Retinal Microvasculature Emerges as a Key Differentiator Between MS and NMOSD

A new study published in the Journal of Neurology highlights distinctive retinal microvascular patterns that can help differentiate multiple sclerosis (MS) from neuromyelitis optica spectrum disorder (NMOSD)—two neurologic conditions that often present with overlapping features such as optic neuritis but demand sharply divergent treatment approaches.

Using advanced imaging tools, researchers conducted a cross-sectional study involving 658 participants, including 167 individuals with MS, 221 with NMOSD, and 270 age- and sex-matched healthy controls. A total of 1,277 eyes were examined using swept-source optical coherence tomography (OCT) and OCT angiography (OCTA), enabling detailed analysis of both retinal structure and microvascular integrity.

The investigators quantified several microvascular metrics, including vessel density, perfusion, and microdensity across the superficial and deep vascular complexes (SVC and DVC), as well as choroidal vascular volume (CVV) and stromal volume (CSV). Structural metrics included the thickness of the retinal nerve fiber layer (RNFL) and the ganglion cell–inner plexiform layer (GCIPL).

Findings revealed disease-specific patterns in retinal vascular changes, especially when stratifying eyes based on history of optic neuritis (ON). In eyes affected by ON, NMOSD patients exhibited significantly lower SVC metrics compared to MS patients, indicating more severe superficial microvascular damage. Conversely, in non-ON eyes, MS patients showed greater microvascular loss than those with NMOSD—a reversal of the ON-associated pattern. Most of these differences reached statistical significance (p < 0.017).

Moreover, the extent of microvascular decline was closely linked to neurological disability as measured by the Expanded Disability Status Scale (EDSS). In ON eyes, the rate of SVC decline with increasing EDSS scores was steeper in NMOSD than in MS, while in non-ON eyes, MS patients demonstrated a more pronounced EDSS-related vascular loss. These interaction effects were significant across most comparisons (interaction p < 0.05).

To translate these findings into clinical utility, the authors developed both logistic regression and machine learning models to distinguish MS from NMOSD using the OCTA-derived microvascular and structural data. A logistic regression model incorporating only microvascular features achieved an area under the curve (AUC) of 0.900, indicating high diagnostic accuracy. Among various machine learning classifiers tested, the support vector machine (SVM) performed best, achieving an AUC of 0.912 and an overall classification accuracy of 84.5%.

The results position retinal microvascular integrity as a promising non-invasive biomarker capable of aiding differential diagnosis between MS and NMOSD, particularly in cases where conventional imaging or serological markers may be inconclusive. According to the authors, the combined use of OCT and OCTA provides not only insight into structural damage but also captures the vascular underpinnings of neuroinflammation in a disease-specific manner.

These findings support the growing role of retinal imaging in neurology and highlight how OCTA-based diagnostic models can offer both precision and practicality in clinical settings, especially when calibrated to account for optic neuritis history.