MDMA-Assisted Therapy for PTSD: Evidence, Regulation and Research Integrity
Psychedelic-Assisted Medicine: Science and Practice
Learning objectives for this lesson:
- Explain why MDMA was proposed as an aid to psychotherapy for post-traumatic stress disorder, and describe the MDMA-assisted therapy protocol used in the phase 3 trials, including the co-therapist team, the three experimental sessions with preparation and integration, and the inner-directed approach.
- Explain how the CAPS-5 measures the severity and diagnosis of post-traumatic stress disorder, and classify trial outcomes as response, loss of diagnosis or remission.
- Summarize the design, results and safety findings of the MAPP1 and MAPP2 trials, and interpret their effect sizes and numbers needed to treat, including the effect of the denominator chosen.
- Evaluate the evidence on durability and explain why MDMA-assisted therapy cannot be compared directly with trauma-focused psychotherapies on the evidence available as of October 2026.
- Explain the issues behind the FDA's 2024 decision, including functional unblinding, expectancy, adverse event recording and the role of psychotherapy, and set out the sponsor's and investigators' responses to each.
- Describe the 2024 retractions, the status of the program as of October 2026 and Australia's 2023 authorization, with their dates.
- Explain why altered states raise the stakes for participant protection, describe the misconduct at a Canadian trial site disclosed in 2019, factually and from published sources, and explain the safeguards that followed.
- Apply safeguards for touch, consent, video review, team composition, overnight monitoring and reporting to the review of a trial protocol.
This course was developed by Dr. Kiffer G. Card.
The MDMA-Assisted Therapy Model and How Outcomes Are Measured
Content note
This section describes post-traumatic stress disorder and the trauma histories of trial participants, which included childhood abuse and sexual violence, and it mentions suicidal ideation recorded in trials. Support is available at any time from the 9-8-8 Suicide Crisis Helpline (call or text 9-8-8), the Hope for Wellness Help Line for Indigenous people across Canada (1-855-242-3310) and, for people in British Columbia affected by sexual violence, VictimLinkBC (1-800-563-0808, call or text).
Introduction
A participant in MAPP1, the earlier of the sponsor's two phase 3 trials of MDMA-assisted therapy, spent about 18 weeks in treatment. Each participant met a pair of therapists for three preparatory conversations, took capsules on three mornings about a month apart and spent each of those days in a quiet room with the same two therapists, and returned for nine further conversations (Mitchell et al., 2021). Neither the participant nor the therapists were told whether the capsules contained MDMA or an inactive placebo. A separate clinician who knew nothing of the treatment then interviewed the participant by video and scored the symptoms of PTSD. Lesson 7 described models of psychedelic-assisted therapy in general, including the inner-directed MAPS approach; this section describes the specific protocol tested in the MDMA trials, because every result in Sections 2 to 4 depends on it.
Learning Objectives
- 1.1 Explain why investigators proposed MDMA as an aid to psychotherapy for post-traumatic stress disorder (PTSD), and distinguish the human evidence from the animal evidence behind that proposal.
- 1.2 Describe the structure of a course of MDMA-assisted therapy in the phase 3 trials: preparation, three experimental sessions with split doses, and integration.
- 1.3 Describe the co-therapist team, the inner-directed approach and the treatment manual's rules on touch.
- 2.1 Explain how the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5) produces a total severity score and a diagnosis, and how the trials defined response, loss of diagnosis and remission.
- 2.2 Calculate a CAPS-5 total severity score from cluster scores and classify a change in score using the trial definitions.
1.1 PTSD and the Case for a New Treatment
Post-traumatic stress disorder is a psychiatric condition that can follow exposure to actual or threatened death, serious injury or sexual violence. The FDA's briefing document for the 2024 advisory committee described it as a disabling condition characterized by intrusive memories, hyperarousal and avoidant behaviour, with high rates of co-occurring depression and substance use disorders and a high risk of suicidal ideation and behaviour (FDA, 2024). A person who survived a serious assault might, for example, relive the event in nightmares, avoid the street where it happened, feel cut off from friends and blame themselves, and startle at small noises. HSCI 214 develops the classification of trauma and PTSD; this lesson uses the diagnosis as the trials did.
1.1.1 The Treatments Already Available
Two kinds of treatment were established before MDMA-assisted therapy reached phase 3. Trauma-focused psychotherapies, such as prolonged exposure, cognitive processing therapy and eye movement desensitization and reprocessing (EMDR), help the person approach the traumatic memory and the beliefs attached to it in a planned way; in prolonged exposure, for example, a person repeatedly recounts the trauma in session and gradually approaches avoided places. The MAPP trial reports describe these therapies as the gold standard treatment (Mitchell et al., 2021, 2023). When the FDA reviewed the MDMA application in 2024, the selective serotonin reuptake inhibitors sertraline and paroxetine were the two medications approved for PTSD in the United States, and the FDA noted that response rates rarely exceed 60 percent and that fewer than 20 to 30 percent of patients achieve full remission (FDA, 2024, citing Berger et al., 2009). Many people also leave psychotherapy early: a meta-analysis of randomized trials found a pooled dropout rate of 16 percent, higher in therapies with a trauma focus (Lewis et al., 2020). Section 2 returns to these figures.
1.1.2 Why MDMA Was Proposed
MDMA is an entactogen that acts mainly by causing the release of serotonin through the serotonin transporter, and it also affects noradrenaline and dopamine (Mitchell et al., 2021; ICER, 2024); Lesson 3 described this pharmacology. The case for using it in PTSD rests on three kinds of evidence of differing strength. The animal findings conflict: in mice, MDMA enhanced the extinction of conditioned fear, an effect linked to increased brain-derived neurotrophic factor in the amygdala, while a study in rats found that it impaired fear extinction (Mitchell et al., 2021; Casey & Heifets, 2026). In a few human imaging studies, mostly of healthy volunteers, MDMA reduced amygdala activity, including responses to negative emotional images (Mitchell et al., 2021; Casey & Heifets, 2026). Clinicians have also described participants who could stay emotionally engaged while revisiting traumatic experiences without becoming overwhelmed. The MAPP1 authors suggested that MDMA might open a "window of tolerance" for processing traumatic material and that its prosocial effects might strengthen the therapeutic alliance (Mitchell et al., 2021). These are hypotheses about mechanism that rest mainly on early animal and human studies: a 2026 review judged the mechanisms by which MDMA might help in PTSD poorly defined (Casey & Heifets, 2026), research on them is evolving, and Lesson 4 examined the gaps between such findings and clinical benefit.
The sponsor hypothesized that MDMA works by facilitating the effects of psychotherapy, and no trial in its program tested MDMA without psychotherapy (FDA, 2024). Every arm of the sponsor's trials therefore received psychotherapy, a design choice that Section 3 shows became central to the regulatory debate.
1.2 The Structure of a Course of Treatment
The phase 3 trials, known as MAPP1 and MAPP2, used an identical schedule of sessions and doses (FDA, 2024). The schedule has three phases, which Figure 9.1 lays out on a single timeline and Table 9.1 summarizes.
1.2.1 Preparation
Before any drug was given, each enrolled participant met the two-person therapy team for three 90-minute preparatory sessions focused on building alliance and trust and on how to respond to memories and feelings that might arise (Mitchell et al., 2021). Psychiatric medications were tapered and stopped under medical supervision, for at least five half-lives plus one week, before the baseline assessment. Randomization took place only after preparation was complete.
During preparation the participant and therapists also make explicit agreements (Mithoefer, 2017). The participant agrees to refrain from harm to self, others or property and to remain in the treatment area until the session ends, including any overnight stay the protocol requires. Both therapists agree to stay in the room for the whole session, apart from brief absences by one at a time, and to keep daily telephone contact for up to a week after each session.
1.2.2 The Experimental Sessions
The treatment period contained three experimental sessions, each about eight hours long and spaced about four weeks apart. After a 10-hour fast, a urine drug screen and measurements of blood pressure, temperature and heart rate, the participant received a divided dose in the morning: 80 mg of MDMA followed by a supplemental 40 mg in the first session, and 120 mg followed by 60 mg in the second and third (Mitchell et al., 2021, 2023). The supplemental dose or the increase to 120 mg could be withheld for tolerability or at the participant's request. Participants in the placebo arm received inactive capsules on the same schedule. These doses are reported as facts about the trials.
The MAPP1 protocol originally required participants to stay overnight at the site after each experimental session, and the treatment manual describes an overnight stay accompanied by a trained attendant (Mithoefer, 2017). A March 2018 amendment added a sub-study in which a participant judged stable at the end of the session could be escorted home if a chosen support person stayed overnight (FDA, 2024). Section 4 considers what overnight monitoring protects against.
1.2.3 Integration
Each experimental session was followed by three 90-minute integration sessions, the first on the following morning and the other two over the next three to four weeks, so that a full course contained nine (Mitchell et al., 2021). They help the participant describe and understand the medication session, particularly the experience of revisiting the trauma. The FDA observed that most of the active psychotherapeutic work appeared to occur in these sessions and that the manual left their content largely to individual therapists (FDA, 2024). Table 9.1 brings the three phases together.
Table 9.1. The components of one course of MDMA-assisted therapy in the MAPP1 and MAPP2 protocols (Mitchell et al., 2021, 2023; Mithoefer, 2017; FDA, 2024).
| Component | Number and length | Main purpose |
|---|---|---|
| Screening and medication taper | Variable (taper 0 to 103 days in MAPP1) | Confirm eligibility; stop psychiatric medications before baseline |
| Preparatory sessions | 3 sessions of 90 minutes | Alliance, trust, history, agreements, guidance on difficult material |
| Experimental (medication) sessions | 3 sessions of about 8 hours, about 4 weeks apart | Divided dose of MDMA or placebo with inner-directed support from two therapists |
| Integration sessions | 9 sessions of 90 minutes (3 after each experimental session) | Make sense of the session and carry insights into daily life |
| Blinded outcome interviews | 4 (baseline, after sessions 1 and 2, week 18) | CAPS-5 and functional impairment scored by independent raters |
The structure in Table 9.1 places three drug exposures inside about 18 hours of non-drug psychotherapy and about 24 hours of supervised medication sessions. That proportion matters when the effect of the drug is separated from the effect of the therapy, a question taken up in Sections 2 and 3.
1.3 The Therapy Team and the Inner-Directed Approach
The manual for MDMA-assisted psychotherapy, written by Michael Mithoefer with contributors from the Multidisciplinary Association for Psychedelic Studies (MAPS), first issued in 2005 and revised up to version 8.1 in August 2017, defines the approach used in all the MAPS-sponsored trials (Mithoefer, 2017). Its central features are a two-person team, a non-directive therapeutic stance and a set of rules about touch.
1.3.1 The Co-Therapist Team
Each participant worked with a team of two co-therapists, "generally one female and one male," for all preparatory, experimental and integration sessions (Mithoefer, 2017). In MAPP1 every therapist held at least a master's degree, at least one member of each team had to be licensed to provide psychotherapy under local rules, and all therapists completed the sponsor's five-part training: a 15-hour online course, a five-day training course, three days of experiential learning, one day of role play and 52 hours of supervision (Mitchell et al., 2021). Two people in the room share the work of an eight-hour session and each witnesses the other's conduct. Section 4 shows that a second therapist did not prevent the misconduct at one Canadian site, where MAPS provisionally found that both co-therapists had departed from the manual.
1.3.2 An Inner-Directed, Non-Directive Stance
The manual asks therapists to follow the participant's lead and describes a non-directive approach that "emphasizes invitation" (Mithoefer, 2017). Its organizing idea is the participant's inner healing intelligence, which it defines as "a person's innate capacity to heal the wounds of trauma," and it describes the MDMA and the therapists as facilitators of that capacity. During a medication session, the therapist might sit in silence and listen empathically, or offer more active support such as helping the participant talk through a memory that has arisen, depending on what the participant appears to need (FDA, 2024). The flip cards below define the manual's working vocabulary.
1.3.3 Touch in the Manual
The manual treats touch as a possible aid to healing with conditions attached: "any touch that has sexual connotations or is driven by the therapist's needs" has "no place in therapy" (Mithoefer, 2017). During preparation, the therapists make clear that the participant need not be touched at all, ask whether the participant wants touch offered or always wants to be asked first, and agree on a word such as "Stop" that they will always obey unless touch is needed to protect the participant from physical harm. The manual also warns that withholding nurturing touch when it is indicated "may even be perceived by the participant as abuse by neglect," so it does not ask therapists to avoid touch altogether. Section 4 examines how these rules worked in practice.
1.3.4 Checking That the Manual Was Followed
The manual states that adherence would be "measured by independent review of session videos against adherence criteria" (Mithoefer, 2017). Critics later argued that the therapy remained poorly standardized, a concern ICER heard from several interviewees and noted also affects many psychotherapy trials (ICER, 2024).
2.1 Measuring PTSD with the CAPS-5
The efficacy results of both phase 3 trials rest on one instrument, the Clinician-Administered PTSD Scale for DSM-5 (CAPS-5), and knowing how it is scored explains what a "10-point change" or "loss of diagnosis" means in Section 2.
2.1.1 Structure and Scoring
The CAPS-5 is a structured interview of 30 items conducted by a trained clinician (National Center for PTSD, 2026). The interviewer first identifies an index trauma, the single event the symptom questions refer to, and then asks about 20 symptoms that correspond to the four DSM-5 symptom clusters listed in Table 9.2. The remaining items ask about onset, duration, distress, impairment, overall severity and the dissociative subtype of PTSD, in which a person also experiences depersonalization or derealization (Weathers et al., 2018). For each symptom, the interviewer combines its frequency and intensity into a single severity rating from 0 to 4: 0 means absent, 1 mild or subthreshold, 2 moderate or threshold, 3 severe or markedly elevated, and 4 extreme or incapacitating (Weathers et al., 2018).
Table 9.2. The four DSM-5 symptom clusters assessed by the 20 core CAPS-5 items, with the number of symptoms needed for a diagnosis.
| Cluster | What it covers (examples) | Items | Severity range | Symptoms needed for diagnosis |
|---|---|---|---|---|
| B: Intrusion | Intrusive memories, nightmares, flashbacks | 5 | 0 to 20 | At least 1 |
| C: Avoidance | Avoiding memories or reminders of the event | 2 | 0 to 8 | At least 1 |
| D: Negative alterations in cognition and mood | Self-blame, persistent negative emotions, feeling detached | 7 | 0 to 28 | At least 2 |
| E: Alterations in arousal and reactivity | Hypervigilance, exaggerated startle, poor sleep | 6 | 0 to 24 | At least 2 |
A symptom counts toward a diagnosis only if its severity rating is 2 or higher, which Weathers and colleagues call the SEV2 rule (Weathers et al., 2018). A diagnosis also requires that the disturbance has lasted more than one month and causes clinically significant distress or impairment (National Center for PTSD, 2026). The total severity score is the sum of the 20 symptom ratings, so it runs from 0 to 80. In its initial validation in two samples of veterans, the total score showed high interrater reliability (intraclass correlation 0.91) and good test–retest reliability (0.78) (Weathers et al., 2018).
2.1.2 Severity Bands and the Trial Definitions of Change
The MAPP2 report used the following severity bands for the total score: 0 to 10 asymptomatic, 11 to 22 mild, 23 to 34 moderate, 35 to 46 severe and 47 or more extreme (Mitchell et al., 2023). MAPP1 required a baseline score of 35 or more and MAPP2 a score of 28 or more (FDA, 2024). Before MAPP1 began, the sponsor and the FDA agreed on three categories of improvement (FDA, 2024; Mitchell et al., 2023):
- A responder had a reduction of 10 points or more in total severity from baseline. ICER noted in 2024 that no agreed definition of a clinically meaningful change on the CAPS-5 existed and that this threshold came from the sponsor's agreement with the FDA (ICER, 2024).
- A participant with loss of diagnosis had a reduction of 10 points or more and no longer met the diagnostic criteria for PTSD.
- A participant in remission no longer met the diagnostic criteria and had a total severity score of 11 or less. The trial reports describe remission as loss of diagnosis with a total of 11 or less, which amounts to the same thing for anyone who met the entry thresholds (Mitchell et al., 2021, 2023).
The categories are nested: a participant in remission after a reduction of 10 points or more also counts as a responder and as having lost the diagnosis.
2.1.3 Who Did the Scoring
In both trials, the CAPS-5 was administered by a centralized pool of independent raters who interviewed participants by live video (Mitchell et al., 2021, 2023). The raters were blinded to the study design, the visit number and the participant's allocation, and no rater assessed the same participant repeatedly, so a rater could not compare answers across visits. Participants were asked not to tell raters which treatment they thought they had received, and diagnostic agreement between raters in MAPP1 had a kappa of 0.94 (Mitchell et al., 2021). Functional impairment, the secondary outcome, was rated with the Sheehan Disability Scale (SDS).
Independent raters protect the measurement from the raters' expectations, but the participant remains the source of the answers. ICER also reported concerns that the CAPS-5, which asks about a single index trauma and was developed mainly in male veterans, may miss changes outside that trauma; some participants it interviewed described improvement on the index trauma while other difficulties worsened, although ICER could not estimate how often (ICER, 2024). Section 3 develops both concerns.
2.2 Scoring and Classifying a Participant
The scoring rules and trial definitions above can be applied to a single participant, which shows how the total score and the diagnosis can move differently.
A hypothetical participant has these cluster totals at baseline: intrusion 11 (five items rated 2 or more), avoidance 6 (two items), cognition and mood 15 (five items) and arousal 12 (five items). The total is 11 + 6 + 15 + 12 = 44, in the severe band and close to the mean baseline score of 44.1 in MAPP1 (Mitchell et al., 2021). All four cluster requirements are met.
At week 18 the cluster totals are 4 (one intrusion item rated 2 or more), 2 (one avoidance item), 6 (two cognition and mood items) and 5 (one arousal item). The new total is 4 + 2 + 6 + 5 = 17, a reduction of 27 points. The arousal cluster now has one symptom at the SEV2 threshold, below the two required, so the diagnosis is no longer met. The participant has lost the diagnosis and is not in remission, because 17 is above 11.
Use Calculator 9.1 to repeat the worked example and explore how the categories behave. It assumes the duration and impairment criteria are met.
Using Calculator 9.1, start from the hypothetical baseline preset (total 44). Find week-18 scores that make the participant a responder who still meets the diagnostic criteria, then scores that produce remission, then a total below 22 with the criteria still met. What does the third set show about the difference between the total score and the diagnosis?
Summary
MDMA-assisted therapy in the phase 3 trials was a structured course of about 18 weeks: three preparatory sessions, three eight-hour experimental sessions with divided doses of MDMA or placebo, and nine integration sessions, all delivered by the same two-person team following an inner-directed manual with explicit rules on touch. The rationale for adding MDMA combined conflicting animal findings, a few imaging studies and clinical observation, which remain hypotheses about mechanism, and the sponsor presented the drug as a facilitator of psychotherapy. Outcomes were measured with the CAPS-5, whose total severity score sums 20 symptom ratings and whose diagnosis depends on cluster counts, scored by blinded independent raters. Section 2 uses these definitions to read the results of MAPP1 and MAPP2.
Key Takeaways
- 1.1 PTSD can follow exposure to actual or threatened death, serious injury or sexual violence, and trauma-focused psychotherapies and two SSRIs were established before MDMA reached phase 3, although many people do not reach remission or leave therapy early. The case for MDMA combined conflicting animal findings on fear extinction, a few imaging studies showing reduced amygdala responses and clinical observation, and these remain hypotheses about mechanism. The sponsor presented MDMA as a facilitator of psychotherapy, so every arm of the sponsor's trials received therapy.
- 1.2 A course lasted about 18 weeks: three 90-minute preparatory sessions, three experimental sessions of about eight hours, about four weeks apart, with divided doses (80 mg plus 40 mg, then 120 mg plus 60 mg), and nine 90-minute integration sessions. Psychiatric medications were tapered before baseline, explicit agreements were made in preparation, and the placebo arm followed the same schedule with inactive capsules.
- 1.3 The same two co-therapists, generally one female and one male, delivered every session under a manual that follows the participant's lead and centres on the participant's inner healing intelligence. Touch, as nurturing touch or focused bodywork, was agreed in preparation while the participant was sober, was never required, and ended at a stop word except to protect the participant from physical harm.
- 2.1 The CAPS-5 total severity score sums 20 symptom ratings of 0 to 4, so it runs from 0 to 80, while the diagnosis depends on how many symptoms in each cluster reach a rating of 2 (the SEV2 rule). A responder improved by 10 points or more, loss of diagnosis added no longer meeting the criteria, and remission added a total of 11 or less. Blinded independent raters scored the interviews by live video.
- 2.2 Adding the four cluster totals gives the total severity score, and applying the cluster rule shows whether the diagnosis is still met. Because the categories are nested and the diagnosis depends on cluster counts, a participant can lose the diagnosis without reaching remission, and a single symptom moving from 2 to 1 can remove the diagnosis while changing the total by one point.
Reflection
A hypothetical friend with PTSD has read that MDMA-assisted therapy is "just taking ecstasy with a therapist in the room." In the MAPP phase 3 protocol, each participant had three 90-minute preparatory sessions with a two-person therapy team; three eight-hour experimental sessions about four weeks apart, with a divided dose of 80 mg plus 40 mg in the first session and 120 mg plus 60 mg in the second and third; and nine 90-minute integration sessions, all following an inner-directed manual with explicit rules on touch, and outcomes were scored by blinded independent raters using the CAPS-5. Explain to your friend how this treatment differs from taking MDMA alone. Then identify one feature of the protocol that you think most protects the participant and one feature that depends most heavily on the therapists' judgement, and explain your choices.
The treatment in the trials was a four-month course of psychotherapy in which MDMA was given on three days. Before any dose, the participant spent three sessions with the same two therapists building trust, discussing what might come up and agreeing on rules, including whether and how they could be touched and a stop word. Each medication day lasted about eight hours, with both therapists present, a divided dose given under medical checks and an inner-directed stance in which the therapists follow the participant's lead. Each was followed by three integration sessions, the first the next morning, to make sense of the experience. The drug was intended to make this therapy more effective, and the comparison arm received the same therapy with a placebo.
The feature I think most protects the participant is the agreement about touch reached in preparation, while the participant is sober, because it sets limits before the drug impairs judgement. The feature most dependent on the therapists is the content of the integration sessions, which the FDA noted the manual did not standardize, so what a participant receives there depends heavily on the particular team.
Minimum 20 characters required.
Question 1: In the MAPP1 protocol, how was MDMA given in the second and third experimental sessions?
Question 2: A hypothetical participant's CAPS-5 total falls from 44 at baseline to 17 at week 18, and only one arousal symptom is now rated 2 or higher. How do the MAPP definitions classify this participant?
Question 3: What does the MAPS treatment manual mean by a participant's inner healing intelligence?
Question 4: Which statement about the CAPS-5 raters in the MAPP trials is accurate?
The Phase 3 Trials: MAPP1 and MAPP2
Content note
This section reports suicidal ideation, self-harm and suicidal behaviour recorded as adverse events in the trials. Support is available at any time from the 9-8-8 Suicide Crisis Helpline (call or text 9-8-8) and the Hope for Wellness Help Line for Indigenous people across Canada (1-855-242-3310).
Introduction
The MDMA program was planned around two confirmatory trials. In a 2017 meeting the FDA and the sponsor agreed that the application would rest on two phase 3 trials with identical designs, run one after the other at largely the same sites (FDA, 2024). The first, MAPP1, enrolled people with severe PTSD between 2018 and 2020; the second, MAPP2, enrolled people with moderate or severe PTSD between 2020 and 2022. This section reads the two trials with the checklist Lesson 1 introduced: design, comparator, blinding, sample, follow-up, and funding and conflicts of interest. Section 3 then shows how the FDA and others weighed the same results.
Learning Objectives
- 3.1 Describe the design, participants, comparator and sponsorship of the MAPP1 and MAPP2 phase 3 trials.
- 3.2 Report the primary results of both trials, calculate a standardized mean difference, and interpret the improvement seen in the placebo-with-therapy arm.
- 3.3 Calculate the absolute difference, relative risk and number needed to treat from the trials' responder categories, and explain how the choice of denominator changes them.
- 3.4 Summarize the safety findings of the two trials, including suicidality, and identify what the trials did not measure.
- 4.1 Evaluate the long-term follow-up data and their limits.
- 4.2 Explain why the MAPP results cannot be compared directly with trials of trauma-focused psychotherapy.
3.1 Design and Participants
Both trials randomized participants one to one to MDMA-assisted therapy or to placebo with identical therapy, used the session schedule described in Section 1, and measured the change in CAPS-5 total severity from baseline to 18 weeks as the primary outcome (Mitchell et al., 2021, 2023). Table 9.3 sets out their main features side by side.
Table 9.3. Main features of the two phase 3 trials of MDMA-assisted therapy for PTSD.
| Feature | MAPP1 (Mitchell et al., 2021) | MAPP2 (Mitchell et al., 2023) |
|---|---|---|
| Registration | NCT03537014 | NCT04077437 |
| Recruitment | November 2018 to May 2020 | August 2020 to May 2022 |
| Sites | 15 (11 United States, 2 Canada, 2 Israel) | 13 (11 United States, 2 Israel) |
| Randomized and analyzed | 90 (46 MDMA, 44 placebo); 9 of 91 confirmed participants were in Canada | 104 (53 MDMA, 51 placebo) |
| Baseline CAPS-5 required | 35 or more (severe) | 28 or more (moderate or severe) |
| Mean baseline CAPS-5 | 44.1 | 39.0 |
| Assigned female at birth | 65.6% | 71.2% |
| Race other than White | 23.3% | 33.7% (and 26.9% Hispanic or Latino) |
| Mean duration of PTSD | 14.1 years | 16.2 years |
| Lifetime MDMA use reported at entry | 32.2% | 46.2% |
| Did not complete | 4 MDMA, 7 placebo (did not complete all three sessions) | 1 MDMA, 8 placebo (discontinued the trial) |
3.1.1 Who Could Take Part
Participants had long-standing PTSD with heavy comorbidity: more than 90 percent had major depression in both trials, most reported developmental trauma (trauma in childhood), and more than 80 percent reported suicidal ideation at some point in their lives (Mitchell et al., 2021, 2023). Exclusions were also extensive. The trials excluded people with a primary psychotic disorder, bipolar I disorder, dissociative identity disorder, a current personality disorder or an eating disorder with active purging; people with a current serious risk of suicide (a history of attempts was allowed); people with uncontrolled hypertension, a history of arrhythmia or a prolonged QT interval; and people with a recent substance use disorder, apart from milder alcohol or cannabis use disorders (FDA, 2024). All psychiatric medications were stopped before baseline. MAPP2 also excluded people who had used ecstasy more than 10 times in the previous 10 years. The results therefore apply to a narrower group than people with PTSD in general care.
3.1.2 The Comparator and the Funding
The comparator was an inactive placebo given with the same therapy. At an end-of-phase 2 meeting in 2016, the FDA was concerned that participants would recognize their allocation and suggested a comparator with noticeable effects, such as niacin or a low dose of MDMA; the sponsor argued that low-dose MDMA had increased anxiety in earlier participants and that niacin or stimulants could worsen PTSD, and the two did not reach agreement on the adequacy of the blind (FDA, 2024). MAPP1 was sponsored by MAPS, which supplied the MDMA and funded the trial from private donations, and was organized by MAPS Public Benefit Corporation (MAPS PBC), a company wholly owned by MAPS that designed the study, monitored and analyzed the data and approved the paper; several authors were employed by MAPS or MAPS PBC (Mitchell et al., 2021). MAPS PBC later became Lykos Therapeutics.
3.2 Primary and Secondary Results
Both trials met their primary and key secondary endpoints. The figures below come from the published reports, which used a mixed model for repeated measures; the least squares (LS) means it produces are model-based estimates that account for repeated measurements over time.
3.2.1 MAPP1
In MAPP1, the model-estimated difference between arms in the change in CAPS-5 total severity at 18 weeks was 11.9 points (95% confidence interval 6.3 to 17.4) in favour of MDMA-assisted therapy (Mitchell et al., 2021). Among participants who completed all three sessions, the mean change was −24.4 points (standard deviation 11.6) with MDMA and −13.9 points (standard deviation 11.5) with placebo. The standardized effect size was d = 0.91. Functional impairment on the Sheehan Disability Scale also improved more with MDMA (d = 0.43).
3.2.2 MAPP2
In MAPP2, the LS mean change in CAPS-5 was −23.7 points (95% confidence interval −26.94 to −20.44) with MDMA-assisted therapy and −14.8 points (−18.28 to −11.28) with placebo, a difference of −8.9 points (−13.70 to −4.12) and an effect size of d = 0.7 (Mitchell et al., 2023). Functional impairment improved by 3.3 points compared with 2.1 (d = 0.4). ICER's meta-analysis of the two trials estimated a pooled difference of −10.2 points (95% confidence interval −13.80 to −6.56) and a pooled d of 0.8 (ICER, 2024).
3.2.3 Expressing the Difference as an Effect Size
A standardized mean difference, Cohen's d, expresses the difference between groups in standard deviation units so that results on different scales can be compared.
Among MAPP1 completers the mean changes were −24.4 and −13.9, a difference of 10.5 points. The report gives a pooled standard deviation of 11.55 (Mitchell et al., 2021). Then d = 10.5 ÷ 11.55 = 0.91, matching the published value. By common convention, a d of 0.8 or more is called large, although that label says nothing about whether the comparison was fair.
3.2.4 The Placebo-with-Therapy Arm
The placebo arm improved by about 14 points in both trials. Within-group effect sizes were 2.1 with MDMA and 1.2 with placebo in MAPP1, and 1.95 and 1.25 in MAPP2 (Mitchell et al., 2021, 2023). Possible explanations, which can act together, include the therapy itself, many hours of attention from two therapists, the natural course of PTSD, regression to the mean (people selected for high scores tend to score lower when measured again) and expectation. The MAPP2 authors suggested that the placebo-arm change might reflect the standalone value of the inner-directed therapy and called for head-to-head comparisons with established psychotherapies (Mitchell et al., 2023). The size of the placebo response is also a reminder that the between-group difference, about 9 to 12 points, is the estimate of what MDMA added.
3.3 Responders, Remission and the Number Needed to Treat
The categories defined in Section 1 (2.1 Measuring PTSD with the CAPS-5) translate the score changes into outcomes that matter to patients. Table 9.4 gives the counts reported in the trial papers, and Figure 9.2 shows the loss-of-diagnosis and remission rates.
Table 9.4. Responder categories at 18 weeks among participants with a final assessment (Mitchell et al., 2021, 2023). Absolute differences and numbers needed to treat (NNT) were calculated for this lesson.
| Trial and outcome | MDMA | Placebo | Absolute difference | NNT (unrounded) |
|---|---|---|---|---|
| MAPP1 loss of diagnosis | 28/42 (67%) | 12/37 (32%) | 34 points | 2.9 |
| MAPP1 remission | 14/42 (33%) | 2/37 (5%) | 28 points | 3.6 |
| MAPP2 responder | 45/52 (87%) | 29/42 (69%) | 17 points | 5.7 |
| MAPP2 loss of diagnosis | 37/52 (71%) | 20/42 (48%) | 24 points | 4.2 |
| MAPP2 remission | 24/52 (46%) | 9/42 (21%) | 25 points | 4.0 |
Lesson 8 introduced the number needed to treat, the number of people who would need to be treated with the intervention, compared with the comparator, for one additional person to reach the outcome. It is the reciprocal of the absolute difference in proportions.
In MAPP2, 37 of 52 participants lost the diagnosis with MDMA (0.7115) and 20 of 42 with placebo (0.4762). The absolute difference is 0.7115 − 0.4762 = 0.2353, the relative risk is 0.7115 ÷ 0.4762 = 1.49, and the NNT is 1 ÷ 0.2353 = 4.25. The trial report gives an NNT of four, which corresponds to rounding to the nearest whole number; the more cautious convention of rounding up gives five (Mitchell et al., 2023).
Which denominator should be used? The percentages in Table 9.4 use the participants included in the analysis. Eight of 51 placebo participants in MAPP2 left the trial, compared with one of 53 in the MDMA arm, and one further placebo participant completed the trial but was not included in the analysis (Mitchell et al., 2023). If everyone randomized is counted and the ten participants who were not analyzed are treated as not having lost the diagnosis, the rates become 37 of 53 (0.698) and 20 of 51 (0.392), and the NNT falls to 3.3. Neither choice is neutral. Treating dropouts as failures is the conservative choice when dropout is similar in both arms, but here it lowers the placebo rate much more than the MDMA rate, and the higher dropout in the placebo arm may itself reflect disappointment after participants recognized their allocation (Mitchell et al., 2023). Use Calculator 9.2 to repeat Worked Example 9.3 and to see how each choice changes the result.
In Calculator 9.2, load the MAPP1 remission preset (14 of 42 against 2 of 37). Change the placebo count from 2 to 4, and then to 6. How much does the NNT change when two or four more placebo participants reach remission? Then load the two MAPP2 loss-of-diagnosis presets and write one sentence explaining to a patient why the same trial can be summarized with an NNT of 3.3 or 4.2.
3.4 Safety in the Trials
The trials paid special attention to adverse events of special interest designated with the FDA: suicidality, cardiac events that could indicate QT prolongation, and events bearing on abuse potential, the regulatory term for a drug's likelihood of non-medical use, which Lesson 11 explains (Mitchell et al., 2021). The four tabs summarize what was found and what was not measured.
Nearly all participants in both arms had at least one treatment-emergent adverse event. In MAPP2, effects more common with MDMA included muscle tightness (58.5% against 25.5% with placebo), nausea (45.3% against 21.6%), decreased appetite (35.8% against 9.8%) and excessive sweating (34.0% against 5.9%); most were mild or moderate and transient (Mitchell et al., 2023). Five participants with MDMA (9.4%) and two with placebo (3.9%) had a severe adverse event, and there were no serious adverse events. MAPP1 reported a similar profile (Mitchell et al., 2021).
MDMA produced expected, dose-dependent and transient rises in blood pressure and heart rate during sessions (Mitchell et al., 2021, 2023). In MAPP1, two participants had a temperature of 38.1 °C after a session. In MAPP2, cardiac events in the MDMA arm, mainly palpitations, were mild, and one participant with a history of hypertension who was not taking antihypertensive medication had a moderate rise in blood pressure. The FDA later judged that the cardiac safety profile was not well characterized and that the assessment of the QT interval was incomplete (FDA, 2024). Lesson 6 discussed cardiovascular risk with MDMA.
Suicidal ideation was common at entry, as expected in severe PTSD. In MAPP1, the only serious adverse events occurred in two placebo participants (two events of suicidal behaviour and one of suicidal ideation leading to self-admission to hospital), and suicidality events of special interest occurred in three MDMA participants (6.5%) and five placebo participants (11.4%) (Mitchell et al., 2021). In MAPP2, two participants in each arm had suicidality events of special interest, and one MDMA participant without suicidal ideation at baseline developed active ideation with some intent (Mitchell et al., 2023). ICER's meta-analysis found very low certainty that MDMA-assisted therapy does not increase suicidal ideation (relative risk 0.89, 95% confidence interval 0.64 to 1.24), and cited an independent review that found discrepancies between the publications and the trial registry records for four serious adverse events (ICER, 2024, citing Colcott et al., 2024). Lesson 6 developed suicidality signals in psychedelic trials.
Some effects were outside the trials' data. Positive or neutral drug effects, such as euphoria, were not recorded as adverse events, which the FDA said limited the assessment of abuse potential, and liver function tests were not collected in phase 3 (FDA, 2024). No adverse events of non-medical use, misuse, dependence or diversion were reported during the trials, but in the follow-up study 13 participants from the MDMA arms and 7 from the placebo arms reported using illicit MDMA after the parent trial (FDA, 2024). Section 3 returns to these gaps.
The safety data describe a selected, closely monitored population over 18 weeks and leave open questions about rarer events, longer follow-up and use outside trial conditions. A 2024 systematic review and meta-analysis of MDMA-assisted therapy across indications found its side effects largely transient and mild or moderate, and it judged that no included randomized trial adequately followed the CONSORT Harms 2022 recommendations for reporting harms (Colcott et al., 2024).
4.1 Durability
The primary outcome was measured about two months after the last experimental session, which shows short-term benefit. Durability, meaning whether the benefit lasts, was assessed in a separate observational study, MPLONG (NCT05066282), in which participants from the phase 3 trials attended a single follow-up visit at least six months after their parent trial ended (FDA, 2024). Table 9.5 shows the FDA's summary.
Table 9.5. Least squares mean change in CAPS-5 from baseline at the end of the parent trial and at the MPLONG follow-up visit, in participants who completed a follow-up assessment (FDA, 2024, Table 8). Values are rounded to one decimal place, and the further change comes from the FDA's unrounded figures, so in the MAPP1 placebo row −15.96 and −15.61 give +0.3 although the rounded entries differ by 0.4.
| Group | Participants | End of parent trial | Follow-up visit | Further change |
|---|---|---|---|---|
| MAPP1, MDMA | 27 | −24.8 | −30.3 | −5.5 |
| MAPP1, placebo | 29 | −16.0 | −15.6 | +0.3 |
| MAPP2, MDMA | 44 | −24.4 | −28.0 | −3.6 |
| MAPP2, placebo | 37 | −15.4 | −16.1 | −0.7 |
The pattern in Table 9.5 suggests that benefit was maintained and even grew in the MDMA arms. The FDA noted that the persistence of the difference "might suggest" a true treatment effect, but it also concluded that MPLONG "was not properly planned, designed, and implemented to evaluate durability" (FDA, 2024). About a quarter of participants were lost between the parent trial and the follow-up visit; the median interval was 23.4 months for MAPP1 participants and 8.5 months for MAPP2 participants; MAPP1 participants had been told their allocation before they enrolled; and some participants had used other treatments in the interval, including ketamine and 5-MeO-DMT. An earlier pooled analysis of long-term follow-up from the phase 2 trials was retracted in 2024 for reasons described in Section 3. The FDA's final guidance on psychedelic trials, issued in July 2026, describes blinded long-term follow-up, typically of 12 months, with prespecified criteria for retreatment as the most informative design, and the MDMA program reviewed in 2024 had no such follow-up (FDA, 2026).
4.2 Comparison with Trauma-Focused Psychotherapies
A person with PTSD and a clinician choosing a treatment want to know whether MDMA-assisted therapy works better than the trauma-focused psychotherapies already recommended. ICER's 2024 review found no head-to-head trials of MDMA-assisted therapy against these psychotherapies (ICER, 2024). As of October 2026, randomized comparisons had been registered, such as a small trial in veterans of MDMA-assisted cognitive processing therapy against cognitive processing therapy alone, which was still recruiting (ClinicalTrials.gov, 2025a), so the question remains open and indirect comparisons are easy to misread. The box below sets out the main reasons.
Why indirect comparisons mislead
The MAPP trials compared MDMA with therapy against the same therapy with placebo, so their between-group effect size estimates what the drug added to an active treatment. Many trials of trauma-focused psychotherapy compare the therapy with a waitlist or usual care, and ICER noted that such inactive comparators may inflate effect sizes (ICER, 2024). In a network meta-analysis of 90 trials, eye movement desensitization and reprocessing (standardized mean difference −2.07) and trauma-focused cognitive behavioural therapy (−1.46) had large effects compared with waitlist, with evidence of moderate to low quality (Mavranezouli et al., 2020). The populations also differ: the MAPP trials stopped all psychiatric medications and excluded many medical and psychiatric conditions. Outcome definitions and timing differ as well; ICER compared remission of about 40 percent with MDMA across the MAPP trials with 53 percent in a meta-analysis of cognitive behavioural therapy trials and cautioned that the definitions of remission were not the same (ICER, 2024).
4.2.1 Dropout and Tolerability
Dropout is one area where MDMA-assisted therapy has looked favourable. In a large trial among 916 United States veterans, 55.8 percent left prolonged exposure and 46.6 percent left cognitive processing therapy before completing it, although both therapies produced substantial improvement and prolonged exposure was slightly better by a margin judged not clinically significant (Schnurr et al., 2022). A meta-analysis found a pooled dropout of 16 percent across randomized trials of psychological therapies for PTSD (Lewis et al., 2020). In the MAPP trials, about 5 percent of participants left the MDMA arm, and the risk of dropout was lower than in the placebo arm (relative risk 0.32); ICER noted that intensive formats of trauma-focused therapy have also achieved a pooled attrition of 5 percent, and that the dropout difference within the MAPP trials may partly reflect functional unblinding (ICER, 2024).
4.2.2 What the Assessors Concluded
ICER rated the evidence comparing MDMA-assisted therapy with trauma-focused psychotherapies as insufficient, and its independent appraisal committee voted unanimously (15 to 0) that the evidence was not adequate to demonstrate a superior net health benefit (ICER, 2024). The MAPP2 investigators wrote that the superiority of MDMA-assisted therapy over SSRIs cannot be assumed without a direct comparison, and, separately, that head-to-head studies are needed to evaluate whether the manualized therapy offers more than the first-line cognitive behavioural therapy and prolonged exposure (Mitchell et al., 2023). A 2026 systematic review and meta-analysis of randomized trials of MDMA-assisted therapy for PTSD rated the overall certainty of the evidence very low and called for larger, higher-quality trials with active controls and long-term follow-up (Fares-Otero et al., 2026). The accordion below adds three further points of comparison.
MAPP participants had severe, long-standing PTSD with depression and childhood trauma, but they were medically screened, free of psychiatric medication and willing to attend an intensive protocol. Veterans in the Schnurr et al. (2022) trial were recruited from outpatient clinics at 17 Veterans Affairs medical centres.
Low dropout in the MDMA arms is a genuine strength of the data, and its interpretation is complicated by the higher dropout of placebo participants who could tell they had not received MDMA.
In the network meta-analysis, eye movement desensitization and reprocessing and trauma-focused cognitive behavioural therapy showed sustained effects at one to four months of follow-up (Mavranezouli et al., 2020). The durability evidence the FDA reviewed in 2024 for MDMA-assisted therapy came from one observational follow-up study, with the limits described in 4.1 Durability.
Summary
MAPP1 and MAPP2 were randomized, double-blind phase 3 trials in which MDMA-assisted therapy reduced CAPS-5 scores by about 9 to 12 points more than the same therapy with placebo (d of 0.91 and 0.7), with more participants losing the diagnosis or reaching remission. The placebo arm also improved substantially. Numbers needed to treat of about three to six depend on the outcome and on how dropouts are counted. Safety data showed transient cardiovascular effects and no excess of suicidality in a screened population, with gaps in what was recorded. Follow-up suggests durable benefit, with important limits, and ICER's 2024 review found no direct comparison with trauma-focused psychotherapy. A 2026 meta-analysis rated the overall certainty of the evidence very low (Fares-Otero et al., 2026), and this research is still evolving. Section 3 follows these data through the 2024 FDA review.
Key Takeaways
- 3.1 MAPP1 (90 participants with severe PTSD) and MAPP2 (104 participants with moderate or severe PTSD) randomized participants one to one to MDMA-assisted therapy or to the same therapy with an inactive placebo. Participants had long-standing PTSD with heavy comorbidity, but extensive exclusions and the stopping of psychiatric medications mean the results apply to a narrower group than people with PTSD in general care. MAPS funded the trials, and MAPS PBC designed, monitored and analyzed them.
- 3.2 Both trials met their primary endpoints, with between-group differences of 11.9 and 8.9 CAPS-5 points and effect sizes of d = 0.91 and d = 0.7. The placebo-with-therapy arm improved by about 14 points, so the drug's contribution is estimated by the between-group difference, which is much smaller than the full change in the MDMA arm.
- 3.3 The number needed to treat is the reciprocal of the absolute difference in proportions; for loss of diagnosis in MAPP2 it is 4.25 among participants analyzed and 3.3 when everyone randomized is counted. Numbers needed to treat of about three to six depend on the outcome chosen and on whether participants who left the trial are counted, because dropout was much higher in the placebo arms.
- 3.4 MDMA caused common, mostly mild and transient effects such as muscle tightness and nausea, and transient rises in blood pressure and heart rate. There was no excess of suicidality in screened participants, a finding ICER rated as very low certainty, and positive effects such as euphoria and liver tests were not recorded in phase 3.
- 4.1 In the MPLONG follow-up, benefit appeared to be maintained or to grow in the MDMA arms. The FDA concluded that the study was not designed to evaluate durability, because about a quarter of participants were lost, intervals varied widely, MAPP1 participants knew their allocation and some used other treatments in between.
- 4.2 ICER's 2024 review found no head-to-head trials of MDMA-assisted therapy against trauma-focused psychotherapies, and differences in comparators, populations and outcome definitions make indirect comparisons misleading. Low dropout in the MDMA arms is a genuine strength whose interpretation is complicated by unblinding, and ICER rated the comparative evidence insufficient.
Reflection
In MAPP2, the least squares mean change in CAPS-5 total severity at 18 weeks was −23.7 points with MDMA-assisted therapy and −14.8 points with placebo plus the same therapy (d = 0.7). Among participants with a final assessment, 37 of 52 in the MDMA arm and 20 of 42 in the placebo arm no longer met the diagnostic criteria (number needed to treat 4.2). If everyone randomized is counted (53 and 51) and the ten participants who were not analyzed are treated as still meeting the criteria, the number needed to treat is 3.3; nine of the ten were in the placebo arm. Write a short paragraph for a clinician colleague that summarizes what these figures show, states which number needed to treat you would report and why, and explains what the 14.8-point improvement in the placebo arm implies for how the benefit of MDMA should be described.
MAPP2 found that adding MDMA to a manualized inner-directed therapy reduced PTSD symptoms by about nine CAPS-5 points more than the same therapy with placebo, a moderate to large standardized effect (d = 0.7), and that roughly one additional person lost the diagnosis for every four treated. I would report the completer-based figure of 4.2 as the main estimate, with 3.3 as a sensitivity analysis, because counting dropouts as failures is not neutral here: nine of the ten participants who were not analyzed were in the placebo arm, eight of them because they left the trial, possibly because they could tell they had not received MDMA, so that approach makes the drug look better partly through differential dropout. The 14.8-point improvement with placebo shows that much of the change in either arm came from the therapy, attention, time or expectation, so the benefit should be described as the drug's addition to an intensive psychotherapy, about nine points, which is much smaller than the 23.7-point change seen in the MDMA arm. The trial does not tell us how MDMA would perform with other therapies or without one.
Minimum 20 characters required.
Question 1: In MAPP2, the least squares mean change in CAPS-5 was −23.7 with MDMA and −14.8 with placebo plus the same therapy. What does the placebo-arm change most directly imply?
Question 2: In MAPP2, 37 of 52 MDMA participants and 20 of 42 placebo participants with a final assessment lost the diagnosis. What is the unrounded number needed to treat?
Question 3: Why did the FDA conclude that the MPLONG follow-up study could not establish durability with confidence?
Question 4: Why can the MAPP effect sizes not be compared directly with effect sizes for trauma-focused psychotherapy from meta-analyses?
The 2024 FDA Decision and After
Content note
This section refers to unethical conduct, including sexual misconduct, by therapists at a Canadian trial site (described in Section 4) and to suicidality in trials. Support is available at any time from the 9-8-8 Suicide Crisis Helpline (call or text 9-8-8), the Hope for Wellness Help Line for Indigenous people across Canada (1-855-242-3310) and, for people in British Columbia affected by sexual violence, VictimLinkBC (1-800-563-0808, call or text).
Introduction
On 4 June 2024, an advisory committee of independent experts met to advise the United States Food and Drug Administration (FDA) on whether to approve MDMA, under its generic name midomafetamine, for use with psychological intervention in adults with PTSD. The committee voted 2 to 9 that the data showed the treatment was effective and 1 to 10 that its benefits outweighed its risks (Pharmaceutical Executive, 2024). Two months later the FDA declined to approve the application. Supporters saw the decision as a failure to recognize an effective treatment for a serious condition; critics saw it as a fair response to trials whose design and conduct made the results hard to trust. This section sets out the evidence on each disputed point together with the responses of the sponsor and investigators, so that the decision can be judged on its merits. Lesson 11 develops drug regulation in general; this section applies it to one application.
Learning Objectives
- 5.1 Trace the regulatory history of MDMA-assisted therapy from the first investigational application to the 2024 advisory committee meeting.
- 5.2 Explain the issues the FDA put to its advisory committee, including functional unblinding, expectancy, durability, the role of psychotherapy and the recording of adverse events, and report the committee's votes.
- 5.3 Present each criticism of the trials together with the sponsor's and investigators' responses, and judge what each side's evidence can establish.
- 6.1 Describe the complete response letter and the 2024 retractions, and explain what a retraction does and does not say.
- 6.2 State the status of the program as of October 2026, including the FDA's final guidance on psychedelic trials and the reported resubmission.
- 6.3 Describe Australia's 2023 authorization and explain how rescheduling a substance for prescriber use differs from approving a medicine.
5.1 The Regulatory Path to 2024
The sponsor, MAPS, opened an investigational new drug application with the FDA in 2001, ran phase 1 and phase 2 studies, and moved toward phase 3 in 2016 (FDA, 2024). Several regulatory steps along the way shaped the trials described in Section 2.
- At an end-of-phase 2 meeting in 2016, the FDA raised the concern that participants would be able to tell whether they had received MDMA, and the two sides did not agree on the adequacy of the blind (FDA, 2024).
- A special protocol assessment, a process in which a sponsor seeks the FDA's agreement that a trial's design could support approval, ended in agreement on the MAPP1 protocol on 28 July 2017. The FDA's briefing notes that such an agreement does not guarantee approval (FDA, 2024).
- The FDA granted Breakthrough Therapy designation, an expedited program for drugs for serious conditions with promising early clinical evidence, on 15 August 2017 (FDA, 2024).
- After both trials were complete, MAPS PBC submitted a new drug application on 11 December 2023 (FDA, 2024). By the time of the advisory committee meeting, the company had been renamed Lykos Therapeutics.
Interactive 9.1 places these steps on a single timeline with the trial results, the integrity events described in Section 4 and the decisions made outside the United States. Select a category to narrow the list, and select an event to read what happened and where the information comes from.
Two features of the timeline in Interactive 9.1 matter for what follows. The FDA's concern about blinding was on record from 2016, before either phase 3 trial began, and the integrity events at the Canadian site were disclosed in 2019, while MAPP1 was enrolling. Both resurfaced in 2024.
5.2 What the FDA Asked Its Advisory Committee
An advisory committee gives the FDA independent expert advice; its votes are not binding, and the agency makes the decision. For this meeting the FDA's briefing document stated that the application included "two positive studies" with clinically meaningful improvement that appeared durable for several months, and it then set out the uncertainties it wanted the committee to weigh (FDA, 2024).
5.2.1 Functional Unblinding and Expectancy
Lesson 5 explained functional unblinding: when a drug has effects that participants can feel, a nominally double-blind trial becomes open in practice. The FDA wrote that MDMA's effects on mood, sensation and cognition make trials "nearly impossible to blind" and that in-session monitors could probably also deduce a participant's allocation (FDA, 2024). At the FDA's request, MAPP2 added a blinding survey at the end of the trial, summarized in Table 9.6.
Table 9.6. Participants' beliefs about their allocation at the end of MAPP2 (FDA, 2024, Table 9; Mitchell et al., 2023).
| Belief | MDMA arm (n = 52) | Placebo arm (n = 44) |
|---|---|---|
| Received MDMA: "I am positive" | 41 (78.8%) | 2 (4.5%) |
| Received MDMA: "I think" | 8 (15.4%) | 7 (15.9%) |
| Cannot tell | 2 (3.8%) | 2 (4.5%) |
| Received placebo: "I think" | 0 (0.0%) | 14 (31.8%) |
| Received placebo: "I am positive" | 1 (1.9%) | 19 (43.2%) |
| Correct belief | 49 (94.2%) | 33 (75.0%) |
Across both arms, 82 of 96 respondents (85 percent) correctly identified their allocation. The FDA argued that expectation bias could therefore have acted on both arms, inflating improvement among participants who knew they had received MDMA and possibly deflating it among those who knew they had not, and that its size "cannot be quantified" (FDA, 2024). It also noted that the placebo arm improved, so disappointment does not appear to have made placebo participants worse. A 2026 systematic review of 112 randomized trials of psychedelics for psychiatric disorders found functional unblinding pervasive among participants and raters, with blinding failure above 85 percent in the MDMA trials that used an inert placebo, and found that few trials assessed blinding or expectancy (Orsini et al., 2026).
5.2.2 The Other Issues
The briefing document listed four further concerns (FDA, 2024). Durability rested on the follow-up study described in Section 2 (4.1 Durability), with its losses to follow-up, variable timing and intervening treatments. The contribution of psychotherapy "has not been characterized," because every arm received it and the integration sessions were not standardized. The safety data were incomplete in specific ways: the cardiac profile was not well characterized, liver function was not tested in phase 3, and positive or neutral effects such as "euphoria" or "elated mood" were not recorded as adverse events, which limited the assessment of abuse potential and of how long impairment lasts. Finally, because participants are impaired and suggestible for several hours, the FDA described the controls it would require if the drug were approved: a risk evaluation and mitigation strategy (REMS) restricting use to certified settings, with at least two health care providers on site (one licensed) monitoring each patient for at least eight hours, discharge to an accompanying adult, follow-up after each session and a patient registry.
Committee members raised possible bias and misconduct, unanswered questions about abuse potential, gaps in the data and limited evidence for the benefit of the psychological approach (Pharmaceutical Executive, 2024). Lykos's chief executive officer, Amy Emerson, said the company was "disappointed in today's vote given the urgent unmet need in PTSD" and remained "committed to working with the FDA to address outstanding questions" (Pharmaceutical Executive, 2024).
5.3 Weighing the Criticisms and the Responses
Two other assessments informed the debate. The Institute for Clinical and Economic Review (ICER), an independent organization in the United States that evaluates the evidence and value of new treatments, rated the evidence for MDMA-assisted therapy against the same therapy with placebo as insufficient in its final report of 27 June 2024 (ICER, 2024). Its rating rested partly on interviews with trial participants, a therapist and others, which it reported while acknowledging that people raising concerns also held strong views. Each tab below sets one criticism beside the responses on record.
Criticism. Nearly all participants who received MDMA knew it (Table 9.6), about 40 percent of participants across the trials had used MDMA before, and ICER heard from multiple experts that the investigators, therapists and patients involved in the trials held very strong prior beliefs in the treatment's value, which heightens the risk of bias when the blind fails (ICER, 2024; FDA, 2024).
Responses. The MAPP2 investigators acknowledged that expectancy was not measured and argued that three observations support its mitigation: the arms did not separate after the first session, placebo-arm dropouts did not cluster after the first session, and not every participant guessed correctly. They also relied on blinded independent raters (Mitchell et al., 2023). An inactive placebo had been agreed with the FDA under the special protocol assessment (FDA, 2024). In 2026 MAPS founder Rick Doblin said that "despite partial functional unblinding, the therapists provided high-quality care" to both arms and that he believed the data were valid (MAPS, 2026).
What the evidence can show. The investigators' observations make a large expectancy effect less likely without excluding a moderate one, and the critics' evidence shows the blind failed without showing how much of the effect it explains.
Criticism. The FDA had advised that adverse events relevant to abuse potential be documented, yet the trials did not capture positive or neutral effects such as euphoria, leaving no verbatim descriptions of participants' experiences and no data on the onset and duration of effects (FDA, 2024). ICER heard reports of severe negative outcomes that were not attributed to the treatment, and it could not determine their frequency (ICER, 2024).
Responses. The MAPP1 investigators wrote that anticipated effects such as euphoria, stimulation and feelings of closeness "were intentionally not solicited as adverse events to avoid biasing the collection of adverse event data" (Mitchell et al., 2021). The FDA observed that the sponsor likely did not treat these effects as adverse events because adverse events are defined as untoward (FDA, 2024). In 2026, Doblin stated that adverse events and outcomes were reported accurately, while acknowledging that positive adverse events were not gathered at the time (MAPS, 2026).
What the evidence can show. The disagreement concerns definitions and completeness more than accuracy: both sides agree that positive effects were not recorded, and they differ on whether that mattered.
Criticism. ICER reported hearing that some participants felt pressured to report good outcomes and suppress bad ones, that some participants had close relationships with those running the trials, and that some were kept out of the long-term follow-up study (ICER, 2024).
Responses. The principal investigator of the MAPP trials replied in a public comment to ICER: "This is untrue. No participants were prevented from entering the phase 3 long-term follow-up study." ICER later received correspondence from one participant that appeared to show an exclusion, which it could not fully investigate and which the participant said had been sent to the FDA (ICER, 2024).
What the evidence can show. ICER stated that it was "not able to assess the frequency of misreporting of benefits and/or harms" (ICER, 2024). Interviews can show that problems occurred for some participants; they cannot show how common the problems were or whether they changed the results.
Criticism. Because every arm received the same therapy, the trials cannot show how much the therapy contributes, whether a different therapy would work as well, or whether the drug would work without it (FDA, 2024).
Responses. The sponsor's position from the outset was that MDMA facilitates psychotherapy, and the MAPP2 investigators suggested that the large placebo-arm improvement may reflect the standalone value of the manualized therapy (Mitchell et al., 2023). MAPS co-executive director Betty Aldworth argued in 2026 that the drug evaluation and regulation system "was not designed to evaluate drugs delivered as part of therapy" (MAPS, 2026).
What the evidence can show. This issue is a matter of design. The FDA's 2026 guidance notes that a factorial design, which randomizes drug and therapy separately, may be useful for answering it (FDA, 2026).
Table 9.7 condenses the four tabs into a single matrix and adds, for each concern, the kind of study that could resolve it.
Table 9.7. The main criticisms of the MDMA trials, the responses on record and the evidence that could resolve each.
| Concern | Raised by | Response on record | What could resolve it |
|---|---|---|---|
| Functional unblinding and expectancy | FDA; ICER | Arms did not separate early; independent raters; inactive placebo agreed with FDA | Active comparator; expectancy measured before randomization; blinding surveys of participants and staff |
| Durability | FDA; ICER | Follow-up shows maintained or increased benefit | Blinded follow-up of about 12 months with prespecified retreatment criteria |
| Role of psychotherapy | FDA; committee members | MDMA is designed to facilitate therapy | Factorial designs that separate drug and therapy |
| Incomplete safety data | FDA | Anticipated effects not solicited to avoid bias | Systematic recording of all drug effects, liver tests and a complete QT assessment |
| Trial conduct and reporting pressure | ICER interviewees | Principal investigator denied exclusions from follow-up | Independent audit; trial staff and therapists without ties to the sponsor |
6.1 The Complete Response Letter and the Retractions
On 9 August 2024 the FDA issued a complete response letter, the formal notice that an application cannot be approved in its present form. Lykos reported that the letter stated that the FDA could not approve the treatment on the data submitted and asked for another phase 3 trial (Lykos Therapeutics, 2024); MAPS later summarized the letter as recommending further research on the durability of response, the characterization of safety and the mitigation of bias (MAPS, 2026). Lykos said it would request a meeting to ask the FDA to reconsider (Fierce Biotech, 2024).
The following day, the journal Psychopharmacology retracted three papers from the phase 2 program: a pooled analysis of six phase 2 trials used to design phase 3 (Mithoefer et al., 2019), a long-term follow-up of the same trials (Jerome et al., 2020), and an analysis of how stopping antidepressants affected response (Feduccia et al., 2021). Each notice states that the editors retracted the paper "after they were informed of protocol violations amounting to unethical conduct at the MP4 study site," that the authors "were aware of these violations at the time of submission" but did not disclose them or remove the site's data, and that the authors "did not fully declare a potential competing interest" (Psychopharmacology, 2024a, 2024b, 2024c). MP4 was the Vancouver site described in Section 4.
The authors' positions differed. In each notice one author agreed with the retraction but not with the wording of the notice and several authors disagreed; in the first notice one author neither agreed nor disagreed, and in the first two some authors did not respond (Psychopharmacology, 2024a, 2024b, 2024c). Lykos said it disagreed and believed corrections would have been sufficient, that it had filed a complaint with the Committee on Publication Ethics, and that it regretted not disclosing the violations to the journal, although it had reported them to the FDA, Health Canada and the ethics boards (Fierce Biotech, 2024). MAPS's own statement places those reports shortly after the participant's complaints reached it in October 2018, and its public disclosure in May 2019 (MAPS, 2019). It said that reanalyses excluding the four participants treated at the site were consistent with the originals, and it noted that another journal had chosen a correction for a paper using the same data (Fierce Biotech, 2024). The retracted papers reported phase 2 analyses and were not the studies on which the application rested (Fierce Biotech, 2024), and as of October 2026 neither phase 3 trial report had been retracted.
What a retraction does and does not say
A retraction removes a paper from the scientific record; a correction amends it and leaves it in place. The Psychopharmacology notices give two grounds: undisclosed knowledge of unethical conduct at a site whose data were included, and incomplete declaration of competing interests. They do not state that the pooled results were wrong. The scientific consequence is that the phase 2 pooled analyses, including the pooled long-term follow-up of the phase 2 trials, can no longer be cited as evidence, and the case shows that what authors disclose to journals about a site's conduct is itself part of research integrity.
6.2 The Status of the Program as of October 2026
After the complete response letter, the company was renamed Resilient Pharmaceuticals, and as of October 2026 the trial registry records for both phase 3 trials and for the Vancouver phase 2 study list Resilient Pharmaceuticals as their sponsor (HCPLive, 2026; ClinicalTrials.gov, 2025b). In July 2026 the FDA finalized its guidance "Psychedelic Drugs: Considerations for Clinical Investigations," which replaces the June 2023 draft (FDA, 2026). The guidance states that alternatives to an inert placebo, such as other psychoactive drugs that mimic some aspects of the subjective experience, may be considered; asks for central raters blinded to allocation and visit number; suggests blinding questionnaires and an expectancy questionnaire before randomization and at the end of treatment; notes that factorial designs may help to separate the contributions of drug and psychotherapy; describes blinded long-term follow-up, typically of 12 months, with prespecified criteria for retreatment as the most informative design; asks that all central nervous system effects, including euphoria, be documented; and expects two session monitors, a lead monitor with graduate-level training and clinical experience in psychotherapy who is licensed to practise independently and an assistant monitor with a nursing or bachelor's degree and at least one year of clinical experience (FDA, 2026).
On 9 August 2026, the industry publication Psychedelic Alpha reported that Resilient had resubmitted its new drug application without a new phase 3 trial. MAPS responded the next day; Doblin stated that "Resilient's new NDA is being submitted without a new Phase 3 study," and MAPS noted that it has no active role in Resilient's program (MAPS, 2026; HCPLive, 2026). As of October 2026, neither the company nor the FDA had publicly confirmed that the FDA had accepted the resubmission or set a review date. In Canada, no MDMA product has market authorization as of October 2026, and access outside trials depends on the Special Access Program, which Lesson 11 describes. Readers should check the current status, which may have changed.
6.3 Australia's 2023 Authorization
Australia took a different route. On 3 February 2023 the Therapeutic Goods Administration (TGA) announced that from 1 July 2023 MDMA for PTSD, and psilocybin for treatment-resistant depression, would move to Schedule 8 of the Poisons Standard for those uses only, so that psychiatrists could prescribe them (TGA, 2023). A psychiatrist must first obtain approval from a human research ethics committee and then be authorized by the TGA under its Authorised Prescriber scheme. When the change took effect, and still as of October 2026, no product containing MDMA was included in the Australian Register of Therapeutic Goods, so the TGA had not evaluated any MDMA product for safety, quality or efficacy; authorized psychiatrists may use unapproved products (TGA, 2023, n.d.). The TGA stated that the evidence suggested the substances may be effective in certain patients, particularly with psychotherapy in controlled settings, while describing the evidence base as limited (TGA, 2023). As of 31 July 2026, 63 psychiatrists had been authorized to prescribe MDMA; the TGA did not report patient numbers (TGA, 2026).
The Australian decision was a rescheduling that allows specialist prescribing under ethics oversight. The FDA application sought approval of a specific product for marketing in the United States, under restrictions, which requires substantial evidence of effectiveness and adequate safety data. The two decisions therefore answered different questions. The Australian announcement also came seven months before MAPP2 was published. Lesson 11 compares Australia's model with other access routes.
Australia's TGA allowed authorized psychiatrists to prescribe MDMA for PTSD from 1 July 2023, and the FDA declined to approve midomafetamine on 9 August 2024. Using 5.2 What the FDA Asked Its Advisory Committee through 6.3 Australia's 2023 Authorization, write a paragraph that explains how both decisions could be reasonable on largely the same trial data. Name two features of the Australian route that address concerns the FDA raised, and one concern that neither route resolves.
Summary
The FDA's 2024 review acknowledged two positive phase 3 trials and identified uncertainties about functional unblinding and expectancy, durability, the contribution of psychotherapy and incomplete safety data. Its advisory committee voted against effectiveness and benefit-risk, and a complete response letter followed on 9 August 2024, the day before three phase 2 papers were retracted over undisclosed unethical conduct at a Canadian site. The sponsor and investigators answered each criticism, and some disagreements concern definitions and design more than facts. As of October 2026 the renamed sponsor was reported to have resubmitted without a new phase 3 trial, while Australia has allowed authorized psychiatrists to prescribe MDMA since 2023. Section 4 examines the Canadian site and the protections that followed.
Key Takeaways
- 5.1 MAPS opened an investigational new drug application in 2001. The FDA raised its concern about blinding in 2016, agreed the MAPP1 protocol under a special protocol assessment and granted Breakthrough Therapy designation in 2017, and received the new drug application in December 2023. Both the blinding concern and the Vancouver disclosure predated the 2024 review.
- 5.2 The FDA's briefing accepted two positive studies but asked its advisory committee to weigh functional unblinding and expectancy (85 percent of MAPP2 respondents correctly identified their allocation), durability, the uncharacterized contribution of psychotherapy and incomplete safety data, and it described the restrictions it would require if the drug were approved. On 4 June 2024 the committee voted 2 to 9 on effectiveness and 1 to 10 on benefit-risk.
- 5.3 The sponsor and investigators answered each criticism: early non-separation of the arms and blinded raters against unblinding, a deliberate choice not to solicit anticipated effects as adverse events, a denial of exclusions from follow-up, and the view that MDMA facilitates therapy. Their evidence makes a large expectancy effect less likely without excluding a moderate one, and several disagreements turn on definitions and design more than on facts.
- 6.1 The FDA issued a complete response letter on 9 August 2024, and Lykos reported that it asked for another phase 3 trial. The next day Psychopharmacology retracted three phase 2 papers because the authors had not disclosed known unethical conduct at the MP4 site or fully declared competing interests; the notices did not state that the results were wrong, and as of October 2026 the phase 3 reports had not been retracted.
- 6.2 The company, renamed Resilient Pharmaceuticals, was reported in August 2026 to have resubmitted without a new phase 3 trial, and as of October 2026 acceptance had not been publicly confirmed. The FDA's final guidance of July 2026 addresses comparators, blinding and expectancy, factorial designs, blinded 12-month follow-up, the recording of all central nervous system effects and session monitors. As of October 2026, no MDMA product had market authorization in Canada.
- 6.3 From 1 July 2023, Australia allowed psychiatrists with ethics committee approval and TGA authorization to prescribe MDMA for PTSD, and 63 had been authorized by 31 July 2026. Rescheduling for specialist prescribing differs from approving a product, which requires substantial evidence of effectiveness and adequate safety data, so the two decisions answered different questions.
Reflection
Imagine that you sat on the FDA advisory committee on 4 June 2024. You know that (1) two phase 3 trials met their primary endpoints, with between-group effect sizes of d = 0.91 and d = 0.7; (2) in the MAPP2 blinding survey, 94 percent of MDMA participants and 75 percent of placebo participants correctly identified their allocation; (3) positive effects such as euphoria were not recorded as adverse events, which the investigators said was done to avoid biasing adverse event collection; (4) a follow-up visit at least six months later suggested that benefit was maintained, but about a quarter of participants were lost and some had used other treatments; and (5) every arm received the same therapy, so the contribution of the therapy was not characterized. Explain how you would vote on whether the data show effectiveness, giving due weight to both the critics' and the investigators' arguments, and name the single piece of new evidence that would most change your vote.
I would vote that the data do not yet show effectiveness to the standard required for approval, while recognizing that they are encouraging. Two randomized trials agreed with each other, and the investigators reasonably point out that the arms did not separate after the first session and that blinded independent raters scored the outcome. Against that, almost everyone knew their allocation, expectancy was not measured, and participants and therapists held strong beliefs about the treatment, so I cannot tell how much of a nine to twelve point difference reflects the drug. The follow-up data point in the same direction but lost a quarter of participants and mixed in other treatments. The adverse event issue weighs less on effectiveness and more on safety and abuse potential, and here I accept that both sides agree positive effects were not recorded. The evidence most likely to change my vote would be a new phase 3 trial with an active comparator that produces some noticeable effects, expectancy measured before randomization, and blinded follow-up for about 12 months. If that trial showed a similar difference, I would vote yes.
Minimum 20 characters required.
Question 1: How did the FDA's Psychopharmacologic Drugs Advisory Committee vote on 4 June 2024?
Question 2: In the MAPP2 blinding survey, what share of participants who received MDMA correctly believed they had received it?
Question 3: The MAPP1 investigators did not record euphoria or feelings of closeness as adverse events. What reason did they give?
Question 4: Why did Psychopharmacology retract three phase 2 papers in August 2024?
Research Integrity and Participant Protection
Content note
This section describes sexual misconduct by a therapist toward a participant in a Canadian clinical trial, and physical contact during a trial session that departed from the treatment protocol. It does so factually and without graphic detail. Support is available at any time from VictimLinkBC for people in British Columbia affected by sexual violence (1-800-563-0808, call or text), the 9-8-8 Suicide Crisis Helpline (call or text 9-8-8) and the Hope for Wellness Help Line for Indigenous people across Canada (1-855-242-3310).
Introduction
Sections 2 and 3 asked whether the evidence for MDMA-assisted therapy can be trusted. This section asks whether the people who took part were protected, a question connected to the first through the retractions described in Section 3. The course's practice standard is that safety and ethics come first, with consent obtained while the person is sober, explicit agreement about any touch before the session, zero tolerance for sexual or romantic contact, and supervision of therapists. Lesson 12 develops the ethics of altered states, including power, transference and the therapist's own inner work; this section concentrates on the protections that research protocols, sponsors and regulators can build.
Learning Objectives
- 7.1 Explain why the altered state produced by MDMA, combined with participants' trauma histories, raises particular risks for participant protection.
- 7.2 Describe, from published sources, the misconduct at the Vancouver site of a MAPS phase 2 trial, how it came to light and how the sponsor responded.
- 7.3 Compare the safeguards for touch and consent set out in the treatment manual, by Health Canada in 2022 and by the FDA, and identify their limits.
- 7.4 Evaluate video recording as a protection, including what it can and cannot prevent.
- 7.5 Evaluate team composition, supervision and overnight monitoring as protections, including what each can and cannot prevent.
- 7.6 Identify routes by which a participant can raise a concern outside the therapy team, and explain the limits of routes that operate only after the fact.
- 8.1 Apply these safeguards to the review of a trial protocol by a research ethics board.
7.1 Why Altered States Raise the Stakes
The FDA described MDMA as producing "profound alterations in mood, sensation, suggestibility, and cognition" and noted that its subjective effects can persist for several hours, "rendering patients in an impaired and vulnerable state that necessitates safety monitoring" (FDA, 2024). Experts interviewed by ICER described how MDMA breaks down barriers and heightens suggestibility, and warned that this creates "a substantial risk with any therapists who might choose to take advantage of patients" (ICER, 2024).
Several features of MDMA-assisted therapy add to that risk. Participants are people with PTSD, most of whom reported developmental trauma (Mitchell et al., 2021, 2023), and some have been harmed by people who held power over them. Sessions last about eight hours in a closed room, and the manual permits touch. Participants may also hesitate to complain about a treatment they hope will help them or about a community they belong to; ICER reported that some participants felt that negative reports could lead to their being shunned (ICER, 2024). None of these features makes misconduct likely, and together they mean that safeguards must work even when a participant cannot, in the moment, protect their own interests.
Figure 9.3 shows the protections discussed in this section as layers. Each layer has weaknesses, and harm reaches a participant when the weaknesses of several layers line up.
7.2 The Vancouver Site: What Happened and How It Came to Light
The account below draws on the sponsor's public statement, its 2022 update and reporting by CBC News. It does not name the people involved; the sources do, and the participant has chosen to speak publicly.
7.2.1 The Study and the Disclosure
MP4 was a randomized, double-blind phase 2 pilot study of MDMA-assisted therapy for treatment-resistant PTSD, conducted in Vancouver between 2014 and 2016; it enrolled six participants before it was terminated (ClinicalTrials.gov, 2025b). On 24 May 2019, MAPS published a statement disclosing ethical violations by two of its former investigators at this site, who served as co-therapists (MAPS, 2019). According to the statement, a participant enrolled in December 2014 and completed her active treatment in October 2015. In June 2015 she moved near the therapists and continued to receive therapy from them. Around late 2015 or early 2016, while she was still enrolled in the study's follow-up, one of the therapists began a sexual relationship with her. The other therapist reportedly learned of the relationship and tried unsuccessfully to stop it, but did not report it to MAPS or to any regulator. Her participation in the study was ended at her long-term follow-up visit in June 2016.
7.2.2 How It Came to Light
The misconduct reached the sponsor through the participant. In January 2018 she filed complaints with British Columbia's regulatory colleges, and in October 2018 she forwarded them to MAPS PBC (MAPS, 2019). MAPS stated that it had had no prior knowledge and that trial monitoring had not revealed the misconduct. Shortly afterwards it reported the matter to the FDA, Health Canada and the ethics review committee, said it would not work with either therapist again, and paid the participant $15,000 (Canadian) on a compassionate basis to help fund her therapy while her legal action against the therapists continued.
In November 2021 a journalist who had viewed video recordings of the participant's sessions interviewed MAPS staff, and MAPS opened a compliance review two days later. In its March 2022 update, MAPS stated that it had provisionally found that the two therapists "substantially deviated from the MDMA-assisted Therapy Treatment Manual on several occasions" during treatment, and it notified the FDA, Health Canada and the ethics committees (MAPS, 2019, updated 2022). CBC News reported that the recordings, from the participant's second experimental session in 2015, showed extended physical contact by both therapists, including holding her down, and that MAPS staff had not viewed the videos until November 2021, about six years after they were made (CBC News, 2022). A researcher interviewed by CBC noted that the manual permits only nurturing touch and focused bodywork.
7.2.3 Responses and Outcomes
CBC News (2022) also reported that the participant's civil lawsuit against the therapists had been settled out of court, and that the RCMP had recommended criminal charges after her complaint, which Crown prosecutors declined to approve. The therapist with whom the relationship occurred was not a licensed psychologist, and because British Columbia did not regulate therapists or counsellors at the time of the report, no regulatory body could investigate him; the other therapist, a psychiatrist, was the subject of a complaint to the College of Physicians and Surgeons of British Columbia. The therapist acknowledged a sexual relationship after the sessions had ended and disputed the participant's account of it. The participant told CBC that the conduct shown in the video was "not therapy" and, as CBC reported, she believes that if these drugs are legalized for use in psychotherapy there needs to be a strong system in place to keep patients safe. The case later led to the retractions described in Section 3, and ICER concluded that "boundaries, including sexual boundaries, were severely crossed with at least one patient in a Phase II trial" (ICER, 2024).
The safeguards that were in place, and why they were not sufficient
MAPS listed the protections in place at the time: two providers in every session, video recording of visits, study monitoring and clinical supervision, and it stated that none had been sufficient (MAPS, 2019). Read against Figure 9.3, each layer had a gap. The second provider was also involved in the departures from the manual. The recordings existed but were not reviewed for six years. Monitoring and supervision did not detect the conduct, and the relationship developed outside the sessions after the participant moved near the therapists, where session-based safeguards did not reach. One therapist was outside any professional regulator. The only route that worked was the participant's own complaint, made years after the events.
7.3 Touch and Consent
Consent is valid only if the person understands what is proposed and agrees freely, and a person under the effects of MDMA is not in a position to give that agreement reliably. For that reason, the course standard, the treatment manual and Health Canada all place the decision about touch before the drug is given. The manual requires therapists to discuss touch during preparation, to make clear that the participant need not be touched at all, and to agree a stop word that is always obeyed except to protect the participant from physical harm (Mithoefer, 2017). Health Canada's 2022 expectations for trials of psychedelic-assisted psychotherapy state that consent for touch should be obtained before the investigational product is given and again at the time; that any physical contact must be non-intimate and non-sexual; and that touch is optional unless it is needed to prevent imminent harm, such as catching someone who is falling (Health Canada, 2022). MAPS's code of ethics prohibits sexual contact between providers and participants (MAPS, 2019, updated 2022). The flip cards summarize the elements a protocol should contain.
Two limits remain. Consent agreed in advance cannot anticipate every situation in an eight-hour session, which is why the second check and the stop word matter. And consent rules only govern conduct that someone is in a position to observe, which leads to the questions of recording and team composition.
7.4 Video Recording and Its Review
The MAPS trials recorded sessions for several purposes: to assess therapists' adherence to the manual, to train therapists, and to let participants review their own sessions (Mithoefer, 2017). The manual asks therapists to explore during the consent interview how the participant feels about being recorded, to explain that recordings will be used only for research or training, and to turn the cameras off temporarily if the participant asks. The Vancouver case showed the central weakness of recording as a safeguard: a recording deters and detects misconduct only if someone independent watches it within a reasonable time, and these were not watched for six years (CBC News, 2022).
Health Canada's 2022 expectations treat recording as optional. If sessions are recorded, consent must be obtained first, the protocol must state why recording is needed, how the recordings will be used, how often they will be reviewed and how long they will be kept, and the participant may withdraw consent at any time (Health Canada, 2022). The requirement to state a review frequency addresses the weakness the Vancouver case exposed. Recording also has costs: sessions capture people in vulnerable states describing trauma, so storage and access need tight control, and ICER reported that some participants had difficulty obtaining recordings they had been promised (ICER, 2024). Table 9.8 compares the safeguards on recording and the other layers across the main documents.
Table 9.8. Safeguards for participant protection in the treatment manual (2017), Health Canada's expectations for trials (2022) and the FDA's proposed risk controls (2024) and final trial guidance (2026).
| Safeguard | Treatment manual (Mithoefer, 2017) | Health Canada (2022) | FDA (2024 proposal; 2026 guidance) |
|---|---|---|---|
| People in the room | Two co-therapists, generally one female and one male, both present except for brief absences | At least two therapists present; the same therapists throughout the trial for each participant | At least two providers on site, one licensed (2024); a lead monitor and an assistant monitor for the whole session (2026) |
| Qualifications | Appropriate training and experience; in MAPP1, a master's degree or above and one licensed member per team | Trained and licensed where applicable; unlicensed members under direct supervision of a licensed therapist; physician available on site | Lead monitor with graduate-level training and clinical experience in psychotherapy, licensed to practise independently; assistant with a nursing or bachelor's degree and one year of clinical experience; an on-call physician able to reach the site within 15 minutes if the lead is not a physician (2026) |
| Touch | Discussed in preparation; optional; stop word obeyed | Consent before dosing and in the moment; non-intimate and non-sexual; optional except to prevent imminent harm | Not addressed in the 2026 guidance, which asks that consent forms describe the increased vulnerability and suggestibility of a session |
| Recording | Recorded with consent for adherence review and training; cameras may be paused | Optional; consent first; purpose, use, review frequency and retention stated; consent may be withdrawn | Not addressed in the 2026 guidance |
| After the session | Overnight stay with a trained attendant where the protocol requires; daily calls for a week | Lodging available on dosing days; check-in visits in the following days and weeks | Monitoring for at least eight hours; discharge to an accompanying adult; follow-up and a patient registry (2024) |
7.5 Team Composition, Supervision and Overnight Monitoring
7.5.1 Two Therapists, and Their Independence
Every document in Table 9.8 requires at least two people in the room. Health Canada adds that therapists should be licensed by a regulatory body where applicable, that unlicensed team members should work under the direct supervision of a licensed therapist, and that a physician should be readily available on site for the duration of the drug's effects (Health Canada, 2022). The FDA's 2026 guidance expects a lead monitor with graduate-level training and clinical experience in psychotherapy who is licensed to practise independently, and an assistant monitor with a nursing or bachelor's degree and at least one year of clinical experience in a licensed mental health care setting (FDA, 2026). Licensure matters because it brings a therapist under a regulator that can investigate complaints, which was not possible for one therapist in the Vancouver case. A second person in the room protects the participant only to the extent that this person is willing and able to intervene and report. When both members of a team are involved in a departure from the protocol, as MAPS provisionally found at the Vancouver site, the two-person safeguard fails, so a protocol needs a route by which concerns can travel beyond the team.
7.5.2 Supervision
In MAPP1, therapist training included 52 hours of supervision, and adherence was to be checked by independent review of recordings (Mitchell et al., 2021; Mithoefer, 2017). Supervision, in which an experienced clinician regularly reviews a therapist's work, gives a place to examine the strong feelings that intensive sessions can evoke in therapists; Lesson 12 develops transference and countertransference. Supervision depends on what the therapist chooses to bring to it, so it works best alongside independent review of recordings.
7.5.3 Overnight Monitoring and Discharge
The manual describes an overnight stay at the treatment site after each session, accompanied by a trained attendant, with both participant and attendant able to reach the therapists (Mithoefer, 2017). From 2018, a MAPP1 sub-study allowed participants judged stable to go home if a support person stayed with them overnight, and neither phase 3 trial found that staying overnight changed outcomes (FDA, 2024; Mitchell et al., 2021, 2023). The FDA's 2024 proposal would require monitoring for at least eight hours and discharge only to an accompanying adult (FDA, 2024). An attendant who is not one of the session therapists adds an independent observer while the participant remains open and suggestible. The Vancouver relationship developed outside the sessions, so protection also has to extend past discharge, through integration sessions, check-in calls and clear rules about contact after the trial.
7.6 Reporting and Accountability
In the Vancouver case, roughly three years passed between the conduct and the complaint that reached the sponsor, and the team member who reportedly knew of it did not report it (MAPS, 2019). In its disclosure, MAPS stated that it had recently published a formal code of ethics, which it was building into its training program and which every study staff member was required to sign; it also announced plans for a patients' bill of rights, a formal grievance process, support resources for people making complaints and an internal review committee (MAPS, 2019). Health Canada requires sponsors to report serious unexpected adverse drug reactions throughout a trial (Health Canada, 2022), and the 2024 retractions show that disclosure to journals is also part of accountability.
A participant needs at least one route for raising a concern that does not pass through the people the concern is about. In a Canadian trial, those routes can include the sponsor, the research ethics board that approved the trial, Health Canada, the regulatory college of any licensed team member and, for criminal conduct, the police; support services such as VictimLinkBC can help a person decide what to do. The Vancouver case also shows the limits of after-the-fact routes: college complaints and a police investigation came years later and did not lead to charges. Protection therefore depends on safeguards that operate during the trial, together with routes that are explained at consent and repeated at follow-up.
8.1 Reviewing a Protocol
Research ethics boards review trial protocols before enrolment and can require changes. Case 9.1 asks you to act as a member of such a board.
A hypothetical research ethics board in British Columbia receives a protocol for a multi-site trial of MDMA-assisted therapy for PTSD that follows the MAPP1 design described in Section 1: three preparatory sessions, three eight-hour sessions with divided doses of MDMA or placebo, nine integration sessions, a two-person therapy team that may offer nurturing touch and focused bodywork, video recording of sessions, and an overnight stay after each session. The board and the site are hypothetical; the design and the standards come from the sources cited in this lesson.
Interactive 9.2 presents the board's five decisions. For each, choose the option you would require and read the feedback.
The strongest options in Interactive 9.2 share a feature: each places a decision or a check where it does not depend on the goodwill of the therapists in the room. That principle runs through Health Canada's expectations and the course's practice standard.
A hypothetical research team in British Columbia has been invited to join a future multi-site trial of MDMA-assisted therapy for PTSD under Health Canada's 2022 expectations. Draft a one-page plan for the team that covers both questions this lesson has raised. For participant protection, state how consent to touch will be obtained and recorded, the purpose and review schedule for any video recording, the composition and supervision of the therapy team, the arrangements after each session, and at least two routes by which a participant can raise a concern outside the team. For the credibility of the evidence, state how the trial will record all drug effects (including positive ones), measure expectancy and blinding, and follow participants for 12 months, in light of the FDA's concerns in Section 3.
Summary
MDMA produces hours of impairment and suggestibility in people who often have histories of trauma, so participant protection cannot rely on participants defending their own interests during a session. At the Vancouver site of a MAPS phase 2 trial, the safeguards in place (two providers, recording, monitoring and supervision) did not prevent departures from the manual during sessions or a sexual relationship that began afterwards, and the misconduct came to light only through the participant's own complaint years later. Since then, the sponsor has adopted a code of ethics, Health Canada has set expectations on touch, recording, licensure and medical oversight, and the FDA has proposed controls on monitoring (2024) and set out expectations for session monitors in trials (2026). Each safeguard has limits, so protection depends on layers that do not all rest on the same people.
Key Takeaways
- 7.1 MDMA produces hours of impairment and heightened suggestibility, and trial participants often had histories of developmental trauma, spent long sessions in closed rooms where touch was permitted and might hesitate to complain. Safeguards must therefore work even when a participant cannot protect their own interests in the moment, and harm occurs when gaps in several layers of protection line up.
- 7.2 At MP4, a Vancouver phase 2 site, both co-therapists substantially deviated from the manual during sessions, and one later began a sexual relationship with the participant while she was still enrolled in follow-up. The misconduct reached MAPS only through her complaints in 2018, MAPS disclosed it in May 2019, and the session recordings were not viewed until November 2021. Two providers, recording, monitoring and supervision were all in place, and none was sufficient.
- 7.3 The manual, Health Canada and the course standard place the decision about touch before the drug is given, with a stop word and touch optional except to prevent harm; Health Canada adds a second check at the time and requires any contact to be non-intimate and non-sexual. Consent agreed in advance cannot anticipate every situation, and consent rules govern only conduct that someone is in a position to observe.
- 7.4 A recording deters and detects misconduct only if someone independent watches it within a reasonable time, which did not happen at the Vancouver site. Health Canada treats recording as optional but requires a stated purpose, review frequency and retention, and recordings of people in vulnerable states need tight control of storage and access.
- 7.5 A second therapist protects a participant only if that person is willing and able to intervene and report, and licensure brings each therapist under a regulator that can investigate. Supervision depends on what therapists bring to it, so it works best with independent review of recordings, and an independent overnight attendant or discharge to an accompanying adult extends protection past the session, although rules on contact after the trial are also needed.
- 7.6 A participant needs at least one route for raising a concern that does not pass through the people the concern is about, such as the sponsor, the research ethics board, Health Canada, a regulatory college or the police. After-the-fact routes came years later in the Vancouver case, so routes must be explained at consent and repeated at follow-up, alongside safeguards that operate during the trial.
- 8.1 A research ethics board reviewing a protocol can require consent to touch obtained while sober and checked again, timely review of recordings by someone outside the team, licensed and supervised two-person teams, monitored discharge, and zero tolerance for sexual or romantic contact with reporting routes outside the team. The strongest options place each decision or check where it does not depend on the goodwill of the therapists in the room.
Reflection
MAPS stated that, at its Vancouver phase 2 site in 2015, four safeguards were in place: two providers in every session, video recording of sessions, study monitoring and clinical supervision, and that none had been sufficient. CBC News later reported that the session recordings were not viewed by MAPS staff until November 2021, and MAPS provisionally found that both co-therapists had substantially deviated from the treatment manual. Choose two of the four safeguards. For each, explain why it did not protect the participant in this case, and describe one specific change, drawn from Health Canada's 2022 expectations for psychedelic trials or from this course's practice standard, that would make it more effective.
Two providers in every session failed because the second provider was also involved in the departures from the manual, and, according to MAPS, later learned of the sexual relationship and did not report it. A second person protects a participant only if that person will intervene and report. Health Canada's expectation that unlicensed team members work under the direct supervision of a licensed therapist helps, and I would add a written duty on every team member to report concerns to someone outside the team, such as the sponsor or the research ethics board, with the participant also told of that route at consent.
Video recording failed because no one watched the recordings for about six years, so they neither deterred nor detected anything. Health Canada now expects a protocol that records sessions to state how often recordings will be reviewed. I would require a reviewer outside the therapy team to watch a sample of each team's sessions within a few weeks, with a clear process for acting on what they see. Neither change would by itself have prevented a relationship that began outside sessions, which is why rules on contact during follow-up also matter.
Minimum 20 characters required.
Question 1: How did the misconduct at the Vancouver site first reach MAPS?
Question 2: Under Health Canada's 2022 expectations for trials of psychedelic-assisted psychotherapy, when should consent for touch be obtained?
Question 3: What weakness of video recording as a safeguard did the Vancouver case reveal?
Question 4: Which arrangement after an experimental session fits both the treatment manual and the FDA's 2024 risk-control proposal?
Final Assessment
Bringing It All Together
MDMA-assisted therapy for PTSD, as tested in the MAPP1 and MAPP2 trials, is a structured course of about 18 weeks in which MDMA is given on three days within a manualized, inner-directed psychotherapy delivered by a two-person team. Both trials found that the drug added about nine to twelve CAPS-5 points of improvement to the same therapy with placebo, with more participants losing the diagnosis or reaching remission, and the placebo-with-therapy arm also improved substantially. Safety data from screened, closely monitored adults showed transient cardiovascular effects and no excess of suicidality, but omitted positive drug effects and liver tests.
The FDA's 2024 review accepted that the trials were positive and found the results difficult to interpret because nearly every participant who received MDMA knew it, expectancy was unmeasured, durability rested on an observational follow-up, and the therapy's contribution was not characterized. Its advisory committee voted against effectiveness and benefit-risk, and a complete response letter followed. The sponsor and investigators answered each concern, and some disagreements turn on definitions and design. Australia, meanwhile, has allowed authorized psychiatrists to prescribe MDMA since 1 July 2023, and in August 2026 the renamed sponsor was reported to have resubmitted without a new phase 3 trial.
The misconduct at the Vancouver site of a MAPS phase 2 trial, disclosed in 2019, showed that safeguards which look sufficient on paper can all fail together. It led to retractions, a code of ethics and clearer expectations from Health Canada and the FDA on touch, recording, licensure and monitoring, and it places participant protection beside evidence quality as a test of whether psychedelic-assisted therapy is ready for wider use.
Key Takeaways from this lesson
- MDMA was proposed as an aid to psychotherapy for PTSD on conflicting animal findings, a few imaging studies and clinical observation, and MDMA-assisted therapy in the phase 3 trials combined three preparatory sessions, three eight-hour experimental sessions with divided doses and nine integration sessions, delivered by the same two-person team under an inner-directed manual.
- The CAPS-5 total severity score sums 20 symptom ratings from 0 to 80, while the diagnosis depends on how many symptoms in each cluster reach a rating of 2, so the score and the diagnosis can change differently.
- MAPP1 and MAPP2 found between-group effect sizes of d = 0.91 and d = 0.7, about nine to twelve CAPS-5 points, in favour of MDMA over the same therapy with placebo. The placebo-with-therapy arm improved by about 14 points, so the drug's contribution is estimated by the between-group difference, which is much smaller than the full change in the MDMA arm. Numbers needed to treat of about three to six depend on the outcome chosen and on whether participants who left the trial are counted, because dropout was much higher in the placebo arms, and safety data showed transient cardiovascular effects and no excess of suicidality in screened participants, with positive drug effects and liver tests not recorded.
- Follow-up data suggest durable benefit, but losses to follow-up, variable timing, unblinding and intervening treatments limit what they show. ICER's 2024 review found no head-to-head trials of MDMA-assisted therapy against trauma-focused psychotherapies, and differences in comparators make indirect comparisons misleading.
- The FDA declined approval on 9 August 2024 after its advisory committee voted 2 to 9 on effectiveness and 1 to 10 on benefit-risk, citing functional unblinding, expectancy, durability, the therapy's role and incomplete safety data, each of which the sponsor and investigators answered.
- Three phase 2 papers were retracted in August 2024 for undisclosed unethical conduct at the Vancouver site and incomplete competing-interest declarations, and as of October 2026 the phase 3 reports had not been retracted. As of October 2026 the renamed sponsor was reported to have resubmitted without a new phase 3 trial, while Australia has allowed authorized psychiatrists to prescribe MDMA since 1 July 2023 under a rescheduling that is not a product approval.
- Hours of impairment and suggestibility in people with trauma histories raise the stakes for participant protection. At the Vancouver site, disclosed in 2019, two providers, recording, monitoring and supervision all failed together, and the case led to a code of ethics and clearer expectations from Health Canada and the FDA.
- Protecting participants in altered states requires consent to touch obtained while sober, independent and timely review of recordings, licensed and supervised teams, monitored discharge and reporting routes that do not depend on the therapists involved.
Core Concepts Reviewed
Section 1: PTSD, the MAPP treatment protocol, co-therapist teams, the inner-directed approach and inner healing intelligence, rules on touch, the CAPS-5 and its SEV2 rule, and the definitions of response, loss of diagnosis and remission.
Section 2: the design and participants of MAPP1 and MAPP2, least squares mean differences and standardized effect sizes, the placebo-with-therapy response, absolute differences and numbers needed to treat, safety and suicidality findings, durability, and comparison with trauma-focused psychotherapies.
Section 3: the regulatory path, the advisory committee, functional unblinding and expectancy, adverse event recording, the role of psychotherapy, risk evaluation and mitigation strategies, the complete response letter, the 2024 retractions, the 2026 status and Australia's 2023 authorization.
Section 4: vulnerability in altered states, the 2019 disclosure concerning the Vancouver site, consent to touch, video review, team composition and supervision, overnight monitoring and discharge, reporting routes and protocol review by a research ethics board.
The final reflection asks you to bring the evidence, the regulatory debate and the lessons on participant protection together in a policy recommendation.
Reflection
In a hypothetical request, a provincial health ministry asks whether British Columbia should allow MDMA-assisted therapy for PTSD through a regulated route similar to Australia's, where, since 1 July 2023, psychiatrists who hold ethics committee approval and authorization from the national regulator may prescribe MDMA for PTSD although, as of October 2026, no MDMA product had been evaluated or registered. Drawing on all four sections of this lesson (the MAPP treatment model and CAPS-5 outcomes; the MAPP1 and MAPP2 results, durability and comparison with trauma-focused psychotherapy; the FDA's 2024 decision with the criticisms and the sponsor's and investigators' responses; and the participant-protection lessons of the Vancouver site), write a brief recommendation. State what the evidence does and does not show, and set out the conditions you would attach to any access, including at least one condition about evidence collection and at least one about participant protection.
Two randomized trials found that adding MDMA to an intensive, manualized psychotherapy reduced PTSD symptoms by about nine to twelve CAPS-5 points more than the same therapy with placebo, with numbers needed to treat of about three to six for response, loss of diagnosis or remission, and with transient cardiovascular effects and no excess of suicidality in carefully screened adults. The evidence does not show how much of that difference survives functional unblinding and expectancy, how long benefit lasts under blinded follow-up, what the therapy contributes, or how MDMA-assisted therapy compares with trauma-focused psychotherapies, and its safety data omit positive drug effects and liver tests. The FDA declined approval in 2024 for these reasons, while the sponsor and investigators argue the results are valid despite partial unblinding.
A limited route could be defensible for people whose PTSD has not responded to established treatments, if it were tied to evidence and protection, although neither MAPP trial required earlier treatment to have failed, so this group was not tested as such. I would require prescribing by authorized psychiatrists with ethics oversight, a registry recording all effects (positive and negative), blinding-independent outcomes and 12-month follow-up, and public reporting. For protection, I would require two licensed or supervised therapists, consent to touch obtained while sober with a stop word, recordings reviewed on a schedule by someone outside the team, discharge only to an accompanying adult after at least eight hours, zero tolerance for sexual or romantic contact, and a complaint route outside the clinic.
Minimum 30 characters required.
Final Knowledge Assessment
Question 1: Which feature of the MAPP trials did the FDA identify as making their efficacy results most difficult to interpret?
Question 2: Which description matches one complete course of treatment in the MAPP phase 3 protocol?
Question 3: Why does the inner-directed treatment model complicate the regulatory evaluation of MDMA itself?
Question 4: A hypothetical participant's CAPS-5 total falls from 40 at baseline to 28 at week 18, and every cluster still has enough symptoms rated 2 or more. How is the participant classified?
Question 5: Which statement about Australia's decision that took effect on 1 July 2023 is accurate?
Question 6: What did the MPLONG follow-up data suggest, and with what main limitation?
Question 7: In MAPP2, the remission NNT among participants with a final assessment was about 4.0. If all randomized participants are counted and those without a final assessment are treated as not in remission, how would the NNT change?
Question 8: According to MAPS and CBC News, which safeguard existed at the Vancouver site but did not work as intended?
Question 9: How did the sponsor respond to the August 2024 retractions?
Question 10: Which pair of trial features most limits the generalization of the MAPP results to people with PTSD in general care?
Question 11: What did ICER conclude in its final evidence report of June 2024?
Question 12: Which statement about therapists matches Health Canada's 2022 expectations for trials of psychedelic-assisted psychotherapy?
Question 13: Which statement describes the status of MDMA-assisted therapy for PTSD in the United States as of October 2026?
Question 14: Why does this course require that agreement about touch be reached while the participant is sober?
Question 15: Which design described in the FDA's 2026 guidance most directly addresses the question of what the therapy contributes?
Glossary: Key Terms, People & Frameworks
📚 Reference page, available throughout the lesson
Definitions of the terms, frameworks and people introduced in this lesson, in the order of the sections where they first appear.