Abstract
BACKGROUND: Early diagnosis and differentiation of demyelinating diseases of the central nervous system (CNS) are essential for timely treatment but challenged with the limited availability of non-invasive, real-time methods to assess the architecture of the CNS. This study compared the retinal layers between patients with CNS demyelinating diseases and the general population and evaluated the predictive value of these structures in prospective diagnosis.</p>
METHODS: This UK Biobank study, incorporating optical coherence tomography images, analyzed patients with CNS demyelinating diseases identified at recruitment and during follow-up. Cross-sectionally, baseline retinal structures were compared, with diagnostic models constructed following cross-validation. Cox regression was used to assess the risk of future diagnosis in the prospective cohort.</p>
RESULTS: 34,230 individuals were included, comprising 61 diagnosed patients. The thickness of the macular ganglion cell-inner plexiform layer (mGCIPL; p = 4.91 × 10-10), papillary retinal nerve fiber layer (pRNFL; p = 9.92 × 10-6), and non-central macular subfields (p value ranging from 1.16 × 10-2 to 1.18 × 10-10) were significantly thinner in the patient group. A diagnostic model incorporating mGCIPL, pRNFL and outer temporal macular thickness achieved the area under the curve of 0.779. During follow-up, 96 patients were newly diagnosed. Multivariable Cox regression revealed thinner mGCIPL (HR: 0.960, 95 % CI: 0.936 to 0.984, p = 0.001) and thinner outer nasal macula (HR: 0.990 95 % CI: 0.983 to 0.997, p = 0.006) as high risk predictors of future diagnosis.</p>
CONCLUSIONS: Retinal structure can serve as non-invasive biomarker for CNS demyelinating diseases and has prospective diagnostic value in identifying pre-clinical and sub-clinical patients.</p>