基于无创指标对肝移植术后急性肾功能损伤 预测模型的建立与验证

Development and Validation of a Prediction Model for Acute Kidney Injury after Liver Transplantation Based on Noninvasive Indicators

摘要: 目的：探究无创指标对肝移植术后急性肾功能损伤(AKI)的预测价值，建立预测模型并验证。方法：本研究为回顾性研究，回顾性分析2022年1月至2025年6月于青岛大学附属医院器官移植中心行同种异体肝移植供受者临床资料，2022年1月至2024年12月供受者作为训练集，2025年1月至2025年6月供受者作为验证集。根据肝移植术后7天内是否发生AKI进行分组，获取临床资料并通过公式计算相关无创指标评分。采用独立样本t检验、Mann-Whitney U检验、χ²检验或Fisher精确检验进行两组间差异比较。通过Logistic回归筛选危险因素并建立预测模型，绘制列线图、校准曲线、临床决策曲线、临床影响曲线对模型进行性能评价，并对模型进行验证。结果：多因素分析结果显示，Agile3+评分(OR = 3.37, p < 0.01)、FIB-4评分(OR = 4.12, p < 0.01)、受者年龄(OR = 1.07, p < 0.01)、受者白蛋白(OR = 0.94, p = 0.02)均为肝移植术后急性肾功能损伤的独立影响因素。用以上危险因素构建肝移植术后AKI的预测模型，受试者工作特征(ROC)曲线的曲线下面积为0.74，时间外部验证ROC曲线下面积为0.64，临床决策曲线及临床影响曲线提示模型有一定预测价值。结论：基于无创指标构建肝移植术后AKI的预测模型具有一定风险预测价值，可对临床决策提供参考，用于临床实践，需要进一步优化及验证。

Abstract: Objective: To evaluate the predictive value of non-invasive indices for acute kidney injury (AKI) after liver transplantation and to develop and validate a prediction model. Methods: This was a retrospective study that reviewed the clinical data of donors and recipients who underwent allogeneic liver transplantation at the Organ Transplantation Center of the Affiliated Hospital of Qingdao University between January 2022 and June 2025. Donors and recipients from January 2022 to December 2024 were included as the training cohort, while those from January 2025 to June 2025 constituted the validation cohort. Patients were divided into groups according to the occurrence of AKI within 7 days after liver transplantation. Clinical variables were collected, and non-invasive index scores were calculated using standard formulas. Between-group differences were assessed using the independent-samples t test, Mann-Whitney U test, chi-square test, or Fisher’s exact test, as appropriate. Logistic regression analysis was performed to identify risk factors and construct a prediction model. Model performance was evaluated using a nomogram, calibration curve, decision curve analysis, and clinical impact curve, and the model was subsequently validated. Results: Multivariate analysis revealed that the Agile3+ score (OR = 3.37, p < 0.01), FIB-4 (OR = 4.12, p < 0.01), recipient age (OR = 1.07, p < 0.01), and recipient serum albumin level (OR = 0.94, p = 0.02), were independent predictors of postoperative AKI. The prediction model constructed using these risk factors demonstrated an area under the ROC curve of 0.74, and the time-based external validation yielded an AUC of 0.64. Decision curve analysis and the clinical impact curve suggested that the model has potential clinical predictive value. Conclusion: A prediction model for AKI after liver transplantation based on non-invasive indicators provides meaningful risk stratification and may assist clinical decision-making; however, further optimization and validation are required before clinical application.