Ketamine and rapid acting antidepressants: Are we ready to cure, rather than treat depression?

Abstract

Depression is a leading cause of disability, with often chronic course of illness and high treatment resistance in a large proportion of patients. In the current short perspective paper, we present evidence supporting the presence of synaptic-based chronic stress pathology (CSP) in depression and across a number of psychiatric disorders. We summarize the synaptic connectivity model of CSP, and briefly review related preclinical and clinical evidence, while providing appropriate references for more comprehensive reviews and alternative models. We then underscore some gaps in the literature and provide various tips for future directions.

Introduction: Chronic Stress Pathology Appears to be a Common Pathway Across Several Disorders

Over the past few decades, it has become increasing evident that depression may share common pathways with several other psychiatric disorders. A key evidence is that slow acting antidepressants (SAADs), such as monoaminergic drugs, have shown some efficacy in treating major depressive disorder (MDD), bipolar depression, posttraumatic stress disorder (PTSD), generalized anxiety disorder (GAD), obsessive compulsive disorder (OCD), and several other pain and stress-related disorders [1–3]. This shared response among several disorders appears to also extend to rapid acting antidepressants (RAADs), with initial evidence supporting the efficacy of ketamine in depression and other stress-related disorders [4–6]. Importantly, both SAADs and RAADs are thought to induce therapeutic response by increasing neurotrophic factors and synaptogenesis, changes that are believed to reverse an underlying pathological synaptic loss (i.e., reduced density and strength) [7, 8]. Therefore, further supporting the presence of a common pathology is the fact that biological markers that are thought to reflect synaptic loss are also common across MDD, PTSD, GAD, and other stress-related disorders (e.g., prefrontal cortical thinning or hippocampal volume reduction) [9–18]. These disorders also share common biopsychosocial predisposing factors, from genetic variants to early life stress to trauma history. Additional supporting evidence of common pathology is the high co-morbidity and symptom overlap across these stress-related disorders [19].

These various lines of evidence are believed to be directly related to chronic stress pathology (CSP), a common feature across these neuropsychiatric disorders. Chronic stress could be a cause (e.g., traumatic event leading to MDD and/or PTSD), an outcome (e.g., prolonged suffering from GAD, OCD, or any of these stress-related disorders), or both (e.g., vicious cycle of trauma-PTSD) [20]. In this model, at least a subgroup of patients suffers from CSP, as manifested by regional synaptic dysconnectivity (i.e., loss and/or gain in density/strength) [8, 20].

The CSP model presents synaptic disruption in depression, as a converging pathway for various systems; including genetic/epigenetic predispositions, increased inflammation, HPA axis alterations, monoaminergic deficits and glutamatergic excitotoxicity [21]. These body systems were repeatedly implicated in the pathology and treatment of depression [8, 20, 21]. In our working model, abnormalities in these various systems converge on disrupting synaptic connectivity, which in turn affect brain circuitry and networks leading to symptoms of depression and other stress-related disorders [7, 8, 22]. The interaction between an individual biopsychosocial predisposition and their disease-inducing insults (e.g., inflammation, traumatic stress, etc.) results in differing patterns of synaptic dysconnectivity, which in turn translates in differing alterations in brain network and constellation of symptoms. For example, combat exposure, as an insult, may lead to PTSD, depression, or full recovery. In our model, the clinical presentation is the outcome of the specific brain network disruption [20, 22], which is in turn the result of the interaction between predisposition and disease-inducing insults.

Preclinical Evidence: Synaptic Connectivity is a Common Target for Treating Chronic Stress Pathology

Acute stress response appears to be beneficial to the brain and promotes resilience [23]. In contrast, chronic stress response have been repeatedly shown in animals to be detrimental to the brain and to lead to behavioral disturbances consistent with various symptoms observed in stress-related disorders [24, 25]. Well-replicated evidence demonstrated chronic stress-induced synaptic loss (density/strength) in the prefrontal cortex and hippocampus, as well as synaptic gain (density/strength) in the nucleus accumbens and basolateral amygdala [24, 26].

The synaptic loss is associated with amino-acid based pathology (ABP). ABP is thought to be triggered by stress-induced disturbance in amino-acid neurotransmission, consistent with increased extracellular glutamate leading to neuronal excitotoxicity. In animal studies, ABP was related to inefficient glutamate uptake by astrocytes, low glutamate and GABA (γ-aminobutyric acid) neurotransmission, inactive brain derived neurotrophic factor (BDNF), dendritic shrinkage, and reduced synaptic density and strength. These ABP changes were also related to increased inflammation and impaired hypothalamic-pituitary-adrenal (HPA) axis [8, 20].

The synaptic gain is associated with monoamine-based pathology (MBP). MBP is believed to be the result of stress-induced monoaminergic transmission disturbance, with evidence showing increase dopamine as initial step in a cascade of events including co-release of BDNF leading to increased synaptogenesis in the nucleus accumbens. The magnitude and type of stressor appear to affect the nature of the monoaminergic disturbance. Additionally, only a subgroup of susceptible animals develops MBP following stress (reviewed in [8, 20]).

While individuals may have both ABP and MBP, extrapolating from the animal models of stress, it is predicted that a subpopulation of patients suffers from more prominent MBP [27, 28]. It is also believed that those who initially experience primarily MBP and synaptic gain may also later develop ABP due to the vicious cycle of untreated mental illness leading to chronic stress-induced synaptic loss, even in the amygdala and nucleus accumbens [20].

Clinical Evidence: Gray Matter Integrity and Functional Connectivity are Putative Markers of Synaptic Connectivity

Gray matter volumetric alterations in MDD and other stress-related disorders have long been interpreted as in vivo indicators of underlying synaptic loss or gain (i.e., density and strength) [29]. This hypothesis is supported by preclinical evidence directly relating synaptic density to regional brain volumes as measured by MRI [30]. More recently, accumulating evidence supported the presence of comparable association between functional global connectivity and each of synaptic density [31] and synaptic neurotransmission [32].

In stress-related disorders , there is well replicated evidence of reduced hippocampal volume and prefrontal cortical thickness [33, 34] putatively in the ABP subgroup, as well as pilot evidence of increased volume in the nucleus accumbens or amygdala [27, 35], presumably in the MBP subgroup. Notably, antidepressants appear to reverse this stress related gray matter alterations [27, 36, 37]. Similarly, convergent evidence supports the presence of altered functional global brain connectivity in MDD [32, 38–43] and in several psychiatric disorders with a considerable chronic stress component [11, 12, 39, 44–46]. Antidepressants, particularly ketamine, were shown to reverse this functional dysconnectivity [32, 36, 38, 47]. Together, these data highlight the potential utility of gray matter morphometry and functional connectivity as putative biomarkers of CSP. However, it is important to underscore that to-date these biomarkers were evident is some but not all studies, presumably due to methodological variabilities as well as ABP vs. MBP heterogeneity. Moreover, the individual effect size of these biomarkers is low, precluding their utility in clinical settings as univariate markers. It remains to be seen in future studies whether a biologically based stratification along with multivariate approaches would address the inconsistency and low effect size limitations.

Literature Gaps: Is it Time to Focus on Curing, rather than Treating Chronic Stress Pathology?

Are we ready to cure rather than treat depression? Unfortunately, the answer is we are not there yet and much work is still needed. Ketamine and other RAADs were found to normalize the stress-induced synaptic dysconnectivity in animals, showing both prefrontal synaptic gain and nucleus accumbens synaptic loss within 24h of treatment [48, 49]. Yet, these synaptic normalization effects are thought to be transient and relapse within 10 days [50]. Similarly, early human studies showed rapid normalization of functional connectivity, as well increased hippocampal but reduced nucleus accumbens volumes within 24h of successful ketamine treatment [27, 32, 38, 47]. These neurobiological changes are accompanied by RAAD effects. Unfortunately, patients also often relapse within 1–2 weeks of single treatment [51].

A major gap of preclinical literature is the lack of focus on the molecular basis of post-ketamine relapse. Understanding the biological relapse mechanisms may unravel new approaches that ultimately could help curing, rather than simply treating depression. Current clinical standards are to administer ketamine repeatedly to maintain response, e.g., twice per week in the induction phase. Neurobiologically, the relapse in symptoms is thought to be due to reversal of ketamine-induced synaptic normalization, although confirmatory data are scarce. Investigating the reversibility of treatment-related synaptic changes is essential to truly understand the etiological mechanisms triggering and sustaining depression. Unfortunately, the neurobiological literature has been largely limited to the acute effects of ketamine within 24h. Few state-of-the-art studies examined the effects of ketamine up to 2 weeks [52, 53]. However, these studies discontinued the “insults” (cortisone or stress) following ketamine treatment. Here, it is important to note that there is extensive evidence supporting the reversibility of stress-induced synaptic loss in these otherwise normal animals, often within weeks of withholding the insults [23, 24]. Therefore, to model human depression and the rapid relapse after single ketamine infusion, future studies should maintain the insult (e.g., stress or cortisol) and investigate the processes involved in the synaptogenic relapse. Equally important, studies investigating the mechanisms underlying the reversibility of stress-induced synaptic loss may also provide clues to the etiology of depression and perhaps assist in providing a cure.

Another related gap is that many of the elegant preclinical studies have conducted the bulk [54, 55] or all [56, 57] of their behavioral assessment of the antidepressant-like effects within few hours of ketamine administration, a period during which the animal may still be under the influence of ketamine and its metabolites. More importantly, these acute effects do not model the clinically meaningful antidepressant effects of ketamine, which are sustained for 1–7 days. The limitations of the “depression under influence” data are twofold, 1) the biological results may include non-specific ketamine effects that are not relevant to its RAAD effects in humans, and 2) the examined behavioral outcomes may be affected by the intoxication, rather than the sustained antidepressant properties. In our view, the clinical relevance of any antidepressant effects during intoxication is highly questionable. Therefore, we believe future preclinical studies would benefit from collecting data at least 24h post administration, when ketamine has been fully metabolized.

The above-mentioned gaps are not just limited to preclinical studies, human research is also lagging in conducting interventional mechanistic studies. To date, the majority of human biological studies are observational in nature, where biomarkers are acquired before, during, or after treatment. These studies are more informative than cross-sectional investigations, but the evidence remains correlational in nature. To provide causal evidence, the field should aim for interventional mechanistic studies, particularly those that are focusing on translating the preclinical-based models to humans, as well those that attempt to unravel the mechanisms of relapse. Comparable studies have been conducted in healthy volunteers, e.g., pretreatment with the glutamate inhibitor lamotrigine to ascertain the causal relationship between glutamate neurotransmission and the ketamine-induced global brain connectivity [32]. The value of this type of studies is highlighted by the fact that the RAAD effects of ketamine were initially demonstrated through an interventional mechanistic study [58]. More recently, another interventional mechanistic study in depressed patients discovered that pretreatment with the immunosuppressant rapamycin significantly prolongs the response to ketamine, while demonstrating the inability of systemic inhibition of the mechanistic target of rapamycin (mTOR) to block the acute RAAD effects [51]. Compared to placebo pretreatment, one oral dose of rapamycin tripled the response rates and quadrupled remission rates at 2 weeks post single ketamine infusion [51]. Both studies may be considered high-risk high-grain or difficult to fund and/or conduct. They are also pilot in nature and more suitable for exploring neurobiological models, than for confirming clinical efficacy of a treatment; therefore, replication is essential. Yet, these studies underscore the need for investigating preclinical models in humans and highlight the unexpected findings that resulted from careful assessment of brain mechanisms. The ketamine mechanistic discovery study [58] opened the door for a number of confirmatory clinical studies [59] and a wealth of preclinical back translation opportunities to gain new insight into the biology and treatment of depression [60, 61]. It is hoped that the rapamycin results [51] would successfully replicate in humans and that their back translation to animal studies will help identifying the etiological mechanisms of relapse and, ultimately provides new approaches for curing depression. Similarly, a recent mechanistic human study demonstrated a previously unknown role for opioid modulation for successful ketamine treatment [62]. The field would benefit from more clinical mechanistic studies, particularly if those were accompanied with biological assessment of target engagement.

Conclusion: The Brain Works in Mysterious Ways

Tremendous progress has been achieved in terms of gaining new knowledge and having better understanding of the putative mechanisms of CSP, and subsequently depression and other stress-related disorders. Yet, a large part of this knowledge remains tilted toward preclinical evidence with sparse but reproducible suggestive human findings.

Biological signatures of depression have been reported in human studies from neuroimaging biomarkers to omics evidence. However, the reproducibility, the specificity, the effect size, and the ubiquity of these findings across all depressed patients have often been challenging.

In general, the biological revolution in psychiatric research has not yet fully translated into clinical biomarkers of diagnosis, of disease progress, or even of treatment targets. The main clinically relevant success was in terms of biological treatments. Yet, the biological treatments currently used in clinics are mostly the product of serendipity followed by development of new drugs with comparable molecular mechanism of action. Unfortunately, this approach has worked for SAAD, but failed repeatedly with RAAD (e.g., lanicemine and rapastinel). To date, successful rational drug development pipelines remains the exception (e.g. brexanolone), rather than the rule (e.g., mGluRs).

Taking notes of the progress made and potential shortcomings of our approach to clinical neuroscience, below are few suggestions that may expedite our journey towards ultimately curing depression and other mental illnesses:

1. We should underscore the urgent need for clinically useful products, particularly biomarkers of treatment targets and disease progress. The literature is full of storytelling papers, yet the findings are not always easy to replicate and if they do, their effect size is often too small to be clinically meaningful. For example, the association between depression and hippocampal volume is reproducible, but it is neither robust (i.e., not consistently evident in small samples) nor specific (i.e., shared with most psychiatric disorder) [63]. The clinical neuroscience field has generated incomprehensible number of findings, it is time to attend to integrating them into knowledge and to prioritize producing clinical products over generating more stories (e.g., focusing on the robustness and reproducibility of identified biomarkers).

2. We should expand the research portfolio beyond cross sectional investigations in small cohorts (e.g., n < 100 subjects per group). After seven decades of biological research, it is becoming increasingly evident that a single genetic (e.g., serotonin transporter polymorphism), anatomical (e.g., hippocampal volume), functional (e.g., amygdala activity or connectivity), or molecular pathology (e.g., low serotonin or high glutamate) is unlikely to fully or even meaningfully account for the neurobiology of clinical depression; or for that matter any other neuropsychiatric disorder. We should aim to concurrently target multiple markers and embrace multivariate pattern analyses along with predictive models and network approaches, while reducing our reliance on univariate interpretive statistics. Machine learning approaches combined with large data collection have made substantial progress in other fields. It is our hope that clinically informed features engineering and integration of these new tools would lead to new biomarkers and novel treatments (e.g., [36]).

3. We should encourage longitudinal studies – regardless of sample size, particularly interventional mechanistic investigations that may provide causal evidence rather than simple correlations. The field of clinical neuroscience research is full of associative evidence, it is now time to conduct the high-risk high-gain human studies. Basic studies are indispensable. However, the generated preclinical models are considered putative and should be demonstrated in humans. Inference across species is at best speculative and the literature is full of examples where these models do not hold true in humans; e.g., recent failure of rapastinel in human studies, despite achieving all preclinical targets both biologically and behaviorally [64]. Human mechanistic studies, accompanied with assessment of target engagement, may be difficult to do, but they are necessary if we are to make two-ways translational progress.

4. While not all unexpected findings will replicate, history has taught us the value and impact of careful assessment of evidence and readiness to rigorously pursue new directions. From penicillin to tricyclic antidepressants, progress in medicine, especially in psychiatry, has been full of examples where serendipity combined with astute clinical research have led to novel treatments and greater understanding of underlying pathology.

5. Call it as it is. Preclinically, if chronic stress produces synaptic loss, then let’s not label it depression or PTSD. Clinically, if hippocampal volume reduction is neither sensitive (approximately 1% less than control), nor specific (evident in MDD and PTSD, among other disorders), then let’s not call it a biomarker of MDD or PTSD [63]. These behavior-first approaches, where the biological correlates are investigated, have not yielded much biological research progress. Perhaps it is time to aim for a biology-first approach, where the behavioral correlates of a biological disturbance are investigated. Psychiatric diagnoses served an important clinical purpose. Rather than inventing an alternative behavior-first approach, such as RDoC, it may be time to create from the ground up a novel biology-first approach designed specifically for clinical neuroscience.

Highlights.

• Chronic stress pathology (CSP) is characterized by patterns of synaptic dysconnectivity

• CSP dysconnectivity is reversible within a month of discontinuing stress in rodents

• The effects of ketamine on CSP dysconnectivity are also short lived

• The field should focus on the reversibility of these synaptic changes

• Mechanisms of CSP reversibility may unravel new approaches toward curing depression

Uncommon Abbreviations:

ABP

amino acid-based pathology

CSP

chronic stress pathology

MBP

monoamine-based pathology

RAAD

rapid acting antidepressant

SAAD

slow acting antidepressant