In a previous study differences in cortical appearance in patients with Alzheimer’s Disease (AD) were assessed in vivo¹ and the nature of these changes were investigated histologically². A diffuse hypointense band was found on MRI in the frontal lobes of AD patients (57% of cases), but not in controls. Histological correlation showed colocalization of the diffuse band with iron and myelin². Differences between patients with an early disease onset (EOAD, before age 65) and patients with a late disease onset (LOAD, after age 65) are previously reported, including a more rapid cognitive decline³ and a more severe post-mortem pathology⁴ in EOAD patients. Even differences on MRI have been found: increased cortical phase shifts suggesting increased iron accumulation are reported for EOAD compared to LOAD patients⁵. However, less is known about the origin of these differences on MRI and the involved pathology.

Therefore, this study investigated the correlation between ultra-high field MRI and AD pathology in the frontal cortex to further investigate the role of iron and myelin on MRI contrast and their changes in subsets of AD patients, containing EOAD and LOAD patients.

Post-mortem tissue samples of the frontal cortex, containing the medial frontal gyrus, from LOAD patients (N=10), EOAD patients (N=11) and non-demented controls (N=13) were obtained from the Netherlands Brain Bank. Ultra-high field MRI scans were made on a 7T horizontal bore Bruker system equipped with a 23mm volume coil and Paravision 5.1 imaging software (Bruker Biospin, Ettlingen, Germany). MGE scans were acquired from each brain sample with TR=75.0ms, TE=12.5, 23.2, 33.9 and 44.6ms, FA=25° at 100µm isotropic resolution with 20 signal averages. The acquired magnitude images were assessed using pre-defined criteria based on previous 7T MRI studies of the human cortex to define normal and abnormal cortex, including the presence of cortical layers, focal hypointensities, and a diffuse hypointense band². Using the scoring criteria and values (Figure 1), a total score was calculated with maximum of 8/8 indicating an abnormal cortex and minimum score of 0/8 indicating a normal cortex. All samples have been scanned, but analysis has only been completed in a sub-set (numbers mentioned in the respective figures). The same brain samples as used for MRI were used for histology; consecutive sections were stained for myelin, iron, Aβ, tau, microglia, and astrocytes. After automatic non-linear registration of the histological images to the MRI images using elastix⁶, the correlation between MRI and histology was determined using spatial correlation of the pixels intensities but after normalizing all images using histogram equalization. Additionally, the myelin architecture of layer IV of the frontal cortex was investigated using light microscopy and quantified using standardized stereological techniques (MCID Core Digital Imaging Software)⁷.

The AD subjects showed differences on MRI and histology compared to non-demented control subjects. AD subjects were more frequently scored as abnormal on MRI, characterized by the presence of hypointensities and absence of cortical layers. Moreover, a diffuse hypointense band was frequently found in AD subjects, but not in controls. Both AD groups were scored higher than controls, but EOAD subjects were more affected than LOAD subjects (Figure 2,3 first panel).

Correlation analysis showed that the diffuse hypointense band on MRI correlated most with an increased band-shaped intensity in the iron and myelin staining (Figure 3,4). The diffuse band showed less spatial correlation with the Aβ, tau, microglia or astrocyte staining. Further analysis of the myelin staining showed that the myelin architecture in layer IV of the frontal cortex of non-demented control subjects was characterized by an organized raster-like structure with myelin fibers transversing all cortical layers and intracortical fibers running parallel to the cortical layers (Figure 5). In LOAD subjects already more disorganization compared to non-demented controls was found, noticeable as a less clear raster-like structure and less fibers transversing all cortical layers. In EOAD subjects the myelin architecture was completely changed compared to non-demented control; the myelin architecture showed a messy, disorganized structure characterized by crowded reticular fibers. Stereometric analysis of the fiber trajectories showed significant differences between all three groups.

Using MRI, we can clearly distinguish different subtypes of AD. These differences in imaging phenotype distinguishes the EOAD from the LOAD patients, which is mirrored in the histology and underlying pathology, mostly in differences in myelin and iron organization. Clinically, it is increasingly recognized that AD is a heterogeneous disease with multiple underlying mechanistic causes. The roles of iron and myelin integrity in the disease process are however poorly understood, and SWI could play a crucial role in classifying subtypes of AD for further mechanistic studies.