Axonal loss in the multiple sclerosis spinal cord revisited

Preventing chronic disease deterioration is an unmet need in people with multiple sclerosis, where axonal loss is considered a key substrate of disability. Clinically, chronic multiple sclerosis often presents as progressive myelopathy. Spinal cord cross-sectional area (CSA) assessed using MRI predicts increasing disability and has, by inference, been proposed as an indirect index of axonal degeneration. However, the association between CSA and axonal loss, and their correlation with demyelination, have never been systematically investigated using human post mortem tissue. We extensively sampled spinal cords of seven women and six men with multiple sclerosis (mean disease duration= 29 years) and five healthy controls to quantify axonal density and its association with demyelination and CSA. 396 tissue blocks were embedded in paraffin and immuno-stained for myelin basic protein and phosphorylated neurofilaments. Measurements included total CSA, areas of (i) lateral cortico-spinal tracts, (ii) gray matter, (iii) white matter, (iv) demyelination, and the number of axons within the lateral cortico-spinal tracts. Linear mixed models were used to analyze relationships. In multiple sclerosis CSA reduction at cervical, thoracic and lumbar levels ranged between 19 and 24% with white (19-24%) and gray (17-21%) matter atrophy contributing equally across levels. Axonal density in multiple sclerosis was lower by 57-62% across all levels and affected all fibers regardless of diameter. Demyelination affected 24-48% of the gray matter, most extensively at the thoracic level, and 11-13% of the white matter, with no significant differences across levels. Disease duration was associated with reduced axonal density, however not with any area index. Significant association was detected between focal demyelination and decreased axonal density. In conclusion, over nearly 30 years multiple sclerosis reduces axonal density by 60% throughout the spinal cord. Spinal cord cross sectional area, reduced by about 20%, appears to be a poor predictor of axonal density.