From Analysis to Action: Implementing lessons from FDA AdComm's MDMA review
Despite the widely recognised need for PTSD treatments and inspiring patient stories supporting MDMA-assisted therapy, an overwhelming majority of the FDA’s Advisory Committee found the efficacy data unsatisfactory and concluded that the benefits didn't outweigh the risks due to methodological and misconduct issues. Although the panel's rejection doesn't guarantee the FDA will disapprove this treatment, it clearly highlights what regulators will scrutinise when evaluating psychedelic therapies. For Lykos, the decision was disappointing, but for others, it has provided invaluable insights into what will be needed to bring psychedelic treatments to market. This article distils the key actionable lessons in designing and delivering psychedelic trials to meet regulatory standards.
Methodological Issues
The silver lining (at least, for the designers of future trials) is that most issues were methodological and can be rectified with appropriately conducted trials. One point of unanimous agreement among all panellists was the trial's lack of adequate control, leading to uncertainties about efficacy. Many reiterated the plethora of data problems, emphasising that while each issue in isolation might be tolerable, their cumulative impact raised significant concerns. Let’s break down these issues.
Blinding
Since functional unblinding of patients was typically anticipated and acknowledged as a limitation, panellists were particularly concerned with expectancy rather than blinding integrity. Although large effect sizes in studies often alleviate regulatory concerns about bias, the ‘hype’ around psychedelics made regulators more cautious. Consequently, the highlighted issues revolved around the unknown role of expectation bias and the unblinding of therapists, in addition to patients. As solutions, drug developers are advised to:
Assess the extent of participant blinding.
Utilise blinded central raters.
Evaluate the impact of expectation and demonstrate that differences between study arms aren't solely due to expectation bias.
Conduct robust long-term follow-up to assess the drug vs. placebo response, as expectancy effects are expected to diminish over time.
It's worth noting that while some panellists advocated for blinding therapists—suggesting different therapists for dosing and integration sessions—this approach may negatively affect the patient-therapist rapport, a factor that has been shown to correlate with patient outcomes. However, data are currently insufficient to substantiate this claim.
Psychological intervention
While the FDA does not regulate psychotherapy, regulators expressed significant concerns about the design and implementation of the psychological intervention. This was primarily due to the challenge of distinguishing the effects of psychotherapy from those of the drug. Quantifying the contribution of psychotherapy to overall efficacy was deemed essential. However, achieving this required standardising its delivery—a daunting task given the inherent variability of talking therapy and differences among therapists, made more challenging in this case due to the flexibility in approach permitted by the MAPS/Lykos therapy manual.
It was also noted that given the uncertainties in the data, if MDMA was approved, the FDA couldn’t label it for use on its own. Labelling regulations allow specification that a drug should be used only with another mode of therapy. However, the agency needed evidence that psychotherapy was necessary to achieve the desired effect and the adjunctive treatment needed to be standard of care.
To disentangle the effects of the drug and psychological intervention:
Use 2x2 factorial design and evaluate the effect of the drug alone.
Measure the integrity and quality of psychotherapy between different arms of the study.
Use an evidence-based, standard of care therapy model.
Have robust therapist recruitment, training and monitoring processes.
Measure therapist fidelity to therapy model. (Digital tools are now available to facilitate this cost-effectively, especially in large scale trials, such as the one provided by our collaborator, mpathic.)
If using a psychotherapy model that’s not evidence-based:
Clearly describe the psychological model, the content and the approach of therapy sessions in the manual.
Ensure there is limited flexibility in the manualised psychotherapy to reduce variability.
Compare the psychological intervention used to other psychotherapeutic approaches.
Participant recruitment
Concerns were raised about the applicability of the trial results to the general population due to selection bias and lack of diversity. To address these issues, it is crucial to:
Monitor recruitment for any selection bias due to past drug use, lifestyle modifications needed for the trial, and recruitment methods.
Make efforts to align the percentage of participants with prior drug use with population-level figures.
Ensure the inclusion of racial and ethnic minorities. (London is one of the best places for patient recruitment due to its concentrated and diverse population, that’s why we set up our research facility in Central London.)
Durability of effect
Long-term follow-up studies in psychiatry are particularly challenging due to factors such as patient retention, changes in treatment plans, and inconsistent data collection. Lykos's follow-up study encountered numerous issues in these areas, making it difficult to isolate the drug's effect from many confounding variables. To address these issues, consider below:
Pre-specify study design with well-defined controls
Have multiple follow-up visits with consistent frequency
Provide appropriate incentives and support systems to reduce dropout and non-compliance
Monitor the use of intercurrent interventions (e.g. treatments that are not part of the trial) by participants, including any off-label treatments
2. Missing Data
The Advisory Committee pointed out holes in the data provided by Lykos. Some key data points the committee noted as absent:
Acute effect size: Measure whether the intensity of acute subjective experience during drug administration correlates to outcomes. (We recommend using both subjective measures such as patient and clinician-reported outcomes and objective measures such as biomarker data where possible)
Adverse effects: Have a clear systematic pathway for reporting treatment-emergent, adverse effects and harms, including ‘positive’ ones such as euphoria to understand abuse potential.
Safety: Ensure all potentially relevant safety data points are collected, such as those on hepatotoxicity, hyponatremia, drug-drug interactions, and cardiovascular effects.
Other: The panel was also interested in understanding what happened to adverse events that didn’t resolve within 7 days, and what was evaluated when patients were safe to leave the research sites following dosing.
3. Misconduct
During the nine-hour session, misconduct emerged as the most pressing topic for discussion and fell into two categories.
Therapists
The alarming occurrence of patient boundary violations in Lykos’s Phase 2 trial raised legitimate concerns among both panellists and the public. There is rightful apprehension about the potential recurrence of such incidents on a larger scale once this treatment is rolled out more widely. Given that these violations occurred within the controlled environment of a clinical trial, the gravity of the situation is even more profound. So, what measures does the FDA AdComm panel suggest to safeguard patients?
Have two therapists in the room.
Be mindful of the relationship between the two and the power dynamics that can affect reporting and accountability.
Provide strict training to therapists, and use stringent, objective therapist selection criteria.
Have an independent organisation to monitor therapists.
When there are numerous patients struggling to access one therapist, mandating two licensed therapists in the room appears impractical. Not only is this unrealistic in real-world scenarios post-approval, but it also poses challenges for cost-effectiveness evaluations. If the requirement for two therapists aims at enhancing safety, we believe this can still be achieved with one therapist, supported by appropriate monitoring structures such as digital tools enabling real-time monitoring and reducing the need for manual processing of session videos—a costly and time-consuming process.
Research
The committee showed keen interest in discussing misconduct related to research practices such as lack of equipoise and researcher bias. There were allegations of investigators discouraging participation in the long-term follow-up study and therapists discouraging reports of adverse events. Given regulators' apprehension about the psychedelic movement, such misconduct carries the risk of major reputational damage to the sector. Drug developers, CROs and research facilities must be particularly vigilant in preventing such occurrences and ensuring the highest quality data and reporting standards.
4. Additional Considerations
Patient perspective
Considering the many clinical, regulatory, and ethical challenges these trials pose, patient experience will play a crucial role in informing regulatory decisions. Patients who felt unsupported or veterans who felt mistreated negatively impacted the committee's opinion of Lykos’s application. Given that these trials target mental health patients, it is essential that patient experience remains central, not only from an ethical standpoint but also from a regulatory perspective.
Health economics
Stakeholders, including the American Psychological Association and ICER, have emphasized the need for head-to-head trials comparing psychedelic treatments with existing standard of care treatments. Surprisingly, no commercial clinical trial has yet addressed this. Conducting such comparisons will help regulators better assess the risk-benefit ratio of psychedelics and aid health payers in determining whether it is worth funding these treatments, which has significant implications for their accessibility.
To advance the discussion on the cost-effective delivery of psychedelic treatments within current healthcare systems (E.g. our paper evaluating psilocybin-assisted therapy for depression), developers should also test their drugs as adjuncts to established therapy models like CBT and exposure therapy. This approach will not only provide regulators with data on real-world applications but also produce valuable comparative data to demonstrate efficacy and address the clinical and operational concerns that payors will be seeking to manage.
Concluding remarks
The FDA Advisory Committee's review underscores the critical need for rigorously designed and consistently delivered trials, which is why we established Clerkenwell Health to lead in the design and delivery of psychedelic clinical development. While the committee's response may have initially seemed discouraging, it is promising that the concerns raised were about trial quality rather than the efficacy of the drug itself, providing a roadmap to the improvement of further clinical research in this space. This significant interaction between psychedelics and regulators has provided invaluable lessons. We are here to support drug developers, from trial design to therapist training, in implementing these lessons and to ensure novel psychiatric treatments receive the best possible support in clinical development.
