A three-subtype prognostic classification based on base excision repair and oxidative stress genes in lung adenocarcinoma and its relationship with tumor microenvironment.

Unrepaired DNA damage is the initiation of mutation and tumor-specific biological characteristics. Oxidative stress and base excision repair (BER) are the two main pathways to cope with oxidative DNA damage, which is closely related to the heterogeneity of Lung adenocarcinoma (LUAD), but their relationship with tumor biological characteristics is unclear, and a molecular subtyping based on comprehensive BER and oxidative stress gene expression is lacking. 501 samples from The Cancer Genome Atlas (TCGA) were classified into three subtypes based on genes related to BER and oxidative stress through hierarchical agglomerative cluster analysis. By integrating the nearest template prediction (NTP), four GEO datasets and 52 samples from our institution were analyzed for validation. Bioinformatic analysis was performed to define the diverse molecular characteristics, mutation background, tumor microenvironment, and prognosis. Three subtypes with distinct gene signatures were identified: relatively high BER and low oxidative stress gene expression (C1), low BER gene and high oxidative stress gene expression (C2), and high expression of both BER and oxidative stress genes (C3). C2 was characterized by a low mutation frequency in TP53 (29%) and a high mutation frequency in EGFR (20%), whereas a high frequency of mutation was seen in C3 in STK11 and KEAP1 genes. Additionally, differentially expressed genes among the three subtypes were particularly enriched in immune-related pathways, and the abundance of immune cells and Immunophenoscore were significantly higher in C2, while the Tumor Immune Dysfunction and Exclusion (TIDE) score was lower in C2, indicating a better response to immunotherapy. C2 was also associated with an improved survival outcome compared with C1 and C3, and this finding was validated in 978 samples from four independent GEO datasets and 52 samples at our institution by the NTP algorithm. The three-subtype classifications based on BER and oxidative stress gene expression offers potential for predicting the survival and response to immunotherapy of LUAD patients.