This article is mentioned in:

Introduction

Ovarian cancer (OC) is a fatal malignancy of the female reproductive system (1). In 2018, there were 22,240 new cases of ovarian malignancies, and 14,070 OC-associated mortality cases (2). Furthermore, the incidence and mortality rates of malignant ovarian tumors in China reached 52.1/100,000 and 22.5/100,000, respectively, in 2015 (3). Due to the non-specific symptoms of OC, 70% of Chinese patients are diagnosed with a later stage of the disease at first diagnosis (4).

Previous studies demonstrated that ~38% of patients with cancer display major emotional distress, including anxiety and depression (5–7). In patients with OC, the incidence of depression and anxiety is of 82 and 92%, respectively (8). In addition, the degree of depression or anxiety is highly variable in patients with OC (9).

At present, the majority of studies have focused on how depression affects the prognosis of OC (10–12); however, the pathogenesis of depression in OC remains poorly investigated, particularly at a molecular level. By using a miRNA-mRNA regulation network, Wu et al (13) identified 12 miRNA-mRNAs pairs (miR-629-5p-FGF1, miR-629-5p-AKT3, miR-629-5p-MAGI2, miR-933-BDNF, miR-933-MEF2A, miR-23b-3p-TJP1, miR-23b-3p-JMJD1, miR-23b-3p-APAF1, miR-23b-3p-CAB39, miR-1265-CDKN1B, miR-33b-3p-CDKN1B and miR-33b-3p-F2R) that could be associated with the development of major depressive disorder (MDD) in patients with OC. Furthermore, Rahman et al (14) identified 34 differentially expressed genes (DEGs) associated with depression in patients with OC and demonstrated that CXCL12, ARL4C, NQO2 were associated with worse survival in patients with OC. As depressed patients display higher mortality rates (15) and since a different mental status can lead to different clinical outcomes (16) and prognosis (17), it is crucial to further understand the molecular mechanisms underpinning the onset of depression in patients with OC.

The present study aimed to identify DEGs in patients with OC and high or low-grade depression using bioinformatics analyses, and to determine the potential hub genes and pathways that may serve critical roles in the onset of depression in patients with OC.

Materials and methods

Study design

The flowchart for this study is presented in Fig. 1. DEGs were first identified in GSE9116 dataset (18,19) according to the following criteria: P<0.05 and absolute log2 value of fold change >1 (|log2(FC)|>1). The top three up- and downregulated DEGs were collected as potential key genes for further analysis. Subsequently, Gene Ontology (GO) (https://www.webgestalt.org/) and pathways enrichment analyses (https://toppgene.cchmc.org/) were performed using WebGestalt (20) and ToppGene (21) tools separately, and the protein-protein interaction (PPI) and pathway crosstalk networks were constructed. The gene-pathway network was constructed by mapping the hub genes into hub pathways extracted from PPI and pathway crosstalk network according to the criterion of nodal degree > average. Subsequently, the core genes with pathways were extracted from the gene-pathway network according to the criterion node degree > average. Eventually, the core genes and the top three up- and downregulated DEGs, which were defined as key genes, were studied by expression and survival analysis.

Figure 1. Workflow for the identification of core genes promoting depression in patients with ovarian cancer. FC, fold change; DEGs, differentially expressed genes.

Microarray dataset

The GSE9116 dataset describing the gene expression profile of OC (18,19) and established on the platform of GPL96 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GPL96), was downloaded from Gene Expression Omnibus (GEO) database (22). To the best of our knowledge, this dataset is the only one that studied the impact of depression on gene expression in patients with OC. This dataset contains ten samples of primary OC (grade 3 serous cancer or serous adenocarcinoma), including five samples from patients with low-grade depression (LD) and five samples from patients with high-grade depression (HD). The samples and psychosocial data were collected from patients who were diagnosed with ovarian epithelial cancer, peritoneal cancer, or cancer of the fallopian tube, and who underwent primary surgical resection of ovarian carcinoma (14). A patient with a Center for Epidemiologic Studies Depression Scale score (23) ≥16 and a Social Provisions Scale-Attachment score (24) ≤15 was defined as having high-grade depression. Otherwise, patients were defined as having low-grade depression (18).

Identification of DEGs

The GSE9116 dataset was divided into two groups: HD and LD groups. Genes were considered differentially expressed between HD and LD groups if they exhibited a |log2(FC)|>1 and P<0.05 calculated using GEO2R (16) with the limma package (25). The top three up- and downregulated DEGs were considered as potential key genes for further analysis.

GO enrichment analysis

GO analysis was performed using WebGestalt tool (20) on the DEGs. Enriched biological terms for cellular components (CC), biological process (BP) and molecular functions (MF) with a P<0.05 were identified.

PPI network and identification of hub genes

The Search Tool for the Retrieval of Interacting Genes (STRING) (26) database and Cytoscape (version 3.5.1) (27) were used to identify hub genes. The PPI network was searched by gene symbols and the minimum required interaction score was set at 0.7 to ensure high confidence in the results. Nodes that were not connected to the major network were removed to reduce the error detection rate. The CentiScaPe plug-in (28) was used to calculate the degree of each node. Nodes were considered as hub if their degree was larger than the average.

Pathway enrichment and crosstalk analysis

For pathway enrichment, DEGs were mapped using the Kyoto Gene and Genome Encyclopedia (KEGG) database (29) and BioCarta (30) using the online analysis tool (21). Pathways with a P<0.05 were considered significant.

Pathway crosstalk analysis was conducted on the pathways defined above based on the assumption that two pathways are considered crosstalking if they share a proportion of genes (31,32). Briefly, pathways containing <3 genes were excluded as pathways with too few genes may contain insufficient biological information. Two indicators named Jaccard Coefficient (33) and Overlapping Coefficient (34), were used to measure the overlap between two pathways and were calculated as follows: Jaccard Coefficient = |(A∩B)/(A∪B) and Overlapping Coefficient = [(|A∩B|)/(min(|A|,|B|)], where A and B represent the list of genes in the two pathways, and min indicated the minimum number of genes between |A| and |B|. Subsequently, pairs of pathways with more than one overlapping gene were retained. Once the pathway crosstalking network was obtained, a subnetwork representing the hub pathways was identified according to the criterion nodal degree > average.

Comprehensive gene-pathway analysis

By mapping hub genes into the subnetwork of pathway crosstalk using KEGG (29) and BioCarta (30), a comprehensive gene-pathway network was obtained to further investigate the association between genes and pathways. The core genes and pathways in the gene-pathway network were identified according to the criterion nodal degree > average.

Expression and survival analysis of key genes

The expression of core genes in the sub gene-pathway network and of the top three up and downregulated DEGs were extracted from the GSE9116 dataset and evaluated for regulations. Furthermore, to analyze the effect of the key genes on patients' overall survival (OS), progression-free survival (PFS) and post-progression survival (PPS), patients were stratified according to the expression of the key genes and evaluated using Kaplan-Meier analysis (35). Kaplan-Meier analysis provides a survival assessment of prognosis-related genes in OC patients whose data obtained from The Cancer Genome Atlas (36) and 14 GSE profiles [GSE14764 (37), GSE15622 (38), GSE18520 (39), GSE19829 (40), GSE23554 (41), GSE26193 (42,43), GSE26712 (44,45), GSE27651 (46), GSE30161 (47), GSE3149 (48), GSE51373 (49), GSE63885 (50), GSE65986 (51) and GSE9891 (52)]. Patients were grouped by median gene expression and hazard ratio (HR) with 95% confidence intervals (CI) and log-rank P-values were calculated. P<0.05 was considered to indicate a significant difference.

Results

Identification of DEGs

According to the criteria P<0.05 and |Log2(FC)|>1, 93 DEGs were identified when comparing the HD and LD groups. A total of 84 DEGs were upregulated and nine were downregulated. Plasminogen activator, tissue type (PLAT), activating transcription factor 3 (ATF3) and cellular communication network factor 2 (CTGF) were the top three upregulated genes with the highest change in expression. Heat shock 70 kDa protein 1B (HSPA1B), endonuclease G (ENDOG) and EPS8 like 1 (EPS8L1) were the top three downregulated genes. A heatmap with logFC values of the 93 DEGs is presented in Fig. 2.

Figure 2. Heatmap of all differentially expressed genes from the GSE9116 dataset. The groups indicated at the top of the figure in blue and red represent groups with low-grade depression and high-grade depression, respectively. The numbers in the expression key represent the log2 value of reads for each gene.

GO enrichment analysis

GO enrichment analysis was performed on the 93 DEGs using WebGestalt (Fig. 3). In terms of BP, DEGs were primarily involved in ‘biological regulation’, ‘metabolic process’, ‘multicellular organismal process’ and ‘response to stimulus’. In terms of CC, DEGs were enriched in ‘membrane’, ‘nucleus’, ‘endomembrane system’ and ‘vesicle’. In terms of MF, DEGs mainly participated in ‘protein binding’, ‘ion binding’ and ‘hydrolase activity’.

Figure 3. Results of GO enrichment analysis for ‘Biological Process’, ‘Cellular Component’ and ‘Molecular Function’ categories. The number on each bar indicates the number of enriched genes annotated with the corresponding GO term. Go, Gene Ontology.

Construction of the PPI network and identification of hub genes

Using DEGs and the STRING database, a PPI network containing 34 nodes and 55 edges was obtained (Fig. 4A). The degree of each node was calculated using CentiScaPe (Table I). A total of 16 hub genes of which degree > average were extracted from the PPI network (Fig. 4B). This subnetwork was divided into two clusters that contained eight and seven genes, respectively. These two clusters were connected by C-C motif chemokine ligand 2 (CCL2) (Fig 4B).

Figure 4. PPI network, pathway crosstalk and gene-pathway analyses of the DEGs from the GSE9116 dataset. (A) PPI network of candidate genes. Red, core genes. Orange, top three up-regulated genes. Blue, other DEGs. (B) Subnetwork of PPI network for hub genes of which nodal degree > average. Red, core genes. Orange, top three up-regulated genes. Blue, other DEGs. (C) Pathway crosstalk analysis of DEGs. Edge thickness represents the average value of Jaccard and Overlapping Coefficient. (D) Subnetwork of pathway crosstalk extracted by the criterion nodal degree > average. (E) Comprehensive gene-pathway network established by mapping the hub genes to the subnetwork. The arrow direction between gene and pathway was determined from Kyoto Gene and Genome Encyclopedia database and BioCarta. Red circle, core genes. Blue circle, other DEGs. Green square, pathways. (F) Subnetwork of gene-pathway collected according to the criterion nodal degree > average; Red circle, core genes. Green square, pathways. ATF3, activating transcription factor 3; CCL2, C-C motif chemokine ligand 2; CTGF, cellular communication network factor 2; DEGs, differentially expressed genes; FOS, Fos proto-oncogene, AP-1 transcription factor subunit; PLAT, plasminogen activator, tissue type; PPI, protein-protein interaction; SERPINE1, serpin family E member 1; SERPING1, serpin family G member 1.

Table I. Characteristics of genes identified from protein-protein interaction network.

Table I.

Characteristics of genes identified from protein-protein interaction network.

Rank	Gene	Degree of each node
Overall average value		3.2352
1	FOS	9
2	JUNB	7
3	DUSP1	6
4	EGR1	6
5	FOSB	6
6	SERPINE1	6
7	SPARC	5
8	TIMP3	5
9	JUND	5
10	SEPP1	4
11	SERPING1	4
12	ZFP36	4
13	ATF3	4
14	COL3A1	4
15	DCN	4
16	CCL2	4
17	LUM	3
18	CTGF	2
19	CXCL12	2
20	CTSB	2
21	HLA-DPA1	2
22	CHPT1	2
23	PLA2G7	2
24	PLA2G16	2
25	SMARCA4	1
26	CST3	1
27	KLC1	1
28	PNOC	1
29	TMEM176A	1
30	TMEM176B	1
31	PRELP	1
32	DUSP6	1
33	PLOD2	1
34	PLAT	1

Pathway enrichment and crosstalk analysis

Pathway enrichment analysis was performed using ToppGene database. The results indicated that the top five significantly enriched pathways were ‘Ensemble of genes encoding extracellular matrix and extracellular matrix-associated proteins’, ‘Ensemble of genes encoding core extracellular matrix including ECM glycoproteins, collagens and proteoglycans’, ‘Complement and coagulation cascades’, ‘Pertussis toxin-insensitive CCR5 Signaling in Macrophage’ and ‘Rheumatoid arthritis’ (Table II).

Table II. Results from pathway enrichment analysis.

Table II.

Results from pathway enrichment analysis.

Pathways	Source	P-value	Genes in the pathway
Ensemble of genes encoding extracellular matrix and extracellular matrix-associated proteins	BIOCARTA	1.19×10−08	COL3A1, CRISPLD2, FBLN5, SPARC, MXRA5, DCN, EGFL6, TIMP3, PLXNC1, PRELP, SERPINE1, XCL2, CST3, SERPING1, CCL2, PLAT, CTGF, LUM, CTSB, XCL1, PLOD2, CXCL12
Ensemble of genes encoding core extracellular matrix including ECM glycoproteins, collagens and proteoglycans	BIOCARTA	1.73×10−05	COL3A1, CRISPLD2, FBLN5, SPARC, MXRA5, DCN, PRELP, CTGF, LUM
Complement and coagulation cascades	KEGG	6.90×10−05	SERPINE1, SERPING1, PLAT, C4A, C4B
Pertussis toxin-insensitive CCR5 signaling in Macrophage	BIOCARTA	1.20×10−04	FOS, CCL2, CXCL12
Rheumatoid arthritis	KEGG	1.29×10−04	ATP6V1B1, HLA-DPA1, FOS, CCL2, CXCL12
Ensemble of genes encoding ECM-associated proteins including ECM-affiliated proteins, ECM regulators and secreted factors	BIOCARTA	1.85×10−04	EGFL6, TIMP3, PLXNC1, SERPINE1, XCL2, CST3, SERPING1, CCL2, PLAT, CTSB, XCL1, PLOD2, CXCL12
Genes encoding enzymes and their regulators involved in the remodeling of the extracellular matrix	BIOCARTA	3.07×10−04	TIMP3, SERPINE1, CST3, SERPING1, PLAT, CTSB, PLOD2
MAPK signaling pathway	KEGG	4.66×10−04	FOS, DUSP1, DUSP6, JUND, FGFR3, HSPA1A, HSPA1B
Pertussis	KEGG	7.77×10−04	FOS, SERPING1, C4A, C4B
Antigen processing and presentation	KEGG	8.16×10−04	HLA-DPA1, CTSB, HSPA1A, HSPA1B
Genes encoding proteoglycans	BIOCARTA	9.00×10−04	DCN, PRELP, LUM
IL-17 signaling pathway	KEGG	1.65×10−03	FOS, FOSB, JUND, CCL2
Lectin Induced Complement Pathway	BIOCARTA	1.87×10−03	C4A, C4B
Fibrinolysis Pathway	BIOCARTA	1.87×10−03	SERPINE1, PLAT
Ether lipid metabolism	KEGG	1.88×10−03	PLA2G7, PLA2G16, CHPT1
Estrogen signaling pathway	KEGG	2.00×10−03	FOS, HSPA1A, HSPA1B, GABBR1
AGE-RAGE signaling pathway in diabetic complications	KEGG	2.08×10−03	COL3A1, EGR1, SERPINE1, CCL2
Classical Complement Pathway	BIOCARTA	2.56×10−03	C4A, C4B
Platelet Amyloid Precursor Protein Pathway	BIOCARTA	2.56×10−03	SERPINE1, PLAT
B Cell Survival Pathway	BIOCARTA	3.36×10−03	FOS, JUND
Staphylococcus aureus infection	KEGG	3.51×10−03	HLA-DPA1, C4A, C4B
Genes encoding structural ECM glycoproteins	BIOCARTA	4.29×10−03	CRISPLD2, FBLN5, SPARC, MXRA5, CTGF
Complement Pathway	BIOCARTA	4.73×10−03	C4A, C4B
Osteoclast differentiation	KEGG	5.52×10−03	FOS, FOSB, JUNB, JUND

For the pathway crosstalk analysis, 17 out of 24 pathways that contained ≥2 genes met the crosstalk analysis criteria and were selected to construct the network (Fig. 4C). The thickness of the edges indicated the average values of Jaccard and the Overlapping Coefficient that represented the overlapping level of genes between two pathways. By selecting the nodes with degree > average, a subnetwork of pathway crosstalk with 7 nodes and 11 edges was constructed (Fig. 4D).

Comprehensive gene-pathway analysis

After mapping the hub genes onto the subnetwork of pathways provided by KEGG and BioCarta, a potential gene-pathway network including 7 essential pathways and 11 hub genes was constructed (Fig. 4E). The results from this network demonstrated that CCL2 and Fos proto-oncogene, AP-1 transcription factor subunit (FOS) participated in most of the pathways. Based on the number of genes involved, the top three pathways were ‘Ensemble of genes encoding extracellular matrix and extracellular matrix-associated proteins’, ‘Ensemble of genes encoding ECM-associated proteins including ECM-affiliated proteins, ECM regulators and secreted factors’ and ‘IL-17 signaling pathway’ (Fig. 4E).

To identify the main nodes (including genes and pathways) in the gene-pathway network, nodes with degree > average were selected (Fig. 4F). The results demonstrated that CCL2, FOS, serpin family G member 1 (SERPING1) and serpin family E member 1 (SERPINE1) and the four pathways ‘Ensemble of genes encoding ECM-associated proteins including ECM-affiliated proteins, ECM regulators and secreted factors’, ‘Ensemble of genes encoding extracellular matrix and extracellular matrix-associated proteins’, ‘MAPK signaling pathway’ and ‘IL-17 signaling pathway’ were identified and may serve a crucial role in the development of depression in patients with OC.

Association between key genes and the six top DEGs

The evaluation of the interaction between the four key genes (CCL2, FOS, SERPING1 and SERPINE1) and the six top DEGs (upregulated; PLAT, ATF3 and CTGF and downregulated; HSPA1B, ENDOG and EPS8L1) in the PPI (Fig 4A) demonstrated that HSPA1B, ENDOG, EPS8L1 were not found since they were not connected to the major network. Furthermore, PLAT was interplayed with SERPINE1, CTGF interacted with CCL2 and ATF3 was connected to FOS.

Expression and survival analysis of key genes

As presented in Fig. 5, all core genes and top three upregulated DEGs were significantly upregulated in the HD group, which was not the case for the top three downregulated DEGs compared with the LD group.

Figure 5. Expression of the core genes CCL2, FOS, SERPINE1 and SERPING1 and the top three up and downregulated genes PLAT, ATF3, CTGF, HSPA1B, ENDOG and EPS8L1 obtained from GSE9116. All core genes were upregulated in the high-grade depression group. ATF3, activating transcription factor 3; CCL2, C-C motif chemokine ligand 2; CTGF, cellular communication network factor 2; DEGs, differentially expressed genes; ENDOG, endonuclease G; EPS8L1, EPS8 like 1; FOS, Fos proto-oncogene, AP-1 transcription factor subunit; HSPA1B, heat shock protein family A (Hsp70) member 1B; PLAT, plasminogen activator, tissue type; PPI, protein-protein interaction; SERPINE1, serpin family E member 1; SERPING1, serpin family G member 1.

The results from survival analysis indicated that the top 3 upregulated DEGs had similar effects as the core genes, which was not the case for the top 3 downregulated DEGs (Fig. 6). In particular, higher expression of ATF3 (HR, 1.19; 95% CI, 1.03–1.37; P=0.019) and SERPINE1 (HR, 1.27; 95% CI, 1.11–1.46; P=0.00056) and lower expression of ENDOG (HR, 0.83; 95% CI, 0.73–0.95; P=0.0061) and EPS8L1 (HR, 0.82; 95% CI, 0.72–0.94; P=0.004) were significantly associated with worse OS. Furthermore, ATF3 (HR, 1.19; 95% CI, 1.03–1.36; P=0.016), CTGF (HR, 1.28; 95% CI, 1.12–1.47; P=0.00033), HSPA1B (HR, 1.15; 95% CI, 1.01–1.52; P=0.04), EPS8L1 (HR, 1.17; 95% CI, 1.12–1.34; P=0.0021) and all core genes, including CCL2 (HR, 1.24; 95% CI, 1.09–1.41; P=0.0013), FOS (HR: 1.21, 95%CI: 1.06–1.37; P=0.0048), SERPINE1 (HR, 1.38, 95% CI, 1.21–1.57; P=1.1e-06) and SERPING1 (HR, 1.22; 95% CI, 1.05–1.41; P=0.0084), were significantly associated with a lower PFS. However, lower expression of HSPA1B (HR, 0.67; 95% CI, 0.56–0.8; P=0.000016), CCL2 (HR, 0.83; 95% CI, 0.7–0.99; P=0.038) and SERPING1 (HR, 0.8; 95% CI, 0.67–0.97; P=0.023) and high expression of SERPINE1 (HR, 1.42; 95% CI, 1.18–1.72; P=0.00026) were significantly associated with worse PPS. The other genes had no significant effect on OS, PFS or PPS.

Figure 6. Survival analysis of the core genes; CCL2, FOS, SERPINE1 and SERPING1, and the top three up/downregulated genes; PLAT, ATF3, CTGF, HSPA1B, ENDOG and EPS8L1. Vertical dotted line represents no effect for the pooled relative risk estimation. If the confidence intervals for individual gene overlap with this line, it means that the survival between low and high expression is not significant. ATF3, activating transcription factor 3; CCL2, C-C motif chemokine ligand 2; CTGF, cellular communication network factor 2; DEGs, differentially expressed genes; ENDOG, endonuclease G; EPS8L1, EPS8 like 1; FOS, Fos proto-oncogene, AP-1 transcription factor subunit; HSPA1B, heat shock protein family A (Hsp70) member 1B; OS, overall survival; PFS, progression-free survival; PLAT, plasminogen activator, tissue type; PPI, protein-protein interaction; PPS, post-progression survival; SERPINE1, serpin family E member 1; SERPING1, serpin family G member 1.

Discussion

The incidence of cancer continues to rise annually (53). With the advancement of psychological research, increasing attention is being paid to the mental health of patients with malignant tumors (54). Depression is a common psychological issue observed in cancer patients. It can reduce the efficacy of treatments, delay the recovery time and reduce the quality of life of patients (55). At similar stages of cancer cell differentiation or treatments, patients with depression may have a worse prognosis compared with non-depressed patients (56). It has been reported that patients with OC whom exhibit a positive attitude have an improved quality of life and prognosis compared with patients with negative emotions (57). It is therefore crucial to determine the underlying molecular mechanisms linking depression with poorer prognosis for patients with ovarian cancer.

Whereas the original study of GSE9116 used a promoter-based bioinformatics strategy to investigate the effect of β-adrenergic signaling (18) and focused on the genetic interaction locus in the human interleukin (IL)-6 promoter (single nucleotide polymorphisms rs1800795) (19), the present study aimed to identify key genes and pathways associated with depression in patients with OC using PPI networks, pathway crosstalk and gene-pathway analyses. The results from this study may provide an improved understanding underlying the link between depression in and a poorer prognosis in patients with OC.

The results from the present study demonstrated that CCL2 may induce depression in patients with OC through ECM-related and IL-17 signaling pathways. Previous studies indicated that high expression of CCL2 is associated with depression. Leighton et al (58) performed a meta-analysis, which included 4688 subjects (765 depressed healthy patients vs. 1528 not depressed healthy patients, and 742 depressed patients with inflammatory illness vs. 1653 not depressed patients with inflammatory illness) and reported that CCL2 blood levels were significantly higher in depressed patients with or without inflammatory illness compared with the control. Another meta-analysis reported similar results for CCL2 (59). In addition, CCL2 was upregulated in cancerous cells, and this required the participation of IL-17 (60), and CCL2 induced tumor cell proliferation and stimulated tumor cell migration and invasion into the surrounding extracellular matrix in order to promote tumor cell intravasation into the circulation (61).

The present study demonstrated an association between FOS expression and depression via MAPK and IL-17 signaling pathways in patients with OC. Yu et al (62) reported that human depression is associated with cognitive deficits, and that FOS is upregulated in cognitively impaired mice. Kung et al (63) demonstrated that FOS expression is higher in preproenkephalin-knockout mice, and induced anxiety and depression-like symptoms of post-traumatic stress disorder compared with wild-type mice. In addition, it was reported that FOS expression is regulated by the MAPK pathway (64,65). Furthermore, IL-17 can mediate inflammatory responses via FOS activation (66) and enhance the recruitment of activated FOS in synergy with IL-6 (67).

The results from the present study indicated that SERPINE1 and SERPING1 could induce depression in patients with OC through ECM-related pathways. Both SERPINE1 and SERPING1 belong to the SERPINs family of serine protease inhibitors that regulate proteases involved in fibrinolysis, coagulation, inflammation, cell mobility, cellular differentiation and apoptosis (68). SERPINE1, which is known as plasminogen activator inhibitor-1 (PAI-1), is described as a major physiological inhibitor of endogenous plasminogen activator which inhibits fibrin degradation, promotes fibrin deposition on blood vessel walls and stimulates smooth muscle cell proliferation (69). Furthermore, SERPINE1 may be involved in the pathogenic and therapeutic mechanisms of MDD. Yamamoto et al (70) reported that stress induced SERPINE1 upregulation in a tissue- and cell type-specific manner. In addition, previous studies demonstrated that both women (71) and men (72) with MDD present higher blood PAI-1 levels compared with healthy subjects. Although these findings were not described in patients with OC, they are similar to the results from the present study. Previous studies indicated that SERPINE1 is a stress-associated gene and that its genetic variants may contribute to the causes of depression and the acute therapeutic response to selective serotonin reuptake inhibitors in MDD (73). However, the promoter polymorphisms of SERPINE1 gene were not associated with Alzheimer's disease-related depression, but they may be associated with the response to antidepressant treatments (74). Therefore, the mechanism by which SERPINE1 induces depression in patients with OC requires further investigation. SERPING1 is also referred to as a complement system C1 inhibitory gene [complement component 1 inhibitor (C1INH)], which is located on chromosome 11q.11-q13.1 on GeneMap (75). SERPING1 has been reported to be associated with plasma protein supplementation and to be a member of the serine protease inhibitor gene family (76). Numerous studies have reported that SERPING1 is downregulated in patients with hereditary angioedema and depression (77–79). Furthermore, Ditzen et al (80) demonstrated that SERPING1 expression levels were decreased in the cerebrospinal fluid of depressed patients compared with healthy controls. Since the results from the present study demonstrated that patients with OC and depression exhibited higher SERPING1 expression, SERPING1 may cause depression through other pathways in cancer microenvironment. In addition, SERPINE1 and SERPING1 are associated with extracellular matrix regulation (68). The plasminogen activator/plasmin system serves a crucial role in ECM degradation, and SERPINE1 is a physiological inhibitor of plasminogen activators (81). Furthermore, SERPINE1 upregulation promotes the deposition of ECM components (82), and SERPING1 interacts with extracellular matrix components to inhibit protease activity (83). In addition, interactions between SERPING1 and extracellular matrix components may result in an increase in C1INH concentration at inflammation sites (84). However, as SERPINE1/SERPING1 interaction with ECM remains unclear, further investigation is required.

ATF3 and CTGF upregulation have been demonstrated to be associated with depression. Green et al (85) reported that ATF3 upregulation in nucleus accumbens decreases emotional reactivity and increases depression-like behavior. Similarly, Turner et al (86) demonstrated that CTGF expression in human amygdala is significantly increased in patients with major depressive disorder compared with healthy subjects. In addition, CTGF administration increases depression-like behavior in outbred rats (86). Only few studies have focused on PLAT association with depression. The results from the present study demonstrated that PLAT, ATF3 and CTGF were co-expressed with SERPINE1, FOS and CCL2. PLAT, ATF3 and CTGF may therefore induce depression through SERPINE1, FOS and CCL2 in patients with OC. However, the underlying mechanisms remain unknown and require further investigation.

Previous studies reported that depression is detrimental to cancer patients' survival (87,88). The results from the present study revealed that upregulation of CCL2, FOS, SERPINE1 and SERPING1 was associated with worse survival. However, Wojnarowicz et al (89) demonstrated that CCL2 protein expression is not correlated with overall or disease-free survival. Conversely, Mahner et al (90) reported that reduced FOS expression is associated with unfavorable PFS and OS in patients with epithelial ovarian carcinoma. However, it has been found that elevated tumor SERPINE1 levels are associated with a poor prognosis and reduced disease-free survival in patients with ovarian carcinoma (91,92). To the best of our knowledge, no study has investigated the association between SERPING1 and survival in patients with OC. However, Peng et al (93) demonstrated that lower SERPING1 mRNA levels predicted worse OS and disease-free survival in patients with prostate cancer compared with healthy controls; however, Mejia et al (94) reported that OS is significantly improved in a mice model of malaria treated with SERPING1 compared with controls. Since it is not clear whether these studies included patients with depression or whether the core genes serve similar functions in different types of cancer, determining how CCL2, FOS, SERPINE1 and SERPING1 could influence patients' survival requires further investigation.

In conclusion, the present study demonstrated that the four core genes CCL2, FOS, SERPINE1 and SERPING1 were upregulated in the HD group, which suggested that they may promote depression and worsen survival in patients with OC through four pathways (‘Ensemble of genes encoding ECM-associated proteins including ECM-affiliated proteins’, ‘ECM regulators and secreted factors’, ‘Ensemble of genes encoding extracellular matrix and extracellular matrix-associated proteins’ and ‘MAPK signaling pathway and IL-17 signaling pathway’). These findings may provide novel markers and methods for predicting and treating depression in patients with OC; however, the determination of specific mechanisms requires further investigated.

Acknowledgements

Not applicable.

Funding

No funding was received.

Availability of data and materials

All data generated or analyzed during this study are included in the published article.

Authors' contributions

YY, YL and WZ designed the study and drafted the manuscript. WW and KW performed the literature search, data extraction and statistical analysis. All authors contributed to the interpretation of the data and the redaction of the manuscript. The final version of the manuscript was reviewed and approved by all authors.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References