Alzheimer’s Prevention Initiative: A Plan to Accelerate the Evaluation of Presymptomatic Treatments

Abstract

There is an urgent need to find effective presymptomatic Alzheimer’s disease (AD) treatments that reduce the risk of AD symptoms or prevent them completely. It currently takes too many healthy people, too much money and too many years to evaluate the range of promising presymptomatic treatments using clinical endpoints. We have used brain imaging and other measurements to track some of the earliest changes associated with the predisposition to AD. We have proposed the Alzheimer’s Prevention Initiative (API) to evaluate investigational amyloid-modifying treatments in healthy people who, based on their age and genetic background, are at the highest imminent risk of developing symptomatic AD using brain imaging, cerebrospinal fluid (CSF), and cognitive endpoints. In one trial, we propose to study AD-causing presenilin 1 [PS1] mutation carriers from the world’s largest early-onset AD kindred in Antioquia, Colombia, close to their estimated average age at clinical onset. In another trial, we propose to study apolipoprotein E (APOE)ε4 homozygotes (and possibly heterozygotes) close to their estimated average age at clinical onset. The API has several goals: 1) to evaluate investigational AD-modifying treatments sooner than otherwise possible; 2) to determine the extent to which the treatment’s brain imaging and other biomarker effects predict a clinical benefit—information needed to help qualify biomarker endpoints for use in pivotal prevention trials; 3) to provide a better test of the amyloid hypothesis than clinical trials in symptomatic patients, when these treatments may be too little too late to exert their most profound effect; 4) to establish AD prevention registries needed to support these and other presymptomatic AD trials; and 5) to give those individuals at highest imminent risk of AD symptoms access to the most promising investigational treatments in clinical trials.

Introduction

In this article, we note the urgent need to find effective presymptomatic Alzheimer’s disease (AD) treatments, which we define as an intervention intended to postpone the onset, reduce the risk of, or completely prevent AD symptoms. We suggest that the greatest roadblock to the development of effective presymptomatic treatments may be the scientific means and financial incentives needed to evaluate the range of promising treatments. We briefly summarize relevant findings from our longitudinal study of cognitively normal people at three levels of risk for late-onset AD, which led us to propose how brain imaging and other biomarkers could be used to rapidly evaluate presymptomatic treatments in proof-of-concept clinical trials. We note the need for humility when it comes to predicting how these biomarkers will respond to AD-slowing treatments in clinical trials and point out that regulatory agencies are unlikely to approve a presymptomatic treatment based solely on biomarker endpoints until evidence is provided to show that a presymptomatic treatment’s biomarker effects are reasonably likely to predict a clinical benefit. Finally, we describe our Alzheimer’s Prevention Initiative (API), in which we intend to evaluate investigational amyloid-modifying treatments in cognitively normal people who, based on their age and genetic background, are at the highest imminent risk for symptomatic AD. The API is intended to evaluate treatments sooner than otherwise possible, to provide a better test of the amyloid hypothesis, to provide the evidence needed to show that an AD-modifying treatment’s biomarker effects are reasonably likely to predict a clinical benefit, and give individuals at the highest imminent risk of symptomatic AD access to promising treatments in prevention trials.

“Presymptomatic (or Preclinical) AD Treatments:” A Proposed Definition

We have recently defined presymptomatic (or preclinical) AD treatments” as those interventions that are initiated before apparent cognitive decline and are intended to reduce the chance of developing AD-related symptoms^[1]. The proposed term refers to an intervention whether it is started before or after biological evidence of the underlying disease (which may be hard to define), and whether it postpones the onset, partially reduces the risk of, or completely prevents symptomatic AD. We have introduced this term based on the United States Food and Drug Administration’s (FDA’s) stated view that it would not approve a treatment for the “prevention” of AD unless trials were able to demonstrate that a treatment prevented the onset of symptoms for the rest of a person’s life, an impractically high hurdle to overcome. We believe that it will be easier to show that an intervention meets our proposed criterion as a presymptomatic or preclinical AD treatment and support regulatory agency approval. Our definition is also consistent with the research criteria for “preclinical AD treatment” recently proposed by a working group for the National Institute on Aging (NIA) and Alzheimer’s Association^[2].

Background

Alzheimer’s disease is an unacceptable problem due to the toll it takes on patients and family caregivers, and the current and project financial impact on society^[3,4]. A large number of healthy lifestyle interventions have been suggested but not yet proven to postpone the onset and reduce risk the risk of developing AD symptoms^{[5,6,7,8,9,10,11,12]}. An even modestly effective therapy could have a significant public health benefit. For instance, a treatment that postponed the onset of AD symptoms by only five years without increasing life-expectancy might be able to reduce the number of clinically affected patients by half^[4]. Meantime, a growing number of investigational disease-modifying treatments are in preclinical and clinical development^[13], including but not limited to a large number of medication and immunization therapies intended to interfere with the production and accumulation of certain amyloid-β (Aβ) species. If, as many but not all researchers believe, the amyloid hypothesis is correct^[14], if the treatment is targeting the Aβ species critically involved in the predisposition to symptomatic AD, if it is sufficiently safe and well tolerated, and if it is started sufficiently early, it might be possible to substantially reduce the risk of symptomatic AD and maybe even prevent it completely.

Unfortunately, it takes too many healthy people, too much money, and too many years—longer than a drug product’s patent life—to evaluate presymptomatic treatments using clinical endpoints. For instance, in order to determine whether cholesterol-lowering, blood pressure-lowering or hormonal treatments reduced the risk of symptomatic AD if they were started in middle-age, when epidemiological studies suggest that they may have their most profound effect, nearly 50,000 healthy middle-aged research participants would be needed for a two-year placebo-controlled randomized clinical trial (RCT), such that a sufficient number of people developed symptomatic AD to detect a significant treatment effect. While there have been a small number of large, time-consuming prevention trials in older people^{[15,16,17,18]}, a new paradigm is needed to evaluate the range of presymptomatic treatments.

We believe that brain imaging or other biomarker measurements of AD are needed to rapidly evaluate presymptomatic treatments without having to study thousands of healthy volunteers or wait many years to characterize and compare clinical endpoints in the investigational and placebo treatment groups. Brain imaging and other biomarker methods continue to be further developed and used to detect and track changes associated with the clinical progression of AD, and several of these methods have been used to detect and track similar changes in the presymptomatic stages of AD^{[19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34]}. Based on findings from the AD Neuroimaging Initiative (ADNI) and other longitudinal studies, researchers have established standardized procedures for the acquisition of brain imaging data and biological samples, to provide common data sets that have helped researchers further develop, test and compare their data analysis methods, and to provide sample size estimates for the use of biomarker endpoints and enrichment strategies for clinical trials, particularly in patients in the symptomatic stages of AD^[35]. To date, the best established biomarkers of AD are fluorodeoxyglucose positron emission tomography (FDG PET) measurements of decline in the regional cerebral metabolic rate for glucose (rCMRgl), volumetric magnetic resonance imaging (MRI) measurements of regional or whole brain shrinkage, PET measurements of fibrillar amyloid-β (Aβ) burden, and low cerebrospinal fluid (CSF) Aβ₄₂ levels, alone or in combination with high CSF total tau or phospho-tau levels^[24,1].

As noted below, we have suggested that biomarker endpoints could be used in proof-of-concept RCTs to rapidly evaluate presymptomatic treatments in cognitively normal people at increased risk for AD, and we have provided preliminary sample sizes estimates for some of these trials^[36,20]. Unfortunately industry partners are unlikely to provide financial support for these studies until the biomarker endpoints are qualified for use in pivotal trials. While the field needs the scientific means (i.e., biomarker endpoints) to rapidly evaluate presymptomatic treatments, it also needs the right financial incentive (i.e., regulatory agency qualification of the biomarker endpoints for use in the accelerated approval of presymptomatic AD treatments).

Regulatory agencies are unlikely to provide accelerated approval for a presymptomatic treatment based solely on biomarker (i.e., surrogate marker) endpoints without additional evidence to show that a treatment’s biomarker effects are “reasonably likely” to predict a clinical benefit^{[37,38,39,40,41]}. We believe that each of the most promising biomarker measurements should be included in clinical trials of AD-modifying treatments in order to show the extent to which an AD-modifying treatment moves the biomarkers, the extent to which the treatment moves the biomarkers in the right direction, and the extent to which a treatment’s biomarker effects predict a clinical benefit^[24,1].

Regulatory agencies may require evidence from presympomatic AD trials themselves to demonstrate the presymptomatic AD treatment’s biomarker effects are reasonably likely to predict a clinical benefit. On one hand, biomarkers are needed to evaluate presymptomatic treatments in a rapid and cost-effective way. On the other hand, clinically proven treatments are needed to help qualify biomarkers for use as reasonably likely surrogate endpoints. Among other things, the API is intended to help resolve this apparent catch-22. We propose to characterize and compare the effects of an amyloid-modifying treatment on FDG PET, volumetric MRI, fibrillar amyloid PET and CSF endpoints in cognitively normal people who, based on their age and genetic background, are at the highest imminent risk of symptomatic AD. If, after two years, the treatment fails to move one or more of the biomarkers in the right direction, the Data Safety Monitoring Board (DSMB) would declare futility and the research participants would be eligible to participate in another trial. If, however, the treatment does have the predicted biomarker effects, the trial would be continued long enough to detect an effect on pre-specified cognitive endpoints.

Before we describe the API in more detail, we briefly summarize relevant biomarker and cognitive findings from our ongoing longitudinal study of cognitively normal people with two copies, one copy and no copies of the apolipoprotein E (APOE) ε4 allele, the major late-onset AD susceptibility gene^[42].

Preliminary Findings

In 1994, we initiated a longitudinal brain imaging study of cognitively normal, initially late-middle-aged, cognitively normal APOEε4 homozygotes, heterozygotes, and non-carriers, reflecting three levels of genetic risk for late-onset AD, who were initially matched for their gender, age and educational level. Our subjects are followed every two years using FDG PET, volumetric MRI, and a battery of neuropsychological and clinical tests. In the last few years, they have begun to be following using fibrillar amyloid-β (Aβ) PET measurements, additional MRI measurements, and additional cognitive assessments; their DNA has been used to provide genome-wide association data, and they have begun to provide serum and plasma samples for use in ongoing and future analyses; and most recently, some have begun to provide CSF samples.

Of particular relevance to the API, we have demonstrated associations between APOE ε4 gene dose (i.e., three levels of genetic risk for AD) and a) baseline reductions and longitudinal declines in FDG PET measurements of the regional cerebral metabolic rate for glucose (rCMRgl)^{[43,19,44,20,21,22,45,46,47,48,49,50,51]}, b) longitudinal declines in volumetric MRI measurements of whole brain shrinkage^{[52,53,54,45]}, c) PET measurements of the magnitude and spatial extent of fibrillar amyloid burden^[23], and d) cross-sectional and longitudinal measurements of the decline in long-term verbal memory^{[55,56,57,58,59,60,61,62]}. Based on our findings, we have estimated the number of cognitively normal late-middle-aged APOE ε4 homozygotes or heterozygotes needed to evaluate presymptomatic AD treatments using the FDG PET, volumetric MRI and cognitive endpoints, and we developed voxel-based image-analysis algorithms with improved power to detect and track AD while addressing the Type I error associated with multiple regional comparisons^[36,63]. Meantime our amyloid PET findings suggest that fibrillar Aβ burden is “on the way up” in cognitively normal APOE ε4 homozygotes and heterozygotes in their 50s and 60s (before fibrillar amyloid levels reaches the plateau observed in symptomatic patients^[64]), that it will be possible to evaluate the differential effects of amyloid-modifying treatments in those carriers with more or less fibrillar Aβ (in case a presymptomatic treatment in normal people with significant Aβ burden is too little too late), and that it will be possible to assess the ability of these treatments to slow down further fibrillar Aβ deposition.

The Alzheimer’s Prevention Initiative (API)

The API proposes to evaluate investigational amyloid-modifying treatments in healthy people who, based on their age and genetic background, are at the highest imminent risk of developing AD symptoms using brain imaging, cerebrospinal fluid (CSF), and cognitive endpoints^[1]. This complements the newly established criteria for preclinical AD^[2], as well as other presymptomatic/preclinical AD treatment trials proposed by the Dominantly Inherited Alzheimer’s Network (DIAN)^[65], the AD Cooperative Study (ADCS), and others^[66,67].

The API has several goals: 1) to evaluate investigational AD-modifying treatments sooner than otherwise possible; 2) to determine the extent to which the treatment’s effects on brain imaging and other biomarkers predicts a clinical benefit—information needed to help qualify biomarker endpoints for use in pivotal prevention trials; 3) to provide a better test of the amyloid hypothesis than clinical trials in symptomatic patients, when these treatments may be too little too late to exert their most profound effect; 4) to establish AD prevention registries needed to support these and other presymptomatic AD trials; and 5) to give those individuals at highest imminent risk of AD symptoms access to the most promising investigational treatments in clinical trials.

In one trial, we propose to study cognitively normal AD-causing presenilin 1 [PS1] E280A mutation carriers, at least 35 years of age (i.e., within 10 years of the carriers’ estimated median age at clinical onset), from the world’s largest early-onset AD kindred, located in Antioquia, Colombia^[1]. This extraordinary kindred, which has been followed for more than 20 years by Dr. Francisco Lopera and his colleagues, and includes about 5,000 people^{[68,69,70,71,72,73,74,75,76,77,78,33,79,80,81,82,83]}, with a sufficient number of presymptomatic carriers in the targeted age group to make it possible to relate a treatment’s effects on both biomarker and clinical endpoints within 2–5 years. In the proposed trial, PS1 mutation carriers would be randomized to active treatment or placebo, non-carriers would be assigned to placebo, and genetic test findings would not be disclosed to the family members simply because they are participating in this trial^[84]. In the other trial, we propose to study cognitively normal 60–80 year-old APOE ε4 homozygotes (and possibly heterozygotes), close to their estimated median age of clinical onset^[1]. Including heterozygotes would depend on the safety and tolerability data for the chosen treatment, but would allow for both an increase in the available samples and an increase in the generalizablity of our findings^[85,86].

For each subject group, we have proposed to conduct a 24 month double-blind, randomized, placebo-controlled trial using fibrillar amyloid PET, FDG PET, volumetric MRI, CSF, and cognitive endpoints^[1]. Biological fluids and other MRI measurements would be used to permit exploratory studies. If after two years, the treatment is not associated with predicted effects on one or more of the biomarkers, the DSMB would declare futility, the trial would be discontinued, and the participants would be eligible to participate in a trial of the next most promising AD-modifying treatment. If, however, the treatment is associated with predicted biomarker effects, the trial would be continued to assess effects on our compound cognitive endpoint.

To support these and other presymptomatic AD trials, we plan to establish two AD prevention registries. We aim to enroll 2,000 members of the PS1 E280A kindred (along with DNA samples, PS1 E280A mutation testing and baseline cognitive assessments), about one-third of whom are projected to be mutation carriers, into the Colombian Registry by 2012. We aim to enroll several hundred thousand individuals in the US-based Alzheimer’s Prevention Registry, some of whom will be invited to provide saliva samples for DNA acquisition and APOE screening. In anticipation of the Colombian trial, we have begun to acquire and analyze brain imaging and CSF samples in the PS1 mutation carriers and non-carriers, a cyclotron/radiochemistry facility is being installed, and we have used data from the ongoing longitudinal study to estimate the sample sizes needed to detect a clinical effect using a composite cognitive endpoint.

While we believe that there is an opportunity to advance the evaluation of presymptomatic AD treatments, there is a responsibility to get it right. In October 2009, we hosted a meeting of 40 scientific advisors to get their input and help us to refine our proposal. In January 2010, we hosted a meeting with industry representatives to get their input and further refine our proposal^[86]. In January 2011, we again met with industry representatives, academic advisors, and FDA and European Medicines Agency (EMA) officials, who provided thoughtful and encouraging feedback^[85]. We have been communicating with pharmaceutical company leaders to explore their interest, the availability and timing of the most promising amyloid-modifying treatments, and to explore the scientific and logistical issues needed to prepare for our proposed trials. Selection of the drug will be made with the assistance of an independent academic advisory board, input from the affected kindred regarding potential benefits versus known adverse effects and previous use in human subjects, and will depend on factors such as target engagement, preclinical and clinical safety and tolerability data, dosing information, availability of the drug product, and in-kind industry support. We have proposed a mix of industry, philanthropic and federal funding, and we have indicated our intent to release the data to the public after the study is completed to help advance the use of biomarker and cognitive enrichment strategies and endpoints in future presymptomatic AD trials.

The proposed API treatment trials will help further develop the biomarker endpoints needed to evaluate a range of presymptomatic AD treatments, and will provide critically needed evidence to support the use of biomarker endpoints in the accelerated approval of presymptomatic AD treatments (i.e., to suggest that a treatment’s biomarker effects may be reasonably likely to predict a clinical benefit in these or other populations). We believe that by helping to determine the extent to which the best established brain imaging and CSF biomarkers of AD budge in response to treatment, the extent to which they move in the predicted direction, and the extent to which the treatment’s biomarker effects are associated with subsequent clinical benefit will help provide regulatory agencies the evidence they need to begin to consider approving presymptomatic treatments solely on biomarker endpoints in future trials. Moreover, these presymptomatic treatment trials will not only provide a better test of the amyloid hypothesis than clinical trials using the same treatment in symptomatic AD patient, but will also provide research participants at the highest imminent risk of symptomatic AD access to the most promising and suitable investigational treatments sooner than otherwise possible.