sPDGFRβ and neuroinflammation are associated with AD biomarkers and differ by race: The ASCEND Study

Abstract

INTRODUCTION

There remains an urgent need to identify preclinical pathophysiological mechanisms of Alzheimer's disease (AD) development in high‐risk, racially diverse populations. We explored the relationship between cerebrospinal fluid (CSF) markers of vascular injury and neuroinflammation with AD biomarkers in middle‐aged Black/African American (B/AA) and non‐Hispanic White (NHW) participants.

METHODS

Adults (45–65 years) with a parental history of AD were enrolled (n = 82). CSF and blood biomarkers were collected at baseline and year 2.

RESULTS

CSF total tau (t‐tau), phosphorylated tau (p‐tau), and amyloid beta (Aβ)40 were elevated at year 2 compared to baseline. CSF soluble platelet‐derived growth factor receptor β (sPDGFRβ) levels, a marker of pericyte injury, correlated positively with t‐tau, p‐tau, Aβ40 markers of vascular injury, and cytokines at baseline and year 2. CSF sPDGFRβ and tau were significantly lower in B/AA than NHW.

DISCUSSION

Vascular dysfunction and neuroinflammation may precede cognitive decline and disease pathology in the very early preclinical stages of AD, and there are race‐related differences in these relationships.

Highlights

• Cerebrospinal fluid (CSF) Alzheimer's disease (AD) biomarkers changed over 2 years in high‐risk middle‐aged adults.

• Markers of vascular dysfunction were associated with the CSF biomarkers amyloid beta and tau.

• AD biomarkers were lower in Black compared to non‐Hispanic White individuals.

• Markers of vascular dysfunction were lower among Black individuals.

1. INTRODUCTION

The increasing prevalence of Alzheimer's disease (AD) remains unaddressed due to the absence of effective disease‐modifying therapies, and there remains a significant and urgent need to identify preclinical pathophysiological mechanisms responsible for AD development in high‐risk, racially diverse populations. As pathophysiological mechanisms leading to the development of AD begin many years before clinical manifestations, ¹ there is a need to study biomarkers that reflect underlying pathological processes that precede the emergence of clinical symptoms in AD. ¹ , ² Accumulation of these biomarkers, defined by amyloid beta (Aβ) deposition, formation of fibrillar tau, and neurodegeneration (AT[N]), is a progressive and continuous process that begins years before the onset of clinical symptoms, often during midlife. ²

Multifactorial putative mechanisms driving the neuropathology of AD, originally described by the amyloid cascade ³ and cholinergic ⁴ hypotheses, have led to the development of emerging complementary theories, including those that implicate vascular dysfunction, ⁵ angiotensin II, ⁶ and inflammation. ⁷ Neurovascular dysfunction, including reduced cerebral blood flow and blood–brain barrier (BBB) breakdown, occur in early stages of AD. ⁸ , ⁹ Markers of endothelial injury are elevated in cerebrospinal fluid (CSF), serum, and pathological studies in AD. ¹⁰ , ¹¹ , ¹² Renin–angiotensin system (RAS) dysregulation is implicated in AD pathology. ⁶ , ¹³ Increased angiotensin‐converting enzyme (ACE1) activity and decreased ACE2 activity have been reported in CSF and post mortem brains in AD and are associated with AD pathology. ¹¹ , ¹⁴ Increased ACE1, proposed to play a role in Aβ degradation, is associated with higher levels of angiotensin II, which has multifactorial deleterious effects on several pathways associated with AD, including hypertension, vascular dysfunction, alterations in BBB function, and inflammation. ¹⁴ , ¹⁵ Angiotensin II–induced cytokine release is a likely contributor to AD pathology through amplification of neuroinflammation and activation of microglia. ⁷ , ¹⁶ , ¹⁷

Pericytes appear to be crucial in maintaining vascular integrity and regulating cerebral blood flow and BBB permeability; ¹⁸ , ¹⁹ endothelial cell injury and damage to pericytes play a key role in AD pathology. ²⁰ , ²¹ Pericytes and endothelial‐expressed matrix metalloproteinases (MMPs) and tissue inhibitor of metalloproteinases (TIMPs) are involved in vascular homeostasis and dysregulated in AD. ²² , ²³ In AD post mortem brain tissue, there is significant pericyte loss and a reduction in platelet‐derived growth factor receptor β (PDGFRβ) levels. ²⁴ , ²⁵ Pericytes in culture shed soluble PDGFRβ (sPDGFRβ) under hypoxic conditions or exposure to Aβ peptides. ²⁶ Elevated CSF levels of sPDGFRβ have been associated with BBB breakdown within the hippocampus in mildly cognitively impaired participants ¹⁵ and were reported to predict cognitive decline in early stages of AD independently of changes in Aβ and tau. ²⁷ We have previously shown that CSF sPDGFRβ levels were elevated from clinical AD cases and correlated with CSF albumin, a marker of BBB breakdown, and levels of CSF total tau (t‐tau) and phosphorylated tau (p‐tau). ²⁸

Black/African American adults (B/AAs) are 64% more likely to develop AD than non‐Hispanic White adults (NHWs). ²⁹ The higher incidence and prevalence of AD among B/AAs are often attributed to sociocultural issues and biological and genetic factors related to higher cardiovascular risk, such as diabetes and hypertension. More work is needed to examine race‐associated differences in AD biomarkers as these differences may further contribute to health disparities in AD diagnosis and potential success of preventive interventions.

Recent studies in B/AA and NHW adults found that CSF levels of t‐tau and p‐tau are lower in cognitively impaired B/AA older adults ³⁰ , ³¹ and cognitively unimpaired B/AA middle‐aged adults, ³² compared to NHWs, and these cognitive changes in B/AA adults are associated with smaller changes in CSF tau. ³² We previously reported baseline race‐related differences in Aβ and tau in cognitively unimpaired middle‐aged B/AA and NHW adults with a parental history of AD. ³² In this study, we explored the relationship between CSF markers of vascular injury and neuroinflammation with established markers of disease pathology, and investigated racial differences in a diverse middle‐aged at‐risk cohort over a 2‐year follow‐up period.

2. METHODS

2.1. Study design and sample

The Association Between Cardiovascular Risk and Preclinical Alzheimer's Disease Pathology (ASCEND) Study was a 2‐year observational study of cognitively unimpaired or mildly impaired, middle‐aged B/AA and NHW adults at risk for AD due to parental history. Baseline results were reported previously. ³² Briefly, we enrolled 82 middle‐aged (≥45 years) adults with a biological parent with either autopsy‐confirmed or probable AD as defined by National Institute of Neurological and Communicative Disorders and Stroke and Alzheimer's Disease and Related Disorders Association criteria ³³ and verified using the validated Dementia Questionnaire ³⁴ and medical records when available. Most participants were recruited from the Alzheimer's Disease Research Center, where the parent with AD underwent a full medical work‐up by a physician with expertise in dementia diagnoses.

Exclusion criteria included contraindication for lumbar puncture (LP); history of significant neurologic disease, head trauma, or major depression within the last 2 years; history of alcohol or substance abuse; diagnosis of AD, mild cognitive impairment (MCI), or residence in a skilled nursing facility; use of investigational medication; and unwillingness to fast. The Montreal Cognitive Assessment (MoCA) was used to assess cognitive impairment. MoCA scores >26 were considered normal, and scores between 18 and 25 were categorized as MCI. ³⁵ The ASCEND study included three annual visits (baseline, year 1, and year 2). Participants underwent LP at baseline and year 2 and a blood draw on all visits.

Medical history and sociodemographic information were collected via a self‐report questionnaire. As race is a social construct and collected by self‐report, terms and concepts around race and ethnicity are not universal. In the United States some individuals self‐identify race as Black, while others identify as African American. To be most inclusive, we are using both terms as they best reflect the racial identities of the study population.

RESEARCH IN CONTEXT

1. Systematic review: We reviewed the literature using traditional (e.g., PubMed, GoogleScholar) sources. Pathophysiologic mechanisms leading to the development of Alzheimer's disease (AD) often begin during mid‐life, especially among high‐risk individuals. Several pathologic mechanisms beyond amyloid and cholinergic processes are implicated in AD pathology, including vascular and renin angiotensin system dysfunction and inflammation. Although Black Americans have a higher incidence and prevalence of AD than non‐Hispanic White Americans, recent studies have found lower levels of cerebrospinal fluid (CSF) amyloid beta and tau. Identification of preclinical pathologic mechanisms leading to AD in racially diverse groups is urgently needed.

2. Interpretation: In a cohort of cognitively unimpaired middle‐aged adults with a parental history of AD, we found CSF AD markers were related to markers of vascular dysfunction, blood–brain barrier breakdown, and inflammation over 2 years. We also found evidence of changes in established AD markers over the 2‐year period, with differences by race remaining over time.

3. Future directions: Further work is warranted to better understand the role of cerebral vascular dysregulation in early pathologic changes leading to AD and how these changes may differ by race.

2.2. CSF and blood collection and analyses

After an 8 hour overnight fast, participants underwent LP to collect CSF for Aβ and tau, and markers of vascular dysfunction and inflammatory cytokines and chemokines, as previously described. ³² Participants also underwent blood draw for analysis of (1) RAS function, including ACE activity; (2) plasma inflammatory markers; and (3) apolipoprotein E (APOE) genotyping.

CSF Aβ40, Aβ42, t‐tau, and p‐tau concentration were measured by Lumipulse technology ³⁵ , ³⁶ (Fujirebio). Cut‐offs for normal values were: Aβ1‐42 > 526 pg/mL; t‐tau < 409 pg/mL; p‐tau < 50.2 pg/mL; and Aβ42/40 < 0.072. ³⁶ CSF sPDGFRβ level was determined using a commercially available sandwich enzyme‐linked immunosorbent assay (Invitrogen Catalog # EHPDGFRB; ThermoFisher Scientific) following the manufacturer's protocol as previously described. ²⁵

Cytokines and chemokines (interleukin [I]L‐1α, IL‐1β, IL‐4, IL‐6, IL‐7, IL‐8, IL‐9, IL‐10, tumor necrosis factor alpha [TNF‐α], transforming growth factor alpha [TGF‐α], interferon [IFN]‐γ, TNF receptor [TNFR]1, TNFR2, macrophage‐derived chemokine [MDC], monocyte chemoattractant protein 1 [MCP‐1], and fractalkine [CX3CL1]) were measured in CSF and plasma using the MILLIPLEX MAP Human Cytokine/Chemokine Magnetic Bead Panel (HCYTOMAG‐60K; Merck‐Millipore). Endothelial injury markers (intercellular adhesion molecule 1 [ICAM‐1] and vascular cellular adhesion molecule 1 [VCAM‐1]) were measured using MILLIPLEX MAP Human Neurodegenerative Magnetic Bead Panel 3 (HNDG3MAG‐36K; Merck‐Millipore). C‐reactive protein (CRP) and serum amyloid protein (SAP) were measured using MILLIPLEX MAP Human Cardiovascular Disease Magnetic Bead Panel 3 (HCVD3MAG‐67K; Merck‐Millipore). Matrix metalloproteinase (MMP)‐1, MMP‐2, and MMP‐9 were measured using MILLIPLEX MAP Human MMP Magnetic Bead Panel 2 (HMMP2MAG‐55K; Merck‐Millipore). Tissue inhibitor of metalloproteinase (TIMP)‐1 and TIMP‐2 were measured using MILLIPLEX MAP Human TIMP Magnetic Bead Panel 1 (HTMP1MAG‐54K; Merck‐Millipore. All kits were run on the Luminex 200 platform. Assays were conducted following the manufacturer's protocol.

ACE1 activity was measured in CSF and serum using an ACE1‐specific fluorescence resonance energy transfer (FRET) peptide substrate (Abz‐FRK[Dnp]‐P; Enzo Life Sciences) ³⁷ and ACE2 activity was measured using the ACE2‐specific FRET substrate ([Mca‐APK][Dnp]; Enzo Life Sciences), ³⁸ as previously described.

APOE genotypes were used as a potential covariate and were determined by real‐time polymerase chain reaction using TaqMan SNP Genotyping Assays (Applied Biosystems Inc.) unique for each APOE single nucleotide polymorphism, rs429358 (Assay ID C 3084793 20) and rs7412 (Assay ID C 904973 10), according to manufacturer's protocol. Participants were not recruited based on APOE genotype.

2.3. Statistical analyses

Data normality of continuous variables was assessed by histogram and Shapiro–Wilk test. Demographic variables were summarized using descriptive statistics. Paired t test for normally distributed data and paired Wilcoxon signed rank test for non‐normally distributed data were used to compare change from baseline to year 2. Differences in categorical variables were analyzed by a chi‐square test. Differences for continuous variables between B/AAs and NHWs were compared using a two‐sample t test for normally distributed data and Mann‐Whitney U test for variables with non‐normal distribution. Effect sizes (ES) were calculated using Cohen d. Multiple linear regressions were used to investigate associations between ACE1 and ACE2 activity, CSF sPDGFRβ level, AD biomarkers, and markers of inflammation, with Bonferroni‐corrected P‐values. To control for confounding factors, age, sex, race, and comorbidities were included in all regression models. Associations between blood and CSF biomarkers were examined using Pearson correlations for normally distributed variables and Spearman correlations for variables with non‐normal distributions. Statistical analyses were performed by SAS version 9.4 (SAS Institute), with an alpha set at 0.05.

3. RESULTS

3.1. Demographics

Baseline demographic and clinical data were previously described. ³² In brief, the mean age of participants was 59 ± 7 years, 38% identified as B/AA (n = 30), and 66% were female (n = 53). Medical history included high cholesterol (57%), high blood pressure (41%), and diabetes (2%). While 31% of participants reported having no history of the aforementioned vascular risk factors, 40% reported having one risk factor, 26% reported two risk factors, and 2% reported a history of all three (i.e., high cholesterol, hypertension, and diabetes). A higher proportion of B/AAs compared to NHWs had a history of hypertension (57% vs. 34%, respectively; P = 0.047). Forty‐seven percent (n = 38) were APOE ε4 positive (n = 9 with two alleles), with no differences by race (P = 0.883). MoCA scores ranged from 21 to 30, with 31% of participants scoring <26, suggestive of MCI.

3.2. Disease‐related changes in CSF AD biomarkers over time and according to race

We previously presented cross‐sectional baseline measurements of CSF Aβ and tau in these participants. ³² Here, we present changes in CSF AD biomarkers over a subsequent 2‐year interval from baseline and in relation to race (Table 1); Figure 1 summarizes the major findings. Cut‐offs for normal values are based on those established by Gobom et al. ³⁶ as delineated in the Methods section.

TABLE 1.

CSF biomarkers by race over 2 years.

	Total (N = 81)	Black/African American (n = 21)	Non‐Hispanic White (n = 46)	P‐value *	Time
AD biomarkers
Aβ1‐40 (pg/mL)	9488.38 ± 2842.4 10606.10 ± 3125 †	8400.38 ± 2493.4 8546.94 ± 2305.2	9970.02 ± 2851.0 11372.30 ± 3061.3	0.0336 0.0014	Baseline Year 2
Aβ1‐42 (pg/mL)	557.60 ± 242.7 517.44 ± 257.3	684.19 ± 200.8 656.13 ± 239.6	703.11 ± 261.3 744.05 ± 256.5	0.20 0.24	Baseline Year 2
Aβ42/40	0.07 (0.05–0.07) 0.06 (0.05–0.06) †	0.09 (0.08–0.09) 0.08 (0.07–0.09) †	0.08 (0.07–0.08) 0.07 (0.06–0.07) †	0.09 0.12	Baseline Year 2
p‐tau (pg/mL)	31.90 (32.3–43.9) 33.30 (35.3–49.9) †	25.00 (22.6–32.9) 26.90 (21.46–35.7) †	35.50 (33.5–45.7) 36.00 (36.90–53.3) †	0.0103 0.0046	Baseline Year 2
t‐tau (pg/mL)	266.00 (264.5–354.8) 259.00 (265.3–363.9)	186.00 (165.7–244.1) 174.50 (134.8–294.7)	276.0 (278.8–374.1) 263.00 (276.8–379.9)	0.0012 0.0021	Baseline Year 2
Vascular markers
sPDGFRβ (pg/mL)	474.57 ± 157.9 462.14 ± 133.8	375.4 ± 151.6 378.9 ± 92.0	499.2 ± 166.4 480.6 ± 131.6	0.0055 0.0063	Baseline Year 2
ICAM‐1 (pg/mL)	296.87 (295.7–398.4) 302.94 (289.5–412.6)	340.58 (268.45–463.57) 276.39 (227.2–424.0)	273.65 (278.3–389.78) 306.30 (291.2–429.8)	0.55 0.69	Baseline Year 2
VCAM‐1 (ng/mL)	25.27 (24.3–31.5) 27.10 (25.1–31.8)	20.40 (16.1–23.9) 18.50 (12.0–41.0)	27.16 (25.0–32.5) 28.38 (26.7–33.9)	0.0040 0.0094	Baseline Year 2
ACE1 (nmol/min)	5562.74 ± 1683.8 5678.58 ± 1790.8	5325.52 ± 1053.4 5349.7 ± 1344.0	5753.34 ± 1768.8 5806.8 ± 1859.1	0.46 0.40	Baseline Year 2
ACE2 (nmol/min)	2172.28 ± 481.3 2268.09 ± 452.8	2245.69 ± 466.41 2291.2 ± 381.3	2203.82 ± 447.83 2277.5 ± 464.5	0.79 0.92	Baseline Year 2
Cytokines and chemokines
IL‐10 (pg/mL)	7.66 ± 2.6 6.52 ± 1.9	7.16 ± 3.1 7.66 ± 5.2	7.11 ± 2.3 5.73 ± 2.0	0.79 0.04	Baseline Year 2
IL‐9 (pg/mL)	4.04 ± 1.3 3.44 ± 1.5 †	3.57 ± 1.4 3.12 ± 1.7	3.94 ± 1.6 3.69 ± 1.9	0.40 0.33	Baseline Year 2
IL‐7 (pg/mL)	3.12 ± 0.6 3.02 ± 1.5	2.59 ± 0.4 2.28 ± 1.2	3.04 ± 0.6 3.22 ± 2.2	0.011 0.17	Baseline Year 2
IL‐8 (pg/mL)	76.71 ± 20.4 57.97 ± 12.2 †	78.67 ± 14.1 70.53 ± 32.3	76.23 ± 20.6 53.52 ± 11.3	0.63 0.004	Baseline Year 2
TNFα (pg/mL)	2.90 ± 1.1 2.77 ± 0.6	2.98 ± 1.0 2.38 ± 0.9	2.93 ± 0.8 2.79 ± 1.0	0.86 0.16	Baseline Year 2
MCP‐1 (pg/mL)	6024.32 ± 556.7 5290.64 ± 379.0 †	6105.8 ± 659.1 5365.2 ± 333.4	5793.1 ± 756.2 5185.3 ± 440.6	0.11 0.14	Baseline Year 2
TGFα (pg/mL)	10.05 ± 1.5 –	10.11 ± 1.9 –	10.04 ± 2.16 –	0.90 –	Baseline Year 2
MDC (pg/mL)	127.52 ± 64.0 –	124.6 ± 75.0 –	50.82 ± 7.5 –	0.5 –	Baseline Year 2
Fractalkine (pg/mL)	– 135.58 ± 19.2	– 125.5 ± 23.9	– 132.6 ± 19.0	– 0.24	Baseline Year 2
MMPs and TIMPs
MMP‐1 (pg/mL)	9.35 ± 6.7 8.38 ± 5.7 †	8.52 ± 6.4 8.57 ± 6.0	9.24 ± 6.1 8.48 ± 5.6	0.66 0.95	Baseline Year 2
MMP‐2 (ng/mL)	17.35 ± 4.0 16.48 ± 4.0	15.22 ± 6.1 15.21 ± 6.2	18.0.9 ± 5.7 16.53 ± 2.8	0.07 0.26	Baseline Year 2
MMP‐9 (pg/mL)	17.72 ± 11.9 12.63 ± 8.3 †	20.57 ± 17.0 12.71 ± 6.1	15.96 ± 6.6 12.90 ± 8.8	0.11 0.94	Baseline Year 2
TIMP‐1 (ng/mL)	36.54 ± 7.4 38.14 ± 12.0	36.63 ± 9.1 38.39 ± 20.8	36.75 ± 6.9 36.96 ± 6.0	0.95 0.68	Baseline Year 2
TIMP‐2 (ng/mL)	– 40.08 ± 8.8	– 37.91 ± 14.4	– 39.78 ± 5.8	– 0.48	Baseline Year 2
TNF receptor superfamily
TNFR1 (pg/mL)	– 519.89 ± 158.0	– 464.4 ± 2007.5	– 541.7 ± 131.5	– 0.08	Baseline Year 2
TNFR2 (pg/mL)	– 789.26 ± 232.1	– 748.3 ± 299.7	– 804.5 ± 203.5	– 0.41	Baseline Year 2

Note: Results are reported as mean ± standard deviation or median (95% CI). Two‐sample t tests for normally distributed data and paired Wilcoxon signed rank test for non‐normally distributed data were used to test differences by race at each time point. Paired t tests for normally distributed data and paired Wilcoxon signed rank test for non‐normally distributed data were used to test differences by time and within each race by time.

Abbreviations: Aβ, amyloid beta; ACE, angiotensin converting enzyme; AD, Alzheimer's disease; CI, confidence interval; CSF, cerebrospinal fluid; ICAM, intercellular adhesion molecule; IL, interleukin; MCP, monocyte chemoattractant protein; MDC, macrophage‐derived chemokine; MMP, matrix metalloproteinase; p‐tau, phosphorylated tau; sPDGFRβ, soluble platelet‐derived growth factor receptor β; TGF, tumor growth factor; TIMP, tissue inhibitor matrix metalloproteinase; TNF, tumor necrosis factor; TNFR, tumor necrosis factor receptor; t‐tau, total tau; VCAM, vascular cell adhesion molecule.

^* P‐values represent differences by race at each time point.

^† P < 0.05 versus baseline.

All bolded values are < 0.05

FIGURE 1.

In a middle‐aged diverse cohort of persons with a parental history of AD, there were significant changes in AD CSF biomarkers over time, including increases in p‐tau and Aβ1‐4₀. CSF AD biomarkers (p‐tau and Aβ1‐40) were lower among Black and African American participants compared to non‐Hispanic Whites. The CSF vascular markers sPDGFRβ and VCAM‐1 were also lower among Black and African American participants compared to non‐Hispanic Whites. Aβ, amyloid beta; AD, Alzheimer's disease; CRP, C‐reactive protein; CSF, cerebrospinal fluid; IL, interleukin; MCP, monocyte chemoattractant protein; MMP, matrix metalloproteinase; p‐tau, phosphorylated tau; SAP, serum amyloid protein; sPDGFRβ, soluble platelet‐derived growth factor beta; TNF, tumor necrosis factor; VCAM‐1, vascular cell adhesion molecule‐1.

CSF levels of p‐tau increased from baseline to year 2, with moderate effect sizes for the cohort overall (ES = 0.600; P < 0.001), B/AAs (ES = 0.722; P = 0.022), and NHWs (ES = 0.626; P < 0.001). No differences over time for t‐tau were found. As previously reported, ³² baseline CSF levels of p‐tau (ES = 0.650) and t‐tau (ES = 0.853) were lower among B/AA participants than NHW, and these differences remained at year 2 (ES = 0.693 and 0.697, respectively). CSF levels of Aβ_1‐40 increased from baseline to year 2 (ES = 0.777; P < 0.001). As previously reported, ³² baseline CSF levels of Aβ1‐40 were lower among B/AA participants than NHW (ES = 0.574) and these differences remained at year 2 (ES = 0.980). No differences by race were found in levels of Aβ1‐42. CSF levels of Aβ42/40 were significantly lower at year 2 compared to baseline for the cohort overall (ES = 0.587; P < 0.001), and for B/AAs (ES = 1.184; P = 0.002) and NHWs (ES = 0.722; P < 0.001).

Participants who were APOE ε4 positive (i.e., having at least one APOE ε4 allele) had lower Aβ1‐42 (600 vs. 798 pg/mL; P = 0.001) and Aβ_42/40 (0.068 vs. 0.084; P = 0.0008) compared to those without the ε4 allele. CSF levels of p‐tau and t‐tau did not differ by APOE ε4 allele when analyzed across the cohort or within groups by race. No associations between AD biomarkers and age or sex were found.

Median p‐tau and t‐tau levels were within normal range (<50.2 pg/mL and <409 pg/mL, respectively). However, p‐tau levels were higher than AD cut‐off limits in 12% of participants at baseline and 17% at year 2, while t‐tau levels were higher than AD cut‐off limits in 26% of participants at baseline and 27% at year 2.

While mean Aβ1‐42 was >526 pg/mL at baseline, the cut‐off for AD association ³⁶ adopted for this study, the number of participants with values lower than the cut‐off limits at baseline (35%) increased to 46% at year 2 (P < 0.001). Mean Aβ42/40 was 0.08 (95% confidence interval [CI] 0.07–0.08) at baseline and 0.07 (95% CI 0.06–0.07) at year 2 (ES = 0.60; P < 0.0001). A higher proportion of participants had a lower (< 0.072) Aβ42/40 at baseline compared to their year 2 measurements (76% vs. 93%; P = 0.013). Ten percent of participants had abnormal (i.e., resembling AD‐associated) levels of all three AT(N) AD biomarkers at both baseline and year 2.

3.3. CSF markers of vascular injury differed according to race and were related to markers of neuroinflammation and AD pathology

Mean levels of CSF markers of vascular injury and neuroinflammation at baseline and year 2 are listed in Table 1. There were no differences over time in vascular markers. When stratified according to race, CSF levels of sPDGFRβ were lower among B/AA participants compared to NHW at both baseline (ES = 0.720; P = 0.01) and year 2 (ES = 0.831; P = 0.006), with large effect sizes. In addition, CSF levels of VCAM‐1 were lower among B/AA participants compared to NHW at both baseline (ES = 0.734; P = 0.008) and year 2 (ES = 0.970; P = 0.002) with large effect sizes. No associations between the APOE ε4 allele and CSF levels of vascular markers were found.

CSF levels of the inflammatory cytokines IL‐9, IL‐8, and MCP‐1 were lower at year 2 compared to baseline (ES = 0.575, 0.648, and 0.973, respectively; P < 0.001). CSF IL‐10 (ES = 0.611; P = 0.04) and IL‐8 (ES = 0.885; P = 0.004) were significantly higher among B/AAs compared to NHWs at year 2. The matrix metalloproteinases MMP‐1 and MMP‐9 were significantly lower than baseline at year 2 (ES = 0.336 and 0.393, respectively; P ≤ 0.018). No other differences by race were found. At both baseline and year 2, individuals with at least one APOE ε4 allele had significantly lower levels of MMP‐2 (ES = 0.677 and 0.470 at year 2; P = 0.009) and TIMP‐1 (ES = 0.631 and 0.373; P ≤ 0.05) and higher levels of MMP‐9 (ES = 0.470 and 0.287; P ≤ 0.05).

CSF sPDGFRβ level, a marker of pericyte injury, was positively associated with CSF p‐tau (β = 0.380; P = 0.003) and t‐tau (β = 0.466; P < 0.001) at baseline, and at year 2 follow‐up (β = 0.356; P = 0.008 and β = 0.485; P < 0.001, respectively) after controlling for age, sex, and race (Figure 2A–D). CSF sPDGFRβ was also positively related to CSF levels of Aβ1‐40 at baseline (β = 0.439; P = 0.001) and year 2 (β = 0.454; P = 0.001), after controlling for age, sex, and race (Figure 2E–F). No associations were observed between CSF sPDGFRβ and Aβ1‐42 at baseline or year 2 (Figure 2G–H). Lower than normal levels of Aβ1‐42 (i.e., <526 pg/mL) were associated with higher changes in sPDGFRβ (mean difference 62.3 pg/mL; P = 0.041; ES = 0.596). CSF sPDGFRβ was higher among individuals with abnormal AT(N) biomarkers compared to those with at least one biomarker within normal range at year 2 (581.2 ± 182 vs. 443.7 ± 122 pg/mL; P = 0.039; ES = 0.710).

FIGURE 2.

CSF sPDGFRβ was associated with CSF AD biomarkers over time. At both baseline and year 2, CSF sPDGFRβ was positively associated with CSF phosphorylated tau (A)–(B), total tau (C)–(D), Aβ peptide (E)–(F), and Aβ1–42 (G)–(H). Multiple linear analyses controlling for age, sex, race, and education were used. Aβ, amyloid beta; AD, Alzheimer's disease; CSF, cerebrospinal fluid; p‐tau, phosphorylated tau; sPDGFRβ, soluble platelet‐derived growth factor beta; t‐tau, total tau.

CSF sPDGFRβ was positively associated with markers and mediators of vascular endothelial injury: VCAM‐1 (β = 0.51; P < 0.0001), ACE‐1 activity (β = 0.36; P = 0.0032), TIMP‐1 (β = 0.40; P = 0.001), and MMP‐2 (β = 0.46; P = 0.0001) at baseline (Figure 3A,C,E). These associations remained at year 2 (Figure 3B,D,F). CSF sPDGFRβ was positively associated with IL‐9 at baseline and year 2 (β = 0.260; P = 0.047 and β = 0.334; P = 0.016; Figure 3G–H). Additionally, CSF sPDGFRβ positively related to TIMP‐2 (β = 0.32; P = 0.015), TNFR‐1 (β = 0.41; P = 0.0013), and TNFR‐2 (β = 0.39; P = 0.0024) at year 2 (Figure 4A–C); these measures were not assessed at year 1.

FIGURE 3.

CSF sPDGFRβ was associated with CSF markers and mediators of vascular endothelial injury over time. At both baseline and year 2, CSF sPDGFRβ was positively associated with CSF ACE‐1 activity (A)–(B), and VCAM‐1 (C)–(D), IL‐9 (E)–(F), and TIMP‐1 (G)–(H). Multiple linear analyses controlling for age, sex, race, and education were used. ACE, angiotensin‐converting enzyme; CSF, cerebrospinal fluid; IL, interleukin; sPDGFRβ, soluble platelet‐derived growth factor beta; TIMP, tissue inhibitor of metalloproteinase; VCAM, vascular cell adhesion molecule.

FIGURE 4.

CSF sPDGFRβ was associated with CSF mediators of vascular endothelial injury at year 2. CSF sPDGFRβ was positively associated with CSF TIMP 2 (A), TNFR1 (B), and TNFR2 (C). These measures were not analyzed at baseline. Multiple linear analyses controlling for age, sex, race, and education were used. CSF, cerebrospinal fluid; IL, interleukin; sPDGFRβ, soluble platelet‐derived growth factor beta; TIMP, tissue inhibitor of metalloproteinase; TNFR, tumor necrosis factor receptor; VCAM‐1, vascular cell adhesion molecule‐1.

3.4. Circulating levels of vascular markers remain relatively stable

Plasma and serum vascular and inflammatory biomarkers are listed in Table 2. There were no differences in circulating vascular markers from baseline to year 2; however, VCAM‐1 was significantly higher at year 1 than baseline (ES = 0.300; P = 0.016). VCAM‐1 was significantly lower among B/AAs compared to NHWs at baseline (ES = 0.530) and year 1 (Cohen d = 0.283), with no differences at year 2.

TABLE 2.

Plasma/serum biomarkers by race over 2 years.

	Total (N = 81)	Black/African American (N = 21)	Non‐Hispanic White (N = 46)	P‐value *	Time
Vascular biomarkers
ICAM‐1 (ng/mL)	138.02 (125.6–168.8) 132.48 (122.2–150.8) 123.94 (114.5–138.5)	136.85 (129.6–148.8) 137.61 (111.4–151.0) 135.11 (113.8–151.2)	137.20 (122.7–190.0) 129.53 (119.8–158.2) 123.98 (112.7–136.5)	0.75 0.87 0.32	Baseline Year 1 Year 2
VCAM‐1 (ng/mL)	842.95 (760.2–953.8) 891.06 (955.8–1741.18) † 866.51 (852.4–1729.9)	807.55 (614.7–883.3) 746.59 (504.6–1581.6) 827.73 (0–3827.3)	863.46 (808.4–1074.2) 930.16 (963.3–2011.5) 896.74 (1028.6–2828.8)	0.02 0.03 0.32	Baseline Year 1 Year 2
ACE1 (nmol/min)	24121.97 ± 10953.8 25072.49 ± 13399.1 24599.05 ± 12979.2	23160.0 ± 11818.4 22151.4 ± 12306.5 24576.4 ± 12446.7	25636.0 ± 10422.7 25916.7 ± 13893.9 25259.9 ± 11153.1	0.36 0.59 0.82	Baseline Year 1 Year 2
ACE2 (nmol/min)	1501.74 ± 963.3 1511.97 ± 1044.8 1479.92 ± 1038.5	1273.6 ± 924.7 1223.8 ± 772.4 1416.2 ± 968.6	1554.8 ± 893.4 1629.6 ± 1127.5 1584.5 ± 1152.2	0.21 0.09 0.55	Baseline Year 1 Year 2
Cytokines and chemokines
IL‐10 (pg/mL)	11.30 ± 8.6 9.29 ± 4.5 † 9.37 ± 4.1 †	9.23 ± 5.0 8.75 ± 4.28 9.23 ± 5.1	12.09 ± 8.5 9.53 ± 4.7 † 9.75 ± 0.7 †	0.09 0.53 0.68	Baseline Year 1 Year 2
IL‐9 (pg/mL)	1.64 (0–38.8) 4.44 (4.7–11.6) 4.71 (4.8–10.6)	4.16 (0–72.2) 5.66 (3.1–10.3) 5.62 (2.1–10.5)	4.67 (0–65.0) 3.52 (3.84–13.8) 4.52 (4.3–11.8)	0.94 0.83 0.72	Baseline Year 1 Year 2
IL‐7 (pg/mL)	5.38 ± 2.2 4.64 ± 1.9 † 4.94 ± 2.0 ‡	5.98 ± 2.5 5.51 ± 2.5 5.19 ± 2.0	4.86 ± 1.6 4.28 ± 1.5 † 4.99 ± 2.3 ‡	0.03 0.02 0.71	Baseline Year 1 Year 2
IL‐8 (pg/mL)	5.36 (10.8–21.3) 5.10 (8.8–26.5) 4.59 (8.8–22.6)	4.28 (4.7–14.9) 4.15 (5.3–20.4) 3.51 (4.6–16.0)	7.03 (12.0–29.0) 5.29 (7.2–32.6) 4.94 (8.9–27.6)	0.08 0.42 0.19	Baseline Year 1 Year 2
TNFα (pg/mL)	6.70 ± 3.2 6.02 ± 3.0 † 5.91 ± 3.0 †	6.33 ± 1.7 5.54 ± 1.8 † 6.0 ± 1.1 †	6.68 ± 3.7 6.25 ± 3.34 6.1 ± 3.5	0.58 0.38 0.92	Baseline Year 1 Year 2
MCP1 (pg/mL)	198.00 ± 81.3 198.05 ± 88.3 171.94 ± 59.8 †	224.4 ± 84.4 203.26 ± 65.1 196.4 ± 49.8	176.8 ± 74.4 195.67 ± 97.6 165.4 ± 9.7 ‡	0.002 0.75 0.05	Baseline Year 1 Year 2
TGFα (pg/mL)	3.4 (2.9–13.0) – –	3.47 (0–27.3) – –	2.75 (2.44–9.96) – –	0.41 – –	Baseline Year 1 Year 2
IFNγ (pg/mL)	7.42 (0–1077.3) 6.74 (7.3–21.1) 7.15 (7.0–14.8)	7.99 (0–657.0) 6.16 (0.2–34.0) 5.92 (5.4–16.8)	9.61 (0–1523.1) 6.75 (6.0–19.6) 6.82 (5.5–16.9)	0.43 0.91 0.78	Baseline Year 1 Year 2
CRP (pg/mL)	7.16 (7.5–15.3) 6.93 (3.3–27.7) 5.43 (3.3–47.7)	13.17 (4.2–50.4) 10.95 (6.3–20.6) 10.49 (8.4–19.7)	4.94 (5.0–9.2) 5.18 (0–34.3) 4.77 (6.0–10.1)	0.002 0.084 0.03	Baseline Year 1 Year 2
SAP (pg/mL)	9.57 ± 3.5 8.64 ± 2.8 † 8.65 ± 3.2 †	10.09 ± 4.5 9.55 ± 3.0 9.99 ± 2.7	8.87 ± 2.5 8.23 ± 2.7 † 8.39 ± 3.5	0.04 0.03 0.04	Baseline Year 1 Year 2
MDC (pg/mL)	1163.42 ± 533.9 1149.00 ± 353.4 1171.80 ± 441.3	1373.3 ± 627.0 1173.98 ± 403.7 1281.7 ± 540.1	1061.8 ± 430.4 1137.64 ± 332.4 1100.8 ± 384.4	0.006 0.71 0.11	Baseline Year 1 Year 2
Fractalkine (pg/mL)	– 163.06 ± 58.9 170.43 ± 54.8	185.3 ± 25.5 183.09 ± 84.5 178.2 ± 69.7	155.9 ± 57.7 153.96 ± 40.7 167.2 ± 51.9	0.24 0.16 0.51	Baseline Year 1 Year 2
IL‐6 (pg/mL)	– 4.75 (10.7–35.3) 4.36 (11.3–31.8)	– 3.44 (4.3–27.4) 2.44 (0–27.4)	– 5.04 (8.8–44.2) 7.69 (10.6–40.8)	– 0.55 0.18	Baseline Year 1 Year 2
IL‐1α (pg/mL)	– 30.23 (21.3–60.3) 15.37 (17.9–55.0)	– 16.15 (6.8–57.5) 23.25 (0–111.5)	– 20.23 (17.5–72.2) 17.98 (18.7–49.8)	– 0.87 0.60	Baseline Year 1 Year 2
IL‐1β (pg/mL)	– 2.98 ± 1.4 2.73 ± 125	– 2.80 ± 1.1 2.55 ± 1.2	– 3.10 ± 1.5 2.95 ± 1.3	– 0.56 0.41	Baseline Year 1 Year 2
IL‐2 (pg/mL)	– 2.33 ± 1.7 2.72 ± 1.8	– 1.85 ± 0.6 2.61 ± 0.3	– 2.67 ± 2.2 3.38 ± 2.4	– 0.37 0.43	Baseline Year 1 Year 2
IL‐4 (pg/mL)	– 58.69 (161.5–514.3) 71.75 (172.0–495.9)	– 53.99 (52.3–464.3) 49.53 (0–363.0)	– 63.83 (130.7–647.5) 88.8 (163.4–621.4)	– 0.60 0.08	Baseline Year 1 Year 2
IP‐10 (pg/mL)	– 304.13 ± 307.0 266.83 ± 200.1	– 230.39 ± 104.1 242.3 ± 79.9	– 337.65 ± 360.1 254.6 ± 131.9 ‡	– 0.07 0.66	Baseline Year 1 Year 2
TNF receptor superfamily
TNFR1 (pg/mL)		– 128.81 ± 77.9 110.3 ± 57.8	– 126.89 ± 76.6 122.6 ± 61.7	– 0.93 0.48	Baseline Year 1 Year 2
TNFR2 (pg/mL)		– 922.93 ± 294.5 928.3 ± 270.8	– 927.92 ± 350.9 891.1 ± 210.5	– 0.96 0.57	Baseline Year 1 Year 2

Note: Results are reported as mean ± standard deviation or median (95% CI). Two‐sample t‐tests for normally distributed data and paired Wilcoxon signed rank test for non‐normally distributed data were used to test differences by race at each time point. Least squares mean analyses were used to test differences over time, differences by race at each time point.

Abbreviations: ACE, angiotensin‐converting enzyme; CI, confidence interval; CRP, C‐reactive protein; ICAM, intercellular adhesion molecule; IFN, interferon; IL, interleukin; IP, interferon gamma‐induced protein; MCP, monocyte chemoattractant protein; MDC, macrophage‐derived chemokine; MMP, matrix metalloproteinase; SAP, serum amyloid protein; TGF, tumor growth factor; TNF, tumor necrosis factor; TNFR, tumor necrosis factor receptor; VCAM, vascular cell adhesion molecule.

^* P‐values represent differences by race at each time point.

^† P < 0.05 versus baseline.

^‡ P < 0.05 versus year 1.

All bolded values are < 0.05

3.5. Circulating levels of markers of neuroinflammation decreased over time

IL‐10 was significantly lower than baseline at year 1 (ES = 0.309; P = 0.0043) and year 2 (ES = 0.325; P = 0.0058). No significant differences between year 1 and year 2 were found. Among NHWs, IL‐10 was significantly lower than baseline at year 1 (ES = 0.315; P = 0.0072) and year 2 (ES = 0.332; P = 0.0135), with no significant differences between year 1 and year 2. No differences over time among B/AAs were found.

IL‐7 was significantly lower at year 1 than baseline (ES = 0.359; P = 0.0037), and year 2 was significantly lower than year 1 (ES = 0.291; P = 0.0135). No significant differences between baseline and year 2 were found. Among NHWs, IL‐7 was significantly lower at year 1 compared to baseline (ES = 0.438; P = 0.029), and year 2 was significantly higher than year 1 (ES = 0.319; P = 0.022), with no differences between baseline and year 2. No differences over time among B/AAs were found. IL‐7 was significantly higher among B/AAs compared to NHWs at baseline (ES = 0.541) and year 1 (ES = 0.656), with no differences at year 2.

TNF‐α was significantly lower than baseline at year 1 (ES = 0.283; P = 0.020) and year 2 (ES = 0.320; P = 0.008). No differences between year 1 and year 2 were found. Among B/AAs, TNF‐α was significantly lower than baseline at year 1 (ES = 0.561; P = 0.011) and year 2 (ES = 0.556; P = 0.010), with no differences between years 1 and 2. No differences over time between NHWs were found.

MCP‐1 was significantly lower at year 2 compared to baseline (ES = 0.365; P = 0.0009) and year 1 (ES = 0.441; P = 0.0009). MCP‐1 was significantly lower at year 2 compared to year 1 among B/AAs (ES = 0.458; P = 0.005) and compared to baseline among NHWs (ES = 0.620; P = 0.010). MCP‐1 was significantly higher among B/AAs compared to NHWs at baseline (ES = 0.774) and year 2 (ES = 0.420).

SAP was significantly lower at year 1 (ES = 0.284; P = 0.0126) and year 2 (ES = 0.233; P = 0.0472) compared to baseline. Among NHWs, SAP was significantly lower than baseline at year 1 (ES = 0.366; P = 0.300), with no significant differences in year 2. No differences between year 1 and year 2 were found. SAP was significantly higher among B/AAs compared to NHWs at all three time points (ES = 0.421, 0.475, and 0.416, respectively).

There were no significant differences over time for CRP, a marker of general systemic inflammation. However, CRP was significantly elevated (2–3‐fold increase) among B/AA participants compared to NHW at both baseline and year 2, with medium to large effect sizes (ES = 0.757 and 0.442, respectively). IP‐10 was collected at year 1 and year 2 only. No differences over time were found among the cohort; however, IP‐10 was significantly lower at year 2 compared to year 1 among NHWs (ES = 0.290; P = 0.0272).

CRP was significantly lower among individuals with at least one copy of the APOE ε4 allele at baseline (3.91 vs. 9.24 pg/mL; ES = 0.405; P = 0.013), year 1 (4.75 vs. 8.62 pg/mL; ES = 0.127; P = 0.042), and year 2 (4.11 vs. 9.19 pg/mL; ES = 0.341; P = 0.017). SAP was significantly lower among individuals with at least one copy of the APOE ε4 allele at baseline (8.66 vs. 10.34 pg/mL; ES = 0.491; P = 0.038) and year 2 (7.65 vs. 9.34 pg/mL; ES = 0.552; P = 0.024).

CSF TNF‐α was positively associated with levels of plasma TNF‐α at baseline (r = 0.405; P = 0.021), with no association at year 2. CSF levels of VCAM‐1 were positively associated with plasma VCAM‐1 at both baseline (rho = 0.329; P = 0.029) and year 2 (rho = 0.330; P = 0.035). No associations between CSF and blood levels of other variables were found. 

4. DISCUSSION

Here we report longitudinal data from a 2‐year follow‐up period of previously reported baseline results from ASCEND in cognitively unimpaired or mildly impaired B/AA and NHW middle‐aged adults with a parental history of AD. ³² Our data demonstrate changes in established AD biomarkers with moderate to large effect sizes, consistent with early disease manifestations. We also highlight significant race‐related differences in disease‐related and vascular markers within the cohort that require further investigation.

In this cohort of at‐risk middle‐aged adults, we found changes consistent with increased risk in CSF AD biomarkers between baseline and year 2. While median CSF t‐tau remained within normal limits over the 2‐year period, there was a significant increase in p‐tau from baseline to year 2. CSF Aβ42/40 decreased and Aβ1‐40 increased over 2 years. Although Aβ1‐40 is less frequently discussed as a biomarker of AD, several studies have found higher CSF Aβ1‐40 in MCI, prodromal AD, and AD; ³⁹ , ⁴⁰ , ⁴¹ , ⁴² however, other studies found no differences with progression to AD. ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ As this cohort is middle aged with low to moderate vascular risk factors, further work is needed to better understand the contribution of Aβ1‐40 to AD risk.

Half of participants at baseline and three fifths of participants at year 2 had below‐normal Aβ1‐42 levels, indicative of a worsening AD profile. Lower CSF Aβ1‐42 is considered one of the earliest indicators of AD neuropathologic change. ⁴⁷ , ⁴⁸ As these findings suggest the higher AD risk inferred by having a first‐degree relative with AD may be associated with higher amyloid burden, a more comprehensive understanding of the interplay of family history, genetics, and vascular risk factors among high‐risk populations is urgently needed. Ten percent of ASCEND participants met the definition for AD with respect to AT(N) biomarker levels. ² Together, these data may suggest AD pathological changes in this middle‐aged, at‐risk cohort, although this cohort is unlikely to have significant evidence of disease. While this cohort has an increased dementia risk due to parental history, the development of dementia is not inevitable. Approximately half of participants in this study were APOE ε4 positive, consistent with other cohorts of adult children of a parent with an AD diagnosis. However, it is important to note we cannot rule out the possibility of parental history of vascular or mixed dementia. We are continuing to follow these individuals over time to better understand the progression of AD biomarkers in high‐risk, diverse cohorts.

No changes in CSF vascular markers over 2 years were found. There were decreases in CSF and plasma markers of inflammation and CSF MMPs from baseline to year 2, which was unexpected considering increasing age and vascular comorbidities are associated with higher vascular dysfunction and inflammation. Changes in these markers over time were not associated with age or comorbidities. Decreases in inflammation could be a product of participation in a research study, as participants had access to education events, which included information on the role of inflammation and risk reduction.

Our baseline findings were the first to report lower levels of t‐tau and p‐tau based on race in middle‐aged adults with an increased risk for AD. ³² We now demonstrate tau levels remained lower among B/AA compared to NHW over a 2‐year period, consistent with recent analyses of racial differences in AD biomarkers in a population with cognitive impairment. ³¹ , ⁴⁹ We also found that CSF markers of vascular dysfunction were lower in B/AA compared to NHW at baseline and year 2. This supports a recent study in which cardiovascular comorbidities were not associated with CSF tau levels in B/AA participants. ⁴⁹ No association with APOE ε4 was found in this study; this might be explained by the cognitively unimpaired or MCI composition of this cohort. ⁵⁰

If there are differences in AD biomarkers by race, it will be increasingly important to understand appropriate cut‐offs for diagnosis and treatments. As potential treatments under development target tau and Aβ, it is important to establish whether certain treatments may be less effective for some communities. Within the research context, there is emerging evidence for differences by race in CSF biomarkers. However, there is insufficient biological data, especially CSF, in racial and ethnic minorities. ⁵¹ More work is needed globally to better understand these emerging differences along with drivers of AD pathophysiologic processes in diverse populations.

Circulating inflammatory biomarkers were higher among B/AAs compared to NHWs, aligning with other research finding higher systemic inflammation among B/AAs compared to NHWs. ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ Higher chronic inflammation among B/AAs has been linked to effects of systemic racism and discrimination, likely due to increased activation of cortisol pathways ⁵⁶ , ⁵⁸ , ⁵⁹ and is thought to be a risk factor for cognitive changes and performance on cognitive function assessments in B/AA adults. ⁵³ , ⁶⁰ However, few studies consider the potential effects of predisposing psychosocial factors, such as stress and discrimination, on the relationship between inflammation and cognitive function. Further work is needed to understand the role of peripheral and central processes in AD risk among diverse populations.

In this cohort of middle‐aged adults with a parental history of AD, 36% of participants had one APOE ε4 allele and 11% were homozygous, higher than the general population (9%–23% and 2%–3%, respectively). ⁶¹ The frequency of APOE ε4 genotype is generally higher among B/AAs (19%) compared to NHWs (14%), ⁶¹ although there were no differences in this study. CSF Aβ42 and Aβ42/40 were significantly lower in persons with at least one APOE ε4 allele, which may relate to additional AD risk in persons with a parent with AD. While APOE ε4 genotype confers an increased genetic risk factor, not all persons who are ε4 positive develop AD, even with a family history. Considering other known risk factors, many of which are vascular, other possible outcomes include the development of vascular dementia or a mixed phenotype.

We also report associations between CSF markers of vascular injury and neuroinflammation in relation to markers of established disease pathology, indicative of early disease changes. CSF sPDGFRβ was positively associated with markers and mediators of vascular injury and brain cytokines and correlated with disease pathology. Interestingly, CSF sPDGFRβ, ICAM‐1, and ACE‐1 were lower in B/AA compared to NHW for reasons that are not yet clear. These findings are surprising considering the higher proportion of B/AAs with a history of hypertension compared to NHWs in this cohort. These data suggest vascular injury and neuroinflammation are related to early changes in AD pathology in a cognitively unimpaired or mildly impaired middle‐aged at‐risk cohort.

Markers of vascular injury, particularly sPDGFRβ, were strongly associated with levels of p‐tau, t‐tau, and Aβ1‐40, but surprisingly not Aβ1‐42. However, persons with low Aβ1‐42 had a larger increase in sPDGFRβ compared to individuals with normal Aβ1‐42, supporting a previous study showing CSF sPDGFRβ level correlated with t‐tau and p‐tau in AD. ²⁵ These data suggest pericyte dysfunction occurs early in the disease process, but further studies are required to confirm whether CSF sPDGFRβ changes are related to, or independent from, changes in CSF tau and Aβ.

Our data indicate sPDGFRβ is associated with markers of vascular endothelial activation, mediators of endothelial dysfunction, and IL‐9. These findings support previous observations linking vascular dysfunction and central nervous system inflammation in the early pre‐symptomatic phases of dementia. In a study of persons with MCI and AD, raised CSF levels of vascular markers and cytokines were associated with t‐tau and p‐tau, cortical thinning, and poorer Mini‐Mental State Examination scores; the associations were stronger in Aβ‐positive individuals. ¹¹ Pericyte loss and BBB breakdown have been associated with elevated levels of brain cytokines in early stages of AD, ⁶² and serum markers of vascular injury and inflammation increased over time with a decline in MCI compared to those that remained MCI‐stable. ⁶³

Our findings identified mediators of vascular homoeostasis, including MMP‐2 and TIMP‐1 and TIMP‐2 were positively correlated with sPDGFRβ, supporting other work suggesting TNFR dysregulation may be associated with tau pathology and may play a neurodegenerative role in AD pathology. ⁶⁴ , ⁶⁵ It is important to note that in a sample size of 81, we are unable to rule out other factors that may mediate relationships between sPDGFRβ and other CSF markers of AD pathology. Further work in larger studies with diverse populations is needed to better understand the role of sPDGFRβ in early AD pathology.

In conclusion, in a cohort of cognitively unimpaired or mildly impaired middle‐aged adults at risk of developing AD, we found evidence of changes in established CSF biomarkers at 2‐year follow‐up consistent with early changes in the development of AD (CSF t‐tau/p‐tau and Aβ40). CSF markers of vascular dysfunction, including pericyte damage and endothelial injury, were associated with changes in MMP‐2 and TIMPs, and cytokines. Elevated sPDGFRβ was related to t‐tau, p‐tau, and changes in Aβ40 but not Aβ42. Interstitial fluid tau, sPDGFRβ, ACE‐1, and ICAM‐1 were lower in B/AA compared to NHW, which is worth noting with respect to what are perceived normative ranges for patients of different racial backgrounds. These data indicate that cerebral vascular dysregulation may be a very early event that occurs in the development of AD, detectable in a mid‐life cognitively unimpaired or mildly impaired cohort, and that race impacts these relationships, warranting further research.

CONFLICT OF INTEREST STATEMENT

H.Z. has served on scientific advisory boards and/or as a consultant for Abbvie, Acumen, Alector, ALZPath, Annexon, Apellis, Artery Therapeutics, AZTherapies, CogRx, Denali, Eisai, Nervgen, Novo Nordisk, Passage Bio, Pinteon Therapeutics, Red Abbey Labs, reMYND, Roche, Samumed, Siemens Healthineers, Triplet Therapeutics, and Wave; has given lectures in symposia sponsored by Cellectricon, Fujirebio, Alzecure, Biogen, and Roche; and is a co‐founder of Brain Biomarker Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incubator Program (outside submitted work). B.B., H.H., W.T.H., P.G.K., J.S.M., D.D.V., L.Z., and W.W. have nothing to disclose. Author disclosures are available in the supporting information.

CONSENT STATEMENT

Each subject was informed of the testing protocols and potential risks and benefits of participation. All participants provided written informed consent before participation.

Supporting information

Supporting Information

Butts B, Huang H, Hu W, et al. sPDGFRβ and neuroinflammation are associated with AD biomarkers and differ by race: The ASCEND Study. Alzheimer's Dement. 2024;20:1175–1189. 10.1002/alz.13457