Background: Acute kidney injury (AKI) in cirrhotic patients is associated with high morbidity and mortality. Serum creatinine (sCr) has limited utility for early detection and etiological differentiation. Novel biomarkers and predictive models offer potential to address these limitations. Objectives: To comprehensively evaluate the diagnostic performance of novel biomarkers and predictive models for the early detection and etiological differentiation of AKI in patients with liver cirrhosis. Design: Systematic review conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Data sources and methods: A comprehensive literature search was performed in PubMed / Medline , EMBASE , and the Cochrane Library from inception to August 31, 2025. Studies evaluating biomarkers or models for AKI prediction or phenotyping in adult cirrhotic patients were included. Study selection, data extraction, and quality assessment (using Quality Assessment of Diagnostic Accuracy Studies-2, QUADAS-2) were performed independently by reviewers. Results: A total of 33 studies were included. For the early prediction of AKI, serum cystatin C (Cys C) demonstrated superior performance to sCr, with an area under the receiver operating characteristic curve (AUROC) of up to 0.85. Urinary neutrophil gelatinase-associated lipocalin (NGAL) exhibited strong predictive capability and was most reliable in differentiating acute tubular necrosis from hepatorenal syndrome (HRS), achieving an AUROC up to 0.87. In contrast, kidney injury molecule-1, interleukin-18, and liver-type fatty acid binding protein, showed only moderate or inconsistent performance across most studies. Seven studies developed predictive models by integrating clinical variables with biomarkers, some of which employed machine learning techniques. However, the clinical applicability of these models is currently constrained by significant heterogeneity and limited external validation. Conclusion: Serum Cys C and urinary NGAL offer distinct advantages over sCr in the early diagnosis and phenotypic differentiation of AKI. Although prediction models show promise, their routine clinical application requires further standardization and extensive external validation. Trial registration: PROSPERO (CRD420251126410).
Chen et al. (Thu,) studied this question.