Early prediction of acute kidney injury after liver transplantation by scoring system and decision tree

BACKGROUND AND AIMS: Early detection of acute kidney injury (AKI) is crucial for the prognosis of patients after liver transplantation (LT). This passage aims to analyze the perioperative clinical markers of AKI after LT and establish predictive models based on clinical variables for early detection of AKI after LT. METHODS: We prospectively collected 109 patients with LT, and compared the differences of perioperative clinical markers between the AKI group and non-AKI group. The scoring system and decision tree model were established through the risk factors. Another 163 patients who underwent LT in the same center from 2017 to 2018 were retrospectively collected to verify the models. RESULTS: In multiple comparisons of risk factors of post-LT AKI, pre-operative factors were excluded automatically, intraoperative and post-operative factors including operating time, intraoperative hypotension time, post-operative infection, the peak of post-operative AST, and post-operative shock were the independent risk factors for post-LT AKI. The scoring system established with the risk factors has good predictive power (AUC = 0.755) in the validation cohort. The decision tree also shows that post-operative shock was the most important marker, followed by post-operative infection. CONCLUSION: Five intraoperative and post-operative factors are independently associated with post-LT AKI rather than pre-operative factors, which indicates that operation technique and post-operative management may more important for the prevention of post-LT AKI. The scoring system and decision tree model could complement each other, and provide quantitative and intuitive prediction tools for clinical practice of early detection of post-LT AKI.