To evaluate white matter microstructural changes in NMOSD using diffusion tensor imaging (DTI) across brain and spinal cord regions, and to identify DTI parameters most suitable as disease-monitoring biomarkers. 

 NMOSD primarily affects optic nerves & spinal cord, but evidence increasingly supports broader CNS involvement. DTI enables detection of microstructural abnormalities beyond conventional MRI-visible lesions by quantifying fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD). Comprehensive synthesis across anatomical regions is needed to identify optimal DTI biomarkers

 Systemic review was conducted searching PubMed, Europe PMC, and ClinicalTrials.gov (2015–2025) for studies comparing DTI parameters between NMOSD patients and healthy controls (HCs). Studies reporting extractable DTI metrics from defined anatomical regions were included. Pooled Hedges' g was calculated using inverse-variance weighted random-effects meta-analysis where ≥2 studies provided extractable means and SDs; single-study regions were reported&Risk of bias was assessed using a modified Newcastle-Ottawa Scale. 

Nine studies (159 NMOSD, 157 HCs; 2012–2022) were included; risk of bias was low-moderate. Pooled meta-analysis across three regions showed: cervical spinal cord had the largest FA reduction (3 studies; g = −1.80, p<0.001)optic radiation demonstrated robust FA reduction (3 studies; g = −0.84, p<0.001) and corpus callosum showed moderate FA reduction (2 studies; g = −0.58, p=0.03). MD and RD were consistently elevated across all regions, consistent with demyelination. In single-study analyses, normal-appearing white matter (g = −0.84) and association tracts/thalamic radiation also showed significant DTI abnormalities, indicating subclinical network-level involvement beyond visible lesions.  

 NMOSD causes microstructural white matter damage extending well beyond MRI-visible lesions. Cervical spinal cord & optic radiation show the most reproducible and largest pooled DTI abnormalities, supporting their use as sensitive, region-specific biomarkers for monitoring. NAWM involvement across multiple tracts suggests subclinical network damage that warrants prospective longitudinal investigation. Future studies with larger cohorts&standardized acquisition protocols are needed to enable robust multi-region meta-analysis