肺恶性肿瘤静脉血栓栓塞症列线图模型的建立及验证

Establishment and verification of nomogram model of venous thrombo embolism in pulmonary malignancy

目的构建肺恶性肿瘤患者的静脉血栓栓塞症(Venous thrombo embolism,VTE)列线图模型,并对其进行验证,以准确预测肺癌人群VTE风险概率。方法基于2020年1月1日至2023年1月31日发生VTE的122例肺癌患者,按1∶4比例随机匹配同期住院的488例未发生VTE的肺癌患者。最终纳入的610例肺癌患者按7∶3随机分为428例建模集(VTE:85例,无VTE:343例),以及182例验证集(VTE:37例,无VTE:145例)。回顾性分析其一般信息、基础病史、实验室指标、肿瘤及肺部因素、治疗方案等数据。用建模集进行单因素分析及多因素Logistic回归筛选变量构建列线图模型,并在建模集与验证集中检测模型区分度、校准度及临床适用性。结果单因素分析结果显示:BMI≥24、患有糖尿病基础病史是肺癌患者发生VTE的不良危险因素;D-D、LDH、CA125、PT、ALP是VTE发生的危险指标;腺癌、Ⅲ-Ⅳ期/广泛期是肺癌患者VTE发生的不良肿瘤因素;化疗及中心静脉导管留置与较高的VTE发生率显著相关。多因素Logistic回归分析结果显示:BMI、糖尿病、D-二聚体、乳酸脱氢酶、病理类型、临床分期、化疗、中心静脉置管是肺癌VTE的独立危险因素。基于建模集多变量Logistic回归分析确定的独立危险因素构建列线图模型。建模集与验证集受试者工作特征(Receiver operating characteristic,ROC)曲线分别为0.887和0.853,证明有良好区分度。建模集与验证集的校准曲线显示预测概率和实际概率非常吻合,Hosmer-lemeshow检验P值分别为0.814和0.186。临床决策曲线显示该模型可使临床肺癌患者明显获益。结论肺恶性肿瘤静脉血栓栓塞症列线图模型对预测肺癌患者VTE风险有一定价值,可根据其制定预防性抗凝策略。

Objectives To construct and text the nomogram model of lung cancer patients combined with Venous thrombus embolism(VTE),that it could accurately predict the incidence of VTE risk in the lung cancer population.Methods The data based from 122 lung cancer patients combined with VTE,between January 2020 and January 2023,there were randomly matched in a 1∶4 ratios to 488 patients without VTE during the same period of hospitalization.610 lung cancer patients were randomly divided into modeling set of 428 patients(VTE:85,non-VTE:343)and validation set of 182 patients(VTE:37,non-VTE:145)in the ratio of 7∶3.The data of general information,basic medical history,laboratory parameters,tumour and pulmonary factors and treatment regimen were retrospectively analysed.Univariate analysis and multivariate Logistic regression were used to screen the variablesin the modeling set to constructa nomogrammodel,and the model differentiation,calibration and clinical applicability were tested in the modeling and validation set.Results The results of univariate analysis showed that BMI≥24 and having a basic medical history of diabetes were adverse risk factors for VTE in lung cancer patients.D-D,LDH,CA125,PTand ALP were risk indicators for VTE occurrence.Adenocarcinoma and stageⅢ-Ⅳ/extensive stage were adverse tumor factors for the occurrence of VTE in lung cancer patients.Chemotherapy and central venous catheter indwelling were significantly associated with a higher incidence of VTE.Multivariate logistic regression analysis showed that BMI,diabetes,D-dimer,lactate dehydrogenase,pathological type,clinical stage,chemotherapy and central vein catheterization were independent risk factors for VTE of lung cancer.Construct a column chart model based on independent risk factors determined through multivariate logistic regression analysis of the modeling set.The receiver operating characteristic(ROC)curves of the modeling and validation sets were 0.887 and 0.853,respectively,indicating good discrimination.The calibration curves of the modeling set and the validation set showed a very good agreement between the predicted probability and the actual probability,with the P-values of the Hosmer Limeshow test being 0.814 and 0.186,respectively.The clinical decision curve showed that the model can significantly benefit clinical lung cancer patients.Conclusions Pulmonary malignancy-venous thromboembolism nomogram model is valuable for predicting the incidence of VTE,and it may use to deploy prophylactic anticoagulation strategies in lung cancer patients.