Risk Prediction of Acute Kidney Injury in Patients with HBV-Related Acute-on-Chronic Liver Failure

Background: Acute kidney injury (AKI) is a common and severe complication in patients with acute-on-chronic liver failure (ACLF), significantly increasing morbidity and mortality. Although several biomarkers have been proposed for the early detection of AKI, their routine clinical application remains limited by accessibility, cost, and inconsistent performance. Therefore, early identification of patients at high risk of AKI remains a major clinical challenge. Methods: This retrospective single-center study included patients with HBV-related ACLF. Clinical data were analyzed to identify independent risk factors for AKI, and a prediction model was developed. Model performance was assessed using ROC, calibration, and decision curve analyses, and patients were stratified according to risk. Results: Among 162 patients with HBV-related ACLF, 32 (19.8%) developed AKI. Patients who developed AKI had significantly higher baseline levels of blood urea nitrogen (BUN) and C-reactive protein (CRP). Multivariate analysis identified BUN (OR = 1.076, P = 0.022) and CRP (OR = 1.031, P = 0.010) as independent predictors of AKI. A prediction model incorporating these variables demonstrated acceptable discrimination (AUROC = 0.77, P < 0.001), good calibration, and favorable clinical utility on decision curve analysis. Restricted cubic spline analysis further demonstrated a nonlinear association between BUN and AKI risk, whereas CRP showed a linear relationship. Conclusion: We developed a simple BUN–CRP–based model for early prediction of AKI in patients with HBV-related ACLF. The model demonstrated acceptable discrimination and calibration and may provide a practical tool for early risk stratification by integrating markers of renal dysfunction and systemic inflammation. However, external validation in larger multicenter cohorts is warranted before routine clinical application. Keywords: acute kidney injury, acute-on-chronic liver failure, hepatitis b virus, blood urea nitrogen, c-reactive protein, prediction model