This double-blind, placebo-controlled study (n=154) of ketamine (35mg/70kg) (or placebo) with (and without) positive self-regard training (automated self-association training) finds that the combination can extend the positive antidepressant (MADRS) effects of ketamine, whilst the effects of ketamine alone was not distinguishable from placebo 30 days later.
Papers cited by this study that are also in Blossom
Objective
Intravenous ketamine, which displays rapid antidepressant properties, is posited to reverse depression by rapidly enhancing neuroplasticity. The authors tested whether an automated, computer-based approach could efficiently leverage enhanced neuroplasticity to extend the durability of rapid clinical response.
Methods
A total of 154 adults (ages 18-60) with treatment-resistant unipolar depression were randomized in a double-blind, parallel-arm design to receive an active/active treatment combination (ketamine plus active “automated self-association training” [ASAT]; N=53) or one of two control arms that lacked either the active drug component (saline plus active ASAT; N=51) or the active behavioral component (ketamine plus sham ASAT; N=50). One day after a single infusion of intravenous ketamine (0.5 mg/kg over 40 minutes) or inert placebo (saline), active ASAT-targeting self-worth through automated “evaluative conditioning” training delivered by computer-or sham ASAT (consisting of identical computer tasks that included no positive or self-referential stimuli) was given, delivered twice daily over 4 consecutive days (eight sessions, ≤20 minutes per session). The prespecified primary outcome measure throughout the main (30-day) study period was score on the Montgomery-Åsberg Depression Rating Scale (MADRS).
Results
Ketamine rapidly and significantly reduced depression scores at 24 hours postinfusion (group-by-time interaction: standardized beta [β]=−1.30, 95% CI=−1.89, −0.70; t=−4.29, df=150). In intent-to-treat linear mixed models, depression scores in the ketamine+ASAT group remained significantly and stably low over the 30-day study period relative to those of the saline+ASAT group (β=−0.61, 95% CI=−0.95, −0.28; t=−3.62, df=148). By contrast, depression scores following ketamine+sham treatment followed a significant, increasing linear trajectory from 24 hours to 30 days, approaching the levels observed in the saline+ASAT group (group-by-time interaction relative to the saline+ASAT group: β=0.015, 95% CI=0.003, 0.03; t=2.35, df=568).
Conclusions
After priming the brain with ketamine, training positive self-associations could provide an efficient, low-cost, portable, noninvasive, and highly dissemination-ready strategy for leveraging and extending ketamine’s rapid antidepressant effects.
Depression remains highly prevalent and costly, with low remission rates and high relapse despite existing treatments; only a minority of patients receive minimally adequate care. Two recent developments have generated clinical interest: rapid-acting pharmacological agents such as intravenous ketamine, which can reduce symptoms within 2–24 hours, and automated, computer-based interventions designed to alter affective processing to increase access, lower cost, and reduce burden. However, ketamine's antidepressant effects after a single infusion typically dissipate within days to weeks, and repeated infusions raise safety and feasibility concerns. The researchers designed a study to test a synergistic biobehavioural strategy: use a single ketamine infusion to transiently increase neuroplasticity and thereby create a ‘‘window of opportunity’’ during which brief, fully automated training could consolidate adaptive cognitive-affective learning and extend ketamine's antidepressant benefit. Specifically, they developed an automated self-association training (ASAT) intervention grounded in evaluative conditioning to promote positive implicit self-worth, and compared the combined ketamine+ASAT approach against two control conditions that removed either the pharmacological or behavioural component, enabling a direct test of whether ketamine and ASAT act complementarily to produce rapid and durable symptom improvement.
Design and participants: The trial was a double-blind, parallel-arm randomised clinical trial enrolling 154 adults aged 18–60 with treatment-resistant unipolar major depressive disorder, defined by DSM-5 criteria, moderate-to-severe depression (MADRS score ≥25), low self-esteem on standard scales, and at least one failed adequate trial of an FDA-approved antidepressant in the current episode. Existing treatments were required to be stable from ≥4 weeks before screening through the 30-day study. Enrollment continued until 150 participants met minimal completion criteria (infusion day, 24-hour assessment, and ≥75% of ASAT sessions); 148 participants completed all eight ASAT sessions per protocol. Randomisation used permuted blocks, stratified by sex and degree of treatment resistance, with a 2:1 allocation to ketamine versus saline and further assignment to ASAT versus sham. The three arms were: ketamine+active ASAT (N=53), saline+active ASAT (N=51), and ketamine+sham ASAT (N=50). Interventions: Infusions were administered in a hospital setting by blinded nurses; ketamine was given as 0.5 mg/kg intravenous over 40 minutes, and placebo was 50 mL 0.9% saline. Safety monitoring and four-hour postinfusion observation were provided; adverse events were reported as mild-to-moderate and time-limited. ASAT began 24 hours postinfusion and consisted of eight computerised sessions (twice daily for 4 consecutive days; each session ~15–20 minutes, with ≥20-minute intersession intervals). Active ASAT used evaluative conditioning to pair self-referential stimuli (e.g., “I,” participant headshots) with positive verbal and pictorial unconditioned stimuli, presented both supraliminally and subliminally, together with incidental tasks (lexical decision, rapid mouse-tracking) to reinforce implicit positive self-associations. Sham ASAT matched the task format and nonspecific factors but used neutral/unrelated stimuli to avoid self-referential positive learning. Assessments and outcomes: The prespecified primary outcome was depressive symptom severity measured by the Montgomery-Åsberg Depression Rating Scale (MADRS), administered by a single experienced rater at screening (eligibility only), preinfusion baseline (infusion −1 hour), 24 hours postinfusion (prior to ASAT), pre-ASAT (infusion +24 hours/124 hours?), and at days 5, 12, 21, and 30. Dichotomous outcomes (for descriptive purposes) included 24-hour responders (≥50% reduction in MADRS) and remitters (MADRS ≤9). A one-year naturalistic follow-up and additional neurocognitive/neuroimaging outcomes were planned but not reported here. Statistical analysis: Intent-to-treat linear mixed-effects regression models were used with continuous MADRS as the outcome, days since infusion as a random within-subject effect, and treatment allocation as a fixed effect. Two primary analyses were specified: an infusion-phase contrast (ketamine vs. saline) using preinfusion and 24-hour assessments, and an ASAT-phase contrast (saline+ASAT vs. ketamine+sham vs. ketamine+ASAT) using 24-hour through day-30 assessments, with preinfusion MADRS as a covariate in the latter. Models included random intercepts and slopes for participants to account for repeated measures and missing data (missingness was minimal, 5.1% of observations). Continuous variables were standardised; results are reported as standardised beta coefficients (b), which index standard-deviation changes akin to Cohen's d. Sensitivity analyses tested robustness to covariates (sex, age, treatment resistance severity, psychotropic medication use). Power calculations targeted 80% power to detect moderate effects, with the study intentionally allocating more resources to the combined active arm.
A total of 154 patients (ages 18-60) were randomized. All patients met DSM-5 criteria for major depressive disorder (see the online supplement), and all patients reported moderate to severe levels of depression (a score $25 on the Montgomery-Åsberg Depression Rating Scale [MADRS]), lower-than-normative self-reported self-esteem (i.e., scoring outside of one standard deviation from the normative mean on the Cognitive Triad Inventory"self" subscaleand/or the Rosenberg Self-Esteem Scale), and at least one unsuccessful adequate trial of an antidepressant medication approved by the U.S. Food and Drug Administration (FDA) in the current depressive episode (assessed with the Antidepressant Treatment Response Questionnaire [ATRQ]). Any existing depression treatment regimens were required to be stably maintained beginning $4 weeks before screening (which equated to roughly 6 weeks before infusion date) and throughout the 30-day trial. See the online supplement for further rationale and details of all inclusion and exclusion criteria, screening assessments, and study ethical oversight. See Figurefor the CONSORT flow diagram, which reflects high adherence and retention, and Tablefor descriptive patient characteristics. To achieve the initial recruitment target of a sample of 150 patients, the study enrolled and randomized 154 patients, continuing enrollment until a final sample was acquired that included the full target of 150 patients who completed, at a minimum, an infusion day, a 24-hour assessment visit, and $75% of intended training (active or sham ASAT) visits, and were thus considered to have received their ASAT allocation (and labeled accordingly in the CONSORT flow diagram). All but two participants who surpassed this 75% threshold for their ASAT allocation received the full, eightsession dose of intended ASAT visits, per protocol (N5148 for full, per-protocol completers who received all allocated ASAT sessions). Thus, 96% of randomized participants were fully adherent with all intervention sessions, and 97% of all intended ASAT sessions were delivered. A three-arm study design was selected to allocate all available resources toward the active/active (ketamine1 ASAT) treatment arm relative to two crucial comparator groups, comprising each intervention component in the absence of the other-a conservative test of the active/active biobehavioral intervention's combined impact. An inactive/ inactive (no-treatment) arm was forgone in order to maximize statistical power for these more conservative and critical comparisons. An a priori power analysis was based on a principal interest in clinically meaningful, moderate (or larger) effects. The target sample sizes were selected to yield 80% power to detect moderate effects based on infusion phase comparisons (see "Statistical Analysis" below) of 100 ketamine and 50 saline patients (d$0.49 and r$0.31 using an alpha of 0.05), and based on comparisons of 50 patients in each of the ketamine1ASAT, ketamine1sham, and saline1 ASAT conditions during the ASAT phase, where effects of d$0.57 would be detectable with 80% power (using an alpha of 0.05). Power for primary mixed-effect analyses comparing treatment groups, which increases with additional repeated measures, was anticipated to be higher than for these simplified, single-point contrasts, making these power calculations conservative.
Create a free account to open full-text PDFs.
Newport, D. J., Carpenter, L. L., Mcdonald, W. M. et al. · American Journal of Psychiatry (2015)
Abdallah, C. G., Averill, L. A., Gueorguieva, R. et al. · Neuropsychopharmacology (2020)
Dakwar, E., Levin, F. R., Hart, C. L. et al. · American Journal of Psychiatry (2020)
Dakwar, E., Nunes, E. V., Hart, C. L. et al. · American Journal of Psychiatry (2019)
Wilkinson, S. T., Taeho, G., Rhee et al. · Psychotherapy and Psychosomatics (2021)
Murrough, J. W., Collins, K. A., Fields, J. et al. · Translational Psychiatry (2015)
Murrough, J. W., Iosifescu, D. V., Chang, L. C. et al. · American Journal of Psychiatry (2013)
Papers in Blossom that reference this study
Tucciarone, J. M., Bandeira, I. D., Blasey, C. et al. · American Journal of Psychiatry (2026)
Aepfelbacher, J., Panny, B., Price, R. · Biological Psychiatry (2024)
Lii, T. R., Smith, A. E., Flohr, J. R. et al. · Nature Mental Health (2023)
Van Elk, M., Fried, E. I. · Therapeutic Advances in Psychopharmacology (2023)
Goodwin, G. M., Malievskaia, E., Fonzo, G. A. et al. · American Journal of Psychiatry (2023)
Price, R., Wallace, M. L., Mathew, S. J. et al. · JAMA Network Open (2023)
Sample and adherence: A total of 154 participants were randomised into three arms (ketamine+ASAT N=53; saline+ASAT N=51; ketamine+sham N=50). Per-protocol, 148 participants completed all eight ASAT sessions; overall, 96% of randomised participants were fully adherent and 97% of intended ASAT sessions were delivered. Missing outcome observations were minimal (5.1%). Adverse events recorded were characterised as mild to moderate and time-limited, consistent with prior ketamine trials. Infusion phase (pre-ASAT): Ketamine produced a rapid and statistically significant reduction in MADRS scores at 24 hours compared with saline. The group-by-time interaction yielded a standardised beta b = −1.30 (95% CI −1.89 to −0.70; t = 24.29; df = 150; p < 0.0001). Clinically, 24-hour response rate was 52% in ketamine-treated participants versus 25% in saline-treated participants; remission rates were 28% versus 4%. The number needed to treat (NNT) for response was 3.7 and for remission 4.2. ASAT phase (24 hours to day 30): In intent-to-treat linear mixed models spanning 24 hours through day 30, the ketamine+ASAT group maintained significantly and stably lower MADRS scores than the saline+ASAT reference group (b = −0.61, 95% CI −0.95 to −0.28; t = −3.62; df = 148; p = 0.0004), with no significant group-by-time interaction (b = 0.009, 95% CI −0.004 to 0.021; t = 1.39; p = 0.164), indicating durability across the 30-day window. By contrast, the ketamine+sham group showed a significant, increasing linear trajectory of symptoms from 24 hours to 30 days (i.e., waning benefit) approaching the saline+ASAT group's levels; the group-by-time interaction for ketamine+sham relative to saline+ASAT was b = 0.015 (95% CI 0.003 to 0.03; t = 2.35; df = 568; p = 0.019). Point estimates at day 30: Post hoc pairwise contrasts at day 30 produced small effect-size point estimates favouring ketamine+ASAT: versus saline+ASAT b = −0.38 (95% CI −0.78 to 0.027) and versus ketamine+sham b = −0.31 (95% CI −0.75 to 0.13). Sensitivity analyses adjusting for prespecified covariates (sex, age, treatment resistance severity, concomitant psychotropic medications) upheld the primary findings. No significant moderating interactions were observed between covariates and treatment allocation.
The researchers conclude that brief, fully automated ASAT extended the rapid antidepressant effect of a single ketamine infusion across a 30-day acute follow-up period. They replicated ketamine's well-established rapid antidepressant effect at 24 hours and found that when ketamine was followed by active ASAT, depressive symptoms remained stably low relative to the saline+ASAT group. In contrast, ketamine followed by sham ASAT showed the expected pattern of rapid improvement followed by gradual symptom return over weeks, consistent with prior single-infusion ketamine studies. The authors interpret these results as evidence that ketamine may open a ‘‘window of opportunity’’ during which behavioural learning—here, implicit positive self-association training—can be efficiently taken up and consolidated, thereby prolonging symptomatic benefit. They note that saline+ASAT participants also showed MADRS scores that remained below pretreatment baseline across 30 days, suggesting either some standalone effect of ASAT on implicit cognition or nonspecific effects of the clinical research context (or both). The three-arm design (including both no-ketamine and no-behavioural-intervention comparators) allowed a conservative test of the combined intervention's necessity and provided novel evidence for additive or complementary action of ketamine and ASAT, but it does not permit inference about efficacy compared with no treatment. Limitations and uncertainties the authors acknowledge include functional unblinding for the infusion condition, which may have amplified ketamine's observed effects via expectancy; by contrast, ASAT condition blinding was adequate, so expectancy is unlikely to explain ASAT's extending effect. The study was not designed to address dose–response relationships, optimal scheduling, or whether repeated ketamine doses, additional ASAT 'boosters', or targeting other cognitive-affective associations would yield larger or more durable effects. Generalisability may be limited by the trial sample's demographic composition and eligibility criteria; applicability to patients with differing degrees of treatment resistance, bipolar depression, or moderate–severe substance use disorders is uncertain. Finally, although ASAT was delivered in the lab for this initial test, the authors suggest its automated, low-cost, portable nature supports potential scalability and dissemination, and they highlight high adherence (97%) and prior acceptability data for automated therapies as promising for future translation. Implications discussed by the authors include pursuing replication, exploring different dosing and delivery schedules (including booster sessions and repeated pharmacological priming), assessing broader or modular cognitive targets beyond self-worth, and conducting longer-term naturalistic follow-up to characterise durability and time to relapse. They also emphasise the broader strategy of exploiting transiently enhanced neuroplastic states induced by rapid-acting agents to introduce protective learning as a potential scalable approach to treating depression.
The authors argue that, given the substantial global burden of depression and the need for efficient, scalable treatments, the combined approach of a single ketamine infusion followed by brief, automated ASAT may represent a promising, low-cost, portable method to rapidly induce and extend symptom relief. They caution that replication is required, but suggest that ketamine and automated training techniques could be dissemination-ready tools to capitalise on transient neurobiological states (enhanced plasticity) as clinical windows in which to introduce protective learning, meriting further research and development.
a Patients are labeled as having received their allocated infusion if they initiated the infusion. For one single patient, who was allocated to the keta-mine1sham ASAT arm, the infusion was discontinued after administration of approximately two-thirds of the intended dose, at the patient's request. Patients are labeled as having received their allocated ASAT condition (active or sham) if they completed $75% of the intended eight sessions. All but two of the patients labeled as having received their ASAT allocation completed the entire set of eight sessions, per protocol.
In this primary clinical report from the trial, the focus is on the combined intervention's impact on MADRS score, the study's prespecified primary clinical endpoint for the acute (30-day) study period, to assess overall depression severity across the 30-day period when all other psychiatric treatments were held constant. A single, experienced master'slevel clinical rater administered the MADRS. The scale was administered at screening (used for eligibility determinations only), preinfusion baseline (infusion 21 hour; the clinical rater then left and was not present during any intervention procedure), infusion 124 hours (prior to the first session of active/ sham ASAT), and at 5, 12, 21, and 30 days. A 1-year exploratory naturalistic follow-up, utilizing remote assessments acquired strictly through a distinct self-report symptom scale (the Quick Inventory of Depressive Symptomatology-Self-Report), remains under way, and data are not included in the present report. Neurocognitive/neuroimaging assessments of ketamine's impact are undergoing analysis and will be included in forthcoming publications, along with secondary outcomes; however, one such cognitive measure (a "target engagement" measure of implicit self-associations) is analyzed in the online supplement. At the final in-person visit (at 30 days), the adequacy of both patient and rater blinding, for both infusion and ASAT allocation, was assessed; see the online supplement.
A randomization table using permuted blocks of three or six, stratified by 1) biological sex assigned at birth and 2) modest versus severe treatment resistance (severe was defined as having had at least three unsuccessful FDA-approved antidepressant trials in the episode, per the ATRQ), was generated using SAS (SAS Institute; Cary, N.C.) by an independent statistical consultant at study start and was maintained by the study pharmacist, who, at the point of randomization, conveyed the ASAT allocation (but not the infusion allocation) to a single study staff member assigned to that patient. This single staff member was tasked with opening the correct computer program to deliver the ASAT sessions to that patient, but was not involved in, or present for, any outcome ratings. Blinding of all other study personnel was maintained until after the final patient in the trial had completed their day 30 assessment. Participants were randomized in a 2:1 ratio to receive ketamine (0.5 mg/kg) or saline (50 mL 0.9% sodium chloride), administered over a 40-minute infusion, as in previous studies (e.g., 51). All infusions were administered by blinded, licensed nurses in a medical hospital setting with immediate, emergency access to an Advanced Cardiovascular Life Support-certified team, oversight by a blinded study physician (R.H.H.), and safety and adverse event monitoring sustained for 4 hours after infusion. Adverse events (see Tablein the online supplement) were mild to moderate, time limited, and consistent with the safety profile described in numerous previous trials using identical infusion procedures.
Drawing on previous work utilizing "evaluative conditioning" to influence participants' preferences and "liking" (39, 40), active ASAT was designed to leverage Pavlovian conditioning in order to promote positive implicit selfassociations and self-worth. Eight 15-to 20-minute sessions (delivered in a private research office setting, twice daily for 4 consecutive days, spaced by an intersession interval $20 minutes) were initiated 1 day postinfusion (after the 24-hour postinfusion assessment). In each of the eight sessions, both verbal and pictorial stimulus pairings were presented, both supraliminally ($250 ms) and subliminally (12 ms), to practice and reinforce implicit associations between positive traits (e.g., "sweet," "attractive," photos of smiling actors; the unconditioned stimuli [US]) and selfreferential stimuli (e.g., "I," participant headshots; the conditioned stimuli [CS]). Incidental tasks such as a "lexical decision" task (indicating whether targets were real words or random letter strings) and a rapid mousetracking task (clicking as fast as possible on the position of stimuli) were used to enhance engagement and to promote semantic and visual processing of stimuli. Similar forms of evaluative conditioning have been found to alter implicit self-esteem, mood reactivity to stress (52), and pathological behaviors (anxious avoidance, selfinjury). Sham ASAT consisted of the exact same computer tasks, but with predominantly neutral rather than positive US and non-self-relevant CS (words related to "others," pictures of gender-matched strangers), designed to eliminate the possibility of unintended self-referential or iatrogenic negative learning, while providing a credible "brain-training" paradigm that controlled for all nonspecific factors (e.g., exposure to lexical and pictorial stimuli with a range of affective valences; incidental task demands; and time spent in the research setting) and was effective in facilitating adequate patient blinding (see Tablein the online supplement). See the online supplement for further technical details on the active and sham ASAT procedures.
Intent-to-treat linear mixed-effects regression models were applied with continuous MADRS scores as the outcome, days since infusion as a random, continuous within-subject effect, and treatment allocation as a fixed, between-subject effect. Separate models were designed to test two discrete hypothesized linear patterns, based on 1) the infusion phase of intervention (two-way split: ketamine vs. saline), using preinfusion and 24-hour assessments only (assessments collected prior to onset of active/sham ASAT); and 2) the ASAT phase of intervention (three-way split: saline1ASAT vs. ketamine1sham vs. ketamine1ASAT), using assessment points from 24 hours (as a pre-ASAT baseline for these trajectories) through day 30. In the infusion phase, we hypothesized rapid symptom decreases in the ketamine arm (relative to saline). For descriptive and clinical characterization purposes only, to facilitate comparisons with previous studies, dichotomous outcomes were defined for 24-hour responders ($50% decrease from preinfusion MADRS baseline score) and remitters (MADRS score #9). In the ASAT phase, the saline1ASAT arm was the reference group for hypothesis tests, and models included preinfusion MADRS scores as a covariate. We hypothesized a stable (main) effect extending from 24 hours to day 30 in the ketamine1ASAT arm (relative to saline1ASAT), with no group-by-time interaction. By contrast, for the ketamine1sham group, we hypothesized a linearly increasing trajectory of symptoms from 24 hours to day 30 (i.e., a group-by-time interaction relative to the reference group), consistent with the dissipation of symptom improvements within 1-2 weeks following a single infusion of ketamine (in isolation) that has been reported in previous ketamine trials. All models included a random intercept and slope for participant to model patient-level trajectories over time and automatically account for missing data, which were minimal (5.1% of all intended observations). For interpretability, continuous variables were standardized. We report standardized beta coefficients (b) as a measure of effect size reflecting the number of standard deviations in the dependent measure that correspond to a one-unit change in the independent measure (e.g., group contrast), akin to Cohen's d, with 95% profile likelihood confidence intervals. Analyses were performed using R, version 3.6. Sensitivity analyses (see the online supplement) probed for the robustness of findings when including the following covariates selected a priori: sex, age, treatment resistance (moderate vs. severe), and use of concomitant psychotropic medications (dichotomized as yes/no).
Infusion phase. Consistent with previous reports, ketamine rapidly reduced MADRS total depression scores at 24 hours postinfusion (group-by-time interaction: standardized beta [b]521.30, 95% CI521.89, 20.70; t524.29, df5150, p,0.0001) (Figure). This moderate to large linear effect corresponded to a 52% response rate and a 28% remission rate for the ketamine-treated participants, compared with a 25% response and 4% remission rate for the saline-treated participants (the number needed to treat for response was 3.7, and for remission, 4.2). ASAT phase. In intent-to-treat linear mixed models, depression scores in the ketamine1ASAT group remained stably low over the 30-day acute phase relative to those in the saline1ASAT group (b520.61, 95% CI520.95, 20.28; t523.62, df5148, p50.0004) (Figure), with no corresponding group-by-time interaction (b50.009, 95% CI520.004, 0.021; t51.39, df5568, p50.164), suggesting durability of effect over the 30-day window. By contrast, depression scores in the ketamine1sham group followed an increasing linear trajectory from 24 hours to 30 days, as the postketamine effect gradually waned and depression score levels approached those of the saline1ASAT group (group-by-time interaction relative to the saline1ASAT group: b50.015, 95% CI50.003, 0.03; t52.35, df5568, p50.019). While the above intent-to-treat statistical tests of linear trajectories over the full 30-day period were used for statistical hypothesis tests, pairwise post hoc contrasts at day 30 were used to generate effect size point estimates for further description of clinical impact at this specific time point. These effect size point estimates suggested a small effect favoring ketamine1ASAT relative to both saline1ASAT (b520.38, 95% CI520.78, 0.027) and ketamine1sham (b520.31, 95% CI520.75, 0.13). All the findings reported above were robustly upheld in sensitivity analyses controlling for covariates (see the online supplement). No significant interaction (moderating) effects were observed between covariates and treatment allocation for either study phase (see the online supplement).
In this study, we found that automated self-association training-a novel, low-cost, fully automated, noninvasive, brief (eight sessions, #20 minutes per session), computerbased intervention-extended the rapid antidepressant effect of a single ketamine infusion for at least 30 days. In one of the largest ketamine randomized controlled trial samples to date, we replicated the well-established finding that intravenous ketamine exerts a rapid antidepressant effect in patients with treatment-resistant depression. Ketamine infusion appeared also to open a clinical window of opportunity, enabling the efficient uptake of positive self-representations (which were evident at the level of implicit cognition following ASAT; see the online supplement) to protect against the return of depression over the subsequent month (Figure). Symptoms in the saline1ASAT group, which received identical ASAT procedures in the absence of pretreatment with ketamine, were elevated relative to those in the ketamine1ASAT group throughout the 30-day follow-up period. Of note, symptoms following saline1ASAT also remained stably below pretreatment baseline, which could be attributable to a stand-alone impact of ASAT procedures (which were effective at altering implicit cognition, as noted above), or to an enduring, nonspecific impact of the overarching clinical research context (e.g., repeated depression assessments and staff interactions), or both. A second comparator group, treated with ketamine followed by sham ASAT, followed the expected trajectory when a single ketamine infusion has been given as a stand-alone treatmentrapid decrease followed by gradual return of depressive symptoms over the ensuing weeks (4-6). To our knowledge, this represents the first study to test a biobehavioral pairing of ketamine with a behavioral intervention that has included both a no-ketamine and a no-behavioral-intervention control condition. Likewise, clinical trials testing a wide range of posited synergistic treatments (e.g., behavioral treatments paired with "psychoplastogens" or other brain-based targeted approaches, e.g., neuromodulation) have routinely neglected one or the other of these two critical comparators. Our design affords pivotal, novel evidence for the combined benefit of ketamine's acute impact and the introduction of protective learning via ASAT, relative to patients' pretreatment baseline symptoms and to either intervention component when given in isolation-providing a conservative test of efficacy. However, our design cannot provide estimates of the impact of the combined intervention (nor of each intervention component alone) relative to no intervention at all. While we expect that a range of traditional and nontraditional behavioral treatments might work synergistically with rapid-acting pharmacotherapy, here we focused specifically on our novel ASAT approach, for both practical and scientific reasons. Practically, ASAT is fully automated and portable, and designed to be efficient, fitting well within the brevity (i.e., about 1 week) of ketamine's posited window of opportunity. By contrast, a full course of either traditional or computer-based cognitive-behavioral therapy (CBT) typically requires 12 to 16 1-hour sessions delivered weekly, and meta-analyses suggest that optimal computer-based CBT requires supplemental therapist coaching. Here, 30-40 minutes of ASAT for 4 consecutive days yielded effect sizes for ketamine1ASAT (relative to unimodal treatment conditions) that were moderate across the entire follow-up interval, but small in point estimates at the final assessment point (day 30). Further work will be needed to understand the patient perspective on the cost-benefit ratio of such gains visà-vis other, conventional treatment options (e.g., psychotherapy) and to assess adherence in clinical settings. However, we observed exceptional adherence (97%) with ASAT sessions in our research context, which is consistent with the high adherence and high ratings for acceptability and satisfaction observed previously for other fully automated therapies, across both research and clinical contexts. Scientifically, a robust evidence base suggests that implicit and explicit forms of cognition are distinct entities, and that implicit cognition (e.g., implicit self-concept) could be a more robust predictor of future behavior than explicit thought content. ASAT is designed to efficiently and directly target implicit cognition, yielding an innovative approach with the potential to profoundly shape future behavior and experience. Furthermore, by targeting a unitary and welldefined implicit cognitive mechanism, ASAT limits the influence of heterogeneous and nonspecific factors that are likely influential in many traditional behavioral interventions, and for which the learning that occurs within a brief window may be more difficult to predict and constrain (e.g., positive vs. negative therapy session experiences). Broadly, evaluative conditioning techniques have a large and robust literature in healthy control subjects, and in clinical samples, they have previously shown replicated acute clinical effects even in an at-home (smartphone) delivery modality. Although in this initial test of ASAT we administered the training in a research lab environment, the ability to both achieve and maintain gains via portable, highly dissemination-ready techniques is a crucial criterion for overcoming intractable barriers to access to care, particularly for underserved communities, representing a critical future direction for this work. Given budgetary constraints, all available resources were allocated to determining the advantage of the active/active biobehavioral treatment combination over its component parts delivered in isolation, providing a conservative test of efficacy focused on establishing the necessity of each active component, but negating the ability to determine intervention efficacy in relation to no treatment. Functional unblinding was evident for the infusion condition (see Tableand further discussion in the online supplement); clinical effects observed for ketamine versus saline may thus have been amplified by expectancy. Importantly, because adequate ASAT condition blinding was achieved (see Tablein the online supplement), such effects are unlikely to have factored strongly into the subsequent extending effect of ASAT-the primary, novel focus of the present work. While ASAT achieved its hypothesized effect of extending the initial rapid improvements induced by ketamine, it did not produce incremental improvements over and above ketamine's rapid effect, and response and remission rates leave substantial room for further improvement of efficacy. Our design cannot directly inform key clinically relevant questions regarding dose-response effects, treatment schedule optimization, and the trade-off between treatment efficiency and obtaining (and sustaining) maximal benefit. We erred on the side of front-loading and maximizing ASAT administration during the acute postketamine window of opportunity, within the perceived confines of feasibility and patient burden, while preserving spaced practice (i.e., 20-minute intersession breaks) in an effort to enhance learning. While this initial effort focused on quantifying 30-day symptom trajectories following this highly efficient (5-day) intervention package, future work involving repeated ketamine doses, higher overall doses of ASAT, and/or subsequent ASAT booster sessions would expand the clinical relevance of the present work. Our ongoing, exploratory, naturalistic 1-year follow-up could yield further insights regarding durability and time to relapse, but for ethical reasons we did not ask patients to refrain from freely making treatment changes and additions during follow-up. Future work could also assess the feasibility and impact of targeting additional cognitive-affective associations beyond self-worth (e.g., perceptions of others, the future, etc.), which might enhance effect sizes and/or enable a personalized, modular approach that could be applicable transdiagnostically. Finally, although clinical trial eligibility criteria were designed to emulate real-world treatment-seeking depressed samples (e.g., stable medications maintained rather than discontinued, and few exclusions based on medical or psychiatric comorbidities), the results may not robustly generalize to patients with distinct characteristics, including those with greater or lesser degrees of treatment resistance, with bipolar depression, or with comorbid moderate to severe substance use disorders. Critically, we are also unable to elucidate what the intervention's efficacy and feasibility might be in samples with greater racial and ethnic diversity than ours had.
The global burden of depression is extremely high and is expected to continue to increase within the current context of a significant pandemic. There is an urgent need for novel treatment approaches, particularly those that can provide relief efficiently and at scale. If the present results can be replicated, this novel, integrative treatment may provide a method to urgently bring relief and to efficiently extend this relief via safe, low-cost, portable techniques. Alongside our fully automated ASAT intervention, ketamine infusion may likewise be more dissemination ready than many alternatives, with a favorable safety profile for isolated infusions and wide international medical usage in both anesthetic and subanesthetic applications. Additional efforts are warranted to create, and then exploit, rapidly induced neurobiological states (such as enhanced neuroplasticity) as clinical windows of opportunity in which to introduce new, protective learning.