基于NLR建立列线图模型预测肝硬化门静脉血栓形成风险

Establishment of Nomograph Model Based on NLR to Predict the Risk of Portal Vein Thrombosis in Liver Cirrhosis

目的分析肝硬化门静脉血栓(PVT)形成的相关危险因素,构建相关列线图模型预测肝硬化PVT形成的风险。方法回顾性分析2014年1月-2021年10月在我院住院诊治的1000例肝硬化患者的临床资料,将2020年12月前收治的设定为建模集(n=810),之后收治的设定为验证集(n=190)。在建模集中以是否有门静脉血栓形成分为PVT组(259例)和N-PVT组(551例);运用多因素Logistic回归分析筛出肝硬化PVT形成的独立危险因素,重点分析中性粒细胞和淋巴细胞比值(NLR)与肝硬化PVT形成的相关性,基于NLR构建用于预测肝硬化PVT形成风险的列线图模型,并对该模型进行内外部验证。结果建模集中PVT组的脾切除史占比、PLT、MPV、NLR、MLR、PLR、门静脉主干内径以及脾静脉直径均高于N-PVT组(P<0.05);两组年龄、性别、Child-Pugh分级、白蛋白及糖尿病病史比较,差异无统计学意义(P>0.05)。多因素Logistic回归分析显示,脾切除史、NLR、门静脉主干内径以及脾静脉直径为肝硬化患者PVT形成的独立预测因素(P<0.05),其中NLR每增加1个单位,肝硬化患者PVT形成风险增加0.527倍(OR=1.527,95%CI:1.373~1.698,P=0.000);以脾切除史、NLR、门静脉主干内径以及脾静脉直径构建预测肝硬化PVT形成的列线图模型;列线图模型校准曲线显示,建模集和验证集肝硬化患者PVT形成的预测值与实际观测值符合度良好(P>0.05),ROC曲线下面积(AUC)为0.769(95%CI:0.732~0.805),验证集联合预测的AUC为0.789(95%CI:0.712~0.866)。决策曲线分析结果显示,在大多数合理阈值概率范围内,建模集和验证集中脾切除史、NLR、门静脉主干内径以及脾静脉直径4个指标预测肝硬化PVT形成均具有良好的净收益率,并且基于NLR联合其余3项指标预测的总体净收益率高于单一指标。结论基于NLR构建的列线图模型可用于准确预测肝硬化PVT形成风险。

Objective To analyze the risk factors of portal vein thrombosis(PVT)in liver cirrhosis,and to construct a relevant nomogram model to predict the risk of PVT in liver cirrhosis.Methods The clinical data of 1000 patients with liver cirrhosis who were hospitalized in our hospital from January 2014 to October 2021 were retrospectively analyzed.The patients admitted before December 2020 were set as the modeling set(n=810),and the patients admitted after December 2020 were set as the validation set(n=190).In the modeling set,the patients were divided into PVT group(259 patients)and N-PVT group(551 patients)according to whether there was portal vein thrombosis.Multivariate logistic regression analysis was used to screen out the independent risk factors of PVT formation in liver cirrhosis,focusing on the correlation between neutrophil-to-lymphocyte ratio(NLR)and PVT formation in liver cirrhosis.Based on NLR,a nomogram model was constructed to predict the risk of PVT formation in liver cirrhosis,and the model was verified internally and externally.Results The proportion of splenectomy history,PLT,MPV,NLR,MLR,PLR,portal vein diameter and splenic vein diameter in PVT group were higher than those in N-PVT group(P<0.05).There was no significant difference in age,gender,Child-Pugh classification,albumin and history of diabetes between the two groups(P>0.05).Multivariate Logistic regression analysis showed that splenectomy history,NLR,main portal vein diameter and splenic vein diameter were independent predictors of PVT formation in patients with cirrhosis(P<0.05).For each unit increase in NLR,the risk of PVT formation in patients with cirrhosis increased by 0.527 times(OR=1.527,95%CI:1.373-1.698,P=0.000).A nomogram model for predicting PVT formation in liver cirrhosis was constructed based on the history of splenectomy,NLR,portal vein diameter and splenic vein diameter.The calibration curve of the nomogram model showed that the predicted values of PVT formation in patients with cirrhosis in the modeling set and the validation set were in good agreement with the actual observed values(P>0.05).The area under the ROC curve(AUC)was 0.769(95%CI:0.732-0.805),and the AUC of the combined prediction of the validation set was 0.789(95%CI:0.712-0.866).The results of decision curve analysis showed that within most reasonable threshold probability ranges,the four indicators of splenectomy history,NLR,main portal vein diameter and splenic vein diameter in the modeling set and validation set had good net rate of return for predicting PVT formation in cirrhosis,and the overall net rate of return based on NLR combined with the other three indicators was higher than that based on a single indicator.Conclusion The nomogram model based on NLR can be used to accurately predict the risk of PVT in liver cirrhosis.