Introduction
The most common cause of dementia is Alzheimer’s disease (AD), which is characterized by the accumulation of amyloid beta plaques and neurofibrillary tau tangles [
1]. The ATN classification provides a framework to diagnose AD based on biomarkers providing an indication of these pathologic changes [
2]. In this framework, individuals are classified by the presence or absence of amyloid (A), hyperphosphorylated tau (T), and neurodegeneration (N), resulting in eight possible ATN profiles. We previously showed that in cognitively normal individuals with subjective cognitive decline (SCD), A+ was associated with a higher risk of dementia [
3].
According to the amyloid cascade hypothesis, the accumulation of amyloid in the brain initiates a series of events including the formation of neurofibrillary tau tangles and neuronal cell loss, eventually resulting in cognitive decline [
4]. Therefore, it is assumed that individuals become A+ before turning T+ or N+. Only a few studies evaluated the temporal ordering of ATN biomarker abnormality in a longitudinal manner. One study in a mixed population of cognitively normal individuals and individuals with mild cognitive impairment (MCI) indeed found that most often, A became abnormal first, yet also described there were multiple routes, specifically A → T → N, A → N → T, T → A → N and N → A → T [
5].
In the ATN classification, biomarkers are treated as dichotomous variables. With research interest shifting to the very early stages of AD, “grey zone” amyloid burden and subthreshold amyloid accumulation were found to be associated with memory decline, showing additional value of amyloid burden in the perithreshold range [
6‐
8]. One study in a population consisting of cognitively normal and MCI individuals investigated change from A− to A+ rather than accumulation rate and found that lower baseline cognition and APOE ε4 carriership were predictive of changing to A+ [
9]. Investigating determinants of amyloid accumulation, specifically in the early stages of disease, therefore has clinical relevance.
The aims of this study were to (1) identify determinants of change in amyloid status, (2) describe changes in ATN profiles over time, and (3) evaluate change in amyloid status as predictor of cognitive decline, in cognitively normal individuals.
Discussion
In our sample of cognitively normal individuals with SCD, we found that biomarker abnormality increased over a 2.5-year period. There was considerable variability in the order of biomarkers becoming abnormal in the ATN classification, suggesting no fixed order. Change from A− to A+ was associated with steeper decline in tests of attention and executive function.
We showed that the number of individuals with an abnormal biomarker status increased for both A, T, and N, over a time course of 2.5 years. Most of these individuals changed from a negative to positive biomarker status, yet of note, a smaller number of individuals changed from positive to negative. There are not many longitudinal studies investigating changes in the ATN classification; hence, the phenomenon of change from a positive to a negative status has not yet received much attention. One study investigating amyloid burden excluded individuals who were amyloid positive at one time point, and negative at the next, but did not specify the number of individuals [
9]. Another excluded the 5% of individuals with borderline amyloid PET burden, reducing the risk of individuals crossing the threshold due to small changes [
5]. Other studies investigating amyloid accumulation rate as continuous measure showed a negative slope in some individuals but did not address the possibility of reverting amyloid status and explained the negative slope by random variation, noise, or actual clearance of amyloid [
7,
8]. One potential explanation for the five individuals changing from a positive to negative amyloid status by visual rating
(positive-negative) could be that these scans were false-positive at baseline. While the quantitative measures might not have changed much over time, scan could be visually assessed differently at the two time points due to an imperfect intra-rater agreement [
24‐
26]. This is part of clinical practice, especially in early disease stages. Therefore, acquiring follow-up scans is highly useful in individuals with equivocal scans with grey zone amyloid burden. Of note, one could also argue that also the
negative-positive scans could be the result of rater variability and that their changing from negative to positive does not necessarily reflect clinical relevance. However, a substantial portion of these individuals had grey zone amyloid burden at baseline, which is already associated with a changed slope in memory function, as shown by our group previously [
6]. We found that this group has a steeper decline in performance on Stroop I and III, which are indicative of attention and executive functioning. This is in line with other studies showing that amyloid burden in the subthreshold range is associated with cognitive decline and highlights the clinical relevance of grey zone amyloid burden [
7,
8]. In general, an amyloid status based on visual assessment is not identical to an amyloid status based on a threshold for quantitative measures. Quantitative measures are not directly affected by rater variability and could therefore be interpreted as more consistent. However, quantitative measures of amyloid burden are often averaged over a larger ROI. If a scan is visually assessed as A+ based on a relatively small area, this does not necessarily translate in a higher average BP
ND in the total ROI, which could be a potential cause of differences between the two approaches. Overall, we found a relatively high degree of changing biomarkers in a short time frame. These results add to the literature suggesting the clinical relevance of changing from a negative to a positive amyloid status.
When we compared ATN profiles over time, we found 44% of individuals changed to a different ATN profile during 2.5 years of follow-up. Data on changing biomarkers enable the evaluation of the actual sequence of biomarker abnormality. Of note, most (11/17) individuals followed a different sequence than the overall accepted hypothesis of A becoming abnormal first, then T and N last [
2]. In our sample, individuals changed to T+ or N+ while still being A− or changed to N+ before T+. These findings are in line with those of a former study investigating change in ATN profiles, which also found multiple sequences [
5]. There are several possible explanations for these observations. First, amyloid could already be accumulating in the subthreshold range in individuals changing to T+ or N+, but before A+, suggesting the pathological process has started just below the detection threshold. An alternative explanation is the suggestion of the dual-pathway hypothesis, in which amyloid and tau accumulation are both the result of a common upstream event, not necessarily causally related to each other [
27]. Finally, there could be mixed pathology, resulting in N+ due to other diseases than AD, hence not related to a specific ordering of events. Overall, the number of individuals with the A−T−N− profile became smaller and the number of individuals with non-AD pathologic change (A−T+N−, A−T−N+, A−T+N+) became larger at follow-up. In a previous study by our group, but also in other studies, these profiles did not have a higher risk of cognitive decline or clinical progression to MCI or dementia [
3,
28].
When we evaluated determinants of change to amyloid positivity, we found APOE ε4 carriers had a higher baseline amyloid burden, a higher risk of transition from A− to A+ and a higher annual amyloid accumulation rate. Several studies confirm a relationship between ε4 carriership and a higher accumulation rate [
29‐
31], although not all [
7,
32]. The relationship between ε4 carriership and a higher risk of change from A− to A+ has also been confirmed [
9,
33]. We add to these results with the finding that ε4 carriership is also associated with risk of change in a sample of cognitively normal individuals with SCD. We did not find evidence for an association with any of the other factors examined, such as baseline age, sex, or education level. In apparent contrast with former studies [
9,
30], we did not find a relationship between a lower baseline cognitive performance and subsequent amyloid accumulation. Reasons for this inconsistency could be that an inclusion criterion for our study is normal performance at baseline and that variability in baseline cognition is small. Therefore, relationships with amyloid accumulation may be obscured. In short, our results suggest A− individuals who are ε4 carrier are still at risk of progression to A+.
Limitations of our study include that our sample size was relatively small. With a larger sample size, our study would have had more power to detect more subtle determinants of changes in A status. The results of our analyses examining changing amyloid status as predictor of cognitive decline should also be interpreted with caution and replicated in larger samples. Furthermore, we used [
18F]flortaucipir PET as measure of tau burden. We pragmatically used Gaussian mixture modeling of [
18F]flortaucipir to obtain a threshold, although there might be other approaches. Nevertheless, our approach resulted in a percentage of T+ which lies within the range of T+ described in other studies in cognitively normal individuals [
3,
34‐
36]. Of note, during the recruitment of individuals for the [
18F]flortaucipir PET scan, we slightly oversampled A+ individuals. Because substantial tau pathology within A− cognitively normal individuals is not expected to be present, we selected more A+ individuals for the [
18F]flortaucipir PET in order to have a broader spectrum of amyloid and tau pathology. Therefore, our results might not reflect the true prevalence of amyloid and tau pathology in cognitively normal individuals and the results might not be directly generalizable to the general population. Another factor that potentially impacts generalizability is the fact that the individuals in our sample were mainly recruited at a memory clinic. Strengths include the longitudinal nature of the study with the availability of biomarkers, diagnoses, and cognition with substantial duration of follow-up. Furthermore, we used dynamic scan protocols which enabled us to calculate BP
ND, which is a more accurate measure of amyloid and tau load than the semi-quantitative SUVr. Another strength is our use of [
18F]flortaucipir for the definition of “T,” since it does not suffer from off-target binding to amyloid plaques or TDP-43 and correlates well with Braak neurofibrillary tangle stages [
37].
Concluding, we showed biomarker status changes in cognitively normal individuals with SCD. There was considerable variability in the sequence of ATN biomarkers becoming abnormal, suggesting that there is not one (causal) order of events. Changing from a negative to positive amyloid status was associated with APOE ε4 carriership and predicted subtle cognitive decline, suggesting the potential clinical relevance of amyloid burden in the negative range.
Declarations
Competing interests
Jarith Ebenau reports no disclosures relevant to the manuscript.
Denise Visser reports no disclosures relevant to the manuscript.
Lior Kroeze reports no disclosures relevant to the manuscript.
Mardou van Leeuwenstijn reports no disclosures relevant to the manuscript.
Argonde van Harten reports no disclosures relevant to the manuscript.
Albert Windhorst reports no disclosures relevant to the manuscript.
Sandeep Golla reports no disclosures relevant to the manuscript.
Ronald Boellaard reports no disclosures relevant to the manuscript.
Philip Scheltens has acquired grant support (for the institution) from Biogen. In the past 2 years, he has received consultancy/speaker fees (paid to the institution) from Probiodrug Biogen, EIP Pharma, Merck AG.
Frederik Barkhof is a consultant for Biogen-Idec, Bayer-Schering, Merck-Serono, Roche, NovartisIXICO and Combinostics; has received sponsoring from European Commission-Horizon 2020, National Institute for Health Research-University College London Hospitals Biomedical Research Centre, TEVA, Novartis, and Biogen; and serves on the editorial boards of Radiology, Brain, Neuroradiology, Multiple Sclerosis Journal, and Neurology.
Bart van Berckel has received research support from EU-FP7, CTMM, ZonMw, NWO, and Alzheimer Nederland. BvB has performed contract research for Rodin, IONIS, AVID, Eli Lilly, UCB, DIAN-TUI, and Janssen. BvB was a speaker at a symposium organized by Springer Healthcare. BvB has a consultancy agreement with IXICO for the reading of PET scans. BvB is a trainer for GE. BvB only receives financial compensation from Amsterdam UMC.
Wiesje van der Flier Research programs have been funded by ZonMW, NWO, EU-FP7, EU-JPND, Alzheimer Nederland, CardioVascular Onderzoek Nederland, Health~Holland, Topsector Life Sciences & Health, stichting Dioraphte, Gieskes-Strijbis fonds, stichting Equilibrio, Pasman stichting, Biogen MA Inc., Boehringer Ingelheim, Life-MI, AVID, Roche BV, Fujifilm, Combinostics. WF holds the Pasman chair. WF has performed contract research for Biogen MA Inc and Boehringer Ingelheim. WF has been an invited speaker at Boehringer Ingelheim, Biogen MA Inc., Danone, Eisai and WebMD Neurology (Medscape). WF is consultant to Oxford Health Policy Forum CIC, Roche, and Biogen MA Inc. WF was associate editor at Alzheimer’s Research & Therapy (2020-2021); she is associate editor of Brain (2021-). All funding is paid to her institution.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.