Abstract
INTRODUCTION
Recent evidence suggests that exposure to the stress of racism may increase the risk of dementia for Black Americans.
METHODS
The present study used 17 years of data from a sample of 255 Black Americans to investigate the extent to which exposure to racial discrimination predicts subsequent changes in serum Alzheimer's Disease Research Center (ADRC) biomarkers: serum phosphorylated tau181(p‐tau181), neurofilament light (NfL), and glial fibrillary acidic protein (GFAP). We hypothesized that racial discrimination assessed during middle age would predict increases in these serum biomarkers as the participants aged into their 60s.
RESULTS
Our findings indicate that exposure to various forms of racial discrimination during a person's 40s and early 50s predicts an 11‐year increase in both serum p‐tau181 and NfL. Racial discrimination was not associated with subsequent levels of GFAP.
DISCUSSION
These findings suggest that racial discrimination in midlife may contribute to increased AD pathology and neurodegeneration later in life.
Highlights
A 17‐year longitudinal study of Black Americans.
Assessments of change in serum p‐tau181, neurofilament light, and glial fibrillary acidic protein.
Exposure to racial discrimination during middle age predicted increases in p‐tau181 and neurofilament light.
Education was positively related to both p‐tau181 and exposure to racial discrimination.
Keywords: Alzheimer's disease, Black Americans, dementia, neurodegeneration, neurofilament light, racial discrimination, serum phosphorylated tau181
1. INTRODUCTION
Given dramatic population increases in older age groups, Alzheimer's disease (AD) and related dementias (ADRD) have become a public health priority throughout the world. Estimates are that 13.5 million individuals will suffer AD in the United States by 2050, 1 with Black Americans being disproportionately at risk compared to non‐Hispanic White Americans (NHWA). Unfortunately, research to date has not been very informative regarding the reasons for the elevated risk of AD seen among Black Americans. 2 , 3 Part of the answer, however, may be linked to the fact that Black Americans are members of a racialized group. Based on skin color, hair texture, and facial features, they are subject to negative stereotypes, patronization, exclusion, harassment, and other forms of racial mistreatment. 4 An abundance of recent research has documented the way that chronic exposure to such events takes a toll on biological systems and compromises health. Several studies have shown, for example, that exposure to racial mistreatment promotes onset of chronic illness, 5 , 6 an increased speed of biological aging, 7 and risk for stroke and mortality. 8
Building upon these findings, some have argued that the impact of stress and trauma likely extends to the brain and onset of dementia. 2 , 3 , 7 For example, high levels of stress are associated with reductions in hippocampal 9 , 10 and prefrontal cortex 11 volumes. These brain structures are critical for episodic memory and executive function and are among the first to show neurodegenerative changes in ADRD. 12 Thus, the stress of racial discrimination might be viewed as a potential contributor to the higher prevalence of dementia seen among Black Americans. In support of this idea, several recent studies have reported a link between exposure to racial discrimination and ADRD. 7 , 13 , 14 , 15 However, the underlying biological mechanisms by which stress and discrimination contribute to cognitive decline are not yet understood. For example, it is not clear whether racial discrimination impacts biomarkers of AD pathology, more general biomarkers of neurodegeneration, or both.
RESEARCH IN CONTEXT
Systematic review: Recent evidence suggests that exposure to the stress of racism may increase the risk of dementia seen among Black Americans. However, the underlying biological mechanisms by which stress and discrimination contribute to Alzheimer's disease and related dementia pathology are not understood.
Interpretation: Our findings indicate that exposure to various forms of racial discrimination during a person's 40s and early 50s predicts an increase over the next decade in both phosphorylated tau (p‐tau181) and neurofilament light (NfL). Discrimination was not associated with subsequent levels of glial fibrillary acidic protein (GFAP).
Future directions: In future studies, we plan to investigate the way that other race‐related stressors such as economic hardship, neighborhood disadvantage, as well as biological weathering and accelerated aging, contribute to the elevated levels of Alzheimer's disease and related dementias seen among African Americans.
Utilizing 17 years of data from 255 Black Americans participating in the Family and Community Health Study (FACHS), the present study investigates the extent to which chronic exposure to racial discrimination during middle age predicts subsequent changes in serum ADRD biomarkers. We examined serum phosphorylated tau181 (p‐Tau181), a marker of Alzheimer's pathology, neurofilament light (NfL), a non‐specific marker of neurodegeneration, and glial fibrillary acidic protein (GFAP), a marker of astroglia activation. We expected that exposure to chronic racial discrimination during middle‐age, assessed between 2002 and 2008, would show little association with the serum biomarkers assessed in 2008 when the participants were still relatively young (aged 40–50) and ADRD pathology is not highly prevalent. However, past research has shown that early exposure to racial discrimination predicts later onset of a wide variety of chronic illnesses, 5 , 6 speed of epigenetic aging, 7 and mortality. 8 We hypothesized that a similar pattern likely exists for ADRD and corresponding biomarkers.
To test this idea, the present study investigated the extent to which persistent exposure to discrimination during middle age predicts increases in the biomarkers of AD and neurodegeneration as the sample participants aged into their 60s. While it was presumed that p‐Tau181, NfL, and GFAP would rise with age across the 11 year period from 2008 to 2019, these increases were expected to be more rapid among those who had experienced chronic racial discrimination during middle age. We tested this prediction by examining the extent to which, after adjusting for various control variables including age, discrimination assessed between 2002 and 2008 predicted increases in the three biomarkers of dementia between 2008 and 2019.
2. METHODS
2.1. Participants
We used psychosocial and demographic data collected at Wave 3 (2002), Wave 4 (2005), Wave 5 (2008), and Wave 8 (2019) from the caregivers in the Family and Community Health Study (FACHS). FACHS is a longitudinal study of several hundred Black American families that was initiated in 1997. All of the families had a 5th grader at study inception. Using a stratified random sampling procedure, the sampling strategy was designed to generate families representing a range of socioeconomic statuses and neighborhood settings. Details regarding the recruitment are described by Simons et al. 16 Lengthy psychosocial interviews were conducted with the primary caregiver (usually the mother) and the target child. At Wave 1, about half of the sample resided in Georgia (n = 422) and the other half in Iowa (n = 467). Beginning in Wave 3, in an effort to increase the number of males in the sample, the primary caregiver's romantic partner was included in the data collection. They remained in the study sample throughout subsequent waves of data collection even if they were no longer married to, or romantically involved with, the primary caregiver.
Two‐hundred fifty‐five individuals participated in the blood draws and psychosocial interviews at both Waves 5 and 8. These individuals comprise the sample for the analyses in the present study. Comparisons of these participants with those who participated only in Wave 1 did not reveal significant differences with regard to either demographic characteristics (e.g., household income, education, or chronological age). Those rare instances of missing data were handled by FIML. 17 The protocol and all study procedures were approved by the Institutional Review Board at the University of Georgia.
2.2. Blood collection
In Waves 5 (2008) and 8 (2019), blood draws were added to the data collection protocol. Given the logistics of scheduling home visits by phlebotomists, only individuals still residing in Georgia or Iowa were identified as eligible for the Wave 5 and Wave 8 blood draws; approximately two‐thirds of these individuals agreed to participate. At each of these waves, a certified phlebotomist visited the home and collected four tubes of blood (30 mL) from each consenting participant. Non‐fasting blood draws were performed in participants’ homes in the morning. Serum samples were spun in the field using portable centrifuges. Sera was then removed and frozen at −80 C the same day if collected in Iowa. Georgia samples were packed in dry ice the same day of the blood draw and mailed overnight to the Iowa Psychiatric Genetic Laboratory at the University of Iowa, where all samples were stored at −80 C. Serum levels of the biomarkers did not differ by site.
2.3. Serum blood assays
In early 2023, 500ul serum samples collected at Wave 5 and Wave 8 were sent to the University of Minnesota Advanced Research and Diagnostics Laboratory (ARDL) to assay the ADRD‐related blood biomarkers on the Quanterix SiMoA HD‐X analyzer (Quanterix Corporation, Lexington, MA, United States). Aβ1‐42 and 1‐40, NfL, and GFAP were assayed using the Simoa Neurology 4‐Plex E Advantage kit (N4PE, item #103670). Most of the Aβ1‐42 and 1‐40 levels were below the limit of detection, so only NfL and GFAP data are presented. The interassay laboratory coefficients of variation (CVs) for NfL were 11.94%, 11.19%, and 13.53% at mean concentrations of 21.36, 446.45, and 11.42 pg/mL; CVs for GFAP were 9.80%, 8.50%, and 8.51% at mean concentrations of 151.37, 3370.83, and 67.71 pg/mL. Serum tau phosphorylated at position 181 (p‐tau181) was measured using the Simoa p‐tau181 Advantage V2 kit (item #103714). The interassay laboratory CVs were 5.6%, 5.2%, and 9.8% at mean concentrations of 38.4, 864.8, and 16.4 pg/mL.
2.4. Measurement of racial discrimination
At Waves 3 (2002), 4 (2005), and 5 (2008) participants completed 13 items from the Schedule of Racist Events (eTable 1). 18 These items assessed the frequency (1 = never, 4 = frequently) with which individuals have experienced discriminatory events during the preceding year. The items focus on events such as being the victim of disrespectful treatment by a store owner or salesclerks, racial slurs, being hassled by the police, exclusion from social activities, and not being expected to do well because of being a Black American. Scores for each item were averaged across waves. This was done to avoid problems with missing data when a participant failed to answer one or more discrimination items at a particular wave. The summation process becomes incomputable when items are missing. Overall elevation across the time period can be calculated; however, by aggregating scores across items within each wave. The mean item score approach addresses this quandary. This approach offers a reasonably precise means of estimating missing values and replicating the sample reliability of an additive scale. 19 , 20
2.5. Covariates
Participant demographics including chronological age, education, and gender were obtained via self‐report and included as covariates as these variables are often associated with ADRD. Body mass index (BMI; kg/m2) was determined at Wave 8 and included as a coviarate because increasing BMI has been associated with lower levels of p‐tau181, NfL, and GFAP. 21
2.6. Statistical analysis
Consistency of report of racial discrimination at Waves 3 (2002), 4 (2005), and 5 (2008) were assessed using Pearson's correlation. Log‐transformations were used to obtain more normalized distributions of the serum p‐Tau181, NfL, and GFAP — all of which were skewed to the right. Graphs showing the distribution of the variables after log transformation at Wave 8 are shown in eFigure 1, eFigure 2, and eFigure 3. Mplus 8.9 with maximum likelihood estimation (MLE) and bootstrapping with 1000 resamples were used in all analyses.
To assess the robustness of our findings, we utilized three methods for assessing the relationship between mean midlife discrimination (three waves between 2002 and 2008) and change in the ADRD serum biomarkers between 2008 (Wave 5) and 2019 (Wave 8), adjusting for age, education, gender, and BMI. First, we calculated change scores (Δ) for the study variables using the residuals obtained from the regression of Wave 8 scores on Wave 5 scores (residual method). The residual score isolates the unique change in Wave 8 that is not accounted for by its values at Wave 5 and other factors included in the model. Second, we calculated the difference between the biomarkers at Wave 8 and Wave 5 (subtraction method). The resulting value represents the raw change in each biomarker over the specified period. Third, we fitted a regression model to estimate the rate of change of the serum biomarkers (slope) from Wave 5 to Wave 8 while controlling for the baseline serum biomarker level at Wave 5 (slope method). The beta coefficient for each biomarker at Wave 5 in this model represents the baseline level, and the beta coefficient for the predictor of interest indicates the additional change in each biomarker from Wave 5 to Wave 8. In additional sensitivity analyses, we examined associations of racial discrimination at each individual wave, rather than averaging across waves, with the serum biomarkers.
Finally, structural equation modeling (SEM), with full information maximum likelihood (FIML) to account for missing cases, was used to examine the effect of midlife racial discrimination on each of the three serum biomarkers while considerating associations of the biomarkers with each other. The covariates age, gender, education, and BMI were also included in the two SEM models. One of the advantages of SEM is that that it can be used to create latent variables, and thereby allowed us to extend the sensitivity analysis presented in the multivariate regression. More specifically, use of SEM allowed us to treat level of discrimination across midlife as a latent variable, thereby allowing assessment of whether different approaches to calculating discrimination (averaging across waves vs. formation of a latent variable) resulted in similar associations between racial discrimination and change in the serum biomarkers. Using Full Information Maximum Likelihood (FIML) to address missing data, Model B treated elevated discrimination as a latent variable, with discrimination at Waves 3, 4, and 5 serving as observed indicators of this latent construct. Thus, our SEM models assessed the extent to which the positive association between chronic discrimination and each of the biomarkers was robust across varying methods for assessing exposure to chronic discrimination (averaging vs. latent variable), and after taking into account the associations of the biomarkers with each other. Notably, the results from the two SEM models are consistent with our prior results. They continue to show a significant association of elevated discrimination with change in Tau181 and NfL, whereas elevated discrimination fails to show an association with GFAP.
3. RESULTS
3.1. Participant characteristics
Table 1 presents basic demographic information for the study sample at Wave 5. As shown in the table, the sample consisted of 212 females and 43 males with a mean age of 46.07 (SD = 6.25). Of these individuals, 33.1% had family incomes below the federal poverty level; 19.3% had less than a 12th grade education and 80.7% received a 12th grade education or higher. About 38% were married and 17.3% were cohabitating. The majority (69.8%) lived in large urban areas, 14.1% lived in the suburbs, and 16.1% lived in rural areas.
TABLE 1.
Demographics for the study sample in 2008 at Wave 5 (n = 255).
| Variables | % or Mean ± SD |
|---|---|
| Women | 83.1% |
| Age (years) | 46.07 ± 6.25 |
| Education | |
| < 12th grade | 19.3% |
| High school | 44.1% |
| College graduate | 36.6% |
| Geographic location | |
| Large urban | 69.8% |
| Suburbs | 14.1% |
| Rural areas | 16.1% |
| Marital status | |
| Married | 38.0% |
| Cohabitating | 17.3% |
| At poverty level | 33.1% |
| 150% of poverty level | 45.3% |
3.2. Correlation of racial discrimination across waves in midlife
Higher scores indicated higher sustained levels of perceived racial discrimination across Waves 3 (2002), 4 (2005), and 5 (2008) (α ≥ 0.90). There was substantial continuity in reported discrimination across waves. For example, discrimination at Wave 3 strongly correlated with discrimination at Wave 4 (r = 0.563) and Wave 5 (r = 0.582); discrimination at Wave 4 also strongly correlated with discrimination at Wave 5 (r = 0.603).
3.3. Correlations between chronic racial discrimination and serum biomarkers
Table 2 shows the extent to which exposure to chronic discrimination across midlife (Waves 3, 4, and 5) correlated with serum p‐Tau181, NfL, and GFAP levels assessed at Wave 5, Wave 8, and Δ from W5 to W8. There were no correlations between racial discrimination and increased levels of the serum biomarkers at Wave 5, when the study participants were roughly 50 years of age. Indeed, the correlation for p‐Tau181 was even in the wrong direction (r = −0.136, p < 0.05). By Wave 8, however, when the participants were roughly 60 years of age, elevated discrimination during middle age significantly correlated with higher levels of both p‐Tau181 (r = 0.158, p ≤ 0.012) and NfL (r = 0.143, p ≤ 0.023).
TABLE 2.
Correlations between mean level of discrimination across middle age, between 2002 and 2008, and serum biomarkers of AD and neurodegeneration at Wave 5 (2008) and Wave 8 (2019) among 255 Black Americans.
| Pearson correlation with mean level of discrimination | Partial correlation with mean level of discrimination | ||
|---|---|---|---|
| Serum biomarker | W5 | W8 | W8 controlling for W5 |
| p‐Tau181 | −0.136 * | 0.158 * | 0.193 ** |
| NfL | 0.005 | 0.143 * | 0.176 ** |
| GFAP | −0.004 | −0.010 | −0.009 |
Abbreviations: GFAP, glial fibrillary acid protein; NfL, neurofilament light chain; p‐Tau181, phosphorylated tau 181; W5, wave 5; W8, wave 8.
* p < 0.05.
** p < 0.01.
The last column of Table 2 shows a significant partial correlation between chronic discrimination and serum p‐Tau181 at Wave 8 (r = 0.193, p ≤ 0.002) after controlling for the effect of p‐tau181 at Wave 5. In addition, there was a significant partial correlation between chronic discrimination and NfL at Wave 8 (r = 0.176, p ≤ 0.005) after controlling for NfL at Wave 5. In contrast to these findings, chronic discrimination was not significantly associated with GFAP at Wave 5 or Wave 8. Additionally, the partial correlation between discrimination and GFAP after controlling for GFAP at Wave 5 was also not significant.
3.4. Correlations between study variables
Next, we ran correlations between the study variables and change in each of the three biomarkers between Waves 5 and 8, as well as chronic discrimination assessed across Waves 3, 4, and 5, and covariates measured at Wave 8 (Table 3). Change in p‐Tau181 did not significantly correlate with either change in NfL or GFAP. However, there was a strong correlation between change in NfL and change in GFAP (r = 0.632, p < 0.01). Age correlated with increases in NfL and GFAP between waves 5 and 8. Higher education correlated with higher discrimination and higher serum p‐Tau181. Finally, BMI at Wave 8 correlated with increases in p‐Tau181 between Waves 5 and 8. Although not shown, BMI at Wave 5 also correlated with change in p‐Tau181 (r = 0.177, p ≤ 0.005), although change in BMI across waves 5 to 8 was not associated with changes in p‐Tau181.
TABLE 3.
Correlation matrix for study variables (N = 255).
| Parameter | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 |
|---|---|---|---|---|---|---|---|---|
| 1. Mean Discrim | — | |||||||
| 2. ΔpTauW58 | 0.191 ** | — | ||||||
| 3. ΔNfLW58 | 0.176 ** | 0.009 | — | |||||
| 4. ΔGFAPW58 | −0.009 | −0.089 | 0.632 ** | — | ||||
| 5. Education | 0.151 * | 0.150 * | −0.020 | −0.034 | — | |||
| 6. BMIW8 | −0.059 | 0.142 * | −0.044 | −0.056 | −0.003 | — | ||
| 7. Gender | 0.091 | 0.041 | 0.168 * | 0.013 | −0.025 | −0.093 | — | |
| 8. AgeW8 | 0.112 † | 0.097 | 0.400 ** | 0.290 ** | 0.200 ** | −0.077 | 0.096 | — |
| Mean | 1.856 | 0.001 | −0.003 | −0.001 | 12.956 | 35.206 | 0.168 | 57.016 |
| SD | 0.523 | 0.721 | 0.456 | 0.410 | 2.164 | 9.217 | 0.375 | 6.296 |
Abbreviations: AgeW8, age at wave 8; BMIW8, body mass index at wave 8; Mean Discrim, the mean of discrimination at waves 3, 4, and 5; SD, Standard Deviation; ΔGFAPW58, change in glial fibrillary acidic protein (GFAP) from wave 5 to wave 8; ΔNfLW58, change in neurofilament light (NfL) from wave 5 to wave 8; ΔpTauW58, change in phosphorylated tau 181 (p‐Tau181) from wave 5 to wave 8.
† p < 0.1.
* p < 0.05.
** p < 0.01.
3.5. Regression analyses examining associations between chronic discrimination and change in the serum biomarkers
Chronic discrimination was significantly associated with change in p‐Tau181 regardless of whether the residual (b = 0.173, p ≤ 0.01), subtraction (b = 0.241, p ≤ 0.01), or slope (b = 0.165, p ≤ 0.01) analytic method was used for calculating change (Table 4). The only difference between the three methods was that BMI was only associated with change in p‐Tau181 using the residual or slope methods, but not the subtraction method.
TABLE 4.
Regression models examining the effects of mean level of discrimination and the covariates on change in p‐Tau181, NfL, and GFAP from wave 5 to wave 8 using either the residual, subtraction, or slope analytic approaches to assess change in the biomarker (N = 255).
| Panel A | ΔpTau181W5‐8 | ΔNfLW5‐8 | ΔGFAPW5‐8 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Residual approach | β | t‐Value | p‐Value | β | t‐Value | p‐Value | β | t‐Value | p‐Value |
| Mean discrimination | 0.173 | 2.79 | 0.006 | 0.139 | 2.42 | 0.016 | −0.031 | −0.50 | 0.620 |
| Age | 0.064 | 1.03 | 0.305 | 0.397 | 6.82 | 0.000 | 0.311 | 5.00 | 0.000 |
| Education | 0.113 | 1.80 | 0.073 | −0.118 | −2.02 | 0.045 | −0.092 | −1.47 | 0.142 |
| BMI at Wave 8 | 0.161 | 2.62 | 0.009 | 0.005 | 0.09 | 0.931 | −0.036 | −0.59 | 0.554 |
| Gender | 0.037 | 0.60 | 0.552 | 0.114 | 2.00 | 0.047 | −0.020 | −0.33 | 0.744 |
| R‐squared | 0.081 | 0.205 | 0.095 | ||||||
| Panel B | pTau181W8‐5 | NfLW8‐5 | GFAPW8‐5 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Subtraction approach | β | t‐Value | p‐Value | β | t‐Value | p‐Value | β | t‐Value | p‐Value |
| Mean discrimination | 0.241 | 3.86 | <0.001 | 0.152 | 2.49 | 0.013 | −0.009 | −0.14 | 0.889 |
| Age | −0.003 | −0.04 | 0.965 | 0.241 | 3.91 | <0.001 | 0.192 | 3.02 | 0.003 |
| Education | 0.013 | 0.21 | 0.832 | −0.137 | −2.21 | 0.028 | −0.152 | −2.39 | 0.018 |
| BMI at Wave 8 | 0.101 | 1.64 | 0.102 | 0.030 | 0.50 | 0.616 | 0.036 | 0.59 | 0.555 |
| Gender | −0.036 | −0.59 | 0.556 | 0.110 | 1.81 | 0.071 | 0.057 | 0.91 | 0.365 |
| R‐Squared | 0.067 | 0.108 | 0.054 | ||||||
| Panel C | pTau181W8 | NfLW8 | GFAPW8 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Slope approach | β | t‐Value | p‐Value | β | t‐Value | p‐Value | β | t‐Value | p‐Value |
| Mean discrimination | 0.165 | 2.67 | 0.008 | 0.101 | 2.24 | 0.026 | −0.028 | −0.61 | 0.544 |
| Age | 0.065 | 1.06 | 0.290 | 0.386 | 7.66 | 0.000 | 0.259 | 5.30 | <0.001 |
| Education | 0.114 | 1.83 | 0.068 | −0.081 | −1.77 | 0.078 | −0.052 | −1.09 | 0.277 |
| BMI at Wave 8 | 0.159 | 2.64 | 0.009 | −0.009 | −0.21 | 0.833 | −0.045 | −0.97 | 0.335 |
| Gender | 0.039 | 0.64 | 0.523 | 0.090 | 2.02 | 0.044 | −0.034 | −0.73 | 0.466 |
| pTau181 at Wave 5 | 0.190 | 3.04 | 0.003 | ||||||
| NfL at Wave 5 | 0.441 | 8.86 | <0.001 | ||||||
| GFAP at Wave 5 | 0.573 | 11.18 | <0.001 | ||||||
| R‐squared | 0.123 | 0.520 | 0.494 | ||||||
Abbreviations: BMI, body mass index; GFAP, glial fibrillary acidic protein; NfL, neurofilament light; pTau181, phosphorylated tau 181; SD, Standard Deviation; W5, Wave 5; W8, Wave 8; Δ, change.
Similar to the findings for p‐Tau181, chronic discrimination was also significantly associated with change in NfL using either the residual (b = 0.139, p ≤ 0.05), subtraction (b = 0.152, p ≤ 0.05), or slope (b = 0.101, p ≤ 0.05) analytic method for calculating change (Table 4). In all three models, increasing age was associated with increasing NfL levels. Education was inversely associated with NfL using the the residual or subtraction methods, but not using the slope analytic methods. In contrast to p‐Tau181 and NfL, there were no associations between chronic discrimination and change in GFAP.
3.6. Sensitivity analyses for regression modeling
In additional analyses, instead of averaging the discrimination scores across Waves 3, 4, and 5, we examined the discrimination score at each individual wave in relation to change in the serum biomarkers. Although discrimination at each wave correlated with serum p‐Tau181 and NfL at Wave 8 as well as with change in levels between waves 5 and 8, the correlations were not as large as when level of discrimination averaged across waves was used. These results suggest that the accumulation/average of discrimination over time is a stronger predictor of p‐Tau181 and NfL compared to assessment of discrimination at a single timepoint.
3.7. Structural equation modeling
In additional analyses, we used structural equation modeling (SEM) to examine the effect of chronic discrimination on each of the three serum biomarkers while considering the associations of the biomarkers with each other (Figure 1). The covariates age, gender, education, and BMI were also adjusted in the two SEM models. Thus, our SEM models assessed the extent to which the positive association between chronic discrimination and each of the biomarkers was robust across varying methods for assessing exposure to chronic discrimination (averaging vs. latent variable), and after considering the associations of the biomarkers with each other. Notably, the results from the two SEM models are consistent with our prior results. They continue to show a significant association of elevated discrimination with change in p‐Tau181 and NfL, whereas elevated discrimination failed to show an association with GFAP.
FIGURE 1.
Structural equation models using either mean (Model A) or latent variable (Model B) methods to assess chronic discrimination across Waves 3, 4, and 5.
Further, the SEM models allowed us to explore the possibility that the impact of racial discrimination on a particular serum biomarker might be indirect through its association with another marker. For example, the ATN model of AD progression 22 posits a stepping‐stone approach where p‐Tau181 leads to neurodegeneration. This suggests that the effect of discrimination on NfL or GFAP might be indirect through (or mediated by) tau. However, the models presented in Figure 1 provide no support for this idea. The figures show that change in p‐Tau181 has only a small association with change in NfL and GFAP, and in both cases the correlation is negative. Change in NfL and GFAP, however, are strongly related to each other. The path model shows, however, that taking these correlations between the biomarkers into account has virtually no effect upon the association between chronic discrimination and change in the biomarker levels.
3.8. Individual discrimination items: A supplemental analysis
Finally, as supplemental analysis, we took each of the 13 items included in the discrimination index, averaged each of them across Waves 3‐5, and correlated them with change in the ADRD serum biomarkers. The results, as shown in Table 5, provide further support for the contention that discrimination is related to subsequent increases in p‐Tau181 and NfL. Nine of the 13 items are significantly related to change in p‐tau181 and 8 to change in NfL, with the remaining items all showing a positive, though insignificant, association with these two biomarkers. Consistent with the previous analyses reported above, none of the discrimination items show a significant association with change in GFAP.
TABLE 5.
Correlations for individual discrimination items.
| Cumulative scales | ||||
|---|---|---|---|---|
| Items | Variables | ΔpTau181 | ΔNfL | ΔGFAP |
| Item 1 | Has someone said something derogatory or insulting to you just because of your race or ethnic background? | 0.136 * | 0.168 ** | 0.035 |
| Item 2 | Has a store owner, sales clerk, or person working at a place of business treated you in a disrespectful way just because of your race or ethnic background? | 0.144 * | 0.112 | −0.070 |
| Item 3 | How often have the police hassled you just because of your race or ethnic background? | 0.084 | 0.118 | 0.025 |
| Item 4 | How often has someone ignored you or excluded you from some activities just because of your race or ethnic background? | 0.111 | 0.116 | −0.021 |
| Item 5 | Has someone suspected you of doing something wrong just because of your race or ethnic background? | 0.127 * | 0.126 * | −0.042 |
| Item 6 | Has someone yelled a racial slur or racial insult at you because of your race or ethnic background? | 0.155 * | 0.153 * | −0.002 |
| Item 7 | Has someone threatened to harm you physically just because of your race or ethnic background? | −0.031 | 0.106 | −0.035 |
| Item 8 | Have you encountered people who are surprised that you, given your race or ethnic background, did something really well? | 0.214 ** | 0.128 * | 0.039 |
| Item 9 | Have you been treated unfairly just because of your race or ethnic background? | 0.207 ** | 0.143 * | −0.035 |
| Item 10 | Have you encountered people who did not expect you to do well just because of your race or ethnic background? | 0.206 ** | 0.126 * | −0.031 |
| Item 11 | Has someone discouraged you from trying to achieve an important goal just because of your race or ethnic background? | 0.093 | 0.145 * | 0.015 |
| Item 12 | Have close friends of yours been treated unfairly just because of their race or ethnic background? | 0.149 * | 0.190 ** | 0.053 |
| Item 13 | Have members of your family been treated unfairly just because of their race or ethnic background? | 0.181 ** | 0.095 | −0.030 |
Abbreviations: ΔGFAP, change in glial fibrillary acidic protein (GFAP) from Wave 5 to Wave 8; ΔNfL, Change in neurofilament (NfL) from Wave 5 to Wave 8; ΔpTau181, change in p‐Tau181 from Wave 5 to Wave 8.
* p < 0.05.
** p < 0.01.
4. DISCUSSION
Currently, there is little consensus regarding the factors that account for Black Americans’ disproportionate risk for dementia compared to non‐Hispanic Whites in the United States. It is clear, however, that these racial differences are not simply a consequence of genetic differences. 2 , 3 Some have suggested that research regarding the factors that account for the high prevalence of dementia among Black Americans might best begin by investigating the stressors, hurdles, and deprivations associated with the everyday reality of living as a racialized group in American society. 2 , 3 Barnes, 2 for example, recently called for longitudinal studies to investigate the extent to which the challenges that characterize daily life for Black Americans are linked to biomarkers of ADRD and neurodegeneration. The present study is in keeping with this appeal. Our findings indicate that, among black American participants, chronic exposure to various forms of racial discrimination between the ages of 40 and 50 predicts an 11‐year increase in both serum p‐Tau181 and NfL. These findings support the hypothesis that unique life stressors encountered by Black Americans in midlife become biologically embedded and contribute to AD pathology and neurodegeneration later in life.
The AT(N) model of AD progression 22 posits a stepping‐stone approach such that amyloid‐beta plaque deposition leads to neurofibrillary tangles and ultimately neurodegeneration and cognitive decline. Somewhat surprisingly, our SEM analysis showed no evidence that increases in serum p‐Tau181 mediated, or partially mediated, the effect of discrimination on NfL.
It is possible that the absence of this effect in our analyses is a result of the age of our sample. Perhaps change in p‐Tau181 will begin to predict change in NfL as the sample enters their 60s and 70s. Alternatively, this finding may reflect differences in the particular aspects of neurodegeneration reflected by different biomarkers, resulting in more nuanced associations across different biomarkers. For example, racial discrimination may impact neurodegeneration through vascular pathways involving hypertension, atherosclerosis, or inflammation that exert their effect earlier in life, even in the absence of changes in tau. Notably, although the effect was not mediated by p‐Tau181, our results showed a persistent association between exposure to chronic discrimination and change in NfL. This finding suggests that racial discrimination contributes to greater neurodegeneration and is consistent with previous studies that have reported high levels of stress are associated with reductions in hippocampal 9 , 10 and prefrontal cortex 11 volumes.
Unlike p‐Tau181 and NfL, racial discrimination was not associated with serum GFAP levels, even despite the fact that NfL and GFAP were highly correlated. The reasons for this finding are not clear. One possibility is that GFAP captures different aspects of the cardiovascular pathway than NfL. Past studies indicate that GFAP is strongly associated with stroke and cardiac arrest, 23 but it may be less sensitive to the cardiovascular consequences of discrimination such as inflammation and atherosclerosis. 24 The results may suggest that discrimination is associated with greater beta‐amyloid deposition (reflected in serum p‐Tau181) and greater neuronal loss (reflected in NfL), but not necessarily greater astrogliosis (reflected in GFAP). This may suggest important complexity in processes associated with development of AD/ADRD and warrants additional consideration in future studies.
The major strengths of the current study were its use of longitudinal data with a Black American sample and the incorporation of both psychosocial variables and blood‐based biomarkers of dementia and neurodegeneration. Of course, there are also limitations. One of the major weaknesses was the absence of amyloid biomarkers. Unfortunately, a substantial proportion of our participants had non‐detectable levels of amyloid, likely due to the use of serum. However, several studies have shown that blood p‐Tau181 is a marker of amyloid pathology measured by amyloid PET or CSF. 17 , 25 , 26 , 27 , 28 Finally, while the current study investigated the role of racial discrimination as a stressor that contributes to the development of AD and neurodegeneration, it is likely that other race‐related stressors also increase the chances of ADRD. For example, past research has documented the effect of economic hardship, 29 , 30 neighborhood disadvantage, 31 , 32 , 33 and loneliness 29 on the development of chronic illness and accelerated aging among Black Americans. There is some evidence that these conditions, similar to racial discrimination, increase the chances of dementia as well. 6 Although such adverse and challenging circumstances would be expected to foster poor health and dementia among all racial/ethnic groups, they tend to be more prevalent among Black Americans. 34 All of this underscores the importance of investigating the everyday challenges and circumstances experienced by Black Americans as a strategy for identifying the factors that explain their increased risk for dementia. 2 , 3
CONFLICT OF INTEREST STATEMENT
M.M.M. has served on scientific advisory boards and/or has consulted for Acadia, Biogen, Eisai, LabCorp, Lilly, Merck, PeerView Institute, Roche, Siemens Healthineers, and Sunbird Bio. The other authors have no conflict of interest to report. Author disclosures are available in the supporting information.
CONSENT STATEMENT
All participants provided written informed consent. The study was approved by the Institutional Review Board at the University of Georgia.
Supporting information
Supporting Information
Supporting Information
ACKNOWLEDGMENTS
The authors express their gratitude to the study participants and staff involved in the data collection and data management of the Family and Community Health Study and the Center for Family Research. The funding sources had no role in the design and conduct of the study, the interpretation of the data, or preparation of this manuscript. This work was supported by the National Institute on Aging (RF1 AG077386 and R01 AG055393), and the National Heart, Lung, Blood Institute (R01 HL118045).
Simons RL, Ong ML, Lei M‐K, et al. Racial discrimination during middle age predicts higher serum phosphorylated tau and neurofilament light chain levels a decade later: A study of aging black Americans. Alzheimer's Dement. 2024;20:3485–3494. 10.1002/alz.13751
Contributor Information
Ronald L. Simons, Email: rsimons@uga.edu.
Michelle M. Mielke, Email: mmielke@wakehealth.edu.
REFERENCES
- 1. Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413‐446. doi: 10.1016/S0140-6736(20)30367-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Barnes LL. Alzheimer disease in African American individuals: increased incidence or not enough data? Nat Rev Neurol. 2022;18(1):56‐62. doi: 10.1038/s41582-021-00589-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Gleason CE, Zuelsdorff M, Gooding DC, et al. Alzheimer's disease biomarkers in Black and non‐Hispanic White cohorts: a contextualized review of the evidence. Alzheimers Dement. 2022;18(8):1545‐1564. doi: 10.1002/alz.12511 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Carter RT. Racism and psychological and emotional injury: recognizing and assessing race‐based traumatic stress. The Couns Psychol. 2007;35(1):13‐105. doi: 10.1177/0011000006292033 [DOI] [Google Scholar]
- 5. Geronimus AT, Hicken M, Keene D, Bound J. “Weathering” and age patterns of allostatic load scores among blacks and whites in the United States. Am J Public Health. 2006;96(5):826‐833. doi: 10.2105/AJPH.2004.060749 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Siddiqi A, Shahidi FV, Ramraj C, Williams DR. Associations between race, discrimination and risk for chronic disease in a population‐based sample from Canada. Soc Sci Med. 2017;194:135‐141. doi: 10.1016/j.socscimed.2017.10.009 [DOI] [PubMed] [Google Scholar]
- 7. Simons RL, Ong ML, Lei MK, et al. Changes in loneliness, BDNF, and biological aging predict trajectories in a blood‐based epigenetic measure of cortical aging: a study of older Black Americans. Genes. 2023;14(4):842. doi: 10.3390/genes14040842 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Lawrence WR, Jones GS, Johnson JA, et al. Discrimination experiences and all‐cause and cardiovascular mortality: multi‐ethnic study of atherosclerosis. Circ Cardiovasc Qual Outcomes. 2023;16(4):e009697. doi: 10.1161/CIRCOUTCOMES.122.009697 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Piccolo LR, Noble KG, Pediatric Imaging, Neurocognition, and Genetics Study. Perceived stress is associated with smaller hippocampal volume in adolescence. Psychophysiol. 2018;55(5):e13025. doi: 10.1111/psyp.13025 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Schoenfeld TJ, McCausland HC, Morris HD, Padmanaban V, Cameron HA. Stress and loss of adult neurogenesis differentially reduce hippocampal volume. Biol Psychiatry. 2017;82(12):914‐923. doi: 10.1016/j.biopsych.2017.05.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Ansell EB, Rando K, Tuit K, Guarnaccia J, Sinha R. Cumulative adversity and smaller gray matter volume in medial prefrontal, anterior cingulate, and insula regions. Biol Psychiatry. 2022;72(1):57‐64. doi: 10.1016/j.biopsych.2011.11.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Maruszak A, Thuret S. Why looking at the whole hippocampus is not enough—a critical role for anteroposterior axis, subfield and activation analyses to enhance predictive value of hippocampal changes for Alzheimer's disease diagnosis. Front Cell Neurosci. 2014;8:95. doi: 10.3389/fncel.2014.00095 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Barnes LL, Lewis TT, Begeny CT, Yu L, Bennett DA, Wilson RS. Perceived discrimination and cognition in older African Americans. J Int Neuropsychol Soc. 2012;18(5):856‐865. doi: 10.1017/S1355617712000628 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Coogan P, Schon K, Li S, Cozier Y, Bethea T, Rosenberg L. Experiences of racism and subjective cognitive function in African American women. Alzheimers Dement (Amst). 2020;12(1):e12067. doi: 10.1002/dad2.12067 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Zahodne LB, Kraal AZ, Sharifian N, Zaheed AB, Sol K. Inflammatory mechanisms underlying the effects of everyday discrimination on age‐related memory decline. Brain Behav Immun. 2019;75:149‐154. doi: 10.1016/j.bbi.2018.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Simons RL, Lei MK, Klopack E, Beach SRH, Gibbons FX, Philibert RA. The effects of social adversity, discrimination, and health risk behaviors on the accelerated aging of African Americans: further support for the weathering hypothesis. Soc Sci Med. 2021;282:113169. doi: 10.1016/j.socscimed.2020.113169 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Lee T, Shi D. A comparison of full information maximum likelihood and multiple imputation in structural equation modeling with missing data. Psychol Methods. 2021;26(4):466‐485. doi: 10.1037/met0000381 [DOI] [PubMed] [Google Scholar]
- 18. Landrine H, Klonoff EA. The schedule of racist events: a measure of racial discrimination and a study of its negative physical and mental health consequences. J Black Psychol. 1996;22(2):144‐168. 10.1177/00957984960222002 [DOI] [Google Scholar]
- 19. Downey RG, King CV. Missing data in Likert ratings: a comparison of replacement methods. J Gen Psychol. 1998;125:175‐191. doi: 10.1080/00221309809595542 [DOI] [PubMed] [Google Scholar]
- 20. Roth PL, Switzer FS, Switzer DM. Missing data in multiple item scales: a Monte Carlo analysis of missing data techniques. Organ Res Methods. 1999;2:211‐232. doi: 10.1177/109442819923001 [DOI] [Google Scholar]
- 21. Mielke MM, Dage JL, Frank RD, et al. Performance of plasma phosphorylated tau 181 and 217 in the community. Nat Med. 2022;28(7):1398‐1405. doi: 10.1038/s41591-022-01822-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Jack CR,Jr , Bennett DA, Blennow K, et al. NIA‐AA research framework: toward a biological definition of Alzheimer's disease. Alzheimers Dement. 2018;14(4):535‐562. doi: 10.1016/j.jalz.2018.02.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Humaloja J, Lahde M, Ashton N, et al. COMACARE Study Groups. GFAP and tau protein as predictors of nerurological outcome after out‐of‐hospital cardiac arrest: a post hoc analysis of the COMCARE trial. Resusciation. 2022;170:141‐149. doi: 10.1016/j.resuscitation.2021.11.033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Rosa G, Giannotti C, Martella L, et al. Brain aging, cardiovascular diseases, mixed dementia, and frailty in the oldest old: gran phenotype to clinical expression. J Alzheimer's Dis. 2020;75:1083‐1103. doi: 10.3233/JAD-191075 [DOI] [PubMed] [Google Scholar]
- 25. Anderson E. Black in White Space: The Enduring Impact of Color in Everyday Life. University of Chicago Press; 2022. [Google Scholar]
- 26. James SA. John Henryism and the health of African‐Americans. Cult Med Psychiatry. 1994;18(2):163‐182. doi: 10.1007/BF0137944 [DOI] [PubMed] [Google Scholar]
- 27. Brody GH, Yu T, Beach SRH. Resilience to adversity and the early origins of disease. Dev Psychopathol. 2016;28:1347‐1365. doi: 10.1017/S0954579416000894 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Mielke MM, Hagen CE, Xu J, et al. Plasma phospho‐tau181 increases with Alzheimer's disease clinical severity and is associated with tau‐ and amyloid‐positron emission tomography. Alzheimers Dement. 2018;14(8):989‐997. doi: 10.1016/j.jalz.2018.02.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Karikari TK, Pascoal TA, Ashton NJ, et al. Blood phosphorylated tau 181 as a biomarker for Alzheimer's disease: a diagnostic performance and prediction modelling study using data from four prospective cohorts. Lancet Neurol. 2020;19(5):422‐433. doi: 10.1016/S1474-4422(20)30071-5 [DOI] [PubMed] [Google Scholar]
- 30. Beach SRH, Klopack ET, Carter SE, et al. Do loneliness and per capita income combine to increase the pace of biological aging for Black Adults across late middle age? Int J Environ Res Public Health. 2022;19(20):13421. doi: 10.3390/ijerph192013421 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Simons RL, Lei MK, Klopach E, Berg M, Zhang Y, Beach SS. Re (Setting) epigenetic clocks: an important avenue whereby social conditions become biologically embedded across the life course. J Health Soc Behav. 2021;62(3):436‐453. doi: 10.1177/00221465211009309 [DOI] [PubMed] [Google Scholar]
- 32. Lei MK, Simons RL. The association between neighborhood disorder and health: exploring the moderating role of genotype and marriage. Int J Environ Res Public Health. 2021;18(3):898. doi: 10.3390/ijerph18030898 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Lei MK, Berg MT, Simons RL, Beach SRH. Neighborhood structural disadvantage and biological aging in a sample of Black middle age and young adults. Soc Sci Med. 2022;293:114654. doi: 10.1016/j.socscimed.2021.114654 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Hill‐Briggs F, Ephraim PL, Vrany EA, et al. Social determinants of health, race, and diabetes population health improvement: Black/African Americans as a population exemplar. Curr Diab Rep. 2022;22(3):117‐128. doi: 10.1007/s11892-022-01454-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information
Supporting Information