Small Vessel Disease Phenotype Associated With Monoallelic NOTCH3 Loss-of-Function Variants

Abstract

Background and objectives: Monoallelic cysteine-altering NOTCH3 (NOTCH3^cys) variants cause the adult-onset small vessel disease cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), and biallelic NOTCH3 loss-of-function (NOTCH3^lof) variants cause a rare, childhood-onset small vessel disease. Whether monoallelic NOTCH3^lof variants also cause a small vessel disease is subject of debate. The aim of this study was to delineate the small vessel disease phenotype of individuals with a monoallelic NOTCH3^lof variant and to compare it with CADASIL.

Methods: In this observational study, monoallelic NOTCH3^lof cases were ascertained in Genome Aggregation Database (gnomAD); UK Biobank; 6 clinical centers from Europe, Asia, and the United States; and literature. In gnomAD, NOTCH3^lof allele frequency was determined. In UK Biobank, normalized white matter hyperintensity volume (nWMHv), peak width of skeletonized mean diffusivity (PSMD), lacune count, and stroke were compared among NOTCH3^lof cases, NOTCH3^cys cases, and controls. In clinical NOTCH3^lof cases, white matter hyperintensities, lacune count, and stroke incidence were assessed, and skin vessel wall pathology was analyzed using immunohistochemistry and electron microscopy.

Results: In gnomAD, 306 NOTCH3^lof variants were identified (allele frequency 0.6/1,000). In UK Biobank, 102 NOTCH3^lof cases were ascertained (median age 58 years, range 40-69, 55% female). NOTCH3^lof cases had an increased nWMHv (Δ0.44 mm³, p < 0.001) and PSMD (Δ0.19 × 10^-4, p = 0.017) compared with controls. nWMHv and PSMD in NOTCH3^lof cases were comparable to NOTCH3^cys cases; however, in contrast to NOTCH3^cys cases, NOTCH3^lof cases did not have an increased stroke risk compared with controls. Clinically ascertained NOTCH3^lof cases (n = 69, median age 50 years, range 20-94, 54% female) often had white matter hyperintensities (28/32, 88%) while lacunes (12/32, 38%) and stroke (11/69, 15%) were predominantly seen in cases with cardiovascular risk factors and at advanced age. Skin vessels of NOTCH3^lof cases more frequently showed abundant vessel wall collagen deposition compared with NOTCH3^cys cases and controls (37% vs 10% [p = 0.016] and 5% [p < 0.001] of vessels).

Discussion: We conclude that monoallelic NOTCH3^lof variants cause a small vessel disease that (1) remains subclinical in most cases but may be exacerbated by cardiovascular risk factors and aging, and (2) is distinct from CADASIL regarding vessel pathology and disease severity. These findings will guide counseling and management of individuals in whom a NOTCH3^lof variant is found.

Figure 1. Neuroimaging Findings in UK Biobank NOTCH3^lof Cases

(A–E) Representative brain T2 fluid-attenuated inversion recovery images illustrating the spectrum of small vessel disease neuroimaging abnormalities in UK Biobank NOTCH3^lof cases. All NOTCH3^lof cases in UK Biobank had white matter hyperintensities, 1 had lacunes, and none had microbleeds on susceptibility-weighted imaging. WMHs were typically periventricular and were present as small to medium-sized punctate foci in the deep white matter (A–C), including in the anterior temporal lobe (B), but not in the external capsules. There were 2 older NOTCH3^lof cases with vascular risk factors, who had confluent WMHs (D, E). One case (D) was a 68-year-old man with a history of dyslipidemia, diabetes, and smoking, and the other (E) was a 70-year-old man with a history of smoking who also had a lacune in the left frontal lobe and a lacune in the globus pallidus (arrowheads). NOTCH3^lof = NOTCH3 loss-of-function; WMH = white matter hyperintensity.

Figure 2. NOTCH3^lof Variants Are Associated With White Matter Hyperintensity Volume and PSMD, but Not With Risk of Stroke in UK Biobank

(A, B) Violin plots showing the residuals of linear regression models after correction for age, sex, and cardiovascular risk factors. (A) WMH volume did not differ significantly between NOTCH3^lof cases and moderate-risk NOTCH3^cys cases (p = 0.981), whereas it was increased compared with low-risk NOTCH3^cys cases (p = 0.024), matched controls (p < 0.001), and the total of 42,950 individuals with MRI in UK Biobank (p < 0.001). (B) PSMD in NOTCH3^lof cases did not differ significantly from that in moderate-risk NOTCH3^cys cases (p = 0.568), whereas it was increased compared with matched controls (p < 0.001). (C) Bar charts showing the percentage of individuals with stroke per genotype. Logistic regression models did not show a difference in the risk of stroke between NOTCH3^lof cases and controls (OR 1.3, 95% CI 0.4–4.3, p = 0.636). Moderate-risk NOTCH3^cys cases did have an increased risk of stroke compared with controls (OR 4.4, 95% CI 2.1–9.3, p < 0.001). NOTCH3^cys = cysteine-altering NOTCH3; NOTCH3^lof = NOTCH3 loss-of-function; ns = not significant; OR = odds ratio; PSMD = peak width of skeletonized mean diffusivity; WMH = white matter hyperintensity. Values are significant at ***p < 0.001, **p < 0.01, and *p < 0.05.

Figure 3. White Matter Hyperintensities and Late-Onset Lacunes in a Dutch Pedigree With a NOTCH3^lof Variant

(A) Pedigree of a Dutch family with a NOTCH3 c.1321C>T; p.(Arg441*) variant. The index' medical history reported a “stroke-like event” at age 46 years, without evidence of infarction on brain CT and MRI. The other siblings were asymptomatic. Except for 1 sibling with well-controlled hypertension (III-1), the siblings did not have cardiovascular risk factors. The father had a high cardiovascular risk factor burden, with over 50 pack-years of smoking, daily alcohol use, and hypertension. He was diagnosed with a normal pressure hydrocephalus, attributed to a neurotrauma at age 53 years. Up to age 67 years, he had normal scores on cognitive tests, including 30/30 on the MMSE. None of his family members had a history of stroke or dementia. (B) Representative brain T1 and T2 fluid-attenuated inversion recovery images. The father (II-8) had confluent deep white matter and periventricular white matter hyperintensities; atrophy; and 7 lacunes located in the thalamus, external capsules, and cerebellum (white arrowheads) at age 70 years. All the affected siblings had white matter hyperintensities but no other small vessel disease markers and no stroke or cognitive decline. (C) Schematic representation of the NOTCH3 mRNA. The NOTCH3 c.1321C>T; p.(Arg441*) variant causes a premature termination codon in exon 8. (D) Genomic DNA and primary patient fibroblast mRNA analysis. The NOTCH3 c.1321C>T; p.(Arg441*) variant that was identified in the genomic DNA was not present in the reversed-transcribed fibroblast mRNA (cDNA), indicating degradation of the mutant allele through nonsense-mediated decay. cDNA = complementary DNA; HT = hypertension; LU = lung cancer; mRNA = messenger RNA; NOTCH3^lof = NOTCH3 loss-of-function; PD = Parkinson disease; S = smoking.

Figure 4. Skin Vessels of NOTCH3^lof Cases Have Increased Deposition of Collagen Fibrils

(A) Representative images of NOTCH3 immunohistochemistry on skin blood vessels in a control, a NOTCH3^lof case (NOTCH3 c.1321C>T; p.(Arg441*)), and a NOTCH3^cys case (NOTCH3 c.505C>T; p.(Arg169Cys)). NOTCH3 staining of skin vessels from NOTCH3^lof cases was similar to that in controls while the NOTCH3^cys case shows the typical increased and granular NOTCH3 vessel wall staining of CADASIL. (B) Representative electron microscopy images of a dermal vascular unit of a NOTCH3^lof case showing a thick, amorphous layer of collagen between the endothelial cells and the mural cells (open arrowheads). The image of the NOTCH3^cys case shows the CADASIL-pathognomonic GOM deposits (black arrowhead). Additional images of skin vessels are included in eAppendices 12–13. (C) Bar chart showing the mean percentage of vessels with abundant collagen fibrils in the basal membrane, after qualitative assessment by 2 independent blinded observers (median per patient = 19 vessels, range 5–26). One dot represents 1 patient. On average, over one-third of skin vessels of NOTCH3^lof cases (37%) showed a large quantity of collagen fibrils, which was rarely seen in skin vessels of controls (5%, p < 0.001) and NOTCH3^cys cases (10%, p = 0.016). BM = basal membrane; CADASIL = cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy; E = endothelial cell; L = lumen; M = mural cell; N = nucleus; NOTCH3^cys = cysteine-altering NOTCH3; NOTCH3^lof = NOTCH3 loss-of-function. Values are significant at ***p < 0.001 and *p < 0.05.