ACAT1/ACAT2 as biomarkers for idiopathic pulmonary fibrosis: insights from transcriptomics analysis and machine learning

Idiopathic pulmonary fibrosis (IPF) is a devastating interstitial lung disease characterized by dysregulated inflammation and progressive lung scarring, ultimately culminating in mortality due to respiratory complications. Metabolic reprogramming in IPF progression has been extensively studied, yet the role of tryptophan metabolism remains poorly understood. This study aimed to identify key tryptophan metabolism-related genes (TMGs) as diagnostic biomarkers for idiopathic pulmonary fibrosis (IPF) and elucidate their cell-type-specific roles in IPF pathogenesis. Public bulk RNA sequencing of human IPF datasets and single-cell RNA-seq (scRNA-seq) data were analyzed to assess the expression level of TMGs. Machine learning algorithms (Boruta, LASSO, SVM-RFE, XGBoost, random forest) identified critical TMGs. Validation included immunohistochemistry, qRT-PCR, Western blot, and immunofluorescence in bleomycin-induced mouse models and human IPF lung tissues. Further scRNA-seq analysis characterized cell-type-specific TMG expression and metabolic activity. ACAT1 and ACAT2 were identified as core TMGs significantly downregulated in IPF, as identified by transcriptomic profiling and machine learning. ScRNA-seq revealed reduced AT2 cell abundance and suppressed TMG activity in IPF. Correlation analyses linked ACAT1/ACAT2 to metabolic pathways (e.g., amino acid metabolism) and fibrosis-related processes. Mouse and human validation confirmed decreased ACAT1/ACAT2 expression in fibrotic lungs, correlating with AT2 cell dysfunction. Inhibition of ACAT1/ACAT2 using Avasimibe significantly aggravated fibrosis. ACAT1 and ACAT2 are novel diagnostic biomarkers and potential therapeutic targets for IPF, with their suppression in AT2 cells reflecting metabolic dysregulation central to fibrosis progression. This study underscores the therapeutic potential of targeting AT2 cell metabolic homeostasis to mitigate IPF. Multi-omics and machine learning approaches provide a robust framework for identifying metabolic biomarkers in complex diseases.