Integrated multiomics methodology and in vitro experiments for the detection of key genes involved in amiodarone-induced pulmonary fibrosis

Amiodarone is a first-line antiarrhythmic drug, but its long-term use causes irreversible amiodarone-induced pulmonary fibrosis (AIPF) with no reliable early biomarkers. The aim of this study was to identify key pathogenic genes of AIPF via integrated multiomics and in vitro validation. Network toxicology was used to identify overlapping genes between amiodarone targets and pulmonary fibrosis-associated genes. A machine learning framework that integrates 127 algorithms was established with cross-validation using independent Gene Expression Omnibus datasets. Molecular docking, molecular dynamics simulation, single-cell RNA sequencing, and spatial transcriptomics were performed to verify the candidate genes. An in vitro AIPF model was constructed using amiodarone-treated A549 cells. Eighty-four overlapping genes were identified and enriched in necroptosis and wingless/integrated (Wnt) signaling pathways. SERPING1 was prioritized as the primary core biomarker, while lipocalin 2 (LCN2) was identified as a secondary candidate gene with prominent inter-cohort spatial heterogeneity. Both proteins were stably bound to amiodarone and were expressed predominantly in lung injury-associated cell populations. Spatial transcriptomics revealed that SERPING1 expression was weakly positively correlated with fibrosis, whereas LCN2 exhibited variable, cohort-dependent expression correlation patterns across datasets. In vitro experiments confirmed the dose-dependent downregulation of SERPING1 expression in amiodarone-exposed cells. In this study, SERPING1 was prioritized as a candidate gene potentially associated with amiodarone-induced pulmonary fibrosis by integrating network toxicology, cross-dataset machine learning, molecular simulation, single-cell and spatial transcriptomic analyses, and in vitro validation. LCN2 represents a secondary candidate but hampered by inter-cohort heterogeneity and awaits further experimental validation. Rather than establishing entirely new disease drivers, these findings extend previous network-based AIPF studies by providing multiomics localization evidence and in-vitro support, particularly for SERPING1 downregulation in amiodarone-exposed epithelial cells. However, further validation in animal models and clinical AIPF samples is needed.