Integrative single-cell and machine learning framework reveals prognostic fibroblast subtypes and constructs a fibroblast-related risk signature in lung adenocarcinoma.

Lung adenocarcinoma (LUAD) is a major subtype of non-small cell lung cancer and continues to contribute substantially to global cancer mortality. Within the tumor ecosystem, cancer-associated fibroblasts (CAFs) are key stromal components that significantly influence LUAD progression. However, their phenotypic diversity and clinical implications remain incompletely elucidated. We integrated two single-cell RNA sequencing datasets (GSE171145 and GSE189357) to delineate the transcriptional landscape and developmental trajectory of fibroblasts in LUAD. A fibroblast-related signature (FRS) was developed by intersecting fibroblast-specific markers with differentially expressed genes from the TCGA-LUAD cohort, followed by univariate Cox analysis and machine learning modeling. A total of 101 combinations of ten machine learning algorithms were evaluated. The prognostic value of the FRS was validated across multiple GEO datasets. We further investigated its associations with immune infiltration, genomic alterations, and therapeutic response. The core gene TIMP1 was subjected to in vitro and clinical validation. We identified pronounced fibroblast heterogeneity in LUAD, with distinct differentiation trajectories revealed by pseudotime analysis. The constructed FRS exhibited robust prognostic performance across cohorts and was significantly correlated with immunosuppressive features, tumor mutation burden, and predicted immunotherapy outcomes. Clinically, the FRS served as an independent prognostic indicator and showed favorable calibration when combined with TNM stage in a nomogram. TIMP1, one of the top-ranked risk genes in univariate Cox analysis, was confirmed to be upregulated in tumor samples and to promote cell invasion and proliferation in vitro, supporting its functional role in LUAD progression. This study developed a fibroblast-based prognostic signature through integrative single-cell and bulk transcriptomic analyses. The FRS effectively stratifies LUAD patients and highlights the dynamic roles of fibroblasts in shaping tumor progression, providing potential biomarkers and therapeutic targets.