Assessment of Small Vessel Density Changes in Cerebral Autosomal Dominant Arteriopathy With Subcortical Infarcts and Leukoencephalophy (CADASIL) by High-Resolution Black-Blood MRI

Abstract

Background: Direct assessments of cerebral small vessels in Cerebral Autosomal Dominant Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL) remain a challenge.

Purpose: To investigate changes of cerebral small vessels in CADASIL using iso-0.5 mm black-blood MRI.

Study type: Case control study.

Population: Thirty-six genetically confirmed CADASIL patients (23 female, 43 ± 12.48 years) and 35 matched healthy controls (27 female, 40 ± 11.57 years).

Field strength/sequence: 3T using a T1-weighted turbo spin-echo with variable flip angles sequence.

Assessment: Vessel density images (VDIs) were derived from black-blood MRI by using a semi-automatic pipeline with a Jerman filter. The differences in VDI were assessed between CADASIL and control groups. The relationships between changes in VDI and cognitive performance and disease burden were studied in the CADASIL group.

Statistical tests: Two-tailed independent samples t-tests were employed to assess the difference in VDI between CADASIL and control groups. Generalized linear mixed-effect models were used to assess the associations of VDI with cognitive performance and disease burden. Voxel-wise analyses were performed to further explore the associations of regional VDI with cognitive performance and disease burden after FDR correction.

Results: Reduced mean VDI was found in gray matter of CADASIL patients (1.31 ± 0.06) compared to controls (1.35 ± 0.03), which was significantly associated with lower MoCA scores (β = 52.89, SE = 12.99, 95% CI [26.38, 79.40]), and higher cerebral small vessel disease (cSVD) burden scores (β = -14.34, SE = 3.22, 95% CI [-20.91, -7.76]) in CADASIL patients. Voxel-wise analyses revealed reduced regional VDI in regions of the temporal pole, insula, cingulate cortex, and orbitofrontal cortex in CADASIL patients.

Data conclusion: The VDI technique based on high-resolution black-blood MRI demonstrated changes in regional VDI in CADASIL patients and offers a noninvasive imaging tool to advance the understanding of the mechanisms underlying cSVD.

Evidence level: 3.

Technical efficiency: Stage 2.

Keywords: Black‐blood MRI; CADASIL; cerebral small vessel disease; vessel density imaging.

FIGURE 1 |

A schematic of semi-automatic Black-Blood Processing Pipeline. (A) (B) The high-resolution BB image was first skull-stripped and de-noised using non-local means filter. (C) Small Vessel Masks were obtained using Jerman vesselness filter with optimal thresholds. (D) The resulting Binary Vessel masks were spatially smoothed into vessel density image (VDI). (E) Finally, the resulting VDIs were normalized to the MNI152 template space for cross-subject analysis.

FIGURE 2 |

Example binary black blood masks. (A–C) Three representative HC subjects. (D–F) Three representative CADASIL patients. Homogeneous distribution of cerebral small vessels can be seen in HC while that in the frontal and temporal areas are visibly sparser in CADASIL patients.

FIGURE 3 |

Comparison of mean vessel density between HC and CADASIL. (A) GM. (B) WM. p < 0.5/7 = 0.0071 is considered significant (*).

FIGURE 4 |

Comparison of voxel-based morphometry (VBM) measures between HC and CADASIL. (A) Total intracranial volume. (B) Gray matter volume. (C) White matter volume. p < 0.05/3 = 0.017 was considered significant difference.

FIGURE 5 |

Correlations between GM VDI with SVD burden (A) and MoCA (B), p < 0.05/(1 × 3) = 0.017 was considered significant difference.

FIGURE 6 |

Significant Clusters identified in voxel-wise analyses of VDI between CADASIL and HC. CADASIL patients showed significantly reduced VDI in temporal pole, OFC, rectus, parahippocampal areas, insula, olfactory bulb, red nucleus, thalamus, and cingulate gyrus, whereas increased VDI was found in supplementary motor areas and white matter tracts. Clusters showing positive correlation were colored red to light yellow whereas clusters showing negative correlations were colored cyan. FDR-cluster corrected p < 0.001 was considered significant.

FIGURE 7 |

Significant Clusters identified in voxel-wise analyses of VDI in CADASIL. (A) Significant clusters where asymptomatic CADASIL patients showing significant higher VDI compared to symptomatic patients including the calcarine, middle occipital gyrus, inferior frontal gyrus, precentral gyrus, and the superior frontal gyrus; (B) Significant clusters where lower VDI predicts higher global WMH lesions including the right temporal gyrus, cingulate, precuneus, right parietal gyri, and paracentral lobule; (C) Significant clusters where lower VDI predicts a higher overall SVD burden including the fusiform, temporal pole, parahippocampal area, inferior-frontal gyrus, NAcc, olfactory bulb, and the caudate. Clusters showing positive correlation were colored red to light yellow whereas clusters showing negative correlations were colored cyan. FDR-cluster corrected p < 0.001 was considered significant.