Ferroptosis is not the tidy, programmed self-destruction of apoptosis. It is a messier, iron-driven form of cell death in which lipid peroxides accumulate in cell membranes until they rupture. Discovered formally just over a decade ago, ferroptosis has since been implicated in cancer, cardiovascular disease, neurodegeneration, and inflammatory disorders. What makes it especially relevant to endometriosis is the disease’s own biochemistry. Endometriotic lesions shed blood into the peritoneal cavity month after month, flooding the local environment with free iron. Studies have documented that this iron overload can trigger ferroptotic damage in granulosa cells of the ovary and even in early embryos, contributing to the infertility that so often accompanies the disease. Retrograde menstruation—the reflux of endometrial tissue through the fallopian tubes into the pelvic cavity—is widely accepted as the initiating event of endometriosis, and ferroptosis may facilitate the survival and implantation of ectopic endometrial cells as they migrate and colonize new territory.

The research team, led by Buhaiqiemu Kadeer and corresponding author Zhifang Chen, assembled their evidence from the Gene Expression Omnibus, a public repository of gene expression data maintained by the National Center for Biotechnology Information. They cross-referenced these endometriosis datasets with the FerrDb database, a manually curated catalog of genes that regulate or mark ferroptosis. From this intersection they extracted the raw material for a multilayered computational investigation that pooled an array of modern bioinformatic techniques: differential expression analysis with the Limma package, weighted gene co-expression network analysis to find groups of genes acting in concert, functional enrichment through Gene Ontology, KEGG pathway, and Metascape frameworks, and a protein-protein interaction network to map which gene products physically and functionally connect.

The initial screen was striking. The analysis revealed 18 differentially expressed genes tied to ferroptosis in endometriotic tissue, along with 36 genes from co-expression modules that clustered with the disease state. When the researchers overlapped and filtered these candidates, eleven core genes emerged. From there, the team deployed two complementary machine learning strategies to distill the list further: LASSO regression, a statistical technique that penalizes complexity and prunes away weak predictors, and the Random Forest algorithm, an ensemble method that builds hundreds of decision trees and votes on the most informative features. Both approaches converged on the same six genes—BRD7, OSBPL9, AGPS, NRAS, PEX12, and NCOA4—suggesting that this sextet carries genuine signal rather than statistical noise.

Each of the six genes tells a biologically coherent story about why iron metabolism and lipid chemistry might matter in endometriosis. NCOA4, or nuclear receptor coactivator 4, is the cargo receptor that delivers ferritin—the cell’s iron storage cage—to autophagy machinery for destruction, a process aptly named ferritinophagy. When NCOA4 ramps up, stored iron is released into the cytoplasm, raising the intracellular iron pool and pushing cells closer to the ferroptotic threshold. AGPS encodes an enzyme at the heart of ether lipid biosynthesis, and ether lipids are among the most vulnerable substrates for the peroxidation reactions that drive ferroptosis. PEX12, a peroxisomal biogenesis factor, connects to peroxisome function, an organelle intimately involved in very-long-chain fatty acid and lipid metabolism. OSBPL9 belongs to the oxysterol-binding protein family, tying the picture to cholesterol and sterol transport, while NRAS—a well-known signaling oncogene—and BRD7, a chromatin regulator with anti-inflammatory and tumor-suppressive roles, link ferroptosis to the proliferation and inflammatory signaling characteristic of endometriotic lesions.

To test whether these six genes could actually tell a sick sample from a healthy one, the researchers built a multi-gene logistic regression model and evaluated its diagnostic performance. The model showed high diagnostic value for distinguishing endometriosis from non-diseased tissue, positioning the six-gene panel as a candidate blood- or tissue-based biomarker signature. Because endometriosis currently lacks any validated non-invasive molecular test, a gene expression signature with strong discriminatory power would represent a meaningful advance, potentially allowing earlier intervention before lesions, adhesions, and fertility damage become entrenched.

But the study went beyond diagnosis to probe the immune landscape of the disease. Using CIBERSORT, a computational tool that estimates the proportions of different immune cell types within bulk tissue expression data, the team found that alterations in the endometriotic immune microenvironment appear closely related to the six hub genes. This is a significant observation because endometriosis is increasingly understood as an inflammatory and immune-dysregulated condition. Lesions are infiltrated by macrophages, and immune cells in the peritoneal cavity often fail to clear refluxed endometrial cells effectively. The correlation between ferroptosis-related gene expression and immune cell composition suggests that iron-driven cell death and immune dysfunction may feed into one another, each amplifying the other in a vicious cycle that sustains lesion growth and inflammation.

The final layer of the analysis reached upward in the regulatory hierarchy, from genes to the microRNAs that control them. By constructing a miRNA-target regulatory network, the researchers predicted three key microRNAs—hsa-mir-125a-5p, hsa-mir-218-5p, and hsa-mir-124-3p—as potential upstream modulators of the hub genes. These small RNA molecules fine-tune protein production by binding messenger RNAs, and each of the three has independent credentials in reproductive biology. MiR-124-3p has been shown to influence endometrial receptivity by altering epithelial cell polarity and adhesion, and it is upregulated in the endometrium and serum of women with chronic endometritis. MiR-125a-5p has been studied in ischemia-reperfusion injury and inflammatory contexts, while miR-218-5p participates in neuroinflammatory regulation. Together, the three microRNAs offer a second tier of candidate biomarkers and potential drug targets, since manipulating a single microRNA can influence multiple genes in a pathway simultaneously.

The authors are careful to frame their work as hypothesis-generating. All of the data derive from publicly available, previously published and fully de-identified datasets, so no new biological samples were collected, and the computational predictions will need validation in independent cohorts of patients, ideally using clinical specimens and prospective study designs. Nonetheless, the convergence of machine learning, network biology, and immunology around a single gene panel offers a template for how complex, poorly understood diseases can be interrogated computationally before expensive clinical trials begin. The study was funded by research projects examining NCOA4-mediated iron autophagy and ferroptosis in endometriosis, indicating that the team intends to pursue these mechanisms experimentally.

If the six-gene signature and its three regulatory microRNAs hold up under experimental and clinical scrutiny, the implications could extend well beyond diagnostics. Genes like NCOA4 and AGPS sit at control points of iron handling and lipid peroxidation that are already attracting pharmaceutical interest, with ferroptosis-inducing and ferroptosis-inhibiting compounds under investigation in oncology. Repurposing or refining such agents for endometriosis could open an entirely new therapeutic avenue for a disease whose current treatments—hormonal suppression and surgery—manage symptoms but rarely cure the condition. For millions of women waiting for an answer and an effective treatment, a molecular map of ferroptosis in their lesions may prove to be the first genuinely new lead in decades.

Subject of Research: Ferroptosis-related genes as diagnostic and predictive biomarkers in endometriosis identified through integrated bioinformatic analysis

Article Title: Diagnostic and Predictive Values of Ferroptosis-Related Genes in Endometriosis Based on Integrated Bioinformatic Analysis

Article References: Kadeer, B., Wufuer, S., Maimaitimin, A., Ablat, N., Li, X., & Chen, Z. (2026). Diagnostic and Predictive Values of Ferroptosis-Related Genes in Endometriosis Based on Integrated Bioinformatic Analysis. Reproductive Sciences. https://doi.org/10.1007/s43032-026-02142-3

Image Credits: AI Generated

DOI: 10.1007/s43032-026-02142-3

Keywords: endometriosis, ferroptosis, biomarkers, NCOA4, BRD7, AGPS, NRAS, PEX12, OSBPL9, microRNA, bioinformatics, immune infiltration