Abstract Number: Esoc2026a1390 Less Is More: The Notch3-SVD Staging System Versus Serum Neurofilament Light as a Predictor for Disease Progression in Cadasil

Abstract Background and aims Disease progression in the genetic small vessel disease (SVD) CADASIL is highly variable, and was recently captured using the NOTCH3-SVD staging system: a tool for uniform disease staging using routine clinical and neuroimaging measures. Previously, serum Neurofilament Light-chain (sNfL) has been shown to predict disease progression in CADASIL, but requires expensive ultrasensitive assays. Here, we examined the association between sNfL and disease progression in a 2-year follow-up study of CADASIL patients, accounting for the NOTCH3-SVD stages. Methods sNfL values were determined in 55 patients at baseline and 2-year follow-up, using Simoa. Associations of incident ischemic stroke, lacunes, and microbleeds, with baseline sNfL (model 1), sNfL and age (model 2), and sNfL, age and NOTCH3-SVD stage (model 3) were determined using generalized linear models, and expressed as odds ratios (OR, for 1 sd or stage). Results Baseline sNfL values were predictive for 2-year incident lacunes and microbleeds, but not for stroke (model 1). The predictive effect of sNfL decreased after adjusting for age (model 2). In a combined model with NOTCH3-SVD stage and age (model 3), sNfL values were not predictive of incident events, whereas NOTCH3-SVD stage was predictive of incident stroke, incident lacunes, and incident microbleeds. Conclusions In CADASIL, a clinico-neuroradiological staging system outperforms the ultrasensitive biomarker sNfL in disease prediction. This study illustrates that the predictive value of the sNfL is mediated through age and NOTCH3-SVD stage in CADASIL, and highlights the importance of accounting for NOTCH3-SVD stage in CADASIL biomarker studies. Conflict of interest G. Gravesteijn, R.J. Hack, M.N. Cerfontaine, M.J.A Koel-Simmelink, C. Teunissen, J.W. Rutten, S.A.J. Lesnik Oberstein: nothing to disclose relevant to this project.