Resting-state BOLD fMRI-derived cerebrovascular reactivity for detecting crossed cerebellar diaschisis.

Abstract

Purpose: To study cerebellar cerebrovascular reactivity (CVR) derived from resting-state (RS) blood oxygen level-dependent (BOLD) as a potential imaging biomarker for crossed cerebellar diaschisis (CCD).

Methods: Thirty-nine patients with symptomatic unilateral cerebrovascular occlusion and eighteen age-matched healthy subjects underwent a simultaneous PET/MR scan. CCD was diagnosed as a contralateral cerebellar metabolic reduction present on ¹⁸F-FDG PET semi-quantitatively. RS BOLD-CVR maps were derived by extracting signals within specific frequency bands of RS BOLD. The cerebellar asymmetry index (CAI) calculated from RS BOLD-CVR and ¹⁸F-FDG PET data were compared using receiver operating characteristic and Bland-Altman analysis. Neurological function was determined with NIHSS and mRS.

Results: In our cohort, CCD was diagnosed in 21 of 39 (53.8%) patients on ¹⁸F-FDG PET. CCD+ patients exhibited larger infarct volume, more severe supratentorial CVR impairment, and worse functional status than CCD- patients at admission and 3.5 years follow-up. The area under the curve for RS BOLD-CVR CAI was 0.862 (95% confidence interval: 0.747-0.978) with a cutoff value of 8.2%, achieving a sensitivity of 81.0% and a specificity of 83.3% to detect CCD. The Bland-Altman analysis showed no significant proportional bias between PET CAI and CVR CAI (P = 0.848). CVR CAI was correlated with NIHSS (ρ = 0.352, p = 0.033) and mRS (ρ = 0.360, p = 0.029) at admission after adjusted for age and infarct volume.

Conclusion: RS BOLD-CVR demonstrates good sensitivity and specificity to detect diaschisis-related cerebellar cerebrovascular reactivity in patients with symptomatic unilateral cerebrovascular occlusion. Furthermore, participants with cerebellar CVR impairment and metabolic reduction tended to have poorer neurological function.