Abstract
Retinol, a pivotal regulator of cellular growth and apoptosis, has garnered substantial attention for its intricate involvement in cancer development. To explore Vitamin A's impact on lung adenocarcinoma (LUAD), we utilized comprehensive datasets from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) to dissect the intricate role of retinol in cancer progression. The unsupervised clustering analysis, grounded on retinol metabolism gene sets, divided patients into two distinct clusters, with cluster 1 exhibiting significantly inferior survival outcomes. Through differential analysis, we uncovered 349 differentially mutated and 394 differentially expressed genes between these clusters. Leveraging these discoveries, we built a seven-gene signature model that precisely predicted poorer survival for patients with a higher risk score, which was subsequently validated in four independent GEO cohorts, demonstrating its robustness and reliability. Our drug sensitivity analysis further revealed that high-risk patients were more susceptible to gefitinib and erlotinib. Notably, leveraging gene dependency scores and RNA-seq data from LUAD cell lines, we identified Phosphoribosylaminoimidazole Carboxylase And Phosphoribosylaminoimidazolesuccinocarboxamide Synthase (PAICS) as a potential therapeutic target. Single-RNA sequencing confirmed PAICS's predominant expression in cancer cells, and functional assays underscored its oncogenic role in promoting cell proliferation, migration, and invasion. These novel findings offer profound insights into potential therapeutic avenues for LUAD patients with poor prognoses, paving the way for future research endeavors. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s43657-025-00223-y.