基于CT图像特征预测肺腺癌Ki-67表达水平的可行性研究

Feasibility of predicting Ki-67 expression levels in lung adenocarcinoma based on CT morphological characteristics

目的 :探讨基于CT图像特征预测肺腺癌Ki-67表达水平的可行性。方法 :选取经病理证实为肺腺癌且行免疫组化Ki-67表达水平检测的患者60例,分为Ki-67低表达组(33例)和Ki-67高表达组(27例)。采用Discovery750HD CT设备进行图像采集,测量2组患者的病灶大小、CT值,并观察肺腺癌的病变位置、发生类型及CT征象。采用Logistic回归分析筛选预测Ki-67表达水平的影响因子,同时绘制ROC曲线联合多指标评价CT图像特征预测Ki-67表达水平的诊断效能。数据采用SPSS 22.0软件进行统计学分析。结果:Ki-67低表达组的表达水平为(26.21±10.83)%,Ki-67高表达组的表达水平为(66.67±9.20)%,2组间存在统计学差异(P<0.05)。CT图像特征中毛刺征、淋巴结肿大、血管集束征与肺腺癌Ki-67高表达有关(P<0.05),支气管充气征与肺腺癌Ki-67低表达有关(P<0.05),静脉期CT值在2组间具有统计学差异(P<0.05)。单变量Logistic回归分析结果表明毛刺征、支气管充气征、淋巴结肿大、血管集束征及静脉期CT值是Ki-67表达水平的影响因子。多变量Logistic回归分析结果表明支气管充气征是Ki-67表达水平的独立保护因子,毛刺征和淋巴结肿大是独立危险因子。ROC曲线联合毛刺征、支气管充气征、淋巴结肿大、血管集束征等征象和静脉期CT值预测Ki-67表达水平的AUC为0.851。结论:CT图像特征中毛刺征、支气管充气征、淋巴结肿大、血管集束征和静脉期CT值可以作为评价肺腺癌Ki-67表达水平的指标,为临床评价肺腺癌增殖活性和生物学行为提供影像学诊断依据。

Objective To explore the feasibility of predicting Ki-67 expression levels in lung adenocarcinoma based on CT morphological characteristics.Methods Sixty patients having pathologically confirmed lung adenocarcinoma and undergoing immunohistochemical evaluation of Ki-67 expression levels were selected and divided into a low Ki-67 expression group(33 patients)and a high Ki-67 expression group(27 patients).Discovery 750 HD CT device was used for image acquisition to measure the lesion size and CT values of the two groups,and to observe the location,type and CT signs of lung adenocarcinoma.The influencing factors for predicting Ki-67 expression levels were screened with Logistic regression analysis,while ROC curves were plotted to combine multiple indicators to evaluate the diagnostic efficacy of CT features in predicting Ki-67 expression levels.SPSS 22.0 software was employed for data statistical analysis.Results The expression level was(26.21±10.83)%in the low Ki-67 expression group and(66.67±9.20)%in the high Ki-67 expression group,with a statistical difference between the 2 groups(P<0.05).Among the CT image features burr sign,lymph node enlargement and vessel convergence sign were associated with high Ki-67 expression in lung adenocarcinoma(P<0.05),while air bronchogram sign was associated with low Ki-67 expression(P<0.05),and CT values in the venous phase were statistically different between the two groups(P<0.05).Univariate Logistic regression analysis showed that burr sign,air bronchogram sign,lymph node enlargement,vessel convergence sign and venous CT values were the factors influencing the levels of Ki-67 expression,while multivariate Logistic regression analysis proved that air bronchogram sign was an independent protective factor for Ki-67 expression levels,and burr sign and lymph node enlargement were independent risk factors.The ROC curve indicated combined burr sign,air bronchogram sign,lymph node enlargement,vessel convergence sign and venous CT value had the AUC being 0.851 when used to predict Ki-67 expression levels.Conclusion The burr sign,air bronchogram sign,lymph node enlargement,vessel convergence sign and venous CT values in CT image features can be used as indicators to evaluate the levels of Ki-67 expression in lung adenocarcinoma,providing a diagnostic imaging basis for clinical evaluation of the proliferative activity and biological behavior of lung adenocarcinoma.