Abstract
Background and Objectives
The NOTCH3-SVD staging system was developed to characterize NOTCH3-related small vessel disease (SVD), but it has not been validated in cohorts carrying a single pathogenic variant. We applied this system to Taiwanese individuals with the NOTCH3 p.R544C variant to evaluate its clinical relevance and prognostic value.
Methods
We enrolled individuals carrying the NOTCH3 p.R544C variants from 2 sources: the Taiwan Precision Medicine Initiative, a hospital-based volunteer cohort undergoing genetic screening, and the Taiwan CADASIL Registry, which includes individuals with symptomatic SVD and confirmed NOTCH3 pathogenic variants. Participants were classified using the NOTCH3-SVD staging system, ranging from stage 0 (premanifest stage) to stage 4B (end stage). Baseline characteristics were compared across stages. Multivariable models were used to identify factors associated with prior stroke or cognitive impairment. Stroke-free survival was analyzed using Kaplan-Meier curves and Cox proportional hazards models. Cognitive decline, assessed by Mini-Mental State Examination, was evaluated using a generalized estimating equation.
Results
Among 260 individuals (median age 62 years; 49% male), the median stage was 2A. Higher stages were positively associated with prior stroke, cognitive impairment, gait disturbance, and psychiatric symptoms and inversely associated with headache (all p values < 0.05). Fewer years of education (OR 0.90, 95% CI 0.83–0.98, p = 0.012), hypertension (OR 2.34, 95% CI 1.18–4.67, p = 0.016), and higher NOTCH3-SVD stage (OR 3.70 per 1-substage increase, 95% CI 2.61–5.25, p < 0.001) were significantly associated with prior stroke or cognitive impairment. During a median follow-up of 1.9 years, individuals with stage ≥2B had a higher risk of incident stroke than those with stage <2B (annual risk 6.7% vs 2.0%, log-rank p = 0.023; adjusted hazard ratio 3.38; 95% CI 1.10–10.4, adjusted for age and hypertension). MMSE scores declined progressively over 2 years in individuals with stage ≥2B, whereas those with stage <2B remained cognitively stable (p for interaction = 0.024).
Discussion
The NOTCH3-SVD staging system effectively stratified disease burden and predicted incident stroke and cognitive decline in individuals with NOTCH3 p.R544C, with stage ≥2B indicating a higher risk.
Introduction
Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is the most common monogenic cause of cerebral small vessel disease (SVD),1 caused by missense variants that alter cysteine residues within the 34 epidermal growth factor–like repeat (EGFr) domains of the NOTCH3 protein.2 The clinical presentation of CADASIL varies depending on the location of the cysteine-altering NOTCH3 variant.3 Variants in EGFr domains 1–6 are more strongly associated with the classical CADASIL phenotype,4 characterized by recurrent strokes, cognitive decline, migraine with aura, psychiatric symptoms, and motor disabilities.4 By contrast, variants in EGFr domains 7–34 are typically associated with milder symptoms or may remain asymptomatic until late adulthood,4,5 reflecting the marked phenotypic heterogeneity of NOTCH3 cysteine-altering variants.
To better characterize this clinical and radiologic spectrum, one international cohort study6 recently proposed the NOTCH3-SVD staging system, which classifies disease into 5 main stages (0–4) and 9 substages (1 A–4B). This system provides a standardized framework for disease classification and monitoring in both research and clinical settings. However, variant-specific prognostication still warrants further investigation. Although the study included 22 independent cohorts worldwide, few were from Asian populations, where prevalent variants such as p.R544C and p.R75P are more common and often associated with hemorrhagic manifestations.7,8 Moreover, this staging system was developed primarily based on neuroimaging findings and functional dependence, with limited integration of clinical manifestations. Therefore, the applicability of this stating system to specific NOTCH3 variants and its correlation with clinical outcomes remains to be validated.
In Taiwan, the NOTCH3 p.R544C variant, a mutational hotspot located between EGFr domains 13 and 14, accounts for approximately 70% of CADASIL cases.9 Patients with stroke who harbored this variant were more likely to present with ischemic strokes of the small vessel occlusion type10,11 and intracerebral hemorrhage.12,13 Population-based analyses from the Taiwan Biobank further indicate that approximately 0.9% of community-dwelling Taiwanese carry this variant, with carriers exhibiting more than a twofold increased risk of stroke compared with noncarriers,11,14 highlighting its clinical relevance even among asymptomatic or mildly affected individuals. Despite its high prevalence and potential clinical impact, the natural history of this variant remains poorly characterized because of limited longitudinal data.
In this study, we applied the NOTCH3-SVD staging system6 to Taiwanese individuals carrying the NOTCH3 p.R544C variant to characterize clinical and imaging heterogeneity and to identify predictors of disease progression.
Methods
Study Participants
We included individuals carrying the NOTCH3 p.R544C variant from 2 distinct hospital-based cohorts beginning in February 2019. The first source was the prospective Taiwan CADASIL Registry (TCR) cohort, which has consecutively enrolled individuals with genetically confirmed cysteine-altering NOTCH3 variants at National Taiwan University Hospital (NTUH) since February 2019.15,16 Genetic testing was offered based on either (1) neuroimaging features suggestive of SVD, with or without associated clinical manifestations (e.g., stroke, cognitive impairment, gait disturbance, psychiatric symptoms, or migraine); or (2) a family history of cysteine‐altering NOTCH3 variants. For this study, individuals from the TCR were excluded if they carried NOTCH3 variants other than p.R544C or underwent genetic testing solely based on a family history of a confirmed index case without clinical symptoms.
The second source was the Taiwan Precision Medicine Initiative (TPMI),17 a nationwide project launched by Academia Sinica in collaboration with 16 medical centers across Taiwan in July 2019.18 The TPMI enrolled outpatient volunteers who provided consent for genetic testing using residual blood samples and for access to their electronic medical records. Genetic profiling was performed using Han Chinese–optimized single-nucleotide polymorphism arrays that included the NOTCH3 p.R544C variant.18 As a hospital-based screening program, TPMI enrolled individuals regardless of NOTCH3-associated clinical manifestations at enrollment. Individuals carrying the NOTCH3 p.R544C variant who were subsequently invited for detailed evaluations at the neurology clinics of Taichung Veterans General Hospital (TCVGH)19 or NTUH were included in this study.
Clinical Assessment
Baseline demographic and clinical data were collected at enrollment, including age, sex, years of education, Mini-Mental State Examination (MMSE) and modified Rankin Scale (mRS) scores, vascular risk factors, and parental and sibling history of stroke. CADASIL-related clinical manifestations were assessed using standardized questionnaires, including stroke, cognitive impairment, gait disturbance, psychiatric symptoms, and headache history. A stroke event, either ischemic or hemorrhagic, was defined as a new cerebrovascular event confirmed by computed tomography or magnetic resonance imaging (MRI) demonstrating a corresponding infarct or intracranial hemorrhage (ICH). Cognitive impairment was defined based on education-specific MMSE cutoff: <24 for individuals with more than 6 years of education, and <21 for those with 6 years or fewer.20-22 Cognitive decline was defined as a transition from intact cognition at baseline to impairment at the most recent follow-up.
Neuroimaging Assessment
Brain MRI scans were performed during outpatient visits after enrollment in both cohorts using harmonized imaging protocols between the 2 hospitals. The required sequences included T1-weighted, gradient-echo or susceptibility-weighted imaging (SWI), diffusion-weighted imaging (DWI), and T2 fluid-attenuated inversion recovery sequences for neuroimaging analysis. All scans were independently reviewed by senior investigators with over 10 years of experience in image interpretation (Y.W.C. and C.H.C. at NTUH; H.C.C. at TCVGH).
MRI markers of SVD, including white matter hyperintensities, cerebral microbleeds (CMBs), and lacunes, were assessed according to the STandards for ReportIng Vascular changes on nEuroimaging criteria.23 DWI-positive lesions were defined as hyperintense on DWI with corresponding hypointensity or isointensity on the apparent diffusion coefficient maps. Cerebral macrohemorrhages were defined as bleeds >10 mm in diameter on SWI.24 The severity of periventricular white matter (PVWM) and deep white matter (DWM) lesions was graded using the Fazekas scale,25 and involvement of the anterior temporal lobe and external capsule was also evaluated.
Longitudinal Follow-Up and NOTCH3-SVD Staging
Participants underwent regular outpatient follow-up at each hospital, with MRI and MMSE assessments conducted every 1–2 years. Incident stroke, deaths, or loss to follow-up were documented. Clinical and outcome data were collected and locked on December 31, 2024, which defined the end of follow-up (administrative censoring date) in this study.
The NOTCH3-SVD staging system,6 comprising 9 substages within 5 main stages (0–4), was applied at baseline to evaluate phenotypic variation and to predict stroke-free and cognitive decline–free survival. Stage 0 denotes carriers without SVD features (Fazekas DWM 0, no lacunes). Stage 1A and 1B represent early disease (Fazekas DWM 1 and ≥2). Stage 2A and 2B indicate intermediate disease (≥1 and ≥5 lacunes). Stage 3A and 3B reflect advanced disease (mRS scores 3 and 4) while stage 4A and 4B correspond to end-stage disease (mRS scores 5 and 6).
Cognitive decline–free survival was defined as remaining free from cognitive decline during follow-up, excluding individuals with cognitive impairment at baseline.
Statistical Analysis
Continuous variables were presented as mean ± SD or median with interquartile range (IQR) according to their distribution and categorical variables as counts and percentages. Group comparisons across NOTCH3-SVD stages were conducted using one-way analysis of variance or the Kruskal-Wallis test for continuous variables and the χ2 test for categorical variables, as appropriate. The Cochran-Armitage trend test was applied to evaluate trends in clinical manifestations across increasing stages.
Two multivariable models were constructed: Model 1 included age, education, vascular risk factors (hypertension, diabetes, hyperlipidemia, and smoking), sibling history of stroke, and MRI markers (number of lacunes, number of CMBs, Fazekas DWM score, and external capsule involvement); Model 2 replaced individual imaging variables with the ordinal NOTCH3-SVD stage. Factors associated with prior stroke or cognitive impairment were first identified using univariable logistic regression; variables with p < 0.05 were entered into forward stepwise multivariable models (entry p < 0.1, removal p > 0.05). Highly correlated variables were excluded to avoid multicollinearity.
The NOTCH3-SVD stage was further dichotomized as <2B vs ≥2B, based on the overall median stage of 2A. Stroke-free survival and cognitive decline–free survival after baseline assessment were compared using Kaplan-Meier analysis with log-rank tests. Cox proportional hazards models were applied to estimate hazard ratios and 95% confidence intervals (CIs) for incident stroke, both unadjusted and adjusted for age26,27 and hypertension.16,27 Participants were followed until the earliest of incident stroke, death, loss to follow-up, or administrative censoring. Those remaining event-free were right-censored in survival analyses.
To further assess cognitive trajectories, longitudinal changes in MMSE scores were visualized using locally estimated scatterplot smoothing curves by stage. A generalized estimating equation (GEE) model with a time-by-stage interaction term was used to evaluate stage-dependent cognitive decline. Multivariable logistic regression models were also applied to identify predictors of cognitive decline, incorporating variables with p < 0.05 in univariable analyses, along with age26,27 and hypertension,16,27 which were included regardless of statistical significance.
All statistical tests were two-tailed, with p values <0.05 considered statistically significant. Statistical analyses were performed using Statistical Analysis System software (version 9.4) and RStudio (version 2025.05.0 + 496).
Standard Protocol Approvals, Registrations, and Patient Consents
This study was approved by the Institutional Review Board of the NTUH (No. 201807044RIND) and TCVGH (CE16270B-1 and CE23035B). Consents were obtained from the participants who signed the informed consent form after a thorough explanation of the study and its protocols.
Data Availability
Deidentified data are available on reasonable request.
Results
Characteristics of Individuals With NOTCH3 p.R544C Across NOTCH3-SVD Stages
We enrolled 145 individuals from the TPMI cohort and 167 from the TCR. After excluding 51 TCR participants (due to the lack of the R544C variant or being asymptomatic relatives) and 1 TPMI participant with incomplete baseline MRI, 260 individuals (median age 62 [IQR 55.5–67.0] years; 49% male) carrying the NOTCH3 p.R544C variant were included. All participants completed baseline brain MRI and were evaluated using the NOTCH3-SVD staging system (Figure 1). Baseline clinical and imaging characteristics of both cohorts are summarized in eTable 1.
Figure 1. Flowchart Illustrating the Inclusion and Exclusion of Participants.
TCR = Taiwan CADASIL Registry; TPMI = Taiwan Precision Medicine Initiative.
Increasing NOTCH3-SVD stage was associated with older age, fewer years of education, lower MMSE scores, and a higher prevalence of hypertension, hyperlipidemia, smoking history, and parental and sibling history of stroke (all p < 0.05) (Table 1). At baseline, a substantial proportion of individuals exhibited clinical symptoms, including stroke (35%), cognitive impairment (12%), gait disturbance (37%), psychiatric symptoms (23%), and headache (29%). The prevalence of all symptoms increased with advancing stages (all p < 0.001), except for headache, which was more frequent at lower stages (p = 0.008).
Table 1.
Baseline Demographic and Neuroimaging Characteristics of Individuals With the NOTCH3 p.R544C Variant Across NOTCH3-SVD Stages
| Variables | Overall (n = 260) | Stage 0 (n = 19) | Stage 1 (n = 101) | Stage 2 (n = 109) | Stage 3 (n = 31) | p Value |
| Age at enrollment, y | 62.0 (55.5–67.0) | 45.0 (34.0–55.0) | 60.0 (55.0–63.0) | 63.0 (57.0–70.0) | 71.0 (67.0–74.0) | <0.001 |
| Male sex (%) | 128 (49) | 7 (37) | 46 (46) | 54 (50) | 21 (68) | 0.113 |
| Education, y | 12.0 (12.0–16.0) | 16.0 (12.0–16.0) | 14.0 (12.0–16.0) | 12.0 (9.0–16.0) | 12.0 (6.0–14.0) | 0.005 |
| MMSE score | 28.0 (26.0–29.0) | 29.0 (28.0–30.0) | 29.0 (28.0–30.0) | 28.0 (27.0–29.0) | 20.0 (14.0–24.0) | <0.001 |
| Medical history (%) | ||||||
| Hypertension | 125 (48) | 1 (5.3) | 36 (36) | 65 (60) | 23 (74) | <0.001 |
| Diabetes mellitus | 50 (19) | 1 (5.3) | 15 (15) | 25 (23) | 9 (29) | 0.089 |
| Hyperlipidemia | 141 (54) | 4 (21) | 50 (50) | 70 (64) | 17 (55) | 0.003 |
| Ever smoking | 43 (17) | 2 (11) | 10 (9.9) | 22 (20) | 9 (29) | 0.041 |
| Alcohol consumption | 11 (4.2) | 0 (0) | 3 (3.0) | 8 (7.3) | 0 (0) | 0.157 |
| Coronary artery disease | 22 (8.5) | 0 (0) | 5 (5.0) | 15 (14) | 2 (6.5) | 0.058 |
| Atrial fibrillation | 15 (5.8) | 0 (0) | 6 (5.9) | 5 (4.6) | 4 (13) | 0.226 |
| Parental history of stroke | 109 (42) | 2 (11) | 33 (33) | 60 (55) | 14 (45) | <0.001 |
| Sibling history of stroke | 57 (22) | 0 (0) | 11 (11) | 37 (34) | 9 (29) | <0.001 |
| Clinical manifestation (%) | ||||||
| Prior stroke | 92 (35) | 0 (0) | 8 (7.9) | 60 (55) | 24 (77) | <0.001 |
| Cognitive impairment | 30 (12) | 0 (0) | 4 (4.0) | 8 (7.3) | 18 (62) | <0.001 |
| Gait disturbance | 95 (37) | 0 (0) | 20 (20) | 46 (42) | 29 (94) | <0.001 |
| Psychiatric symptoms | 61 (23) | 1 (5.3) | 14 (14) | 28 (26) | 18 (58) | <0.001 |
| Headache | 76 (29) | 7 (37) | 34 (34) | 34 (31) | 1 (3.2) | 0.008 |
| MRI characteristics | ||||||
| DWI lesion (%) | 20 (7.7) | 0 (0) | 0 (0) | 12 (11) | 8 (26) | <0.001 |
| Macrohemorrhage (%) | 39 (15) | 1 (5.3) | 6 (5.9) | 20 (18) | 12 (39) | <0.001 |
| CMB number | 2.0 (0.0–14.0) | 0.0 (0.0–0.0) | 0.0 (0.0–1.0) | 8.0 (2.0–25.0) | 30.0 (13.0–75.0) | <0.001 |
| Lacune number | 1.0 (0.0–6.0) | 0.0 (0.0–0.0) | 0.0 (0.0–0.0) | 4.0 (2.0–8.0) | 9.0 (6.0–15.0) | <0.001 |
| Fazekas PVWM score | 2.0 (2.0–3.0) | 0.0 (0.0–0.0) | 2.0 (1.0–2.0) | 3.0 (2.0–3.0) | 3.0 (3.0–3.0) | <0.001 |
| Fazekas DWM score | 2.0 (1.0–3.0) | 0.0 (0.0–0.0) | 2.0 (1.0–2.0) | 3.0 (2.0–3.0) | 3.0 (3.0–3.0) | <0.001 |
| Anterior temporal involvement (%) | 58 (22) | 1 (5.3) | 13 (13) | 38 (35) | 6 (19) | <0.001 |
| External capsule involvement (%) | 100 (38) | 0 (0) | 15 (15) | 61 (56) | 24 (77) | <0.001 |
Abbreviations: CMB = cerebral microbleed; DWI = diffusion-weighted imaging; DWM = deep white matter; MMSE = Mini-Mental State Examination; PVWM = periventricular white matter; SVD = small vessel disease.
Values are presented as median (interquartile range) or count (percentage).
Regarding MRI findings, higher stages were associated with a greater number of CMBs and lacunes, higher Fazekas PVWM and DWM scores, and increased frequency of DWI-positive lesions and cerebral macrohemorrhage (all p < 0.001). The frequency of anterior temporal pole and external capsule involvement rose significantly with advancing stages (p < 0.001). Differences in MRI features across the NOTCH3-SVD substage are illustrated in Figure 2.
Figure 2. MRI Characteristics Across NOTCH3-SVD Substages.
(A) Prevalence of anterior temporal pole and external capsule involvement, DWI lesions, and macrohemorrhage. (B) Quantitative imaging markers, including the numbers of CMBs and lacunes, as well as Fazekas PVWM and DWM scores. CMBs = cerebral microbleeds; DWI = diffusion-weighted imaging; DWM = deep white matter; PVWM = periventricular white matter; SVD = small vessel disease.
NOTCH3-SVD Staging in TPMI and TCR Cohorts
The distribution of NOTCH3-SVD stages differed between the TPMI and TCR cohorts (eFigure 1A), with an overall median stage of 2A. TPMI participants were mostly at lower stages, while TCR participants were more frequently at higher stages (median stage 1B vs 2B).
Age-stratified analysis revealed a progressive increase in NOTCH3-SVD severity with advancing age in both cohorts (eFigure 1B). In TPMI, most individuals aged <60 years were in stages between 0 and 2A, with none reaching stage ≥3A. By contrast, the TCR cohort exhibited a more severe staging profile across all age groups, including those younger than 60 years. Among those aged ≥70 years in the TCR, 30 of 43 (70%) were classified as stage ≥2B, with a considerable proportion reaching stage ≥3A.
Correlation Between NOTCH3-SVD Stage and CADASIL-Related Clinical Symptoms
The prevalence of prior stroke, cognitive impairment, psychiatric symptoms, and gait disturbance increased significantly with higher NOTCH3-SVD stage (all p for trend ≤0.001), while headache was more common at lower stages (p for trend = 0.002; Figure 3). When stratified by cohort, the prevalence of prior stroke and gait disturbance increased significantly across stages in both TPMI and TCR (prior stroke: TPMI p for trend <0.001, TCR p = 0.010; gait disturbance: TPMI p = 0.039, TCR p < 0.001). By contrast, stage-related increases in cognitive impairment and psychiatric symptoms were observed only in the TCR cohort (p for trend <0.001 and p = 0.021, respectively). Headache declined significantly with higher stages in the TCR (p for trend = 0.013), but not in TPMI (p for trend = 0.633).
Figure 3. Correlation Between NOTCH3-SVD Stages and CADASIL-Related Clinical Symptoms.
(A–E) Left panels show the overall prevalence of clinical symptoms by NOTCH3-SVD stage. Right panels display stratified results by cohort (TPMI in orange; TCR in blue-green). The prevalence of (A) prior stroke, (B) cognitive impairment, (C) gait disturbance, and (D) psychiatric symptoms increased significantly with advancing NOTCH3-SVD stage, while (E) headache was more frequent at lower stages. Cochran-Armitage trend test p values are shown for the overall cohort and for each subgroup (TPMI and TCR). SVD = small vessel disease; TCR = Taiwan CADASIL Registry; TPMI = Taiwan Precision Medicine Initiative.
Factors Associated With Prior Stroke or Cognitive Impairment
As listed in Table 2, univariate analyses identified several factors significantly associated with prior stroke or cognitive impairment, indicative of overt clinical disease. These included older age, fewer years of education, hypertension, diabetes mellitus, hyperlipidemia, history of smoking, alcohol consumption, sibling history of stroke, greater numbers of CMBs and lacunes, macrohemorrhage, higher Fazekas PVWM and DWM scores, anterior temporal and external capsule involvement, and a higher NOTCH3-SVD stage (all p < 0.05).
Table 2.
Factors Associated With Either Prior Stroke or Cognitive Impairment in Individuals Carrying the NOTCH3 p.R544C Variant
| Variables | Univariate | Multivariable | ||||
| OR (95% CI) | p Value | Model 1a | p Value | Model 2b | p Value | |
| Adjusted OR (95% CI) | Adjusted OR (95% CI) | |||||
| Age at enrollment, y | 1.09 (1.06–1.13) | <0.001 | ||||
| Male sex | 1.08 (0.66–1.79) | 0.743 | ||||
| Education, y | 0.87 (0.81–0.93) | <0.001 | 0.92 (0.84–1.00) | 0.040 | 0.90 (0.83–0.98) | 0.012 |
| Hypertension | 4.56 (2.67–7.79) | <0.001 | 2.18 (1.09–4.36) | 0.027 | 2.34 (1.18–4.67) | 0.016 |
| Diabetes mellitus | 2.07 (1.11–3.86) | 0.022 | ||||
| Hyperlipidemia | 2.24 (1.34–3.74) | 0.002 | ||||
| Coronary artery disease | 0.86 (0.35–2.13) | 0.745 | ||||
| Atrial fibrillation | 1.02 (0.35–2.95) | 0.975 | ||||
| Ever smoking | 2.21 (1.14–4.28) | 0.019 | ||||
| Alcohol consumption | 7.42 (1.57–35.08) | 0.011 | ||||
| Parental history of stroke | 1.20 (0.73–1.99) | 0.469 | ||||
| Sibling history of stroke | 4.25 (2.27–7.93) | <0.001 | ||||
| CMB number | 1.04 (1.02–1.05) | <0.001 | ||||
| Macrohemorrhage | 19.69 (6.73–57.58) | <0.001 | ||||
| Lacune number | 1.39 (1.27–1.52) | <0.001 | 1.21 (1.10–1.33) | <0.001 | ||
| Fazekas PVWM score | 5.08 (3.19–8.09) | <0.001 | ||||
| Fazekas DWM score | 4.60 (3.02–6.98) | <0.001 | 2.02 (1.20–3.40) | 0.008 | ||
| Anterior temporal involvement | 1.20 (0.67–2.18) | 0.538 | ||||
| External capsule involvement | 9.79 (5.48–17.49) | <0.001 | 3.23 (1.59–6.61) | 0.001 | ||
| NOTCH3-SVD stage | 3.87 (2.78–5.40) | <0.001 | 3.70 (2.61–5.25) | <0.001 | ||
Abbreviations: CMB = cerebral microbleeds; DWM = deep white matter; OR = odds ratio; PVWM = periventricular white matter; SVD = small vessel disease.
Model 1 used stepwise selection (entry p < 0.1, removal p > 0.05) from age, years of education, hypertension, diabetes mellitus, hyperlipidemia, ever smoking, sibling history of stroke, number of CMBs, number of lacunes, Fazekas DWM score, and external capsule involvement. The final model included years of education, hypertension, number of lacunes, Fazekas DWM score, and external capsule involvement.
Model 2 used stepwise selection (entry p < 0.1, removal p > 0.05) from age, years of education, hypertension, diabetes mellitus, hyperlipidemia, ever smoking, sibling history of stroke, and NOTCH3-SVD stage. The final model included years of education, hypertension, and NOTCH3-SVD stage.
In Model 1, the following were independently associated with prior stroke or cognitive impairment: fewer years of education (OR 0.92, 95% CI 0.84–1.00, p = 0.040), hypertension (OR 2.18, 95% CI 1.09–4.36, p = 0.027), number of lacunes (OR 1.21, 95% CI 1.10–1.33, p < 0.001), Fazekas DWM score (OR 2.02, 95% CI 1.20–3.40, p = 0.008), and external capsule involvement (OR 3.23, 95% CI 1.59–6.61, p = 0.001).
In Model 2, which replaced imaging variables with the NOTCH3-SVD stage, fewer years of education (OR 0.90, 95% CI 0.83–0.98, p = 0.012), hypertension (OR 2.34, 95% CI 1.18–4.67, p = 0.016), and higher NOTCH3-SVD stage (OR 3.70 per 1-substage increase, 95% CI 2.61–5.25, p < 0.001) remained significantly associated with prior stroke or cognitive impairment.
Predictors of Incident Stroke and Stroke-Free Survival
In the cohort of 260 NOTCH3 p.R544C carriers, 57 were lost to follow-up. The remaining 203 participants (87 from TPMI, 116 from TCR) were followed for a median of 1.9 years (IQR 1.0–2.7). Two participants died during follow-up. Five homozygous carriers were identified, all from the TCR cohort, with 3 at stage 2B and 2 at stage 2A; none of them developed stroke or cognitive decline.
During follow-up, 16 participants experienced an incident stroke, including 11 (69%) with infarctions and 5 (31%) with ICH. The overall incidence rate of stroke was approximately 4.2 per 100 person-years (2.97 for infarction, 1.35 for ICH). As illustrated in eFigure 2, the 3-year cumulative incidence of infarction was 15.7% among individuals with stage ≥2B compared with 3.7% among those with stage <2B (relative risk [RR] 4.20, 95% CI 1.12–15.77), and 4.5% vs 2.2% for ICH (RR 2.06, 95% CI 0.34–12.39).
Kaplan-Meier analysis showed that the individuals with stage ≥2B had a significantly higher risk of incident stroke than those with stage <2B (annual risk 6.7% vs 2.0%; log-rank p = 0.023; Figure 4A). In Cox proportional hazards models, stage ≥2B was associated with a threefold higher risk of incident stroke after adjustment for age and hypertension (adjusted HR [aHR] = 3.36, 95% CI: 1.09–10.32, p = 0.035), although this association was attenuated after additional adjustment for prior stroke (Table 3).
Figure 4. Kaplan-Meier Analysis for Stroke-Free and Cognition Decline–Free Survival Stratified by NOTCH3-SVD Stage (<2B vs ≥2B).
(A) Over a median follow-up of 1.9 years, individuals with NOTCH3-SVD stage ≥2B had significantly lower stroke-free survival compared with those with stage <2B (log-rank p = 0.023). (B) No significant difference in cognitive decline–free survival was observed between the 2 groups over a median follow-up of 2.0 years (log-rank p = 0.251). SVD = small vessel disease.
Table 3.
Cox Proportional Hazards Analysis of Incident Strokes in Individuals Carrying the NOTCH3 p.R544C Variant
| Modela | Variables | HR (95% CI) | p Value |
| Model 1 | NOTCH3-SVD stage | 1.49 (1.02–2.18) | 0.039 |
| Model 2 | NOTCH3-SVD stage ≥2B | 3.36 (1.09–10.32) | 0.035 |
| Model 3 | Prior stroke | 7.81 (1.55–39.3) | 0.013 |
| NOTCH3-SVD stage | 1.17 (0.78–1.77) | 0.451 | |
| Model 4 | Prior stroke | 7.58 (1.55–38.14) | 0.013 |
| NOTCH3-SVD stage ≥2B | 1.85 (0.60–5.75) | 0.285 |
Abbreviations: OR = odds ratio; SVD = small vessel disease.
Total follow-up cases = 203, incident stroke cases = 16.
All models were adjusted for age and hypertension, with additional covariates specified for each model.
In cause-specific Cox models (eTable 2), higher NOTCH3-SVD stage was significantly associated with an increased risk of incident infarction after adjustment for age and hypertension (aHR = 1.67, 95% CI 1.04–2.68), but not with ICH (aHR = 1.21, 95% CI 0.62–2.35). When dichotomized at stage ≥2B, no significant associations were observed, likely due to limited statistical power from the small number of ICH events.
In the subgroup without prior stroke (n = 113; incident stroke cases = 2), NOTCH3-SVD stage was not significantly associated with incident stroke after adjustment for age and hypertension (aHR = 1.79, 95% CI 0.42–7.66, eTable 3), although the model was likely unstable because of the small number of events. The median NOTCH3-SVD stage was significantly higher in participants with prior stroke [2B (2 A–3A), n = 90] than in those without [1B (1 A–2A), p < 0.001].
Predictors of Cognitive Decline and Longitudinal Cognitive Trajectories
Among 260 individuals, 163 underwent at least 1 repeat MMSE assessment. After excluding 21 participants with baseline cognitive impairment, 142 were included in the final analysis for cognitive decline. Over a median follow-up of 2.0 years (IQR 1.4–2.3), 9 individuals developed cognitive decline. Kaplan-Meier analysis revealed a nonsignificant higher risk of cognitive decline in those with stage ≥2B compared with stage <2B (annual risk 6.6% vs 2.9%, log-rank p = 0.251; Figure 4B).
In univariate logistic regression, prior stroke (OR 5.81, 95% CI 1.16–29.07, p = 0.032) and higher NOTCH3-SVD stage (OR 1.87 per 1-substage increase, 95% CI 1.09–3.20, p = 0.023; eTable 4) were significantly associated with cognitive decline. However, these associations were attenuated after adjustment for age and hypertension (eTable 5).
Among participants with longitudinal MMSE assessments, those with stage ≥2B exhibited a progressive decline in MMSE scores over 2 years, while those with stage <2B remained cognitively stable. LOESS-smoothed curve illustrated this divergence in cognitive trajectories. A significant interaction between time and NOTCH3-SVD stage was observed in the GEE model (p for interaction = 0.024; Figure 5), indicating stage-dependent cognitive deterioration over time.
Figure 5. Longitudinal MMSE Trajectories Stratified by NOTCH3-SVD Stage 2B Threshold.
LOESS-smoothed curves with 95% CIs illustrate longitudinal changes in MMSE scores over a 2-year follow-up. Participants with NOTCH3-SVD stage ≥2B (n = 142) exhibited a progressive decline in MMSE scores, whereas those with stage <2B remained relatively stable. A significant time-by-stage interaction was observed (p for interaction = 0.024, GEE model). GEE = generalized estimating equation; LOESS = locally estimated scatterplot smoothing; MMSE = Mini-Mental State Examination; SVD = small vessel disease.
Discussion
In this study, we identified 3 key findings among Taiwanese carriers of the NOTCH3 p.R544C variant. First, the NOTCH3-SVD staging system was significantly associated with clinical manifestations, most of which worsened with advancing stage, whereas headache showed an inverse trend. Second, higher NOTCH3-SVD stages were independently associated with prior stroke or cognitive impairment. Third, the staging system also predicted incident stroke and cognitive decline, with stage ≥2B emerging as a clinically relevant threshold.
Notably, these stage-dependent patterns were observed not only in the symptomatic TCR cohort but also in the predominantly asymptomatic TPMI cohort. Despite the low prevalence of overt symptoms at baseline, TPMI participants with higher stages exhibited increased rates of prior stroke and gait disturbance. This consistency across both cohorts highlights the value of the NOTCH3-SVD staging system to capture subclinical disease progression and to identify individuals at higher risk, even among asymptomatic or paucisymptomatic carriers.
In contrast to a UK cohort in which migraine affected up to 75% of patients with symptomatic CADASIL and represented the initial symptom in 68%,28,29 fewer than 3% of individuals in a Taiwanese cohort presented with migraine,9 suggesting possible ethnic or variant-related differences. In this study, only 1 participant from the TCR cohort reported aura-like symptoms, and 1 from the TPMI cohort met the diagnostic criteria for migraine with aura. Given these very few cases, no meaningful association with outcomes could be assessed.
Of interest, headache was more prevalent in the asymptomatic TPMI cohort and declined with advancing stage, particularly in the symptomatic TCR cohort. Consistent with Western data, our findings showed that migraine was more frequently reported at earlier stages of CADASIL28-30; however, its clinical significance among individuals without overt clinical disease remains uncertain. Current evidence does not support a clear association between migraine and either disease severity29 or increased stroke risk in CADASIL.28
The apparent remission of migraine at more advanced stages may reflect ischemia-related alterations in cerebrovascular physiology. Migraine, particularly with aura, is linked to cortical spreading depression (CSD), characterized by a propagating wave of neuronal depolarization followed by transient hyperemia and subsequent prolonged oligemia.31 Recurrent or chronic ischemic injury may lead to permanent alterations in vascular reactivity and neurovascular coupling, thereby reducing susceptibility to CSD or vasodilation and, consequently, diminishing migraine occurrence in later stages of the disease.32
Regarding predictors of incident stroke among individuals with the NOTCH3 p.R544C variant, prior stroke emerged as a strong predictor in the adjusted model. This finding may be partly influenced by the composition of the TCR cohort, which predominantly includes symptomatic individuals, potentially amplifying the effect of prior stroke. Previous studies have shown that patients with TIA or ischemic stroke and a higher SVD burden were at greater risk of recurrence.33,34 Beyond aligning with previous findings, our results further suggested that advancing NOTCH3-SVD stage may independently confer a higher risk of incident stroke, even among neurologically intact p.R544C carriers.
Moreover, we found that stage ≥2B appeared to represent a clinically meaningful threshold for risk stratification. In our cohort, imaging markers showed a clear threshold-like escalation around stage 2B rather than a smooth linear increase across all stages. Individuals at this stage, corresponding to a lacune count >5, had a significantly higher risk of incident stroke and exhibited greater cognitive decline over time, supported by a significant time-by-stage interaction in longitudinal MMSE trajectories. These findings were consistent with prior studies showing that lacune burden predicts incident infarction and is associated with executive dysfunction and functional deterioration over a three-year period in NOTCH3 variant carriers.35
Our analysis applies the NOTCH3‑SVD staging system to individuals with the p.R544C variant, demonstrating its close association with clinical severity and outcomes. By integrating 2 complementary cohorts with longitudinal imaging and cognitive data, we validated the utility of the staging system in a genetically homogeneous population and demonstrated its prognostic value in both symptomatic and asymptomatic carriers.
This study has several limitations. First, hospital-based recruitment, particularly in the TPMI cohort, may have introduced selection bias, while loss to follow-up among younger and less affected TPMI participants (eTable 6) could have resulted in attrition bias and underestimation of long-term risks in milder cases. Second, the small sample size and the use of binary or semiquantitative imaging markers may have limited the detection of linear trends observed in the original validation study.6 Alternatively, this finding may indicate that imaging progression does not necessarily follow a linear trajectory across different NOTCH3 variants. Therefore, while the NOTCH3-SVD staging system provides a variant-agnostic framework, its applicability and prognostic performance may vary across genotypes. Further validation in carriers of other NOTCH3 variants and diverse populations is warranted to determine the generalizability of our findings. Finally, cognitive outcomes were limited by the short follow-up duration, the small number of incident events, and reliance on the MMSE score, which is insensitive to early executive dysfunction in CADASIL. Future studies incorporating more sensitive assessments, such as the Trail Making Test, are warranted.
The NOTCH3-SVD staging system effectively captures the clinical and radiologic heterogeneity among individuals carrying the p.R544C variant. Stage ≥2B was associated with an increased risk of incident stroke at a median follow-up of 1.9 years and with MMSE decline within 2 years. These findings suggest that this staging system may be useful for clinical monitoring and risk stratification, particularly in East Asian populations. Further validation across other NOTCH3 variants and with longer follow-up is needed to confirm its prognostic significance.
Acknowledgment
The authors thank all individuals involved in this study, especially the participants and investigators from the Taiwan Precision Medicine Initiative.
Glossary
- aHR
adjusted HR
- CADASIL
cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy
- CSD
cortical spreading depression
- CMB
cerebral microbleed
- DWI
diffusion-weighted imaging
- DWM
deep white matter
- GEE
generalized estimating equation
- ICH
intracranial hemorrhage
- IQR
interquartile range
- MMSE
Mini-Mental State Examination
- mRS
modified Rankin Scale
- NTUH
National Taiwan University Hospital
- PVWM
periventricular white matter
- RR
relative risk
- SVD
small vessel disease
- SWI
susceptibility-weighted imaging
- TCR
Taiwan CADASIL Registry
- TCVGH
Taichung Veterans General Hospital
- TPMI
Taiwan Precision Medicine Initiative
Author Contributions
Y-C. Shen: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. C-H. Chen: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. H-C. Chen: analysis or interpretation of data. Y-W. Cheng: drafting/revision of the manuscript for content, including medical writing for content; study concept or design. C-Y. Chang: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data. P-H. Kuo: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. Y-M. Chen: drafting/revision of the manuscript for content, including medical writing for content. W-J. Lee: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data. T. Sung-Chun: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design.
Study Funding
This work was supported by the Ministry of Science and Technology, Taiwan (Grants 113-2314-B-002-102, 113-2314-B-002-278, 112-2314-B-005-003-MY3, and 111-3114-Y-001-001); Taichung Veterans General Hospital, Taiwan (Grants TCVGH-YM1100108, TCVGH-YM1110101, and TCVGH-YM1120107); and National Taiwan University Hospital, Taiwan (Grant 109F005-110-O).
Disclosure
The authors report no relevant disclosures. Go to Neurology.org/NG for full disclosures.
References
- 1.Chabriat H, Joutel A, Dichgans M, Tournier-Lasserve E, Bousser MG. Cadasil. Lancet Neurol. 2009;8(7):643-653. doi: 10.1016/s1474-4422(09)70127-9 [DOI] [PubMed] [Google Scholar]
- 2.Boston G, Jobson D, Mizuno T, Ihara M, Kalaria RN. Most common NOTCH3 mutations causing CADASIL or CADASIL-like cerebral small vessel disease: a systematic review. Cereb Circ Cogn Behav. 2024;6:100227. doi: 10.1016/j.cccb.2024.100227 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Hack RJ, Gravesteijn G, Cerfontaine MN, et al. Three-tiered EGFr domain risk stratification for individualized NOTCH3-small vessel disease prediction. Brain. 2023;146(7):2913-2927. doi: 10.1093/brain/awac486 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Rutten JW, Dauwerse HG, Gravesteijn G, et al. Archetypal NOTCH3 mutations frequent in public exome: implications for CADASIL. Ann Clin Transl Neurol. 2016;3(11):844-853. doi: 10.1002/acn3.344 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rutten JW, Hack RJ, Duering M, et al. Broad phenotype of cysteine-altering NOTCH3 variants in UK Biobank: CADASIL to nonpenetrance. Neurology. 2020;95(13):e1835-e1843. doi: 10.1212/wnl.0000000000010525 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gravesteijn G, Rutten JW, Cerfontaine MN, et al. Disease severity staging system for NOTCH3-associated small vessel disease, including CADASIL. JAMA Neurol 2025;82(1):49-60. doi: 10.1001/jamaneurol.2024.4487 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ishiyama H, Kim H, Saito S, et al. Pro-hemorrhagic cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy associated with NOTCH3 p.R75P mutation with low vascular NOTCH3 aggregation property. Ann Neurol. 2024;95(6):1040-1054. doi: 10.1002/ana.26916 [DOI] [PubMed] [Google Scholar]
- 8.Chen CH, Chu YT, Chen YF, et al. Comparison of clinical and neuroimaging features between NOTCH3 mutations and nongenetic spontaneous intracerebral haemorrhage. Eur J Neurol. 2022;29(11):3243-3254. doi: 10.1111/ene.15485 [DOI] [PubMed] [Google Scholar]
- 9.Liao YC, Hsiao CT, Fuh JL, et al. Characterization of CADASIL among the han Chinese in taiwan: distinct genotypic and phenotypic profiles. PLoS One. 2015;10(8):e0136501. doi: 10.1371/journal.pone.0136501 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Tang SC, Chen YR, Chi NF, et al. Prevalence and clinical characteristics of stroke patients with p.R544C NOTCH3 mutation in Taiwan. Ann Clin Transl Neurol. 2019;6(1):121-128. doi: 10.1002/acn3.690 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lee YC, Chung CP, Chang MH, Wang SJ, Liao YC. NOTCH3 cysteine-altering variant is an important risk factor for stroke in the Taiwanese population. Neurology. 2020;94(1):e87-e96. doi: 10.1212/wnl.0000000000008700 [DOI] [PubMed] [Google Scholar]
- 12.Chen CH, Tang SC, Cheng YW, et al. Detrimental effects of intracerebral haemorrhage on patients with CADASIL harbouring NOTCH3 R544C mutation. J Neurol Neurosurg Psychiatry. 2019;90(7):841-843. doi: 10.1136/jnnp-2018-319268 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chen CH, Cheng YW, Zhang R, et al. Intracerebral hemorrhage in patients with CADASIL: additive impact of the NOTCH3 R544C variant and hypertension? Stroke. 2025;56(8):2159-2166. doi: 10.1161/strokeaha.124.050484 [DOI] [PubMed] [Google Scholar]
- 14.Lin HJ, Chen CH, Su MW, et al. Modifiable vascular risk factors contribute to stroke in 1080 NOTCH3 R544C carriers in Taiwan Biobank. Int J Stroke. 2024;19(1):105-113. doi: 10.1177/17474930231191991 [DOI] [PubMed] [Google Scholar]
- 15.Cheng YW, Chen CH, Chen YF, et al. Prevalence and characteristics of pathogenic variants in Taiwanese patients with cerebral small vessel disease. Neurol Genet 2025;11(3):e200258. doi: 10.1212/nxg.0000000000200258 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chen CH, Liao YC, Cheng YW, Chung CP, Lee YC, Tang SC. High daily diastolic blood pressure predicts incident stroke, lacune, and cerebral microbleeds in CADASIL. J Stroke. 2024;26(1):112-115. doi: 10.5853/jos.2023.02880 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Taiwan Precision Medicine Initiative. Academia Sinica; 2025. Accessed November 9, 2025. tpmi.ibms.sinica.edu.tw/www/ [Google Scholar]
- 18.Yang HC, Kwok PY, Li LH, et al. The Taiwan Precision Medicine Initiative provides a cohort for large-scale studies. Nature. 2025;648(8092):117-127. doi: 10.1038/s41586-025-09680-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tung H, Chou CC, Chen HM, et al. White matter hyperintensities and cognitive functions in people with the R544C variant of the NOTCH3 gene without stroke or dementia. Neurology. 2024;103(9):e209941. doi: 10.1212/wnl.0000000000209941 [DOI] [PubMed] [Google Scholar]
- 20.Cui GH, Yao YH, Xu RF, et al. Cognitive impairment using education-based cutoff points for CMMSE scores in elderly Chinese people of agricultural and rural Shanghai China. Acta Neurol Scand. 2011;124(6):361-367. doi: 10.1111/j.1600-0404.2010.01484.x [DOI] [PubMed] [Google Scholar]
- 21.Li H, Jia J, Yang Z. Mini-mental state examination in elderly Chinese: a population-based normative study. J Alzheimers Dis. 2016;53(2):487-496. doi: 10.3233/jad-160119 [DOI] [PubMed] [Google Scholar]
- 22.Yang Z, Holt HK, Fan JH, et al. Optimal cutoff scores for Alzheimer's disease using the Chinese version of mini-mental state examination among Chinese population living in rural areas. Am J Alzheimers Dis Other Demen. 2016;31(8):650-657. doi: 10.1177/1533317516662336 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Duering M, Biessels GJ, Brodtmann A, et al. Neuroimaging standards for research into small vessel disease-advances since 2013. Lancet Neurol. 2023;22(7):602-618. doi: 10.1016/s1474-4422(23)00131-x [DOI] [PubMed] [Google Scholar]
- 24.Ungvari Z, Tarantini S, Kirkpatrick AC, Csiszar A, Prodan CI. Cerebral microhemorrhages: mechanisms, consequences, and prevention. Am J Physiol Heart Circ Physiol. 2017;312(6):H1128-h1143. doi: 10.1152/ajpheart.00780.2016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Fazekas F, Barkhof F, Wahlund LO, et al. CT and MRI rating of white matter lesions. Cerebrovasc Dis. 2002;13(suppl 2):31-36. doi: 10.1159/000049147 [DOI] [PubMed] [Google Scholar]
- 26.Zhang R, Ouin E, Grosset L, et al. Elderly CADASIL patients with intact neurological status. J Stroke. 2022;24(3):352-362. doi: 10.5853/jos.2022.01578 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Li WX, Liu ZY, Yao M, Zhu YC. Modifiable vascular risk factors for outcomes in patients with CADASIL: a literature review and meta-analysis of cross-sectional analyzes. J Stroke Cerebrovasc Dis. 2023;32(12):107437. doi: 10.1016/j.jstrokecerebrovasdis.2023.107437 [DOI] [PubMed] [Google Scholar]
- 28.Tan RY, Markus HS. CADASIL: migraine, encephalopathy, stroke and their inter-relationships. PLoS One. 2016;11(6):e0157613. doi: 10.1371/journal.pone.0157613 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Guey S, Mawet J, Hervé D, et al. Prevalence and characteristics of migraine in CADASIL. Cephalalgia. 2016;36(11):1038-1047. doi: 10.1177/0333102415620909 [DOI] [PubMed] [Google Scholar]
- 30.Szymanowicz O, Korczowska-Łącka I, Słowikowski B, et al. Headache and NOTCH3 gene variants in patients with CADASIL. Neurol Int. 2023;15(4):1238-1252. doi: 10.3390/neurolint15040078 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Liem MK, Oberstein SA, van der Grond J, Ferrari MD, Haan J. CADASIL and migraine: a narrative review. Cephalalgia 2010;30(11):1284-1289. doi: 10.1177/0333102410370870 [DOI] [PubMed] [Google Scholar]
- 32.Gryglas A, Smigiel R. Migraine and stroke: what's the link? What to do? Curr Neurol Neurosci Rep. 2017;17(3):22. doi: 10.1007/s11910-017-0729-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Lau KK, Li L, Schulz U, et al. Total small vessel disease score and risk of recurrent stroke: validation in 2 large cohorts. Neurology. 2017;88(24):2260-2267. doi: 10.1212/wnl.0000000000004042 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Haidegger M, Klock N, Kneihsl M, et al. Recurrent cerebrovascular events after recent small subcortical infarction. J Neurol. 2024;271(8):5055-5063. doi: 10.1007/s00415-024-12460-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Ling Y, De Guio F, Duering M, et al. Predictors and clinical impact of incident lacunes in cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy. Stroke. 2017;48(2):283-289. doi: 10.1161/strokeaha.116.015750 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Deidentified data are available on reasonable request.