肺癌并发肺栓塞患者预测模型的构建和内部验证

Construction and internal verification of prediction model for patients with lung cancer complicated with pulmonary embolism

目的探讨肺癌并发肺栓塞(PE)患者的影响因素及临床预测模型的构建。方法选择新疆医科大学附属肿瘤医院2016年1月至2021年12月150例肺癌并发PE患者为PE组,同时间匹配600例肺癌患者为对照组。收集两组一般资料及临床资料,分析肺癌并发PE患者的影响因素,构建列线图模型,采用受试者操作特征曲线、校准曲线和决策曲线分析对模型进行评价。结果两组病理类型、分期、合并肺炎、合并胸腔积液、糖尿病史、冠状动脉疾病史、高脂血症病史、中心静脉置管史、手术史、血型、呼吸困难、吸烟、化疗史比较,差异有统计学意义(P<0.05)。PE组白细胞计数、血小板计数(PLT)、中性粒细胞计数、中性粒细胞/淋巴细胞比值、血小板/淋巴细胞比值(PLR)、D-二聚体(D-D)、凝血酶原时间(PT)、纤维蛋白原(FIB)、乳酸脱氢酶(LDH)、甘油三酯、总胆固醇、癌胚抗原(CEA)、糖类抗原125(CA125)、鳞状细胞癌抗原、pH值高于对照组,而白蛋白(Alb)、二氧化碳分压(PCO_(2))、氧分压低于对照组,差异有统计学意义(P<0.05)。单因素logistic回归分析结果显示,病理类型、合并肺炎、冠状动脉疾病史、高脂血症病史、手术史、pH值、呼吸困难、PLT、PLR、D-D、PT、FIB、B型利钠肽原、Alb、LDH、CEA、CA125、化疗史、PCO_(2)对肺癌并发PE患者有影响(P<0.05)。LASSO回归分析结果显示,PLT、PLR、D-D、PT、CA125、PCO_(2)、呼吸困难、合并肺炎、化疗史为筛选的特征变量。多因素logistic回归分析结果显示,合并肺炎、PLT、D-D、化疗史是肺癌患者并发PE的独立危险因素(OR=5.065、1.005、1.343、16.240,P<0.05)。列线图模型预测肺癌并发PE患者的曲线下面积为0.918,灵敏度为0.861,特异度为0.840。校准曲线结果显示,模型校准度好、有较好的预测能力。决策曲线分析结果显示,当阈值概率<76%时,净收益>0,模型进行风险评估有临床意义。结论构建的列线图模型可以较好地预测肺癌并发PE患者的发生风险,有助于患者的个体化治疗,为及时采取有效治疗措施提供依据。

Objective To investigate the influencing factors of patients with lung cancer complicated with pulmonary embolism(PE)and the construction of clinical prediction model.Methods A total of 150 patients with lung cancer complicated with PE from the Affiliated Tumor Hospital of Xinjiang Medical University from January 2016 to December 2021 was selected as PE group,and 600 patients with lung cancer were matched as control group at the same time.General data and clinical data of two groups were collected,the influencing factors of patients with lung cancer complicated with PE were analyzed,a nomogram model was constructed,and the model was evaluated by receiver operating characteristic curve,calibration curve,and decision curve analysis.Results There were statistically significant differences in pathological type,stage,combined pneumonia,combined pleural effusion,history of diabetes,history of coronary artery disease,history of hyperlipidemia,history of central venous catheter,history of operation,blood type,dyspnea,smoking,and history of chemotherapy between two groups(P<0.05).White blood cell count,blood platelet count(PLT),neutrophil count,neutrophil/lymphocyte ratio,platelet/lymphocyte ratio(PLR),D-dimer(D-D),prothrombin time(PT),fibrinogen(FIB),lactic dehydrogenase(LDH),triglyceride,total cholesterol,carcino-embryonic antigen(CEA),carbohydrate antigen 125(CA125),squamous cell carcinoma antigen,and pH value of PE group were higher than those of control group,while albumin(Alb),partial pressure of carbon dioxide(PCO_(2)),and partial pressure of oxygen were lower than those of control group,and the differences were statistically significant(P<0.05).The results of univariate logistic regression analysis showed that pathological type,combined pneumonia,history of coronary artery disease,history of hyperlipidemia,history of operation,pH value,dyspnea,PLT,PLR,D-D,PT,FIB,type B natriuretic peptide,Alb,LDH,CEA,CA125,history of chemo-therapy,and PCO_(2) had an impact on patients with lung cancer complicated with PE(P<0.05).LASSO regression analysis showed that PLT,PLR,D-D,PT,CA125,PCO_(2),dyspnea,combined pneumonia,history of central venous catheterization,smoking,and history of chemotherapy were the characteristic variables of screening.Multivariate logistic regression analysis showed that pneumonia,PLT,D-D,and history of chemotherapy were independent risk factors for lung cancer patients complicated with PE(OR=5.065,1.005,1.343,16.240,P<0.05).The area under the curve of patients with lung cancer complicated with PE predicted by nomogram model was 0.918,the sensitivity was 0.861,and the specificity was 0.840.The calibration curve results showed that the model had good calibration degree and good prediction ability.The results of decision curve analysis showed that when the threshold probability was<76%,the net benefit was>0,and the risk assessment model had clinical significance.Conclusion The constructed nomogram model can better predict the risk of patients with lung cancer complicated with PE,which is helpful for individualized treatment of patients and provides a basis for timely and effective treatment measures.