The Value Of ¹⁸F-FDG PET/CT Imaging In Predicting Epidermal Growth Factor Receptor Mutation Status In Pulmonary Adenocarcinoma

Objective:To study the differences of ¹⁸F-FDG PET/CT metabolic parameters and the differences of radiomic features extracted from ¹⁸F-FDG PET images and CT images between EGFR mutation and wild type in lung adenocarcinoma.Elucidating the diagnostic values of metabolic parameters and radiomic features in predicting EGFR mutations in order to provide a reference for individualized treatment of patients with lung adenocarcinoma through non-invasive examination.Method:We retrospectively reviewed 125 pulmonary adenocarcinoma patients who underwent EGFR mutation testing after surgical treatment and pretreatment ¹⁸F-FDG PET/CT from June 2016 to September 2017.All patient PET/CT images were retrieved and SUVmax,SUVmean,SUVpeak,MTV,TLG of tumor lesions were automaticlly measured on the AW4.6 workstation using the PETVCAR software application.?1?Analyze the differences of the clinical and CT features between EGFR mutants and wild type,and the differences of SUVmax,SUVmean,SUVpeak,MTV,TLG between EGFR mutants and wild type,K-RAS mutants and wild type,EGFR mutants and K-RAS mutants.ROC Curves of different conventional parameters were analyzed.?2?We chose 115 patients whose tumor lesions were mainly based on solid components from 125 pulmonary adenocarcinoma patients.Delineate the contour of tumor lesions in the largest cross sections of PET and CT images using image J software.Then extract PET image features and CT image features by different image feature extraction algorithms.The extracted features were dimensioned by LASSO algorithm.The differences of radiomic features between EGFR mutant and wild type were analyzed.The correlation between PET conventional parameters and radiomic features were compared.ROC curves of radiomic features were analyzed.Result:The patients with EGFR mutations were mainly non-smokers.The histological subtype of EGFR mutations was mainly acinar.The SUVmax,SUVmean,SUVpeak,TLG of EGFR mutations were significantly lower than wild-types.The SUVmax,SUVmean and SUVpeak of K-RAS mutations were significantly higher than wild-types.The SUVmax,SUVmean and SUVpeak of K-RAS mutations were significantly higher than EGFR mutations.The SUVmean of primary lung lesions were the highest parameter to predict EGFR mutant status in pulmonary adenocarcinoma.But there was no significant difference between SUVmean with SUVmax,SUVpeak,MTV and TLG.In the radiomic features of PET images,concavity,gray mean,homogeneity,energy,correlation,entropy,contrast was significantly different between EGFR mutated and wild-types.In the radiomic features of CT images,gray span was significantly different between EGFR mutated and wild-types.Concavity was moderately related to MTV,TLG,homogeneity was moderately related to MTV,TLG.Correlation was strongly related to entropy,so as to entropy and homogeneity,contrast and homogeneity.Entropy was moderately related to gray mean,so as to gray mean and correlation.The remaining parameters were weakly correlated or uncorrelated.In addition to concavity,the accuracy of predicting EGFR mutated status using radiomics of PET images was higher than metabolic parameters.The accuracy of predicting EGFR mutated status using radiomics of CT images was slightly higher than metabolic parameters.Conclusion:?1?EGFR mutation status can be predicted by SUVmax,SUVmean,SUVpeak and TLG in patients with lung adenocarcinoma;?2?To some extent,K-RAS mutation can affects the prediction accuracy of EGFR gene mutation status using ¹⁸F-FDG PET metabolic parameters;?3?The prediction accuracy of EGFR mutation status of lung adenocarcinoma with the majority of solid components by the radiomic features of PET imaging was higher than metabolic parameters.The prediction accuracy of EGFR mutation status of lung adenocarcinoma with the majority of solid components by the radiomic features of CT imaging was slightly higher than metabolic parameters.?4?The glucose metabolism in the wild-type lesion was less regular,and more uniform in the local area,while the glucose metabolism in the mutant lesion of EGFR was opposite.The glucose metabolism of EGFR mutated lesion at the edge of the lesion was more nonuniform.