AUTHOR=Mo Hanjun , Ye Fang , Chen Danxia , Wang Qizhe , Liu Ru , Zhang Panpan , Xu Yaxin , Cheng Xuelin , Mei Zhendong , Zheng Yan , Dai Yuxiang , Jiang Sunfang , Ge Junbo 

TITLE=A Predictive Model Based on a New CI-AKI Definition to Predict Contrast Induced Nephropathy in Patients With Coronary Artery Disease With Relatively Normal Renal Function

JOURNAL=Frontiers in Cardiovascular Medicine

VOLUME=8

YEAR=2021

URL=https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2021.762576

DOI=10.3389/fcvm.2021.762576

ISSN=2297-055X

ABSTRACT=<p><bold>Background:</bold> Contrast induced nephropathy (CIN) is a common complication in patients receiving intravascular contrast media. In 2020, the American College of Radiology and the National Kidney Foundation issued a new contrast induced acute kidney injury (CI-AKI) criteria. Therefore, we aimed to explore the potential risk factors for CIN under the new criteria, and develop a predictive model for patients with coronary artery disease (CAD) with relatively normal renal function (NRF).</p><p><bold>Methods:</bold> Patients undergoing coronary angiography or percutaneous coronary intervention at Zhongshan Hospital, Fudan University between May 2019 and April 2020 were consecutively enrolled. Eligible candidates were selected for statistical analysis. Univariate and multivariate logistic regression analyses were used to identify the predictive factors. A stepwise method and a machine learning (ML) method were used to construct a model based on the Akaike information criterion. The performance of our model was evaluated using the area under the receiver operating characteristic curves (AUC) and calibration curves. The model was further simplified into a risk score.</p><p><bold>Results:</bold> A total of 2,009 patients with complete information were included in the final statistical analysis. The results showed that the incidence of CIN was 3.2 and 1.2% under the old and new criteria, respectively. Three independent predictors were identified: baseline uric acid level, creatine kinase-MB level, and log (N-terminal pro-brain natriuretic peptide) level. Our stepwise model had an AUC of 0.816, which was higher than that of the ML model (AUC = 0.668, <italic>P</italic> = 0.09). The model also achieved accurate predictions regarding calibration. A risk score was then developed, and patients were divided into two risk groups: low risk (total score &lt; 10) and high risk (total score ≥ 10).</p><p><bold>Conclusions:</bold> In this study, we first identified important predictors of CIN in patients with CAD with NRF. We then developed the first CI-AKI model on the basis of the new criteria, which exhibited accurate predictive performance. The simplified risk score may be useful in clinical practice to identify high-risk patients.</p>