Personalized use of ketamine and esketamine for treatment-resistant depression

Abstract

A large and disproportionate portion of the burden associated with major depressive disorder (MDD) is due to treatment-resistant depression (TRD). Intravenous (R,S)-ketamine (ketamine) and intranasal (S)-ketamine (esketamine) are rapid-acting antidepressants that can effectively treat TRD. However, there is variability in response to ketamine/esketamine, and a personalized approach to their use will increase success rates in the treatment of TRD. There is a growing literature on the precision use of ketamine in TRD, and the body of evidence on esketamine is still relatively small. The identification of reliable predictors of response to ketamine/esketamine that are easily translatable to clinical practice is urgently needed. Potential clinical predictors of a robust response to ketamine include a pre-treatment positive family history of alcohol use disorder and a pre-treatment positive history of clinically significant childhood trauma. Pre-treatment versus post-treatment increases in gamma power in frontoparietal brain regions, observed in electroencephalogram (EEG) studies, is a promising brain-based biomarker of response to ketamine, given its time of onset and general applicability. Blood-based biomarkers have shown limited usefulness, with small-effect increases in brain-derived neurotrophic factor (BDNF) being the most consistent indicator of ketamine response. The severity of treatment-emergent dissociative symptoms is typically not associated with a response either to ketamine or esketamine. Future studies should ensure that biomarkers and clinical variables are obtained in a similar manner across studies to allow appropriate comparison across trials and to reduce the signal-to-noise ratio. Most predictors of response to ketamine/esketamine have modest effect sizes; therefore, the use of multivariate predictive models will be needed.

Introduction

Treatment-resistant depression

Globally, major depressive disorder (MDD) is a leading cause of disability affecting ~300 million people worldwide [1, 2]. In the United States alone, the annual cost of MDD is estimated at $ 326.2 billion [3]. A large and disproportionate part of the burden associated with MDD is due to treatment-resistant depression (TRD), which is typically defined as an individual with depression who has not responded therapeutically to at least two trials of antidepressants of appropriate dose and duration [4, 5]. Using this definition, about 31% of patients with MDD meet criteria for TRD [6], and the total direct and indirect costs associated with a TRD case are between 53 and 91% higher than the cost of a non-TRD case [6, 7].

TRD, sometimes used interchangeably with ‘difficult-to-treat depression’, poses significant challenges to its management [8, 9]. When compared to non-TRD, TRD is associated with higher rates of hospitalization, emergency room visits, suicide attempts and completion, depression relapse, and other negative outcomes [4, 6, 7, 9, 10]. In addition, TRD is linked to the use of a greater number of current medications and higher prevalence of co-occurring psychiatric and medical comorbidities (which increases the risks of side effects and drug-drug interactions), and more challenging social situations (including greater unemployment, lower productivity, poorer quality of life [4, 6, 7, 9, 10]). Greater efficacy in managing TRD will lead to substantial individual and societal benefits.

Ketamine and esketamine

In the 2000s, randomized placebo-controlled trials in patients with major depressive symptoms reliably observed the rapid-antidepressant effects of intravenous (IV) administration of subanesthetic doses of (R,S)-ketamine (ketamine) [11, 12], which became the first and prototypical representative of a drug resulting in rapid-acting antidepressant actions. The first clinical trial reporting on antidepressants effects of ketamine was a placebo-controlled published in 2000 by Berman et al. (2000). The study had seven completers (six diagnosed with MDD and one with bipolar depression) [11]. Although all participants had a history of recurrent depressive episodes, they did not meet criteria for TRD. The second clinical trial was published by Zarate et al., (2006) and included 18 participants (all with TRD) in a placebo controlled double-blind crossover study [12]. In these two first studies, there was a statistical difference in antidepressant effects between ketamine and placebo, which emerged within 2–4 h after the start of the single 40 min infusion, persisted for at least 3–7 days, and encouraged further and larger clinical trials.

Broadly, rapid-acting antidepressants are expected to exert antidepressant effects within hours to days, in contrast to conventional antidepressants that typically require weeks to produce clinically significant antidepressant effects [13, 14]. The faster onset of therapeutic action of rapid-acting antidepressants has multiple benefits including faster recognition of whether the antidepressant treatment will be helpful or not, shortening of the personal suffering and of impaired productivity associated with depression, and reduction of the risk of suicide [15].

The observation that IV ketamine had rapid-acting antidepressant effects incentivized further research into other routes of ketamine administration, the mechanisms underlying its antidepressant effects and the development of new rapid-acting antidepressants [13, 14]. Racemic ketamine is composed of S (+) and R (−) ketamine enantiomers, which are mirroring pairs of compounds with the same connectivity and molecular formula but in a mirror image arrangement [16]. It was later observed that intranasal (IN) (S)-ketamine (esketamine), the S (+) enantiomer of ketamine, also had clinically significant antidepressant effects in TRD, when used in conjunction with oral antidepressants [17, 18]. IN esketamine was approved for TRD by the Food and Drug Administration in 2019. However; preliminary evidence indicates that IN esketamine, when compared to IV ketamine, may require a greater number of treatments and longer time to reach response and remission [19], and may have lower antidepressant effects, based on self-report measures [20]. Despite this preliminary evidence, IN esketamine is considered a rapid-acting antidepressant given its faster onset of action when compared to conventional monoaminergic-acting antidepressants.

In unipolar depression, ketamine is most frequently used concurrently with oral antidepressants; however, ketamine may also be utilized as monotherapy (i.e., without a concurrent antidepressant) [21]. Preliminary evidence has found that the efficacy is similar whether a concurrent antidepressant is used or not [21]. On the other hand, IN esketamine is virtually only utilized in conjunction with an oral antidepressant [22].

There have been preliminary positive studies using other routes of administration of ketamine and esketamine including IN ketamine [23], IV esketamine [24, 25], intramuscular (IM) ketamine [26], subcutaneous ketamine [26–28] and esketamine [29], and oral esketamine and ketamine [30]. However, this review focusses on the personalized use of IV ketamine and IN esketamine, which are the most studied and clinically available forms. Table 1 provides an overview of the treatment with IV ketamine and IN esketamine.

Table 1.

Comparisons of treatment with intravenous (IV) (R,S)-ketamine (ketamine) and intranasal (IN) (S)-ketamine (esketamine).

	IV Ketamine	IN Esketamine
History	Introduced in the market in 1970 (in the United States) Initially used as an anesthetic First trials using IV ketamine for depression were published in the 2000s	Introduced in the market in 1997 (in Germany) Initially used as an anesthetic First trials using IN esketamine for depression were published in the late 2010s Approved by FDA as an adjunct treatment for TRD in 2019, and as an adjunct treatment for MDD with acute suicidal ideation or behavior in 2020
Dose	0.5 mg/kga	28 mg, 56 mg or 84 mg given in spray devices of 28 mg
Duration	The infusion lasts 40 minutes. The total treatment (including pre- and post-treatment assessments and monitoring) typically last about 90 to120 minutes	28 mg are administered 5 min apart until the desired dose is reached. The total treatment (including pre- and post-treatment assessments and monitoring) typically last about 150 to180 minutes
Treatment schedule	Divided in induction phase and maintenance phase. Induction phase typically consists of six treatments given twice weekly over three weeksa	Divided in induction phase and maintenance phase. Induction phase typically consists of eight treatments given twice weekly over four weeks
Common acute side effectsb [159, 160]	Dizziness = 33% Dissociation = 28% Increased blood pressure = 28% Nausea = 20% Sedation = 18% Anxiety = 15%	Dizziness = 29% Dissociation = 41% Increased blood pressure = 10% Nausea = 28% Sedation = 23% Anxiety = 13%
Potential advantages	Greater and more reliable bioavailability (IV delivery) IV ketamine includes (R)-ketamine (arketamine), which has shown independent antidepressant-like properties in preclinical tests [16, 124, 161] Preliminary evidence that IV ketamine may need a lower number of treatments and shorter time to reach response and remission [19] and may have greater antidepressant effects, based on self-report measures [20] More customized dose, which is based on weight	FDA-approved, which facilitates insurance coverage IN administration is less invasive and more convenient than IV administration Clinical availability is increasing

Abbreviations: FDA United States Food and Drug Administration.

^aWe report the most typical dose and treatment schedule for IV ketamine during induction phase. However, in real community settings, the dose, frequency and number of treatments vary considerably.

^bOnly randomized controlled trials were included. There is a need for head-to-head comparisons.

Personalized medicine and TRD

Personalized medicine, often used interchangeably with the term ‘precision medicine’, is the practice of medicine that uses unique individual characteristics (e.g., clinical/environmental variables, biomarkers) to guide customized prevention and/or treatment of medical conditions [31–33]. There has been an increasing interest in personalized medicine given its potential to deliver better outcomes when compared to conventional medicine, which typically focuses on larger groups of individuals with the same diagnosis [31–33]. In conventional medicine, large clinical trials have been considered the gold standard, but these studies have mostly focused on average response with little exploration of the heterogeneity in patients and magnitude of response [34]. Personalized medicine, on the other hand, is based on the premise that there is substantial heterogeneity within a specific diagnosis and that the systematic study of a large group of individuals allows the identification and in-depth study of smaller subsets of patients with more homogenous characteristics such as similar response to a specific treatment [34].

A personalized medicine approach fits well with the treatment of TRD. First, TRD likely consists of a heterogeneous group of patients. Second, there is a reasonable armamentarium to treat TRD, and therapeutic interventions involving different mechanisms of action are currently available. However, there are many challenges in the development of personalized medicine including the need to systematically study clinical and biological variables in large samples and replicability difficulties. Oncology and hematology are examples of medical field areas that have implemented personalized medicine in clinical care in their use of targeted and immunotherapy approaches [35, 36]. However, in most medical specialties, including psychiatry, the systematic application of personalized medicine in clinical practice is still relatively limited. In this review, we discuss the growing literature on the personalized use of ketamine and esketamine in TRD with a greater focus on the evidence from studies with higher level of evidence according to the Evidence-Based Medicine Pyramid, particularly systematic reviews and meta-analyses. The main target audience of this review are mental health providers and clinicians. Initially, we will focus on the discussion of clinical markers of response to ketamine/esketamine as clinical variables are the main modalities and sometimes the only ones collected by clinicians in real life treatment settings. Then, we will summarize the biological markers of response, facilitating the understanding of the complex and broad body of research in personalized use of ketamine/esketamine. We will also briefly discuss neurocognitive markers of response, a less studied area in the field of response to ketamine/esketamine.

Markers of treatment response

To appropriately personalize the treatment of TRD, it is important to understand the concept of marker (sometimes referred to as correlate) of treatment response, which is a clinical or biological variable associated/correlated with probability and/or magnitude of therapeutic benefit with a specific treatment. In short, the marker is the independent variable (the feature) while the treatment response is the dependent variable (the outcome) [37, 38]. Useful markers of treatment need to have external validity, i.e., to have generalizability/replicability to other samples with similar characteristics [33, 37, 38]. Predictors on the other hand are variables that forecast outcomes a priori in longitudinal and hypothesis-driven studies, which have been virtually inexistent in ketamine/esketamine research. Therefore, this review will use the term “marker of response”.

Clinical variables versus biomarkers

The two most investigated categories of markers of treatment response are clinical variables and biomarkers. Clinical variables refer to characteristics/attributes that have a range of possible values and are easily obtained in clinical practice. They include, for example, sociodemographic variables (e.g., age, biological sex, race/ethnicity, income, education), depression-related variables (e.g., severity of depression, duration of current depressive episode, depression clinical subtype), psychiatric/medical comorbidities and family history. Clinical variables typically can be acquired with limited resources. On the other hand, their accuracy may be affected, in variable degrees, by clinicians’ and patients’ subjective factors and biases such as memory, perceived distress, social acceptability, and education/language barriers [39–42]. In addition, clinical variables do not necessarily represent underlying biological processes.

A biomarker is a biological measure that can be objectively and quantifiably measured and ideally a measure of relevant underlying biological processes for a disease and/or treatment [41, 43, 44]. Despite potential challenges with measurement properties of biological variables [45], biomarkers are broadly more accurately and reliably quantified than most clinical variables, facilitating the replicability of findings in independent samples [46, 47]. On the other hand, biomarkers require resources and may not necessarily be biologically linked to what patients are experiencing or feeling, which is usually the reason the individual is seeking help [43, 48]. Biomarkers can be further subdivided into brain-based biomarkers (directly measure brain structure and/or function) and peripheral biomarkers (collected from peripheral sites such as blood, urine, and saliva) [49–51]. Certain biomarkers may also be heritable endophenotypes, which are intermediate measures between genetic risk and the clinical phenotype [52, 53].

Table 2 provides an overview of the advantages and disadvantages of the use of clinical variables and biomarkers (brain-based and peripheral) in psychiatry.

Table 2.

Potential advantages and disadvantages of the use of clinical variables and biomarkers to predict response to treatments.

	Clinical Variables	Biomarkers
	Require less financial and technological resources to be obtained	More accurately and reliably quantified than most clinical variables
Potentialadvantages	Represent the patients’ feelings and experiences	May represent relevant underlying neurobiological processes
	Psychiatry diagnoses and study end-points currently use clinical symptoms/measures	The discovery of predictive biomarkers may facilitate the development of new and more effective medications
	Accuracy may be affected by subjective factors and biases such as memory, perceived distress, and social acceptability	Require more financial and technological resources
Potential disadvantages	Can be difficult to obtain accurate information from individuals with cognitive problems or education/language barriers	Some biomarkers might be hard to translate to clinical practice due to cost and accessibility
	Clinical variables do not necessarily represent underlying biological processes	Potentially affected by factors such as fasting status, time of the day, concurrent medications, medical comorbidities, and sample processing

Pre-treatment (baseline) versus post-treatment (longitudinal) markers

Temporally, there are markers of treatment response: pre-treatment (baseline) and post-treatment (longitudinal) [41]. Pre-treatment markers are obtained before the treatment is given to the individuals and indicate those persons who are more or less likely to respond to the future intervention.

Post-treatment markers are those measured after the start of the treatment, are examined longitudinally (i.e., pre- vs post-treatment), and linked to dynamic changes that correlate with a higher likelihood of detecting changes after the initiation of treatment [41].

Post-treatment can be further divided in early post-treatment (peri-treatment) and typical post-treatment. Early post-treatment markers are the symptoms that frequently occur during the treatment and include symptoms such as treatment-emergent dissociation or increases in blood pressure. Early post-treatment markers are pharmacologically correlated with the blood concentration ketamine. On the other hand, typical post-treatment markers happen after the treatment is concluded and are likely related to downstream biological changes that caused by ketamine/esketamine.

Non-specific markers of poor treatment response

Evidence suggests that some markers are linked to poorer outcomes in a wide variety of currently available antidepressant treatments including conventional oral antidepressants, psychotherapeutic interventions, different neuromodulatory treatments, and rapid-acting antidepressants. These non-specific markers of poor treatment response include pre-treatment clinical variables such as limited/dysfunctional social support [51, 54], unemployment [55], lower educational level [51], longer duration of current episode [55–58], greater number of failed antidepressant treatments [55, 59, 60], presence of co-occurring psychiatric disorders including anxiety disorders, substance use disorder, and persistent depressive disorder [51, 54, 55, 57] and lack of early symptomatic or functional improvement after the treatment [55, 61]. Pre-treatment biomarkers such as smaller volume of the hippocampus [51, 62], hypoactivity of the anterior cingulate cortex [51, 62] and greater white matter damage in different brain regions [63–67] have also been consistently linked to poorer response to different treatment modalities.

It is impossible to know with certainty whether these clinical and biomarker variables are going to be also markers of poorer response to future antidepressant treatments; however, their association with worse response to currently available antidepressant treatments has been consistently described. Hopefully, new antidepressants treatments will be able to address the unmet scientific need of better managing individuals with markers of poor response to currently available therapies.

Non-specific markers of poor response may indicate the need for more intensive treatments and/or the combination of interventions; however, they have limited usefulness informing differential response to treatments (i.e., treatment A versus treatment B).

Personalized use of ketamine in TRD

The first studies on specific markers of antidepressant response to ketamine were published in the late 2000s [68–70]. Since then, a substantial body of research has been produced, and the sample sizes of the studies have also increased [41, 71]. Some of the most studied clinical variables include pre-treatment body mass index [23, 72–77], pre-treatment anxiety symptoms/features [78–85], and post-treatment dissociative symptoms [23, 72–74, 76, 77, 86] (further discussed below). The most investigated peripheral and brain-based biomarkers have been, respectively, post-treatment changes in blood levels of brain-derived neurotrophic factor (BDNF) [41] and post-treatment changes in functional connectivity (FC) [71]. Here we will summarize clinical, biomarker and neurocognitive markers of response to ketamine with a particular focus on variables that have been replicated in independent samples.

Clinical markers

Pre-treatment

Positive family history of alcohol use disorder (AUD) in first-degree relatives has been the most consistently replicated pre-treatment markers of response to ketamine. The first study to observe that a positive family history of AUD was associated with a greater response to ketamine was conducted by Phelps et al., (2009) who examined data on 23 participants with TRD treated with a single open-label infusion. The authors observed that individuals with a positive family history of alcohol dependence (n = 12) had a more robust response to ketamine between 80 and 230 minutes post-ketamine than those without the family history of alcohol dependence (n = 11) [68]. A later study by the same group combined individual-level analysis of additional four clinical trials and also found that a positive family history of AUD was linked to a significantly better response 1 day (β = −0.37, p = 0.001, n = 82) and 7 days (β = −0.41, p < 0.001, n = 71), but only a trend effect at an earlier time point of 230 minutes (β = −0.17, p = 0.080, n = 108), after a single ketamine infusion [87]. In addition, two studies conducted by other groups investigated the impact of positive family history of AUD on response to ketamine [78, 88]. Both treated patients with open-label infusions in naturalistic designs (i.e., real clinical settings). Permoda-Osip et al. (2014) studied 42 patients with bipolar depression who received a single ketamine infusion while Thomas et al. (2018) provided six ketamine infusions to 50 individuals with unipolar or bipolar depression who were resistant to ECT and/or pharmacological interventions [78]. The two studies found that individuals with a positive family history had numerically higher response rates than those with a negative family history of AUD, but statistical significance was reached only in the Permoda-Osip et al. study [88].

There have been studies linking other pre-treatment clinical variables with a more robust response to ketamine including higher body mass index [87, 89], positive history of clinically significant childhood trauma [90, 91], positive personal history of AUD [88], greater anhedonia scores [78], presence of anxious features [81], and absence of melancholic features [85]. However, there have been mixed results with negative studies for each of these variables: body mass index [78, 82, 83, 92–96], childhood trauma [68, 87], personal history of AUD [12, 78, 87, 93], anhedonia scores [84], anxious features [78–80], and non-melancholic depression [97, 98].

Despite one study showing that IV ketamine had greater antisuicidal effects (i.e., not antidepressant effects) in individuals with bipolar depression than those with unipolar depression [99], several studies did not find significant differences in antidepressant effects between participants with bipolar and unipolar depression [78, 82, 96, 100]. A standardized meta-analytical comparison of antidepressant response rates 1 day after a single infusion of IV ketamine found similar rates for unipolar (45%) and bipolar (48%) depression (odds ratio [95% CI] = 0.90 (0.38, 2.16), p = 0.83) [41]. One study that examined individuals with bipolar depression further compared participants with bipolar 1 disorder and bipolar 2 disorder and failed to find statistically significant differences in response (response rates of, respectively, 37% and 31%, p = 0.63) [101]. Non-specific pre-treatment clinical markers of poorer response have also been described in studies with ketamine including a history of previous psychiatric hospitalization [82, 83] or previous suicide attempt [87], and worse socioeconomic status [82].

With respect to the published work examining the impact of biological sex on antidepressant response to ketamine, Rybakowski et al. (2017) reported that men had a better antidepressant response to ketamine than women 1 day post-treatment (response rates: men = 77% (n = 10/13); women = 42.5 (n = 17/40); p = 0.031). However, the vast majority of studies did not find differences in response between men and women [12, 64, 79, 82, 83, 88, 100, 102–105].

There is evidence suggesting that some concurrent medications may affect ketamine’s antidepressant efficacy. However, broadly these data have been mixed. Five studies (with ten or more participants) studied the impact of benzodiazepines on response to ketamine [78, 79, 82, 106, 107]. Three studies observed that benzodiazepines negatively affected ketamine’s antidepressant effects either by reducing the overall efficacy [79, 106], or by delaying the time to reach response/remission, and/or shortening the time to depression relapse [107]. However, two investigations failed to observe a negative impact of concurrent benzodiazepines [78, 82]. One possible explanation for the mixed data is that higher doses of benzodiazepines are needed to exert interactive effects with ketamine’s antidepressant effects, as indicated by Frye et al., (2015) [106] and Andrashko et al. [79].

A small randomized controlled trial (RCT) with 12 completers conducted by Williams et al., 2019 found that premedication with 50 mg of naltrexone (a mu-opioid receptor antagonist) reduced the antidepressant effects of one infusion of ketamine when compared to premedication with placebo [108]. However, two other studies found that modulation of the opioid system did not change ketamine’s antidepressant effects [109, 110]. Several investigations also examined whether a specific antidepressant or combination of antidepressants can increase ketamine’s efficacy; however, data are mixed, and most are negative (i.e., did not find statistically significant differences) indicating that likely most antidepressants are compatible with IV ketamine [12, 79, 82, 83, 105, 111].

Finally, in bipolar depression, ketamine is given to patients in conjunction with mood stabilizer(s) (i.e., not as monotherapy). Some studies investigated whether the mood stabilizer used by the patient affected ketamine’s antidepressants effects, and overall, they failed to find a differences between the different types of mood stabilizers (e.g., lithium, valproate, antipsychotics, combinations of mood stabilizers) [83, 101, 112, 113].

Post-treatment

The most examined post-treatment clinical markers of response to ketamine has been the transitory dissociative (8 studies) [23, 72–74, 76, 77, 86, 114] symptoms after initiation of treatment (8 studies). Dissociative symptoms are commonly reported by individuals treated with ketamine/esketamine and include distortions on perception of the body such as feelings of being disconnected from it (depersonalization), abnormal perception of the environment such as feelings that the world is distorted or not real (derealization), and distortions in the perception of time.

In all studies, dissociative symptoms were measured by the Clinician-Administered Dissociative States Scale (CADSS), which provides a total score and sub-scores in three dimensions (depersonalization, derealization, and amnesia) [86, 115]. Two studies observed a statistical association between greater dissociative symptoms and response to ketamine [76, 77] (with an explained variance in antidepressant response of 12 and 21% [116]). However, most investigations (6 out 8 studies) failed to find statistical associations for dissociative [23, 72–74, 86, 114] symptoms.

Although overall dissociation (i.e., total CADSS scores) is not associated response to ketamine [117], a report by Niciu et al. (2018) indicated that the sub-score derealization was associated with response [86] while Chen et al. (2020) observed a relationship between happiness after infusion (measured by the Visual Analog Scale for Happiness) was related to better response to ketamine [118]. These findings may warrant further consideration of scales/subscales as positive/euphoria feelings after administration of ketamine. There is also mixed evidence indicating that greater increases in blood pressure is be linked to better response to ketamine [76, 119].

Biomarkers

Pre-treatment

Pre-treatment biomarkers, when compared to post-treatment biomarkers, have shown more limited usefulness in understanding response to ketamine [41]. This has been true for both peripheral and brain-based biomarkers.

A systematic review and meta-analysis of blood-based biomarkers combining 56 studies found that baseline levels of blood-based biomarkers, including neurotrophic factors and inflammatory markers were not consistently associated with response to ketamine [41]. Some of the standard mean differences (SMD) (95% CI) for the comparison between responders and non-responders were: BDNF (11 studies): −.04 (− 0.30, .23), p = 0.77; Vascular endothelial growth factor (VEGF, 3 studies): .08 (−0.25, .42) p = 0.62; Interleukin 6 (6 studies): −0.04 (−0.36, .29) p = 0.83; and Tumor necrosis factor α: −0.10 (−0.40, 0.21), p = 0.54 [41]. A study by Permoda-Osip et al. (2013) with 20 participants diagnosed with bipolar depression found that higher pre-treatment vitamin B12 was associated with a better response to a single IV infusion of ketamine 7 days post-treatment [120]. However, a subsequent study that analyzed 49 individuals with unipolar depression and 34 with bipolar depression did not observe a relationship between pre-treatment vitamin B12 and magnitude of response to a single IV infusion of ketamine 230 min, 1 day or 7 days post-treatment in unipolar depression or bipolar depression [121].

In terms of brain-based biomarkers, a systematic review including 69 studies found that the most promising baseline marker of response to ketamine was associated with greater baseline fractional anisotropy in the cingulum (suggesting that individuals with lower white matter damage in the cingulum have more antidepressant benefits with the medication) [71]. However, this finding was observed in only two independent samples treated with ketamine [63, 64] and likely is a non-specific markers of poor response since a less healthy cingulum white matter has also been linked to poor response to oral antidepressants [65–67].

Several other pre-treatment peripheral (e.g., levels of metabolites of the tryptophan-kynurenine pathway, genetic markers, levels of amino acids and derivates) and brain-based biomarkers (e.g., blood flow of brain regions examined with arterial spin labeling, brain levels of molecules/receptors measured by magnetic resonance spectroscopy and/or positron emission tomography) were investigated. However, there were no consistent replication of the association between these biomarkers and response to ketamine [41, 71].

Post-treatment

A systematic review and meta-analysis found that responders, but not non-responders, to ketamine had statistically significant increases in BDNF after treatment. However, the effect size of this association was small (SMD (95% CI) = 0.26 (0.03, 0.48), p = 0.02, 11 studies) and showed substantial variability across samples [41].

Another systematic review observed that the three most consistent brain-based post-treatment markers of response to ketamine were: (1) post-treatment increased gamma power in electrophysiological studies, (2) post-treatment increases in functional connectivity within the prefrontal cortex, and (3) measures suggestive of post-ketamine increases in activation within the striatum using distinct neuroimaging modalities, specifically, task-based magnetic resonance imaging, positron emission tomography and arterial spin labeling [71]. Each of these findings were observed in three independent samples but post-treatment increased gamma power is the most consistent since there were fewer negative studies for this result [71]. This could have broad utility, as post-treatment increased gamma power can be measured by EEG, a widely available and portable modality.

Several other post-treatment peripheral (e.g., levels of ketamine and ketamine metabolites, changes in levels of metabolites of the tryptophan-kynurenine pathway, changes in levels of amino acids and derivates) and brain-based biomarkers (e.g., changes in blood flow of brain regions examined with arterial spin labeling, changes in brain levels of molecules/receptors measured by magnetic resonance spectroscopy and/or positron emission tomography) were investigated. Two studies noted that lower levels of (2 R,6 R)-hydroxynorketamine were associated with greater response to ketamine [122, 123]; however, the sample sizes were relatively small and further replication is needed.

Pre-clinical [124, 125] and clinical studies (particularly when ketamine is given orally) [126] indicate that metabolites such as norketamine and (2S,6S;2 R,6 R)-hydroxynorketamine (HNK) may be involved in the rapid antidepressant effects of ketamine, particularly in early response. Although some studies in humans treated with IV ketamine failed to observe statistically significant relationships between early improvement and levels of norketamine (other metabolites were not consistently investigated) [112, 122, 123, 127–129], the relationship between the levels of these metabolites and early improvement should be further explored in clinical studies.

Neurocognitive markers

A type of marker of response that has been less investigated than clinical markers and biomarkers are neurocognitive markers. To the best of our knowledge, only studies using IV ketamine have investigated neurocognitive markers. Here we examine the available evidence for this modality of marker.

Pre-treatment

Results for pre-treatment neurocognitive markers have had mixed. A study by Murrough et al., (2013) with treatment-resistant unipolar depression found that responders (n = 16) 24 hours after one infusion of ketamine had slower pre-treatment processing speed than non-responders (n = 9) (F = 8.42; p = 0.008) [130]. This finding was replicated by the same group in a larger study (t = 2.3, p = 0.027). Shiroma et al., (2014) observed that lower pre-treatment attention was linked to higher response to ketamine (F_1,13 = 5.59, p = 0.034) [131]. Grunebaum et al., (2017) observed that poorer pre-treatment memory encoding in individuals with bipolar depression and suicidal thoughts (n = 7) was correlated with better response to one treatment with ketamine (ρ = 0.79, p = 0.04) [132]. Broadly, these findings suggest that individuals with greater cognitive impairment have better antidepressant response to ketamine.

However, four studies observed findings that contradict this conclusion [133–136]. Shiroma et al., (2020) found a correlation between better pre-treatment complex working memory and greater improvement in MADRS scores after five IV ketamine infusions (F_1,49 = 7.66, p < 0.01) [133]. Dai et al., (2022) observed that higher total score on the Montreal Cognitive Assessment (MoCA) was correlated with a more robust reduction of depression scores after repeated infusions (F_1,20 = 5.176, p = 0.034) [134]. Zavaliangos-Petropulu et al., (2023) found that better pre-treatment picture sequence performance was related to greater reduction in depression scores (R = −.27, p = 0.037) [135]. Finally, Singh et al. (2024) observed that better baseline language performance was associated with greater improvement in MADRS scores after three open label infusions of ketamine ($\widehat{\beta }$  = −.97 (95% CI = −1.74 to −0.20), p = 0.02) [136].

There were also studies that failed to find statistically significant relationships between pre-treatment neurocognitive variables and response ketamine [137, 138].

Post treatment

Four studies observed that post-ketamine improvements in neurocognitive domains were predictive of antidepressant response [130, 135, 136, 139]; but not all investigations controlled the improvement in neurocognition for reductions in depressive scores.

Phillips et al. (2022) observed that cognitive changes that were attributed to improvement in verbal memory were associated with greater response to ketamine (F_1,26 = 5.18, p = 0.029) [139]. Zavaliangos-Petropulu et al., (2023) found that improvements in inhibition and complex attention, measured by the Flanker Test, were associated with decreases in the Hamilton Depression Rating Scale (HDRS) (r = −0.2, p = 0.045) after serial ketamine treatments [135]. Singh and colleagues (2024) found an improvement in language only in non-remitters only (p < 0.001) but not in non-remitters (p = 0.08) after three open label infusions [136]. Finally, Murrough et al., (2013) observed that early post-ketamine worsening of cognitive performance (from baseline to 40 min) was associated with lower response rates 1 day post-ketamine (Fisher’s Exact Test 2-sided: p = 0.027) [130].

A better understanding of the specific neurocognitive domains that are markers of response to ketamine and replication of findings by independent research groups is needed.

Personalized use of esketamine in TRD

When compared to ketamine, the number of studies examining markers of response to esketamine are substantially more limited. In addition, most studies investigating biological markers of esketamine have also included sub-samples that received ketamine and report only the aggregate results (i.e., participants who received ketamine and esketamine are analyzed together, therefore, not differentiating the markers for the two medications). In this context, to date there is limited evidence in markers of response to esketamine, particularly with respect to biomarkers.

Clinical markers

The largest study investigating clinical markers of response to esketamine was conducted by Turkoz et al. (2023), who combined the sample of two RCTs for a total of 310 participants treated with intranasal (IN) esketamine [55]. The authors found that younger individuals (18–44 years old), when compared to older individuals (45–64 years old), and individuals with greater depression severity, according to the Montgomery–Åsberg Depression Rating Scale (MADRS), had a greater response to esketamine. However, the effect sizes were modest and studies investigating other samples failed to observe the effect of age and greater depression severity on response to esketamine [58, 60, 140]. A retrospective study of 70 patients treated with six subcutaneous esketamine doses found that individuals with comorbid anxiety disorder (47% of the sample) had a better response to treatment [58]. However, the large analysis by Turkoz et al. (2023) (n = 310) observed an association in the opposite direction with individuals with greater anxiety symptoms (according to General Anxiety Disorder-7 scale) or anxiety comorbidity having a worse response to IN esketamine [55].

Pre-treatment clinical markers that have had positive associations, at least in some studies, with response to ketamine, were also examined in investigations of esketamine. Pre-treatment positive family history of AUD [55], BMI [55, 58, 60] and history of sexual childhood trauma [141] have not been linked to better response to esketamine.

Non-specific pre-treatment clinical markers of poorer response have also been described in studies with esketamine including unemployment [55], greater number of failed antidepressant treatments [55, 60], greater pre-treatment anxiety symptoms and anxiety disorder [55], duration of current episode [55, 58], greater level of treatment resistance [58], and lack of early post-treatment functional improvement [55].

Post-treatment

Two large studies, with partially overlapping samples but distinct analytical strategies, investigated the association between dissociative symptoms and response to IN esketamine and failed to observe a statistically significant relationship [142, 143].

Biomarkers

Pre-treatment

A genome-wide association study (GWAS) with 527 patients from European ancestry found that genetic polymorphisms nearby IRAK3 (p = 3.57 × 10⁻⁸), which encodes a protein that regulates immune signal transduction pathways and NME7 (p = 1.73 × 10⁻⁶), which encodes a protein involved in microtubule nucleation, were associated with greater response to IN esketamine [144]. Importantly, investigations using GWAS usually need large samples [145] and this study with 527 was underpowered. Rotroff et al. (2016) conducted a metabolomic study that examined the pre-treatment plasma levels of >400 brain functional metabolites and failed to find statistically significant relationships with response to one infusion of IV esketamine [146].

With respect to brain-based biomarkers, two studies (one using resting-state functional MRI [147] and one using arterial spin labeling [148]) examined samples that largely overlapped from individuals who were treated with esketamine or ketamine (aggregate results combining the two medications were presented). The first study found that lower pre-treatment FC between the right lateral prefrontal cortex and the subgenual anterior cingulate cortex was associated with greater response to esketamine/ketamine [147] while the second study observed that lower pre-treatment cerebral blood flow in the thalamus was linked to greater response to esketamine/ketamine [148].

Post-treatment

There have been two studies examining post-treatment peripheral biomarkers of response to esketamine and both failed to observe statistically significant associations with post-treatment changes on serum BDNF levels [149] and post-treatment changes on >400 metabolites [146]. In both studies individuals treated with esketamine and ketamine were analyzed together.

There have been two studies investigating post-treatment brain-based biomarkers, and both reported positive results. Gartner et al. (2019) observed that post-treatment increases in FC between the right lateral prefrontal cortex and subgenual anterior cingulate cortex correlated with improvement of depressive symptoms [147]. Gartner et al. (2022) found that post-treatment increases in cerebral blood flow in the thalamus were associated with greater improvement of depression after esketamine/ketamine [148]. Again, in both studies individuals treated with esketamine and ketamine were also analyzed together.

Contraindications and potential safety concerns

In addition to markers of greater response to ketamine and esketamine, treatment selection in TRD should also consider contraindications and safety concerns. In some situations, ketamine and esketamine are contraindicated or their safety has not been well-established and, therefore, considering alternative treatments is appropriate (e.g., electroconvulsive therapy (ECT), repetitive transcranial magnetic stimulation (rTMS), vagus nerve stimulations (VNS), and optimized oral pharmacological regimens). Contraindications to ketamine and esketamine include two main groups of conditions. The first situation is when an increase in blood pressure would be hazardous (e.g., aneurismal vascular disease, arteriovenous malformation, intracerebral hemorrhage). Importantly, in some of these conditions, ECT may be also pose a significant risk in individuals since the transient seizure may also cause significant increases in blood pressure. Another group of contraindications is hypersensitivity either to ketamine, esketamine or any of the excipients in the formulations.

In addition to contraindications, there are some situations where safety has been not well-established. Despite preliminary reports indicating that ketamine may be safe and effective in psychotic depression [150], the available data are still limited and most clinical trials exclude participants with psychosis. Therefore, other antidepressant treatments should be initially considered if psychotic features are present. In such patients, oral pharmacological regimens that include antipsychotics are generally recommended [151]. ECT may also be a good option since meta-analyses have indicated that pre-treatment psychotic symptoms are a marker of response to ECT [56, 152]. Similarly, initial evidence suggests that patients with AUD and/or substance use disorder may benefit from ketamine and esketamine [153, 154]. However, their safety and efficacy in participants with severe substance use disorders is limited. In patients with clinically significant comorbid substance use or history, ketamine and esketamine should be initially avoided, and treatment plans should include interventions focused on the alcohol and substance misuse.

Future directions

Similar to many other antidepressant treatments, most markers of response to ketamine and esketamine have modest effect sizes. As a result, it is unlikely that a single variable will be able to accurately inform treatment selection in real clinical practice. Therefore, the use of predictive models that combine several variables to forecast future response (multivariate predictive models) will be needed, which will require the study of larger sample sizes. The investigation of larger sample sizes also provides results that are representative and more likely to be replicated. There are successful examples of data sharing and multicenter studies that need to be further incentivized [41, 55, 73, 93, 140, 155–157]. The use of measurement-based care and focused biomarkers imbedded in the electronic medical record will not only allow the systematic study of larger samples but also the examination of real-world samples that have ecological validity [158]. It is also important to further investigate the specific use of ketamine/esketamine in non-TRD samples, which could reduce the rates of treatment-resistance.

Although analyses of markers of response are already conducted by some studies, clinical trials could routinely report, in addition to overall efficacy results, more detailed analyses of markers of response to ketamine and esketamine. It is also important to ensure that biomarkers and clinical variables are obtained in the similar manner across studies (e.g., serum vs plasma, time of collection, clinical measure used to examine outcome and marker) to allow appropriate comparison across trials and to increase signal-to-noise ratio.

Studies comparing differential markers of response between ketamine and esketamine, and other common treatments in TRD are needed. The comparison of differential markers between two or more antidepressant interventions could also minimize findings involving non-specific markers of poorer response.

In addition, although the mechanism of action of ketamine and esketamine may share similarities, it is unclear how much clinical and biological markers are shared by the two medications. Future investigations could investigate, in addition to specific markers of treatment response, further therapeutic information such as dose, treatment schedule and markers and predictors of potential side effects. Finally, investigation of biomarkers could focus on neurobiological targets that have a direct and meaningful relationship to mechanisms of action of ketamine and esketamine [53].

Conclusions

Approximately half of patients with TRD respond to ketamine and esketamine, and more research is needed on how to better identify the patients that are likely to respond to these medications. Most research on clinical and biological markers of response has focused on ketamine, and further research is needed for esketamine. A positive family history of AUD in first-degree relatives is the most consistent clinical markers of response to ketamine and needs to be further investigated both in individuals treated with ketamine and esketamine. Another clinical variable that warrants further investigation is pre-treatment presence/severity of childhood trauma. Brain-based biomarkers, particularly electrophysiological modalities such as EEG, have shown a better prediction or response to ketamine and esketamine than peripheral biomarkers and can be better understood in the context of the central pharmacological effects of ketamine. Predictive models combining clinical and biological (especially, brain-based) markers are promising, and investigation of larger samples is needed to develop such models.

Author contributions

GCM contributed to conceptualization and design of this review, participated in the selection of the included manuscripts and the interpretation of available literature, wrote the original and revised drafts, and reviewed and edited the original and revised manuscripts. ID assisted GCM in the elaboration of the original and revised drafts, reviewed and edited the original and revised manuscripts. FSG contributed to conceptualization and design of this review, contributed to the selection of the included manuscripts and interpretation of available literature, reviewed and edited the original and revised manuscripts. CAZ contributed to conceptualization and design of this review, contributed to the selection of the included and the interpretation of available literature, reviewed and edited the original and revised manuscripts. TDG contributed to conceptualization and design of this review, to the selection of the manuscripts included in this review and the interpretation of available literature, supervised and wrote the original and revised drafts, and reviewed and edited the revised and edited manuscripts.

Competing interests

GCM, ID, FSG, CAZ, and TDG have no consulting or grant funding to disclose in the prior 3 years. TDG was supported by NIH/NIMH R01MH107615 and U.S. Department of Veterans Affairs Merit Awards 1I01BX004062 and 1I01BX006018. Funding for this work was provided in part by the Intramural Research Program at the National Institute of Mental Health, National Institutes of Health (IRP-NIMH-NIH; ZIAMH002857). The contents of this manuscript do not represent the views of the U.S. Department of Veterans Affairs, the National Institutes of Health, the Department of Health and Human Services, or the United States Government. TDG is listed as an inventor in patents and patent applications related to the pharmacology and use of a ketamine metabolites, including (2 R,6 R)-hydroxynorketamine, in the treatment of depression, anxiety, anhedonia, suicidal ideation, and post-traumatic stress disorders. TDG has assigned his patent rights to the University of Maryland, Baltimore, but will share a percentage of any royalties that may be received by the University of Maryland, Baltimore. CAZ is listed as a co-inventor on a patent for the use of ketamine in major depression and suicidal ideation; as a co-inventor on a patent for the use of (2 R,6 R)-hydroxynorketamine, (S)-dehydronorketamine, and other stereoisomeric dehydroxylated and hydroxylated metabolites of (R,S)-ketamine metabolites in the treatment of depression and neuropathic pain; and as a co-inventor on a patent application for the use of (2 R,6 R)-hydroxynorketamine and (2S,6S)-hydroxynorketamine in the treatment of depression, anxiety, anhedonia, suicidal ideation, and post-traumatic stress disorders. He has assigned his patent rights to the U.S. government but will share a percentage of any royalties that may be received by the government.