nomogram模型在HBV相关ACLF预后危险因素中的应用及评估效能分析

Application of nomogram model in prognosis risk factors in HBV related acute-on-chronic liver failure and its assessment efficiency analysis

目的探讨乙型肝炎病毒(HBV)相关慢加急性肝衰竭(ACLF)短期预后危险因素诺模图(nomogram)模型建立的方法和其生存分析的评估效能。方法回顾性分析该院2016年1月至2020年12月收治的124例HBV相关ACLF患者临床资料,将124例患者分为建模组64例及验证组60例,再将建模组患者根据跟踪3个月的临床结果分为风险组(死亡或肝移植23例)和稳定组(41例),对比风险组及稳定组一般资料、实验室结果、并发症发生率、腹部面积比(LAAR)、终末期肝病评估模型(MELD)评分等因素,分析其独立风险因素,并将其代入nomogram模型变量,采用R软件形成回归系数,并形成nomogram模型。与MELD评分对比建模组和验证组nomogram模型生存分析评估效能差异。结果单因素分析显示,风险组患者腹腔积液、肝性脑病、感染的发生率高于稳定组(P<0.05);总胆红素(TBIL)、国际正常化比值(INR)、血钠、肌酐值与稳定组比较差异有统计学意义(P<0.05);影像学检查显示,风险组LAAR低于稳定组,而MELD评分高于稳定组,差异均有统计学意义(P<0.05)。多因素分析结果显示,感染(HR=1.054,95%CI:1.001~1.194)、MELD评分(HR=1.123,95%CI:1.075~1.158)、LAAR(HR=1.003,95%CI:0.906~1.103)为独立风险因素,以此建立nomogram模型,建模组、验证组的曲线下面积(AUC)高于MELD模型,差异有统计学意义(P<0.05)。约登指数计算nomogram模型评测最佳临界值为73.6,灵敏度为83.4%,特异度为77.1%。结论HBV相关ACLF短期预后危险因素主要为感染、LAAR及MELD评分,通过建立nomogram模型可准确地对HBV相关ACLF短期生存情况进行评估。

Objective To investigate the nomogram model establishment method for the risk factors of hepatitis B virus(HBV)related acute-on-chronic liver failure(ACLF)and its assessment efficiency on the survival analysis.Methods The clinical data in 124 cases of HBV-related ACLF were retrospectively analyzed.The patients were randomly divided into the modeling group(64 cases)and verification group(60 cases).The patients of the modeling group were divided into the risk group(death or liver transplantation,23 cases)and stable group(41 cases)according to the clinical results of 3-month-follow-up.The general data,laboratory results,occurrence rate of complications,LAAR,MELD score,etc.were compared between the risk group and stable group.Their independent risk factors were analyzed,which were substituted into the Nomogram model variables.The R software was adopted to form the regression coefficients and the nomogram model.The differences in the survival analysis evaluation efficiency of nomogram model and MELD score were compared between the modeling group and verification group.Results The univariate analysis showed that the incidence rates of complications such as ascites,hepatic encephalopathy and infection in the risk group were higher than those in the stable group.The values of TBIL,INR,blood Na and Cr had statistical differences compared with those in the stable group(P<0.05).The imaging examination showed that LAAR in the risk group was lower than that in the stable group,while the MELD score was higher than that in the stable group,and the differences were statistically significant(P<0.05).The multivariate analysis showed that infection(HR=1.054,95%CI:1.001-1.194),MELD score(HR=1.123,95%CI:1.075-1.158)and LAAR(HR=1.003,95%CI:0.906-1.103)were the independent risk factors,on this account,the nomogram model was established.The areas under the ROC curve(AUC)of the modeling group and verification group were higher than that of the MELD model,and the difference was statistically significant(P<0.05).The evaluation optimal critical value of nomogram model calculated by Youden index was 73.6,the sensitivity was 83.4%,and the specificity was 77.1%.Conclusion The short-term prognostic risk factors of HBV-related ACLF are mainly the infection,LAAR and MELD scores.The nomogram model establishment could accurately assess the short-term survival situation of HBV-related ACLF.