Inflammation blood profiles across neurodegenerative diseases
Maura Malpetti, Peter Swann, Leonidas Chouliaras, Nicholas J. Ashton, Kamen A Tsvetanov, Duncan Street, David J Whiteside, Thomas E Cope, George Savulich, Timothy Rittman, Ajenthan Surendranathan, Alexander G Murley, Katherine Stockton, W Richard Bevan‐Jones, Maria A Prats‐Sedano, Li Su, John T O'Brien, James B. Rowe- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Brain inflammation is an important pathogenic mechanism in many dementias, occurring early in the disease and being predictive of clinical decline. However, data on blood markers of inflammation across different dementia subtypes are limited. Here we assess inflammatory patterns of serum cytokines from patients with Alzheimer’s disease (AD), Lewy‐body dementia (DLB), frontotemporal dementia (FTD), progressive supranuclear palsy (PSP), corticobasal syndrome (CBS) and motor neurone disease (MND).
Method
Blood samples were obtained from 422 participants (29 controls, 46 patients with AD, 35 with mild cognitive impairment (MCI), 38 with DLB, 52 with behavioural variant FTD, 51 with primary progressive aphasia, 58 with PSP, 53 with CBS, 60 with MND. Serum assays for 41 inflammatory markers used the MesoScale Discovery V‐Plex‐Human Cytokine 36 plex panel plus five additional cytokine assays. A Principal Component Analysis (PCA) across all participants was used to identify leading multivariate components, or profiles, of inflammation. Analysis of variance and Bayesian pairwise t‐tests were performed on the resulting components to compare each patient cohort to controls.
Result
Fifteen cytokines were undetectable in the majority (>50%) of participants and were thus excluded. The PCA of remaining 26 cytokines identified 3 components (explaining 18.3%, 8.5% and 6.4% variance, respectively). Component 1 was strongly represented by pro‐inflammatory cytokines (Figure 1, left). Kruskal–Wallis one‐way analyses of variance on the first component detected significant differences across the groups (χ2(8) = 24.3, p = 0.0021), and the Bayesian pairwise t‐tests identified significant differences between each patient cohort and controls, except for people with MCI (Figure 1, right). Component 2 did not differ between patients and controls. Component 3 split the cytokines into two subgroups, and was mainly loaded by patients with MCI, AD and MND (χ2(8) = 64.6, p<0.0001).
Conclusion
This data‐driven approach identified similar profiles of pro‐inflammatory responses across multiple neurodegenerative dementia types. Further analyses will clarify differences between diagnoses, relationships of the pro‐inflammatory patterns with clinical severity and brain changes captured by neuroimaging (PET and MRI).