Множественный склероз Мисс диагностика в церебральной автономной доминантной артериопатии с субкортическими инфарктами и лейкоэнцефалопатией: исследование генотипа-миспенотипа.

Introduction

Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL), the most common inherited small vessel disease, is caused by pathogenic variants in the NOTCH3 gene on chromosome 19. It is characterized by migraine, recurrent ischemic strokes, and progressive cognitive decline. 1 Despite these characteristic features, CADASIL can be misdiagnosed as multiple sclerosis (MS). 2-4 The diagnostic confusion stems from overlapping clinical and radiologic features such as migraine, transient neurologic symptoms, and widespread white matter lesions on MRI, which, when combined with an absent or unrecognized stroke history, can push clinicians toward an MS diagnosis. 2,3

The diagnostic challenge is further compounded by the fact that CADASIL may initially lack classic temporal pole or external capsule involvement, leading to misclassification as demyelinating disease. 4 In addition, cysteine-altering NOTCH3 variants have been linked to increased susceptibility to autoimmune disorders, which may reinforce an erroneous MS diagnosis. 5 A subsequent CADASIL diagnosis is often made after progressive ischemic events, characteristic white matter changes, or as clear family history emerges. 2-5

Understanding why CADASIL patients are first diagnosed with MS and which factors prompt correction to a diagnosis of CADASIL is essential for improving diagnostic accuracy. Misdiagnosis not only delays appropriate genetic counseling and vascular risk management but also may expose patients to unnecessary therapies. In this study, we aimed to estimate the proportion of patients with CADASIL who had a prior MS misdiagnosis within a multicenter tertiary-care cohort to examine the clinical, radiologic, and genetic features associated with diagnostic confusion and identify the drivers of initial misdiagnosis. 6

Methods

Baseline Assessment

Baseline was defined as the date CADASIL was confirmed by genetic testing (n = 98) or skin biopsy (n = 9). At baseline, demographics (age, sex, and race), medical history (stroke, hypertension, hyperlipidemia, diabetes, and smoking), and family history (CADASIL, MS, migraine, stroke, or dementia) were collected. Functional status was assessed using the modified Rankin Scale ([mRS], 0–6, from no symptoms to death). 7 Disease severity was staged using the NOTCH3 Small Vessel Disease (N3SVD) system, which incorporates deep white-matter hyperintensity burden (Fazekas 0–2), lacune count (1–4 vs ≥5), and functional decline (mRS 3–6). 8 Genetic testing for NOTCH3 was performed when CADASIL was clinically suspected, not at the time of the MS misdiagnosis. For patients who underwent genetic testing, NOTCH3 variants were classified by exon and epidermal growth factor–like repeat (EGFr) domain and stratified into high-risk, medium-risk, low-risk, or undetermined-risk categories according to published criteria. 9

The unified electronic medical record (Epic) was reviewed to determine whether a diagnosis of MS had been documented at any time prior to CADASIL confirmation. Patients with such a diagnosis were classified as the MS misdiagnosis group, while all others comprised the non-misdiagnosed group. Baseline demographic, clinical, neuroimaging, functional, and genetic variables were compared descriptively between these groups. Among patients with a prior MS misdiagnosis, we additionally abstracted the facility at which the MS misdiagnosis was first documented and the Mayo Clinic site at which CADASIL was ultimately confirmed.

All available brain MRI studies were re-reviewed by a single-board–certified neuroradiologist to systematically adjudicate CADASIL-associated and MS-associated imaging features, including the presence and symmetry of anterior temporal pole and external capsule white matter hyperintensities (WMHs), as well as MS typical markers (periventricular, juxtacortical, infratentorial, corpus callosum lesions, and Dawson fingers). Cerebral microbleeds (CMBs) were assessed on susceptibility-sensitive sequences (SWI or T2*-GRE) when available and categorized by burden (0 vs 1–10 vs > 10 CMBs, with studies lacking adequate susceptibility imaging classified as ungradable).

Longitudinal Assessment in the MS Misdiagnosed Subgroup

Within the MS misdiagnosed subgroup, MS-specific treatment exposures (agents and duration) were recorded, the interval between the initial MS misdiagnosis and CADASIL confirmation was calculated. To assess radiographical progression, brain MRIs were retrieved at 2 prespecified timepoints: (1) the scan closest to the date of MS diagnosis and (2) the scan closest to the date of CADASIL confirmation. Across these timepoints, deep WMH burden, lacune count, and functional status were compared using the N3SVD staging system.

Statistical Analysis

Baseline characteristics were summarized for the 2 groups (those with and without MS misdiagnosis). Continuous variables are presented as means and SDs for normally distributed data or as the medians (minimum-maximum) for skewed data. Categorical variables are summarized as counts and percentages (n, %).

Multivariable logistic regression was performed with MS misdiagnosis as the dependent variable. Core covariates were selected a priori based on clinical plausibility and included age at CADASIL diagnosis, sex, race (modeled as Asian vs non-Asian due to sparse counts among other subgroups), and prior stroke history. To reduce overfitting given the limited number of events, each imaging or genetic predictor was evaluated in a separate adjusted model rather than combining multiple candidate predictors in a single model. Effect sizes are reported as odds ratios (ORs) with 95% CIs. Given the small sample size of the misdiagnosed subgroup, statistical comparisons should be regarded as exploratory and interpreted with caution.

Standard Protocol Approvals, Registrations, and Patient Consents

This was a multicenter, retrospective cohort study in genetically or histopathologically confirmed CADASIL patients, identified through consecutive case ascertainment between 1999 and 2024 using International Classification of Diseases code (I67.850) searches in the electronic medical record (Epic Inc., Verona, WI) followed by manual verification across the 3 Mayo Clinic sites (Mayo Clinic FL [MCF], Mayo Clinic Rochester [MCR], and Mayo Clinic Arizona [MCA]). The study was approved by the Mayo Clinic Institutional Review Board (institutional review board # 24–013764), with a waiver of informed consent.

Data Availability

To ensure scientific accuracy, data integrity, and protection of human participants, the corresponding author will review and consider reasonable requests for data sharing. Any sharing will only occur if all institutional and regulatory requirements are met, appropriate data use agreements are in place when necessary, and full human participant protections (including HIPAA adherence, deidentification, and institutional review board considerations) are maintained. If a request cannot be fulfilled because of privacy, legal, or ethical constraints, a justification will be provided, and aggregate or summary-level data may be offered as an alternative.

Results

Of 107 patients with CADASIL (mean age 57.5 ± 13.4 years), 11 (10.2%) had a prior MS misdiagnosis ( Table 1 ). The initial MS diagnosis was made outside the Mayo system in 72% (8/11) of patients and at Mayo-affiliated sites in 27% (3/11) of patients. CADASIL was ultimately confirmed at MCF 45% (5/11), MCR 36% (4/11), and MCA 18% (2/11).

Table 1 Demographic and Clinical Characteristics at the Time of CADASIL Diagnosis Variable MS misdiagnosis (n = 11) No MS misdiagnosis (n = 96) Age, mean (SD) 55.7 (12.8) 57.8 (13.5) Sex, n (%)  Male 3 (27.3) 33 (34.4)  Female 8 (72.7) 63 (65.6) Race, n (%)  White 7 (63.6) 88 (91.7)  African American 0 (0.0) 2 (2.1)  Asian 4 (36.4) 3 (3.1)  Other 0 (0.0) 3 (3.1) Family history, n (%)  Multiple sclerosis 1 (9.1) 13 (13.5) Stroke/migraine/dementia, n (%)  None 2 (18.2) 20 (20.8)  Stroke only 3 (27.3) 19 (19.8)  Dementia only 1 (9.1) 4 (4.2)  Migraine only 0 (0.0) 12 (12.5)  Stroke + dementia + migraine 3 (27.3) 20 (20.8)  CADASIL 6 (54.5) 63 (65.6) Medical history, n (%)  Stroke 4 (36.4) 62 (73.8)  Hypertension 7 (63.6) 45 (46.9)  Diabates mellitus 2 (18.2) 15 (15.6)  Hyperlipidemia 8 (72.7) 63 (65.6) Smoking status, n (%)  Never 6 (60.0) 51 (56.7)  Former 3 (30.0) 36 (40.0)  Active 1 (10.0) 3 (3.3)  History of lumbar puncture, n (%) 10 (90.9) 25 (26.0)  CSF oligoclonal bands, n (%) 1 (10) 0 (0) Modified Rankin scale-median (min-max) 1 (0–3) 1 (0–4) Baseline N3SVD stages, n (%)  0 0 (0) 4 (4.2)  1A 1 (9.1) 24 (25.0)  1B 6 (54.5) 61 (63.5)  2A 0 (0.0) 0 (0.0)  2B 0 (0.0) 0 (0.0)  3A 4 (36.4) 5 (5.2)  3B 0 (0.0) 2 (2.1) EGFr risk categories a , n (%)  High-risk 4 (40) 56 (63.6)  Medium-risk 6 (60) 27 (30.7)  Low-risk 0 (0.0) 4 (4.5)  Unknown-risk 0 (0.0) 1 (1.1) Abbreviations: CADASIL = Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy; EGFr = epidermal growth factor–like repeat (domain); MS = multiple sclerosis; mRS = modified Rankin Scale;N3SVD = NOTCH3 small vessel disease stage. Age is presented as mean (SD). Ordinal and continuous variables are reported as median (minimum–maximum), and categorical variables as number (%). a EGFr risk categories were defined according to the 3-tiered classification system proposed by Hack et al. 9 based on the NOTCH3 cys variant frequency odds ratio (NVFOR), which represents the odds that a NOTCH3 cys variant is located in a specific EGFr domain in CADASIL cohorts relative to the odds in NOTCH3 cys-positive individuals from population databases. Risk categories were defined as high-risk EGFr (NVFOR >5), medium-risk EGFr (NVFOR 0.20–5), and low-risk EGFr (NVFOR <0.20). EGFr domains with fewer than 10 variant counts were classified as unknown risk.

Demographic and Clinical Characteristics at the Time of CADASIL Diagnosis

Abbreviations: CADASIL = Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy; EGFr = epidermal growth factor–like repeat (domain); MS = multiple sclerosis; mRS = modified Rankin Scale;N3SVD = NOTCH3 small vessel disease stage.

Age is presented as mean (SD). Ordinal and continuous variables are reported as median (minimum–maximum), and categorical variables as number (%).

EGFr risk categories were defined according to the 3-tiered classification system proposed by Hack et al. 9 based on the NOTCH3 cys variant frequency odds ratio (NVFOR), which represents the odds that a NOTCH3 cys variant is located in a specific EGFr domain in CADASIL cohorts relative to the odds in NOTCH3 cys-positive individuals from population databases. Risk categories were defined as high-risk EGFr (NVFOR >5), medium-risk EGFr (NVFOR 0.20–5), and low-risk EGFr (NVFOR <0.20). EGFr domains with fewer than 10 variant counts were classified as unknown risk.

Compared with patients who were correctly diagnosed with CADASIL at initial presentation, those who were initially misdiagnosed with MS were more frequently Asian (36.4% vs 3.1%), more likely to have undergone lumbar puncture (90.9% vs 26.0%), and less likely to have a history of stroke (36.4% vs 73.8%). Among the 10 misdiagnosed patients who underwent lumbar puncture, CSF oligoclonal bands were present in 1 patient. This patient did not have sufficient clinical or radiographic evidence to support a confirmed CADASIL-MS codiagnosis. 10,11 By the time of CADASIL confirmation, patients with a prior MS misdiagnosis had higher median mRS scores and were at a higher N3SVD stage than those diagnosed correctly at initial presentation.

At the time of CADASIL diagnosis ( Table 2 ), anterior temporal pole WMH was present in 45.5% of patients with prior MS misdiagnosis compared with 82.3% of those without misdiagnosis. When present, anterior temporal pole WMH was symmetric in nearly all cases in both groups (5/5, 100% vs 77/79, 97.5%). External capsule WMHs were common in both groups. However, among patients with external capsule WMH, symmetric involvement was less frequent in the misdiagnosed group than in those without MS misdiagnosis (14.3% vs 86.5%). CMB burden was low in the MS-misdiagnosed group, with 90.9% of patients having no CMBs, 9.1% having 1–10 CMBs, and none had >10 CMBs. By contrast, among patients without prior MS misdiagnosis, 65.9% had no CMBs, 24.2% had 1–10 CMBs, and 9.9% had >10 CMBs (5.2% were ungradable). Furthermore, CMBs involving the anterior temporal pole or external capsule were observed only in the non-misdiagnosed group (9.4%) and were absent in patients with prior MS misdiagnosis. Among the 9 of 11 patients with prior MS misdiagnosis had spinal MRI, none of which showed spinal cord lesions. Figure 1 shows the neuroimaging findings from a misdiagnosed patient in our cohort, highlighting imaging overlap with MS and the absence of typical CADASIL features.

Table 2 Neuroimaging Characteristics at the Time of CADASIL Diagnosis Variable MS misdiagnosis (n = 11) No MS misdiagnosis (n = 96) General neuroimaging markers-median (min-max)  (1) Deep white matter hyperintensity 2 (1–3) 2 (0–3)  (2) Lacunes 1 (0–13) 0 (0–11)  (3) Cerebral microbleeds   0 (%) 10 (90.9) 60 (65.9)   1–10 (%) 1 (9.1) 22 (24.2)   >10 (%) 0 (0) 9 (9.9)   Ungradable (%) 0 (0) 5 (5.2) CADASIL-related neuroimaging markers, n (%)  (1) External capsule WMH 7 (63.6) 74 (77.1)  (2) Symmetric external capsule WMH 1 (14.3) 64 (86.5)  (3) Anterior temporal pole WMH 5 (45.5) 79 (82.3)  (4) Symmetric anterior temporal pole WMH 5 (45.5) 77 (80.2)  (5) Cerebral microbleeds in anterior temporal pole or external capsule 0 (0.0) 9 (9.4) MS-related neuroimaging markers, n (%)  (1) Periventricular WMH 10 (90.9) 90 (98.9)  (2) Juxtacortical WMH 10 (90.9) 91 (94.8)  (3) Infratentorial WMH 2 (18.2) 68 (70.8)  (4) Spinal cord WMH 0 (0.0) 3 (3.1)  (5) Corpus callosum WMH 3 (27.3) 26 (27.1)  (6) Dawson fingers 7 (63.6) 51 (53.1) Abbreviations: DWMH = deep white matter hyperintensity; MS = multiple sclerosis; WMH = white matter hyperintensity. Quantitative imaging variables are reported as median (minimum–maximum), and categorical imaging features as number (%).

Neuroimaging Characteristics at the Time of CADASIL Diagnosis

Abbreviations: DWMH = deep white matter hyperintensity; MS = multiple sclerosis; WMH = white matter hyperintensity.

Quantitative imaging variables are reported as median (minimum–maximum), and categorical imaging features as number (%).

Neuroimaging Findings in a Patient With Prior MS Misdiagnosis Compared With Typical CADASIL and Multiple Sclerosis

CADASIL and MS share several imaging features that may lead to misdiagnosis. Representative images from patient 1 (top row, A–D), who was initially treated for MS but was subsequently diagnosed with CADASIL, demonstrate this overlap. On sagittal T2 FLAIR imaging (C), the periventricular lesions may resemble MS. However, closer inspection shows extension to the ependymal margin without infratentorial WMH (D). Notably, patient 1 (A and B) lacks classic CADASIL features, such as external capsule WMH (E) and anterior temporal white matter WMH (F), which are demonstrated in patient 2. Patients 3 (G) and 4 (H) are examples of patients with known MS who have true Dawson fingers and infratentorial demyelination. Panels A, B, E, and F show axial T2 FLAIR sequences; panels C, D, G, and H show sagittal T2 FLAIR sequences. CADASIL = Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy; FLAIR = fluid-attenuated inversion recovery; MS = multiple sclerosis; WMH = white matter hyperintensity.

In adjusted logistic regression models, absence of prior stroke (adjusted OR 4.60; 95% CI 1.02–20.68), Asian race (adjusted OR 20.74; 95% CI 3.21–134.18), asymmetric external capsule WMH (adjusted OR 53.34; 95% CI 4.06–701.08), and absence of infratentorial WMH (adjusted OR 12.48; 95% CI 1.90–82.01) were independently associated with MS misdiagnosis ( Table 3 ).

Table 3 Demographic, Clinical, and Radiographic Factors Associated With MS Misdiagnosis in Patients With CADASIL Variable Unadjusted OR (95% CI) Adjusted OR (95% CI) Age at CADASIL diagnosis 1.01 (0.96–1.05) 1.00 (0.95–1.05) Male sex 0.72 (0.18–2.88) 1.01 (0.22–4.75) Asian race 17.71 (3.29–95.29) 20.74 (3.21–134.18) Absence of stroke history 4.05 (1.10–14.93) 4.60 (1.02–20.68) Asymmetric external capsule WMH a 38.40 (4.17–353.36) 53.34 (4.06–701.08) Asymmetric anterior temporal pole WMH 4.86 (1.34–17.64) 4.33 (0.99–19.04) Absence of infratentorial WMH 10.87 (2.22–52.63) 12.48 (1.90–82.01) Presence of medium-risk EGFr domain variant b 3.39 (0.88–12.99) 1.39 (0.20–9.52) Abbreviations: CADASIL = Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy; EGFr = epidermal growth factor–like repeat; MS = multiple sclerosis; WMH = white matter hyperintensity. Odds ratios (OR) and 95% CIs were estimated using logistic regression with MS misdiagnosis as the dependent variable. Unadjusted ORs were derived from univariable models. Adjusted ORs were calculated using prespecified multivariable models that included age at CADASIL diagnosis, male sex, Asian race, and prior stroke history. To minimize overfitting due to the limited number of misdiagnosis events (n = 11), each imaging or genetic factor was evaluated in a separate adjusted model. a Analyses for external capsule symmetry were restricted to participants with external capsule WMH (N = 81; misdiagnosis events = 7). Because standard maximum-likelihood estimation was unstable in this subset when sex was included, the adjusted model for external capsule asymmetry was controlled for age, Asian race, and prior stroke history only. b EGFr analyses were restricted to participants with genetic testing (N = 98).

Demographic, Clinical, and Radiographic Factors Associated With MS Misdiagnosis in Patients With CADASIL

Abbreviations: CADASIL = Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy; EGFr = epidermal growth factor–like repeat; MS = multiple sclerosis; WMH = white matter hyperintensity.

Odds ratios (OR) and 95% CIs were estimated using logistic regression with MS misdiagnosis as the dependent variable. Unadjusted ORs were derived from univariable models. Adjusted ORs were calculated using prespecified multivariable models that included age at CADASIL diagnosis, male sex, Asian race, and prior stroke history. To minimize overfitting due to the limited number of misdiagnosis events (n = 11), each imaging or genetic factor was evaluated in a separate adjusted model.

Analyses for external capsule symmetry were restricted to participants with external capsule WMH (N = 81; misdiagnosis events = 7). Because standard maximum-likelihood estimation was unstable in this subset when sex was included, the adjusted model for external capsule asymmetry was controlled for age, Asian race, and prior stroke history only.

EGFr analyses were restricted to participants with genetic testing (N = 98).

The median time to correct the MS misdiagnosis to CADASIL was 14.45 months (range 1.18–138.07). During this interval, although there was no significant clinical and radiologic progression ( Table 4 ), 5 patients (45.5%) received unnecessary MS-directed treatments that resulted in adverse events. The regimens (eTable 1), 80% of which were disease-modifying therapies, included glatiramer acetate (n = 2), interferon β-1a (n = 1), interferon β-1b with subsequent transition to dimethyl fumarate and later diroximel fumarate (n = 1), and methylprednisolone (n = 1). Documented adverse events included injection site reactions in 2 patients.

Table 4 Clinical and Radiographical Progression in CADASIL Patients Following Multiple Sclerosis Misdiagnosis Variable At MS misdiagnosis (Timepoint 1) At CADASIL diagnosis (Timepoint 2) DWMH-median (min-max) 2 (1–3) 2 (1–3) Lacunes-median (min-max) 0 (0–9) 1 (0–13) mRS-median (min-max) 1 (0–3) 1 (0–3) N3SVD stage, n (%)  0 0 (0.0) 0 (0.0)  1A 2 (18.2) 1 (9.1)  1B 3 (27.3) 6 (54.5)  2A 3 (27.3) 0 (0.0)  2B 0 (0.0) 0 (0.0)  3A 3 (27.3) 4 (36.4)  3B 0 (0.0) 0 (0.0) Abbreviations: DWMH = deep white matter hyperintensity; mRS = modified Rankin Scale; N3SVD = NOTCH3 small vessel disease. This table shows clinical and radiographical measures at 2 time points: at MS misdiagnosis (time point 1) and after correction to a CADASIL diagnosis (time point 2), with a median interval of 14.45 mo between the 2 assessments (range 1.18–138.07 mo). Quantitative variables are reported as median (min–max); categorical variables are reported as n (%).

Clinical and Radiographical Progression in CADASIL Patients Following Multiple Sclerosis Misdiagnosis

Abbreviations: DWMH = deep white matter hyperintensity; mRS = modified Rankin Scale; N3SVD = NOTCH3 small vessel disease.

This table shows clinical and radiographical measures at 2 time points: at MS misdiagnosis (time point 1) and after correction to a CADASIL diagnosis (time point 2), with a median interval of 14.45 mo between the 2 assessments (range 1.18–138.07 mo). Quantitative variables are reported as median (min–max); categorical variables are reported as n (%).

Among the 11 patients with prior MS misdiagnosis, 10 had available NOTCH3 genetic testing, while 1 was diagnosed by skin biopsy without genotyping. Misdiagnosis was observed in both high-risk and medium-risk genetic variant groups, with no cases in low-risk or unknown-risk categories ( Figure 2 ). Medium-risk variants were more frequent in the misdiagnosed group (60% vs 30.7%), and the interval from MS misdiagnosis to CADASIL confirmation was longer in the medium-risk group (mean 48.85 months, median 10.20 months [range 1.18–138.07]) than in the high-risk group (mean 28.54 months, median 8.59 months [range 1.58–95.40]), with the single skin biopsy–confirmed patient without genotyping corrected after 20.01 months. Half of the misdiagnosed patients carried medium-risk variants clustered in EGFr domains 13, 13.5, and 14 (eFigure 1). Notably, among the 5 patients who received treatment, 4 had genetic testing. Of these, 3 (75%) carried medium-risk variants, while 1 had a high-risk variant.

Distribution of Multiple Sclerosis Misdiagnosis Across EGFr Risk Categories

Stacked bars (x-axis) depict the number of genetically or histopathologically confirmed CADASIL patients (left y-axis) stratified by NOTCH3 EGFr risk category (high, medium, low, and unknown risk) 9 and by MS misdiagnosis status (no MS misdiagnosis, MS misdiagnosis untreated, and MS misdiagnosis treated). Misdiagnosis occurred exclusively in the high-risk and medium-risk EGFr groups, with no cases observed in low-risk or unknown-risk categories. Medium-risk variants were proportionally more frequent among misdiagnosed patients compared with non-misdiagnosed patients. Blue dots (right y-axis) represent the individual time intervals (months) between initial MS misdiagnosis and subsequent confirmation of CADASIL among misdiagnosed patients. CADASIL = Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy; EGFr = epidermal growth factor–like repeat (domain); MS = multiple sclerosis.

EGFr-stratified neuroimaging patterns are summarized in eTable 2. Within the medium-risk stratum, patients with a prior MS misdiagnosis (n = 6) rarely demonstrated anterior temporal pole WMH (16.7% vs 70.4%) or infratentorial WMH (16.7% vs 74.1%) compared with medium-risk patients without misdiagnosis. Although external capsule WMH was observed in 50.0% (3/6) of misdiagnosed patients, none exhibited symmetric involvement (0% vs 86.4%).

Discussion

We found that nearly 1 in 10 patients with CADASIL had previously been misdiagnosed with MS. In prespecified adjusted models, Asian race, absence of prior stroke history, presence of asymmetric external capsule WMH, and absence of infratentorial WMH on MRI brain were independently associated with MS misdiagnosis. However, the small number of misdiagnosis events resulted in wide CIs, and these findings should be interpreted with caution pending confirmation in other populations and larger samples.

Beyond the effects of diagnostic delay, misdiagnosis created the potential for treatment-related harm. While no significant clinical or radiologic progression was observed during the typically short interval between misdiagnosis and diagnostic correction, 45.5% of misdiagnosed patients received MS-directed treatments for a median of 10 months, incurring financial costs and avoidable risks. 2,3 Most therapies for MS were disease-modifying agents, whereas only 1 patient had corticosteroid exposure. This low rate of steroid use is consistent with guideline recommendations that corticosteroids be reserved for clinically significant acute relapses and ought not be used in mild or unrecognized episodes. 10 Two individuals experienced injection-site reactions from first-line injectable agents such as glatiramer acetate and interferon β formulations. Patients were also at risk of additional adverse events from these therapies, including flu-like symptoms, hepatotoxicity, autoimmune complications, recurrent infections, and in rare cases, progressive multifocal leukoencephalopathy. 10,12,13 These risks underscore the critical importance of avoiding misdiagnosis and unnecessary treatment exposure.

Genetic analysis showed that medium-risk EGFr variants were more frequent among misdiagnosed patients in unadjusted comparisons, but this association did not persist after prespecified adjustment. Half of the misdiagnosed patients carried medium-risk variants clustered in EGFr domains 13, 13.5, and 14. Among the 4 treated patients who underwent genetic analysis, 3 (75%) carried medium-risk variants. 9 In addition, all 4 had NOTCH3 variants located outside the classic high-risk EGFr 1–6 domains. 14 Owing to their intermediate penetrance and variable expressivity, 9 medium-risk genetic variants may result in atypical phenotypes that complicate the recognition of CADASIL and predispose patients to MS misdiagnosis.

The observed association between MS misdiagnosis and the Asian race may reflect underlying differences in CADASIL genetic epidemiology and phenotype expression. Population-based genetic studies demonstrate higher frequencies of pathogenic NOTCH3 variants in East and South Asians compared with the general population. 8,15 Moreover, CADASIL in Asian patients often lacks classic radiologic phenotype, such as anterior temporal pole involvement, and may present later in life with confluent WMHs and cognitive impairment, increasing the risk of misdiagnosis. 16-18

Beyond clinical factors, atypical neuroimaging patterns contribute to diagnostic confusion. In adjusted models, asymmetric external capsule involvement and the absence of infratentorial lesions were independently associated with MS misdiagnosis, a pattern that does not fit the classic radiologic picture of CADASIL. 1 EGFr-stratified imaging patterns further support this mechanism. Medium-risk misdiagnosed cases rarely demonstrated anterior temporal pole involvement and never showed symmetric external capsule lesions, and infratentorial WMH were uncommon. Furthermore, a high CMB burden (>10) and region-specific CMB distribution were observed only in correctly diagnosed patients, suggesting that prominent microbleeds favor a small vessel disease diagnosis rather than MS. 1,4,10 Together, these atypical or incomplete CADASIL markers create a plausible pathway to misclassification, particularly in late-onset or stroke-naïve presentations where confluent supratentorial WMH may be interpreted within an MS framework.

While rare, a true codiagnosis of CADASIL and MS is possible. 19,20 This is supported by evidence that cysteine-altering NOTCH3 variants could increase autoimmune susceptibility. 5 In a retrospective study conducted across 7 MS tertiary centers, CSF oligoclonal bands were absent in 100% of patients with CADASIL misdiagnosed as MS, but present in 83% of those with confirmed codiagnosis of neurogenetic disease and MS. 2 Similarly, in our cohort, CSF oligoclonal bands were absent in 91% patients misdiagnosed with MS. These findings highlight the role of CSF oligoclonal bands in distinguishing true codiagnosis from misdiagnosis.

In our study, all patients treated for MS were older than 50 years, an age group in which a new diagnosis of MS is uncommon (∼5% cases). 21 The presence of clinical and radiographical mimics, such as cerebral small vessel disease (including CADASIL), migraine-related white matter changes, and certain inflammatory conditions in the same age group increases risk of misdiagnosis. Therefore, before diagnosing and treating suspected late-onset MS, it is recommended that physicians obtain additional supportive evidence according to 2024 revision of McDonald criteria. 11 First, a spinal cord lesion should be investigated because its presence strongly supports an MS diagnosis while being generally absent in cerebrovascular disease. Notably, among misdiagnosed patients in our cohort who underwent spinal MRI, no spinal cord lesions were identified, highlighting the absence of a key MS-supportive imaging marker. Second, advanced MRI biomarkers such as the central vein sign and paramagnetic rim lesions are highly specific for MS and seldom seen in mimics. Because our cohort lacked high-resolution 3T/7T MRI brain imaging, these markers could not be assessed. Third, while the CSF kappa-free light chain (kFLC) index is an accepted alternative to traditional oligoclonal bands (OCBs) for demonstrating CNS inflammatory activity, it was not measured in our cohort. Nevertheless, OCBs were negative in 91% of misdiagnosed patients, underscoring the value of CSF immunologic markers in distinguishing MS from its mimics. 11

Building on this study, a practical next step would be to validate prospectively targeted CADASIL testing within MS clinics. The decision to test could be guided by 2 clinical perspectives. First, consider testing when supportive MS markers are absent, particularly in late onset presentations with negative OCB or kFLC and no spinal cord demyelinating lesions. Second, prioritize testing when specific small vessel disease features are present, such as migraine with or without aura, prior TIA or lacunar stroke, cognitive symptoms, vascular risk factors, lacunes, or WMH involving the anterior temporal pole or external capsule. As a standardized alternative, clinics could apply the Pescini CADASIL Scale and use a prespecified threshold such as 14 or higher to trigger NOTCH3 sequencing. 6

This study has several notable strengths. Although CADASIL is rare, we assembled a genetically confirmed cohort of more than 100 patients, representing one of the larger series reported to date. The centralized multicenter design improves generalizability by encompassing diverse clinical presentations and practice patterns across diverse regions of the United States while ensuring standardized data collection. However, because all sites are tertiary-care centers, referral biases remain, and generalization of results must be interpreted with caution. Incorporating detailed genotypic classification by EGFr domain risk provides important biological context to diagnostic patterns and highlights how variant location may influence clinical recognition. In addition, our review of radiologic data and functional outcomes offers a comprehensive assessment of the downstream consequences of misdiagnosis, linking diagnostic delay to both disease stage and exposure to inappropriate MS-directed therapies.

Certain limitations of our study should be acknowledged. While our cohort size is large for a rare disease, it remains modest compared with the broad clinical heterogeneity of CADASIL, which may limit our ability to detect more subtle between-group differences. The retrospective design introduces potential bias due to incomplete or inconsistent documentation. Because initial clinical and radiologic data at the time of MS labeling were often incomplete or predated contemporary diagnostic standards, we were unable to reliably reclassify cases under 2024 McDonald criteria and thus cannot fully exclude historical miscoding or diagnostic uncertainty. The small size of the MS-misdiagnosed subgroup limits the robustness of our statistical analyses. In addition, the short interval between misdiagnosis and CADASIL confirmation in some patients constrained our ability to detect meaningful clinical or radiologic changes. Although Mayo Clinic serves a national and international referral population, the racial and ancestral composition of this tertiary-care cohort is not population representative and race-related associations, particularly with Asian ancestry, should therefore be interpreted cautiously. These limitations highlight the need for multicenter collaborative studies pooling CADASIL cases across programs to increase sample size, validate these preliminary associations, and refine risk stratification for misdiagnosis.

In conclusion, in this multicenter cohort study, we found that 1 in 10 patients with CADASIL was initially misdiagnosed with MS, a diagnostic error associated with Asian race, absence of prior stroke history, asymmetric external capsule WMHs, and absence of infratentorial lesions on brain MRI. At the time of CADASIL confirmation, patients with a prior MS misdiagnosis were more likely to be at an advanced N3SVD disease stage and had greater functional disability, and many were exposed to unnecessary and potentially harmful MS-directed therapies. Medium-risk NOTCH3 variants were more frequent among misdiagnosed patients. EGFr-stratified imaging suggests that these variants may be linked to atypical radiologic patterns that reduce recognition of CADASIL and increase diagnostic ambiguity. Together, these results underscore the critical need for clinicians to maintain a high index of suspicion for CADASIL in patients with atypical or late onset demyelinating presentations and to judiciously apply advanced diagnostic tools to ensure accurate diagnosis and appropriate management.

Glossary

Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalopathy

cerebral microbleed

epidermal growth factor–like repeat

kappa-free light chain

Mayo Clinic Arizona

Mayo Clinic Florida

Mayo clinic Rochester

modified Rankin Scale

multiple sclerosis

NOTCH3 Small Vessel Disease

oligoclonal band

odds ratio

white matter hyperintensity

Author Contributions

D.D. Mendis: 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. S.K. Vallamchetla: 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. I.V. Marin Collazo: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. S.J.S. Sandhu: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. J.F. Meschia: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data.

Study Funding

Mayo Clinic Foundation for Medical Education and Research.

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/NG for full disclosures.

References

Associated Data

Data Availability Statement

To ensure scientific accuracy, data integrity, and protection of human participants, the corresponding author will review and consider reasonable requests for data sharing. Any sharing will only occur if all institutional and regulatory requirements are met, appropriate data use agreements are in place when necessary, and full human participant protections (including HIPAA adherence, deidentification, and institutional review board considerations) are maintained. If a request cannot be fulfilled because of privacy, legal, or ethical constraints, a justification will be provided, and aggregate or summary-level data may be offered as an alternative.