
==== Front
Medicine (Baltimore)
Medicine (Baltimore)
MD
Medicine
0025-7974
1536-5964
Lippincott Williams & Wilkins Hagerstown, MD

39029052
MD-D-23-10741
00054
10.1097/MD.0000000000038875
3
5600
Research Article
Diagnostic Accuracy Study
Identification and analysis of biomarkers associated with oxidative stress and ferroptosis in recurrent miscarriage
Xie Jinxia MD 243160566@qq.com
ab
Zhu Hongli MD 1259533657@qq.com
c
Zhao Shaozhi MD teszsz@163.com
c
Ma Yongqin MD 3559200247@qq.com
c
Shi Panpan MD 463825883@qq.com
c
Zhan Xuxin MD zhanxuxin3968@163.com
c
Tian Wenyan PhD, MD ab*
https://orcid.org/0000-0001-9883-5054
Wang Yingmei PhD, MD wangyingmei@tmu.edu.cn
ab
a Department of Gynecology and Obstetrics, Tianjin Medical University General Hospital, Tianjin, China
b Tianjin Key Laboratory of Female Reproductive Health and Eugenics, Tianjin Medical University General Hospital, Tianjin, China
c Xi’an Gynecology and Obstetrics Hospital, Xi’an People’s Hospital (Xi’an Fourth Hospital), Affiliated Guangren Hospital, School of Medicine, Xi’an Jiaotong University, Xi’an, China.
* Correspondence: Wenyan Tian, Tianjin Medical University General Hospital, No. 154 Anshan street, Heping district, Tianjin 300052, China (e-mail: tianwenyan1108@163.com).
19 7 2024
19 7 2024
103 29 e3887515 1 2024
18 6 2024
19 6 2024
Copyright © 2024 the Author(s). Published by Wolters Kluwer Health, Inc.
2024
https://creativecommons.org/licenses/by-nc/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial License 4.0 (CCBY-NC), where it is permissible to download, share, remix, transform, and buildup the work provided it is properly cited. The work cannot be used commercially without permission from the journal.

Recurrent miscarriage (RM) has a huge impact on women. Both oxidative stress and ferroptosis play an important role in the pathogenesis of RM. Hence, it was vital to screen the ferroptosis oxidation-related biomarkers for the diagnosis and treatment of RM. We introduced transcript data to screen out differentially expressed genes (DEGs) in RM. Ferroptosis oxidation-related differentially expressed genes were obtained by overlapping DEGs and oxidative stress related genes with correlations >0.9 with ferroptosis-related genes. Least Absolute Shrinkage and Selectionator operator regression and support vector machine based recursive feature elimination algorithm were implemented to screen feature genes. The biomarkers associated with ferroptosis oxidation were screened via receiver operating characteristic curve analysis. We finally analyzed the competing endogenous RNAs regulatory network and potential drugs of biomarkers. We identified 1047 DEGs in RM. Then, 9 ferroptosis oxidation-related differentially expressed genes were obtained via venn diagram. Subsequently, 8 feature genes (PTPN6, GJA1, HMOX1, CPT1A, CREB3L1, SNCA, EPAS1, and TGM2) were identified via machine learning. Moreover, 4 biomarkers associated with ferroptosis oxidation, including PTPN6, GJA1, CPT1A, and CREB3L1, were screened via receiver operating characteristic curve analysis. We constructed the ‘227 long noncoding RNAs-4 mRNAs-36 microRNAs’ network, in which hsa-miR-635 was associated with CREB3L1 and PTPN6. There were 11 drugs with therapeutic potential on 3 biomarkers associated with ferroptosis oxidation. We also observed higher expression of CPT1A and CREB3L1 in RM group compared to the healthy control group by quantitative real-time reverse transcription polymerase chain reaction. Overall, we obtained 4 biomarkers (PTPN6, GJA1, CPT1A, and CREB3L1) associated with ferroptosis and oxidative stress, which laid a theoretical foundation for the diagnosis and treatment of RM.

bioinformatics
biomarkers
ferroptosis oxidation
recurrent miscarriage
OPEN-ACCESSTRUE
==== Body
pmc1. Introduction

When the body is in a pregnant state, if the time is <12 weeks, but there are no <3 or more miscarriages, it can be considered as recurrent natural abortion.[1] Analyzing existing research data can clarify that having 2 miscarriages can lead to recurrent miscarriage (RM).[2] Moreover, when the body undergoes an abortion, it not only poses a serious threat to physical health, but also poses a serious threat to mental health.[3] At this point, the body’s own negative emotions can further exacerbate the situation. There are many factors that may lead to this type of miscarriage in the body, including hormonal disorders and immune dysfunction.[4] Numerous therapies have been tested, including those that target immunological mechanisms (polyvalent immunoglobulins, tiny doses of corticosteroids), endocrine systems (progesterone, for example), or hemostasis (LMWH, aspirin).No treatment appears to be efficient as of now.[5]

Foreign research scholar Helmut proposed the concept of oxidative stress (OS) after in-depth research, which means that the relationship between antioxidant defense and oxidant production cannot be maintained in a state of motherhood, which can damage the biological system.[6] Existing research data shows that systemic OS and placenta can play a crucial role in the pathogenesis of RM.[7] A researcher[8] pointed out that if the body does not have a normal mitochondrial respiratory chain, it can lead to oxidative stress, which can have an impact on the pregnancy process of the body and increase the likelihood of recurrent miscarriage. Ishii et al[9] stated that placental oxidative stress causes placental angiodysplasia and recurrent abortions through inflammation, and it can also cause embryopathy in early pregnancy and cause termination of growth through excessive apoptosis.

Since the birth of this term, researchers have been conducting research and analysis on it, and conducting intense discussions. Many metabolic pathways can have important effects on this process, such as lipid and sugar metabolism, ferroptosis treatment, etc.[10] Studies have demonstrated that the existence of crosstalk pathways between oxidative stress and ferroptosis. For instance, glutathione, a critical regulator of ferroptosis, is needed to mitigate the rise of reactive oxygen species, while GPX4 can suppress ferroptosis by activating the nuclear factor kappa-B pathway and lowering reactive oxygen species production.[11] However, the role of ferroptosis combined with oxidative stress in RM has not been clarified.

In this study, for the first time, publicly available data from the gene expression omnibus (GEO) database as well as various bioinformatics tools were used to identify and analyze oxidative stress and ferroptosis related biomarkers and their potential mechanisms of action in RM, providing reference for clinical staff in developing treatment plans.

2. Materials and methods

2.1. Data collection

In total, GSE165004 and GSE26787 datasets of endometrial tissue samples, including RM and healthy control, were collected from GEO. The GSE165004 dataset (Agilent-039494 SurePrint G3 Human GE v2 8 × 60K Microarray 039381) consisted of 24 RM and 24 HC cases. The GSE89632 dataset (5 HC vs 5 RM samples),which form Affymetrix Human Genome U133 Plus 2.0 Array, was used to verify the expression level of biomarkers. In addition, 1093 oxidative stress related genes (OSRGs) were derived from GeneCards database (www.enecards.or/) (The correlation score > 7). Four hundred eighty-four ferroptosis related genes were derived from FerrDb database.

2.2. Analysis of differential genes

The “limma” R package[12] was applied to excavate differentially expressed genes (DEGs) between RM group and HC group in GSE165004. The P value < .05 & |log2fold change| > 0.5 was determined as the significance criteria. Volcano plot and heatmap were applied to show DEGs in RM.

2.3. The correlation and function enrichment analysis

The correlation coefficients between OSRGs and ferroptosis related genes was calculated via ‘Hmisc’ package. Based on the correlation coefficient >0.9, ferroptosis related OSRGs were extracted. Ferroptosis oxidation-related differentially expressed genes (FOR-DEGs) were obtained by overlapping DEGs in 2 groups and ferroptosis related OSRGs. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of FOR-DEGs were conducted via “clusterProfiler” (version 4.4.4) package.[13] The P value < .05 was selected as criterion.

2.4. Machine learning methods

Least Absolute Shrinkage and Selectionator operator regression analysis and support vector machine based recursive feature elimination algorithm were utilized to screen feature genes. Biomarkers associated with ferroptosis oxidation were obtained through taking the intersection genes of 2 machine learning algorithms. Moreover, receiver operating characteristic curves were plotted to evaluate the diagnostic value of the biomarkers by “pROC” package.[14]

2.5. Clinical nomogram model

The nomogram containing biomarkers were drawn via “rms” to predict morbidity of RM patients. Evaluation of the predictive effect was done by the calibration curve.

2.6. Gene set enrichment analysis (GSEA) and ingenuity pathway analysis (IPA)

The Spearman correlation between biomarkers and all genes was analyzed. Based on ranking of correlation coefficients, GSEA was conducted to explore the potential GO items and KEGG pathways associated with biomarkers through “clusterProfiler” package.[15] In order to explore the inhibition or activation state of the biological pathway for biomarkers, IPA was performed (P < .05). Z-score > 0 was considered as the activation state and Z-score < 0 was considered as the inhibition state.

2.7. Construction of competing endogenous RNAs (ceRNA) network

The microRNAs (miRNAs) of biomarker associated with ferroptosis oxidation were predicted by TargetScan database (https://www.targetscan.org/vert_80/) and TarBase database (https://dianalab.e-ce.uth.gr/html/diana/web/index.phpr=tarbasev8), respectively. Overlapping miRNAs–mRNAs pairs were selected for subsequent analysis. The long noncoding RNAs targeting miRNAs were predicted by ENCORI database (https://starbase.sysu.edu.cn/) and miRNet database(https://www.mirnet.ca/upload/MirUploadView.xhtml), respectively. Overlapping long noncoding RNAs were selected for subsequent analysis. Moreover, the “lncRNAs-mRNAs-miRNAs” network was constructed via Cytoscape software[16] to optimize the results.

2.8. Potential drug prediction

In order to explore the potential therapeutic drugs for biomarkers associated with ferroptosis oxidation in BM, the targeting drugs of biomarkers were identified through the The Drug Gene Interaction Database (version.4.2.0) (https://www.dgidb.org).

2.9. The analysis and validation of the expression of biomarkers

In order to confirm the expression of biomarkers, we implemented quantitative real-time reverse transcription polymerase chain reaction (RT-qPCR). Human decidual tissue were obtained from patients with their knowledge and consent from Xi’an People’s Hospital (Xi’an No. 4 Hospital) hospital, and this study was approved by the Xi’an People’s Hospital (Xi’an No. 4 Hospital) hospital ethics committee. Total RNA of 20 samples was separated by the TRIzol (Ambion, Austin, TX) based on the manufacturer’s guidance. The inverse transcription of total RNA into cDNA was implemented by using the First-strand-cDNA-synthesis-kit (Servicebio, Wuhan, China) based on the producer’s indication. Then, RT-qPCR was carried out utilizing the 2×Universal Blue SYBR Green qPCR Master Mix (Servicebio, Wuhan, China) under the direction of the manual. The primer sequences for PCR were tabulated in Table 1. GAPDH was used as an internal reference gene, and the expression was calculated according to the 2−ΔΔCt method.[17]

Table 1 Sequence of primers used in the RT-qPCR experiments.

Gene name	Primer sequences	
PTPN6-F	ATCACCTATCCCCCAGCCAT	
PTPN6-R	GCAAGTACAGGCAGTTGGTG	
GJA1-F	GAGATCCCTGCCCACATCAG	
GJA1-R	GGACACCACCAGCATGAAGA	
CPT1A-F	ATCAATCGGACTCTGGAAACGG	
CPT1A-R	TCAGGGAGTAGCGCATGGT	
CREB3L1-F	GCCTTGTGCTTTGTTCTGGT	
CREB3L1-R	AGGGGGTCTTCCTTCACAGT	
GAPDH-F	CGAAGGTGGAGTCAACGGATTT	
GAPDH-R	ATGGGTGGAATCATATTGGAAC	

2.10. Statistical analysis

All P values of statistical results were based on two-sided statistical tests, and a P value < .05 was considered statistically significant.

3. Results

3.1. Identification of OSRGs in RM

One thousand forty-seven DEGs were identified in RM, including 557 down-regulated and 490 up-regulated genes (Fig. 1A and B). Subsequently, 9 FOR-DEGs were obtained via Venn diagram, including PTPN6, GJA1, CYBB, HMOX1, CPT1A, CREB3L1, SNCA, EPAS1, and TGM2 (Fig. 1C). To further probe the function of FOR-DEGs, functional enrichment analysis was conducted. GO results indicated that these FOR-DEGs were principally involved in “negative regulation of mast cell activation involved in immune response,” “negative regulation of secretion by cell,” and “cellular response to decreased oxygen levels” (Fig. 1D). Additionally, the KEGG analysis demonstrated that these FOR-DEGs were mainly enriched in the “AMPK signaling pathway” and “Glucagon signaling pathway” (Fig. 1E).

Figure 1. Identification of oxidative stress related genes (OSRGs) in recurrent miscarriage (RM). (A and B) The volcano plot (A) and heatmap (B) of differentially expressed genes (DEGs) between RM group and healthy control (HC) group in the GSE165004 dataset. (C) The venn diagram of 9 ferroptosis oxidation-related differentially expressed genes (FOR-DEGs). (D) The Gene Ontology (GO) results of FOR-DEGs. The green bar represents GO Biological Process (BP) terms, the blue bar represents GO Cellular Components (CC) terms. (E) The Kyoto Encyclopedia of Genes and Genomes (KEGG) results of FOR-DEGs.

3.2. Screening of biomarkers associated with ferroptosis oxidation

To further dig out the key genes, Least Absolute Shrinkage and Selectionator operator regression analysis was performed on 9 FOR-DEGs to unearth the optima. Ultimately, 8 feature genes were obtained, including PTPN6, GJA1, HMOX1, CPT1A, CREB3L1, SNCA, EPAS1, and TGM2 (Fig. 2A and B). Meanwhile, 8 feature genes were retained by support vector machine based recursive feature elimination algorithm, including PTPN6, GJA1, HMOX1, CPT1A, CREB3L1, SNCA, EPAS1, and TGM2 (Fig. 2C and D). Subsequently, we combined 8 overlapping genes between these 2 methods, including PTPN6, GJA1, HMOX1, CPT1A, CREB3L1, SNCA, EPAS1, and TGM2 (Fig. 2E). The area under curve values of 4 genes (PTPN6, GJA1, CPT1A, and CREB3L1) were >0.72, indicating an excellent diagnostic accuracy (Fig. 2F). We verified the diagnostic value of these 4 genes in the validation set GSE26787 (Fig. 2G). Thus, PTPN6, GJA1, CPT1A, and CREB3L1 were defined as biomarkers associated with ferroptosis oxidation in RM. To evaluate the diagnostic ability of biomarkers associated with ferroptosis oxidation, the nomogram containing biomarkers was generated (Fig. 2H). The calibration curve proved that the performance of the model was effective (Fig. 2I).

Figure 2. Screening of biomarkers associated with ferroptosis oxidation. (A and B) Least Absolute Shrinkage and Selectionator operator (LASSO) regression analysis was performed on 9 FOR-DEGs. (C and D) Eight feature genes were retained by support vector machine based recursive feature elimination (SVM-RFE) algorithm. (E) The Venn diagram of 8 overlapping genes. (F) The area under curve (AUC) values of 4 genes (PTPN6, GJA1, CPT1A, and CREB3L1). (G) The diagnostic value of PTPN6, GJA1, CPT1A, and CREB3L1 in the GSE26787 dataset. (H) The nomogram model containing 4 biomarkers. (I) The calibration curve of the nomogram model.

3.3. GSEA analysis based on biomarkers

To further study the potential roles of PTPN6, GJA1, CPT1A, and CREB3L1 in RM, we performed single-gene GSEA on biomarkers. The results showed that PTPN6 were mainly enriched in the “Oxidative phosphorylation” and “cytosolic large ribosomal subunit” (Fig. 3A and B). GJA1 was mainly enriched in the “positive regulation of cell activation” and “Cytokine-cytokine receptor interaction” (Fig. 3C and D). CREB3L1 was mainly enriched in the “cytosolic ribosome” and “Cytokine-cytokine receptor interaction” (Fig. 3E and F). CPT1A was mainly enriched in the “structural constituent of ribosome” and “Cell cycle” (Fig. 3G and H). Classical pathway analysis of IPA indicated that the biomarkers associated with ferroptosis oxidation were related to 25 pathways, such as “neurovascular coupling signaling pathway” and “IL-10 signaling” (Fig. 4A). The regulatory relationship of “S100 Family Signaling Pathway” (Z-score was the largest) was displayed in Figure 4B.

Figure 3. Gene set enrichment analysis (GSEA) based on 4 biomarkers. (A–H) The single-gene GSEA (GO and KEGG) on PTPN6 (A and B), GJA1 (C and D), CREB3L1 (E and F), and CPT1A (G and H).

Figure 4. The ingenuity pathway analysis (IPA) results of DEGs. (A) The column plot of IPA classical pathway enrichment results. Blue bars indicate that the corresponding pathway is inhibited, and orange bars indicate that it is activated. (B) The schematic graph of regulatory relationship of S100 Family Signaling Pathway. In a network where red is up-regulated genes and green is down-regulated genes, the network blue molecules in the network will be repressed, while orange molecules will be activated.

3.4. The ceRNA network analysis

To explore the regulatory mechanism of PTPN6, GJA1, CPT1A, and CREB3L1, the ceRNA network was constructed. We obtained 36 miRNAs and 227 lncRNAs. The “lncRNAs-mRNAs-miRNAs” network was constructed (Fig. 5A). The network had 267 nodes and 591 edges, in which hsa-miR-635 was associated with CREB3L1 and PTPN6. The hsa-miR-5582-3p and hsa-miR-33a-3p synchronously affected the expression of CPT1A.

Figure 5. Construction of the regulation network and validation of the biomarker expression. (A) The graph of the competing endogenous RNAs (ceRNA) network analysis. Green nodes indicate biomarkers, red nodes indicate targeted microRNAs (miRNAs), orange nodes indicate long noncoding RNAs (lncRNAs), blue lines indicate mRNA-miRNA relationship pairs, and gray lines indicate miRNA-lncRNA relationship pairs. (B) The graph of biomarkers-drug network constructed from 11 drugs with therapeutic potential for PTPN6, GJA1, and CPT1A. Purple nodes are biomarkers and green nodes are small molecule drugs. (C) The differential expression of biomarkers in RM group compared to the HC group in GSE165004 and GSE26787 datasets. (D) The differential expression of biomarkers in clinical BM samples versus control samples (**P < .01, ***P < .001, ****P < .0001).

3.5. Biomarkers-drug network

We explored the potential therapeutic drugs for PTPN6, GJA1, and CPT1A. There were 11 drugs with therapeutic potential on 3 biomarkers associated with ferroptosis oxidation (Fig. 5B), including LABETALOL, CARVEDILOL, BLEOMYCIN, PROPYLTHIOURACIL, ATENOLOL, PERHEXILINE, EPIGALOCATECHIN GALLATE, CHEMBL472004, SORAFENIB, CHEMBL510966, and TOFACITINIB.

3.6. Expression validation of the biomarkers

As illustrated in Figure 5C, we observed higher expression of PTPN6, CPT1A, and CREB3L1 in RM group compared to the HC group in GSE165004 and GSE26787 datasets. However, GJA1 was down-regulated in RM group. We further verified expression trend by RT-qPCR experiments (Fig. 5D). In agreement with the results of the public database data analysis, the expression of CPT1A and CREB3L1 were notably markedly over-expressed, and GJA1 was markedly reduced in clinical RM samples versus control samples. However, due to the heterogeneity of the sample PTPN6 showed the opposite trend.

4. Discussion

Foreign researchers have conducted research and analysis on it, and proposed corresponding guidance guidelines.[18] If the body has at least 2 or more pregnancy losses, it can be considered as recurrent pregnancy loss. After in-depth research, researchers have pointed out that oxidative stress can have a significant impact on this type of abortion.[7] Ferroptosis in placental tissues was involved in the pathogenesis of preeclampsia, spontaneous preterm birth, and gestational diabetes mellitus.[19] we obtained 4 biomarkers (PTPN6, GJA1, CPT1A, and CREB3L1) associated with ferroptosis oxidation. The protein tyrosine phosphatases (PTP) has been recognized as a major regulator of inflammation.[20] There are many proteins that can play an important role in controlling inflammation or transduction of dead cell signals, one of which is PTPN6.[21] Foreign researchers such as Speir have conducted in-depth research and pointed out that Ptpn6 can play a very important role in physiological activities in the human body, playing a dual role, not only in regulating IL-1 α/β The expression of Ripk1 has an impact and can also regulate tumor necrosis factor. Research has shown that it can effectively maintain Ripk1 function, thereby effectively inhibiting the process of cell death.[22] If the body has a high expression level, it can affect colon cancer cells and regulate processes such as cell migration.[23] Researcher Mok et al[24] pointed out through in-depth research that multiple tissues in invasive ovarian epithelial cancer have high levels of PTPN6 transcript expression, suggesting a close relationship between ovarian cancer and the expression of this gene. After thorough analysis of biological information and conducting a series of verifications, it was found that for postmenopausal women, PTPN6 can be used to determine whether osteoporosis will occur.[25] In this study, we observed higher expression of PTPN6 in RM group compared to the HC group in GSE165004 and GSE26787 datasets. Single-gene GSEA showed that PTPN6 was mainly enriched in the “oxidative phosphorylation” and “cytosolic large ribosomal subunit.” or cancer, research and analysis of oxidative phosphorylation 4 gene markers, such as IKZF3 and PTPN6, can predict the overall survival rate of patients. In normal tissues, if the above genes are identified, cancer can be identified, and based on this, patients can be divided into different groups.[26] Endometrial biopsy was performed on the control group and RM group. Analyze gene expression levels using microarrays. Clarify the differentially expressed genes related downregulation pathways of RM and recurrent implantation failure.[27] We hypothesize that this gene regulates the disease process by participating in these 2 signaling pathways.

Multiple connexin proteins are expressed in human tissues, including GJA1, commonly referred to as connexin 43 due to its molecular weight in thousands of daltons, forms GJA1 = connexin 43s and dominates in the decidua and trophoblast.[28] Functional enrichment analysis shows that GJA1 is mainly enriched in “positive regulation of cell activation” and “cytokine cytokine receptor interaction.” Changes in the structure and function of connexin (Cx) can regulate the permeability of GJ channels and the selectivity of permeable substances, thereby affecting material exchange and signal transmission between cells, leading to abnormal cellular function.[29] During the implantation of fertilized embryos, GJIC plays a crucial role in influencing not only the placental trophoblast, affecting its differentiation and development, but also the decidual tissue.[30] Cx43 can affect placental formation and also participate in the process of embryo implantation.[31] Researchers have found through in vitro experiments that if Cx43 has a high expression level during the differentiation of the cytotrophoblast and the formation of the syncytiotrophoblast, it can have a positive promoting effect on gap junction intercellular communication.[32] After in-depth research, researchers have found that women with recurrent early pregnancy loss have significantly lower levels of GJA1 expression compared to normal pregnant women.[33] For women with miscarriage, decidual cells obtained through curettage have significantly lower levels of GJ expression in women who terminate pregnancy but have no other abnormalities.[34] If GJA1 is knocked out in the endometrial stroma, the embryo will not continue to grow, and it can be considered that GJA1 can significantly affect the pregnancy process in the early stages of pregnancy.[35] Researchers such as He et al[36] found that recurrent pregnancy loss women have significantly lower levels of Cx43 expression. The above was in agreement with the results of the public database data analysis and the validation of the biomarkers in clinical RM samples versus control samples. We hypothesize that downregulation of this gene may also regulate or influence the development of RM by the same mechanism.

The unfolded response protein (UPR) family includes multiple proteins, one of which is the CREB3L1, which was located on chromosome 11 and contained the bZIP structural domain and the DNA-binding structural domain.[37] The UPR pathway was activated in the endoplasmic reticulum under stress.[38]

After activation of the UPR pathway in the endoplasmic reticulum, CREB3L1 formed an effector fragment that enters the nucleus and acted on the target gene by binding to the cAMP response element, promoting apoptosis and inhibiting mitosis.[39] In breast cancer, CREB3L1 was thought to be a novel metastasis suppressor gene that acted primarily by inhibiting tumor invasion and angiogenesis.[40] Existing studies have shown that this protein has a significantly higher expression level in glioma tissue, and if the malignancy of the tumor in the body is higher, the protein expression is also higher.[41] Endoplasmic reticulum stress can affect the expression of this protein, thereby regulating cellular physiological processes, including proliferation.[41] We observed higher expression of CREB3L1 in RM group compared to the HC group in GSE165004 and GSE26787 datasets. Consistent with the results of public database data analysis, the expression level of CREB3L1 is significantly excessive. We conjecture that the up-regulated expression of this gene may play an important regulatory role in the development of RM, which, of course, requires further validation and attention Pan, Z., et al demonstrated that CREB3L1 maintained the CAF (cancer-associated fibroblasts)-like property of anaplastic thyroid carcinoma cells by activating the extracellular matrix (ECM) signaling.[42] Our functional enrichment analysis found that CREB3L1 was mainly enriched in the “cytosolic ribosome” and “Cytokine-cytokine receptor interaction.” By utilizing the KEGG pathway annotation, it can be clarified that coagulation cascades, complements, etc can play a very important role in inflammatory cancer related cells. Inflammatory cancer related cells can activate CREB3L1, leading to glycolysis.[43]

Human CPT1A is a key enzyme located in the inner membrane of mitochondria, which can regulate fatty acids, regulate their oxidation process, transfer fatty acids, and allow them to enter the mitochondria. It can transfer fatty acids and allow them to enter the mitochondria.[44] CPT1A was up-regulated in many cancers, including prostate, lung, stomach, and breast cancer.[45] we observed higher expression of CPT1A in RM group compared to the HC group in GSE165004 and GSE26787 datasets. In agreement with the results of the public database data analysis, the expression of CPT1A was notably markedly over-expressed in clinical trial. We speculated that the up-regulated expression of this gene may play a different regulatory role in the development of RM, and this new specific mechanism requires further verification and attention. CPT1A was mainly enriched in the “structural constituent of ribosome” and “Cell cycle.” In order to investigate and analyze the relationship between endometrial cancer and polycystic ovary syndrome, this article uses bioinformatics to analyze the common gene features, and uses KEGG pathway to clarify the high correlation between “ribosomal subunits” and “mitochondrial translation extension.”[46]

By using IPA pathway analysis, it can be determined that multiple pathways can be related to ferroptosis oxidation, such as the “S100 family signaling pathway,” “neurovascular coupling signaling pathway” and “IL-10 signaling.” For S100 protein, it can play many roles in biological organisms, not only controlling intracellular functions, but also controlling extracellular functions, such as cell migration and proliferation processes. It can also exert regulatory effects on protein phosphorylation, energy metabolism, etc.[47] Research data shows that members of this type of protein family have significantly higher expression levels in malignant tumor cells, such as S100A9, S100A1, etc.[48] After in-depth research, some researchers have found that astrocytes can regulate neurovascular coupling at the level of small arteries.[49]

EGCG is a major active ingredient in green tea and has great research value. Numerous studies have shown that EGCG has excellent antioxidant activity[50] and good effects on gene expression signaling[51] and other cellular functions. We predict that this drug targets GJA1, which may be a breakthrough for this biomarker as a new therapeutic target, and the specific targeting mechanism needs further attention and exploration. Tofacitinib, through Janus kinase (JAK), can cut off the proliferation of T lymphocytes, and the killing of target cells, thus stopping the autoimmune attack related to T lymphocytes.[52] Tofacitib targets the JAK-transcription pathway and inhibits the release and synthesis of downstream inflammatory cytokines and mediators, resulting in therapeutic effects.[53] We predict that this drug targets PTPN6, which may be a breakthrough for this biomarker as a new therapeutic target.

5. Conclusions

This study is the first to use publicly available data from the GEO database and various bioinformatics tools to identify and analyze OS and ferroptosis-related biomarkers and their potential mechanisms of action in RM, providing reference for clinical diagnosis of RM and laying the foundation for clinical staff to develop treatment plans.

Acknowledgments

The authors are grateful for the technical support of Tianjin Medical University General Hospital and Xi’an People’s Hospital (Xi’an Fourth Hospital).

Author contributions

Conceptualization: Jinxia Xie.

Data curation: Yongqin Ma.

Formal analysis: Shaozhi Zhao.

Methodology: Hongli Zhu.

Project administration: Yingmei Wang.

Software: Hongli Zhu.

Supervision: Wenyan Tian.

Validation: Panpan Shi.

Writing – original draft: Jinxia Xie.

Writing – review & editing: Xuxin Zhan.

Abbreviations:

ceRNA competing endogenous RNAs

Cx connexin

DEGs differentially expressed genes

FOR-DEGs ferroptosis oxidation-related differentially expressed genes

GEO gene expression omnibus

GO gene ontology

GSEA gene set enrichment analysis

IPA ingenuity pathway analysis

JAK Janus kinase

KEGG Kyoto Encyclopedia of Genes and Genomes

OS oxidative stress

OSRGs oxidative stress related genes

PTP protein tyrosine phosphatases

RM recurrent miscarriage

RT-qPCR quantitative real-time reverse transcription polymerase chain reaction

UPR the unfolded response protein

Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper.

The study was conducted in accordance with the Declaration of Helsinki, and approved by the institutional review board and Ethics Committee of Xi’an People’s Hospital (Xi’an Fourth Hospital) (Ethical Approval Number: 20233027).

The authors have no funding and conflicts of interest to disclose.

All data generated or analyzed during this study are included in this published article [and its supplementary information files].

How to cite this article: Xie J, Zhu H, Zhao S, Ma Y, Shi P, Zhan X, Tian W, Wang Y. Identification and analysis of biomarkers associated with oxidative stress and ferroptosis in recurrent miscarriage. Medicine 2024;103:29(e38875).
==== Refs
References

[1] Chen Y Hu J . ATP6V1G3 acts as a key gene in recurrent spontaneous abortion: an integrated bioinformatics analysis. Med Sci Monit. 2020;26 :e927537.33028803
[2] Bhattacharya S Townend J Bhattacharya S . Recurrent miscarriage: are three miscarriages one too many? Analysis of a Scottish population-based database of 151,021 pregnancies. Eur J Obstet Gynecol Reprod Biol. 2010;150 :24–7.20207064
[3] de Jong PG Kaandorp S Di Nisio M Goddijn M Middeldorp S . Aspirin and/or heparin for women with unexplained recurrent miscarriage with or without inherited thrombophilia. Cochrane Database Syst Rev. 2014;2014 :CD004734.24995856
[4] Larsen EC Christiansen OB Kolte AM Macklon N . New insights into mechanisms behind miscarriage. BMC Med. 2013;11 :154.23803387
[5] de Moreuil C Alavi Z Pasquier E . Hydroxychloroquine may be beneficial in preeclampsia and recurrent miscarriage. Br J Clin Pharmacol. 2020;86 :39–49.31633823
[6] Forman HJ Zhang H . Targeting oxidative stress in disease: promise and limitations of antioxidant therapy. Nat Rev Drug Discov. 2021;20 :689–709.34194012
[7] Zejnullahu VA Zejnullahu VA Kosumi E . The role of oxidative stress in patients with recurrent pregnancy loss: a review. Reprod Health. 2021;18 :207.34656123
[8] Ishii T Miyazawa M Takanashi Y . Genetically induced oxidative stress in mice causes thrombocytosis, splenomegaly and placental angiodysplasia that leads to recurrent abortion. Redox Biol. 2014;2 :679–85.24936442
[9] Namli Kalem M Akgun N Kalem Z Bakirarar B Celik T . Chemokine (C-C motif) ligand-2 (CCL2) and oxidative stress markers in recurrent pregnancy loss and repeated implantation failure. J Assist Reprod Genet. 2017;34 :1501–6.28707148
[10] Jiang X Stockwell BR Conrad M . Ferroptosis: mechanisms, biology and role in disease. Nat Rev Mol Cell Biol. 2021;22 :266–82.33495651
[11] Yu Y Yan Y Niu F . Ferroptosis: a cell death connecting oxidative stress, inflammation and cardiovascular diseases. Cell Death Discov. 2021;7 :193.34312370
[12] Colaprico A Silva TC Olsen C . TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data. Nucleic Acids Res. 2016;44 :e71.26704973
[13] Yu G Wang LG Han Y He QY . clusterProfiler: an R package for comparing biological themes among gene clusters. Omics. 2012;16 :284–7.22455463
[14] Robin X Turck N Hainard A . pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinf. 2011;12 :77.
[15] Kumar L Futschik ME . Mfuzz: a software package for soft clustering of microarray data. Bioinformation. 2007;2 :5–7.18084642
[16] Shannon P Markiel A Ozier O . Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13 :2498–504.14597658
[17] Livak KJ Schmittgen TD . Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001;25 :402–8.11846609
[18] Yu M Long Y Wang Y Zhang R Tao L . Effect of levothyroxine on the pregnancy outcomes in recurrent pregnancy loss women with subclinical hypothyroidism and thyroperoxidase antibody positivity: a systematic review and meta-analysis. J Matern Fetal Neonatal Med. 2023;36 :2233039.37433649
[19] Sun F Cui L Qian J . Decidual stromal cell ferroptosis associated with abnormal iron metabolism is implicated in the pathogenesis of recurrent pregnancy loss. Int J Mol Sci . 2023;24 :7836.37175543
[20] Adhikari A Martel C Marette A Olivier M . Hepatocyte SHP-1 is a critical modulator of inflammation during endotoxemia. Sci Rep. 2017;7 :2218.28533521
[21] Kiratikanon S Chattipakorn SC Chattipakorn N Kumfu S . The regulatory effects of PTPN6 on inflammatory process: reports from mice to men. Arch Biochem Biophys. 2022;721 :109189.35307366
[22] Speir M Nowell CJ Chen AA . Ptpn6 inhibits caspase-8- and Ripk3/Mlkl-dependent inflammation. Nat Immunol. 2020;21 :54–64.31819256
[23] Liu G Zhang Y Huang Y Yuan X Cao Z Zhao Z . PTPN6-EGFR protein complex: a novel target for colon cancer metastasis. J Oncol. 2022;2022 :7391069.35186080
[24] Mok SC Kwok TT Berkowitz RS Barrett AJ Tsui FW . Overexpression of the protein tyrosine phosphatase, nonreceptor type 6 (PTPN6), in human epithelial ovarian cancer. Gynecol Oncol. 1995;57 :299–303.7774833
[25] Deng YX He WG Cai HJ . Analysis and validation of hub genes in blood monocytes of postmenopausal osteoporosis patients. Front Endocrinol (Lausanne). 2021;12 :815245.35095774
[26] Qu G Xu Y Lu Z . Prognostic signature development on the basis of macrophage phagocytosis-mediated oxidative phosphorylation in bladder cancer. Oxid Med Cell Longev. 2022;2022 :4754935.36211821
[27] Liaqat Ali Khan N Nafee T Shao T . Dysregulation in multiple transcriptomic endometrial pathways is associated with recurrent implantation failure and recurrent early pregnancy loss. Int J Mol Sci . 2022;23 :16051.36555686
[28] Nevin RL . Mefloquine gap junction blockade and risk of pregnancy loss. Biol Reprod. 2012;87 :65.22837476
[29] Totland MZ Rasmussen NL Knudsen LM Leithe E . Regulation of gap junction intercellular communication by connexin ubiquitination: physiological and pathophysiological implications. Cell Mol Life Sci. 2020;77 :573–91.31501970
[30] Dunk CE Gellhaus A Drewlo S . The molecular role of connexin 43 in human trophoblast cell fusion. Biol Reprod. 2012;86 :115.22238282
[31] Laird DW . Life cycle of connexins in health and disease. Biochem J. 2006;394 (Pt 3 ):527–43.16492141
[32] Cronier L Defamie N Dupays L . Connexin expression and gap junctional intercellular communication in human first trimester trophoblast. Mol Hum Reprod. 2002;8 :1005–13.12397213
[33] Nair RR Jain M Singh K . Reduced expression of gap junction gene connexin 43 in recurrent early pregnancy loss patients. Placenta. 2011;32 :619–21.21669459
[34] Kara F Cinar O Erdemli-Atabenli E Tavil-Sabuncuoglu B Can A . Ultrastructural alterations in human decidua in miscarriages compared to normal pregnancy decidua. Acta Obstet Gynecol Scand. 2007;86 :1079–86.17712648
[35] Laws MJ Taylor RN Sidell N . Gap junction communication between uterine stromal cells plays a critical role in pregnancy-associated neovascularization and embryo survival. Development. 2008;135 :2659–68.18599509
[36] He X Chen Q . Reduced expressions of connexin 43 and VEGF in the first-trimester tissues from women with recurrent pregnancy loss. Reprod Biol Endocrinol. 2016;14 :46.27535546
[37] Koumenis C Wouters BG . “Translating” tumor hypoxia: unfolded protein response (UPR)-dependent and UPR-independent pathways. Mol Cancer Res. 2006;4 :423–36.16849518
[38] Saito A Kanemoto S Kawasaki N . Unfolded protein response, activated by OASIS family transcription factors, promotes astrocyte differentiation. Nat Commun. 2012;3 :967.22828627
[39] Denard B Lee C Ye J . Doxorubicin blocks proliferation of cancer cells through proteolytic activation of CREB3L1. Elife. 2012;1 :e00090.23256041
[40] Ward AK Mellor P Smith SE . Epigenetic silencing of CREB3L1 by DNA methylation is associated with high-grade metastatic breast cancers with poor prognosis and is prevalent in triple negative breast cancers. Breast Cancer Res. 2016;18 :12.26810754
[41] Yan Z Hu Y Zhang Y Pu Q Chu L Liu J . Effects of endoplasmic reticulum stress-mediated CREB3L1 on apoptosis of glioma cells. Mol Clin Oncol. 2022;16 :83.35251634
[42] Pan Z Xu T Bao L . CREB3L1 promotes tumor growth and metastasis of anaplastic thyroid carcinoma by remodeling the tumor microenvironment. Mol Cancer. 2022;21 :190.36192735
[43] Hu B Wu C Mao H . Subpopulations of cancer-associated fibroblasts link the prognosis and metabolic features of pancreatic ductal adenocarcinoma. Ann Transl Med. 2022;10 :262.35402584
[44] Bonnefont JP Djouadi F Prip-Buus C Gobin S Munnich A Bastin J . Carnitine palmitoyltransferases 1 and 2: biochemical, molecular and medical aspects. Mol Aspects Med. 2004;25 :495–520.15363638
[45] Melone MAB Valentino A Margarucci S Galderisi U Giordano A Peluso G . The carnitine system and cancer metabolic plasticity. Cell Death Dis. 2018;9 :228.29445084
[46] Miao C Chen Y Fang X Zhao Y Wang R Zhang Q . Identification of the shared gene signatures and pathways between polycystic ovary syndrome and endometrial cancer: an omics data based combined approach. PLoS One. 2022;17 :e0271380.35830453
[47] Donato R Cannon BR Sorci G . Functions of S100 proteins. Curr Mol Med. 2013;13 :24–57.22834835
[48] McKiernan E McDermott EW Evoy D Crown J Duffy MJ . The role of S100 genes in breast cancer progression. Tumour Biol. 2011;32 :441–50.21153724
[49] Mapelli L Gagliano G Soda T Laforenza U Moccia F D’Angelo EU . Granular layer neurons control cerebellar neurovascular coupling through an NMDA receptor/NO-dependent system. J Neurosci. 2017;37 :1340–51.28039371
[50] Alam M Ali S Ashraf GM Bilgrami AL Yadav DK Hassan MI . Epigallocatechin 3-gallate: from green tea to cancer therapeutics. Food Chem. 2022;379 :132135.35063850
[51] Kim YJ Kim KS Lim D . Epigallocatechin-3-gallate (EGCG)-inducible SMILE inhibits STAT3-mediated hepcidin gene expression. Antioxidants (Basel). 2020;9 :514.32545266
[52] Palmroth M Kuuliala K Peltomaa R . Tofacitinib suppresses several JAK-STAT pathways in rheumatoid arthritis in vivo and baseline signaling profile associates with treatment response. Front Immunol. 2021;12 :738481.34630419
[53] Malemud CJ . The role of the JAK/STAT signal pathway in rheumatoid arthritis. Ther Adv Musculoskelet Dis. 2018;10 :117–27.29942363
