Development and validation of a rating scale for perivascular spaces on 3T MRI.

BACKGROUND AND PURPOSE: To develop and validate a novel perivascular space rating scale, based on single axial slices in the basal ganglia and the centrum semiovale on T1-weighted and FLAIR images obtained on a 3T MRI scanner. METHODS: 414 community dwelling older adults age 70-90 were assessed. The number of perivascular spaces in the slices 2 mm (basal ganglia) and 37 mm (centrum semiovale) above the anterior commissure were counted. The construct validity of the scale was tested by examining associations with age, sex, vascular risk factors and neuroimaging markers of small vessel disease; white matter hyperintensities, lacunes and cerebral microbleeds. Associations with cross sectional global and domain specific cognition were also examined. RESULTS: The rating scale had excellent inter-rater reliability (intraclass correlation coefficient in basal ganglia 0.82 and centrum semiovale 0.96), good intra-rater reliability (ICC in basal ganglia 0.72 and centrum semiovale 0.87) and reasonable concurrent validity with an existing perivascular spaces scale (Spearman rho = 0.49, p < .001). There was a median of four basal ganglia and zero centrum semiovale perivascular spaces. Basal ganglia perivascular spaces were more common in men and associated with the other neuroimaging markers. Perivascular spaces in either location were not independently associated with global or domain specific cognitive impairment. CONCLUSION: The new rating scale is easy to use, quick, has good psychometric properties and performs better than existing scales in a community dwelling older cohort. Further studies are needed to validate the scale in more diverse cohorts with greater cerebrovascular burden.

Neuroimaging and neuropathology indices of cerebrovascular disease burden: A systematic review.

OBJECTIVE: To systematically review the literature on the use of both neuroimaging and neuropathologic indices of cerebrovascular disease (CVD) burden, as estimation of this burden could have multiple benefits in the diagnosis and prognosis of cognitive impairment and dementia. METHODS: MEDLINE and EMBASE databases were searched (inception to June 2017) to obtain and then systematically review all pertinent neuroimaging and neuropathology studies, where an index of CVD was developed or tested. RESULTS: Twenty-five neuroimaging articles were obtained, which included 4 unique indices. These utilized a limited range of CVD markers from mainly structural MRI, most commonly white matter hyperintensities (WMH), cerebral microbleeds, and dilated perivascular spaces. Weighting of the constituent markers was often coarse. There were 7 unique neuropathology indices, which were heterogeneous in their regions sampled and lesions examined. CONCLUSION: There is increasing interest in indices of total CVD burden that incorporate multiple lesions, as traditional individual markers of CVD such as WMH only provide limited information. Neuropathologic indices are needed to validate neuroimaging findings. The studies clearly demonstrated proof of concept that information from multiple imaging measures of CVD provide more information, including a stronger association with cognitive impairment and dementia, than that provided by a single measure. There has been limited exploration of the psychometric properties of published indices and no comparison between indices. Further development of indices is recommended, including the use of data from diffusion tensor and perfusion imaging.