摘要
目的:通过分析长径5~15 mm的孤立性肺结节(SPN)患者的临床、实验室检查及影像学征象等资料,建立并验证长径5~15 mm的SPN良恶性预测模型。方法:回顾性搜集2012年12月-2018年3月在云南省肿瘤医院经手术切除并获取明确病理诊断的长径5~15 mm的SPN病例共958例,分析患者的临床特征(年龄、性别、体质指数、吸烟史、呼吸系统疾病史、恶性肿瘤病史、家族肿瘤史),实验室检查(癌胚抗原、糖类抗原CA125、糖类抗原CA724、中性粒细胞淋巴细胞比值(NLR))和影像学征象(结节位置、最大径和最小径、边缘、磨玻璃成分、空洞、空泡征、空气支气管征、血管集束征、胸膜牵拉征、钙化、肺气肿背景、纤维条索影)。采用随机分组将患者按7:3比例分为训练集(670例)和验证集(288例)。通过单因素及多因素logistics回归分析筛选出与长径5~15 mm的SPN良恶性相关的独立危险因素,构建列线图预测模型。将验证集的数据代入该模型进行验证,绘制受试者操作特征(ROC)曲线和校准曲线,评估模型预测价值。结果:训练集670例长径5~15 mm的SPN患者中,良性313例(46.72%),恶性357例(53.28%)。多因素logistic回归分析筛选得出年龄、空泡征、血管纠集征、空气支气管征、胸膜牵拉、含GGO成分以及NLR等7个因素是诊断长径5~15 mmSPN良恶性的独立预测因子(P<0.05),构建的预测模型为:P(长径5~15mm的SPN恶性概率预测值)=ex/(1+ex),X=-3.14282+(0.03737×年龄)+(0.51374×空泡)+(1.87897×血管纠集征)+(0.90573×空气支气管征)+(1.15055×胸膜牵拉)+(1.57413×含GGO成分)-(0.84182×NLR)。结果显示,该模型对长径5~15 mm的SPN患者良恶性判断的ROC曲线下面积为0.860(95%CI:0.832,0.888),敏感度为90.8%,特异度68.4%,符合率为80.3%。利用验证集数据验证该模型,得到ROC曲线下面积为0.864(95%CI:0.820~0.907),敏感度为77.2%,特异度为86.3%,符合率为81.6%,两者间差异无统计学意义(P>0.05)。结论:患者的年龄、NLR、空泡征、血管纠集征、空气支气管征、胸膜牵拉、含GGO成分是预测长径5~15 mm的SPN良恶性的独立预测因子。本研究构建的模型具有良好的诊断效力,对临床医生判断长径5~15mm的SPN的良恶性有一定帮助。
Objective:To establish and validate the benign and malignant predicting model in patients with 5~15 mm solitary pulmonary nodules(SPN)by analyzing the clinical,laboratory and radiologic features.Methods:A total of 958 patients with 5~15 mm SPN who underwent surgical resection with definite postoperative pathological diagnosis from December 2012 to March 2018 in Yunnan Cancer Hospital were retrospectively collected.The clinical features,preoperative laboratory examination data and radiologic characteristics were analyzed,including age,gender,height,weight,smoking history,history of respiratory disease,history of malignancy,family tumor history,carcinoembryonic antigen,carbohydrate antigen 125,carbohydrate antigen 724,and neutrophil to lymphocyte ratio(NLR),location,maximum and minimum diameter in lung window,edge,ground-glass opacity(GGO)composition,cavity,vacuole,air bronchogram,vascular convergence,pleura indentation,calcification,emphysema background,and fiber strip.All cases were randomly divided into development set(670 cases)and validation set(288 cases)in a 7:3 ratio.A new prediction model was established by univariate and multivariate logistic regression analysis.Then,the validation set data were used to validate this new model,draw the receiver operating characteristic(ROC)curve and calibration curve,to assess the predictive value of the model.Results:357 malignant cases and 313 benign cases were included in development set.Logistic Regression analysis showed that age,vacuole,vascular convergence,air bronchogram,pleura indentation,GGO composition and NLR were independent predictors of malignancy in patients with 5~15 mm SPN(P<0.05).We calculated a prediction model for the probability of malignancy as follow:P=ex/(1+ex),X=-3.14282+(0.03737×age)+(0.51374×vacuole)+(1.87897×air bronchogram)+(1.15055×pleura indentation)+(1.57413×GGO component)-(0.84182×NLR).The area under the ROC curve was 0.860(95%CI:0.832,0.888)for development set,sensitivity 90.8%,specificity 68.4%,accuracy 80.3%.The area under the ROC curve was 0.864(95%CI:0.820,0.907)for validation set,sensitivity 77.2%,specificity 86.3%,accuracy 81.6%.There was no significant difference between the two groups(P>0.05).Conclusion:Age,NLR,vacuole,vascular convergence,air bronchogram,pleura indentation,GGO composition were independent predictors of malignancy in patients with 5~15 mm SPN.This prediction model showed good diagnosis efficiency in external validation and could be applied to make decision for patients with 5~15 mm SPN.
作者
晏睿滢
李华秀
丁莹莹
李振辉
陈小波
高德培
Yan Rui-ying;Li Hua-xiu;Ding Ying-ying(Department of Radiology,Yunnan Cancer Hospital,the Third Affiliated Hospital of Kunming Medical University,Kunming 650118,China)
出处
《放射学实践》
北大核心
2020年第3期352-359,共8页
Radiologic Practice
基金
云南省科技厅-昆明医科大学应用基础研究联合专项资金面上项目[2017FE467(-084)].
作者简介
晏睿滢(1993-),女,云南宣威人,硕士,住院医师,主要从事胸部肿瘤影像诊断工作;通讯作者:高德培,E-mail:gaodepei311@sohu.com
