Studies have highlighted the significant role of focal adhesion signaling in cancer. Nevertheless, its specific involvement in the pathogenesis of endometrial cancer and its clinical significance remains uncertain. We analyzed TCGA-UCEC and GSE119041 datasets with corresponding clinical data to investigate focal adhesion-related gene expression and their clinical significance. A signature, "FA-riskScore," was developed using LASSO regression in the TCGA cohort and validated in the GSE dataset. The FA-riskScore was compared with four existing models in terms of their prediction performance. We employed univariate and multivariate Cox regression analyses towards FA-riskScore to assess its independent prognostic value. A prognostic evaluation nomogram based on our model and clinical indexes was established subsequently. Biological and immune differences between high- and low-risk groups were explored through functional enrichment, PPI network analysis, mutation mining, TME evaluation, and single-cell analysis. Sensitivity tests on commonly targeted drugs were performed on both groups, and Connectivity MAP identified potentially effective molecules for high-risk patients. qRT-PCR validated the expressions of FA-riskScore genes. FA-riskScore, based on FN1, RELN, PARVG, and PTEN, indicated a poorer prognosis for high-risk patients. Compared with published models, FA-riskScore achieved better and more stable performance. High-risk groups exhibited a more challenging TME and suppressive immune status. qRT-PCR showed differential expression in FN1, RELN, and PTEN. Connectivity MAP analysis suggested that BU-239, potassium-canrenoate, and tubocurarine are effective for high-risk patients. This study introduces a novel prognostic model for endometrial cancer and offers insights into focal adhesion's role in cancer pathogenesis.
Keywords: Disease subtyping; Endometrial cancer; Machine learning; Precision medicine; Prognosis evaluation.
© 2024. The Author(s), under exclusive licence to Society for Reproductive Investigation.