Screening of Critical Genes Involved in Metastasis and Prognosis of High-Grade Serous Ovarian Cancer by Gene Expression Profile Data

Abstract

We aimed to identify critical genes for metastasis and prognosis of high-grade serous ovarian cancer (HGSOC). The mRNA data of GSE73064 including 15 HGSOC samples (5 primary tumor cells, 5 ascites tumor cells, and 5 metastasis tumor cells) were applied for the present analysis. The data were first preprocessed using robust multiarray average (RMA) method, and time-series gene expression analyses were performed using short time-series expression miner (version 1.3.11). Then, functional analysis for clustered genes was conducted, and protein–protein interaction (PPI) network was constructed. Subnetwork module was also analyzed. After that, we analyzed the miRNA regulatory network. Finally, survival analysis was performed. In this study, PPBP, ITGB2, and ITGAM were genes with higher degrees in the PPI network. miR-4469, miR-6734-5p, and miR-1304-5p were several of important miRNAs in the miRNA-target regulatory network. In addition, survival analysis showed that nine genes such as CLEC5A, FPR1, and CXCL12 might be important for HGSOC prognosis. PPBP, ITGB2, and ITGAM may be critical for HGSOC metastasis. CLEC5A, FPR1, and CXCL12 may be important for the prognosis of HGSOC. Furthermore, miR-4469, miR-6734-5p, and miR-1304-5p may be important miRNAs associated with HGSOC.

1. Introduction

High-grade serous ovarian cancer (HGSOC), the most aggressive type of ovarian cancer (OC), accounts for most death resulted from gynecological cancers (Reiner et al., 2017). Worldwide, OC is the seventh most common cancers in females, and more than 239,000 patients are diagnosed with OC each year (Ferlay et al., 2015). Although initial treatment for OC has higher response rate, most patients develop chemotherapy resistance eventually (Coleman et al., 2013). In the past few decades, there has been no significant change in the overall survival of HGSOC (Bowtell et al., 2015). Therefore, it is significant to identify effective treatment for HGSOC to improve the overall survival rate of it.

Understanding of the important genes or miRNAs for the development of HGSOC may provide clues for the finding of effective therapy. Some previous studies have indicated that some genes or miRNAs are significantly associated with HGSOC. For example, Zaid et al. (2013) indicated that fibroblast growth factor receptor 4 (FGFR4) could be regarded as a prognostic marker for HGSOC. Kannan et al. (2014) showed that CDKN2D-WDFY2 might be a clinical biomarker for HGSOC detection. Cesaratto et al. (2016) suggested that BNC2 might be a suppressor gene in HGSOC. FRS2 was identified as an oncogene in HGSOC by the study of Luo et al. (2015). In addition, miR-29a played tumor-suppressive roles in OC (Creighton et al., 2012). miR-34c might be a tumor suppressor in HGSOC and might have potential therapeutic advantages in this cancer (Zhifeng et al., 2014). Yu et al. (2014) showed that miR-1236-3p inhibited cell invasion and migration in HGSOC via targeting zinc-finger E-box binding homeobox 1 (ZEB1). Many previous studies reported the critical genes or miRNAs for HGSOC metastasis, but these studies were mainly focused on primary and metastasis tumor samples (Lalarukh et al., 2010; Ruifen et al., 2014; Tucker et al., 2014), and the study of ascites tumor, as intermediate stage of primary tumor and metastasis tumor, is limited. Ascites progression is related to the development of HGSOC in patients (Emma et al., 2013). Thus, not only primary tumor and metastasis tumor are important for the study of HGSOC metastasis but also ascites tumor is significant for it.

In this study, we aimed to identify critical genes for HGSOC metastasis and prognosis by combining primary tumor, ascites tumor, and metastasis tumor samples. First, the mRNA data of GSE73064 including15 HGSOC samples (5 primary tumor cells, 5 ascites tumor cells, and 5 metastasis tumor cells) were applied for the analysis. The data were preprocessed and time-series gene expression analyses were performed. Then, functional analysis for clustered genes was conducted, and protein–protein interaction (PPI) network was constructed. Subnetwork module was also analyzed. After that, we analyzed the miRNA regulatory network. Finally, survival analysis was performed and survival-related mRNAs in subnetwork modules were selected.

2. Materials and Methods

2.1. Data source

The mRNA data of GSE73064, the platform of which was GPL570 [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array, were downloaded from Gene Expression Omnibus database. There were 15 HGSOC samples (5 primary tumor cells, 5 ascites tumor cells, and 5 metastasis tumor cells) in these data.

2.2. Data preprocessing

The downloaded expression profile data (cel format) were preprocessed using R (3.4.3) package Affy (version 1.56.0, http://bioconductor.org/packages/release/bioc/html/affy.html). Data normalization was performed using robust multiarray average (RMA) method in Affy. The probe ID was mapped to gene symbol using annotation files in the platform. The probe not mapping to gene symbol was removed. For different probes mapping to the same symbol, the mean values of different probes were applied as the final expression value of this gene.

2.3. Identification of time-series genes

Short time-series expression miner (STEM) is a tool used to analyze gene expression data of short time series (Ernst and Bar-Joseph, 2006). Cluster analysis for gene expression profile of different time points was performed using STEM (version 1.3.11, www.cs.cmu.edu/~jernst/stem). The parameters were set as following: number of genes not less than 20, correlation coefficient of gene expression higher than 0.7, p-value less than 0.05, and Homo sapiens as gene annotation source. Genes with consistent trends were clustered together for the subsequent analyses.

2.4. Functional analysis for clustered genes

Gene Ontology (GO)–biological process (BP) (Ashburner et al., 2000) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway (Kanehisa and Goto, 2000) were analyzed to obtain two-color module gene sets using DAVID (Da et al., 2009) (version 6.8, https://david.ncifcrf.gov). p-Value <0.05 and count ≥2 were applied as screening criteria.

2.5. PPI network for clustered target genes and subnetwork analysis

The interaction between proteins encoded by two-color module genes was predicted based on STRING (version 10.5) database (Szklarczyk et al., 2017). The PPI network was constructed using Cytoscape (version 3.6.1) (Kohl et al., 2011). Degree centrality was applied to analyze node score in the network. Higher node scores indicated the importance of this node in the network, more likely to be key nodes. MCODE (version 1.4.2) plug-in (Bandettini et al., 2012) in Cytoscape software was used to analyze the subnetwork for the PPI network. The threshold value was >5.

2.6. miRNA regulatory network analysis

Based on ENCODE miRTarBase data in Enrichr database (Chen et al., 2013), upstream miRNAs that regulated the genes in subnetwork modules were predicted. The results with p-value <0.05 were selected. miRNA-target regulatory relationships were sort out, and the miRNA-target regulatory network was constructed.

2.7. Survival analysis

Disease (ovarian) information in The Cancer Genome Atlas (TCGA) database (http://firebrowse.org) was searched, and the data of 307 tumor samples were selected to use in the study. Combining clinical survival information of samples in TGCA database, the samples were divided into two groups (high expression and low expression groups) based on the expression values of genes in tumor group using R package Survival (version 2.42-6, https://cran.r-project.org/web/packages/survival/index.html). Log-rank statistical test was carried out. Survival-related mRNAs in subnetwork modules were selected with setting p < 0.05 as significant threshold value.

3. Results

3.1. Data preprocessing

After preprocessing, 20,192 genes were obtained. As presented in Figure 1, after normalization, the median of the data was on the same horizontal line, indicating good effect.

FIG. 1.

The box plot of expression profiles for the two set of data before and after normalization. Dark grey: normal samples; light grey: disease samples; lateral axis: sample name.

3.2. Identification of time-series genes

Four significantly enriched clusters were screened through time-series gene expression analysis (Fig. 2). The same color indicated the same trends; therefore, profile 10 and profile 14 were combined as dark grey-profile (494 genes) and profile 5 and profile 1 were combined as light grey-profile (303 genes). Collectively, these 797 genes were identified as time-series genes relative with HGSOC.

FIG. 2.

STEM cluster analysis. The square represents the different cluster gene sets obtained by STEM, and black polygonal line in square indicates the overall trend of all gene expressions in this gene set. The number in the left bottom of square indicates significant p-value in this gene set. Square in color represents significant difference. Square with the same color represents the same trend of the gene sets. STEM, short time-series expression miner.

3.3. Functional analysis for clustered genes

After functional analysis for mRNA in dark grey-profile and light grey-profile, pathways and GO-BP results associated with these genes were obtained. The top 10 results for dark grey-profile and light grey-profile are shown in Figure 3. Most genes in dark grey-profile were significantly enriched in Staphylococcus aureus infection pathway (e.g., FPR1), rheumatoid arthritis pathway (e.g., CXCL12), and lysosome pathway (e.g., CXCL12, Fig. 3A). Most genes in light grey-profile were significantly enriched in response to virus and negative regulation of viral genome replication (e.g., ISG20, Fig. 3B).

FIG. 3.

The results of pathways and GO-BP involved by dark grey-profile genes (A) and light grey-profile genes (B). Lateral axis: the number of enriched genes; vertical axis: names; gray line: p-value. GO, Gene Ontology; BP, biological process.

3.4. PPI network analysis and subnetwork analysis

There were 369 nodes and 1417 interaction pairs in the PPI network (Fig. 4). Five modules were obtained using Cytoscape plug-in MCODE with the threshold score of >5. Module 1 (score = 19.59) included 40 nodes and 382 interaction pairs, module 2 (score = 15) included 15 nodes and 105 interaction pairs, module 3 (score = 10.545) included 12 nodes and 58 interaction pairs, module 4 (score = 7.529) included 35 nodes and 128 interaction pairs, and module 5 (score = 7.25) included 9 nodes and 29 interaction pairs. The degrees of genes in the PPI network and modules are presented in Table 1. PPBP, ITGB2, and ITGAM were highlighted in the PPI network with high degrees.

FIG. 4.

PPI network for clustered genes. Circle: dark grey-profile genes; triangle: light grey-profile genes. PPI, protein–protein interaction.

Table 1.

The Degrees of Genes in the Protein–Protein Interaction (Top 30) and Submodule Networks

Module 1		Module 1		Module 2		Module 3		Module 4		Module 5
Name	Degree	Name	Degree	Name	Degree	Name	Degree	Name	Degree	Name	Degree
C3AR1	37	C3AR1	37	LYZ	27	CTSD	20	HLA-DRA	32	RSAD2	12
PPBP	35	PPBP	35	GGH	27	CTSS	18	HLA-DQB1	27	ISG15	11
ITGB2	32	ITGB2	32	HEXB	19	CFP	17	HLA-DQA1	27	IFIT1	11
HLA-DRA	32	ITGAM	30	CTSC	19	CTSH	15	HLA-DPB1	27	IFIT2	10
ITGAM	30	CXCL12	27	PYCARD	16	TIMP2	15	HLA-DPA1	27	HERC5	9
HLA-DQB1	27	CYBB	26	MAN2B1	15	DOCK2	14	FN1	22	IFITM1	7
HLA-DQA1	27	CCL4	26	MNDA	14	LTF	14	LCP2	16	ISG20	7
HLA-DPB1	27	FCER1G	26	RNASE2	14	PTX3	14	ADRB2	16	IFI44	5
HLA-DPA1	27	BST1	25	NPC2	14	OSCAR	14	PTPRC	16	IFI44L	5
CXCL12	27	PLAU	24	GRN	14	LCN2	12	MSLN	16
LYZ	27	GPR84	24	ADA2	14	SLPI	11	ARRB2	15
GGH	27	BDKRB1	23	FABP5	14	DNASE1L1	11	AP2S1	13
CYBB	26	CLEC5A	22	PLAC8	14			VCAN	12
CCL4	26	CD53	22	VAT1	14			APOE	11
FCER1G	26	FPR1	22	TRAPPC1	14			THY1	11
BST1	25	CD33	21					CDH2	11
VAV1	24	CD93	21					SPP1	11
PLAU	24	LAIR1	21					SH3KBP1	11
GPR84	24	OLR1	21					VWF	10
BDKRB1	23	MCEMP1	21					FCGR3B	9
CD3D	22	ATP8B4	21					CFD	8
CLEC5A	22	ITGAL	20					BIN1	8
FN1	22	CCL5	20					DAB2	8
CD53	22	CCR5	20					SPARCL1	8
FPR1	22	TNFRSF1B	20					PROS1	8
CD33	21	LILRB2	20					FAM20A	8
CD93	21	PSAP	20					LGALS1	8
LAIR1	21	CXCL10	19					F13A1	7
OLR1	21	CXCL11	19					FERMT3	7
		SIRPA	19					CD52	7
		CCR1	18					SRGN	7
		CX3CL1	18					FOLR2	7
		C5AR1	18					LYPD3	7
		GPR183	18					STON1	7
		ADORA3	18					VNN3	7
		FPR3	18
		HCAR3	18
		SUCNR1	18
		SLC2A5	17
		CD300A	16

3.5. miRNA regulatory network analysis

A total of 24 miRNAs were predicted, and 85 miRNA-target regulatory pairs were obtained. There were 29 genes from dark grey-profile and 13 genes from light grey-profile in this regulatory relation pairs. The miRNA-target regulatory network is presented in Figure 5. VAT1, OLR1, and SLC2A5 were targeted by more miRNAs than other genes. miR-1-3p and miR-146a-5p targeted more genes than other miRNAs in this regulatory network. For example, ISG20 and CXCL12 were targeted by miR-1-3p, and CXCL12 was targeted by miR-146a-5p. miR-4469, miR-6734-5p, and miR-1304-5p were also several of important miRNAs in the miRNA-target regulatory network.

FIG. 5.

The miRNA-target regulatory network. light grey circle: dark grey-profile genes; dark grey circle: light grey-profile genes; triangle: miRNA.

3.6. Screening of prognosis-related genes

A total of 111 subnetwork genes were extracted to screen prognosis-related genes. We obtained nine genes (ISG20, CLEC5A, ADORA3, CD300A, BDKRB1, THY1, FPR1, CXCL12, and SPP1) that were associated with the prognosis of HGSOC. Kaplan–Meier survival curves for these nine genes are shown in Figure 6.

FIG. 6.

Kaplan–Meier survival curves for nine genes (ISG20, CLEC5A, ADORA3, CD300A, BDKRB1, THY1, FPR1, CXCL12, and SPP1). Dotted line: below the median (lower expression group); Solid line: above the median (higher expression group).

4. Discussion

As the most aggressive type of OC, the effective treatment of HGSOC is urgently needed. In the present study, PPBP, ITGB2, and ITGAM are the genes with higher degrees in the PPI network. miR-4469, miR-6734-5p, and miR-1304-5p were several of important miRNAs in the miRNA-target regulatory network. In addition, survival analysis showed that nine genes such as CLEC5A, FPR1, and CXCL12 might be prognosis-related genes for HGSOC.

In the present PPI network, PPBP, ITGB2, and ITGAM were hub genes with higher degrees. PPBP, also known as CXCL7 (a member of the CXC subfamily of chemokines), promotes invasion and growth of cholangiocarcinoma cells (Guo et al., 2017). Hironobu et al. (2010) suggested that chemokines/chemokine receptors were associated with cancer metastasis and might be regarded as new targets for the treatment of metastatic cancer. The data of Zhuo et al. (2008) showed that CXCL7 might be significant for the metastasis of breast cancer. CXCL7/CXCR2 (receptor of CXCL7) axis is correlated with the metastasis of breast cells to malignant phenotype (Ju, 2012).

ITGB2 (CD18), one chain of αMβ2, can form multiple heterodimers, which play important roles in immune response and cellular adhesion (Arnaout, 1990). Cell adhesion is significantly involved in cancer metastasis (Kannagi et al., 2010).

ITGAM (CD11B) is the second chain of αMβ2, which is involved in the innate immune system (Solovjov et al., 2005). ITGAM plays role in adhesion, spreading, and migration of cells with the presence of CD18 (Solovjov et al., 2005). The blockage of CD18 or CD11B can result in significant decreases of microparticle-mediated metastasis of tumor cells (Jingwei et al., 2013). The β2 subunit of the LFA-1 (CD18/CD11a) plays modulatory role in the metastatic development of colorectal cancer to the liver through harming activation of liver endothelium (Benedicto et al., 2017).

CD11B/Gr1 is involved in colorectal cancer liver metastasis (Su et al., 2014). CD11B is correlated with tumor metastasis stage in gastric cancer (Okita et al., 2014). Thus, PPBP, ITGB2, and ITGAM may be critical for HGSOC metastasis.

It is reported that CLEC5A was correlated with the metastasis of prostate cancer (Marín-Aguilera et al., 2015). Szu-Ting et al. (2012) showed that blockade of CLEC5A attenuated inflammation of central nervous system. Joyce-Shaikh et al. (2010) indicated that in autoimmune joint inflammation, the blockade of CLEC5A reduced the clinical signs. Factors associated with inflammation (e.g., IL-6) can be applied to predict the prognosis of gastric cancer (Chang et al., 2014). Besides, CLEC3A plays role in the progression of breast invasive ductal cancer and can be used to predict prognosis of this cancer (Ni et al., 2018).

FPR1 is involved in the development of many cancers or tumors, such as lung cancer (Morris et al., 2016), breast cancer (Vecchi et al., 2018), and neuroblastoma (Snapkov et al., 2016). FPR2 promotes metastasis of cells and invasion of gastric cancer and can be used for predicting the prognosis of this cancer (Hou et al., 2017).

In rectal cancer, CXCL12 and CXCR4 are correlated with poor prognosis (Susumu et al., 2010). Furthermore, CXCL12 is related to prognosis in non-small cell lung cancer (Suzuki et al., 2008) and esophageal squamous cell carcinoma (Uchi et al., 2016). CXCL12-CXCR4 plays important roles in OC cell migration, growth, invasion, and metastasis (Wang et al., 2014; Qing et al., 2015) and may be potential therapeutic targets for OC (Salomonnson et al., 2013). Our present survival analysis showed that CLEC5A, FPR1, and CXCL12 were prognosis-related genes for HGSOC. Thus, combining with the previous studies, we inferred that CLEC5A, FPR1, and CXCL12 were critical genes for the prognosis of HGSOC.

In a previous study, Cao et al. (2017) showed that miR-4469 suppressed the metastasis of breast cancer by targeting CDK3. Kang et al. (2016) indicated that miR-6734 was involved in apoptosis and cell cycle arrest in colon cancer cells. Cheng-gang et al. (2017) suggested that miR-1304 could inhibit cell growth in non-small cell lung cancer through heme oxygenase-1. No more studies about the associations between these three miRNAs and cancers were reported. In our present study, miR-4469, miR-6734-5p, and miR-1304-5p were several of important miRNAs in the miRNA-target regulatory network. Thus, we inferred that miR-4469, miR-6734-5p, and miR-1304-5p might be significant miRNAs associated with HGSOC. But further verifications about this infer are needed.

In summary, PPBP, ITGB2, and ITGAM may be critical for HGSOC metastasis. CLEC5A, FPR1, and CXCL12 may be important for the prognosis of HGSOC. Furthermore, miR-4469, miR-6734-5p, and miR-1304-5p may be important miRNAs associated with HGSOC. However, because of the limitations existed in this study, for instance, small sample size and without experimental verification, further studies are needed.

Footnotes

Author Disclosure Statement

The authors declare they have no financial conflicts of interest.

Funding Information

This work was supported by the Traditional Chinese Medicine Science and Technology Development Plan Project of Jining City (grant number ZYY2015022).

References

Arnaout

M.A.

1990. Structure and function of the leukocyte adhesion molecules CD11/CD18. Blood. 75, 1037–1050.

Ashburner

, Ball

C.A.

, Blake

J.A.

, et al. 2000. Gene ontology: Tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29.

Bandettini

W.P.

, Kellman

, Mancini

, et al. 2012. MultiContrast Delayed Enhancement (MCODE) improves detection of subendocardial myocardial infarction by late gadolinium enhancement cardiovascular magnetic resonance: A clinical validation study. J. Cardiovasc. Magn. Reson. 14, 83–83.

Benedicto

, Marquez

, Herrero

, et al. 2017. Decreased expression of the β 2 integrin on tumor cells is associated with a reduction in liver metastasis of colorectal cancer in mice. BMC Cancer. 17, 827.

Bowtell

D.D.

, Steffen

B.H.

, Ahmed

A.A.

, et al. 2015. Rethinking ovarian cancer II: Reducing mortality from high-grade serous ovarian cancer. Nat. Rev. Cancer. 15, 668–79.

Cao

, Xiao

, Huang

, et al. 2017. CDK3, target of miR-4469, suppresses breast cancer metastasis via inhibiting Wnt/β-catenin pathway. Oncotarget. 8, 84917–84927.

Cesaratto

, Grisard

, Coan

, et al. 2016. BNC2is a putative tumor suppressor gene in high-grade serous ovarian carcinoma and impacts cell survival after oxidative stress. Cell Death Dis. 7, e2374.

Chang

W.J.

, Du

, Zhao

, et al. 2014. Inflammation-related factors predicting prognosis of gastric cancer. World J. Gastroenterol. 20, 4586–4596.

Chen

E.Y.

, Tan

C.M.

, Kou

, Duan

, Wang

2013. Enrichr: Interactive and collaborative HTML5 gene list enrichment;analysis tool. BMC Bioinformatics.14, 128.

10.

Coleman

R.L.

, Monk

B.J.

, Sood

A.K.

, et al. 2013. Latest research and treatment of advanced-stage epithelial ovarian cancer. Nat. Rev. Clin. Oncol. 10, 211–224.

11.

Creighton

C.J.

, Anadulce

H.H.

, Anders

, et al. 2012. Integrated analyses of microRNAs demonstrate their widespread influence on gene expression in high-grade serous ovarian carcinoma. PLoS One. 7, e34546.

12.

W.H.

, Sherman

B.T.

, and Lempicki

R.A.

2009. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57.

13.

Emma

, Tan

D.S.P.

, and Kaye

S.B.

2013. Meeting the challenge of ascites in ovarian cancer: New avenues for therapy and research. Nat. Rev. Cancer. 13, 273–282.

14.

Ernst

, and Bar-Joseph

2006. STEM: A tool for the analysis of short time series gene expression data. BMC Bioinformatics. 7, 191.

15.

Ferlay

, Soerjomataram

, Dikshit

, et al. 2015. Cancer incidence and mortality worldwide: Sources, methods and major patterns in GLOBOCAN 2012. Int. J. Cancer. 136, 9.

16.

Guo

, Jian

, Jia

, et al. 2017. CXCL7 promotes proliferation and invasion of cholangiocarcinoma cells. Oncol. Rep. 37, 1114–1122.

17.

Hironobu

, Tomoyuki

, Minoru

, et al. 2010. [Chemokine and chemokine receptor related to cancer metastasis]. Gan Kagaku Ryoho Cancer Chemotherapy. 37, 2052–2057.

18.

Hou

X.L.

, Ji

C.D.

, Tang

, et al. 2017. FPR2 promotes invasion and metastasis of gastric cancer cells and predicts the prognosis of patients. Sci. Rep. 7, 3153.

19.

Jingwei

, Wenqian

, Yi

, et al. 2013. Innate immune cell-derived microparticles facilitate hepatocarcinoma metastasis by transferring integrin α(M)β2 to tumor cells. J. Immunol. 191, 3453.

20.

Joyce-Shaikh

, Bigler

M.E.

, Chao

C.C.

, et al. 2010. Myeloid DAP12-associating lectin (MDL)-1 regulates synovial inflammation and bone erosion associated with autoimmune arthritis. J. Exp. Med. 207, 579–89.

21.

2012. The CXCL7/CXCR2 axis and the migration of breast cells toward the malignant phenotype. J. Clin. Oncol. 30, 181.

22.

Kanehisa

, and Goto

2000. KEGG: Kyoto Encyclopaedia of Genes and Genomes. Nucleic Acids Res. 28, 27–30.

23.

Kang

M.R.

, Park

K.H.

, Yang

J.O.

, et al. 2016. miR-6734 Up-Regulates p21 gene expression and induces cell cycle arrest and apoptosis in colon cancer cells. PLoS One. 11, e0160961.

24.

Kannagi

, Izawa

M.T.

, Miyazaki

, et al. 2010. Carbohydrate-mediated cell adhesion in cancer metastasis and angiogenesis. Cancer Sci. 95, 377–384.

25.

Kannan

, Coarfa

, Rajapakshe

, et al. 2014. CDKN2D-WDFY2 is a cancer-specific fusion gene recurrent in high-grade serous ovarian carcinoma. PLoS Genet. 10, e1004216.

26.

Kohl

, Wiese

, and Warscheid

2011. Cytoscape: Software for visualization and analysis of biological networks. Methods Mol. Biol. 696, 291–303.

27.

Lalarukh

, Ayse

, Whittaker

J.C.

, et al. 2010. The clonal evolution of metastases from primary serous epithelial ovarian cancers. Int. J. Cancer. 124, 1579–1586.

28.

C.G.

, Pu

M.F.

, Li

C.Z.

, et al. 2017. MicroRNA-1304 suppresses human non-small cell lung cancer cell growth in vitro by targeting heme oxygenase-1. Acta Pharmacol. Sin. 38, 110–119.

29.

Luo

L.Y.

, Kim

, Cheung

H.W.

, et al. 2015. The tyrosine kinase adaptor protein FRS2 is oncogenic and amplified in high-grade serous ovarian cancer. Mol. Cancer Res. 13, 502–509.

30.

Marín-Aguilera

, Reig Ò, Lozano

J.J.

, et al. 2015. Molecular profiling of peripheral blood is associated with circulating tumor cells content and poor survival in metastatic castration-resistant prostate cancer. Oncotarget. 6, 10604–10616.

31.

Morris

, Vachani

, Pass

H.I.

, et al. 2016. Whole Blood FPR1 mRNA expression predicts both non-small cell and small cell lung cancer. Int. J. Cancer. 11, S34–S34.

32.

, Peng

, Yang

F.L.

, et al. 2018. Overexpression of CLEC3A promotes tumor progression and poor prognosis in breast invasive ductal cancer. Onco Targets Ther. 11, 3303–3312.

33.

Okita

, Tanaka

, Ohira

, et al. 2014. Role of tumor-infiltrating CD11b + antigen-presenting cells in the progression of gastric cancer. J. Surg. Res. 186, 192–200.

34.

Qing

, Bu-Lang

, Xue-Jing

, et al. 2015. CXCL12-CXCR4 axis promotes proliferation, migration, invasion, and metastasis of ovarian cancer. Oncol. Res. 22, 247–58.

35.

Reiner

A.T.

, Tan

, Agreiter

, et al. 2017. EV-Associated MMP9 in high-grade serous ovarian cancer is preferentially localized to annexin V-binding EVs. Dis. Markers. 2017, 1–9.

36.

Ruifen

, Xiaolin

, Qing

, et al. 2014. miR-145 inhibits tumor growth and metastasis by targeting metadherin in high-grade serous ovarian carcinoma. Oncotarget. 5, 10816–10829.

37.

Salomonnson

, Stacer

A.C.

, Ehrlich

, et al. 2013. Imaging CXCL12-CXCR4 signaling in ovarian cancer therapy. PLoS One. 8, e51500.

38.

Snapkov

, Öqvist

C.O.

, Figenschau

, et al. 2016. The role of formyl peptide receptor 1 (FPR1) in neuroblastoma tumorigenesis. BMC Cancer. 16, 490.

39.

Solovjov

D.A.

, Elzbieta

, and Plow

E.F.

2005. Distinct roles for the alpha and beta subunits in the functions of integrin alphaMbeta2. J. Biol. Chem. 280, 1336–1345.

40.

Y.L.

, Zhao

, Gordon-Weeks

, et al. 2014. Tumor-derived CCL2 promotes recruitment of CD11b/Gr1midmyeloid cells to sustain colorectal cancer liver metastasis. Cancer Res. 73, 2871–2871.

41.

Susumu

, Yuji

, Koji

, et al. 2010. Stromal CXCR4 and CXCL12 expression is associated with distant recurrence and poor prognosis in rectal cancer after chemoradiotherapy. Ann. Surg. Oncol. 17, 2051–2058.

42.

Suzuki

, Mohamed

, Nakajima

, et al. 2008. Aberrant methylation of CXCL12 in non-small cell lung cancer is associated with an unfavorable prognosis. Int. J. Oncol. 33, 113.

43.

Szklarczyk

, Morris

J.H.

, Cook

, et al. 2017. The STRING database in 2017: Quality-controlled protein–protein association networks, made broadly accessible. Nucleic Acids Res. 45, D362–D368.

44.

Szu-Ting

, Ren-Shyan

, Ming-Fang

, et al. 2012. CLEC5A regulates Japanese encephalitis virus-induced neuroinflammation and lethality. PLoS Pathogens. 8, e1002655.

45.

Tucker

S.L.

, Kshipra

, Herbrich

S.M.

, et al. 2014. Molecular biomarkers of residual disease after surgical debulking of high-grade serous ovarian cancer. Clin. Cancer Res. 20, 3280–3288.

46.

Uchi

, Takeuchi

, Matsuda

, et al. 2016. CXCL12 expression promotes esophageal squamous cell carcinoma proliferation and worsens the prognosis. BMC Cancer. 16, 514.

47.

Vecchi

, Zoia

MAP

, Santos

T.G.

, et al. 2018. Inhibition of the AnxA1/FPR1 autocrine axis reduces MDA-MB-231 breast cancer cell growth and aggressiveness in vitro and in vivo. Biochim. Biophys. Acta Mol. Cell Res. 1865, 1368–1382.

48.

Wang

, Liu

, Wei

, et al. 2014. Effect of the LPA-mediated CXCL12-CXCR4 axis in the tumor proliferation, migration and invasion of ovarian cancer cell lines. Oncol. Lett. 7, 1581–1585.

49.

, Shi

, Xiaolin

, et al. 2014. miR-1236-3p represses the cell migration and invasion abilities by targeting ZEB1 in high-grade serous ovarian carcinoma. Oncol. Rep. 31, 1905.

50.

Zaid

T.M.

, Tsz-Lun

, Thompson

M.S.

, et al. 2013. Identification of FGFR4 as a potential therapeutic target for advanced-stage, high-grade serous ovarian cancer. Clin. Cancer Res. 19, 809–820.

51.

Zhifeng

, Jaeyeon

, Lin

, et al. 2014. Functional analysis of miR-34c as a putative tumor suppressor in high-grade serous ovarian cancer. Biol. Reprod. 91, 113.

52.

Zhuo

, Minghuan

, Fred

, et al. 2008. Increased invasion through basement membrane by CXCL7-transfected breast cells. Am. J. Surg. 196, 690–696.