Multiple sclerosis (MS) is a chronic demyelinating disease in which thalamic demyelination and degeneration contribute to neurological disability. We hypothesized that clinical symptom severity can be predicted by quantitative MRI measure of thalamic integrity.

We used T1w MPRAGE and T2w TSE acquired on Siemens 3T systems in the Mellen Center Clinical Practice Registry, segmented using FreeSurfer (v7.3.2) SynthSeg, and calculated thalamic volume and T1w/T2w ratio with established scaling methods. T2 lesion volume was calculated using deep learning algorithm. Extracted data were analyzed using a cumulative link mixed-effects model with random intercepts to predict ordinal PDDS categories (Low = 0–2, Mid = 3–4, High = 5–8). Analyses included patients with ≥3 longitudinal visits and minimum transient PDDS ≥1. Predictors included normalized thalamic T1w/T2w ratio, T2 lesion volume (T2LV), mean thalamic volume, age, sex, and disease duration.

Among 717 subjects (2,827 MRI–clinical pairs; 26.6% male; mean age 48.4 ± 10.6 and disease duration 15.5 ± 10.2 years), greater T2LV was significantly associated with worse PDDS (OR = 1.82, p < 0.001). Lower thalamic volume predicted greater disability (OR = 0.67, p = 0.038). The thalamic T1w/T2w ratio showed a positive, nonsignificant trend toward higher PDDS (OR = 1.22, p = 0.08). Age (OR = 1.09/year, p < 0.001) and disease duration (OR = 1.04/year, p = 0.009) were significant covariates; sex was not (p = 0.45).

In this large cohort, MRI-derived metrics of injury were robust correlates of disability. Thalamic atrophy and T2 lesion burden independently predicted higher PDDS. T1w/T2w was not associated with disability similar to prior findings. Substantial between-subject variability underscores disability heterogeneity and supports subject-specific modeling with future work extending this framework to additional longitudinal functional/cognitive measures.