Combining Serum Cystatin C And Urinary N-acetyl-β-D-glucosidase To Establish A Clinical Warning Model Of Acute Kidney Injury After Cardiac And Aortic Surgery

Purpose and significance of the studyAcute kidney injury(AKI)is the most common complication in the perioperative period of cardiovascular surgery and is associated with high mortality and poor outcomes.Early diagnosis of AKI is important due to the lack of effective targeted therapies.Clinical warning models based on Combination of non-homologous renal biomarkers are expected to improve the diagnostic efficacy of AKI.However,no conclusive combination of markers has been established.Serum cystatin C(s Cys C)and urinary N-acetyl-β-D-glucosidase(u NAG)are widely used biomarkers in clinical practice,representing glomerular filtration and tubular damage,respectively.This study is the first to prospectively analyse the predictive efficacy of s Cys C and urinary u NAG in predicting postoperative AKI of cardiac and aortic surgery and to establish an early warning model to provide new ideas for optimal management of AKI.Ideas and methods of the studyPart I: Consecutive patients admitted to our hospital for surgical procedures for thoracic aortic dissection(TAD)were tested for immediate postoperative s Cys C and u NAG.Risk factors for postoperative AKI were analysed using univariate and multifactorial logistic regression.The weights of each clinical factor were evaluated by random forest(RF).The predictive efficacy of s Cys C,u NAG and the combination of both in predicting AKI after TAD was assessed by the area under the subject operating characteristic curve(ROC-AUC),and the predictive cut-off points for both were explored.Part 2: Patients undergoing cardiac surgery in our hospital were prospectively enrolled.s Cys C and u NAG were measured in the immediate postoperative period.Independent risk factors for the postoperative AKI were analyzed.The weight of each clinical variable was evaluated by RF,and the predictive efficacy of s Cys C,u NAG and the combination of both in postoperative AKI of cardiac surgery was assessed by ROC-AUC.Part 3: Patients enrolled in parts 1 and 2 were used as the modelling cohort,and the patients continued to be included to create the validation cohort.Lasso regression was utilized to select predictors to establish warning models for AKI after TAD and postoperative AKI for cardiovascular surgery,respectively.The predictive performance and clinical value of the AKI warning model based no the combination of s Cys C and u NAG was evaluated by ROC-AUC,calibration curve,clinical decision curve(DCA),incremental discrimination improvement(IDI),net reclassification index(NRI)and external validation of the validation set.The results of the study were translated into practical application by developing a risk nomogram and a computational software.ResultsPart I: A total of 621 patients were included,and the incidence of AKI after TAD was50.6%.Multifactorial analysis showed that baseline creatine(P<0.01),the history of hypertension(P=0.028),bilateral renal artery involvement(P<0.05),cardiopulmonary bypass(CPB)time(P<0.01),preoperative D-dimer(P<0.01),postoperative s Cys C(P<0.01)and u NAG(P<0.01)were independent risk factors for AKI after TAD.The ROC-AUC was 0.864 and 0.802 for s Cys C and u NAG respectively,with a prediction cut-off of 1.395 mg/L and31.90 U/g Cre respectively.The combination of the two using binary logistic improved the prediction by increasing the ROC-AUC to 0.885(P<0.05).Part 2: A total of 689 patients were included and the incidence of postoperative AKI of cardicac surgery was 29.2%.Age(P<0.01),the history of hypertension(P<0.01),CPB time(P<0.01),postoperative s Cys C(P<0.01),u NAG(P<0.01),N-terminal pro-B-type natriuretic peptide(NT-pro BNP)(P<0.01)and cardiac troponin I(c TNI)(P<0.01)were the cardiac independent risk factors for postoperative AKI,while high preoperative serum albumin(P<0.01)and high haemoglobin(P<0.01)were protective factors.s Cys C and u NAG alone were not prominent predictors of AKI,with ROC-AUC of 0.730 and 0.650 respectively,while the combination of the two improved the predictive efficacy,with an ROC-AUC of 0.758(P<0.05).Part 3: The clinical warning model based on the combination of s Cys C and u NAG significantly improved the predictive efficacy of AKI after TAD and the postoperative AKI of cardiac surgery(ROC-AUC increased to 0.909 and 0.830,respectively,P<0.05),and could effectively improve the clinical benefit of patients(NRI increased by 12.2% and 11.3%,and IDI increased by 15.5% and 5.0%,respectively).The joint models fitted well through external validation.The nomograms and calculation procedures based on the models had ideal predictive performance and good clinical application.ConclusionThe clinical warning models of AKI based on the combination of the functional marker s Cys C and the renal tubular injury marker u NAG significantly improve the identification of postoperative AKI of cardiac and aortic surgery,and the nomograms as well as calculation procedures based on them are accurate and intuitive.Results from our study provides a new means of optimising the diagnosis and management of AKI.