BACKGROUND: The predictive value of plasma and urine neutrophil gelatinase-associated lipocalin (NGAL) for use of renal replacement therapy (RRT) and acute kidney injury (AKI) is not established in patients with severe sepsis.

METHODS: This was a prospective observational study in three general intensive care units (ICUs) in adult ICU patients with severe sepsis needing fluid resuscitation and a sub-study of the 6S trial. Plasma and urine were sampled at baseline and NGAL was measured using particle-enhanced turbidimetric immunoassay (The NGAL Test). Outcome measures were use of RRT in ICU, development of AKI according to the Kidney Disease: Improving Global Outcomes plasma creatinine criteria within 48 h and 90-day mortality.

RESULTS: Two-hundred- twenty-two patients had samples taken (211 had plasma and 162 urine sampled); simplified acute physiology score II was 54 (39-66). Forty patients (18%) had RRT in the ICU, 91 patients had AKI at enrollment; of the remaining 131 patients 24% developed AKI during the first 48 h, and 55% had died at 90 days. Areas under receiver-operating characteristics curve (AuROC) for predicting use of RRT in ICU were 0.70 (95% confidence interval 0.61-0.78) and 0.62 (0.51-0.73) for plasma and urine NGAL, respectively. AuROC of plasma and urine NGAL for AKI were 0.66 (0.54-0.77) and 0.71 (0.59-0.82), respectively, and for 90-day mortality 0.55 (0.47-0.63) and 0.61 (0.53-0.70), respectively. Combining NGAL values with plasma creatinine did not improve AuROCs.

CONCLUSION: In ICU patients with severe sepsis, plasma and urine NGAL had low predictive power for use of RRT, AKI and 90-day mortality. These results were supported by sensitivity and exploratory analyses.