Prospective associations of interleukin‐6 and APOE allele with cognitive decline in biracial community‐dwelling older adults: The Chicago Health and Aging Project (CHAP)

Abstract

INTRODUCTION

It is unclear whether inflammation, that is, high interleukin‐6 (IL‐6) levels, and genetic risk, that is, apolipoprotein E (APOE) ε4 allele, have a compounding effect on cognitive decline (CD).

METHODS

We analyzed a subset of participants from the longitudinal cohort study, Chicago Health and Aging Project, comprising 1120 biracial community‐dwelling older adults (60% Black and 62% women), and mean follow‐up = 6.4 years. We ran adjusted mixed‐effects models on2 longitudinal CD.

RESULTS

In APOE ε4 carriers, higher serum IL‐6 was not associated with the rate of CD (β = –0.0091 [standard deviation (SD) = 0.0165, p = 0.5800]). Conversely, in non‐ε4 carriers, compared to the lower tertile, those with the upper tertile of serum IL‐6 levels experienced significantly accelerated CD (β = –0.0257 [SD = 0.0084, p = 0.0023]).

DISCUSSION

Even without the largest genetic risk factor for late‐onset Alzheimer's disease/Alzheimer's disease and related dementias (AD/ADRD), elevated serum IL‐6 still accelerate the rate of CD in non‐APOE ε4 carriers. Hence, interventions ameliorating inflammation may prevent AD/ADRD.

Highlights

Interleukin‐6 (IL‐6) and the apolipoprotein E (APOE) ε4 allele have been separately associated with an increased risk for cognitive decline, but their interaction remains unclear. In ε4 carriers, IL‐6 was not associated with cognitive decline. However, even without the biggest genetic risk factor for Alzheimer's disease (AD), that is, APOE ε4, elevated serum IL‐6 still could confer accelerated rate of cognitive decline, with a detrimental effect half of that imposed by APOE ε4 alone. We found no racial differences in these associations. These findings contribute complementary evidence on non–APOE ε4‐dependent and non‐AD biological pathways through which cognitive decline can still be accelerated in non‐APOE ε4 carriers and highlight a specific subgroup of older adults who are at a higher risk of AD and thus may benefit from anti‐inflammatory interventions.

BACKGROUND

Alzheimer's disease (AD) is a complex and multi-etiological disease, conferred by risk factors including genetics and inflammation. Apart from the canonical neuropathological hallmarks of the accumulation of amyloid beta (Aβ) plaques and tau protein tangles evidenced in the brain of patients with AD, neuroinflammation has gained traction in recent years as an understudied hallmark of AD.

Misfolded and aggregated proteins can trigger innate immune responses aimed at clearing plaques and tangles, leading to the release of inflammatory mediators, such as interleukin-6 (IL-6).^1,2 Among the triggering factors and aggravators, IL-6 is a pleiotropic cytokine produced in response to stressors and stimuli, including social–environmental stressors,^3,4 and Aβ plaques in the brain. ^1,2 IL-6 can trigger a cascade of molecular events that lead to the production of more inflammatory cytokines and further the activation of microglia and astrocytes. Hence, apart from its roles in acute responses to stressors, IL-6 also controls chronic inflammatory responses.⁵ Furthermore, chronic low-grade inflammation can lead to dysregulations in the intricate molecular control mechanisms, contributing to neuronal inflammation and eventual neuronal cell death. Hence, chronic low-grade inflammation caused by IL-6 predisposes cognitively normal older adults to hasten cognitive decline and in the long term developing AD.⁶

The apolipoprotein E (APOE) ε4 allele is the single largest genetic risk factor for late-onset AD/AD and related dementias (ADRD), with APOE ε4 carriers having an increased risk of developing late-onset AD/ADRD. Notably, a recent study showed that compared to APOE ε3 homozygotes, almost all APOE ε4 homozygotes had significantly higher levels of AD biomarkers.⁷ Among several mechanisms, APOE ε4 has been postulated to contribute to AD pathogenesis via eliciting and aggravating neuroinflammation.⁸ Specifically, in the central nervous system, the glial cells, predominantly astrocytes, express the apoE protein. The apoE protein can exacerbate the inflammatory response to stressors, leading to even greater neuroinflammation and neuronal cell death. This hypothesis is substantiated by animal studies showing that humanized APOE ε4 mice had increased IL-6 levels. ^9,10

Both IL-6 and the APOE ε4 allele, independently, have been implicated in cognitive decline (CD) in humans. However, there is a scarcity of studies examining both IL-6 and APOE ε4 concurrently as risk factors for CD in humans, particularly investigating how APOE ε4 influences the effects of IL-6 on CD. The combined detrimental effects of IL-6 and APOE ε4 on CD could indicate an even greater neurodegenerative burden and neuronal damage than their independent effects. Therefore, further investigation is critical to disentangle the complex relationship among APOE ε4, IL-6, and CD.

African Americans, that is, Blacks, have a higher risk for and steeper increase in incident AD/ADRD in the United States in the next 40 years,^11–13 which could be in part attributed to Blacks having a higher frequency of the APOE ε4 allele (due to genetic inheritance/ancestry)¹⁴ and higher IL-6 levels, due to social–environmental stressors, including experiencing systemic racism.¹⁵ However, extant studies on the longitudinal associations between either the APOE ε4 allele or IL-6 and CD have predominantly focused on White Americans,^16,17 hindering generalizability to minorities, including Blacks. Furthermore, many blood-based biomarker studies recruited human subjects mostly from clinical settings, which have been demonstrated to exhibit distinct demographics and thus differential risk than community-dwelling older adults.

It is thus imperative to investigate potential interactions between the APOE ε4 allele and IL-6 in a more racially diverse and community-dwelling population. In this paper, analyzing our biracial cohort comprising solely community-dwelling older adults, we first tested whether the genetic predisposition of AD, through the presence of the APOE ε4 allele, increases the rate of CD in people with chronic, low-grade inflammation, as indicated by blood-based IL-6 levels. Due to established racial differences in the exposure variables and the potential differences in the associations, we additionally performed a priori exploratory race-stratified analyses.

METHODS

Study design, setting, and population

The Chicago Health and Aging Project (CHAP) is a prospective population-based cohort study designed to assess bio-psycho-social and structural risk factors for age-related chronic conditions in older adults, with a specific focus on AD/ADRD.¹⁸ Briefly, we started recruitment in 1993, enrolling participants based on four geographically defined Chicago neighborhoods that had substantial proportions of non-Hispanic Black and White residents. The only two inclusion criteria for the cohort were living in the study catchment area and having a minimum age of 65 years at enrollment.

During in-home assessments, research assistants administered questionnaires and neurocognitive tests every 3 years and up to six times throughout the study period. Blood specimen collection and processing have previously been described in detail.¹⁹ In brief, over the course of the study, oversampling for Black participants, approximately one third of CHAP participants were selected for a clinical assessment for AD when they also provided blood samples. Upon laboratory processing, we stored the samples in −80°C freezers within 3 hours of collection. There were 1327 participants with serum samples assayed. For this paper, we analyzed data from 1120 participants with baseline levels of IL-6 assayed and APOE ε4 allele genotyped and at least two follow-up cognitive testings.

APOE carrier status

The APOE ε4 carrier status was ascertained with two single nucleotide polymorphisms (SNPs): rs7412 and rs429358.²⁰ The genotyping was performed at the Broad Institute Center for Genotyping using the hME Sequenom MassARRAY platform. Genotyping call rates were 100% for SNP rs7412 and 99.8% for SNP rs429358. Both SNPs were in Hardy–Weinberg equilibrium with p values of 0.0833 and 0.7925, respectively. Based on these two SNPs, we created an indicator variable for participants with and without the APOE ε4 allele.²¹ APOE ε4 carriers were participants with at least one copy of the APOE ε4 allele, that is, ε2/ε4, ε3/ε4, and two copies of the APOE ε4 allele, that is, ε4/ε4. The second group, the non-APOE ε4 carriers, were those without any copy of the APOE ε4 allele, that is, ε2/ε2, ε2/ε3, and ε3/ε3 genotypes.

Quantification of serum IL-6 levels

At the end of the in-home assessments, research assistants transported the blood samples on dry ice to a Rush biorepository. After centrifugation, the serum was extracted, aliquoted and bio-banked in −80°C freezers. Previously unthawed samples were transported in dry ice for biomarker detection at Quanterix. The serum IL-6 levels were measured using CorPlex Human Cytokine 10-plex panel 1 assay (CPX). SIMOA and HD-X were used for IL-6 quantification. IL-6 has a lower limit of detection of 0.037 pg/mL (range 0.008–0.067 pg/mL) and coefficient of variation of 11%.²²

RESEARCH IN CONTEXT

• Systematic review: Interleukin-6 (IL-6) and the apolipoprotein E (APOE) ε4 allele, separately, have been associated with an increased risk for cognitive decline in humans. Animal studies have shown that mice with the APOE ε4 allele have higher IL-6 levels. However, it remains unclear whether the increased risk for cognitive decline is accelerated in APOE ε4 carriers, and whether these associations differ by race. Hence, we attempted to address, for the first time, these four important AD risk factors, namely APOE ε4 status (i.e., carriers vs. non-carriers), serum IL-6 levels, race, and cognitive decline, within a single longitudinal human cohort study.

• Interpretations: In APOE ε4 carriers, surprisingly, IL-6 was not associated with cognitive decline. However, even without the biggest genetic risk factor for Alzheimer's disease (AD), that is, APOE ε4, elevated serum IL-6 still could confer accelerated rate of cognitive decline, with a detrimental effect half of that imposed by APOE ε4 alone. On the contrary, non-APOE ε4 carriers without elevated serum IL-6 levels had the slowest decline and thus the lowest risk of cognitive decline in the long term. We found no racial differences in these associations.

• Future directions: These findings may implicate precision medicine, highlighting a specific subgroup of older adults who are at a higher risk of cognitive decline and AD, and thus may benefit from anti-inflammatory interventions. Specifically, even without the APOE ε4 allele, which constitutes most of the population, heightened systemic inflammation still accelerate cognitive decline. In light of recent findings showing overwhelming biological penetrance of the APOE ε4 allele on canonical AD biomarkers and pathologies, our finding is significant in contributing complementary evidence on non–APOE ε4-dependent and non-AD biological pathways through which cognitive decline can still be accelerated in non-APOE ε4 carriers. Replications of our findings presented here, especially in cohorts including more diverse racial groups, for example, Hispanics, is needed.

Global and domain-specific tests of cognition

The four tests included two tests of episodic memory, one test of executive function, and the Mini-Mental State Examination. The rationale for selecting these four tests was due to the simplicity and feasibility of administration during in-home assessments for a population-wide epidemiological study. After centering and scaling each test to the cohort's baseline means and standard deviations (SDs), we derived standardized scores for both a global measure of cognitive function and domain-specific cognitive tests. Individual domain-specific cognitive tests were based on two tests for memory and one executive function–based speed test. The global measure of cognitive function was calculated by averaging the four cognitive tests, which we described in detail in our previous publication.^21,23,24

Covariates

Pertinent covariates collected during the baseline assessment were controlled for. They included the age at first blood sample collection (centered at 75 years), biological sex (males or females), race (non-Hispanic Black and non-Hispanic White), education (measured in the number of years of schooling completed, centered at 12 years), body mass index (BMI), and chronic health conditions (i.e., hypertension, diabetes, stroke, heart condition, cancer, and hip fracture). The centering for age and education was decided using their approximate means.

Statistical analyses

Serum IL-6 levels were positively skewed and above zero. Hence, we made a log₁₀ transformation with geometric mean and its 95% confidence interval. Descriptive comparisons between non-APOE ε4 carriers and APOE ε4 carriers were based on t tests for untransformed characteristics, chi-square test for frequencies, and Wilcoxon rank test for serum IL-6 levels. All regression models were adjusted for age at first blood assay (centered at 75 years), education (centered at 12 years), male sex, non-Hispanic Black race/ethnicity, BMI, and common chronic health conditions.

We used a linear mixed effects regression model to examine the association of baseline serum IL-6 levels with longitudinal change in cognitive function. These mixed models included random intercepts and slopes and allow us to measure both within- and across-participant variability. Time since baseline blood assessment was measured in years, capturing the annual rate of change in cognitive function over time. We operationalized serum IL-6 levels in two different ways. First, we used log-transformed continuous biomarker levels. Second, we divided serum IL-6 levels into tertiles, to examine the association of higher tertiles of IL-6 levels with CD.

We also performed two sets of analyses; we first added an interaction term in a three-way interaction model (i.e., log₁₀ transformed continuous IL-6 levels/tertiles x APOE ε4 x time). Upon detecting a significant three-way interaction effect, we then performed stratifications of the total sample by the APOE ε4 allele and examined two-way interaction models (i.e., log₁₀ transformed continuous IL-6 levels/ tertiles x time).

Due to established racial differences in the exposure variables and thus, the potential differences in associations, we additionally performed exploratory race-based stratified analyses repeated for all statistical models, while considering the sample size and power limitations.

All regression models were performed using SAS 9.4, and graphical representations were performed with the R program.²⁵ A P value of 0.05 was considered statistically significant.

RESULTS

Baseline demographics

Of the total 1120 participants (APOE ε4 carriers = 378 and non-APOE ε4 carriers = 742), most baseline characteristics (see Table 1) did not show a significant difference between the carriers and non-carriers, except for race (higher Black for APOE ε4 carriers; n = 247, 65% vs. lower Black in non-APOE ε4 carriers; n = 424, 57%), global cognitive function (mean [SD] = 0.29 [0.62] in non-APOE ε4 carriers and 0.17 [0.64] in APOE ε4 carriers), and time in study (mean [SD] = 6.7 [3.8] in non-APOE ε4 carriers and in 5.9 [3.4] in APOE ε4 carriers).

TABLE 1 Baseline demographics by APOE ε4 carrier status.

Variable	Total sample N = 1120a	Non-APOE ε4 carriers N = 742	APOE ε4 carriers N = 378	p valueb
Age (years), mean (SD)	77.2 (6.0)	77.5 (6.1)	76.8 (5.6)	0.061
Education (years), mean (SD)	12.6 (3.5)	12.7 (3.5)	12.5 (3.4)	0.46
Female, n (%)	699 (62)	466 (63)	233 (62)	0.75
Black, n (%)	671 (60)	424 (57)	247 (65)	0.010*
Chronic conditions, mean (SD)	1.3 (1.0)	1.3 (1.0)	1.3 (1.0)	0.57
BMI (kg/m2), mean (SD)	27.6 (5.5)	27.7 (5.6)	27.4 (5.3)	0.34
Global cognitive function, mean (SD)	0.25 (0.63)	0.29 (0.62)	0.17 (0.64)	0.006**
Time in study (years), mean (SD)	6.4 (3.7)	6.7 (3.8)	5.9 (3.4)	<0.001***
IL-6 (pg/mL), median (IQR)	2.6 (1.7–4.4)	2.6 (1.7–4.4)	2.5 (1.6–4.3)	0.44
IL-6 tertile, n (%)				0.21
1	379 (34)	239 (32)	140 (37)
2	382 (34)	264 (36)	118 (31)
3	359 (32)	239 (32)	120 (32)

Supplementary text and Table S1a in supporting information show group differences (and lack thereof) in demographics in the tertile subgroups.

2-way interaction models

Tables S2a&b in supporting information show significant two-way interaction between log₁₀ IL-6 levels and time, with beta estimate (SD, p value) = −0.0211 (0.0092, 0.0223), and between IL-6 tertiles and time −0.0167 (0.0077, 0.0302).

3-way interaction models

Continuous log₁₀-transformed IL-6 levels

There was no statistically significant three-way interaction among log₁₀ IL-6 levels, APOE ε4 carrier status, and time, with 0.0316 (0.0209, 0.1307; Table 2A).

TABLE 2A Total sample: three-way interaction model with the interaction term “Log₁₀ transformed IL-6 levels x APOE ε4 x time” and association with 12-year cognitive decline.

	Total sample estimate (SD, p value) N = 1120
Log10 IL-6, pg/mL	−0.0168 (0.0598, 0.7784)
APOE ε4	−0.0526 (0.0576, 0.3611)
Time	−0.0219 (0.0084, 0.0087*)
Log10 IL-6 x time	−0.0297 (0.0108, 0.0062**)
Log10 IL-6 x APOE ε4	−0.0693 (0.1032, 0.5020)
APOE ε4 x time	−0.0572 (0.0110, < 0.0001***)
Log10 IL-6 x APOE ε4 x time	0.0316 (0.0209, 0.1307)

IL-6 tertiles

There was no statistically significant three-way interaction among the middle tertile of IL-6 levels, APOE ε4 carrier status, and time, with 0.0117 (0.0162, 0.4699) compared to those in the lower tertile (Table 2B).

TABLE 2B Total sample: three-way interaction model with the interaction term “IL-6 tertiles x APOE ε4 x time” and association with 12-year cognitive decline.

	Total sample estimate (SD, p value) N = 1120
IL-6 tertile 1	Reference
IL-6 tertile 2	−0.0333 (0.0491, 0.4975)
IL-6 tertile 3	−0.0387 (0.0505, 0.4438)
APOE ε4	−0.0810 (0.0584, 0.1658)
Time	−0.0247 (0.0087, 0.0047*)
IL-6 tertile 1 x time	Reference
IL-6 tertile 2 x time	−0.0040 (0.0090, 0.6594)
IL-6 tertile 3 x time	−0.0270 (0.0094, 0.0040**)
IL-6 tertile 1 x APOE ε4	Reference
IL-6 tertile 2 x APOE ε4	0.0245 (0.0845, 0.7722)
IL-6 tertile 3 x APOE ε4	−0.0373 (0.0848, 0.6599)
APOE ε4 x time	−0.0571 (0.0110, < 0.0001***)
IL-6 tertile 1 x APOE ε4 x time	Reference
IL-6 tertile 2 x APOE ε4 x time	0.0117 (0.0162, 0.4699)
IL-6 tertile 3 x APOE ε4 x time	0.0331 (0.0166, 0.0457*)

There was a statistically significant three-way interaction among the upper tertile IL-6 levels, APOE ε4 carrier status, and time, with 0.0331 (0.0166, 0.0457) compared to the lower tertile.

Models stratified by APOE ε4 carrier status

Log₁₀-transformed IL-6 levels

Regardless of APOE ε4 carrier status, we detected no significant association of baseline serum IL-6 levels with baseline level of global cognition. APOE ε4 carriers = −0.0955 (0.0878, 0.2775); non-APOE ε4 carriers = −0.0147 (0.0593, 0.8045; Table 3A).

TABLE 3A Stratified samples: two-way interaction model with the interaction term “Log₁₀-transformed IL-6 levels x time” and association with 12-year cognitive decline, stratified by APOE ε4 carrier status.

	APOE ε4 carriers estimate (SD, p value) N = 378	Non-APOE ε4 carriers estimate (SD, p value) N = 742
Log10 IL-6, pg/mL	−0.0955 (0.0878, 0.2775)	−0.0147 (0.0593, 0.8045)
Time	−0.0974 (0.0176, < 0.0001***)	−0.0155 (0.0081, 0.0565)
Log10 IL-6 x time	0.0044 (0.0211, 0.8368)	−0.0275 (0.0097, 0.0047**)

Longitudinally, compared to non-APOE ε4 carriers, APOE ε4 carriers had a significantly accelerated average annual rate of CD, with −0.0155 (0.0081, 0.0565) and −0.0974 (0.0176, 0.0000), respectively.

For examining longitudinal associations and making within-group comparisons, for every one unit increase in log₁₀-transformed IL-6 levels, non-APOE ε4 carriers with elevated IL-6 had almost triple the rate of CD, with an additional −0.0275 (0.0097, 0.0047). In contrast, log IL-6 levels did not accelerate the rate of CD in APOE ε4 carriers, 0.0044 (0.0211, 0.8368).

Comparing groups, the magnitude of accelerated CD is noteworthy: The average rate of CD was approximately half that of APOE ε4 carriers.

IL-6 tertiles

In non-APOE ε4 carriers with the lowest tertile of IL-6 level, the rate of CD was −0.0177 (0.0084, 0.0363; Table 3B). Compared to the reference group, non-APOE ε4 carriers with elevated IL-6, that is, those having the upper tertile of IL-6, had an additional rate of decline of −0.0257 (0.0084, 0.0023). Interestingly, non-APOE ε4 carriers with moderate IL-6 at the middle tertile, −0.0036 (0.0081, 0.6533), did not experience accelerated CD. Neither the middle nor the upper tertiles were significant in the APOE ε4 carriers.

TABLE 3B Stratified samples: two-way interaction model with the interaction term “IL-6 tertile x time” and association with 12-year cognitive decline, stratified by APOE ε4 carrier status.

	APOE ε4 carriers estimate (SD, p value) N = 378	Non-APOE ε4 carriers estimate (SD, p value) N = 742
IL-6 tertile 1	Reference	Reference
IL-6 tertile 2	−0.0219 (0.0719, 0.7604)	−0.0246 (0.0487, 0.6130)
IL-6 tertile 3	−0.0930 (0.0712, 0.1923)	−0.0343 (0.0502, 0.4951)
Time	−0.1005 (0.0173, < 0.0001a)	−0.0177 (0.0084, 0.0363c)
IL-6 tertile 1 x time	Reference	Reference
IL-6 tertile 2 x time	0.0078 (0.0163, 0.6323)	−0.0036 (0.0081, 0.6533)
IL-6 tertile 3 x time	0.0091 (0.0165, 0.5800)	−0.0257 (0.0084, 0.0023b)

Comparing the groups, similar to our findings using the operationalization of IL-6 levels on a continuous log₁₀-transformed scale, the accelerated rate of CD in non-APOE ε4 carriers with upper IL-6 tertile was almost half that of APOE ε4 carriers (Figure 1).

[IMAGE OMITTED. SEE PDF]

Exploratory race-stratified analyses

See supplementary text and Tables 4A, 4B, 4C, and 4D.

TABLE 4A Total and stratified samples: three-way and two-way interaction models with the interaction term “IL-6 tertiles x APOE ε4 x time” and association with 12-year cognitive decline, stratified a priori by race—Black subgroup.

	Total Black sample estimate (SD, p value)	Black APOE ε4 carriers estimate (SD, p value)	Black non-APOE ε4 carriers estimate (SD, p value)
	N = 671	N = 247	N = 424
IL-6 lower tertile	Reference	Reference	Reference
IL-6 middle tertile	−0.0689 (0.0638, 0.2807)	−0.0753 (0.0836, 0.3688)	−0.0632 (0.0651, 0.3317)
IL-6 upper tertile	−0.0188 (0.0684, 0.7837)	−0.1205 (0.0821, 0.1434)	−0.0144 (0.0699, 0.8369)
APOE ε4	−0.0316 (0.0748, 0.6730)	–	–
Time	−0.0208 (0.0115, 0.0700)	−0.0612 (0.0192, 0.0015**)	−0.0242 (0.0102, 0.0174*)
IL-6 lower tertile x time	Reference	Reference	Reference
IL-6 middle tertile x time	−0.0042 (0.0118, 0.7236)	0.0078 (0.0213,0.7147)	−0.0033 (0.0096, 0.7289)
IL-6 upper tertile x time	−0.0243 (0.0129, 0.0592)	−0.0041 (0.0213,0.8475)	−0.0228 (0.0107, 0.0333*)
IL-6 lower tertile x APOE ε4	Reference	Reference	Reference
IL-6 middle tertile x APOE ε4	−0.0029 (0.1058, 0.9784)	–	–
IL-6 upper tertile x APOE ε4	−0.0970 (0.1077, 0.3678)	–	–
APOE ε4 x time	−0.0505 (0.0143, 0.0004**)	–	–
IL-6 lower tertile x APOE ε4 x time	Reference	Reference	Reference
IL-6 middle tertile x APOE ε4 x time	0.0160 (0.0201, 0.4274)	–	–
IL-6 upper tertile x APOE ε4 x time	0.0199 (0.0209, 0.3411)	–	–

TABLE 4B Total and stratified samples: three-way and two-way interaction models with the interaction term “IL-6 tertiles x APOE ε4 x time” and association with 12-year cognitive decline, stratified a priori by race—White subgroup.

	Total White sample estimate (SD, p value) N = 449	White APOE ε4 carriers estimate (SD, p value) N = 131	White non-APOE ε4 carriers estimate (SD, p value) N = 318
IL-6 lower tertile	Reference	Reference	Reference
IL-6 middle tertile	0.0227 (0.0771, 0.7690)	0.0692 (0.1364, 0.6125)	0.0322 (0.0724, 0.6568)
IL-6 upper tertile	−0.0617 (0.0753, 0.4130)	−0.0506 (0.1382, 0.7152)	−0.0534 (0.0709, 0.4520)
APOE ε4	−0.1709 (0.0939, 0.0693)	–	–
Time	−0.0303 (0.0128, 0.0178*)	−0.1435 (0.0239, < 0.0001***)	−0.0159 (0.0132, 0.2295)
IL-6 lower tertile x time	Reference	Reference	Reference
IL-6 middle tertile x time	0.0002 (0.0146, 0.9882)	0.0090 (0.0271, 0.7410)	0.0007 (0.0140, 0.9588)
IL-6 upper tertile x time	−0.0299 (0.0143, 0.0370*)	0.0271 (0.0277, 0.3280)	−0.0286 (0.0137, 0.0374*)
IL-6 lower tertile x APOE ε4	Reference	Reference	Reference
IL-6 middle tertile x APOE ε4	0.0679 (0.1408, 0.6299)	–	–
IL-6 upper tertile x APOE ε4	0.0441 (0.1390, 0.7510)	–	–
APOE ε4 x time	−0.0700 (0.0178, 0.0001**)	–	–
IL-6 lower tertile x APOE ε4 x time	Reference	Reference	Reference
IL-6 middle tertile x APOE ε4 x time	0.0015 (0.0280, 0.9565)
IL-6 upper tertile x APOE ε4 x time	0.0597 (0.0281, 0.0339*)

TABLE 4C Total and stratified samples: three-way and two-way interaction models with the interaction term “Log₁₀-transformed IL-6 levels x APOE ε4 x time” and association with 12-year cognitive decline, stratified a priori by race—Black subgroup.

	Total Black sample estimate (SD, p value) N = 671	Black APOE ε4 carriers estimate (SD, p value) N = 247	Black non-APOE ε4 carriers estimate (SD, p value) N = 424
Log10 IL-6, pg/mL	0.0211 (0.0787, 0.7888)	−0.1192 (0.1003, 0.2360)	0.0198 (0.0804, 0.8053)
APOE ε4	0.0119 (0.0720, 0.8690)	–	–
Time	−0.0175 (0.0106, 0.0997)	−0.0594 (0.0201, 0.0032**)	−0.0210 (0.0095, 0.0271*)
Log10 IL-6 x time	−0.0278 (0.0147, 0.0577)	−0.0032 (0.0280, 0.9094)	−0.0246 (0.0121, 0.0425*)
Log10 IL-6 x APOE ε4	−0.1582 (0.1292, 0.2211)	–	–
APOE ε4 x time	−0.0528 (0.0140, 0.0002**)	–	–
Log10 IL-6 x APOE ε4 x time	0.0297 (0.0266, 0.2640)	–	–

TABLE 4D Total and stratified samples: three-way and two-way interaction models with the interaction term “Log₁₀-transformed IL-6 levels x APOE ε4 x time” and association with 12-year cognitive decline, stratified a priori by race—White subgroup.

	Total White sample estimate (SD, p value) N = 449	White APOE ε4 carriers estimate (SD, p value) N = 131	White non-APOE ε4 carriers estimate (SD, p value) N = 318
Log10 IL-6, pg/mL	−0.0594 (0.0925, 0.5209)	−0.0470 (0.1732, 0.7867)	−0.0476 (0.0873, 0.5856)
APOE ε4	−0.1568 (0.0963, 0.1043)	–	–
Time	−0.0274 (0.0121, 0.0244c)	−0.1380 (0.0237, < 0.0001a)	−0.0130 (0.0126, 0.3022)
Log10 IL-6 x time	−0.0320 (0.0170, 0.0593)	0.0075 (0.0342, 0.8265)	−0.0288 (0.0162, 0.0764)
Log10 IL-6 x APOE ε4	0.0471 (0.1729, 0.7853)	–	–
APOE ε4 x time	−0.0694 (0.0182, 0.0001b)	–	–
Log10 IL-6 x APOE ε4 x time	0.0414 (0.0345, 0.2293)	–	–

DISCUSSION

Our study presented several notable findings. First, in APOE ε4 carriers, serum IL-6 was not associated with CD. However, it is worth noting that in the non-APOE ε4 carriers, compared to those having lower serum IL-6 levels, those having elevated serum IL-6 levels experienced significantly accelerated CD. Specifically, compared to non-APOE ε4 carriers with the lower tertile of IL-6 levels, non-APOE ε4 carriers who had the upper tertile of IL-6 levels (but not those in the middle IL-6 tertile) had significantly accelerated rate of CD, to the extent of approximately half the rate of decline of that of APOE ε4 carriers. These results did not differ by race. Hence, without the largest genetic risk factor for AD, that is, APOE ε4, elevated serum IL-6 still could confer accelerated rate of CD, with a similar detrimental effect half of that imposed by APOE ε4 alone. The slowest CD was observed in non-carriers who had the lowest levels/tertiles of serum IL-6 levels. Hence, non-APOE ε4 carriers without elevated serum IL-6 levels, that is, those without a genetic predisposition and late-life stressors, have the lowest risk of CD and possibly the lowest risk of AD/ADRD.

We initially hypothesized that the ɛ4 allele may exert its detrimental effects via inflammatory mechanisms and thus the presence of the ɛ4 allele will compound the rate of CD conferred by high IL-6 levels. However, contrary to our hypothesis, in the presence of genetic effects imposed by the APOE ε4 allele, serum IL-6 levels did not seem to exert an additional impact in affecting CD. It is plausible that APOE ε4 elicits its detrimental effects via alternative pathways instead, with potential candidates including lipid and cholesterol dysregulations, as one of the main functions of apoE protein include clearing cholesterol and lipids.¹⁰ There are also other candidates; apoE protein may instead interact with amyloid, tau, and/or neurodegeneration markers, including elevated glial fibrillary acidic protein, total tau, and neurofilament light chain. A study showed that by age 65, nearly all APOE ε4 homozygote participants show abnormal levels of canonical AD biomarkers, evidenced by abnormal cerebrospinal fluid Aβ1-42 levels and 75% had positive amyloid scans.⁷ On the contrary, serum IL-6 is a non-canonical and non-specific AD biomarker indicative of systemic inflammatory processes that originate both from the periphery and the brain. Taken together these complementary findings, our findings suggest that instead of a systemic marker indicative of both peripheral and central nervous system inflammation, that is, IL-6, the APOE ε4 allele may interact with neurodegeneration and neuroinflammation more specific to the brain.

In non-APOE ε4 carriers, elevated serum IL-6 accelerated CD. It is noteworthy that there seemed to be a threshold effect of IL-6 with CD. The findings were not significant when serum IL-6 levels were operationalized as continuous log₁₀ IL-6 nor in the middle tertile, but were significant in only the upper tertile. Taken together, our findings suggested that to affect CD in the absence of APOE ε4 allele, serum IL-6 levels may need to be above a certain threshold level to exert its detrimental effects on CD. Hence, cognitive effects of serum IL-6 levels depend not only on APOE ε4 allele, but also how high IL-6 levels are. These findings highlight a specific subgroup of older adults who are more at risk and who could potentially benefit from anti-inflammatory interventions. Given the findings on almost full penetrance of APOE ε4 homozygotes on canonical AD biomarkers,⁷ our findings provide complementary yet important evidence on non-APOE ε4 and non-canonical AD-dependent processes that accelerate cognitive decline.

One of the possible sources of heterogeneity in previous literature could be race-based differences in both the APOE ε4 allele frequency and serum IL-6 levels. Due to genetic ancestry, compared to Whites, Blacks have a higher APOE ε4 allele frequency. Systemic issues faced by Blacks in their daily lives, including discrimination, systemic racism, and access to health care, could contribute to elevated serum IL-6 levels.^15,26 Hence, some studies^27–30 suggest that these differences potentially contribute differently to the pathophysiology of dementia. Here, we showed that despite the lack of a significant difference in IL-6 levels, given significant difference in APOE ε4 carriers in Blacks versus Whites, all the associations in Blacks versus Whites did not seem to differ. Conversely, two previous cross-sectional studies showed that blood-based IL-6 was significantly associated with CD in Blacks but not Whites.^30–32 Furthermore, two other longitudinal studies showed no racial differences on the effect of inflammation on CD.^33,34 However, it is worth noting that these previous studies did not examine the interaction effects of blood-based IL-6 with APOE ε4 allele, let alone in a sample comprising a substantial percentage of Blacks. Owing to these mixed results of previous studies, our cohort offers a unique opportunity to interrogate this issue, showing for the first time no racial differences in the interaction effects. In all, there is a need for replication of these findings in other cohorts with a more diverse racial representation.

A few potential issues may have limited our interpretations. Due to sample availabilities, we examined serum IL-6 at a single timepoint and lacked other inflammatory marker levels. Future analyses of additional cytokines across timepoints are needed. Additionally, there might have been a potential statistical power issue with our a priori race stratifications, which might have rendered reduced power to detect statistical significance and a complicated four-way interaction model infeasible. Conversely, interpreting a four-way interaction model is inherently challenging and may not be feasible. Although it would have been useful to perform sensitivity analyses on APOE ε4 homozygotes, the requirement of a larger sample size for interaction and stratification analyses deter us from performing such analyses. Our study lacked other genetic characterization, such as TREM2, which impacts neuroinflammation. Recent studies have suggested that both APOE and TREM2 are required for microglia to cluster around the Aβ plaque and clear apoptotic neurons.³⁵ Similarly, we also did not have IL-6 genotyping; several studies have shown that the G/G IL-6 genotype, predominantly found in Blacks,^36,37 could result in elevated IL-6 production. Future studies should build on our findings here and examine the roles of/control for these other genotypes.

Our study has several notable strengths. In stark contrast to previous studies, our cohort comprised solely community-dwelling older adults who were relatively healthy. Participants had an average of one chronic condition and were relatively free of cardiovascular and cardiometabolic diseases. These are two critical factors that have been shown to influence inflammatory markers^26,30 and thus could have potentially confounded findings of previous studies. For example, a previous study analyzed a cohort comprising ≈ 50% of participants who had hypertension and/or cardiovascular conditions.³⁰ Despite controlling for the confounding effects of chronic conditions on IL-6 levels in the statistical model, residual confounding effect may remain, as the chronic conditions were inherent to the participants and thus likely have affected other variables. Second, this study addressed the issue of scarcity of studies examining inflammation and cognitive function in cohorts with substantial Black participants, thus enhancing the generalizability of findings. Third, in previous studies, even when minorities, such as Blacks, were included, race was frequently operationalized as a covariate and not as a core variable, thus impeding a direct comparison of racial differences in the associations. In this study, with 60% of Black population, we conducted both total and subgroup analyses to examine racial differences in the associations.

In all, our data showed that the detrimental effects of serum IL-6 levels were contingent upon APOE ε4 carrier status. Specifically, in the presence of the APOE ε4 allele, regardless of IL-6 levels, APOE ε4 carriers experienced accelerated CD at a similar rate. However, even without any APOE ε4 allele, which constitutes most of the population, heightened systemic inflammation still accelerate cognitive decline. In light of recent findings showing overwhelming biological penetrance of the APOE ε4 allele on canonical AD biomarkers and pathologies,⁷ our finding is significant in contributing complementary evidence on non–APOE ε4-dependent and non-AD biological pathways through which cognitive decline can still be accelerated in at-risk non-APOE ε4 carriers. Though the main effect of APOE ε4 in different races was validated, race-stratified exploratory analyses did not show race-based differential interaction effects with serum IL-6 levels. This study thus highlights the complexity of the inter-relationships among race, APOE ε4 allele, serum IL-6 levels, and CD, while for the first time attempting to address myriad important AD risk factors within a single study. Consequently, this study has pertinent implications. Despite the APOE ε4 allele being a non-modifiable risk factor for AD/ADRD, serum IL-6, an indicator of chronic low-grade inflammation, is a modifiable risk factor that could be ameliorated by modulating stressors. Hence, strategies aimed at mitigating IL-6 levels may have therapeutic potential. Indeed, several existing non-pharmacological interventions have previously been conducted, demonstrating potential to improve IL-6 levels, including diet combined with resistance training,³⁸ mindfulness practices,³⁹ and horticultural therapy.⁴⁰ In all, this study not only furthered our understanding of the neurodegenerative disease processes, but it also identified a high-risk subgroup of vulnerable older adults. Hence, findings may inform the development of more targeted therapies and interventions, which is a step forward in precision medicine in AD/ADRD.

ACKNOWLEDGMENTS

The authors thank all the participants in the Chicago Heath and Aging Project. They also thank the staff of the Rush Institute of Healthy Aging. This study was supported by the National Institutes on Aging of the National Institutes of Health under Award Numbers: P30AG072972, T32AG050061, R01AG051635, RF1AG057532, and R01AG073627. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

CONFLICT OF INTEREST STATEMENT

K.D. and K.B.R. are funded by NIH research grants and report no conflicts of interest. All other authors report no conflicts of interest. Author disclosures are available in the journal. Author disclosures are available in the supporting information.

CONSENT STATEMENT

The institutional review board of the Rush University Medical Center approved the CHAP study protocols, and all participants provided written consent for population interviews, blood collection, and clinical evaluations.

DATA AVAILABILITY STATEMENT

The institutional review board of the Rush University Medical Center approved the study protocols, and all participants provided written consent for blood and DNA collection, population interviews, and clinical evaluations. Data that support study findings are available through data request and a data use agreement from our research resource data portal, .

References

Huell M, Strauss S, Volk B, Berger M, Bauer J. Interleukin‐6 is present in early stages of plaque formation and is restricted to the brains of Alzheimer's disease patients. Acta Neuropathol (Berl). 1995;89:544‐551.

West PK, Viengkhou B, Campbell IL, Hofer MJ. Microglia responses to interleukin‐6 and type I interferons in neuroinflammatory disease. Glia. 2019;67(10):1821‐1841.

Brydon L, Edwards S, Mohamed‐Ali V, Steptoe A. Socioeconomic status and stress‐induced increases in interleukin‐6. Brain Behav Immun. 2004;18(3):281‐290.

Kershaw KN, Lewis TT, Roux AVD, et al. Self‐reported experiences of discrimination and inflammation among men and women: the multi‐ethnic study of atherosclerosis. Health Psychol. 2016;35(4):343.

Gabay C. Interleukin‐6 and chronic inflammation. Arthritis Res Ther. 2006;8(2):1‐6.

Hüll M, Fiebich B, Lieb K, et al. Interleukin‐6‐associated inflammatory processes in Alzheimer's disease: new therapeutic options. Neurobiol Aging. 1996;17(5):795‐800.

Fortea J, Pegueroles J, Alcolea D, et al. APOE4 homozygozity represents a distinct genetic form of Alzheimer's disease. Nat Med. 2024:1‐8. Published online.

Verghese PB, Castellano JM, Holtzman DM. Apolipoprotein E in Alzheimer's disease and other neurological disorders. Lancet Neurol. 2011;10(3):241‐252.

Lynch JR, Tang W, Wang H, et al. APOE genotype and an ApoE‐mimetic peptide modify the systemic and central nervous system inflammatory response. J Biol Chem. 2003;278(49):48529‐48533.

Kloske CM, Wilcock DM. The important interface between apolipoprotein E and neuroinflammation in Alzheimer's disease. Front Immunol. 2020;11:754.

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.

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.

Matthews KA, Xu W, Gaglioti AH, et al. Racial and ethnic estimates of Alzheimer's disease and related dementias in the United States (2015‐2060) in adults aged≥ 65 years. Alzheimers Dement. 2019;15(1):17‐24.

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.

Arriola KJ, Lewis TT, Pearce B, et al. A randomized trial of race‐related stress among African Americans with chronic kidney disease. Psychoneuroendocrinology. 2021;131: [eLocator: 105339].

Singh‐Manoux A, Dugravot A, Brunner E, et al. Interleukin‐6 and C‐reactive protein as predictors of cognitive decline in late midlife. Neurology. 2014;83(6):486‐493.

Weaver J, Huang MH, Albert M, Harris T, Rowe J, Seeman TE. Interleukin‐6 and risk of cognitive decline: macArthur studies of successful aging. Neurology. 2002;59(3):371‐378.

Bienias JL, Beckett LA, Bennett DA, Wilson RS, Evans DA. Design of the Chicago health and aging project (CHAP). J Alzheimers Dis. 2003;5(5):349‐355.

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.

Hixson JE, Vernier D. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31(3):545‐548.

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.

IL‐6 Assays. Quanterix. Published October 3, 2023. Accessed March 18, 2024. https://www.quanterix.com/simoa‐assay‐kits/il‐6/

Wilson RS, Bennett D, Bienias J, De Leon CM, Morris M, Evans D. Cognitive activity and cognitive decline in a biracial community population. Neurology. 2003;61(6):812‐816.

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.

R Core Team R, R: A language and environment for statistical computing. Published online 2013.

Stowe RP, Peek MK, Cutchin MP, Goodwin JS. Plasma cytokine levels in a population‐based study: relation to age and ethnicity. J Gerontol Ser Biomed Sci Med Sci. 2010;65(4):429‐433.

Tan Z, Beiser A, Vasan R, et al. Inflammatory markers and the risk of Alzheimer disease: the Framingham Study. Neurology. 2007;68(22):1902‐1908.

Ravaglia G, Forti P, Maioli F, et al. Blood inflammatory markers and risk of dementia: the Conselice Study of Brain Aging. Neurobiol Aging. 2007;28(12):1810‐1820.

Koyama A, O'Brien J, Weuve J, Blacker D, Metti AL, Yaffe K. The role of peripheral inflammatory markers in dementia and Alzheimer's disease: a meta‐analysis. J Gerontol Ser Biomed Sci Med Sci. 2013;68(4):433‐440.

Windham BG, Simpson BN, Lirette S, et al. Associations between inflammation and cognitive function in African Americans and European Americans. J Am Geriatr Soc. 2014;62(12):2303‐2310.

Boots EA, Castellanos KJ, Zhan L, et al. Inflammation, cognition, and white matter in older adults: an examination by race. Front Aging Neurosci. 2020;12: [eLocator: 553998].

Goldstein FC, Zhao L, Steenland K, Levey AI. Inflammation and cognitive functioning in African Americans and Caucasians. Int J Geriatr Psychiatry. 2015;30(9):934‐941.

Yaffe K, Lindquist K, Penninx B, et al. Inflammatory markers and cognition in well‐functioning African‐American and white elders. Neurology. 2003;61(1):76‐80.

Beydoun MA, Dore GA, Canas JA, et al. Systemic inflammation is associated with longitudinal changes in cognitive performance among urban adults. Front Aging Neurosci. 2018;10:313.

Gratuze M, Leyns CE, Holtzman DM. New insights into the role of TREM2 in Alzheimer's disease. Mol Neurodegener. 2018;13:1‐16.

Delaney NL, Esquenazi V, Lucas DP, Zachary AA, Leffell MS. TNF‐α, TGF‐β, IL‐10, IL‐6, and INF‐γ alleles among African Americans and Cuban Americans. Hum Immunol. 2004;65(12):1413‐1419. Report of the ASHI minority workshops: part IV.

Jamie WE, Edwards RK, Ferguson RJ, Duff P. The interleukin‐6− 174 single nucleotide polymorphism: cervical protein production and the risk of preterm delivery. Am J Obstet Gynecol. 2005;192(4):1023‐1027.

Daly RM, O'Connell SL, Mundell NL, Grimes CA, Dunstan DW, Nowson CA. Protein‐enriched diet, with the use of lean red meat, combined with progressive resistance training enhances lean tissue mass and muscle strength and reduces circulating IL‐6 concentrations in elderly women: a cluster randomized controlled trial. Am J Clin Nutr. 2014;99(4):899‐910.

Ng TKS, Fam J, Feng L, et al. Mindfulness improves inflammatory biomarker levels in older adults with mild cognitive impairment: a randomized controlled trial. Transl Psychiatry. 2020;10(1):21.

Ng TKS, Gan DR, Mahendran R, Kua EH, Ho RC. Social connectedness as a mediator for horticultural therapy's biological effect on community‐dwelling older adults: secondary analyses of a randomized controlled trial. Soc Sci Med. 2021;284: [eLocator: 114191].