Predictive models for one-year renal function improvement in patients with atherosclerotic renal artery stenosis (ARAS) following renal artery revascularization: a real-world study from China

The efficacy of interventional treatment for atherosclerotic renal artery stenosis (ARAS) in improving renal function remains controversial. The aim of this study was to develop a predictive model for estimating the probability of renal function response within one year following renal artery intervention in patients with ARAS. We retrospectively analyzed ARAS patients with renal artery intervention from January 2020 to December 2024 at Peking University First Hospital consecutively. Candidate predictors were selected using least absolute shrinkage and selection operator (LASSO) regression. A multivariable logistic regression model and a support vector machine (SVM) model were developed and internally validated using nested cross-validation. Discrimination was assessed by the area under the precision-recall curve (AUPRC), and calibration by the Brier score. 224 cases were enrolled, among which 47 patients exhibited renal function response following interventional treatment. LASSO regression identified four predictors (age, baseline renal function, diabetes, and bilateral renal artery stenosis) for model development. According to the results of internal validation, the mean AUPRC of logistic regression model was 0.701 (95% CI: 0.580, 0.834) and SVM model was 0.753 (95% CI: 0.642, 0.864). The logistic regression model demonstrated a mean Brier score of 0.150 (95% CI: 0.118, 0.182), whereas the SVM model with a mean Brier score of 0.148 (95% CI: 0.105, 0.193), indicating good calibration and stable performance. After further external validation, the models may serve as a risk stratification and post-procedural prognostic counseling aid for both patients and physicians.