Prediction of in-stent restenosis (ISR) is clinically important for patients with peripheral artery disease in their superficial femoral arteries (SFA) who have been treated with stenting. The aim of this study was to construct and validate a predictive model for ISR after SFA stenting based on a series of clinical and ultrasonic parameters. This retrospective study included 381 patients who were treated with self-expanding bare nitinol stents in their SFA at our hospital between January 1, 2018, and January 1, 2022. These patients were randomly allocated to a training cohort (266 patients) or a validation cohort (115). Clinical and ultrasonic parameters related to ISR (>50%) in the SFA at 12 months were derived by univariable and multivariable logistic regression analyses to create a nomogram model predictive of risk of ISR. Receiver operating characteristic (ROC) curve analyses were used to assess the recognition ability of the model. A calibration curve was used to evaluate the model's calibration ability, and decision curve analysis was used to validate the nomogram's clinical utility. Logistic regression analyses showed that sex, echogenicity of the target plaque, preoperative arterial runoff score, preoperative popliteal artery flow rate, lesion length, and residual diameter were risk factors for ISR; these parameters were used to construct the nomogram model. Internal and external validation showed that the areas under the ROC curves were 0.82 (95%CI: 0.77-0.87) and 0.70 (95%CI: 0.60-0.79), respectively, suggesting good recognition ability of the model. Additionally, calibration curves for the predictive model indicated good calibration, and decision curve analysis demonstrated clinical utility of the model. This novel nomogram that predicts ISR after SFA stenting demonstrated excellent discriminatory power, calibration capacity, and clinical usefulness.
Read full abstract