Introduction
The preclinical stages of Alzheimer’s disease (AD) have become the main focus of therapeutic clinical trials given the assumption that better outcomes can be achieved with changes in the course of the disease before the appearance of cognitive symptoms [
1,
2]. The International Working Group and the American Alzheimer’s Association have recently characterized preclinical AD as the presence of amyloid-β (Aβ) and tau abnormalities in cognitively normal persons [
1]. Although these individuals show greater rates of disease progression than cognitively normal biomarker-negative individuals, most of them remain stable over typical clinical trial periods [
1,
3]. Therefore, a critical next step proposed by the working groups was to identify, from among individuals with preclinical AD, those with the highest likelihood of disease progression within time frames acceptable for clinical trial designs taking into account financial, medical, and ethical considerations [
1].
Its slow rate of change makes the use of cognition as the primary outcome of clinical trials in individuals with preclinical AD difficult because it leads to prohibitively high sample sizes and long follow-up times [
4]. Hence, surrogate measurements of disease progression using established biomarkers of neurodegeneration might provide a useful alternative for such trials [
5]. In fact, few studies have tested changes in structural magnetic resonance imaging (MRI) as a possible surrogate marker for preclinical AD clinical trials. For example, a recent study by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) has suggested that to test for structural changes on MRI in a population of individuals with preclinical AD enriched on the basis of Aβ and tau biomarker positivity, sample size estimates are prohibitively large [
3]. In this regard, changes in [
18F]fluorodeoxyglucose ([
18F]FDG) uptake have been suggested for following disease progression and monitoring therapeutic effects [
6‐
11]. Indeed, [
18F]FDG positron emission tomography (PET) is one of the most important biomarkers of AD with applications in the research and clinical settings. However, the characteristics of [
18F]FDG uptake as a surrogate marker of preclinical AD in clinical trials are not fully known. Indeed, the inclusion of cognitively normal individuals harbouring Aβ and tau, and with abnormal [
18F]FDG uptake could be argued as an interesting strategy for these trials, since such individuals have a higher probability of developing further neurodegeneration and cognitive symptoms [
1]. However, it is reasonable to suggest that the absence of baseline neurodegeneration might offer a more favourable pathophysiological scenario for preventive therapies aiming to mitigate disease progression [
2].
Recent literature indicates that Aβ and tau pathologies may arise independently and, at some pathophysiological point, synergistically potentiate imminent neurodegeneration in preclinical AD [
2,
12‐
14]. Notably, this framework infers the existence of thresholds for Aβ and tau pathologies associated with the triggering of their deleterious synergy in potentiating neurodegeneration. However, the fact that the majority of Aβ-positive plus tau-positive individuals with preclinical AD remain pathophysiologically stable over long periods suggests that these thresholds are greater than their respective individual thresholds for abnormality [
1]. Thus, determination of the Aβ and tau biomarker thresholds associated with imminent neurodegeneration might provide complementary information in individuals with preclinical AD destined to develop AD-related progression.
In this study, we tested the hypothesis that the combination of optimized Aβ and tau thresholds predictive of neurodegeneration might provide an alternative framework with a high statistical power for testing the efficacy of the emerging disease-modifying therapies. This framework has the potential to select preclinical individuals on the verge of AD-related progression, considering metabolic changes as indices of neurological decline.
Discussion
In summary, our results suggest that the Aβ and tau thresholds associated with imminent AD-related metabolic decline in individuals with preclinical AD are higher than their respective thresholds for abnormality. In addition, we showed that the use of these thresholds and the use of voxel-wise changes in [
18F]FDG uptake as a surrogate marker as inclusion criteria provide an alternative framework with a high statistical power for 2-year clinical trials in preclinical AD. Overall, we propose the new concept of Aβ and tau thresholds for imminent neurodegeneration as a valuable approach to population enrichment in clinical trials in individuals with preclinical AD and a high probability of developing AD-related neurodegeneration within a short time. Specifically, we demonstrated that a 2-year clinical trial in cognitively normal individuals using brain Aβ plus CSF p-tau thresholds more than one SD higher than their standard values together with regional [
18F]FDG uptake as a surrogate marker would require as few as 100 individuals to test a hypothetical 25% drug effect. These results contrast with a recent study performed by ADNI in individuals with preclinical AD defined as those with Aβ plus tau positivity which showed that a 2-year clinical trial measuring changes in cognition or structural MRI would require more than 2,000 individuals to test a hypothetical 25% drug effect with 80% power [
3].
It is important to emphasize that the thresholds proposed here are based on dynamic biomarker changes and indicate metabolic decline over short time frames, rather than the presence or absence of pathological proteinopathies, which are best assessed with post-mortem correlations [
28]. It is also important to mention that in our understanding the term “biomarker abnormality” refers to the presence of brain pathology and should be defined by thresholds that are better associated with post-mortem studies. Within the biomarker abnormality spectrum, we suggest that the existence of thresholds for imminent disease progression will identify from among individuals with biomarker abnormalities those with the highest probability of progression to a given outcome.
Since the proposed model optimizes neurodegeneration as a function of Aβ and tau levels, we may argue that this model has immediate applicability to provide a framework with a high statistical power for use with the emerging anti-tau and anti-amyloid therapies. Interestingly, the highest rates of metabolic decline shown by individuals with preclinical AD were found in the limbic structures within the cingulate, orbitofrontal, and medial and basal temporal cortices. Importantly, this pattern of metabolic dysfunction is not considered part of normal ageing [
29,
30]. In fact, metabolic decline in the limbic structures has been proposed as an early abnormality associated with AD [
31,
32]. In addition, the fact that this brain circuit is affected by tangles or plaques [
29,
30], rather than by the normal ageing process [
33‐
35], further supports the notion that cognitively normal subjects with these baseline biomarker signatures are on the AD pathway. Interestingly, in contrast with [
18F]florbetapir, CSF Aβ levels failed to model imminent metabolic decline in our population, which is in line with previous observations showing that Aβ PET depicts metabolic decline better than CSF Aβ in this preclinical AD population [
13]. Although CSF Aβ well-represents the disease status, brain fibrillar Aβ deposition seems to be better associated with imminent disease progression [
13,
36].
Declines in [
18F]FDG uptake as a function of the hallmark AD proteins were best described by a sigmoidal rather than a linear association. This pattern of relationship is consistent with the most accepted models of AD progression [
22], which assume “ceiling effects” for Aβ and tau. Therefore, our results further emphasize that this ceiling effect should be considered in studies testing the relationship between these proteinopathies and downstream neurodegeneration. For example, studies using linear functions for testing the association between proteins and brain hypometabolism in AD patients might not show any correlation [
13], since it is likely that most demented patients have already reached the plateau of the relationship.
Methodological aspects limit the external validity of the present results. Our population was composed of self-selected individuals and might not represent the general elderly cognitively normal population. Therefore, it would be highly desirable to replicate our results in a larger population-based study. However, it is important to mention that the use of a conservative multiple comparison correction threshold of 0.001 helped to avoid false-positives results in our analysis. It is also important to emphasize that change in cognition is always the most desirable outcome for a therapeutic clinical trial aiming to mitigate AD progression. However, due to the methodological limitations in the use of cognition in individuals with preclinical AD and the advances in in vivo biomarkers, increasing our understanding of the use of brain imaging in the assessment of AD progression is of paramount importance. Importantly, individuals with preclinical AD in this study were considered those with normal cognition but Aβ and tau abnormalities rather than those who would certainly have developed dementia over time. The 2-year follow-up was not sufficient to draw definite conclusions regarding clinical progression from cognitively normal to dementia.
Partial volume correction did not translate into differences in our final results. Therefore, we present the PET data without corrections for volume effects. Biomarker thresholds are invariably subject to idiosyncrasies and as such might vary slightly depending on the analytical method used. However, in this study, biomarker levels predictive of rapid metabolic decline were at least one SD higher than the threshold for the presence of brain pathology in an elderly cognitively normal population. Notably, the thresholds associated with imminent disease progression would invariably be expected to be higher than those used to determine an abnormal biomarker status. It is important to mention that the use of a hypothetical drug effect of 25% used in our analysis was the same as that used in previous studies in AD [
3,
24,
25,
27]. Although study population enrichment with two biomarker modalities might potentially have a high economic cost, it is important to emphasize that an increasingly large number of observational studies using multiple biomarkers for Aβ and tau have indicated that this approach could serve as a screening tool to select individuals for therapeutic trials. Despite the current lack of an effective disease-modifying therapy for AD, the promising recent results with anti-Aβ drugs such as aducanumab suggest the need for sensitive methods to detect disease progression.
To conclude, our results show that Aβ and tau biomarker thresholds associated with imminent neurodegeneration are higher than their respective thresholds for abnormality. In addition, the determination of these thresholds may provide complementary information for selecting individuals with preclinical AD who will most likely show pathophysiological progression over time frames compatible with clinical trials.
Acknowledgements
Data collection and sharing for this project was funded by the Alzheimer’s Disease Neuroimaging Initiative (ADNI; National Institutes of Health grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). ADNI is funded by the National Institute on Aging and the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association, Alzheimer’s Drug Discovery Foundation, Araclon Biotech, BioClinica, Biogen, Bristol-Myers Squibb Company, CereSpir, Eisai Inc., Elan Pharmaceuticals, Eli Lilly and Company, EuroImmun, F. Hoffmann-La Roche and its affiliated company Genentech, Fujirebio, GE Healthcare, IXICO, Janssen Alzheimer Immunotherapy Research & Development, Johnson & Johnson Pharmaceutical Research & Development, Lumosity, Lundbeck, Merck, Meso Scale Diagnostics, NeuroRx Research, Neurotrack Technologies, Novartis Pharmaceuticals Corporation, Pfizer, Piramal Imaging, Servier, Takeda Pharmaceutical Company, and Transition Therapeutics. The Canadian Institutes of Health Research providies funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health (
www.fnih.org). The grantee organization is the Northern California Institute for Research and Education, and the study was coordinated by the Alzheimer’s Disease Cooperative Study at the University of California, San Diego. ADNI data are disseminated by the Laboratory of Neuro Imaging at the University of Southern California.
Compliance with ethical standards
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.