The effects of ketamine and classic hallucinogens on neurotrophic and inflammatory markers in unipolar treatment-resistant depression: a systematic review of clinical trials

Rossi and colleagues conducted a systematic review in accordance with PRISMA guidelines. Electronic searches were performed in PubMed, Web of Science, Embase, SciELO and LILACS up to 5 May 2022 using a search string combining terms for ketamine, classical hallucinogens and a range of neurotrophic and inflammatory markers. No language restrictions were applied. Inclusion criteria required clinical trials enrolling adult patients with unipolar TRD, administration of ketamine or a classical hallucinogen, and pre- and post-treatment measurement of neurotrophins and/or inflammatory biomarkers in any biological matrix. Studies with mixed diagnostic samples beyond unipolar TRD (with some exceptions for comorbid anxiety) were excluded. Two reviewers independently screened titles/abstracts and full texts, resolving disagreements by consensus or third-party adjudication. Extracted variables included study design, TRD definition, participant characteristics (sample size, sex, age, drug-free or add-on status), treatment details (drug, dose, regimen), biomarkers and assay methods, depression scales, and main statistical results including p values and whether corrections for multiple comparisons were applied. Quality assessment was performed by two independent reviewers using National Heart, Lung and Blood Institute Study Quality Assessment Tools: the ‘‘Before-After (Pre-Post) Studies’’ tool for open-label trials and the ‘‘Quality Assessment of Controlled Intervention Studies’’ tool for randomized controlled trials (RCTs). Where disagreements arose, a third author made the final decision. The authors synthesised study characteristics and findings in narrative form and in summary tables. If details were not clearly reported in the extracted text (for example exact randomisation methods or some sample descriptors), the review did not invent them and noted such reporting deficiencies in the quality assessment.

Study selection and characteristics The search yielded 3,313 records, of which 26 were read in full and 13 trials met the inclusion criteria. Two trials evaluated a classical hallucinogen (oral ayahuasca, 1 mL/kg) but derived from the same original randomized, double-blind, placebo-controlled trial. The remaining 11 trials administered ketamine, most commonly as a single 0.5 mg/kg intravenous infusion. Across included studies total sample sizes per study ranged from 22 to 73 (mean 48.3 ± 17.6) and the aggregate sample was reported as 617 participants, of whom 415 (67%) were TRD patients. Considering only TRD patients, samples ranged 16–71 (mean 31.1 ± 13.8). Study designs included open-label trials, RCTs with parallel groups and placebo-controlled crossover trials. Classic hallucinogens (ayahuasca) Two reports from the same randomized, double-blind, placebo-controlled trial evaluated peripheral biomarkers 48 hours after a single ayahuasca dose. In the combined sample of TRD patients and healthy controls, ayahuasca-treated volunteers (n = 35) had higher serum BDNF at day 2 compared with placebo (n = 34), effect size d = 0.53 (p = 0.03, uncorrected). In TRD patients, serum BDNF at day 2 was negatively correlated with depressive symptom scores (MADRS; rho = -0.55, p ≤ 0.05, uncorrected). A separate analysis from the same trial found a decrease in plasma C-reactive protein (CRP) at day 2 after ayahuasca (adjusted for BMI, d = -1.37, p = 0.04, corrected), with CRP reductions in TRD patients correlating with lower MADRS scores (rho = 0.57, p ≤ 0.05, uncorrected). IL-6 showed no significant change. Ketamine and neurotrophic biomarkers Findings for BDNF and other growth factors after ketamine were mixed across studies. Machado-Vieira et al. (open-label, n = 23) reported no significant plasma BDNF changes across repeated early timepoints up to 230 minutes. Duncan et al. (open-label, n = 30) observed a significant increase in plasma BDNF at 230 minutes (p < 0.05, corrected). Haile et al. (randomized, active placebo-controlled; 0.5 mg/kg ketamine versus midazolam) found that responders had higher plasma BDNF at 240 minutes (p = 0.032, corrected) and plasma BDNF at 240 minutes was strongly negatively correlated with MADRS at multiple timepoints; these associations remained significant after controlling for age, BMI and gender. Allen et al. (open-label comparing ketamine and ECT) reported an increase in serum BDNF at 1 week in ketamine responders (p = 0.03, uncorrected, d = 1.08) but no sustained elevation at later times. Kadriu et al. reported no change in FGF-23. Kiraly et al., using a multiplex panel, found no change in BDNF with ketamine but reported decreases in G-CSF (240 min, p = 0.038, uncorrected) and PDGF-AA (24 h, p = 0.024, uncorrected); lower pretreatment FGF-2 was associated with response at 24 h (r = -0.565, p = 0.0009, uncorrected). A randomized crossover trial by Medeiros et al. (n = 39) detected no drug, time or interaction effects for plasma BDNF or VEGF. Jiang et al., assessing up to four infusions, reported no significant changes in mature BDNF or S100B. Ketamine and inflammatory biomarkers Results for cytokines and CRP were also heterogeneous. Yang et al. (open-label) reported that among responders IL-1β decreased at 230 minutes and 1 day (p = 0.013, corrected) and IL-6 decreased at 230 minutes and 3 days (p < 0.001, corrected). Kiraly et al. observed decreases at 240 minutes for several markers (IL-6, IL-1α, IL-13, IP-10; p values reported as significant) and at 24 h an increase in IL-7 with decreases in IL-8; these changes were reported as not correlated with clinical response. Chen et al. (randomized, double-blind, parallel; n = 71, add-on) compared 0.2 mg/kg, 0.5 mg/kg and placebo and found time-dependent changes: IL-6 was lower at early timepoints versus later days, and TNF-α was lower at 240 minutes compared with later days. Notably, in the 0.5 mg/kg group TNF-α at 40 and 240 minutes was lower than baseline, and the 40-minute decrease in TNF-α correlated with reductions in MADRS at days 4–5. Mkrtchian et al. (randomized, crossover) reported no significant main effects on plasma CRP. Overall, reductions in IL-6 were reported in two of three studies that measured it; TNF-α reductions were detected in one of three studies; IL-1β reductions in one of two; and neither of the two studies measuring CRP found consistent ketamine effects. Quality assessment Risk-of-bias concerns were substantial. Of the RCTs, only one reported an intention-to-treat analysis (intention-to-treat refers to analysing participants in the groups to which they were randomised regardless of adherence). Most RCT reports did not describe randomisation methods or report a priori sample size calculations. Open-label trials frequently lacked complete reporting on participant selection and enrolment, and only one study reported an a priori sample size calculation. Several biomarker analyses were performed without correction for multiple comparisons in some studies, increasing the risk of false-positive findings.

Rossi and colleagues interpret the assembled evidence as inconsistent and insufficient to support any peripheral neurotrophic or inflammatory marker as a reliable biomarker of clinical response to ketamine or classical hallucinogens in unipolar TRD. They note that ayahuasca produced increases in BDNF and reductions in CRP at 48 hours in the single trial reported, with biomarker changes correlating with reductions in depressive symptoms; however, both ayahuasca reports derive from the same original sample and thus replication is lacking. For ketamine, only three of seven studies measuring BDNF reported significant post-treatment increases, and correlations between BDNF changes and symptom improvement were observed in only one trial. Other neurotrophic factors (G-CSF, PDGF-AA, FGF-2) showed isolated associations but evidence is sparse and sometimes uncorrected for multiple comparisons. Inflammatory findings were likewise mixed: some trials reported post-ketamine decreases in IL-6, IL-1β or TNF-α in subsets of participants or timepoints, whereas others did not, and CRP results were generally null. The authors highlight several reasons for the heterogeneity and lack of confirmatory results. Peripheral biomarker levels may not reflect central nervous system (CNS) concentrations or processes. Methodological heterogeneity across trials—in assay methods, use of plasma versus serum, timing of blood draws, single-dose versus repeated administrations, add-on versus drug-free designs, and different statistical approaches including variable correction for multiple comparisons—complicates cross-study comparisons. Small sample sizes and incomplete reporting (for example missing details about randomisation, blinding of assays, and sample selection) increase the risk of type I and II errors. They also discuss biological complexity, pointing to possible depression subtypes (for example an inflammatory TRD subtype) and the multifactorial nature of BDNF regulation, which reduce the likelihood that a single peripheral measure will robustly index treatment response. The review authors therefore caution against overinterpretation of isolated positive findings and recommend that biomarker measurement in this field be pursued in adequately powered, well-controlled, ideally multi-centre trials with standardised protocols for sample collection and assay.

The authors conclude that, based on current clinical-trial evidence, none of the evaluated peripheral neuroplastic or immunological parameters can be used confidently as biomarkers of symptom improvement in unipolar TRD treated with classical hallucinogens or ketamine. They recommend that future biomarker studies be conducted in large-scale, multi-centre trials with sufficient sample sizes and statistical power to avoid continued contradictory and inconclusive results.