Optimal combination of CSF phosphorylated‐tau forms to predict Alzheimer's disease pathological burden as measured by PET.

Background: Recently, we demonstrated the sensitivity of a novel CSF mass spectrometer‐based assay, which enabled the assessment of multiple phospho‐tau (pTau) forms (pT181, pS199, pS202, pT205, pT217, pT231, and pS396) and two non‐phosphorylated peptides. Initial results suggested that there may be disease associated differences across pTau species. Here, we aimed to investigate the associations between these CSF pTau forms and Alzheimer's disease (AD) core pathologies as measured by PET. Methods: We included 352 participants from the BioFINDER‐2 cohort with available CSF mass‐spectrometry quantification and who underwent [18F]flutemetamol and [18F]RO948 PET to assess amyloid‐β (Aβ) and tau burden, respectively (Table1). Aβ burden was determined globally and in an early accumulating region (precuneus). Tau burden was determined for a temporal meta‐ROI and Braak stages (I‐II, III‐IV, V‐VI). The ratio of tau phosphorylated to non‐phosphorylated peptides was additionally determined for pS199, pS202, pT205, pT217. CSF pTau‐forms (ratio‐)levels were log‐transformed. To assess the contribution of Aβ and tau burden on pTau‐forms (ratio‐)levels, we ran multivariable linear regression models including both Aβ and tau PET measures as independent variables, covarying for age and sex. Next, we determined the parsimonious model (i.e. minimal number of predictors within 2 Akaike points of optimal model) of pTau‐forms (ratio‐)levels using R package MuMIn to predict Aβ and tau PET burden in the different ROIs. Results: In the whole cohort, all CSF pTau‐forms peptide concentrations, except for pS396, were associated with both Aβ and tau PET burden, though to different extents (Figure1). pT217 and pT231 were more strongly associated with Aβ, while pS199 and pT205 were more strongly associated with tau. In Aβ‐positive subjects (Figure1C/D), the association with tau increased across pTau‐forms. For the multivariable parsimonious models, pTau217 consistently showed the strongest contribution to predicting amyloid and tau burden across ROIs (Table2&3). pT231 was also selected when early amyloid burden was assessed. pT205 was only selected to predict tau burden in early ROIs when corrected for the non‐phosphorylated peptide (Table3). In addition, sex significantly contributed to predicting tau‐PET burden across ROIs. Conclusion: We provide further evidence that measurement of multiple phosphorylated‐tau species could be used to track AD disease progression. [ABSTRACT FROM AUTHOR]