The EPAD PoC platform was developed as an open-ended resource for adaptive, multi-arm drug development proof-of-concept studies for early and accurate decisions on the ongoing development of drug candidates, drug combinations, or other interventions for the secondary prevention of Alzheimer’s dementia.
The platform was designed with a world-leading consortium of 39 partners to efficiently deliver early, accurate results. EPAD planned to study new drugs in a well-designed Phase II PoC trial with clinical endpoints prior to Phase III, utilising the power of adaptive design and Bayesian statistics.
The platform was supported by three key strategies: the EPAD register, the cohort and trial.
Diagnosis: The Longitudinal Cohort Study (LCS) included subjects with evidence for Alzheimer’s disease pathology.
Severity: The study population spaned the full continuum from no symptoms (i.e., preclinical Alzheimer’s disease) through the late stages of prodromal Alzheimer’s dementia or MCI due to Alzheimer’s dementia, but included nobody who already met criteria for a diagnosis of dementia.
Biological fingerprint: The requirement for Alzheimer’s disease pathology entails the use of CSF. The LCS also included some subjects who had some evidence for Alzheimer’s disease pathology but who did not meet standardised cut-offs for amyloid positivity (i.e. who are in a “grey zone” pathologically) and some control subjects who were amyloid negative.
Genetics: The LCS population comprised sporadic and not dominantly inherited Alzheimer’s disease. The population selection process involved some enrichment for ApoE4+ genotype, but the full population included subjects who were both ApoE4+ and -.
Cognitive: A cognitive battery that is appropriate for the full spectrum of cognitive status in the LCS population and that is sensitive to change over time was developed and wwas used to follow LCS subjects at regular intervals. The availability of such longitudinal run-in data for subjects who are eligible for the PoC trial was a key advantage of EPAD, as it provided individual data on the rate of decline that can served as a baseline set of assessments and increase the study’s power to detect change due to treatment.
Behavioural: Assessments of depression, anxiety, and sleep were proposed.
Functional: A functional assessment was included. At this time, the Amsterdam IADL Scale was being considered, and actigraphy was included.
Imaging: MRI, including fMRI, was included and other types added.
Tissue: See above.
A full list of all outcome measures is available at the ClinicalTrials.gov EPAD LCS study record.
The EPAD project included more than 40 Trial Delivery Centres (TDC) that conducted the LCS. These centres were highly experienced and underwent further training for standardisation as part of EPAD.
Research participants were followed up every 6 months during the first year and annually thereafter. The frequency of follow-up in the PoC was be determined during the protocol development process according to the intervention owners requirements.
All LCS data are available in accordance with the access rights provided for in the EPAD Project Agreement.
The study design
Key features of the PoC study design were:
- Use of a cognitive measure as the primary endpoint.
- Frequent interim analyses for success and futility.
- Advancement of interventions that achieve an effect on an intermediate phenotype or biomarker of target engagement to the clinical stage of the study (note: interventions that have already demonstrated target engagement may bypass this step).
- Bayesian statistical models used to adapt on the cognitive measure to achieve faster randomisation to doses/drugs that appear more effective overall or in specific subpopulations.
- Efficient use of the accruing data, e.g., by utilizing all assessments within a longitudinal model.
- Randomisation to drug or placebo for each intervention, with analysis using combined data from all placebo subjects to increase the power of the analyses.
A standard set of efficacy and safety measures were assessed in all subjects.
The key adaptive features were adaptive randomisation of subjects and the ability to discontinue a drug treatment for futility or evidence for early success in the PoC, i.e. efficacy data indicating readiness for Phase III. Adaptive randomisation could be used to preferentially assign subjects to doses of a drug that appeared more efficacious at an interim analysis. It was also possible that subpopulations of subjects, defined by clinical, biomarker, or genetic criteria, could be preferentially randomised to a specific drug based on evidence for greater efficacy of that drug in that subpopulation at an interim analysis. Other adaptations could be considered if warranted by the needs of a drug or drug owner.
Essentially, the addition of each new drug to the PoC entailed an appendix in which the specific treatment parameters and assessments were described. Such appendices or amendments were developed jointly by the EPAD Clinical Development Executive Committee – the group responsible for overall study design – and the intervention owner.
The primary analyses utilised Bayesian statistics. This approach was more powerful for determining the probability of success of a given drug at multiple interim analyses and therefore led to faster decision-making.
The Clinical Candidate Selection Committee (CCSC) evaluated all nominated compounds or other interventions. To be eligible, interventions had scientific support for potential efficacy in the prevention of Alzheimer’s dementia or delay in progression of symptoms, and adequate preclinical and clinical safety to support advancement to Phase II.
Novel small molecules, repurposed molecules, biologics, vaccines and combinations of these were considered.
All interventions selected for potential use in the PoC were evaluated on the following criteria:
- Scientific rationale
- Pharmacology requirements
- Plans for full development/regulatory issues
- Develop-ability/CMC Quality guidelines
Any intervention that was be considered ready for PoC in development for Alzheimer’s dementia had already the data required to be eligible for inclusion.
We worked closely with the intervention owner, in advance of study initiation, to ensure that the timing of inclusion of the drug worked. We ensured that interventions were added to the study on the expected timeline.
A single academic entity (University of Edinburgh) was the sponsor of the EPAD PoC trial and took full responsibility for ensuring compliance across the project.
Members of the CCSC were required to sign non-disclosure agreements before reviewing nominations. The committee included novody with potential conflicts of interest.
Testing combinations was a specific goal of EPAD and it was expected that combinations of therapies were proposed for inclusion. The same criteria was used for selecting combinations as for selecting single therapies.
The EPAD Project Agreement included confidentiality provisions that were also applicable to employees of EPAD members. However, if requested by the intervention owner, all CCSC members signed confidentiality agreements before reviewing confidential information. Confidential information were submitted and maintained using secure processes with limited access. Submission of a full nomination only occurred after informal discussion with the CCSC to ensure that an intervention had all data sufficient for evaluation.