Ketamine for treatment-resistant major depressive disorder: Double-blind active-controlled crossover study

Abstract

Background:

The N-methyl-D-aspartate antagonist ketamine has rapid onset antidepressant activity in treatment-resistant depression (TRD).

Aims:

To evaluate mood rating, safety and tolerability data from patients with TRD treated with ketamine and the psychoactive control fentanyl, as part of a larger study to explore EEG biomarkers associated with mood response.

Methods:

We evaluated the efficacy and safety of intramuscular racemic ketamine in 25 patients with TRD, using a double-blind active-controlled randomized crossover design. Ketamine doses were 0.5 and 1 mg/kg, and the psychoactive control was fentanyl 50 mcg, given at weekly intervals.

Results/outcomes:

Within 1 h of ketamine dosing, patients reported reduced depression and anxiety ratings, which persisted for up to 7 days. A dose–response profile for ketamine was noted for dissociative side effects, adverse events and changes in blood pressure; however, changes in mood ratings were broadly similar for both ketamine doses. Overall, 14/25 patients (56%) were responders (⩾50% reduction at 24 h compared with baseline) for either ketamine dose for the Hospital Anxiety and Depression Scale (HADS), and 18/25 (72%) were responders for the HADS-anxiety scale. After fentanyl, only 1/25 (HADS-depression) and 3/25 (HADS-anxiety) were responders. Ketamine was generally safe and well tolerated in this population.

Conclusions:

Our findings add to the literature confirming ketamine’s activity against depressive and anxiety symptoms in patients with TRD.

Introduction

Berman’s seminal publication on the rapid-onset antidepressant activity of low-dose ketamine in treatment-resistant depression (TRD) identified a new therapeutic option for this patient group (Berman et al., 2000). Since that time, multiple research groups have confirmed racemic ketamine’s antidepressant and anxiolytic activities (Johnston et al., 2023). A meta-analysis of 28 acute dosing ketamine studies in patients with TRD reported positive results for both treatment response and remission compared with control treatments (Bahji et al., 2022). In addition, a nasal spray formulation of esketamine was approved by the FDA for use in TRD in 2019 (Kim et al., 2019).

A range of biomarker studies have been published to elucidate the neurobiological mechanisms for ketamine’s activity (Kadriu et al., 2020). We recently reported on the effects of ketamine on EEG changes in patients with generalized anxiety disorder and social anxiety disorder (Shadli et al., 2018). Decreases in medium–low (theta) frequency at right frontal sites correlated with the effect of ketamine on the Fear Questionnaire in patients with Social Anxiety Disorder. We have hypothesized that the effects of ketamine on EEG biomarkers linked to trait neuroticism will be associated with EEG (McNaughton and Glue, 2020). To expand this biomarker work, we have recently recruited and evaluated patients with TRD, post-traumatic stress disorder (PTSD), obsessive-compulsive disorder and specific phobia. This paper reports mood rating, safety and tolerability data from patients with TRD treated with ketamine; EEG biomarker data are still being analysed and will be reported elsewhere.

Materials and methods

The protocol and consent forms for this study were approved by the Central Health and Disability Ethics Committee (19/CEN/21), and the study was registered with the Australian and New Zealand Clinical Trial Registry (ACTRN12619000311156). The protocol included recruitment of patients with TRD, PTSD, OCD and spider phobia in separate cohorts, to evaluate the effects of ketamine on EEG biomarkers linked to trait neuroticism (McNaughton and Glue, 2020); only the TRD cohort is reported in this paper. This was a randomized double-blind psychoactive-controlled study in patients with treatment-resistant (Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5)) major depressive disorder (failure to respond to at least two conventional antidepressants and psychotherapy). Patients were interviewed using a structured clinical interview (Sheehan et al., 1998). Patient inclusion criteria included having a Montgomery-Asberg Depression Rating Scale (MADRS; Montgomery and Asberg 1979) score of ⩾20, being aged between 18 and 50 years, having good overall health and having had an inadequate response to prior treatment. Patients were permitted to have comorbid anxiety spectrum disorders. Exclusion criteria included evidence of severe or chronic medical disorders, past or current diagnoses of schizophrenia, bipolar disorder or current psychotic symptoms, current significant suicidal ideation, patients who were pregnant or lactating, substance use disorder or dependence in the last 6 months, and prior history of seizures or head injury. Patients provided signed informed consent before screening and were assessed as suitable to participate based on a review of medical history, safety laboratory tests (complete blood count, electrolytes, pregnancy test for patients who were capable of becoming pregnant), negative urine drug screening and vital signs. Patients were asked to provide a referral from a GP or psychiatrist who knew them well and could confirm the medical diagnosis and course of treatment. Patients were permitted to remain on current medication regimens and to continue with ongoing psychotherapy; however, no new treatments were to be started or changed.

Study treatments included single doses of racemic ketamine 0.5 mg/kg, 1.0 mg/kg or fentanyl 50 mcg (psychoactive control). These were administered as intramuscular injections in the deltoid muscle, according to the technique proposed by the New Zealand Ministry of Health (2020). Study drugs were given according to a computer-generated random code with balanced randomization, using a three-way within-subject double-blind active-controlled cross-over design. There were three dosing sessions, each session separated by at least 1 week. A 10-min relaxation EEG test was obtained pre-dose, and 2 h and 24 h after each dosing session to assess the timing of EEG changes in response to study treatments (data to be presented elsewhere). Mood ratings and assessments of safety and tolerability were collected up to 168 h after each dose. Patients were monitored in the research clinic for a minimum of 2 h post-dose, with vital signs obtained pre-dose and at 15, 30, 45, 60, 90 and 120 min post-dose. Mood assessments included the MADRS (Montgomery & Asberg, 1979) pre-dose, and the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983) pre-dose, at 60 and 120 min, and 24, 72 and 168 h post-dose. The choice of including the HADS scale as the main depression and anxiety rating scale was made after study registration but before final ethics committee approval and was to evaluate the use of a patient-rated outcome scale for these endpoints. Responder analyses (patients with reductions in HADS-anxiety and -depression scores ⩾50%) were evaluated at 24 h post-dose. Safety and tolerability assessments included reported adverse events (AEs) throughout the study, assessment of bladder symptoms using the bladder pain/interstitial cystitis scale (BPIC; Humphrey et al., 2012), and Clinician-Administered Dissociative States Scale (CADSS; Bremner et al., 1998) scores pre-dose, 30 and 60 min post-dose. Because of our previous experience in treating patients with ketamine, we implemented a protocol of administering 4 mg of oral ondansetron 1 h prior to dosing, to reduce the incidence of nausea and vomiting. Maintenance of blinding in participants and raters was not assessed.

We assessed cognition using orientation questions and Trail Making tests because changes in cognition (memory impairment and executive functioning) have been reported when ketamine is used recreationally at high doses (Strous et al., 2022). Before patients were discharged from the research clinic, 2 h after dosing, we assessed their level of orientation, and recorded blood pressure and heart rate to check these were ⩽120% of baseline, that they were able to walk unassisted, were feeling physically well and not significantly sedated or distressed. If we had any concerns, we kept them in the clinic and reassessed them. Blinded safety data were reviewed during the study by an independent Data Safety Monitoring Board.

The two HADS subscales were the primary outcome measures. We used data from Zarate (2006) to calculate sample size (with 20 subjects, mean difference ketamine vs placebo = 46.4%, SD 20.3, alpha = 0.05, statistical power 100%). Post hoc statistical power estimates were also obtained from the analysis of variance (ANOVA) outputs. Summary statistics were calculated and reported for demographics, vital signs and rating scale data. Categorical variables were reported using counts and percentages. Repeated measures ANOVA was used to assess the effect of drug treatment on HADS depression and anxiety subscale scores. The frequency of AEs by treatment arm was analysed by chi-square tests.

Results

We screened 50 patients and enrolled 29 in the study. Of the 29 patients enrolled, 25 completed the study. One patient was withdrawn because of a COVID-19 infection, one patient had untreated hypertension and was withdrawn on safety grounds, and two patients were withdrawn as they no longer met entry criteria. The patients who entered the study comprised 12 females and 13 males. The mean age of patients was 32 years (range 19–54 years). All subjects met DSM-5 criteria for major depressive disorder, which was treatment resistant. The mean (SD) number of failed antidepressants prior to enrolment was 3.5 (1.6), and 6/25 subjects had failed various augmentation strategies. The mean (SD) duration of the current depressive episode was 10.2 (9.6) years (range 1–37 years). No patient had trialled electroconvulsive therapy during their current depressive episode. There was significant co-morbidity, with 11 patients having generalized anxiety disorder, 8 with social anxiety disorder, 11 with PTSD, 3 with agoraphobia and 1 with obsessive-compulsive disorder.

Mean (SD) pre-dose MADRS scores were 23.0 (7.3), 24.3 (7.6) and 22.0 (6.5) for the fentanyl, ketamine 0.5 mg/kg and ketamine 1 mg/kg dose groups, respectively. Mean (SEM) HADS scores by treatment group are shown in Figure 1. Mean baseline scores for both scales were consistent with clinically significant depression and anxiety (HADS subscale scores ⩾11). The reduction in HADS-depression scores was greater for both ketamine doses compared with the fentanyl treatment arm (Figure 1(a)). The duration of response was longer for the ketamine 1 mg/kg arm compared with the 0.5 mg/kg arm. Repeated measures ANOVA showed statistically significant effects of time (F(5,120) = 11.6, p < 0.001), treatment (F(2,48) = 5.9, p = 0.005) and a treatment by time interaction (F(10,240) = 2.7, p = 0.003). Post hoc statistical power estimates from the ANOVA were 77.7% for treatment, 100% for time and 81.4% for the treatment-by-time interaction. The reduction in HADS-anxiety scores was also greater for both ketamine doses compared with the fentanyl treatment arm, with similar response profiles for both ketamine dose arms (Figure 1(b)). Repeated measures ANOVA showed statistically significant effects of time (F(5,120) = 41.7, p < 0.001), treatment (F(2,48) = 10.1, p < 0.001) and a treatment by time interaction (F(10,239) = 2.1, p = 0.024). Post hoc statistical power estimates from the ANOVA were 96.7% for treatment, 100% for time and 57.8% for the treatment by time interaction. Overall, 14/25 patients (56%) were responders (⩾50% reduction at 24 h compared with baseline) for either ketamine dose for the HADS-depression scale, and 18/25 (72%) were responders for the HADS-anxiety scale. After fentanyl, only 1/25 (HADS-depression) and 3/25 (HADS-anxiety) were responders.

Figure 1.

Effect of fentanyl and ketamine doses on the mean (SEM) HADS-depression: (a) and HADS-anxiety scores and (b) in patients with TRD.

HADS: Hospital Anxiety and Depression Scale; TRD: treatment-resistant depression.

Safety and tolerability

Blood pressure changes after dosing are shown in Figure 2(a). There were negligible changes after fentanyl, 15 min after ketamine dosing, mean change in systolic blood pressure was 11 and 17 mmHg for the ketamine 0.5 and 1.0 mg/kg dose groups, respectively, and 7 and 14 mmHg for diastolic blood pressure. Blood pressure values trended downwards by 60 min. All patients reported dissociative symptoms after ketamine dosing, starting approximately 3–5 min after each intramuscular (IM) injection, with peak intensity around 15–30 min, and then slowly decreasing. CADSS scores were highest after the 1.0 mg/kg dose of ketamine (Figure 2(b)) with the peak at 30 min post-dose. There were smaller increases after ketamine 0.5 mg/kg and negligible dissociative effects with fentanyl.

Figure 2.

Effect of fentanyl and ketamine doses on CADSS dissociation scores (a) and systolic and diastolic blood pressures (b) in patients with TRD.

CADSS: Clinician-Administered Dissociative States Scale; TRD: treatment-resistant depression.

AEs are shown in Table 1. Fentanyl had the highest proportion of patients reporting no AEs and had numerically the greatest number reporting drowsiness. After ketamine dosing, patients reported statistically significantly higher rates of blurred vision, light headedness, numb lips and nausea of mild–moderate intensity. The duration of side effects was less than 2 h in all patients. Because of pre-dosing with ondansetron, it is possible that rates of nausea and vomiting were reduced. Mean BPIC scores taken pre-dose were 2.8, 4.0 and 2.7 for the fentanyl, ketamine 0.5 mg/kg and ketamine 1 mg/kg dose groups, respectively. No patients experienced a serious AE.

Table 1.

AEs by study treatment.

AE	Fentanyl	Ketamine 0.5	Ketamine 1.0	Chi2/p
No AEs	13	5	4	9.4/p = 0.009
Blurred vision	0	9	16	19.9/p < 0.001
Drowsy	8	3	6	2.9/p = 0.24
Light headed	3	13	5	11.1/p = 0.004
Numb lips	1	8	9	8.3/p = 0.02
Nausea	1	2	6	5.3/p = 0.07
Vomiting	0	0	2	0.5/p = 0.77
Tingling sensation	2	2	4	1.1/p = 0.57
Dry mouth	0	3	3	1.3/p = 0.53
Nystagmus	0	0	2	0.5/p = 0.77

AEs: adverse events.

Discussion

The rapid-onset activity of ketamine on symptoms of anxiety and depression in patients with TRD has been reported from multiple research groups (Bahji et al., 2022; Johnston et al., 2023), so the positive mood responses in this paper are unsurprising. The novel aspects of this study include the route of ketamine administration (intramuscular injection), the dose–response data and the use of a patient-scored rating scale, the HADS. As expected, we have confirmed ketamine’s antidepressant and anxiolytic activity in patients with TRD. IM ketamine dosing was effective, with an acceptable tolerability profile. Changes in mood ratings after ketamine dosing showed no or small differences in mood responses between doses of 0.5 and 1 mg/kg, except possibly for more durable antidepressant effects for the higher dose. The HADS scale appears to be a suitable tool to assess the anxiolytic and antidepressant effects of ketamine in patients with TRD.

Although most published data on the use of ketamine in patients with TRD is via intravenous (IV) administration, IM dosing of ketamine was safe and well tolerated in the present study. Most published data using IM dosing has been from case reports or small case series (Andrade, 2017). Some advantages of IM administration compared with an IV infusion are the reduced staff time and equipment required compared with an IV infusion. The bioavailability of IM racemic ketamine appears to be very close to values after IV administration (92–93%; Glue et al., 2021). Although oral dosing of ketamine may eventually be the preferred route of administration, IM dosing may be a useful alternative route to consider.

Published dose–response data for ketamine in TRD are limited. Loo (Loo et al., 2016), who compared IV, IM and subcutaneous dosing routes for doses of 0.1 to 0.5 mg/kg, reported dose–responses for all routes of administration. More recently, the same research group (Loo et al., 2023) reported that ketamine dosed flexibly (0.5–0.9 mg/kg) had greater effects on depression ratings than ketamine at a fixed dose of 0.5 mg/kg. Chilukuri reported that mood responses to ketamine doses of 0.25 mg/kg and 0.5 mg/kg in patients with MDD were equivalent (Chilukuri et al., 2014). Fava (Fava et al., 2020) reported that IV racemic ketamine doses of 0.5 and 1 mg/kg were more effective in patients with TRD than lower ketamine doses or midazolam (active placebo). In the present study, IM ketamine doses of 0.5 and 1 mg/kg produced similar acute changes in both the HADS-depression and -anxiety scales (Figure 1). The only dose-related difference was the apparent longer duration of antidepressant effects for the higher ketamine dose (Figure 1(a)). In a recent meta-analysis of ketamine in anxiety disorders, we identified relatively minor differences between ketamine doses of 0.5 and 1 mg/kg on changes in several anxiety rating scales (Whittaker et al., 2021), which may indicate that the 0.5 mg/kg dose is a reasonable initial dose for this patient group. One difference in the magnitude of changes in HADS depression and anxiety scales was the greater mean reduction in anxiety scores at 24 h post-dose (−37 to −40%) compared with the reduction in mean depression scores (−26 to −28%) (Figure 1). This could be a chance finding or might suggest that anxiety symptoms are more responsive to ketamine than depressive symptoms. More data are needed to address questions about ketamine dose–response and differential symptom responsiveness in patients with TRD.

The MADRS and Hamilton Depression Scales are the most commonly used instruments to evaluate mood responses to ketamine in patients with TRD (Yavorsky et al., 2023). The HADS is another validated scale (Bjelland et al., 2002), and although it has been used to evaluate the antidepressant effects of ketamine (Irwin et al., 2013; Jafarinia et al., 2016; Moitra et al., 2016; Voute et al., 2023; Zhou et al., 2021), this has been mainly in patients with medical or surgical comorbidities. In this study, the HADS was quick and easy to administer. It is important to consider the patients’ perspectives of mood and anxiety changes after study treatments, without the moderating effects of a research rater. When patients repeat the HADs regularly throughout the study, they can see change (or its absence) for themselves in a quantitative format. We think it is a helpful strategy for patients to feel included in the research process and more useful than the nebulous ‘I feel so much better’.

We acknowledge a number of limitations relating to this research. Fentanyl was not an ideal psychoactive control, as shown by its very different side effect profile compared with ketamine, and a complete absence of dissociative symptoms. In this sense, it is not much different to low-dose midazolam. Methodologically, it would be useful to identify an alternative psychoactive control that had a similar side effect profile to ketamine, to assist with blinding. We did not formally assess the blinding of patients and raters. We did not compare the psychometric performance of the HADS against other mood rating scales; however, it appears to be sensitive to the effects of ketamine in patients with TRD. We used a balanced crossover design to manage potential carry-over effects on mood ratings across the three study periods. To further evaluate this, we analysed HADS scores from all subjects’ visit 1 data (Supplemental Figure 1). The baseline (pre-dose) scores were very similar (HADS-D range 13.1–13.6, HAD-A 13.0–14.1) and the response profiles for anxiety and depression ratings over time were very similar to Figure 1.

In conclusion, our data add to the literature confirming ketamine’s activity against depressive and anxiety symptoms in patients with TRD. We await with interest the final EEG data, to explore associations between ketamine response and EEG changes, as a potential response biomarker for ketamine in internalizing disorders.