Why does ARIA‐E appear bright? A quantitative model linking brain tissue composition and T2 FLAIR hyperintensities

Background Amyloid related imaging abnormalities (ARIA) detected by magnetic resonance imaging (MRI) are associated with a class of therapeutic anti‐amyloid‐β antibodies under investigation in Alzheimer’s disease. ARIA‐E, manifested as bright signals on the T2 fluid attenuated inversion recovery (FLAIR) sequence, are suggestive of parenchymal vasogenic edema and/or sulcal effusions, although, solid experimental support is still missing [1]. By employing a novel voxel‐based model of the MRI signal, we test whether the extravasation of fluid and plasma proteins in the brain tissue and/or sulcal spaces can indeed explain the T2 FLAIR hyperintensities characteristic for ARIA‐E. Method The modelled MRI voxel consists of myelin (my), intracellular (ic) and extracellular (ec) compartments (Figure 1). The material properties and MRI parameters (T1 and T2 relaxation times) of the physiological system come from experiments on protein solutions and human brain (Table 1). The movement of water molecules between the ic and ec compartments is assumed to be fast and to generate a well‐mixed intra/extra‐cellular (iec) compartment. The net T2 FLAIR signal within the voxel is assumed to result from the water in the my and iec compartments undergoing slow exchange and having independent magnetic relaxation. We assess how expanding the ec space with a protein‐rich fluid affects the T1 and T2 relaxation times of the ec and iec compartments and the total FLAIR signal within the voxel. Result Our model predicts that a two‐fold enlargement of the ec space, corresponding to a 12% increase in water volume, can lead to a substantial parenchymal hyperintensity on T2 FLAIR. The presence of a protein‐rich fluid, e.g. plasma, in the ec space cannot yield parenchymal hyperintensities in the absence of ec expansion. Conversely, replacing the cerebrospinal fluid from the sulci (normally appearing dark on T2 FLAIR) with protein‐rich fluid results in a signal comparable to normal grey matter and the apparent loss of sulci (Figure 2). Conclusion By linking brain tissue composition to FLAIR signal changes, our model could help unravel the pathophysiological mechanisms of ARIA‐E and, therefore, provide a useful tool for testing other potential mechanisms underpinning brain MRI abnormalities. Reference: (1) van Dyck, Biological Psychiatry, 2018, 83(4):311‐319.