Abstract Body: Background: The patterns of neural functional changes induced by chronic hypoperfusion resulting from intracranial arterial stenosis and their relationship to cognitive performance are not yet fully understood, with unilateral asymptomatic middle cerebral artery stenotic-occlusive disease (MCAs/o) serving as an ideal model for investigation.
Methods: We established a discovery-validation research pipeline to identify functional abnormalities in the hypoperfused regions of MCAs/o patients and quantify them into a machine learning-based data-driven biomarker, i.e. hypoperfusion-based functional abnormalities (HFA). Pseudocontinuous arterial spin labeling imaging was used to identify hypoperfused regions (HypoR) in the discovery cohort. Functional changes within HypoR, including regional homogeneity (ReHo) and functional connectivity, were analyzed and quantitatively integrated using a logistic regression machine learning model to calculate the HFA after feature extraction. The correlation between HFA and cognitive scores was then assessed, and the findings were validated in the independent cohort (Figure 1).
Results: The discovery cohort included 44 MCAs/o patients (LMCAs/o: n=22, age 56±11 years, 13 male; RMCAs/o: n=22, age 49±14 years, 16 male) and 35 matched normal controls (age 52±11 years, 15 male), in which MCAs/o patients showed poorer performance on bilateral hand-grooved pegboard tests (GPT) compared to controls (p<0.05). We identified 45 regions of interest (ROIs) in the stenotic hemisphere with significantly elevated arterial transit time, defining them as hypoperfused areas. We found that MCAs/o patients had increased ReHo(FDR-p<0.05) and reductions in HypoR~non-HypoR connections(FDR-p<0.05), coupled with increased HypoR~HypoR connections(FDR-p<0.05). HFA was then calculated based on functional features, which effectively identified early functional changes in MCAs/o patients(AUC=0.994) and correlated with bilateral GPT(p<0.001) (Figure 2). Validation in an independent cohort(20 MCAs/o, 16 controls) confirmed these findings(AUC=0.948; HFA~bilateral GPT p<0.01, Figure 3).
Conclusions: Chronic hypoperfusion caused by unilateral MCAs/o can lead to a specific pattern of functional changes characterized by increased local neural activity. The quantified biomarker of this pattern can be used to monitor the decline in cognitive function. Our study provides a foundation for further research into the neural damage caused by chronic hypoperfusion.