A novel risk signature for predicting brain metastasis in patients with lung adenocarcinoma

Abstract

Background: Brain metastasis (BM) are a devastating consequence of lung cancer. This study was aimed to screen risk factors for predicting BM. Methods: Using an in vivo BM preclinical model, we established a series of lung adenocarcinoma (LUAD) cell subpopulations with different metastatic ability. Quantitative proteomics analysis was used to screen and identify the differential protein expressing map among subpopulation cells. Q-PCR and Western-blot were used to validate the differential proteins in vitro. The candidate proteins were measured in LUAD tissue samples (n = 81) and validated in an independent TMA cohort (n = 64). A nomogram establishment was undertaken by performing multivariate logistic regression analysis. Results: The quantitative proteomics analysis, qPCR and Western blot assay implied a five-gene signature that might be key proteins associated with BM. In multivariate analysis, the occurrence of BM was associated with age ≤ 65 years, high expressions of NES and ALDH6A1. The nomogram showed an area under the receiver operating characteristic curve (AUC) of 0.934 (95% CI, 0.881-0.988) in the training set. The validation set showed a good discrimination with an AUC of 0.719 (95% CI, 0.595-0.843). Conclusions: We have established a tool that is able to predict occurrence of BM in LUAD patients. Our model based on both clinical information and protein biomarkers will help to screen patient in high-risk population of BM, so as to facilitate preventive intervention in this part of the population.