Multivariate Pattern Analysis of Perfusion and Oxygenation Impairment in Asymptomatic Carotid Artery Stenosis

Hemodynamic impairment may contribute to stroke risk and cognitive decline in asymptomatic internal carotid artery stenosis (ICAS). Therefore, multimodal MRI-based quantification of hemodynamic impairment could inform improved treatment decisions. While gross interhemispheric hemodynamic imbalances have been reported in ICAS, identifying more spatially resolved patterns of disease-related alterations may be promising to harness the full potential of hemodynamic MRI.

In this feasibility study, we investigated the spatial topography of ICAS-related impairments by applying scaled subprofile model principal component analysis (SSM-PCA) to cerebral blood flow (CBF), relative oxygen extraction fraction (rOEF), and oxygen extraction capacity (OEF^max) data of 21 unilateral ICAS patients and 25 healthy controls (HC).

We found spatially extended, partly overlapping disease-related patterns for CBF and OEF^max, but not rOEF. CBF (area under the curve [AUC] = 0.95) but not OEF^max (AUC = 0.72) SSM-PCA scores distinguished ICAS patients and HC better than interhemispheric lateralizations (AUC = 0.75/0.73). SSM-PCA scores were only partly explained by interhemispheric lateralization (R² = -0.27/0.38), indicating complementary information. Critically, ICAS patients with higher OEF^max SSM-PCA scores (z ≥ 1) demonstrated higher stenotic degrees and lower cognitive performance (p < 0.05) without differing in interhemispheric lateralization (p > 0.05).

We demonstrated the feasibility of SSM-PCA in ICAS and obtained novel insights into complex hemodynamic impairment patterns and their association with cognitive function.