Predicting Cardiac Surgery-Associated Acute Kidney Injury Using a Combination of Clinical Risk Scores and Urinary Biomarkers

Abstract

Background Prediction, early diagnosis, and therapy of cardiac surgery-associated acute kidney injury (CSA-AKI) are challenging. We prospectively tested a staged approach to identify patients at high risk for CSA-AKI combining clinical risk stratification and early postoperative quantification of urinary biomarkers for AKI. Methods All patients, excluding those on chronic hemodialysis, undergoing scheduled surgery with cardiopulmonary bypass between August 2015 and July 2016 were included. First, patients were stratified by calculating the Cleveland clinic score (CCS) and the Leicester score (LS). In high-risk patients (defined as LS > 25 or CCS > 6), urinary concentrations of biomarkers for AKI ([TIMP-2]*[IGFBP-7]) were evaluated 4 hours postoperatively. CSA-AKI was observed until postoperative day 6 and classified using the Kidney Disease: Improving Global Outcomes criteria. Results AKI occurred in 352 of613 patients (54%). In high-risk patients, AKI occurred more frequently than in low-risk patients (66 vs. 49%; p = 0.001). In-hospital mortality after AKI stage 2 (15%) or AKI stage 3 (49%) compared with patients without AKI (1.8%; p = 0.001) was increased. LS was predictive for all stages of AKI (area under the curve [AUC] 0.601; p < 0.001) with a poor or fair accuracy, while CCS was only predictive for stage 2 or 3 AKI (AUC 0.669; p < 0.001) with fair accuracy. In 133 high-risk patients, urinary [TIMP-2]*[IGFBP-7] was significantly predictive for all-stage AKI within 24 hours postoperatively (AUC 0.63; p = 0.017). However, for the majority of AKI (55%), which occurred beyond 24 hours postoperatively, urinary [TIMP-2]*[IGFBP-7] was not significantly predictive. Sensitivity for all-stage AKI within 24 hours was 0.38 and specificity was 0.81 using a cutoff value of 0.3. Conclusion CSA-AKI is a relevant and frequent complication after cardiac surgery. Patients at high risk for CSA-AKI can be identified using clinical prediction scores, however, with only poor to fair accuracy. Due to its weak test performance, urinary [TIMP-2]*[IGFBP-7] quantification 4 hours postoperatively does not add to the predictive value of clinical scores.