Integrating network toxicology, interpretable machine learning, and molecular docking to unravel key molecular networks in diethyl phthalate-related rheumatoid arthritis

Diethyl phthalate (DEP) is widely used in everyday consumer products and has been associated with an increased risk of rheumatoid arthritis (RA), but its potential mechanisms remain unclear. This study elucidated the potential computational links between DEP and the molecular networks implicated in RA pathogenesis. Employing a network toxicology framework, we systematically delineated DEP-related molecular networks and screened key RA targets by integrating target prediction, RA transcriptomic data mining, interpretable machine learning, and molecular docking. Following batch correction, we conducted differential expression analysis and weighted gene co-expression network analysis (WGCNA), followed by functional enrichment of the intersecting targets. Core genes were prioritized via multiple machine-learning algorithms interpreted using SHapley Additive exPlanations (SHAP). Finally, molecular docking and molecular dynamics (MD) simulations were performed to computationally assess structural binding stability. A total of 69 targets overlapping between DEP and RA were identified. Among the constructed models, the LASSO-RF sequential framework achieved optimal discrimination, with an apparent AUC of 0.9998 (95% CI, 0.999–1.000) in the training set and robust cross-cohort validation in GSE55235 (AUC = 0.830; 95% CI, 0.647–1.000) and GSE77298 (AUC = 0.839; 95% CI, 0.655–1.000). We identified six hub genes ( NLRP1, ADORA2A, XBP1, PPP1CA, SAE1, and PIN1 ), which enrichment analyses linked to immune inflammation, programmed cell death, and metabolic regulation. Notably, molecular docking suggested thermodynamically plausible interactions between DEP and four of these targets: ADORA2A, NLRP1, PIN1, and PPP1CA. A 100-ns MD simulation further supported the conformational stability of the DEP–PIN1 complex. Our computational framework underscores a putative link between DEP and critical pathways governing adenosine signaling and inflammasome activation. While limited by the lack of empirical exposure data and experimental validation, this study offers valuable hypothesis-generating insights. These findings establish a foundation for subsequent research to elucidate the mechanistic interplay between environmental factors and RA risk.