosages of 7.5 and 15 mg daily. While these differences in pindolol regimes may cause variability in the results, the two groups appear comparable, as repeated administration of 7.5 mg of pindolol daily to healthy volunteers (Martinez et al, 2001) produced preferential occupancy similar to that produced by a single dose of 10 mg of pindolol (Figure 1). All the results reported above remain significant when plasma pindolol concentration is introduced as a covariate (data not shown).

It is necessary to consider whether the preferential occupancy induced by pindolol may be a methodological artefact of the PET procedure, rather than reflecting a true increase in binding by pindolol at the RN. The various methodological caveats, such as partial volume effects, were discussed by Martinez et al (2001), who concluded that it was unlikely that preferential occupancy is artefactual. Our data as well as data from other units indicate that high occupancies of the 5-HT1A receptors (up to 70%) can be reached by antagonist compounds without preferential occupancy being found (Andree et al, 2003; Rabiner et al, 2002b), supporting the view that preferential occupancy is not due to a non-linearity in partial volume effects.

We found some evidence of a decrease in delivery of the radioligand in the RN compared to cortical regions in healthy volunteers treated with pindolol, but not in patients. Although the simplified reference tissue model assumes that BP and RI are independent parameters, some co-dependence has been noted previously. To examine the possibility that decreases in RI could lead to decreases in BP (and hence artifactual occupancy), we conducted simulations in which RI was varied and the effect on BP was estimated. The results, presented in Figure 3, indicate that a decline in RI does not lead to a decrease in the estimated BP (if anything the effect is in the opposite direction). Preferential occupancy effect has also been found by Martinez using an alternative quantification method (arterial plasma derived input function (Martinez et al, 2001)).

Figure 3
figure 3

Simulation of the effects of a change in RI on the value of BP estimated by a simplified reference tissue model. (a) Simulated time–activity curves for the reference region (thick line) and five runs for a region of interest (thin line). (b) Results of RI and BP estimation by a simplified reference region model, at various simulated levels of RI. The simulations were conducted using a compartmental model with two tissue compartments in the target tissue and one compartment in the reference tissue. The following values of the kinetic constants (in m−1, k1=0.1, k2=0.5, k3=0.25, k4=0.05) were used in the target tissue. The values for the reference tissue were identical, except that k3 and k4 were set to 0. Decreases in RI were simulated as decreases in k1 and k2 in the target tissue.

Full size image

Finally, a main effect of treatment was found in the examination of the effects of pindolol on the metabolism of [11C]WAY 100635, with the depressed patients who received a 10 mg single dose having a significantly slower clearance of the radioligand from the plasma. This may be due to an interaction between pindolol and the liver enzymes metabolizing [11C]WAY 100635. However, changes in the profile of the radioligand in the plasma are likely to produce global effects on radioligand binding, rather than the regional differences seen in above, and we do not expect these changes to account for the loss of preferential occupancy we found.

In conclusion, we have shown that preferential occupancy of the 5-HT1A autoreceptors by pindolol is attenuated in depressed patients on SSRIs. This phenomenon may be an effect of SSRI treatment, or of the pathophysiology of depression, and as such may have important implications in clinical practice of antidepressant treatment augmentation.