This double-blind controlled trial (n=61) found that high-dose LSD-assisted therapy (100μg + 200μg) reduced depression symptoms more than low-dose LSD (25μg + 25μg) in patients with moderate-to-severe major depressive disorder (MDD), with benefits lasting up to 12 weeks and similar side effects between groups.
Papers cited by this study that are also in Blossom
van Elk, M., Yaden, D. B. · Neuroscience and Biobehavioral Reviews (2022)
Background
This trial aimed to assess the efficacy of lysergic acid diethylamide (LSD)-assisted therapy in patients with moderate-to-severe major depressive disorder.
Methods
This was a randomized, parallel, double-blind, low-dose controlled trial (Clinicaltrials.gov: NCT03866252). Patients were randomly assigned in a 1:1 ratio to receive supportive psychotherapy and either 100 μg + 200 μg LSD or 25 μg + 25 μg LSD in two dosing sessions. The primary endpoints were the changes in scores on the Inventory of Depressive Symptomatology, in the Clinician-Rated (IDS-C) version (assessed by the treating therapist) and the Self-Rated (IDS-SR) version, from baseline to 2 weeks after the second administration. The IDS scores were also assessed 6 and 12 weeks after the second administration.
Findings
Thirty-one patients were randomized to the low-dose group, and 30 were randomized to the high-dose group. At the primary endpoint, least-squares mean change (LSM) in IDS-SR scores was −3.9 in the low-dose and −11.8 in the high-dose group (difference: −7.9; 95% CI, −16.0 to 0.3; effect size: −0.5; p = 0.059). LSM in IDS-C scores was −3.6 in the low-dose and −12.9 in the high-dose group (difference: −9.2; CI, −17.1 to −1.3; effect size: −0.6; p = 0.023; corrected <0.05). However, significance was not reached after adjusting for baseline depression scores (p = 0.086). Both outcomes remained numerically consistent up to the final follow-up at 12 weeks. Adverse events were comparable between groups.
Conclusions
The findings of this exploratory study support further investigation of LSD-assisted therapy in depression in a larger phase 3 trial.
LSD (lysergic acid diethylamide) is a classic hallucinogen that primarily acts as a serotonin-2A receptor agonist. Early uncontrolled studies from the mid-20th century suggested that a small number of LSD administrations could produce sustained antidepressant effects, but those studies had substantial methodological limitations. More recent controlled trials with related compounds such as psilocybin have shown antidepressant and anxiolytic effects after one or two high-dose administrations, and a previous trial reported lasting decreases in anxiety and associated depressive symptoms following two high doses of LSD in a palliative population. The precise mechanisms mediating these effects remain uncertain, and blinding has been a recurrent challenge in trials of hallucinogens.
The trial was powered with a target of at least 50 completers, yielding an intended enrolled sample of approximately 60 participants to allow for dropouts. The primary efficacy analyses used a restricted maximum likelihood mixed model for repeated measures (MMRM) on least-squares-mean (LSM) changes from baseline, with treatment, time point, and their interaction as fixed factors and an unstructured covariance. The Benjamini–Hochberg method was applied to control overall type I error across the two primary outcomes. Sensitivity analyses included MMRM adjusted for covariates (age, sex, weight, baseline antidepressant use and baseline score), two multiple-imputation approaches for missing data (pattern-mixture and reference-based), and an ANOVA approach specified in the original analysis plan. The full analysis set comprised randomised patients who received at least one dose and had valid baseline and at least one post-baseline assessment.
Sensitivity analyses yielded qualitatively similar directions of effect but with smaller estimated treatment effects and loss of statistical significance for several outcomes after covariate adjustment. Specifically, the adjusted LSM change for IDS-C at the primary endpoint was −6.8 (95% CI −14.7 to 1.0, SES −0.47, p = 0.086) and for IDS-SR was −6.1 (95% CI −14.0 to 1.8, p = 0.129). Baseline depression score was a significant predictor in adjusted models. Multiple imputation approaches did not materially alter these adjusted results. An ANOVA-based analysis generally agreed with the MMRM findings, although the IDS-SR at the primary endpoint reached significance in the ANOVA analysis.
Lysergic acid diethylamide (LSD) is a hallucinogenic drug that acts as a serotonin-2A receptor agonist.Several studies in the 1950s and 1960s suggested that a few administrations of LSD induce sustained antidepressant effects.However, firm evidence is lacking because of substantial limitations of these early studies. Modern trials on the related compound psilocybin indicated the CONTEXT AND SIGNIFICANCE This study investigates the effectiveness of LSD-assisted therapy for patients suffering from moderate-to-severe major depressive disorder. Depression is a widespread and debilitating condition, and exploring new treatment options is critical due to the limitations of existing therapies. The trial compared high-dose-LSD-assisted therapy with low-dose LSD combined with supportive psychotherapy, measuring changes in depressive symptoms over 12 weeks. Results showed that patients receiving higher doses of LSD experienced greater reductions in depression scores, suggesting potential therapeutic benefits. These findings highlight the promise of LSD-assisted therapy as a novel approach to treating depression and support further research in larger clinical trials to confirm its efficacy and safety for broader clinical use. efficacy of one or two doses in patients with major depressive disorder (MDD)and efficacy on depressive symptoms in terminally ill patients.Another recent trial demonstrated long-lasting decreases in anxiety and associated depressive symptoms after two administrations of high doses of LSD.The mechanism of action of these antidepressant and anxiolytic effects has not been elucidated. Some studies described associations between specific mental states and treatment outcomes,which might indicate that psychological processes mediate the clinical effects. Compared with classic antidepressants, these drugs are fast acting and have sustained antidepressant effects after limited administrations, thus suggesting advantages with regard to chronic side effects that often pose a challenge for patient compliance with conventional antidepressants.The treatment of MDD with LSD has not yet been investigated in a randomized trial. Here, we assessed efficacy and safety using a randomized, double-blind trial that compared two administrations of low doses of LSD with two high doses (Figure). A low dose was used as a comparator because of previous shortcomings of inactive control conditions regarding blinding in trials with hallucinogenic drugs.This is consistent with the current Food and Drug Administration recommendations for clinical trials on hallucinogens and is intended to improve blinding and reduce potential nocebo effects.
Create a free account to open full-text PDFs.
Holze, F., Vizeli, P., Ley, L. et al. · Neuropsychopharmacology (2020)
Rucker, J., Young, A. H., Jelen, L. A. et al. · Journal of Psychopharmacology (2016)
Goodwin, G. M., Aaronson, S. T., Alvarez, O. et al. · New England Journal of Medicine (2022)
Schindowski, E. M., Jungwirth, J., Schuldt, A. et al. · EClinicalMedicine (2023)
Davis, A. K., Barrett, F. S., May, D. G. et al. · JAMA Psychiatry (2021)
Carhart-Harris, R. L., Bolstridge, M., Rucker, J. et al. · Lancet Psychiatry (2016)
Griffiths, R. R., Johnson, M. W. · Journal of Psychopharmacology (2016)
Ross, S., Bossis, A. P., Guss, J. et al. · Journal of Psychopharmacology (2016)
Holze, F., Gasser, P., Müller, F. et al. · Biological Psychiatry (2023)
Holze, F., Gasser, P., Müller, F. et al. · BJPsych Open (2024)
Raison, C. L., Sanacora, G., Woolley, J. D. et al. · JAMA (2023)
Grob, C. S., Danforth, A. L., Chopra, G. S. et al. · JAMA Psychiatry (2011)
Holze, F., Ley, L., Müller, F. et al. · Neuropsychopharmacology (2022)
Studerus, E., Gamma, A., Vollenweider, F. X. · PLOS ONE (2010)
Bonson, K. R. · Neuropsychopharmacology (1996)
Papers in Blossom that reference this study
Belinger, L., Rieser, N., Engeli, E. et al. · European Neuropsychopharmacology (2026)
Mertens, L. J., Betzler, F., Brand, M. et al. · Psychotherapy and Psychosomatics (2026)
Beichmann, K., Catzeflis, P., Aicher, H. D. et al. · Therapeutic Advances in Psychopharmacology (2026)
Avram, M., Menegaux, A., Müller, F. et al. · Cell Reports (2026)
Erne, L., Mueller, L., Straumann, I. et al. · Translational Psychiatry (2026)
Strengths cited by the authors include a clinically representative sample in terms of chronicity, comorbidity and previous treatments, the use of an active low-dose comparator to improve blinding relative to inactive controls, and a relatively long follow-up of 12 weeks after the last dose. They acknowledge important limitations: incomplete blinding (with binary blinding assessment only), clinician-rated IDS-C assessments conducted by the treating therapist rather than an independent rater (which may introduce bias), possible over-reporting of predefined adverse events because case report forms listed common complaints, absence of an inactive comparator, exclusion of patients with acute suicide risk or very severe illness limiting generalisability, relatively small sample size, lack of multiplicity adjustment for secondary endpoints, and limited ethnic diversity. Given these caveats, the authors characterise the results as promising but not definitive and recommend further investigation in a larger Phase III trial.
Three hundred and nineteen people were assessed for eligibility, and 61 were enrolled (Figure). The most common reasons for screening failures were insufficient German language skills (n = 83), subsequent loss of interest/non-responsiveness (n = 70), and not meeting the criteria for a depressive disorder (n = 40). Of the 61 patients enrolled, 31 were allocated to the low-dose group and 30 were allocated to the high-dose group (Table). A total of 12 patients dropped out of the study (Table). Six patients dropped out after the first administration. Three of these dropouts were related to drug administrations. Two patients in the high-dose group experienced acute drug effects that led them to not wish to take the second dose (namely dysphoria and anxiety). One patient in the high-dose group dropped out after the first dosing session because he did not experience any acute effects. All remaining patients agreed to increase the dose and completed the second administration. One patient in each group was lost to follow-up after the second dosing session.
Twenty-nine patients in the high-dose group and 27 patients in the low-dose group were available for the full analysis set. Of those, the mean Inventory of Depressive Symptomatology, Clinician-Rated (IDS-C) score at screening was 37.2 in the high-dose group and 34.6 in the low-dose group (Table). Inventory of Depressive Symptomatology, Self-Rated (IDS-SR) scores were comparable. The mean duration of depression diagnosis was 16.8 years, and 39% of the patients had comorbid anxiety disorders. In the low-dose group, 19% had prior consumption of a hallucinogen, compared with 17% in the high-dose group (with a mean of 3.4 and 3.2 consumption events, respectively). Almost all participants had attended psychotherapy at least once, 66% had been treated as inpatients, 84% received previous treatments with antidepressants with a mean of three different antidepressants, and 41% were on an antidepressant at enrollment. Antidepressants were tapered off in 21 cases. Additionally, in five cases bupropion, was suspended only on the days of dosing sessions and resumed the following day as usual, as it was not considered to induce relevant interactions with the study drug. During the follow-up period, treatment with antidepressants was re-initiated in two patients in the high-dose group and three patients in the low-dose group. Treatment with an antidepressant was newly initiated in one patient in the high-dose group. Mean ± standard deviation IDS-C scores at baseline were 28.9 ± 8.8 in the low-dose group and 35.9 ± 9.7 in the highdose group. IDS-SR scores were comparable. At the primary endpoint, the least-squares mean change (LSM) change IDS-SR score from baseline was -11.8 in the high-dose group and -3.9 in the low-dose group. The LSM difference in change between conditions was -7.9, 95% confidence interval (CI) of -16.0 to 0.3, standardized effect size (SES) (Cohen's d) of -0.52, and p value of 0.059. The LSM change IDS-C score from baseline was -12.9 in the high-dose group and -3.6 in the low-dose group with an LSM difference in change of -9.2, 95% CI of -17.1 to -1.3, SES of -0.63, and p value of 0.023. This result remained significant after adjusting for multiple comparisons (p = 0.046).
Results for the secondary endpoints are shown in Figureand Table(see also Table). Findings were in the same direction as the primary outcomes. Decreases in IDS-C scores were sustained at weeks 13 and 19. However, decreases in IDS-SR and Symptom Checklist-90 (SCL-90) scores only reached statistical significance at week 13, and changes in Beck Depression Inventory (BDI) scores reached significance at weeks 13 and 19. The unadjusted scores of all efficacy outcomes are shown in Figure. Response rates at the primary endpoint for the IDS-C and the IDS-SR were 40% and 44% in the high-dose group to 20% for both measurements in the lowdose group. The corresponding remission rates were 32% and 36% in the high-dose group, while the low-dose group achieved 20% for both measurements. Response and remission rates were consistently higher in the highdose group across all outcomes and at all time points (see Table).
The mean five dimensions of altered states of consciousness (5D-ASC) total score was 16.0 and 12.6 at low-dose sessions 1 and 2 and 34.7 and 37.9 at highdose sessions 1 and 2, respectively (Figure). In session 1, participants in the high-dose group correctly guessed their assignment 72% of the time, while those in the low-dose group guessed correctly 89% of the time. Therapists identified patients in the high-dose group correctly 69% of the time and in the low-dose group 96% of the time. After session 2, 92% of the participants in the high-dose group correctly identified their assignment, compared with 81% in the low-dose group. Therapists guessed correctly in 80% of cases for the high-dose condition and 92% for the low-dose condition.
Twenty-nine patients in the high-dose group and 28 patients in the low-dose group were available for the safety analysis set. Adverse events (AEs) and List of Complaints (LC) scores were comparable between groups (Table). See Tables S3-S8 for a complete list of all AEs. At session 1, two participants in the high-dose group reported short episodes of paranoid thinking. One participant reported flashbacks at night after both sessions. The described AEs were only present at the session days. Please see the legend of Tablefor more details. After the sessions, the most common AEs were disturbances in attention, headache, and fatigue. Two patients in the highdose group experienced acute drug effects at session 1 as distressing and reported the temporary aggravation of psycholog-ical complaints afterward. One of those patients reported increased passive suicidal ideations (without suicidal intent) for 3 days after the session. It is noteworthy that such thoughts had been previously documented in this patient. Both recovered but eventually dropped out of the study. Four serious adverse events (SAEs) occurred after administering the study drug, and three were judged to be possibly related to the intervention. All three concerned the aggravation of depression with subsequent hospitalization. Of those, two occurred in the low-dose group (one after session 1, one after session 2) and one occurred in the high-dose group (after session 1). However, intervals between the administration of the study drug and the worsening of the condition (≥2 weeks), as well as subsequent hospitalization were relatively long (24, 31, and 102 days, respectively). Moreover, there were other possible causes present (see Tablefor more details). Therefore, hospitalization was unlikely to be associated with the drug administration. Coincidentally, the study team noticed that a patient who was allocated to the low-dose group and had withdrawn consent after the first session was hospitalized and died of suicide 7 months after dropping out of the study. Because this incidence occurred after study participation, it was not documented as an SAE. The session of this patient was uncomplicated. She reported increased restlessness and anxiety in the night after the session which lasted for 3 days but was released from the study in a stable condition 23 days after the session. The independent rater assessed all reported events as unlikely related to the study drug. At sessions, patients in the low-dose group reported a mean of complaints on the LC of 5.5 and 6.1, respectively. In the highdose group, 9.5 and 9.8 complaints were reported, respectively (see Table). At the end-of-study visit, laboratory values were not statistically significantly different between groups (Table). Blood pressure and heart rate at the dosing sessions were nominally higher in the high-dose group, but this difference did not reach statistical significance (Table). Occasionally, study participants in both groups showed elevated diastolic and systolic blood pressure values as well as increased heart rate (see Tablefor more details). One participant in the high-dose group had a systolic pressure of >180 mmHg. Vital signs were not altered at the end-of-study visit (Table). The percentage of patients with aggravation of suicidal ideations was numerically lower in the high-dose group and there were no significant differences between groups (Table).
The results of the sensitivity analyses qualitatively coincided with the efficacy analyses. However, the estimated treatment effects were reduced, and statistical significance was not reached at several time points in all three models. In particular, the primary outcomes did not reach significance after adjustment for covariates (Table). The LSM change score for the IDS-C was -6.8 (95% CI, -14.7 to 1.0, SES of -0.47, p = 0.086). The LSM change score for the IDS-SR was -6.1 (95% CI, -14.0 to 1.8, p = 0.129). The effect of baseline depression scores was significant in both models (IDS-C: coefficient = -0.39, t 49 = -2.51, p = 0.015; IDS-SR: coefficient = -0.40, t 50 = -2.62, SES of -0.41, p = 0.012). No statistically significant effects of age, sex, weight, or antidepressant use at baseline were observed (Table). Additional multiple imputations of missing data did not notably alter the results of the adjusted sensitivity analysis (Table). The analysis using ANOVA yielded results largely consistent with the mixed model for repeated measure approach (see Table). However, a notable difference was the IDS-SR at the primary endpoint, which reached statistical significance in the ANOVA analysis.
This trial investigated the antidepressant efficacy of LSD-assisted therapy using two high doses versus two low doses of LSD. The primary analysis demonstrated statistically significant improvements in one of the two primary outcomes. This result, which remained significant after correction for multiple comparisons, indicates an overall positive outcome for the trial. All secondary endpoints at all time points were in the same direction, suggesting a consistent treatment effect. However, several outcomes, including the IDS-SR at the primary endpoint (p = 0.059) did not reach statistical significance. The sensitivity analyses confirmed the results of the primary analyses qualitatively. After adjustment for covariates, however, differences in outcomes between groups were reduced overall, and several outcomes, including the IDS-C at the primary endpoint (p = 0.086), did not reach statistical significance. These alterations were mainly driven by baseline imbalances in depression scores between groups. Therefore, effects of regression to the mean might have influenced the results. Effect sizes resulting from this sensitivity analysis were still comparable with those seen in a recent, larger trial on psilocybin for MDD,and our study might have been underpowered to detect these effects (please see below for other possible factors). Overall, these findings are promising but should be interpreted with caution given this lack of robustness. Most AEs were nonspecific, and the incidence was comparable between groups. Two participants dropped out of the study The table shows the number of subjects with adverse events. Percentages are shown in parentheses. Brackets show the number of events per subject. Plus-minus values are means ± standard deviation. Frequencies were calculated by considering dropouts at the respective interval. Patients could be counted in more than one category. a Coincidentally, the study team noticed that a patient who was allocated to the low-dose group and had withdrawn consent after the first session was hospitalized and died of suicide 7 months after dropping out of the study. This incident is not listed because only patients who were enrolled at the time point of the respective event are considered. b Both patients dropped out of the study due to negatively experienced acute effects. after their first high-dose session because they reported that acute effects were too distressing. According to our judgment, these were the clinically most important adverse events associated with the investigated treatment. A recent large phase 2 trial on psilocybin reported SAEs that were related to suicidal ideations and behavior in the moderate-and high-dose groups.In the present trial, there were no SAEs related to suicidal ideations and no significant changes in assessments of suicidal ideations. The dropout rate in our study (19%) was higher than in recent psilocybin trials (6%-14%).However, factors beyond the study drug itself may have contributed to this difference, and our safety data indicate that LSD was generally well tolerated. The present trial's strengths include a clinically representative sample with respect to the duration of illness, common comorbid conditions, and various pretreatments. Other strengths include the comparison with a low-dose group and a relatively long follow-up period of 12 weeks after the last administration. This contrasts with previous trials on psilocybin that used shorter follow-ups of 2-6 weeks,no comparators, 7 or inactive comparators.As with trials on classic antidepressants,blinding has not been assessed in most studies with hallucinogenic drugs.If blinding was assessed, then unblinding of >97% was reported.To tackle this issue, two previous studies used a small dose of 1 mg psilocybin as a control condition.By comparison, the dose of 25 μg LSD that was used in the present study, which roughly corresponds to 5 mg psilocybinwas considerably higher. Participants and therapists made correct guesses between 69% and 96% of the cases, suggesting some improvement in blinding. On the other hand, the low dose that was used in this trial resulted in substantial acute effects to such an extent that was unexpected based on our previously reported dose-response data.It seems possible that this caused overlapping pharmacological effects between the low-dose and the high-dose groups, which might have led to improvements beyond the placebo effect in some patients. However, improvements in the low-dose group were overall very moderate. The inclusion of inactive and active low-dose comparators in trials with hallucinogens would be beneficial to further disentangle unspecific factors from interventions. A trial on psilocybin for the treatment of MDD (Clinicaltrials.gov: NCT04670081) has compared 25 mg with 5 mg psilocybin and active placebo and will provide valuable data to clarify this issue. The present phase 2 trial suggests that treatment with two high doses (100 μg and 200 μg) of LSD improved symptoms of MDD beyond treatment with two low doses (25 μg and 25 μg). However, several outcomes did not reach statistical significance after adjusting for covariates, which limits the findings. Compared with trials that used inactive controls, blinding was improved by the inclusion of a low-dose group. Nonetheless, there was still considerable unblinding, and treatment diffusion might have occurred in some cases. Therefore, adequate blinding in trials with hallucinogenic drugs remains a challenge. LSD could be used safely within the framework of this study. Compared with psilocybin, LSD has a longer duration of action.This prolonged effect makes clinical application more resource intensive. It remains to be resolved whether this extended duration offers clinical advantages. Furthermore, it is yet to be determined if there are other relevant differences among hallucinogenic drugs in terms of therapeutic potential. The findings of this exploratory trial support further investigations of LSD as a treatment option for major depression in a larger phase 3 trial.
This trial has several limitations. Blinding was not effective in a considerable number of cases. This might have resulted in nocebo effects in the low-dose group or might have biased ratings made by clinicians. Besides that, blinding was assessed using binary options, but the certainty of the guess was not rated. The clinician-rated IDS questionnaire was assessed by the therapist who also conducted the treatment and not by an independent rater. Outcome assessments conducted by unblinded treating clinicians have been demonstrated to introduce bias.Furthermore, the most common AEs that were reported in this trial were also those that were assessed on the case report forms. Therefore, this approach might have led to over-reporting. Our safety analysis is further limited by the lack of an inactive comparator group. We did not include patients with acute risk for suicide or very severe disorders. Therefore, the results might not be generalizable to other samples. Further limitations of the present trial include the relatively small sample size, lack of adjustment for multiplicity in secondary endpoints, and lack of a sample with ethnic diversity.
Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Felix Mu ¨ ller (felix.mueller@upk.ch).
The trial was in accordance with the Declaration of Helsinki and International Conference on Harmonization Guidelines in Good Clinical Practice. Approval was obtained from the Ethics Committee of Northwest Switzerland, Swiss Federal Office for Public Health, and Swiss Agency for Therapeutic Products (Swissmedic). All participants provided written informed consent. This was an investigator-initiated, 19-week, randomized, double-blind, low-dose controlled, parallel-group phase 2 trial. The study was conducted at the Department of Psychiatry, University of Basel, Switzerland between November 26, 2019, and December 1, 2022.
Patients were recruited via advertisement, referral, and word of mouth. They were eligible for inclusion in the study if they met Diagnostic and Statistical Manual of Mental Disorders, 5th edition, criteria for MDD (assessed with SCID-5), and were at least 25 years old (this limit was chosen due to the increased risk of psychotic disorders in younger individuals). An additional requirement was a moderate-to-severe degree of MDD, assessed with the Inventory of Depressive Symptomatology, Clinician-rated version (IDS-C; score between 24 and 46 points), and Inventory of Depressive Symptomatology, Self-Rated version (IDS-SR; score between 26 and 48). Other inclusion criteria were the following: • Sufficient understanding of the study procedures and risks associated with the study. • Participants must be willing to adhere to the study procedures and sign the consent form. • Patients must be willing to discontinue psychotropic medications in cases where drug-drug interactions are possible (the washout phase will be at least five times the drug's half-life [typically 3-7 days] up to two weeks before the first study day). • If they are in ongoing psychotherapy, participants may continue to see their outside therapist, provided they sign a release for communication with the therapist. • Participants must also refrain from using any psychoactive drugs, except long-term medication or caffeine or nicotine, within 24 hours before each session. They must agree not to use nicotine for at least 2 hours before and 6 hours after each dose of LSD. They must agree not to ingest alcohol-containing beverages for at least one day before each LSD treatment session. Drugs taken 24 hours before the LSD treatment session may result in rescheduling the treatment session to another date. • Participants must be willing not to drive a vehicle or operate machines within 24 hours after LSD administration. Other exclusion criteria were the following: • Weight < 42 kg. • Known hypersensitivity to LSD. • Requiring ongoing concomitant therapy with a psychotropic drug (other than anxiolytic and pain control medication) and being unable or unwilling to comply with the washout period. • Women who are pregnant or breastfeeding or intend to become pregnant during the study. • Lack of safe contraception, defined as: Female participants of childbearing potential not using or not willing to use a medically reliable method of contraception for the entire study duration, such as oral, injectable, or implantable contraceptives, or intrauterine contraceptive devices, or not using any other method considered sufficiently reliable by the investigator in individual cases. Female participants who are surgically sterilized, have undergone hysterectomy, or are post-menopausal for longer than two years are not considered to be of childbearing potential. • Somatic disorders incompatible with safe administration of LSD, including central nervous system involvement of cancer, se- vere cardiovascular disease, untreated hypertension, severe liver disease (liver enzymes increased by more than five times the upper limit of the normal range) or severely impaired renal function (estimated creatinine clearance <30 ml/min). • Preparation session: Patients should be informed about the structure of the LSD session, safety measures, possible effects of LSD (including distressing effects), the importance of introspection and openness to the experience, and the possibility of destabilization or worsening of symptoms at and after the session. They should be advised to reduce stress before and after the session and to avoid major decisions following the session. Concerns and fears should be addressed, and expectations should be discussed. • Integration session: Patients should reflect on and openly discuss their experiences from the LSD session. They should be encouraged to write a detailed experience report. At the fourth visit, fears and concerns related to the second LSD session should be discussed. • LSD sessions: Patients should be advised to have a light breakfast (however, food intake on the session day was not standard- ized). Upon arrival, their current state and any fears or concerns should be discussed. During the session, patients should be encouraged to lie on a couch while therapists should provide reassurance, facilitate introspection, and assist with practical needs such as accompanying them to the restroom.
An independent pharmacy prepared the study medication and performed the computer-generated block randomization and blinding. Block sizes of four were used. The different LSD dosages were prepared in identical-looking vials. The codes kept in opaque envelopes until study completion. LSD base (> 99% purity) was formulated as a drinking solution (water/alcohol) in vials that contained 25 or 100 μg LSD in 1 ml of 96% ethanol. Patients in the high-dose group were assigned one vial (containing 100 μg LSD) for the first session and two vials (containing 100 μg LSD each) for the second session, allowing dose ascension. Patients in the low-dose group were assigned one vial (containing 25 μg LSD) for the first session and two vials (one vial containing 25 μg LSD, one vial containing 0 μg LSD) for the second session. Thus, for blinding purposes, the default vial (first vial) given in session 1 and session 2 contained 25 or 100 μg LSD (label: 0.025 or 0.1mg LSD), and the additional (second) vial given only in session 2 contained 0 or 100 μg LSD (label: Session 2, Dose 2, Placebo or 0.1mg LSD).
Depressive symptomatology was assessed at weeks 2, 5, 9, 13, and 19 using the IDS-C (rated by the treating therapist), IDS-SR, and Beck Depression Inventory-II (BDI). Additionally, a broader spectrum of psychopathology was assessed using the Symptom Checklist-90-R. The primary, prespecified endpoints were the changes in IDS-C and IDS-SR scores from baseline at week 2 to week 9 compared between the low-and high-dose groups. Further secondary endpoints comprised changes in IDS-C and IDS-SR total scores from baseline to weeks 5, 13, and 19 and changes in BDI scores and the SCL-90 global severity index from baseline to weeks 5, 9, 13, and 19 (please see Supplement for more details on the psychometric scales).
After each dosing session, participants rated acute drug effects on the 5 Dimensions of Altered States of Consciousness questionnaire (5D-ASC). To assess blinding, participants and their respective therapists were asked to guess the group assignment (binary answers) after each dosing session. Safety At each visit and each dosing session, patients were asked by the respective therapists if they had experienced adverse events (AEs) of any form. When AEs occurred, they were reported by the investigators on a standard form. All AEs were coded according to the Medical Dictionary for Regulatory Activities (MedDRA; version 26.0). Serious adverse events (SAEs) were re-assessed by an independent rater. Adverse events (AEs) were assessed with regard to duration, intensity, frequency, relation to study drug, action taken, and outcome. Adverse events were defined as serious adverse events (SAEs) if the event resulted in death, was immediately lifethreatening, led to hospitalization or prolongation of hospitalization, caused persistent or clinically significant disability or incapacity, or resulted in congenital abnormality or birth defect. In addition to the assessments described for AEs, SAEs were assessed with regard to time interval from administration, relevant medical history, concomitant medication, lab findings and other investigations. The relationship of AEs to the study drug was assessed according to the following criteria: • Definitely ○ Temporal relationship ○ Improvement after dechallenge (only taken into consideration if applicable to reaction) ○ Recurrence after rechallenge (or other proof of drug cause) • Probably ○ Temporal relationship ○ Improvement after dechallenge (only taken into consideration if applicable to reaction) ○ No other cause evident • Unlikely: Any assessable reaction that does not fulfil the above conditions • Not related: Causal relationship can be ruled out To facilitate documentation, the case report forms also contained identical standard forms that contained common, predefined adverse events (headache, nausea, fatigue, dizziness, and disturbances in attention) for all visits and sessions. At the dosing sessions, participants were asked about AEs in the morning and in the evening. Furthermore, the participants completed the List of Complaints questionnaire (LC)immediately before and after drug administration. The LC contains 66 predefined complaints and was used to capture the side effects of the drug. This which involved measuring the symptoms present in the morning (before intake) and comparing them to symptoms reported after drug intake, allowed for controlling of pre-existing symptoms. Safety assessments also comprised vital signs (taken once at screening, once at the end-of-study visit, and every 2 h during the dosing sessions) and laboratory tests (hemogram, creatinine, alanine aminotransferase) at screening and the end-of-study visit. An automatic oscillometric device was used to measure heart rate and blood pressure (Omron Healthcare Europe NA, Hoofdoorp Netherlands). Vital signs were measured once at screening, at the end-of-study visit, and at the sessions directly before drug administration and every two hours after administration for at least 10 hours. Laboratory analyses (hemogram, creatinine, alanine aminotransferase) at screening and the end-of-study visit were conducted using standard assays and according to Good Laboratory Practice by the Laboratory Medicine Department of the University Hospital Basel. Several other assessments were taken during the study, including multiple other questionnaires administered at the study visits, actigraphy recordings between visit 2 and visit 6, MRI scans before session 1 and several days after session 2, additional blood sample at visit 5 and 8, and a structured interview at visit 8. At the session days, one additional questionnaire was administered prior to drug administration. All these data will be published separately. In this publication, we primarily report the main objectives of the study, i.e., clinical outcomes and safety data. Please note that the State-Trait Anxiety Inventory (STAI) was used in the study to assess anxiety symptoms. However, a baseline assessment was not available for this scale. Given the significant imbalance in depression scores and general psychopathology, it was deemed unsuitable to utilize this scale in our analyses.
All questionnaire scores were calculated per patient and questionnaire. IDS IDS-C and IDS-SR measure depressive symptoms occurring over the last seven days.Each questionnaire contains 30 questions with a 4-point scale ranging from 0 to 3. Scores on the IDS-C range from 0 to 84, with 0-11 corresponding to no depression, 12-23 corresponding to mild depression, 24-36 corresponding to moderate depression, and >36 corresponding to severe or very severe depression. IDS-SR scores range from 0 to 84. Answers were summed up for the calculation of the IDS-C and IDS-SR total scores. Questions 9b and 9c of the IDS-C were not included. Question 9b of the IDS-SR was not included. BDI BDI measures depressive symptoms occurring over the last 14 days.The questionnaire contains 21 questions with a 4-point answer scale ranging from 0 to 3. Scores range from 0 to 63, with higher scores indicating more severe depression. For the calculation of the total score, all answers were summed up.
The SCL-90 evaluates a broad range of symptoms of psychopathology.The SCL-90 contains 90 questions with a 5-point answer scale ranging from 0 to 4. Scores range from 0 to 360, with higher scores indicating more severe symptoms. The SCL-90-R-GSI score was calculated by adding all answers and dividing the result by the number of answered questions.
The LC consists of 66 items which assess mental and somatic complaints.The LC consists of 66 items with binary yes or no answers. The LC was completed directly before and 12 hours after administration. Items already answered with ''yes'' before the drug administration were not considered. All items were converted into codes according to the Medical Dictionary for Regulatory Activities (MedDRA; version 26.0). In some cases, this resulted in the summation of different items under the same MedDRA code. Namely, faintness, exhaustibility, and fatigue were summarized as fatigue; shortness of breath and respiratory distress were summarized as dyspnoea; blush, and two distinct German terms for hot flush (''aufsteigende Hitze'', ''Hitzewallungen'') were summarized as hot flush; inner restlessness, and restless legs were summarized as restlessness; hypersensitivity to odors, hypersensitivity to heat, and hypersensitivity to cold were summarized as sensory processing sensitivity. For one item of the LC (erotic thoughts), there was no corresponding MedDRA code.
The 5D-ASC measures different aspects of the acute subjective effects of hallucinogenic drugs.The questionnaire contains 90 questions with a visual analog scale ranging from 0 to 100. The scores are calculated as follows (
The trial was powered based on a pilot study on psilocybin for moderate-to-severe treatment-resistant MDD with two moderate-tohigh doses of psilocybin.This open-label, uncontrolled trial reported an effect size (Hedge's g) of 3.2 two weeks after treatment (reflecting the change from baseline on the short of the IDS). As this study did not provide a control group, the contribution of non-treatment effects (e.g., placebo effects, regression to the mean) is unknown. Therefore, we estimated the actual treatment effect size at around 1.0. Based on this estimation, a sample size of 17 was calculated to achieve >80% power at an alpha of 0.05 (two-sided independent t-test). To account for uncertainties in this analysis, the target sample size was increased to reach a sample of at least 50 patients who complete the primary endpoint. Dropouts were expected and replaced, resulting in a sample size of approximately 60 enrolled patients. Given the actual sample size of the full analysis set, the smallest outcome estimate reaching statistical significance was an effect size of 0.52 (Cohen's d). For the analysis of the efficacy outcomes, the original predefined analysis plan specified an ANOVA with correction for multiple comparison across all time points of the respective outcome, without joint correction across both primary outcomes. However, a mixed model for repeated measures (MMRM) was ultimately used due to its statistical advantages (e.g., handling of missing data). Additionally, the final approach incorporated a joint correction for multiple comparisons across both primary outcomes, diverging from the initial plan. While an alternative gatekeeping procedure was initially used in the revised analysis, it was then decided that a non-hierarchical approach better aligned with the original analysis plan (please see below). We recognize that such methodological alterations may introduce analytical bias. To address this concern, robustness checks were conducted, comparing results from both the original and revised approaches. To compare the MMRM with the ANOVA approach, all outcomes were additionally analyzed using the originally planned approach. Least squares means (LSM) changes from baseline were used to ensure consistent estimates of change over time. Consistent with the MMRM model, we also included age, sex, weight, use of antidepressants, as baseline scores of the respective outcome as covariates in an ANCOVA to control for potential influencing factors. Results were adjusted using Tukey post hoc tests. A restricted maximum likelihood mixed model for repeated measures (MMRM) on least square mean (LSM) change from baseline was used for the assessments. The model included treatment, assessment time point, and the interaction of both as fixed factors. An unstructured covariance structure was used. Model assumptions were tested. The assessment of outcomes was based on the full analysis set (i.e., all randomized patients who received at least one dose of the study drug and had been validly assessed at baseline [week 2] and at week 5 or 9). For the primary endpoints, data from weeks 5 and 9 were included in the MMRM. To control the overall type I error rate of this primary analysis, the Benjamini-Hochberg method was used. Additional secondary analyses comprised IDS-C, IDS-SR, BDI, and SCL-90 scores at all time points. Although all secondary endpoints were primarily intended to provide validation for the primary endpoints, we additionally applied adjustments for multiple comparisons (Benjamini-Hochberg method) in the analysis of these outcomes. Response and remission rates, expressed as percentages, were calculated for all outcomes and at all timepoints. Response was defined as 50% symptom reduction from baseline, and remission was defined according to established cutoff scores of the respective questionnaires (IDS-C: ≤11; IDS-SR: ≤13; BDI: ≤9; SCL-90: ≤0.5).Furthermore, three post hoc sensitivity analyses were conducted. The first was a repetition of the MMRM analyses with adjustments for the covariates age, sex, weight, use of antidepressantsat screening, and baseline value of the respective outcome. The second and the third were multiple imputations of missing data at weeks 5 and 9 (see below table) using a pattern mixture method under a missing-at-random assumption and a reference-based approach under a missing not-at-random assumption with the lowdose group as a reference. This last analysis was conducted on all randomized patients. As a measure of the acute effects of the drug, the total score of the 5D-ASC was calculated and reported as the mean for each group and dosing session.
The success of blinding was assessed by calculating the percentage of correct guesses by participants and therapists separately for each dosing session. We primarily report on the blinding assessment conducted after the first session.The safety analysis set comprised all patients who received at least one dose of the study drug and had at least one post-administration safety assessment. All AEs that occurred after administration of the investigational drug are reported. Adverse events were summarized as the number of subjects with AEs. Additionally, the mean sum scores of all complaints on the List of Complaints were calculated for each group and dosing session. Laboratory values were grouped according to deviations from normal ranges, and changes at the end-of-study visit were compared between groups. Vital parameters at screening and the end-of-study visit and maximal vital parameters at the dosing sessions were grouped in proportions (> 90, > 100, and > 110 mmHg for diastolic blood pressure; > 140, > 160, and > 180 mmHg for systolic blood pressure; > 80, > 100, and > 120 beats/min for heart rate) and compared between groups at the dosing sessions and the end-of-study visit. Vital signs and laboratory values were compared using Fisher's exact tests. For post hoc analyses, alterations of suicidal ideations between groups were assessed due to the observation of suicide ideation and behavior after exposure to moderate-to-high doses of psilocybin in patients with MDD.Therefore, alterations of suicidal ideations between groups were assessed in the present trial. To this end, aggravation from baseline on item 18 (which assesses suicidal ideations) of the IDS-C was tested for differences between conditions using Fisher's exact tests. The last-observation-carried-forward method was used to impute missing data after week 5. With the exception of the ANOVA analyses, which were conducted using R version 4.4.3, 32 SAS 9.4 software was used for the analyses.Significance was assumed at p < 0.05 (two-sided). The study was registered at ClinicalTrials.gov (NCT03866252) and monitored regularly and independently. Statistical results are reported in the results section and detailed further in the Supplement.
ClinicalTrials.gov Identifier: NCT03866252. Please cite this article in press as: Mu ¨ ller et al., Efficacy and safety of low-versus high-dose-LSD-assisted therapy in patients with major depression: A randomized trial, Med (2025),