Electrophysiological Biomarkers of Antidepressant Response to Ketamine in Treatment-Resistant Depression: Gamma Power and Long-Term Potentiation

Abstract

Over the last two decades, the discovery of ketamine’s antidepressant properties has galvanized research into the neurobiology of treatment-resistant depression. Nevertheless, the mechanism of action underlying antidepressant response to ketamine remains unclear. This study reviews electrophysiological studies of ketamine’s effects in individuals with depression as well as healthy controls, with a focus on two putative markers of synaptic potentiation: gamma oscillations and long-term potentiation. The review focuses on: 1) measures of gamma oscillations and power and their relationship to both acute, psychotomimetic drug effects as well as delayed antidepressant response in mood disorders; 2) changes in long-term potentiation as a promising measure of synaptic potentiation following ketamine administration; and 3) recent efforts to model antidepressant response to ketamine using novel computational modeling techniques, in particular the application of dynamic causal modeling to electrophysiological data. The latter promises to better characterize the mechanisms underlying ketamine’s antidepressant effects.

1. Introduction

It has been three decades since seminal work implicated the NMDA subtype of glutamate receptors in the pathophysiology of depression (Trullas and Skolnick, 1990). A decade later, ketamine’s antidepressant effects were first reported in a controlled trial (Berman et al., 2000), galvanizing research into the neurobiology underpinning mood disorders and leading to increased focus on the glutamatergic system’s putative role in both major depressive disorder (MDD) (Bernard et al., 2011; Choudary et al., 2005) and bipolar depression (Eastwood and Harrison, 2010). A wealth of studies shave since explored the efficacy of ketamine and its metabolites as a clinical treatment option for mood disorders (Diazgranados et al., 2010; Zanos et al., 2019; Zarate et al., 2006), and interest in targeting the glutamatergic system has grown exponentially (Ohgi et al., 2015).

A growing body of evidence now demonstrates that a single infusion of subanesthetic-dose ketamine rapidly (within hours) relieves depressive symptoms in individuals with both MDD (Fava et al., 2018; Murrough et al., 2013a; Murrough et al., 2013b; Zarate et al., 2006) and bipolar depression (Diazgranados et al., 2010; Zarate et al., 2012), including those who are treatment-resistant. Studies have also observed rapid antidepressant effects following intranasal ketamine administration (Lapidus et al., 2014), which led the FDA to approve esketamine (the intranasally-administered S-enantiomer of ketamine) as a conjunctive treatment for treatment-resistant depression in 2019. Repeat infusions of ketamine have been shown to maintain ketamine’s antidepressant effects far beyond that of a single administration (Murrough et al., 2013b; Phillips et al., 2019). Additional studies have also explored ketamine’s effectiveness in other clinical domains, including obsessive-compulsive disorder (Rodriguez et al., 2013), post-traumatic stress disorder (Feder et al., 2014), suicidal ideation (Murrough et al., 2015), and social anxiety disorder (Taylor et al., 2018). Given the broad range of its therapeutic effects, understanding the mechanism of action underlying rapid antidepressant response to ketamine could ultimately help expedite the development of novel, rapid-acting therapeutics to more effectively treat depressive symptoms without the associated psychotomimetic side effects and abuse potential of ketamine (Krystal et al., 1994).

Ketamine is a noncompetitive N-methyl-D-aspartate (NMDA) receptor antagonist (Hirota and Lambert, 1996). Early work examining the link between glutamatergic system dysfunction and depression focused on the role of NMDA receptor activation in long-term potentiation (LTP) in the hippocampus (Trullas and Skolnick, 1990), while more recent work has focused on ketamine’s specific role in enhancing LTP outside the hippocampus (Kirk et al., 2010; Sumner et al., 2019). However, the pharmacological profile of ketamine and its underlying mechanism of action appear to extend beyond simply modulating glutamate transmission and include direct and indirect high affinity antagonistic binding properties at the NMDA receptor as well as α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) throughput modulation (Maeng et al., 2008; Zanos et al., 2016). Subanesthetic-dose ketamine administration leads to immediate presynaptic disinhibition of glutamatergic neurons, producing a glutamate surge (Moghaddam et al., 1997). This surge is thought to result from NMDA receptor blockade by ketamine of fast-spiking γ-aminobutyric acid (GABA)-ergic interneurons, leading to local inhibition of interneuron tonic firing and subsequent disinhibition of pyramidal neurons (Homayoun and Moghaddam, 2007; Widman and McMahon, 2018). Due to NMDA receptor blockade on post-synaptic excitatory neurons, excess synaptic glutamate is primarily taken up by AMPA receptors (Duman et al., 2019; Moghaddam et al., 1997), thereby activating neuroplasticity-related signaling pathways including mammalian target of rapamycin complex 1 (Li et al., 2010; Li et al., 2011) and brain-derived neurotrophic factor (Liu et al., 2012a), both of which result in increased synaptic potentiation and synaptogenesis. This synaptic potentiation and synaptogenesis has been posited to ultimately form the foundation of ketamine’s antidepressant effectiveness (Autry et al., 2011; Duman and Aghajanian, 2012; Nosyreva et al., 2013).

This paper will focus on two promising neuroimaging markers of ketamine response in electrophysiology: gamma oscillations (approximately 30–60 Hz) and LTP, both of which are putative measures of synaptic potentiation (Table 1). The manuscript will also briefly discuss how new computational modeling methods can be applied to better characterize ketamine’s effects in electrophysiological data (Table 2). Here, an attempt has been made to synthesize the disparate literature on these electrophysiological markers, including providing hypotheses about ketamine’s specific role in generating these markers and their relationship to antidepressant response. The literature is summarized in three sections: ketamine and gamma oscillations and power, ketamine and LTP, and computational modeling of ketamine effects. Key limitations and areas where further research is needed are noted wherever appropriate.

Table 1.

Gamma Oscillations/Power, Long-Term Potentiation, and Ketamine Response

Authors	Ketamine Dosage	Time of measurement (acute or delayed)	Study Sample	Study Design/Task	Modality	Placebo controlled	Ketamine Findings
Cornwell et al., 2012	Subanesthetic dose infusion	Delayed effects measured 6 to 7 hours post-infusion	MDD, N=20	Somatosensory stimulation	MEG	No	Increased gamma band (30–50 Hz) power in responders compared with non-responders
de la Salle et al., 2016	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=21	Resting state, eyes closed	EEG	Yes	Increased gamma band (30–60 Hz) current density estimates across the scalp and in regions of the default mode and salience network
Gilbert et al., 2018	Subanesthetic dose infusion	Delayed effects measured 6 to 9 hours post-infusion	HV, N=18 and MDD, N=18	Somatosensory stimulation	MEG	Yes	Increased gamma band (30–58 Hz) evoked responses in bilateral somatosensory cortex and frontal cortex
Hong et al., 2010	Subanesthetic dose infusion	Acute effects measured immediately following infusion	HV, N=10	Paired click auditory task	EEG	Yes	Increased gamma band (40–85 Hz) power spectrum density at CZ electrode site
Lazarewicz et al., 2010	Anesthetic dose infusion	Acute effects measured during infusion	Mice, N=20	Paired click auditory task	Implanted EEG	No	Increased background gamma band (30–80 Hz) power, increased gamma band evoked power, and increased induced gamma band responses
Maksimow et al., 2006	Subanesthetic followed by anesthetic dose infusion	Acute effects measured during infusion	HV, N=8	Resting state and following sedation	EEG	Active propofol anesthesia control	Increased gamma band (30.2–48 Hz) activity and gamma spindles during anesthetic dosing
Moran et al., 2015	Anesthetic dose infusion	Acute effects measured during infusion	Rats, N=10	Locomotor activity in open-field arena	EEG	Yes	Increased gamma band (40–75 Hz) power, particularly in medial prefrontal cortex
Muthukumaraswamy et al., 2015	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=19	Resting state, eyes open	MEG	Yes	Increased low (30–49 Hz) and high (51–99 Hz) gamma band activity
Nugent et al., 2019a	Subanesthetic dose infusion	Delayed effects measured 6 to 9 hours post-infusion	HV, N=25 and MDD, N=35	Resting state, eyes closed	MEG	Yes	Increased gamma band (30–50 Hz) power in both groups; for MDDs, higher post-ketamine gamma band power was associated with better response
Nugent et al., 2019b	Subanesthetic dose infusion	Delayed effects measured 6 to 9 hours post-infusion	HV, N=25 and MDD, N=31	Somatosensory stimulation	MEG	Yes	Difference in gamma band (30–50 Hz) power between ketamine and placebo sessions significantly correlated with antidepressant response
Shaw et al., 2015	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=20	Visuomotor paradigm	MEG	Yes	Increased gamma band amplitude in primary motor (60–90 Hz) and visual (30–80 Hz) cortices with reduced peak gamma frequency in visual cortex
Sumner et al., 2019	Subanesthetic dose infusion	Delayed effects measured 3 to 4 hours post-infusion	MDD, N=30	Visual grating stimulation	EEG	Active remifentanil hydrochloride control	Increased visual P2 component long-term potentiation compared with placebo
Zacharias et al., 2020	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=17	Resting state, eyes closed	EEG	Yes	Marginally increased gamma band (30–50 Hz) activity

MDD: individuals with major depressive disorder; HV: healthy volunteers; MEG: magnetoencephalography; EEG: electroencephalography

Table 2.

Dynamic Causal Modeling of Ketamine Effects

Authors	Ketamine Dosage	Time of measurement (acute or delayed)	Study Sample	Study Design/Task	Modality	Regions Included	Modeled Ketamine Effects
Gilbert et al., 2018	Subanesthetic dose infusion	Delayed effects measured 6 to 9 hours post-infusion	HV, N=18 and MDD, N=18	Somatosensory stimulation	MEG	Primary somatosensory cortex and frontal cortex	Increased top-down NMDA receptor drive between frontal cortex and primary somatosensory cortex; reduced AMPA receptor drive associated with antidepressant response two weeks following ketamine administration
Moran et al., 2015	Anesthetic dose infusion	Acute effects measured during infusion	Rats, N=10	Locomotor activity in open-field arena	EEG	Hippocampus and medial prefrontal cortex	Increased bottom-up AMPA receptor drive and reduced NMDA receptor drive from hippocampus to medial prefrontal cortex coupled with reduced top-down NMDA receptor drive; decreased intrinsic NMDA estimates in hippocampus and increased local feedback inhibition onto hippocampal pyramidal cells
Muthukumaraswamy et al., 2015	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=19	Resting state, eyes open	MEG	Parietal and frontal cortices	Reduced top-down AMPA receptor and NMDA receptor drive between frontal and parietal cortices; decreased gain on parietal superficial pyramidal cells associated with self-reported blissful state
Rosch et al., 2019	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=18	Auditory oddball paradigm	EEG	Primary auditory cortex, superior temporal gyrus, and inferior frontal gyrus	Increased inhibitory drive in primary auditory cortex and superior temporal gyrus; decreased inhibitory drive in inferior frontal gyrus
Schmidt et al., 2012	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=19	Auditory oddball paradigm	EEG	Primary auditory cortex, superior temporal gyrus, and inferior frontal gyrus	Increased bottom-up drive between primary auditory cortex and superior temporal gyrus for deviant tones; bottom-up drive associated with self-reported impaired control and cognition
Shaw et al., 2019	Subanesthetic dose infusion	Acute effects measured during infusion	HV, N=20	Visual grating stimulation	MEG	Visual cortex	Reduced gain on superficial pyramidal cells coupled with increased gain on inhibitory interneurons; shortened decay time of NMDA receptors, increased decay time on GABA receptors, and increased cortico-thalamic projection from layer IV into the thalamus relay populations
Sumner et al., 2019	Subanesthetic dose infusion	Delayed effects measured 3 to 4 hours post-infusion	MDD, N=30	Visual grating stimulation	EEG	Middle occipital gyrus, inferior temporal cortex, and superior parietal cortex	Increased bottom-up drive between middle occipital gyrus and both inferior temporal cortex and superior parietal cortex; asymmetrical increased and decreased top-down drive between these same regions

MDD: individuals with major depressive disorder; HV: healthy volunteers; MEG: magnetoencephalography; EEG: electroencephalography

2. Ketamine and Gamma: Altering Excitation/Inhibition Balance

Within psychiatry, a growing body of evidence suggests that alterations in the ratio of cortical excitation/inhibition balance could underlie a host of disorders, including depression (Fogaça and Duman, 2019; Fuchs et al., 2017; Godfrey et al., 2018; Kato et al., 2019). Short-lived gamma frequency oscillations have been shown to emerge from the coordinated interaction of excitation and inhibition in modeling work (Brunel and Wang, 2003; Buzsáki and Wang, 2012; Economo and White, 2012; Ray and Maunsell, 2015) as well as in animal work (Csicsvari et al., 2003; Whittington et al., 1995). In addition, the magnitude of gamma oscillations has been shown to be modulated by slower rhythms, including delta (approximately 1–4 Hz), theta (approximately 5–8 Hz), and alpha (approximately 9–14 Hz) frequencies (Buzsáki and Wang, 2012). Recent empirical work has explored the link between depression and gamma oscillations more directly, with several studies demonstrating alterations in gamma oscillations in both MDD and bipolar disorder (Bi et al., 2018; Lee et al., 2010; Liu et al., 2012b; Liu et al., 2014). These changes potentially reflect alterations in glutamatergic excitation-GABAergic inhibition concomitant with depression (Fee et al., 2017; Lloyd et al., 1989), suggesting that aberrant gamma oscillations might be a putative biomarker of mood disorders, including MDD (for a review see (Fitzgerald and Watson, 2018)). As an extension of this work, gamma rhythms have been found to be correlated with the generation of action potentials (Nir et al., 2007; Watson et al., 2018), suggesting that pharmacological interventions that increase gamma via synaptic potentiation might serve to alleviate depressive symptoms by increasing cortical excitation.

Acute ketamine administration, which leads to pyramidal cell disinhibition downstream of NMDA receptor antagonism (Homayoun and Moghaddam, 2007; Widman and McMahon, 2018), should lead to robust changes in gamma oscillations. In fact, this finding has been well-replicated across species, perhaps driven by reduced NMDA receptor-mediated input to fast-spiking parvalbumin-expressing GABAergic interneurons (Anver et al., 2011; Carlén et al., 2012; Hong et al., 2010; Lazarewicz et al., 2010; Maksimow et al., 2006; Sohal et al., 2009). In addition, acute ketamine administration has been found to decrease lower-frequency oscillations, including delta (Hong et al., 2010) and theta (Lazarewicz et al., 2010) oscillations, in line with predictions from modeling work. While it is outside the scope of this review to focus on lower-frequency oscillations and ketamine-mediated antidepressant response, this is certainly an area for further evaluation. Instead, the present review examines changes in gamma oscillations in response to acute (effects around the drug’s half-life) and delayed (effects occurring after the drug’s half-life as well as psychotomimetic side effects) ketamine administration, with a focus on potential mechanistic explanations for these effects drawn from electrophysiological studies of ketamine administration in both healthy individuals and those with mood disorders.

In both humans and animals, recent electrophysiological work found robust increases in gamma oscillations and measures of gamma power following acute ketamine administration at therapeutic, subanesthetic levels (de la Salle et al., 2016; Hong et al., 2010; Maksimow et al., 2006; Muthukumaraswamy et al., 2015; Shaw et al., 2015; Zacharias et al., 2020) as well as at higher doses where schizophrenia-like effects become more manifest (Lazarewicz et al., 2010; Moran et al., 2015), though these effects may not translate to increases in gamma coherence (Witkin et al., 2017). In healthy human controls receiving subanesthetic-dose ketamine, increased gamma oscillations/power were found in resting-state networks (de la Salle et al., 2016; Maksimow et al., 2006; Muthukumaraswamy et al., 2015; Zacharias et al., 2020) and in motor and visual areas during a visuomotor task (Shaw et al., 2015). Given that acute ketamine administration, even at subanesthetic doses, leads to transient dissociative and psychotomimetic effects that resemble the positive and negative symptoms of schizophrenia (Krystal et al., 1994), one potential hypothesis for this acute increase in gamma is that it reflects dysfunctions of conscious integration, that is, the dissociative and psychotomimetic effects that occur during ketamine administration (Pinault, 2008). A second hypothesis is that sustained and elevated gamma power post-ketamine might constitute pyramidal cell noise, disrupting information processing within pyramidal cell assemblies (Moran et al., 2015). Evidence from animal work supports this second assertion, as theta-gamma coupling within the hippocampus was enhanced at a low ketamine dose (25 mg/kg, used to test a schizophrenia model), but disrupted at higher doses (Caixeta et al., 2013). These findings point to acute increases in gamma synchrony that reflect both a drug-induced dissociative state and associated changes in pyramidal cell information processing. What has yet to be determined, however, is how these acute gamma oscillations relate to antidepressant response in specific mood disorders, or how they might directly relate to the dissociative clinical symptoms experienced during ketamine administration.

To date, no studies have assessed ketamine’s acute effects on gamma oscillations in mood disorders. However, several studies have assessed its delayed effects on both gamma power and antidepressant response (Cornwell et al., 2012; Gilbert et al., 2018; Nugent et al., 2019a; Nugent et al., 2019b); such studies have measured electrophysiological changes in gamma power during an antidepressant response window outside of ketamine’s acute, dissociative drug effects, which dissipate by about one hour post-infusion in most individuals (Luckenbaugh et al., 2014). These studies all found enhanced gamma power several hours (approximately six to nine hours) post-ketamine administration, including within large-scale brain networks during rest (Nugent et al., 2019a) and within primary and secondary somatosensory cortices following tactile stimulation of the index finger (Cornwell et al., 2012; Gilbert et al., 2018; Nugent et al., 2019b).

With regard to antidepressant response, Cornwell and colleagues found that ketamine responders (defined as those experiencing a greater than 50% reduction in depressive symptoms post-ketamine) showed a uniform increase in gamma power (a putative metric of cortical excitability and synaptic potentiation) post-ketamine infusion compared to baseline (Cornwell et al., 2012), but that non-responders showed no such changes. The same study also found a positive correlation between gamma power and plasma levels of norketamine, a major active metabolite of ketamine with a relatively long half-life (Cornwell et al., 2012). One limitation of the study, however, was that ketamine was administered open-label, meaning that there was no control for expectancy effects. To address this short-coming, Nugent and colleagues replicated these same findings of increased gamma power using a double-blind, placebo-controlled study design, providing further evidence that delayed electrophysiological measures of gamma power might be an important biomarker of antidepressant response to ketamine (Nugent et al., 2019b). Building on this work, Nugent and colleagues subsequently found that baseline levels of resting-state gamma power within large-scale brain networks in individuals with MDD moderated the relationship between post-ketamine changes in gamma power and antidepressant response. In particular, they found that depressed participants with lower baseline gamma power levels who experienced large increases in gamma power post-ketamine had better antidepressant response than those with higher baseline gamma levels who also experienced large increases in gamma power post-ketamine (Nugent et al., 2019a).

Taken together, these findings suggest that there might be some optimal level (a “sweet spot”) of gamma power that is key to mediating antidepressant response following ketamine administration. Within the field of psychiatry, these findings also provide a potential bridge between gamma oscillations, ketamine’s antidepressant effects, depression, and schizophrenia – the idea of an optimal homeostatic balance of excitation/inhibition. This notion would be consistent with the idea that increases (e.g., in schizophrenia) and decreases (e.g., in depression) in gamma power beyond a homeostatic ideal could be pathological, with ketamine-mediated upregulation of gamma power providing antidepressant relief only in depressed individuals who have reduced gamma power levels at baseline (reflecting lower levels of excitation-inhibition). Further work is needed to examine the robustness of these findings.

3. Ketamine and Long-Term Potentiation

The pivotal role that NMDA receptors play in LTP and synaptic plasticity was first described almost 40 years ago, focusing on LTP within the hippocampus (Collingridge et al., 1983; Harris et al., 1984). Subsequently, the first experimental test of NMDA antagonists used an animal model of depression (inescapable stress) to examine how reduced NMDA neurotransmission within an NMDA receptor-rich area of the hippocampus could ameliorate depressive symptomatology, thus providing the first evidence that NMDA receptor antagonism was associated with effective antidepressant response (Trullas and Skolnick, 1990). Since then, the role that NMDA receptors play in LTP was found to extend outside of the hippocampus to anatomically early areas within the visual (Artola and Singer, 1987; Kirk et al., 2010) and auditory cortices (Kirk et al., 2010) using paradigms that include rapid, repetitive presentations of sensory stimuli (e.g., flashing visual checkerboards, trains of auditory tones). In addition, several studies have demonstrated reduced LTP in the visual and auditory cortices of individuals with mood disorders (Normann et al., 2007; Ostermann et al., 2012; Yeap et al., 2009). Taken together, these studies suggest that up-regulation of LTP via ketamine administration might serve as an important mechanistic process mediating ketamine’s antidepressant effects.

While the importance of NMDA receptors in LTP is well known, only a few electrophysiological studies have explored ketamine-mediated effects on LTP in humans, and only one has focused on antidepressant response in MDD (Sumner et al., 2019). Sumner and colleagues examined LTP as a marker of neural plasticity within a delayed time window of three to four hours post-ketamine infusion and found enhanced LTP in the visual system following ketamine administration; specifically, the P2 event-related potential (ERP) amplitude was increased post-ketamine compared to an active placebo, remifentanil (Sumner et al., 2019). However, the authors found no evidence of a relationship between ERP changes and antidepressant response, leaving open the question of whether ERP effects generally track with antidepressant response. Given that this is the only study to date demonstrating ketamine-mediated effects on LTP in humans with depression, more work is needed to determine whether measures of LTP in primary visual and auditory cortices could serve as effective biomarkers of ketamine-mediated antidepressant response.

4. Computational Modeling of Ketamine’s Effects

Recent advances in computational modeling have shifted focus onto translational efforts to bridge neuroscience and psychiatry. These novel computational methods attempt to characterize abnormalities in brain systems that underlie psychiatric diseases, including depression. Though in its infancy, such work could potentially enable the identification, categorization, and prediction of dimensional processes in depression and other psychiatric conditions (Ferrante et al., 2019). Progress in this field has led to recent attempts to use computational methods to characterize ketamine’s effects in animal models (Moran et al., 2015) as well as in both healthy controls (Muthukumaraswamy et al., 2015; Schmidt et al., 2012; Shaw et al., 2019) and individuals with mood disorders (Gilbert et al., 2018; Sumner et al., 2019). These studies have all used dynamic causal modeling (DCM), which fits a biologically-plausible model of neural dynamics to measured electrophysiological signals. While it is outside the scope of this review to describe the sets of coupled differential equations that govern dynamics in DCM, model inversion—the fitting of parameterized mean-field neuronal models to electrophysiological data features—results in parameter estimates that govern unobservable neuronal states such as receptor-mediated connectivity between cell populations and decay times of specific receptor types. These models have been used to measure parameter effects using ketamine dosing levels that exacerbate schizophrenia symptoms (Moran et al., 2015) as well as lower, antidepressant-level doses (Gilbert et al., 2018; Muthukumaraswamy et al., 2015; Rosch et al., 2019; Schmidt et al., 2012; Shaw et al., 2019; Sumner et al., 2019). Below, we focus first on ketamine’s effects in animals and healthy humans, then discuss ketamine’s antidepressant effects.

In animal models of schizophrenia, ketamine was found to amplify bottom-up AMPA receptor drive and attenuate NMDA receptor drive between the hippocampus and medial prefrontal cortex (the two regions modeled in the task) in a dose-dependent fashion; ketamine also attenuated top-down NMDA receptor drive from medial prefrontal cortex to hippocampus in a dose-dependent fashion (Moran et al., 2015) (Figure 1). In addition, ketamine decreased intrinsic estimates of NMDA in the hippocampus in a dose-dependent fashion, while also increasing local feedback inhibition onto hippocampal pyramidal cells (Moran et al., 2015). In humans, acute, subanesthetic-dose ketamine attenuated top-down AMPA and NMDA receptor drive between the frontal and parietal cortices in healthy individuals at rest, and also decreased the gain on parietal superficial pyramidal cells (Muthukumaraswamy et al., 2015). During a visuomotor task, ketamine decreased the gain on superficial pyramidal cells while increasing the gain on inhibitory interneurons within the visual cortex (Shaw et al., 2019). Ketamine administration was also found to shorten the decay time of NMDA receptors in the visual cortex, lengthen the decay time of GABA_A receptors in the visual cortex, and increase the cortico-thalamic projection from layer VI in the visual cortex into the thalamus relay populations (Shaw et al., 2019). Finally, using auditory mismatch negativity tasks, ketamine was found to amplify the bottom-up drive between the primary auditory cortex and the superior temporal gyrus for deviant tones (Schmidt et al., 2012) and to increase the inhibitory drive within both these regions while decreasing inhibitory drive in the inferior frontal gyrus (Rosch et al., 2019).

Figure 1.

Dynamic Causal Modeling of Ketamine Effects on AMPA, NMDA, and GABA Parameters. Dynamic Causal Modeling (DCM) has been used to model bottom-up and top-down drive in brain regions important for somatosensory, locomotor, auditory, and visual tasks, in addition to resting state network regions. Using models that include parameters for AMPA, NMDA, and GABA receptor signaling, acute ketamine administration has been found to attenuate bottom-up and top-down NMDA receptor drive, in addition to shortening the decay time of NMDA receptors in lower-level regions of the processing hierarchy. In addition, acute ketamine has been found to increase the bottom-up drive of AMPA receptors, decrease the top-down drive of AMPA receptors, and lengthen the decay time of GABA receptors in lower-level regions. In contrast, delayed measures of receptor-mediated signaling suggest that ketamine increases both bottom-up and top-down NMDA receptor drive, perhaps related to upregulation and sensitization effects. In addition, while both NMDA and AMPA receptor drive are associated with antidepressant response on the day of infusion, only changes in AMPA receptor drive were associated with longer-term changes in antidepressant response at two weeks following ketamine administration. R1: region 1; R2: region 2; C: control; K: ketamine.

Interestingly, of these parameters, only two have been associated with clinical symptoms: the pyramidal gain parameter and the bottom-up drive from primary auditory cortex to superior temporal gyrus. Muthukumaraswamy and colleagues found that the pyramidal gain parameter was associated with self-reported blissful state following ketamine administration (Muthukumaraswamy et al., 2015), strongly suggesting that this parameter is associated with the drug’s psychotomimetic effects. Schmidt and colleagues found that the bottom-up drive between the primary auditory cortex and the superior temporal gyrus was associated with individuals’ subjective ratings of impaired control and cognition, again suggesting that this parameter is associated with psychotomimetic effects. As an overarching summary, ketamine administration seems to alter NMDA and AMPA receptor drive/connectivity in key regions important for both resting networks and task-specific networks, and some of these parameters are associated with dissociative and psychotomimetic effects during acute administration.

Modeling delayed ketamine-mediated effects in individuals with MDD, ketamine was found to amplify the top-down NMDA receptor drive between frontal cortex and primary somatosensory cortex during a somatosensory stimulation task (Gilbert et al., 2018). Given that ketamine is an NMDA antagonist, this unexpected result is perhaps attributable to upregulation and sensitization effects. In terms of antidepressant response, both reduced NMDA and AMPA receptor drive have been associated with better antidepressant response, though only changes in AMPA receptor drive were associated with longer-term changes in antidepressant response two weeks following ketamine administration (Gilbert et al., 2018). This second finding is in keeping with the hypothesis that changes in AMPA throughput mediate antidepressant response. Using a visual grating task to measure changes in LTP, ketamine was also found to broadly amplify bottom-up drive between the middle occipital gyrus and both the inferior temporal cortex and superior parietal cortex, while asymmetrically attenuating and amplifying top-down drive between these same regions (Sumner et al., 2019). This latter study used a second-level Bayesian modeling extension of DCM to assess ketamine’s effects and did not directly measure associations between modeled parameter estimates and antidepressant response.

In summary, recent advances in computational modeling offer promise for uncovering how ketamine alters receptor-mediated connectivity between cell populations and even decay times of specific receptor types. However, much research is still needed in this area, particularly efforts to explore ketamine-mediated effects in depression, which could help better characterize differences between those who respond to ketamine versus those who do not. More work is also needed to model acute drug effects in mood disorders, which could potentially help differentiate parameters that govern psychotomimetic effects versus those that mediate antidepressant response.

5. Conclusions

Electrophysiological measures of both gamma oscillations/power and LTP hold considerable promise for uncovering the potential mechanistic processes mediating ketamine’s antidepressant effects. As reviewed above, acute increases in gamma oscillations have been identified following ketamine administration, although additional work is needed to determine whether this marker is associated with antidepressant response in addition to dissociative and psychotomimetic drug effects. However, delayed estimates of gamma power are potentially promising biomarkers of ketamine response and synaptic potentiation that highlight important differences between responders and non-responders. Nevertheless, more recent findings showing that baseline gamma might moderate the relationship between responders and non-responders suggests that future studies are needed to better characterize baseline gamma in depression. While this review focused on gamma in particular, additional work is also needed to better characterize ketamine’s effects on other frequencies and on cross-frequency coupling, examining how these measures correlate with antidepressant response.

As described above, studies of LTP similarly suggest that it is a strong marker of neural plasticity, but limited work has focused on how it is altered following ketamine administration. Future studies are needed to better characterize the role of LTP in depression and to measure how it changes following ketamine administration. In this context, novel computational methods are particularly promising tools to better characterize brain networks and connectivity metrics in mood disorders, in addition to providing stronger mechanistic accounts of how these are altered post-ketamine administration. Additional research is needed to better assess modeled parameter changes associated with direct antidepressant response to ketamine. Such knowledge could ultimately help develop next-generation antidepressants by providing a more complete, mechanistic understanding of ketamine’s antidepressant effects.

Highlights.

• Electrophysiological studies of ketamine’s effects are reviewed

• Review includes ketamine response in depression and healthy subjects

• Markers of synaptic potentiation reviewed include gamma and long-term potentiation

• Dynamic causal modeling of ketamine response is discussed