Accuracy of Neutrophil Gelatinase-Associated Lipocalin (NGAL) in Diagnosis and Prognosis in Acute Kidney Injury: A Systematic Review and Meta-analysis

Background Neutrophil gelatinase-associated lipocalin (NGAL) appears to be a promising biomarker for the early diagnosis of acute kidney injury (AKI); however, a wide range in its predictive value has been reported. Study Design Meta-analysis of diagnostic test studies using custom-made standardized data sheets sent to each author. Setting & Population Different clinical settings of AKI. Selection Criteria for Studies MEDLINE, EMBASE, and CENTRAL databases and congress abstracts were searched for studies reporting the value of NGAL to predict AKI. Index Tests Plasma/serum and urine NGAL within 6 hours from the time of insult (if known) or 24-48 hours before the diagnosis of AKI if the time of insult was not known. Reference Tests The primary outcome was AKI, defined as an increase in serum creatinine level > 50% from baseline within 7 days or contrast-induced nephropathy (creatinine increase > 25% or concentration > 0.5 mg/dL in adults or > 50% increase in children within 48 hours). Other outcomes predicted using NGAL were renal replacement therapy initiation and in-hospital mortality. Results Using a hierarchical bivariate generalized linear model to calculate the diagnostic odds ratio (DOR) and sample size–weighted area under the curve for the receiver-operating characteristic (AUC-ROC), we analyzed data from 19 studies and 8 countries involving 2,538 patients, of whom 487 (19.2%) developed AKI. Overall, the DOR/AUC-ROC of NGAL to predict AKI was 18.6 (95% CI, 9.0-38.1)/0.815 (95% CI, 0.732-0.892). The DOR/AUC-ROC when standardized platforms were used was 25.5 (95% CI, 8.9-72.8)/0.830 (95% CI, 0.741-0.918) with a cutoff value > 150 ng/mL for AKI compared with 16.7 (95% CI, 7.1-39.7)/0.732 (95% CI, 0.656-0.830) for “research-based” NGAL assays. In cardiac surgery patients, the DOR/AUC-ROC of NGAL was 13.1 (95% CI, 5.7-34.8)/0.775 (95% CI, 0.669-0.867); in critically ill patients, 10.0 (95% CI, 3.0-33.1)/0.728 (95% CI, 0.615-0.834); and after contrast infusion, 92.0 (95% CI, 10.7-794.1)/0.894 (95% CI, 0.826-0.950). The diagnostic accuracy of plasma/serum NGAL (17.9 [95% CI, 6.0-53.7]/0.775 [95% CI, 0.679-0.869]) was similar to that of urine NGAL (18.6 [95% CI, 7.2-48.4]/0.837 [95% CI, 0.762-0.906]). We identified age to be an effective modifier of NGAL value with better predictive ability in children (25.4 [95% CI, 8.9-72.2]/0.930 [95% CI, 0.883-0.968]) compared with adults (10.6 [95% CI, 4.8-23.4]/0.782 [95% CI, 0.689-0.872]). NGAL level was a useful prog