Urine interleukin-18 in prediction of acute kidney injury: a systemic review and meta-analysis.

BACKGROUND: Interleukin-18 (IL-18) mediates ischemic acute tubular necrosis; it has been proved as a rapid, reliable, and affordable test marker for the early detection of acute kidney injury (AKI), but its predictive accuracy varies greatly.

METHODS: MEDLINE and EMBASE, Cochrane Library, Ovid, and Springerlink (from inception to November 15, 2013) were searched for relevant studies (in English) investigating diagnostic accuracy of urine IL-18 to predict AKI in various clinical settings. The text index was increasing or increased urine IL-18 level and the main outcome was the development of AKI, which was primarily based on serum creatinine level [using risk, injury, failure, loss and end-stage renal disease (RIFLE), acute kidney injury network, or modified pediatric RIFLE criteria in pediatric patients]. Pooled estimates of diagnostic odds ratio (OR), sensitivity and specificity were calculated. Summary receiver operating characteristic curves were used to calculate the measures of accuracy and Q point value (Q*). Remarkable heterogeneity was explored further by subgroup analysis based on the different clinical settings.

RESULTS: We analyzed data from 11 studies of 3 countries covering 2,796 patients. These studies were marked by limitations of threshold and non-threshold effect heterogeneity. Across all settings, the diagnostic OR for urine IL-18 level to predict AKI was 5.11 [95% confidence interval (CI) 3.22-8.12], with sensitivity and specificity respectively at 0.51 and 0.79. The area under the ROC curve of urine IL-18 level to predict AKI was 0.77 (95% CI 0.71-0.83). Subgroup analysis showed that urine IL-18 level in pediatric patients (<18 years) and early AKI predictive time (<12 h) were more effective in predicting AKI, with diagnostic ORs of 7.51 (2.99-18.88), 8.18 (2.19-30.51), respectively.

CONCLUSION: Urine IL-18 holds promise as a biomarker in the prediction of AKI but has only moderate diagnostic value.