# Identification of potential ferroptosis-related biomarkers in proliferative vitreoretinopathy based on machine learning

## TL;DR

This study identifies TIMP1 and STAT3 as potential ferroptosis-related biomarkers in proliferative vitreoretinopathy, linking them to immune changes and disease progression.

## Contribution

Novel identification of ferroptosis-related genes TIMP1 and STAT3 as potential biomarkers for PVR, linking them to immune infiltration and oxidative stress pathways.

## Key findings

- TIMP1 and STAT3 show strong discriminative ability between PVR and control samples.
- TIMP1 and STAT3 expression correlates with immune cell subsets in PVR.
- GPX4 protein is upregulated in a rabbit PVR model compared to controls.

## Abstract

Proliferative vitreoretinopathy (PVR) is a blinding retinal condition often linked to retinal detachment, eye trauma, and complications following intraocular surgery. Although oxidative stress and epithelial–mesenchymal transition (EMT) are well-recognized contributors to PVR pathogenesis, whether ferroptosis-related pathways are involved in this process remains unclear.

Differentially expressed genes (DEGs) were identified from the GSE28133 dataset and intersected with ferroptosis-related genes curated from FerrDb to obtain ferroptosis-related differentially expressed genes (FRDGs). Functional enrichment analyses, protein–protein interaction network construction, and machine learning approaches (LASSO regression and SVM-RFE) were applied to identify key candidate genes. Immune infiltration was analyzed using CIBERSORT. Experimental validation was performed using an in vitro EMT model of retinal pigment epithelial cells and a rabbit PVR model.

Functional enrichment analyses indicated that FRDGs were mainly involved in wound healing, tissue remodeling, oxidative stress responses, and ferroptosis-related pathways. TIMP1 and STAT3 were identified as ferroptosis-associated candidate genes with strong discriminative ability between PVR and control samples in the discovery dataset. Immune infiltration analysis revealed distinct immune cell profiles in PVR samples and significant correlations between TIMP1 and STAT3 expression and multiple immune cell subsets. Experimental validation confirmed upregulation of TIMP1 and STAT3 in the EMT model. In addition, Western blot analysis demonstrated significantly increased GPX4 protein expression in retinal tissues from the rabbit PVR model compared with controls.

This study identifies TIMP1 and STAT3 as ferroptosis-associated candidate genes in proliferative vitreoretinopathy and highlights potential links among ferroptosis-related regulatory pathways, immune microenvironment alterations, and PVR pathogenesis. These findings provide a foundation for further mechanistic studies to clarify the role of ferroptosis in PVR.