Screening for significant atherosclerotic renal artery stenosis with a regression model in patients undergoing transradial coronary angiography/intervention.

OBJECTIVE: Early detection of atherosclerotic renal artery stenosis (ARAS) is clinically important with respect to blood pressure control, prevention of renal insufficiency, and even improving survival. We investigated whether the presence of significant ARAS (luminal diameter narrowing ≥70%) could be predicted using a logistic regression model before coronary angiography/intervention.
METHODS: Initially, we developed a logistic regression model for detecting significant ARAS based upon clinical and angiographic features and biochemical measurements in a cohort of 1813 patients undergoing transfemoral coronary and renal angiography. This model was then prospectively applied to an additional 495 patients who received transradial renal angiography to ascertain its predictive accuracy for the presence of significant ARAS.
RESULTS: Multivariate regression analysis revealed that older age (≥65 years), resistant hypertension, type 2 diabetes, creatinine clearance (Ccr) ≤60 ml/min, and multivessel coronary disease were independent predictors for significant ARAS. A logistic regression model for detecting ARAS by incorporating conventional risk factors and multivessel coronary disease was generated as: P/(1-P)=exp(-2.618+1.112[age≥65 years]+1.891[resistant hypertension]+0.453[type 2 diabetes]+0.587[Ccr≤60 ml/min]+2.254[multivessel coronary disease]). When this regression model was prospectively applied to the additional 495 patients undergoing transradial coronary and renal angiography, significant ARAS could be detected with a sensitivity of 81.2%, specificity of 88.9%, and positive and negative predictive accuracies of 53.8% and 96.7%, respectively.
CONCLUSIONS: The logistic regression model generated in this study may be useful for screening for significant ARAS in patients undergoing transradial coronary angiography/intervention.