Reliability and validity of noninvasive imaging of internal carotid arterypseudo-occlusion

Background and Purpose-Our study evaluated noninvasive tests for the diagno sis of atheromatous internal carotid artery (ICA) pseudo-occlusion. Methods-Twenty patients (17 men, 3 women; mean age +/-SD, 64.3+/-11.6 years ) with angiographically proven atheromatous ICA pseudo-occlusion (20 vessel s) were prospectively examined with MR angiography (MRA; 2D and 3D time-of- flight techniques), color Doppler-assisted duplex imaging (CDDI) and power- flow imaging (PFI) with and without an intravenous ultrasonic contrast agen t. As a control group, 13 patients (13 men; mean+/-SD age, 63.0+/-9.0 years ) with angiographically proven ICA occlusion (13 vessels) were studied with the same techniques. For the determination of interobserver agreement (kap pa statistics), the findings of each diagnostic technique were read by 2 bl inded and independent observers who were not involved in patient recruitmen t and initial data acquisition. Specificity and sensitivity were calculated for all noninvasive techniques (observer consensus) in comparison to the s tandard of reference (intra-arterial angiography). Results-Interobserver reliabilities were kappa = 0.86 for intra-arterial an giography, kappa = 0.90 for unenhanced CDDI, kappa = 0.93 for enhanced CDDI , kappa = 0.93 for unenhanced PFI, kappa = 1.0 for enhanced PFI, kappa = 0. 93 for 2D MRA, and kappa = 0.77 for 3D MRA, respectively (P<0.0001). Specif icities and sensitivities were 0.92 and 0.70 for unenhanced CDDI, 0.92 and 0.83 for enhanced CDDI, 0.92 and 0.95 for unenhanced PFI, 1.0 and 0.94 for enhanced PFI, 1.0 and 0.65 for 2D MRA, and 0.89 and 0.47 for 3D MRA, respec tively. Conclusions-Advanced ultrasonographic techniques, especially PFI (with only 1 false-positive diagnosis of occlusion in the present series), can provid e reliable and valid data to differentiate between ICA pseudo-occlusion and complete occlusion. In contrast, time-of-flight MRA at its present state i s not capable of predicting minimal residual flow within a nearly occluded ICA.