Abstract
Tumor evolution involves genetic, transcriptional, and phenotypic alterations that shape cancer cell behavior and interactions with the microenvironment. While single-cell technologies have advanced our understanding of this process, spatial dynamics remain incompletely characterized. Here, whole-exome sequencing (WES), imaging mass cytometry (IMC), and spatial transcriptomics (ST) were integrated to study molecular evolution and immune responses in two lung adenocarcinoma (LUAD) mouse models: a genetically engineered model (129S4/Sv-Kras(LSL-G12D), termed 129S4 K) and a carcinogen-induced precancer model (129S4 U). Compared to 129S4 K, 129S4 U tumors exhibited higher mutational, neoantigen but lower copy number variation (CNV) burdens at matched developmental timepoints, consistent with findings of higher mutational burden in human smoking-related LUAD than nonsmoking LUAD. We profiled over 1.4 million spatial single cells from 284 IMC regions of interest and 51,531 spatial transcriptomic spots from 156 lesions across 141 mice. Macrophage abundance increased with tumor progression, while CD8 T-cell and B-cell densities declined in late-stage LUAD. 129S4 U showed greater immune infiltration in both tumor and adjacent normal tissue, higher T-cell cytotoxicity signature score in line with its higher mutational and neoantigen burdens. LUAD progression was marked by early morphological shifts and late-stage changes in cell states and interactions. These data define spatial and genetic landscapes of LUAD development and provide a framework for investigating immune evolution and therapeutic strategies in early carcinogenesis.