MS clinical types are obsolete

[frodo]

Since Charcot times, neurologist classify the MS types by its clinical evolution. This is so deeply stablished that even when NMO was considered a subtype of MS (it was called optic-spinal MS), the WHO four main "types" were still RRMS, PPMS, SPMS and the old PRMS.

Obviously with a so unspecific classification little work can be done. Finally there is some work towards a pathological classification via MRI.

Redefining multiple sclerosis phenotypes using MRI

https://www.medrxiv.org/content/10.1101/19011080v1

Abstract

Background:

There are 4 courses of multiple sclerosis (MS): clinically-isolated syndrome (CIS), relapsing-remitting MS (RRMS), primary-progressive MS (PPMS) and secondary-progressive MS (SPMS). We aimed to achieve a further sophistication in the definition of MS phenotypes by identifying patient subgroups who accumulate magnetic resonance imaging (MRI) abnormalities with similar patterns. We assessed whether data-driven subtyping predicted clinical outcome and response to experimental treatments.

Methods:

In this retrospective study, we included longitudinal data from 8,545 people with MS who had 31,451 visits from 14 double-blind randomised controlled trials and three observational cohorts. We included cross-sectional data from 14,928 healthy volunteers. For each visit, we processed brain MRI scans. We obtained 18 MRI variables, includ that included the volume of the cortex of each lobe and deep grey matter, cerebellar grey matter and wite matter, total lesion volume, cerebral white matter, brainstem, and T1/T2 ratio in regions of the normal appearing white matter (or NAWM). We trained a machine learning algorithm, called SuStaIn, on 14 datasets to identify data-driven subtypes. We then tested it in three external, independent datasets.we assessed the external validity of our model by testing if we could predict 24-week confirmed Expanded Disability Status Scale (EDSS) progression, disease activity, and the reduction in EDSS worsening in each subtype on treatment vs placebo. To investigate whether there was a difference in the treatment response between the SuStaIn subtypes we used linear mixed effect modles

Findings:

We identified three data-driven subtypes with a distinct neuroanatomical spread of abnormality and termed subtypes according to areas they affected early in disease course: cortex-first (44% of patients), NAWM-first (30%), and the lesion-first (26%). Data-driven subtyping and staging, but not clinical classifications or EDSS at baseline, was associated with time to EDSS progression (βSubtype=0.04 and βstage=-0.06, p<0.01 for all). In external datasets, compared to the cortex-first, the lesion-first subtype had a shorter time to EDSS progression (hazard ratios range=1.75 to 1.82 across trials, p<0.01) and higher disease activity (p<0.05 for relapse rate and gad-enhancing lesions). In three progressive MS trials the lesion-first subtype showed the greatest treatment response (64% more than the cortex-first, p<0.001).

Interpretation:

Data-driven MS subtypes and stages, when compared with clinical classification or baseline EDSS better predicts the subsequent clinical course and treatment response. Data-driven subtyping has the potential to prospectively enrich clinical trials with patients who are more likely to respond to treatments. Funding: International Progressive MS Alliance.