18F-FDG PET/CT代谢参数联合肿瘤标志物对肺腺癌EGFR基因突变的预测价值

 

目的 探讨¹⁸F-FDG正电子发射断层显像计算机断层扫描(positron emission tomography/computed tomography, PET/CT)代谢参数联合肿瘤标志物对肺腺癌EGFR基因突变的预测价值。方法 选取2010-01至2018-08于武警特色医学中心PET/CT检查诊断为原发性肺癌并经组织病理学证实为肺腺癌的患者,提取患者的基本临床资料、PET/CT影像学特征、PET/CT代谢参数、肿瘤标志物水平及EGFR基因突变数据,根据EGFR是否突变将所有患者分为突变型组与野生型组。通过单因素logistic回归分析,将对EGFR基因突变有预测价值的相关参数纳入预测模型,绘制ROC受试者工作曲线,计算其Cutoff值,对比各个模型的AUC值,建立可预测EGFR基因突变的模型,并分析该模型的预测效能。结果 共纳入肺腺癌105例,EGFR突变型32例(30.5%),EGFR野生型73例(69.5%),多因素分析显示两组之间的基本临床资料(性别χ²=5.74,P=0.017)、吸烟(χ²=4.60,P=0.032)、CT影像特征[密度χ²=5.77,P=0.016)、毛刺征χ²=2.06,P=0.015)、肺内转移(χ²=2.91,P=0.088)]、PET/CT代谢参数[SUVmean(t=2.82,P=0.015)]、肿瘤标志物[CEA(t=-2.48,P=0.016)]具有统计学意义(P<0.05),是影响EGFR基因突变的独立影响因素,并以该独立影响因素建立了4种EGFR基因突变预测模型,相应的预测准确度分别为78.1%、81.0%、78.1%和81.9%,4种预测模型均具有预测的统计学意义(P<0.05)。结论 与EGFR基因突变有关的单因素影响因子包括吸烟、性别、SUVmean、密度、毛刺征、肺内转移、CEA,而根据这些影响因子建立的预测模型4将有助于临床分析肺癌患者是否发生EGFR基因的突变,从而指导临床进行靶向药物的个体化治疗。

18F-FDG PET/CT metabolic parameters combined with tumor markers for prediction of EGFR gene mutations in lung adenocarcinoma

Objective To investigate the predictive value of ¹⁸F-FDG PET/CT metabolic parameters combined with tumor markers for EGFR gene mutations in lung adenocarcinoma. Methods Patients who were diagnosed with primary lung cancer by PET/CT and histopathologically confirmed as cases of lung adenocarcinoma between January 2010 and August 2018 were selected. The basic clinical data, PET/CT imaging features, PET/CT metabolic parameters, tumor marker levels and EGFR gene mutation data were collected.These patients were divided into the mutation group and wild type group according to the occurrence of mutation. Based on univariate logistic regression analysis, the related parameters of predictive value for EGFR gene mutations were included in the prediction model.The working curve of ROC was drawn for these subjects, the cut-off value calculated, the AUC value of each model compared, and a model for predicting EGFR gene mutations was established before the prediction efficiency of the model was analyzed. Results A total of 105cases of lung adenocarcinoma were included, including 32 cases of mutant EGFR (30.5%) and 73 cases of wild-type EGFR (69.5%).Multivariate analysis showed that the difference in gender(χ²=5.74,P=0.017), rate of smoking(χ²=4.60,P=0.032), density(χ²=5.77,P=0.016), burr sign(χ²=2.06,P=0.015), intra-pulmonary metastasis(χ²=2.91,P=0.088), SUVmean(t=2.82,P=0.015), and CEA(t=-2.48,P=0.016) was statistically significant between the two groups (P<0.05). Four prediction models were established for EGFR gene mutations, and the corresponding prediction accuracy was 78.1%, 81.0%, 78.1%, and 81.9%, respectively. The four prediction models were of statistical significance (P<0.05). Conclusions The single factors related to EGFR gene mutations include smoking, sex, SUVmean, density, burr sign, intrapulmonary metastasis and CEA. The four prediction models based on these factors will help to analyze the occurrence of EGFR gene mutations in patients with lung cancer and guide the individualized treatment with targeted drugs.