Tracking reactive astrocytes in autosomal dominant Alzheimer disease with plasma GFAP and multi‐modal PET imaging.

Background: We have proposed astrogliosis as a "first wave" of response to AD pathology with diverging longitudinal changes of reactive astrocytes and amyloid‐β positron emission tomography (PET) retention in pre‐symptomatic autosomal dominant AD (ADAD). It has been suggested that plasma glial fibrillary acidic protein (GFAP) could be a marker of neuroinflammation in AD brain, although this hypothesis remains unproven. We therefore compared plasma GFAP levels with 11C‐deprenyl (DED) with a multi‐modal PET design in ADAD mutation carriers (MC) and non‐carriers (NC). Method: Twenty‐three individuals from families with known ADAD mutations were included (MC = 9; NC = 14). All individuals underwent PET imaging with 11C‐DED, 11C‐PIB and 18F‐FDG, for assessing monoamine oxidase B as a marker of reactive astrocytes, amyloid‐beta deposition, and glucose metabolism, respectively. Plasma sampling was performed after a median of 3 (interquartile range 1.5‐5.5) months. GFAP levels were measured using single molecule array technology. Eleven individuals had follow‐up imaging investigations and plasma sampling after a median of 2.7 (interquartile range 2.5‐2.9) years. Result: Plasma GFAP levels and 11C‐PIB binding increased significantly from pre‐symptomatic to symptomatic stage while 11C‐DED binding and 18F‐FDG uptake significantly decreased towards estimated onset of clinical symptoms in MC. Cross‐sectionally, plasma GFAP showed negative correlations with [18F]FDG uptake in widespread cortical areas, in contrast to no correlations with 11C‐DED or 11C‐PIB binding. 11C‐DED binding showed a positive correlation with 18F‐FDG uptake. Across the follow‐up interval, plasma GFAP levels showed a four‐fold greater intraindividual variation relative to 11C‐DED binding variation (longitudinal Δ variance = 1.67 vs 0.39 z‐scores, respectively). When averaging plasma GFAP values and tracer binding across consecutive time points for the same individual, strong negative correlations were found between plasma GFAP and 11C‐DED binding and 18F‐FDG uptake, but no significant correlation was found between plasma GFAP levels and 11C‐PIB binding in MC. Conclusion: Divergent plasma GFAP levels and PET 11C‐DED brain binding in pre‐symptomatic ADAD carriers may reflect different but associated underlying processes, including different types of reactive astrocytes. However, the levels of plasma GFAP show large intraindividual variation over time and the assessment of reactive astrocytes based solely on cross‐sectional plasma GFAP levels poses great challenges and warrants further investigations. [ABSTRACT FROM AUTHOR]