Association between antemortem plasma and structural MRI biomarkers and postmortem tau pathology in the Mayo Clinic...

Antemortem plasma and structural MRI biomarkers of Alzheimer's disease (AD) are increasingly utilized for diagnosis and prognosis but their ability to predict regional tau burden remains relatively unexplored. We aimed to evaluate predictive models based on antemortem AD plasma and structural MRI for prediction of regional tau pathology in postmortem brain tissue.

Autopsy data from 62 participants in the Mayo Clinic Study of Aging (MCSA) were analyzed (mean age of 84.6 years; 32% female; 1.4 years of mean time of last plasma draw to death). Tau pathology in the hippocampus and parietal cortex was quantified using two immunohistochemical markers: AT8, an early tangle maturity marker, and 2E9, an advanced tangle maturity marker. Three antemortem plasma markers including phosphorylated tau protein at threonine 181 (p-tau181), total amyloid-β (Aβ) 42 relative to Aβ40 ratio (Aβ42/40), and glial fibrillary acidic protein (GFAP) and three regional structural MRI markers including the hippocampus volume adjusted for total intracranial volume (HVA), parietal thickness (PTH) and parietal volume divided by total intracranial volume (PVTIV) were utilized as predictors. Weighted linear regression models using weights based on time from last plasma draw to death were adjusted for age, sex and cardiovascular and metabolic conditions (CMC) score for patients. We report estimated adjusted R² and partial R² for predicting postmortem tau pathology in the hippocampus and parietal cortex stained by AT8 and 2E9.

MRI models (adjusted R² of 0.28-0.66) showed better performance at predicting tau pathology than plasma models (adjusted R² of 0.19-0.36). The highest adjusted R² was 0.66 for predicting hippocampal 2E9 using HVA, followed by an adjusted R² of 0.6 for hippocampal AT8. Predictions for parietal AT8 and 2E9 were lower with an adjusted R² of 0.28-0.29 based on MRI biomarkers and 0.24-0.36 based on plasma biomarkers. The adjusted R² values for both regions were higher for AT8 compared to 2E9 prediction for plasma biomarker models.

Our study demonstrates the greater sensitivity of regional MRI to tau pathology in comparison to plasma biomarkers. Future studies should explore newer MRI methods and additional plasma biomarkers more specific to tau pathology such as microtubule binding region for prediction of antemortem tau pathology.