Integrated analysis of therapeutic strategies and prognostic factors in advanced lung adenocarcinoma: Retrospective study with emphasis on gene assays, multimodality treatment approaches and predictive machine learning models

Patients with advanced lung adenocarcinoma have a range of treatment options, including targeted therapy and gene assay-guided chemotherapy. The aim of the present study was to investigate the prognostic factors and treatment-related variables influencing overall survival (OS), and to develop and validate machine learning models to predict OS in patients with advanced lung adenocarcinoma. Data on the clinical features and treatment strategies of patients with advanced lung adenocarcinoma were collected, and the effects of these variables on OS were analyzed. A total of 575 patients were included in a retrospective analysis. Among these patients, 38.4% had bone metastases, 17.9% had liver metastases, 36.0% had brain metastases and 28.9% had lung metastases. Smoking status, wild-type EGFR and older age were identified as poor prognostic factors. Patients with EGFR mutations demonstrated a prolonged OS. However, differences in OS were not observed among patient subgroups stratified by programmed death-ligand 1 expression and by common EGFR mutation subtypes, namely exon 19 deletion and L858R. First-line treatment with the tyrosine kinase inhibitor afatinib was associated with improved OS compared with that of patients treated with erotinib or gefitinib. In addition, combination therapy with the angiogenesis inhibitor bevacizumab had a positive impact on OS. Finally, several machine learning models were validated to predict OS using clinical and molecular features, and their performance was assessed using the concordance index. These findings highlight the importance of molecular profiling and individualized treatment strategies in optimizing OS for patients with advanced lung adenocarcinoma. Furthermore, the validated machine learning models may serve as useful tools for risk stratification and personalized prognostic assessment to support clinical decision-making.