Modest Impact of Serial Measurements of Acute Kidney Injury Biomarkers in an Adult Intensive Care Unit

<b><i>Background/Aims:</i></b> Acute kidney injury (AKI) biomarkers have been developed with the aim of being able to detect kidney damage earlier than the detection process based on serum creatinine levels. However, single time-point measurements appear to furnish insufficient information for detecting and predicting AKI in intensive care unit patients who are frequently affected by multiple and transient/persistent renal insults. We evaluated whether serial measurements enable the prediction of AKI outcomes in such patients. <b><i>Methods:</i></b><i></i> Serial measurements of AKI biomarkers, including plasma and urinary neutrophil gelatinase-associated lipocalin, urinary L-type fatty acid-binding protein, and urinary N-acethyl-β-D-glucosaminidase, at intensive care unit (ICU) admission (d1) and 24 h later (d2) were performed for critically ill adult patients in a mixed ICU. We assessed whether each biomarker could predict newly developed AKI, recovery from AKI, worsening of AKI, and hospital mortality. <b><i>Results:</i></b><i></i> Among the enrolled 272 patients, 33 were determined to show newly developed AKI after ICU admission, 58 showed worsening of AKI, 57 recovered from AKI, and 38 died in the hospital. ROC analysis showed that biomarkers at day 2 provided no significant additional benefit in predicting the above-mentioned AKI outcomes compared with those at day 1. However, net reclassification improvement analysis demonstrated that adding day 2 biomarkers to the clinical model comprising clinical variables along with day 1 biomarkers significantly improved the prediction of these AKI outcomes. <b><i>Conclusion:</i></b><i></i> Serial measurement of AKI biomarkers used in clinical models could contribute to the prediction of AKI outcomes in a heterogeneous cohort of adult mixed ICU patients, although its reliability seemed to be modest.