Introduction

Major depressive disorder (MDD) is a leading cause of disease burden worldwide [1]. Serotonin (5-HT) is one of the main neurotransmitters in brain and of central interest in the pathophysiology and treatment of MDD. One reason is that selective serotonin reuptake inhibitors (SSRIs) or serotonin–norepinephrine reuptake inhibitors (SNRIs) have been shown to increase synaptic 5-HT concentrations via inhibition of the 5-HT transporter (5-HTT) in the animal brain [2] and are the current first choice among therapeutics for MDD due to the combined efficacy and tolerability profile of these drugs. However, the therapeutic outcome of current MDD treatment strategies with SSRIs or SNRIs is suboptimal and there is a large need for more effective treatments [3, 4].

Vortioxetine is a recent developed antidepressant, which was approved in 2013 by EU and US authorities for the treatment of MDD in adults [5, 6]. In addition to 5-HTT inhibition, vortioxetine has in preclinical studies been characterized as a 5-HT3, 5-HT7, and 5-HT1D receptor antagonist, a partial agonist at the 5-HT1B receptor and a full agonist at the 5-HT1A receptor [6,7,8,9]. Moreover, it has been shown that the mechanism of action of vortioxetine in animals differ from SSRIs. In rat microdialysis studies, vortioxetine has been shown to induce an around twofold higher increase in extracellular 5-HT concentration when compared to SSRIs, and it has been suggested that the mechanism for antidepressant effect of vortioxetine may include binding to one or several of the 5-HT receptor subtypes [6, 10, 11]. Importantly, species differences in the binding of vortioxetine to some of the 5-HT receptor subtypes has to be taken into account, e.g., higher affinity for 5-HT1B and 5-HT1D receptors and lower affinity for 5-HT1A and 5-HT7 receptors for rat than for recombinant human receptors [6,7,8,9]. It is therefore of interest to evaluate the mechanism of action of vortioxetine directly in the primate brain in vivo.

Positron emission tomography (PET) imaging studies have previously demonstrated that clinically relevant doses of SSRIs occupy ~60–90% of the 5-HTT in healthy subjects [12] and in treated MDD patients [13, 14]. Clinically relevant doses of vortioxetine have shown slightly lower occupancies, typically around 60%, of the 5-HTT in the human brain [15, 16]. This potential difference in 5-HTT occupancy between clinically relevant doses of SSRIs and vortioxetine might relate to a different mechanism of action as described above. Although the level of target occupancy is important for dose selection in initial clinical studies [17], the 5-HTT occupancy of vortioxetine in the nonhuman primate (NHP) brain has not been reported so far.

We have previously demonstrated that the binding of [11C]AZ10419369, a 5-HT1B receptor partial agonist, is significantly reduced in the primate brain following a high but not clinically relevant dose of escitalopram [18]. A likely interpretation is that [11C]AZ10419369 binding is sensitive to drug-induced increases in 5-HT concentration. Previous rodent microdialysis studies have demonstrated that the increases in 5-HT concentration (two- to fivefold from baseline) induced by a clinically relevant dose of vortioxetine [10] are comparable to increases induced by high doses of SSRIs [19, 20]. It can thus be anticipated that [11C]AZ10419369 binding may also be sensitive to increases in 5-HT concentration induced by a clinically relevant dose of vortioxetine.

It is worth noting that vortioxetine has nanomolar range affinity to recombinant human 5-HT1B receptors [7]. Consequently, in addition to increases in 5-HT concentration, vortioxetine might also alter [11C]AZ10419369 binding by direct occupancy of the 5-HT1B receptors. More recently, [11C]Cimbi-36, a 5-HT2A receptor agonist radioligand, has been reported to have similar sensitivity as [11C]AZ10419369 to increases in 5-HT concentration [21]. Vortioxetine has fivefold higher affinity for the recombinant human 5-HT1B receptor (Ki = 33 nmol/L) than the 5-HT2A receptor (Ki = 180 nmol/L) (Table 1) [7]. Therefore, evaluation of the effect of vortioxetine on [11C]Cimbi-36 binding may help differentiate between a direct effect of vortioxetine on [11C]AZ10419369 binding and an indirect effect from increased 5-HT concentration. Specifically, if there is no effect on [11C]Cimbi-36 binding, it would support direct vortioxetine binding to the 5-HT1B receptors.

Table 1 Binding affinities of vortioxetine and citalopram to recombinant human and rat 5-HTT and 5-HT receptors reported in vitro
Full size table

The aim of the present study was to examine the mechanism of action of vortioxetine in the NHP brain. PET measurements with [11C]MADAM were performed to estimate the 5-HTT affinity of vortioxetine in vivo in the NHP brain. The effect of vortioxetine and citalopram, a reference SSRI with high affinity for the 5-HTT (Table 1) [22,23,24], on the binding of [11C]AZ10419369 to the 5-HT1B receptor were then compared at clinically relevant and comparable doses. Finally, the effect of vortioxetine on [11C]Cimbi-36 binding to the 5-HT2A receptor was evaluated.

Materials and methods

Subjects

The study was approved by the Animal Research Ethical Committee of the Northern Stockholm region (Dnr N185/14). Five female rhesus monkeys (Macaca mulatta) with body weight: 7.7 ± 2.7 kg (mean ± standard deviation) were included. The caring and experimental procedures were performed according to the ‘‘Guidelines for planning, conducting and documenting experimental research’’ (Dnr 4820/06-600) of Karolinska Institutet and the ‘‘Guide for the Care and Use of Laboratory Animals: Eighth Edition’’ [25].

Preparation of radioligands

[11C]MADAM [26], [11C]AZ10419369 [27], and [11C]Cimbi-36 [28] were prepared according to procedures reported previously.

Study design

A total of 32 PET measurements (10 with [11C]MADAM, 18 with [11C]AZ10419369, and 4 with [11C]Cimbi-36) were performed on 16 experimental days. On each experimental day, following a baseline PET measurement, a consecutive PET measurement was conducted after pretreatment with vortioxetine or citalopram. The two PET measurements were performed ~3 h apart.

There were four stages in the experiments. First, to support that the 5-HTT affinity of citalopram in rhesus monkeys is comparable to that previously reported for cynomolgus monkeys [29], one rhesus monkey (NHP1) underwent PET measurements with [11C]MADAM before and after administration of 2.0 mg/kg of citalopram. Second, to determine the 5-HTT affinity of vortioxetine in rhesus monkeys, four monkeys (NHP2–NHP5) underwent PET measurements with [11C]MADAM before and after administration of one dose of vortioxetine (0.1, 0.3, 1.0, or 3.0 mg/kg, respectively). Third, to compare the effects of citalopram and vortioxetine on 5-HT1B receptor binding, three monkeys (NHP1, NHP3, and NHP4), each underwent PET measurements with [11C]AZ10419369 before and after administration of 0.3 mg/kg of citalopram, 0.3 mg/kg of vortioxetine or 1.0 mg/kg of vortioxetine. These doses were selected based on literature data and the comparison in the first part of the present study to assure clinically relevant 5-HTT occupancy levels. Finally, to evaluate the effect of vortioxetine on 5-HT2A receptor binding, two monkeys (NHP2 and NHP3) underwent PET measurements with [11C]Cimbi-36 before and after administration of 1.0 mg/kg of vortioxetine.

Pretreatment administration

Citalopram was formulated in phosphate-buffered saline (PBS). Vortioxetine was formulated in a mixture of 10% hydroxypropyl beta cyclodextrin dissolved in PBS. All drug solutions were infused (1 mL/kg) over 30 min, starting 45 min before injection of radioligand.

PET experimental procedures

Anesthesia was initiated by intramuscular injection of ketamine hydrochloride (~10 mg/kg) and maintained by a mixture of sevoflurane (2–8%), oxygen and medical air. PET measurements were conducted in the High Resolution Research Tomograph (HRRT); a six min transmission measurement (using a single 137Cs source) was followed by the acquisition of list-mode data for 123 min after a bolus injection of radioligand. Eleven venous blood samples (at −55, −14.5, −5, 15, 30, 45, 60, 75, 90, 105, and 120 min after injection of radioligand) were collected for determination of plasma drug concentrations of vortioxetine and citalopram.

Determination of plasma drug concentrations

Plasma drug concentrations for vortioxetine and citalopram were determined using ultra performance liquid chromatography (UPLC) followed by tandem mass spectrometry (MS/MS) detection according to procedures reported previously [29] and as described in Supplementary Materials and Methods. The mean of the plasma drug concentration at eight time points after injection of radioligand and during time of PET data acquisition was used to represent plasma drug concentration in each PET experiment.

Magnetic resonance imaging (MRI)

Each monkey was examined by MRI to define brain volumes of Interest (VOIs). T1-weighted MRI images were acquired on a GE 1.5 Tesla Signa MRI scanner (Milwaukee, WI) using a 3D spoiled gradient recalled (SPGR) protocol: repetition time 21 ms, flip angle 35°, FOV 12.8, matrix 256 × 256 × 128, 128 × 1.0 mm2 slices.

Image data analysis

Preprocessing of MRI images included manual reorientation to the anterior-posterior commissure (AC-PC) plane, manual removal of non-brain tissue and N4 bias field correction. PET images were preprocessed according to previously reported methods [30] with reconstructed image frames binned as: 9 × 10, 2 × 15, 3 × 20, 4 × 30s, 4 × 60, 4 × 180, and 17 × 360 s. Following imaging data analyses were performed using PMOD (version 3.604; PMOD Technologies, Zurich, Switzerland).

Each monkey’s baseline summed PET image (average of time frames corresponding to 15–69 min for [11C]MADAM, 5–18 min for [11C]AZ10419369 and 12–63 min for [11C]Cimbi-36) was co-registered to its individual MRI brain image by the Rigid matching algorithm in the PMOD Fuse It Tool (PFUSEIT). The resulting transformation matrices were applied to the corresponding two PET measurements performed on the same day.

Twelve VOIs were defined based on the NeuroMaps atlas in the INIA19 rhesus template [31], including four regions for the basal ganglia: putamen, caudate nucleus (CN), ventral striatum (VS), and globus pallidum (GP); two neocortical regions: frontal cortex (FC) and occipital cortex (OC); three limbic regions: anterior cingulated cortex (ACC), amygdala and hippocampus; and VOIs for thalamus, midbrain, and cerebellum. Each monkey’s brain MRI image was normalized to the INIA19 rhesus template with the Deformable matching algorithm in the PFUSEIT and the resulting normalization matrix was used to inversely transform the template VOIs into individual MRI space. In addition, the dorsal raphe nuclei (DRN) was manually delineated on each monkey’s coregistrated summed [11C]MADAM PET image in a sagittal plane, including 5–6 slices from the level of the superior colliculus to the level of the inferior colliculus [32].

Quantification

For each VOI, a decay-corrected time–activity curve (TAC) was generated from the co-registered dynamic PET data. Binding potential (BPND) values were calculated using the simplified reference tissue model (SRTM) [33], with cerebellum as the reference region, as has been validated in previous studies [34,35,36,37].

Relative change in BPND values (∆BPND) (%) was calculated using the following equation:

$$\Delta {\mathrm{BP}}_{{\mathrm{ND}}}\left( \% \right) = \frac{{{\mathrm{BP}}_{{\mathrm{ND}}}{\mathrm{Pretreatment}} - {\mathrm{BP}}_{{\mathrm{ND}}}{\mathrm{Baseline}}}}{{{\mathrm{BP}}_{{\mathrm{ND}}}{\mathrm{Baseline}}}} \times 100,$$
(1)

For illustration purposes, parametric BPND images were generated using SRTM2 [38] and normalized to the INIA19 rhesus template [31] by using the methods described in Supplementary Materials and Methods.

Relationship between pretreatment drugs and 5-HTT occupancy

According to the law of mass action, the relationship between radioligand binding and the concentration of drug at equilibrium can be described by a one-site binding hyperbola, as expressed by the following equation:

$${\mathrm{Decrease}}\,{\mathrm{in}}\,{\mathrm{BP}}_{{\mathrm{ND}}}\left( \% \right) = I_{{\mathrm{max}}} \times \frac{{{\mathrm{drug}}\,{\mathrm{dose}}\,\left( {{\mathrm{or}}\,C_{\mathrm{P}}} \right)}}{{{\mathrm{drug}}\,{\mathrm{dose}}\,\left( {{\mathrm{or}}\,C_{\mathrm{P}}} \right) + {\mathrm{ID50}}\,\left( {{\mathrm{or}}\,K_{\mathrm{i}}} \right)}},$$
(2)

where Imax is the maximal inhibition (%), CP is the plasma drug concentration and ID50 or Ki corresponds to the drug dose or the plasma drug concentration at which 5-HTT occupancy is 50%, respectively [29, 39].

Based on Eq. (2), the drug-induced decrease in [11C]MADAM BPND (%) (equal to the ∆BPND (%) with opposite sign) in putamen and CN were plotted against the values of corresponding drug dose or plasma drug concentration [29]. The mean plasma drug concentration of 8 blood samples taken during the PET measurement was used to represent the plasma drug concentration during the PET measurement period. An unconstrained Imax value was applied to the model [29], and the ID50 or Ki value was calculated using GraphPad Prism (version 6.05; GraphPad Software Inc., La Jolla, CA, USA).

Statistical analysis

A paired t-test was used to assess changes in parameters between the two PET measurements performed on the same day. All statistical analyses were performed in GraphPad Prism. The threshold of significance was set as P < 0.05 (two-tailed).

Results

Radiochemistry

The 32 PET measurements in rhesus monkeys were conducted according to plans and the three radioligands were synthesized to high specific radioactivity. There were no statistically significant differences in injected radioactivity or injected mass at time of injection between baseline and pretreatment conditions (Supplementary Table 1).

Plasma concentrations of pretreatment drugs

In the measurements conducted to estimate 5-HTT occupancy using [11C]MADAM, following administration of 2.0 mg/kg citalopram, the plasma drug concentration was 952 nmol/L. Following administration of vortioxetine at 0.1, 0.3, 1.0mg/kg and 3.0 mg/kg, the plasma drug concentration was 16, 43, 192, and 562 nmol/L, respectively.

In the measurements to evaluate the effects of pretreatment drugs on [11C]AZ10419369 binding, the time course of plasma concentrations of citalopram or vortioxetine are presented in Supplementary Fig. 1. The group mean of plasma drug concentration was 119, 38, and 114 nmol/L for 0.3 mg/kg of citalopram, 0.3 mg/kg of vortioxetine, and 1.0 mg/kg of vortioxetine, respectively. In the measurements with [11C]Cimbi-36, the group mean of plasma drug concentration was 157 nmol/L following administration of 1.0 mg/kg of vortioxetine.

5-HTT binding after administration of pretreatment drugs

After administration of 2.0 mg/kg of citalopram in a single rhesus monkey, there was a marked decrease in [11C]MADAM binding. The 5-HTT occupancy was 56% in putamen and 63% in CN. This level is comparable to previous data in cynomolgus monkeys (60%, with ID50 = 0.059 mg/kg and Imax = 62%) [29].

There was a decrease of [11C]MADAM binding in all four monkeys examined after administration of vortioxetine (Fig. 1). The estimated ID50 and Ki values for vortioxetine were 0.25 ± 0.09 mg/kg (95% CI = 0.03–0.47; R2 = 0.89; Imax = 69%) and 38.9 ± 12.1 nmol/L (95% CI = 10.2–67.7; R2 = 0.91; Imax = 68%), respectively.

Fig. 1
figure 1

Decreases in [11C]MADAM binding potential (BPND) in caudate nucleus (CN, open square symbol, □) and putamen (Put, plus sign, +) after intravenous administration of vortioxetine in four nonhuman primates. The solid lines represent fitting of an unconstrained one-site binding hyperbola for both regions. The dotted lines represent the estimated maximal inhibition (Imax). The corresponding serotonin transporter (5-HTT) occupancy is shown along the right y-axis. a Doses of vortioxetine versus decreases in BPND. b Plasma concentrations of vortioxetine versus decreases in BPND

Full size image

5-HT1B receptor binding after administration of pretreatment drugs

After administration of 0.3 mg/kg of citalopram, there were numerical increases in BPND values for [11C]AZ10419369 binding to the 5-HT1B receptor across several brain regions, with a trend level of significance for the FC (Table 2, Figs. 2, 3, and Supplementary Fig. 2). Interestingly, there was a small but significant reduction in BPND for the DRN.

Table 2 Effect of citalopram and vortioxetine on mean regional binding potential (BPND) values of [11C]AZ10419369 and [11C]Cimbi-36 in rhesus monkeys
Full size table
Fig. 2
figure 2

Mean parametric binding potential (BPND) images of [11C]AZ10419369 or [11C]Cimbi-36 derived by SRTM2 at baseline and after pretreatment with citalopram or vortioxetine (n = 3 for each pretreatment condition measured by [11C]AZ10419369 and n = 2 for vortioxetine 1.0 mg/kg measured by [11C]Cimbi-36). The images were normalized to the INIA19 rhesus template. a Axial view of images at the level of superior neocortex. b Axial view of images at the level of midbrain. ACC anterior cingulated cortex, FC frontal cortex, HC hippocampus, MB midbrain, OC occipital cortex, Put putamen

Full size image
Fig. 3
figure 3

Individual [11C]AZ10419369 or [11C]Cimbi-36 BPND values before and after pretreatment with citalopram or vortioxetine in 12 pre-defined volumes of interest, including putamen (Put), caudate nucleus (CN), ventral striatum (VS), globus pallidum (GP), frontal cortex (FC), occipital cortex (OC), anterior cingulated cortex (ACC), amygdala (Amyg), hippocampus (HC), thalamus (Thal), midbrain (MB), and dorsal raphe nuclei (DRN). a Citalopram 0.3 mg/kg and [11C]AZ10419369 (n = 3). b Vortioxetine 0.3 mg/kg and [11C]AZ10419369 (n = 3). c Vortioxetine 1.0 mg/kg and [11C]AZ10419369 (n = 3). d Vortioxetine 1.0 mg/kg and [11C]Cimbi-36 (n = 2)

Full size image

Following administration of 0.3 or 1.0 mg/kg of vortioxetine, the binding of [11C]AZ10419369 to the 5-HT1B receptor was reduced in a dose-dependent fashion when compared to baseline (Table 2, Figs. 2 and 3, and Supplementary Fig. 2). After 0.3 mg/kg of vortioxetine, the reduction in BPND values was statistically significant in six brain regions (CN, VS, GP, OC, thalamus, and midbrain). After 1.0 mg/kg of vortioxetine, the reduction in BPND values was statistically significant in all 12 examined brain regions (Table 2).

5-HT2A receptor binding after administration of vortioxetine

Following administration of 1.0 mg/kg of vortioxetine, there was no conspicuous change in [11C]Cimbi-36 binding across different brain regions in the two monkeys (Table 2, Figs. 2 and 3, and Supplementary Fig. 3).

Discussion

In the current PET study, we evaluated the mechanism of action of vortioxetine by comparing the effect of clinically relevant doses of vortioxetine and citalopram on serotonergic PET markers. A main observation was that vortioxetine significantly reduced [11C]AZ10419369 binding to the 5-HT1B receptor in several cortical and subcortical regions while citalopram reduced [11C]AZ10419369 binding only in DRN. Moreover, there was no significant effect of vortioxetine on [11C]Cimbi-36 binding to the 5-HT2A receptor though [11C]Cimbi-36 binding has been demonstrated to have similar sensitivity for the 5-HT concentration as [11C]AZ10419369 binding. In conclusion, the large and dose-dependent decreases in [11C]AZ10419369 binding induced by vortioxetine suggest direct drug binding to the 5-HT1B receptor. The observations also suggest that vortioxetine significantly occupies the 5-HT1B receptor when administered at clinically relevant doses. This finding appears to differentiate vortioxetine from the reference SSRI citalopram. However, a difference in their ability to elevate 5-HT concentration cannot be excluded.

The Ki value determined for vortioxetine binding to the 5-HTT (39 ± 12 nmol/L) was modestly higher than previously reported for humans (16–20 nmol/L) [15, 16]. Similar level of discrepancy between monkey and human 5-HTT Ki values have in previous PET studies been reported for escitalopram [12, 29] and venlafaxine [13, 40]. Species differences in pharmacokinetics or the different route of drug administration may have contributed to these modest discrepancies [29, 40, 41]. The estimated Ki value for 5-HTT binding in the monkey brain was therefore considered to be in line with previous human results.

It is crucial to select adequate doses when comparing the effects of vortioxetine and citalopram. We first confirmed that the 5-HTT affinity of citalopram is comparable between rhesus and cynomolgus monkeys [29]. Since the 5-HTT affinity of citalopram and vortioxetine was estimated using the same methodology, it is feasible to use the in vivo affinity data to select doses, which achieve comparable 5-HTT occupancy [18, 29, 42]. In the measurements of binding to the 5-HT1B receptor, the doses of 0.3 mg/kg of citalopram and 1.0 mg/kg of vortioxetine were chosen as they should correspond to about 80% 5-HTT occupancy, which is considered a representative effective occupancy level at clinical treatment with SSRIs [13, 14]. A lower dose of vortioxetine (0.3 mg/kg) was selected to achieve about 55% 5-HTT occupancy, and to represent the lower end of reported 5-HTT occupancy values at the clinical dose range for vortioxetine (5–10 mg/day) [5, 15, 16].

The clinical relevance of the selected doses was further supported by the plasma concentration of current pretreatment drugs (119 nmol/L for 0.3 mg/kg of citalopram; 38 and 114 nmol/L for 0.3, and 1.0 mg/kg of vortioxetine, respectively). The concentration values are comparable to clinical data for 20–60 mg/day of citalopram (130–400 nmol/L) [43] and 5–20 mg/day of vortioxetine (30–110 nmol/L) [5]. It can thus be assumed that the current results can be extrapolated to human studies at which a therapeutic antidepressant effect has been documented.

In the current study, in contrast to citalopram, clinically relevant doses of vortioxetine significantly decreased [11C]AZ10419369 binding in several brain regions. With regard to the mechanism of action of these two drugs, the divergent effects might originate from a difference in the ability to change the 5-HT concentration or, alternatively, a difference in affinity for the 5-HT1B receptor. The decreases in [11C]AZ10419369 binding induced by 1.0 mg/kg of vortioxetine were two- to fourfold higher than those previously reported for a high dose of an SSRI [18] and comparable to the decreases induced by 5.0 mg/kg of fenfluramine, one of the strongest 5-HT releaser [21]. Moreover, [11C]Cimbi-36 binding has been reported to have similar sensitivity as [11C]AZ10419369 to the 5-HT release induced by fenfluramine [21]. Accordingly, if the vortioxetine induced reductions in [11C]AZ10419369 binding mainly originate from increases in 5-HT concentration, it would be expected that 1.0 mg/kg of vortioxetine could significantly reduce [11C]Cimbi-36 binding. The lack of effect of vortioxetine on [11C]Cimbi-36 binding in the NHP brain suggests that the increases in 5-HT concentration induced by clinically relevant doses of vortioxetine might not be detectable by [11C]AZ10419369 binding. Thus, the observed reductions in [11C]AZ10419369 binding are more likely explained by direct binding of vortioxetine to the 5-HT1B receptors. This interpretation is also in line with the markedly different binding affinities to the 5-HT1B receptors between these two drugs reported in vitro (Table 1) [7,8,9, 22,23,24]. In summary, the current results support that there might be direct drug occupancy of 5-HT1B receptors at the administered doses of vortioxetine.

It is important to note that the lack of effect of vortioxetine on [11C]Cimbi-36 binding did not exclude the possibility that vortioxetine induces larger increases in 5-HT concentration than citalopram in the NHP brain. Vortioxetine has been reported to increase extracellular 5-HT concentration about twice as large as for an SSRI in the rat hippocampus [10]. Considering the low statistical power in the current study due to a small sample size, the current results might reflect that the 5-HT release induced by administration of vortioxetine was below the threshold that could be reliably measured by the study protocol using [11C]Cimbi-36. Future studies with a larger sample size to evaluate the effect of 1.0 mg/kg of vortioxetine on [11C]Cimbi-36 binding might further corroborate our findings. In addition, it cannot be excluded that the effect of vortioxetine on [11C]Cimbi-36 binding might differ between NHPs and humans.

The current results suggest that vortioxetine binds to the 5-HT1B receptor at clinically relevant doses. The 5-HT1B receptors are widely distributed throughout the brain. The receptors are predominantly located on axon terminals where they serve as autoreceptors on serotonergic neurons and heteroreceptors on non-serotonergic neurons. Importantly, the functional role of the autoreceptors is to regulate the release of 5-HT [44,45,46]. On the basis of a series of experimental studies, the 5-HT1B receptor has been proposed to be involved in the pathophysiology of depression and to be a potential target for drug treatments [44,45,46,47,48]. The occupancy of vortioxetine at the 5-HT1B receptor might also contribute to cognitive benefits, which in clinical studies has been reported to be independent of the antidepressant effect [49, 50]. Although desensitization of hippocampal 5-HT1B autoreceptors following vortioxetine administration has been shown to be related to antidepressant effects in rodent models [51], the contribution of 5-HT1B receptor occupancy by vortioxetine in relation to its antidepressant effect in patients remains to be clarified. Future studies are warranted, including application of the current PET paradigm in human subjects [52, 53].

Considering the affinity of vortioxetine for multiple neuroreceptors [6, 10, 11], it is challenging to dissect the clinical effect of 5-HT1B receptor binding from the summed effect of the engaged targets. While several selective 5-HT1B receptor compounds have been reported to demonstrate antidepressant-like effects in different animal models of depression [54,55,56], none of them have been examined in clinical studies [46]. Extended work with compounds that are selective for the 5-HT1B receptor may not only aid identification of mechanisms of action of vortioxetine but also, in a wider sense, assessment of the 5-HT1B receptor as a potential target for development of novel antidepressant drugs.

A potential limitation of the current PET study is the use of anesthesia. Anesthetic doses of ketamine might influence the binding of radioligand to 5-HTT and result in overestimation of ID50 or Ki values [29, 57]. Ketamine has also been reported to increase [11C]AZ10419369 binding in the NHP brain, but had no effect on decreases in [11C]AZ10419369 binding induced by fenfluramine, a 5-HT releaser [57]. So far, there have been no studies evaluating the effect of sevoflurane on [11C]AZ10419369 binding or the effect of anesthesia on [11C]Cimbi-36 binding. Importantly, as the same anesthesia regimen was applied across the experiments, potential anesthesia effects were similar between citalopram and vortioxetine in 5-HTT occupancy studies and between [11C]AZ10419369 and [11C]Cimbi-36 pretreatment studies using vortioxetine. Accordingly, potential anesthesia effects were not likely to affect the main observations of the current study.

Another potential limitation of the current study is the inclusion of female monkeys only. In addition to higher prevalence of MDD in females than in males, gender differences in the 5-HT system have been reported [58, 59]. Therefore, it cannot be excluded that gender differences influenced the current results, although we are not aware of PET studies indicating gender differences in drug-induced 5-HTT occupancy or 5-HT release.

In conclusion, our results reveal that the effect of clinically relevant doses of vortioxetine and citalopram on [11C]AZ10419369 binding to the 5-HT1B receptor was different in the NHP brain. The results might mainly originate from a difference in 5-HT1B receptor affinity between these two drugs although it cannot be excluded that a difference in their ability to elevate 5-HT concentration may also contribute. These observations suggest that vortioxetine binds to the 5-HT1B receptor at clinically relevant doses. Future studies are warranted to evaluate the role of the 5-HT1B receptor in the therapeutic effects of vortioxetine and as a potential target for the development of novel antidepressant drugs.

Funding and disclosure

This work was sponsored by the H. Lundbeck A/S. Dr. Takano is currently an employee and share-holder of Takeda Pharmaceutical Company. Dr. Christoffer Bundgaard is currently an employee at Lundbeck and share-holder of Eli Lilly & Company. Dr. Benny Bang-Andersen is currently an employee and share-holder of Lundbeck. Dr. Connie Sanchez is currently an employee and share-holder of Alkermes and a share-holder of Lundbeck. Dr. Farde is an employee and share-holder of AstraZeneca. He has received compensation for being an external scientific reviewer at the University of Helsinki, Finland. Dr. Finnema is currently an employee and share-holder of AbbVie. Dr. Yang, Dr. Stepanov, Dr. Amini, Mr. Martinsson, and Dr. Halldin declare no competing interests.