Integration of genetic risk, cardiometabolic diseases, and cognitive impairment among African American adults

Abstract

Background:

African American (AA) adults have a high burden of late-life cognitive impairment (CI) and dementia but remain underrepresented in genetic epidemiology studies. Genetic risk and cardiometabolic diseases (CMDs) contribute to dementia risk. This study investigated whether genetic susceptibility and CMDs were associated with a composite CI outcome and whether CMDs modified these associations.

Methods:

In AA participants within the REasons for Geographic and Racial Differences in Stroke (REGARDS) study, we assessed the association of a dementia polygenic risk score (PRS), APOE ε4 carrier status, and three prevalent CMDs: stroke, coronary artery disease (CAD), and type 2 diabetes (T2D) with a composite outcome of CI and dementia as a contributing cause of death (DCCD).

We used logistic regression adjusted for age, sex, education, income, body mass index, smoking status, alcohol intake, physical activity, hypertension, low-density lipoprotein, and C-reactive protein. Interaction terms were included to assess whether CMDs modified the associations between genetic risk and the composite outcome.

Results:

Of 8,838 participants, 516 (5.84 %) developed CI or had DCCD. In fully adjusted models, high polygenic risk (highest vs lowest PRS tertile) was associated with increased odds of the composite outcome [odds ratio (OR): 1.42; 95 % confidence interval (CI): 1.12–1.78], as was APOE ε4 carrier status (OR: 1.46; 95% CI: 1.21–1.78). Among CMDs, stroke (OR: 1.45; 95% CI: 1.04–2.02) and T2D (OR: 1.31; 95% CI: 1.06–1.61) were significantly associated with increased odds of the composite outcome. However, the association between genetic risk and the composite outcome did not significantly differ by CMD status.

Conclusion:

Genetic risk and CMDs independently contributed to dementia-related outcomes, indicating their relevance in understanding dementia risk among AA adults.

Introduction

Life expectancy in the US has increased significantly, with the population aged ≥65 projected to reach 83.7 million by 2050.¹ This demographic shift has been accompanied by a growing burden of late-life cognitive impairment (CI), which often precedes dementia. Importantly, risk can differ across populations and African Americans (AAs) experience a disproportionately higher incidence and prevalence of CI and dementia compared to White adults. The reasons for these disparities have not been fully explained, but genetic susceptibility and cardiometabolic diseases (CMDs; e.g., type 2 diabetes (T2D), coronary artery disease (CAD)) are known to contribute to the complex underpinnings of dementia risk.^2,3 While many studies have examined these risk factors individually,^3–5 few have explored their combined effects.

The apolipoprotein E (APOE) ε4 allele is a well-established genetic risk factor for Alzheimer’s disease (AD) and related dementias. For instance, a study conducted in a European ancestry population (EUR) reported that APOE ε4 carriers had a 3.16-fold higher hazard of dementia [95 % confidence interval (CI): 3.00–3.33] compared to non-carriers.⁶ Further, evidence suggests that the magnitude of this association may vary by ancestry. Data from the Health and Retirement Study (HRS) showed that the odds of dementia among APOE ε4 carriers was lower in AAs (odds ratio (OR): 1.74; 95 % CI: 1.15–2.63) than in White adults (OR: 2.30; 95 % CI: 1.86–2.85).⁷

Although APOE remains a key risk factor,⁸ CI and dementia are widely considered to be polygenic. While genome-wide association studies (GWAS) have identified numerous loci associated with dementia-related phenotypes,^9–11 these efforts have largely focused on EUR populations.^10,11 The underrepresentation of AA populations in GWAS limits the ability to identify ancestry-relevant risk variants and to develop accurate downstream tools like polygenic risk scores (PRS) for use in these populations. Chen et al. constructed a dementia PRS and found that CMDs independently increased dementia risk and may further increase risk in individuals with high genetic susceptibility.⁶ However, as these findings were primarily observed in EUR, it remains unclear whether similar associations exist in AAs.

Therefore, in AA participants from the REasons for Geographic and Racial Differences in Stroke (REGARDS) study, we (1) assessed the independent effects of genetic risk-captured by both the APOE ε4 allele and a non-APOE dementia PRS and CMDs (stroke, CAD, and T2D) on the composite outcome of CI and dementia as a contributing cause of death (DCCD) and (2) examined whether CMDs modified these associations.

Methods

Study population

The REGARDS study is a national, longitudinal cohort of 30,239 AA and White adults aged ≥45 years across the US enrolled 2003–2007 across the lower 48 continental US. Details of the study design, recruitment, and data collection have been described elsewhere.¹²

Exposure

Genetic data were available for 8,838 AA participants. Genotyping was performed using the Illumina Infinium Multi-Ethnic AMR/AFR Extended BeadChip arrays (MEGA chip; Illumina Inc., San Diego, CA). Imputation was conducted using the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine release 2 (Freeze 8) reference panel. Detailed quality control procedures have been described previously.¹³ We applied the PRS developed by Chen et al.,⁶ (https://www.pgscatalog.org/publication/PGP000402/) which comprises 27 dementia-associated variants, to AAs in the REGARDS study. The APOE ε4 carrier status was determined based on genotypes at rs429358 and rs7412, with participants categorized as carriers (ε2/ε4, ε3/ε4, ε4/ε4) or non-carriers (ε2/ε2, ε2/ε3, ε3/ε3).

CMDs included stroke, CAD, and T2D. Prevalent stroke was defined by self-reported history stroke at baseline. CAD was defined as self-reported history of myocardial infarction (MI), coronary artery bypass grafting, angioplasty, or stenting, or evidence of prior MI based on electrocardiogram. T2D was defined as fasting glucose ≥126 mg/dL, non-fasting glucose ≥200 mg/dL, or self-reported use of diabetes medication or insulin. The ‘any CMD’ category was defined by having a history of at least one CMD.

Outcome

The primary outcome was a composite measure of CI and DCCD. Participants with either incident CI or DCCD were classified as having the composite outcome (CI/DCCD). Incident CI was assessed using the Enhanced Cognitive Battery and defined as impairment in at least one memory test (Word List Learning or Delayed Recall) and one fluency test (Animal Fluency or Letter F Fluency) (Supplement Methods). DCCD was determined using NDI records, identified by ICD-10 codes F00-F03, G30, G31.0-G31.1, and R54. Participants who developed DCCD without prior CI were classified as DCCD cases.

Covariates

Consistent with prior literature⁶ and their established associations with genetic risk, CMDs, and CI, we included age, sex, income, education, body mass index (BMI), smoking status, alcohol intake frequency, physical activity (PA), hypertension, low-density lipoprotein (LDL), and high-sensitivity C-reactive protein (hs-CRP).

Statistical analysis

The PRS was constructed using the PLINK ‘–score’ function, summing risk alleles weighted by their effect sizes from Chen et al. The PRS was z-standardized and then categorized into tertiles (low, medium, high). Logistic regression models were used to estimate the association between CMDs and genetic risk including both PRS and APOE ε4 genotype with CI/DCCD, reporting ORs and 95 % CIs. Model 1 adjusted for age and sex, while Model 2 further adjusted for income, education, BMI, smoking status, alcohol intake frequency, PA, hypertension, LDL, and hs-CRP. Interaction terms were included to determine whether CMDs modified the association between genetic risk and CI/DCCD. All analyses were performed in SAS 9.4 (SAS Institute Inc., Cary, NC). A significance level of p < 0.10 was used for interaction terms to account for the lower statistical power typically observed in interaction analyses.¹⁴

Results

The analytic sample included 8,838 AAs, of whom 516 (5.8 %) developed CI (n = 224) or had DCCD (n = 278), or experienced both (n = 14). APOE ε4 carrier status was present in 36.6 % (n = 3,237) of participants. CMDs were prevalent, with 6.5 % having stroke, 13.5 % having CAD, and 29.1 % having T2D (Supplement Table 1).

In the fully adjusted model, participants with high polygenic risk had 42 % higher odds of CI/DCCD compared to those with low genetic risk (OR: 1.42; 95 % CI: 1.12–1.78), while no association was observed for the intermediate risk group (OR: 1.10, 95 % CI: 0.86–1.39). APOE ε4 carrier status was also associated with increased odds of CI/DCCD (OR: 1.46; 95 % CI: 1.21–1.78). Among CMDs, stroke (OR: 1.45; 95 % CI: 1.04–2.02), T2D (OR: 1.31; 95 % CI: 1.06–1.61), and the presence of any CMD (OR: 1.30; 95 % CI: 1.07–1.59) were associated with increased odds of CI/DCCD (Table 1). We did not observe significant effect modification by CMDs on the associations of polygenic risk or APOE ε4 status with CI/DCCD (Figs. 1.1 and 1.2).

Table 1.

Logistic regression to assess the association of genetic risk, APOE ε4 genotype, and CMDs with the composite outcome of incident cognitive impairment and dementia as a contributing cause of death (CI/DCCD) (N = 8,838).

	Model 1 OR (95 % CI)	Model 2 OR (95 % CI)
Non-APOE dementia PRS
Low	Ref	Ref
Intermediate	1.10 (0.87, 1.38)	1.10 (0.86, 1.39)
High	1.39 (1.12, 1.74)	1.42 (1.12, 1.78)
APOE ε4 carrier status	1.45 (1.21, 1.74)	1.46 (1.21, 1.78)
One CMD
Stroke	1.40 (1.02, 1.92)	1.45 (1.04, 2.02)
T2D	1.25 (1.03, 1.51)	1.31 (1.06, 1.61)
CAD	1.16 (0.91, 1.49)	1.20 (0.92, 1.56)
Any CMD	1.26 (1.05, 1.51)	1.30 (1.07, 1.59)

Any CMD: presence of at least one of the cardiometabolic diseases - stroke, CAD, or T2D.

Non-APOE dementia PRS: tertiles based on distribution - low (lowest genetic risk), intermediate (intermediate genetic risk), and high (highest genetic risk).

Model 1 adjusted for age and sex.

Model 2 further adjusted for income, education levels, BMI, smoking status, alcohol intake frequency, physical activity, hypertension, LDL, and hs-CRP.

Only complete observations were included in models.

Figs. 1.1 and 1.2.

Multivariable ORs and 95 % CI of joint effect of CMDs and genetic predisposition (PRS and APOE ε4) on the composite outcome of incident cognitive impairment and dementia as a contributing cause of death (CI/DCCD). CMDs: presence of at least one of the cardiometabolic diseases - stroke, CAD, or T2D. Non-APOE dementia PRS: tertiles based on distribution - low (lowest genetic risk), intermediate (intermediate genetic risk), and high (highest genetic risk). Model 1 adjusted for age and sex. Model 2 further adjusted for income, education levels, BMI, smoking status, alcohol intake frequency, physical activity, hypertension, LDL, and hs-CRP.

Discussion

In this study of AAs, we found that participants with high polygenic risk, as defined by the highest tertile of PRS and those with APOE ε4 carrier status, had increased odds of CI/DCCD. Stroke, T2D, and the presence of any CMD were each independently associated with increased odds of CI/DCCD. The current findings support the role of PRS and APOE ε4 as strong genetic risk factors across populations, with weaker associations observed here for AAs compared to the EUR literature.

Regarding the association between a dementia PRS and CI/DCCD observed here, the effect size was smaller than in previous studies. Chen et al. found a stronger association between a dementia PRS and dementia risk in EUR,⁶ consistent with a broader trend of weaker predictive performance of EUR-derived PRS in non-European populations. Similarly, in the HRS, Bakulski et al. examined the performance of an AD PRS constructed from a EUR meta-analysis GWAS (excluding the APOE region and using SNPs with p < 0.01) in both White and AAs.⁷ A one standard deviation increase in the PRS was associated with higher odds of incident dementia among White adults (1.10; 1.00–1.20), but showed a weak, non-significant association in AAs (1.06; 0.97–1.30).⁷

Consistent with prior literature, we observed a significant association between APOE ε4 and dementia-related outcomes in AAs. In the EUR, Chen et al. reported an approximately threefold increased risk of dementia among APOE ε4 carriers,⁶ while our effect estimate was about half that magnitude but remained significant. A recent multi-ancestry study also showed that APOE*4 dosage was associated with increased dementia risk across diverse populations, though the magnitude of association differed by ancestry.⁸ These differences likely reflect heterogeneity in local ancestry at the APOE locus.

Stroke and T2D were associated with increased odds of CI/DCCD, supporting the well-known vascular and metabolic pathways underlying dementia related conditions.^5,6 While prior studies have explored the joint effect of genetic risk and CMDs in EUR,^6,15 our study expands on this by examining how CMDs may modify the relationship between genetic risk and dementia outcomes in AAs. We did not observe significant effect modification by CMD status, and prior studies in EUR have yielded mixed results. For instance, Chen et al. reported that individuals with APOE ε4 or high PRS combined with CMDs showed significantly higher dementia risk compared to those without CMDs or low genetic risk.⁶ In contrast, Tai et al. found no significant interaction between cardiometabolic multimorbidity and polygenic risk,¹⁵ suggesting that these risk factors contribute to dementia independently. This variability may reflect differences in allele frequencies, linkage disequilibrium patterns, and PRS derivation between EUR and AA populations, which may modify dementia risk.

Our study has a few limitations. First, the PRS was derived from a GWAS conducted in EUR. While GWAS in African ancestry populations are emerging, ancestry-specific PRS for AAs remain limited. As a result, our findings may be impacted by using a PRS not optimized for this population. Nonetheless, the observed associations suggest some degree of cross population relevance and emphasize the need to expand GWAS in underrepresented populations. Second, we used prevalent CMDs due to limited incident CMDs available in REGARDS. Finally, phenotyping relied in part on self-reported data, which may introduce misclassification and measurement error.

Despite these limitations, our study has several strengths. Incorporating DCCD into the composite outcome enhances the assessment of dementia-related burden in REGARDS since the study has not yet classified incident dementia. Additionally, leveraging both the non-APOE PRS and APOE ε4 carrier status provides a nuanced evaluation of genetic susceptibility, incorporating both well-established causal variants and polygenic contributions to dementia risk. Lastly, the use of the REGARDS cohort, which includes extensive clinical, genetic, and metabolic data in a large sample of AAs, further enhances the robustness of our findings.

In conclusion, our findings demonstrate the independent contributions of genetic risk and CMDs to CI/DCCD. The lack of a significant interaction suggests these factors may act through distinct pathways. Future studies should further investigate these mechanisms in dementia-related outcomes, particularly in AA populations, as PRS continue to evolve to better represent diverse ancestries.

Supplementary Material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jstrokecerebrovasdis.2025.108535.