Can post-mortem MRI be used as a proxy for in vivo? A case study

Abstract

Post-mortem in situ MRI has been used as an intermediate between brain histo(patho)logy and in vivo imaging. However, it is not known how comparable post-mortem in situ is to ante-mortem imaging. We report the unique situation of a patient with familial early-onset Alzheimer's disease due to a PSEN1 mutation, who underwent ante-mortem brain MRI and post-mortem in situ imaging only 4 days apart. T1-weighted and diffusion MRI was performed at 3-Tesla at both time points. Visual atrophy rating scales, brain volume, cortical thickness and diffusion measures were derived from both scans and compared. Post-mortem visual atrophy scores decreased 0.5-1 point compared with ante-mortem, indicating an increase in brain volume. This was confirmed by quantitative analysis; showing a 27% decrease of ventricular and 7% increase of whole-brain volume. This increase was more pronounced in the cerebellum and supratentorial white matter than in grey matter. Furthermore, axial and radial diffusivity decreased up to 60% post-mortem whereas average fractional anisotropy of white matter increased approximately 10%. This unique case study shows that the process of dying affects several imaging markers. These changes need to be taken into account when interpreting post-mortem MRI to make inferences on the in vivo situation.

Keywords: MRI; ante-mortem; immunohistopathology; in vivo; post-mortem.

Graphical Abstract

Figure 1

MRI and corresponding (immuno)histopathology of the presented case. Hyperphosphorylated tau (p-tau; AT8) shows many neuritic plaques, threads and neurofibrillary tangles in the medial frontal gyrus. Amyloid-beta (Aβ; 6F/3D) deposits are mostly found as homogenous compact plaques in the medial frontal gyrus. Haematoxylin-eosin staining of the cerebellum shows agonal changes, visualized by shrunken eosinophilic cytoplasm of the Purkinje cell and retraction artefacts enclosing cells in both grey and white matter. Bordered areas represent the image-region of the consecutive column. Bar is applicable to all (immuno)histopathology images and represents 2 µm, 144 µm and 23 µm from left to right in consecutive columns.

Figure 2

Ante-mortem and post-mortem T1-weighted images of the brain. 3D T1-weighted images of the brain, 4 days ante-mortem (A) and 9 h post-mortem (B) registered in the same space. The fusion of both images (C) clearly shows an increase in brain volume post-mortem, particularly evident by smaller ventricles and the swollen cerebellum.

Figure 3

Scatterplot of the ante-mortem versus post-mortem cortical thickness. Scatterplot of the ante-mortem versus post-mortem cortical thickness (A) in 68 different brain regions (Desikan et al., 2006), using the longitudinal stream in FreeSurfer (Reuter et al., 2012). Pearson’s r = 0.975, P < 0.001.

Figure 4

Scatterplots of cortical thickness versus immunohistopathology. The scatterplots of ante-mortem and post-mortem cortical thickness versus immunohistopathology for amyloid-beta (Aβ) immunoreactivity (A) and hyperphosphorylated tau (p-tau) immunoreactivity (B). Both scatterplots show the same shape for the two time points. The symbol legend for brain regions on the far right is applicable to both plots.