Prediction of EGFR mutations by conventional CT-features in advanced pulmonary adenocarcinoma

Objective: This study assessed the ability of conventional computed tomography (CT) features (including primary tumors, metastatic lesions, lymph nodes, and emphysema) to predict epidermal growth factor receptor (EGFR) mutations in advanced pulmonary adenocarcinoma.

Methods: Patients who were diagnosed with advanced pulmonary adenocarcinoma between January 2017 and August 2017 and had undergone a chest CT and EGFR mutation testing were enrolled in this retrospective study. Qualitative and quantitative CT-features and clinical characteristics evaluated in this study included: primary tumor location, size, and morphology (including degree of lobulation, density, calcification, cavitation, vacuole sign, and air bronchogram), size and distribution of lung and pleural metastatic nodules, size and status of hilar and mediastinal lymph nodes, emphysema, organs with distant metastasis, and patient age, sex, and smoking history.

Results: Of 201 patients, 107 (53.23%) were EGFR-mutation positive. The multivariate logistic regression indicated that EGFR mutations were significantly associated with smaller lymph nodes, a lower percentage of deep lobulation of the primary tumor and partial fusion of lymph nodes, and absence of emphysema. The area under the curve of logistic regression model for predicting EGFR mutations was 0.898.

Conclusions: Conventional CT-features, including emphysema, degree of primary tumor lobulation, and lymph node size and status, help to predict the presence or absence of EGFR mutations in advanced pulmonary adenocarcinoma. Additionally, these same CT-features demonstrated that the CT manifestations of the EGFR mutant group were of relatively lower malignancy and less invasive as compared to the wild-type EGFR group.