Abstract
BACKGROUND: While DNA methylation profiling from peripheral blood mononuclear cells (PBMCs) has demonstrated utility in cancer risk prediction, notably for non-small cell lung cancer (NSCLC), its prognostic value for survival stratification in Chinese lung adenocarcinoma (LUAD) patients remains unestablished. This study addresses whether PBMC-derived methylation signatures can discriminate clinical outcomes in EGFR-mutation LUAD subgroups. METHODS: We performed genome-wide methylation analysis of PBMCs from LUAD patients using the Infinium Methylation EPIC 850 K array. Clinical characteristics were associated with overall survival (OS) through Cox regression. Prognostic differentially methylated positions (DMPs) were identified via Lasso regression, followed by the construction of risk-score models. Functional enrichment (KEGG/GO) and tissue microarray-based immunohistochemistry (IHC) for FKBP4 expression (n = 90 LUAD samples) were conducted. Analyses were conducted in R 4.4.1 with curated Bioconductor packages. RESULTS: In the retrospective cohort of 174 Chinese LUAD patients (April 2014-September 2019), PBMC analysis of 128 cases revealed 12 hypomethylated DMPs were associated with OS. EGFR-mutant patients (n = 66) showed 325 significant DMPs (|Δβ|≥ 0.06, P ≤ 0.01), with four DMPs (cg05802998, cg19313959, cg00685115, cg15224444) independently predicting OS. The cg19313959 located in the TSS1500 region of FKBP4 gene (Δβ = 0.21) demonstrated the strongest methylation shift. Reduced FKBP4 protein expression was associated with improved survival (HR = 0.42, 95%CI 0.24-0.72). In EGFR-wildtype patients (n = 51), three prognostic DMPs emerged from 2,531 candidates. EGFR mutation-specific prognostic scoring models were established successfully, while pathway analyses revealed divergent biological processes between EGFR subgroups. CONCLUSION: In this epigenome-wide study based on PBMCs in Chinese patients with LUAD, methylation signatures dependent on EGFR mutations and predictive of survival were identified.