Duplex criteria for determination of in-stent stenosis after angioplasty and stenting of the superficial femoral artery

Abstract

Endovascular intervention is considered first-line therapy for most superficial femoral artery (SFA) occlusive disease. Duplex ultrasound (DU) criteria for SFA in-stent stenosis and correlation with angiographic data remain poorly defined. This study evaluated SFA-specific DU criteria for the assessment of SFA in-stent stenosis. From May 2003 to May 2008, 330 limbs underwent SFA angioplasty and stenting and were monitored by serial DU imaging. Suspected stenotic lesions underwent angiography and intervention when appropriate. Data pairs of DU and angiographically estimated stenosis <or=30 days of each other were analyzed. Seventy-eight limbs met these criteria, and 59 underwent reintervention. In-stent peak systolic velocity (PSV), the ratio of the stented SFA velocity/proximal SFA velocity, changes in ankle-brachial indices (ABIs), and the percentage of angiographic stenosis were examined. Linear regression and receiver operator characteristic (ROC) curve analyses were used to compare angiographic stenosis with PSV and velocity ratios (Vrs) to establish optimal criteria for determining significant in-stent stenosis. Mean follow-up was 16.9 +/- 8.3 months. Of the 59 limbs that underwent reintervention, 37 (63%) were symptomatic, and 22 (37%) underwent reintervention based on DU findings alone. Linear regression models of PSV and Vr vs degree of angiographic stenosis showed strong adjusted correlation coefficients (R(2) = 0.60, P < .001 and R(2) = 0.55, P < 0.001, respectively). ROC curve analysis showed that to detect a >or=50% in-stent stenosis, a PSV >or=190 had 88% sensitivity, 95% specificity, a 98% positive predictive value (PPV), and a 72% negative predictive value (NPV); for Vr, a ratio of >1.50 had 93% sensitivity, 89% specificity, a 96% PPV, and a 81% NPV. To detect >or=80% in-stent stenosis, a PSV >or=275 had 97% sensitivity, 68% specificity, a 67% PPV, and a 97% NPV; a Vr ratio >or=3.50 had 74% sensitivity, 94% specificity, a 77% PPV, and a 88% NPV. Combining a PSV >or=275 and a Vr >or=3.50 to determine >or=80% in-stent stenosis had 74% sensitivity, 94% specificity, a 88% PPV, and a 85% NPV; odds ratio was 42.17 (95% confidence interval, 10.20-174.36, P < .001) to predict >or=80% in-stent stenosis. A significant drop in ABI (>0.15) correlated with a >62% in-stent stenosis, although the adjusted correlation coefficients was low (R(2) = 0.31, P = .02). PSV and Vr appear to have a significant role in predicting in-stent stenosis. To determine >or=80% stenosis, combining PSV >or=275 cm/s and Vr >or=3.50 is highly specific and predictive.