Abstract
INTRODUCTION:
Longitudinal changes in Alzheimer disease (AD) biomarkers, including cerebrospinal fluid (CSF) analytes, amyloid uptakes from positron emission tomography (PET), structural outcomes from magnetic resonance imaging (MRI), and cognition, have not been compared between African Americans (AAs) and Whites.
METHODS:
A total of 179 AAs and 1180 Whites who were cognitively normal at baseline and had longitudinal data from at least one biomarker modality were analyzed for the annual rates of change.
RESULTS.
CSF Aβ42/Aβ40 declined more slowly (p=0.0390), and amyloid (PET) accumulated more slowly (p=0.0157), in AAs than Whites. CSF Aβ42 changed in opposite directions over time between AAs and Whites (p=0.0039). The annual increase in CSF total tau and pTau181 for AAs was about half of that for Whites.
DISCUSSION:
Longitudinal racial differences in amyloid biomarkers are observed. It will be important to comprehensively and prospectively examine the effects of APOE genotype and sociocultural factors on these differences.
Keywords: Alzheimer disease, CSF biomarkers, Imaging biomarkers, Longitudinal, Racial Difference
Introduction
Alzheimer disease (AD) is an irreversible neurodegenerative disease that affects ~6 million Americans and cases are projected to more than double the number by 20501. Solutions to this public crisis mandate prevention and/or treatment options that work for all, including under-represented groups (URG). In the absence of truly efficacious treatments that can modify the progression of AD, prevention may be the only viable option that may contain the public health crisis. In order to design prevention trials for AD, biomarkers are important to establish appropriate inclusion/exclusion criteria and to track disease progression. Decades of biomarker studies have established the validity of an array of AD biomarkers to detect Aβ and neurofibrillary tangles (NFTs) in the brain as well as associated neuronal death and neurodegeneration, including the molecular imaging of cerebral fibrillar β-amyloid with positron emission tomography (PET) using the [11C] benzothiazole tracer, Pittsburgh Compound-B (PiB)2 and other tracers (18F-AV45), cerebrospinal fluid (CSF) analytes3–4, tau PET regional uptakes5, and magnetic resonance imaging (MRI)-based brain structural measures. These studies further converge to demonstrate that the neuropathological course of AD begins years or even decades prior to symptom onset6–9, and the biomarker changes follow a temporal cascade with early Aβ accumulation and deposition in the brain, followed by formation of NFTs, neuronal death and brain structural changes9–11. However, almost all of these important findings are based on predominantly White cohorts, and the few biomarker studies that reported racial differences in AD biomarker changes12–13 are all based on cross-sectional data. Understanding of longitudinal racial differences in AD biomarkers is crucial for the optimal design of prevention/intervention trials to assure that treatments benefit URG.
The objective of this study is to evaluate longitudinal differences between African Americans (AA) and Whites in all major AD biomarkers, including CSF, amyloid PET, and structural MRI.
Methods
Participants
This study included middle-aged and older adults enrolled in the longitudinal studies of memory and aging at the Washington University (WU) Knight Alzheimer Disease Research Center (ADRC). Details of recruitment have been described13. Briefly, participants were community-dwelling individuals recruited from the greater St. Louis, Missouri, metropolitan area. Individuals were excluded if they had illnesses that could interfere with longitudinal follow-up, prevent participation in neuroimaging, or adversely affect cognition (e.g. metastatic cancer). The inclusion criteria for the current study were 1) availability of longitudinal data for at least one of the following biomarker modalities: CSF, amyloid PET, structural MRI, and cognition, and 2) normal cognition at the baseline biomarker or cognitive assessments. Notice that not all individuals had longitudinal data on all biomarkers and cognition. All participants provided written informed consent at recruitment. WU Human Research Protection Office approved procedures.
Demographics
Age, sex, years of education, and body mass index (BMI) were obtained. Race and family history (FH) of dementia in first-degree relatives were self-reported by the participants. Socioeconomic data were also collected using the Hollingshead index14.
Clinical and cognitive assessments
The clinical and cognitive assessment protocols of the Knight ADRC are consistent with that of the National Alzheimer Coordinating Center Uniform Data Set (UDS)15. The NACC UDS includes standard diagnostic criteria for dementia and its differential diagnoses16, and uses the global Clinical Dementia Rating™ (CDR™)17 to operationalize the presence or absence of dementia, and when present, the severity of dementia. A neuropsychological test battery18 was administered, and included episodic memory, working memory, semantic knowledge, executive function and attention, and visuospatial ability. Nine cognitive tests were shared by most participants: the Mini-Mental State Examination19, Animal Fluency (60 seconds)20, Wechsler Adult Intelligence Scale (WAIS-R) Digit Symbol21, Boston Naming Test22, and Logical Memory Delayed Recall21, Free and Cued Selective Reminding23, and Trailmaking Test A and B24. A cognitive composite score was constructed by averaging Z-scores of the nine tests.
APOE genotypes
Details of the APOE genotyping protocols have been described previously16. We dichotomized APOE ε4 status as positive or negative, indicating presence of one or two APOE ε4 alleles, or none.
CSF sample collection and analysis
Participants underwent lumbar puncture at ~8 am after overnight fasting and ~20–30mL of CSF was collected via gravity drip. Samples were gently inverted to avoid possible gradient effects, briefly centrifuged at low speed, and aliquoted into polypropylene tubes prior to freezing at –84°C until assays. Aβ42, Aβ40, total tau (Tau), and tau phosphorylated at position 181 (pTau181) were measured with a single lot of reagents for all samples with an automated immunoassay platform (LUMIPULSE G1200, Fujirebio, Malverne, PA) according to manufacturer specifications. The assay performance has been examined in preclinical25 and clinical26 AD cohorts by comparing biomarker measures with amyloid PET27.
Structural brain MRI and amyloid PET scan collection and processing
Structural MRI scans were processed by following a protocol similar to that of the Alzheimer’s Disease Neuroimaging Initiative (ADNI). Details of the structural brain MRI and amyloid PET protocols are provided elsewhere13,28. Regional volumes and cortical thickness from the MRI scans were obtained via the FreeSurfer image analysis suite29. A standardized uptake value ratio (SUVR) with correction for partial volume effects was calculated30 for the FreeSurfer Region-of-Interest (ROIs) for PiB or 18F-AV45. The cerebellum was chosen as the reference region. A composite measure of global Aβ 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. Values from this global summary were converted into centiloid units31 to harmonize tracer and data processing differences using previously published equations31,32.
Statistical analyses
We implemented random intercept and random slope models33 for each biomarker. These models allowed fixed intercepts and slopes specific to each race, and their random variation among participants. We further examined the effects of major AD risk factors, including baseline age, sex, APOE ε4 status, FH, years of education, Hollingshead index, and BMI as additional fixed effects, and compared the rates of change between racial groups after adjustments. We examined whether APOE ε4 status may modify longitudinal racial differences. Due to the exploratory nature of the study that aimed to generate critical hypotheses on longitudinal racial differences in AD biomarkers for future testing, no rigorous multiplicity adjustments were employed. A rigorous power analysis was provided, however, to assess the sample sizes necessary for a future study to test the generated hypotheses by powering the detection of the observed longitudinal racial differences in AD biomarkers34. All computations were conducted using the statistical programming language R (version 3.3.1)35 and the R package lme4 36.
Results
Characteristics of biomarker sub-cohorts
The baseline demographics and APOE ε4 status by race for each of the longitudinal sub-cohorts as defined by the marker modalities (CSF, amyloid PET, structural MRI, and cognition) are presented in Table 1. A total of 1359 participants (179 AAs, 1180 Whites) were CDR 0 at their baseline clinical or biomarker assessments, and completed the longitudinal assessments with at least one of the four modalities since baseline. Of the 179 AAs, longitudinal data were available on 37 participants for CSF biomarkers, 70 participants for MRI outcomes, 47 participants for amyloid PET uptakes, and 177 participants for cognitive tests. Of the 1180 Whites, longitudinal data were available on 330 participants for CSF biomarkers, 524 participants for MRI outcomes, 374 participants for amyloid PET uptakes, and 1135 participants for cognitive tests. Across the sub-cohorts, the median baseline age was about 64–71 years (Table 1). The proportion of participants with a family history of AD or dementia ranged from 54% to 71%. There were no statistically significant differences in baseline age and FH between AAs and Whites within each sub-cohort. There were no differences between AAs and Whites in the proportion of female participants (59%–71%) and the proportion of APOE ε4 carriers (32%–40%) except for in the sub-cohort with cognitive testing, in which AAs had a higher proportion. Although the median education is 16 years for both AAs and Whites across all sub-cohorts, Whites had higher mean years of education than AAs in the sub-cohort of MRI (Table 1). AAs had a higher BMI than Whites within each sub-cohort, and a lower socioeconomic status than Whites based on the Hollingshead index.
Table 1:
Baseline and longitudinal characteristics of the four cohorts defined by modalities
| Variable | CSF biomarkers | Amyloid PET | Structural MRI | Cognitive tests | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||||||||
| N | AA (n=37) | White (n=330) | p | N | AA (n=47) | White (n=374) | p | N | AA (n=70) | White (n=524) | p | N | AA (n=177) | White (n=1135) | p | |
| Age | 367 | 64.1 (58.7~68.6) | 65.9 (57.9~70.4) | 0.2004 | 421 | 65.2 (58.4~69.8) | 67.3 (60.1~72.5) | 0.2993 | 594 | 67.6 (60.9~72.4) | 68.1 (60.8~73.5) | 0.5324 | 1312 | 69.2(65.7~74.9) | 70.7(65.3~77.5) | 0.0952 |
| Education (yrs) | 367 | 16(13~18) | 16(15~18) | 0.1554 | 421 | 16(12.5~18) | 16(14~18) | 0.0752 | 594 | 16(13~18) | 16(14~18) | 0.0396 | 1312 | 16(13~18) | 16(13~18) | 0.8871 |
| BMI | 291 | 30.3(25.2~33.9) | 27.1(24.3~29.7) | 0.0108 | 396 | 30.1(25.3~33.6) | 26.9(23.9~29.9) | 0.0026 | 437 | 30.8(26.9~33.6) | 27.1(24.2~30.2) | 2.92E-05 | 710 | 31.01(27.1~35.1) | 27.1(24.2~30.2) | 3.54E-13 |
| Follow-up (yrs) | 367 | 5.7(3.4~9.0) | 6.4(3.7~9.8) | 0.213 | 421 | 4.2(3.2~8.0) | 5.9(3.2~8.4) | 0.3283 | 594 | 4.6(3.2~7.6) | 6(3.2~9.0) | 0.1709 | 1312 | 5.1(3.1~9.0) | 7.2(3.6~12.1) | 1.29E-05 |
| Sex | 367 | 1 | 421 | 0.7493 | 594 | 0.4348 | 0.0016 | |||||||||
| F | 22(59.5) | 194(58.8) | 31(66.0) | 234(62.6) | 46(65.7) | 316(60.3) | 1245 | 126(71.2) | 665(58.6) | |||||||
| M | 15(40.5) | 136(41.2) | 16(34.0) | 140(37.4) | 24(34.3) | 208(39.7) | 51(28.8) | 470(41.4) | ||||||||
| APOE ε4 status | 362 | 1 | 413 | 0.6256 | 584 | 0.6868 | 1245 | 0.0286 | ||||||||
| Negative | 24(64.9) | 212(65.2) | 33(70.2) | 242(66.1) | 47(68.1) | 335(65.1) | 103(59.9) | 735(68.5) | ||||||||
| Positive | 13(35.1) | 113(34.8) | 14(29.8) | 124(33.9) | 22(31.9) | 180(34.9) | 69(40.1) | 338(31.5) | ||||||||
| Family history | 361 | 0.7007 | 413 | 0.393 | 585 | 0.0734 | 0.2039 | |||||||||
| No | 12(33.3) | 95(29.2) | 16(35.6) | 108(29.3) | 29(42.6) | 162(31.3) | 1176 | 77(46.1) | 411(40.7) | |||||||
| Yes | 24(66.7) | 230(70.8) | 29(64.4) | 260(70.7) | 39(57.4) | 355(68.7) | 90(53.9) | 598(59.3) | ||||||||
| Hollingshead | 367 | 0.005 | 421 | 0.00006 | 594 | 0.00003 | 1312 | 0.00004 | ||||||||
| 1 | 5(13.5) | 106(32.1) | 4(8.5) | 113(30.2) | 7(10) | 169(32.4) | 29(16.4) | 330(29.1) | ||||||||
| 2 | 20(54.1) | 150(45.4) | 27(57.4) | 179(47.9) | 36(51.4) | 224(42.7) | 73(41.2) | 420(37) | ||||||||
| 3 | 5(13.5) | 53(16.1) | 6(12.8) | 58(15.5) | 13(18.6) | 91(17.4) | 40(22.6) | 249(21.9) | ||||||||
| 4 | 5(13.5) | 20(6.1) | 8(17.0) | 24(6.4) | 11(15.7) | 37(7.1) | 28(15.8) | 129(11.4) | ||||||||
| 5 | 2(5.4) | 1(0.3) | 2(4.3) | 0(0) | 3(4.3) | 3(0.6) | 7(4.0) | 7(0.6) | ||||||||
continuous characteristics are summarized as median (inter-quartile range) and compared between race by Wilcoxon rank sum test. Categorical characteristics are summarized as count (percentages) and compared between race by Fisher’s exact test or Chi-squared test as appropriate.
Longitudinal follow-up of sub-cohorts
Across the 3 biomarker sub-cohorts, the median duration of longitudinal follow-up ranged from 4.2 to 5.7 years for AAs and from 5.9 to 6.4 years for Whites (Table 1). The median duration of follow-up for cognitive tests was 5.1 years for AAs and 7.2 years for Whites. There was no statistically significant racial difference in the duration of follow-up in the sub-cohorts except for cognitive test sub-cohort. More than half of the participants for each race had at least 3 longitudinal assessments in all sub-cohorts with the exception of AAs in the CSF sub-cohort (18 out of 37) and the amyloid PET sub-cohort (19 out of 47). Supplemental Table 1 presents the sample sizes for each race as functions of the number of serial assessments across the four sub-cohorts as defined by modalities.
Baseline racial differences
The baseline levels of biomarkers and cognition in AAs and Whites were summarized (Table 2). No racial differences were observed for CSF Aβ42 and Aβ42/Aβ40, but amyloid PET centiloid was lower in AA than Whites (p=0.04). There was a trend towards lower levels of CSF pTau181 in AAs than Whites. Compared to Whites, AAs had smaller hippocampal volumes and cortical thickness and scored slightly lower on the cognitive composite.
Table 2.
Baseline levels of biomarkers and cognition as a function of race
| Marker | Race | n | Mean | SD | p-value for comparing AAs and Whites |
|---|---|---|---|---|---|
| CSF Aβ42 | AA | 37 | 810.946 | 414.368 | 0.4472 |
| White | 330 | 855.848 | 331.254 | ||
| CSF Aβ42/Aβ40 | AA | 37 | 0.084 | 0.016 | 0.1845 |
| White | 330 | 0.079 | 0.02 | ||
| CSF Tau | AA | 37 | 249.243 | 142.617 | 0.1324 |
| White | 330 | 305.793 | 222.879 | ||
| CSF pTau181 | AA | 37 | 31.443 | 15.62 | 0.0530 |
| White | 330 | 39.051 | 23.243 | ||
| Amyloid PET centiloid scale | AA | 47 | 2.454 | 10.987 | 0.0444 |
| White | 374 | 9.631 | 24.073 | ||
| MRI hippocampal volume | AA | 70 | 7174.131 | 1022.229 | 0.0050 |
| White | 524 | 7538.447 | 1014.925 | ||
| MRI cortical thickness | AA | 70 | 4.392 | 0.302 | 0.0009 |
| White | 524 | 4.501 | 0.251 | ||
| Cognitive Composite | AA | 177 | 0.073 | 0.226 | 0.0123 |
| White | 1135 | 0.12 | 0.235 |
Longitudinal racial differences
The annual rates of longitudinal change for each marker in AAs and Whites were evaluated (Table 3 and Figure 1). AAs had a slower decline in CSF Aβ42/Aβ40 (−0.0004/year ± standard error [SE] 0.0002/year) than Whites (−0.0009/year ± 0.0001/year; p=0.0390). Amyloid PET centiloid, which is inversely associated with CSF Aβ42/Aβ40, also demonstrated a slower rate of accumulation in AAs than Whites (0.85/year ± 0.30/year versus 1.61/year ± 0.10/year; p=0.0157). The estimated annual rate of change in CSF Aβ42 was −4.29 pg/mL (SE=1.44 pg/mL) for Whites, but +9.93 pg/mL (SE=4.68 pg/mL) for AAs, resulting in a statistically significant difference (p=0.0039). CSF Tau and pTau181 increased more slowly in AA than Whites, but the differences were not statistically significant (p=0.1715 and p=0.1395 for Tau and pTau181, respectively). There were no significant racial differences of longitudinal changes in hippocampal volume, cortical thickness, or performance on cognitive tests.
Table 3:
Estimated annual rates of longitudinal changes (i.e., the slope) for each race, SE, and the difference between the two races, and the P value for testing the difference
| Marker | RACE | Slope Estimate | Slope SE | Slope difference Estimate | Slope difference SE | P |
|---|---|---|---|---|---|---|
| CSF Aβ42 | AA | 9.9343 | 4.6779 | 14.2276 | 4.8956 | 0.0039 |
| White | −4.2933 | 1.4435 | ||||
| CSF Aβ42/Aβ40 | AA | −0.0004 | 0.0002 | 0.0005 | 0.0002 | 0.0390 |
| White | −0.0009 | 0.0001 | ||||
| CSF Tau | AA | 4.0369 | 3.3816 | −4.8758 | 3.5579 | 0.1715 |
| White | 8.9127 | 1.1059 | ||||
| CSF pTau181 | AA | 0.6948 | 0.3285 | −0.5122 | 0.3454 | 0.1395 |
| White | 1.2070 | 0.1069 | ||||
| Amyloid PET centiloid scale | AA | 0.8475 | 0.2956 | −0.7624 | 0.3135 | 0.0157 |
| White | 1.6099 | 0.1043 | ||||
| MRI hippocampal volume | AA | −60.2642 | 9.3131 | 4.5767 | 9.8167 | 0.6413 |
| White | −64.8409 | 3.1040 | ||||
| MRI cortical thickness | AA | −0.0154 | 0.0023 | −0.0007 | 0.0024 | 0.7610 |
| White | −0.0147 | 0.0007 | ||||
| Cognitive Composite | AA | −0.0123 | 0.0022 | 0.0031 | 0.0023 | 0.1747 |
| White | −0.0154 | 0.0007 |
Figure 1:
Spaghetti plots as functions of age, along with the model-fitted longitudinal trajectories (AB40= Aβ40, pTau=pTau181, AB42= Aβ42)
Adjusted analyses that also included tracer as a covariate revealed that amyloid accumulation as measured by amyloid PET centiloid remained slower for AA compared to Whites (1.30/year ± 0.26/year versus 1.86/year ± 0.11/year; p=0.0326). The adjusted estimate for the annual rate of change in CSF Aβ42/Aβ40 was −0.0005 ± 0.0001 for AAs and −0.0009 ± 0.0001 for Whites, and the difference was no longer statistically significant (p=0.09). The racial difference in the annual rate of change remained statistically significant for CSF Aβ42 (p=0.037) with an adjusted estimate of 9.63±5.15 pg/mL per year for AAs and −1.62±1.97 pg/mL per year for Whites.
Role of APOE ε4 in longitudinal racial difference
The AA APOE ε4 positive group, but not APOE ε4 negative group, had declining Aβ42/Aβ40 (p=0.014 and p=0.39, respectively; Supplemental Table 2). For Whites, both APOE ε4 positive and negative groups had declining Aβ42/Aβ40. The White participants’ annual decrease in CSF Aβ42 appeared to be primarily from APOE ε4 positive participants (slope=−11.48±2.40 pg/mL, p=2.73E-06) in comparison to APOE ε4 negative participants (slope=−0.94 ±1.74 pg/mL, p=0.59). The unexpected annual increase in CSF Aβ42 for AAs appeared to be primarily from APOE ε4 negative participants (slope=18.85±6.09 pg/mL, p=0.002) in comparison to APOE ε4 positive participants (slope=−5.68±6.92 pg/mL, p=0.41), but there was no significant interaction between APOE ε4 and race on the rate of change (p=0.1494). All groups accumulated amyloid over time as measured by amyloid PET centiloid. However, AAs had a slower rate than Whites (1.160 versus 2.357, p=0.028) in APOE ε4 positive participants, but not in APOE ε4 negative participants (p=0.28), and no significant interaction between APOE ε4 and race was observed.
Sample sizes for a future study to detect the observed longitudinal racial differences
For CSF Aβ42, Aβ42/Aβ40, Tau, pTau181, and amyloid PET centiloid, fairly large racial differences (in the range of ~50%) were observed in the annual rates of change. Table 4 presents the sample sizes required to detect these differences with at least 80% power in a future longitudinal study, with Bonferroni’s adjustment on multiplicity, i.e., at a significance level of 1%, and a sample size ratio of AAs to Whites either 1:1 or 1:2. These power analyses were based on the random intercept and random slope model and an asymptotic standard normal test to compare the rates of change between the races33, and assumed longitudinal assessments every two years for a total of 6 years. A total of 220 AAs for a 1:1 design but only 165 AAs are required for a 1:2 design to detect the longitudinal racial differences in CSF Aβ42/Aβ40. Slightly fewer AAs are required to adequately power studies of amyloid PET centiloid (198 for the 1:1 design, 148 for the 1:2 design). Larger cohorts are required to power future studies of CSF Tau and pTau181—approximately 400 AAs for the 1:1 design.
Table 4:
Sample sizes necessary to detect the observed longitudinal racial differences (i.e., the annual rates of change) in a future longitudinal study (6 years follow-up with assessments every 2 years, 1:1 or 1:2 sample size ratio assumed between AAs and Whites)
| Marker | Observed slope difference (AA-White) | Variance for the random slope | Variance for the random error | AAs needed for a 1:1 design | AAs needed for a 1:2 design |
|---|---|---|---|---|---|
| CSF Aβ42 | 14.2275 | 16 | 130 | 128 | 96 |
| CSF Aβ42/Aβ40 | 0.0004964 | 0.0008743 | 0.0055696 | 220 | 165 |
| Amyloid PET centiloid | −0.7624 | 1.546 | 7.095 | 198 | 148 |
| CSF Tau | −4.876 | 17.22 | 47.28 | 402 | 301 |
| CSF pTau181 | −0.5122 | 1.633 | 6.488 | 425 | 319 |
Discussion
A fundamental question in designing and analyzing clinical trials on AD is whether AD pathophysiology is the same for AAs and Whites. Especially for designing prevention trials on AD, well established imaging and CSF biomarkers are critical for estimating the risk profiles of individuals for enrollment, and for demonstrating the efficacy of the treatments. Thus far, only a few biomarker studies have examined racial differences between AAs and Whites. All were cross-sectional and included both symptomatic and asymptomatic participants. Some reported mixed findings in cross-sectional racial difference in PET PiB SUVR37–39. Others found that CSF Tau and pTau181 were lower in AAs than Whites12–13.
In this preliminary study we reported the first set of findings on longitudinal racial differences among cognitively normal AAs and Whites in all major validated AD biomarkers. We found slower rate of accumulation of amyloid in the brain, and slower rate of decline of CSF Aβ42/ Aβ40, in AAs than Whites. We also found significant longitudinal racial difference in CSF Aβ42 with AAs showing an unexpected increase. We did not find statistically significant longitudinal racial differences in CSF Tau (p=0.1715) and pTau181 (p=0.1395), although the estimated annual increase in both markers for AAs was only roughly half of that for Whites. Further, some of the racial differences in amyloid biomarkers may be modified by APOE ε4 status.
Given that it has been well established that CSF and imaging biomarkers predict clinical and cognitive outcomes4–5,8, our findings are consistent with a recent large study on racial differences in the risk of developing AD dementia40. The study analyzed longitudinal clinical data from 1229 AAs and 6679 Whites and found that overall, cognitively normal older AAs had a lower, but not significantly different, risk of AD dementia than Whites, supporting our findings that cognitively normal AAs had a reduced magnitude on the rates of change for several CSF and imaging biomarkers. Further, the study40 found that the racial difference in the risk of AD dementia was statistically significant only among APOE ε4 positive participants, also consistent with our observation that AAs had a slower rate of increase than Whites in the amyloid accumulation only among APOE ε4 positive participants.
Our baseline analyses suggested a trend towards lower baseline levels of CSF Tau and pTau181 in AAs than Whites, but the differences were not statistically significant. Because our study had a much smaller sample size of AAs than others12–13, the inconsistency with previous findings12–13 may be due to the limited statistical power in the current study. Our power analyses indicated a much larger sample size is needed to detect the longitudinal racial differences in CSF Tau and pTau181. Taken together, our findings highlight the importance to clearly differentiate the racial differences between cross-sectional and longitudinal designs. Whereas the latter focuses on the longitudinal changes that can facilitate cross-sectional inferences, the former cannot be interpreted in the context of within-subject changes.
Our findings generate several interesting hypotheses, and may have important implications, both in understanding the validity of established biomarkers and the natural history of AD as a function of ethnoracial factors and in designing prevention trials on AD. First, the findings of slower longitudinal changes among asymptomatic AAs in CSF and PET amyloid biomarkers, along with the surprising observation that CSF Aβ42 increased over time among AAs, suggest that AAs and Whites may have very different time windows on amyloid metabolism and different longitudinal trajectories, and that the temporal ordering of these biomarkers, as hypothesized in the literature41 and recently summarized in the ATN framework42, may not be shared between AAs and Whites. This could have implications for preventive approaches, including different age windows for preventions. Second, whereas cross-sectional racial differences as reported in CSF Tau and pTau18112–13 may necessitate future prevention trials to adopt race-specific cutoffs for participant enrollment in Tau-based interventions, the longitudinal racial differences in amyloid we found, may indicate different response rates or effect size for any amyloid-based interventions, and hence different statistical power across races. Third, our findings suggest that for future biomarker studies including prevention trials, longitudinal analyses need to allow the potential modifying effect of race on the rate of change. Importantly, except for the fact that baseline levels of these biomarkers often predict the subsequent rate of longitudinal changes, it remains unknown about the mechanism of the observed longitudinal racial differences. Additional and much larger longitudinal studies with adequate representation of AAs and URG, as recommended by the International Society to Advance Alzheimer’s Research and Treatment, Alzheimer’s Association43, are urgently needed to fully test these hypotheses. Finally, our findings on longitudinal racial differences in AD biomarkers must be interpreted with caution. We must consider the growing scientific research related to race as a social construct, taking into considerations the impact of racial identity, lived experiences, real and perceived experiences of racism may have on racial differences. There is evidence that sociocultural factors related to residential and school segregation, access and quality of health care, occupational safety, ability to build and maintain wealth, experiences with the legal system and violence exposure are linked directly to structural and systemic racism1. More importantly, URGs often live with chronic comorbid health conditions that are associated with a greater prevalence of AD. There is an association of AD with social and structural determinants of health linked directly to environment, education, poverty, and experiences with adversity. This highlights the need for health disparities and equity research to use what the Alzheimer’s Association calls “a lifecourse perspective” to examine the differences in AD across racial groups in future studies. Hence, the AD research community faces a major challenge to develop and standardize comprehensive assessments of lifecourse Social Determinants of Heath (SDOH) that can help interpret and understand the possible racial differences in AD biomarkers.
The biggest limitation of the study is the small sample sizes of AAs, which prevent a comprehensive analysis of longitudinal racial differences and their modification factors, along with the covariates that may relate directly to the willingness and readiness of AAs to participate in research, the perceptions and even experience related to quality of healthcare. Another limitation is the retrospective and convenience nature of the AAs and Whites cohorts, which may not represent the general population. Whereas we have provided adjusted analyses that indicated largely consistent longitudinal racial differences after adjusting for the effects of all major AD risk factors, including baseline age, sex, APOE ε4 status, FH, years of education, Hollingshead index, and BMI, these analyses do not erase the highly significant differences in socioeconomic status (i.e., Hollingshead index) between AAs and Whites in our cohort (Table 1). Hence, selection bias, including enrollment factors and referral sources44 and SDOH, and their possible confounding with our observed longitudinal racial differences in AD biomarkers, cannot be ruled out in our inferences. Prospectively designed longitudinal AD biomarker studies that appropriately balance the major SDOH between races will be needed.
Conclusions
We found longitudinal racial differences in amyloid biomarkers from CSF and PET imaging. The main findings need to be fully replicated in larger prospectively designed longitudinal biomarker studies with comprehensive assessments of SDOH that may help interpret the possible longitudinal racial differences. Hence, findings from our study are not definite, yet the hypotheses they generated, if confirmed, may inform the design of future prevention trials of AD on the optimal biomarker target, time window of intervention, outcome measures, inclusion/exclusion criteria, and statistical power, all of which may have to be race specific.
Supplementary Material
Research in Context.
Systematic review: All relevant articles on PubMed relating to longitudinal racial differences in cerebrospinal (CSF) and imaging biomarkers of Alzheimer disease (AD) among cognitively normal individuals were searched. Whereas several evaluated cross-sectional racial differences in AD biomarkers that are appropriately cited, very few reported longitudinal differences in these biomarkers.
Interpretation: Our findings suggest that the longitudinal changes in some of the AD biomarkers may depend on race, consistent with reported cross-sectional differences between African Americans and Caucasians. These results, if confirmed, may have important implications in the design and analyses of future prevention trials of AD.
Future directions: The main findings need to be fully replicated in larger prospectively designed longitudinal biomarker studies. Further, the underlying mechanism for the observed longitudinal racial differences remains unknown, and Social Determinants of Heath (SDOH) may help interpret these differences. Hence, future studies with comprehensive assessments of SDOH will be needed.
Study Funding and Acknowledgements.
This study was supported by National Institute on Aging (NIA) grant R01 AG067505 and R01 AG053550 (Dr. Xiong) and NIA grants P50 AG005681, P01AG026276, and P01 AG0399131 (Dr. Morris).
Glossary:
- AD
Alzheimer Disease
- CSF Aβ42/Aβ40
The ratio of concentrations of CSF amyloid-β peptide 42 and 40
- CSF total tau
Concentration of total tau in the CSF
- CSF pTau181
concentration of tau phosphorylated at 181 in the CSF
- PiB
[11C] benzothiazole tracer, Pittsburgh Compound-B
- SUVR
standardized uptake value ratio
- CI
confidence interval
- SE
standard error
- IQR
interquartile range
Footnotes
Disclosure
Drs. Xiong, Schindler, Fagan, Benzinger, Hassenstab, Balls-Berry, Moulder, and Morris all have received research funding from the National Institute on Aging of the National Institutes of Health that was made to their institutions.
Dr. Hassenstab also has received BrightFocus grant that was made to his institution.
Dr. Morris received royalties or licenses for CDR registration, and received support for attending meetings and/or travel (Srinivasan 40th Oration, India; World Congress of Neurology; Cure Alzheimer’s Board meeting; CBR Intl’ Advisory Board).
Dr. Xiong consults for Diadem. There are no conflicts.
Dr. Schindler consults for National Institute on Aging Alzheimer Disease Center Clinical (ADC) Task Force, to me National Centralized Repository for Alzheimer Disease.
Dr. Fagan has received research funding from the National Institute on Aging of the National Institutes of Health, Biogen, Centene, Fujirebio and Roche Diagnostics. She is a member of the scientific advisory boards for Roche Diagnostics, Genentech and AbbVie and also consults for Diadem, DiamiR and Otsuka Pharmaceuticals. Dr. Fagan also consults for Seimens Healthcare Diagnostics. There are no conflicts.
Dr. Benzinger consults for Biogen. There are no conflicts.
Dr. Hassenstab consults for Lundbeck, Eisai, Roch, and Parabon Labs. There are no conflicts.
Dr. Morris consults for Barcelona Betabrain Research Center, BBRC SAB meeting, Barcelona Centre for Brain Research meeting, Bangalore, India. There are no conflicts.
Dr. Schindler received payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from University of Wisconsin and St. Luke’s Hospital. There are no conflicts.
Dr. Benzinger received payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events from Biogen. There are no conflicts.
Dr. Hassenstab received payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events: (seminar speaker) from Alzheimer’s Therapeutic Research Institute (ATRI), University of Southern California. There are no conflicts.
Dr. Balls-Berry received payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events: (Keynote lecture) from University of Kansas Medical Center Diversity Black History Month Research Day Kansas City, Kansas; (Keynote lecture) from INSciTS: International Network for the Science of Team Science: Building the Knowledge Base for Effective Team Science; and (Norman R. Seay Lecture) from The Knight Alzheimer’s Disease Research Center (Knight ADRC), Washington University School of Medicine, Missouri 2018). There are no conflicts.
Dr. Morris received payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events: (Grand Rounds lecture) from Montefiore, NY. There are no conflicts.
Drs. Xiong, Luo, Benzinger, Hassenstab, and Balls-Berry all served on Data Safety Monitoring Board or Advisory Board for FDA or NIH-funded studies. There are no conflicts.
Dr. Schindler is a Member of the Board of Directors, Alzheimer’s Association Greater Missouri Chapter. Dr. Balls-Berry is President of the Board of Directors for Health Literacy Media. Dr. Morris is a member of Cure Alzheimer’s Board.
Avid Radiopharmacueticals and Life Molecular Imaging have provided reagents and technology transfer agreements to Dr. Benzinger’s institution for the production of radiopharmaceuticals.
Dr. Grant and Ms. Agboola have nothing to disclose.
References
- 1.Alzheimer’s Association. Alzheimer’s Disease Facts and Figures (https://www.alz.org/media/Documents/alzheimers-facts-and-figures.pdf) 2021.
- 2.Klunk WE, Engler H, Nordberg A, et al. Imaging brain amyloid in Alzheimer’s disease with Pittsburgh Compound-B. Ann Neurol. 55: 2004; 306–319. [DOI] [PubMed] [Google Scholar]
- 3.Fagan AM, Xiong CJ, Jasielec MS, et al. Longitudinal Change in CSF Biomarkers in Autosomal-Dominant Alzheimer’s Disease. Science Translational Medicine. 2014;6(226). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Xiong CJ, Jasielec MS, Weng H, et al. Longitudinal relationships among biomarkers for Alzheimer disease in the Adult Children Study. Neurology. 2016;86(16):1499–1506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Morris JC, Price JL. Pathologic correlates of nondemented aging, mild cognitive impairment, and early-stage Alzheimer’s disease. J Mol Neurosci. 2001;17(2):101–118. [DOI] [PubMed] [Google Scholar]
- 7.Bateman RJ, Xiong C, Benzinger TL, et al. Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N Engl J Med. 2012;367(9):795–804. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Vos SJB, Xiong CJ, Visser PJ, et al. Preclinical Alzheimer’s disease and its outcome: a longitudinal cohort study. Lancet Neurol. 2013;12(10):957–965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Price JL, McKeel DW Jr., Buckles VD, et al. Neuropathology of nondemented aging: presumptive evidence for preclinical Alzheimer disease. Neurobiol Aging. 2009;30(7):1026–1036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jack CR, Knopman DS, Jagust WJ, et al. Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol. 2013;12(2):207–216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Luo J, Agboola F, Grant E, et al. Sequence of Alzheimer disease biomarker changes in cognitively normal adults: A cross-sectional study. Neurology. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Howell JC, Watts KD, Parker MW, Wu J, Kollhoff A, Wingo TS, Dorbin CD, Qiu D, Hu WT. Race modifies the relationship between cognition and Alzheimer’s disease cerebrospinal fluid biomarkers. Alzheimer’sRes Ther. 2017;9(1):88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Morris JC, Schindler SE, McCue LM, Moulder KL, Benzinger TLS, Cruchaga C, Fagan AM, Grant E, Gordon BA, Holtzman DM, Xiong C. Assessment of Racial Disparities in Biomarkers for Alzheimer Disease. JAMA Neurol. 2019; 76(3):264–273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hollingshead AB (1957) Two factor index of social position. Yale University Press, New Haven. [Google Scholar]
- 15.Morris JC, Weintraub S, Chui HC, et al. The uniform data set (UDS): Clinical and cognitive variables and descriptive data from Alzheimer disease centers. Alz Dis Assoc Dis. 2006;20(4):210–216. [DOI] [PubMed] [Google Scholar]
- 16.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. Alzheimers & Dementia. 2011;7(3):263–269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Morris JC. The Clinical Dementia Rating (CDR) - Current Version and Scoring Rules. Neurology. 1993;43(11):2412–2414. [DOI] [PubMed] [Google Scholar]
- 18.Xiong C, Roe CM, Buckles V, Fagan AM, Holtzman D, Balota D, Duchek J, Storandt M, Mintun M, Grant E, Snyder AZ, Head D, Benzinger TLS, Mettenburg J, Csernansky J, Morris JC. Role of family history for Alzheimer biomarker abnormalities in the adult children study. Archives of Neurology 2011; 68(10):1313–1319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–198. [DOI] [PubMed] [Google Scholar]
- 20.Rosen WG. Verbal fluency in aging and dementia. Journal of Clinical Neuropsychology. 1980;2(2):135–146. [Google Scholar]
- 21.Wechsler D. Wechsler Adult Intelligence Scale. New York: Psychological Corporation; 1955. [Google Scholar]
- 22.Goodglass H, Kaplan E. Boston Diagnostic Aphasia Examination Booklet, III, Philadelphia, PA: Lea & Febiger, 1983. [Google Scholar]
- 23.Grober E, Buschke H, Crystal H, Bang S, & Dresner R. Screening for dementia by memory testing. Neurology 1988;3:900–903. [DOI] [PubMed] [Google Scholar]
- 24.Armitage SG. An analysis of certain psychological tests used for the evaluation of brain injury. Psychological Monographs 1945;60 (1, Whole No. 177), 1–48. [Google Scholar]
- 25.Keshavan A, Wellington H, Chen Z, et al. Concordance of CSF measures of Alzheimer’s pathology with amyloid PET status in a preclinical cohort: A comparison of Lumipulse and established immunoassays. Alzheimers Dement (Amst). 2020;12(1):e12097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Leitao MJ, Silva-Spinola A, Santana I, et al. Clinical validation of the Lumipulse G cerebrospinal fluid assays for routine diagnosis of Alzheimer’s disease. Alzheimers Research & Therapy. 2019;11(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Alcolea D, Pegueroles J, Munoz L, et al. Agreement of amyloid PET and CSF biomarkers for Alzheimer’s disease on Lumipulse. Annals of clinical and translational neurology. 2019;6(9):1815–1824. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Xiong C, Luo J, Agboola F, et al. A harmonized longitudinal biomarkers and cognition database for assessing the natural history of preclinical alzheimer disease from young adulthood and for for designing prevention trials. Alzheimer’s & dementia. 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.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]
- 30.Su Y, Flores S, Wang G, Hornbeck RC, Speidel B, Joseph-Mathurin N, Vlassenko AG, Gordon BA, Koeppe RA, Klunk WE, Jack CR Jr., Farlow MR, Salloway S, Snider BJ, Berman SB, Roberson ED, Brosch J, Jimenez-Velazques I, van Dyck CH, Galasko D, Yuan SH, Jayadev S, Honig LS, Gauthier S, Hsiung GR, Masellis M, Brooks WS, Fulham M, Clarnette R, Masters CL, Wallon D, Hannequin D, Dubois B, Pariente J, Sanchez-Valle R, Mummery C, Ringman JM, Bottlaender M, Klein G, Milosavljevic-Ristic S, McDade E, Xiong C, Morris JC, Bateman RJ, Benzinger TLS, Comparison of Pittsburgh compound B and florbetapir in cross-sectional and longitudinal studies. Alzheimers Dement (Amst) 11, 180–190 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Klunk WE, Koeppe RA, Price JC, Benzinger TL, Devous MD Sr., Jagust WJ, Johnson KA, Mathis CA, Minhas D, Pontecorvo MJ, Rowe CC, Skovronsky DM, Mintun MA, The Centiloid Project: standardizing quantitative amyloid plaque estimation by PET. Alzheimers Dement 11, 1–15 e11–14 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Su Y, Flores S, Hornbeck RC, Speidel B, Vlassenko AG, Gordon BA, Koeppe RA, Klunk WE, Xiong C, Morris JC, Benzinger TLS, Utilizing the Centiloid scale in cross-sectional and longitudinal PiB PET studies. NeuroImage. Clinical 19, 406–416 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics. 1982;38(4):963–974. [PubMed] [Google Scholar]
- 34.Xiong C, Zhu K, Yu K, Miller JP. Statistical Modeling in Biomedical Research: Longitudinal Data Analysis. in Handbook of Statistics: Epidemiology and Medical Statistics, Edited by Rao CR, Philip Miller J, and Rao DC Amsterdam: Elsevier B.V., 2008. [Google Scholar]
- 35.R: A language and environment for statistical computing [computer program]. Vienna, Australia: R Foundation for Statistical Computing; 2017. [Google Scholar]
- 36.Bates D, Machler M, Bolker BM, Walker SC. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015;67(1):1–48. [Google Scholar]
- 37.McDonough IM. Beta-amyloid and cortical thickness reveal racial disparities in preclinical Alzheimer’s disease. Neuroimage Clin. 2017. Sep 21;16:659–667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Deters KD, Napolioni V, Sperling RA, Greicius MD, Mayeux R, Hohman T, Mormino EC. 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–e1500. doi: 10.1212/WNL.0000000000011599. Epub 2021 Feb 10. PMID: 33568538; PMCID: PMC8032379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Gottesman RF, Schneider AL, Zhou Y, et al. The ARIC-PET amyloid imaging study: Brain amyloid differences by age, race, sex, and APOE. Neurology. 2016;87(5):473–480. doi: 10.1212/WNL.0000000000002914 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Xiong C, Luo J, Coble D, Agboola F, Kukull W, Morris JC. Complex interactions underlie racial disparity in the risk of developing Alzheimer’s disease dementia. Alzheimers Dement. 2020. Apr;16(4):589–597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Jack CR, Knopman DS, Jagust WJ, et al. Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol. 2013;12(2):207–216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Jack CR, Bennett DA, Blennow K, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers & Dementia. 2018;14(4):535–562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Babulal GM, Quiroz YT, Albensi BC, Arenaza-Urquijo E, Astell AJ, Babiloni C, Bahar-Fuchs A, Bell J, Bowman GL, Brickman AM, Chételat G, Ciro C, Cohen AD, Dilworth-Anderson P, Dodge HH, Dreux S, Edland S, Esbensen A, Evered L, Ewers M, Fargo KN, Fortea J, Gonzalez H, Gustafson DR, Head E, Hendrix JA, Hofer SM, Johnson LA, Jutten R, Kilborn K, Lanctôt KL, Manly JJ, Martins RN, Mielke MM, Morris MC, Murray ME, Oh ES, Parra MA, Rissman RA, Roe CM, Santos OA, Scarmeas N, Schneider LS, Schupf N, Sikkes S, Snyder HM, Sohrabi HR, Stern Y, Strydom A, Tang Y, Terrera GM, Teunissen C, Melo van Lent D, Weinborn M, Wesselman L, Wilcock DM, Zetterberg H, O’Bryant SE; International Society to Advance Alzheimer’s Research and Treatment, Alzheimer’s Association. Perspectives on ethnic and racial disparities in Alzheimer’s disease and related dementias: Update and areas of immediate need. Alzheimers Dement. 2019. Feb;15(2):292–312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.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. Alzheimers Dement. 2019;15(12):1533–1545. [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.