Flow-metabolism uncoupling in patients with asymptomatic unilateral carotid artery stenosis assessed by multi-modal magnetic resonance imaging

Abstract

Oxygen extraction (OEF), oxidative metabolism (CMRO₂), and blood flow (CBF) in the brain, as well as the coupling between CMRO₂ and CBF due to cerebral autoregulation are fundamental to brain's health. We used a clinically feasible MRI protocol to assess impairments of these parameters in the perfusion territories of stenosed carotid arteries. Twenty-nine patients with unilateral high-grade carotid stenosis and thirty age-matched healthy controls underwent multi-modal MRI scans. Pseudo-continuous arterial spin labeling (pCASL) yielded absolute CBF, whereas multi-parametric quantitative blood oxygenation level dependent (mqBOLD) modeling allowed imaging of relative OEF and CMRO₂. Both CBF and CMRO₂ were significantly reduced in the stenosed territory compared to the contralateral side, while OEF was evenly distributed across both hemispheres similarly in patients and controls. The CMRO₂-CBF coupling was significantly different between both hemispheres in patients, i.e. significant interhemispheric flow-metabolism uncoupling was observed in patients compared to controls. Given that CBF and CMRO₂ are intimately linked to brain function in health and disease, the proposed easily applicable MRI protocol of pCASL and mqBOLD imaging might serve as a valuable tool for early diagnosis of potentially harmful cerebral hemodynamic and metabolic states with the final aim to select clinically asymptomatic patients who would benefit from carotid revascularization therapy.

Keywords: Carotid stenosis; arterial spin labeling; cerebral blood flow; cerebral metabolic rate of oxygen consumption; multi-parametric quantitative blood oxygenation level dependent imaging; perfusion.

Figure 1.

Calibration of pCASL-based CBF. Left inset: Maps of mean CBF in gray matter derived from pCASL in 9 young healthy subjects (top) and from ¹⁵O-water-PET in 13 young subjects (bottom) overlaid on a T1-weighted image in MNI space, thus allowing parcellation of gray matter space into several neuroanatomical areas (Table S1). Comparison of mean CBF from pCASL (black bars) and PET (white bars) shows good correspondence of absolute values except for region 17 (visual cortex). This discrepancy between ¹⁵O-water-PET and pCASL in visual cortex has been reported. *Two-sample t-test: p < 0.05.

Figure 2.

Multi-parametric imaging in carotid stenosis. (a) Template of combined vascular territories of the anterior and middle cerebral artery defining the left (red) and right (blue) anterior circulation as shown earlier. (b) Parameter maps of R2′, rCBV, rOEF, CBF and rCMRO₂ in a healthy elderly control (top row) and a left-sided carotid stenosis patient (bottom row) within gray matter in MNI space. All parameters are reasonably symmetric in the healthy subject. However, in the patient, there are reductions of CBF and rCMRO₂ on the side of the stenosis, whereas R2′, rCBV and the resulting rOEF maps are quite symmetrical.

Figure 3.

Comparison of averaged hemodynamic and metabolic parameters within the anterior circulation in healthy controls and patients. Paired scatter plots depicting mean R2′, rCBV, rOEF, CBF and rCMRO₂ values between the (a) left versus right hemispheres in control subjects, and (b) affected (i.e. ipsilateral to the stenosis) versus unaffected (i.e. contralateral) sides of the anterior circulation in patients, respectively. Significant intra-subject side differences were observed only for CBF and rCMRO₂ in patients (paired t-test, p < 0.001 for both parameters). The red dashed line indicates the mean parameter values of the respective side.

Figure 4.

Profiles of parameter maps in the affected versus unaffected hemispheres in patients. Maps of rOEF, CBF and rCMRO₂ from patients were flipped along the mid-sagittal axis so that all hemispheres ipsilateral to the stenosis (i.e. affected side) were located on the same side and could be compared to the contralateral side (which was presumed to be the unaffected side). Top row shows the patients' averaged rOEF, CBF and rCMRO₂ maps normalized to MNI space. The blue and red vertical lines across the images indicate the affected and unaffected hemispheres, respectively, where the profile A–B is on the affected side and the profile C–D is on the unaffected side. Bottom row: Exemplary profiles of affected (blue) and unaffected (red) hemispheres were created within an axial slice at MNI coordinate z=17 along the y-direction (y_MNI) at x=−48 (affected side, blue lines and corresponding plots from A–B profile in the top row) and x=48 (unaffected side, red lines and corresponding plots from C–D profile in the top row).

Figure 5.

Interhemispheric flow-metabolism coupling. Comparison of rCMRO₂-CBF coupling across gray matter areas in the left (red) and right (blue) hemispheres in (a) two exemplary controls and (b) two representative patients. In (a), the rCMRO₂-CBF coupling values in controls are shown for the left and right hemispheres, with red and blue circles, respectively. In (b), the rCMRO₂-CBF coupling values in patients are shown for the affected (i.e. ipsilateral to the stenosis which is the right hemisphere) and unaffected hemispheres (i.e. contralateral to the stenosis which is the left hemisphere), with blue and red circles, respectively. Note that the slope ratios of the linear fitting curves for the rCMRO₂-CBF coupling in each hemisphere differ more in patients (2.05 and 2.82, which correspond to 38% difference between the two patients) compared to the control subjects (0.92 and 1.01, which correspond to 10% difference between the two subjects). (c) The individual rCMRO₂-CBF couplings (i.e. ratios between the fitting slopes shown in (a) and (b)) were significantly different between controls and patients (two-sample t-test: p = 0.04).