One of the challenges in multiple sclerosis (MS) research is to improve prediction of disease progression. While information from conventional MRI of the brain is essential in the diagnosis of MS, it only allows prognosis to some extent. Given the complexity of the disease, a combined analysis of structural and functional MRI changes appears more promising to identify markers associated with disease progression.
We thus investigated how multimodal MRI can contribute to predicting disease progression in a single-centre cohort of patients with MS (PwMS).
We analyzed multimodal MRI-data from 123 PwMS (71 women; age (years): M=37.2, SD=10.4; nCIS=16; nRRMS=98; nPMS=9). All patients had undergone clinical and 3T MRI evaluations between 2015 and 2016 (baseline, BL) and clinical re-evaluations 2 years later (SD=1.0; follow-up, FU). Brain-volume, T2 lesion load, fractional anisotropy (FA) and resting-state functional connectivity (rsFC) of the default-mode (DMN) and sensorimotor network (SMN) at BL were correlated with the patients’ disease severity score progression (absolute Expanded Disability Status Scale (EDSS) score change from BL to FU).
Across the entire cohort, median EDSS scores were significantly higher at FU (Med=1; IQR=2.5) than at BL (Med=1; IQR=2; p=0.04), with a low rate of disease severity score progression (assessed by EDSS BL – EDSS FU/ FU duration; Med = 0; IQR = 0.5). Neither normalized brain volume (NBV) nor T2 lesion load, extracted mean scores of whole brain FA or rsFC within DMN or SMN significantly correlated with disease severity score progression. Whole brain voxel-based analyses (controlled for age and disease duration) indicated trends for decreased FA within the corpus callosum (CC) and the corticospinal tract (CST) and decreased rsFC within the anterior cingulate cortex (ACC) and the hand motor area to be associated with disease progression. Subsequent ROI analyses revealed a significant decrease in mean FA in the CC genu (p=0.024), the CC forceps minor (p=0.020) and right CST (p=0.020) related to disease progression. Moreover, ROI analyses showed a decrease in mean rsFC in the left hand motor area (p=0.012) and the ACC (p=0.005) with increased disease progression.
Our results show that even within a relatively low rate of clinical disease progression over short term FU, subtle microstructural and functional changes may represent more sensitive predictors compared to gross morphological measures obtained from conventional MRI.