基于炎症因子中性粒/淋巴细胞比值与红细胞分布宽度的预测模型对乙肝相关慢加急性肝衰竭的预后评估

Prognostic evaluation of a prediction model based on neutrophil/lymphocyte ratio and red blood cell distribution width of inflammatory factors on hepatitis B virus-related acute-on-chronic liver failure

摘要:

目的： 拟基于炎症因子中性粒/淋巴细胞比值（neutrophil/lymphocyte ratio，NLR）及红细胞分布宽度（red blood cell distribution，RDW）建立乙肝相关慢加急性肝衰竭（HBV-related acute-on-chronic liver failure，HBV-ACLF）预后的新型预测模型。方法： 纳入577例患者的一般临床资料及实验室数据，采用终末期肝病模型（model for end-stage liver disease，MELD）评分为9~40分的554例患者为建模队列。使用单因素及多因素COX回归分析与HBV-ACLF患者预后相关的独立危险因素建立预后评估模型，再分别在3家医院使用ROC曲线验证新模型对HBV-ACLF患者90 d预后预测的准确性。使用SPSS 22.0软件分析处理数据。结果： 多因素分析显示RDW、NLR、国际化标准比值（international normalized ratio，INR）、肌酐（creatinine，Cr）、总胆红素（total bilirubin，TBIL）为HBV-ACLF患者90 d预后的独立危险因素（P<0.05）。根据COX回归分析建立如下预测模型：COX_RNTIC=0.073×RDW+0.027×NLR+0.004×TBIL+0.236×INR+0.005×Cr（P=0.000），该模型截断值为3.59（特异度为84.86%，灵敏度为78.48%）。使用ROC曲线检测预测能力，分别为RNTIC（0.864，95%CI=0.833~0.892），高于MELD评分（0.737，95%CI=0.698~0.773）、NLR（0.705，95%CI=0.665~0.743）及RDW（0.677，95%CI=0.637~0.716）（P=0.000）。在验证队列中，RNTIC模型在3家中心同样展示出优于MELD评分及NLR、RDW的死亡预测能力。结论： 基于炎症因子NLR及RDW建立的HBV-ACLF短期预后预测模型与MELD评分相比具有更好的预测价值，是值得信赖的临床预测模型。

Abstract:

Objective: To establish a new prediction model combing the inflammatory markers including neutrophil/lymphocyte ratio (NLR) and red blood cell distribution width (RDW) with several hematological testing indicators to assess the prognosis of patients with hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF). Methods: Clinical data and laboratory testing indicators of 577 patients from three hospitals were collected in this study. The model for end-stage liver disease (MELD) score was used to establish the new model cohort of 554 patients with MELD score between 9 points and 40 points. Univariate and multivariate COX regression analysis were used to identify the independent risk factor associated with the prognosis of patients with HBV-ACLF, so as to establish the prognostic assessment model. And ROC curves were applied to validate the new model in predicting the 90-day prognosis in patients with HBV-ACLF in three hospitals, respectively. SPSS 22.0 software was employed for data analyses. Results: Multivate COX regression analysis showed that RDW, NLR, international normalized ratio (INR), and creatinine (Cr) and total bilirubin (TBIL) were independent factors of 90-day prognosis in patients with HBV-ACLF (P<0.05). The prediction model was established according to the multivariate Cox regression analysis, COX_RNTIC=0.073×RDW+0.027×NLR+0.004×TBIL+0.236×INR+0.005×Cr (P=0.000), with a cut-off value of 3.59 (sensitivity: 78.48%, specificity: 84.86%). ROC curve was used to detect the predictive ability and the results showed that RNTIC (0.864, 95%CI=0.837-0.903) was better than MELD score (0.737, 95%CI=0.698-0.773), NLR (0.705, 95%CI=0.665-0.743) and RDW (0.677, 95%CI=0.637-0.716) (P=0.000). In the validation cohort, RNTIC model demonstrated a better predictive value of death than RDW, NLR, and MELD score in three hospitals. Conclusion: The short-term prognostic prediction model of HBV-ACLF which is established on the basis of inflammatory markers of RDW and NLR has a better predictive value when compared with MELD score, and is a reliable clinical predictive model.