肝移植术后急性肾损伤的危险因素和预后探析及列线图预测模型的构建

Risk factors and prognosis of acute kidney injury after liver transplantation and constructing a nomogram prediction model

目的探究肝移植术后急性肾损伤(acute kidney injury, AKI)的危险因素及对受者预后的影响, 构建临床预测模型。方法回顾性分析2018年1月至2020年8月在北京协和医学院南京鼓楼医院行肝移植的220例受者的临床资料, 根据术后是否发生AKI分为AKI组(93例)和无AKI组(127例)。分析比较两组受者的临床资料, 将两组间单因素分析差异具有统计学意义的变量纳入多因素分析, 得出影响肝移植术后急性肾损伤发生的独立危险因素;将得出的独立危险因素作为独立预测因子拟合预测模型, 并构建可视化的列线图, 同时对预测模型的区分度和校准度进行评价;比较两组间气管插管拔除时间、ICU停留时间、连续性肾脏替代治疗(continuous renal replacement therapy, CRRT)率、住院时间、在院死亡率、出院时估算的肾小球滤过率(eGFR)、慢性肾衰竭(chronic renal failure, CRF)发生率、再入院次数等术后指标, 同时进行AKI组和无AKI组、AKI 0/1级和AKI 2/3级间的生存分析。结果肝移植术后AKI的发病率为42.3%;受者年龄(OR=1.036, 95%CI:1.001~1.073)、术前血肌酐浓度(OR=1.030, 95%CI:1.011~1.049)、血小板计数(OR=0.992, 95%CI:0.985~0.999)、Child-Pugh C级(OR=2.678, 95%CI:1.031~6.952)、术后腹腔感染(OR=2.271, 95%CI:1.120~4.603)和腹腔出血(OR=3.869, 95%CI:1.016~14.720)是肝移植术后发生AKI的独立危险因素;列线图预测模型的曲线下面积(AUC)/C-index为0.789, Brier评分为0.183, 表现出良好的区分度和校准度, 根据列线图评分结果将AKI风险大于50%的受者纳入高风险组, 小于50%的受者纳入低风险组, 低风险组的术后生存优于高风险组(P<0.001);AKI组比无AKI组气管插管拔除时间(P=0.003)更晚, ICU停留(P<0.001)和住院时间(P=0.001)更长, CRRT使用率(P<0.001)和在院死亡率(P<0.001)更高, 出院时eGFR(P<0.001)更低以及CRF的发生率(P=0.004)更高, 无AKI组的术后生存优于AKI组(P=0.048), AKI 0/1级的术后生存优于AKI 2/3级(P=0.002)。结论受者高龄、术前血肌酐高、血小板低、肝功能差, 术后并发腹腔感染和腹腔出血会增加肝移植术后AKI发生的风险, 基于以上危险因素构建的列线图预测模型具有良好的临床应用价值。

Objective To explore the risk factors of acute kidney injury(AKI)after liver transplantation(LT),examine its prognostic impact and construct a clinical prediction model.Methods Clinical data are retrospectively reviewed for 220 LT recipients.They are divided into two groups of AKI(93 cases)and non-AKI(127 cases)according to the occurrence of AKI post-LT.Clinical data of two groups are compared.The variables with statistically significant inter-group differences in univariate analysis are included for multivariate analysis for obtaining the independent risk factors for AKI post-LT.Then the independent risk factors are employed for fitting a prediction model and a visual nomogram is constructed.At the same time,discrimination and calibration of the prediction model are evaluated.Extubation time,length of intensive care unit(ICU)stay,continuous renal replacement therapy(CRRT)rate,length of hospital stay,in-hospital mortality,estimated glomerular filtration rate(eGFR)at discharge,incidence of chronic renal failure(CRF)and readmission times are compared between two groups.Survival analysis is also performed between AKI and non-AKI groups and AKI 0/1 and AKI 2/3 stages.Results The incidence of AKI post-LT is 42.3%.Age(OR=1.036,95%CI:1.001~1.073),preoperative serum creatinine level(OR=1.030,95%CI:1.011~1.049),platelet count(OR=0.992,95%CI:0.985~0.999),Child-Pugh class C(OR=2.678,95%CI:1.031~6.952),postoperative abdominal infection(OR=2.271,95%CI:1.120~4.603)and abdominal hemorrhage(OR=3.869,95%CI:1.016~14.72)are independent risk factors for AKI post-LT.The AUC/C-index of nomogram prediction model is 0.789 with a Brier score of 0.183,showing decent discrimination and calibration.According to the nomogram score,the recipients with a risk of AKI>50%are included into high-risk group while those with a risk of AKI<50%into low-risk group.Postoperative survival of low-risk group is better than that of high-risk group(P<0.001).Compared with non-AKI group,AKI group had a later extubation time(P=0.003),a longer length of ICU stay(P<0.001)and hospital stay(P=0.001),a higher rate of CRRT usage(P<0.001)and in-hospital mortality(P<0.001),a lower eGFR at discharge(P<0.001)and a higher incidence of CRF(P<0.001).Postoperative survival of non-AKI group was better than that of AKI group(P=0.048).Postoperative survival of patients with AKI 0/1 is better than that of those with AKI 2/3(P=0.002).Conclusions Advanced age,high preoperative serum creatinine,low preoperative platelet,poor preoperative liver function,postoperative abdominal infection and abdominal hemorrhage may elevate the risks of AKI post-LT.And the nomogram prediction model based upon the above risk factors has a high value of clinical application.