Pattools‑implemented methylation vector analysis reveals aberrant subtype‑specific methylation in lung cancer across tissue and plasma cfDNA.

INTRODUCTION: Lung cancer is characterised by high mortality and encompasses various subtypes with markedly different treatment outcomes. While individual subtypes have been studied, comprehensive genome-wide and fine-scale DNA methylation analyses across subtypes remain unexplored. This study aims to identify true subtype-specific aberrant methylation by developing a novel method that eliminates cell-origin and immune confounding, moving beyond traditional approaches. METHODS: We assembled an in-house cohort of 115 tissue samples across five groups (CTL, LUAD, LUSC, LCC, SCLC) with detailed clinicopathological annotations. Matched plasma cfDNA samples (n = 24; 9 LUAD, 7 LUSC, 8 healthy) were collected for translational assessment. We developed an MV analysis method implemented in the open-source toolkit pattools to differentiate cell-specific fragments in bulk BS-seq data and identify subtype-specific methylation vector regions (SMVRs). RESULTS: Genome-wide methylation profiling revealed that subtype-specific signals originate from distinct cell types-alveolar epithelium in LUAD, head and neck epithelium in LUSC, fibroblasts in LCC, and endocrine cells in SCLC-with additional immune infiltration influences. Using pattools-based MV analysis, we identified 500 key SMVRs per subtype. In matched plasma cfDNA, these tissue-derived SMVRs demonstrated promising discriminatory power for LUAD (AUC = 0.98, 95% CI: 0.95 to 1.00) and LUSC (AUC = 0.88, 95% CI: 0.79 to 0.97) in binary classification, and AUCs of 0.95 (95% CI: 0.90 to 1.00), 0.75 (95% CI: 0.62 to 0.88), and 0.63 (95% CI: 0.48 to 0.78) for LUAD, LUSC and healthy samples, respectively, in multi-class classification. CONCLUSION: The MV analysis in pattools effectively uncovers subtype-specific aberrant methylation signals, offering potential for precise diagnosis and subtyping in tissue and liquid biopsy. Independent validation in larger cohorts is required before clinical translation.