Objectives Vascular cognitive impairment (VCI) is a heterogeneous group of cerebrovascular diseases secondary to large and small vessel disease. We hypothesised that biomarkers obtained early in the disease could identify a homogeneous subpopulation with small vessel disease.

Methods We obtained disease markers in 62 patients with VCI that included neurological findings, neuropsychological tests, multimodal MR and cerebrospinal fluid measurements of albumin ratio, matrix metalloproteinases (MMPs), amyloid-β_1–42 and phosphorylated-τ181. Proton MR spectroscopic imaging showed ischaemic white matter and permeability of the blood-brain barrier (BBB) was measured with dynamic contrast-enhanced MRI. We constructed a 10-point Binswanger disease score (BDS) with subjective and objective disease markers. In addition, an objective set of biomarkers was used for an exploratory factor analysis (EFA) to select patients with BD. Patients were followed for an average of 2 years to obtain clinical consensus diagnoses.

Results An initial BDS of 6 or greater was significantly correlated with a final diagnosis of BD (p<0.05; area under the curve (AUC)=0.79). EFA reduced nine objective biomarkers to four factors. The most predictive of BD was the factor containing the inflammatory biomarkers of increased BBB permeability, elevated albumin index and reduced MMP-2 index (factor 2; AUC=0.78). Both measures independently predicted a diagnosis of BD, and combining them improved the diagnostic accuracy.

Conclusions Biomarkers predicted the diagnosis of the BD type of subcortical ischaemic vascular disease. Using pathophysiological biomarkers to select homogeneous groups of patients needs to be tested in targeted treatment trials.