Abstract
Lung adenocarcinoma (LUAD) is the most common and lethal cancer worldwide. Radiotherapy (RT) is widely used at all stages of LUAD, and the development of immunotherapy substantially enhances the survival of LUAD patients. Although the emerging treatments for LUAD have improved prognosis, only a small fraction of patients can benefit from clinical therapies. Thereby, approaches assessing responses to RT and immunotherapy in LUAD patients are essential. After integrating the analysis of RT, immunization, mRNA, and clinical information, we constructed a signature based on 308 tumor-infiltrating B lymphocyte-specific genes (TILBSig) using a machine learning method. TILBSig was composed of 6 B cell-specific genes (PARP15, BIRC3, RUBCNL, SP110, TLE1, and FADS3), which were highly associated with the overall survival as independent factors. TILBSig was able to differentiate better survival compared with worse survival among different patients, and served as an independent factor for clinical characteristics. The low-risk TILBSig group was correlated with more immune cell infiltration (especially B lineages) and lower cancer stem cell characteristics than the high-risk group. The patients with lower risk scores were more likely to respond to RT and immunotherapy. TILBSig served as an excellent predicator for prognosis and response to immunotherapy and RT in LUAD patients.