A novel ferro-aging-related gene signature for prognosis and targeted drug prediction in cervical cancer

Abstract Objective Cervical cancer (CESC) remains a leading cause of cancer-related mortality in women. Recent studies have identified ferro-aging as a genuine biological pathway in primates, representing an iron-dependent metabolic process. However, cellular ageing and the decline in overall physiological function play a crucial role in tumour progression. Based on this, we have established a prognostic model related to ferro-aging and evaluated its predictive and diagnostic value in cervical cancer. Methods We integrated data from the The Cancer Genomics Atlas (TCGA), the Gene Expression Online Resource (GEO) and single-cell RNA sequencing datasets. Following the identification of ferro-aging-related gene (FARGs) based on the literature, we constructed a prognostic model using machine learning algorithms and analysed model features via Shapley-based interpretability (SHAP) analysis. The Scissor algorithm was employed to evaluate cell populations associated with high- and low-risk groups. Studies on intercellular communication and spatial transcriptomics revealed the process of tumour microenvironment remodelling. Finally, potential therapeutic compounds were identified through AI-assisted virtual screening and molecular docking. Results A total of eight FAR model genes were ultimately identified: DUOX1, E2F1, IL1B, KEAP1, MAP3K14, SLAMF8, TFRC and TNFAIP3 . This study constructed and validated a novel risk profile model comprising these eight ferro-aging-related genes; this model demonstrated moderate prognostic discriminatory power in both internal and external cohorts, with the tumour microenvironment in the high-risk group exhibiting an immunosuppressive state. Using single-cell RNA sequencing to assess cellular heterogeneity, macrophages, epithelial cells and CD8 + T cells were identified as key factors in CESC progression. Pseudotime analysis was performed to visualize the expression levels of FARGs along the pseudotime axis and explore how FARGs participate in the dynamic process of EndMT. Spatial transcriptomics was employed to map the distribution and expression of FARGs within epithelial cells in the tumour microenvironment. Intercellular communication intensity was increased in the low-risk group, with epithelial cells in this group signalling to macrophages via the APP-CD74 pathway. Drug screening indicated that BRD-A56371469 is likely to bind to DUOX1. Conclusions The FARG signature is a robust independent prognostic indicator for CESC. The integrated pipeline offers insights into ferro-aging biology and suggests novel therapeutic candidates.