Introduction

Depression is the psychiatric disorder with the greatest impact on the global disability burden.1 This is because it is common2 and because many individuals fail to respond to available treatments.3 Moreover, currently available medications typically require weeks to months to exert their beneficial effects, which are commonly lost within the first year of treatment.4 Identifying novel treatment targets and developing new tools that can rapidly ease the burden of depression to individuals and society is thus of high priority.

Glutamate signaling has an important role in the regulation of affective processes.5, 6, 7 A postmortem study reported elevated tissue glutamate levels in depression in the anterior cingulate cortex8 and magnetic resonance spectroscopy study in the anterior cingulate cortex9 and occipital cortex detected elevated glutamate levels in individuals with major depressive disorder (MDD).10 Elevations in glutamate may produce neuroadaptations in neural structure and function, including alterations in the regulation of glutamate receptors.11, 12 There is, however, also magnetic resonance spectroscopy literature showing lower glutamate and glutamate/glutamine levels in individuals with depression.13, 14

The metabotropic glutamatergic receptor 5 (mGluR5) is a receptor target for glutamate that may show depression-related changes in regulation. MGluR5 is a G-protein coupled receptor located presynaptically and postsynaptically15, 16, 17, 18 and is present on neurons both at the cell surface and intracellularly.19, 20, 21 Preclinical literature suggests that changes in mood states are associated with changes in mGluR5 levels. Knock out of mGluR5 in mice is associated with depressive-like behavior,22 and reductions in mGluR5 protein occur in animal models of depression.23, 24 In human postmortem brain tissue, reductions in mGluR5 protein levels have been reported in depressed individuals in the cerebellum25 and prefrontal cortex (PFC; BA9).26 A different postmortem study, however, failed to demonstrate significant alterations in the anterior cingulate mGluR5 density in MDD.27 In vivo, a positron emission tomography (PET) imaging study with [11C]ABP688 reported lower mGluR5 density in MDD in several regions.26 Chronic, but not acute, antidepressant treatment appears to increase mGluR5 expression,28 raising the possibility that therapeutic effects might be mediated by mGluR5 regulation.

In the present study, we compared ketamine effects upon mGluR5 availability in depressed patients and healthy comparison subjects. Ketamine is an N-methyl-D-aspartate (NMDA) glutamate receptor antagonist that produces clinically meaningful improvement in depression symptoms within 24 h of administration in a majority of patients.29, 30 Previous work in rodent models have shown evidence suggesting that rapid changes in glutamate cycling are related to the rapid antidepressant effects of ketamine and other rapidly acting agents.31, 32

Using [11C]ABP688, a PET tracer specific for the negative allosteric modulator (NAM) site on the mGluR5, our prior study demonstrated a rapid and large (20%) reduction in [11C]ABP688 binding after ketamine infusion in healthy humans.11 As glutamate does not compete with radioligands that target the NAM site on mGluR5 (Novartis, personal communication; Lin, in preparation), the ketamine-induced decrease in ligand binding could be an indicator of reduction in mGluR5 availability (that is, internalization) in response to glutamate released following ketamine administration11 or a glutamate-induced conformational change in the receptor that reduces the likelihood of radioligand binding at the NAM site. Presently, we examined ketamine effects on mGluR5 availability in depressed and healthy subjects. We hypothesized that the administration of ketamine would lead to an immediate reduction in mGluR5 availability and that the magnitude of the reduction of mGluR5 in the PFC or hippocampus would predict improvement in depressive symptoms in the depressed cohort. Secondary exploratory analyses investigated whether the reduction in mGluR5 availability during infusion is maintained at 24 h post-ketamine, the time point with highest antidepressant effect.

Materials and methods

Subjects

The Yale University Institutional Review Board, the Yale Radiation Safety Committee and the Yale-New Haven Hospital Radiation Safety Committee approved this study. After providing informed consent, inclusion criteria were assessed by: a physical examination, routine blood tests, and psychiatric and neurological examination. A drug screen (urine), electrocardiogram, and a pregnancy test (for female subjects) were performed as part of subject screening, before radiotracer administration. Inclusion criteria for both cohorts were: (1) Subjects are between 18 and 60 years and (2) English speaking. Inclusion criteria for healthy controls (HCs) were: (1) No current, or history of, any Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) diagnosis as assessed by structured clinical interview for DSM (SCID); (2) No first-degree relative with history of psychotic, mood or anxiety disorder; (3) No regular medication use within the past 2 months and no history of psychiatric medication use; (4) No lifetime drug abuse or dependence disorder (including for alcohol and nicotine). Inclusion criteria for depressed subjects were: (1) Currently in a depressed mood state as defined by Montgomery–Åsberg Depression Rating Scale (MADRS) severity scores, (2) Primary diagnosis of MDD as assessed by DSM-IV SCID, (3) No other current DSM-IV diagnosis, besides anxiety disorder, (4) No use of psychiatric medication in the past month preceding the study. (5) No significant suicidal ideation. (6) No lifetime drug abuse or dependence disorder (including for alcohol and nicotine).

Thirty-three subjects met eligibility criteria for the study (16 control and 17 depressed). Of those, one HC subject was withdrawn from the study after the baseline PET scan owing to high blood pressure (prior to ketamine administration); two HCs could not tolerate the full ketamine dose owing to psychotomimetic effects and were unable to continue with the scanning procedures, thus their data were discarded; two subjects with depression were excluded for abnormally rapid plasma radiotracer metabolism, which prevented an accurate fit of the plasma data; and one subject exhibited too much motion during the ketamine scan for accurate analysis. In total, 13 contemporaneous healthy non-smoking volunteers (6 males, 7 females) and 14 unmedicated mild to moderately depressed subjects (4 males, 10 females) completed this study (mean age: 33.5±13.2 years) (Table 1). Data from the first 10 HCs (same day assessment only) were published previously.11

Table 1 Demographic characteristics
Full size table

Psychiatric assessments

A structured clinical interview (SCID-Non-Patient) and psychiatric history were conducted at screening. MADRS33 and Beck Depression Inventory (BDI-II)34 were used to assess subjects’ depressive symptoms both during intake and on PET scan day (before and 30 min and 24 h after ketamine administration). The effects of ketamine on the subject’s mental state were assessed using the Clinician Administered Dissociative State Scale35 and Profile of Mood States.36

Positron emission tomography

[11C]ABP688 of high E/Z ratio (70:1; n=20) is produced from reacting [11C]methyl iodide with desmethyl-ABP688 in the presence of tetrabutylammonium hydroxide using the loop method developed by Nabulsi and colleagues.37 The average radiochemical and chemical purities is 97% and 93% average, respectively. The E/Z ratio of [11C]ABP688 conformers in the final PET drug product (range: 42/1–98/1) was determined for each produced batch by analytical radio high-performance liquid chromatography area percent. The E-isomer exhibits a higher KD in vivo.38 PET imaging was performed and analyzed as described previously.11

Subjects participated in three scans: two on same day—baseline [11C]ABP688 scan and [11C]ABP688 scan with ketamine infusion—and [11C]ABP688 scan 24 h after ketamine infusion. The radiotracer [11C]ABP688 was administered as a bolus in all scans. After the baseline [11C]ABP688 scan, subjects were given a short break (~1 h). A second [11C]ABP688 dose was administered (as a bolus over 1 min), and following successful radioligand administration, ketamine was immediately administered. The study was designed as such to decrease subject burden in case of equipment failure. In addition, the design ensures the immediate effects of ketamine (rapid glutamate release) are captured. Blood pressure, pulse and peripheral capillary oxygen saturation (SPO2) were acquired before and after ketamine administration and during the ketamine infusion (at 5–10 min intervals). Twelve (of the 14) depressed and 10 (of the 13) HC subjects returned for the 24-h scan. Of these scans, five (two from the depressed cohort, three from the HC cohort) were removed from analysis because arterial lines could not be placed.

The Yale New Haven Hospital Pharmacy provided racemic ketamine, which was administered intravenously11, 39, 40 (initial bolus: 0.23 mg kg−1 over 1 min, followed by constant infusion: 0.58 mg kg−1 over 1 h). The dose administered here is considered a psychotomimetic dose and not an antidepressant dose.

T1-weighted magnetic resonance images (MRIs) were acquired on a 3-T Trio imaging system (Siemens Medical Systems, Erlangen, Germany) and used to delineate anatomical regions on the PET (coregistered to the MRI). The MRI voxel size was 1 × 1 × 1 mm3.

Input function measurement

Prior to PET imaging, catheters were inserted in the forearm veins and radial artery for radioisotope injection and arterial blood sampling, respectively, as previously described.11 Radioactivity was analyzed as previously described.41 After correction for dispersion and delay,42 the automated and manual plasma concentration values were combined and convolved with a Gaussian function (full-width half-maximum=24 s) for smoothing.

A high-performance liquid chromatography assay of arterial blood samples (five samples at 0, 4, 12, 30, and 60 min) provided unmetabolized parent compound levels.43 These levels were fit with a Hill function.44 To calculate the input function, the fit parent fraction curve and merged plasma counts were multiplied. A combination of a straight line (before the peak) and the sum of three exponentials (after the peak) was used to fit the combined data in order to generate the metabolite-corrected arterial input function. Free fraction (fP) measurements were considered unreliable.45

Image analysis

Image analysis routines were implemented in MATLAB (The MathWorks, Natick, MA, USA) as previously described.11 Regions of interest were probabilistic, as determined by nonlinear registration of previously drawn regional atlases to the subject’s MRI as previously described.46 Cortical regions of interest were gray matter masked by voxel-wise multiplication (regional probability multiplied by the probability of that the voxel is in gray matter, as assessed by Statistical Parametric Mapping, SPM5; Institute of Neurology, University College of London, London, England). In this work, the striatum was defined as a weighted (by volume) average binding in the dorsal caudate, dorsal putamen and ventral striatum. A semiautomated technique was used to coregister the subject’s mean PET image to the MRI.47 The mean activity of a region, which was weighted by regional label probabilities, within each PET frame was then calculated to generate time activity curves.

Outcome measure calculation

An unconstrained two-tissue compartment mode was used to calculate regional outcome measures.11, 48 Given the unreliable fP values and a non-ideal reference region,49 VT (volume of distribution: ratio, at equilibrium, of the ligand concentration in the region of interest to that in the plasma50) was used as the main outcome measure. Percentage of change in binding was then calculated as ((VT,baseline−VT,ketamine or 24_hour_scan)/VT,baseline) × 100.

Statistical analysis

To determine the significance of measured VT differences due to ketamine administration or between groups or sexes at baseline, a linear mixed-effects model with region as a fixed effect was used. The dependence structure among regions and images from the same subject was modeled using the Kronecker product between unrestricted symmetry (to model the correlation among all regions) and compound symmetry (to model the correlation between two scans). The interaction term between region and scan was examined and, if appropriate, included in the model. To model change in subjects’ vital signs following ketamine administration, linear mixed models for longitudinal data were also used. The dependence structure used for these models was compound symmetry. The paired comparisons of scores from subjective reports from two time points were performed using Wilcoxon’s signed-rank test. Both unadjusted P-values and false discovery rate corrected values, based on Benjamini and Hochberg method, are provided for these paired comparisons. Tests were all two-sided. All analyses were performed using R 3.0.2 (http://www.r-project.org/) and SAS 9.2 (SAS, Cary, NC, USA).

Results

All mean results are presented as an average±s.d.

Vital signs/subjective report

Significant increases in blood pressure and heart rate were observed during the ketamine scan. Systolic and diastolic blood pressure and heart rate were significantly (P<0.05) higher than baseline until ~30 min after ketamine, similar to our previous report.11 SPO2 values differed from baseline only at the 4-min time point.

There were statistically significant, large reductions in depression scores following ketamine administration (45±70% on MADRS, 52±29% on BDI-II; all Ps<0.001) and 24 h after ketamine (53±69% on MADRS, 49±35% on BDI-II; all Ps<0.001) in the MDD group, which is similar to previously reported results29, 30 (Table 1).

There was a significant increase in Clinician Administered Dissociative State Scale scores during ketamine administration in both the HC and MDD groups (amnesia: F(2,22)=15.08, P<0.001; depersonalization: F(2,22)=24.48, P<0.001; derealization: F(2,22)=25.85, P<0.001), which returned to baseline at 24 h. No between-group differences were observed, however, suggesting that the immediate psychotomimetic effects of ketamine were similar between control and depressed groups (all Fs<1.97, all Ps>0.16).

Tracer metabolism/clearance

No significant differences were observed between scans in injected dose (568±143 MBq), specific activity (329.8±208.6 MBq nmol−1), injected mass (0.7±0.7 μg) or delivery rate of tracer from arterial plasma to brain tissue (K1),50 in either cohort (listed values represent averages over all scans). Clearance rates (injected dose divided by the calculated area under the metabolite-corrected arterial input function51) were not significantly different across time points in the depressed cohort (baseline: 93.0±12.5 l h−1, ketamine: 87.1±12.9 l h−1, 24 h: 91.7±14.4 l h−1, P=0.37). However, there was a significant decrease in clearance between the baseline and ketamine scans (−11.3±21.6% difference, on average) in the HC cohort. The average clearance values for the HC cohort were: baseline: 104.1±36.6 l h−1, ketamine: 89.5±29.2 l h−1, 24 h: 83.2±17.7 l h−1, P=0.03 (linear mixed-effects model across times).

Baseline analysis

Baseline mGluR5 availability (VT) was significantly lower in the depressed group as compared with the healthy cohort in all regions assessed (Figure 1, P=0.01, linear mixed-effects model across all regions).

Figure 1
figure 1

Average volume of distribution (VT) across healthy control (n=13, black) and depressed (n=14, white) subjects. All individual regional differences were significant in post hoc analysis. Error bars represent s.d. across subjects.

PowerPoint slide

Full size image

Ketamine-induced change in [11C]ABP688 binding

There were no significant between-group differences in the amount of ketamine injected (P=0.16). As indicated in Figure 2, significant binding (VT) reductions from baseline were observed on average during the ketamine scan and the 24-h scan both within the depressed group and the HCs (P<0.01, linear mixed-effects model, in both cases). No significant differences were observed between the ketamine scan and the 24-h scan in either group (P>0.8, in both cases). The linear mixed-effects model did not indicate a diagnosis by scan interaction (P=0.46), showing that the binding reductions (VT; from baseline to ketamine scan and from baseline to 24-h scan) were not significantly different between the depressed and control groups. Figures 2c and d show the average percentage of difference in VT from baseline. Across all subjects and regions, the percentage of reductions from baseline in the control and depressed cohort, respectively, were: 19±22% and 14±9% in the ketamine scan and 31±30% and 14±13% in the 24-h after ketamine scan. As indicated by this and Figures 2c and d, variability was particularly high for the 24-h after ketamine scan in controls (gray bars).

Figure 2
figure 2

Top: Average volume of distribution (VT) across (a) healthy control (n=13 total, 7 at 24 h) and (b) depressed (n=14 total, 10 at 24 h) subjects. Average values from the baseline (black bars), ketamine challenge (white bars) and 24 h after ketamine (gray bars) scans are shown. Regions are organized from left to right in order of the highest to the lowest mean baseline binding. Differences in all regions shown are statistically significant (P<0.05, uncorrected) in post hoc testing. Error bars represent s.d. across subjects. For comparison, the depressed and control cohort data are displayed with the same scale. Bottom: Average percentage of reduction in VT from baseline. Percentage of differences were measured across all subjects within a region and averaged. All average percentage of differences are negative. Absolute values are shown for easier visualization. Error bars represent s.d. across subjects. For comparison, the control (c) and depressed (d) cohort data are displayed with the same scale.

PowerPoint slide

Full size image

To examine this variability further, Figure 3 shows the changes in VT for each individual subject. On average, there was a significant decrease in binding during the ketamine scan, and this pattern was similar for individual brain regions. However, two HC subjects (Figure 3 left) showed increases in binding. Further, there was high variability in the 24-h scan in the HCs (subject binding differences from baseline ranged from −75±13% to 14±3%).

Figure 3
figure 3

Average volume of distribution (VT) within a subject. Each subject is represented by a single line. Plotted values are the average VT values across all regions.

PowerPoint slide

Full size image

Associations with mood during ketamine PET scan

We examined whether there was an association between ketamine-induced changes in mGluR5 availability and mood in the PFC and hippocampus. Given the sample size and exploratory nature of the analyses, we did not correct for multiple comparisons. Exploratory analyses (four in total: MADRS and BDI-II as measures and PFC and hippocampus as regions) revealed that reduction in mGluR5 availability in the hippocampus was associated with a reduction in total depressive symptoms assessed by the MADRS (r=0.52, P=0.035, Figure 4). The questions that MADRS and BDI-II include have several latent factors, including psychic anxiety and dysphoric apathy.52, 53 In our data, the change in mGluR5 availability in the hippocampus was strongly correlated with the psychic anxiety symptoms on both the MADRS (r=0.6, P=0.031, Figure 4) and the BDI-II (r=0.68, P=0.010). We did not observe similar associations between change in mGluR5 availability and the dysphoric symptoms.

Figure 4
figure 4

Relationship between ketamine-induced change in Montgomery–Åsberg Depression Rating Scale (MADRS) Score (total (blue) and psychic anxiety (red)) and percentage of change in volume of distribution (VT) in the hippocampus. Each symbol represents a different subject.

PowerPoint slide

Full size image

Discussion

The current study highlights the acute reduction in mGluR5 availability in vivo in humans following administration of ketamine, consistent with the capacity of this drug to enhance glutamate release.54 Reductions in mGluR5 availability persisted for 24 h following ketamine administration, suggesting that glutamate release triggers persisting adaptations in mGluR5s, perhaps through internalization. Interestingly, hippocampal mGluR5 changes during infusion correlated with ketamine’s rapid antidepressant effects, in particular the latent factor of psychic anxiety symptoms.

At baseline, mGluR5 availability (measured as VT) in the depressed cohort was significantly lower than in the controls. Lower mGluR5 availability in the depressed cohort was consistent with a previous report of lower mGluR5 levels in individuals with MDD in several brain regions and reduced PFC mGluR5 protein expression in the postmortem brains of depressed subjects.26

In this work, ketamine-induced reductions in mGluR5 availability were observed in both the depressed and control cohorts, and these reductions persisted 24 h after the ketamine infusion. The immediate decreases in mGluR5 availability are likely due to the rapid surge in extracellular glutamate leading to mGluR5 downregulation or internalization. The glutamate surge hypothesis is consistent with rodent data showing that a single administration of ketamine rapidly, but only transiently, induces increases in glutamate efflux54 and cycling.31, 32 The hypothesis is also in line with other recent studies demonstrating that activation of postsynaptic AMPA receptors is a critical component related to the rapid antidepressant-like effects associated with several additional compounds.55, 56 Evidence exists from other studies suggesting that activation of the AMPA receptors leads to changes in the mGluR5 signaling and facilitation of downstream effects,57 which could have a role in the current work.

Given the apparently similar ketamine-induced decrease in mGluR5 availability and similarity in the psychotomimetic response, it appears that the ketamine-induced surge in glutamate does not differ between these groups. Previous studies have found both elevated8, 9, 10 and reduced13, 14 levels of total tissue glutamate levels to be associated with MDD. These findings may suggest that some difference in ketamine-induced release would be expected between the MDD and healthy comparison subjects. However, it is critical to note that these studies are providing measures of total tissue glutamate levels and do not truly inform us about glutamate transmission to any extent, thus making it very difficult to predict the effect on stimulated glutamate release.

As noted above, the allosteric relationship between the radioligand used and the endogenous neurotransmitter glutamate is different in our study as compared with many published PET studies where the radiotracer interacts with the orthosteric site and thus is subject to direct competition by endogenous neurotransmitter. In our study, however, the radioligand is a NAM that binds to a site in the transmembrane domain of the receptor and not to the N-terminal orthosteric site where glutamate binds. Consistent with its different sites of binding, glutamate does not directly impact the binding of [11C]ABP688 in a membrane preparation but it does lead to receptor internalization (Lin et al., in preparation). Thus the measured reduction in receptor availability is potentially due to rapid receptor internalization. It is also possible that the rapid surge in glutamate leads to a conformational change of the receptor, reducing the ability of the radioligand to bind to the receptor. The overall result would be an observation of lower receptor availability. Overall, the decrease in ligand binding observed here immediately following treatment likely represents an indirect in vivo measure of the ketamine-induced glutamate surge. However, the exact mechanism leading to mGluR5 downregulation in the current study cannot be assessed with PET imaging; nonetheless, our data suggest that long-term reduced mGluR5 availability may have an important role in relieving symptoms of anxiety specifically.

The glutamate surge has been shown to be transient in nature.32, 54, 58 Thus the persistent mGluR5 downregulation at 24 h is unlikely to reflect continued exposure of these receptors to elevated glutamate levels. We hypothesize that the persistent reduction in mGluR5 availability reflects an imbalance between mGluR5 internalization and the rate of recycling of internalized receptor to the plasma membrane within the 24-h period between assessments. Preclinical studies are in line with this interpretation. mGluR5 receptors are anchored to the cytoskeleton via the proteins Homer and Shank. mRNA levels for Homer decline as early as 90 min after ketamine infusion, with downregulation of Homer 1B (although upregulation of Homer 1A has been reported).59 Persistent downregulation of mGluR5 may be among the persisting neuroadaptations to acute ketamine administration that are associated with the persistence of antidepressant effects beyond the brief occupancy of NMDA receptors by ketamine.56

It is possible that we are underestimating the ketamine-induced decrease in mGluR5 density during infusion. In studies of test–retest reliability of mGluR5 radioligand binding, when test and retest studies are conducted on the same day, we have consistently found an increase in binding in the retest (afternoon) scan60, 61 (average percentage of increase in VT: 27±32%). In prior work, we also reported a significant correlation between percentage of change in VT from baseline and percentage of change in K1/clearance from baseline using two tracers that bind to mGluR5—[11C]ABP688 and [18F]FPEB.61 We have hypothesized that the significant correlation between the percentage of change in VT and the percentage of change in clearance/K1 was due to the diurnal variation in glutamate underlying all three parameters: K1 (as there are multiple pathways by which glutamate influences blood flow in the brain62), clearance, and VT changes. This may be the case in this study as well: namely, the post-ketamine surge in glutamate may similarly affect tracer binding, K1 and clearance, as a similar correlation was observed (ρ=0.3/0.6, P=0.04/<0.01 for correlation of VT with K1/clearance across all subjects and all time points). Given that diurnal glutamate changes appear to be in the opposite direction of ketamine-induced changes, it is likely that we are underestimating the ketamine-induced decreases in mGluR5 binding by ~25%, on average. Further, the ultimate goal of this study was to examine the change in specific binding (that is, mGluR5 binding) due to ketamine. However, in this study, change in total binding (VT=specific plus nonspecific binding) was evaluated. The ratio of nonspecific to specific binding in the cerebellum was estimated by Kågedal et al.63 to be 1.33. As this ratio has only been measured in controls, it is unknown whether the ratio is different in a depressed cohort. Assuming that it is not, based on cerebellar VT, the non-specific binding of the depressed subjects is estimated to be ~1.3 and ~1.1 ml cc−1 for the controls, on average. Therefore, in the high binding regions of Figure 1 (all regions shown excluding the cerebellum), the non-specific binding is approximately one-third of the total binding (~27–38%). Ketamine is not expected to change the non-specific binding. Therefore, following ketamine treatment, when specific binding is decreased, the non-specific binding is expected to become a greater portion of the total binding (~34–44%, in this case). Taken together, this means that examining percentage of change in VT underestimates the total effect on specific binding. Given the values above, this underestimation is ~7% for the depressed cohort and ~10% for the HCs. As such, given these assumptions, using specific binding rather than total binding (VT) should not greatly affect overall conclusions.

We observed that individuals who exhibited the greatest reduction in mGluR5 density during ketamine infusion also exhibited the greatest response, primarily on symptoms related to psychic anxiety (Figure 4), immediately after treatment. Although still preliminary, the results are consistent with prior evidence suggesting a model in which the ketamine-induced glutamate surge, and presumably mGluR5 internalization, is a critical step relating to the rapid acting antidepressant effects of the drug.64 Briefly, blockade of NMDA receptors on tonically firing GABAergic interneurons is thought to lead to glutamatergic disinhibition and thus a surge of glutamate in the PFC. This surge increases AMPA receptor activation. which, coupled with ketamine’s inhibition of extrasynaptic NMDA receptors, initiates and enables postsynaptic activation of signaling pathways related to neuroplasticity. These pathways include those involving mammalian target of rapamycin (mTOR) and brain-derived neutrotrophic factor (BDNF).30, 65 These neuroplasticity pathways are hypothesized to underlie ketamine’s antidepressant response. mGluR5 antagonism and internalization have also been associated with rapid, although not long lasting, antidepressant effects in preclinical studies.66, 67 Furthermore, the functional interplay between mGluR5 and NMDA was previously related to antidepressant activity,68 potentially through facilitation of BDNF signaling pathways. A recent, phase 2b study (n=333), examining the efficacy of a selective mGluR5 NAM (basimglurant) as adjunctive therapy for previously treatment non-responsive MDD, failed to show the drug to produce a statistically significant difference in the primary end point (mean change in clinician-rated MADRS) compared with a placebo group.69 However, the study was complicated by very high placebo response rates and secondary analyses (at 1.5 mg) showed statistically significant differences in basimglurant adjuvant therapy versus placebo in multiple other scales, including the patient-rated MADRS and Clinical Global Impression-Improvement mean score. Thus the findings suggest that future studies are required to more fully understand the clinical utility of this psychopharmacological approach.

mGluR5 antagonism has also been shown to be effective in reducing anxiety symptoms.70, 71, 72, 73 The rapid anxiolytic effects of mGluR5 antagonists such as 2-methyl-6-(phenylethynyl)-pyridine do not appear to rely on new protein synthesis but may partially involve facilitation of mTOR signaling pathways and regulation of glutamate neurotransmission.70, 74 Sustained effects of mGluR5 antagonism, however, as seen at the 24 h time point, may rely on new protein synthesis and mTOR signaling pathways that lead to normalization of synaptic plasticity,75, 76 as well as other factors that have yet to be elucidated.

Recently, it has been suggested that a metabolite of ketamine, (2S,6S;2R,6R)-hydroxynorketamine (HNK) may be responsible for the antidepressant response as in preclinical studies HNK was necessary and sufficient for this response.55 The metabolite pathway is independent of NMDA receptors, acting through AMPA receptors to activate BDNF and mTOR pathways. However, even this independent pathway would affect mGluR5 downstream. Evidence from other studies suggests that activation of the AMPA receptors leads to changes in the mGluR5 signaling and facilitation of downstream effects,57 which could have a role in the current work. The exact mechanism leading to mGluR5 downregulation in the current study cannot be assessed with PET imaging; nonetheless, our data suggest that mGluR5 may have an important role in relieving symptoms of anxiety specifically.

Limitations

There are several limitations of this study. First, our study lacks a placebo control group. Ketamine is an increasingly well-known therapeutic agent, and individuals with MDD may have the expectation of reduction in symptoms after ketamine administration; some of the antidepressant response might be due to these expectations. However, the reduction in depressive symptoms was associated with a change in receptor availability, suggesting that the change in symptoms was not solely due to expectation but potentially due to the therapeutic effects of the glutamate surge and mGluR5 downregulation. Second, hemodynamic ketamine effects may alter tracer delivery and washout. The kinetic models used assume equilibrium conditions;50 however, heart rate and blood pressure were elevated (transiently) during ketamine administration. Furthermore, ketamine has been associated with increased blood flow in frontal regions,77, 78, 79 measured with functional MRI, although the effect reported was not great and isolated to a few regions. In this study, however, the delivery of tracer (K1) did not significantly change post-ketamine, and clearance only differed post-ketamine in the control group. Thus, given the magnitude of the previously reported effects and the lack of difference in K1 over the whole scan, even if ketamine were to cause transient changes in K1, these are unlikely to affect outcome. Further, the outcome VT, by incorporating the metabolite-corrected arterial input function, accounts for other potential changes induced by ketamine (for example, metabolism). Third, our study lacks long-term follow up. Repeat scanning about 2 weeks after ketamine administration, at the time when the antidepressant response wanes, would help determine the time course of mGluR5 re-insertion into the membrane and whether this corresponds to the increases in depressive symptoms. Fourth, only nonsmokers were studied in the present experiment. The extent of ketamine-induced changes on mGluR5 availability (or [11C]ABP688 binding) in smokers has yet to be determined. Although previous work showed that there is downregulation in mGluR5 with chronic tobacco exposure80 and both mGluR5 NAMs and NMDA antagonists appear to reduce smoking-related behaviors, whether there are smoking-specific differences in the glutamate surge (and mGluR5 downregulation) after ketamine administration is not known. Thus the current results might only apply to the nonsmoking MDD population. Finally, the ketamine dose administered here is higher than the typical dose used for clinical purposes. Thus the extent of change in receptor availability might be greater in this study as compared with the general practice.

Conclusion

We conducted a novel examination to evaluate ketamine-induced changes in mGluR5 availability using PET in healthy and depressed cohorts. Significant decreases in [11C]ABP688 binding, a tracer specific for mGluR5, were observed in both cohorts. It is likely that this decreased binding reflects mGluR5 downregulation/internalization. Exploratory analyses suggest that mGluR5 downregulation (and/or the likely surge in glutamate) has a role in the reduction of depressive symptomatology. These data support the development and implementation of mGluR5 therapies for the relief of depressive symptomatology.