Serotonergic Neurotransmission in Limbic Regions May Reflect Therapeutic Response of Depressive Patients: A PET Study With 11C-WAY-100635 and 18F-MPPF

Abstract Background Central serotonin (5-hydroxytryptamine [5-HT]) neurotransmission has been implicated in the etiology of depression. Most antidepressants ameliorate depressive symptoms by increasing 5-HT at synaptic clefts, but their effect on 5-HT receptors has yet to be clarified. 11C-WAY-100635 and 18F-MPPF are positron emission tomography (PET) radioligands for 5-HT1A receptors. While binding of both ligands reflects 5-HT1A receptor density, 18F-MPPF biding may also be affected by extracellular 5-HT concentrations. This dual-tracer PET study explored the neurochemical substrates underlying antidepressant effects in patients with depression. Methods Eleven patients with depression, including 9 treated with antidepressants, and 16 age- and sex-matched healthy individuals underwent PET scans with 11C-WAY-100635 and 18F-MPPF. Radioligand binding was determined by calculating the nondisplaceable binding potential (BPND). Results Patients treated with antidepressants showed significantly lower 18F-MPPF BPND in neocortical regions and raphe nuclei, but not in limbic regions, than controls. No significant group differences in 11C-WAY-100635 BPND were found in any of the regions. Significant correlations of BPND between 11C-WAY-100635 and 18F-MPPF were observed in limbic regions and raphe nuclei of healthy controls, but no such associations were found in antidepressant-treated patients. Moreover, 18F-MPPF BPND in limbic regions was significantly correlated with the severity of depressive symptoms. Conclusions These results suggest a diversity of antidepressant-induced extracellular 5-HT elevations in the limbic system among depressive patients, which is associated with the individual variability of clinical symptoms following the treatment.

To date, there have been several reports on the in vivo evaluation of 5-HT 1A receptors in patients with depression. Compared with controls, some studies documented that the patients showed decreased 11 C-WAY-100635 binding (Sargent et al., 2000;Meltzer et al., 2004;Hirvonen et al., 2008), while other studies showed increased 11 C-WAY-100635 binding (Parsey et al., 2006;Sullivan et al., 2009;Kaufman et al., 2015). Binding of 18 F-MPPF to 5-HT 1A receptors was reportedly lower in drug-naive patients with depression than in healthy controls and was elevated by administration of a selective serotonin reuptake inhibitor (SSRI) (Lothe et al., 2012). Thus, PET-assessed alterations of 5-HT 1A receptors in depression remain elusive. This raises the concern as to whether changes in extracellular 5-HT concentrations or 5-HT 1A receptor densities underlie depressive symptoms and affect PET findings.
To the best of our knowledge, no clinical examinations evaluating depression by simultaneous usage of both 11 C-WAY-100635 and 18 F-MPPF have been conducted. The purpose of this study was to explore alterations of the serotonergic system by PET with dual 5-HT 1A receptor radioligands and determine their association with clinical symptoms in the same individual patients with depression. In addition to the assessment of 5-HT 1A receptor density by 11 C-WAY-100635-PET, drug-induced alterations of extracellular 5-HT levels could modify the interaction of 18 F-MPPF with the receptors, possibly eliciting uncoupling of the 2 radioligand bindings. We hypothesized that antidepressant-induced alterations of serotonergic neurotransmission in limbic regions, which are known to be associated with mood and emotion, differentially affect binding of these 2 radioligands and correlate with the amelioration of depressive symptoms.

Participants
Eleven patients with depression and 16 age-and sex-matched healthy controls were included in this study. The patients were recruited from 3 affiliated psychiatric clinics. Ten of the patients met the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria for major depressive disorder, and 1 met the criteria for dysthymic disorder. For the patients, lifetime psychiatric disorders were evaluated based on the Mini-International Neuropsychiatric Interview. Exclusion criteria included other psychiatric disorders, existence of current suicidal ideation, severe current physical illness, or administration of antidepressant drugs other than SSRIs or serotonin noradrenalin reuptake inhibitors (SNRIs). Of the 11 patients, 6 concurrently received treatment with SSRIs (20 mg/ day paroxetine for 2 patients and 50-75 mg/day sertraline for 4 patients). The other 3 patients received treatment with an SNRI (30-60 mg/day duloxetine). As described elsewhere (Pae and Patkar, 2007), repeated daily dosage led to a steady state of the plasma drug concentrations long before PET scans. The remaining 2 patients were antidepressant naive. The controls were free of any somatic, neurological, or psychiatric disorders. We evaluated the severity of the depressive symptoms of all participants using the Beck Depression Inventory-II (BDI-II) (Beck et al., 1996). The patients were also evaluated by the Japanese version of the 17-item Hamilton Depression Rating Scale (HAM-D) (Hamilton, 1960). We excluded 3 healthy controls from the statistical analysis of the image data because they showed depressive tendencies with BDI-II scores exceeding 10 points (supplementary Figure 1). This study was approved by the Radiation Drug Safety Committee and the Institutional Review Board of the National Institute of Radiologic Sciences of Japan. Written informed consent was obtained from all participants. The study was registered with the University Hospital Medical Information Network Clinical Trials Registry (UMIN000012050).

Brain PET Scans
We used 11 C-WAY-100635 (228.3 ± 10.8 MBq, molar activity (A m ): 147.8 ± 100 GBq/μmol) for assessing 5-HT 1A receptor density. We also employed 18 F-MPPF (186.4 ± 96 MBq, A m : 174.8 ± 6.5 GBq/ μmol) for evaluating extracellular 5-HT concentration in addition to receptor density. Out of 27 participants, 25 individuals received PET scans first with 18 F-MPPF and then with 11 C-WAY-100635, whereas 1 healthy control and 1 patient with depression underwent 2 PET scans in reverse order. The mean interval between the 2 PET sessions was 18 days, ranging from 7 to 68 days. After i.v. injection, 3-dimensional dynamic images were acquired using a PET scanner (Eminence SET-3000GCT/X; Shimadzu, Kyoto, Japan) for 90 and 60 minutes for 11 C-WAY-100635 and 18 F-MPPF, respectively. All PET images were reconstructed using the filtered back-projection method (Gaussian filter, kernel 5 mm; reconstructed in-plane resolution, 7.5 mm full width at half maximum; voxel size, 2 × 2 × 2.6 mm) with correction for attenuation, randoms, and scatter. Head motion during the scans was corrected on the emission images after attenuation correction with μ-maps realigned to each frame of the emission images (Wardak et al., 2010).

Brain Imaging Analyses
T1-weighted magnetic resonance (MR) images were acquired using a 3-T MR imaging scanner (Vero, Siemens, Germany; TE 1.9 mms, TR 2300 mms, TI 900 mms, flip angle 9°, field of view 250 mm, acquisition matrix 256 × 256, slice thickness 1 mm). Dynamic PET data were spatially normalized based on the transformation parameters from the MR images, and time activity curves in cerebellar gray matter, receptor-poor regions, were acquired from both radioligands. The mask of cerebellar gray matter was defined with a probabilistic cerebellar atlas excluding the vermis (Diedrichsen et al., 2009). We then estimated the nondisplaceable binding potential (BP ND ) using a multilinear reference tissue model (MRTM2) (Ichise et al., 2003), with the cerebellar gray matter as a reference region in the naive PET data (Hahn et al., 2010). There are also previous studies defining a reference region in cerebellar gray matter (Hirvonen et al., 2007(Hirvonen et al., , 2008 based on a finding of a single individual exhibiting marked radioligand retention in cerebellar gray matter. Another report demonstrated no significant difference in 5-HT 1A binding between the uses of cerebellar gray and white matter references (Hahn et al., 2010). We chose cerebellar gray matter as reference tissue in consideration of the substantial difference in the radioligand delivery to the brain tissue between gray and white matters. We applied a multilinear reference tissue model to the dynamic data of 11 C-WAY-100635 because it is a suitable reference tissue approach for BP ND estimation of 11 C-WAY-100635 (Zanderigo et al., 2013). For this analysis, k 2 ʹ was calculated as the average of k 2 ' values determined with MRTM in the insula, hippocampus, and parahippocampal gyrus, which are enriched with the target receptors. The cerebellum was employed as a receptor-less region. Because the BP ND value was comparably small, we applied Motmot to the BP ND estimation of 18 F-MPPF for good linear fitting (Ichise et al., 2003). Each of the parametric images was reregistered to the individual MR images, then spatially normalized using the Diffeomorphic Anatomical Registration Through Exponentiated Lie Algebra algorithm included in the statistical parametric mapping package (SPM12; Wellcome Trust Centre for Neuroimaging, London, UK) ( Figure 1). Based on the normalized BP ND images, we calculated the mean BP ND values in neocortical regions, including the frontal and lateral temporal cortices, which postsynaptically express 5-HT 1A receptors at a relatively low level. BP ND was also determined in limbic regions, including the medial temporal and anterior cingulate cortices, which are enriched with postsynaptic 5-HT 1A receptors. In addition, we calculated BP ND in raphe nuclei, which possess presynaptic 5-HT 1A receptors. Four cortical regions of interest (ROIs) were defined by the automated anatomical labeling atlas (Tzourio-Mazoyer et al., 2002). Each ROI intersected with the patients' normalized gray matter masks. Because the raphe nuclei could not be differentiated from surrounding tissue on MR images, a circular ROI 6 mm in diameter was placed on 5 contiguous axial slices to be centered at the highest radio signal in the dorsal midbrain. The ROI size was determined in light of the actual diameter of the median and dorsal raphe nuclei approximating 5 mm and the spatial resolution of the reconstructed PET images (7.5 mm full width at half maximum) in reference to previous work (Kranz et al., 2012). Analyses of all PET data were conducted using PMOD 3.7 (PMOD Technologies Ltd.).

Statistical Analyses
For the statistical analyses, we excluded 2 depressive patients without antidepressant treatment to focus on the treatment-related neuroimaging changes (supplementary Figure 1) and displayed data from these individuals in the graph plots for preliminary comparisons. Unpaired t tests were used to evaluate group comparisons with demographic parameters of age, BDI-II scores, and Fisher's exact test for sex (statistical significance: P < .05). We performed unpaired t tests to examine the group differences in the estimated BP ND values of either 11 C-WAY-100635 or 18 F-MPPF between the patients and healthy controls. To evaluate the mutual relationship between the estimated BP ND values of 11 C-WAY-100635 and 18 F-MPPF, correlation analyses were performed. Among the patients with depression, we evaluated the relationships between severity of depression by HAM-D scores and the estimated BP ND values of both radioligands, respectively. Because estimated BP ND values of 11 C-WAY-100635 and 18 F-MPPF, and HAM-D scores showed normal distribution as examined by the Shapiro-Wilk test, correlations between these parameters were examined by t test of Pearson correlation coefficient (statistical significance: P < .05). Statistical analyses were conducted using the Statistical Package for the Social Sciences (SPSS) version 22 (SPSS Inc., Chicago, IL, USA).

Demographic Characteristics
The demographic and psychological profiles of the participants in this study are shown in Table 1. There were no significant differences in age or gender between the patients and healthy controls.
The patients showed significantly higher BDI-II scores than those of the healthy controls (P < .05).

Group Comparisons of 11C-WAY-100635 and 18F-MPPF
BP ND values of 11 C-WAY-100635 and 18 F-MPPF in the patients versus healthy controls are shown in Figure 2 and supplementary Table 1. Significantly lower 11 C-WAY-100635 BP ND values were seen in the frontal cortex of the antidepressant-treated patients than those of the healthy controls (P < .05). By contrast, the antidepressant-treated patients exhibited lower 18 F-MPPF BP ND values in the frontal, lateral temporal, and anterior cingulate cortices and raphe nuclei than the healthy controls (P < .05). These group differences in the 18 F-MPPF binding remained statistically significant in the frontal and lateral temporal cortices and raphe nuclei after applying Bonferroni's correction for multiple comparisons (P < .01). Figure 3 demonstrates the mutual relationships between the estimated BP ND values of 11 C-WAY-100635 and 18 F-MPPF in each of the 2 groups. Significant positive correlations between the 2 radioligand bindings were observed in all brain regions (lateral temporal  cortex, r = 0.66, P = .015; medial temporal cortex, r = 0.67, P = .012; anterior cingulate cortex, r = 0.59, P = .034; raphe nuclei, r = 0.58, P = .037) of the healthy controls except the frontal cortex (r = 0.55, P = .054). However, no brain regions in the antidepressant-treated patients showed such significant correlations (frontal cortex, r = 0.15, P = .70; lateral temporal cortex, r = 0.21, P = .59; medial temporal cortex, r = 0.15, P = .71; anterior cingulate cortex, r = 0.25, P = .51; raphe nuclei, r = 0.17, P = .67).

DISCUSSION
To the best of our knowledge, this is the first study to compare the binding of the 2 PET radioligands for 5HT 1A receptors in patients with depression. There were no differences in 11 C-WAY-100635 BP ND values between patients with depression and healthy controls in most of the brain regions, implying that there were no marked changes in the density of 5-HT 1A receptors. However, 18 F-MPPF BP ND values in the neocortex and raphe nuclei of antidepressant-treated patients were significantly lower than in controls (Figure 2). 11 C-WAY-100635 and 18 F-MPPF exert different properties when binding to 5-HT 1A receptors. The binding of both ligands is considered to reflect 5-HT 1A receptor density, while 18 F-MPPF binding could also be influenced by extracellular 5-HT molecules. Accordingly, the positive correlations between the BP ND values of 11 C-WAY-100635 and 18 F-MPPF in most brain areas of healthy controls (Figure 3) suggest that PET data with both radioligands provide indicators for the density of 5-HT 1A receptors under a normal physiological condition. Meanwhile, the correlation between the 2 radioligand bindings appeared to be weakened in the antidepressant-treated patients. To ensure this finding, we statistically examined the difference in the coefficients of these correlations across brain regions between the patients and controls by unpaired t test and found that the coefficient values were significantly smaller in the patients than in the controls (P = .05; supplementary Figure 2), conceivably resulting from drug-induced increases in extracellular 5-HT levels and consequent decreases in 18 F-MPPF BP ND .
It is noteworthy that the difference in 18 F-MPPF binding between the treated depressive cases and controls was less remarkable in the medial temporal and anterior cingulate cortices, indicating that limbic extracellular 5-HT levels might not be markedly elevated in response to the treatment in a significant subset of patients. This was previously documented in the literature (Van Dyke et al., 2019;Pu et al., 2021). This finding could be associated with the efficacy of antidepressants, because patients with higher limbic 18 F-MPPF binding, which might stem from less profound drug-induced increases in 5-HT levels, exhibited higher scores on the depressive scale. Hence, the serotonergic system in the limbic region of patients with depression differentially responded to anti-depressive agents. This appears to be critically linked to the severity of depression during treatment.
Notably, the present study demonstrated that BP ND of 18 F-MPPF in limbic regions, especially the medial temporal cortex, exhibited significant correlations with the severity of depression. Limbic regions, including the medial temporal and anterior cingulate cortices, have been known to contain abundant 5-HT 1A receptors and diminished serotonergic neurotransmissions in these regions have been implicated in depressive symptoms (Saulin et al., 2012;Wang et al., 2016). Our findings suggest that a marked response to antidepressant medication may be elicited by a noticeable increase in extracellular 5-HT, which is represented by relatively low 18 F-MPPF binding to 5-HT 1A receptors. The Sequenced Treatment Alternatives to Relieve Depression trial indicated that a proportion of patients were resistant to sequential antidepressant treatments, highlighting the need for an objective indicator of drug efficacies at an initial stage of therapy (Rush et al., 2006). The current findings also imply that evaluation of 5-HT status in limbic regions using 18 F-MPPF early during therapeutic intervention in serotonergic transmission serves as a predictor of subsequent effects of individual symptoms. Accordingly, those who are potentially resistant to 5-HT-enhancing agents may be engaged by other approaches exemplified by augmentation therapies and non-drug treatments.
The reduction of 5-HT neurotransmission is critically involved in the molecular and neurochemical etiology of depression, which causes upregulation of postsynaptic 5-HT 1A receptors (Albert, 2012;Underwood et al., 2012). Previous studies have reported that untreated depressive patients with depression without antidepressant medication showed elevated 5-HT 1A receptor binding of 11 C-WAY-100635 relative to healthy controls (Parsey et al., 2006;Metts et al., 2019). Accordingly, the 2 depressive patients without antidepressant medication examined in this study showed a tendency of high 11 C-WAY-100635 BP ND compared with healthy controls in each of the brain regions. Antidepressant treatment appears to enhance synaptic 5-HT neurotransmission, resulting in reversal of the synaptic 5-HT 1A receptor density as assessed by 11 C-WAY-100635. Parsey and colleagues also reported no significant differences in 11 C-WAY-100635 binding between combined antidepressant-treated and untreated patients with depression and healthy controls (Parsey et al., 2006). Despite the putative antidepressant-induced normalization of 5-HT 1A receptor density, there were no significant correlations between the estimated 11 C-WAY-100635 BP ND values and the severity of the depressive symptoms in treated patients. This indicates that the density of 5-HT 1A receptors assessed by 11 C-WAY-100635-PET may not offer a neurochemical index of therapeutic efficacy.
In raphe nuclei, 18 F-MPPF binding in patients with depression treated with antidepressants was significantly lower than that of healthy controls. 5-HT 1A autoreceptors mainly distribute to the raphe nuclei and inhibit synaptic 5-HT release Courtney and Ford, 2016). Antidepressant medication induces 5-HT 1A autoreceptor internalization, presumably reducing 18 F-MPPF-accessible receptors. Previous studies have reported that antidepressant administration can be associated with 5-HT 1A autoreceptor internalization in raphe nuclei, which was consistent with the current findings Sibon et al., 2008). Depressive severity was not associated with the estimated BP ND values of the 2 radioligands in this area. As autoreceptor internalization in raphe nuclei may not directly reflect synaptic 5-HT concentrations in the projection areas, the relationship between depressive symptoms and 5-HT 1A receptor alterations may be elusive.
Several limitations need to be considered in this study. First, the statistical significance of the correlations between the 2 radioligand bindings and between the limbic 18 F-MPPF binding and HAM-D scores did not survive Bonferroni correction for multiple comparisons, primarily due to the relatively small sample size. However, such corrections are usually of importance for examining a single positive finding among many negative results. By contrast, a large portion of the t tests and correlation analyses performed here showed uncorrected significance. Nonetheless, the positive findings in the correlational analysis might be considered trends. Second, the number of antidepressant-naive and remitted patients was insufficient. Third, because the patients showed mainly mild or moderate disease severity, 5-HT 1A receptor density and extracellular 5-HT concentration in severely depressive and antidepressant-resistant patients are yet to be investigated. Because the severity of depressive symptoms was assessed at the time of the first PET scan, there was a possibility of minor mood fluctuations towards the second PET scan, notwithstanding the continuous treatment during this period and the relatively short interval (<2 weeks) between the 2 scans. The current data could also be affected by several factors other than antidepressant treatments, including nonpharmacological therapy given to the treated patients for a certain period as the participants of this study were not evaluated by PET with 11 C-WAY-100635 and 18 F-MPPF before antidepressant administration.
Additionally, most patients with depression in the current study received antidepressant medication at the time of the PET scans, impeding the separation between the disease-and treatment-associated alterations of the 5-HT 1A radioligand binging. The present assay was conducted under the assumption that the disease-associated changes in the receptor density could be evaluated by PET with 11 C-WAY-100635 because the binding of this radioligand is not susceptible to drug-induced elevations of extracellular 5-HT concentrations. However, this issue would be preferably addressed by PET assays of drug-naïve patients. Despite no direct interaction between antidepressants used in this study (SSRI or SNRI) and 5-HT 1A receptors, indirect influences of these drugs on the receptors via enhanced stimulation by 5-HT would be more appropriately identified by analyzing 18 F-MPPF binding before and after the initiation of the antidepressant therapy, although the performance of PET in untreated cases with depression might cause a delay in the commencement of the therapy. The antidepressant-provoked change in 18 F-MPPF binding and its association with clinical outcome should also be assessed, preferably in a longitudinal study consisting of examinations at baseline, as well as during treatment. Such analysis would be conducted with a larger sample size to assess the influences of the antidepressant types (e.g., SSRI vs SNRI) on PET findings. Moreover, we did not obtain detailed information on the histories of smoking, drinking, and intake of drugs other than antidepressants, although we briefly confirmed that each healthy control was neither a heavy smoker nor a heavy alcohol consumer upon the visit to the PET facility. Accordingly, we did not consider these factors as determinants of the exclusion criteria or confounding parameters in the statistical analysis.
This dual radioligand study was conducted based on the viewpoint that 18 F-MPPF binding reflected extracellular 5-HT concentrations as well as 5-HT 1A receptor density, in contrast to 11 C-WAY100635 binding solely reflecting receptor density. Because the inhibition constant (K i ) of 18 F-MPPF for 5-HT 1A receptors is 3.3 nM, much higher than that of 11 C-WAY100635 (0.8 nM) (Zhuang et al., 1994), 18 F-MPPF binding is more susceptible to extracellular 5-HT levels than 11 C-WAY100635 (Zimmer et al., 2002). We could not entirely rule out the possibility that the difference in binding potentials between the 2 ligands may arise from the reactivity of these compounds to off-target molecules, although WAY100635 and MPPF are known to display >100-fold and 50-fold selectivity, respectively, for 5-HT 1A receptors relative to other diverse CNS receptors, including non-1A 5-HT receptor subtypes (Gozlan et al., 1995;Kung et al., 1996). It is also likely that antidepressant treatment alters the nondisplaceable retention of 18 F-MPPF. Furthermore, 5-HT 1A receptors may be present in both high-affinity and low-affinity states (Watson et al., 2000), with the ratio between these states presumably being susceptible to changes in extracellular 5-HT levels caused by an antidepressant. 18 F-MPPF could also have higher selectivity for the receptors in the high-affinity state than 11 C-WAY100635. Besides these possibilities, it should also be considered that the drug-induced increase in extracellular 5-HT could promote internalization of the receptor, and 18 F-MPPF might not readily access the internalized receptors (Udo de Haes et al., 2005), unlike 11 C-WAY100635 (Gozlan et al., 1995). If 18 F-MPPF binding is affected by the affinity state and internalization of 5-HT 1A receptors elicited by 5-HT reuptake inhibition, the binding may still provide an indirect index for the extracellular 5-HT levels.
In conclusion, the combined use of 11 C-WAY-100635 and 18 F-MPPF can facilitate detailed in vivo assessments of the serotonergic system in depression. Serotonergic alterations in neocortical regions and raphe nuclei had less association with the severity of depressive symptoms. Serotonergic dysfunction in limbic regions is suggested to be critically significant in the pathophysiology of depression and the therapeutic effects of antidepressant treatment. Because adequate 5-HT release, induced by antidepressant medication in limbic regions, is clinically important for the improvement of depression, 18 F-MPPF binding in limbic regions may enable evaluation of the neurochemical responses to psychotropic treatment.

Supplementary Materials
Supplementary data are available at International Journal of Neuropsychopharmacology (IJNPPY) online.