Cortical axonal loss is associated with both gray matter demyelination and white matter tract pathology in progressive multiple sclerosis: Evidence from a combined MRI-histopathology study

Background: Neuroaxonal degeneration is one of the hallmarks of clinical deterioration in progressive multiple sclerosis (PMS). Objective: To elucidate the association between neuroaxonal degeneration and both local cortical and connected white matter (WM) tract pathology in PMS. Methods: Post-mortem in situ 3T magnetic resonance imaging (MRI) and cortical tissue blocks were collected from 16 PMS donors and 10 controls. Cortical neuroaxonal, myelin, and microglia densities were quantified histopathologically. From diffusion tensor MRI, fractional anisotropy, axial diffusivity (AD), radial diffusivity (RD), and mean diffusivity (MD) were quantified in normal-appearing white matter (NAWM) and white matter lesions (WML) of WM tracts connected to dissected cortical regions. Between-group differences and within-group associations were investigated through linear mixed models. Results: The PMS donors displayed significant axonal loss in both demyelinated and normal-appearing (NA) cortices (p < 0.001 and p = 0.02) compared with controls. In PMS, cortical axonal density was associated with WML MD and AD (p = 0.003; p = 0.02, respectively), and NAWM MD and AD (p = 0.04; p = 0.049, respectively). NAWM AD and WML AD explained 12.6% and 22.6%, respectively, of axonal density variance in NA cortex. Additional axonal loss in demyelinated cortex was associated with cortical demyelination severity (p = 0.002), explaining 34.4% of axonal loss variance. Conclusion: Reduced integrity of connected WM tracts and cortical demyelination both contribute to cortical axonal loss in PMS.


Quantification of neuro-axonal degeneration and cortical pathology
Per section two regions of interest (ROIs) of approximately 8 mm2 each, were randomly selected in sixlayered and non-curved parts of the cortex. Afterwards ROIs were classified as containing lesioned or cortical NAGM by an experienced rater (S.K.).
Images of Bielschowsky staining were acquired using a Leica DM/RBE photomicroscope (Leica, Heidelberg, Germany) at 200x magnification. Images were stitched to obtain one image per ROI including all cortical layers. Analyses of axonal density was performed manually in ImageJ/Fiji (version 1.52a, https://imagej.net/Fiji). 1 Vertical lines running from the pial surface to the WM were superimposed on every image i.e. ROI. Axonal density was obtained by counting the intersection of axons with either a vertical line or with horizontal lines that were computed along the entire length of the vertical line every 500µm, to obtain a density measure without bias of fiber orientation ( Figure 1B).
Axons intersecting with both lines or with one line twice were only counted once. Axonal density was expressed as axon number per mm.
Neuronal density and volume were quantified using a Leica DMR microscope (Leica Microsystems, Wetzlar, Germany) in combination with stereoinvestigator software (MBF Bioscience, Williston, VT; Figure 1D). The optical fractionator tool was used to obtain density measures of neurons unbiased for orientation, shape and size in thick sections. 2 The nucleus top was used as a unique point for counting, while sampling in a random systematic manner throughout the ROIs in 3D at 630x magnification (oil lens). Per section, 103±30 NeuN + cells were counted on average, the coefficient of error was 0.10±0.02 (Schmitz and Hof). 3 The nucleator tool was used to quantify neuronal volume while sampling with the optical fractionator, using the nucleolus as a unique point. 4 Microglia density was quantified using a Leica Ctr 5000 microscope (Leica Microsystems, Wetzlar, Germany) with a Nuance spectral imaging device and associated Nuance spectral imaging software (Nuance version 3.0.2, Caliper Kife Sciences, Inc, a Perkin Elmer Company, Hopkinton, MA).
Separate images were acquired of cortical layers 1 and 2, layer 3 and layer 5 and 6, at 200x magnification. To obtain robust measures of microglia density two images were acquired per cortical layer and results were averaged. Spectral information of Iba-1 immunostaining was used to compute a corresponding mask and microglia density was expressed as percentage of stained area of the image area.
Myelin density was quantified within ROIs using images of PLP stained sections that were acquired using a Leica DM/RBE photomicroscope (Leica, Heidelberg, Germany) at 50x magnification.
Images were stitched to obtain a single image of the entire section. ImageJ was used to superimpose the ROIs onto the images. A mask of the myelin staining was computed and myelin density was expressed as percentage of stained area of the ROI area.
A group-level probabilistic WM tract atlas, based on an in vivo healthy control dataset (23 males, 37 females, mean age=50±7), was finally constructed by non-linear co-registration of each subjects' binarized tracts to Montreal Neurological Institute (MNI) template space. The resulting atlas was used