Comparing diagnostic accuracy of biomarkers for acute kidney injury after major surgery: A PRISMA systematic review and network meta-analysis.

Abstract

The timely identification of patients at risk of acute kidney injury (AKI), along with early prevention, real-time monitoring, and prompt intervention, plays a crucial role in enhancing patient prognosis after major surgery. We conducted a comprehensive search across multiple databases, including Web of Science, EMBASE, MEDLINE, China National Knowledge Infrastructure, and Cochrane Library. Each study's risk of bias was independently evaluated as low, moderate, or high, utilizing criteria adapted from Quality Assessment of Diagnostic Accuracy Studies 2. The analysis was performed using STATA V.17.0 and R software V.3.4.1. Diagnostic tests were ranked based on the dominance index. We performed meta-analyses to calculate odds ratios (ORs) and 95% confidence intervals (CIs) individually. We then carried out a network meta-analysis to compare the performances of these biomarkers. Fifteen studies were included in this analysis. The meta-analysis findings revealed that among all the biomarkers assessed, serum cystatin C (s-CysC) (hierarchical summary receiver operating characteristic curve [HSROC] 82%, 95% CI 0.78-0.85) exhibited the highest HSROC value. The network meta-analysis demonstrated that urinary kidney injury molecule-1 (u-KIM-1) and s-CysC displayed relatively higher sensitivity and specificity, respectively. In subgroup analyses, u-KIM-1 in the urine output (OU) group (OR 303.75, 95% CI 3.39-1844.88), s-CysC in the non-OU group (OR 10.31, 95% CI 3.09-26.2), interleukin-18 in the noncardiac surgery group (OR 46.20, 95% CI 0.48-307.68), s-CysC in the cardiac group (OR 12.42, 95% CI 2.9-35.86), u-KIM-1 in the retrospective group (OR 243.00, 95% CI 1.73-1582.11), and s-CysC in the prospective group (OR 8.35, 95% CI 2.34-21.15) had the best diagnostic accuracy. However, it is important to note that existing publication bias may reduce the reliability of the above-mentioned results. The biomarker of s-CysC has the highest HSROC value to predicting acute kidney injury after major surgery in meta-analysis and relatively higher specificity in network meta-analyses. u-KIM-1 exhibited relatively higher sensitivity, with best diagnostic accuracy in the OU and retrospective group in the subgroup analysis.