Plasma and urine neutrophil gelatinase-associated lipocalin in the diagnosis of new onset acute kidney injury in critically ill patients

INTRODUCTION Neutrophil gelatinase-associated lipocalin (NGAL) has been demonstrated to be a useful early diagnostic biomarker of acute kidney injury (AKI) where the timing of the insult is certain. However, NGAL is not well validated in adult critical care practice because of indeterminate timing of injury. Therefore, we sought to establish the predictive ability of both urine and plasma NGAL to detect AKI in ICU patients. METHOD This prospective observational study was performed in a busy large district general hospital mixed surgical-medical ICU in Reading, UK. Consecutive adult admissions to the ICU, with absence of chronic kidney disease, renal transplant or AKI as defined by RIFLE criteria were included. Blood and urine specimens were collected at admission and every 24 hours until 72 hours and tested for NGAL. The purpose of the study was to assess whether urinary NGAL (uNGAL) or plasma NGAL (pNGAL) can predict the occurrence of AKI at an earlier point of time than the conventional markers, that is creatinine and urine output as is used in RIFLE criteria. RESULTS Over a 12-month period, 194 patients were enrolled. In total, 59 (30.4%) patients developed AKI. The admission pNGAL and uNGAL were significantly higher in the patients who developed AKI compared to the non-AKI patients (436 ng/mL (240, 797) versus 168 ng/mL (121.3, 274.3) P <0.001 and 342 ng/mL (61.5, 1,280) versus 34.5 ng/mL (11.5, 107.75) P <0.001 respectively). Hospital mortality was higher in the AKI group (17% versus 4%). Plasma NGAL performed fairly on admission (AUROC 0.77) and thereafter performance improved at 24 and 48 hours (AUROC 0.88 and 0.87) following ICU admission. Urine NGAL had a fair predictive value on admission (AUROC 0.79) and at 24 hours (AUROC 0.78) and was good at 48 hours (AUROC 0.82). CONCLUSIONS In critically ill patients without pre-existing kidney disease, both pNGAL and uNGAL measured at admission can predict AKI (defined by RIFLE criteria) occurrence up to 72 hours post-ICU admission and their performance (AUROC) was fair. The accuracy of NGAL appeared to improve slightly as patients progressed through their ICU stay. Serial measurements of NGAL (both pNGAL and uNGAL) may be of added value in an ICU setting to predict the occurrence of AKI.