Associations between Aβ‐PET, CSF pTau, and plasma GFAP, pTau181, pTau231 in memory clinic patients
Marina Bluma, Marco Bucci, Nicholas J. Ashton, Irina Savitcheva, Konstantinos Chiotis, Anna Matton, Guglielmo Di Molfetta, Lana Grötschel, Kaj Blennow, Henrik Zetterberg, Agneta K Nordberg- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Previous studies yielded contradictory findings regarding the association of plasma biomarkers with established biomarkers of tau and amyloid (Aβ) pathology. Here, we aim to examine the relationship between Aβ‐PET and CSF pTau with plasma biomarkers in a cohort of memory clinic patients.
Methods
Plasma GFAP, pTau181, pTau231, and CSF pTau181 were measured in 124 patients (age = 65±8(mean±SD), 69F/55M, MMSE = 25.6±3), admitted to the Memory Clinic,Theme Inflammation and Aging, Karolinska University Hospital, Stockholm,Sweden, and referred to [18F]flutemetamol Aβ‐PET after extensive clinical assessment.PET images were preprocessed with rPOP pipeline for PET‐only datasets. Plasma GFAP was measured using N4PE assay (Quanterix), whereas pTau181 and pTau231 with in‐house Single molecule array (Simoa) assays developed at the University of Gothenburg, and CFS pTau181 ‐ with commercial ELISA.To assess the relationship between the biomarkers we ran multiple linear regression (adjusted for sex and age,standardized variables),dominance analysis to compare the relative importance of the predictors, and mediation analysis to clarify interrelations between them.
Results
We found significant(p<0.05) associations of plasma GFAP(βAβ‐PET = 0.42,βpTau = 0.22), pTau231(βAβ‐PET = 0.38,βpTau = 0.39), and pTau181(βAβ‐PET = 0.35,βpTau = 0.3) with Aβ‐PET and CSF pTau, although when both Aβ‐PET and CSF pTau were added into the multivariate regression the effect of CSF pTau was attenuated by the effect of Aβ‐PET for GFAP(βAβ‐PET = 0.45) and pTau181(βAβ‐PET = 0.27), but not for pTau231(βAβ‐PET = 0.24,βpTau = 0.25). Dominance analysis showed that the relative importance of the variables (i.e., Aβ‐PET, demographics (sex, age), and CSF pTau) for prediction of GFAP values followed the order: Aβ‐PET>demographics>CSF pTau; of pTau181: Aβ‐PET>CSF pTau>demographics; and of pTau231: CSF pTau>Aβ‐PET>demographics. Mediation analysis revealed that the effect of CSF pTau on plasma GFAP and pTau181 was fully mediated by Aβ‐PET, whereas in case of pTau231 we could observe partial mediation, where both direct effect of CSF pTau and indirect effect through Aβ‐PET were significant.
Conclusion
In a clinical cohort of patients, we found differing associations between plasma GFAP, pTau181, and pTau231 with Aβ‐PET and CSF pTau.Amyloid burden contributed the most in explaining the variance in GFAP and pTau181, but not pTau231, where the contribution of CSF pTau was higher.These results suggest plasma GFAP and pTau181 as potential markers of pathological processes associated with amyloid pathology, whereas pTau231 ‐ with both amyloid and tau pathology.