Abstract
We have previously reported surrogate biomarkers of VEGF pathway activities with the potential to provide predictive information for anti-VEGF therapies. The aim of this study was to systematically evaluate a new VEGF-dependent gene signature (VDGs) in relation to molecular subtypes of ovarian cancer and patient prognosis. Using microarray profiling and cross-species analysis, we identified 140-gene mouse VDGs and corresponding 139-gene human VDGs, which displayed enrichment of vasculature and basement membrane genes. In patients who received bevacizumab therapy and showed partial response, the expressions of VDGs (summarized to yield VDGs scores) were markedly decreased in post-treatment biopsies compared with pre-treatment baselines. In contrast, VDGs scores were not significantly altered following bevacizumab treatment in patients with stable or progressive disease. Analysis of VDGs in ovarian cancer showed that VDGs as a prognostic signature was able to predict patient outcome. Correlation estimation of VDGs scores and molecular features revealed that VDGs was overrepresented in mesenchymal subtype and BRCA mutation carriers. These findings highlighted the prognostic role of VEGF-mediated angiogenesis in ovarian cancer and proposed a VEGF-dependent gene signature as a molecular basis for developing novel diagnostic strategies to aid patient selection for VEGF-targeted agents.
Similar content being viewed by others
Introduction
High-grade serous ovarian carcinoma (HGS-OvCa) is the predominant form of ovarian cancer and the most lethal gynecological malignancy, with approximately 140,000 deaths per year globally1,2,3. The majority of patients are diagnosed as advanced, disseminated disease and the survival rate is dismal4,5. In addition, recent analyses from The Cancer Genome Atlas (TCGA) research network have identified four molecular subtypes of HGS-OvCa, namely Differentiated, Immunoreactive, Mesenchymal and Proliferative, indicating that a high level of intertumoral heterogeneity may also impact on patient outcome6. As a result, despite our increased understanding of the physiopathology underpinning HGS-OvCa, its clinical management has not been appreciably improved over the past decades7. The current standard treatment of HGS-OvCa is aggressive surgical debulking followed by multi-cycles of platinum-based combination chemotherapy. Although many patients display a transient response, the vast majority eventually relapse and suffer from recurrent disease without efficacious treatment regimen8,9. Therefore, there is a compelling need to develop novel therapeutic strategies that can effectively control advanced-stage HGS-OvCa10,11.
VEGF-mediated tumor angiogenesis has been prominently implicated in the progression of ovarian cancer and hence represents one of the most promising targets12,13,14,15. Triggered by the remarkable preclinical efficacy of bevacizumab (Avastin), a humanized VEGF blocking monoclonal antibody, in a range of solid tumor types, a series of clinical studies have been conducted to evaluate bevacizumab in patients with newly-diagnosed or recurrent ovarian cancer. Two large prospective randomized phase III trials (GOG-0218 and ICON7) met their primary objective, demonstrating significantly improved progression-free survival (PFS) with bevacizumab administered with front-line chemotherapy compared with chemotherapy alone16,17. Two further randomized phase III clinical trials (OCEANS and AURELIA) have proved the efficacy of bevacizumab in recurrent ovarian cancer18,19. Based on these results, bevacizumab received European and FDA regulatory approval for use in combination with chemotherapy to treat advanced-stage ovarian cancer. Nevertheless, the increased PFS did not translate into a significant improvement in overall survival (OS) and robust biomarkers for predicting bevacizumab efficacy are currently lacking, which impedes patient selection and the optimal use of bevacizumab in ovarian cancer20,21.
We have previously employed gene expression profiling analysis and identified surrogate markers of VEGF inhibition as potential guides for the selection of patients who likely benefit from anti-VEGF therapy. The selected gene set was able to inform on VEGF downstream bioactivity and predict clinical outcome in breast cancer following bevacizumab treatment22. In this study, through further characterization of angiogenesis-related gene transcripts in mouse models and human samples, we established a novel VEGF-dependent gene signature and investigated its correlation with molecular subtypes of ovarian cancer and patient prognosis.
Results
Identification of a VEGF-dependent gene signature
In order to generate a faithful VEGF-dependent gene signature, we systematically profiled the transcriptional changes induced by VEGF neutralization in a transgenic murine model of highly vascularized pancreatic neuroendocrine tumors, using two well-established microarray platforms in two independent experiments (Fig. 1A). As we reported previously22, anti-VEGF treatment displayed merely anti-vascular but not anti-proliferative effects at day 7, which was selected as the time point to characterize the specific gene expression response of the tumor vasculature due to VEGF blockade. Affymatrix and Agilent microarray analysis identified 386 and 207 genes with a significant decrease (adjusted P value < 0.05) in transcript abundance, respectively (Supplementary Tables 1 and 2). We focused on the 140 genes detected by both platforms to further minimize false-positives associated with genome-wide profiling assays (Supplementary Table 3) and termed these genes VDGs (VEGF-dependent gene signature). Notably, we observed no corresponding upregulation of gene expression with one only exception Oxct1, consistent with the physical elimination of tumor vascular endothelial cells (Fig. 1B). Functional annotation demonstrated that the VDGs was enriched for endothelial or basement membrane specific genes (such as Cdh5 and Collagens) and was implicated in blood vessel morphogenesis (Fig. 1C; Supplementary Table 4). In addition, the VDGs was composed of both candidate proximal biomarkers of VEGF inhibition (e.g. Esm1, Nid2 and Prnd) and more distal downstream surrogate markers of the subsequent vessel loss (e.g. Cdh5, Plvap and Flt1), as we discovered previously22.
Validation of the VEGF-dependent gene signature
Next, we sought to validate the VDGs as an indicator of VEGF signaling activity in various tumor models. The expressions of VDGs were summarized to yield VDGs scores by using single sample gene set enrichment analysis. Firstly, we analyzed transcriptional responses to long-term anti-VEGF treatment in samples from a murine genetic intestinal tumor model ApcMin23. As expected, anti-VEGF administration was sufficient to induce a significant decrease in the mean value of VDGs scores (Fig. 2A). Secondly, in an orthotopic 66c14 mouse breast cancer model, the VDGs was similarly downregulated by short-term anti-VEGF treatment (Fig. 2B). Thirdly, anti-VEGF exposure resulted in a statistically significant reduction of VDGs scores in an established subcutaneous human breast carcinoma tumor model (MDA-MB-231) (Fig. 2C). Therefore, regardless of the different tumor models and independent of the length of anti-VEGF treatment, the VDGs consistently reflected the anti-vascular consequences of VEGF signaling inhibition and likely correlated with the VEGF pathway bioactivity in tumor samples.
Importantly, we assessed the dynamic changes of VDGs scores in serial clinical specimens collected from breast cancer patients treated with neoadjuvant bevacizumab. In this cohort, twenty treatment-naïve patients were biopsied and received one cycle of bevacizumab followed by six cycles of bevacizumab plus combination chemotherapy prior to surgery24. The 139 human orthologs of the VDGs exhibited a significant downregulation in post-bevacizumab tissues compared with matched pretreatment biopsies (Fig. 2D). Of more interest, when the patients were stratified into two subgroups based on the Response Evaluation Criteria in Solid Tumors (RECIST), VDGs scores were markedly decreased by bevacizumab treatment specifically in responders (partial response), but were not significantly altered in nonresponders (stable or progressive disease) (Fig. 2D; Supplementary Figure 1). Additionally, we observed a trend towards higher baseline VDGs scores in responders than nonresponders (Supplementary Figure 2), suggesting that elevated VEGF pathway bioactivity might be associated with the clinical benefit from bevacizumab therapy. Taken together, the new VDGs that we identified in preclinical models enables the accurate detection of an evolutionary conserved vascular response to VEGF signaling inhibition in clinical tumor samples.
The VDGs predicts patient prognosis in HGS-OvCa
Although bevacizumab has been approved for treating advanced-stage ovarian cancer, predictive biomarkers to improve patient selection are needed. Considering the tempting hypothesis that higher levels of VEGF downstream bioactivity potentially correlate with increased responses to bevacizumab, we reasoned that it might be important to comprehensively characterize the VDGs in the context of ovarian cancer. To this end, we investigated the VDGs in a gene set of ovarian tumor samples, in which epithelial and stromal components had been microdissected and profiled separately. Both ssGSEA and GSEA indicated that the VDGs was significantly enriched in the microdissected stroma components in comparison to paired tumor cells (Fig. 3A). In addition, we analyzed gene expression profiles of nine pairs of ovarian tumors and matched patient-derived xenografts (PDXs), in which mouse cells should competitively substitute human stroma including endothelial cells. The VDGs scores were accordingly decreased in PDXs (Fig. 3B). Our analyses also indicated that the VDGs scores increased upon chemotherapy or tumor metastasis (Supplementary Figure 3), although the clinical significance of these findings was not yet clear.
To assess the VDGs as a prognostic biomarker, we segregated the 486 TCGA HGS-OvCa samples into two clusters, i.e. the ‘VDGs low’ group and the ‘VDGs high’ group, according to the median VDGs scores. We found that patients classified as ‘VDGs high’ had significantly shorter survival than ‘VDGs low’ patients (hazard ratio: 1.30; 95% CI: 1.01 to 1.68) (Fig. 3C). To independently corroborate the prognostic value of VDGs, we performed meta-analysis of 681 HGS-OvCa expression profiles across four large published datasets. Each of the four HGS-OvCa cohorts displayed a trend for decreased overall survival in the ‘VDGs high’ group versus ‘VDGs low’ group but did not achieve statistical significance (Supplementary Figure 4). However, when we combined all four datasets together, the difference in overall survival between the two groups was statistically significant (hazard ratio: 1.30; 95% CI: 1.06 to 1.60) (Fig. 3D). Interestingly, the VDGs did not predict the responsiveness to chemotherapy (Supplementary Figure 5), but was found to correlate with residual disease after debulking surgery (Fig. 3E), suggesting that it could be a potential molecular biomarker for accurate identification of patients at high risk of suboptimal cytoreduction. We conclude that the VDGs is of prognostic value in patients with HGS-OvCa.
The VDGs is enriched in mesenchymal ovarian tumors
Previous studies have identified four HGS-OvCa subtypes, namely Differentiated, Mesenchymal, Immunoreactive and Proliferative, which exhibit distinct transcriptional, biological and clinical characteristics6,25. We speculated that these molecular subtypes might have differential expression of VDGs and accordingly differ in their response to anti-VEGF agents. Therefore, we explored the potential differences in transcriptional levels of the 139 VDGs genes among the four TCGA subtypes. Indeed, the vast majority of VDGs genes appeared to be relatively highly expressed in the Mesenchymal cluster (Fig. 4A), in which the summarized VDGs scores were significantly increased (Fig. 4B). Similar results were obtained with Tothill and Crijns cohorts (Fig. 4C), strongly indicating that the VDGs was enriched in mesenchymal ovarian tumors. When we stratified the TCGA specimens into two groups based on the median VDGs scores, 100% (111/111) of the Mesenchymal tumors were classified as ‘VDGs high’, in comparison to 40% (61/153), 33% (28/85) and 31% (43/137) of the Differentiated, Immunoreactive and Proliferative samples, respectively (Fig. 4D). Of note, even in non-mesenchymal tumors, higher VDGs scores still tended to predict poorer patient outcome (Fig. 4E). Lastly, the BRCA1/BRCA2 status was available for 316 TCGA patients (Supplementary Table 5) and the VDGs was overrepresented in BRCA mutation carriers (Fig. 4F).
Discussion
In this study, we extended previous work of exploiting gene expression profiling approach for accurate detection of VEGF downstream biological activity and designed a more stringent analytical protocol to identify and validate a reliable VEGF-dependent gene signature in preclinical tumor models and in human patients. Although the VDGs overlapped with previously developed VEGF-related signatures in general22,26, the refined gene list presented here was physiologically relevant on the basis of suitable in vivo models and integration of gene profiling results from two independent microarray platforms. Furthermore, we presented a detailed analysis of the VDGs in ovarian cancer by systematically interrogating TCGA HGS-OvCa expression data and four additional genome-wide transcriptome cohorts in the public domain. As the VDGs directly reflected vascular-specific anti-VEGF downstream effects in various experimental models and clinical dataset, an exquisite delineation of its relation to molecular subtypes and disease prognosis in ovarian cancer may advance our understanding of the VEGF pathway and provide a promising diagnostic strategy for identifying a patient subpopulation likely to derive benefit from anti-VEGF treatment. Further optimization and validation of the VDGs in both retrospective and prospective clinical studies will pave the way to developing a biomarker-based companion assay for predicting responsiveness to bevacizumab in the clinical management of ovarian cancer patients.
Tumor angiogenesis is characterized by extensive molecular regulation, aberrant endothelial morphology, constant vessel remodeling and loss of hierarchical architecture27,28. The complexity and heterogeneity of tumor vasculature not only make pathological assessments of microvascular dependency challenging, but also impede biomarker discovery for pharmacological perturbations of tumor endothelium. As a result, overall microvascular density has variable prognostic capacity in predicting outcome of cancer patients29,30. Similarly, neither regulators of endothelial proliferation nor expression of vascular specific genes consistently correlate with bevacizumab efficacy across indications31,32. Our results, together with other reports22, collectively suggest that transcriptome-level characterization of the distinct tumor vascular compartment directly responsive to VEGF signaling inhibition may better inform on the biological activity and functional significance of VEGF-mediated tumor angiogenesis. This drug-gene signature-disease connection is reminiscent of the “Connectivity Map” concept33,34, which assists to elucidate the complex mechanisms underlying a biological pathway or pharmacological modulation. Hence, the system biology approach may be extremely powerful in understanding the intricate tumor microvasculature.
Ovarian cancer is a heterogeneous disease and its predominant form, HGS-OvCa, contains at least four molecular subtypes, i.e. Differentiated, Mesenchymal, Immunoreactive and Proliferative. We have previously provided evidence that tumor-infiltrating stromal cells had a profound effect on the expression patterns of HGS-OvCa as well as patient prognosis35. Endothelial cells are considered as important constituents of cancer-supporting stroma and the VDGs is indeed enriched in the noncancerous stromal components of ovarian tumors. Therefore, tumor vasculature, in conjunction with other stromal composition, may also contribute to defining molecular subtypes of HGS-OvCa. Alternatively, the VDGs and VEGF-regulated endothelium could independently mark a specific HGS-OvCa subpopulation with unique pathophysiological properties. Although a simple stratification of HGS-OvCa cases based on median VDGs scores is proved to be prognostically predictive, prospectively refined gene sets and tailored threshold may further improve the performance of this signature classifier. In addition, previous studies have identified multiple prognostic signatures in ovarian cancer36,37,38,39,40,41 and it would be interesting to determine whether the VDGs led to the same or different patient stratification in comparison to published models.
Our observation that the VDGs is overrepresented in the Mesenchymal subtype is particularly intriguing. Mechanistically, the intimate interplay between endothelial cells and fibroblasts/pericytes may promote vigorous angiogenic sprouting in mesenchymal tumors. Conceivably, the Mesenchymal subtype of ovarian cancer would derive increased benefit from anti-angiogenic therapy. Supporting this hypothesis, a preliminary analysis of response to bevacizumab in relation to the molecular classifications has indicated an improvement of PFS for patients in the Mesenchymal subtype of ovarian cancer42. It is noteworthy that a similar distinct subtype with mesenchymal features has been identified in multiple other tumor types including colon cancer, breast cancer and glioblastoma43,44,45,46,47. Therefore, it would be interesting to determine whether the VDGs is enriched in mesenchymal tumors and whether the enrichment translates to anti-VEGF efficacy across different cancers. Along this line, recent retrospective analysis of AVAglio data showed that both mesenchymal and proneural tumors derived a PFS benefit from bevacizumab compared with placebo in glioblastoma48.
In conclusion, we proposed a VEGF-dependent gene signature as a molecular basis for developing novel diagnostic and therapeutic strategies. Our data indicated the prognostic value of VDGs in ovarian cancer. The VDGs enrichment in ovarian cancer with mesenchymal signatures or BRCA mutations suggested that these patients might most likely gain sustained benefit from bevacizumab therapy, which requires further investigation in retrospective and prospective studies.
Methods
Mouse models and treatment regimens
RIP-TβAg mice, ApcMin mice and Beige Nude mice were housed and cared for according to guidelines from the Institutional Animal Care and Use Committee (IACUC) of Renji Hospital and all the animal experiments were approved by Renji Hospital IACUC. B20-4.1.1 (anti-VEGFA) and anti-Ragweed were dosed by intraperitoneal injection at 5 mg/kg twice weekly as described previously22.
Microarray analysis and gene signature derivation
Tumor RNA was prepared with RNeasy Plus Mini Kit (Qiagen) according to the manufacturer’s protocol. Total RNA was subjected to microarray analysis with Affymetrix Mouse Genome 430 2.0 Array or Agilent Whole Mouse Genome Oligo Microarray. Five biological replicates per treatment group were included for statistical analyses. All microarray analyses were performed using the R Statistical language. Affymetrix microarray probe-level data were normalized and summarized using robust multiarray average procedure (RMA)49. Agilent data were lowess normalized and log transformed and the mean was used to calculate gene level summaries. Differential gene expression was analyzed with linear models for microarray data (Limma)50. The VEGF-dependent gene signature was derived from genes significantly (adjusted P value < 0.05) downregulated in anti-VEGF treated tumors compared to controls.
HGS-OvCa microarray datasets
The microarray datasets of HGS-OvCa used in current study are publicly available and as described in our previous report35. Combined and filter TCGA gene expression data were downloaded from https://tcga-data.nci.nih.gov/docs/publications/ov_2011/. The BRCA mutation status of TCGA samples was based on http://www.cbioportal.org/. The four patient cohorts (Tothill, Crijns, Bonome and Yoshihara) for meta-analysis have been described previously51,52,53,54 and processed data were downloaded from a recent paper37. Other microarray datasets, including GSE9890, GSE15622, GSE30587 and GSE5692051,55,56,57, were obtained from the Gene Expression Omnibus.
Molecular subtypes of ovarian tumors
Molecular subtypes of ovarian tumors were identified as previously described35. Briefly, TCGA HGS-OvCa samples were classified based on non-negative matrix factorization (NMF) consensus clustering. To minimize the impact of outlier samples on the identification of subtype markers, the silhouette width was computed to filter out expression profiles with negative values. Significance analysis of microarrays (SAM) was performed to identify genes significantly differentially expressed across the four subtypes. These genes were trained by prediction analysis for microarrays (PAM) to achieve the lowest prediction error, which resulted in the 749-gene subtype-specific signature. This signature was applied to Tothill and Crijns cohorts, followed by consensus-based NMF analysis for molecular subtyping. Heatmaps were generated using GenePattern58.
GSEA and ssGSEA
The Gene Set Enrichment Analysis (GSEA) software was downloaded from the Broad Institute GSEA portal and GSEA was performed as described59, using the VDGs as the input gene set. Single sample GSEA (ssGSEA) was applied to generate compound scores for VDGs, in which gene expression values were ranked for a given sample and an enrichment score was calculated based on the normalized rank difference in Empirical Cumulative Distribution Functions (ECDF) of the genes in the signature and the remaining genes60.
Survival analysis
The ssGSEA compound scores of VDGs were computed for each sample. The patients were dichotomized into a high-score and a low-score group, using the median VDGs score as the threshold value. Overall survival curves were calculated using the Kaplan–Meier method and statistical significance was assessed using the log-rank test. The analyses were conducted with the R Bioconductor ‘survival’ package.
Statistical analysis
Gene ontology and pathway analyses were performed in DAVID Bioinformatics Resources with GOTERM ALL categories61,62. Terms were defined as significantly enriched if they contained at least five counts and had a P value of < 0.001 and an FDR of < 5%. For microarray analyses, an empirical Bayes method was used to adjust P values for multiple comparisons. In all experiments, comparisons between two groups were based on two-sided Student’s t test and one-way analysis of variance (ANOVA) was used to test for differences among more groups followed by post-hoc Tukey analysis for multiple comparisons. P values of < 0.05 were considered statistically significant.
Additional Information
How to cite this article: Yin, X. et al. A VEGF-dependent gene signature enriched in mesenchymal ovarian cancer predicts patient prognosis. Sci. Rep. 6, 31079; doi: 10.1038/srep31079 (2016).
References
Siegel, R. L., Miller, K. D. & Jemal, A. Cancer statistics, 2016. CA Cancer J Clin 66, 7–30 (2016).
Bowtell, D. D. The genesis and evolution of high-grade serous ovarian cancer. Nat Rev Cancer 10, 803–808 (2010).
Liu, J. & Matulonis, U. A. New strategies in ovarian cancer: translating the molecular complexity of ovarian cancer into treatment advances. Clin Cancer Res 20, 5150–5156 (2014).
Coleman, R. L., Monk, B. J., Sood, A. K. & Herzog, T. J. Latest research and treatment of advanced-stage epithelial ovarian cancer. Nat Rev Clin Oncol 10, 211–224 (2013).
Jayson, G. C., Kohn, E. C., Kitchener, H. C. & Ledermann, J. A. Ovarian cancer. Lancet 384, 1376–1388 (2014).
Cancer Genome Atlas Research, N. Integrated genomic analyses of ovarian carcinoma. Nature 474, 609–615 (2011).
Morgan, R. J., Jr. et al. Epithelial ovarian cancer. J Natl Compr Canc Netw 9, 82–113 (2011).
Berns, E. M. & Bowtell, D. D. The changing view of high-grade serous ovarian cancer. Cancer Res 72, 2701–2704 (2012).
Pignata, S. et al. Chemotherapy in epithelial ovarian cancer. Cancer Lett 303, 73–83 (2011).
Banerjee, S. & Kaye, S. B. New strategies in the treatment of ovarian cancer: current clinical perspectives and future potential. Clin Cancer Res 19, 961–968 (2013).
Bookman, M. A. et al. Better therapeutic trials in ovarian cancer. J Natl Cancer Inst 106, dju029 (2014).
Yamamoto, S. et al. Expression of vascular endothelial growth factor (VEGF) in epithelial ovarian neoplasms: correlation with clinicopathology and patient survival and analysis of serum VEGF levels. Br J Cancer 76, 1221–1227 (1997).
Byrne, A. T. et al. Vascular endothelial growth factor-trap decreases tumor burden, inhibits ascites and causes dramatic vascular remodeling in an ovarian cancer model. Clin Cancer Res 9, 5721–5728 (2003).
Jayson, G. C., Kerbel, R., Ellis, L. M. & Harris, A. L. Antiangiogenic therapy in oncology: current status and future directions. Lancet (2016).
Shaw, D., Clamp, A. & Jayson, G. C. Angiogenesis as a target for the treatment of ovarian cancer. Curr Opin Oncol 25, 558–565 (2013).
Burger, R. A. et al. Incorporation of bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 365, 2473–2483 (2011).
Perren, T. J. et al. A phase 3 trial of bevacizumab in ovarian cancer. N Engl J Med 365, 2484–2496 (2011).
Aghajanian, C. et al. OCEANS: a randomized, double-blind, placebo-controlled phase III trial of chemotherapy with or without bevacizumab in patients with platinum-sensitive recurrent epithelial ovarian, primary peritoneal, or fallopian tube cancer. J Clin Oncol 30, 2039–2045 (2012).
Pujade-Lauraine, E. et al. Bevacizumab combined with chemotherapy for platinum-resistant recurrent ovarian cancer: The AURELIA open-label randomized phase III trial. J Clin Oncol 32, 1302–1308 (2014).
Secord, A. A., Nixon, A. B. & Hurwitz, H. I. The search for biomarkers to direct antiangiogenic treatment in epithelial ovarian cancer. Gynecol Oncol 135, 349–358 (2014).
Colombo, N., Conte, P. F., Pignata, S., Raspagliesi, F. & Scambia, G. Bevacizumab in ovarian cancer: Focus on clinical data and future perspectives. Crit Rev Oncol Hematol 97, 335–348 (2016).
Brauer, M. J. et al. Identification and analysis of in vivo VEGF downstream markers link VEGF pathway activity with efficacy of anti-VEGF therapies. Clin Cancer Res 19, 3681–3692 (2013).
Moser, A. R. et al. ApcMin: a mouse model for intestinal and mammary tumorigenesis. Eur J Cancer 31A, 1061–1064 (1995).
Yang, S. X. et al. Gene expression profile and angiogenic marker correlates with response to neoadjuvant bevacizumab followed by bevacizumab plus chemotherapy in breast cancer. Clin Cancer Res 14, 5893–5899 (2008).
Konecny, G. E. et al. Prognostic and therapeutic relevance of molecular subtypes in high-grade serous ovarian cancer. J Natl Cancer Inst 106 (2014).
Tobin, N. P. et al. An Endothelial Gene Signature Score Predicts Poor Outcome in Patients with Endocrine-Treated, Low Genomic Grade Breast Tumors. Clin Cancer Res 22, 2417–2426 (2016).
Chung, A. S. & Ferrara, N. Developmental and pathological angiogenesis. Annu Rev Cell Dev Biol 27, 563–584 (2011).
Chung, A. S., Lee, J. & Ferrara, N. Targeting the tumour vasculature: insights from physiological angiogenesis. Nat Rev Cancer 10, 505–514 (2010).
Uzzan, B., Nicolas, P., Cucherat, M. & Perret, G. Y. Microvessel density as a prognostic factor in women with breast cancer: a systematic review of the literature and meta-analysis. Cancer Res 64, 2941–2955 (2004).
Cheng, S. H. et al. Prognostic role of microvessel density in patients with renal cell carcinoma: a meta-analysis. Int J Clin Exp Pathol 7, 5855–5863 (2014).
Lambrechts, D., Lenz, H. J., de Haas, S., Carmeliet, P. & Scherer, S. J. Markers of response for the antiangiogenic agent bevacizumab. J Clin Oncol 31, 1219–1230 (2013).
Jubb, A. M. et al. Impact of vascular endothelial growth factor-A expression, thrombospondin-2 expression and microvessel density on the treatment effect of bevacizumab in metastatic colorectal cancer. J Clin Oncol 24, 217–227 (2006).
Lamb, J. et al. The Connectivity Map: using gene-expression signatures to connect small molecules, genes and disease. Science 313, 1929–1935 (2006).
Lamb, J. The Connectivity Map: a new tool for biomedical research. Nat Rev Cancer 7, 54–60 (2007).
Zhang, S. et al. Stroma-associated master regulators of molecular subtypes predict patient prognosis in ovarian cancer. Sci Rep 5, 16066 (2015).
Kang, J., D’Andrea, A. D. & Kozono, D. A DNA repair pathway-focused score for prediction of outcomes in ovarian cancer treated with platinum-based chemotherapy. J Natl Cancer Inst 104, 670–681 (2012).
Verhaak, R. G. et al. Prognostically relevant gene signatures of high-grade serous ovarian carcinoma. J Clin Invest 123, 517–525 (2013).
Lu, J., Wu, D., Li, C., Zhou, M. & Hao, D. Correlation between gene expression and mutator phenotype predicts homologous recombination deficiency and outcome in ovarian cancer. J Mol Med (Berl) 92, 1159–1168 (2014).
Jin, N. et al. Network-based survival-associated module biomarker and its crosstalk with cell death genes in ovarian cancer. Sci Rep 5, 11566 (2015).
Zhou, M. et al. Comprehensive analysis of lncRNA expression profiles reveals a novel lncRNA signature to discriminate nonequivalent outcomes in patients with ovarian cancer. Oncotarget (2016).
Zhou, M. et al. Characterization of long non-coding RNA-associated ceRNA network to reveal potential prognostic lncRNA biomarkers in human ovarian cancer. Oncotarget 7, 12598–12611 (2016).
Winterhoff, B. J. N. et al. Bevacizumab and improvement of progression-free survival (PFS) for patients with the mesenchymal molecular subtype of ovarian cancer. J Clin Oncol 32, (suppl), abstr 5509 (2014).
Isella, C. et al. Stromal contribution to the colorectal cancer transcriptome. Nat Genet 47, 312–319 (2015).
Calon, A. et al. Stromal gene expression defines poor-prognosis subtypes in colorectal cancer. Nat Genet 47, 320–329 (2015).
Taube, J. H. et al. Core epithelial-to-mesenchymal transition interactome gene-expression signature is associated with claudin-low and metaplastic breast cancer subtypes. Proc Natl Acad Sci USA 107, 15449–15454 (2010).
Phillips, H. S. et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression and resemble stages in neurogenesis. Cancer Cell 9, 157–173 (2006).
Verhaak, R. G. et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR and NF1. Cancer Cell 17, 98–110 (2010).
Sandmann, T. et al. Patients With Proneural Glioblastoma May Derive Overall Survival Benefit From the Addition of Bevacizumab to First-Line Radiotherapy and Temozolomide: Retrospective Analysis of the AVAglio Trial. J Clin Oncol 33, 2735–2744 (2015).
Irizarry, R. A. et al. Exploration, normalization and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249–264 (2003).
Smyth, G. K., Yang, Y. H. & Speed, T. Statistical issues in cDNA microarray data analysis. Methods Mol Biol 224, 111–136 (2003).
Tothill, R. W. et al. Novel molecular subtypes of serous and endometrioid ovarian cancer linked to clinical outcome. Clin Cancer Res 14, 5198–5208 (2008).
Crijns, A. P. et al. Survival-related profile, pathways and transcription factors in ovarian cancer. PLoS Med 6, e24 (2009).
Bonome, T. et al. A gene signature predicting for survival in suboptimally debulked patients with ovarian cancer. Cancer Res 68, 5478–5486 (2008).
Yoshihara, K. et al. Gene expression profile for predicting survival in advanced-stage serous ovarian cancer across two independent datasets. PLoS One 5, e9615 (2010).
Ahmed, A. A. et al. The extracellular matrix protein TGFBI induces microtubule stabilization and sensitizes ovarian cancers to paclitaxel. Cancer Cell 12, 514–527 (2007).
Brodsky, A. S. et al. Expression profiling of primary and metastatic ovarian tumors reveals differences indicative of aggressive disease. PLoS One 9, e94476 (2014).
Ricci, F. et al. Patient-derived ovarian tumor xenografts recapitulate human clinicopathology and genetic alterations. Cancer Res 74, 6980–6990 (2014).
Reich, M. et al. GenePattern 2.0. Nat Genet 38, 500–501 (2006).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 102, 15545–15550 (2005).
Barbie, D. A. et al. Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1. Nature 462, 108–112 (2009).
Huang da, W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 4, 44–57 (2009).
Huang da, W., Sherman, B. T. & Lempicki, R. A. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res 37, 1–13 (2009).
Acknowledgements
This work was supported by the National Natural Science Foundation of China (81472537 to G Zhuang, 81502597 to Y Jing and 81472426 to W Di), the Grants from the State Key Laboratory of Oncogenes and Related Genes (91-14-18 and 91-15-12 to G Zhuang; SB201510 to Z Zhang), the grants from Shanghai Jiao Tong University School of Medicine (14XJ10020 to G Zhuang and DLY201505 to W Di), the Shanghai Institutions of Higher Learning (Eastern Scholar to G Zhuang), Shanghai Rising-Star Program (16QA1403600 to G Zhuang), Shanghai Municipal Commission of Health and Family Planning (2013ZYJB0202 and 15GWZK0701 to W Di).
Author information
Authors and Affiliations
Contributions
G.Z. and W.D. designed and supervised the project. X.Y., X.W., B.S., Y.J., Q.L., M.-C.C., Z.G., Q.Y., Z.Z., J.L. and H.L. performed all the analyses. G.Z. wrote the manuscript with the help of all authors.
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Electronic supplementary material
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Yin, X., Wang, X., Shen, B. et al. A VEGF-dependent gene signature enriched in mesenchymal ovarian cancer predicts patient prognosis. Sci Rep 6, 31079 (2016). https://doi.org/10.1038/srep31079
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/srep31079
- Springer Nature Limited
This article is cited by
-
Research Progresses in Cancer Stem Cells of Three Common Fertility-Related Female Malignancies
Pathology & Oncology Research (2019)
-
Vascular endothelial growth factor receptor 2 (VEGFR2) correlates with long-term survival in patients with advanced high-grade serous ovarian cancer (HGSOC): a study from the Tumor Bank Ovarian Cancer (TOC) Consortium
Journal of Cancer Research and Clinical Oncology (2019)
-
Recommendations for biomarker testing in epithelial ovarian cancer: a National Consensus Statement by the Spanish Society of Pathology and the Spanish Society of Medical Oncology
Clinical and Translational Oncology (2018)
-
Characterisation of tumour microvessel density during progression of high-grade serous ovarian cancer: clinico-pathological impact (an OCTIPS Consortium study).
British Journal of Cancer (2018)
-
The Prognostic 97 Chemoresponse Gene Signature in Ovarian Cancer
Scientific Reports (2017)