Lack of agreement between transcranial Doppler and Xe/CT cerebral blood flow in the detection of cerebrovascular reserves in patients with carotid occlusive disease
Introduction: Transcranial Doppler ultrasonography (TCD) has been used to identify patients with compromised cerebrovascular reserve(CVR). Patients with compromised CVR have been shown to have an increased risk for stroke. Transcranial Doppler, however, has not been compared to the high resolution quantitative cerebral blood flow (CBF) method of stable xenon enhanced computed tomography (Xe/CT). Methods: Cerebrovascular reserves before and after acetazolamide administration (1 gram, i.v.) were measured by TCD insonation of the middle cerebral artery (MCA) and Xe/CT CBF. Thirty-one patients with carotid occlusive disease were studied. Calculations were based on two methods; for the first method Xe/CT CBF values were averaged over 4 levels of the MCA, the second method used the value for the MCA level with the lowest percent change. The threshold for identifying compromised CVR versus non-compromised CVR was 0% change(two standard deviations below normal)for both the TCD and the Xe/CT CBF analyses. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated. Results: For the average MCA territory analyses TCD compared to Xe/CT CBF had a 42% sensitivity, a 90% specificity, a 50% positive predictive value and a 86% negative predictive value. The comparison of TCD to Xe/CT using the lowest MCA territory analysis for the Xe/CT CBF values also gave a sensitivity of 42%. The specificity and positive predictive value were both 100% and the negative predictive value was 72%. Discussion/Conclusion: The sensitivity of TCD compared to Xe/CT CBF indicates a significant lack of agreement in identifing those patients with compromised CVR. The negative predictive value indicates that the number of patients identified with positive CVR (i.e. negative for compromised reserves)is not the same as with Xe/CT CBF. Physicians should consider this lack of correlation when using TCD to identify patients with compromised CVR. A direct prospective comparison to PET or a clinical prediction of stroke by Xe/CT needs to be done in order to determine the best technology.