Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Identification of Novel Hypoxia Subtypes for Prognosis Based on Machine Learning Algorithms

Identification of Novel Hypoxia Subtypes for Prognosis Based on Machine Learning Algorithms Hindawi Journal of Oncology Volume 2022, Article ID 1508113, 12 pages https://doi.org/10.1155/2022/1508113 Research Article Identification of Novel Hypoxia Subtypes for Prognosis Based on Machine Learning Algorithms 1 2 3 3 4 4 Jiawei Wang, Tuo Li, Songquan Wei, Gengye Zhao, Cong Ye, Qiuping Ma, 4 1 Jinchun Ma , and Xiaoyan Cheng Department of Obstetrics and Gynecology, Nantong Maternity and Child Health Care Hospital Aliated to Nantong University, Nantong 226018, Jiangsu Province, China Department of Endocrinology, Second Aliated Hospital of Naval Medical University, Shanghai 200003, China „e „ird Aliated Hospital of Guangzhou Medical University, Guangzhou 510000, China Department of Obstetrics and Gynecology, Soochow University Aliated Taicang Hospital („e First People’s Hospital of Taicang), Suzhou, Jiangsu 215400, China Correspondence should be addressed to Jinchun Ma; 1452212929@qq.com and Xiaoyan Cheng; 1012770174@qq.com Received 6 July 2022; Revised 21 July 2022; Accepted 25 July 2022; Published 12 September 2022 Academic Editor: Mingjun Zheng Copyright © 2022 Jiawei Wang et al. ƒis is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Objective. A reduced level or tension or the deprivation of oxygen is termed hypoxia. It is common for tumours to outgrow their natural source of nutrients, which causes hypoxia in some tumour regions. Hypoxia aŒects ovarian cancer (OC) in several ways. Methods. In this study, the expression patterns of prognostic hypoxia-related genes were curated, and consensus clustering analyses were performed to determine hypoxia subtypes in OC included in ƒe Cancer Genome Atlas cohort. Two hypoxia-related subtypes were observed and considered for further investigation. ƒe analyses of diŒerentially expressed genes (DEGs), gene ontology, mutation, and immune cell infraction were performed to explore the underlying molecular mechanisms. Results. In total, 377 patients with OC were classi˜ed into two subgroups based on the subtype of hypoxia. ƒe clinical outcome was considerably poor for patients with hypoxia subtype 2. DEG and protein-protein interaction analyses revealed that the expression levels of CLIP2 and SH3PXD2A were low in OC tissues. Immune cell infarction analysis revealed that the subtypes were associated with the tumour microenvironment (TME). Conclusion. Our ˜ndings established the existence of two distinctive, complex, and varied hypoxia subtypes in OC. Findings from the quantitative analysis of hypoxia subtypes in patients improved our un- derstanding of the characteristics of the TME and may facilitate the development of more ešcient treatment regimens. are more likely to develop OC [3, 4]. Currently, the 1. Introduction standard-line therapy for OC comprises cytoreductive Ovarian cancer (OC) is by far the deadliest type of gynae- surgery and chemotherapy (usually paclitaxel and carbo- cological cancer and the ˜fth leading cause of cancer- platin) to remove excess tissue [5]. However, even though associated death among females [1]. Early diagnosis of chemotherapy is occasionally eŒective in treating early-stage OC is challenging owing to the absence of disease-speci˜c cancer, many patients ultimately develop chemoresistance symptoms. Subsequently, a majority of women are di- and experience recurrence [6]. Chemoresistance is driven by agnosed with OC at an advanced stage [2]. Long-term ex- molecular and genetic changes that are unknown, and this posure to steroid hormones contributes to several risk lack of mechanistic insight hinders its prevention and factors. Even though hormone synthesis slows down after prediction. [7, 8] Owing to this, novel therapeutic techniques menopause, women who have been exposed to these hor- are needed to avoid chemoresistance and increase the mones continuously and chronically throughout their lives success rate of therapy. 2 Journal of Oncology Table 1: Hypoxia-related genes. Hypoxia is reportedly associated with chemoresistance via several pathways. Altering the metabolism of cancer cells Gene symbol Gene symbol is one of the ways through which hypoxia may cause che- PSMB6 PSMA1 moresistance in patients with cancer. OC cells, when ex- PSMB5 PSMA8 posed to hypoxia, are subjected to metabolic HIGD1A PSMC6 reprogramming, which alters the glycolytic pathway and EGLN2 PSMD9 enhances resistance to carboplatin [9]. Hypoxia in OC is PSMD1 RBX1 associated with altered levels of circulating microRNAs PSMA7 PSMC5 (miRNAs), and these miRNA expression patterns are linked HIF1AN PSMB1 PSMC2 PSMB8 to a greater risk of OC development. However, research on PSMD3 PSMA4 the mechanisms underlying hypoxia in OC is insufficient. EP300 VHL Immunotherapy is considered a potentially viable VEGFA HIF3A treatment option since it has high degrees of specificity, ELOC WTIP long-term benefits, and minimal adverse effects. Owing to PSMC3 EGLN3 extensive variability, including clinical and pathological PSME4 ARNT parameters, molecular characteristics, and the immune cell UBE2D2 PSMD12 milieu, among other factors, the response rate to immune UBC PSMA6 checkpoint blockade treatment in patients with OC remains PSMD11 EGLN1 as low as 15% [10–12]. Given the heterogeneity of OC, the PSMD10 PSMB3 accurate identification of the specific advantages of im- PSMB10 PSMD8 PSMD5 CUL2 munotherapy in patients is essential for its further ad- ELOB PSMA3 vancement [13]. In this study, two distinct hypoxia subtypes PSME2 PSMC1 were investigated, each characterised by distinct immune CREBBP SEM1 infiltrates and immune responses. Additionally, an immune UBB EPAS1 scoring system was developed for patients with OC, which PSMD6 PSMA5 yielded a thorough understanding of the characteristics of PSMD13 PSMA2 the tumour microenvironment (TME) and prompted the PSMB11 EPO development of efficacious treatment modalities. CA9 PSME3 PSMF1 PSMD7 AJUBA PSMB9 2. Material and Methods UBE2D1 PSME1 2.1. Data Resources. -e Cancer Genome Atlas (TCGA, PSMD14 HIF1A PSMC4 CITED2 https://cancergenome.nih.gov/) project was used to collect PSMB2 UBA52 and process the molecular data of 377 individuals who had PSMB4 UBE2D3 been diagnosed with OC. -e GDC data portal (https:// LIMD1 PSMD4 portal.gdc.cancer.gov) was used to obtain the transcriptomic PSMD2 RPS27A profiles (HTSeq-fragments per kilobase of exon per million PSMB7 mapped fragments (FPKM)) and clinical data for TCGA-OC dataset. -e Ensembl IDs were translated to gene symbols, 2.3. Identification of Differently Expressed Genes (DEGs). and the FPKM values were transformed into transcripts per -e significance analysis built into the empirical Bayes million [14]. techniques used as a part of the limma package was used to detect DEGs. -e cut-off values for selecting the relevant 2.2. Identification of Hypoxia Subtypes Using Consensus DEGs were a P-value <0.01 and a |logFC|> 1. Additionally, Clustering. Using the ConsensusClusterPlus tool, subtypes using the cBioPortal web platform (https://www.cbioportal. of hypoxia were determined. Hypoxia-related genes are org), we created a network of DEGs and their coexpression listed in Table 1. To properly classify OC samples, a con- genes [16,17]. sensus matrix was developed using consensus clustering. Consistent with the partitioning around the medoids al- gorithm and using the Pearson correlation coefficient as the 2.4. Gene Ontology (GO) and Pathway Enrichment Analysis. distance measure, 500 bootstraps were provided, with each -e data were evaluated using functional enrichment comprising patients with OC included in TCGA cohort. -e analysis to confirm the fundamental function of putative number of clusters was determined to be two–eight. Ad- targets. GO is a technique extensively used to annotate genes ditionally, a consensus clustering approach was adopted to with functions, including cellular components (CC), bi- classify the genes immunologically related to the prognosis. ological pathways (BP), and molecular function (MF). ClusterProfiler version 3.18.0 in R was used to examine the -e consistency matrix and the consistency cumulative distribution function were selected as the methods for op- GO function of putative targets and enrich the Kyoto En- timal classification [15]. cyclopedia of Genes and Genomes (KEGG) pathway to gain Journal of Oncology 3 addition, we used the ggplot2 and pheatmap functions of the a deeper understanding of how mRNA contributes to the onset and advancement of cancer. -e boxplot and heatmap R package [21]. were drawn using the ggplot2 and pheatmap functions of R software, respectively [18]. 2.10. Quantitative Reverse-Transcription Polymerase Chain Reaction (qRT-PCR). Total RNA was extracted from para- 2.5. Mutation Analysis. Using TCGA dataset (https://portal. neoplastic and tumour tissues of patients with OC using the gdc.com), we retrieved the RNA-seq expression patterns, TRIzol reagent (Sigma-Aldrich, St. Louis, MO, USA). genetic mutation, and relevant clinical data of 376 patients. Furthermore, RNA from each sample (2 μg) was subjected to -e maftools package of R software was used to retrieve data qRT-PCR using the FastStart Universal SYBR Green on mutations, which were further visualised by the program. Master (Roche, Germany) on an ABI QuantStudio5 Q5 real- Genes with a higher mutational frequency detected in 376 time PCR system (-ermo Fisher Scientific, USA). After- patients in the histogram are demonstrated. ward, we used cDNA as a template in 20 μL reaction volume (containing 10 μL of a PCR mixture, 0.5 μL of reverse and forward primers, 2 μL of cDNA template, and an appropriate 2.6. Protein-Protein Interaction (PPI) Enrichment Analysis. volume of water). We conducted PCR as follows: cycling An enrichment study of PPI was performed using the began with an initial DNA denaturation step at 95 C for 30 s, Metascape database for each gene list that was provided. ° ° ° followed by 45 cycles at 94 C for 15 s, 56 C for 30 s, and 72 C Only the physical interactions observed in the STRING for 20 s. Each sample was assessed in triplicates. Using the −ΔΔCT (with a score greater than 0.132) and BioGRID were con- 2 method, readings from the threshold cycle (CT) were sidered. -e resultant network included the subset of pro- obtained and further standardised to the levels of glycer- teins that physically interacted with at least one other aldehyde 3-phosphate dehydrogenase in each sample. -e member in the list. -e molecular complex detection mRNA expression levels were compared to those in para- (MCODE) algorithm 10 is used to determine the network cancerous tissue controls. -e primer pair sequences cor- components that are densely connected when the number of responding to the target genes are presented in Table 2. proteins in the network ranges between 3 and 500 [19]. 3. Results 2.7. Gene Expression Validation and Survival Analysis of Hub 3.1. Characterisation of Two Distinct Subtypes of OC Hypoxia. Genes. To further confirm the significant role of hub genes -e mRNA expression profiles of hypoxia-associated genes in the pathogenesis and prognosis of OC, we used the Gene in OC tissues were obtained from the TCGA cohort and used Expression Profiling Interactive Analysis (GEPIA) database in this investigation. Patients with OC were clustered using to retrieve information on the expression of these genes and consensus clustering methods in line with the expression their prognostic significance. -e GEPIA database, an in- profiles of prognostic hypoxia-related genes. -e stability of teractive online platform for analysing gene expression, clustering was analysed with k-values ranging from 2 to 8. As contains data on 8,587 normal samples and 9,736 tumour a direct consequence of this, selecting k � 2 was the best samples [20]. alternative. Two distinct immune subtypes, immune subtype 1 (n � 134) and immune subtype 2 (n � 242), were identified 2.8. Cox Analysis. To define the appropriate terms to gen- in patients with OC. Survival analysis revealed that patients with subtype 2 had a poorer outcome (Figure 1(b)). erate the nomogram, both univariate and multivariate Cox regression analyses were used. Using the “forestplot” R package, we generated a forest plot to display the P-value, 3.2. Determination of DEGs in Subtypes. -e limma program HR, and 95% confidence interval (CI) for each variable. We was used to conduct the analysis. -e results demonstrated created a nomogram based on the findings of a multivariate that 375 DEGs, including one gene that was considerably Cox proportional hazard analysis to accurately predict the 1- upregulated and 374 genes that were downregulated. -e year overall recurrence. volcano plot of gene expression profile data in each dataset is presented in Figure 1(c). -e heatmap of the top DEGs is 2.9. Immune Cell Infarction Analysis. We used immunee- presented in Figure 1(d). conv, an R software package that incorporates the two most recent algorithms, ssGSEA and CIBERSORT, to validate the outcomes of the immune score assessment. -ese algorithms 3.3. GO Enrichment Analysis and KEGG Pathways of DEGs. are benchmarked and have distinct advantages. SIGLEC15, -e potential mRNA targets were analysed using the GO TIGIT, PDCD1LG2, HAVCR2, PDCD1, LAG3, CTLA4, and database. -e findings obtained from the MF, CC, and BP of CD274 were determined to produce transcripts that are putative targets clustered, based on the clusterProfiler important for immune checkpoints, and the expression program in R software, revealed a substantial enrichment of levels of these eight genes were measured. R foundation for DEGs in functions such as the modulation of synapse statistical computing (version 4.0.3) was used for imple- structure or activities, modulation of synapse organization, menting the aforementioned analytical techniques. In modulation of small GTPase and mediated signal 4 Journal of Oncology Table 2: Primers of CLIP2, SH3PXD2A and GAPDH. Forward primer sequence Reverse primer sequence Gene (5′-3) (5′-3′) CLIP2 TTAGCGGACAACAGGCTGAC GCTGGAGCTCCTCGATTTCA SH3PXD2A GACTGTACTGCTTAGGGGTGC CCGCTCTCGTTCTTCTCGAT GAPDH AATGGGCAGCCGTTAGGAAA GCCCAATACGACCAAATCAGAG consensus matrix k=2 1.0 0.8 0.6 0.4 0.2 HR = 1.35 (1.03−1.78) P = 0.029 0.0 0 5 10 15 Time G1 G2 (a) (b) USP34 Down−regulation POM121 ATG9A None Up−re CNO gulation T1 USP22 group group SEC24C MCM3AP G1 TBC1D14 2 CDC42BPB G2 PRRC2B ATP1A1 ZNF770 DLG5 0 ASAP2 BCL9L 40 −1 HSPG2 ZFP36L2 −2 −3 DCHS1 F13A1 THBS2 −1 0 1 Log (fold change) (c) (d) Figure 1: Continued. −Log P−value Survival probability Journal of Oncology 5 GO term regulation of synapse structure or activity regulation of synapse organization regulation of small GTPase mediated signal transduction regulation of extracellular matrix organization −log10 (p.adjust) regulation of embryonic development regulation of cell morphogenesis regulation of Ras protein signal transduction positive regulation of transcription of Notch receptor target peptidyl−lysine modification neuron projection guidance morphogenesis of an epithelial sheet Count gastrulation extracellular structure organization extracellular matrix organization establishment or maintenance of cell polarity cell−substrate adhesion cardiac septum development axonogenesis axon guidance Ras protein signal transduction 0.02 0.04 0.06 0.08 Enrichment Ratio (e) Figure 1: (a) A heatmap illustrating the sample clustering when consensus k � 2, based on the expression profile of prognostic immune- related genes. (b) Analysis of survival using the Kaplan-Meier method for the clusters. (c) -e fold change values and the P-adjust parameters were used to construct the volcano plot. Upregulated genes are represented by red dots; downregulated genes are represented by blue dots; non-significant genes are represented by grey dots. (d) -e heatmap of differential gene expression. (e) -e KEGG signalling pathways with significant enrichment illustrate the main biological activities of significant candidate mRNAs. -e gene ratio is indicated by the abscissa, and the enriched pathways are indicated by the ordinate. Analysis of putative mRNA targets using the gene ontology (GO) database. Altered in 102 (100%) of 102 samples. TP53 92 TTN 34 CSMD3 16 FAT3 12 RYR2 12 MYH4 11 TOP2A 11 MUC16 10 APOB 9 FLG 9 Groups Missense_Mutation Frame_Shift_Ins Frame_Shift_Del In_Frame_Ins Splice_Site In_Frame_Del Nonsense_Mutation Multi_Hit (a) Figure 2: Continued. (%) 6 Journal of Oncology Altered in 161 (94.71%) of 170 samples. 0 152 TP53 89 TTN 38 MUC16 13 CSMD3 12 PRUNE2 11 USH2A 11 FLG 10 FLG2 10 KMT2C 10 SYNE1 9 Missense_Mutation Frame_Shift_Del Frame_Shift_Ins In_Frame_Del Nonsense_Mutation Splice_Site In_Frame_Ins Multi_Hit (b) TP53 : [Somatic Mutation Rate: 89.41%]NM_000546 P53_TAD P53 P53_tetramer 0 100 200 300 393 Frame_Shift_Del Nonsense_Mutation Splice_Site Missense_Mutation Frame_Shift_Ins In_Frame_Del (c) Figure 2: An oncoprint depicting the landscape of somatic mutations observed in ovarian cancer (OC) samples from (a) hypoxia subtype 1 and (b) hypoxia subtype 2. (c) Lollipop charts of the mutated TP53 gene; the figure caption depicts the somatic mutation rate; the subheadings depict the name of the somatic mutation. transduction modulation of the extracellular matrix orga- were selected for further analysis. Many hub genes were nisation (Figure 1(e)). observed to be enriched in certain pathways, including the PI3K-Akt signalling pathway (Figure 4(a)). 3.4. Mutation State in Subtypes. We examined how single- nucleotide polymorphisms were distributed among the OC 3.6. Analysis and Validation of Hub Genes. -e screening of samples. Overall, genetic mutations in immune subtypes 1 the GEPIA database revealed that CLIP2 and SH3PXD2A and 2 were observed in 102 and 161 OC samples, respectively displayed substantial differences in expression between tu- (Figures 2(a) and 2(b)). Lollipop charts of the mutated TP53 mour and normal specimens in OC (Figures 4(b) and 4(c)). gene, the figure caption displays the somatic mutation rate, -e findings of GEPIA for overall survival (OS) revealed that and the subheadings depict the name of somatic mutation patients with OC were categorised into high- and low- (Figure 2(c)). expression groups. We confirmed that the overexpression of CLIP2 and SH3PXD2A was associated with a significantly poor OS in patients. 3.5. Establishment of the PPI Network and Module Analysis. -e Metascape database served as the foundation for the establishment of a PPI network of DEGs (Figure 3(a)). -e 3.7. Survival Analysis. -e one-year survival rate for patients two most significant modules, one comprising upregulated with OC may be predicted using the nomogram. We genes and the other comprising downregulated genes, were established a calibration curve for the OS based on the extracted from this PPI network using MCODE. Hub genes nomogram model in the discovery subgroup. -e univariate (%) Journal of Oncology 7 (a) GCN1 MAT2A GPATCH8 HYOU1 THBS2 LRP1 EIF4G1 FBLN2 F13A1 ACACA SNRNP200CNOT1 LAMA5 MMP14 ATXN1L CLIP2 ITGB1 HSPG2 EPB41 NID2 RNF123 KIF1A IGF2 KIF1B ATP1A1 PTPRF SKI IRS1 ERBB2 PTK7 HTT TRIO MTOR AMOT ABL1 CRB2 ULK1 CCDC85C CTDSP2 MYH10 BECN1 AMOTL1 BCOR HMG20A TRRAP TP53BP1 RBL2 ZNF609 SH3PXD2A SH3PXD2B EP400 SETD1B (b) Figure 3: (a) Protein-protein interaction (PPI) network comprising the differentially expressed genes (DEGs) and their co-expressed genes. (b) Hub genes among DEGs. and multivariate analyses showed that CLIP2 and Checkpoint analysis revealed that hypoxia subtype 2 has SH3PXD2A expressions functioned independently as a risk a higher expression in CD274, HAVCR2, ODCD1LG2, and factor for OC (Figures 4(d) and 4(e)). SIGLEC15(Figure 5(b)). Finally, ssGSEA revealed that CLIP2 and SH3PXD2A expression was positively correlated with immune cells, such as Tem and natural killer (NK) cells 3.8. Two Hypoxia Subtypes with Different Immune Infiltrates (Figures 5(c) and 5(d)). and Immune Responses. Using the CIBERSORT algorithm, the landscape of tumour-infiltrating lymphocytes was ob- tained, and 21 types of immune cell profiles of patients with 3.9. Evaluation of Gene Expression in OC. To validate the glioma were determined from TCGA. -e proportion of expression of the CLIP2 and SH3PXD2A genes in the tu- cells such as na¨ıve B cells and CD8 T cells differed sig- mour and nontumour adjacent tissues, the relative mRNA nificantly between the hypoxia subtypes (Figure 5(a)). expression levels of CLIP2 and SH3PXD2A in both tumour 8 Journal of Oncology PI3K−Akt signaling pathway Tight junction ECM−receptor interaction p.adjust extracellular matrix structural constituent 0.020 0.015 phosphoprotein binding 0.010 0.005 phosphatidylinositol 3−kinase binding focal adhesion Counts basement membrane Swr1 complex extracellular structure organization extracellular matrix organization cell−substrate adhesion CLIP2 0.05 0.10 0.15 0.20 0.25 (num (T)=426; num (N)=88) GeneRatio (a) (b) Overall Survival Logrank p=0.012 HR (high)=1.4 p (HR)=0.012 n (high)=212 n (low)=212 0 50 100 150 Months Low CLIP2 TPM SH3PXD2A High CLIP2 TPM (num (T)=426; num (N)=88) (c) (d) 0 102030405060708090 100 Overall Survival Points Logrank p=0.04 C−index : 0.592 (0.542−1) HR (high)=1.3 Age p−value = p < 0.001 35 40 45 50 55 60 65 70 75 80 85 90 p (HR)=0.042 n (high)=212 BLACK AMERICAN INDIAN n (low)=212 Race WHITE ASIAN Total Points 0 102030405060708090 100 110 Linear Predictor −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 50 100 150 1−year survival Pro Months 0.95 0.9 Low SH3PXD2A TPM High SH3PXD2A TPM (e) (f) Figure 4: Continued. 1.0 1.0 Percent survival 0.0 0.2 0.4 0.6 0.8 Percent survival 0.0 0.2 0.4 0.6 0.8 0123456 Journal of Oncology 9 Uni_cox Pvalue Hazard Ratio (95% CI) Mult_cox p.value Hazard Ratio (95% CI) CLIP2 0.00820 1.18641 (1.0452,1.34669) CLIP2 0.62242 1.04446 (0.87842,1.24188) SH3PXD2A 0.00565 1.20239 (1.05524,1.37006) SH3PXD2A 0.47977 1.06167 (0.8993,1.25336) Age 0.00168 1.01948 (1.00728,1.03182) Age 0.02618 1.01693 (1.00199,1.03209) Race 0.01344 0.79663 (0.66521,0.95401) Race 0.02518 0.75928 (0.59661,0.96629) pTNM−stage 0.13286 1.23965 (0.93676,1.64049) pTNM−stage 0.80985 1.04392 (0.73556,1.48154) Grade 0.30641 1.22906 (0.82779,1.82484) Grade 0.24257 1.31434 (0.83102,2.07874) newTumor 0.16939 0.69629 (0.41549,1.16685) newTumor 0.14809 0.67317 (0.39374,1.15091) 0.5 1 1.5 2 0.5 1 1.5 2 Hazard Ratio Hazard Ratio (g) (h) Figure 4: (a) Pop plot of pathway enrichment of hub genes. (b–e) -e level of expression of hub genes and the significance of their predictive value based on data from the Gene Expression Profiling Interactive Analysis (GEPIA) database. (f) Nomograms can predict the 1-year overall survival of patients with OV cancer. (g-h) -e P-value, risk coefficient (HR) and confidence interval analysed by multivariate and univariate Cox regression. * * ns ** ns * nsnsnsnsnsnsns *** ns * ns * * nsns 0.6 0.4 0.2 0.0 category G1 G2 (a) Figure 5: Continued. value B cell naive B cell memory B cell plasma T cell CD8+ T cell CD4+ memory resting T cell CD4+ memory activated T cell follicular helper T cell regulatory (Tregs) T cell gamma delta NK cell resting NK cell activated Monocyte Macrophage M0 Macrophage M1 Macrophage M2 Myeloid dendritic cell resting Myeloid dendritic cell activated Mast cell activated Mast cell resting Eosinophil Neutrophil 10 Journal of Oncology *** *** *** *** Group G1 G2 (b) Tem NK cells NK cells Tem Tcm Eosinophils iDC Tcm P value P value Eosinophils TFH Macrophages Mast cells 0.75 0.75 Mast cells iDC 0.50 0.50 Neutrophils Th17 cells 0.25 0.25 Tgd Macrophages 0.00 0.00 T helper cells Tgd NK CD56dim cells Neutrophils Correlation Correlation DC NK CD56dim cells 0.1 0.1 Th1 cells CD8 T cells 0.2 0.2 CD8 T cells NK CD56bright cells 0.3 0.3 TFH Th2 cells 0.4 0.4 B cells T helper cells TReg DC T cells TReg Th2 cells pDC pDC B cells Cytotoxic cells Th1 cells NK CD56bright cells T cells Th17 cells aDC aDC Cytotoxic cells −0.2 0.0 0.2 0.4 −0.1 0.0 0.1 0.2 0.3 0.4 0.5 Correlation Correlation (c) (d) Figure 5: (a) -e boxplot of immune infarction cells in hypoxia types 1 and 2. (b) -e expression and distribution of immune checkpoint genes in tissues affected by hypoxia types 1 and 2. (c) Barplot of immune cell infarction in high and low CLIP2 expression obtained via single-sample Gene Set Enrichment Analysis (ssGSEA). (d) Barplot of immune cell infarction of high and low SH3PXD2A expression via ssGSEA. and nontumour tissues were determined using qPCR. -e ineffective and have a detrimental effect on patients’ quality average expression level of CLIP2 and SH3PXD2A in OC of life. -us, viable and effective therapies are urgently tissue was considerably less than that in normal tissues needed. [25, 26] A growing body of evidence has illustrated (Figure 6). that the hypoxia microenvironment plays a critical role in immune response and carcinogenesis based on the dys- regulated expression of genes associated with hypoxia. 4. Discussion [27, 28] Most research conducted in the past on hypoxia in OV has focused on a single regulator. Hypoxia- OC is a severe epithelial cancer that predominantly con- induciblefactor-1α, for instance, has been reported to tributes to cancer-associated death among females [22–24]. play an integral role in various processes, including the -e treatment options available for OC are clinically CD274 Immune checkpoint CTLA4 HAVCR2 LAG3 PDCD1 PDCD1LG2 TIGIT SIGLEC15 Journal of Oncology 11 2.0 ** 1.5 1.0 0.5 0 0.0 Normal Tumor Normal Tumor (a) (b) Figure 6: -e expression of CLIP2 and SH3PXD2A determined via polymerase chain reaction (PCR). promotion of OC immunosuppression, tumour metastasis, groups based on their potential response to chemotherapy or and chemoresistance. other immune checkpoint blockades. -us, CLIP2 and In this study, two subtypes of hypoxia were identified SH3PXD2A should be further investigated and could be using consensus clustering analysis, each of which was based novel biomarkers for patients with OC. on the prognostic immune-relevant genes. Particularly, hypoxia subtype 2 displayed a more unfavourable clinical Abbreviations outcome than hypoxia subtype 1. Cancer is a malignant OC: Ovarian cancer neoplasm that may be caused by genetic mutations and TCGA: -e Cancer Genome Atlas variations [29]. Hypoxia subtype 1 was characterised by the DEGs: Differentially expressed genes presence of more prevalent genetic alterations. Alterations in TME: -e tumour microenvironment the expression of several genes, including TP53, have been miRNAs: microRNAs observed to be correlated with the success of immuno- FPKM: Fragments per kilobase of exon per million therapies and exhibit a predictive potential [30]. In the OC mapped fragments samples, the TP53 gene was the first to undergo mutation. In TPM: Transcripts per million hypoxia subtype 1, the TP53 gene was reported to have GO: Gene Ontology a greater incidence of mutations than that in hypoxia CC: Cellular components subtype 2. Our results suggest a difference among the BP: Biological pathways hypoxia subtypes in terms of genetic changes and mutations. MF: Molecular function In this study, we identified 374 genes generated from the KEGG: Kyoto Encyclopedia of Genes and Genomes hypoxia subtypes, which had the potential to influence PPI: Protein-Protein Interaction pathways such as the PI3K-Akt signalling pathway. Hub MCODE: Molecular Complex Detection genes, such as CLIP2 and SH3PXD2A, were selected and GEPIA: Gene Expression Profiling Interactive Analysis used for further investigation. Recently, the expression of CI: Confidence interval SH3PXD2A-AS1 was observed to be related to OC; however, qRT- Quantitative Reverse-Transcription Polymerase the underlying molecular mechanism remains unknown. PCR: Chain Reaction Simultaneously, the absence of SH3PXD2A has been re- NK cells: Natural killer cells ported in the OV area. Nonetheless, further investigation is warranted. -e results of ssGSEA demonstrated that the Data Availability decrease in CLIP2 and SH3PXD2A expression may influence the infiltration levels of immune cells, such as NK cells. -e datasets used and/or analysed during the current study Finally, PCR results confirmed these patterns in OC tissues. are available from the corresponding author upon reason- While this work is a bioinformatics and pcr analysis, more able request. investigation should be performed in clinic for future application. Conflicts of Interest A well-defined hypoxia score has significant benefits over standard prognostic markers used for OC. -erefore, -e authors declare that they have no conflicts of interest. the hypoxia score may be used to compare various hypoxia- modulating components and aid the exploration of how Authors’ Contributions tumour cells interact with the immunological milieu. In addition, it helps stratify patients with OC into various Jiawei Wang and Tuo Li contributed equally to this work. CLIP2 mRNA Levels (relative to Normal) SH3PXD2A mRNA Levels (relative to Normal) 12 Journal of Oncology [16] Q. Wu, B. Zhang, Z. Wang et al., “Integrated bioinformatics Acknowledgments analysis reveals novel key biomarkers and potential candidate small molecule drugs in gastric cancer,” Pathology, Research & -is work was funded by the Nantong Science and Tech- Practice, vol. 215, no. 5, pp. 1038–1048, 2019. nology Program (MS12021009), the Hong Kong Scholars [17] M. E. Ritchie, B. Phipson, D. Wu et al., “Limma powers Program (XJ2016016), and the Shanghai Pujiang Program differential expression analyses for RNA-sequencing and (2020PJD063). microarray studies,” Nucleic Acids Research, vol. 43, no. 7, p. e47, 2015. References [18] L. Zhu, X. Liu, W. Zhang, H. Hu, Q. Wang, and K. Xu, “MTHFD2 is a potential oncogene for its strong association [1] F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre, with poor prognosis and high level of immune infiltrates in and A. Jemal, “Global cancer statistics 2018: GLOBOCAN urothelial carcinomas of bladder,” BMC Cancer, vol. 22, no. 1, estimates of incidence and mortality worldwide for 36 cancers p. 556, 2022. in 185 countries,” CA: A Cancer Journal for Clinicians, vol. 68, [19] Y. Zhou, B. Zhou, L. Pache et al., “Metascape provides a bi- no. 6, pp. 394–424, 2018. ologist-oriented resource for the analysis of systems-level [2] D. Badgwell and R. C. Bast Jr, “Early detection of ovarian datasets,” Nature Communications, vol. 10, no. 1, p. 1523, cancer,” Disease Markers, vol. 23, no. 5-6, pp. 397–410, 2007. [3] A. L. Shafrir, A. Babic, M. Gates Kuliszewski et al., “Estrogen [20] Z. Tang, C. Li, B. Kang, G. Gao, C. Li, and Z. Zhang, “GEPIA: receptor-β expression of ovarian tumors and its association a web server for cancer and normal gene expression profiling with ovarian cancer risk factors,” Cancer Epidemiology, Bio- and interactive analyses,” Nucleic Acids Research, vol. 45, markers & Prevention, vol. 29, no. 11, pp. 2211–2219, 2020. no. 1, pp. W98–W102, 2017. [4] W. Wang, Q. Wu, Z. Wang et al., “Development of a prog- [21] A. M. Newman, C. L. Liu, M. R. Green et al., “Robust enu- nostic model for ovarian cancer patients based on novel meration of cell subsets from tissue expression profiles,” immune microenvironment related genes,” Frontiers in On- Nature Methods, vol. 12, no. 5, pp. 453–457, 2015. cology, vol. 11, Article ID 647273, 2021. [22] D. Matei and K. P. Nephew, “Epigenetic attire in ovarian [5] B. Orr and R. P. Edwards, “Diagnosis and treatment of ovarian cancer: the emperor’s new clothes,” Cancer Research, vol. 80, cancer,” Hematology-Oncology Clinics of North America, no. 18, pp. 3775–3785, 2020. vol. 32, no. 6, pp. 943–964, 2018. [23] -e Lancet Oncology, “Ovarian cancer: breaking the silence,” [6] INVALID CITATION !!! 6, https://community.endnote.com/ +e Lancet Oncology, vol. 13, no. 2, p. 111, 2012. t/invalid-citation/259536. [24] P. B. Ottevanger, “Ovarian cancer stem cells more questions [7] S. B. Hooks, P. Callihan, M. K. Altman, J. H. Hurst, M. W. Ali, than answers,” Seminars in Cancer Biology, vol. 44, pp. 67–71, and M. M. Murph, “Regulators of G-protein signaling RGS10 and RGS17 regulate chemoresistance in ovarian cancer cells,” [25] J. Liu and U. A. Matulonis, “New strategies in ovarian cancer: Molecular Cancer, vol. 9, no. 1, p. 289, 2010. translating the molecular complexity of ovarian cancer into [8] R. C. Bast, B. Hennessy, and G. B. Mills, “-e biology of treatment advances,” Clinical Cancer Research, vol. 20, no. 20, ovarian cancer: new opportunities for translation,” Nature pp. 5150–5156, 2014. Reviews Cancer, vol. 9, no. 6, pp. 415–428, 2009. [26] F. Ricci, M. Broggini, and G. Damia, “Revisiting ovarian [9] M. Alharbi, A. Lai, S. Sharma et al., “Extracellular vesicle cancer preclinical models: implications for a better man- transmission of chemoresistance to ovarian cancer cells is agement of the disease,” Cancer Treatment Reviews, vol. 39, associated with hypoxia-induced expression of glycolytic no. 6, pp. 561–568, 2013. pathway proteins, and prediction of epithelial ovarian cancer [27] K. DePeaux and G. M. Delgoffe, “Metabolic barriers to cancer disease recurrence,” Cancers, vol. 13, no. 14, p. 3388, 2021. immunotherapy,” Nature Reviews Immunology, vol. 21, no. 12, [10] Y. Gao, L. Chen, G. Cai et al., “Heterogeneity of immune pp. 785–797, 2021. microenvironment in ovarian cancer and its clinical signifi- [28] X. Jing, F. Yang, C. Shao et al., “Role of hypoxia in cancer cance: a retrospective study,” OncoImmunology, vol. 9, no. 1, therapy by regulating the tumor microenvironment,” Mo- Article ID 1760067, 2020. lecular Cancer, vol. 18, no. 1, p. 157, 2019. [11] R. S. Riley, C. H. June, R. Langer, and M. J. Mitchell, “Delivery [29] K. L. Kanchi, K. J. Johnson, C. Lu et al., “Integrated analysis of technologies for cancer immunotherapy,” Nature Reviews germline and somatic variants in ovarian cancer,” Nature Drug Discovery, vol. 18, no. 3, pp. 175–196, 2019. Communications, vol. 5, no. 1, p. 3156, 2014. [12] R. Derynck, S. J. Turley, and R. J. Akhurst, “TGFβ biology in [30] H. Sun, S. Y. Liu, J. Y. Zhou et al., “Specific TP53 subtype as cancer progression and immunotherapy,” Nature Reviews biomarker for immune checkpoint inhibitors in lung ade- Clinical Oncology, vol. 18, no. 1, pp. 9–34, 2021. nocarcinoma,” EBioMedicine, vol. 60, Article ID 102990, 2020. [13] D. Hao, J. Liu, M. Chen et al., “Immunogenomic analyses of [31] N. Li and X. Zhan, “Identification of clinical trait-related advanced serous ovarian cancer reveal immune score is lncRNA and mRNA biomarkers with weighted gene co- a strong prognostic factor and an indicator of chemo- expression network analysis as useful tool for personalized sensitivity,” Clinical Cancer Research, vol. 24, no. 15, medicine in ovarian cancer,” +e EPMA Journal, vol. 10, no. 3, pp. 3560–3571, 2018. pp. 273–290, 2019. [14] W. Hu, G. Wang, Y. Chen, L. B. Yarmus, B. Liu, and Y. Wan, “Coupled immune stratification and identification of thera- peutic candidates in patients with lung adenocarcinoma,” Aging, vol. 12, no. 16, pp. 16514–16538, 2020. [15] W. Qi and Q. Zhang, “Identification and validation of im- mune molecular subtypes and immune landscape based on colon cancer cohort,” Frontiers of Medicine, vol. 9, Article ID 827695, 2022. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Oncology Hindawi Publishing Corporation

Identification of Novel Hypoxia Subtypes for Prognosis Based on Machine Learning Algorithms

Identification of Novel Hypoxia Subtypes for Prognosis Based on Machine Learning Algorithms

Abstract

<i>Objective</i>. A reduced level or tension or the deprivation of oxygen is termed hypoxia. It is common for tumours to outgrow their natural source of nutrients, which causes hypoxia in some tumour regions. Hypoxia affects ovarian cancer (OC) in several ways. <i>Methods</i>. In this study, the expression patterns of prognostic hypoxia-related genes were curated, and consensus clustering analyses were performed to determine hypoxia subtypes in OC included in The Cancer Genome Atlas cohort. Two hypoxia-related subtypes were observed and considered for further investigation. The analyses of differentially expressed genes (DEGs), gene ontology, mutation, and immune cell infraction were performed to explore the underlying molecular mechanisms. <i>Results</i>. In total, 377 patients with OC were classified into two subgroups based on the subtype of hypoxia. The clinical outcome was considerably poor for patients with hypoxia subtype 2. DEG and protein-protein interaction analyses revealed that the expression levels of <i>CLIP2</i> and <i>SH3PXD2A</i> were low in OC tissues. Immune cell infarction analysis revealed that the subtypes were associated with the tumour microenvironment (TME). <i>Conclusion</i>. Our findings established the existence of two distinctive, complex, and varied hypoxia subtypes in OC. Findings from the quantitative analysis of hypoxia subtypes in patients improved our understanding of the characteristics of the TME and may facilitate the development of more efficient treatment regimens.

Loading next page...
 
/lp/hindawi-publishing-corporation/identification-of-novel-hypoxia-subtypes-for-prognosis-based-on-lDrGPnVVOl
Publisher
Hindawi Publishing Corporation
ISSN
1687-8450
eISSN
1687-8469
DOI
10.1155/2022/1508113
Publisher site
See Article on Publisher Site

Abstract

Hindawi Journal of Oncology Volume 2022, Article ID 1508113, 12 pages https://doi.org/10.1155/2022/1508113 Research Article Identification of Novel Hypoxia Subtypes for Prognosis Based on Machine Learning Algorithms 1 2 3 3 4 4 Jiawei Wang, Tuo Li, Songquan Wei, Gengye Zhao, Cong Ye, Qiuping Ma, 4 1 Jinchun Ma , and Xiaoyan Cheng Department of Obstetrics and Gynecology, Nantong Maternity and Child Health Care Hospital Aliated to Nantong University, Nantong 226018, Jiangsu Province, China Department of Endocrinology, Second Aliated Hospital of Naval Medical University, Shanghai 200003, China „e „ird Aliated Hospital of Guangzhou Medical University, Guangzhou 510000, China Department of Obstetrics and Gynecology, Soochow University Aliated Taicang Hospital („e First People’s Hospital of Taicang), Suzhou, Jiangsu 215400, China Correspondence should be addressed to Jinchun Ma; 1452212929@qq.com and Xiaoyan Cheng; 1012770174@qq.com Received 6 July 2022; Revised 21 July 2022; Accepted 25 July 2022; Published 12 September 2022 Academic Editor: Mingjun Zheng Copyright © 2022 Jiawei Wang et al. ƒis is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Objective. A reduced level or tension or the deprivation of oxygen is termed hypoxia. It is common for tumours to outgrow their natural source of nutrients, which causes hypoxia in some tumour regions. Hypoxia aŒects ovarian cancer (OC) in several ways. Methods. In this study, the expression patterns of prognostic hypoxia-related genes were curated, and consensus clustering analyses were performed to determine hypoxia subtypes in OC included in ƒe Cancer Genome Atlas cohort. Two hypoxia-related subtypes were observed and considered for further investigation. ƒe analyses of diŒerentially expressed genes (DEGs), gene ontology, mutation, and immune cell infraction were performed to explore the underlying molecular mechanisms. Results. In total, 377 patients with OC were classi˜ed into two subgroups based on the subtype of hypoxia. ƒe clinical outcome was considerably poor for patients with hypoxia subtype 2. DEG and protein-protein interaction analyses revealed that the expression levels of CLIP2 and SH3PXD2A were low in OC tissues. Immune cell infarction analysis revealed that the subtypes were associated with the tumour microenvironment (TME). Conclusion. Our ˜ndings established the existence of two distinctive, complex, and varied hypoxia subtypes in OC. Findings from the quantitative analysis of hypoxia subtypes in patients improved our un- derstanding of the characteristics of the TME and may facilitate the development of more ešcient treatment regimens. are more likely to develop OC [3, 4]. Currently, the 1. Introduction standard-line therapy for OC comprises cytoreductive Ovarian cancer (OC) is by far the deadliest type of gynae- surgery and chemotherapy (usually paclitaxel and carbo- cological cancer and the ˜fth leading cause of cancer- platin) to remove excess tissue [5]. However, even though associated death among females [1]. Early diagnosis of chemotherapy is occasionally eŒective in treating early-stage OC is challenging owing to the absence of disease-speci˜c cancer, many patients ultimately develop chemoresistance symptoms. Subsequently, a majority of women are di- and experience recurrence [6]. Chemoresistance is driven by agnosed with OC at an advanced stage [2]. Long-term ex- molecular and genetic changes that are unknown, and this posure to steroid hormones contributes to several risk lack of mechanistic insight hinders its prevention and factors. Even though hormone synthesis slows down after prediction. [7, 8] Owing to this, novel therapeutic techniques menopause, women who have been exposed to these hor- are needed to avoid chemoresistance and increase the mones continuously and chronically throughout their lives success rate of therapy. 2 Journal of Oncology Table 1: Hypoxia-related genes. Hypoxia is reportedly associated with chemoresistance via several pathways. Altering the metabolism of cancer cells Gene symbol Gene symbol is one of the ways through which hypoxia may cause che- PSMB6 PSMA1 moresistance in patients with cancer. OC cells, when ex- PSMB5 PSMA8 posed to hypoxia, are subjected to metabolic HIGD1A PSMC6 reprogramming, which alters the glycolytic pathway and EGLN2 PSMD9 enhances resistance to carboplatin [9]. Hypoxia in OC is PSMD1 RBX1 associated with altered levels of circulating microRNAs PSMA7 PSMC5 (miRNAs), and these miRNA expression patterns are linked HIF1AN PSMB1 PSMC2 PSMB8 to a greater risk of OC development. However, research on PSMD3 PSMA4 the mechanisms underlying hypoxia in OC is insufficient. EP300 VHL Immunotherapy is considered a potentially viable VEGFA HIF3A treatment option since it has high degrees of specificity, ELOC WTIP long-term benefits, and minimal adverse effects. Owing to PSMC3 EGLN3 extensive variability, including clinical and pathological PSME4 ARNT parameters, molecular characteristics, and the immune cell UBE2D2 PSMD12 milieu, among other factors, the response rate to immune UBC PSMA6 checkpoint blockade treatment in patients with OC remains PSMD11 EGLN1 as low as 15% [10–12]. Given the heterogeneity of OC, the PSMD10 PSMB3 accurate identification of the specific advantages of im- PSMB10 PSMD8 PSMD5 CUL2 munotherapy in patients is essential for its further ad- ELOB PSMA3 vancement [13]. In this study, two distinct hypoxia subtypes PSME2 PSMC1 were investigated, each characterised by distinct immune CREBBP SEM1 infiltrates and immune responses. Additionally, an immune UBB EPAS1 scoring system was developed for patients with OC, which PSMD6 PSMA5 yielded a thorough understanding of the characteristics of PSMD13 PSMA2 the tumour microenvironment (TME) and prompted the PSMB11 EPO development of efficacious treatment modalities. CA9 PSME3 PSMF1 PSMD7 AJUBA PSMB9 2. Material and Methods UBE2D1 PSME1 2.1. Data Resources. -e Cancer Genome Atlas (TCGA, PSMD14 HIF1A PSMC4 CITED2 https://cancergenome.nih.gov/) project was used to collect PSMB2 UBA52 and process the molecular data of 377 individuals who had PSMB4 UBE2D3 been diagnosed with OC. -e GDC data portal (https:// LIMD1 PSMD4 portal.gdc.cancer.gov) was used to obtain the transcriptomic PSMD2 RPS27A profiles (HTSeq-fragments per kilobase of exon per million PSMB7 mapped fragments (FPKM)) and clinical data for TCGA-OC dataset. -e Ensembl IDs were translated to gene symbols, 2.3. Identification of Differently Expressed Genes (DEGs). and the FPKM values were transformed into transcripts per -e significance analysis built into the empirical Bayes million [14]. techniques used as a part of the limma package was used to detect DEGs. -e cut-off values for selecting the relevant 2.2. Identification of Hypoxia Subtypes Using Consensus DEGs were a P-value <0.01 and a |logFC|> 1. Additionally, Clustering. Using the ConsensusClusterPlus tool, subtypes using the cBioPortal web platform (https://www.cbioportal. of hypoxia were determined. Hypoxia-related genes are org), we created a network of DEGs and their coexpression listed in Table 1. To properly classify OC samples, a con- genes [16,17]. sensus matrix was developed using consensus clustering. Consistent with the partitioning around the medoids al- gorithm and using the Pearson correlation coefficient as the 2.4. Gene Ontology (GO) and Pathway Enrichment Analysis. distance measure, 500 bootstraps were provided, with each -e data were evaluated using functional enrichment comprising patients with OC included in TCGA cohort. -e analysis to confirm the fundamental function of putative number of clusters was determined to be two–eight. Ad- targets. GO is a technique extensively used to annotate genes ditionally, a consensus clustering approach was adopted to with functions, including cellular components (CC), bi- classify the genes immunologically related to the prognosis. ological pathways (BP), and molecular function (MF). ClusterProfiler version 3.18.0 in R was used to examine the -e consistency matrix and the consistency cumulative distribution function were selected as the methods for op- GO function of putative targets and enrich the Kyoto En- timal classification [15]. cyclopedia of Genes and Genomes (KEGG) pathway to gain Journal of Oncology 3 addition, we used the ggplot2 and pheatmap functions of the a deeper understanding of how mRNA contributes to the onset and advancement of cancer. -e boxplot and heatmap R package [21]. were drawn using the ggplot2 and pheatmap functions of R software, respectively [18]. 2.10. Quantitative Reverse-Transcription Polymerase Chain Reaction (qRT-PCR). Total RNA was extracted from para- 2.5. Mutation Analysis. Using TCGA dataset (https://portal. neoplastic and tumour tissues of patients with OC using the gdc.com), we retrieved the RNA-seq expression patterns, TRIzol reagent (Sigma-Aldrich, St. Louis, MO, USA). genetic mutation, and relevant clinical data of 376 patients. Furthermore, RNA from each sample (2 μg) was subjected to -e maftools package of R software was used to retrieve data qRT-PCR using the FastStart Universal SYBR Green on mutations, which were further visualised by the program. Master (Roche, Germany) on an ABI QuantStudio5 Q5 real- Genes with a higher mutational frequency detected in 376 time PCR system (-ermo Fisher Scientific, USA). After- patients in the histogram are demonstrated. ward, we used cDNA as a template in 20 μL reaction volume (containing 10 μL of a PCR mixture, 0.5 μL of reverse and forward primers, 2 μL of cDNA template, and an appropriate 2.6. Protein-Protein Interaction (PPI) Enrichment Analysis. volume of water). We conducted PCR as follows: cycling An enrichment study of PPI was performed using the began with an initial DNA denaturation step at 95 C for 30 s, Metascape database for each gene list that was provided. ° ° ° followed by 45 cycles at 94 C for 15 s, 56 C for 30 s, and 72 C Only the physical interactions observed in the STRING for 20 s. Each sample was assessed in triplicates. Using the −ΔΔCT (with a score greater than 0.132) and BioGRID were con- 2 method, readings from the threshold cycle (CT) were sidered. -e resultant network included the subset of pro- obtained and further standardised to the levels of glycer- teins that physically interacted with at least one other aldehyde 3-phosphate dehydrogenase in each sample. -e member in the list. -e molecular complex detection mRNA expression levels were compared to those in para- (MCODE) algorithm 10 is used to determine the network cancerous tissue controls. -e primer pair sequences cor- components that are densely connected when the number of responding to the target genes are presented in Table 2. proteins in the network ranges between 3 and 500 [19]. 3. Results 2.7. Gene Expression Validation and Survival Analysis of Hub 3.1. Characterisation of Two Distinct Subtypes of OC Hypoxia. Genes. To further confirm the significant role of hub genes -e mRNA expression profiles of hypoxia-associated genes in the pathogenesis and prognosis of OC, we used the Gene in OC tissues were obtained from the TCGA cohort and used Expression Profiling Interactive Analysis (GEPIA) database in this investigation. Patients with OC were clustered using to retrieve information on the expression of these genes and consensus clustering methods in line with the expression their prognostic significance. -e GEPIA database, an in- profiles of prognostic hypoxia-related genes. -e stability of teractive online platform for analysing gene expression, clustering was analysed with k-values ranging from 2 to 8. As contains data on 8,587 normal samples and 9,736 tumour a direct consequence of this, selecting k � 2 was the best samples [20]. alternative. Two distinct immune subtypes, immune subtype 1 (n � 134) and immune subtype 2 (n � 242), were identified 2.8. Cox Analysis. To define the appropriate terms to gen- in patients with OC. Survival analysis revealed that patients with subtype 2 had a poorer outcome (Figure 1(b)). erate the nomogram, both univariate and multivariate Cox regression analyses were used. Using the “forestplot” R package, we generated a forest plot to display the P-value, 3.2. Determination of DEGs in Subtypes. -e limma program HR, and 95% confidence interval (CI) for each variable. We was used to conduct the analysis. -e results demonstrated created a nomogram based on the findings of a multivariate that 375 DEGs, including one gene that was considerably Cox proportional hazard analysis to accurately predict the 1- upregulated and 374 genes that were downregulated. -e year overall recurrence. volcano plot of gene expression profile data in each dataset is presented in Figure 1(c). -e heatmap of the top DEGs is 2.9. Immune Cell Infarction Analysis. We used immunee- presented in Figure 1(d). conv, an R software package that incorporates the two most recent algorithms, ssGSEA and CIBERSORT, to validate the outcomes of the immune score assessment. -ese algorithms 3.3. GO Enrichment Analysis and KEGG Pathways of DEGs. are benchmarked and have distinct advantages. SIGLEC15, -e potential mRNA targets were analysed using the GO TIGIT, PDCD1LG2, HAVCR2, PDCD1, LAG3, CTLA4, and database. -e findings obtained from the MF, CC, and BP of CD274 were determined to produce transcripts that are putative targets clustered, based on the clusterProfiler important for immune checkpoints, and the expression program in R software, revealed a substantial enrichment of levels of these eight genes were measured. R foundation for DEGs in functions such as the modulation of synapse statistical computing (version 4.0.3) was used for imple- structure or activities, modulation of synapse organization, menting the aforementioned analytical techniques. In modulation of small GTPase and mediated signal 4 Journal of Oncology Table 2: Primers of CLIP2, SH3PXD2A and GAPDH. Forward primer sequence Reverse primer sequence Gene (5′-3) (5′-3′) CLIP2 TTAGCGGACAACAGGCTGAC GCTGGAGCTCCTCGATTTCA SH3PXD2A GACTGTACTGCTTAGGGGTGC CCGCTCTCGTTCTTCTCGAT GAPDH AATGGGCAGCCGTTAGGAAA GCCCAATACGACCAAATCAGAG consensus matrix k=2 1.0 0.8 0.6 0.4 0.2 HR = 1.35 (1.03−1.78) P = 0.029 0.0 0 5 10 15 Time G1 G2 (a) (b) USP34 Down−regulation POM121 ATG9A None Up−re CNO gulation T1 USP22 group group SEC24C MCM3AP G1 TBC1D14 2 CDC42BPB G2 PRRC2B ATP1A1 ZNF770 DLG5 0 ASAP2 BCL9L 40 −1 HSPG2 ZFP36L2 −2 −3 DCHS1 F13A1 THBS2 −1 0 1 Log (fold change) (c) (d) Figure 1: Continued. −Log P−value Survival probability Journal of Oncology 5 GO term regulation of synapse structure or activity regulation of synapse organization regulation of small GTPase mediated signal transduction regulation of extracellular matrix organization −log10 (p.adjust) regulation of embryonic development regulation of cell morphogenesis regulation of Ras protein signal transduction positive regulation of transcription of Notch receptor target peptidyl−lysine modification neuron projection guidance morphogenesis of an epithelial sheet Count gastrulation extracellular structure organization extracellular matrix organization establishment or maintenance of cell polarity cell−substrate adhesion cardiac septum development axonogenesis axon guidance Ras protein signal transduction 0.02 0.04 0.06 0.08 Enrichment Ratio (e) Figure 1: (a) A heatmap illustrating the sample clustering when consensus k � 2, based on the expression profile of prognostic immune- related genes. (b) Analysis of survival using the Kaplan-Meier method for the clusters. (c) -e fold change values and the P-adjust parameters were used to construct the volcano plot. Upregulated genes are represented by red dots; downregulated genes are represented by blue dots; non-significant genes are represented by grey dots. (d) -e heatmap of differential gene expression. (e) -e KEGG signalling pathways with significant enrichment illustrate the main biological activities of significant candidate mRNAs. -e gene ratio is indicated by the abscissa, and the enriched pathways are indicated by the ordinate. Analysis of putative mRNA targets using the gene ontology (GO) database. Altered in 102 (100%) of 102 samples. TP53 92 TTN 34 CSMD3 16 FAT3 12 RYR2 12 MYH4 11 TOP2A 11 MUC16 10 APOB 9 FLG 9 Groups Missense_Mutation Frame_Shift_Ins Frame_Shift_Del In_Frame_Ins Splice_Site In_Frame_Del Nonsense_Mutation Multi_Hit (a) Figure 2: Continued. (%) 6 Journal of Oncology Altered in 161 (94.71%) of 170 samples. 0 152 TP53 89 TTN 38 MUC16 13 CSMD3 12 PRUNE2 11 USH2A 11 FLG 10 FLG2 10 KMT2C 10 SYNE1 9 Missense_Mutation Frame_Shift_Del Frame_Shift_Ins In_Frame_Del Nonsense_Mutation Splice_Site In_Frame_Ins Multi_Hit (b) TP53 : [Somatic Mutation Rate: 89.41%]NM_000546 P53_TAD P53 P53_tetramer 0 100 200 300 393 Frame_Shift_Del Nonsense_Mutation Splice_Site Missense_Mutation Frame_Shift_Ins In_Frame_Del (c) Figure 2: An oncoprint depicting the landscape of somatic mutations observed in ovarian cancer (OC) samples from (a) hypoxia subtype 1 and (b) hypoxia subtype 2. (c) Lollipop charts of the mutated TP53 gene; the figure caption depicts the somatic mutation rate; the subheadings depict the name of the somatic mutation. transduction modulation of the extracellular matrix orga- were selected for further analysis. Many hub genes were nisation (Figure 1(e)). observed to be enriched in certain pathways, including the PI3K-Akt signalling pathway (Figure 4(a)). 3.4. Mutation State in Subtypes. We examined how single- nucleotide polymorphisms were distributed among the OC 3.6. Analysis and Validation of Hub Genes. -e screening of samples. Overall, genetic mutations in immune subtypes 1 the GEPIA database revealed that CLIP2 and SH3PXD2A and 2 were observed in 102 and 161 OC samples, respectively displayed substantial differences in expression between tu- (Figures 2(a) and 2(b)). Lollipop charts of the mutated TP53 mour and normal specimens in OC (Figures 4(b) and 4(c)). gene, the figure caption displays the somatic mutation rate, -e findings of GEPIA for overall survival (OS) revealed that and the subheadings depict the name of somatic mutation patients with OC were categorised into high- and low- (Figure 2(c)). expression groups. We confirmed that the overexpression of CLIP2 and SH3PXD2A was associated with a significantly poor OS in patients. 3.5. Establishment of the PPI Network and Module Analysis. -e Metascape database served as the foundation for the establishment of a PPI network of DEGs (Figure 3(a)). -e 3.7. Survival Analysis. -e one-year survival rate for patients two most significant modules, one comprising upregulated with OC may be predicted using the nomogram. We genes and the other comprising downregulated genes, were established a calibration curve for the OS based on the extracted from this PPI network using MCODE. Hub genes nomogram model in the discovery subgroup. -e univariate (%) Journal of Oncology 7 (a) GCN1 MAT2A GPATCH8 HYOU1 THBS2 LRP1 EIF4G1 FBLN2 F13A1 ACACA SNRNP200CNOT1 LAMA5 MMP14 ATXN1L CLIP2 ITGB1 HSPG2 EPB41 NID2 RNF123 KIF1A IGF2 KIF1B ATP1A1 PTPRF SKI IRS1 ERBB2 PTK7 HTT TRIO MTOR AMOT ABL1 CRB2 ULK1 CCDC85C CTDSP2 MYH10 BECN1 AMOTL1 BCOR HMG20A TRRAP TP53BP1 RBL2 ZNF609 SH3PXD2A SH3PXD2B EP400 SETD1B (b) Figure 3: (a) Protein-protein interaction (PPI) network comprising the differentially expressed genes (DEGs) and their co-expressed genes. (b) Hub genes among DEGs. and multivariate analyses showed that CLIP2 and Checkpoint analysis revealed that hypoxia subtype 2 has SH3PXD2A expressions functioned independently as a risk a higher expression in CD274, HAVCR2, ODCD1LG2, and factor for OC (Figures 4(d) and 4(e)). SIGLEC15(Figure 5(b)). Finally, ssGSEA revealed that CLIP2 and SH3PXD2A expression was positively correlated with immune cells, such as Tem and natural killer (NK) cells 3.8. Two Hypoxia Subtypes with Different Immune Infiltrates (Figures 5(c) and 5(d)). and Immune Responses. Using the CIBERSORT algorithm, the landscape of tumour-infiltrating lymphocytes was ob- tained, and 21 types of immune cell profiles of patients with 3.9. Evaluation of Gene Expression in OC. To validate the glioma were determined from TCGA. -e proportion of expression of the CLIP2 and SH3PXD2A genes in the tu- cells such as na¨ıve B cells and CD8 T cells differed sig- mour and nontumour adjacent tissues, the relative mRNA nificantly between the hypoxia subtypes (Figure 5(a)). expression levels of CLIP2 and SH3PXD2A in both tumour 8 Journal of Oncology PI3K−Akt signaling pathway Tight junction ECM−receptor interaction p.adjust extracellular matrix structural constituent 0.020 0.015 phosphoprotein binding 0.010 0.005 phosphatidylinositol 3−kinase binding focal adhesion Counts basement membrane Swr1 complex extracellular structure organization extracellular matrix organization cell−substrate adhesion CLIP2 0.05 0.10 0.15 0.20 0.25 (num (T)=426; num (N)=88) GeneRatio (a) (b) Overall Survival Logrank p=0.012 HR (high)=1.4 p (HR)=0.012 n (high)=212 n (low)=212 0 50 100 150 Months Low CLIP2 TPM SH3PXD2A High CLIP2 TPM (num (T)=426; num (N)=88) (c) (d) 0 102030405060708090 100 Overall Survival Points Logrank p=0.04 C−index : 0.592 (0.542−1) HR (high)=1.3 Age p−value = p < 0.001 35 40 45 50 55 60 65 70 75 80 85 90 p (HR)=0.042 n (high)=212 BLACK AMERICAN INDIAN n (low)=212 Race WHITE ASIAN Total Points 0 102030405060708090 100 110 Linear Predictor −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 0 50 100 150 1−year survival Pro Months 0.95 0.9 Low SH3PXD2A TPM High SH3PXD2A TPM (e) (f) Figure 4: Continued. 1.0 1.0 Percent survival 0.0 0.2 0.4 0.6 0.8 Percent survival 0.0 0.2 0.4 0.6 0.8 0123456 Journal of Oncology 9 Uni_cox Pvalue Hazard Ratio (95% CI) Mult_cox p.value Hazard Ratio (95% CI) CLIP2 0.00820 1.18641 (1.0452,1.34669) CLIP2 0.62242 1.04446 (0.87842,1.24188) SH3PXD2A 0.00565 1.20239 (1.05524,1.37006) SH3PXD2A 0.47977 1.06167 (0.8993,1.25336) Age 0.00168 1.01948 (1.00728,1.03182) Age 0.02618 1.01693 (1.00199,1.03209) Race 0.01344 0.79663 (0.66521,0.95401) Race 0.02518 0.75928 (0.59661,0.96629) pTNM−stage 0.13286 1.23965 (0.93676,1.64049) pTNM−stage 0.80985 1.04392 (0.73556,1.48154) Grade 0.30641 1.22906 (0.82779,1.82484) Grade 0.24257 1.31434 (0.83102,2.07874) newTumor 0.16939 0.69629 (0.41549,1.16685) newTumor 0.14809 0.67317 (0.39374,1.15091) 0.5 1 1.5 2 0.5 1 1.5 2 Hazard Ratio Hazard Ratio (g) (h) Figure 4: (a) Pop plot of pathway enrichment of hub genes. (b–e) -e level of expression of hub genes and the significance of their predictive value based on data from the Gene Expression Profiling Interactive Analysis (GEPIA) database. (f) Nomograms can predict the 1-year overall survival of patients with OV cancer. (g-h) -e P-value, risk coefficient (HR) and confidence interval analysed by multivariate and univariate Cox regression. * * ns ** ns * nsnsnsnsnsnsns *** ns * ns * * nsns 0.6 0.4 0.2 0.0 category G1 G2 (a) Figure 5: Continued. value B cell naive B cell memory B cell plasma T cell CD8+ T cell CD4+ memory resting T cell CD4+ memory activated T cell follicular helper T cell regulatory (Tregs) T cell gamma delta NK cell resting NK cell activated Monocyte Macrophage M0 Macrophage M1 Macrophage M2 Myeloid dendritic cell resting Myeloid dendritic cell activated Mast cell activated Mast cell resting Eosinophil Neutrophil 10 Journal of Oncology *** *** *** *** Group G1 G2 (b) Tem NK cells NK cells Tem Tcm Eosinophils iDC Tcm P value P value Eosinophils TFH Macrophages Mast cells 0.75 0.75 Mast cells iDC 0.50 0.50 Neutrophils Th17 cells 0.25 0.25 Tgd Macrophages 0.00 0.00 T helper cells Tgd NK CD56dim cells Neutrophils Correlation Correlation DC NK CD56dim cells 0.1 0.1 Th1 cells CD8 T cells 0.2 0.2 CD8 T cells NK CD56bright cells 0.3 0.3 TFH Th2 cells 0.4 0.4 B cells T helper cells TReg DC T cells TReg Th2 cells pDC pDC B cells Cytotoxic cells Th1 cells NK CD56bright cells T cells Th17 cells aDC aDC Cytotoxic cells −0.2 0.0 0.2 0.4 −0.1 0.0 0.1 0.2 0.3 0.4 0.5 Correlation Correlation (c) (d) Figure 5: (a) -e boxplot of immune infarction cells in hypoxia types 1 and 2. (b) -e expression and distribution of immune checkpoint genes in tissues affected by hypoxia types 1 and 2. (c) Barplot of immune cell infarction in high and low CLIP2 expression obtained via single-sample Gene Set Enrichment Analysis (ssGSEA). (d) Barplot of immune cell infarction of high and low SH3PXD2A expression via ssGSEA. and nontumour tissues were determined using qPCR. -e ineffective and have a detrimental effect on patients’ quality average expression level of CLIP2 and SH3PXD2A in OC of life. -us, viable and effective therapies are urgently tissue was considerably less than that in normal tissues needed. [25, 26] A growing body of evidence has illustrated (Figure 6). that the hypoxia microenvironment plays a critical role in immune response and carcinogenesis based on the dys- regulated expression of genes associated with hypoxia. 4. Discussion [27, 28] Most research conducted in the past on hypoxia in OV has focused on a single regulator. Hypoxia- OC is a severe epithelial cancer that predominantly con- induciblefactor-1α, for instance, has been reported to tributes to cancer-associated death among females [22–24]. play an integral role in various processes, including the -e treatment options available for OC are clinically CD274 Immune checkpoint CTLA4 HAVCR2 LAG3 PDCD1 PDCD1LG2 TIGIT SIGLEC15 Journal of Oncology 11 2.0 ** 1.5 1.0 0.5 0 0.0 Normal Tumor Normal Tumor (a) (b) Figure 6: -e expression of CLIP2 and SH3PXD2A determined via polymerase chain reaction (PCR). promotion of OC immunosuppression, tumour metastasis, groups based on their potential response to chemotherapy or and chemoresistance. other immune checkpoint blockades. -us, CLIP2 and In this study, two subtypes of hypoxia were identified SH3PXD2A should be further investigated and could be using consensus clustering analysis, each of which was based novel biomarkers for patients with OC. on the prognostic immune-relevant genes. Particularly, hypoxia subtype 2 displayed a more unfavourable clinical Abbreviations outcome than hypoxia subtype 1. Cancer is a malignant OC: Ovarian cancer neoplasm that may be caused by genetic mutations and TCGA: -e Cancer Genome Atlas variations [29]. Hypoxia subtype 1 was characterised by the DEGs: Differentially expressed genes presence of more prevalent genetic alterations. Alterations in TME: -e tumour microenvironment the expression of several genes, including TP53, have been miRNAs: microRNAs observed to be correlated with the success of immuno- FPKM: Fragments per kilobase of exon per million therapies and exhibit a predictive potential [30]. In the OC mapped fragments samples, the TP53 gene was the first to undergo mutation. In TPM: Transcripts per million hypoxia subtype 1, the TP53 gene was reported to have GO: Gene Ontology a greater incidence of mutations than that in hypoxia CC: Cellular components subtype 2. Our results suggest a difference among the BP: Biological pathways hypoxia subtypes in terms of genetic changes and mutations. MF: Molecular function In this study, we identified 374 genes generated from the KEGG: Kyoto Encyclopedia of Genes and Genomes hypoxia subtypes, which had the potential to influence PPI: Protein-Protein Interaction pathways such as the PI3K-Akt signalling pathway. Hub MCODE: Molecular Complex Detection genes, such as CLIP2 and SH3PXD2A, were selected and GEPIA: Gene Expression Profiling Interactive Analysis used for further investigation. Recently, the expression of CI: Confidence interval SH3PXD2A-AS1 was observed to be related to OC; however, qRT- Quantitative Reverse-Transcription Polymerase the underlying molecular mechanism remains unknown. PCR: Chain Reaction Simultaneously, the absence of SH3PXD2A has been re- NK cells: Natural killer cells ported in the OV area. Nonetheless, further investigation is warranted. -e results of ssGSEA demonstrated that the Data Availability decrease in CLIP2 and SH3PXD2A expression may influence the infiltration levels of immune cells, such as NK cells. -e datasets used and/or analysed during the current study Finally, PCR results confirmed these patterns in OC tissues. are available from the corresponding author upon reason- While this work is a bioinformatics and pcr analysis, more able request. investigation should be performed in clinic for future application. Conflicts of Interest A well-defined hypoxia score has significant benefits over standard prognostic markers used for OC. -erefore, -e authors declare that they have no conflicts of interest. the hypoxia score may be used to compare various hypoxia- modulating components and aid the exploration of how Authors’ Contributions tumour cells interact with the immunological milieu. In addition, it helps stratify patients with OC into various Jiawei Wang and Tuo Li contributed equally to this work. CLIP2 mRNA Levels (relative to Normal) SH3PXD2A mRNA Levels (relative to Normal) 12 Journal of Oncology [16] Q. Wu, B. Zhang, Z. Wang et al., “Integrated bioinformatics Acknowledgments analysis reveals novel key biomarkers and potential candidate small molecule drugs in gastric cancer,” Pathology, Research & -is work was funded by the Nantong Science and Tech- Practice, vol. 215, no. 5, pp. 1038–1048, 2019. nology Program (MS12021009), the Hong Kong Scholars [17] M. E. Ritchie, B. Phipson, D. Wu et al., “Limma powers Program (XJ2016016), and the Shanghai Pujiang Program differential expression analyses for RNA-sequencing and (2020PJD063). microarray studies,” Nucleic Acids Research, vol. 43, no. 7, p. e47, 2015. References [18] L. Zhu, X. Liu, W. Zhang, H. Hu, Q. Wang, and K. Xu, “MTHFD2 is a potential oncogene for its strong association [1] F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre, with poor prognosis and high level of immune infiltrates in and A. Jemal, “Global cancer statistics 2018: GLOBOCAN urothelial carcinomas of bladder,” BMC Cancer, vol. 22, no. 1, estimates of incidence and mortality worldwide for 36 cancers p. 556, 2022. in 185 countries,” CA: A Cancer Journal for Clinicians, vol. 68, [19] Y. Zhou, B. Zhou, L. Pache et al., “Metascape provides a bi- no. 6, pp. 394–424, 2018. ologist-oriented resource for the analysis of systems-level [2] D. Badgwell and R. C. Bast Jr, “Early detection of ovarian datasets,” Nature Communications, vol. 10, no. 1, p. 1523, cancer,” Disease Markers, vol. 23, no. 5-6, pp. 397–410, 2007. [3] A. L. Shafrir, A. Babic, M. Gates Kuliszewski et al., “Estrogen [20] Z. Tang, C. Li, B. Kang, G. Gao, C. Li, and Z. Zhang, “GEPIA: receptor-β expression of ovarian tumors and its association a web server for cancer and normal gene expression profiling with ovarian cancer risk factors,” Cancer Epidemiology, Bio- and interactive analyses,” Nucleic Acids Research, vol. 45, markers & Prevention, vol. 29, no. 11, pp. 2211–2219, 2020. no. 1, pp. W98–W102, 2017. [4] W. Wang, Q. Wu, Z. Wang et al., “Development of a prog- [21] A. M. Newman, C. L. Liu, M. R. Green et al., “Robust enu- nostic model for ovarian cancer patients based on novel meration of cell subsets from tissue expression profiles,” immune microenvironment related genes,” Frontiers in On- Nature Methods, vol. 12, no. 5, pp. 453–457, 2015. cology, vol. 11, Article ID 647273, 2021. [22] D. Matei and K. P. Nephew, “Epigenetic attire in ovarian [5] B. Orr and R. P. Edwards, “Diagnosis and treatment of ovarian cancer: the emperor’s new clothes,” Cancer Research, vol. 80, cancer,” Hematology-Oncology Clinics of North America, no. 18, pp. 3775–3785, 2020. vol. 32, no. 6, pp. 943–964, 2018. [23] -e Lancet Oncology, “Ovarian cancer: breaking the silence,” [6] INVALID CITATION !!! 6, https://community.endnote.com/ +e Lancet Oncology, vol. 13, no. 2, p. 111, 2012. t/invalid-citation/259536. [24] P. B. Ottevanger, “Ovarian cancer stem cells more questions [7] S. B. Hooks, P. Callihan, M. K. Altman, J. H. Hurst, M. W. Ali, than answers,” Seminars in Cancer Biology, vol. 44, pp. 67–71, and M. M. Murph, “Regulators of G-protein signaling RGS10 and RGS17 regulate chemoresistance in ovarian cancer cells,” [25] J. Liu and U. A. Matulonis, “New strategies in ovarian cancer: Molecular Cancer, vol. 9, no. 1, p. 289, 2010. translating the molecular complexity of ovarian cancer into [8] R. C. Bast, B. Hennessy, and G. B. Mills, “-e biology of treatment advances,” Clinical Cancer Research, vol. 20, no. 20, ovarian cancer: new opportunities for translation,” Nature pp. 5150–5156, 2014. Reviews Cancer, vol. 9, no. 6, pp. 415–428, 2009. [26] F. Ricci, M. Broggini, and G. Damia, “Revisiting ovarian [9] M. Alharbi, A. Lai, S. Sharma et al., “Extracellular vesicle cancer preclinical models: implications for a better man- transmission of chemoresistance to ovarian cancer cells is agement of the disease,” Cancer Treatment Reviews, vol. 39, associated with hypoxia-induced expression of glycolytic no. 6, pp. 561–568, 2013. pathway proteins, and prediction of epithelial ovarian cancer [27] K. DePeaux and G. M. Delgoffe, “Metabolic barriers to cancer disease recurrence,” Cancers, vol. 13, no. 14, p. 3388, 2021. immunotherapy,” Nature Reviews Immunology, vol. 21, no. 12, [10] Y. Gao, L. Chen, G. Cai et al., “Heterogeneity of immune pp. 785–797, 2021. microenvironment in ovarian cancer and its clinical signifi- [28] X. Jing, F. Yang, C. Shao et al., “Role of hypoxia in cancer cance: a retrospective study,” OncoImmunology, vol. 9, no. 1, therapy by regulating the tumor microenvironment,” Mo- Article ID 1760067, 2020. lecular Cancer, vol. 18, no. 1, p. 157, 2019. [11] R. S. Riley, C. H. June, R. Langer, and M. J. Mitchell, “Delivery [29] K. L. Kanchi, K. J. Johnson, C. Lu et al., “Integrated analysis of technologies for cancer immunotherapy,” Nature Reviews germline and somatic variants in ovarian cancer,” Nature Drug Discovery, vol. 18, no. 3, pp. 175–196, 2019. Communications, vol. 5, no. 1, p. 3156, 2014. [12] R. Derynck, S. J. Turley, and R. J. Akhurst, “TGFβ biology in [30] H. Sun, S. Y. Liu, J. Y. Zhou et al., “Specific TP53 subtype as cancer progression and immunotherapy,” Nature Reviews biomarker for immune checkpoint inhibitors in lung ade- Clinical Oncology, vol. 18, no. 1, pp. 9–34, 2021. nocarcinoma,” EBioMedicine, vol. 60, Article ID 102990, 2020. [13] D. Hao, J. Liu, M. Chen et al., “Immunogenomic analyses of [31] N. Li and X. Zhan, “Identification of clinical trait-related advanced serous ovarian cancer reveal immune score is lncRNA and mRNA biomarkers with weighted gene co- a strong prognostic factor and an indicator of chemo- expression network analysis as useful tool for personalized sensitivity,” Clinical Cancer Research, vol. 24, no. 15, medicine in ovarian cancer,” +e EPMA Journal, vol. 10, no. 3, pp. 3560–3571, 2018. pp. 273–290, 2019. [14] W. Hu, G. Wang, Y. Chen, L. B. Yarmus, B. Liu, and Y. Wan, “Coupled immune stratification and identification of thera- peutic candidates in patients with lung adenocarcinoma,” Aging, vol. 12, no. 16, pp. 16514–16538, 2020. [15] W. Qi and Q. Zhang, “Identification and validation of im- mune molecular subtypes and immune landscape based on colon cancer cohort,” Frontiers of Medicine, vol. 9, Article ID 827695, 2022.

Journal

Journal of OncologyHindawi Publishing Corporation

Published: Sep 12, 2022

There are no references for this article.