Introduction
Alzheimer’s disease (AD) comprises a long asymptomatic preclinical stage characterized by pathophysiological changes that start decades before overt clinical manifestations [
1]. Abnormal brain accumulation of β-amyloid (Aβ) is thought to be among the earliest detectable events occurring along the AD continuum, followed by tau aggregation and cerebral atrophy [
1]. PET imaging allows the detection of Aβ plaques across the brain in vivo. In clinical settings, Aβ PET scans are typically assessed visually and categorized as negative or positive. To this end, scans are assessed in several brain regions and categorized as positive if significant Aβ is detected in at least one culprit brain region. According to an investigation of this type, Aβ accumulation can be detected in cognitively unimpaired (CU) individuals decades before the onset of clinical symptoms [
2]. In addition, there is strong evidence suggesting that global Aβ positivity in CU individuals is associated with future cognitive decline [
3‐
5].
Even though the regional pattern of Aβ positivity is usually not taken into consideration in clinical routine, there is evidence pointing that the consideration of this regional pattern is of prognostic value [
6], as Aβ accumulation typically follows a defined spatial–temporal pattern progression across the AD continuum [
7]. Cortical regions, such as the precuneus, insular, cingulate, and orbital cortices, generally show Aβ deposition earlier than temporal or striatal regions [
7]. In line with this, recent literature shows that changes in regional Aβ, as measured by PET quantification, can predict episodic memory decline in CU individuals, particularly in the precuneus [
8,
9], posterior cingulate cortex, and lateral parietal cortices [
9]. Therefore, regional Aβ PET measures may be better suited for predicting cognitive decline than global positivity as defined when any region is read as positive. Specifically, regional Aβ PET analyses could potentially be useful for intervention trials, insofar as Aβ positivity in key brain regions might identify those CU individuals who are at greater risk for developing AD [
10,
11].
The
APOE-ε4 allele represents the major genetic factor for non-autosomal inherited AD, and it is associated with an earlier and higher cerebral Aβ deposition [
12‐
14] which is proportional to the number of
ε4 alleles [
12]. Carrying this allele is also associated with a greater risk of AD dementia, younger age of symptoms onset, and faster cognitive decline [
15]. However, the impact of the
APOE-ε4 allele on cognition in late-/middle-aged CU individuals remains unclear. Previous studies reported a faster cognitive decline in
ε4 carriers than in non-carriers [
16,
17], especially when individuals are Aβ positive [
18‐
20], although another study with a similar design did not confirm such longitudinal association [
21]. Similarly, cross-sectional studies reported worse cognitive performance in CU
ε4 carriers compared to non-carriers [
22,
23]. The relationship between regional Aβ accumulation and
APOE-ε4 status associated with cognition in CU has been scantily explored. In a previous study including 408 CU, Kantarci et al. [
18] showed that
APOE-ε4 carriers displaying Aβ accumulation in frontal, temporal, and parietal lobes presented worse global cognitive performance.
In the present study, we aimed to determine whether the presence of the APOE-ε4 allele modifies the association between global and regional Aβ PET visual reads (VR) and cognitive decline in middle-aged CU individuals. We hypothesized that (1) the APOE-ε4 allele modulates the relationship between regional Aβ burden and cognitive changes in the early asymptomatic stages of the AD continuum, and (2) regional Aβ positivity is predictive of cognitive decline in CU individuals at higher risk of developing AD. We primarily assessed cognitive decline by means of the Preclinical Alzheimer Cognitive Composite (PACC), wherein we separately analysed scores for episodic memory and executive function. We further examined these relationships using regional positivity determinations based on standardized update value ratio (Centiloids) as a quantitative measurement of Aβ deposition and explored whether the studied associations could be related to hippocampal volume as a measure of neurodegeneration.
Discussion
The present study aimed to determine whether regional VR of Aβ PET interacts with APOE-ε4 status to predict cognitive decline in a cohort of middle-aged CU participants at high risk of AD. Specifically, APOE-ε4 carriers who were VR+ in lateral temporal regions and in the striatum displayed a significantly steeper cognitive decline, as measured with the PACC, than non-carriers and VR- individuals. This interaction could not be detected with global VR+ and persisted after controlling for hippocampal volume, as a measure of neurodegeneration. The result in the temporal region was replicated when determining regional positivity with quantitative Centiloids-based cut-off values for positivity (Centiloids+). With regard to specific cognitive domains, we also found significant interactions with APOE-ε4 in temporal regions and the striatum in both VR+ and Centiloids+ on episodic memory and executive function. Again, these associations were not observed with the global measures of amyloid positivity and persisted after controlling for hippocampal volume.
The main novelty of our study resides in that it highlights the value of regional VR of Aβ PET to detect cognitive decline in CU in
APOE-ε4 carriers that, to the best of our knowledge, has not been previously described. These findings are in line with previous reports using regional Aβ positivity using continuous SUVR values [
8,
9,
18].
Visual and Centiloid-based determinations of regional Aβ positivity were in high concordance (> 95%), thus showing that regional VR is a valid and comparable alternative to quantitative methods. This is in agreement with our previous assessment of the inter- and intra-reader agreement of regional VRin all brain regions [
6]. Taken together, regional patterns of VR positivity, readily available in clinical settings, have a similar capacity as Centiloids-based methods, usually used in research, to assess the risk of cognitive decline associated with regional patterns of Aβ positivity. Therefore, such a good agreement may help bridge methodological discrepancies in regional PET quantification between studies.
According to the spatial–temporal sequencing of Aβ accumulation, the regions with positive Aβ VR that showed a significant interaction with
APOE-ε4 status on cognitive decline (temporal regions and the striatum) are typically considered to be accumulating Aβ later in the AD continuum than those in which this interaction was not found (e.g. frontal and PC/PCC) [
7]. Therefore, it is sensible to hypothesize that cognitive decline is closer when Aβ deposition reaches these late-accumulating regions. Supporting this hypothesis, such an interaction was not observed with global measurements of Aβ burden, which are mainly driven by positivity in early regions. In line with our results, recent literature has linked Aβ accumulation in the striatum with a higher risk of cognitive decline among non-demented individuals with elevated cortical Aβ [
40].
Our results show that
APOE-ε4 carriers with Aβ VR+ in late amyloid accumulation regions exhibited a significantly steeper cognitive decline compared with non-carriers. This finding is in agreement with previous reports showing that Aβ and
APOE-ε4 interact to influence short-term decline in preclinical AD [
3‐
5]. It is well established that
APOE-ε4 carriers show an earlier and higher cerebral Aβ deposition. On top of this, literature supports the existence of additional mechanisms that may promote cognitive decline in CU
APOE-ε4 carriers with higher Aβ accumulation. Evidence suggests a higher vulnerability of
APOE-ε4 carriers to the toxic effects of Aβ on neuronal integrity that could impact brain processes such as tau phosphorylation, mitochondrial activity, or neuroinflammation [
41]. However, many of our results remained significant after controlling for hippocampal volume, suggesting that the effect of neurodegeneration could be regionally specific. Finally,
APOE-ε4 carriers may have higher levels of underlying tau pathology and cerebrovascular disease that could be hypothesized as possible mechanisms underlying cognitive decline in those individuals [
41]. In line with this, we found a significant interaction between
APOE-ε4 and VR Aβ accumulation in the striatum with regard to cognitive decline. In agreement with this interpretation, it has been suggested that this association indicates the duration or severity of Aβ burden or signals towards an increase in tau-pathology [
14,
40]. Moreover, it has been suggested that tau might partially mediate the deleterious effect of striatal Aβ on cross-sectional cognition [
40]. However, the loss of significance in regional Centiloids could be related to quantification in this region which is susceptible of contamination by white matter uptake [
6,
7]. Regarding the temporal lobe, a recent study on the sample included here showed that
APOE-ε4 carriers were more prone to Aβ aggregation in temporal areas for any given level of soluble Aβ dyshomeostasis. This finding suggests that
APOE-ε4 facilitates the spread of Aβ in these regions, promoting an earlier co-localization with tau [
42] to trigger AD-related neurodegeneration. Taken together, these studies suggest an interplay between patterns of Aβ and tau spread in determining cognitive decline, which would explain why positivity in regions of late Aβ accumulation is associated with cognitive decline in
APOE-ε4 carriers in the present study. Still, previous studies classified the individuals as globally Aβ positive using quantitative methods mostly available in research settings.
The present work mainly measured cognitive change with the PACC, a gold standard global cognitive composite used in the context of the preclinical stage of AD. This measure combines the two cognitive composites for which we have found significant results, namely episodic memory and executive function. In addition, the PACC was constructed to maximize the sensitivity to detect the earliest Aβ-related cognitive changes [
11]. Regarding specific cognitive domains, we found an interaction between
APOE-ε4 and Aβ aggregation to promote episodic memory and executive function decline in line with previous reports[
18,
19]. We observed that
APOE-ε4 carriers with Aβ VR positivity in the striatum exhibited worse episodic memory performance in the follow-up visit. These results are in alignment with a recent study that found that amyloid accumulation predicts memory decline in 133 CU individuals [
8]. In particular, precuneal Aβ burden predicated immediate and delayed episodic memory performance in the whole population, whereas lateral orbitofrontal Aβ burden predicted working and semantic memory performance in Aβ negative baseline group [
8]. However, the effect of
APOE-ε4 status in this association was not explored. Here, we found a significant association between decline in episodic memory and Aβ aggregation in a larger and younger sample, although exclusively to
APOE-ε4 carriers and late Aβ accumulation regions. With regard to executive function, we found a significant interaction between
APOE-ε4 and regional Aβ accumulation in temporal and parietal regions and in the striatum. This result is in agreement with the previously mentioned higher risk of cognitive decline with Aβ accumulation in the striatum among non-demented individuals [
40]. However, our results demonstrated an additional higher risk for decline in the performance of specific cognitive domains, namely episodic memory and executive function. Therefore, our results show the capacity of regional VR to detect those individuals that have persistent Aβ deposition at the early stages of the AD continuum that, consequently, are at higher risk of cognitive decline.
Some limitations of this work should be considered. First, the number of follow-up visits may be insufficient to interpret results in terms of the temporal evolution of the events. More extended longitudinal studies are required to track the progression of cognitive change related to regional Aβ PET uptake in
APOE-ε4 carriers and non-carriers to improve our understanding of the plausible mechanisms relating the
APOE genotype and AD pathogenesis. In turn, further longitudinal data collection will help to confirm whether regional VR has a similar prognostic value compared to quantification [
6]. Second, replication of the findings presented here in independent cohorts is essential. To overcome these limitations, the second follow-up visit of the longitudinal ALFA+ study is currently ongoing, including both cognition and amyloid PET acquisition, which is being collected in the context of the AMYPAD Consortium [
10]. The replication in larger cohorts will also help addressing the relatively few cases of positivity in our study. Nevertheless, we verified that our results were not driven by the presence of a few extreme cases, as no outliers were detected in our sample using standard methods (cases exceeding the median plus or minus 1.5 times the interquartile range). Third, the lack of AD biomarkers at the baseline cognitive assessment is another relevant limitation for interpreting our findings. Even though we cannot know the amyloid status at baseline, evidence shows that Aβ deposition in preclinical AD is slow and protracted, likely to extend for more than two decades [
43]. Therefore, given such a slow accumulation rate, 4 years are still a relatively short period to observe major changes in deposited Aβ. Last, in our cohort, individuals with relevant medical conditions or neurologic disease were excluded. As a result, our sample is healthier than expected from an age-matched cohort selected from the general population. Further, participants are younger than those in previous studies [
18‐
20]. However, due to the lack of other comorbidities and the young age in our cohort, the present sample represents the most suitable population for identifying subtle changes in the early stages of the AD continuum, such as cognition.
Acknowledgements
This publication is part of the ALFA study. The authors would like to express their most sincere gratitude to the ALFA project participants and relatives without whom this research would have not been possible. Collaborators of the ALFA study are: Alba Cañas, Lidia Canals, Laura Iglesias, Paula Marne, Annabella Beteta, Carme Deulofeu, Maria Emilio, Irene Cumplido, Ruth Domínguez, Sherezade Fuentes, Laura Hernández, Marc Vilanova, Lluís Solsona, Gema Huesa, Jordi Huguet, Tania Menchón, Albina Polo, Sandra Pradas, Aleix Sala-Vila, Anna Soteras, Laura Stankeviciute, Müge Akinci, Eleni Palpatzis, Patricia Genius, Blanca Rodríguez, Marina García, and Paula Ortiz-Romero. Authors would like to thank GE Healthcare for kindly providing [18F]flutemetamol doses of ALFA+ participants.
Declarations
Competing interests
AB-S, GS-B, RC, GS, MS, LEC, AN-B, MM-A, NV-T, GO, CF, OG-R, EMA-U, CM, KF, and JDG report no existing potential conflicts of interest relevant to this article. BvB 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. MS-C has served as a consultant and at advisory boards for Roche Diagnostics International Ltd and has given lectures in symposia sponsored by Roche Diagnostics, S.L.U and Roche Farma, S.A. JLM is currently a full-time employee of H. Lundbeck A/S and previously has served as a consultant or on advisory boards for the following for-profit companies or has given lectures in symposia sponsored by the following for-profit companies: Roche Diagnostics, Genentech, Novartis, Lundbeck, Oryzon, Biogen, Lilly, Janssen, Green Valley, MSD, Eisai, Alector, BioCross, GE Healthcare, and ProMIS Neurosciences.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.