Brain metastases from lung adenocarcinoma are the most common type of brain metastatic tumors in adults. Patients with lung adenocarcinoma have a high probability of developing brain metastases (11%) [1], resulting in poor prognosis with short survival [2]. Lung cancer is one of the most common malignant tumors threatening human life, with a 5-year survival rate of only 10−20% [3].

The epidermal growth factor receptor (EGFR) gene is recognized as the most common driver mutation gene in non-small cell lung cancer (NSCLC) in Asia. Exon 19 deletion (19Del) and exon 21 L858R point mutation (21L858R) account for approximately 85% of EGFR mutations, and both exhibit strong sensitivity to tyrosine kinase inhibitors (TKIs) [4]. Lung cancers with 19Del and 21L858R mutations are often treated uniformly in clinical practice; however, clinical studies with various subpopulation groups have indicated that these two types of lung cancer may not have the same sensitivity to TKIs [5]. Furthermore, an increasing number of studies have demonstrated differences between 21L858R-positive and 19Del-positive patients in terms of molecular structure, drug resistance mechanism, and clinical characteristics [6]. Among patients with advanced lung cancer, although both 21L858R-positive and 19Del-positive patients benefited from molecular targeted therapies, those with 19Del exhibited longer time to progression and longer survival following TKI treatment, as compared with 21L858R-positive patients [7]. Riely et al. reported similar findings, with patients with 19Del demonstrating a survival advantage over patients with 21L858R (median 34 months vs. 8 months; p=0.01) [8]. Therefore, distinguishing between these two EGFR mutation subtypes and developing a more tailored and individualized treatment plan could provide better treatment benefits not only for patients with 21L858R, but also for those with 19Del [9].

Currently, biopsy is the standard method for confirming EGFR gene mutations in NSCLC. However, biopsy is an invasive method that may lead to complications, such as bleeding and infection, and the biopsy specimens are not representative of the whole tumor due to the tumor heterogeneity [10]. Radiomics is a method of extracting quantitative features from standard medical images and transforming them into data that can be manually mined. These data can be applied to auxiliary clinical diagnosis, prognosis assessment, and treatment response prediction. Although a prior study has evaluated the utility of a radiomics model based on contrast-enhanced T1-weighted images (T1WI) to predict EGFR mutations in primary lung cancer lesions [11], further research on predicting mutation subtypes is scarce [12].

Therefore, in this study, we investigated the ability of magnetic resonance imaging (MRI)-based radiomics features extracted from lung adenocarcinoma brain metastases to differentiate between molecular subtypes of EGFR mutations. This could aid clinicians in establishing accurate diagnosis in a noninvasive manner and formulating more reasonable individualized treatment plans.