Background/Objectives: Multiple sclerosis (MS) can cause neurodegeneration leading to accelerated brain atrophy. Brain-predicted age (BA) is an emerging neuroimaging biomarker for neurodegeneration but remains underexplored in MS. This study examines the pathophysiological substrates associated with the brain age gap in MS compared with healthy controls (HCs) through a combination of volumetric, spectroscopic, and diffusion imaging. Methods: This retrospective cross-sectional study included 33 HCs and 124 MS patients. Participants underwent 3T MRI including 3D-T1, MR spectroscopy, magnetization transfer, and diffusion imaging. BA and volumes were estimated from T1-weighted scans using brainageR. Metabolic integrity (total N-acetylaspartate to total creatine ratio, tNAA/tCr) and microstructural damage (magnetization transfer ratio MTR, fractional anisotropy FA) were evaluated independently in normal-appearing tissues. Multivariate linear regression assessed MS diagnosis as an independent predictor of BA metrics, controlling for age, sex, and race. Results: MS patients showed significantly higher predicted brain age (53.3 vs. 31.8 years) and a markedly larger age gap (10.2 vs. −0.1 years) compared to HCs. Beyond macroscopic volume loss, accelerated aging paralleled profound subclinical degradation, including lower neuronal integrity (tNAA/tCr: 2.0 vs. 2.4) and widespread microstructural damage, evidenced by reduced MTR and FA across both normal-appearing gray and white matter. Linear regression confirmed MS diagnosis as an independent predictor of both BA and Age Gap (15.09 and 13.50 years) after adjusting for confounders. Conclusions: MS patients exhibit accelerated biological brain aging, characterized by a significant age gap and concurrent tissue volume loss. The brain age gap in MS extends beyond macroscopic atrophy, capturing underlying subclinical metabolic failure and widespread microstructural degradation in normal-appearing tissues. This positions BA as a robust, multi-dimensional proxy for neuroaxonal pathology.
Nourelden et al. (Mon,) studied this question.