APOE and amyloid‐tau pathology in cognitively unimpaired older adults

Abstract

INTRODUCTION

Apolipoprotein E (APOE) genotype shows well‐established dose‐dependent associations with higher amyloid in cognitively unimpaired (CU) adults. In contrast, associations with tau burden and cognition are less well characterized.

METHODS

We performed a cross‐sectional analysis of CU participants from the Anti‐Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4), Alzheimer's Disease Neuroimaging Initiative (ADNI), Wisconsin Registry for Alzheimer's Prevention (WRAP), and National Alzheimer's Coordinating Center (NACC) within the harmonized multi‐cohort Alzheimer's Disease Sequencing Project Phenotype Harmonization Consortium (ADSP‐PHC) data. A total of 4380 CU participants were included with APOE genotype, amyloid PET, and cognitive data (memory, language, executive, and visuospatial function), including a subset of 758 with tau PET imaging.

RESULTS

Compared with ε33, ε24 (β = 18.340), ε34 (β = 23.850), and ε44 (β = 44.820) showed higher amyloid burden (all p < .001). Similarly, ε24 (OR = 4.180), ε34 (OR = 4.540), and ε44 (OR = 14.150) had higher odds of amyloid positivity (all p < .001). After adjustment for amyloid burden, ε4 carriers had higher tau burden in the entorhinal cortex (β = 0.026; false discovery rate [FDR] ‐adjusted p = 0.041) and amygdala (β = 0.040; FDR‐adjusted p = 0.003). Higher tau burden was associated with lower memory (β = −0.439; FDR‐adjusted p = 0.012) and language performance (β = −0.610; FDR‐adjusted p = 0.020).

DISCUSSION

APOE ε4 showed a strong dose‐dependent association with amyloid, with the highest levels observed among ε4 homozygotes. Associations between APOE and global tau were more modest and appeared to be driven mainly by ε4 homozygotes, while regional analyses showed localized APOE ε4‐related associations in medial temporal regions. Independently, higher tau was associated with lower memory and language performance.

Highlights

• Apolipoprotein E (APOE) ε4 showed dose‐dependent associations with amyloid burden and age‐related amyloid differences.

• APOE‐related global tau associations were modest, mainly in ε44 group, with stronger medial temporal effects among ε4 carriers.

• Higher tau burden was associated with lower memory and language performance, regardless of APOE status.

1. INTRODUCTION

Apolipoprotein E (APOE) ε4 is the strongest genetic risk factor for sporadic Alzheimer's disease (AD). ¹ , ² Carrying one APOE ε4 allele is associated with an approximately two‐ to 3‐fold higher risk of AD, while individuals with two ε4 alleles may have a 10–15‐fold greater risk and develop dementia at an earlier age, supporting a dose‐dependent effect. ³ In contrast, the APOE ε2 allele is generally associated with lower AD risk and later onset of AD‐related biomarker changes. ε2 carriers have an estimated odds ratio of about 0.6, compared to individuals with ε33, the most common APOE genotype. ⁴ APOE and its protein product, apoE, are involved in several biological pathways linked to AD, including amyloid‐β (Aβ) aggregation, fibril formation, and clearance. ⁵ Consistent with these effects, APOE ε4 carriers typically become amyloid positive about 10–15 years earlier than ε33 carriers, in contrast, amyloid positivity tends to occur later among ε2 carriers. ⁶

Recent studies suggest that APOE may also contribute to tau pathology and downstream neurodegeneration. Because Aβ and tau are closely linked in AD, ⁷ part of the relationship between APOE and tau may be mediated through amyloid burden. However, whether APOE has independent associations with tau pathology and cognition remains uncertain, particularly in cognitively unimpaired (CU) individuals. Previous tau positron emission tomography (PET) studies have reported mixed findings. Some studies found that tau abnormalities were primarily associated with age and amyloid burden, with limited independent APOE effects after accounting for amyloid pathology. ⁸ , ⁹ In contrast, Ossenkoppele and colleagues ¹⁰ identified APOE ε4 carrier status as a factor associated with tau PET positivity, and Ghisays and colleagues ¹¹ showed that APOE ε4 modified the relationship between age and entorhinal tau deposition. Young and colleagues ¹² further reported that among amyloid‐positive CU individuals, APOE ε4 was associated with higher regional tau PET burden even after accounting for cross‐sectional amyloid burden. Ossenkoppele and colleagues ¹³ showed that tau PET positivity in temporal composite regions increased with APOE ε4 carrier status in CU individuals in models accounting for amyloid status. APOE ε4 has also been reported to modify the relationship between medial temporal tau pathology and memory performance independently of amyloid burden. ¹⁴

Given these mixed findings, we used harmonized data from four cohorts within the Alzheimer's Disease Sequencing Project Phenotype Harmonization Consortium (ADSP‐PHC). We evaluated associations between APOE genotype and amyloid burden, examined the relative contributions of amyloid burden and APOE genotype to global and regional tau pathology, and assessed associations between tau burden and domain‐specific cognitive performance. We hypothesized that APOE ε4 would show the strongest association with amyloid burden, that amyloid burden would be the primary correlate of tau pathology with smaller APOE‐related tau effects, and that higher tau burden would be associated with lower cognitive performance, particularly memory, independently of APOE status.

2. METHODS

2.1. Study design and participants

This cross‐sectional study used data from ADSP‐PHC Release 3, ¹⁵ which harmonizes demographic, cognitive, imaging, and biomarker data across multiple cohorts to support large‐scale investigations of Alzheimer's disease.

Participant selection is summarized in Figure S1 and described in detail in the Supplementary Methods. From 64,932 participants in the ADSP‐PHC Release 3 extraction, the final amyloid PET sample included 4380 CU participants from the Anti‐Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4) (n = 3310), Alzheimer's Disease Neuroimaging Initiative (ADNI) (n = 691), Wisconsin Registry for Alzheimer's Prevention (WRAP) (n = 236), and National Alzheimer's Coordinating Center (NACC) (n = 143). The tau PET subset included 758 participants from ADNI (n = 421), A4 (n = 333), and NACC (n = 4).

All contributing studies received approval from their local institutional review boards, and all participants provided written informed consent.

RESEARCH IN CONTEXT

1. Systematic review: The authors reviewed the literature using PubMed and Google Scholar to examine relationships among Apolipoprotein E (APOE) genotype, amyloid, tau, and cognition in cognitively unimpaired (CU) individuals. Previous studies from the Anti‐Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4), Alzheimer's Disease Neuroimaging Initiative (ADNI), and A4/LEARN have consistently shown strong APOE‐related effects on amyloid and more modest regional effects on tau. This study extends this literature using a large harmonized Alzheimer's Disease Sequencing Project Phenotype Harmonization Consortium (ADSP‐PHC) dataset with amyloid positron emission tomography (PET), tau PET, and harmonized cognitive measures.

2. Interpretation: Amyloid abnormalities remain the main pathological feature linked to APOE genotype in CU individuals, while tau effects are smaller and primarily involve medial temporal regions. Higher tau burden was associated with lower memory and language performance, regardless of APOE status.

3. Future directions: These findings support the use of APOE genotype as a stratification factor in studies of CU individuals. Longitudinal studies are needed to determine whether APOE‐related amyloid and tau differences predict subsequent cognitive decline and progression to Alzheimer's disease.

2.2. Clinical and cognitive assessments

Cognitive performance was assessed using ADSP‐PHC harmonized domain‐specific scores for memory, executive function, language, and visuospatial function. Briefly, individual cognitive test items were assigned to cognitive domains and harmonized across cohorts using confirmatory factor analysis (CFA) and cocalibration, with common items serving as anchors to place different neuropsychological batteries on a common metric. ¹⁶ Domain‐specific factor scores were estimated for each participant, with higher scores indicating better cognitive performance. The cognitive measures contributing to each domain varied across cohorts and are summarized in Table S1.

Cognitive status classifications were assigned by the individual cohorts and were not harmonized by ADSP‐PHC. Only participants classified as cognitively unimpaired (PHC_Diagnosis = 1) were included. Cohort‐specific criteria used to define CU status are provided in the Supplementary Methods.

2.3. Amyloid and tau PET imaging

Amyloid and tau PET measures were obtained from the ADSP‐PHC MRI‐free PET processing pipeline, with additional processing details provided in the Supplementary Methods. ¹⁷ Amyloid burden was quantified using a global cortical composite, normalized to the whole cerebellum, and converted to Centiloid (CL) units; amyloid positivity was defined as CL ≥ 25. ¹⁸ , ¹⁹

Tau burden was assessed using ¹⁸F‐flortaucipir (FTP) standardized uptake value ratios (SUVRs) normalized to inferior cerebellar gray matter, with global tau represented by a meta‐temporal composite. ²⁰ Regional analyses examined individual regions contributing to the meta‐temporal composite and selected neocortical regions, including the entorhinal cortex, amygdala, inferior temporal cortex, inferior parietal cortex, and precuneus. ¹²

2.4. Data analysis

Regression models were used to examine associations among APOE genotype, amyloid burden, tau pathology, and cognitive performance. Continuous outcomes (CL, global and regional tau burden, and cognitive scores) were analyzed using linear regression, while amyloid positivity was analyzed using logistic regression.

Initial analyses focused on the associations between APOE genotype and amyloid burden using both continuous CL measures and binary amyloid positivity. We then examined associations between amyloid burden and global tau pathology using both amyloid positivity and continuous CL values. Associations between APOE and global tau burden were subsequently tested after adjusting for continuous amyloid burden. The same analyses were repeated in amyloid‐positive individuals only. These analyses were then extended to regional tau measures. Multiple comparisons across regional tau analyses were controlled using false discovery rate (FDR) correction, and both uncorrected and FDR‐corrected p values are reported.

APOE × age and APOE × sex interactions were tested, and spline‐based age models were used to assess potential nonlinear age effects. Model fit was evaluated using likelihood ratio tests (LRTs) and changes in Akaike information criterion (ΔAIC).

Associations between tau burden and domain‐specific cognitive performance were examined using all available cognitive data. Sensitivity analyses included robust and quantile regression models.

All models were adjusted for age, sex, education, and study cohort. APOE genotype was initially modeled as a six‐category variable (ε22, ε23, ε33, ε24, ε34, ε44), with ε33 as the reference group. When overall genotype effects were not significant, ε4 carrier and ε44 homozygous status were evaluated separately.

Group differences were assessed using Kruskal–Wallis tests with Dunn post hoc comparisons and Holm correction, and categorical variables were analyzed using chi‐square tests. Statistical significance was defined as two‐sided p < 0.05. All analyses were performed in Python.

3. RESULTS

3.1. Sample characteristics

Baseline characteristics are summarized in Tables 1 and 2. The amyloid PET sample included 4380 participants (mean [SD] age, 71.3 [5.6] years; 60.1% women; education, 16.7 [2.8] years). The most common APOE genotypes were ε33 (54.6%), ε34 (29.3%), and ε23 (10.2%).

TABLE 1.

Participant characteristics of the amyloid PET analytic sample, overall and by APOE genotype.

Characteristic	Overall	ε22	ε23	ε33	ε24	ε34	ε44	p‐Value
N (%)	4380	14 (0.3)	447 (10.2)	2390 (54.6)	107 (2.4)	1282 (29.3)	140 (3.2)
Age, mean (SD), y	71.3 (5.6)	69.63 (4.12)	72.06 (5.71)	71.66 (5.74)	71.53 (5.65)	70.48 (5.36)	69.92 (5.36)	< 0.001
Women, N (%)	2631 (60.1)	8 (57.1)	257 (57.5)	1450 (60.7)	66 (61.7)	771 (60.1)	79 (56.4)	0.771
Race, N (%)
American Indian or Alaska Native	13 (0.3)	0	0	6 (0.3)	0	6 (0.5)	1 (0.7)
Asian	73 (1.7)	0	4 (0.9)	57 (2.4)	0	11 (0.9)	1 (0.7)
Black or African American	203 (4.6)	0	43 (9.6)	89 (3.7)	8 (7.5)	54 (4.2)	9 (6.4)
Native Hawaiian or Other Pacific Islander	2 (0.0)	0	0	1 (0.0)	0	1 (0.1)	0
White	4028 (92.0)	14 (100.0)	395 (88.4)	2199 (92.0)	97 (90.7)	1195 (93.2)	128 (91.4)
Other/Unknown/Multiple	61 (1.4)	0	5 (1.1)	38 (1.6)	2 (1.9)	15 (1.2)	1 (0.7)
Ethnicity, N (%)								0.048
Hispanic or Latino	176 (4.0)	0	9 (2.0)	115 (4.8)	5 (4.7)	42 (3.3)	5 (3.6)
Not Hispanic or Latino	4180 (96.0)	14 (100.0)	435 (98.0)	2262 (95.2)	101 (95.3)	1233 (96.7)	135 (96.4)
Education, mean (SD), y	16.7 (2.8)	16.86 (2.14)	16.60 (3.01)	16.60 (2.72)	16.60 (2.99)	16.80 (2.69)	16.77 (2.55)	0.699
Memory, mean (SD)	0.74 (0.44)	0.64 (0.32)	0.73 (0.42)	0.75 (0.45)	0.71 (0.44)	0.72 (0.43)	0.72 (0.46)	0.502
Executive Function, mean (SD)	0.86 (0.60)	–	0.89 (0.56)	0.87 (0.60)	0.74 (0.53)	0.86 (0.63)	0.73 (0.50)	0.562
Language, mean (SD)	0.85 (0.56)	–	0.79 (0.49)	0.85 (0.57)	0.82 (0.48)	0.87 (0.57)	0.87 (0.50)	0.822
Visuospatial, mean (SD)	0.49 (0.49)	–	0.49 (0.46)	0.48 (0.51)	0.43 (0.51)	0.52 (0.47)	0.57 (0.39)	0.804
Aβ positive, N (%)	1300 (29.7)	2 (14.3)	76 (17.0)	454 (19.0)	52 (48.6)	614 (47.9)	102 (72.9)	< 0.001
Aβ SUVR (Centiloids), mean (95% CI)	17.91 (16.89–18.93)	4.64 (−2.92–12.20)	6.78 (4.46–9.09)	9.73 (8.59–10.87)	28.15 (21.70–34.60)	32.47 (30.32–34.63)	53.20 (46.40–60.00)	< 0.001

Abbreviations: –, missing data; APOE, apolipoprotein E; Aβ, amyloid‐β; CI, confidence interval; SD, standard deviation; SUVR, standardized uptake value ratio.

TABLE 2.

Participant characteristics of the tau PET analytic sample, overall and by APOE genotype.

Characteristic	Overall	ε22	ε23	ε33	ε24	ε34	ε44	p‐Value
N (%)	758	1 (0.1)	62 (8.2)	370 (48.8)	20 (2.6)	277 (36.5)	28 (3.7)
Age, mean (SD), y	72.3 (6.1)	68.73	72.65 (6.03)	73.18 (6.39)	72.84 (5.58)	71.09 (5.42)	70.89 (5.82)	0.002
Women, N (%)	456 (60.2)	1 (100.0)	36 (58.1)	214 (57.8)	11 (55.0)	173 (62.5)	21 (75.0)	0.414
Race, N (%)
American Indian or Alaska Native	3 (0.4)	0	0	2 (0.5)	0	1 (0.4)	0
Asian	14 (1.8)	0	1 (1.6)	11 (3.0)	0	2 (0.7)	0
Black or African American	36 (4.7)	0	8 (12.9)	13 (3.5)	1 (5.0)	14 (5.1)	0
White	691 (91.2)	1 (100.0)	52 (83.9)	335 (90.5)	18 (90.0)	258 (93.1)	27 (96.4)
Other/Unknown/Multiple	14 (1.8)	0	1 (1.6)	9 (2.4)	1 (5.0)	2 (0.7)	1 (3.6)
Ethnicity, N (%)								0.830
Hispanic or Latino	28 (3.7)	0	2 (3.2)	15 (4.1)	0	11 (4.0)	0
Not Hispanic or Latino	724 (96.3)	1 (100.0)	60 (96.8)	353 (95.9)	20 (100.0)	262 (96.0)	28 (100.0)
Education, mean (SD), y	16.6 (2.5)	18.00	16.37 (2.54)	16.82 (2.50)	15.80 (3.00)	16.34 (2.44)	16.64 (2.86)	0.194
Memory, mean (SD)	0.81 (0.49)	0.89	0.80 (0.47)	0.86 (0.49)	0.63 (0.44)	0.76 (0.48)	0.79 (0.55)	0.055
Executive Function, mean (SD)	0.91 (0.55)	–	0.86 (0.47)	0.94 (0.56)	0.47 (0.49)	0.94 (0.55)	0.68 (0.60)	0.140
Language, mean (SD)	0.86 (0.55)	–	0.75 (0.50)	0.84 (0.54)	0.75 (0.31)	0.95 (0.61)	0.75 (0.37)	0.194
Visuospatial, mean (SD)	0.47 (0.51)	–	0.48 (0.44)	0.45 (0.52)	0.16 (0.64)	0.53 (0.51)	0.33 (0.44)	0.273
Aβ positive, N (%)	384 (50.7)	1 (100.0)	17 (27.4)	139 (37.6)	14 (70.0)	190 (68.6)	23 (82.1)	< 0.001
Aβ SUVR (Centiloids), mean (95% CI)	32.95 (30.14–35.76)	28.00	10.29 (4.44–16.14)	22.46 (18.84–26.08)	36.40 (23.70–49.10)	49.01 (44.28–53.73)	60.61 (44.18–77.04)	< 0.001
Tau SUVR, mean (95% CI)	1.24 (1.23–1.25)	1.18	1.21 (1.19–1.23)	1.23 (1.22–1.25)	1.23 (1.17–1.28)	1.26 (1.24–1.27)	1.31 (1.23–1.39)	< 0.001

Abbreviations: –, missing data; APOE, apolipoprotein E; Aβ, amyloid‐β; CI, confidence interval; SD, standard deviation; SUVR, standardized uptake value ratio.

The tau PET subset included 758 participants (mean [SD] age, 72.3 [6.1] years; 60.2% women), with similar educational attainment and APOE genotype frequencies.

In both samples, age differed statistically across APOE genotype groups (amyloid PET: p < 0.001; tau PET: p = 0.002), although absolute differences were small and unlikely to be clinically meaningful. Sex, education, and cognitive domain scores did not differ significantly across groups (all p ≥ 0.05).

3.2. APOE genotype is associated with amyloid burden and positivity

Descriptive analyses showed marked differences in amyloid burden and amyloid positivity across APOE genotype groups (Table 1; Figure 1A). Mean amyloid burden ranged from 9.730 in ε33 carriers to 53.20 in ε44 carriers, with progressively higher levels across ε4‐containing genotypes (p < 0.001). Amyloid positivity showed a similar pattern, ranging from 17.0% in ε23 carriers to 72.9% in ε44 carriers (p < 0.001). Pairwise comparisons confirmed higher amyloid burden in ε24, ε34, and ε44 compared with ε23 and ε33 (all adjusted p < 0.001). Results for ε22 should be interpreted cautiously because of the small sample size.

FIGURE 1.

Association between APOE genotype, age, and amyloid burden. (A) Distribution of amyloid burden across APOE genotypes, showing progressively greater amyloid burden across ε4‐containing genotypes. (B) Predicted amyloid burden by age according to APOE genotype from covariate‐adjusted regression models, demonstrating larger cross‐sectional age‐associated differences in amyloid burden among ε4‐containing genotypes. (C) Spline‐based sensitivity analysis showing nonlinear cross‐sectional age‐associated differences in amyloid burden according to APOE ε4 carrier status. APO, apolipoprotein E.

After adjustment for age, sex, education, and cohort, APOE genotype remained strongly associated with both continuous amyloid burden (LRT χ²(4) = 666.40; p < 0.001) and amyloid positivity. Compared with ε33, ε24 (β = 18.340), ε34 (β = 23.850), and ε44 (β = 44.820) showed higher amyloid burden (all p < 0.001), while ε22/ε23 showed slightly lower levels (β = −3.460; p = 0.032). Similarly, ε24 (odds ratio [OR] = 4.180), ε34 (OR = 4.540), and ε44 (OR = 14.150) had higher odds of amyloid positivity (all p < 0.001), while ε22/ε23 did not differ from ε33 (OR = 0.840; p = 0.209).

3.3. Age‐related differences in continuous amyloid burden by APOE genotype

A significant APOE genotype × age interaction was observed for amyloid burden (ΔAIC = −37.96; LRT χ²(4) = 45.96; p < 0.001), indicating that age‐associated differences in amyloid burden varied across APOE genotype groups (Figure 1B). Age‐associated differences were minimal in ε22/ε23 carriers (−0.05 CL per year of age), modest in ε33 carriers (0.66), and progressively greater in ε24 (1.05), ε34 (1.69), and ε44 carriers (2.04), consistent with higher amyloid burden at older ages among ε4‐containing genotypes.

Spline‐based models did not support a significant overall nonlinear association between age and amyloid burden (ΔAIC = 3.71; LRT χ²(3) = 2.29; p = 0.514). However, adding APOE ε4 × spline(age) interaction terms significantly improved model fit (ΔAIC = −32.76; LRT χ²(4) = 40.76; p < .001), indicating nonlinear age‐related differences between ε4 carriers and non‐carriers (Figure 1C).

No significant APOE × sex interactions were observed for continuous amyloid burden (LRT χ²(4) = 2.40; p = 0.663) or amyloid positivity (LRT χ²(4) = 2.69; p = 0.610).

3.4. Amyloid burden is associated with tau pathology

Sequential regression models examining global tau burden are presented in Table S2. Descriptive analyses showed clear differences in amyloid burden and amyloid positivity across APOE genotype groups, with progressively higher amyloid levels observed in ε4‐containing genotypes (Table 2).

Tau burden was higher in Aβ‐positive than Aβ‐negative individuals (mean SUVR: 1.269 vs. 1.214; p < .001) (Figure 2A). After adjustment for covariates, amyloid positivity remained associated with higher tau burden (β = 0.060; p < 0.001) (Figure 2B), with no significant AmyloidPos × age interaction (p = 0.980).

FIGURE 2.

Association between amyloid status, age, APOE ε44 status, and global tau burden. (A) Distribution of global tau SUVR according to amyloid status, showing higher tau burden among amyloid‐positive individuals (p < 0.001). (B) Predicted global tau SUVR by age according to amyloid status from covariate‐adjusted regression models, showing higher tau burden across age among amyloid‐positive individuals. (C) Among amyloid‐positive individuals, predicted global tau SUVR by age stratified by APOE ε44 carrier status, showing modestly higher differences in tau burden after adjustment for continuous amyloid burden and covariates. APOE, apolipoprotein E; SUVR, standardized uptake value ratio.

When analyzed continuously, higher CL values were associated with greater tau burden (β = 0.001; p < 0.001). A significant CL × age interaction was observed (p = 0.032), although spline‐based models did not support significant nonlinear age‐related differences in tau burden either before (p = 0.109) or after adjustment for amyloid burden (p = 0.202).

No significant AmyloidPos × sex (p = 0.745) or CL × sex (p = 0.257) interactions were observed for tau burden.

3.5. APOE genotype shows a modest association with tau burden

Tau burden also differed across APOE genotype groups (p < 0.001), with higher values generally observed in ε4 carriers. Mean tau SUVR ranged from 1.21 in ε23 group, and 1.23 in ε33 group, to 1.26 in ε34 group (16.89–18.93) and 1.31 in ε44 group, with ε24 group showing intermediate values (Table 2).

After adjustment for continuous amyloid burden (CL) and covariates, APOE‐related differences in global tau burden were more modest. Compared with ε33 group, only ε44 group had higher tau burden (β = 0.0482; p = 0.032; LRT χ²(1) = 4.74; p = 0.029). Tau burden in ε24 and ε34 groups did not differ significantly from ε33 group. No APOE × age interaction was identified, and spline‐based age models similarly showed no evidence of nonlinear APOE × age effects on tau burden. An APOE ε44 × sex interaction was observed for global tau burden (ΔAIC = −5.42; LRT χ²(1) = 7.42; p = 0.006).

3.6. APOE‐related associations with tau burden among amyloid‐positive individuals

Analyses were restricted to amyloid‐positive participants to further examine APOE‐related differences in tau burden among individuals with elevated amyloid pathology.

Amyloid positivity differed across APOE genotype groups in the tau PET subset, increasing from 27.4% in ε23 carriers and 37.6% in ε33 carriers to 70.0% in ε24 carriers, 68.6% in ε34 carriers, and 82.1% in ε44 carriers (p < 0.001) (Table 2).

Within the amyloid‐positive subset, mean tau SUVR ranged from 1.206 in ε22/ε23 carriers to 1.326 in ε44 carriers, with intermediate values in ε33, ε24, and ε34 carriers.

After adjustment for continuous amyloid burden (CL) and covariates, overall APOE genotype was not significantly associated with global tau burden (LRT χ²(4) = 5.50; p = 0.240). Compared with ε33 carriers, ε44 carriers showed a borderline association with higher tau burden (β = 0.055; p = 0.054) and no significant differences were observed for ε24 or ε34 carriers. Figure 2C illustrates higher estimated tau SUVR values in ε44 carriers than non‐carriers after adjustment for amyloid burden and covariates.

An APOE ε44 × sex interaction was observed for global tau burden (ΔAIC = −5.24; LRT χ²(1) = 7.24; p = 0.007).

3.7. Regional tau analyses reveal spatially selective associations with amyloid and APOE ε4

Given the relatively modest APOE‐related differences in global tau burden, regional analyses were performed to determine whether these associations varied across brain regions. Results are presented for APOE ε4 carrier status (Table 3), although similar patterns were observed for ε44 homozygous status.

TABLE 3.

Associations between APOE ε4 carrier status and regional tau burden in the tau PET subset.

Region	APOE ε4 β (95% CI)	Unadjusted p	FDR‐adjusted p
Entorhinal cortex	0.026 (0.005 to 0.047)	0.016	0.041
Amygdala	0.040 (0.018 to 0.062)	< 0.001	0.003
Inferior temporal cortex	0.005 (−0.016 to 0.026)	0.625	0.637
Inferior parietal cortex	0.004 (−0.013 to 0.022)	0.637	0.637
Precuneus	0.004 (−0.011 to 0.020)	0.596	0.637

Abbreviations: APOE, apolipoprotein E; CI, confidence interval; FDR, false discovery rate; PET, positron emission tomography.

Across all participants, higher continuous amyloid burden was associated with higher tau burden in the entorhinal cortex, amygdala, inferior temporal cortex, inferior parietal cortex, and precuneus (all FDR‐adjusted p < 0.001), with the largest effects observed in medial temporal regions. CL × age interactions were identified in all regions except the inferior temporal cortex (all FDR‐adjusted p ≤ 0.009).

APOE ε4 carrier status was associated with higher tau burden in the entorhinal cortex (β = 0.026; FDR‐adjusted p = 0.041) and amygdala (β = 0.040; FDR‐adjusted p = 0.003), while associations in the inferior temporal cortex, inferior parietal cortex, and precuneus did not remain significant after FDR correction.

Among amyloid‐positive individuals (Table S3), APOE ε4 carrier status remained associated with higher tau burden in the entorhinal cortex (β = 0.047; FDR‐adjusted p = 0.017) and amygdala (β = 0.057; FDR‐adjusted p = 0.009) after adjustment for continuous amyloid burden and covariates. No significant associations were observed in the remaining regions after FDR correction. Continuous amyloid burden remained associated with higher tau burden across all examined regions, with the largest effects again observed in medial temporal regions.

No APOE ε4 × age or APOE ε4 × sex interactions remained significant after FDR correction.

3.8. Tau burden is selectively associated with memory and language performance

Cognitive domain scores were generally similar across APOE genotype groups in this cognitively unimpaired sample (Tables 1 and 2), and APOE genotype was not consistently associated with cognitive performance after covariate adjustment.

After adjustment for covariates, APOE genotype, and amyloid burden, higher tau SUVR was associated with worse memory performance in models using both binary amyloid status (β = −0.497; FDR‐adjusted p = 0.004) and continuous CL measures (β = −0.439; FDR‐adjusted p = 0.012). Similar associations were observed among amyloid‐positive individuals (β = −0.464; FDR‐adjusted p = 0.022). Higher tau burden was also associated with lower language performance in models adjusted for binary amyloid status (β = −0.579; FDR‐adjusted p = 0.020) and continuous CL burden (β = −0.610; FDR‐adjusted p = 0.020), although these associations did not remain significant among amyloid‐positive individuals after FDR correction. Similar findings were observed across sensitivity analyses (Figures S2–S3). Tau burden was not associated with executive function or visuospatial performance (Figures S4–S5).

In the amyloid PET sample, higher amyloid burden was associated with lower memory performance in both binary amyloid status (β = −0.050; FDR‐adjusted p < 0.001) and continuous CL models (β = −0.001; FDR‐adjusted p < .001). No significant associations were observed for executive function, language, or visuospatial performance.

Within the tau PET subset, amyloid burden remained independently associated with lower memory performance after adjustment for tau burden and covariates and no independent associations were identified for executive function, language, or visuospatial performance.

No consistent tau × sex or amyloid × sex interactions were observed across cognitive domains.

4. DISCUSSION

We evaluated relationships among APOE genotype, amyloid burden, tau pathology, and cognition in CU older adults using harmonized data from the ADSP‐PHC. APOE ε4 showed strong dose‐dependent associations with both amyloid burden and amyloid positivity, with the highest levels observed in ε44 carriers and larger age‐related differences among ε4 carriers. Amyloid burden emerged as the primary correlate of tau pathology. The main findings showed that APOE‐related differences in global tau burden were comparatively modest and appeared to be driven primarily by ε44 carriers. Regional analyses showed the clearest APOE‐related tau differences in medial temporal regions, particularly the entorhinal cortex and amygdala driven by ε4 carriers. Higher tau burden was related mainly to lower memory and language performance, while differences in executive and visuospatial domains were not observed.

The strong association between APOE ε4 and amyloid burden is consistent with previous PET studies showing higher amyloid burden and a greater frequency of amyloid positivity among ε4 carriers, particularly ε44 individuals. ¹¹ , ¹² , ¹³ Together with prior work, our findings support a clear APOE dose effect on amyloid pathology, with larger age‐associated differences observed among ε4‐containing genotypes.

APOE‐related differences in tau were smaller than those seen for amyloid and were largely limited to medial temporal regions. This pattern is generally consistent with previous tau PET studies in CU individuals. Prior studies have shown that amyloid burden is the strongest predictor of both cross‐sectional and longitudinal tau accumulation, while APOE ε4 tends to show more modest independent associations with tau after accounting for amyloid burden, with findings most evident in medial temporal regions and among ε44 carriers. ¹⁰ , ¹¹ , ²¹ Young and colleagues ¹² further showed that among amyloid‐positive CU individuals, APOE ε4 was associated with higher regional tau PET burden after accounting for amyloid burden, with the strongest effects observed in medial temporal regions. Ossenkoppele and colleagues ¹³ additionally found that APOE ε4 carrier status was associated with an earlier age of tau PET positivity in CU individuals. Our findings align most closely with these regionally selective tau effects: amyloid burden remained the strongest correlate of tau overall, while APOE‐related tau differences were smaller and largely limited to the entorhinal cortex and amygdala among APOE ε4 carriers, with somewhat stronger regional associations observed among amyloid‐positive individuals.

Cognitive performance was generally similar across APOE genotype groups after covariate adjustment, suggesting that APOE genotype alone had limited associations with cognition in this CU sample. In contrast, higher tau burden was associated with lower memory and language performance independent of amyloid burden, while executive and visuospatial performance were not significantly related to tau. This pattern is consistent with the known distribution of tau pathology in AD, with early involvement of medial temporal regions supporting episodic memory before broader neocortical involvement. ²² Previous studies have similarly linked higher medial temporal tau burden to worse memory performance in CU individuals. Weigand and colleagues ¹⁴ further reported that medial temporal tau pathology interacted with APOE ε4 status to influence memory independently of amyloid burden. Longitudinal studies have also shown that higher medial temporal and neocortical tau burden is associated with greater decline in everyday functioning among amyloid‐positive CU older adults. ²³ Together, these findings suggest that tau pathology may be more closely related to subtle cognitive differences during preclinical AD than APOE genotype itself, with APOE‐related cognitive effects likely reflecting its stronger associations with amyloid burden and comparatively smaller relationships with tau pathology.

We did not observe consistent APOE × sex interactions across amyloid burden, regional tau burden, or cognitive outcomes, although a modest APOE ε44 × sex interaction was identified for global tau burden. These largely negative findings should be interpreted cautiously, as the study was restricted to CU individuals with relatively limited tau pathology and the tau PET subset was modest in size. Previous studies have reported mixed findings regarding sex differences in tau pathology and cognition. ¹³ , ²⁴ , ²⁵

Several limitations should be noted. Our focus on CU individuals was intentional to characterize APOE‐related differences during the preclinical stage of AD; however, this restriction may have limited our ability to detect stronger APOE‐related effects on tau, which may emerge in older or mildly impaired individuals as tau burden increases. In addition, the restricted range of cognitive performance in CU individuals may have attenuated associations between cognition and other study variables. Longitudinal studies across the disease course will be important to clarify these relationships. Second, the sample was drawn from harmonized research cohorts, was dominated by A4 and ADNI participants, and was relatively homogeneous in terms of race and ethnicity and highly educated, which may limit generalizability to more diverse and community‐based populations. This is particularly relevant given reported ethnoracial differences in APOE‐associated genetic risk and amyloid and tau biomarker profiles. ²⁶ Consequently, the generalizability of these findings requires reapplication in more racially, ethnically, and educationally diverse populations. Although PET and cognitive measures were harmonized across cohorts, residual differences in recruitment procedures, imaging acquisition, tracer use, and cognitive assessment may remain. Finally, some APOE genotype groups, particularly ε22, ε24, and ε44, were relatively small, limiting power to detect subtle genotype‐specific effects.

5. CONCLUSION

These findings support amyloid burden as the primary APOE‐linked biomarker abnormality in CU older adults, while highlighting medial temporal tau as a downstream marker more closely related to subtle cognitive differences.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no commercial or financial relationships that could be construed as a potential conflict of interest. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All human subjects provided written informed consent and permission to share their de‐identified data.

Supporting information

Supporting Information

Supporting Information