Biomarkers identify the Binswanger type of vascular cognitive impairment. Issue 8 (August 2019)

• Record Type: Journal Article
• Authors: Barry Erhardt, Erik; Pesko, John C; Prestopnik, Jillian; Thompson, Jeffrey; Caprihan, Arvind; Rosenberg, Gary A
• Is Part Of: Journal of cerebral blood flow & metabolism. Volume 39:Issue 8(2019)
• Page Start: 1602
• Page End: 1612
• Publication Date: 2019-08
• Subjects: Binswanger's disease -- vascular cognitive impairment and dementia -- white matter hyperintensities -- random forests -- biomarkers; Cerebral circulation -- Periodicals; Brain -- Metabolism -- Periodicals; Brain -- Blood-vessels -- Periodicals; Cerebrovascular disease -- Periodicals; 612.824
• DOI: 10.1177/0271678X18762655
• Languages: English
• ISSNs: 0271-678X

Abstract

Binswanger's disease is a form of subcortical ischemic vascular disease (SIVD-BD) with extensive white matter changes. To test the hypothesis that biomarkers could improve classification of SIVD-BD, we recruited 62 vascular cognitive impairment and dementia (VCID) patients. Multimodal biomarkers were collected at entry into the study based on clinical and neuropsychological testing, multimodal magnetic resonance imaging (MRI), and cerebrospinal fluid (CSF) analysis. The patients' diagnoses were confirmed by long-term follow-up, and they formed a "training set" to test classification methods, including (1) subcortical ischemic vascular disease score (SIVDS), (2) exploratory factor analysis (EFA), (3) logistic regression (LR), and (4) random forest (RF). A subsequently recruited cohort of 43 VCID patients with provisional diagnoses were used as a "test" set to calculate the probability of SIVD-BD based on biomarkers obtained at entry. We found that N-acetylaspartate (NAA) on proton magnetic resonance spectroscopy ( 1 H-MRS) was the best variable for classification, followed by matrix metalloproteinase-2 in CSF and blood–brain barrier permeability on MRI. Both LR and RF performed better in diagnosing SIVD-BD than either EFA or SIVDS. Two-year follow-up of provisional diagnosis patients confirmed the accuracy of statistically derived classifications. We propose that biomarker-based classification methods could diagnose SIVD-BD patients earlier, facilitating clinical trials.