Rapid effectiveness of intravenous ketamine for ultraresistant depression in a clinical setting and evidence for baseline anhedonia and bipolarity as clinical predictors of effectiveness

Thomas and colleagues frame the study around an underserved subgroup of depressed patients who have failed extensive prior treatments — antidepressants, augmentation strategies, psychotherapy, and electroconvulsive therapy (ECT) — and describe this population as ultraresistant depression (URD). Previous research indicates that a substantial proportion of major depressive disorder (MDD) patients develop treatment-resistant depression (TRD) and that higher levels of treatment resistance predict poorer outcomes. Intravenous (IV) ketamine, an NMDA receptor antagonist acting on the glutamatergic system, has shown rapid antidepressant and anti‑suicidal effects in TRD in controlled research settings, but evidence is limited for its effectiveness in routine clinical practice and specifically in patients with URD. The authors note prior work suggesting clinical predictors of ketamine response (for example anhedonia, family history of alcohol use disorder, higher body mass index and early response) but highlight the need for replication and investigation in more severely resistant samples. This study set out to measure response and remission rates after a six‑infusion course of IV ketamine delivered in routine clinical settings to patients with URD, and to assess whether previously reported clinical predictors of ketamine effectiveness could be supported in this population. The investigators used a retrospective, open‑label database approach to examine outcomes and applied logistic regression to test clinical and demographic predictors of response, defined primarily as a 50% reduction on standard depression rating scales at any point during treatment. The study therefore aimed to provide real‑world effectiveness data and to explore feasible clinical predictors that could help stratify patients most likely to benefit from ketamine.

The researchers conducted a retrospective chart‑based database study of all patients who received IV ketamine for depression at two community hospitals in Edmonton between 13 May 2015 and 31 December 2016. Fifty subjects were included in the analysis; one additional patient with clinician‑noted improvement was excluded because rating scales were not completed, while four patients with no response documented but missing end‑of‑treatment scales were counted as non‑responders. Ethical approval and a waiver of consent were obtained from local research ethics boards. Inclusion required meeting at least three of four pragmatic clinical criteria: (a) a clinician‑determined major depressive episode (unipolar or bipolar), (b) refractory to at least five psychotropic medications for mood disorder as judged by the treating clinicians, (c) refractory to or unsuitable for ECT (at least six ECT treatments without adequate response), and (d) acute suicidality. Exclusion criteria included active psychosis, active substance abuse or dependence, unstable medical conditions, pregnancy, significant primary personality disorder, dissociative identity disorder, or a history of severe adverse reaction to ketamine. Patients continued concomitant medications; no drug classes were mandated to stop and past substance use in remission was permitted. Ketamine was administered at 0.5 mg/kg IV over 40 minutes by nursing staff with a physician on site, typically 2–3 times per week up to a recommended maximum of six infusions. Monitoring included pre‑, mid‑ and post‑infusion vital signs and safety bloodwork and ECG at baseline. If side effects occurred clinicians could slow or stop the infusion or give low‑dose quetiapine or lorazepam. Baseline and outcome assessments used one or more standard rating scales: the Montgomery–Åsberg Depression Rating Scale (MADRS), Beck Depression Inventory I or II (BDI‑I/II), Hamilton Rating Scale for Depression 17 or 21 (HRSD‑17/21) and the Brief Psychiatric Rating Scale (BPRS). Administration timing was not standardised but was typically at least four hours post‑infusion. A bespoke nursing side‑effect tracking sheet categorised side effects as mild, moderate, or severe. Response was defined as a 50% reduction in BDI‑I, BDI‑II or HRSD‑21 at any time during treatment; remission used accepted cut‑offs for each scale. Statistical analysis compared baseline characteristics between responders and non‑responders using independent t‑tests and z‑tests. Logistic regression models examined predictors of response (and, separately, remission). Three sets of logistic regressions were run to reduce type II errors: (1) categorical clinical variables (age split >50/≤50, gender, family history of alcohol use disorder, suicidality, anxiety, unipolar versus bipolar diagnosis, and anhedonia), (2) continuous variables (BMI, age, and Maudsley Staging Method (MSM) score, the latter measuring level of treatment resistance), and (3) concurrent medication categories (lamotrigine, benzodiazepines, antipsychotics). Missing data were handled by excluding the individual from that specific analysis rather than imputing means. Odds ratios (ORs) from logistic regression quantify the change in odds of response associated with predictors.

Fifty patients meeting the URD criteria were analysed. At baseline the cohort had a high degree of treatment resistance: mean Maudsley Staging Method (MSM) score was 12.1 out of 15, 90% were ECT‑resistant and 86% had prior unsuccessful psychotherapy. Nearly all patients remained on concurrent psychotropic medications during ketamine treatment. Personality trait diagnoses (Cluster B and C) were documented in 44% and 48% respectively, and 64% had a lifetime DSM‑5 anxiety disorder recorded. The overall response rate (≥50% reduction on BDI‑I/II or HRSD‑21 at any time during treatment) was 44% and the remission rate was 16%. The authors report that no patient who showed response at an interim point subsequently lost that response by the end of the observed treatment course. Subscale improvements reached zero (complete remission on that item) for 54.0% of patients who were at least mildly depressed at baseline on the depressed‑mood subscale, and 68.9% of those who were at least mildly suicidal at baseline scored zero on the suicidality subscale during treatment. Baseline comparisons between responders and non‑responders found that concurrent lamotrigine use differed significantly (p=0.015) while bipolarity differed at a trend level (p=0.080). In categorical logistic regression analyses, at least moderate baseline anhedonia alone was associated with a 55% increased likelihood of response (reported as 55% more likely; p=0.072, OR=1.21) and bipolar depression alone with a 54% increased likelihood (p=0.087, OR=1.18). When both factors were present together (bipolar diagnosis plus at least moderate anhedonia), the likelihood of response increased to 73% (p=0.035, OR=3.03). Logistic regressions using continuous variables (BMI, MSM, age) and analyses of concurrent benzodiazepine or antipsychotic use did not identify statistically significant predictors. Repeating logistic regressions with remission as the dependent variable yielded no significant clinical predictors. Regarding tolerability, most patients experienced mild, transient sensations; 82% had mild side effects that did not require intervention. Four of 50 patients experienced adverse events leading to premature cessation of an infusion. The investigators note that all five patients who were on lamotrigine during the treatment course responded, an observation described as underpowered but potentially noteworthy.

The researchers interpret the findings as evidence that a six‑infusion IV ketamine course can be effective and reasonably well tolerated in a highly treatment‑resistant, clinically complex URD population treated in routine hospital settings. They emphasise the severity and heterogeneity of the sample — 90% ECT‑resistant, high rates of comorbid personality traits and anxiety disorders, and a mix of unipolar and bipolar depression — and view the achieved response (44%) and remission (16%) rates as encouraging given this poor‑prognosis group. The authors also highlight notable subscale improvements, particularly in depressed mood and suicidality items. Anhedonia emerged as a clinically relevant predictor of response in this dataset. The authors situate this finding within prior work showing anti‑anhedonic effects of ketamine, and link anhedonia to anterior cingulate cortex dysfunction in neuroimaging and electrophysiological studies; they note that ketamine's putative mechanism of action (synaptogenesis via glutamatergic pathways) may plausibly affect reward circuitry implicated in anhedonia. Bipolarity was also associated with greater likelihood of response in the combined predictor model, and the authors relate this to other evidence of ketamine's efficacy in bipolar depression. They acknowledge, however, that bipolar diagnosis in this study derived from standard clinical documentation rather than structured diagnostic interviews. The discussion addresses the unexpected finding regarding lamotrigine, noting that all five patients on lamotrigine responded; this contrasts with the theoretical expectation that lamotrigine (a glutamate‑release inhibitor) might attenuate ketamine's glutamatergic effects. The authors reference mixed prior results and describe their observation as underpowered and hypothesis‑generating. Limitations are acknowledged: the retrospective, open‑label design lacks controls and is vulnerable to confounding, including concurrent medications and psychosocial interventions; missing or non‑standardised timing of rating scales; limited power for some predictors due to low frequency; and the potential for placebo or expectancy effects (they cite a 28% response rate with midazolam in another study). The diagnostic assessments were based on routine clinical practice rather than standardised instruments, which the authors note both limits precision and arguably reflects real‑world accuracy given long clinical relationships. They call for replication in larger samples and propose that combining clinical predictors with biomarkers (for example fMRI‑defined biotypes or measures of global brain connectivity) could improve stratification and move towards personalised treatment selection. The authors also support the development of registries to aggregate real‑world ketamine outcomes and predictors.

In routine clinical practice, a six‑infusion course of IV ketamine showed effectiveness and short‑term tolerability in a complex, severely treatment‑resistant depressed population receiving multiple concurrent medications. The observed response rate is presented as encouraging for this unmet clinical need. The authors recommend larger collaborative registries and further research combining clinical predictors (such as moderate anhedonia and bipolarity) with biomarkers to improve patient selection and reduce the duration of disabling, ineffective treatments for patients with ultraresistant depression.