Evolution of the pulmonary immune microenvironment from premalignant lesions to invasive lung adenocarcinoma: A spatial transcriptomic study.

e20029 Background: Pulmonary adenocarcinoma develops through a multistep process from premalignant epithelial lesions to invasive disease. While therapeutic advances have improved outcomes in advanced stages, the immune microenvironment during early lung carcinogenesis remains poorly understood. Atypical adenomatous hyperplasia (AAH) is a key precursor lesion whose study may provide insight into early mechanisms of tumor progression and identify biomarkers of aggressive behavior. Methods: We conducted a retrospective multicenter study including patients with pulmonary adenocarcinoma who underwent surgical resection between 2010 and 2025 at three Spanish hospitals (N=23). Regions of interest corresponding to invasive tumor, peritumoral stroma, and AAH were selected from FFPE samples. Spatial transcriptomic profiling was performed using the NanoString GeoMx Digital Spatial Profiler RNA platform with quality control, Q3 normalization, and paired statistical analyses accounting for intra-patient correlation. Penalized regression models were used to derive a gene signature, and associations with frequently mutated genes were explored using public databases. Molecular data were integrated with clinical outcomes. Results: Spatial transcriptomics revealed progressive gene expression changes across lung adenocarcinoma carcinogenesis. Genes upregulated in dysplasia compared with normal tissue (>2-fold change) included C4BPA, SERPINA1, MLPH, NKX2-1, CD24, GDF15, FOXJ1, THBS1, TPSAB1/B2, PECAM1, and MRC1. An eleven-gene expression signature accurately discriminated normal lung, AAH, and invasive adenocarcinoma (AUC 0.996). Signature genes were associated with epithelial differentiation, microenvironment remodeling, angiogenesis, and immune polarization. AAH lesions showed intermediate signature expression between normal tissue and invasive adenocarcinoma, consistent with gradual activation of tumor-associated transcriptional programs. While individual genes such as NKX2-1 correlated with improved overall survival in public datasets (p=0.011), the combined signature was not prognostic for overall survival (p=0.98). Conclusions: Premalignant pulmonary lesions, particularly AAH, already harbor transcriptomic alterations that distinguish them from invasive disease. This signature reflects progressive activation of biological programs during lung adenocarcinoma carcinogenesis and may help identify early biological vulnerabilities to prevent progression to invasive cancer.