The Impact of Childhood Maltreatment on Intravenous Ketamine Outcomes for Adult Patients with Treatment-Resistant Depression

Major depressive disorder and bipolar depression affect a substantial proportion of the population and an estimated 35% of people with depression meet criteria for treatment-resistant depression (TRD), commonly defined as less than 50% improvement after at least two different antidepressant trials. TRD is associated with worse quality of life and higher mortality, so identifying predictors of response to alternative interventions is clinically important. Intravenous ketamine, an N-methyl-D-aspartate receptor (NMDAR) antagonist, produces rapid antidepressant effects in TRD, possibly via glutamate-mediated activation of AMPA receptors, increased brain-derived neurotrophic factor and mTOR signalling, but it is unknown whether childhood maltreatment influences ketamine response. Previous research has associated childhood maltreatment with poorer outcomes on conventional antidepressants, and there is emerging evidence for NMDAR antagonists in trauma-related disorders such as PTSD, but no clear data on how early-life maltreatment affects ketamine efficacy in TRD. O'brien and colleagues set out to examine whether self-reported childhood maltreatment, measured with the Childhood Trauma Questionnaire (CTQ), predicts antidepressant response to IV ketamine in a naturalistic outpatient sample. They tested two related hypotheses: first, that a history of childhood maltreatment would predict an unfavourable response to an acute single infusion and to a course of repeated infusions; and second, that greater maltreatment burden (a higher number of clinically significant maltreatment domains, termed maltreatment load) would be associated with poorer ketamine effectiveness.

This was a naturalistic, clinic-based observational study using de-identified data collected during routine care at a ketamine treatment centre between April 2016 and April 2019 under an IRB waiver. Adults presenting with moderate to very severe depressive symptoms (baseline QIDS-SR > 10) who had failed at least one antidepressant trial were eligible for inclusion in analyses. Two analytic samples were defined: a single-infusion sample (n = 115) assessed 3 or 7 days after one IV ketamine infusion, and a repeated-infusion subsample (n = 63) who completed at least four infusions on a fixed twice-weekly (every 3–4 days) or weekly (every 7 days) schedule. Depressive symptoms were measured with the Quick Inventory of Depressive Symptomatology—Self Report (QIDS-SR) at baseline and at post-infusion visits. Childhood maltreatment before age 18 was assessed with the 28-item Childhood Trauma Questionnaire (CTQ), which yields five subscales (sexual abuse, physical abuse, emotional abuse, physical neglect, emotional neglect) and a total score. The investigators derived a "maltreatment load" (0–5) representing the number of CTQ subscales for which a patient met established cut-offs for clinically significant maltreatment (sexual, physical abuse and physical neglect cut-off ≥ 8; emotional abuse ≥ 10; emotional neglect ≥ 15). IV ketamine was administered in a monitored clinical setting by anaesthesia staff; dosing was weight-based and delivered over 40 minutes to 2 hours. Mean per-infusion doses reported were approximately 0.62 mg/kg in the single-infusion group and 0.70 mg/kg in the repeated-infusion group. The primary outcomes were change in QIDS-SR (baseline minus post-infusion) for single and repeated infusions, with response defined as ≥ 50% reduction in QIDS-SR and remission defined as QIDS-SR < 6. Statistical analyses used repeated measures general linear models (RM-GLM) to test time effects and interactions with CTQ variables. CTQ predictors were entered as continuous or dichotomous variables based on initial correlations with QIDS-SR change. Demographic (age, gender), treatment (ketamine dose, timing of post-infusion assessment, treatment schedule), diagnosis and concurrent psychotropic medication were tested as potential moderators. Correlational analyses and chi-square or t-tests evaluated associations with response and remission. Bayes factors (BF) were reported alongside p-values to express the strength of evidence for alternative versus null hypotheses. An intent-to-treat analysis using last-observation-carried-forward was also performed to compare with per-protocol findings.

Single-infusion sample (n = 115): RM-GLM showed a large and statistically significant reduction in depressive symptoms after one ketamine infusion: mean QIDS-SR fell from 18.63 ± 3.70 at baseline to 13.12 ± 5.13 post-infusion (F(1,114) = 175.70, p < 0.001, η2p = 0.61), with Bayes factor indicating very strong evidence (BF > 1.27 × 10+10). Nineteen percent (n = 22) met response criteria (≥ 50% reduction) and 7% (n = 8) achieved remission (QIDS-SR < 6) after a single infusion. Correlational analyses linked greater maltreatment burden to larger acute symptom reductions: QIDS-SR change correlated with maltreatment load (r = 0.31, p < 0.001; BF10 = 29.07), sexual abuse (SA; r = 0.29, p = 0.001) and physical neglect (PN; r = 0.24, p = 0.01), and with CTQ total and physical abuse (PA) at lower strength. RM-GLM including load showed a significant time × load interaction (F(1,109) = 3.78, p = 0.003, η2p = 0.148; BF10 = 6.37). Follow-up comparisons on change scores indicated patients with maltreatment load of 5 (clinically significant on all five CTQ domains) had larger reductions in QIDS-SR than those with loads 0, 1 or 3 (t > 3.18, p < 0.03). A subscale analysis showed a significant time × SA interaction (F(1,112) = 5.37, p = 0.022); the extracted text does not report the precise mean decrease for the SA ≥ 8 subgroup. Repeated-infusion sample (per-protocol, n = 63): Across five visits (baseline plus four post-baseline assessments) RM-GLM showed a significant time effect (F = 97.60, p < 0.001, η2p = 0.61; BF10 ≈ 1.77 × 10+46). Post-hoc comparisons indicated significant reductions after the first and second infusions, with smaller or non-significant additional change after the third and fourth infusions. After four infusions, 46.03% (n = 29) were responders and 23.81% (n = 15) were in remission. In the repeated-infusion analyses, QIDS-SR change correlated strongly with maltreatment load (r = 0.427, p < 0.001; BF10 = 59.89) and with PN (r = 0.390, p = 0.002), total CTQ, PA and SA; emotional abuse and neglect did not correlate significantly. RM-GLM with load revealed a significant time × load interaction (F[4,248] = 2.40, p = 0.003, η2p = 0.17; BF = 6.57). Examination of change scores showed larger mean QIDS-SR reductions for load 4 (14.67 ± 4.41) and load 5 (15.40 ± 2.30) compared with load 1 (7.38 ± 4.87) and a trend versus load 0 (8.0 ± 4.85). Subscale analyses identified a significant time × PA interaction (F[4,236] = 5.83, p < 0.001, η2p = 0.090) and a time × PN interaction that approached significance (p = 0.052). A model including CTQ total score also yielded a significant time interaction (F[4,244] = 3.08, p = 0.017). Response and remission analyses at Visit 5 found significant associations between higher maltreatment load and remission, and between physical neglect and remission; Bayes factors suggested very strong evidence for the effect of load on remission and strong evidence for PN, with moderate evidence for SA and PA. There were no consistent effects of CTQ variables on response rate. Demographic and treatment variables (age, gender, ketamine dose, schedule), diagnosis and concurrent medications had minimal or inconsistent influence on the time × load interaction. Per-protocol analyses excluded patients who did not return (n = 32) or changed schedules (n = 20); intent-to-treat analyses (last-observation-carried-forward, n = 115) produced results consistent with per-protocol findings. The extraction reports that patients with clinically significant physical neglect were less likely to be excluded (X2 = 6.54, p = 0.011).

Contrary to their initial hypotheses, O'brien and colleagues report that in this naturalistic TRD sample patients with histories of childhood maltreatment not only benefited from IV ketamine but in some analyses showed greater symptom reductions and higher remission rates than those without clinically significant maltreatment. Sexual abuse was particularly associated with greater benefit after a single infusion, while physical abuse and higher maltreatment load were more clearly linked to larger improvements after repeated infusions. The investigators emphasise that maltreatment load—the number of clinically significant CTQ domains—appeared a better predictor of treatment outcome than any single subscale. The authors propose a mechanistic explanation centred on trauma-induced behavioural sensitization. They note that preclinical and clinical evidence implicates NMDAR-dependent processes in the induction and expression of sensitization to stress and that NMDAR antagonists, including ketamine, can block sensitization in animal models and reduce hyperarousal and depressive symptoms in related human conditions. From this perspective, ketamine might counteract a neurobiological expression of early-life sensitization that contributes to resistance to conventional monoaminergic antidepressants. The investigators also highlight pragmatic findings about dosing schedules: improvements were observed after the first and second infusions, with a plateau thereafter, suggesting that an initial twice-weekly schedule for two to three infusions followed by once-weekly maintenance could balance efficacy and burden. The authors acknowledge several important limitations. The sample size—particularly for the highest maltreatment-load subgroups—was modest, limiting power for some subgroup and moderator analyses. Key demographic and clinical variables such as education, socioeconomic status, family history and detailed medication histories were unavailable. The CTQ is a retrospective self-report instrument that omits other relevant early or adult traumas and may be subject to recall bias. The naturalistic clinic setting, lack of randomisation or placebo control, and concurrent psychotropic medications limit causal inference and generalisability. The authors therefore frame their conclusions cautiously and call for further research, including development of objective markers of sensitization that might guide physiologically informed treatment choices.

The study authors conclude that, in this naturalistic TRD cohort, patients with high self-reported childhood maltreatment showed at least comparable and in some analyses greater antidepressant benefit from IV ketamine than patients with low or no maltreatment. They suggest that ketamine could be considered earlier than additional conventional antidepressant augmentation strategies for TRD patients with substantial childhood trauma burden. The authors also propose that an initial course of two to three twice-weekly infusions followed by once-weekly maintenance may optimise response while limiting treatment burden. They note there is no established cut-off for total CTQ score to guide clinical decisions and that their findings require replication in larger, controlled studies.