Non‐amnestic initial symptoms may be less prevalent in African American than White participants with Alzheimer's and mixed Lewy body and Alzheimer's pathology

Abstract

INTRODUCTION

Knowing the relationship between initial clinical symptoms and rate of clinical decline by race is important in dementia care.

METHODS

A retrospective longitudinal cohort study of 2887 participants with autopsy confirmed AD pathology (ADP), and mixed ADP with Lewy body pathology (AD‐LBP) in the National Alzheimer's Coordinating Center database. Two‐sample t‐tests and Fisher's exact tests were used to evaluate outcomes.

RESULTS

African American participants (AAs) were less likely to have initial non‐amnestic symptoms than White participants (6.5% vs 23% p < 0.001). AAs showed a faster rate of clinical decline on the Clinical Dementia Rating‐Sum of Boxes than Whites in the amnestic group (0.30 points per year, 95% confidence interval [CI]: 0.04 to 0.55; p = 0.023). When accounting for biological and social‐medical factors, the difference between the races were not significant (0.16, 95% CI: –0.10 to 0.42; p = 0.22).

DISCUSSION

Larger diverse cohorts are necessary to robustly evaluate interaction between race and clinical factors in determining dementia outcomes.

Highlights

• African American participants were less likely to have initial non‐amnestic symptoms than White participants with Alzheimer's disease and mixed Alzheimer's and Lewy body neuropathology.

• African American participants with initial amnestic symptoms in the National Alzheimer's Coordinating Center cohort also experienced a faster rate of cognitive and functional decline.

• Interaction between race, clinical symptoms, and neuropathology in diverse populations is an important consideration in clinical trial design.

1. BACKGROUND

Alzheimer's disease (AD) and Lewy body dementia (LBD) are among the most common neurodegenerative dementia neuropathologies. ¹ AD is characterized by the presence of amyloid beta (Aβ) plaques and hyperphosphorylated tau (p‐tau) protein, whereas Lewy body pathology (LBP) is characterized by α‐synuclein aggregates (Lewy bodies, Lewy neurites) as the underlying pathology. However, a significant proportion of patients with dementia have a mixed pathology with coexistent AD and LBP. ² , ³

Cognitively, AD dementia commonly presents with early amnestic symptoms ⁴ and LBD with early dysexecutive and visuospatial changes. ⁵ However, atypical presentations with initial language, dysexecutive, or visuospatial changes are not uncommon in AD, ⁶ as are initial amnestic or language changes in LBD. ⁷ , ⁸ , ⁹ Atypical presentations have been reported with younger age ¹⁰ and associated with apolipoprotein E (APOE) ε4 carrier status. ¹⁰ , ¹¹ The nature of initial symptoms has also been reported to impact the rate of clinical decline in AD and AD‐LBD, with those presenting non‐amnestic initial symptoms having a faster overall rate of clinical decline when compared to participants with amnestic initial symptoms. ⁹

Associations of race with health and disease are thought to reflect complex relationships that may range from systemic determinants to individual‐level social exposures impacting health and possibly genetic ancestry in some sub‐populations. ¹² , ¹³ Clearly it is important to recognize how these factors could impact health care delivery to help us develop more impactful clinical interventions in the future. Differences in dementia risk and outcomes by race and ethnicity have been known for some time. Notably, Black or African American (AA) and Hispanics have a higher risk of developing dementia, ¹⁴ whereas the AD risk by APOE ε4 carrier status is lower among AA and Hispanic individuals. ¹⁵ Mixed AD‐LBD is reported as more likely among AA participants as opposed to White participants. ¹⁶ AA participants with dementia have been reported to have a faster rate of clinical decline than White participants. ¹⁴ The limitations of designing and interpreting dementia clinical trials and counseling patients and families without considering a wider racial diversity is also now well recognized. ¹⁷ A significant challenge in the field is from missed and under diagnosis of AD and related dementias from reliance on clinical diagnosis alone. ¹⁸ , ¹⁹ , ²⁰ Discrepancies between clinical diagnoses and neuropathology in AD and LBD are also notably different between races. ²¹ , ²² A careful evaluation of the impact of race and clinical symptoms in a neuropathology‐confirmed dementia cohort has been highlighted as important to understand the underlying causes of these disparities. ²³

Black participants with dementia in the National Alzheimer's Coordinating Center (NACC) data set showed greater cognitive deficits, neuropsychiatric symptoms, and functional dependence. ²⁴ Variations in clinical presentation may impact diagnostic accuracy, given the racial differences in missed diagnosis. Differences in the frequency of initial symptoms by race and ethnicity could therefore be important in understanding early clinical presentations to help early diagnosis and evaluate clinical outcomes. The nature of initial cognitive symptoms has also been related to the subsequent rate of cognitive decline. ⁹ We therefore undertook a retrospective longitudinal cohort study of neuropathologically confirmed participants within the NACC data set, which includes data from multiple sites in the United States. As is typical for many cohort studies in the United States, there is a predominance of White participants in NACC. However, given that it is the largest cohort of neuropathology‐confirmed dementia participants that covers multiple sites in the United States, the number of minority participants in NACC with neuropathology‐confirmed dementia is also among the largest. This made the NACC neuropathology cohort attractive for a more careful analysis.

We aimed to determine if there were differences in the frequency of initial amnestic or non‐amnestic cognitive symptoms observed in AD pathology (ADP), and mixed ADP and Lewy body pathology (AD‐LBP) by race and if differences in amnestic versus non‐amnestic symptoms impact the rate of clinical decline across all races. We hypothesized that participants regardless of race with similar initial cognitive symptoms have similar rates of functional and cognitive decline.

2. METHODS

2.1. Participants and study design

This was a retrospective cross‐sectional study using the NACC dataset. The dataset includes participant information collected from 37 past and present Alzheimer's Disease Research Centers (ADRCs) funded by the National Institute on Aging (NIA). Data from the NACC Uniform Data Set (UDS) between June 2005 and December 2024, were used for the present analysis. Details on data collection and curation have been published previously. ²⁵ Additional details on the cohort relevant to the study is provided in the Supplementary Material section.

2.2. Standard protocol approvals, registrations, and patient consents

Prior to submitting data to the NACC, all contributing ADRCs are required to obtain informed consent from their participants and maintain separate institutional review board (IRB) review and approval from their institution.

2.3. Participant assessment

Data from participants in the current study were collected in the UDS in the following categories: White, Black or African American (AA), American Indian or Alaska Native, Native Hawaiian or Other Pacific Islander, Asian, Other, and Unknown. Neuropathology data related to AD among participants were noted as (1) “Intermediate” or “High” Alzheimer's disease neuropathic change (ADNC), considered sufficient explanation for dementia (per NIA/Alzheimer's Association 2012 guidelines for the neuropathologic assessment of AD ²⁶ without co‐existing LBP or (2) ADNC as above along with any coexistent LB neuropathology (AD‐LBP; brainstem‐predominant, amygdala‐predominant, limbic, neocortical Lewy bodies present, or LBP was noted but the region was unspecified). As there were eight AA participants with LBP without concomitant ADP, this group was not included in this analysis on the impact of race and needs a separate evaluation in future studies. The likelihood of co‐existing vascular pathology for each participant by the Hachinski Ischemic Scale at the time of initial evaluation was also characterized. ²⁷ Supplementary Material details co‐existing vascular pathology burden among all participants in this study, and Figure S1 shows the participant selection flow chart.

All participants had a Clinical Dementia Rating‐Global (CDR‐G) scale ≤1 at initial clinical visit. The CDR‐G scale assesses the current cognitive and functional status of the participant. The CDR‐G ratings range from 0 (no dementia) to 3 (severe dementia), ²⁸ with a CDR‐G of 1 representing the threshold of early dementia. Here, the analysis was limited to participants with CDR‐G ≤1 at initial visit to increase the reliability of the data regarding patients’ initial cognitive symptoms and to ensure that the data were reported when the cognitive symptoms were still early. ⁹

2.4. Initial cognitive symptoms and neuropsychological tests

For each participant, the predominant symptom first recognized as a decline in the subject's cognition was determined by a clinician after interviewing the patient and caregiver and documenting the symptom in the “NACCCOGF” variable in the NACC. Participants were characterized as primary amnestic (if the initial symptom was memory related), executive (if the initial symptoms were related to executive or attention/concentration changes), language (if initial symptom were related to language concerns), and visuospatial (if initial symptom was related to visuospatial concerns), as published previously (details in Supplementary Materials). ¹¹ , ²⁹ In addition, when present and documented at the last clinical visit, the frequency of atypical AD clinical syndrome diagnosis (primary progressive aphasia, posterior cortical atrophy) is provided in the Supplementary Material. The Clinical Dementia Rating‐Sum of Boxes (CDR‐SB) ²⁸ and Mini‐Mental State Examination (MMSE) ³⁰ scores were collected to characterize baseline cognitive and functional deficits, as well as their longitudinal progression. A core battery of neuropsychological measures was administered to all participants at each visit. ³¹ Details of the neurocognitive test panel are provided as Supplementary Material.

RESEARCH IN CONTEXT

1. Systematic review: A literature review was conducted using PubMed. The nature of initial presenting clinical symptoms has not been explored previously between races in Alzheimer's and Lewy body dementia. Prior studies evaluating differences between races on the clinical rate of progression in dementia have been cited.

2. Interpretation: The results of this study note that initial non‐amnestic symptoms are less likely to be reported in African American (AA) participants with Alzheimer's disease and mixed Alzheimer's Lewy body neuropathology. AA participants in this cohort also had a faster rate of clinical decline that was not significant after adjustment for biological and socio‐environmental factors.

3. Future directions: These findings point to the potential for racial diversity among participants in impacting longitudinal outcomes in dementia clinical trials. These findings warrant future studies to replicate these results in other community‐based dementia cohorts and in other neuropathologies.

2.5. Statistical analysis

Continuous measures were graphically assessed for normality. Normally distributed measures were compared using two‐sample t‐tests, and non‐normally distributed measures were compared using Wilcoxon rank sum tests. Categorical factors were compared with Fisher's exact test.

Given the small number of participants in other races (American Indian/Alaska Native, n = 4; Native Hawaiian/Pacific Islander, n = 1; Asian, n = 27; Other, n = 4), the analysis was limited to White and AA participants.

Due to missing data on APOE ε4 status (White amnestic, n = 167; White non‐amnestic, n = 47; AA amnestic, n = 15; AA non‐amnestic, n = 1) and education (White amnestic, n = 7; White non‐amnestic, n = 9; AA amnestic, n = 1; AA non‐amnestic, n = 0), multiple imputations using fully conditional specification were performed on the longitudinal data, creating 10 imputed data sets. Imputed values of baseline factors were created using models that included all baseline measures, in addition to outcome measures.

We noted that the cohort's follow‐up lengths were right‐skewed, with few participants having longitudinal follow‐ups over multiple years. A total of 1154 subjects (40%; 1112 White, 42 AA) died or were lost to follow‐up in the first 3 years. To limit the effects of survivorship bias, we limited longitudinal analysis to the first 3 years of follow‐up. The 3‐year cutoff was chosen because it was close to the median follow‐up length of the AA group, which was more susceptible to the effects of the non‐normal distribution due to its smaller sample size. A sensitivity analysis was performed to assess the effect of truncating the follow‐up data on the longitudinal analysis results.

Linear mixed‐effects (LME) models were fit to evaluate the racial differences in longitudinal outcomes. Models were fit for each imputed data set and pooled according to Rubin's rules. They were stratified by initial symptom and pathology to evaluate between‐group differences in trajectory. Due to the small sample size of non‐amnestic participants in the AA group (AA, n = 8; White, n = 625), non‐amnestic participants were omitted from longitudinal analyses stratified by pathology group. Hierarchical regression was performed by adding the baseline covariates (baseline score, age, sex, education, APOE ε4 status, hypertension, diabetes, and hypercholesterolemia) to the LME models in steps to evaluate their effect on the relationship between race and cognitive performance over time.

In order to assess the reliability of characterization of initial amnestic symptoms, baseline neurocognitive scores were assessed for amnestic and non‐amnestic domains. Neurocognitive score differences, by analysis of variance models, p‐values, and η ² statistics, were examined, and the least squares means with 95% confidence intervals (CIs) were estimated for each score.

All tests were two‐tailed and performed with a significance level of 0.05. η ² > 0.01 was considered a threshold effect size of significance for interpreting neurocognitive variable differences. All analyses were performed using R Statistical Software (version 4.3.1; Vienna, Austria).

2.6. D ata a vailability

NACC data are available on request from https://naccdata.org/.

3. RESULTS

The study included 2923 participants (1999 with AD; 924 with AD‐LBP) with a mean (SD) interval of 7.0 (3.1) years from initial visit to autopsy. Of the participants, 2763 identified as White, 124 as AA, 4 as American Indian/Alaska Native, 1 as Native Hawaiian/Other Pacific Islander, 27 as Asian, and 4 as Other race. Participants included those with 2286 amnestic and 637 non‐amnestic initial symptoms.

Due to small numbers of cases in most racial and ethnic groups when evaluated by initial cognitive symptoms, these subgroups were not included in further detailed analysis. Table 1 provides details on the AA and White group differences summarized below.

TABLE 1.

Patient characteristics.

Characteristic	White, N = 2763	Black/African American, N = 124	p‐value a
Age at baseline, mean (SD)	73 (10)	75 (10)	0.015
Age at onset of cognitive decline, mean (SD)	69 (10)	71 (10)	0.012
Age at death, median (IQR)	80 (73–87)	82 (76–88)	0.006
Sex, n (%)			0.021
Female	1219 (44)	68 (55)
Male	1544 (56)	56 (45)
APOE ε4 carrier status, n (%)			0.16
ε4 ‐ve	1009 (40)	35 (32)
ε4+ve	1540 (60)	73 (68)
Years of education, median (IQR)	16.00 (13.00–18.00)	14.00 (12.00–17.00)	<0.001
Hachinski ischemic score, median (IQR)	1.00 (0.00–1.00)	1.00 (0.00–1.00)	0.003
Hypertension, n (%)	1153 (42)	80 (65)	<0.001
Diabetes, n (%)	196 (7.1)	19 (15)	0.002
Hypercholesterolemia, n (%)	1304 (48)	70 (56)	0.066
CD‐SB, median (IQR)	4.00 (2.00–5.50)	4.00 (2.50–6.00)	0.27
Total MMSE score, median (IQR)	24.0 (21.0–27.0)	23.0 (19.0–26.0)	0.050
Initial symptom, n (%)			<0.001
Amnestic	2138 (77)	116 (94)
Non‐amnestic	625 (23)	8 (6.5)
Pathology, n (%)			0.17
ADP	1881 (68)	92 (74)
AD‐LBP	882 (32)	32 (26)

Abbreviations: SD, standard deviation; IQR, interquartile range; ADP, Alzheimer's disease pathology; AD‐LBP, mixed ADP and Lewy body pathology, CDR‐SB, Clinical Dementia Rating Scale Sum of Boxes, MMSE, Mini mental State Exam

^a Welch two‐sample t‐test; Wilcoxon rank sum test; Fisher's exact test.

Bold values denote p < 0.05.

In this cohort, White participants were more likely to be male (56% vs 46%, p = 0.021) and have higher education years (16 vs 14, p < 0.001) when compared to AA participants. Baseline MMSE scores for White participants also suggested lower clinical impairment at initial visit than for AA participants (24 vs 23, p = 0.050). However, AA participants reported initial cognitive symptoms at an older age than White participants (71 vs 69, p = 0.012). APOE ε4 status was similar between AA and White participants (68% vs 60% p = 0.16).

3.1. Vascular disease comorbidities

AA participants were more likely to have hypertension (65% vs 42%, p < 0.001) and diabetes (15% vs 7.1%, p = 0.002) than White participants at baseline.

3.2. Differences in the frequency of initial amnestic or non‐amnestic cognitive symptoms by race

AA participants were more likely to have initial amnestic symptoms than White participants (94% vs 77%, p < 0.001). In addition, AA participants were more likely to present with initial amnestic than non‐amnestic symptoms in each of the ADP and AD‐LBP groups. APOE ε4 status was similar within the amnestic group in both races (68% vs 63%, p = 0.29) (Table 2). Furthermore, among Whites, the classical clinical diagnoses of primary progressive aphasia (PPA) and posterior cortical atrophy (PCA) were higher among those with initial non‐amnestic symptoms than amnestic initial symptoms (Table S1). Among AAs, there were no autopsy‐proven cases of PPA and PCA. Given the small non‐amnestic participant numbers in the AA group (n = 8), additional inferences were limited.

TABLE 2.

Amnestic and non‐amnestic symptom frequency by race and pathology.

Group	Characteristic	Amnestic	Non‐amnestic	p‐value a
ADP	Race, n (%)			<0.001
	White	1446 (94)	435 (98)
	Black/African American	85 (5.6)	7 (1.6)
AD‐LBP	Race, n (%)			0.007
	White	692 (96)	190 (99)
	Black/African American	31 (4.3)	1 (0.5)

Abbreviations: ADP, Alzheimer's disease pathology; AD‐LBP, mixed ADP and LBP; LBP, Lewy body pathology.

^a Fisher's exact test.

Bold values denote p < 0.05.

3.3. Linear mixed‐effects models for rate of decline comparing amnestic versus non‐amnestic by race

AA participants had a faster rate of clinical progression on the CDR‐SB than White participants in the initial amnestic symptom group (0.30 points per year, 95% CI: 0.04 to 0.55; p = 0.023) (Figure 1 and Table 3). After adjusting for age, sex, education, APOE ε4 carrier status, baseline cognitive score, and vascular risk factors as covariates, this slope difference was no longer statistically significant (Figure 2, Table 4). Hierarchical regression analysis indicated that among the covariates of interest, adjusting for baseline cognitive score and biological factors of age, sex, and APOE ε4 carrier status reduced the estimated slope difference the most (Table 5). As the number of non‐amnestic participants was smaller in the AA group, the inferences on differences in slope of cognitive decline were not made.

FIGURE 1.

Unadjusted linear mixed models evaluating the rate of (A) CDR‐SB and (B) MMSE change across African American versus White patients with amnestic and non‐amnestic initial symptoms, restricted to the first 3 years of follow‐up. Slope differences. CDR‐SB: amnestic β 0.30 [95% CI, 0.04 to 0.55), p = 0.023; non‐amnestic β –0.38 [95% CI, –1.4 to 0.65], p = 0.47; overall β 0.21 [95% CI, –0.04 to 0.46], p = 0.11. MMSE: amnestic β –0.23 [95% CI, –0.64 to 0.19], p = 0.28; non‐amnestic β –0.19 [95% CI, –2.0 to 1.6], p = 0.84; overall β –0.14 [95% CI, –0.56 to 0.27], p = 0.50. CDR‐SB, clinical dementia rating scale sum of boxes; MMSE, mini mental state exam.

TABLE 3.

Unadjusted longitudinal models for CDR‐SB and MMSE trajectories over 3 years across the whole cohort.

		Amnestic		Non‐amnestic		Whole cohort
Model	Characteristic a	Beta (95% CI)	p‐value	Beta (95% CI)	p‐value	Beta (95% CI)	p‐value
CDR‐SB	Slope	1.6 (1.5 to 1.7)	<0.001	1.9 (1.8 to 2.0)	<0.001	1.7 (1.6 to 1.7)	<0.001
	Race
	White	—		—		—
	Black/African American	0.27 (−0.37 to 0.91)	0.41	−0.01 (−2.5 to 2.5)	>0.99	0.29 (−0.33 to 0.91)	0.36
	Slope * Race
	Slope * Black/African American	0.30 (0.04 to 0.55)	0.023	−0.38 (−1.4 to 0.65)	0.47	0.21 (−0.04 to 0.46)	0.11
MMSE	Slope	−1.8 (−1.9 to −1.7)	<0.001	−2.8 (−3.1 to −2.6)	<0.001	−1.9 (−2.0 to −1.9)	<0.001
	Race
	White	—		—		—
	Black/African American	−0.98 (−2.1 to 0.13)	0.083	−0.56 (−5.4 to 4.2)	0.82	−0.76 (−1.9 to 0.37)	0.19
	Slope * Race
	Slope * Black/African American	−0.23 (−0.64 to 0.19)	0.28	−0.19 (−2.0 to 1.6)	0.84	−0.14 (−0.56 to 0.27)	0.50

Abbreviations: AA, African American; CDR‐SB, Clinical Dementia Rating‐Sum of Boxes; CI, confidence interval; MMSE, Mini‐Mental State Examination.

^a African American * Slope indicates the interaction between being in the AA group and the slope.

Bold values denote p < 0.05.

FIGURE 2.

Linear mixed models evaluating rate of (A) CDR‐SB and (B) MMSE change across African American versus White patients, adjusted for baseline covariates. Slope differences. CDR‐SB: amnestic β 0.16 [95% CI, –0.10 to 0.42], p = 0.22; non‐amnestic β –0.67 [95% CI, –1.8 to 0.43], p = 0.23; overall β 0.06 [95% CI, –0.20 to 0.32], p = 0.64. MMSE: amnestic β –0.17 [95% CI, –0.59 to 0.25], p = 0.43; non‐amnestic β 0.14 [95% CI, –1.7 to 2.0], p = 0.88; overall β –0.06 [95% CI, –0.48 to 0.37], p = 0.79. CDR‐SB, clinical dementia rating scale sum of boxes; MMSE, mini mental state exam.

TABLE 4.

Longitudinal models for CDR‐SB and MMSE trajectories over 3 years across the whole cohort, adjusted for baseline covariates.

		Amnestic		Non‐amnestic		Whole cohort
Model a	Characteristicb	Beta (95% CI)c	p‐value	Beta (95% CI)c	p‐value	Beta (95% CI)c	p‐value
CDR‐SB	Race
	White	—		—		—
	Black/African American	0.01 (–0.42 to 0.45)	0.96	0.34 (–1.5 to 2.1)	0.71	0.03 (–0.40 to 0.46)	0.90
	Slope	1.6 (1.5 to 1.6)	<0.001	1.9 (1.8 to 2.1)	<0.001	1.7 (1.6 to 1.7)	<0.001
	Baseline CDR‐SB	1.2 (1.2 to 1.2)	<0.001	1.3 (1.2 to 1.3)	<0.001	1.2 (1.2 to 1.2)	<0.001
	Baseline age	0.00 (–0.01 to 0.00)	0.29	−0.01 (–0.03 to 0.01)	0.22	−0.01 (–0.02 to 0.00)	0.014
	Sex
	Female	—		—		—
	Male	0.02 (–0.13 to 0.17)	0.78	−0.10 (–0.42 to 0.23)	0.57	0.01 (–0.13 to 0.14)	0.94
	Education	−0.01 (–0.03 to 0.02)	0.54	0.03 (–0.03 to 0.09)	0.31	0.00 (–0.02 to 0.03)	0.85
	APOE ε4 carrier	−0.22 (–0.37 to –0.06)	0.007	0.03 (–0.29 to 0.34)	0.87	−0.20 (–0.34 to –0.06)	0.005
	Hypertension	−0.07 (–0.23 to 0.09)	0.39	0.41 (0.06 to 0.76)	0.021	0.02 (–0.13 to 0.16)	0.83
	Diabetes	−0.22 (–0.51 to 0.07)	0.14	−0.18 (–0.82 to 0.46)	0.58	−0.21 (–0.48 to 0.06)	0.12
	Hypercholesterolemia	−0.05 (–0.21 to 0.10)	0.50	−0.28 (–0.62 to 0.05)	0.10	−0.11 (–0.25 to 0.03)	0.13
	Slope difference
	Black/African American * Slope	0.16 (–0.10 to 0.42)	0.22	−0.67 (–1.8 to 0.43)	0.23	0.06 (–0.20 to 0.32)	0.64
MMSE	Race
	White	—		—		—
	Black/African American	0.02 (–0.64 to 0.68)	0.95	−0.69 (–3.4 to 2.1)	0.62	−0.06 (–0.72 to 0.60)	0.86
	Slope	−1.8 (–1.9 to –1.7)	<0.001	−2.9 (–3.2 to –2.7)	<0.001	−2.0 (–2.1 to –1.9)	<0.001
	Baseline MMSE	1.1 (1.0 to 1.1)	<0.001	1.1 (1.0 to 1.1)	<0.001	1.1 (1.0 to 1.1)	<0.001
	Baseline age	0.04 (0.03 to 0.05)	<0.001	0.05 (0.02 to 0.08)	0.002	0.05 (0.04 to 0.06)	<0.001
	Sex
	Female	—		—		—
	Male	−0.10 (–0.33 to 0.13)	0.38	0.02 (–0.55 to 0.59)	0.95	−0.09 (–0.31 to 0.12)	0.41
	Education	−0.02 (–0.06 to 0.02)	0.38	−0.05 (–0.15 to 0.05)	0.32	−0.03 (–0.06 to 0.01)	0.15
	APOE ε4 carrier	0.00 (–0.23 to 0.24)	0.99	−0.42 (–0.97 to 0.12)	0.13	−0.01 (–0.23 to 0.21)	0.93
	Hypertension	−0.04 (–0.29 to 0.20)	0.72	−0.40 (–0.99 to 0.20)	0.19	−0.10 (–0.33 to 0.13)	0.38
	Diabetes	−0.01 (–0.45 to 0.44)	0.97	0.50 (–0.58 to 1.6)	0.36	0.06 (–0.35 to 0.48)	0.77
	Hypercholesterolemia	0.16 (–0.08 to 0.39)	0.19	0.16 (–0.41 to 0.73)	0.58	0.15 (–0.07 to 0.37)	0.19
	Slope difference
	Black/African American * Slope	−0.17 (–0.59 to 0.25)	0.43	0.14 (–1.7 to 2.0)	0.88	−0.06 (–0.48 to 0.37)	0.79

Abbreviations: AA, African American; APOE, apolipoprotein E; CDR‐SB, Clinical Dementia Rating‐Sum of Boxes; CI, confidence interval; MMSE, Mini‐Mental State Examination.

^a African American * Slope indicates the interaction between being in the AA group and the slope.

Bold values denote p < 0.05.

TABLE 5.

Longitudinal models for CDR‐SB trajectories in the amnestic group after adjusting for various covariate groups.

	No adjustment		Stage of disease		Social/environmental		Biological
Characteristic a	Beta (95% CI)	p‐value	Beta (95% CI)	p‐value	Beta (95% CI)	p‐value	Beta (95% CI)	p‐value
Race
White	—		—		—		—
Black/African American	0.27 (–0.37 to 0.91)	0.41	−0.02 (–0.43 to 0.39)	0.93	0.02 (–0.40 to 0.43)	0.94	0.01 (–0.42 to 0.45)	0.96
Slope	1.6 (1.5 to 1.7)	<0.001	1.6 (1.6 to 1.7)	<0.001	1.6 (1.6 to 1.7)	<0.001	1.6 (1.5 to 1.6)	<0.001
Race * Slope
Black/African American * Slope	0.30 (0.04 to 0.55)	0.023	0.24 (–0.01 to 0.48)	0.057	0.23 (–0.01 to 0.48)	0.065	0.16 (–0.10 to 0.42)	0.22
Baseline CDR‐SB			1.2 (1.2 to 1.3)	<0.001	1.2 (1.2 to 1.3)	<0.001	1.2 (1.2 to 1.2)	<0.001
Education (years)					−0.01 (–0.03 to 0.02)	0.64	−0.01 (–0.03 to 0.02)	0.54
Hypertension					−0.02 (–0.18 to 0.13)	0.75	−0.07 (–0.23 to 0.09)	0.39
Diabetes					−0.20 (–0.47 to 0.08)	0.17	−0.22 (–0.51 to 0.07)	0.14
Hypercholesterolemia					−0.10 (–0.25 to 0.06)	0.21	−0.05 (–0.21 to 0.10)	0.50
Age at baseline							0.00 (–0.01 to 0.00)	0.29
Sex
Female							—
Male							0.02 (–0.13 to 0.17)	0.78
APOE ε4 carrier							−0.22 (–0.37 to –0.06)	0.007

Abbreviations: AA, African American; APOE, apolipoprotein E; CI, confidence interval; CDR‐SB, Clinical Dementia Rating‐Sum of Boxes.

^a African American * Slope indicates the interaction between being in the AA group and the slope.

Bold values denote p < 0.05.

3.4. Linear mixed‐effects models for CDR‐SB total score across neuropathology groups and race

When evaluated with each of the ADP and AD‐LBP groups, there were no differences by race on the rate of CDR‐SB progression or rate of MMSE progression (Table 6).

TABLE 6.

Longitudinal models for CDR‐SB trajectories over three years in ADP and AD‐LBP groups.

	ADP		AD‐LBP
Characteristic a	Beta (95% CI)	p‐value	Beta (95% CI)	p‐value
Race
White	—		—
Black/African American	0.00 (–0.52 to 0.51)	0.99	0.13 (–0.67 to 0.93)	0.75
Slope	1.7 (1.7 to 1.8)	<0.001	1.5 (1.4 to 1.6)	<0.001
Baseline CDR‐SB	1.2 (1.2 to 1.2)	<0.001	1.2 (1.2 to 1.3)	<0.001
Baseline age	−0.01 (–0.02 to 0.00)	0.010	0.00 (–0.02 to 0.01)	0.44
Sex
Female	—		—
Male	0.04 (–0.13 to 0.21)	0.64	−0.04 (–0.27 to 0.19)	0.72
Education	0.00 (–0.03 to 0.03)	0.86	0.01 (–0.03 to 0.05)	0.57
E4 carrier	−0.20 (–0.38 to –0.03)	0.019	−0.13 (–0.37 to 0.10)	0.27
Hypertension	0.04 (–0.14 to 0.22)	0.64	−0.06 (–0.31 to 0.20)	0.67
Diabetes	−0.13 (–0.45 to 0.20)	0.45	−0.43 (–0.90 to 0.05)	0.079
Hypercholesterolemia	−0.05 (–0.22 to 0.13)	0.60	−0.25 (–0.49 to –0.01)	0.045
Slope difference
Black/African American * Slope	−0.03 (–0.34 to 0.29)	0.87	0.23 (–0.23 to 0.69)	0.32

Abbreviations: AA, African American; ADP, Alzheimer's disease pathology; AD‐LBP, mixed ADP and Lewy body pathology; CDR‐SB, Clinical Dementia Rating‐Sum of Boxes; CI, confidence interval; MMSE, Mini‐Mental State Examination.

^a African American * Slope indicates the interaction between being in the AA group and the slope.

Bold values denote p < 0.05.

3.5. Neurocognitive score baseline differences in White and AA groups

Among participants with neurocognitive scores at the first visit, the comparative cognitive test performance profile of participants for each of the four initial cognitive symptoms was evaluated. A summary of neurocognitive results showing notable differences between the initial cognitive symptom groups is shown in Figures S2 and S3, and the comprehensive results for each pathology group are presented in Tables S1 and S2.

As expected, the initial amnestic symptom group had significantly lower scores on Logical Memory delayed recall across all pathology groups (Figure S2).

The initial non‐amnestic symptom group scored higher in immediate and delayed recall, but lower in the Digit Span test, Category Fluency (animals/vegetables), and the Boston Naming Test (BNT) among the White participants. The amnestic group scored higher than the non‐amnestic group on the wechsler adult intelligence scale IV (WAIS‐IV) digit symbol test among White participants, but lower among AA participants.

When evaluating differences between races on baseline cognitive scores, it is worth noting that the 95% confidence intervals (CI) for AA participants’ mean baseline cognitive scores are much wider than those for White participants, making comparisons based only on mean values challenging. For example, in the amnestic group, White participants had a CI width of 0.39 for immediate recall and 0.37 for delayed recall, whereas AA participants had a CI width of 1.7 for immediate recall and 1.6 for delayed recall (Figure S3). This difference is especially striking in the non‐amnestic group, with immediate recall and delayed recall at 5.9 and 5.6, respectively, for AA participants compared to 1.1 for both in White participants. However, the standard deviations (SDs) of scores between the AA and White participants are similar (amnestic group: immediate recall, 4.3 vs 4.1, delayed recall, 4.1 vs 3.9; non‐amnestic group: immediate recall, 2.4 vs 4.6, delayed recall, 2.4 vs 4.5). The wider CIs observed for test scores in the AA group are therefore likely indicative of the smaller sample size rather than a higher variance of scores between participants, as the SDs of scores are similar between AA and Whites.

4. DISCUSSION

In this study, we evaluated racial differences in the frequency of initial amnestic or non‐amnestic cognitive symptoms with ADP and AD‐LBP neuropathology, and their impact on rates of clinical decline. We determined that AA participants were less likely to present with initial non‐amnestic than amnestic symptoms in both “pure” and mixed AD with LBP neuropathology. Given the limited published data among neuropathology‐confirmed AA dementia participants, this information on initial clinical presentations and their differences between races is significant, but it needs corroboration in additional diverse community‐based autopsy cohorts.

4.1. Racial differences in initial symptoms

As there is a limited body of research on racial differences in initial cognitive symptoms reported in AD and LBD, we considered some likely reasons for our key result. Age and APOE ε4 carrier status are known factors for non‐amnestic presentations in AD, which are often seen with younger age at onset and a lower frequency of APOE ε4. ¹⁰ , ¹¹ Despite the slightly older age of the AA group in this cohort compared to the White group (76 vs 74 years), the frequency of participants <65 years (early onset of AD) was not significantly lower in the AA group than in the White group (15% vs 20%, p = 0.11). In this cohort, the frequency of APOE ε4 carriers also did not differ by race. Misunderstanding and misclassification of non‐amnestic symptoms as amnestic in the AA group is another possibility, as cultural factors and their impact on clinical presentation and diagnosis of AD and LBD are reported. ²¹ , ²² , ³² When the nature of initial clinical symptoms was corroborated with baseline neurocognitive scores, we saw that the cognitive scores were reflective of amnestic and non‐amnestic groups in both races. It is therefore less likely that AA non‐amnestic initial presentations were misclassified as amnestic on average in this cohort. These results challenge us to evaluate other potential biological and cultural factors and need to be confirmed in cohorts with larger number of Black participants overall to evaluate potential biases in minority recruitment in the NACC cohort and the impact of sample size as a determinant of these results in future investigations.

4.2. Racial differences in clinical trajectories

It is thought that there is probably not a strong difference in trajectories of cognitive change by race in the context of normal aging. ³³ , ³⁴ In the current study with well‐characterized dementia neuropathology, the AA group had a faster rate of decline than the White group in the amnestic group, with the difference in annual CDR‐SB slope being 0.30 points per year. The difference between races on CDR‐SB decline is slightly smaller than the difference in CDR‐SB, at 18 months for the lecanemab Phase III trial, which was 0.45. ³⁵ In this cohort, adjusting for baseline cognitive score and biological factors of age, sex, and APOE ε4 carrier status reduced the estimated slope difference the most.

The impact of baseline cognitive scores on diminishing the difference in rate of decline between the races suggests that the AA group presented in the clinic at a later stage of the disease, when a faster rate of progression on the CDR‐SB is expected. ³⁶ It has been noted previously that AA participants often exhibit greater severity of symptoms at the time of presentation, ²⁴ and are less likely to receive early AD medications such as cholinesterase inhibitors or memantine. ³⁷ In support of this possibility, in our cohort, the AA group had a higher baseline CDR‐SB and lower baseline MMSE score than the White group. The differences in baseline scores persist when examined in each of the pathologies.

A second possibility for the faster rate of decline could be related to lower cognitive reserve. ³⁸ , ³⁹ Higher cognitive reserve may slow the rate of decline earlier in the dementia course. ⁴⁰ Educational attainment is a proxy for cognitive reserve, ³⁹ , ⁴⁰ , ⁴¹ and the AA group had a lower educational attainment than the White group in this cohort. However, in the hierarchical regression, educational attainment was not a significant factor in reducing the estimated slope difference. We also recognize that the current analysis is not able to address racial differences that could be the underlying drivers of the rate of decline, such as additional differences in other sociodemographic variables or social determinants of health.

Taken together, this suggests that future clinical trials could benefit from better stratification of the cohort between amnestic and non‐amnestic initial presentations given differences in rate of decline. ⁹ In addition, accounting for biological factors of age, sex, and APOE ε4 carrier status and awareness of differences in severity of disease between the races at initial visit could help in better interpretation results.

4.3. Stage of dementia on trajectories

Participants in this study were in the mild cognitive impairment (MCI) or early stages of dementia at initial visit. Unlike the results from clinical stages in this study, AA participants with preclinical AD in the Harvard Brain Aging Study had a faster rate of decline on the Preclinical Alzheimer Cognitive Composite‐5 (PACC5) than other groups, despite accounting for educational factors, vascular factors, and Aβ burden. ⁴² However, when neither biomarker nor neuropathology confirmation of an AD diagnosis was used, a slower decline in AA participants than White participants was reported. ⁴³ , ⁴⁴ Taken together, these results suggest that additional features, including stage of neurodegenerative disease (preclinical vs clinical) and better characterization of neuropathology and underlying social and environmental factors, and the nature of the cohort, its recruitment and severity of symtoms at initial visit also need to be considered in understanding the differences in clinical outcomes between racial groups.

4.4. Vascular disease and other comorbidities

An additional possibility is the role of medical co‐morbidity burden, including social and economic deprivation, impacting the rate of decline. ⁴⁵ In the current study, despite there being differences in classical vascular risk factors between the groups that have been well documented in the NACC data, such differences did not contribute significantly to differences in the slope of decline in the hierarchical regression model. Given the current overall results, evaluating the contribution of additional medical and social factors in impacting differences in rate of progression by racial group is a significantly broader research question that needs to be taken up next.

4.5. Limitations and strengths

Differences in cultural beliefs about memory and aging can result in underreporting or misunderstanding of early cognitive symptoms. ⁴⁶ The impact of cultural factors on clinical presentation and diagnosis in clinical diagnosis of AD, DLB, and frontotemporal dementia is also well known. ²¹ , ²² , ³² , ⁴⁷ This highlights the need to report the current state of how large datasets, including NACC, have been capturing data on initial clinical presentations and their differences across races. NACC uses documentation of initial cognitive symptoms data by clinicians trained in dementia evaluation based on patient and/or family reported clinical history. Therefore, findings are impacted by clinician skill and subjectivity. Our validation analysis indicates that the clinician judgement of initial symptoms broadly matched the neurocognitive profile for memory and non‐memory domains, which suggests the clinical relevance of the initial cognitive symptoms data. Our study is also unique in having a large number of autopsy‐confirmed neuropathology diagnoses in the AA and White groups.

The use of the CDR‐G score to determine participants with early stages of dementia or MCI also has some limitations. Some CDR‐G domains (e.g., behavior, language) are not characterized and participants with these symptoms could be much further along in the disease course. Clinician ratings on the CDR may be influenced by differences in informant characteristics ⁴⁸ and caregiver experiences and how they are reported based on cultural expectations of caregiving and aging. ³² In addition, there is inadequate normative data for dementia specific scales in Black samples. ⁴⁹ The higher frequency of non‐amnestic symptoms in Whites may also be influencing the rate of decline, as longitudinal MMSE scores with frontal and language atypical AD variants ⁵⁰ and CDR and MMSE among non‐amnestic initial symptoms have been reported to have faster rates of progression. ⁹ Despite the large size of the initial NACC cohort, sub‐stratification for analysis also reduces the size of some subgroups. Severe co‐existing vascular pathology was minimal and reported rigorously, as noted in the Supplementary Methods and, therefore, is less likely to account for variability in clinical trajectories.

Missing data were addressed by multiple imputation using fully conditional specification by creating multiple imputation data sets. Given the strengths and biases of this NACC cohort (see Supplementary Materials), it is likely that the current results are generalizable to other prospective research cohorts, including clinical trials. These results could improve clinical trial design through real‐world validity in clinical trials that have limited racial diversity, as well as the nature of dementia care for underrepresented groups.

5. CONCLUSIONS

Our findings in this cohort with neuropathology‐confirmed diagnoses indicate that initial non‐amnestic symptoms are less frequent in AA participants compared to White participants. Among those with initial amnestic symptoms, AA participants had a faster rate of decline than White participants in both ADP and AD‐LBP. These results are important in both clinical trial design and individual patient management in these dementias. These results warrant a closer evaluation of the interaction between race and neuropathology in determining clinical outcomes in diverse populations.

AUTHOR CONTRIBUTIONS

Quintasha Beamon conceptualized study, wrote the first draft, and interpreted the data. Audrey Zhu analyzed the data and revised the manuscript for intellectual content. James B. Leverenz interpreted the data and revised the manuscript for intellectual content. Jagan A. Pillai designed and conceptualized study, interpreted the data, and drafted the manuscript for intellectual content.

CONFLICT OF INTEREST STATEMENT

Quintasha Beamon and Audrey Zhu declare no conflicts of interest. James B. Leverenz received research funding from the National Institutes of Health, GE Healthcare, the Lewy Body Dementia Association, Alzheimer's Association, Jane and Lee Seidman Endowed Chair for Advanced Neurological Education, Cleveland Clinic, and Douglas Herthel DVM Memorial Fund. Jagan A. Pillai received research funding from the National Institutes of Health, Department of Defense, Annabelle Foundation, Keep Memory Alive Foundation, and the Iverson Family Endowed Chair for Alzheimer's Disease Research. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All human subjects provided informed consent in this study. All contributing ADRCs are required to obtain written informed consent from their participants and maintain their own separate institutional review board review and approval from their institution before submitting data to the National Alzheimer's Coordinating Center (NACC).

Supporting information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Supporting Information

Beamon Q, Zhu A, Leverenz JB, Pillai JA. Non‐amnestic initial symptoms may be less prevalent in African American than White participants with Alzheimer's and mixed Lewy body and Alzheimer's pathology. Alzheimer's Dement. 2025;21:e70360. 10.1002/alz.70360