Prevalence and Significance of the Vessel-Cluster Sign on Susceptibility-Weighted Imaging in Patients With Severe Small Vessel Disease

BACKGROUND AND OBJECTIVES: Magnetic resonance susceptibility-weighted imaging (SWI) can identify small brain blood vessels that contain deoxygenated blood due to its induced magnetic field disturbance. We observed focal clusters of possible dilated small vessels on SWI in white matter in severe small vessel disease (SVD). We assessed their prevalence, associations with SVD lesions, and vascular reactivity in patients with sporadic SVD and in patients with cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). METHODS: Secondary cross-sectional analysis of a prospective multicenter observational study of patients with either sporadic SVD or CADASIL (INVESTIGATE-SVD) studied with 3 Tesla MRI including blood-oxygen-level-dependent MRI cerebrovascular reactivity (CVR). Two independent raters evaluated SWI sequences to identify “vessel-clusters” in white matter as focal low-signal dots/lines with small vessel appearance (interrater agreement, kappa statistic = 0.66). We assessed per-patient and per-cluster associations with SVD lesion type and severity on structural MRI sequences. We also assessed CVR within and at 2-voxel concentric intervals around the vessel-clusters using contralateral volumes as a reference. RESULTS: Among the 77 patients enrolled, 76 had usable SWI sequences, 45 with sporadic SVD (mean age 64 years [SD 11], 26 men [58%]) and 31 with CADASIL (53 years [11], 15 men [48%]). We identified 94 vessel-clusters in 36 of the 76 patients (15/45 sporadic SVD, 21/31 CADASIL). In covariate-adjusted analysis, patients with vessel-clusters had more lacunes (OR, 95% CI) (1.30, 1.05–1.62), higher white matter hyperintensity (WMH) volume (per-log10 increase, 1.92, 1.04–3.56), and lower CVR in normal appearing white matter (per %/mm Hg, 0.77, 0.60–0.99), compared with patients without vessel-clusters. Fifty-seven of the 94 vessel-clusters (61%) corresponded to noncavitated or partially cavitated WMH on fluid-attenuated inversion recovery, and 37 of 94 (39%) to complete cavities. CVR magnitude was lower than in the corresponding contralateral volumes (mean difference [SD], t, p) within vessel-cluster volumes (−0.00046 [0.00088], −3.021, 0.005) and in the surrounding volume expansion shells up to 4 voxels (−0.00011 [0.00031], −2.140, 0.039; −0.00010 [0.00027], −2.295, 0.028) in vessel-clusters with complete cavities, but not in vessel-clusters without complete cavitation. DISCUSSION: Vessel-clusters might correspond to maximally dilated vessels in white matter that are approaching complete tissue injury and cavitation. The pathophysiologic significance of this new feature warrants further longitudinal investigation.