基于ICG-R15、ALBI、AFP及HBV-DNA的肝切除术后肝衰竭预测模型：一项多中心回顾性研究

Prediction Model of Posthepatectomy Liver Failure Based on ICG-R15, ALBI, AFP and HBV-DNA: A Multicenter Retrospective Study

摘要: 目的：建立并验证基于吲哚菁绿15分钟滞留率(ICG-R15)、ALBI评分、甲胎蛋白(AFP)及HBV-DNA载量的肝切除术后肝功能衰竭(posthepatectomy liver failure, PHLF)发生风险预测模型。方法：回顾性纳入2017年2月至2024年12月安徽医科大学第二附属医院及皖南医科大学第一附属医院接受择期肝切除术的578例患者。以国际肝脏外科研究组(ISGLS)标准定义PHLF，并将B/C级事件作为结局。采用多因素Logistic回归筛选独立预测因子并构建列线图，通过受试者工作特征曲线下面积(AUC)、校准曲线和决策曲线分析(DCA)评价模型性能，并采用1000次Bootstrap重抽样和5折交叉验证进行内部验证。结果：共纳入578例患者，其中72例发生B/C级PHLF (12.5%)。最终纳入ICG-R15、ALBI评分、log₁₀(AFP + 1)及HBV-DNA ≥ 1000 IU/mL 4个独立预测因子。模型开发集AUC为0.844 (95% CI 0.798~0.890)，Bootstrap校正后AUC为0.839，Hosmer-Lemeshow检验P = 0.554，DCA显示在5%~35%阈值范围内具有较好净获益。结论：基于ICG-R15、ALBI评分、AFP及HBV-DNA构建的列线图模型具有良好的区分度、校准度与临床应用价值，可用于肝切除术前的个体化PHLF风险评估。

Abstract: Objective: This paper aims to develop and validate a predictive model for posthepatectomy liver failure (PHLF) based on the indocyanine green retention rate at 15 minutes (ICG-R15), ALBI score, alpha-fetoprotein (AFP), and HBV-DNA level. Methods: This retrospective study included 578 patients who underwent elective hepatectomy between February 2017 and December 2024 at the Second Affiliated Hospital of Anhui Medical University and the First Affiliated Hospital of Wannan Medical University. PHLF was defined according to the International Study Group of Liver Surgery (ISGLS), and grade B/C events were used as the endpoint. Multivariable logistic regression was used to identify independent predictors and construct a nomogram. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis (DCA), with internal validation by 1000-bootstrap resampling and 5-fold cross-validation. Results: Among the 578 enrolled patients, 72 developed grade B/C PHLF (12.5%). Four independent predictors were retained in the final model: ICG-R15, ALBI score, log₁₀(AFP + 1), and HBV-DNA ≥ 1000 IU/mL. The model yielded an AUC of 0.844 (95% CI 0.798~0.890) in the overall cohort and a bootstrap-corrected AUC of 0.839; the Hosmer-Lemeshow test was non-significant (P = 0.554). DCA demonstrated favorable net benefit across threshold probabilities of 5% to 35%. Conclusions: The nomogram based on ICG-R15, ALBI score, AFP, and HBV-DNA demonstrates good discrimination, calibration, and clinical utility, and may serve as a practical tool for individualized preoperative PHLF risk assessment.