Count-based PET method for predicting ischemic stroke in patients with symptomatic carotid arterial occlusion.

PURPOSE To test the ability of a count-based positron emission tomographic (PET) method, without arterial sampling, for the measurement of regional cerebral oxygen extraction fraction (OEF) to predict ischemic stroke in patients with symptomatic carotid arterial occlusion. MATERIALS AND METHODS The outcome analysis of a blinded prospective study designed to determine if increased OEF was an independent predictor of stroke in patients with symptoms and with carotid occlusion was repeated by substituting a count-based method of OEF measurement for the original quantitative technique. The performance of the quantitative and count-based methods was assessed by using Kaplan-Meier cumulative survival functions (log-rank, [p < .05]). Receiver operating characteristic (ROC) curves for both methods were generated. RESULTS Thirteen ipsilateral strokes occurred during a mean follow-up of 3.1 years for 81 patients. All ipsilateral strokes occurred in 50 patients with increased count-based OEF (P = .002, sensitivity 100%, specificity 46%). Sixty-eight patients underwent complete quantitative studies, which allowed comparison of OEF methods. Both the count-based and the quantitative methods were predictive of stroke in this subgroup (P = .005 and .025, respectively). ROC analysis demonstrated a greater area under the curve for the count-based OEF method. CONCLUSION Count-based PET measurement of OEF without arterial sampling accurately predicts stroke in patients with carotid occlusion.