Comparison of PET Oxygen Extraction Fraction Methods for the Prediction of Stroke Risk

PET measurement of increased oxygen extraction fraction (OEF) identifies patients at high risk for subsequent stroke. OEF methodology remains controversial. In this study we compare the sensitivity and specificity of absolute OEF measurements with ipsilateral-to-contralateral ratios of absolute OEF and count-based OEF estimates. Multivariate analyses of OEF methods were performed using data from patients with symptomatic carotid artery occlusion (n = 68). Outcome and receiver operating characteristic (ROC) curve analyses were performed. All 3 methods were predictive of stroke risk in univariate analysis. Only the count-based method remained significant in multivariate analysis. The area under the ROC curve was greatest for the count-based ratio: 0.815 versus 0.769 (absolute) and 0.737 (ratios of absolute). All 3 methods are predictive of stroke risk in patients with unilateral carotid artery occlusion. ROC curve analysis is useful for selecting optimal thresholds for maximal sensitivity and specificity.