Preoperative Cerebral Oxygen Extraction Fraction Imaging Generated from 7T MR Quantitative Susceptibility Mapping Predicts Development of Cerebral Hyperperfusion following Carotid Endarterectomy

BACKGROUND AND PURPOSE:Preoperative hemodynamic impairment in the affected cerebral hemisphere is associated with the development of cerebral hyperperfusion following carotid endarterectomy. Cerebral oxygen extraction fraction images generated from 7T MR quantitative susceptibility mapping correlate with oxygen extraction fraction images on positron-emission tomography. The present study aimed to determine whether preoperative oxygen extraction fraction imaging generated from 7T MR quantitative susceptibility mapping could identify patients at risk for cerebral hyperperfusion following carotid endarterectomy.MATERIALS AND METHODS:Seventy-seven patients with unilateral internal carotid artery stenosis (≥70%) underwent preoperative 3D T2*-weighted imaging using a multiple dipole-inversion algorithm with a 7T MR imager. Quantitative susceptibility mapping images were then obtained, and oxygen extraction fraction maps were generated. Quantitative brain perfusion single-photon emission CT was also performed before and immediately after carotid endarterectomy. ROIs were automatically placed in the bilateral middle cerebral artery territories in all images using a 3D stereotactic ROI template, and affected-to-contralateral ratios in the ROIs were calculated on quantitative susceptibility mapping–oxygen extraction fraction images.RESULTS:Ten patients (13%) showed post-carotid endarterectomy hyperperfusion (cerebral blood flow increases of ≥100% compared with preoperative values in the ROIs on brain perfusion SPECT). Multivariate analysis showed that a high quantitative susceptibility mapping–oxygen extraction fraction ratio was significantly associated with the development of post-carotid endarterectomy hyperperfusion (95% confidence interval, 33.5–249.7; P = .002). Sensitivity, specificity, and positive- and negative-predictive values of the quantitative susceptibility mapping–oxygen extraction fraction ratio for the prediction of the development of post-carotid endarterectomy hyperperfusion were 90%, 84%, 45%, and 98%, respectively.CONCLUSIONS:Preoperative oxygen extraction fraction imaging generated from 7T MR quantitative susceptibility mapping identifies patients at risk for cerebral hyperperfusion following carotid endarterectomy.

Cerebral hyperperfusion following carotid endarterectomy (CEA) has been defined as a substantial increase in ipsilateral cerebral blood flow well above the metabolic demands of brain tissue following surgical repair of carotid stenosis.^1,2 Cerebral hyperperfusion syndrome after CEA is a complication of cerebral hyperperfusion;³ its characteristic features include unilateral headache, pain in the face or eyes, seizures, and focal symptoms secondary to intracerebral hemorrhage or cerebral edema.^1–4 Intracerebral hemorrhage has a low incidence (1%), but patients with this condition have a poor prognosis.⁵ Moreover, several studies have found that post-CEA hyperperfusion, even when asymptomatic, causes slight but diffuse damage to the ipsilateral cerebral cortex and white matter.^3,6,7 This damage that occurs after CEA hyperperfusion is a principal cause of the postoperative cognitive impairment observed in 10% of patients following CEA.^3,6,7Cerebrovascular autoregulatory mechanisms operate through dilation of precapillary resistance vessels that maintain CBF when reductions in cerebral perfusion pressure occur, and this is referred to as stage 1 ischemia.^3,8–10 However, the autoregulatory mechanism provides insufficient compensation for severe decreases in cerebral perfusion pressure, which then leads to decreased CBF, referred to as misery perfusion or stage 2 ischemia.^3,8–10 Thus, misery perfusion, which is defined as marginally sufficient cerebral blood supply relative to cerebral metabolic demand, is a situation with severely impaired cerebral hemodynamics.⁸ This condition occurs in patients with chronic steno-occlusive diseases of the internal carotid artery.⁸The risk factors for cerebral hyperperfusion include high-grade stenosis, poor collateral blood flow, contralateral carotid occlusion, and long-standing hypertension, and they often result in impaired cerebral hemodynamics.^11–14 When normal perfusion pressure is rapidly restored after CEA, hyperperfusion may occur in regions of the brain with impaired autoregulation due to chronic ischemia. This hypothesis is like the “normal perfusion pressure breakthrough” theory of Spetzler et al.^13,15 Indeed, preoperative misery perfusion in the affected cerebral hemisphere is reportedly associated with the development of cerebral hyperperfusion following CEA or carotid stent placement for cervical ICA stenosis.^16,17Misery perfusion is principally detected as an increased oxygen extraction fraction (OEF) on positron-emission tomography.⁸ Several approaches have been attempted to measure OEF by using MR imaging techniques.¹⁸ In general, these techniques use blood oxygen level–dependent effects induced by differences in magnetic susceptibility between oxy- and deoxyhemoglobin to quantify oxygenation in venous structures and/or brain parenchyma.^19–21 Quantitative susceptibility mapping (QSM) is a postprocessing technique for quantifying the magnetic susceptibility of venous structures and brain parenchyma from T2*-weighted magnitude/phase images, which can be easily obtained by commercial scanners.²² Indeed, a recent study has introduced an OEF measurement method based on the QSM technique and has demonstrated that cerebral OEF images generated from QSM at 7T MR imaging correlate with OEF images on PET and provide high sensitivity and high specificity for detecting misery perfusion in the middle cerebral artery territory in patients with unilateral chronic ICA or MCA steno-occlusive disease.²³The purpose of the present study was to determine whether preoperative OEF imaging generated from 7T MR QSM could identify patients at risk for cerebral hyperperfusion following CEA.