Reveal a risk and potential mechanism for keratoconus progression with isotretinoin: Toxicogenomic, transcriptomics and real-world evidence insights

Abstract

Purpose: Limited clinical data and unstable animal models hinder understanding isotretinoin-induced keratoconus. We addressed this by integrating pharmacovigilance with computational approaches.

Methods: Isotretinoin-associated ocular surface adverse events were extracted from the FDA Adverse Event Reporting System (FAERS). Disproportionality analyses detected pharmacovigilance signals, and age confounding bias was assessed with multivariable logistic regression. Network toxicology and transcriptomics defined candidate targets. Enrichment analyses implicated biological processes and pathways. Key targets were refined via random forest, SVM-RFE, and LASSO regression. Single-cell transcriptomic profiling delineated the cell-type-specific expression patterns of the four targets in corneal stromal cells, epithelial cells, and immune cell subsets. Computer-Aided Drug Design examined the molecular binding mechanisms between isotretinoin and the three identified target proteins.

Results: Keratoconus is strongly associated with isotretinoin (log₁₀ROR = 1.32, 95 % CI = 1.10-1.56), supported across multiple algorithms. Age showed no significant association (OR = 0.99, 95 % CI: 0.92-1.04, p = 0.795). 19 overlapping genes identified, predominantly involved in extracellular matrix processes such as cell migration and growth factor binding. Machine learning prioritized key molecular markers: IL1A, THBS1, IGFBP3, and TGFB1. Single-cell transcriptomics delineated IGFBP3 and TGFB1 enrichment in corneal epithelial cells, IL1A in immune cells, and THBS1 in both stromal and immune compartments. Molecular docking and dynamics demonstrate that isotretinoin can stably bind to the three target proteins, maintaining stable binding conformations and interactions throughout the simulation period.

Conclusions: This integrative analysis identifies a robust pharmacovigilance signal between isotretinoin and keratoconus and warrants further clinical and experimental investigation.

Keywords: FAERS; Isotretinoin; Keratoconus; Pharmacovigilance; Single-cell; Transcriptomic analysis.