Association of Plasma Amyloid-β and Dementia Among Black and White Older Adults

INTRODUCTION

An estimated 6.5 million Americans aged 65 and older are living with Alzheimer’s disease (AD) and related dementias (ADRD) [1, 2]. The hallmark of AD pathology includes the accumulation of the amyloid-β (Aβ) plaques in the brain, which slowly begin to form many years before clinical symptoms emerge. Although cerebrospinal fluid (CSF) and PET biomarkers of Aβ are highly accurate in detecting amyloid pathology in vivo, the high cost and invasiveness limit their use in community settings [3, 4]. Plasma Aβ level has emerged as a promising biomarker of AD [5–7]. Furthermore, a lower plasma Aβ_42/40 ratio has also been associated with a higher risk of mild cognitive impairment and all-cause dementia [6, 8]. With the recent Food and Drug Administration (FDA) approval of anti-amyloid drugs that may slow cognitive decline among individuals with early AD [9], using plasma Aβ to identify individuals at high risk before they develop dementia could potentially be a critical first step to ensure timely diagnosis and intervention.

ADRD disproportionately impacts Black older adults in the United States, who may be up to twice as likely to be diagnosed with ADRD compared to White older adults [2, 11]. A population-based study of non-demented elders reported that Black (versus White) older adults had greater amyloid deposition on florbetapir PET [12]. Few studies have investigated plasma Aβ markers in diverse populations, with a limited number of Black participants. While one study found differences in plasma Aβ levels by race, others did not find similar disparities [13, 14]. The study did not separate participants with and without AD, although the association between race and plasma Aβ may differ by AD diagnosis [14]. In addition, many risk factors can modulate the association between amyloid burden and cognitive outcomes, such as vascular risk factors (VRFs, including hypertension, diabetes, obesity, dyslipidemia), apolipoprotein E (APOE) ɛ4 genotype [15, 16]. These risk factors are more prevalent in Black older adults, who may also have more barriers to VRF treatment and, therefore, more uncontrolled or severe VRFs than White older adults [17–19]. The association between amyloid burden and cognitive outcomes may also be modulated by cognitive reserve that is closely intertwined with educational quality and attainment [15, 20]. Lower educational quality and attainment, which can be attributed to factors related to residential segregation and structural racism, may disproportionately affect Black older adults [21, 22]. These findings raise the question of whether the association between plasma Aβ and the risk of dementia differs by race.

In a large cohort of community-dwelling Black and White older adults without dementia, our goal was to compare the plasma Aβ_42/40 by race, adjusting for potential confounders. Another goal of this study was to assess whether the association between plasma Aβ_42/40 and the risk of incident dementia over 15 years differed among Black and White older adults.

METHODS

Statistical analysis

Descriptive statistics were used to compare the baseline characteristics between Black and White participants and by Aβ_42/40 tertile within each race using Mann-Whitney U tests, Kruskal-Wallis tests, and Chi-Square tests as appropriate. We used linear regression to model natural log-transformed plasma Aβ_42/40 by race with and without adjusting for potential confounders. The estimated marginal means for the plasma Aβ_42/40 levels by race were presented by exponentiation of the natural log-transformed plasma levels (back-transformed from the log scale). We selected potential confounders a priori based on previous studies, which included demographics, behavioral and VRFs, renal function, and depression [8, 34]. Given the important role of APOE ɛ4 in Aβ metabolism and its different prevalence by race, we also adjusted for and stratified our analysis by APOE. We then conducted race-stratified analyses using Cox proportional hazards models to assess, within each race, the association between Aβ_42/40 tertile and the risk of dementia with and without adjusting for the above potential confounders, education, baseline cognition, and APOE ɛ4 [8, 35], with censoring at dementia, death, or their last study visit (whichever occurred first). Additional detail on the potential confounders and effect modification in our conceptual model was included in Supplementary Figures 1 and 2. The results were reported as hazard ratios (HR) with 95% confidence intervals. We tested the interaction between race and Aβ_42/40 by tertile. A p < 0.05 was considered statistically significant in all other analyses. Statistical analyses were conducted with SAS (version 9.4).

OPEN IN VIEWER

Fig. 1

The distribution of amyloid-β (Aβ) 42/40 level by race.

OPEN IN VIEWER

Fig. 2

Kaplan-Meier curves of dementia-free survival by amyloid-β (Aβ) 42/40 tertile among 997 Black and White participants. Annotations denoting the number of participants at risk at 0, 5, 10, and 15 years.

RESULTS

Among the 997 participants (mean age 74±2.9 years), 538 (54%) were Black, and 550 (55%) were female. Table 1 shows the baseline characteristics of Health ABC participants by race. Compared to White participants, Black participants were more likely to be female and APOE ɛ4 carriers, drank less alcohol, had higher serum creatinine and greater VRFs, and had lower education, physical activity, and baseline 3MS score. Baseline characteristics were also compared by Aβ_42/40 tertile among White and Black participants in Table 2. Those with lower Aβ_42/40 levels were more likely to be APOE ɛ4 carriers.

DISCUSSION

In a large prospective cohort of dementia-free adults aged 70–79, we found that Black older adults have lower age-adjusted plasma Aβ_42/40 than White older adults. The higher prevalence of APOE ɛ4 carriers among Black older adults did not explain much of the disparity in plasma Aβ_42/40, which was primarily driven by the Aβ_42/40 difference among APOE ɛ4 non-carriers. Furthermore, demographics, vascular and behavioral risk factors, renal function, and depression did not explain the difference in Aβ_42/40 by race. Interestingly, the association between lower plasma Aβ_42/40 and dementia risk over 15 years was found to be stronger among Black older adults.

Despite the recent advances in research on plasma AD biomarkers, including Aβ_42/40, significant gaps remain in the study of plasma Aβ and the association with incident dementia among diverse communities. Prior reports regarding Aβ_42/40 levels in Black and White adults are conflicting. Results from the Washington Heights-Inwood Columbia Aging Project (WHICAP) found that Aβ_42/40 level did not differ among 113 autopsied and 300 clinically evaluated Black, White, and Hispanic participants with and without AD, but the study may have had limited power [14]. Among a diverse cohort of 1,428 participants (mean age 63–69, 10% Black) from the Health and Aging Brain Study-Health Disparities (HABS-HD), higher Aβ_42/40 was reported for Black compared to White older adults [13]. Heterogeneity in age and sample size, especially for Black participants, and the inclusion of participants with AD may contribute to the discrepancies in findings compared to ours. Previous studies comparing Aβ_42/40 across racial groups are also limited by the lack of adjustment for age and APOE ɛ4, two important factors of plasma Aβ that are likely to vary by race [36].

We found evidence of a racial disparity in plasma Aβ_42/40 between Black and White older adults without dementia, which aligned with the prior study on Florbetapir PET Aβ burden among non-demented older adults in the Atherosclerosis Risk in Communities - Neurocognitive Study (ARIC-NCS) [12]. Our finding on the racial difference in plasma Aβ_42/40 existing primarily among the APOE ɛ4 non-carriers is novel and has important implications for the underlying mechanisms contributing to the racial disparity in dementia risk. Despite an observed stronger association between lower Aβ_42/40 and dementia risk in Black than White participants, the interaction did not meet statistical significance. This finding is similar to that from the ARIC study, which showed that the associations between late-life plasma Aβ measures and cognitive outcomes did not differ by race [8].

The mechanisms for disparities in plasma and brain Aβ deposition are unclear. APOE ɛ4, a genetic risk factor of AD with an important role in Aβ metabolism, is more prevalent in Black than White individuals, but most studies showed that the association between APOE ɛ4 and dementia risk is weaker in Black than in White older adults [37–39]. Our findings suggest that differences in APOE ɛ4 prevalence did not appear to be the major factor, and the Aβ difference by race may be primarily driven by differences in Black and White older adults among APOE ɛ4 non-carriers. While we did not find that VRF or other comorbidities explained the difference in plasma Aβ_42/40 by race, it is possible that disproportionate exposure to long-term VRF burden by race, which can start from midlife, may still contribute to this disparity. In addition, the possibility that differential social and structural factors by race may contribute to the disparity of Aβ_42/40 by race invites future studies in diverse communities [40, 41].

Although the difference in plasma Aβ_42/40 by race was statistically significant in our study, it is important to note that the mean difference was small. Nonetheless, previous studies have demonstrated similarly small but significant differences in mean plasma Aβ_42/40 between amyloid PET negative and positive individuals [42], suggesting that even subtle variations in Aβ_42/40 ratios may indicate differences in underlying brain amyloidosis. The potential disparities in amyloid pathology between Black and White older adults may contribute to the disparities in dementia risk by race, particularly when combined with other risk factors, such as VRFs, which have a disproportional impact on different racial groups [16]. In addition, although we adjusted for baseline cognitive function, unaccounted differences in pathologies and brain aging at baseline may still play a role in the observed differences by race, which invites future studies.

Despite the fact that the interaction between race and Aβ_42/40 in association with the risk of dementia was not significant, we had limited power. Nevertheless, our findings suggest that Black older adults may be at a higher risk for dementia compared to White older adults with similar plasma Aβ levels. Emerging evidence now supports the notion that plasma Aβ_42/40 is a sensitive early indicator of dementia risk, preceding amyloid PET [43]. Further research is necessary to better understand the potential benefits of utilizing plasma Aβ_42/40 as a screening tool for early detection in different racial groups, with the goal of lessening racial disparities in dementia risk.

Our study has a number of strengths, including the prospective study design with up to 15 years of follow-up for incident dementia. Plasma biomarkers have now emerged as inexpensive and minimally invasive tools to predict AD risk, and we measured plasma Aβ in a community-based cohort with a large number of Black older adults. This study fills the gap in understanding the racial disparity in plasma Aβ and the association with dementia risk among non-demented Black and White older adults. We were able to control for numerous potential confounders, including those that were included in previous studies. There are some limitations to our study. Dementia diagnosis used in our study is likely less accurate than the diagnosis using comprehensive clinical evaluation, but the misclassification of dementia diagnosis should not differ by Aβ_42/40 tertile. We also did not have information on the etiology of dementia. We used Innogenetic INNO-BIA assays, an immunoassay-based method, to quantify plasma Aβ in a community-based cohort, similar to the ARIC study and the Framingham Heart Study [8, 35]. However, the precision of immunoassay-based methods may not be as sensitive as those obtained by newer mass spectrometry-based methods, such as the immunoprecipitation-mass spectrometry (IP-MS) [44]. Ongoing investigations using more precise assays (e.g., Simoa) to evaluate racial or ethnic differences in plasma Aβ_42/40 are underway [13]. Future studies employing newer and more precise methods are warranted to confirm our findings. The long-term stability of plasma Aβ when frozen for up to 10 years remains to be determined and merits thorough evaluation [45]. Unmeasured confounding from sociocultural and structural factors, VRFs earlier in life, and other comorbid conditions may also contribute to the racial differences in plasma Aβ_42/40. Although we have a large study sample, we may still be underpowered to detect an interaction by race in the association between Aβ_42/40 and dementia risk, the confirmation of which may require a larger study.

In conclusion, non-demented, community-dwelling older Black adults have lower plasma Aβ_42/40 levels than White adults, particularly among APOE ɛ4 non-carriers. Additionally, lower Aβ_42/40 levels are associated with a higher risk of dementia over 15 years, with a stronger association seen among Black participants. These findings suggest a potential explanation for the existing disparity in dementia risk among Black and White older adults and highlight a need for further research on the underlying mechanism of Aβ disparity by race.

AUTHOR CONTRIBUTIONS

Xiaqing Jiang (Conceptualization; Formal analysis; Methodology; Writing – original draft; Writing – review & editing); Amber L. Bahorik (Formal analysis; Methodology; Writing – review & editing); Neill R. Graff-Radford (Data curation; Writing – review & editing); Kristine Yaffe (Conceptualization; Data curation; Funding acquisition; Supervision; Writing – review & editing).