Abstract
Background:
High-grade serous ovarian cancer (HGSOC) is the biggest cause of gynecological cancer-related mortality because of its extremely metastatic nature. This study aimed to explore and evaluate the characteristics of candidate factors associated with the metastasis and progression of HGSOC.
Methods:
Transcriptomic data of HGSOC patients' samples collected from primary tumors and matched omental metastatic tumors were obtained from three independent studies in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database. Differentially expressed genes (DEGs) were selected to evaluate the effects on the prognosis and progression of ovarian cancer using data from The Cancer Genome Atlas (TCGA) database. Hub genes' immune landscapes were estimated by the Tumor Immune Estimation Resource (TIMER) database. Finally, using 25 HGSOC patients' cancer tissues and 10 normal fallopian tube tissues, immunohistochemistry (IHC) was performed to quantify the expression levels of hub genes associated with International Federation of Gynecology and Obstetrics (FIGO) stages.
Results:
Fourteen DEGs, ADIPOQ, ALPK2, BARX1, CD37, CNR2, COL5A3, FABP4, FAP, GPR68, ITGBL1, MOXD1, PODNL1, SFRP2, and TRAF3IP3, were upregulated in metastatic tumors in every database while CADPS, GATA4, STAR, and TSPAN8 were downregulated. ALPK2, FAP, SFRP2, GATA4, STAR, and TSPAN8 were selected as hub genes significantly associated with survival and recurrence. All hub genes were correlated with tumor microenvironment infiltration, especially cancer-associated fibroblasts and natural killer (NK) cells. Furthermore, the expression of FAP and SFRP2 was positively correlated with the International Federation of Gynecology and Obstetrics (FIGO) stage, and their increased protein expression levels in metastatic samples compared with primary tumor samples and normal tissues were confirmed by IHC (P = 0.0002 and P = 0.0001, respectively).
Conclusions:
This study describes screening for DEGs in HGSOC primary tumors and matched metastasis tumors using integrated bioinformatics analyses. We identified six hub genes that were correlated with the progression of HGSOC, particularly FAP and SFRP2, which might provide effective targets to predict prognosis and provide novel insights into individual therapeutic strategies for HGSOC.
Keywords: High-grade serous ovarian cancer, Metastasis, Gene Expression Omnibus, Prognosis, Recurrence, Immune infiltration
Introduction
As the most common cause of death related to gynecological cancer, ovarian cancer (OC) is the fifth cause of cancer death among women worldwide, causing approximately 5% of female cancer deaths.[1] As the most aggressive and common histological subtype of OC, the overall survival (OS) rate of high-grade serous ovarian cancer (HGSOC) remains dismal, with 30% survival at 5 years after diagnosis because early detection is so difficult.[2] The main factor contributing to the high death to incidence rate is the advanced stage of the disease at the time of diagnosis.[3] Many patients are diagnosed after the tumor has already metastasized throughout the peritoneal cavity to other sites, which typically occurs in advanced HGSOC patients.[4] Commonly, HGSOC tumor cells undergo adhesion, migration, and invasion into the peritoneal cavity during metastasis and preferentially adhere to the omentum, which is the most common site of HGSOC metastasis.[5,6] HGSOC progression and metastasis are facilitated by the interactions between cancer cells and various stromal components. The tumor microenvironment (TME) plays an important role in various cancer types, including OC. A vast array of factors such as the immune response, metabolic pathways, and intracellular signaling molecules shape the pathogenicity of the TME, which in turn affects tumor progression and prognosis, and even the efficacy of conventional treatment and immunotherapy.[7] While metastasis plays a crucial role in promoting OC progression and decreasing patient survival rates, the underlying mechanisms of cancer cell spread have yet to be thoroughly explored, and there is still a lack of comprehensive understanding of disease metastasis and progression. Hence, it is of great value to understand the underlying molecular mechanisms of HGSOC metastasis.
In this study, we aimed to identify the gene expression profiles associated with the progression of HGSOC and employed a multistep bioinformatics strategy that used omics information and clinical data.
Methods
Ethical approval
This study was approved by the Ethics Committee of the Peking University People's Hospital (No. 2019HPB034-01). All patients provided the informed consent.
GEO datasets sources and processing
To identify critical genes involved in the metastasis of HGSOC, we searched keywords ((("ovarian cancer" OR "ovarian carcinoma") [all fields] AND "metastasis" [all fields]) AND "Homo sapiens" [porgn]) in NCBI GEO (https://www.ncbi.nlm.nih.gov/geo/). Then, we screened the titles and abstracts of datasets, and the full information of the datasets of interest was further evaluated and finally selected according to the following inclusion criteria: (1) selected datasets should be messenger RNA (mRNA) transcriptome data; (2) only datasets containing more than five matched pairs of primary tumors and metastatic tumor samples from omentum were included; and (3) the pathological type of the primary tumor is high-grade serous carcinoma. Finally, we retrieved the RNA-seq data from four independent datasets in NCBI GEO (GSE98281, GSE137237, GSE133296, and GSE30587).
The Cancer Genome Atlas (TCGA) TARGET Genotype-Tissue Expression (GTEx) dataset analysis
University of California Santa Cruz (UCSC) Xena is an online exploration tool for multi-omic and clinical/phenotype data, which provides easy access to publicly available cancer transcriptome data including OC to explore relationships between genomic and/or phenotypic variables.[8] TCGA TARGET GTEx study (https://xenabrowser.net/) was used to compare differential gene expression among normal ovary tissues from The GTEx and ovarian serous cystadenocarcinoma and recurrent tumors from TCGA to identify the impacts of candidate upregulated and downregulated differentially expressed genes (DEGs) on the progression of OC. We filtered the cohort and only kept the ovary samples cohort. The search term that was used to filter was "ovary". Then, we clicked to add columns for every candidate gene and choose the "sample type" option to compare the expression of the hub genes. To further explore the relationship of hub genes, log2-transformed RNA-sequencing data from 323 HGSOC patients in the TCGA dataset were also obtained through the UCSC Xena platform for correlation analysis (https://xenabrowser.net/datapages/).
Differential gene expression analysis
The DEGs were obtained by conducting differential expression analysis. The downloaded platform and series of matrix file(s) were converted using the R language software (version 4.0.5; R Foundation, Vienna, Austria) and annotation package. Gene differential expression analysis was performed using the limma package in the Bioconductor package (available online:http://www.bioconductor.org/). We divided the samples into two subgroups, namely the primary tumor group (PT) and the metastasis tumor group (MT). By employing absolute |log2 fold change (FC)| >1 and P <0.05, DEGs, including significantly upregulated and downregulated genes between HGSOC primary tumors vs. matched metastasis tissues in every GSE dataset, were identified.
Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment data analysis
With the screened DEGs, Metascape (http://www.metascape.org) was used to assess GO functional annotation of DEGs and candidate genes in biological networks. GO chord was constructed using R language software. KEGG pathway analyses of candidate genes were performed by KOBAS 3.0 online analysis database (http://kobas.cbi.pku.edu.cn/kobas3) to predict their underlying molecular functions. We set P <0.05 as statistically significant.
Kaplan–Meier plotter analysis
Kaplan–Meier plotter (https://kmplot.com) included OC transcriptomic datasets in GEO and TCGA.[9] The OS analyses and progression-free survival (PFS) analyses for upregulated and downregulated candidate genes were performed using Kaplan–Meier plotter with hazard ratios (HRs) with 95% confidence intervals and log-rank P-value cutoffs for each gene at 0.05. Only patients diagnosed with serous ovarian cancer were used for the analysis.
TIMER2.0 database analysis
The Tumor Immune Estimation Resource (TIMER2.0) tool is a comprehensive online resource for the analysis of immune cell infiltration among diverse cancers.[10] TIMER2.0 was used to analyze the correlation of hub genes with the infiltration level of several TME cell types, including cluster of differentiation (CD)8+ T cells, CD4+ T cells, B cells, myeloid dendritic cells, natural killer (NK) cells, macrophages, and cancer-associated fibroblasts (CAFs).
Participant recruitment
For the validation of the hub gene findings, 25 Chinese HGSOC patients were diagnosed and recruited at our hospital from January 2020 to December 2021. Of these, paired samples of the primary tumor and omental metastases were available for 14 patients, while another 6 patients provided primary tumor tissue and 5 patients provided omental metastases. As a control set, 10 samples of fallopian tube tissues without the presence of tumor cells were used. After surgery, patients' tissues were prepared into paraffin sections.
Immunohistochemistry (IHC) staining and H-score
IHC was performed on paraffin sections of OC samples, omental metastatic cancer samples, and normal tissues to characterize the expression profiles of target genes (FAP and SFRP2). Briefly, after deparaffinization and rehydration, OC tissue paraffin sections (4 μm) were subjected to heat-induced epitope retrieval using ethylene diamine tetraacetic acid (EDTA) (C1033; Solarbio, Beijing), and then the sections were treated with a blocking buffer. Primary antibodies against fibroblast activation protein (FAP) (ab207178; Abcam) and secreted frizzled-related protein 2 (SFRP2) (sc-365524; Santa Cruz Biotechnology) were incubated with the prepared sections overnight at 4°C for further staining.
The expression levels were quantified by the histochemistry score (H-SCORE) using ImageJ software (Version v1.8.0, National Institutes of Health, Bethesda, USA). Both the staining intensity and the percentage of cells stained using this method were taken into account. The following formula was used: H-SCORE = Σ (PI × I) = (percentage of cells of weak intensity × 1) + (percentage of cells of moderate intensity × 2) + (percentage of cells of strong intensity × 3). In the formula, "PI" represents the percentage of positive cells among the total number of cells in this position, and "I" represents the intensity of staining, resulting in a score ranging from 0 to 300. Thus, higher H-scores indicate more intense staining in a higher percentage of cells.
Statistical analysis
Statistical processing was performed using Statistical Package for the Social Sciences (SPSS) software (version 21; IBM Corp., Armonk, NY, United States). Student's t-tests were used to compare the expression levels of the PT and MT groups in the GEO dataset. One way-analysis of variance test was used for statistical comparisons among multiple groups of boxplots. The log-rank test was used to calculate HRs and log-rank P-values to compare the survival curves. The correlation of hub gene expression and their relationship with TME cell infiltration levels was determined by Spearman's correlation. P-values <0.05 were considered statistically significant.
Results
Screening of DEGs associated with metastasis of HGSOC
To identify the critical genes involved in the metastasis and progression of HGSOC, we selected four GEO datasets, GSE30587, GSE137237, GSE98281, and GSE133296, according to our workflow. As shown in Figure 1, there were 9, 11, 10, and 10 matched primary tumor and metastases tissues in the above four datasets, respectively. Each expression dataset was analyzed by the limma package and expression data were standardized [Supplementary Figure 1, http://links.lww.com/CM9/B874]. A total of 63 and 2 DEGs were significantly upregulated and downregulated, respectively, in the MT group compared with the PT group in GSE30587 [Supplementary Table 1, http://links.lww.com/CM9/B472]. Furthermore, 1884 DEGs were obtained from GSE137237. Seven hundred sixty-three and 1121 DEGs were significantly upregulated and downregulated, respectively, in the MT group compared with the PT group Supplementary Figure 2A, http://links.lww.com/CM9/B874 and Supplementary Table 2, http://links.lww.com/CM9/B473]. Additionally, 465 DEGs were screened from GSE98281. Among them, 193 and 272 DEGs were significantly upregulated and downregulated, respectively, in the MT group compared with the PT group Supplementary Figure 2B, http://links.lww.com/CM9/B874 and Supplementary Table 3, http://links.lww.com/CM9/B473]. Furthermore, 1284 upregulated genes and 1359 downregulated genes were identified in GSE133296 according to the same principles of data processing mentioned above Supplementary Figure 2C, http://links.lww.com/CM9/B874 and Supplementary Table 4, http://links.lww.com/CM9/B480]. Considering that there were only two DEGs downregulated in the MT group in GSE30587, which would affect the overall data processing, the datasets that were finally included were GSE137237, GSE98281, and GSE133296.
Figure 1.
Workflow diagram of whole study design and analysis. DEGs: Differentially expressed genes; FIGO: International Federation of Gynecology and Obstetrics; GEO: Gene Expression Omnibus; GO: Gene Ontology; GSE: Gene set-based expression; GTEx: TARGET Genotype-Tissue Expression;HGSOC: High grade serous ovarian cancer; IHC: Immunohistochemistry; KEGG: Kyoto Encyclopedia of Genes and Genomes; OC: Ovarian cancer; OS: Overall survival; PFS: Progression-free survival; TCGA: The Cancer Genome Atlas; TIMER: Tumor Immune Estimation Resource.
To clarify the major characteristics of the tumor metastasis process, we selected the upregulated and downregulated DEGs of every dataset with |log2FC| >1.5 to assess GO function enrichment. In GSE137237, the top highly enriched GO terms of upregulated DEGs in the MT group were extracellular matrix organization, vasculature development, connective tissue development, and leukocyte migration specifically, while cilium movement pathways were chiefly enriched in downregulated DEGs [Supplementary Figures 2D–G, http://links.lww.com/CM9/B874]. In GSE98281, the top five enriched GO terms of upregulated DEGs were lymphocyte activation, response to bacterium, regulation of cytokine production, negative regulation of immune system process, and leukocyte differentiation, while the chemical synaptic transmission was obviously enriched in downregulated DEGs [Supplementary Figures 2H–K, http://links.lww.com/CM9/B874]. In GSE133296, cornification and adaptive immune response were distinctly enriched for upregulated DEGs while development of primay sexual characteristic and regulation of phospholipase C activity were enriched in downregulated DEGs [Supplementary Figures 2L–O, http://links.lww.com/CM9/B874]. Briefly, we found the pathways commonly enriched among upregulated genes were focused on the extracellular matrix and immune-related response programs.
Identification of candidate genes and functional enrichment analysis
Venn diagrams showed 18 candidate genes, which consisted of 14 upregulated genes and 4 downregulated genes in the MT group, that were common in three comparison sets [Supplementary Figures 3A,B, http://links.lww.com/CM9/B874]. Overall, the genes that were highly expressed in the MT group have a carcinogenic effect. Therefore, we concentrated on the characteristics of these upregulated genes. The enriched GO terms of these genes were negative regulation of Wnt signaling, fat cell differentiation, extracellular matrix organization, and negative regulation of response to external stimulus [Supplementary Figure 3C, http://links.lww.com/CM9/B874 and Supplementary Table 5, http://links.lww.com/CM9/B623]. Furthermore, the enriched KEGG pathways were peroxisome proliferators-activated receptor (PPAR) signaling pathways and metabolism pathways such as steroid and lipid pathways [Supplementary Figure 3D, http://links.lww.com/CM9/B874 and Supplementary Table 6, http://links.lww.com/CM9/B624]. These enriched pathways indicate that the exchange of materials between tumor cells and the microenvironment might be frequent.
Effects of individual candidate genes on OS and PFS
To explore the potential roles of the 18 candidate DEGs on OS and PFS, Kaplan–Meier survival curves were generated to establish the relationship between the prognostic roles and gene expression levels. Among the 18 DEGs, a total of 14 DEGs were found to be significantly related to OS in the log-rank test while 16 of 18 DEGs were linked with PFS. Among the 14 upregulated candidate genes, we found that the expression of 10 DEGs (ADIPOQ, ALPK2, BARX1, COL5A3, FABP4, FAP, GPR68, ITGBL1, MOXD1, and SFRP2) was associated with OS [Supplementary Figure 4, http://links.lww.com/CM9/B874]. Notably, with the exception of GPR68, the high expression of the other nine genes indicated a poor prognosis of OC patients. In this study, we mainly focused on the upregulated genes with high expression in the MT group, which might be correlated with an unfavorable prognosis. Thus, these nine genes (ADIPOQ, ALPK2, BARX1, COL5A3, FABP4, FAP, ITGBL1, MOXD1, and SFRP2) were worthy of more attention.
For PFS, except CNR2 and TRAF3IP3, the other 12 upregulated candidate genes were associated with short PFS of serous OC patients. Additionally, the high expression level of four DEGs (CADPS, GATA4, STAR, and TSPAN8) were down-regulated in the MT group indicated a good prognosis of patients for both OS and PFS [Supplementary Figures 5A–R, http://links.lww.com/CM9/B874].
Expression validation and recurrence evaluation of candidate genes in TCGA TARGET GTEx study
HGSOC patients usually developed tumor recurrence, which is the main reason for their poor prognosis. We compared gene expression among normal, primary, and recurrent tissues in the TCGA TARGET GTEx dataset and found that 15 of 18 candidate genes were significantly different. More importantly, our results showed that the expression of ALPK2, CD37, CNR2, FAP, GPR68, and SFRP2 were increased during tumor recurrence, while GATA4, STAR, and TSPAN8 were decreased [Figure 2].
Figure 2.
Evaluation of RNA expression of hub genes among normal ovary, primary tumors, and recurrent tumors. (A–N) Differential expression of upregulated genes. (O–R) Differential expression of downregulated genes.
Exploration of hub genes according to OS, PFS, and recurrence
In our analysis, several critical factors involved in the metastatic process would have changed during normal–primary–recurrent progression. Furthermore, it is well known that DEG patterns of metastasis are more likely to play crucial roles in the progression and prognosis of cancer. To precisely identify hub genes, we combined the effects of the candidate genes on OS, PFS, and recurrence, as shown in Tables 1 and 2. We identified hub genes among 14 upregulated candidate genes in the MT group associated with poor OS and PFS: the expression of these genes in recurrent tumors was higher than that in primary tumors and normal tissues. The downregulated genes were associated with good prognosis and lower expression level in recurrent tumors. ALPK2, FAP, SFRP2, GATA4, STAR, and TSPAN8 were markedly identified. Moreover, the Human Protein Atlas database was used to validate the protein expression of these hub genes, as shown in [Supplementary Figure 6A–E, http://links.lww.com/CM9/B874]. ALPK2 and FAP exhibited higher staining in the membranous and cytosolic compartments of HGSOC tissues compared with normal ovaries. At the same time, the association between the expression levels of these six genes in HGSOC patients from the TCGA dataset was explored in [Supplementary Figure 6F, http://links.lww.com/CM9/B874]. ALPK2, FAP, and SFRP2 expression were positively correlated with survival, while GATA4, STAR, and TSPAN8 were negatively correlated.
Table 1.
Survival and expression analyses of the 14 upregulated genes in OC patients.
| Gene | OS (P <0.05) | PFS (P <0.05) | Expression (normal <primary <recurrent) |
|---|---|---|---|
| ADIPOQ | ✓ | ✓ | ✗* |
| ALPK2 | ✓ | ✓ | ✓* |
| BARX1 | ✓ | ✓ | ✗* |
| CD37 | ✗ | ✓ | ✓* |
| CNR2 | ✗ | ✗ | ✓* |
| COL5A3 | ✓ | ✓ | ✗* |
| FABP4 | ✓ | ✓ | ✗* |
| FAP | ✓ | ✓ | ✓* |
| GPR68 | ✗ | ✓ | ✓* |
| ITGBL1 | ✓ | ✓ | ✗* |
| MOXD1 | ✓ | ✓ | ✗* |
| PODNL1 | ✗ | ✓ | ✗* |
| SFRP2 | ✓ | ✓ | ✓* |
| TRAF3IP3 | ✗ | ✗ | ✗* |
*means the difference was statistically significant among the groups; ✓ means high expression of candidate genes indicated a poor prognosis of OC patients; ✗ means expression of candidate genes was not correlated with the prognosis of OC patients; ✓* means expression of candidate genes in recurrent tumors was higher than that in primary tumors and normal tissues.OC: Ovarian cancer; OS: overall survival; PFS: Progression-free survival.
Table 2.
Survival and expression analyses of the four downregulated genes in OC patients.
| Gene | OS (P <0.05) | PFS (P <0.05) | Expression (normal >primary >recurrent) |
|---|---|---|---|
| CADPS | ✓ | ✓ | ✗* |
| GATA4 | ✓ | ✓ | ✓* |
| STAR | ✓ | ✓ | ✓* |
| TSPAN8 | ✓ | ✓ | ✓* |
*means the difference was statistically significant among the groups; ✓ means high expression of candidate genes indicated a poor prognosis of OC patients; ✗ means expression of candidate genes was not correlated with the prognosis of OC patients; ✓* means expression of candidate genes in recurrent tumors was lower than that in primary tumors and normal tissues.OC: Ovarian cancer; OS: overall survival; PFS: Progression-free survival.
Relationship between hub gene expression and the infiltration levels of cells in the TME
The interplay between metastatic tumor cells and different cell types within the TME is very tight in the process of tumor metastasis. Thus, we examined the correlation between the expression of hub genes and the differential abundance of infiltrating cells in the TME, especially that of six immune cell types and CAFs in the TIMER2.0 database. All the hub genes were correlated with tumor purity in OC tissues [Supplementary Figure 7, http://links.lww.com/CM9/B874 and Table 3]. Notably, we observed that three upregulated hub genes (ALPK2, FAP, and SFRP2) presented significant positive associations with the infiltrating levels of CD8+ T cells, B cells, macrophages, NK cells, and CAFs, among which these genes were most strongly correlated with CAFs (correlation coefficient [COR], 0.736–0.830; P <0.001), NK cells (COR, 0.238–0.325; P <0.001), and macrophages (COR, 0.238–0.244; P <0.001). Conversely, downregulated hub genes (GATA4, STAR, and TSPAN8) were negatively correlated with the infiltration of NK cells (COR, –0.290 to –0.330; P <0.001) and CAFs (COR, –0.237 to –0.324; P <0.001) [Supplementary Figure 8, http://links.lww.com/CM9/B874]. This indicated that these hub genes were specifically related to cancer-associated NKs and CAFs in the OC microenvironment.
Table 3.
Relationship between the expression of six hub genes and the infiltration levels of cells in the TME.
| Hub genes | Purity | CD8+ T cell | CD4+ T cell | B cell | DC cell | NK cell | Macrophages | CAF | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| COR | P-value | COR | P-value | COR | P-value | COR | P-value | COR | P-value | COR | P-value | COR | P-value | COR | P-value | |
| ALPK2 | –0.519 | 1.24e-18 | 0.203 | 1.26e-03 | –0.055 | 0.391 | –0.187 | 3.12e-03 | –0.005 | 0.936 | 0.302 | 1.23e-06 | 0.243 | 1.08e-04 | 0.803 | 1.69e-57 |
| FAP | –0.595 | 2.77e-25 | 0.196 | 1.85e-03 | –0.045 | 0.479 | –0.222 | 4.05e-04 | 0.035 | 0.583 | 0.325 | 1.62e-07 | 0.238 | 1.50e-04 | 0.830 | 1.45e-64 |
| SFRP2 | –0.500 | 3.04e-17 | 0.190 | 2.64e-03 | –0.094 | 0.137 | –0.245 | 9.50e-05 | –0.064 | 0.313 | 0.238 | 1.55e-04 | 0.244 | 1.02e-04 | 0.736 | 8.27e-44 |
| GATA4 | 0.138 | 0.0289 | –0.125 | 0.0494 | 0.044 | 0.489 | 0.025 | 0.691 | –0.140 | 0.027 | –0.290 | 3.32e-06 | –0.208 | 9.45e-04 | –0.237 | 1.59e-04 |
| STAR | 0.281 | 6.61e-06 | –0.179 | 4.65e-03 | 0.091 | 0.151 | 0.040 | 0.526 | –0.138 | 0.0298 | –0.297 | 1.88e-06 | –0.200 | 1.52e-03 | –0.300 | 1.46e-06 |
| TSPAN8 | 0.224 | 3.57e-04 | –0.115 | 0.0698 | 0.011 | 0.862 | 0.001 | 0.993 | –0.105 | 0.0984 | –0.330 | 1.02e-07 | 0.007 | 0.917 | –0.324 | 1.7e-07 |
CAF: Cancer associated fibroblasts; CD4: Cluster of differentiation 4; COR: Correlation coefficient; DC: Dendritic cells; NK: natural killer; TME: Tumor environment.
Relationship between hub gene expression and International Federation of Gynecology and Obstetrics (FIGO) stages of OC and the protein expression levels of hub genes in different tissue types
We also evaluated the relationship between gene expression and clinical stage. Interestingly, two hub genes, FAP and SFRP2, were associated with the severity of the epithelial ovarian cancer (EOC) FIGO stage with an increasing trend [Figures 3A–F]. Then, we compared the protein expression of FAP and SFRP2 in control fallopian tube tissues, primary ovarian carcinoma samples, and metastases samples collected from our center [Figure 3G,H]. In addition, stronger positive FAP and SFRP2 staining was generally present in HGSOC metastatic tissues compared with primary tumor tissues and normal controls. The differences in the total scores of FAP and SFRP2 among the tissue types were statistically significant (P = 0.0002 and P = 0.0001)[Figure 3I]. Notably, FAP overexpression was greater in fibroblasts from paired metastatic tissues than those from primary tumor tissue.
Figure 3.
Differential expression of six hub genes in various clinical stages and representative IHC images of FAP and SFRP2 in different tissue types. (A–F) Differential expression of hub genes in various FIGO stages of ovarian cancer. (G–I) IHC staining and H-score of FAP and SFRP2 in different tissue types (fallopian tube, primary cancer tissue, and paired metastatic tissue. Dotted arrow, cytoplasm; solid arrow, CAFs. Bar = 100 μm). CAFs: Cancer-associated fibroblasts; FAP: Fibroblast activation protein; FIGO: International Federation of Gynecology and Obstetrics; IHC: Immunohistochemistry; SFRP2: Secreted frizzled-related protein 2.
Discussion
HGSOC is the most common histological subtype of OC and has a poor 5-year survival rate of 35–40%, accounting for 70–80% of OC deaths.[1] Although surgery and other treatment methods have been improved, the treatment effects and prognosis of HGSOC patients are very poor because of the late diagnosis.[11] Because approximately 70% of OC cases are diagnosed at a late stage when cancer cells are actively metastasizing, understanding OC pathogenesis and the mechanism of its metastasis is crucial for the management of this deadly, highly metastatic disease.[5,12] Although multiple-level studies have been performed, they are not yet mature. Thus, there remains a need to screen novel metastasis-associated factors to better understand this mechanism.
We conducted this study to identify more useful metastatic factors in HGSOC and used bioinformatical methods to analyze the profiles of three datasets (GSE137237, GSE98281, and GSE133296). We compared the differential gene expression of metastatic tumors and primary tumors in every dataset. Our results showed that the prominent features of upregulated DEGs were extracellular matrix and regulation of immune response and angiogenesis. Many researchers have demonstrated that OC metastases from the primary site to omental metastatic tumor sites are facilitated by the interaction between cancer cells and various microenvironment components.[13,14] After OC cell implantation, the resultant inflammation and injury stimulate the peritoneal cells and their associated immune and stromal cells to release cytokines, which further leads to tumor progression.[15]
In our study, we took the intersection of the DEGs of the above three datasets. These datasets shared 18 common DEGs in the MT group: 14 upregulated genes and 4 downregulated genes. Subsequent GO and KEGG analyses focused on upregulated genes to intensively examine metastasis characteristics. We found that the GO terms of 14 candidate genes were particularly enriched in the negative regulation of the Wnt signaling pathway and fat cell differentiation. The Wnt/β-catenin signaling pathway plays a crucial role in tumorigenesis, metastasis, and therapy resistance of OC subtypes, especially HGSOC.[16,17,18] McMellen et al[19] demonstrated that the expression of typical Wnt molecules is dependent on the anatomic/metastatic site, highlighting the importance of the TME and indicating that Wnt signaling activity in HGSOC varies depending on this TME. Recent studies highlighted the understanding of the relevance of altered lipid metabolic pathways contributing to the poor prognosis of HGSOC. Cancer cells utilize various strategies to boost lipid uptake to fulfill their high energetic needs for cell proliferation and altered oncogenic signaling.[20,21] Meanwhile, cancer-associated adipocytes promote the proliferation of cancer cells mediated by adipokines including interleukin-8 (IL-8) and provide energy for rapid tumor growth.[22,23] Conversely, KEGG pathway enrichment of the above genes revealed the PPAR signaling pathway and the metabolism of multiple products, especially lipids. PPAR exerts multiple regulatory functions in the metabolism of lipids, glucose, and amino acids, as well as inflammation, cell proliferation, and apoptosis in cancer.[24] Mukherjee et al[25] reported that FABP4, which is mediated by PPAR, was upregulated in omental metastases, suggesting a pro-tumorigenic role in OC. This is consistent with our results and suggests that new strategies targeting critical enzymes involved in lipid uptake or utilization in cancer cells should be investigated.
To better identify the crucial factors of metastasis, the intersected genes among common DEGs, prognosis-, and recurrence-related genes were filtered. Finally, six hub genes were selected: ALPK2, FAP, and SFRP2 were upregulated in HGSOC metastatic tissues while GATA4, STAR, and TSPAN8 were downregulated. Interestingly, several studies confirmed that the loss of expression and epigenetic silencing of GATA4 could play an important role in ovarian carcinogenesis.[26,27] Abd-Elaziz et al[28] reported that STAR expression appeared to confer a better prognosis of patients with ovarian carcinoma. Furthermore, TSPAN8 could be a therapeutic target in epithelial OC invasion and metastasis.[29] It is known that ALPK2 plays a vital role in cancer by regulating the cell cycle and DNA repair, and the knockdown of ALPK2 could inhibit proliferation and migration and promote apoptosis, arresting the cell cycle of OC cells.[30,31] FAP is overexpressed by fibroblasts present in the microenvironment of many tumors. Various studies have found that high expression of FAP in EOC is associated with poorer clinical outcomes.[32,33] FAP may have novel cell-autonomous effects, suggesting that targeting FAP could have pleiotropic anti-tumor effects. Notably, SFRP2, a potent regulator of Wnt signaling, can stimulate the angiogenesis of cancer and accelerate tumor metastasis formation.[34] Therefore, several researchers highlighted that SFRP2 has a potential role as an anti-angiogenic therapeutic target in cancer. Significantly, Mariani et al[35] reported that FAP and SFRP2 were both overexpressed in bowel metastases compared with primary tumors among patients with OC by RNA sequencing. This was consistent with our findings and confirmed that FAP and SFRP2 should be studied as potential therapeutic targets to decrease the adhesion or proliferation of OC cells. Furthermore, the relationship between SFRP2 and the OC TME has not been reported, and the mechanism and functional experiments of hub genes in OC were ambiguous. Thus, the current research highlighted the value of these hub genes in the progression of OC.
Besides malignant transformed cells, tumors are composed of other cells including fibroblasts, muscle cells, and immune cells, which form the TME. The TME is a remarkable complex ecosystem and its components participate in numerous cancer progresses, such as tumorigenesis, metastasis,[36,37] response to chemotherapy and immune checkpoint blocking therapy,[38] and prediction of clinical outcomes.[39,40] Our results showed that six hub genes presented significant associations with the infiltration levels of NK cells and CAFs. CAFs are the most abundant cell type in the TME and greatly influence immune cell activity, functioning within the TME.[41,42] Furthermore, the involvement of NK cells in OC is multifaceted because they can exert pro-tumor or anti-tumor effects. The functional capability of NK cells may be strongly impaired by the TME.[43] Activated CAFs are known to support tumor growth and secrete chemokines and cytokines such as transforming growth factor-β (TGF-β), which can effectuate immunosuppression and inhibit cytokine secretion from NK cells, thus decreasing its cytotoxicity.[44] Although the molecular mechanism of NK cell–CAF interactions is still not clear, these results demonstrate that targeting tumor-promoting CAFs and exploiting NK cells separately are beneficial as therapeutic strategies. Although there are currently 28 clinical trials registered on Clinical Trials.gov that are investigating NK cell therapy and OC, the data on OC is still limited. It is worth exploring how to exploit NK cell cytotoxicity in tumors and how to target CAFs to enhance the efficacy of cancer therapies and cytotoxic immune cells. Moreover, targeting FAP and SFRP2 by combining the above treatment methods may obtain better therapeutic effects in OC.
Several limitations of our study should be acknowledged. Our study provides evidence that six novel genes are significantly related to the progression of HGSOC patients and may be potential therapeutic targets for precision medicine through data mining. Further functional experiments to reveal their roles in cancer are valuable and crucial.
In conclusion, HGSOC is undoubtedly a challenge for patients, medical practitioners, and researchers because of its lethality. In this study, we identified six hub genes (ALPK2, FAP, SFRP2, GATA4, STAR, and TSPAN8) associated with the metastasis and progression of HGSOC, and explored the relationship between these genes and TME cell infiltration, especially FAP and SFRP2. This study will help us to understand the underlying mechanisms involved in HGSOC and aid in the development of a clinical assay for guiding the therapeutic selection and individualized treatment.
Funding
This work was supported by the grant from Beijing Natural Science Foundation (No. 7222202).
Conflicts of interest
None.
Supplementary Material
Footnotes
Huiping Liu and Ling Zhou contributed equally to this study.
How to cite this article: Liu HP, Zhou L, Cheng HY, Wang S, Luan WQ, Cai E, Ye X, Zhu HL, Cui H, Li Y, Chang XH. Characterization of candidate factors associated with the metastasis and progression of high-grade serous ovarian cancer. Chin Med J 2023;136:2974–2982. doi: 10.1097/CM9.0000000000002328
References
- 1.Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer statistics, 2021. CA Cancer J Clin 2021;71: 7–33. doi: 10.3322/caac.21654. [DOI] [PubMed] [Google Scholar]
- 2.Lheureux S, Braunstein M, Oza AM. Epithelial ovarian cancer: Evolution of management in the era of precision medicine. CA Cancer J Clin 2019;69: 280–304. doi: 10.3322/caac.21559. [DOI] [PubMed] [Google Scholar]
- 3.Nik NN, Vang R, Shih Ie M, Kurman RJ. Origin and pathogenesis of pelvic (ovarian, tubal, and primary peritoneal) serous carcinoma. Annu Rev Pathol 2014;9: 27–45. doi: 10.1146/annurev-pathol-020712-163949. [DOI] [PubMed] [Google Scholar]
- 4.Ferraresi A Girone C Esposito A Vidoni C Vallino L Secomandi E, et al. How autophagy shapes the tumor microenvironment in ovarian cancer. Front Oncol 2020;10: 599915. doi: 10.3389/fonc.2020.599915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Yeung TL, Leung CS, Yip KP, Au Yeung CL, Wong ST, Mok SC. Cellular and molecular processes in ovarian cancer metastasis. A review in the theme: Cell and molecular processes in cancer metastasis. Am J Physiol Cell Physiol 2015;309: C444–C456. doi: 10.1152/ajpcell.00188.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bilbao M, Aikins JK, Ostrovsky O. Is routine omentectomy of grossly normal omentum helpful in surgery for ovarian cancer? A look at the tumor microenvironment and its clinical implications. Gynecol Oncol 2021;161: 78–82. doi: 10.1016/j.ygyno.2020.12.033. [DOI] [PubMed] [Google Scholar]
- 7.Lengyel E. Ovarian cancer development and metastasis. Am J Pathol 2010;177: 1053–1064. doi: 10.2353/ajpath.2010.100105. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Goldman MJ Craft B Hastie M Repečka K McDade F Kamath A, et al. Visualizing and interpreting cancer genomics data via the Xena platform. Nat Biotechnol 2020;38: 675–678. doi: 10.1038/s41587-020-0546-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gyorffy B, Lánczky A, Szállási Z. Implementing an online tool for genome-wide validation of survival-associated biomarkers in ovarian-cancer using microarray data from 1287 patients. Endocr Relat Cancer 2012;19: 197–208. doi: 10.1530/ERC-11-0329. [DOI] [PubMed] [Google Scholar]
- 10.Li T Fu J Zeng Z Cohen D Li J Chen Q, et al. TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res 2020;48: W509–W514. doi: 10.1093/nar/gkaa407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Bowtell DD Böhm S Ahmed AA Aspuria PJ Bast RC Jr Beral V, et al. Rethinking ovarian cancer II: Reducing mortality from high-grade serous ovarian cancer. Nat Rev Cancer 2015;15: 668–679. doi: 10.1038/nrc4019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kostova I, Mandal R, Becker S, Strebhardt K. The role of caspase-8 in the tumor microenvironment of ovarian cancer. Cancer Metastasis Rev 2021;40: 303–318. doi: 10.1007/s10555-020-09935-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Motohara T Masuda K Morotti M Zheng Y El-Sahhar S Chong KY, et al. An evolving story of the metastatic voyage of ovarian cancer cells: Cellular and molecular orchestration of the adipose-rich metastatic microenvironment. Oncogene 2019;38: 2885–2898. doi: 10.1038/s41388-018-0637-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Worzfeld T Pogge von Strandmann E Huber M Adhikary T Wagner U Reinartz S, et al. The unique molecular and cellular microenvironment of ovarian cancer. Front Oncol 2017;7: 24. doi: 10.3389/fonc.2017.00024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.De Nola R Menga A Castegna A Loizzi V Ranieri G Cicinelli E, et al. The crowded crosstalk between cancer cells and stromal microenvironment in gynecological malignancies: Biological pathways and therapeutic implication. Int J Mol Sci 2019;20: 2401. doi: 10.3390/ijms20102401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Nguyen VHL, Hough R, Bernaudo S, Peng C. Wnt/β-catenin signalling in ovarian cancer: Insights into its hyperactivation and function in tumorigenesis. J Ovarian Res 2019;12: 122. doi: 10.1186/s13048-019-0596-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Teeuwssen M, Fodde R. Wnt signaling in ovarian cancer stemness, EMT, and therapy resistance. J Clin Med 2019;8: 1658. doi: 10.3390/jcm8101658. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kotrbová A Ovesná P Gybel' T Radaszkiewicz T Bednaříková M Hausnerová J, et al. WNT signaling inducing activity in ascites predicts poor outcome in ovarian cancer. Theranostics 2020;10: 537–552. doi: 10.7150/thno.37423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.McMellen A, Woodruff ER, Corr BR, Bitler BG, Moroney MR. Wnt signaling in gynecologic malignancies. Int J Mol Sci 2020;21: 4272. doi: 10.3390/ijms21124272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yu C Niu X Du Y Chen Y Liu X Xu L, et al. IL-17A promotes fatty acid uptake through the IL-17A/IL-17RA/p-STAT3/FABP4 axis to fuel ovarian cancer growth in an adipocyte-rich microenvironment. Cancer Immunol Immunother 2020;69: 115–126. doi: 10.1007/s00262-019-02445-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zhao J Zhang X Gao T Wang S Hou Y Yuan P, et al. SIK2 enhances synthesis of fatty acid and cholesterol in ovarian cancer cells and tumor growth through PI3K/Akt signaling pathway. Cell Death Dis 2020;11: 25. doi: 10.1038/s41419-019-2221-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nieman KM Kenny HA Penicka CV Ladanyi A Buell-Gutbrod R, et al. Adipocytes promote ovarian cancer metastasis and provide energy for rapid tumor growth. Nat Med 2011;17: 1498–1503. doi: 10.1038/nm.2492. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zhao G, Cardenas H, Matei D. Ovarian cancer-why lipids matter. Cancers (Basel) 2019;11: 1870. doi: 10.3390/cancers11121870. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tan Y, Wang M, Yang K, Chi T, Liao Z, Wei P. PPAR-α modulators as current and potential cancer treatments. Front Oncol 2021;11: 599995. doi: 10.3389/fonc.2021.599995. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Mukherjee A Chiang CY Daifotis HA Nieman KM Fahrmann JF, et al. Adipocyte-induced FABP4 expression in ovarian cancer cells promotes metastasis and mediates carboplatin resistance. Cancer Res 2020;80: 1748–1761. doi: 10.1158/0008-5472.CAN-19-1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Bubancova I Kovarikova H Laco J Ruszova E Dvorak O Palicka V, et al. Next-generation sequencing approach in methylation analysis of HNF1B and GATA4 genes: Searching for biomarkers in ovarian cancer. Int J Mol Sci 2017;18: 474. doi: 10.3390/ijms18020474. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Baranova I Kovarikova H Laco J Sedlakova I Vrbacky F Kovarik D, et al. Identification of a four-gene methylation biomarker panel in high-grade serous ovarian carcinoma. Clin Chem Lab Med 2020;58: 1332–1340. doi: 10.1515/cclm-2019-1319. [DOI] [PubMed] [Google Scholar]
- 28.Abd-Elaziz M, Moriya T, Akahira J, Suzuki T, Sasano H. StAR and progesterone producing enzymes (3beta-hydroxysteroid dehydrogenase and cholesterol side-chain cleavage cytochromes P450) in human epithelial ovarian carcinoma: Immunohistochemical and real-time PCR studies. Cancer Sci 2005;96: 232–239. doi: 10.1111/j.1349-7006.2005.00040.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Park C.S Kim TK Kim HG Kim YJ Jeoung MH Lee WR, et al. Therapeutic targeting of tetraspanin8 in epithelial ovarian cancer invasion and metastasis. Oncogene 2016;35: 4540–4548. doi: 10.1038/onc.2015.520. [DOI] [PubMed] [Google Scholar]
- 30.Yoshida Y Tsunoda T Doi K Fujimoto T Tanaka Y Ota T, et al. ALPK2 is crucial for luminal apoptosis and DNA repair-related gene expression in a three-dimensional colonic-crypt model. Anticancer Res 2012;32: 2301–2308. doi: NODOI [PubMed] [Google Scholar]
- 31.Zhu X, Yan S, Xiao S, Xue M. Knockdown of ALPK2 inhibits the development and progression of ovarian cancer. Cancer Cell Int 2020;20: 267. doi: 10.1186/s12935-020-01347-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Li M, Cheng X, Rong R, Gao Y, Tang X, Chen Y. High expression of fibroblast activation protein (FAP) predicts poor outcome in high-grade serous ovarian cancer. BMC Cancer 2020;20: 1032. doi: 10.1186/s12885-020-07541-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Mhawech-Fauceglia P Yan L Sharifian M Ren X Liu S Kim G, et al. Stromal expression of fibroblast activation protein alpha (FAP) predicts platinum resistance and shorter recurrence in patients with epithelial ovarian cancer. Cancer Microenviron 2015;8: 23–31. doi: 10.1007/s12307-014-0153-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.van Loon K, Huijbers EJM, Griffioen AW. Secreted frizzled-related protein 2: A key player in noncanonical Wnt signaling and tumor angiogenesis. Cancer Metastasis Rev 2021;40: 191–203. doi: 10.1007/s10555-020-09941-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Mariani A Wang C Oberg AL Riska SM Torres M Kumka J, et al. Genes associated with bowel metastases in ovarian cancer. Gynecol Oncol 2019;154: 495–504. doi: 10.1016/j.ygyno.2019.06.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Hanahan D, Coussens LM. Accessories to the crime: Functions of cells recruited to the tumor microenvironment. Cancer Cell 2012;21: 309–322. doi: 10.1016/j.ccr.2012.02.022. [DOI] [PubMed] [Google Scholar]
- 37.Stolk D, van der Vliet HJ, de Gruijl TD, van Kooyk Y, Exley MA. Positive & negative roles of innate effector cells in controlling cancer progression. Front Immunol 2018;9: 1990. doi: 10.3389/fimmu.2018.01990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Taube JM Galon J Sholl LM Rodig SJ Cottrell TR Giraldo NA, et al. Implications of the tumor immune microenvironment for staging and therapeutics. Mod Pathol 2018;31: 214–234. doi: 10.1038/modpathol.2017.156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Hui L, Chen Y. Tumor microenvironment: Sanctuary of the devil. Cancer Lett 2015;368: 7–13. doi: 10.1016/j.canlet.2015.07.039. [DOI] [PubMed] [Google Scholar]
- 40.Fridman WH, Zitvogel L, Sautès-Fridman C, Kroemer G. The immune contexture in cancer prognosis and treatment. Nat Rev Clin Oncol 2017;14: 717–734. doi: 10.1038/nrclinonc.2017.101. [DOI] [PubMed] [Google Scholar]
- 41.Mhaidly R, Mechta-Grigoriou F. Fibroblast heterogeneity in tumor micro-environment: Role in immunosuppression and new therapies. Semin Immunol 2020;48: 101417. doi: 10.1016/j.smim.2020.101417. [DOI] [PubMed] [Google Scholar]
- 42.Gao Q Yang Z Xu S Li X Yang X Jin P, et al. Heterotypic CAF-tumor spheroids promote early peritoneal metastatis of ovarian cancer. J Exp Med 2019;216: 688–703. doi: 10.1084/jem.20180765. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Nersesian S, Glazebrook H, Toulany J, Grantham SR, Boudreau JE. Naturally killing the silent killer: NK cell-based immunotherapy for ovarian cancer. Front Immunol 2019;10: 1782. doi: 10.3389/fimmu.2019.01782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Malchiodi ZX, Weiner LM. Understanding and targeting natural killer cell-cancer-associated fibroblast interactions in pancreatic ductal adenocarcinoma. Cancers (Basel) 2021;13: 405. doi: 10.3390/cancers13030405. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.