Assessment of plasma endostatin to predict acute kidney injury in critically ill patients

Background: We evaluated whether plasma endostatin predicts acute kidney injury (AKI), need for renal replacement therapy (RRT), or death. Methods: Prospective, observational, multicenter study from 1 September 2011 to 1 February 2012 with data from 17 intensive care units (ICUs) in Finland. Results: A total of 1112 patients were analyzed. We measured plasma endostatin within 2 h of ICU admission. Early AKI (KDIGO stage within 12 h of ICU admission) was found in 20% of the cohort, and 18% developed late AKI (KDIGO criteria > 12 h from ICU admission). Median (IQR) admission endostatin was higher in the early AKI group, 29 (19.1, 41.9) ng/ml as compared to 22.4 (16.1, 30.1) ng/ml for the late AKI group, and 18 (14.0, 23.6) ng/ml for non-AKI patients (P < 0.001). Endostatin level increased with increasing KDIGO stage. Significantly higher endostatin levels were found in patients with sepsis as compared to those without. Predictive properties for AKI, RRT, and mortality were low with corresponding areas under the receiver operating characteristic curve (AUC) of 0.62, 0.67, and 0.59. Sensitivity analyses among patients with chronic kidney disease or sepsis did not improve the predictive ability of endostatin. Adding endostatin to a clinical AKI prediction model (illness severity score, urine output, and age) insignificantly changed the AUC from 0.67 to 0.70 (P = 0.14). Conclusions: Endostatin increases with AKI severity but has limited value as a predictor of AKI, RRT and 90-day mortality in patients admitted to ICU. Moreover, endostatin does not improve AKI risk prediction when added to a clinical risk model.