Markers of neuroinflammation associated with Alzheimer's disease pathology in older adults

Background In vitro and animal studies have linked neuroinflammation to Alzheimer's disease (AD) pathology. Studies on markers of inflammation in subjects with mild cognitive impairment or AD dementia provided inconsistent results. We hypothesized that distinct blood and cerebrospinal fluid (CSF) inflammatory markers are associated with biomarkers of amyloid and tau pathology in older adults without cognitive impairment or with beginning cognitive decline. Objective To identify blood-based and CSF neuroinflammation marker signatures associated with AD pathology (i.e. an AD CSF biomarker profile) and to investigate associations of inflammation markers with CSF biomarkers of amyloid, tau pathology, and neuronal injury. Design/methods Cross-sectional analysis was performed on data from 120 older community-dwelling adults with normal cognition (n = 48) or with cognitive impairment (n = 72). CSF Aβ1–42, tau and p-tau181, and a panel of 37 neuroinflammatory markers in both CSF and serum were quantified. Least absolute shrinkage and selection operator (LASSO) regression was applied to determine a reference model that best predicts an AD CSF biomarker profile defined a priori as p-tau181/Aβ1–42 ratio >0.0779. It was then compared to a second model that included the inflammatory markers from either serum or CSF. In addition, the correlations between inflammatory markers and CSF Aβ1–42, tau and p-tau181 levels were assessed. Results Forty-two subjects met criteria for having an AD CSF biomarker profile. The best predictive models included 8 serum or 3 CSF neuroinflammatory markers related to cytokine mediated inflammation, vascular injury, and angiogenesis. Both models improved the accuracy to predict an AD biomarker profile when compared to the reference model. In analyses separately performed in the subgroup of participants with cognitive impairment, adding the serum or the CSF neuroinflammation markers also improved the accuracy of the diagnosis of AD pathology. None of the inflammatory markers correlated with the CSF Aβ1–42 levels. Six CSF markers (IL-15, MCP-1, VEGFR-1, sICAM1, sVCAM-1, and VEGF-D) correlated with the CSF tau and p-tau181 levels, and these associations remained significant after controlling for age, sex, cognitive impairment, and APOEε4 status. Conclusions The identified serum and CSF neuroinflammation biomarker signatures improve the accuracy of classification for AD pathology in older adults. Our results suggest that inflammation, vascular injury, and angiogenesis as reflected by CSF markers are closely related to cerebral tau pathology.