Multiple sclerosis (MS) subtypes defined by brain magnetic resonance imaging (MRI) scans can predict disability progression and response to treatment, according to study results published in Nature Communications.
The study’s researchers processed brain scans from 19 data sets (randomized controlled trials: n = 16) and used them to train and validate a machine learning model. Unsupervised, the scans were clustered to layer MS-associated brain patterns. External validation using 5 data sets was used to predict disability progression and response to treatment.
The training and validation cohorts comprised 6322 and 3068 patients, respectively. The control and training datasets differed significantly in 13 of the 18 MRI features.
The clustering model identified 3 different evolutionary patterns that were successfully validated in the independent cohort. The clusters were considered cortex-gated, normal-appearing white matter (NAWM), and lesion-gated.
The most common subtypes were cortex-controlled, followed by NAWM-controlled. Stratified by these subtypes, patients with a lesion-controlled subtype had the smallest baseline gray matter volume, the highest baseline lesion load, an expanded disability status scale (EDSS), an increase in lesions, and the longest disease duration (all P <0.001) . Study researchers observed no age or gender differences between subtypes.
Patients with a lesion-controlled subtype had a 30% (95% CI, 5% to 62%; P = 0.01) higher risk of 24-week disability progression than the cortex-controlled subtype in the training dataset and 32% (95% CI, 9% to 59%; P = 0.004) higher risk in the validation cohort. The mean relapse rate was highest among the lesion-guided subtypes for both training sessions (mean 0.56; standard error) [SE], 0.07) and validation cohorts (mean 0.41; SE, 0.03).
These subtypes were with a 24-week disability progression (b, 0.04; SE, 0.01; P = 0.02) and stages (b, -0.06; SE, 0.02; P <0.001) associated.
The subtypes could be used to predict disease progression with a concordance index of 0.55 (standard deviation) [SD]± 0.01), and this index was improved to 0.63 by including clinical features (SD ± 0.01; P <0.01).
Using data from clinical trials, patients with a lesion-controlled subtype on active treatment showed slower deterioration in EDSS compared to placebo in patients with secondary progressive MS (mean difference -66%; SE ± 25.6%; P = 0.009) or primary progressive MS (mean difference -89%; SE ± 44%; P = 0.04).
Although data from several independent cohorts were combined in this study, the patient stratification should be repeated with a prospective study.
These results suggest that there are 3 subtypes of MS phenotypes associated with disability progression and response to active treatment.
Disclosure: Some of the study’s authors stated links with biotech, pharmaceutical, and / or device companies. For a full list of the authors’ claims, see the original article.
Eshaghi A., Young AL, Wijeratne PA, et al. Identification of multiple sclerosis subtypes using unsupervised machine learning and MRI data. Nat Commun. 2021; 12 (1): 2078. doi: 10.1038 / s41467-021-22265-2