Predicting EGFR mutation status in lung adenocarcinoma: development and validation of a computed tomography-based radiomics signature

Patients with epidermal growth factor receptor (EGFR) mutations in lung adenocarcinoma can benefit from targeted therapy. However, noninvasively determination of EGFR mutation status before targeted therapy remains a challenge. This study constructed a nomogram based on a combination of radiomics features with the clinical and radiological features to predict the EGFR mutation status. The least absolute shrinkage and selection operator (LASSO) and Wilcoxon test were used for feature selection. Decision tree (DT), logistic regression (LR), and support vector machine (SVM) classifiers were used for radiomics model building. Used the clinical and radiological features establish clinical-radiology (C-R) model. The C-R model with the best radiomics model to establish clinicalradiological-radiomics (C-R-R) model. The predictive performance of the model was evaluated by ROC and calibration curves, and the clinical usefulness was assessed by a decision curve analysis. The current study showed that twelve radiomics features were significantly associated with EGFR mutations. The best radiomics signature model was obtained using the SVM classifier. The C-R-R model had the best distinguishing ability for predicting the EGFR mutation status, with an AUC of 0.849 (95% CI, 0.805-0.893) and 0.835 (95% CI, 0.761-0.909) in the development and validation cohorts, respectively. Our study provides a non-invasive C-R-R model that combines CT-based radiomics features with clinical and radiological features, which can provide useful image-based biological information for targeted therapy candidates.

Automated summary

This study developed a nomogram combining radiomics features with clinical and radiological features to predict EGFR mutation status in lung adenocarcinoma, achieving high predictive performance and clinical usefulness.