An integrated analysis of lactylation signature reveals PRDX1 as a therapeutic target of keloid pathogenesis

Keloids represent a pathological fibroproliferative disorder with high recurrence rates and limited therapeutic options. This study integrates multi-dataset transcriptomics (GSE158395, GSE188952, GSE92566, GSE173900) and machine learning algorithms (XGBoost, Random Forest, LASSO) to systematically investigate the role of lactylation modification in keloid pathogenesis. We identified 26 lactylation-related differentially expressed genes (15 upregulated, 11 downregulated) enriched in oxidative stress, immune response, and extracellular matrix pathways. Machine learning convergence revealed five lactylation hub genes (PRDX1, CSRP1, IFI16, CALD1, VIM), with PRDX1 exhibiting the highest diagnostic efficacy (AUC = 0.85). Immune infiltration analysis demonstrated significant correlations between hub genes and dysregulated immune cells. Experimental validation confirmed reduced PRDX1 expression in keloid tissues; its knockdown in fibroblasts elevated ROS levels and enhanced proliferation and migration. Regulatory network analysis predicted shared transcription factors (KLF12, NFKB1, MYC) governing hub genes, while drug screening prioritized three clinically actionable compounds (acetaminophen, valproic acid, vorinostat) targeting PRDX1. These findings establish lactylation as a critical regulator of keloid pathogenesis and identify PRDX1 as a promising therapeutic target.