Associations of CSF PDGFRβ With Aging, Blood-Brain Barrier Damage, Neuroinflammation, and Alzheimer Disease Pathologic Changes

Abstract

Background and objectives: Injured pericytes in the neurovascular unit release platelet-derived growth factor β (PDGFRβ) into the CSF. However, it is not clear how pericyte injury contributes to Alzheimer disease (AD)-related changes and blood-brain barrier (BBB) damage. We aimed to test whether CSF PDGFRβ was associated with different AD-associated and age-associated pathologic changes leading to dementia.

Methods: PDGFRβ was measured in the CSF of 771 participants with cognitively unimpaired (CU, n = 408), mild cognitive impairment (MCI, n = 175), and dementia (n = 188) from the Swedish BioFINDER-2 cohort. We then checked association with β-amyloid (Aβ)-PET and tau-PET standardized uptake value ratio, APOE ε4 genotype and MRI measurements of cortical thickness, white matter lesions (WMLs), and cerebral blood flow. We also analyzed the role of CSF PDGFRβ in the relationship between aging, BBB dysfunction (measured by CSF/plasma albumin ratio, QAlb), and neuroinflammation (i.e., CSF levels of YKL-40 and glial fibrillary acidic protein [GFAP], preferentially expressed in reactive astrocytes).

Results: The cohort had a mean age of 67 years (CU = 62.8, MCI = 69.9, dementia = 70.4), and 50.1% were male (CU = 46.6%, MCI = 53.7%, dementia = 54.3%). Higher CSF PDGFRβ concentrations were related to higher age (b = 19.1, β = 0.5, 95% CI 16-22.2, p < 0.001), increased CSF neuroinflammatory markers of glial activation YKL-40 (b = 3.4, β = 0.5, 95% CI 2.8-3.9, p < 0.001), GFAP (b = 27.4, β = 0.4, 95% CI 20.9-33.9, p < 0.001), and worse BBB integrity measured by QAlb (b = 37.4, β = 0.2, 95% CI 24.9-49.9, p < 0.001). Age was also associated with worse BBB integrity, and this was partly mediated by PDGFRβ and neuroinflammatory markers (16%-33% of total effect). However, PDGFRβ showed no associations with APOE ε4 genotype, PET imaging of Aβ and tau pathology, or MRI measures of brain atrophy and WMLs (p > 0.05).

Discussion: In summary, pericyte damage, reflected by CSF PDGFRβ, may be involved in age-related BBB disruption together with neuroinflammation, but is not related to Alzheimer-related pathologic changes.

Figure 1. Scatter-Dot Plot Representing the Correlation Between CSF PDGFRβ and Age in the Whole Sample (n = 771)

CU participants, participants with MCI, and participants with dementia shown in blue, green, and red, respectively. Regression line with 95% CIs is not adjusted for covariates. CU = cognitively unimpaired; MCI = mild cognitive impairment; PDGFRβ = platelet-derived growth factor β.

Figure 2. CSF PDGFRβ and AD Imaging Measures

Scatter-dot plot representing the correlation between CSF PDGFRβ and Aβ-PET SUVR in the neocortical meta-ROI (A), tau-PET SUVR in the temporal meta-ROI (B), weighted cortical thickness in the AD signature meta-ROI (entorhinal, fusiform, inferior temporal, and middle temporal) (C), and volume of white matter lesions (D) in the whole sample. CU participants, participants with MCI, and participants with dementia shown in blue, green, and red, respectively. According to the study protocol, Aβ-PET was not performed in participants with dementia. Regression lines with 95% CIs are not adjusted for covariates. Aβ = β-amyloid; AD = Alzheimer disease; CU = cognitively unimpaired; MCI = mild cognitive impairment; PDGFRβ = platelet-derived growth factor β; ROI = region of interest; SUVR = standardized uptake value ratio.

Figure 3. Scatter-Dot Plot Representing the Correlation Between CSF PDGFRβ and the CSF/Plasma Albumin Ratio (QAlb, A), YKL-40 (B), and GFAP (C)

CU participants, participants with MCI, and participants with dementia shown in blue, green, and red, respectively. Regression lines with 95% CIs are not adjusted for covariates. CU = cognitively unimpaired; GFAP = glial fibrillary acidic protein; MCI = mild cognitive impairment; PDGFRβ = platelet-derived growth factor β.

Figure 4. YKL-40, GFAP, and PDGFRβ as Mediators of the Effect of Age on BBB Damage

Sequential mediation analysis for neuroinflammation markers (YKL-40, A; GFAP, B) and PDGFRβ (A, B) as mediators of the relationship between age (X) and CSF/plasma albumin ratio (QAlb, Y). a₁: effect of X on M1; a₂: effect of X on M2 adjusted for M1; b₁: effect of M1 on Y adjusted for M2 and X; b₂: effect of M2 on Y adjusted for M1 and X; c': direct effect of X on Y; c: total effect of X on Y; d: effect of M1 on M2 adjusted for X. Blue arrow: indirect effect for model X→M1→M2→Y; red arrow: indirect effect for model X→M1→Y; green arrow: indirect effect for model X→M2→Y. Indirect effect (coefficient indicated with b) was considered significant if the 95% CI did not include 0 (shown in bold). Size of the indirect effect on the total effect shown as %. GFAP = glial fibrillary acidic protein; PDGFRβ = platelet-derived growth factor β.