Compelling evidence of neural inter-subject variability has been reported within healthy populations. This is even more true when considering clinical cohorts in which inter-individual differences, leading to atypical neural patterns, are expectedly more pronounced, and considering longitudinal/rehabilitative effects, likely characterized by subject-specific plasticity mechanisms. In this context, standard GLM-based group-level analysis hardly represents individual patterns, requiring careful investigation of subject-specific neural correlates. We aimed to evaluate the inter-subject variability longitudinally in a cohort of people with multiple sclerosis (pwMS) following an ‘Eye movement desensitization and reprocessing’ treatment.

Thirty healthy volunteers (HV) and thirteen pwMS with associated depression were enrolled. fMRI data were acquired twice during an emotion stimulation task on a 3T Siemens Prisma scanner and preprocessed using SPM12. Two contrasts (positive/negative emotions) were derived at subject-level. Group statistics were obtained using GLM-based one-sample t-tests, while normalized individual contrasts t-maps were used to derive threshold-weighted overlap-maps (OM_th-w), assessing the consistency of neural activation.

pwMS were characterized by scattered neural activation patterns resulting in lower consistency. GLM-based analyses did not detect any longitudinal changes for either HV or pwMS. A decrease in consistency was observed comparing the OM_th-w derived longitudinally for both contrasts in HV. An increased consistency was instead observed in pre vs post-rehabilitation for pwMS together with significant improvement (p<0.001) in depressive symptoms.

The investigation of inter-subject variability allowed us to identify an increase in consistency in the neural response of pwMS following rehabilitation coupled with decreased depression symptoms suggesting the relevance of this approach over standard methods.