Objective evaluation of cerebrovascular reactivity for acetazolamide predicts cerebral hyperperfusion after carotid artery stenting: Comparison with region of interest methods

Background and purpose. - Hemodynamic impairments are considered risk factors of cerebral hyperperfusion after carotid artery stenting (CAS); measurement by Single-photon emission computed tomography (SPECT) using a subjective region of interest (ROI) method lacks consistency and reproducibility. Materials and methods. - The present study compared objective perfusion analysis (stereotactic extraction estimation [ SEE] method) with the ROI method for preoperative SPECT to predict the hyperperfusion phenomenon (HPP) after CAS. Preoperative resting asymmetry index (cerebral blood flow [CBF] ratio from the affected to unaffected hemisphere) and cerebrovascular reactivity (CVR) to acetazolamide were measured by N-isopropyl-p-[I-123]odoamphetamine SPECT using the SEE and ROI method in 84 patients. CBF was also measured the day after CAS. Perfusion data with the highest area under the curve (AUC) by receiver-operating characteristic (ROC) analysis was considered a perfusion risk factor of HPP. Multivariate analyses for clinical characteristics and perfusion risk factors were performed to determine predictors of HPP. Results. - The HPP was observed in 10 patients (11.9%). Female sex, contralateral stenosis, and degree of stenosis were significantly associated with HPP development on univariate analysis, and symptomatic stenosis was not found to be a significant factor. On SPECT analysis, CVR in the MCA area by SEE method had the highest AUC (0.981). Multivariate analysis showed that CVR in the MCA area was a significant predictor of HPP (P = 0.041). To predict hyperperfusion, the ROC curve of the CVR showed a cutoff value of -0.60%, sensitivity of 94.6%, and specificity of 100% (P< 0.001). Conclusions. - Objective SEE method had better a predictive capability than ROI method to identify risk of hyperperfusion after CAS. (C) 2018 Elsevier Masson SAS. All rights reserved.