Author Of 1 Presentation
PS07.03 - Predicting disability progression and cognitive worsening in multiple sclerosis with gray matter network measures
In multiple sclerosis (MS), MRI measures at a whole and regional brain level have proven able to predict future disability, albeit to a limited degree. Their modest prognostic ability may reflect how cognitive and neurological functions are served by distributed networks rather than by single brain regions.
We aimed to identify data-driven MRI network-based measures of covarying gray matter (GM) volumes that can predict disability progression.
We used baseline MRI and longitudinal clinical data from 988 patients with secondary progressive MS (SPMS) from a randomized, double-blind, placebo-controlled, multicenter trial (ASCEND). We applied spatial-ICA to baseline structural GM probability maps to identify co-varying GM regions. We computed correlations between the loading of our ICA components and expanded disability status scale (EDSS), 9 hole peg test (9HPT), and symbol digit modalities test (SDMT) scores. We estimated the progression of the EDSS confirmed at 3 months, 6 months, and 1 year, and respectively the 20% and 10% worsening of 9HPT and SDMT. We used Cox proportional hazard models to determine the prognostic value of our ICA-components and conventional MRI measures (whole and deep GM volumes, and white matter lesion load).
We identified 15 networks of co-varying GM patterns that were clinically relevant. At baseline, SDMT and 9HPT scores correlated more strongly with ICA-components than the conventional MRI measures. The highest correlations were with a mainly basal ganglia component (encompassing the thalamus, caudate, putamen, frontal and temporal lobe). EDSS correlated more closely with an ICA-component involving cerebellum, brainstem, temporal and parietal lobes (r= -0.11, p<0.001). Prognostically, the baseline volume of caudate predicted EDSS progression confirmed at 3 months (HR= 0.81, 95%CI [0.68: 0.98], p<0.05), while some GM network-based measures outperformed conventional MRI measures in predicting SDMT and 9HPT worsening. SDMT progression was predicted by 6 ICA-components (component 8 (HR= 1.26, 95% CI [1.08-1.48], p< 0.005, and component 13 (HR= 1.25, 95% CI [1.07:1.46], p<0.005)). Two ICA-components were predictors of 9HPT worsening (HR=1.30, 95% CI [1.06:1.60], p<0.01; and HR= 1.21, 95%CI [1.01:1.45], p<0.05).
Data-driven MRI network-based measures of covarying GM volumes predict disability progression better than volumetric measures of GM and white matter lesion loads. ICA of MRI shows promise as a method that could enrich clinical MS studies with patients more likely to show a treatment response.