Background: The optimal duplex ultrasound (DUS) velocity criteria to determine in-stent carotid restenosis are controversial. We previously reported the optimal DUS velocities for >or=30% in-stent restenosis. This prospective study will further define the optimal velocities in detecting various severities of in-stent restenosis: >or=30%, >or=50%, and 80% to 99%.
Methods: The analysis included 144 patients who underwent carotid artery stenting as a part of clinical trials. All patients had completion arteriograms and underwent postoperative carotid DUS imaging, which was repeated at 1 month and every 6 months thereafter. Patients with peak systolic velocities (PSVs) of the internal carotid artery (ICA) of >or=130 cm/s underwent carotid computed tomography (CT)/angiogram. The PSVs and end-diastolic velocities of the ICA and common carotid artery (CCA) and the PSV of the ICA/CCA ratios were recorded. Receiver operating characteristic curve (ROC) analysis was used to determine the optimal velocity criteria for the diagnosis of >or=30, >or=50, and >or=80% restenosis.
Results: The mean follow-up was 20 months (range, 1-78 months). Available for analysis were 215 pairs of imaging (DUS vs CTA/angiography) studies. The accuracy of CTA vs carotid arteriogram was confirmed in a subset of 22 patients (kappa = 0.81). The ROC analysis demonstrated that an ICA PSV of >or=154 cm/s was optimal for >or=30% stenosis with a sensitivity of 99%, specificity of 89%, positive-predictive value (PPV) of 96%, negative-predictive value (NPV) of 97%, and overall accuracy (OA) of 96%. An ICA EDV of 42 cm/s had sensitivity, specificity, PPV, NPV, and OA in detecting >or=30% stenosis of 86%, 62%, 87%, 60%, and 80%, respectively. An ICA PSV of >or=224 cm/s was optimal for >50% stenosis with a sensitivity of 99%, specificity of 90%, PPV of 99%, NPV of 90%, and OA of 98%. An ICA EDV of 88 cm/s had sensitivity, specificity, PPV, NPV, and OA in detecting >or=50% stenosis of 96%, 100%, 100%, 100%, 53%, and 96%. An ICA/CCA ratio of 3.439 had sensitivity, specificity, PPV, NPV, and OA in detecting >or=50% stenosis of 96%, 100%, 100%, 100%, 58%, and 96%, respectively. An ICA PSV of >or=325 cm/s was optimal for >80% stenosis with a sensitivity of 100%, specificity of 99%, PPV of 100%, NPV of 88%, and OA of 99%. An ICA EDV of 119 cm/sec had sensitivity, specificity, PPV, NPV, and OA in detecting >or=80% stenosis of 99%, 100%, 100%, 100%, 75%, and 99%, respectively. The PSV of the stented artery was a better predictor for in-stent restenosis than the end-diastolic velocity or ICA/CCA ratio.
Conclusion: The optimal DUS velocity criteria for in-stent restenosis of >or=30%, >or=50%, and >or=80% were the PSVs of 154, 224, and 325 cm/s, respectively.