A Tumor-infiltrating B lymphocytes -Related Index Based on Machine-learning Predicts Prognosis and Immunotherapy Response in Lung Adenocarcinoma

Jiale Fang ^1* Siyuan Yu ¹ Wei Wang ^2,3 Cheng Liu ¹ Xiaojia Lv ¹ JiaQi Jin ¹ Xiaomin Han ^1,3,4 Fang Zhou ³ Yukun Wang ^1*

• ¹ Southern University of Science and Technology, Shenzhen, China
• ² Air Force Medical University, Xi'an, Shaanxi Province, China
• ³ Department of Pharmacy, Southern University of Science and Technology Hospital, Shenzhen, China
• ⁴ Baotou Medical College, Baotou, China

Tumor-infiltrating B lymphocytes (TILBs) play a pivotal role in molding the immune microenvironment of tumors (TIME) and in the course of cancer development within lung adenocarcinoma (LUAD). Despite their recognized importance, comprehensive and methodical delineations of TIL-Bs within LUAD are scarce, and the precise molecular pathways through which TILBs modulate the efficacy of immunotherapy remain to be elucidated. The research employed single-cell RNA sequencing from the GSE117570 dataset to pinpoint markers linked to TILBs. A comprehensive machine learning method, utilizing ten distinct algorithms, facilitated the creation of a TILBs-related index (BRI) encompassing the TCGA, GSE31210, and GSE72094 datasets. BRI independently served as a risk determinant in LUAD, evidencing strong and dependable capability in forecasting total survival rates. In the datasets containing outcomes of immunotherapy patients (GSE91061 and GSE126044), we found that patients with lower BRI were more likely to respond to immunotherapy. Furthermore, lower BRI were linked to reduced IC50 values in standard chemotherapy and targeted therapy medications, in contrast to higher BRI. Additionally, the BRI-based survival prediction nomogram demonstrated significant promise for clinical application in predicting the 1-, 3-, and 5-year total survival rates among LUAD patients.