Networks of microstructural damage predict disability in multiple sclerosis

Damage in multiple sclerosis is not confined to visible lesions. Subtle changes in the microstructure of otherwise normal-looking tissue can be picked up from standard clinical sequences — and they do not occur in isolation, but in regions that deteriorate together.

This Journal of Neurology, Neurosurgery & Psychiatry study, senior-authored by Dr Arman Eshaghi, analysed 1,836 people across MS phenotypes, split into a discovery cohort of 843 and a replication cohort of 842. Standardised T1-weighted/T2-weighted ratio maps — a measure sensitive to tissue microstructure that can be derived from routine scans — were computed in brain grey and white matter, and spatial independent component analysis was used to find networks of covarying damage. Outcomes were 24-week confirmed disability progression and time to cognitive worsening on the Symbol Digit Modalities Test.

Eight white matter and seven grey matter networks replicated across both cohorts. Loading in the anterior corona radiata and in a temporo-parieto-frontal component was associated with a higher risk of confirmed disability progression in the discovery cohort (HR 0.85 and 0.83, p < 0.05) and again in replication (HR 0.84 and 0.80, p < 0.005). Independent replication in a separately held cohort is the part that matters: these are prognostic markers that survive being tested on data they were not derived from, computed from sequences hospitals already acquire.

Read the paper: Networks of microstructural damage predict disability in multiple sclerosis.