In sporadic cerebral small vessel disease (SVD), blood neurofilament light chain (NfL) has been associated with disease severity and activity in cross-sectional studies. Baseline blood NfL levels were predictive of cognitive decline and dementia development in longitudinal studies [1-3]. Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is the most common hereditary cerebral SVD and is caused by mutations in the NOTCH3 gene. Moreover, blood NfL levels are associated with magnetic resonance imaging (MRI) markers of CADASIL-induced cerebral SVD, cognitive deficits, and disability [1,4,5]. However, our knowledge of the longitudinal change in blood NfL levels in patients with SVD is limited due to the small number of such studies [6,7]. In CADASIL, SVD has an early onset, occurring even in patients in their 20s, and significant changes are seen even after only 3 years of follow-up [8,9]. Disease progression is known to vary widely among patients with CADASIL [10]. Therefore, CADASIL could be a useful target for investigating longitudinal changes. In this study, we measured serum NfL levels at baseline and after 3 years to investigate the clinical significance of longitudinal changes in serum NfL levels in patients with CADASIL.

This was a single-center prospective study. We consecutively enrolled patients with CADASIL who were followed up regularly in the outpatient clinic at Jeju National University Hospital, Jeju, South Korea, from October 2018 to October 2019. We assessed clinical symptoms, cognitive function, brain MRI findings, and serum NfL levels at baseline and after a 3-year follow-up. We also enrolled an equal number of age- and sex-matched healthy individuals with no history of stroke as controls. This study was approved by the Institutional Review Board of Jeju National University Hospital (IRB File No. JEJUNUH 2018-08-003).

Serum levels of NfL were measured using a commercially available Simoa reagent kit (NF-light^TM Advantage Kit) on the Simoa HD-X Analyzer (Quanterix, Lexington, MA, USA). For cognitive function assessment, the Korean version of the Mini-Mental State Examination (K-MMSE) and the Seoul Neuropsychological Screening Battery (SNSB) were used. The SNSB global cognitive function (GCF) score is a composite score that was developed and validated based on the SNSB-Dementia version, building upon the foundation of the original SNSB. The modified Rankin Scale (mRS) score was used to assess the overall disability status of the patients. The Clinical Dementia Rating Scale (CDR) and sum of boxes (CDR-SB) were used to assess the severity of dementia. The Korean version of the National Institutes of Health Stroke Scale was used to assess neurological deficits related to cerebrovascular presentation in patients with CADASIL. All MRI scans were obtained with a 3-T scanner (Achieva, Philips Healthcare, Best, The Netherlands) and performed at baseline and after 3 years of follow-up. Lacunes, cerebral microbleeds (CMB), white matter hyperintensity (WMH) volume, and cortical thickness were analyzed. We used repeated measures correlations between changes in serum NfL levels and changes in cognitive function or MRI markers during follow-up. Linear mixed-effects models were used additionally for correlating changes in serum NfL levels and cognitive function or MRI markers while adjusting for other variables. For ordinal outcomes, we used multilevel mixed-effects ordered logistic regression models. Detailed methods for serum NfL level measurements, MRI parameters, and statistical analysis were provided in Supplementary Methods.

We examined 63 patients with CADASIL at baseline. However, baseline serum NfL levels could not be measured in three patients because of red blood cell contamination. One patient showed an extreme value and was excluded as an outlier based on the boxplot distribution and high z-score (10.75). Consequently, 59 patients and matched controls were included in the analysis. Compared with healthy participants (control), patients with CADASIL were slightly older and had a more frequent history of hypertension and hypercholesterolemia (Supplementary Table 1). Median serum NfL levels were significantly higher in patients with CADASIL than in healthy participants (17.36 pg/mL [11.93-24.45] vs. 12.73 pg/mL [9.74-20.36], P=0.01). However, the distribution of serum NfL largely overlapped between patients with CADASIL and healthy participants (Supplementary Figure 1). This difference remained significant after adjusting for age, body mass index (BMI), and estimated glomerular filtration rate (eGFR) (P=0.003). The characteristics of the 59 patients are shown in Supplementary Table 2.

At baseline, log-transformed serum NfL levels significantly correlated with age, BMI, eGFR, and serum high-density lipoprotein (HDL) cholesterol levels in patients with CADASIL (Supplementary Figure 2). Among the brain MRI markers, only cortical thickness showed a significant inverse correlation (R²=0.217, P<0.001) (Figure 1). Disability and cognitive scores, including the K-MMSE, GCF, mRS, CDR, and CDR-SB scores, also showed significant correlations with baseline serum NfL levels (Figure 2). Of the 59 patients examined at baseline, 51 completed the 3-year follow-up (median with interquartile range: 36.4 [34.5-37.0] months). There were significant changes in serum NfL levels (median with interquartile range, 17.1 [11.3-23.4] pg/mL vs. 24.2 [13.5-40.0] pg/mL, P<0.001) corresponding to an approximate 14% increase per year. Significant changes were also noted in the K-MMSE, CDR-SB, and GCF scores, and all brain MRI markers during the 3-year follow-up (Supplementary Table 3). Changes in the GCF score significantly correlated with changes in cortical thickness (rho=0.449, P=0.002). Changes in serum NfL levels significantly correlated with changes in GCF score (rho=-0.353, P=0.012), cortical thickness (rho=-0.358, P=0.013), WMH volume (rho=0.442, P=0.003), lacune count (rho=0.279, P=0.034), and CMB count (rho=0.558, P<0.001) in repeated measures correlations (Figure 3). The correlations remained significant after adjusting for age, BMI, eGFR, and serum HDL levels in the linear mixed-effects model (Table 1).

In this longitudinal study, we observed for the first time a significant increase in serum NfL levels in patients with CADASIL over a 3-year follow-up. These changes were independently associated with worsening global cognitive function and progression of MRI markers for SVD, including cortical thickness, WMH volume, lacune count, and CMB count. The study was limited by a small sample size and the presence of a predominant genetic variant (p.Arg544Cys) in >90% of the cases. Therefore, our findings may not be applicable to other patients with CADASIL or those with the common sporadic cerebral SVD.