Abstract
Background: Tyrosine kinase inhibitors (TKIs) provide clinical benefits to the lung cancer patients with epidermal growth factor receptor (EGFR) mutations. However, non-invasively determine EGFR mutation status in patients before targeted therapy remains a challenge. This study aimed to develop and validate a nomogram for preoperative prediction of EGFR mutation status in patients with lung adenocarcinoma.Methods: This study retrospectively collected medical records of 403 patients with histologically confirmed lung adenocarcinoma from January 2016 and June 2020. The patients were divided into development and validation cohorts. The preoperative information on all patients was obtained, including clinical characteristics and computed tomography (CT) features. Multivariate logistic regression analysis was used to develop the predictive model. We combined CT features and clinical risk factors and used them to build a prediction nomogram. The performance of the nomogram was evaluated in terms of calibration, discrimination, and clinical usefulness. The nomogram was further validated in an independent external cohort.Results: The predictive factors incorporated in the personalized prediction nomogram included smoking history (OR, 0.2; 95% CI: 0.1, 0.4; P < 0.001), bubble-like lucency (OR, 2.2; 95% CI: 1.3, 3.8; P = 0.003), pleural attachment (OR, 0.4; 95% CI: 0.2, 0.7, P = 0.001) and thickened adjacent bronchovascular bundles (OR, 3.1; 95% CI: 1.8, 5.3; P < 0.001). Based on these parameters, the prediction model has good discrimination and calibration ability. The area under the curve in the development and validation cohorts were 0.784 (95% CI: 0.733, 0.835) and 0.740 (95% CI: 0.643, 0.838), respectively. Decision curve analysis showed that the model was clinically useful.Conclusions: This study presented a nomogram that contained CT features and clinical risk factors, which could conveniently and non-invasively predict EGFR mutation status in patients with lung adenocarcinoma before surgery.