Alzheimer's Imaging Consortium

Abstract

Background: Fluid biomarkers provide a convenient way to predict AD pathophysiology. However, few studies have focused on determining associations with tau neurofibrillary tangle pathology in the early preclinical AD continuum, relevant to prevention strategies.

Methods: Ninety-nine cognitively unimpaired individuals from the ALFA+ cohort with valid 18F-RO-948 and 18F-flutemetamol PET, T1-weighted MRI, cognition, CSF, and plasma biomarkers were included. Participants were initially categorized into AT stages using CSF-based pre-established cut-off values [1]. Regional SUVR of 18F-RO-948 PET was calculated in entorhinal (BraakI/II), limbic (BraakIII/IV), and neocortical (BraakV/VI) regions using the inferior cerebellum as reference region as well as with the CenTAURz. Regional positivity thresholds per Braak stage were calculated as the median+2SD of the CSF A-T- group. Amyloid PET was quantified using Centiloids. Pearson correlations were calculated between regional 18F-RO-948 SUVRs and AD biomarkers. ROC analyses adjusted for age, sex, and APOE-e4 performed to evaluate the capacity of biomarkers in predicting BraakI/IIPositive. Four progressive PET-derived AT groups were defined using Centiloid and tau PET positivity cut-offs (A-T-, AGZT-, A+T- and A+T+; with A- CL<12, 12=AGZ<38 and A+ CL=38 [2], and T+ BraakI/II>1.35) and between-stage differences in z-scored biomarkers evaluated using a Kruskal-Wallis tests.

Results: Table 1 shows demographic information of participants. Nine(9.09%) participants were BraakI/IIPositive, seven(7.07%) BraakIII/IVPositive and one(1.01%) BraakV/VIPositive. Two BraakIII/IVPositive participants were BraakI/IINegative, deviating from the Braak hierarchical model. CSF biomarker correlations with BraakI/II SUVR (Figure 1-A) ranged from r=0.24(ttau) to r=0.57(ptau217) and plasma (Figure 1-B) from r=0.30(ptau217) to r=0.49(ptau181). Correlations survived adding age+sex+APOE-e4 in the model (Figure 1-C&D). CSF ptau181/Aß42, ptau217 and ptau205 showed an AUC=0.93 to predict BraakI/IIPositive, and plasma ptau181, ptau181/Aß42 and ptau217 had an AUC=0.84. Centiloid positivity threshold for BraakI/IIPositive was 38.14CL. Plasma ptau181, ptau181/Aß42, and CSF ptau205, ptau217, and ptau235 reached a mean z-score>2 for the PET-derived A+T+ group (Figure 2) which was associated with lower cognitive scores for executive function (p=0.03), attention (p=0.05), and the PACC (p=0.01).

Conclusion: 18F-RO-948 PET conformed to the Braak hierarchical model for most tau-positive participants. Fluid AD biomarkers showed moderate associations with tau PET SUVR. Plasma biomarkers showed good capacity to predict BraakI/IIPositive and track fibrillary amyloid and tau pathological changes in the early preclinical AD continuum.