Association of Apolipoprotein E ε 4 With Medial Temporal Tau Independent of Amyloid- β

and amyloid-β PET with [ 18 F]florbetapir. Exclusion other head There 489 eligible selected based on availability of amyloid-PET, tau-PET, and genotyping for APOE ε 4 . Forty-five young adults (<30 years) from the TRIAD cohort were not selected for this and in has deleterious consequences in disease beyond its link with amyloid-β and suggest APOE ε 4 as a potential target for future disease-modifying therapeutic trials targeting tau pathology.

O f genetic risk factors for sporadic Alzheimer disease, 1 the apolipoprotein E ε4 (APOEε4) allele is the most well established. The presence of 1 ε4 allele is linked with earlier development of Alzheimer disease, 2 and homozygosity for APOEε4 is associated with onset of Alzheimer disease 10 years earlier compared with non-ε4 carriers. 3 The APOEε4 allele is associated with increased production of amyloid-β 4 as well as with diminished clearance of cerebral amyloid-β compared with ε3 and ε2 alleles. 5, 6 Consequently, individuals with the APOEε4 genotype demonstrate increased cerebral amyloid-β deposition as measured by amyloid positron emission tomography (PET), 7 with amyloid-β positivity beginning earlier in life in APOEε4 carriers than noncarriers. 8 However, the APOEε4 allele has been implicated in numerous other processes independent of amyloid-β in preclinical models of Alzheimer disease, 9,10 including neuroinflammation and neurodegeneration. In humans, the APOEε4 allele is linked with medial temporal hypometabolism in cognitively normal elderly individuals 11 and individuals with Alzheimer disease 12 independently of amyloid-β burden, although the mechanisms underlying the process are not known. Because of the spatiotemporal association between tau aggregation and neurodegeneration, [13][14][15] aggregation of tau pathology presents a potential pathway for the specific patterns of neurodegeneration observed in APOEε4 carriers.
The goal of this study is therefore to determine whether APOEε4 is associated with cerebral tau pathology, independently of age, sex, clinical status, and amyloid-β deposition. Building on previous reports of specific patterns of neurodegeneration in APOEε4 carriers, 11,12,16 we hypothesize that APOEε4 is associated with tau pathology in medial temporal structures.

Participants Translational Biomarkers in Aging and Dementia
The Translational Biomarkers in Aging and Dementia (TRIAD) cohort aims at describing biomarker trajectories and interactions as drivers of dementia. The TRIAD study was launched in 2017 as part of the McGill Centre for Studies in Aging. We assessed cognitively normal participants (n = 124), participants with mild cognitive impairment (MCI) (n = 50), and participants with Alzheimer disease dementia (n = 50) who underwent amyloid-β PET with fluorine 18-labeled [ 18 F] AZD4694, tau PET with [ 18 F]MK6240, structural magnetic resonance imaging, and genotyping for APOEε4. All participants had detailed clinical assessments including Mini-Mental State Examination, Clinical Dementia Rating (CDR), and cerebrovascular disease risk with the Hachinski Ischemic scale. 17 Cognitively normal control individuals had a CDR of 0, participants with MCI had a CDR of 0.5, and participants with Alzheimer disease had a CDR between 1 and 2, in addition to meeting standard diagnostic criteria. 18 Similar to other longitudinal cohort studies of aging and Alzheimer disease, 19 the TRIAD cohort is enriched for APOEε4 carriers. Inclusion cri-teria for all participants are the ability to speak English or French, good general health (no diseases expected to interfere with study participation over time), absence of claustrophobia, and adequate visual and auditory capacities to follow neuropsychologic evaluation. This study's protocol was approved by McGill University's institutional review board, and informed written consent was obtained from each participants. There was no attempt to match cases between cohorts.

Alzheimer's Disease Neuroimaging Initiative
In this study, we assessed cognitively normal individuals (n = 157), individuals with amnestic mild cognitive impairment (n = 83), and individuals with Alzheimer disease (n = 25) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort who underwent amyloid-β PET with [ 18 F]florbetapir, tau PET with [ 18 F]flortaucipir, structural MRI, and genotyping for APOEε4. Cognitively normal control individuals had a CDR of 0, participants with MCI had a CDR of 0.5, and participants with Alzheimer disease had a CDR of 1 or greater in addition to meeting standard diagnostic criteria. 18 The Alzheimer's Disease Neuroimaging Initiative (ADNI) study was approved by the institutional review boards of all of the participating institutions. Informed written consent was obtained from all participants at each site. Full information regarding the ADNI inclusion and exclusion criteria can be accessed at http://adni.loni.usc.edu/.
Genetic Analyses TRIAD Determination of APOE genotypes for patients recruited at Mc-Gill was performed using the polymerase chain reaction amplification technique, followed by restriction enzyme digestion, standard gel resolution and visualization processes. Full details of this procedure can be found elsewhere. 20 ADNI Determination of APOE genotypes for ADNI patients took place at the University of Pennsylvania Alzheimer Disease Biomarker Laboratory. Complete details of genetic methods used in ADNI can be accessed at http://adni.loni.usc.edu/datasamples/clinical-data/.
Positron Emission Tomography Image Acquisition and Processing TRIAD All participants had a T1-weighted MRI that was used for coregistration. Full details of MRI acquisition and processing is described in the eMethods 1 in the Supplement. The PET scans were acquired with a Siemens High Resolution Research Tomograph. The [ 18 F]MK6240 images were acquired 90 to 110 minutes postinjection, and scans were reconstructed with the ordered subset expectation maximization algorithm on a 4-dimensional volume with 4 frames (4 × 300 seconds). 21 The [ 18 F]AZD4694 images were acquired 40 to 70 minutes following injection, and scans were reconstructed with the ordered subset expectation maximization algorithm on a 4-dimensional volume with 3 frames (3 × 600 seconds). 22 Immediately following each PET acquisition, a 6-minute transmission scan was conducted with a rotating cesium 137 point source for attenuation correction. Additionally, the images underwent correction for dead time, decay, and random and scattered coincidences. T1-weighted images were nonuniformity and field-distortion corrected and processed using an in-house pipeline. Then, PET images were automatically registered to the T1-weighted image space, and the T1-weighted images were linearly and nonlinearly registered to the ADNI template space. Subsequently, a PET nonlinear registration was performed using the linear and nonlinear transformations from the T1-weighted image to the ADNI space and the PET to T1-weighted image registration using advanced normalization tools. The PET images were spatially smoothed to achieve a final resolution of 8 mm full width at half maximum. The PET image partial volume correction was carried out using the PETPVC toolbox. 23 Briefly, the region-based voxelwise correction technique was used to perform partial volume correction using 10 tissue priors with a gaussian kernel with a full width at half maximum of 2.4 mm. The [ 18 F]MK6240 standardized uptake value ratio (SUVR) maps were generated using the inferior cerebellar gray matter as a reference region, and [ 18 F]AZD4694 SUVR maps were generated using the cerebellar gray matter as a reference region. A global [ 18 F]AZD4694 SUVR value was estimated for each participant by averaging the SUVR from the precuneus, prefrontal, orbitofrontal, parietal, temporal, anterior, and posterior cingulate cortices. 24 ADNI Full information regarding acquisition and preprocessing of PET data in ADNI is provided at http://adni.loni.usc.edu/datasamples/pet/. Preprocessed PET images downloaded from ADNI underwent spatial normalization to the ADNI standardized space using the transformations of PET native to MRI native space and MRI native to the ADNI space. Partial volume correction was carried out using the PETPVC toolbox 23  In each cohort, we tested whether APOEε4 is associated with tau pathology independently of amyloid-β, sex, or age using voxelwise multivariate linear regression models.
Baseline demographic and clinical data were assessed using t tests and χ 2 tests. Neuroimaging analyses were carried out using the VoxelStats toolbox (https://github.com/sulantha2006/ VoxelStats), a MATLAB-based analytical framework that allows for the execution of multimodal voxelwise neuroimaging analyses. 26 All neuroimaging analyses described in subsequent paragraphs were repeated using partial volume-corrected data. Other statistical analyses were performed using the R Statistical Software Package, version 3.5.3 (the R Foundation). Given the large number of covariates in the statistical models, model diagnostics were carried out using the car package in R to determine the presence of multicollinearity. We computed the variance inflation factor, a measurement of how much variance in regression coefficients are inflated owing to multicollinearity in the statistical models. 27 In the TRIAD cohort, the voxel-based model outlined here was built to test whether main effects between APOEε4 carriership are associated with [ 18 F]MK6240 uptake independently of [ 18 F]AZD4694 uptake. To ensure that the results were not driven by an effect of clinical status (ie higher frequency of APOEε4 carriers in the MCI and Alzheimer disease groups), we adjusted the model for clinical diagnosis. The model was also adjusted for age. Because APOEε4 is associated with amyloid-PET uptake, amyloid-β was included as a covariate in every analysis. Sex was included as a covariate owing to sex differences in entorhinal tau aggregation 28 and stronger associations between APOEε4 and tau in women. 29 Statistical parametric maps were corrected for multiple comparisons using random field theory, 30 with a cluster threshold of P < .001. The analysis was repeated using partial volumecorrected data. In every brain voxel, the model was of the form: Next, we tested the same hypothesis in the ADNI database, examining whether APOEε4 carriership is associated with [ 18 F]flortaucipir uptake independently of [ 18 F]florbetapir uptake. This model was also adjusted for amyloid-β, sex, age, and clinical status. Statistical parametric maps were corrected for multiple comparisons using random field theory, 30 with a cluster threshold of P < .001. The analysis was repeated using partial volumecorrected data. In every brain voxel, the model was of the form: To better understand the association between APOEε4 and medial temporal tau aggregation, we conducted subgroup analyses, stratifying individuals according to the presence of cog-

Research Original Investigation
Association of Apolipoprotein E ε4 With Medial Temporal Tau Independent of Amyloid-β nitive impairment (ie, in cognitively unimpaired individuals and cognitively impaired individuals). The cognitively impaired groups consisted of the individuals with MCI and AD pooled together. These models were adjusted for amyloid-β, sex, and age. The analyses were repeated using partial volumecorrected data.
To derive an estimate of the association between APOEε4 and medial temporal tau-PET SUVR across both cohorts, we used the Metafor package in R. We fit a meta-analytic fixedeffects model using β weights and standard errors for the estimates from each population, analyzed using the rma function. The same process was repeated for gene-dose and voxelbased morphometry analyses described subsequently.
The P value level of significance was .001, and all tests were 2-sided. Exploratory gene-dose analyses, APOEε4-voxelbased morphometry analyses, APOEε4 × age interaction analyses, APOEε4 × amyloid-PET interaction analyses, and APOEε4 unadjusted for amyloid-PET analyses are described in eMethods 2 in the Supplement.

Results
Demographic and clinical information for both samples examined in this study is summarized in Table 1. Demographic comparisons between cohorts are reported in eTable 1 in the Supplement. Variance inflation factors (VIFs) for all variables were between 1 and 2, indicating that problematic levels of multicollinearity are not present in our analyses. 27 We tested the hypothesis that APOEε4 is associated with greater [ 18 F]MK6240 uptake independently of global [ 18 F]AZD4694 uptake. Voxelwise analyses revealed that AP OEε4 c arriership was assoc iated with inc reased [ 18 F]MK6240 SUVR in the bilateral entorhinal cortex and hippocampus (random field theory corrected at P < .001; signifi-cant clusters: P <.001; t = 4.42; β = 0.33; 95% CI, 0.19-0.49) ( Figure 1A). These results are independent of amyloid-β, clinical diagnosis, age, and sex. Results remained similar when using partial volume-corrected data (eFigure 1A in the Supplement; full model statistics of PVC data presented in Table 2). No statistically significant associations were observed beyond the medial temporal lobes.
We also tested the hypothesis that APOEε4 is associated with greater [ 18 F]flortaucipir uptake independently of global [ 18 F]florbetapir uptake. Voxelwise analyses revealed that APOEε4 carriership was associated with increased [ 18 F]flortaucipir SUVR in the bilateral entorhinal cortex (random field theory corrected at P < .001; significant clusters: t = 4.527; β = 0.13; 95% CI, 0.08-0.19) ( Figure 1B). Results remained similar when using partial volume-corrected data (eFigure 1B in the Supplement; full model statistics of PVC data presented in Table 2). These results are independent of amyloid-β, clinical diagnosis, age, and sex. No statistically significant associations were observed beyond the medial temporal lobes.
Scatterplots of the association between neocortical amyloid-PET SUVR and medial temporal tau-PET SUVR stratified by APOEε4 status are displayed in Figure 2. Density plots are also provided to visualize distribution of the data.

Meta-analytic Estimates
When fitting a fixed-effect rma model to the coefficients and standard errors from the models in both cohorts, we found that the main association of APOEε4 on medial temporal tau-PET SUVR was significant (P < .001, meta-analytic β = 0.22; 95% CI, 0.15-0.29).

Discussion
This study provides evidence from 2 independent cohorts that APOEε4 is associated with increased tau pathology in the entorhinal cortex and hippocampus independently of age, clinical status, sex, and amyloid-β. Our study is in agreement with a growing body of research demonstrating greater vulnerability of the medial temporal lobes to hypometabolism 11,12 and atrophy [31][32][33] in APOEε4 carriers compared with noncarriers, independently of amyloid-β. Because of the topographical concordance between tau pathology and neurodegeneration, 13,14,34 our results suggest greater tau pathology may be responsible for the medial temporal neurodegeneration observed in APOEε4 carriers. Our findings of greater medial temporal tauopathy are consistent with specific neuropsychologic profiles of APOEε4 carriers vs noncarriers. Patients with Alzheimer disease demen-tia who are APOEε4 carriers perform worse on memory tasks than noncarriers at the same disease stage. 35,36 Correspondingly, memory tends to be relatively preserved in ε4-negative patients, while deficits in executive function and processing speed are more severe. 37,38 Patients with Alzheimer disease dementia who do not carry an ε4 allele are also more likely to present with nonamnestic phenotypes. 39 Taken together, these studies support a framework in which medial temporal structures are specifically vulnerable to the deleterious effects of APOEε4.
An outstanding question is why APOEε4's association with tau pathology is restricted to the medial temporal lobe. While pyramidal neurons of the entorhinal cortex, subiculum, and CA1 region of the hippocampus are vulnerable to early tau accumulation in Alzheimer disease, [40][41][42] limited data exist as to how APOEε4 may preferentially (or selectively) affect tau aggregation in these structures. 10 Data from the Allen Brain Atlas suggest that messenger RNA expression of APOE is high-

Cognitively impaired
A, Clusters that remained significant after multiple comparisons correction with random field theory at P < .001 were used to extract tau-PET standardized uptake value ratio (SUVR) values in the TRIAD cohort (left) and ADNI cohort (right). B, Scatterplots displaying associations between medial temporal tau PET and neocortical amyloid PET stratified by APOEε4 genotype in TRIAD (left) and ADNI (right). Density plots are provided along the x and y axes to visualize the distribution of the data for neocortical amyloid PET and medial temporal tau PET SUVR, respectively. In the TRIAD cohort, est in the medial temporal lobes. 43 Apolipoprotein E immunoreactivity is observed in neurons bearing neurofibrillary tangles. 44 Furthermore, greater expression of neuronal APOE is associated with increased tau phosphorylation in transgenic animal models [45][46][47] and human stem cell models. 48 Truncated APOEε4 fragments are also associated with greater tau hyperphosphor ylation and neuronal c y toskeletal disruption. 49,50 Our study builds on studies of tau-PET distribution across the Alzheimer disease spectrum 15 by identifying a unique regional contribution of APOEε4 to tau pathology. Furthermore, neuropathologic 14 and tau-PET 51 studies that have identified medial temporal tauopathy in the absence of amyloid-β suggest that medial temporal tauopathy may be a consequence of aging. Correspondingly, later age at onset of Alzheimer disease dementia is linked to limbic-predominant or memory-predominant clinical presentations. 52 Even in cognitively normal individuals, increased tau pathology in the medial temporal lobe is associated with declines in subjective 53 and objective memory function as well as medial temporal gray matter volume. 54 Our study extends these findings by identifying APOEε4 as a contributor to medial temporal tauopathy, independent of age and amyloid-β.
While the results of our study implicate APOEε4 in the pathogenesis of both pathological hallmarks of Alzheimer disease, 55 APOEε4 is not sufficient for a diagnosis of Alzheimer disease nor to cause dementia. Instead, our study supports a framework in which isocortical/medial temporal (Braak stage 1-2) tau pathology may be a consequence of specific vulnerability factors (such as aging 51,56 or genotype 9 ), while amy-loid-β facilitates the spread of tau pathology from the medial temporal lobe to neocortical regions, 57,58 associated with greater cognitive decline. In fact, significant tau pathology in neocortical regions is seldom observed independently of amyloid-β pathology, 59 although exceptions do exist. 60 Because accepted Alzheimer disease models suggest that amyloid-β accumulation occurs years before tau accumulation measured with cerebrospinal fluid, 61,62 longitudinal imaging studies are needed to clarify APOEε4's association with medial temporal tau pathology across disease stages.

Strengths and Limitations
Some methodologic limitations should be considered when interpreting this study. The first is that this study is not designed to discover a biological mechanism underlying the association between APOEε4 and tau independently of amyloid-β. It is important to mention that both TRIAD and ADNI cohorts are convenience samples of individuals motivated to participate in a study about Alzheimer disease and thus involve recruitment and sampling biases. Future work is needed to determine whether the effects of APOEε4 on tau result in increased phosphorylation, conformational changes, or increased cortical spreading. Future studies should also investigate possible associations between APOEε4 and amyloid-β in relation to tau pathology. Methodologic strengths of this study include large sample sizes as well as a replication in an independent cohort. In particular, replication of results obtained with first-generation and second-generation tau-PET ligands is an important methodological advance.

Conclusions
In summar y, we found that AP OEε 4 is assoc iated with increased tau pathology in medial temporal structures i n d e p e n d e nt o f a my l o i d -β , s ex , a ge , a n d c l i n i c a l status. These results, in combination with preclinical data, 9 , 4  Role of the Funder/Sponsor: Data used in preparation of this article were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. The other funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: Data used in preparation of this article were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: http://adni.loni.usc. edu/wp-content/uploads/how_to_apply/ADNI_