Relationships of Cognitive Measures with Cerebrospinal Fluid but not Imaging Biomarkers of Alzheimer Disease Vary between Black and White Individuals

Samuele Bonomi; Ruijin Lu; Suzanne E Schindler; Quoc Bui; James J Lah; David Wolk; Carey E Gleason; Reisa Sperling; Erik D Roberson; Allan I Levey; Leslie Shaw; Carol Van Hulle; Tammie Benzinger; Morgann Adams; Cecelia Manzanares; Deqiang Qiu; Jason Hassenstab; Krista L Moulder; Joyce E Balls-Berry; Keith Johnson; Sterling C Johnson; Charles F Murchison; Jingqin Luo; Emily Gremminger; Folasade Agboola; Elizabeth A Grant; Russ Hornbeck; Parinaz Massoumzadeh; Sarah Keefe; Donna Dierker; Julia D Gray; Rachel L Henson; Marissa Streitz; Dawn Mechanic-Hamilton; John C Morris; Chengjie Xiong

doi:10.1002/ana.26838

. Author manuscript; available in PMC: 2025 Mar 1.

Published in final edited form as: Ann Neurol. 2023 Dec 19;95(3):495–506. doi: 10.1002/ana.26838

Relationships of Cognitive Measures with Cerebrospinal Fluid but not Imaging Biomarkers of Alzheimer Disease Vary between Black and White Individuals

Samuele Bonomi ^A,^*, Ruijin Lu ^B,^*, Suzanne E Schindler ^A,^C, Quoc Bui ^B, James J Lah ^D,^E, David Wolk ^F, Carey E Gleason ^G,^H,^I, Reisa Sperling ^J, Erik D Roberson ^K, Allan I Levey ^D,^E, Leslie Shaw ^F,^L, Carol Van Hulle ^G,^H, Tammie Benzinger ^C,^M, Morgann Adams ^F, Cecelia Manzanares ^D,^E, Deqiang Qiu ^E, Jason Hassenstab ^A,^C, Krista L Moulder ^A,^C, Joyce E Balls-Berry ^A,^C, Keith Johnson ^N, Sterling C Johnson ^G,^H,^I, Charles F Murchison ^K, Jingqin Luo ^B, Emily Gremminger ^A, Folasade Agboola ^B,^C, Elizabeth A Grant ^B,^C, Russ Hornbeck ^M, Parinaz Massoumzadeh ^M, Sarah Keefe ^M, Donna Dierker ^M, Julia D Gray ^C, Rachel L Henson ^A,^C, Marissa Streitz ^A,^C, Dawn Mechanic-Hamilton ^F, John C Morris ^A,^C, Chengjie Xiong ^B,^C

^ADepartment of Neurology, Washington University School of Medicine in St. Louis, St. Louis, MO, USA.

^BDivision of Biostatistics, Washington University School of Medicine in St. Louis, St. Louis, MO, USA.

^CKnight Alzheimer Disease Research Center, Washington University School of Medicine in St. Louis, St. Louis, MO, USA.

^DDepartment of Neurology, Emory University School of Medicine, Atlanta, Georgia, USA.

^EGoizueta Alzheimer’s Disease Research Center, Emory University, Atlanta, GA

^FDepartment of Neurology, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA

^GDivision of Geriatrics and Gerontology, Department of Medicine, University of Wisconsin School of Medicine and Public Health, Madison, Wisconsin, USA.

^HWisconsin Alzheimer’s Disease Research Center, Madison, Wisconsin, USA.

^IGeriatric Research, Education and Clinical Center, William S. Middleton Memorial Veterans Hospital, Madison, Wisconsin, USA.

^JDepartment of Neurology, Center for Alzheimer Research and Treatment, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA.

^KCenter for Neurodegeneration and Experimental Therapeutics, Alzheimer’s Disease Center, Department of Neurology, University of Alabama at Birmingham, Birmingham, AL, USA

^LDepartment of Pathology and Laboratory Medicine, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, USA

^MMallinckrodt Institute of Radiology, Washington University School of Medicine in St. Louis, St. Louis, MO, USA

^NDepartment of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.

=co-first authors

Author contributions

SB, RL, SS, QB, JL, DW, CG, RS, ER, AL, LS, CH, JM, and CX contributed to the conception and design of the study; SB, RL, SS, QB, JL, DW, CG, RS, ER, AL, LS, CH, TB, MA, CM, DQ, JH, KM, JB, KJ, SJ, CM, JL, EM, FA, EG, RH, PM, SK, DD, JG, RH, MS, DM, JM, and CX contributed to the acquisition and analysis of data; SB, RL, SS, QB, JL, DW, CG, RS, ER, AL, LS, CH, JH, KM, JB, SJ, CM, JL, PM, JM, and CX contributed to drafting the text or preparing figures.

^✉

Corresponding author: Chengjie Xiong (chengjie@wustl.edu), Division of Biostatistics, Washington University School of Medicine in St. Louis, St. Louis, MO, USA

PMCID: PMC10922199 NIHMSID: NIHMS1948578 PMID: 38038976

Abstract

Objective:

Biomarkers of Alzheimer disease vary between groups of self-identified Black and White individuals in some studies. This study examined whether the relationships between biomarkers or between biomarkers and cognitive measures varied by racialized group.

Methods:

Cerebrospinal fluid (CSF), amyloid positron emission tomography (PET), and magnetic resonance imaging measures were harmonized across four studies of memory and aging. Spearman correlations between biomarkers and between biomarkers and cognitive measures were calculated within each racialized group, then compared between groups by standard normal tests after Fisher’s Z-transformations.

Results:

The harmonized dataset included at least one biomarker measurement from 495 Black and 2,600 White participants. The mean age was similar between racialized groups. However, Black participants were less likely to have cognitive impairment (28% versus 36%) and had less abnormality of some CSF biomarkers including CSF Aβ42/40, total tau, p-tau181, and neurofilament light. CSF Aβ42/40 was negatively correlated with total tau and p-tau181 in both groups, but at a smaller magnitude in Black individuals. CSF Aβ42/40, total tau, and p-tau181 had weaker correlations with cognitive measures, especially episodic memory, in Black than White participants. Correlations of amyloid measures between CSF (Aβ42/40, Aβ42) and PET imaging were also weaker in Black than White participants. Importantly, no differences based on race were found in correlations between different imaging biomarkers, or in correlations between imaging biomarkers and cognitive measures.

Interpretation:

Relationships between CSF biomarkers but not imaging biomarkers varied by racialized groups. Imaging biomarkers performed more consistently across racialized groups in associations with cognitive measures.

1. INTRODUCTION

Recent epidemiological studies indicate that the prevalence of dementia is higher in self-identified Black or African American and Hispanic adults compared with Non-Hispanic White individuals (NHW) ^1–3. Since biomarker testing for Alzheimer disease (AD) was not performed in most of these studies, it is unknown whether the higher prevalence of dementia in Black and Hispanic groups reflects a higher rate of AD or is instead due to non-AD causes of dementia such as cerebrovascular diseases or mixed dementias. In a seeming contradiction, several observational research studies that performed AD biomarker testing have reported a lower rate of AD biomarker abnormalities in mostly cognitively normal participants from under-represented groups (URGs) ^4–9. However, individuals from URGs enrolled in research studies are often recruited from different sources than NHW participants and may have differences in social determinants of health (SDOH) and medical comorbidities that may confound analyses ^{6, 10}. Further, participants from URGs are less likely to undergo some types of AD biomarker tests ¹¹, leading to even smaller study sizes that do not permit detailed analysis of the underlying causes of group differences ^{6, 12}. Despite the limitations of studying racial and ethnic differences in research cohorts, participants in these studies are deeply phenotyped with high quality measures and the resulting findings can inform studies in more representative cohorts. Importantly, findings in research cohorts may be helpful in considering more equitable design and analysis of clinical trials ⁵.

Most AD clinical trials have assumed that the relationships between AD biomarkers, and between biomarkers and cognitive measures, are consistent across racialized groups. AD clinical trials now frequently include biomarkers such as the levels of various analytes in the cerebrospinal fluid (CSF) and blood, amyloid positron emission tomography (PET), tau PET, and structural magnetic resonance imaging (MRI) measures. Biomarkers are used to confirm that impaired individuals have AD pathology and to evaluate the effects of treatments on pathology ¹³. Biomarkers can also be used to identify cognitively normal individuals with early AD pathology who may be enrolled in trials of interventions that prevent or delay cognitive impairment ¹⁴. AD clinical trials typically apply a single cut-off for AD biomarker positivity to all individuals regardless of factors such as age, sex, medical conditions, SDOH, recruitment method, or self-identified race. However, it is unclear whether biomarkers consistently reflect AD pathology and cognitive performance across all these various factors, especially if there may be differences in the causes of dementia between groups.

In the current study, we examined whether the relationships between biomarkers, or between biomarkers and cognitive measures, were consistent across racialized groups. Because relatively few Black individuals are enrolled in most AD research studies, combining data from multiple studies was required to provide adequate power for these analyses. Data on CSF biomarkers, amyloid PET, tau PET, and brain MRI were harmonized from four studies of memory and aging based at Washington University, University of Pennsylvania, Emory University, and Harvard University. Spearman correlations between biomarkers and between biomarkers and cognitive measures were calculated within each racialized group, then compared between groups.

2. METHODS

2.1. Participants

The Study of Race to Understand Alzheimer Biomarkers (SORTOUT-AB; NIH/NIA R01 AG067505) was designed to evaluate for potential racial differences in harmonized data collected by multiple research studies of memory and aging in middle-aged and older individuals: the Washington University (WU) Knight Alzheimer Disease Research Center (ADRC), the University of Pennsylvania (UPenn) ADRC, the Emory University Goizueta ADRC (Emory), and the Harvard Aging Brain Study (HABS). The details of recruitment for these studies have been described previously ^{7, 15, 8, 16, 17, 18}. Participants who had conditions that could prevent participation in neuroimaging (e.g., a pacemaker), or that could affect long-term participation (e.g., metastatic cancer), were excluded. All participants had clinical and/or cognitive assessments within 2 years of their CSF or imaging assessments, an interval over which correlations between AD biomarkers remain relatively stable ¹⁹. Because many individuals did not have biomarker data on all modalities, participants with either CSF or imaging assessments (MRI, amyloid PET, tau PET) were included to maximize the sample sizes for estimating each pairwise correlation between biomarkers and cognition. For analyses correlating CSF with imaging biomarkers, only participants who had both CSF and imaging assessments within 2 years were included. All participants provided written informed consent at recruitment from their parent studies. The Washington University Human Research Protection Office approved the current study with additional approvals from the IRBs for other sites.

2.2. Clinical and cognitive assessments

All clinical and cognitive assessment protocols were consistent with the National Alzheimer Coordinating Center (NACC) Uniform Data Set (UDS) ^{20, 21}. Age, sex, years of education, body mass index (BMI), and medical history were collected at study entry and updated at each visit. Race and family history of dementia in first-degree relatives were self-reported by the participants. The Clinical Dementia Rating^®^™ (CDR^®^™)²² was performed at each clinical assessment to determine the presence or absence of dementia, and when present, its severity. Standard criteria were applied to diagnose the likely etiology of dementia²³.

The cognitive battery included measures of episodic memory, working memory, semantic knowledge, executive function and attention, and visuospatial ability. Cognitive measures were harmonized as necessary across UDS version 2 and 3²⁴: the Mini-Mental State Examination (MMSE, harmonized with Montreal Cognitive Assessment [MoCA]), Animal Fluency (60 seconds), Vegetable Fluency, Wechsler Adult Intelligence Scale (WAIS-R) Digit Symbol, Digit Span, Logical Memory Immediate Recall (harmonized with Craft Story Immediate Recall), Logical Memory Delayed Recall (harmonized with Craft Story Delayed Recall), Boston Naming Test (BNT; harmonized with Multilingual Naming Test), Free and Cued Selective Reminding, and Trail Making Test A and B. All scales were oriented such that a higher score indicates better cognition. Z-scores were computed by subtracting the cohort mean from the individual test scores and dividing the difference by the cohort standard deviation (SD). The episodic memory composite was computed by averaging the z-scores from the Craft Story-immediate and Craft Story-delayed tasks. A global cognitive composite score was constructed by averaging Z-scores across all tests, including MMSE, Multilingual Naming Test, Craft Story-immediate, Craft Story-delayed, Digit Span-forward, Digit Span-backward, Animal Naming, Vegetable Naming, and Trail Making Tests.

2.3. Apolipoprotein E genotypes

Details of the apolipoprotein E (APOE) genotyping protocols have been described previously²⁵. The APOE ε4 carrier status was dichotomized as positive or negative, indicating that an individual carried either one or two APOE ε4 alleles, or no APOE ε4 alleles.

2.4. CSF sample collection and analysis

Details of CSF collection at the WU ADRC have been previously described²⁶. CSF samples (20–30 mL) were collected at 8 AM after overnight fasting by gravity drip, briefly centrifuged at low speed, and aliquoted into polypropylene tubes prior to freezing at −80°C. CSF samples from participants enrolled at the Emory and UPenn ADRCs were collected in accordance with protocols for the AD Neuroimaging Initiative (ADNI)²⁷. An automated immunoassay (LUMIPULSE G1200, Fujirebio, Malverne, PA) was used to measure CSF concentrations of β-amyloid (1–42) (Aβ42), β-amyloid (1–40) (Aβ40), total tau (t-tau) and tau phosphorylated at 181 (p-tau181) ²⁸. Concentrations of neurofilament light (NfL) were measured with a commercial ELISA kit (UMAN Diagnostics). All CSF samples from the WU and Emory ADRCs were run at WU. A subset of the CSF samples (n=114) from the UPenn ADRC was selected to represent a wide range of values for all analytes and were run at the same time and with the same reagents as the WU and Emory samples to evaluate and adjust for systematic differences between the UPenn and WU sites. The UPenn and WU values were harmonized by a linear regression based on the values of the bridging samples²⁹.

2.5. Structural brain MRI and amyloid and Tau PET

Scans were centrally re-processed at the Knight Alzheimer Research Imaging (KARI) Program at WU using a protocol similar to that used by the ADNI. Details of the structural brain MRI and amyloid PET protocols are provided elsewhere^{26, 25, 7}. The FreeSurfer image analysis tool was used to obtain regional volumes and cortical thickness from the MRI scans, adjusting for the effect of scanners with a regression analysis³⁰. A standardized uptake value ratio (SUVR) with correction for partial volume effects was calculated for the FreeSurfer regions of interest (ROIs) for PiB, Florbetapir or Florbetaben³¹. The cerebellum was used as the reference region. A composite measure of global amyloid burden was calculated using the averaged SUVR values in the lateral orbitofrontal, medial orbitofrontal, precuneus, rostral middle frontal, superior frontal, superior temporal, and middle temporal regions. Because different tracers (PiB, Florbetapir or Florbetaben) were used in amyloid PET imaging scans, values from this global summary were converted into Centiloid units³² to harmonize tracer and data processing differences using previously published equations^{32, 33}. Tau PET was performed with 18F-AV1451 (Flortaucipir) as previously described, and values from the bilateral entorhinal cortex, amygdala, lateral occipital cortex, and inferior temporal cortex regions were averaged together as a summary measure of tau PET (called Tauopathy)³⁴.

2.6. Statistical analyses

Participant demographics, cognitive status (cognitively normal, CDR=0 or cognitively impaired, CDR>0), APOE ε4 status (carrier or non-carrier), and presence of select medical comorbidities (hypertension or diabetes) that had a prevalence of at least 10% in each racialized group were summarized with the mean and standard deviation for continuous variables or count (percentage) for categorical variables. Racial differences in AD biomarkers were analyzed with Analysis of Covariance (ANCOVA) models³⁵, adjusting for the effects of age, cognitive status, APOE ε4 status, years of education, sex, and presence of each medical comorbidity (hypertension or diabetes).

The primary analyses were conducted to detect potential racial differences in the correlations between biomarkers and between biomarkers and cognitive measures. Because not all participants had data on all biomarker modalities (CSF biomarkers, amyloid PET, tau PET, and structural MRI) and cognitive measures, each pairwise correlation was estimated using the maximum sample size of participants for whom the bivariate data were available. These pairwise correlations were estimated for each racialized group, and then compared with a standard normal test after the Fisher’s Z-transformation³⁶. This test was chosen because Fisher’s Z-transformation can be approximated by a normal distribution regardless of the sample sizes, assuring the validity of our tests for a large number of pairwise correlations estimated from a wide range of sample sizes in Black and White groups. In estimating and comparing the correlations, we conducted both unadjusted analyses and adjusted analyses accounting for the effects of age, cognitive status, APOE ε4 status, years of education, sex, and medical comorbidities (hypertension and diabetes). Because of the large number of comparisons, we adjusted significance for a False Discovery Rate (FDR)³⁷ of 5%.

3. RESULTS

3.1. Cohort characteristics

The cross-sectional study included a total of 495 Black and 2,600 White participants who had available data on at least one of the four biomarker modalities (CSF biomarkers, amyloid PET, tau PET, or structural MRI). Participant characteristics at the clinical assessment closest to the biomarker visit are shown (Table 1). For the 495 Black participants included in the study, CSF biomarker data were available from 266 participants and imaging data were available from 359 participants. For the 2,600 White participants, CSF biomarker data were available from 1,977 participants and imaging data were available from 1,645 participants. A total of 121 Black and 959 White participants had data on both CSF and imaging biomarkers. The average age was 70.94 ± 8.95 years (mean ± standard deviation), similar for Black (70.78 ± 8.24 years) and White participants (70.97 ± 9.07 years). Most participants (95% of Black and 97% of White participants) had completed at least 12 years of education. Similar percentages of Black and White participants carried an APOE ε4 allele (45% versus 44%, p=0.79). Black participants were more likely than White participants to be female (68% versus 53%, p<0.001) and rated as cognitively unimpaired (72% versus 64%, p=0.01). Black participants were more likely than White participants to have a history of hypertension (73% versus 45%, p<0.001) or diabetes (28% versus 10%, p<0.001), but less likely to report family history of dementia (42% versus 54%, p<0.001). Black participants had a higher average BMI than White participants (29.84 ± 6.04 versus 27.02 ± 5.05, p<0.001). Participant characteristics specific to each cohort defined by biomarker modalities (CSF, Imaging, and their overlap) are also shown (Supplementary Table 1).

Table 1.

Demographics, genetic, and clinical status of all study participants at the closest clinical assessment to biomarker assessments

	Either CSF or imaging cohort
	Overall	Black	White
N=	3,095	495	2,600
Gender N (%Female)	1,712 (55%)	337 (68%)	1,375 (53%)
APOE ε4 carrier status N (% ε4 carrier)	1,279 (45%)	203 (45%)	1,076 (44%)
(missing)	223	45	178
CDR 0/0.5/+1 (%>0)	1,564/596/224 (34.4%)	291/78/37 (28.1%)	1,273/518/187 (35.5%)
(missing)	711	89	622
History of hypertension N (%)	1,100 (50%)	266 (73%)	834 (45%)
(missing)	891	133	758
History of diabetes N (%)	259 (13%)	96 (28%)	163 (9.9%)
(missing)	1,108	158	950
Family history of Dementia N (%)	1,062 (52%)	127 (42%)	935 (54%)
(missing)	1,064	191	873
Baseline age (years)	70.94 (8.95)	70.78 (8.24)	70.97 (9.07)
Years of education
<12	91 (3.0%)	25 (5.4%)	66 (2.6%)
>=12	2,906 (97%)	435 (95%)	2,471 (97%)
(missing)	98	35	63
BMI (kg/m²)	27.49 (5.33)	29.84 (6.04)	27.02 (5.05)
(missing)	1,129	163	966
Site
Emory	812 (26%)	122 (25%)	690 (27%)
UPenn	435 (14%)	87 (18%)	348 (13%)
WU	1,667 (54%)	242 (49%)	1,425 (55%)
HABS	181 (5.8%)	44 (8.9%)	137 (5.3%)

Open in a new tab

3.2. Racial differences in the levels of CSF and imaging biomarkers

Biomarker levels in Black and White individuals were compared after adjustment for age, sex, APOE ε4 carrier status, years of education, cognitive status and history of diabetes and hypertension (see Table 2, which also included unadjusted comparisons). A similar adjusted mean concentration of CSF Aβ42 was found in both Black and White participants (650 pg/mL versus 690 pg/mL, p=0.136). However, the adjusted mean concentration of CSF Aβ40 was lower in Black (9,600 pg/mL) than White (10,800 pg/mL) participants (p<0.0001), and the adjusted mean CSF Aβ42/40 level was higher in Black (0.0697) than White (0.0647) participants (p=0.002). The adjusted mean concentration of CSF total tau was lower in Black (364 pg/mL) than White (450 pg/mL) participants (p=0.0003), and a similar difference was found for CSF p-tau181 (50.7 pg/mL versus 61.4 pg/mL, p=0.001). The adjusted mean log transformed value of CSF NfL was lower in Black (6.61) than White (6.79) participants (p<0.0001). In the combined cohort, 36.1% of all individuals were amyloid PET positive. The amyloid PET positive rate was 24.6% for Black participants (18.2% for cognitively normal and 57.1% for cognitively impaired individuals) and 38.3% for White participants (26.8% for cognitively normal and 77.4% for cognitively impaired individuals). There was a trend towards a racial difference in the adjusted mean amyloid PET Centiloid (p=0.051), but no difference in the tau PET summary measure (p=0.83); notably, fewer participants had amyloid and tau PET data, and therefore these comparisons had lower power to detect differences. Black participants had lower adjusted mean hippocampal volumes and cortical thickness than White participants (p=0.001 and p<0.0001, respectively). When cognitively normal and cognitively impaired groups were analyzed separately, there were similar racial differences in biomarkers (Supplementary Table 2).

Table 2.

Adjusted mean levels of biomarkers as a function of race

	Either CSF or imaging cohort
	Black¹	White¹	Unadjusted p-value	Adjusted p-value²
CSF Aβ42 (pg/mL) (N=2227, 1332)³	650 (26)	690 (14)	0.35	0.136
CSF Aβ40 (pg/mL) (N=2227, 1333)³	9600 (280)	10800 (150)	<.0001	<.0001
CSF Aβ42/40 (N=2225, 1331)³	0.0697 (0.0015)	0.0647 (0.0008)	<.0001	0.002
CSF total tau (pg/mL) (N=2220, 1329)³	364 (23)	450 (12)	<.0001	0.0003
CSF ptau-181 (pg/mL) (N=2221, 1328)³	50.7 (3.1)	61.4 (1.7)	<.0001	0.001
CSF NFL (pg/mL) [log] (N=2066, 1277)³	6.61 (0.04)	6.79 (0.02)	<.0001	<.0001
Centiloid SUVR (N=1310, 1125)³	35.3 (3.0)	41.4 (1.9)	<.0001	0.051
Tauopathy (Tau PET) (N=586, 543)³	1.52 (0.05)	1.53 (0.03)	0.13	0.83
MRI Hippocampus (mm³) (N=2002, 1528)³	6871 (64)	7105 (40)	0.007	0.001
Cortical thickness (mm) (N=1280, 1267)³	2.444 (0.010)	2.485 (0.006)	0.003	<.0001

Open in a new tab

⁽¹⁾

: adjusted least-squared mean (SE)

⁽²⁾

: adjusted for age (centered), sex, APOE ε4 carrier status, years of education (centered), cognitive status; history of diabetes, or hypertension

⁽³⁾

: (N unadjusted, N adjusted)

3.3. Correlations of CSF biomarkers and cognition

Unadjusted Spearman correlations between CSF biomarkers and cognitive measures within each racialized group were evaluated (Supplementary Figure 1). Partial Spearman correlations included adjustment for the effects of age, sex, APOE ε4 carrier status, years of education, cognitive status, and each of the medical comorbidities (hypertension and diabetes), and the significance of differences between correlations for Black or White individuals was FDR adjusted for multiplicity (Figure 1). CSF Aβ42 and total tau were positively correlated in Black (r=0.35) but not White participants (r=0.03), which was significantly different (Δ=−0.33, p=0.0001). CSF Aβ42 and p-tau181 were positively correlated in Black (r=0.37) but negatively correlated in White participants (r=−0.07), which was significantly different (Δ=−0.44, p<0.0001). CSF Aβ40 was positively correlated with p-tau181 (r=0.79 for Black and 0.58 for White participants) in both groups, but the magnitude was higher (Δ=−0.21, p<0.0001) in Black participants. CSF Aβ42/40 was negatively correlated with total tau (r=−0.32) and p-tau181 (r=−0.37) in Black participants, but at a smaller magnitude than in White participants (r=−0.49 for total tau, Δ=−0.18, p=0.0142; r=−0.59 for p-tau181, Δ=−0.22, p=0.0007). In both racialized groups, CSF NfL was positively correlated with both total tau and p-tau181, and negatively correlated with Aβ42/40, and did not vary significantly by race. Notably, the correlation between CSF total tau and p-tau was very high and did not vary by race (0.89 for Black and 0.91 for White participants).

Figure 1. — Partial Spearman correlations between different CSF biomarkers and between CSF biomarkers and cognitive measures for Black (top left) and White participants (top right) after adjusting for the effects of age, sex, *APOE* ε4 carrier status, years of education, cognitive status, and comorbidities (hypertension and diabetes). In the bottom left panel, the raw p-values of differences are shown in black (lower triangle) and the significant differences after FDR adjustment between White and Black individuals are shown in colors (top triangle). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

Importantly, no CSF biomarkers were significantly correlated with cognitive measures in Black participants. However, in White participants, all CSF biomarkers were significantly correlated with the global cognitive composite and episodic memory composite with the exception of Aβ40. Hence, there were substantial racial differences in the correlations of CSF biomarkers with cognitive measures. Specifically, CSF Aβ42 and Aβ42/40 were not significantly correlated with the episodic memory composite in Black participants, but were positively correlated in White participants (r = 0.20, Δ=0.27, p=0.0077 for Aβ42; r=0.21, Δ=0.27, p=0.0043 for Aβ42/40). CSF total tau was not significantly correlated with the global cognitive composite or episodic memory composite in Black participants, but was negatively correlated in White participants (r=−0.24 for global cognition; r = −0.21 for episodic memory, Δ=−0.24, p=0.0049 for global cognition; Δ=−0.23, p = 0.0168 for episodic memory). CSF p-tau181 was not significantly correlated with the global cognitive composite or episodic memory composite in Black participants, but was negatively correlated in White participants (r=−0.24 for both, Δ=−0.26, p=0.0016 for global cognition; Δ=−0.24, p=0.0069 for episodic memory).

3.4. Correlations of imaging biomarkers and cognition

Unadjusted Spearman correlations between imaging biomarkers and cognitive measures within each racialized group were evaluated (Supplementary Figure 2). Using the same approach as that taken with CSF biomarkers, partial Spearman correlations between imaging biomarkers and cognitive factors were evaluated (Figure 2). Amyloid PET Centiloid was positively correlated with the tau PET summary measure (Tauopathy) in both Black (r=0.35) and White (r=0.38) participants, and there was no significant difference in associations based on race. The tau PET summary measure was negatively correlated with hippocampal volume in Black (r=−0.11) and White participants (r=−0.20), and there was no significant difference between groups. The correlations between hippocampal volume and cortical thickness were almost identical in both racialized groups (r=0.38).

Figure 2. — Partial Spearman correlations between imaging biomarkers and cognitive measures for Black (top left) and White participants (top right) after adjusting for the effects of age, sex, *APOE* ε4 carrier status, years of education, cognitive status, and comorbidities (hypertension and diabetes). In the bottom left panel, the raw p-values of differences are shown in black (lower triangle) and the significant differences after FDR adjustment between White and Black individuals are shown in colors (top triangle). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

Amyloid PET Centiloid, the tau PET summary measure, hippocampal volume, and cortical thickness were all significantly correlated with the global cognitive composite in both Black and White participants, with the exception of the tau PET summary measure in Black participants (which was available for relatively few participants), and there were no significant racial differences in the correlations. The episodic memory composite was negatively correlated with amyloid PET (r =−0.17), and positively correlated with hippocampal volume and cortical thickness (r= 0.19 and 0.20) for White participants, but differences in the correlations between racialized groups were not significant.

3.5. Correlations between CSF and imaging biomarkers

The correlations between CSF and imaging biomarkers were examined using the sub-cohort with both CSF and imaging biomarkers obtained within a 2-year interval. Unadjusted Spearman correlations between CSF biomarkers and cognitive measures within each racialized group were evaluated (Supplementary Figure 3), and partial Spearman correlations were evaluated as described for the other biomarker comparisons (Figure 3). In Black participants, no CSF biomarkers were correlated with amyloid PET Centiloid or the tau PET summary measure, although this could be related to low statistical power. In the larger group of White participants, all CSF biomarkers except CSF Aβ40 were correlated with amyloid PET Centiloid and the tau PET summary measure. Two significant racial differences in the correlations were observed: CSF Aβ42 and Aβ42/40 were not significantly correlated with amyloid PET Centiloid in Black participants, but were negatively and significantly correlated in White participants (r=−0.46, Δ=−0.41, p=0.0005 for Aβ42; r=−0.62, Δ=−0.49, p<0.0001 for Aβ42/40). When analyses were performed in only cognitively normal participants, racial differences in correlations remained consistent, although some were no longer significant due to loss of statistical power (Supplementary Figure 4–6). There were insufficient numbers of cognitively impaired Black individuals to enable statistically valid analyses in only cognitively impaired individuals.

Figure 3. — Partial Spearman correlations between CSF and imaging biomarkers for Black (top left) and White participants (top right) after adjusting for the effects of age, sex, *APOE* ε4 carrier status, years of education, cognitive status, and comorbidities (hypertension and diabetes). In the bottom left panel, the raw p-values of differences are shown in black (left) and the significant differences after FDR adjustment between White and Black individuals are shown in colors (right). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

4. DISCUSSION

The SORTOUT-AB study harmonized CSF and imaging data from four major AD biomarker studies and included 495 Black and 2,600 White participants with data from at least one biomarker modality. Compared to White participants, Black participants were more likely to be cognitively unimpaired, female, and to have hypertension and diabetes. Even after adjustment for key covariates, Black participants had less abnormality of some CSF biomarkers, including CSF Aβ42/40, total tau, p-tau181, and neurofilament light compared to White participants. There was also a trend toward lower amyloid burden by PET in Black participants compared to Whites. The relationships between biomarkers or between biomarkers and cognitive measures were then examined in Black and White participants and compared. CSF Aβ42/40 was negatively correlated with total tau and p-tau181 in both racialized groups, but at a smaller magnitude in Black individuals. CSF Aβ42/40, total tau, and p-tau181 had weaker correlations with cognitive measures, especially episodic memory, in Black than White participants. Correlations of amyloid measures between CSF (Aβ42/40, Aβ42) and PET imaging were also weaker in Black than White participants. Importantly, no racial differences were found in correlations between different imaging biomarkers, or in correlations between imaging biomarkers and cognitive measures.

The lesser abnormality of CSF biomarkers in Black compared to White participants is consistent with several recent studies ^4–9, but seemingly conflicts with the results of epidemiological studies that have found a higher prevalence of dementia in Black individuals ^1–3. It is possible that the higher rate of dementia in Black individuals is caused by non-AD etiologies, such as cerebrovascular diseases, or mixed dementias ³⁸. Notably, in this study and other similar AD research studies, Black individuals have a higher rate of medical conditions associated with cerebrovascular diseases such as hypertension and diabetes ^{4, 9}. The presence of cerebrovascular disease could potentially reduce the burden of AD brain pathology required to cause clinically significant cognitive impairment. CSF biomarkers may be used to determine eligibility for prevention and treatment trials. Since CSF biomarkers are sometimes used to determine eligibility for prevention and treatment trials, the lower occurrence of abnormal CSF biomarkers in Black participants could lead to fewer Black participants being included in AD clinical trials. To increase the representation of Black individuals, clinical trials may consider setting a lower threshold for biomarker abnormality for Black participants. However, including Black individuals with lower levels of biomarker abnormality would likely select individuals with less AD pathology who may be less responsive to AD-specific treatments, and could lead to the potentially erroneous conclusion that the treatment is not effective in Black individuals. Furthermore, use of race-based cut-offs has sometimes led to unintended consequences including systematic racial discrimination ³⁹. These concerns provide a strong rationale for not using cut-offs that vary by race, even if the intent of these cut-offs is to increase representation of Black individuals in clinical trials.

Race is a social rather than a biological construct and is associated with many social factors that affect risk for AD ³⁸. Clarifying the factors underlying these social experience differences will allow better understanding in future large and population-based biomarker studies. Moreover, understanding risk from the perspective of SDOH will assist in developing personalized and evidence-based approaches, or precision medicine-based approaches in future clinical trials of AD. Doing so would allow studies to develop inclusion/exclusion criteria that are sensitive to these factors. For example, cut-offs for abnormal performance or mild cognitive impairment on cognitive tests often adjust for age and education rather than race³⁸. Using cut-offs based on the factors that underlie the differences between biomarkers and race could improve the categorization of all individuals and avoid unintended consequences related to race-based cut-offs.

The major focus of this study was the relationships between biomarkers or between biomarkers and cognitive measures; racial differences in these relationships has not previously been evaluated. Although the statistical power to detect significant correlations depends on the number of individuals represented, and is therefore much higher in the larger White sub-cohort, the magnitude of significant correlations would not be expected to be affected by sample sizes. Interestingly, correlations between CSF biomarkers varied significantly between Black and White participants, not only in the magnitude of the correlations (e.g., CSF Aβ42/40 versus total tau or p-tau181), but also sometimes in the direction (positive or negative) of the correlations (e.g., Aβ42 versus p-tau181). In contrast, the relationships between different imaging measures of AD did not vary by race. This suggests there are factors associated with race that affect the relationships between CSF biomarkers, but these factors do not affect imaging biomarkers in the same way. PET imaging biomarkers are thought to reflect the lifetime accumulation of pathology, while CSF biomarker levels depend on the production and clearance of proteins at the time of CSF collection and may be affected by CSF dynamics in ways that are not yet well understood ^{40, 41}. It is possible that factors associated with race may influence CSF dynamics; further studies are needed to explore this hypothesis.

Both CSF and imaging biomarkers predicted a global cognitive composite and episodic memory composite in White participants. However, while imaging biomarkers were correlated with cognitive measures in Black participants, there were no significant correlations between CSF biomarkers and cognitive measures in Black participants. Consistent with these findings, there were no significant racial differences in correlations between cognitive measures and imaging measures, but there were significant racial differences in correlations between cognitive measures and CSF biomarkers. Overall, these findings suggest that imaging biomarkers have more consistent associations with cognitive measures across racialized groups. If these findings are confirmed by future studies of large and representative cohorts, they suggest that CSF biomarkers may be less useful in the assessment of AD in Black individuals, especially since normative ranges for CSF tests have been established in cohorts comprised primarily of White individuals. Further, the utility of CSF biomarkers in clinical trials as part of inclusion/exclusion criteria or as part of efficacy endpoints will need to be reassessed in Black participants.

A major strength of this study was the analysis of a multicenter cohort, the relatively large number of Black participants, and the harmonized biomarker data for CSF, amyloid PET, tau PET, and structural MRI. Although the study was relatively large compared to other studies of racial differences in AD biomarkers, analyses of the effects of SODH and medical comorbidities require even larger sample sizes. Future studies with even larger cohorts are needed to evaluate SODH, including quality of education and encompassing sociocultural and social economic factors related to residential and school segregation⁴⁸, access and quality of health care, occupational safety, ability to build and maintain wealth, experiences with the legal system, violence exposure, structural and systemic racism, environment, poverty, and experiences with adversity. Collection of detailed SDOH data is currently underway. Limitations of the study include that research participants are subject to potential survivor bias, especially because Black individuals have a higher rate of medical conditions that may cause death, disability, or more severe dementia that may adversely affect enrollment in research studies ⁴². Additionally, both Black and White participants in this study had high levels of education that are not representative of the general population. Further, Black and White participants may be recruited via different methods, potentially leading to confounds ¹⁰.

Overall, this study suggests that relationships between CSF biomarkers but not imaging biomarkers vary by racialized group and that imaging biomarkers perform more consistently in associations with cognitive measures. These findings raise the possibility that CSF biomarkers may not have the same degree of association and clinical validity in Black individuals as compared to White individuals. The reason why CSF biomarkers perform less consistently than imaging biomarkers in Black individuals remains unclear, but CSF and imaging biomarkers do reflect correlated but distinct measures of AD that may be affected differently by factors associated with race such as medical conditions and SDOH. Future longitudinal studies of large and diverse cohorts with detailed data on medical conditions and SDOH will be essential to further understand these factors.

Supplementary Material

Fig S1

Supplementary Figure 1. Racial differences in correlations between CSF biomarkers and cognitive measures. Unadjusted Spearman correlations between different CSF biomarkers and between CSF biomarkers and cognitive measures for Black (top left) and White participants (top right). In the bottom left panel, the significant differences after FDR adjustment between White and Black individuals are shown in colors (top triangle), and the raw p-values of differences are shown in black (lower triangle). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

NIHMS1948578-supplement-Fig_S1.tif^{(11.5MB, tif)}

Fig S2

Supplementary Figure 2. Racial differences in correlations between imaging biomarkers and cognitive measures. Unadjusted Spearman correlations between imaging biomarkers and cognitive measures and between imaging biomarkers for Black (top left) and White participants (top right). In the bottom left panel, the significant differences after FDR adjustment between White and Black individuals are shown in colors (top triangle), and the raw p-values of differences are shown in black (lower triangle). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

NIHMS1948578-supplement-Fig_S2.tif^{(11.5MB, tif)}

Fig S3

Supplementary Figure 3. Racial differences in correlations between CSF and imaging biomarkers. Unadjusted Spearman correlations between CSF and imaging biomarkers for Black (top left) and White participants (top right). In the bottom left panel, the significant differences after FDR adjustment between White and Black individuals are shown in colors(left), and the raw p-values of the differences (right) are shown in black. In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

NIHMS1948578-supplement-Fig_S3.tif^{(14MB, tif)}

Fig S4

Supplemental Figure 4. Racial differences in correlations between CSF biomarkers and cognitive measures in cognitively normal participants. Unadjusted Spearman correlations between different CSF biomarkers and between CSF biomarkers and cognitive measures for Black (top left) and White participants (top right). In the bottom left panel, the significant differences after FDR adjustment between White and Black individuals are shown in colors (top triangle), and the raw p-values of differences are shown in black (lower triangle). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

NIHMS1948578-supplement-Fig_S4.tif^{(11.5MB, tif)}

Fig S5

Supplemental Figure 5. Racial differences in correlations between imaging biomarkers and cognitive measures in cognitively normal participants. Unadjusted Spearman correlations between imaging biomarkers and cognitive measures and between imaging biomarkers for Black (top left) and White participants (top right). In the bottom left panel, the significant differences after FDR adjustment between White and Black individuals are shown in colors (top triangle), and the raw p-values of differences are shown in black (lower triangle). In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

NIHMS1948578-supplement-Fig_S5.tif^{(11.5MB, tif)}

Fig S6

Supplemental Figure 6. Racial differences in correlations between CSF and imaging biomarkers in cognitively normal participants. Unadjusted Spearman correlations between CSF and imaging biomarkers for Black (top left) and White participants (top right). In the bottom left panel, the significant differences after FDR adjustment between White and Black individuals are shown in colors(left), and the raw p-values of the differences (right) are shown in black. In the bottom right panel, the number of Black (lower left triangle) and White (upper right triangle) individuals in each pairwise comparison is shown.

NIHMS1948578-supplement-Fig_S6.tif^{(14MB, tif)}

Tab S1

NIHMS1948578-supplement-Tab_S1.docx^{(19.5KB, docx)}

Tab S2

NIHMS1948578-supplement-Tab_S2.docx^{(18.3KB, docx)}

Acknowledgements

The work is partly supported by National Institute on Aging (NIA) grants R01 AG067505 and R01 AG053550 (Dr. Xiong), P30 AG066444, P01 AG026276, and P01 AG03991 (Dr. Morris), P30 AG072979 (Dr. Wolk), P30 AG066511 (Dr. Levey), P01 AG036694 (Drs. Reisa Sperling and Keith Johnson), R01 AG054059 (Dr. Gleason), P20 AG068024 (Dr. Roberson), and R24 AG074915-02 (Dr. Balls-Berry). This material is the result of work supported with resources and the use of facilities at the William S. Middleton Memorial VA Hospital in Madison, WI. We want to thank all the participants of WU, UPenn and Emory ADRCs, HABS who contributed to this study and their families.

Footnotes

Potential Conflict of Interest

SS has served on a scientific advisory board for Eisai. DW has served as a paid consultant for Eli Lilly, GE Healthcare and Qynapse, and serves on a DSMB for Functional Neuromodulation. ER serves on a data monitoring committee for Eli Lilly. TB participates as a site investigator in clinical trials sponsored by Avid Radiopharmaceuticals, Eli Lilly and Company, Biogen, Eisai, Jaansen, and Roche. The other Authors declare no Competing Financial or Non-Financial Interests. This work was supported in part by funding from the National Institutes of Health (grant #AG067505). WU has a financial interest in C2N Diagnostics and may financially benefit if the company is successful in marketing its product(s) that is/are related to this research.

Data availability

Anonymized data that support the findings of this study are available from the corresponding author, upon reasonable request from any qualified investigator.

References

1.Manly JJ, Jones RN, Langa KM, et al. Estimating the Prevalence of Dementia and Mild Cognitive Impairment in the US: The 2016 Health and Retirement Study Harmonized Cognitive Assessment Protocol Project. JAMA Neurol. 2022. Dec 1;79(12):1242–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Power MC, Bennett EE, Turner RW, et al. Trends in Relative Incidence and Prevalence of Dementia Across Non-Hispanic Black and White Individuals in the United States, 2000–2016. JAMA Neurol. 2021. Mar 1;78(3):275–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Kornblith E, Bahorik A, Boscardin WJ, Xia F, Barnes DE, Yaffe K. Association of Race and Ethnicity With Incidence of Dementia Among Older Adults. JAMA. 2022. Apr 19;327(15):1488–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Wilkins CH, Windon CC, Dilworth-Anderson P, et al. Racial and Ethnic Differences in Amyloid PET Positivity in Individuals With Mild Cognitive Impairment or Dementia: A Secondary Analysis of the Imaging Dementia-Evidence for Amyloid Scanning (IDEAS) Cohort Study. JAMA Neurol. 2022. Oct 3;79(11):1139–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Deters KD, Napolioni V, Sperling RA, et al. Amyloid PET Imaging in Self-Identified Non-Hispanic Black Participants of the Anti-Amyloid in Asymptomatic Alzheimer’s Disease (A4) Study. Neurology. 2021. Mar 16;96(11):e1491–e500. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Schindler SE, Karikari TK, Ashton NJ, et al. Effect of Race on Prediction of Brain Amyloidosis by Plasma Abeta42/Abeta40, Phosphorylated Tau, and Neurofilament Light. Neurology. 2022. Apr 21. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Morris JC, Schindler SE, McCue LM, et al. Assessment of Racial Disparities in Biomarkers for Alzheimer Disease. JAMA Neurol. 2019. Mar 1;76(3):264–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Howell JC, Watts KD, Parker MW, et al. Race modifies the relationship between cognition and Alzheimer’s disease cerebrospinal fluid biomarkers. Alzheimers Res Ther. 2017. Nov 2;9(1):88. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Garrett SL, McDaniel D, Obideen M, et al. Racial Disparity in Cerebrospinal Fluid Amyloid and Tau Biomarkers and Associated Cutoffs for Mild Cognitive Impairment. JAMA Netw Open. 2019. Dec 2;2(12):e1917363. [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Gleason CE, Norton D, Zuelsdorff M, et al. Association between enrollment factors and incident cognitive impairment in Blacks and Whites: Data from the Alzheimer’s Disease Center. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2019. Dec;15(12):1533–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Howell JC, Parker MW, Watts KD, Kollhoff A, Tsvetkova DZ, Hu WT. Research Lumbar Punctures among African Americans and Caucasians: Perception Predicts Experience. Front Aging Neurosci. 2016;8:296. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Windon C, Iaccarino L, Mundada N, et al. Comparison of plasma and CSF biomarkers across ethnoracial groups in the ADNI. Alzheimer’s & dementia. 2022;14(1):e12315. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Schindler SE, Atri A. The role of cerebrospinal fluid and other biomarker modalities in the Alzheimer’s disease diagnostic revolution. Nat Aging. 2023. May;3(5):460–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Rafii MS, Aisen PS. Detection and treatment of Alzheimer’s disease in its preclinical stage. Nat Aging. 2023. May;3(5):520–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Hanseeuw BJ, Betensky RA, Jacobs HIL, et al. Association of Amyloid and Tau With Cognition in Preclinical Alzheimer Disease: A Longitudinal Study. JAMA Neurol. 2019. Aug 1;76(8):915–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Hu WT, Watts KD, Shaw LM, et al. CSF beta-amyloid 1–42 - what are we measuring in Alzheimer’s disease? Ann Clin Transl Neurol. 2015. Feb;2(2):131–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Dagley A, LaPoint M, Huijbers W, et al. Harvard Aging Brain Study: Dataset and accessibility. Neuroimage. 2017. Jan;144(Pt B):255–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Wolk DA, Mancuso L, Kliot D, Arnold SE, Dickerson BC. Familiarity-based memory as an early cognitive marker of preclinical and prodromal AD. Neuropsychologia. 2013. May;51(6):1094–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Boerwinkle AH, Wisch JK, Chen CD, et al. Temporal Correlation of CSF and Neuroimaging in the Amyloid-Tau-Neurodegeneration Model of Alzheimer Disease. Neurology. 2021. Jul 6;97(1):e76–e87. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Morris JC, Weintraub S, Chui HC, et al. The Uniform Data Set (UDS): clinical and cognitive variables and descriptive data from Alzheimer Disease Centers. Alzheimer Dis Assoc Disord. 2006. Oct-Dec;20(4):210–6. [DOI] [PubMed] [Google Scholar]
21.Weintraub S, Salmon D, Mercaldo N, et al. The Alzheimer’s Disease Centers’ Uniform Data Set (UDS): the neuropsychologic test battery. Alzheimer Dis Assoc Disord. 2009. Apr-Jun;23(2):91–101. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. 1993. Nov;43(11):2412–4. [DOI] [PubMed] [Google Scholar]
23.McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2011. May;7(3):263–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Monsell SE, Dodge HH, Zhou XH, et al. Results From the NACC Uniform Data Set Neuropsychological Battery Crosswalk Study. Alzheimer Dis Assoc Disord. 2016. Apr-Jun;30(2):134–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Luo J, Agboola F, Grant E, et al. Accelerated longitudinal changes and ordering of Alzheimer disease biomarkers across the adult lifespan. Brain. 2022. Dec 19;145(12):4459–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Luo J, Agboola F, Grant E, et al. Sequence of Alzheimer disease biomarker changes in cognitively normal adults: A cross-sectional study. Neurology. 2020. Dec 8;95(23):e3104–e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Shaw LM, Vanderstichele H, Knapik-Czajka M, et al. Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann Neurol. 2009. Apr;65(4):403–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Alcolea D, Pegueroles J, Muñoz L, et al. Agreement of amyloid PET and CSF biomarkers for Alzheimer’s disease on Lumipulse. Ann Clin Transl Neurol. 2019. Sep;6(9):1815–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Henson RL, Volluz K, Saef BA, et al. A methodology for normalizing fluid biomarker concentrations across reagent lots. Alzheimer’s & dementia. 2022;18(S5):e066912. [Google Scholar]
30.Su Y, D’Angelo GM, Vlassenko AG, et al. Quantitative analysis of PiB-PET with FreeSurfer ROIs. PloS one. 2013;8(11):e73377. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Rousset OG, Ma Y, Evans AC. Correction for partial volume effects in PET: principle and validation. J Nucl Med. 1998. May;39(5):904–11. [PubMed] [Google Scholar]
32.Su Y, Flores S, Hornbeck RC, et al. Utilizing the Centiloid scale in cross-sectional and longitudinal PiB PET studies. Neuroimage Clin. 2018;19:406–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Klunk WE, Koeppe RA, Price JC, et al. The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2015. Jan;11(1):1–15 e1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Mishra S, Gordon BA, Su Y, et al. AV-1451 PET imaging of tau pathology in preclinical Alzheimer disease: Defining a summary measure. NeuroImage. 2017. Nov 1;161:171–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Milliken GAJ DE Analysis of Messy Data, Volume III. New York: Chapman and Hall; 2001. [Google Scholar]
36.Graybill FA. Theory and Application of the Linear Model. North Scituate, Ma: Duxbury; 1976. [Google Scholar]
37.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J R Stat Soc B. 1995;57(1):289–300. [Google Scholar]
38.Gottesman RF, Albert MS, Alonso A, et al. Associations Between Midlife Vascular Risk Factors and 25-Year Incident Dementia in the Atherosclerosis Risk in Communities (ARIC) Cohort. JAMA Neurol. 2017. Oct 1;74(10):1246–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Powe NR. Black Kidney Function Matters: Use or Misuse of Race? JAMA. 2020. Aug 25;324(8):737–8. [DOI] [PubMed] [Google Scholar]
40.Patterson BW, Elbert DL, Mawuenyega KG, et al. Age and amyloid effects on human central nervous system amyloid-beta kinetics. Annals of neurology. 2015. Sep;78(3):439–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Karlsson L, Vogel J, Arvidsson I, et al. Cerebrospinal fluid reference proteins increase accuracy and interpretability of biomarkers for brain diseases. bioRxiv. 2023:2023.06.08.544222. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Nianogo RA, Rosenwohl-Mack A, Yaffe K, et al. Risk Factors Associated With Alzheimer Disease and Related Dementias by Sex and Race and Ethnicity in the US. JAMA Neurol. 2022;79(6):584–591. [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

Fig S1

NIHMS1948578-supplement-Fig_S1.tif^{(11.5MB, tif)}

Fig S2

NIHMS1948578-supplement-Fig_S2.tif^{(11.5MB, tif)}

Fig S3

NIHMS1948578-supplement-Fig_S3.tif^{(14MB, tif)}

Fig S4

NIHMS1948578-supplement-Fig_S4.tif^{(11.5MB, tif)}

Fig S5

NIHMS1948578-supplement-Fig_S5.tif^{(11.5MB, tif)}

Fig S6

NIHMS1948578-supplement-Fig_S6.tif^{(14MB, tif)}

Tab S1

NIHMS1948578-supplement-Tab_S1.docx^{(19.5KB, docx)}

Tab S2

NIHMS1948578-supplement-Tab_S2.docx^{(18.3KB, docx)}

Data Availability Statement

Anonymized data that support the findings of this study are available from the corresponding author, upon reasonable request from any qualified investigator.

[R1] 1.Manly JJ, Jones RN, Langa KM, et al. Estimating the Prevalence of Dementia and Mild Cognitive Impairment in the US: The 2016 Health and Retirement Study Harmonized Cognitive Assessment Protocol Project. JAMA Neurol. 2022. Dec 1;79(12):1242–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] 2.Power MC, Bennett EE, Turner RW, et al. Trends in Relative Incidence and Prevalence of Dementia Across Non-Hispanic Black and White Individuals in the United States, 2000–2016. JAMA Neurol. 2021. Mar 1;78(3):275–84. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] 3.Kornblith E, Bahorik A, Boscardin WJ, Xia F, Barnes DE, Yaffe K. Association of Race and Ethnicity With Incidence of Dementia Among Older Adults. JAMA. 2022. Apr 19;327(15):1488–95. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Wilkins CH, Windon CC, Dilworth-Anderson P, et al. Racial and Ethnic Differences in Amyloid PET Positivity in Individuals With Mild Cognitive Impairment or Dementia: A Secondary Analysis of the Imaging Dementia-Evidence for Amyloid Scanning (IDEAS) Cohort Study. JAMA Neurol. 2022. Oct 3;79(11):1139–47. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Deters KD, Napolioni V, Sperling RA, et al. Amyloid PET Imaging in Self-Identified Non-Hispanic Black Participants of the Anti-Amyloid in Asymptomatic Alzheimer’s Disease (A4) Study. Neurology. 2021. Mar 16;96(11):e1491–e500. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Schindler SE, Karikari TK, Ashton NJ, et al. Effect of Race on Prediction of Brain Amyloidosis by Plasma Abeta42/Abeta40, Phosphorylated Tau, and Neurofilament Light. Neurology. 2022. Apr 21. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] 7.Morris JC, Schindler SE, McCue LM, et al. Assessment of Racial Disparities in Biomarkers for Alzheimer Disease. JAMA Neurol. 2019. Mar 1;76(3):264–73. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] 8.Howell JC, Watts KD, Parker MW, et al. Race modifies the relationship between cognition and Alzheimer’s disease cerebrospinal fluid biomarkers. Alzheimers Res Ther. 2017. Nov 2;9(1):88. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Garrett SL, McDaniel D, Obideen M, et al. Racial Disparity in Cerebrospinal Fluid Amyloid and Tau Biomarkers and Associated Cutoffs for Mild Cognitive Impairment. JAMA Netw Open. 2019. Dec 2;2(12):e1917363. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Gleason CE, Norton D, Zuelsdorff M, et al. Association between enrollment factors and incident cognitive impairment in Blacks and Whites: Data from the Alzheimer’s Disease Center. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2019. Dec;15(12):1533–45. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] 11.Howell JC, Parker MW, Watts KD, Kollhoff A, Tsvetkova DZ, Hu WT. Research Lumbar Punctures among African Americans and Caucasians: Perception Predicts Experience. Front Aging Neurosci. 2016;8:296. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Windon C, Iaccarino L, Mundada N, et al. Comparison of plasma and CSF biomarkers across ethnoracial groups in the ADNI. Alzheimer’s & dementia. 2022;14(1):e12315. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Schindler SE, Atri A. The role of cerebrospinal fluid and other biomarker modalities in the Alzheimer’s disease diagnostic revolution. Nat Aging. 2023. May;3(5):460–2. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Rafii MS, Aisen PS. Detection and treatment of Alzheimer’s disease in its preclinical stage. Nat Aging. 2023. May;3(5):520–31. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Hanseeuw BJ, Betensky RA, Jacobs HIL, et al. Association of Amyloid and Tau With Cognition in Preclinical Alzheimer Disease: A Longitudinal Study. JAMA Neurol. 2019. Aug 1;76(8):915–24. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.Hu WT, Watts KD, Shaw LM, et al. CSF beta-amyloid 1–42 - what are we measuring in Alzheimer’s disease? Ann Clin Transl Neurol. 2015. Feb;2(2):131–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] 17.Dagley A, LaPoint M, Huijbers W, et al. Harvard Aging Brain Study: Dataset and accessibility. Neuroimage. 2017. Jan;144(Pt B):255–8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Wolk DA, Mancuso L, Kliot D, Arnold SE, Dickerson BC. Familiarity-based memory as an early cognitive marker of preclinical and prodromal AD. Neuropsychologia. 2013. May;51(6):1094–102. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Boerwinkle AH, Wisch JK, Chen CD, et al. Temporal Correlation of CSF and Neuroimaging in the Amyloid-Tau-Neurodegeneration Model of Alzheimer Disease. Neurology. 2021. Jul 6;97(1):e76–e87. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Morris JC, Weintraub S, Chui HC, et al. The Uniform Data Set (UDS): clinical and cognitive variables and descriptive data from Alzheimer Disease Centers. Alzheimer Dis Assoc Disord. 2006. Oct-Dec;20(4):210–6. [DOI] [PubMed] [Google Scholar]

[R21] 21.Weintraub S, Salmon D, Mercaldo N, et al. The Alzheimer’s Disease Centers’ Uniform Data Set (UDS): the neuropsychologic test battery. Alzheimer Dis Assoc Disord. 2009. Apr-Jun;23(2):91–101. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] 22.Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology. 1993. Nov;43(11):2412–4. [DOI] [PubMed] [Google Scholar]

[R23] 23.McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2011. May;7(3):263–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Monsell SE, Dodge HH, Zhou XH, et al. Results From the NACC Uniform Data Set Neuropsychological Battery Crosswalk Study. Alzheimer Dis Assoc Disord. 2016. Apr-Jun;30(2):134–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Luo J, Agboola F, Grant E, et al. Accelerated longitudinal changes and ordering of Alzheimer disease biomarkers across the adult lifespan. Brain. 2022. Dec 19;145(12):4459–73. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Luo J, Agboola F, Grant E, et al. Sequence of Alzheimer disease biomarker changes in cognitively normal adults: A cross-sectional study. Neurology. 2020. Dec 8;95(23):e3104–e16. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Shaw LM, Vanderstichele H, Knapik-Czajka M, et al. Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects. Ann Neurol. 2009. Apr;65(4):403–13. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] 28.Alcolea D, Pegueroles J, Muñoz L, et al. Agreement of amyloid PET and CSF biomarkers for Alzheimer’s disease on Lumipulse. Ann Clin Transl Neurol. 2019. Sep;6(9):1815–24. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Henson RL, Volluz K, Saef BA, et al. A methodology for normalizing fluid biomarker concentrations across reagent lots. Alzheimer’s & dementia. 2022;18(S5):e066912. [Google Scholar]

[R30] 30.Su Y, D’Angelo GM, Vlassenko AG, et al. Quantitative analysis of PiB-PET with FreeSurfer ROIs. PloS one. 2013;8(11):e73377. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Rousset OG, Ma Y, Evans AC. Correction for partial volume effects in PET: principle and validation. J Nucl Med. 1998. May;39(5):904–11. [PubMed] [Google Scholar]

[R32] 32.Su Y, Flores S, Hornbeck RC, et al. Utilizing the Centiloid scale in cross-sectional and longitudinal PiB PET studies. Neuroimage Clin. 2018;19:406–16. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] 33.Klunk WE, Koeppe RA, Price JC, et al. The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2015. Jan;11(1):1–15 e1–4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Mishra S, Gordon BA, Su Y, et al. AV-1451 PET imaging of tau pathology in preclinical Alzheimer disease: Defining a summary measure. NeuroImage. 2017. Nov 1;161:171–8. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Milliken GAJ DE Analysis of Messy Data, Volume III. New York: Chapman and Hall; 2001. [Google Scholar]

[R36] 36.Graybill FA. Theory and Application of the Linear Model. North Scituate, Ma: Duxbury; 1976. [Google Scholar]

[R37] 37.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J R Stat Soc B. 1995;57(1):289–300. [Google Scholar]

[R38] 38.Gottesman RF, Albert MS, Alonso A, et al. Associations Between Midlife Vascular Risk Factors and 25-Year Incident Dementia in the Atherosclerosis Risk in Communities (ARIC) Cohort. JAMA Neurol. 2017. Oct 1;74(10):1246–54. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Powe NR. Black Kidney Function Matters: Use or Misuse of Race? JAMA. 2020. Aug 25;324(8):737–8. [DOI] [PubMed] [Google Scholar]

[R40] 40.Patterson BW, Elbert DL, Mawuenyega KG, et al. Age and amyloid effects on human central nervous system amyloid-beta kinetics. Annals of neurology. 2015. Sep;78(3):439–53. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] 41.Karlsson L, Vogel J, Arvidsson I, et al. Cerebrospinal fluid reference proteins increase accuracy and interpretability of biomarkers for brain diseases. bioRxiv. 2023:2023.06.08.544222. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] 42.Nianogo RA, Rosenwohl-Mack A, Yaffe K, et al. Risk Factors Associated With Alzheimer Disease and Related Dementias by Sex and Race and Ethnicity in the US. JAMA Neurol. 2022;79(6):584–591. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Relationships of Cognitive Measures with Cerebrospinal Fluid but not Imaging Biomarkers of Alzheimer Disease Vary between Black and White Individuals

Samuele Bonomi, MD

Ruijin Lu, PhD

Suzanne E Schindler, MD, PhD

Quoc Bui, MD, MS

James J Lah, MD, PhD

David Wolk, MD

Carey E Gleason, PhD

Reisa Sperling, MD

Erik D Roberson, MD, PhD

Allan I Levey, MD

Leslie Shaw, PhD

Carol Van Hulle, PhD

Tammie Benzinger, MD, PhD

Morgann Adams, MS

Cecelia Manzanares

Deqiang Qiu, PhD

Jason Hassenstab, PhD

Krista L Moulder, PhD

Joyce E Balls-Berry, PhD, MPE

Keith Johnson, MD

Sterling C Johnson, PhD

Charles F Murchison, PhD

Jingqin Luo, PhD

Emily Gremminger

Folasade Agboola, MPH

Elizabeth A Grant, PhD

Russ Hornbeck, MSc

Parinaz Massoumzadeh, PhD

Sarah Keefe

Donna Dierker

Julia D Gray

Rachel L Henson, MSA

Marissa Streitz, MSW

Dawn Mechanic-Hamilton, PhD

John C Morris, MD

Chengjie Xiong, PhD

Abstract

Objective:

Methods:

Results:

Interpretation:

1. INTRODUCTION

2. METHODS

2.1. Participants

2.2. Clinical and cognitive assessments

2.3. Apolipoprotein E genotypes

2.4. CSF sample collection and analysis

2.5. Structural brain MRI and amyloid and Tau PET

2.6. Statistical analyses

3. RESULTS

3.1. Cohort characteristics

Table 1.

3.2. Racial differences in the levels of CSF and imaging biomarkers

Table 2.

3.3. Correlations of CSF biomarkers and cognition

Figure 1. Racial differences in correlations between different CSF biomarkers and between CSF biomarkers and cognitive measures.

3.4. Correlations of imaging biomarkers and cognition

Figure 2. Racial differences in correlations between different imaging biomarkers and between imaging biomarkers and cognitive measures.

3.5. Correlations between CSF and imaging biomarkers

Figure 3. Racial differences in correlations between CSF and imaging biomarkers.

4. DISCUSSION

Supplementary Material

Acknowledgements

Footnotes

Data availability

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases