直径不大于1厘米肺磨玻璃结节中浸润性腺癌列线图模型的构建

Construction of a Nomogram Model for Invasive Adenocarcinoma in Pulmonary Ground Glass Nodules with Diameter Less Than 1 cm

目的使用临床和CT资料,构建一个可鉴别直径≤1 cm的肺磨玻璃结节(ground-glass nodules,GGNs)中浸润性肺腺癌(invasive pulmonary adenocarcinoma,IPA)和浸润前病变(preinvasive lesion,PIL)的列线图模型并加以验证。方法回顾性分析肺部患有磨玻璃结节的患者的临床资料和CT影像特征,通过二元Logistic回归分析筛选出独立危险因素,使用独立危险因素建立IPA风险预测列线图模型,通过使用校准曲线和受试者工作特性曲线下面积(area under the receiver operating characteristic curve,AUC)来验证和评估模型预测性能。结果182例患者的数据被用来开发模型,男性53例(29.1%),女性131例(70.9%),年龄26~77岁,平均(52.1±11.1)岁,其中51例为PIL,131例为IPA。单因素及多因素分析确定了结节大小、毛刺征、空泡征、血管征、分叶征、结节边界不清和CT值为独立危险因素,以此建立了列线图预测模型。该预测模型的计算公式为:IPA=e^(x)/(1+e^(x)),x=1.502+(0.523×结节大小)+(-1.095×毛刺征)+(-1.972×空泡征)+(-1.544×血管集束征)+(-1.051×分叶征)+(-1.795×边界不清)+(0.003×CT值),其内部验证和外部验证的AUC值分别为0.907和0.947,本模型浸润性腺癌的预测概率与实际概率的校准曲线散点图分布于参考线附近,表明两者具有较好的一致性。结论本研究建立了一个列线图模型,用于预测肺浸润性腺癌风险,为临床医生提供了一个作为术前肺浸润性癌个体化预测的无创性工具,帮助他们更直观的评估直径≤1 cm肺磨玻璃结节的风险程度,为后续和治疗提供参考。

Objective Using clinical and CT data,a nomogram model was constructed and validated for differentiating invasive lung adenocarcinoma from preinvasive lung lesions in ground-glass nodules≤1cm in diameter.Methods The clinical data and CT image characteristics of patients with ground glass nodules in the lungs were analyzed retrospectively.Independent risk factors were screened out by binary Logistic regression analysis,and the IPA risk prediction diagram model was established by using independent risk factors.the predictive performance of the model was verified and evaluated using calibration curves and the area under the receiver operating characteristic curve(AUC).Results Data from 182 patients were used to develop the model,including 53 males(29.1%)and 131 females(70.9%),aged 26 to 77 years with a mean of(52.1±11.1),of which 51 were PIL and 131 IPA.Univariate and multivariate analysis identified nodule size,burr sign,vacuole sign,vascular sign,lobular sign,nodule boundary ambiguity and CT value as independent risk factors,and established a line graph prediction model.The calculation formula of the prediction model is as follows:IPA=e^(x)/(1+e^(x)),x=1.502+(0.523×nodule size)+(-1.095×burr sign)+(-1.972×vacuolar sign)+(-1.544×vascular cluster sign)+(-1.051×phyllographic sign)+(-1.795×borderless)+(0.003×CT value).The AUC values of internal verification and external verification were 0.907 and 0.947,respectively.The scatter plots of calibration curves of predicted probability and actual probability of invasive adenocarcinoma in this model were distributed near the reference line,indicating a good consistency between the two.Conclusion In this study,a nomogram model was established to predict the risk of invasive adenocarcinoma of the lung,providing clinicians with a noninvasive tool for individualized prediction of invasive lung cancer before surgery,helping them more intuitively assess the degree of risk of pulmonary ground glass nodules≤1cm in diameter,and providing reference for follow-up and treatment.