APOE4, Blood Neurodegenerative Biomarkers, and... : JAMA Network Open (original) (raw)

Introduction

Alzheimer disease and related dementias (ADRDs) are multifactorial neurodegenerative disorders defined biologically by the ATN-I framework, characterized by the accumulation of β-amyloid plaques (A), tau tangles (T), neurodegeneration (N), and neuroinflammation (I). The APOE4 allele is the most dominant genetic risk factor for late-onset AD and increases AD risk through several potential mechanisms, resulting in high amyloid burden, accumulation of tau tangles, and microglial dysfunction. Specifically, in the central nervous system, the apolipoprotein E protein is expressed by the glial cells, predominantly in the astrocytes.

Apart from β-amyloid plaque (A) and tau tangle (T) AD biomarkers, AD blood-based neurodegenerative biomarkers include the total tau (t-tau) and glial fibrillary acidic proteins (GFAPs), which are indicative of neurodegeneration (N) and neuroinflammation (I) in the central nervous system, respectively, and neurofilament light (NfL), which reflects non–AD-specific subcortical, large-caliber axonal degeneration. These blood-based neurodegenerative markers have been shown to predict cognitive decline robustly and longitudinally years before cognitive decline in human participants.

However, although recent studies have acknowledged the importance of APOE4 by including it as a covariate, scarce population-based data exist on whether APOE4 modifies the associations of blood-based neurodegenerative biomarkers with cognitive decline, particularly in diverse, biracial populations of community-dwelling older adults without dementia. For example, only 2 studies, both published in the past year, have examined _APOE4_’s modification of the associations of 2 select blood-based AD and neurodegenerative markers, plasma p-tau and NfL levels. One of the studies was a small clinical study on the associations of plasma p-tau181, p-tau231, and GFAP with cognitive decline. The study investigated cross-sectional but not prospective associations of the biomarkers with cognitive decline. In studies on animals with the astrocyte-specific GFAP promoter compared with APOE3-secreting astrocytes, there was a significantly decreased number of neurite outgrowths from neurons cultured with _APOE4_-secreting or APOE-knockout astrocytes. Despite these findings, to our knowledge, no population-based study has investigated the modification of GFAP associations by APOE4 carrier status. Furthermore, to the best of our knowledge, no studies on _APOE4_’s modification of the associations of t-tau levels with cognitive decline, specifically in community-dwelling older adults, have been performed.

Additionally, more diverse demographics are needed to enhance the generalizability of blood-based biomarker studies. Extant studies are among overwhelmingly White populations, excluding members of racial minority groups, who have been estimated to have a higher risk for and the steepest increase in incident ADRD in the next 40 years, as well as a higher frequency of the APOE4 allele due to genetic inheritance and ancestry. Furthermore, many blood-based biomarker studies recruited human participants solely from clinical settings, often with very different demographics and AD risk than real-world, community-dwelling older adults.

To address these gaps, we hypothesized that the combination of APOE4 carrier status and increased blood-based neurodegenerative marker levels, indicated by higher levels of blood-based neurodegeneration (t-tau), axonal injury (NfL), and reactive astrocytes and neuroinflammation (GFAP), would constitute a higher neurodegenerative burden and neuronal damage in APOE4 carriers than noncarriers. We hypothesized that this would be associated with an accelerated rate of cognitive decline in a biracial population of community-dwelling older adults without dementia.

Methods

Study Participants

The Chicago Health and Aging Project (CHAP) enrolled participants based on a door-to-door census in 4 Chicago neighborhoods with substantial proportions of Black older adults and White residents. Inclusion criteria required that participants live in the geographical areas and be aged older than 65 years. CHAP data were collected in triennial cycles, with the baseline cohort starting in 1993 and subsequent cycles in 1997, 2000, 2003, 2006, 2009, and 2012. In each cycle, a stratified random sample based on age, sex, race, and cognitive level was selected for clinical evaluation of incident AD dementia when the participants also provided blood samples. Starting in the fifth cycle, all older adults who consented to blood sample collection also provided blood samples. Participants who had not consented to blood sample collection in the fifth cycle but had consented in subsequent cycles were included. For this study, we restricted the population to individuals fulfilling the following additional criteria: (1) participants had no dementia at study baseline, indicated by the absence of a clinical diagnosis of AD performed as part of the clinical assessment or a cutoff of Mini-Mental State Examination (MMSE) score of 24 or greater for individuals without a clinical diagnosis (ie, for participants who were not randomly selected to participate in the clinical assessment component); (2) participants provided blood samples in 1 cycle; and (3) participants had 1 or more cycles of composite global composite cognitive scores. Across study waves, 1038 participants without dementia fulfilled all criteria and were included in this study (eFigure 1 in Supplement 1). This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. The CHAP study was approved by the Institutional Review Board of Rush University Medical Center, and each participant provided written informed consent.

Quantification of Biomarker Serum Levels

We have described the blood specimen collection, selection, and processing in detail. In mid-2019, previously unthawed blood sample aliquots were shipped in dry ice to Quanterix Corporation. We focused on 3 neurodegenerative biomarkers: t-tau, NfL, and GFAP. They were assayed using single-molecule ultrasensitive immunoassays in duplicates with several control assays in a Quanterix bead-based HD platform and Neurology 4-Plex kit. The coefficient of variation in duplicate samples was 7.3% for t-tau and 3.0% for NfL and GFAP. For our analysis, we found the mean of duplicate measurements for each participant and used the mean level of biomarkers.

APOE Genotypes

APOE genotypes were ascertained using 2 single-nucleotide variants (SNVs; formerly single-nucleotide polymorphisms): rs7412 and rs429358. We genotyped the 2 SNVs at the Broad Institute Center for Genotyping using the homogeneous mass extension Sequenom MassArray platform. Genotyping call rates were 100% for SNV rs7412 and 99.8% for SNV rs429358. Both SNVs were in Hardy-Weinberg equilibrium, with P values of .08 and .79, respectively. Based on these 2 SNVs, we created an indicator variable for participant APOE4 carrier status. Participants with ε2/ε4, ε3/ε4, and ε4/ε4 were classified as APOE4 carriers, and those with ε2/ε2 ε2/ε3, and ε3/ε3 genotypes were classified as non–APOE4 carriers.

Global and Individual Tests of Cognition

We administered 4 brief performance tests of cognitive function at 3-year intervals as part of a structured in-home population interview, including immediate and delayed recall tests, the Symbol Digit Modalities Test, and the MMSE. Immediate and delayed recall of 12 ideas in the East Boston Story provided measures of episodic memory. The oral version of the Symbol Digit Modalities Test was administered to assess perceptual speed, a component of executive function. The MMSE provided a measure of global cognition. Because in a previous factor analysis, all 4 cognitive tests loaded on a single factor that accounted for 75% of the variance, we formed a composite measure of global cognition based on the 4 tests. As previously described, raw scores on each test were converted to z scores using baseline means and SDs of the entire CHAP population, and we found means of z scores to yield the composite global cognition score. We used the composite as the primary outcome because composite measures minimize floor and ceiling artifacts. Individual tests of cognition were also conducted in sensitivity analyses.

Demographic Variables and Other Covariates

During the baseline assessment, we collected demographic measures: age at the time of first blood sample collection, biological sex (male or female), and self-reported race (Black older adults and White older adults) using items from the 1990 US Census Bureau and education (measured in the number of years of schooling completed). Our analysis centered on baseline age at 75 years and education at 12 years and included indicator variables for female sex (reference group, male) and Black older adults (reference group, White older adults). Other covariates included common chronic conditions (ie, heart disease, diabetes, hypertension, and stroke), defined by self-report questions from the Established Populations for the Epidemiologic Study of the Elderly, as well as body mass index (BMI; calculated as weight in kilograms divided by height in meters squared).

Statistical Analysis

Baseline descriptive statistics were computed for demographic and cognitive characteristics, including age, number of formal years of education completed, self-reported race, sex at birth, and global cognition, and stratified by the APOE4 allele. Levels of blood biomarkers were positively skewed and greater than zero. Hence, we made a log10 transformation with geometric means and their 95% CIs. Descriptive comparisons between participants with and without the APOE4 were based on t tests for untransformed characteristics, χ2 tests for frequencies, and the Wilcoxson rank test for serum biomarkers. All regression models were adjusted for age at first blood sample collection (centered at 75 years), education (centered at 12 years), female sex, Black older adult race, BMI, and common chronic conditions, including heart disease, diabetes, hypertension, stroke.

We used linear mixed-effects regression modeling to examine the association of baseline blood-based biomarker levels with the baseline level of cognitive function and longitudinal changes in the annual rate of cognitive decline, with intercept and slope. Model assumptions were checked analytically and graphically and were deemed to be adequately met, including linearity, homoscedasticity, normality of residuals, and multicollinearity.

Time since baseline blood assessment in years captured the annual rate of change in cognitive function over time. We operationalized biomarkers in 2 different ways. First, we used log-transformed continuous biomarker levels. Second, we categorized tertiles of serum biomarker levels with 2 indicator variables for the second and third tertiles (with the first tertile as the reference group) to examine the association of higher tertiles of biomarker levels with cognitive decline.

We also performed 2 separate yet related sets of analyses. First, we ran a 3-way interaction model incorporating the interaction term APOE4 carrier status × time × continuous biomarker levels/tertiles. Subsequently, we stratified the total sample by APOE4 carrier status. Between– and within–APOE4 carrier differences were tested using contrast for estimates from respective regression models.

All regression models were performed using the nlme library, and graphical representations were performed with R statistical software version 4.2 (R Project for Statistical Computing). P values < .05 were considered statistically significant, and P values were 2-sided. We did not perform adjustments for multiple comparisons given that all analyses were based on an a priori hypothesis and contrasts were based on a single regression model for each biomarker. Statistical analyses were conducted from June 2024 to January 2025.

Similarly, using linear mixed-effects regression modeling, we performed 3 sets of sensitivity analyses to test the robustness of conclusions and granularity in the data that may suggest subgroup differences:

After removing baseline cases of mild cognitive impairment, the total sample was further reduced to include 802 participants without any cognitive impairment.
Instead of a composite cognitive measure, individual domains or tests of cognition (ie, episodic memory, perceptual speed and executive function, and MMSE) were evaluated.
Finally, individual vs combined neurodegenerative markers and their differential associations with cognitive decline were assessed.

Results

Sample Characteristics

Among 1038 participants (mean [SD] age, 77.1 [5.9] years; 615 Black [59.2%] and 423 White [40.8%]; 651 female [62.7%]), there was a mean (SD) of 12.8 (3.4) years of education, 343 individuals (33.0%) were APOE4 carriers, and 695 individuals were non–APOE4 carriers (67.0%) (Table 1). There was a mean (SD) of 6.5 (3.9) years of follow-up. The distribution of APOE4 genotypes in the total sample, showing a low number of APOE4 homozygotes, is shown in eTable 1 in Supplement 1. At baseline, APOE4 carriers compared with non–APOE4 carriers were a year younger (mean [SD] age, 76.4 [5.4] years vs 77.4 [6.1] years; P = .02) and had a greater proportion of Black older adults (226 Black participants [66.X%] vs 389 Black participants (56.X%]; P = .003) but had comparable education levels and global cognition scores. APOE4 carriers also had higher mean (SD) baseline levels of 2 neurodegenerative biomarkers compared with noncarriers: 0.82 (2.48) pg/mL vs 0.67 (3.03) pg/mL for t-tau, for a difference of 0.15 pg/mL (P = .002), and 277.8 (195.6) pg/mL vs 251.4 (178.0) pg/mL for GFAP, for a difference of 26 pg/mL (P = .03). Conversely, NfL levels were not significantly different between APOE4 carriers and noncarriers (Table 1; eFigure 2 in Supplement 1).

Interactions Between APOE4 Carrier Status and Associations of Biomarkers With Cognitive Decline

Linear Association

There were 3-way interactions in t-tau and GFAP models (Table 2).For APOE 4 × time × t-tau, the estimate (SD) was −0.03 (0.02) (P = .046), and for APOE 4 × time × GFAP, the estimate (SD) was −0.07 (0.03) (P = .02). Conversely, there were no significant 3-way interactions in the NfL model.

Associations of Biomarker Tertiles

Similar to the linear associations, when each biomarker was operationalized as tertiles, 3-way interaction terms remained significant for both tertiles of t-tau and GFAP (eTable 4 in Supplement 1). Contrary to the linear association, the 3-way interaction model with the middle tertile NfL was significant; for APOE 4 × time × middle tertile NfL, the estimate (SD) was −0.04 (0.02) (P = .006); for the upper tertile, the outcome was not significant (−0.03 [0.02]; P = .07).

Associations of Biomarker Levels With Cognitive Decline by APOE4 Carrier Status

Linear Association

Within groups, the rate of cognitive decline was not accelerated in non–APOE4 carriers but was accelerated among APOE4 carriers per unit increase in t-tau, with an estimate (SD) of −0.04 (0.02) per year (P = .02) (eTable 2 in Supplement 1). Similarly, the rate of cognitive decline was not accelerated in non–APOE4 carriers, but it was accelerated among APOE4 carriers per unit increase in NfL, with an estimate (SD) of −0.06 (0.03) per year (P = .02) (Table 3). For GFAP, as with the other 2 biomarkers, the rate of cognitive decline was not accelerated in non–APOE4 carriers, but it was accelerated among APOE4 carriers per unit increase in GFAP, with an estimate (SD) of −0.12 (0.03) per year (P < .001) (eTable 2 in Supplement 1).

Associations of Biomarker Tertiles

Within groups, the rate of cognitive decline was not accelerated in non–APOE4 carriers or APOE4 carriers with increased t-tau levels (eTable 2 and eFigure 3 in Supplement 1; Figure 1 and Figure 2). Conversely, the rate of cognitive decline was not accelerated in non–APOE4 carriers but was accelerated among APOE4 carriers with NfL in middle and upper tertiles, with an estimate (SD) of −0.05 (0.02) (P = .003) and −0.06 (0.02) (P = .004), respectively (Table 3; Figure 1 and Figure 2; eFigure 3 in Supplement 1). Similarly, the rate of cognitive decline was accelerated in only APOE4 carriers with upper tertile GFAP levels (eTable 2 and eFigure 3 in Supplement 1; Figure 1 and Figure 2). Models also examined cross-sectional associations between the 3 biomarkers and baseline cognitive score and found similar associations (Figure 1).

Open multimedia modal

Figure 1Open multimedia modal

Estimated Baseline Levels of Cognitive Function and Annual Rate of Cognitive Decline

. Estimated measures are given with 95% CIs by serum neurodegenerative biomarker tertile among APOE4 carriers vs noncarriers. Compared with noncarriers, APOE4 carriers had lower baseline levels of cognitive function and accelerated annual rates of cognitive decline, especially those with upper tertiles of biomarkers. Estimates are derived from the interaction of biomarkers with time since baseline in linear mixed-effects regression models adjusting for demographic and chronic health conditions. Whiskers indicate 95% CIs.

Open multimedia modal

Figure 2Open multimedia modal

Longitudinal Cognitive Decline in Lower and Upper Tertiles of Biomarkers

. Models were adjusted for demographic and chronic health conditions to examine the interaction between APOE4 carrier status and the combined associations of biomarkers with global cognitive function.

Sensitivity Analyses

Results for all sensitivity analyses are presented in eTables 3, 4, and 5 in Supplement 1. First, we added a sensitivity analysis on the total sample excluding individuals with mild cognitive impairment or cognitive impairment (eTable 3 in Supplement 1). Results and conclusions remain unchanged. Second, individual domains or tests of cognition were also assessed in addition to a composite cognitive measure (eTable 4 in Supplement 1). We found that associations were driven by memory domains given that there was a lack of associations with executive function regardless of APOE4 status.

Lastly, we performed another set of sensitivity analyses examining associations of multiple high levels of these neurodegenerative biomarkers with cognitive decline (eTable 5 in Supplement 1). Compared with participants with only a single elevated biomarker (the reference group), those with combined biomarker middle tertile (ie, all 3 biomarkers in the middle tertiles) who were also APOE4 carriers experienced an even faster rate of cognitive decline. Participants with combined biomarker upper tertile (ie, all 3 biomarkers in the upper tertiles) who were also APOE4 carriers experienced the fastest (ie, worst) rate of cognitive decline.

Discussion

In this cohort study, compared with non–APOE4 carriers, neurodegenerative serum biomarkers of t-tau, NfL, and GFAP were associated with lower baseline cognitive function levels and a faster cognitive decline rate among APOE4 carriers without dementia. Notably, when considering both baseline neurodegenerative serum biomarkers and the APOE4 carrier status, the difference in cognitive decline between APOE4 carriers and noncarriers increased with higher levels of t-tau, NfL, or GFAP. Taken together, our findings suggest that the combined elevated neurodegenerative burdens imposed by APOE4 and neurodegenerative biomarkers were associated with an accelerated rate of cognitive decline in a biracial population of community-dwelling older adults without dementia compared with associations of either APOE4 or neurodegenerative biomarker alone. Our finding is of great public health significance given that we showed that an individual with the same level of neurodegenerative serum biomarkers could have a very different risk based on their APOE4 carrier status. Hence, risk predictions based on these blood-based neurodegenerative biomarkers alone are likely insufficient. This prompts the need to concurrently account for APOE4 carrier status to more accurately monitor differences in risk of cognitive decline and, by extension, the risk of ADRD in future scientific investigations and clinical trials.

These findings suggest that the influence of the APOE4 allele may occur through several mechanisms. Given that blood t-tau level is a marker for neurodegeneration, a faster cognitive decline in APOE4 carriers with high levels of t-tau suggests the association of APOE4 with neurodegeneration, potentially amplifying tau’s pathological association with cognitive decline. The difference in cognitive decline in individuals with high t-tau levels was almost doubled in APOE4 carriers compared with noncarriers. These findings indicate that the role of APOE4 was more pronounced among individuals at a greater risk for neurodegeneration than those with a lower risk.

NfL is a nonspecific AD biomarker often associated with vascular aspects of dementia and indicative of the neurodegenerative processes of axonal injuries. Our findings demonstrated that APOE4 carriers with higher NfL levels had a significantly greater rate of cognitive decline than noncarriers with the same NfL levels. Hence, older adults free of dementia but with higher neuronal injury had a much greater risk of accelerated cognitive decline and of developing dementia in the presence of the APOE4 allele. As the biomarker levels increased, the rate of cognitive decline was substantially higher among APOE4 carriers than noncarriers. These findings indicate that the influence of the APOE4 allele may also occur through the cerebrovascular pathway of the disease, as observed through the NfL biomarker, which may be downstream of the neuronal injury pathway. Notably, a 2024 study found that almost all APOE4 homozygotes exhibited AD pathology and had significantly higher levels of AD biomarkers, including cerebrospinal fluid Aβ1-42 and p-tau, amyloid-β as measured by positron emission tomography, plasma p-tau, and plasma NfL, suggesting that the APOE4 may serve as a proxy for AD pathology. Hence, taken together with previous findings, our findings suggest that AD-associated pathology, conferred by APOE4, may interact with non–AD-specific pathology, conferred by NfL, in the association with accelerated cognitive decline.

Similarly, we found a significantly faster rate of cognitive decline in APOE4 carriers with higher blood-based GFAP levels. The APOE4 is primarily expressed by the astrocytes and upregulated by reactive microglia in the diseased state. The influence of the APOE4 allele on astrocytes and of microglial dysfunction markers on cognitive decline, especially among humans, could thus be clinically significant. Although mechanistic evidence has been shown in animal models, our findings are, to the best of our knowledge, the first to report such an effect modification of the APOE4 allele on neuroinflammation mediated by astrocytes and microglial dysfunction in the central nervous system in humans, specifically in a study of a large and diverse, biracial, community-based population of humans comprising more than 59% Black older adults and 63% females.

To test the robustness of and enhance the granularity in findings, we performed several sensitivity analyses. Taken together with the main analyses, these sensitivity analyses provided additional granularity to subgroup differences in the associations that were driven by cognitive domains and combinations of neurodegenerative burdens imposed by multiple biomarkers. Our data covered only 2 cognitive domains. Future studies with data on other cognitive domains should further examine variations in the associations. Notably, APOE4 carriers with all 3 biomarkers in the upper tertiles also experienced the fastest (ie, worst) rate of cognitive decline. This finding suggests that multiple neurodegenerative pathways may be additive and associated with an even greater increase in risk of cognitive decline than a single neurodegenerative pathway.

Strengths and Limitations

Our study has several notable strengths. First, our cohort comprised population-based community-dwelling older adults without dementia, which is much more representative of the larger at-risk population than clinical samples, which are typically restricted in demographics and more enriched in AD pathology. Hence, our findings may be more generalizable to the population of cognitively healthy and at-risk community-dwelling older adults, which forms a much larger portion of the older adult population. Second, our sample is unique in its biracial composition, with an overrepresentation of Black older adults, who are at a higher risk of developing AD and ADRD than White older adultss. This racially diverse population composition starkly contrasts with most cohort samples, which consist of predominantly White older adults, thus further enhancing the generalizability of our findings. Third, our cohort had a mean follow-up of 6.5 years, with frequent follow-up waves, enabling us to track cognitive decline over an extended period. Long-term follow-up is essential for tracking cognitive decline in older adults with dementia, among whom this decline could take many more years to manifest than among those with existing cognitive impairment or ADRD. Fourth, to our knowledge, this is the first population-based prospective study to have examined interactions and associations among 3 primary constructs (APOE4 carrier status, blood-based neurodegenerative biomarkers, and measures of cognitive decline), notably investigating 3 serum neurodegenerative biomarkers and the modifying interactions of APOE4 status with them in the same study, thus translating and validating animal findings in human samples. Furthermore, we excluded participants with clinician-diagnosed AD or MMSE-defined cases of probable AD or related dementias at baseline and further supplemented main analyses with sensitivity analyses excluding participants with mild cognitive impairment. This reduced the heterogeneity in biomarker levels and cognitive status, minimizing confounding effects imposed by existing AD comorbidities that would have otherwise skewed findings. Additionally, we performed a series of sensitivity analyses, testing the robustness of and enhancing the granularity in findings.

A few limitations may have also impacted our interpretations. Due to the low numbers of APOE4 homozygotes, we focused analyses on APOE4 carriers vs noncarriers. Owing to sample availabilities, we are restricted to analyzing serum instead of plasma or cerebrospinal fluid samples, focusing on 3 neurodegenerative blood-based biomarkers, and were not able to include other minority populations, such as American Indian, Asian, or Hispanic populations. Comparing blood biomarkers with postmortem neuropathological markers may provide additional details on the potential pathways of the APOE4 association with AD and related dementias. Notably, we assayed serum biomarkers before the advent and validations of various phospho(p)-tau assays; future studies examining these neurodegenerative and neuroinflammation biomarkers and the amyloid and latest p-tau species in plasma, which are more specific to AD pathways, could validate and provide further complementary evidence specific to the AD neurodegenerative process. Indeed, we are currently assessing amyloid β and p-tau biomarkers in our newly collected plasma samples.

Conclusions

Findings in this cohort study highlighted that in APOE4 carriers compared with noncarriers, higher levels of neurodegenerative blood biomarkers were associated with a faster rate of cognitive decline in a biracial population of community-dwelling older adults without dementia. These findings have potential public health and clinical significance. First, our and others’ findings have shown that Black older adults have a greater frequency of the APOE4 allele than White older adults and thus a greater risk of developing ADRD than those without the APOE4 allele. Hence, our samples composition of 59% Black older adults enables the generalizability of the findings. Second, biomarkers of neurodegeneration and neuroinflammation are used for staging and prognosis and as indicators of biological treatment effect. They also have broader applications beyond cognitive decline caused by AD and ADRD and can be broadly applicable to various neurodegenerative diseases. Hence, our findings provide complementary and broader neurodegenerative disease mechanisms in older adults without cognitive impairment of 2 major races in the US. Particularly, serum samples have been widely collected in health care systems. Hence, our findings, assessed using serum samples could enhance the broader adoption of blood-based biomarkers in clinics. Lastly, as ADRD treatments advance, increasing numbers of clinical studies and trials are poised to screen and recruit participants using blood-based biomarkers as a risk-stratification tool, particularly in primary and secondary prevention trials to slow cognitive decline. For such studies, our findings stress the importance of simultaneously including the APOE4 allele genotyping as a standard workflow given _APOE4_’s compounding effects in individuals with higher levels of blood-based neurodegenerative markers and association with an increased rate of cognitive decline and, by extension, risk of developing dementias.

References

1. Jack CR Jr, Andrews JS, Beach TG, et al. Revised criteria for diagnosis and staging of Alzheimer’s disease: Alzheimer’s Association Workgroup. Alzheimers Dement. 2024;20(8):5143–5169. doi:10.1002/alz.13859
2. Strittmatter WJ, Saunders AM, Goedert M, et al. Isoform-specific interactions of apolipoprotein E with microtubule-associated protein tau: implications for Alzheimer disease. Proc Natl Acad Sci U S A. 1994;91(23):11183–11186. doi:10.1073/pnas.91.23.11183
3. Liao F, Yoon H, Kim J. Apolipoprotein E metabolism and functions in brain and its role in Alzheimer’s disease. Curr Opin Lipidol. 2017;28(1):60–67. doi:10.1097/MOL.0000000000000383
4. Krasemann S, Madore C, Cialic R, et al. The TREM2-APOE pathway drives the transcriptional phenotype of dysfunctional microglia in neurodegenerative diseases. Immunity. 2017;47(3):566–581.e9. doi:10.1016/j.immuni.2017.08.008
5. Paasila PJ, Aramideh JA, Sutherland GT, Graeber MB. Synapses, microglia, and lipids in Alzheimer’s disease. Front Neurosci. 2022;15:778822. doi:10.3389/fnins.2021.778822
6. Ferrari-Souza JP, Bellaver B, Ferreira PCL, et al. APOEε4 potentiates amyloid β effects on longitudinal tau pathology. Nat Aging. 2023;3(10):1210–1218. doi:10.1038/s43587-023-00490-2
7. Dincer A, Chen CD, McKay NS, et al. APOE ε4 genotype, amyloid-β, and sex interact to predict tau in regions of high APOE mRNA expression. Sci Transl Med. 2022;14(671):eabl7646. doi:10.1126/scitranslmed.abl7646
8. Zetterberg H. Neurofilament light: a dynamic cross-disease fluid biomarker for neurodegeneration. Neuron. 2016;91(1):1–3. doi:10.1016/j.neuron.2016.06.030
9. Khalil M, Teunissen CE, Otto M, et al. Neurofilaments as biomarkers in neurological disorders. Nat Rev Neurol. 2018;14(10):577–589. doi:10.1038/s41582-018-0058-z
10. Kuhle J, Gaiottino J, Leppert D, et al. Serum neurofilament light chain is a biomarker of human spinal cord injury severity and outcome. J Neurol Neurosurg Psychiatry. 2015;86(3):273–279. doi:10.1136/jnnp-2013-307454
11. Neselius S, Brisby H, Theodorsson A, Blennow K, Zetterberg H, Marcusson J. CSF-biomarkers in Olympic boxing: diagnosis and effects of repetitive head trauma. PLoS One. 2012;7(4):e33606. doi:10.1371/journal.pone.0033606
12. Mielke MM, Hagen CE, Wennberg AMV, et al. Association of plasma total tau level with cognitive decline and risk of mild cognitive impairment or dementia in the Mayo Clinic Study on Aging. JAMA Neurol. 2017;74(9):1073–1080. doi:10.1001/jamaneurol.2017.1359
13. Mattsson N, Andreasson U, Zetterberg H, Blennow KAlzheimer’s Disease Neuroimaging Initiative. Association of plasma neurofilament light with neurodegeneration in patients with Alzheimer disease. JAMA Neurol. 2017;74(5):557–566. doi:10.1001/jamaneurol.2016.6117
14. Mattsson N, Cullen NC, Andreasson U, Zetterberg H, Blennow K. Association between longitudinal plasma neurofilament light and neurodegeneration in patients with Alzheimer disease. JAMA Neurol. 2019;76(7):791–799. doi:10.1001/jamaneurol.2019.0765
15. Mielke MM, Syrjanen JA, Blennow K, et al. Plasma and CSF neurofilament light: relation to longitudinal neuroimaging and cognitive measures. Neurology. 2019;93(3):e252–e260. doi:10.1212/WNL.0000000000007767
16. Preische O, Schultz SA, Apel A, et alDominantly Inherited Alzheimer Network. Serum neurofilament dynamics predicts neurodegeneration and clinical progression in presymptomatic Alzheimer’s disease. Nat Med. 2019;25(2):277–283. doi:10.1038/s41591-018-0304-3
17. Verberk IMW, Laarhuis MB, van den Bosch KA, et al. Serum markers glial fibrillary acidic protein and neurofilament light for prognosis and monitoring in cognitively normal older people: a prospective memory clinic-based cohort study. Lancet Healthy Longev. 2021;2(2):e87–e95. doi:10.1016/S2666-7568(20)30061-1
18. Grasset L, Bouteloup V, Cacciamani F, et alMEMENTO study group. Associations between blood-based biomarkers and cognitive and functional trajectories among participants of the MEMENTO cohort. Neurology. 2024;102(9):e209307. doi:10.1212/WNL.0000000000209307
19. Raffin J. Does physical exercise modify the pathophysiology of Alzheimer’s disease in older persons? JAR Life. 2024;13:77–81. doi:10.14283/jarlife.2024.11
20. Jack CR Jr, Wiste HJ, Algeciras-Schimnich A, et al. Comparison of plasma biomarkers and amyloid PET for predicting memory decline in cognitively unimpaired individuals. Alzheimers Dement. 2024;20(3):2143–2154. doi:10.1002/alz.13651
21. Chatterjee P, Pedrini S, Doecke JD, et alAIBL Research Group. Plasma Aβ42/40 ratio, p-tau181, GFAP, and NfL across the Alzheimer’s disease continuum: a cross-sectional and longitudinal study in the AIBL cohort. Alzheimers Dement. 2023;19(4):1117–1134. doi:10.1002/alz.12724
22. Fortea J, Pegueroles J, Alcolea D, et al. APOE4 homozygozity represents a distinct genetic form of Alzheimer’s disease. Nat Med. 2024;30(5):1284–1291. doi:10.1038/s41591-024-02931-w
23. Snellman A, Ekblad LL, Ashton NJ, et al. Head-to-head comparison of plasma p-tau181, p-tau231 and glial fibrillary acidic protein in clinically unimpaired elderly with three levels of _APOE4_-related risk for Alzheimer’s disease. Neurobiol Dis. 2023;183:106175. doi:10.1016/j.nbd.2023.106175
24. Sun Y, Wu S, Bu G, et al. Glial fibrillary acidic protein-apolipoprotein E (apoE) transgenic mice: astrocyte-specific expression and differing biological effects of astrocyte-secreted apoE3 and apoE4 lipoproteins. J Neurosci. 1998;18(9):3261–3272. doi:10.1523/JNEUROSCI.18-09-03261.1998
25. Rajan KB, Weuve J, Barnes LL, McAninch EA, Wilson RS, Evans DA. Population estimate of people with clinical Alzheimer’s disease and mild cognitive impairment in the United States (2020-2060). Alzheimers Dement. 2021;17(12):1966–1975. doi:10.1002/alz.12362
26. Quiroz YT, Solis M, Aranda MP, et al. Addressing the disparities in dementia risk, early detection and care in Latino populations: highlights from the second Latinos & Alzheimer’s Symposium. Alzheimers Dement. 2022;18(9):1677–1686. doi:10.1002/alz.12589
27. Evans DA, Bennett DA, Wilson RS, et al. Incidence of Alzheimer disease in a biracial urban community: relation to apolipoprotein E allele status. Arch Neurol. 2003;60(2):185–189. doi:10.1001/archneur.60.2.185
28. Rajan KB, Wilson RS, Weuve J, Barnes LL, Evans DA. Cognitive impairment 18 years before clinical diagnosis of Alzheimer disease dementia. Neurology. 2015;85(10):898–904. doi:10.1212/WNL.0000000000001774
29. Rajan KB, Weuve J, Barnes LL, Wilson RS, Evans DA. Prevalence and incidence of clinically diagnosed Alzheimer’s disease dementia from 1994 to 2012 in a population study. Alzheimers Dement. 2019;15(1):1–7. doi:10.1016/j.jalz.2018.07.216
30. Rajan KB, Aggarwal NT, McAninch EA, et al. Remote blood biomarkers of longitudinal cognitive outcomes in a population study. Ann Neurol. 2020;88(6):1065–1076. doi:10.1002/ana.25874
31. Rajan KB, McAninch EA, Aggarwal NT, et al. Longitudinal changes in blood biomarkers of clinical Alzheimer disease in a biracial population sample. Neurology. 2023;100(8):e874–e883. doi:10.1212/WNL.0000000000201289
32. Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31(3):545–548.
33. Wenham PR, Price WH, Blandell G. Apolipoprotein E genotyping by one-stage PCR. Lancet. 1991;337(8750):1158–1159. doi:10.1016/0140-6736(91)92823-K
34. Rajan KB, Barnes LL, Wilson RS, Weuve J, McAninch EA, Evans DA. Blood pressure and risk of incident Alzheimer’s disease dementia by antihypertensive medications and APOE ε4 allele. Ann Neurol. 2018;83(5):935–944. doi:10.1002/ana.25228
35. Rajan KB, Wilson RS, Skarupski KA, Mendes de Leon CF, Evans DA. Gene-behavior interaction of depressive symptoms and the apolipoprotein E {varepsilon}4 allele on cognitive decline. Psychosom Med. 2014;76(2):101–108. doi:10.1097/PSY.0000000000000029
36. Smith A. Symbol Digit Modalities Test (SDMT) Manual (Revised). Western Psychological Services; 1982.
37. Folstein MF, Robins LN, Helzer JE. The mini-mental state examination. Arch Gen Psychiatry. 1983;40(7):812. doi:10.1001/archpsyc.1983.01790060110016
38. Wilson RS, Beckett LA, Barnes LL, et al. Individual differences in rates of change in cognitive abilities of older persons. Psychol Aging. 2002;17(2):179–193. doi:10.1037/0882-7974.17.2.179
39. Wilson RS, Bennett DA, Beckett LA, et al. Cognitive activity in older persons from a geographically defined population. J Gerontol B Psychol Sci Soc Sci. 1999;54(3):155–160. doi:10.1093/geronb/54B.3.P155
40. US Census Bureau. Census of Population and Housing, 1990, Population and Housing Characteristics for Census Tracts and Block Numbering Areas. US Department of Commerce Economics and Statistics Administration; 1993.
41. Cornoni-Huntley J, Ostfeld AM, Taylor JO, et al. Established populations for epidemiologic studies of the elderly: study design and methodology. Aging (Milano). 1993;5(1):27–37. doi:10.1007/BF03324123
42. R Core Team. The R project for statistical computing. Accessed April 1, 2025. https://www.r-project.org/
43. Fernández-Calle R, Konings SC, Frontiñán-Rubio J, et al. APOE in the bullseye of neurodegenerative diseases: impact of the APOE genotype in Alzheimer’s disease pathology and brain diseases. Mol Neurodegener. 2022;17(1):62. doi:10.1186/s13024-022-00566-4
44. Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic review and meta-analysis. Lancet Neurol. 2016;15(7):673–684. doi:10.1016/S1474-4422(16)00070-3
45. Bridel C, Verberk IMW, Heijst JJA, Killestein J, Teunissen CE. Variations in consecutive serum neurofilament light levels in healthy controls and multiple sclerosis patients. Mult Scler Relat Disord. 2021;47:102666. doi:10.1016/j.msard.2020.102666
46. Bridel C, van Wieringen WN, Zetterberg H, et aland the NFL Group. Diagnostic value of cerebrospinal fluid neurofilament light protein in neurology: a systematic review and meta-analysis. JAMA Neurol. 2019;76(9):1035–1048. doi:10.1001/jamaneurol.2019.1534
47. Ma W, Zhang J, Xu J, Feng D, Wang X, Zhang F. Elevated levels of serum neurofilament light chain associated with cognitive impairment in vascular dementia. Dis Markers. 2020;2020(1):6612871. doi:10.1155/2020/6612871
48. Duering M, Konieczny MJ, Tiedt S, et al. Serum neurofilament light chain levels are related to small vessel disease burden. J Stroke. 2018;20(2):228–238. doi:10.5853/jos.2017.02565
49. Shi Y, Yamada K, Liddelow SA, et alAlzheimer’s Disease Neuroimaging Initiative. ApoE4 markedly exacerbates tau-mediated neurodegeneration in a mouse model of tauopathy. Nature. 2017;549(7673):523–527. doi:10.1038/nature24016
50. Belloy ME, Andrews SJ, Le Guen Y, et al. APOE genotype and Alzheimer disease risk across age, sex, and population ancestry. JAMA Neurol. 2023;80(12):1284–1294. doi:10.1001/jamaneurol.2023.3599
51. Rajan KB, Mcaninch EA, Wilson RS, Dhana A, Evans-Lacko S, Evans DA. Statin initiation and risk of incident Alzheimer disease and cognitive decline in genetically susceptible older adults. Neurology. 2024;102(7):e209168. doi:10.1212/WNL.0000000000209168
52. Ng TKS, Beck T, Desai P, et al. Prospective associations of interleukin-6 and APOE allele with cognitive decline in biracial community-dwelling older adults: the Chicago Health and Aging Project (CHAP). Alzheimers Dement (Amst). 2024;16(4):e70002. doi:10.1002/dad2.70002

eFigure 1. Flow diagram indicating participant inclusion and flow

eFigure 2. Box plots of serum neurodegenerative biomarkers stratified by race and APOE4 carrier status

eFigure 3. An alternative way of presentation: longitudinal cognitive decline in the lower and upper tertiles of serum neurodegenerative biomarkers among APOE4 carriers vs noncarriers

eTable 1. Distribution of APOE4 genotypes in the total sample ( N= 1038)

eTable 2. Associations of concentrations of biomarkers with annual rate of cognitive decline, stratified by APOE4 carriers status

eTable 3. Sensitivity analyses on subset of cognitively unimpaired sample with baseline MCI and AD cases excluded: associations of concentrations of biomarkers with annual rate of cognitive decline, stratified by APOE4 carriers status

eTable 4. Total and stratified samples: 3-way and 2-way interaction models

eTable 5. Spearman correlation coefficients of the biomarkers, cross-tabulations of biomarker tertiles, distributions of different combinations of biomarker tertiles, and total and stratified models examining combined biomarker tertiles

Data Sharing Statement