LIMB-GIRDLE MUSCULAR DYSTROPHY I: P.11Clinical outcome study in dysferlinopathy: random forest approach to assess the relationship between baseline muscle MRI and longitudinal functional outcome measures

Abstract

The Jain clinical outcome study (COS) is an international study of 203 adults with dysferlinopathy in 8 countries. Patients undergo six visits over three years, during which physiotherapy and medical assessments as well as muscle MRI were performed. Our aim is to analyse baseline muscle pathology by MRI in a large cohort of dysferlinopathy patients and to examine the relationship between baseline muscle MRI findings and the results of longitudinal functional outcome measures over one year. We aim to identify which muscles may predict a rapid decline over one year. Lower limb muscle MRI has been performed in 182 patients with a confirmed diagnosis of dysferlinopathy in 14 different centres, using 1.5T or 3T scanners from different manufacturers (Philips, General Electrics, Siemens). 74 patients also underwent upper limb muscle MRI at baseline. Muscles were scored on axial T1-weighted sequences with the semi quantitative Mercuri visual scale modified by fisher. Physiotherapy assessments included muscle strength (manual muscle testing; hand held dynamometry) and functional ability evaluations (adapted north star, brooke test, timed tests) and patient reported outcome measure (ACTIVLIM). We will present the outcome of the analysis between semi-quantitative MRI at baseline and disease progression over one year as measured by the north star assessment for dysferlinopathy, the 10 metre walk and the ACTIVLIM using the Random Forest approach (package ``random forest SRC'' in R). Random-forest studies will be used to identify whether combinations of muscle involvement at baseline can predict a change of more than 10% in muscle function tests at a one year visit. Muscle MRI is useful for the diagnosis of patients with muscle disorders, but also to identify patients at risk of having a quick progression of their symptoms.