Preoperative plasma growth-differentiation factor-15 for prediction of acute kidney injury in patients undergoing cardiac surgery

Background Growth-differentiation factor-15 (GDF-15) is an emerging humoral marker for risk stratification in cardiovascular disease. Cardiac-surgery-associated acute kidney injury (CSA-AKI), an important complication in patients undergoing cardiac surgery, is associated with poor prognosis. The present secondary analysis of an observational cohort study aimed to determine the role of GDF-15 in predicting CSA-AKI compared with the Cleveland-Clinic Acute Renal Failure (CC-ARF) score and a logistic regression model including variables associated with renal dysfunction. Methods Preoperative plasma GDF-15 was determined in 1176 consecutive patients undergoing elective cardiac surgery. Patients with chronic kidney disease stage 5 were excluded. AKI was defined according to Kidney-Disease-Improving-Global-Outcomes (KDIGO) - creatinine criteria. The following variables were screened for association with development of postoperative AKI: age, gender, additive Euroscore, serum creatinine, duration of cardiopulmonary bypass, duration of surgery, type of surgery, total circulatory arrest, preoperative hemoglobin, preoperative oxygen-supplemented cerebral oxygen saturation, diabetes mellitus, hemofiltration during ECC, plasma GDF-15, high sensitivity troponin T (hsTNT), and N-terminal prohormone of B-type natriuretic peptide (NTproBNP). Results There were 258 patients (21.9 %) with AKI (AKI stage 1 (AKI-1), n = 175 (14.9 %); AKI-2, n = 6 (0.5 %); AKI-3, n = 77 (6.5 %)). The incidence of AKI-1 and AKI-3 increased significantly from the lowest to the highest tertiles of GDF-15. In logistic regression, preoperative GDF-15, additive Euroscore, age, plasma creatinine, diabetes mellitus, and duration of cardiopulmonary bypass were independently associated with AKI. Inclusion of GDF-15 in a logistic regression model comprising these variables significantly increased the area under the curve (AUC 0.738 without and 0.750 with GDF-15 included) and the net reclassification ability to predict AKI. Comparably, in receiver operating characteristic analysis the predictive capacity of the CC-ARF score (AUC 0.628) was improved by adding GDF-15 (AUC 0.684) but this score also had lower predictability than the logistic regression model. In random forest analyses the predictive capacity of GDF-15 was especially pronounced in patients with normal plasma creatinine. Conclusion This suggests that preoperative plasma GDF-15 independently predicts postoperative AKI in patients undergoing elective cardiac surgery and is particularly helpful for risk stratification in patients with normal creatinine. Trial registration NCT01166360 on July 20, 2010. Electronic supplementary material The online version of this article (doi:10.1186/s13054-016-1482-3) contains supplementary material, which is available to authorized users.