Development of interventions for the secondary prevention of Alzheimer's dementia: the European Prevention of Alzheimer's Dementia (EPAD) project

Summary

Alzheimer's dementia affects more than 40 million people worldwide with substantial increases in prevalence anticipated. Interventions that either modify risk or reduce the development of early disease could delay the onset of dementia or reduce the rate of cognitive and functional decline. The European Prevention of Alzheimer's Dementia (EPAD) is a public–private consortium, funded by the Innovative Medicines Initiative, designed to increase the likelihood of successful development of new treatments for the secondary prevention of Alzheimer's dementia. EPAD will help with testing of different agents in this pre-dementia population through four components: improvement of access to existing cohorts and registries, development of the EPAD Registry of approximately 24 000 people who might be at increased risk of developing Alzheimer's dementia, establishment of the EPAD Longitudinal Cohort Study of 6000 people at any one time, and establishment of an adaptive, proof-of-concept trial including 1500 participants at any given time. The need for EPAD and its key design elements are described, and we discuss EPAD in relation to similar projects in progress. These parallel efforts reflect the need for a coordinated, worldwide battle against dementia, in which EPAD will play a crucial role.

Introduction

When memantine was launched in 2003, researchers could not have foreseen that no new drugs would be approved for either the secondary prevention or symptomatic treatment of Alzheimer's dementia in the next 13 years.¹ This failure has occurred despite major advances in the understanding of the biology of Alzheimer's disease and substantial investments by the pharmaceutical industry to develop drugs for both old and new targets.

In the past 10 years, the extent of the problem of dementia on a human, societal, and economic level has been expressed with ever-greater vigour and clarity. A 2012 WHO report estimated that 36 million people were living with dementia and this number would rise to 115 million by 2050.² The total cost of disease will rise from €515 billion per year to €1·6 trillion per year by 2050. Age is the principal, unmodifiable risk factor for dementia with incidence doubling every 5 years from 1% incidence at 65 years. If life expectancy is constant, delaying onset of dementia by 5 years overall could theoretically reduce the number of people with dementia by 50% (figure 1).³

Although primary prevention could have striking, long-term effects by reducing the prevalence of modifiable risks at population level, the difficulty of effecting behavioural change, the unpredictable consequences of primary prevention, and untested hypotheses related to population level risk factors mean primary prevention will be insufficient to prevent the worldwide burden of dementia. Modification of the pathological process after the onset of disease, but before the onset of symptoms—ie, secondary prevention—is of huge importance. Initiation of disease processes leading to symptomatic and functional decline occurs decades before the onset of dementia. Although secondary prevention is a tractable solution, accurate identification of the presence of disease processes in an individual with few or no symptoms, the ability to calculate the probability of an individual of developing dementia, and availability of effective interventions are needed.

These preconditions should be tested in proof of concept (PoC) studies and are now the focus of major, coordinated, worldwide research efforts. The European Prevention of Alzheimer's Dementia (EPAD) Consortium is a partnership between public and private organisations funded by the Innovations Medicine Initiative (IMI). Close and integrated efforts between clinicians, the public, and academics with expertise, and resources from the commercial sector are needed for the project. The emphasis on collaboration eliminates the usual conflicts that might arise between individual academics and the pharmaceutical industry through a comprehensive and legally binding partnership agreement.

The EPAD project builds on promising findings from many trials of secondary prevention in people with prodromal dementia, preclinical disease, or in other high-risk populations using both behavioural and pharmacological approaches. Definitions of pre-dementia states are emerging, consisting of distinct categories from preclinical (evidence of Alzheimer's disease pathology with no symptoms), prodromal Alzheimer's disease (also referred to as mild cognitive impairment due to Alzheimer's disease), and dementia. However, this nomothetic approach to an exceptionally complex disease from a phenotypic and biological perspective is suboptimal from a modelling perspective. A definition of individuals at high probability of developing dementia needs the application of empirically derived tests to ensure that patients exposed to interventions are most likely to benefit and, therefore, tolerate any potential harmful effects of the interventions. Stratification in the Alzheimer's Prevention Initiative (API) and Dominantly Inherited Alzheimer's Network Trial (DIAN-TU) work is based on genetic factors, apolipoprotein E risk variants, and autosomal dominant mutations. However, we propose in EPAD that prediction models that incorporate three factors are most likely to achieve accurate individualised trajectories of decline to create a spectrum of probability as opposed to stratified probability. The three factors that should be taken into account in prediction models are risk factors (genetic and other unmodifiable risks—eg, age and modifiable risks), disease marker expression (biomarker measurement), and patient symptoms (including the cognitive profile of an individual). The principal benefit of this individualised, multifactorial approach is that each individual can be placed on a scale of probability of decline in the proposed spectral model; whereas in a stratified model, people within each strata share a similar probability of decline. This three-factor model of risk is the focus of planned disease modelling efforts of the EPAD project with collection of data to generate these models from the extensive EPAD-Longitudinal Cohort Study (LCS). Such approaches are in progress in EPAD and in smaller, secondary prevention, mid-life research efforts such as PREVENT⁴ and ALFA.⁵

The EPAD-LCS is a pivotal element of EPAD and has four aims: to be a readiness cohort for the EPAD-PoC study, to provide run-in data for the pre-randomisation period in the EPAD-PoC study, to gather longitudinal data for Alzheimer's disease modelling of probability of decline, and to generate probability models and place individuals on the probability spectrum. Populations from this spectrum can then be defined by their response to intervention in the EPAD-PoC study (figure 2). Selection for the EPAD-PoC study will be based on upper and lower limits for risk as defined in the numerous appendices (one per intervention in the trial) to the EPAD-PoC master protocol. Thresholds for cognitive test scores, biomarker values (eg, hippocampal atrophy or amyloid burden as measured by positron emission tomography [PET] scanning), or genetic risk could be used as criteria for inclusion or exclusion. Finally, identification of the subpopulation most likely to respond to the intervention allows for the selection of the ideal candidate subpopulation to recruit for subsequent phase 3 confirmatory trials. Through the EPAD project, we will provide the empirical evidence needed for precision medicine and to improve disease models for pre-dementia Alzheimer's disease in the EPAD-LCS (complemented by EPAD-PoC data). We will assemble a broad population of people in the EPAD-LCS who are well characterised on a spectrum of probability and therefore act as a readiness cohort for the EPAD-PoC study, and place individuals from this population into the EPAD-PoC study-specific strata.

In addition to focusing on the early stages of the Alzheimer's disease spectrum, EPAD will aim to improve the likelihood of successful drug development through other design elements of the EPAD-PoC study. Success in phase 3 drug development depends on the ability to select the appropriate population, endpoints, doses, and sample sizes to confirm clinical efficacy of treatments. Traditionally, data from phase 2 trials inform these choices. Much drug development in Alzheimer's disease has involved the initiation of large phase 3 programmes based on biomarker findings before data for clinical effects were available. The EPAD-PoC study will first investigate drugs based on biomarker effects, but success in the EPAD-PoC study will be defined as achieving a clear, demonstrable clinical benefit. Drugs deemed successful in the EPAD-PoC study will, therefore, be more likely to achieve clinical and regulatory success in phase 3 (figure 3).

To develop the drug candidates in EPAD most rapidly and efficiently, the EPAD-PoC study will employ a Bayesian adaptive design that learns from data accrued as the trial progresses. Frequent interim analyses, done in accordance to predefined algorithms and blinded to all trial personnel, allow adaptive randomisation of individuals to interventions that appear to show the greatest clinical efficacy, and potentially in subpopulations defined by clinical status, biomarkers, or genetics. These interim analyses are used to test for early signals of drug success or futility; the results of the interim analyses are only revealed to trial personnel if such a signal is detected.

Planned discontinuation of additional enrolment, once such a signal is detected, minimises patient exposure to treatments that are unlikely to be useful and speeds up progress to phase 3 testing and, ultimately, to successful registration and clinical use. Using Bayesian adaptive design statistics, which allows for the use of all available data and the updating of the longitudinal model at each interim analysis, and using a shared placebo group, which overall reduces the number of required subjects in the trial, can increase the efficiency of the EPAD-PoC study.