APOE4 impact on soluble and insoluble tau pathology is mostly influenced by amyloid-beta

Claudia Cicognola, Gemma Salvadó, Ruben Smith, Sebastian Palmqvist, Erik Stomrud, Tobey Betthauser, Sterling Johnson, Shorena Janelidze, Niklas Mattsson-Carlgren, Oskar Hansson, Alexa Pichet Binette
doi: https://doi.org/10.1101/2024.09.20.24314064

Abstract

The APOE4 allele is the strongest genetic risk factor for sporadic Alzheimer’s disease (AD). While APOE4 is strongly associated with amyloid-beta (Aβ) accumulation, its relationship with tau accumulation is less understood. Studies evaluating the role of APOE4 on tau accumulation have shown conflicting results, particularly regarding the independence of these associations from Aβ load. To clarify the relations between APOE4, Aβ and tau, we examined three independent longitudinal cohorts (the Swedish BioFINDER-1, BioFINDER-2 and WRAP cohorts) in which participants had cross-sectional and longitudinal measures of tau tangles (tau-PET; temporal meta-ROI and entorhinal) or soluble p-tau (p-tau217), Aβ-PET and APOE genotype. The study included a total of 1370 cognitively unimpaired (CU) and 449 mild cognitive impairment (MCI) subjects, followed longitudinally with tau-PET and p-tau217. APOE4 carriers accounted for 40.2-50% of the cohorts. Different linear regressions (cross-sectional) and linear mixed-effect models (longitudinal) with tau measures as outcomes were fitted to test the effect of APOE4 as independent predictor, as well as in combination with baseline Aβ load or the interaction term between APOE4 and Aβ load. All models included age, sex and cognitive status as covariates.

We found no independent effects of the APOE4 carriership on insoluble tau aggregates in either cohort (BioFINDER-2 or WRAP), both on cross-sectional and longitudinal tau-PET levels in the temporal meta-ROI, when Aβ was present in the model (p=0.531-0.949). Aβ alone was the best predictor of insoluble tau accumulation, and there was no interaction between APOE4 and Aβ on tau-PET. Similarly, no independent effects of the APOE4 carriership on baseline (p=0.683-0.708) and longitudinal (p=0.188-0.570) soluble p-tau217 were observed when Aβ was included in the model in BioFINDER-1 and WRAP. No interaction between APOE4 and Aβ on soluble p-tau217 was observed. Furthermore, mediation analysis revealed that Aβ load fully mediated most associations between APOE4 and tau (46-112%, either cross-sectional or longitudinal tau-PET or soluble p-tau217). In the largest cohort (BioFINDER-2), looking at APOE4 groups based on the number of ε4 alleles, we found an interaction between APOE4 homozygotes only and Aβ on tau-PET levels at baseline and over time.

In conclusion, although APOE4 is strongly associated with Aβ aggregation, it seems to be minimally associated with longitudinal changes in soluble or insoluble p-tau levels at a given level of Aβ pathology, confirming the primacy of Aβ in driving tau pathology.

Introduction

The APOE4 allele is the strongest genetic risk factor for sporadic Alzheimer’s disease (AD) [1–4]. Associations between APOE4 genotype and amyloid β (Aβ) are also well-established, with carriers being more likely to be Aβ-positive and accumulating Aβ earlier in life than non-carriers [5–8]. Following the deposition of Aβ plaques in the brain, the deposition of neocortical tau aggregates ensues years later [9]. However, associations between APOE and longitudinal accumulation of soluble and insoluble hyperphosphorylated tau (p-tau) remain unclear. Some studies suggest that the presence of the ε4 allele potentiates the effects of Aβ on accumulation of insoluble tau aggregates and tau accumulation might be accelerated in ε4 carriers at lower Aβ levels [10–15]. However, others found no additional effect of APOE4 on tau deposition beyond the effects on Aβ [16, 17]. It is important to clarify the role of APOE4 on the two core pathologies of AD, considering recent efforts to develop APOE-targeting therapies. By an improved understanding of which AD pathology is most influenced by APOE4 allele, and at which levels, we can better define the optimal target populations and outcomes in trials evaluating APOE-modifying therapies [18, 19]. For example, if only Aβ accumulation, and not tau accumulation, is associated with APOE genotype, APOE-targeting therapies should preferably be evaluated in primary (or potentially secondary) prevention trials. However, if APOE4 has a strong direct effect on tau accumulation as well, trials could also include later disease stages such as cognitively impaired AD populations.

In the present study, we therefore aimed to assess if the presence of the APOE4 allele modified the associations between insoluble tau (as measured with tau-PET) or soluble p-tau217 (as measured in CSF or plasma) and Aβ pathology, and whether APOE4 had Aβ-independent effects on cross-sectional and longitudinal tau deposition. We examined cognitively unimpaired (CU) individuals and patients with mild cognitive impairment (MCI) from three independent longitudinal cohorts (BioFINDER-1, BioFINDER-2, WRAP). We put emphasis on the effects of APOE4 on longitudinal tau measures, which allowed us to adjust for the fact that different participants are at different disease stages at baseline.

Materials and methods

Participants

The study participants of the first two cohorts belonged to the ongoing prospective Swedish BioFINDER study, namely BioFINDER-1 (NCT01208675) and BioFINDER-2 (NCT03174938), as previously described [20–22] (www.biofinder.se). All participants were recruited at the memory clinics of Skåne University Hospital and the Hospital of Ängelholm, Sweden. We included CU individuals and patients with MCI in the present study, for whom enrollment and follow-up visits were conducted between 2007 and 2015 for BioFINDER-1 and from 2017 to 2023 for BioFINDER-2. Briefly, participants included were 40 to 100 years old, had a Mini-Mental State Examination (MMSE) between 27 and 30 or 26 and 30 depending on age for cognitively unimpaired (CU) participants and between 24 and 30 for MCI, and were proficient in Swedish such that an interpreter was not necessary for cognitive testing. Exclusion criteria included severe somatic disease, current alcohol/substance misuse, refusing lumbar puncture or neuroimaging. Classification of cognitively unimpaired (CU) individuals followed the National Institute on Aging-Alzheimer’s Association criteria, i.e. not meeting criteria for Mild Cognitive Impairment (MCI) or dementia [23]. The CU group included cognitively healthy controls or individuals with subjective cognitive decline, who performed within normal ranges on a comprehensive cognitive test battery [24]. In BioFINDER-2, a group of younger controls (20-40 years old, n=41) was also included, which by design enrolled a high proportion of APOE4 carriers (39%). In BioFINDER-2, MCI diagnosis was defined as performing below -1.5[standard deviation in any cognitive domain, employing a regression-based norm that accounted for age and education and z-scores based on Aβ-negative controls [25]. Cognitive domain z-scores were computed by averaging the z-scores of tests related to attention/executive function, verbal ability, memory, and visuospatial function. In BioFINDER-1, MCI determination was made by a senior neuropsychologist following comprehensive neuropsychological testing, as detailed elsewhere [26].

The Wisconsin Registry for Alzheimer’s Prevention (WRAP) cohort includes participants who were cognitively normal at enrollment, recruited from the population, enriched for parental history of AD and were between 40 and 65 years at baseline. Participants underwent neuropsychological evaluations on a biennial basis. More information about the recruitment of these participants can be found in [27]. For the present study, we refer to two different samples (“WRAP PET” and “WRAP plasma”) which were analyzed separately and share some participants, but not all as not everyone had both tau-PET and plasma measurements. For this analysis, baseline was considered as time of first tau-PET or plasma p-tau217 measurement, and by then a small proportion of participants had progressed to MCI. Cognitive status at each visit was determined by a consensus diagnosis and diagnosis of MCI was assigned based on National Institute on Aging-Alzheimer’s Association criteria without reference to biomarkers [28]. Complete inclusion and exclusion criteria are described elsewhere [29]. The WRAP PET sample included all participants who had completed Aβ- and tau-PET and T1-weighted MRI scans by April 2023. The WRAP plasma sample included participants who had an Aβ-PET and plasma p-tau217 levels, which was measured from visits that took place between August 2011 and January 2020.

All participants provided written informed consent prior to participation in the studies. The BioFINDER-studies received ethical approval from the Regional Ethical Committee in Lund, Sweden and the University of Wisconsin Institutional Review Board approved all WRAP study procedures. All research was completed in accordance with the Declaration of Helsinki.

Image acquisition and processing

In the BioFINDER cohorts, PET images were were acquired as 4×5 min time frames on digital GE Discovery MI scanners. Aβ PET was done 90-110 minutes after the injection of ∼185 MBq [ F]flutemetamol and tau PET was done 70-90 minutes post injection of ∼370 MBq [18F]RO948. Briefly, images were motion corrected, averaged and co-registered to the closest T1 MPRAGE scan acquired on a Siemens MAGNETOM Prisma 3T MRI scanner at 1×1×1 mm isometric resolution. MRI scanning parameters have been described in detail previously. Cortical parcellation of the MRIs were performed using FreeSurfer, version 6.0. Standardized uptake value ratios (SUVRs) were then calculated using the cerebellum as reference region for Aβ and the inferior cerebellar cortex reference region for tau [30]. For the analysis, Aβ-PET measures were considered both as continuous SUVR and as binarized data using a cutoff derived from mixture modeling in the BioFINDER-2 cohort (1.03 SUVR) [31]. A neocortical meta-region of interest (ROI) for Aβ-PET (prefrontal, lateral temporal, parietal, anterior cingulate, and posterior cingulate/precuneus) was calculated [31]. For tau-PET, the regions of interest were (1) a temporal meta-ROI (bilateral entorhinal cortex, inferior and middle temporal cortices, fusiform gyrus, parahippocampal cortex, and amygdala) [30] and (2) the entorhinal cortex, given its previous associations with the APOE4 allele [32–34].

In the WRAP PET cohort, all participants underwent Aβ- ([11C]-PiB; Pittsburgh compound B) and tau- ([18F]-MK-6240) PET imaging at the University of Wisconsin-Madison. Details regarding radiopharmaceutical production, acquisition protocols, as well as image reconstruction, processing and quantification of [11C]-PiB and [18F]-MK-6240 PET data have been described previously [35]. Aβ burden was assessed as a global average [11C]-PiB distribution volume ratio (DVR; Logan graphical analysis of the dynamic 0-70 minute acquisition using the cerebellum grey matter as the reference region), taken across eight bilateral cortical ROIs. A previously established threshold of DVR > 1.2 was used to define [11C]-PiB positivity [36] which corresponded to a centiloid of 21.6. For reporting, DVR values were linearly translated to centiloids [29]. Tau burden on the [18F]-MK-6240 tau-PET scan was quantified using SUVRs (70-90 min, inferior cerebellar grey matter reference region), in brain regions corresponding to the Harvard Oxford atlas. A temporal meta-ROI encompassing the amygdala, parahippocampal gyrus, inferior and middle temporal gyri was calculated, as well as the parahippocampal gyrus (corresponding to the entorhinal cortex in this atlas).

CSF and plasma measurements

CSF samples in BioFINDER-1 were collected and handled according to a standardized protocol [37]. Samples were taken from non-fasting individuals by lumbar puncture at 3 different centers [38]. The CSF was centrifuged (2000g, +4°C, 10 minutes), aliquoted into 1 mL polypropylene tubes (Sarstedt AG & Co, Nümbrecht, Germany), and stored at −80°C. Samples went through 1 freeze-thaw cycle before the analysis when 200 μL was further aliquoted into LoBind tubes (Eppendorf Nordic A/S, Hørsholm, Denmark). Plasma samples from WRAP plasma were collected in EDTA tubes (BD 366643; Franklin Lakes, New Jersey, USA). Samples were mixed gently by inverting 10–12 times and were centrifuged 15 min at 2000g at room temperature within one hour of collection. Plasma samples were aliquoted into 2 mL cryovials (Wheaton Cryoelite W985863; Millville, New Jersey, USA). Aliquoted plasma was frozen at −80°C within 90 min and stored. CSF and plasma p-tau217 were measured using a Mesoscale Discovery immunoassay developed by Lilly Research Laboratories at Lund University. Samples were analyzed as previously described [39]. Briefly, samples were assayed in duplicates with biotinylated-IBA493 used as a capture antibody and SULFO-TAG-4G10-E2 as the detector. The assay was calibrated with a synthetic p-tau217 peptide.

Statistics

Statistical analysis and data visualization were performed with R software version 4.2.3. P values <0.05 were considered significant. Participants were dichotomized in APOE4 carriers and non-carriers based on whether they had at least one □4allele or no □4allele, respectively. In complementary analyses, we showed similar results based on the number of □4alleles (0 as non-carriers, 1 as APOE4 heterozygotes, 2 as APOE4 homozygotes) instead of the binary APOE4 status, with a main focus on BioFINDER-2 given that its large sample size allowed to repeat all analyses. Separate linear regression models were used to determine the associations between baseline tau-PET and CSF/plasma p-tau217 measures with APOE4, Aβ and the interaction between APOE4 and Aβ measures at baseline. Similarly, linear mixed-effect models were used to assess associations between longitudinal tau measures and the same parameters (APOE4, Aβ, and their interaction), using random slope and intercept. The “time” variable was calculated in years based on the dates when tau follow-up measures were performed. IDs that had missing data in one of the variables included in the lm and lme models were removed. Overall, we constructed four linear models for cross-sectional data and linear mixed-effect for longitudinal tau data as follows: 1) with only APOE4 as predictor; 2) with only baseline Aβ PET as predictor; 3) with APOE4 and baseline Aβ PET as predictors; 4) with the interaction APOE4 x baseline Aβ PET as predictor. All continuous variables were normalized using the “scale()” function in R prior to this analysis. Aβ was used in continuous measures in all statistical models and was dichotomized only for visual representation. Lastly, mediation analysis was used to investigate the indirect effect of APOE4 on tau variables through Aβ mediation. The analysis was conducted using the “mediation” package in R, using bootstrapping with 1000 resamples to obtain bias-corrected confidence intervals. All analyses were adjusted for age, sex, and cognitive status. For cross-sectional analyses, group differences were assessed with ANOVA, with Tukey’s Honest Significant Difference (HSD) post-hoc test for pairwise group comparisons.

Data Availability

Pseudonymized BioFINDER data can be shared to qualified academic researchers after request for the purpose of replicating procedures and results presented in the study. In line with the EU General Data Protection Regulation (GDPR) legislation, a data transfer agreement must be established to share data. The agreement will include terms of how data is stored, protected, and accessed, and will define what the receiver can or cannot do. The agreement also must be approved by the Swedish Ethical Review Authority and Region Skåne. The data from the Wisconsin Registry for Alzheimer’s Prevention can be requested through online submission processes.

Results

Study cohorts and demographics

The study included a total of 1819 subjects, of which 1133 had longitudinal measurements. Samples came from three different cohorts, in which longitudinal measures of tau PET or soluble p-tau217 were available: the Swedish BioFINDER-2 study (baseline tau-PET, n=1048 with baseline tau PET; longitudinal tau-PET, n=629), the Swedish BioFINDER-1 study (baseline CSF p-tau217, n=237; longitudinal p-tau217, n=178), and WRAP, here subdivided in a “WRAP PET” cohort (baseline tau-PET, n=403; longitudinal tau-PET, n=199) and a “WRAP plasma” cohort (baseline plasma p-tau217, n=131; longitudinal p-tau217, n=127; Table 1, Supplementary table 1-2). In the BioFINDER studies, 31.9-40.9% of patients had MCI while the WRAP cohort was composed almost entirely of CU participants. APOE4 carriers accounted for 40.2-50% of the cohorts. APOE4 heterozygotes represented between 33 and 42% of the sample in the different cohorts, and APOE4 homozygotes between 5 and 8% (Table 1). The detailed APOE genotypes across cohorts are listed in Supplementary Table 1. Across all cohorts, the Aβ-positive participants had elevated tau levels at baseline, but there was no difference in tau levels between the APOE4 carriers or non-carriers within the Aβ-positive or negative groups (Supplementary Figure 1). The longitudinal subsets were followed up for an average time of 2.6-5.2 years, depending on the cohort. All demographic characteristics are detailed in Table 1, and in Supplementary Tables 2-3 where participants are broken down by APOE4 status and cognitive status.

View this table:
Table 1. Characteristics of the BioFINDER and WRAP cohorts

APOE4 effects on deposition and accumulation of insoluble tau are explained by the presence of Aβ

We first evaluated whether APOE4 carriership was associated with the deposition (cross-sectional) and accumulation (longitudinal) of insoluble tau measured with tau-PET in BioFINDER-2 and WRAP PET, and, if present, whether such associations were influenced by the levels of Aβ pathology.

Looking at cross-sectional tau-PET in the temporal meta-ROI, in both cohorts, APOE4 carriership was only associated with higher tau-PET burden in models that did not include other main predictors. In all models that included Aβ-PET load, Aβ was always associated with tau-PET levels, the effect of APOE4 was no longer significant and there were no interactions between Aβ and APOE4 (Table 2, Figure 1a-c). Further, in both cohorts the model explaining the highest variance was the one including only Aβ-PET (R2=0.364 in BioFINDER-2 and 0.333 in WRAP) and there were no improvements of the models by adding APOE4 as covariate or the interaction between Aβ and APOE4 (Table 2). Results focusing on tau-PET in the entorhinal cortex were different: in BioFINDER-2, there was a significant interaction between APOE4 carriership and Aβ (b=0.166, p<0.001), such that for a given level of Aβ, APOE4 carriers had higher tau-PET burden in the entorhinal cortex than non-carriers (Figure 1b and Supplementary Table 4). However, such an interaction was not present in WRAP PET, where results were similar as those in the temporal meta-ROI (Supplementary Table 4).

Figure 1.
Figure 1. APOE4 and Aβ interaction on cross-sectional and longitudinal insoluble tau (tau PET SUVR) in BioFINDER-2 and WRAP PET

a-c) Scatter plot for visualization of the relationship between tau-PET SUVR and Aβ-PET SUVR at baseline divided by APOE □4status in BioFINDER-2 (a-b) and WRAP PET (c). d-f) Longitudinal tau-PET SUVR, groups divided by baseline Aβ PET status combined with APOE4 genotype in BioFINDER-2 (c-d) and WRAP PET (d). For visualization purposes, we created 4 groups based on Aβ status and APOE □4status, but continuous Aβ-PET SUVR was used in the statistical models. All statistical results are reported in Table 2.

View this table:
Table 2. Effects of APOE4 and Aβ on cross-sectional and longitudinal insoluble tau (tau-PET SUVR)

When using longitudinal tau-PET in the temporal meta-ROI as the outcome (BioFINDER-2 and WRAP PET), APOE4 carriership was significantly associated with faster accumulation insoluble tau aggregates only in BioFINDER-2, in model in which APOE4 genotype was the main predictor (Table 2). Conversely, Aβ load at baseline was a significant predictor of insoluble tau accumulation in all models where it was included (b=0.064-0.097, p<0.001). The models including only Aβ as predictor had the best model fit (represented by the lowest Akaike information criterion, Table 2) in both cohorts. Similar to the cross-sectional results, there was no interaction effect of APOE x Aβ x time on insoluble tau accumulation in any of the cohorts (p=0.642 and 0.996, Table 2). We additionally compared the longitudinal trajectories of tau accumulation between 4 groups defined by baseline Aβ-PET status combined with APOE4 genotype (Aβ-□4-, Aβ-□4+, Aβ+□4-, Aβ+□4+; Figure 1c-e). Using Aβ-□4-participants as the reference group, the only significant differences in trajectories were between the Aβ+ and Aβ-groups (p<0.001). There were no differences in trajectories between the Aβ-□4- and the Aβ-□4+ groups or between the Aβ+□4+ and the Aβ+□4-groups across the two cohorts (p=0.171-0.683; Figure 1c-e). Looking at associations with longitudinal tau-PET uptake in the entorhinal cortex showed similar results than the temporal meta-ROI, with a main effect of Aβ on longitudinal tau-PET, and no significant effect of APOE4 in models including both Aβ and the genotype both in BioFINDER-2 and WRAP PET (Supplementary Table 4).

Mediation analyses showed that Aβ was fully mediating the relationship between APOE4 and baseline tau-PET levels in the temporal meta-ROI (93-99% of the total effect) as well as between APOE4 and longitudinal change in tau-PET (48-95% of the total effect) in both cohorts (p=0.203-0.949 for direct effect of APOE4, Supplementary Figure 2a-d). On cross-sectional levels of tau-PET in entorhinal cortex, Aβ had a partial mediating effect in BioFINDER-2 (p=0.019 for the direct effect of APOE4), but a fully mediating effect on longitudinal levels (86%, p<0.001, Supplementary Figure 3).

Soluble p-tau217 levels are differentially moderated by APOE4 at baseline and longitudinally

Next, we repeated similar analyses assessing if APOE4 carriership affected the levels and accumulation of tau, focusing on soluble p-tau 217 in CSF (BioFINDER-1) or plasma (WRAP) as outcome.

In BioFINDER-1, APOE4 was associated with higher baseline p-tau217 levels when it was the only main predictor (b=0.468, p<0.001, Table 3), but had no significant effect when Aβ-PET was present in the model. However, Aβ-PET was a significant predictor of p-tau217 in all models (b=0.612-0.679, p<0.001, Table 3), and there were no interactions between APOE4 and Aβ-PET on p-tau217 levels (Figure 2a). In the WRAP plasma cohort, unlike all other cross-sectional results, APOE4 was not associated with p-tau217 levels when it was the sole predictor, and there was a significant interaction between APOE4 and Aβ-PET (b=-0.258, p=0.030), such that at a given level of Aβ-PET, APOE4 non-carriers had higher p-tau217 levels (Table 3, Fig. 2b). However, this interaction was no longer significant when removing an extreme outlier (an APOE4 non-carrier who had the highest Aβ- and tau-PET levels [b=-0.143, p=0.285]). Aβ-PET was associated to p-tau217 in all models where it was included (p<0.001, Table 3).

Figure 2.
Figure 2. APOE4 and Aβ interaction on cross-sectional and longitudinal soluble p-tau217 in BioFINDER-1 and WRAP plasma

a-b) Scatter plot for visualization of the relationship between CSF (a) and plasma (b) p-tau217 and Aβ-PET SUVR at baseline divided by APOE4 status in BioFINDER-1 (a) and WRAP plasma (b). c-d) Longitudinal CSF (c) and plasma (d) p-tau217, groups divided by baseline Aβ-PET status combined with APOE4 genotype in BioFINDER-1 (c) and WRAP plasma (d). For visualization purposes, we created 4 groups based on Aβ status and APOE4 status, but continuous Aβ-PET SUVR was used in the statistical models. All statistical results are reported in Table 3.

View this table:
Table 3. Effects of APOE4 and Aβ on cross-sectional and longitudinal soluble p-tau217

When using longitudinal soluble p-tau217 levels as the outcome, the only significant association with longitudinal p-tau217 in BioFINDER-1 was with baseline Aβ-PET SUVR in the model where it was the main predictor (b=0.026, p=0.024, Table 3). When APOE4 alone or both APOE4 and Aβ-PET were present in the model, none of the longitudinal associations were significant, but the effect of Aβ-PET SUVR was at trend level (p=0.072, Table 3). Of note, if using instead dichotomous Aβ-PET status (+/-) as the predictor, Aβ-PET was associated with p-tau217 in all models where it was included (b=0.032-0.038, p<0.001-0.002, Supplementary Table 5). In WRAP, APOE4 was associated with longitudinal p-tau217 only when it was the main predictor (b=0.058, p=0.009, Table 3). Aβ-PET was associated to p-tau217 in all models where it was included (p<0.001, Table 3). In both cohorts, there were no interactions between APOE4 and Aβ-PET load on p-tau217 levels over time. The longitudinal trajectories of soluble p-tau217 in the 4 groups defined by dichotomous Aβ-PET status (+/-) combined with APOE4 genotype showed no difference between the Aβ-□4- and the Aβ-□4+ group or between the Aβ+□4+ and Aβ+□4-group (p=0.585-0.673 in BioFINDER-1; p=0.174-0.377 in WRAP, Figure 2c-d).

Mediation analyses showed that baseline Aβ-PET fully mediated the association between APOE4 and cross-sectional p-tau217 levels in both cohorts and with longitudinal p-tau217 in the BioFINDER-1 cohort (110-112% of the total effect; Supplementary Figure 4a-c). However, APOE4 had a significant direct effect on longitudinal p-tau217 in WRAP (b=0.008, p=0.034), and Aβ-PET had a partial mediating effect (b=0.007, p=0.006, 46% of mediation Supplementary Figure 4d).

Differential associations based on number of APOE4 alleles in BioFINDER-2

Lastly, we investigated similar associations based on the number of APOE4 alleles (0, 1 or 2) instead of the dichotomous APOE4 carriership. Such analyses were focussed on BioFINDER-2 given the largest sample size, which allowed to have 83 APOE4 homozygotes. Looking at cross-sectional association with tau-PET, results were different between the temporal meta-ROI and the entorhinal cortex. In the former, there was a significant interaction between Aβ-PET and the APOE4 homozygotes compared to non-carriers or APOE4 heterozygotes (Fig. 3a, Supplementary Table 6). The interaction between Aβ-PET and the APOE4 heterozygotes and non-carriers was not significant (Supplementary Table 6). Overall, for a given level of Aβ, only the APOE4 homozygotes had higher tau burden compared to the two other groups. In the entorhinal cortex, the interaction between Aβ-PET and the APOE4 was significant for all APOE4 comparisons (Fig. 3b, Supplementary Table 6). Looking at associations with longitudinal tau-PET in the same regions, in the temporal meta-ROI, results were consistent with the interactions seen at the cross-sectional level: the APOE4 homozygotes accumulated faster tau compared to the two other groups for a given level of Aβ (APOE4 homozygotes*Aβ*time compared to non-carriers or heterozygotes with p=0.036 and p=0.002, Figure 3c, Supplementary Table 6). In the entorhinal cortex, there was not any significant associations with the genotypes, neither APOE4 groups*time or APOE4 groups*Aβ*time across the different models. The main effect on longitudinal tau uptake in the entorhinal cortex was higher levels of Aβ at baseline (Aβ*time: p<0.001, Figure 3d, Supplementary Table 6). All other cohorts had a smaller sample size and thus were not investigated in detail beyond APOE4 dichotomization. Still, in all cross-sectional measures, none of the other cohorts showed significant interactions between the three APOE4 groups (Supplementary Figure 5).

Figure 3.
Figure 3. Interaction between APOE4 groups based on the number of ε4 alleles and Aβ on cross-sectional and longitudinal insoluble tau-PET in BioFINDER-2

a-b) Scatter plot for visualization of the relationship between tau-PET SUVR (temporal meta-ROI in a and entorhinal cortex in b) and Aβ-PET SUVR split between three APOE4 groups. 0 corresponds to non-carriers, 1 corresponds to APOE4 heterozygotes; 2 corresponds to APOE4 homozygotes c-d) Longitudinal tau-PET SUVR, groups divided by baseline Aβ PET status combined with the three different APOE4 groups. For visualization purposes, we created 6 groups based on Aβ status and number of □4 alleles, but continuous Aβ-PET SUVR was used in the statistical models. All statistical results are reported in Supplementary Table 5.

Discussion

Our study aimed to investigate the interplay of APOE4 genotype and Aβ on both baseline tau burden and longitudinal accumulation in the context of the AD continuum. Our results, which were very consistent in three cohorts across cross-sectional and longitudinal tau-PET or soluble p-tau measures, challenge the hypothesis that APOE4 significantly influences tau accumulation in the temporal meta-ROI independently of Aβ load. In cross-sectional analyses, in the largest cohort we observed a significant interaction between Aβ and APOE4 carriership on tau-PET uptake in the medial temporal lobe. In longitudinal analyses, APOE4 carriership was not a significant predictor of accumulation of insoluble tau aggregates or soluble p-tau217 when Aβ load was also included as a predictor in the models across all cohorts. The tau accumulation trajectories of the Aβ+/ε4+ for group was not significantly steeper than the ones of the Aβ+/ε4-group. No major, direct role of APOE4 on tau was further supported by the mediation analyses, which showed that in almost all cases, Aβ fully mediated the effects of APOE on tau deposition and accumulation. Still, we should mention that when looking at groups split based on the number of APOE4 alleles in the largest cohort, the APOE4 homozygotes showed a significant interaction with Aβ on tau-PET in the temporal meta-ROI, both at the cross-sectional and longitudinal level, which might point to different associations in this population particularly at risk of AD. Taken together, these findings suggest that the main way APOE4 carriership might affect tau accumulation is through Aβ load, both at baseline and over time.

The current findings converge with those of previous studies where the effect of the ε4 allele on the baseline load or the accumulation (i.e., change in tau-PET) of insoluble tau in the neocortical temporal regions were entirely mediated by Aβ load [16, 40]. Likewise, another study has shown that ε4 carriers did not have higher rates of accumulation of insoluble tau aggregates [17, 41]. A recent study on plasma NTA tau (a N-terminal soluble tau fragment associated to AD) showed that ε4 carriers do not have higher plasma tau concentrations compared to non-carriers [42]. However, results differ slightly when looking at insoluble tau in the medial temporal lobe. In BioFINDER-2, the largest cohort included in this paper, we found an interaction such that for a given level of Aβ, APOE carriers had higher tau levels than non-carriers. Such a result is also in line with previous studies done across the AD continuum [11, 32, 33]. We did not see this interaction in the WRAP cohort, probably due to the lower tau levels in that cohort, composed of mostly cognitively normal older adults. Still, we are cautious not to infer Aβ-independent pathway through which APOE4 affects insoluble tau accumulation [10, 15], given that we did not see an Aβ-independent over time, nor when looking beyond APOE4 dichotomization. The two previous studies suggesting an APOE4 effect independent of Aβ have several differences from ours, including smaller sample sizes and the use of different statistical methods (dichotomization of tau and Aβ measures, logistic regression). Dichotomization of Aβ, specifically, might mask the fact that the APOE4 carriers might have higher amounts of Aβ pathology since the aggregation of Aβ starts at an earlier age, which is an important consideration in interpreting the literature.

In BioFINDER-2, when investigating three groups based on the number of APOE4 alleles, the different associations seen in the temporal meta-ROI compared to the medial temporal lobe (an early region of tau deposition) might give credence to the recent idea that APOE4 homozygotes are a “distinct genetic form of AD” [5]. In the temporal lobe composite region, both at baseline and over time, the APOE4 homozygotes diverged both from heterozygotes and non-carriers, while the two latter groups behaved similarly. For the same level of Aβ, only the homozygotes showed higher tau aggregates or faster accumulation of aggregates, potentially hinting at more advanced pathology specifically in this genetic form. On the other hand, in the medial temporal lobe, at baseline, we observed a more “dose-dependent” effect of the number of APOE4 interacting with Aβ. However, over time, the larger effect on accumulation of tau aggregates in this early tau region was higher levels of Aβ at baseline, and not APOE genotype. We postulate that the “dose-dependent” effect on baseline levels might rather represent individuals being at different stages of the disease.

The current results have important implication for therapeutics targeting APOE [19], in such that these therapies might likely affect the rate of Aβ aggregation, but not directly tau aggregation. We thus speculate that APOE therapies might work better if initiated during preclinical AD stages to stop or reduce the accumulation of Aβ, before Aβ has become widespread and drives tau accumulation in the neocortex. Supporting this notion, a recent study based on the ADNI cohort found that APOE4 was an important factor in the progression from A-T- to A+T-, but only had a marginal effect in the progression from A+T- to A+T+ [43]. Consistent with these findings on tau, a study that examined effects of APOE4 on cognition also did not find any effects of APOE4 on cognition after accounting for Aβ load [44].

Here we provided a comprehensive assessment of tau dynamics based on a large sample size, longitudinal data, three independent cohorts and overall consistent results across different tau-PET tracers and soluble p-tau217 measures (both CSF and plasma). We acknowledge that the cohorts studied here were on average younger and were enriched by design to include a higher proportion of APOE4 carriers, which might explain part of the differences with previous literature. The current results based on the three APOE4 groups are also derived from only one cohort, and it will be interesting to see if similar associations are present in other large samples. The observational nature of the study does not allow to establish direct causation of our findings. Almost all participants included in the current study were White, and the Aβ-dependent or independent effect of APOE4 on tau will be important to study in ethnically diverse populations, particularly given the different impact of APOE on AD pathology in different ethnicities [45].

In summary, our study sheds light on the intricate relationships between APOE4, Aβ, and tau in AD. In showing that overall Aβ load surpasses the impact of APOE4 on tau accumulation, our findings strengthen the links between APOE4 and Aβ and further investigating the genetic and molecular pathways underlying such links might help to find therapeutics to postpone the disease onset. We found that most effect of APOE4 on tau accumulation in humans is mediated by Aβ load, with potential differences in APOE4 homozygotes, which will need to be investigated further in other large samples.

Funding

Work at the authors’ research center was supported by European Research Council (ADG-101096455), Alzheimer’s Association (ZEN24-1069572, SG-23-1061717), GHR Foundation, Swedish Research Council (2022-00775, 2021-02219, 2018-02052), ERA PerMed (ERAPERMED2021-184), Knut and Alice Wallenberg foundation (2022-0231), Strategic Research Area MultiPark (Multidisciplinary Research in Parkinson’s disease) at Lund University, Swedish Alzheimer Foundation (AF-980907, AF-994229, AF-939981, AF-994075), Swedish Brain Foundation (FO2021-0293, FO2023-0163, FO2024-0284), Parkinson foundation of Sweden (1412/22), Cure Alzheimer’s fund, Rönström Family Foundation, WASP and DDLS Joint call for research projects (WASP/DDLS22-066), Konung Gustaf V:s och Drottning Victorias Frimurarestiftelse, Skåne University Hospital Foundation (2020-O000028), Regionalt Forskningsstöd (2022-1259, 2021-1013) and Swedish federal government under the ALF agreement (2022-Projekt0080, 2022-Projekt0107). The precursor of 18F-flutemetamol was sponsored by GE Healthcare. The precursor of 18F-RO948 was provided by Roche. G.S. received funding from the European Union’s Horizon 2020 Research and Innovation Program under Marie Sklodowska-Curie action grant agreement number 101061836, an Alzheimer’s Association Research Fellowship (AARF-22-972612), the Alzheimerfonden (AF-980942), Greta och Johan Kocks research grants and travel grants from the Strategic Research Area MultiPark (Multidisciplinary Research in Parkinson’s Disease) at Lund University. The data contributed from the WRAP study were supported by grants from the U.S. National Institutes of Health AG027161 and AG021155.The funding sources had no role in the design and conduct of the study; in the collection, analysis, interpretation of the data; or in the preparation, review, or approval of the manuscript.

Competing Interests

OH has acquired research support (for the institution) from AVID Radiopharmaceuticals, Biogen, C2N Diagnostics, Eli Lilly, Eisai, Fujirebio, GE Healthcare, and Roche. In the past 2 years, he has received consultancy/speaker fees from Alzpath, BioArctic, Biogen, Bristol Meyer Squibb, Eisai, Eli Lilly, Fujirebio, Merck, Novartis, Novo Nordisk, Roche, Sanofi and Siemens. SP has acquired research support (for the institution) from ki elements / ADDF and Avid. In the past 2 years, he has received consultancy/speaker fees from Bioartic, Biogen, Esai, Lilly, and Roche. RS has received a speaker fee from Roche. NMC has a consultancy agreement with Biogen. SCJ has consultancy agreements with Enigma Biomedical and with AlzPath. All other authors report no competing interests.

References

  1. 1.↵
    Corder, E.H., et al., Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science, 1993. 261(5123): p. 921–3.
    OpenUrlAbstract/FREE Full Text
  2. 2.
    Farrer, L.A., et al., Effects of age, sex, and ethnicity on the association between apolipoprotein E genotype and Alzheimer disease. A meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium. Jama, 1997. 278(16): p. 1349–56.
    OpenUrl
  3. 3.
    Genin, E., et al., APOE and Alzheimer disease: a major gene with semi-dominant inheritance. Mol Psychiatry, 2011. 16(9): p. 903–7.
    OpenUrlCrossRefPubMedWeb of Science
  4. 4.↵
    Lambert, J.C., et al., Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet, 2013. 45(12): p. 1452–8.
    OpenUrlCrossRefPubMed
  5. 5.↵
    Fortea, J., et al., APOE4 homozygozity represents a distinct genetic form of Alzheimer’s disease. Nat Med, 2024. 30(5): p. 1284–1291.
    OpenUrl
  6. 6.
    Jansen, W.J., et al., Prevalence Estimates of Amyloid Abnormality Across the Alzheimer Disease Clinical Spectrum. JAMA Neurol, 2022. 79(3): p. 228–243.
    OpenUrl
  7. 7.
    Belloy, M.E., et al., APOE Genotype and Alzheimer Disease Risk Across Age, Sex, and Population Ancestry. JAMA Neurol, 2023. 80(12): p. 1284–1294.
    OpenUrl
  8. 8.↵
    Betthauser, T.J., et al., Multi-method investigation of factors influencing amyloid onset and impairment in three cohorts. Brain, 2022. 145(11): p. 4065–4079.
    OpenUrlCrossRef
  9. 9.↵
    Mattsson-Carlgren, N., et al., Aβ deposition is associated with increases in soluble and phosphorylated tau that precede a positive Tau PET in Alzheimer’s disease. Sci Adv, 2020. 6(16): p. eaaz2387.
    OpenUrlFREE Full Text
  10. 10.↵
    Ferrari-Souza, J.P., et al., APOEε4 potentiates amyloid β effects on longitudinal tau pathology. Nat Aging, 2023. 3(10): p. 1210–1218.
    OpenUrl
  11. 11.↵
    Therriault, J., et al., APOEε4 potentiates the relationship between amyloid-β and tau pathologies. Mol Psychiatry, 2021. 26(10): p. 5977–5988.
    OpenUrl
  12. 12.
    Jack, C.R., et al., Predicting future rates of tau accumulation on PET. Brain, 2020. 143(10): p. 3136–3150.
    OpenUrlCrossRefPubMed
  13. 13.
    Baek, M.S., et al., Effect of APOE ε4 genotype on amyloid-β and tau accumulation in Alzheimer’s disease. Alzheimers Res Ther, 2020. 12(1): p. 140.
    OpenUrl
  14. 14.
    Koychev, I., et al., Prediction of rapid amyloid and phosphorylated[Tau accumulation in cognitively healthy individuals. Alzheimers Dement (Amst), 2020. 12(1): p. e12019.
    OpenUrl
  15. 15.↵
    Steward, A., et al., ApoE4 and Connectivity-Mediated Spreading of Tau Pathology at Lower Amyloid Levels. JAMA Neurol, 2023.
  16. 16.↵
    Salvadó, G., et al., Differential associations of APOE-ε2 and APOE-ε4 alleles with PET-measured amyloid-β and tau deposition in older individuals without dementia. Eur J Nucl Med Mol Imaging, 2021. 48(7): p. 2212–2224.
    OpenUrl
  17. 17.↵
    Ghisays, V., et al., Brain imaging measurements of fibrillar amyloid-β burden, paired helical filament tau burden, and atrophy in cognitively unimpaired persons with two, one, and no copies of the APOE ε4 allele. Alzheimers Dement, 2020. 16(4): p. 598–609.
    OpenUrlCrossRef
  18. 18.↵
    Liu, C.C., et al., Apolipoprotein E and Alzheimer disease: risk, mechanisms and therapy. Nat Rev Neurol, 2013. 9(2): p. 106–18.
    OpenUrlCrossRefPubMed
  19. 19.↵
    Jackson, R.J., B.T. Hyman, and A. Serrano-Pozo, Multifaceted roles of APOE in Alzheimer disease. Nat Rev Neurol, 2024.
  20. 20.↵
    Palmqvist, S., et al., Cognitive effects of Lewy body pathology in clinically unimpaired individuals. Nature Medicine, 2023. 29(8): p. 1971–1978.
    OpenUrlCrossRef
  21. 21.
    Palmqvist, S., et al., Prediction of future Alzheimer’s disease dementia using plasma phospho-tau combined with other accessible measures. Nat Med, 2021. 27(6): p. 1034–1042.
    OpenUrl
  22. 22.↵
    Palmqvist, S., et al., Discriminative Accuracy of Plasma Phospho-tau217 for Alzheimer Disease vs Other Neurodegenerative Disorders. JAMA, 2020.
  23. 23.↵
    Jack, C.R., Jr., et al., NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement, 2018. 14(4): p. 535–562.
    OpenUrlCrossRefPubMed
  24. 24.↵
    Palmqvist, S., et al., Cognitive effects of Lewy body pathology in clinically unimpaired individuals. Nat Med, 2023. 29(8): p. 1971–1978.
    OpenUrlCrossRef
  25. 25.↵
    Borland, E., et al., The age-related effect on cognitive performance in cognitively healthy elderly is mainly caused by underlying AD pathology or cerebrovascular lesions: implications for cutoffs regarding cognitive impairment. Alzheimers Res Ther, 2020. 12(1): p. 30.
    OpenUrl
  26. 26.↵
    Petrazzuoli, F., et al., Brief Cognitive Tests Used in Primary Care Cannot Accurately Differentiate Mild Cognitive Impairment from Subjective Cognitive Decline. J Alzheimers Dis, 2020. 75(4): p. 1191–1201.
    OpenUrl
  27. 27.↵
    Johnson, S.C., et al., The Wisconsin Registry for Alzheimer’s Prevention: A review of findings and current directions. Alzheimers Dement (Amst), 2018. 10: p. 130–142.
    OpenUrl
  28. 28.↵
    Albert, M.S., et al., The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement, 2011. 7(3): p. 270–9.
    OpenUrlCrossRefPubMedWeb of Science
  29. 29.↵
    Cody, K.A., et al., Characterizing brain tau and cognitive decline along the amyloid timeline in Alzheimer’s disease. Brain, 2024.
  30. 30.↵
    Pichet Binette, A., et al., Amyloid-associated increases in soluble tau relate to tau aggregation rates and cognitive decline in early Alzheimer’s disease. Nat Commun, 2022. 13(1): p. 6635.
    OpenUrlCrossRefPubMed
  31. 31.↵
    Ossenkoppele, R., et al., Amyloid and tau PET-positive cognitively unimpaired individuals are at high risk for future cognitive decline. Nat Med, 2022. 28(11): p. 2381–2387.
    OpenUrlCrossRefPubMed
  32. 32.↵
    La Joie, R., et al., Association of APOE4 and Clinical Variability in Alzheimer Disease With the Pattern of Tau- and Amyloid-PET. Neurology, 2021. 96(5): p. e650–e661.
    OpenUrlCrossRef
  33. 33.↵
    Salvado, G., et al., Differential associations of APOE-epsilon2 and APOE-epsilon4 alleles with PET-measured amyloid-beta and tau deposition in older individuals without dementia. Eur J Nucl Med Mol Imaging, 2021. 48(7): p. 2212–2224.
    OpenUrl
  34. 34.↵
    Therriault, J., et al., Association of Apolipoprotein E epsilon4 With Medial Temporal Tau Independent of Amyloid-beta. JAMA Neurol, 2020. 77(4): p. 470–479.
    OpenUrl
  35. 35.↵
    Betthauser, T.J., et al., In Vivo Characterization and Quantification of Neurofibrillary Tau PET Radioligand (18)F-MK-6240 in Humans from Alzheimer Disease Dementia to Young Controls. J Nucl Med, 2019. 60(1): p. 93–99.
    OpenUrlAbstract/FREE Full Text
  36. 36.↵
    Mueller, K.D., et al., Proper names from story recall are associated with beta-amyloid in cognitively unimpaired adults at risk for Alzheimer’s disease. Cortex, 2020. 131: p. 137–150.
    OpenUrlCrossRef
  37. 37.↵
    Palmqvist, S., et al., Accuracy of brain amyloid detection in clinical practice using cerebrospinal fluid beta-amyloid 42: a cross-validation study against amyloid positron emission tomography. JAMA Neurol, 2014. 71(10): p. 1282–9.
    OpenUrl
  38. 38.↵
    Blennow, K., et al., Cerebrospinal fluid and plasma biomarkers in Alzheimer disease. Nat Rev Neurol, 2010. 6(3): p. 131–44.
    OpenUrlCrossRefPubMedWeb of Science
  39. 39.↵
    Janelidze, S., et al., Detecting amyloid positivity in early Alzheimer’s disease using combinations of plasma Aβ42/Aβ40 and p-tau. Alzheimers Dement, 2022. 18(2): p. 283–293.
    OpenUrlCrossRef
  40. 40.↵
    Young, C.B., et al., APOE effects on regional tau in preclinical Alzheimer’s disease. Mol Neurodegener, 2023. 18(1): p. 1.
    OpenUrlCrossRef
  41. 41.↵
    Smith, R., et al., The accumulation rate of tau aggregates is higher in females and younger amyloid-positive subjects. Brain, 2020. 143(12): p. 3805–3815.
    OpenUrlCrossRef
  42. 42.↵
    Koivumäki, M., et al., Blood biomarkers of neurodegeneration associate differently with amyloid deposition, medial temporal atrophy, and cerebrovascular changes in APOE ε4-enriched cognitively unimpaired elderly. Alzheimers Res Ther, 2024. 16(1): p. 112.
    OpenUrl
  43. 43.↵
    Altmann, A., et al., Towards cascading genetic risk in Alzheimer’s disease. Brain, 2024.
  44. 44.↵
    Insel, P.S., O. Hansson, and N. Mattsson-Carlgren, Association Between Apolipoprotein E epsilon2 vs epsilon4, Age, and beta-Amyloid in Adults Without Cognitive Impairment. JAMA Neurol, 2021. 78(2): p. 229–235.
    OpenUrl
  45. 45.↵
    Royse, S.K., et al., Apolipoprotein E and Alzheimer’s disease pathology in African American older adults. Neurobiol Aging, 2024. 139: p. 11–19.
    OpenUrl
Back to top
Posted September 23, 2024.
Download PDF

Supplementary Material

Email
APOE4 impact on soluble and insoluble tau pathology is mostly influenced by amyloid-beta
Claudia Cicognola, Gemma Salvadó, Ruben Smith, Sebastian Palmqvist, Erik Stomrud, Tobey Betthauser, Sterling Johnson, Shorena Janelidze, Niklas Mattsson-Carlgren, Oskar Hansson, Alexa Pichet Binette
medRxiv 2024.09.20.24314064; doi: https://doi.org/10.1101/2024.09.20.24314064
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area