This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Buckanovich, R. J. Articles by Coukos, G. Search for Related Content PubMed PubMed Citation Articles by Buckanovich, R. J. Articles by Coukos, G.

Tumor Vascular Proteins As Biomarkers in Ovarian Cancer

Ronald J. Buckanovich, Dimitra Sasaroli, Anne O'Brien-Jenkins, Jeffrey Botbyl, Rachel Hammond, Dionysios Katsaros, Raphael Sandaltzopoulos, Lance A. Liotta, Phyllis A. Gimotty, George Coukos

From the Center for Research on Reproduction and Women's Health, Abramson Family Cancer Research Institute, Department of Medicine Division of Hematology-Oncology, and Department of Biostatistics and Epidemiology, University of Pennsylvania, Philadelphia, PA; University of Michigan, Ann Arbor, MI; Center for Applied Proteomics and Molecular Medicine, George Mason University, Fairfax, VA; Department of Obstetrics and Gynecology, University of Turin, Turin, Italy; and Molecular Biology and Genetics Program, Democritus University of Thrace, Komotini, Greece

Address reprint requests to George Coukos, MD, PhD, 1315 BRB II/III, 421 Curie Blvd, Philadelphia, PA, 19104; e-mail: gcks{at}mail.med.upenn.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Purpose: This study aimed to identify novel ovarian cancer biomarkers and potential therapeutic targets through molecular analysis of tumor vascular cells.

Methods: Immunohistochemistry-guided laser-capture microdissection and genome-wide transcriptional profiling were used to identify genes that were differentially expressed between vascular cells from human epithelial ovarian cancer and healthy ovaries. Tumor vascular markers (TVMs) were validated through quantitative real-time polymerase chain reaction (qRT-PCR) of immunopurified tumor endothelial cells, in situ hybridization, immunohistochemistry, and Western blot analysis. TVM expression in tumors and noncancerous tissues was assessed by qRT-PCR and was profiled using gene expression data.

Results: We identified a tumor vascular cell profile of ovarian cancer that was distinct from the vascular profile of normal ovary and other tumors. We validated 12 novel ovarian TVMs. These were expressed by immunopurified tumor endothelial cells and localized to tumor vasculature. Select TVMs were found to be specifically expressed in ovarian cancer and were absent in all normal tissues tested, including female reproductive tissues with physiologic angiogenesis. Many ovarian TVMs were expressed by a variety of other solid tumors. Finally, overexpression of any one of three ovarian TVMs by vascular cells was associated with decreased disease-free interval (all P < .005).

Conclusion: We have identified for the first time the molecular profile of ovarian tumor vasculature. We demonstrate that TVMs may serve as potential biomarkers and molecular targets for ovarian cancer and a variety of other solid tumors.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Gene expression profiling is a convenient method to uncover tumor-specific genes that may function as biomarkers or therapeutic targets. So far, important efforts have been made regarding the cancer cell as the main source of biomarkers.^1,2 However, emerging studies argue that tumor components other than tumor cells may contribute significantly to cancer molecular signatures. In landmark studies, tumor neovessels have been shown to be different from vessels of normal tissues,³ expressing unique genes.^4,5 Cellular and molecular changes characterizing the angiogenic switch are essential for tumor growth beyond a few millimeters.⁶ Furthermore, specialization of the vasculature may even precede tumor establishment at metastatic sites.⁷ These observations suggest that molecular changes in tumor vasculature might provide sensitive markers of tumor initiation or metastasis. Similarly, it is expected that genes expressed uniquely by the vasculature of tumors may provide important therapeutic targets.

Ovarian cancer critically depends on blood supply for expansive growth, and increased angiogenesis is associated with rapid tumor recurrence and decreased survival.^8,9 Unique among solid tumors, targeting tumor angiogenesis by blocking vascular endothelial growth factor (VEGF) as monotherapy has produced tangible clinical results in ovarian cancer,^10-14 suggesting that vascular development is central to ovarian cancer biology. Thus, we hypothesized that ovarian cancer is an appropriate tumor to evaluate for vasculature-specific genes to uncover novel tumor biomarkers.

To date, the expression profile of vascular cells in ovarian cancer remains unreported. We used immunohistochemistry-guided laser-capture microdissection (immuno-LCM) and transcriptional profiling¹⁵ to characterize tumor vascular cells (TVCs) from 21 epithelial ovarian cancers and four normal ovaries. From the cancer vascular profile, we validated 12 novel tumor vascular markers (TVMs) and demonstrated that three TVMs have prognostic value. Several of these could function as tumor biomarkers or therapeutic targets; they were expressed at high levels in ovarian cancer and were either absent or expressed at significantly lower levels in normal tissues.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Patient Tissues
Stage III epithelial ovarian cancer (n = 46 categorized as either serous, mucinous, endometrioid [n = 33], or clear-cell/undifferentiated [n = 13] and stage II/III invasive ductal breast cancer [n = 5]) frozen specimens from untreated patients were provided by the University of Turin (Turin, Italy). Fresh tumors stage II/III lung adenocarcinoma (n = 5) and metastatic melanomas (n = 5) for cell purification and/or molecular studies were collected at the University of Pennsylvania (Philadelphia, PA). Normal human tissues, including ovary, were provided by the Cooperative Human Tissue Network, for which clinical information is unavailable. All specimens were processed in compliance with institutional review board and Health Insurance Portability and Accountability Act requirements.

Immuno-LCM and RNA Isolation and Amplification
Anti-hCD31 (BD Pharmingen, San Diego, CA) mixed with anti-CD146 (Chemicon, Temecula, CA), and secondary antibody (Biotin anti-mouse secondary antibody; Vector, Burlingame, CA) were used (1:10) in phosphate-buffered saline (PBS) containing RNA Protector (1:10; Roche, Basel, Switzerland). Immuno-LCM^15-18 was performed using Microcut (MMI, Glattbrugg, Switzerland) to isolate up to 2 x 10³ immunopositive cells per specimen. RNA was isolated using the TRIzol microprotocol, and RNA was amplified using Messageamp (Ambion, Austin, TX), as detailed elsewhere.¹⁵ After two rounds of amplification, cRNA was biotin-labeled (12 to 24 hours, ENZO [Farmingdale, NY] RNA labeling kit) and purified using RNA cleanup (Qiagen, Valencia, CA).

Gene Expression
Biotin-cRNA was fragmented, hybridized to HG-U133 A/B arrays (Affymetrix, Santa Clara, CA) and scanned as recommended in the Affymetrix GeneChip protocol (www.affymetrix.com). Quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) was performed as detailed elsewhere.¹⁵ Primers are provided in Table A1 (online only). All transcripts were confirmed using 3% agarose gel electrophoresis. In all cases, expression was normalized to beta-actin (ß-actin). Experiments comparing tumors, normal tissues, and angiogenic tissues were also normalized to either CD31 or VE-cadherin, to control for vascular density. In situ hybridization (ISH) as performed using a protocol provided by K.W. Kinzler (Johns Hopkins, Annapolis, MD).⁴ Digoxigenin (DIG) -labeled antisense RNA probes were transcribed using DIG RNA labeling reagents and T7 RNA polymerase (Roche) from 500- to 600–base pair PCR templates containing T7 promoters in the antisense primers. Primers are detailed in Table A2 (online only). We used two rounds of amplification and detected hybridized probe with streptavidin-Cy3 conjugate (Arcturus, Mountain View, CA; 1:200, 5 minutes). Slides were washed and mounted with fluorescent mounting media containing 4',6-Diamidino-2-phenylindole (DAPI; Vector). Double immunofluorescence and immunohistochemistry was performed as previously described.¹⁹ Primary antibodies against CD31, STC2 (Genway, San Diego, CA), F2RL1, and DR6 were incubated for 2 hours. Western blot was performed using 300 µg of protein from patient sera and probed with goat antihuman DR6 antibody. All antibodies used are detailed in Table A3 (online only).

Cells
CD146⁺, VE-cadherin⁺, CD45^– tumor endothelial cells (TECs) were isolated by fluorescence-associated cell sorting (FACS), after Ficol gradient centrifugation, from mechanically dispersed fresh human epithelial ovarian cancer, lung adenocarcinoma or melanoma specimens using fluorescein isothiocyanate (FITC) anti–VE-cadherin antibody (Medsystems Diagnostics, Vienna, Austria), biotinylated anti-CD146 coupled with streptavidin phycoerythrin (PE) Cy7, and APC-anti-CD45 (BD Pharmingen).^19,20 Normal murine serum (10%; Sigma, St Louis, MO) and 25 mg/mL anti-mouse Fc receptor (BD Pharmingen) were added before incubation to avoid nonspecific binding. FACS-isolated TECs were more than 95% pure (data not shown). TECs were analyzed immediately by qRT-PCR. Human umbilical-vein endothelial cells (HUVECs) were cultured in endothelial basal medium with 10% bovine serum and 100 ng/mL VEGF. Peripheral-blood mononuclear cells (PBMCs) were obtained freshly through leukapheresis from healthy donors and analyzed immediately.

Bioinformatics and Statistical Analyses
Present/absent indicators (gene expressed/not expressed) were produced with MAS5.0 Suite (Affymetrix). All samples were normalized with the median defined as 1.0. Analysis of expression profiles used Genespring software (Agilent, Santa Clara, CA). Statistical significance for mRNA expression differences between tissue types was determined using a two-tailed t test, and differences were characterized by fold changes defined as the ratio of the normalized mean expression of the tumor to the normal vascular samples. A heat map condition tree was developed using the Genespring hierarchical clustering algorithm. Descriptive statistics were performed with SPSS statistics software (SPSS Inc, Chicago, IL). Means are presented with SEs. Nonparametric analysis based on the ranks of the expression level for tumor and normal samples was developed in SAS 9.1.²¹ The log-rank statistic was used to compare the Kaplan-Meier disease-free interval (DFI) distribution of six tumors with the highest TVM expression values (highest tertile) to the remaining 11 tumors using SAS 9.1.²¹

Archived Gene Expression Datasets
Data used for expression of the TVM in normal and tumor tissue samples are available via the Gene Expression Omnibus (GEO; National Center for Biotechnology Information [NCBI], http://www.ncbi.nlm.nih.gov/geo) with series numbers GSE3526 and GSE2109, respectively. All CEL files were similarly processed using the rate monotonic algorithm.²²

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Identification of TVMs
We used optimized immuno-LCM coupled with RNA amplification and genome-wide gene expression profiling¹⁵ to profile tumor vasculature cells from 21 stage III ovarian cancers and four normal ovaries. Unsupervised hierarchical clustering was used to determine whether the transcriptional profile of microdissected TVCs differed from that of normal vascular cells. We included 17,920 genes after elimination of genes for which the difference between tumor and normal mean expression level was less than its SE. TVCs were accurately distinguished from normal vascular cells (Fig 1A).

View larger version (57K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 1. Analysis of tumor vascular cell arrays. (A) Unsupervised hierarchical clustering of ovarian tumor vascular cell and normal (Nml) vascular cell profiles. (B) List of tumor vascular markers identified using three independent criteria. Genes in bold font have previously been identified in vascular cells. Genes highlighted in blue are further analyzed in this article.

Consistent with our hypothesis that the tumor vascular profile of ovarian cancer would be distinct, several of the identified markers had not been previously reported as TVMs. We selected 12 new putative TVMs for further validation: adlican (matrix-remodeling associated 5,MXRA5); C11orf8 (metallophosphoesterase domain containing 2, MPPED2); collagen type-XI alpha-1 (COL11A1); death receptor-6 (DR6; TNF receptor superfamily member-21, TNFRSF21); epidermal growth factor-like-domain multiple-6 (EGFL6); four jointed box-1 (FJX1); coagulation factor-II (thrombin) receptor-like 1 (F2RL1); frizzled homolog-10 (FZD10); glycoprotein M6B (GPM6B); leucine zipper putative tumor suppressor-1 (LZTS1); olfactomedin-like 2B (OLFML2B); and stanniocalcin-2 (STC2).

Validation and Localization of TVMs
To confirm TVM expression by TVCs we immunopurified VE-cadherin⁺, CD146⁺, CD45^– TECs by FACS from five fresh ovarian cancer specimens. TVM mRNA expression was evaluated by qRT-PCR in TECs as well as in HUVECs and PBMCs. All TVM were detected in purified TEC, confirming the vascular origin of the genes (Fig 2A). Adlican, C11orf8, EGFL6, FZD10, and LZTS1 were absent in HUVECs, whereas the remaining TVMs were expressed at similar or lower levels in HUVECs than in TECs. The known tumor vascular marker TEM1, tested as control, was expressed in TECs but not in HUVECs. Most TVMs were absent or expressed at the lowest level of detection in PBMCs.

View larger version (49K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 2. (A) Expression of tumor vascular markers (TVMs) by quantitative real-time polymerase chain reaction (qRT-PCR) in purified tumor endothelial cells (TECs), human umbilical-vein endothelial cells (HUVECs) and peripheral-blood mononuclear cells (PBMCs). (B) TVM expression in ovarian tumor vasculature (in situ hybridization). Controls: sense-strand RNA (Sense); normal ovary, antisense probe (Nml Ov); and RNAse-pretreated tumor (RNAse). DAPI, 4',6-Diamidino-2-phenylindole.

For several of the TVMs for which commercial antibodies are available, we performed immunofluorescence or immunohistochemistry to confirm protein expression within vascular structures. Immunofluorescence demonstrated colocalization of STC-2 with CD31 (endothelial localization; Fig 3A). Immunohistochemistry demonstrated expression of F2RL1 in a periendothelial location and in stromal cells in tumors, whereas staining in normal ovary was rare or absent. DR6 protein was diffusely expressed in tumor vascular structures and faintly in tumor stroma. DR6 protein was expressed in the vasculature of normal ovaries, although at significantly lower levels (Fig 3B).

View larger version (88K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 3. Expression of tumor vascular marker (TVM) protein. (A) Immunofluorescent colocalization of STC2 and CD31 in ovarian tumors. Merge includes 4',6-Diamidino-2-phenylindole (DAPI) nuclear staining. (B) Immunohistochemical localization of F2RL1 and DR6 protein. (C, D) Western blot of DR6 protein in control normal human sera (NHS) and sera from ovarian cancer patients (OC), with densitometric quantitation of expression.

Ovarian TVMs Are Novel Tumor Biomarkers
Because TVMs are expressed specifically by tumors harboring TECs, we hypothesized that TVMs could potentially function as tumor biomarkers. An important prerequisite of tumor biomarkers is that they be overexpressed in tumor relative to normal tissue from the same organ. All 12 TVMs tested were overexpressed in ovarian cancers (n = 25, stage III, independent set) relative to normal ovaries (n = 5; P < .05 for all TVMs) by qRT-PCR. In fact, most TVMs were expressed at the lowest limits of detection in normal ovaries (Fig 4A).

View larger version (33K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 4. Quantitative real-time polymerase chain reaction analysis of tumor vascular marker expression in (A) ovarian cancer and normal ovaries, (B) ovarian cancer and tissues with physiologic angiogenesis, and (C) ovarian cancer and a panel of normal human tissues. Expression for each normal tissue was normalized against ovarian cancer, which was defined as 100% for each gene.

More generally, tumor biomarkers are characterized by their overexpression in tumor tissue relative to other normal tissues. Several TVMs, including adlican, COL11A1, EGFL6, F2RL1, FZD10, LZTS1, and OlfML2B showed limited or no expression by qRT-PCR in normal tissues, but were expressed at high levels in ovarian cancer. In contrast, C11orf8, DR6, FJX1, GPM6B, and STC2 were expressed in at least one normal tissue (Fig 4C).

As an independent analysis, we examined the expression profile of TVMs in normal tissues in a publicly available gene expression profile data set that included 44 normal tissue types from five male and five female human donors. Results were concordant with the above qRT-PCR data, showing little or no TVM expression in normal ovary (Fig 5A). C11orf8, CL11A1, EGFL6, FJX1, F2RL1, FZD10, LZTS1, and STC2 were also absent in almost all normal human tissues, with few exceptions. Adlican, DR6, and GPM6B were detected in many normal tissues including reproductive organs and the nervous system. Similar results were obtained using publicly available serial analysis of gene expression (SAGE) data sets (data not shown).

View larger version (72K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 5. Heat map of tumor vascular marker expression in (A) normal human tissues and (B) different solid tumor types. Average expression is reported for each gene/tissue in the corresponding box. N indicates the number of different samples available for the specified tissue. Data represent 355 samples from five female and five male donors.

We independently tested expression of TVM by qRT-PCR in lung cancer, mesothelioma, breast cancer, melanoma, and ovarian cancer (Fig 6A). As expected, most TVMs were expressed at quantitatively different levels in different types of malignancy. These differences persisted when expression was normalized to CD31 or VE-cadherin, to control for vascular density.

View larger version (41K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 6. Tumor vascular marker (TVM) expression in different tumors. (A) Expression of TVM by quantitative real-time polymerase chain reaction (qRT-PCR) in different tumor types (whole tissue). (B) Expression of TVMs by qRT-PCR in immunopurified VE-cadherin+, CD146+, CD45– tumor endothelial cells from melanoma, lung cancer, and ovarian cancer. (C) Kaplan-Meier disease-free survival distributions of patients classified on the basis of TVM expression.

TVMs Are Prognostic Markers
Finally for each TVM, the Kaplan-Meier DFI distributions were computed for patients with high TVM expression (in the highest tertile of TVM expression in microdissected TVCs) compared with those with lower TVM expression (Fig 6C). The expression of STC2, EGFL6, and FZD10 in the highest tertile was associated with significantly decreased DFI (all P < .005). By contrast, the survival curves of patients with highest tertile and lower expression of the vascular markers CD31 and VE-cadherin were not significantly different (not shown), suggesting that the observed effect with TVMs is not simply a result of increased content of endothelial cells.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
This study identified for the first time the expression profile of ovarian cancer vasculature in situ, and showed that high-throughput genome-wide discovery efforts addressing tumor vasculature can unveil interesting candidate genes for further testing as tumor biomarkers or therapeutic targets. The immuno-LCM isolation used in this study has the advantage of capturing TECs in their natural microenvironment, and, on the basis of the use of frozen samples, allowed us to analyze the largest number of samples to date in tumor vasculature profiling. Our protocol also allows for the inclusion of pericytes, and our work suggests that pericytes are likely to be one source of TVMs. Seventy genes were identified as TVMs including several known TVMs or TVM family members. We validated 12 novel TVMs; these were expressed by immunopurified TECs, and expression in vivo was localized to vascular structures. Thus, the approach undertaken was robust and can provide a detailed molecular map of tumor vasculature.

Although ovarian vascular cells expressed numerous genes identified in other tumor vascular profiles, including several collagen molecules, nidogen 2, Thy-1, FolH1, and TEM7, the majority of the TVMs identified have not been reported in vascular profiling studies from colon cancer, glioblastoma multiforme, or breast ductal carcinoma in situ.^4,23,24 This could be due in part to differences in the techniques utilized, the number of samples analyzed, or true heterogeneity of vascular endothelium among tumors. Analysis of TVM expression from immunopurified TVCs from ovarian cancer, lung cancer, and melanoma suggested significant quantitative differences among TVCs from the three tumor types. These results argue that the distinct molecular profiles observed to date are due in part to TEC specialization in different tumor types.

Many of the TVMs identified were markedly overexpressed in ovarian cancer relative not only to normal ovary but also to a comprehensive panel of normal organs. By highly sensitive qRT-PCR, adlican, COL11A1, EGFL6, F2RL1, FZD10, LZTS1, and OlfML2B were detected at the lowest limits, or not at all, in normal tissues tested. These findings were confirmed by SAGE data and by comprehensive gene expression array data including 44 different normal tissues. Thus, some of the TVMs validated may be considered for further preclinical testing as bona fide tumor biomarkers or therapeutic targets. Our Western blot analysis suggests that secreted proteins such as DR6, adlican, and COL11A1 could provide useful serum tumor biomarkers. More sensitive/quantitative enzyme-linked immunosorbent assay analyses are necessary to test this in a larger population of patients. Sensitive antibodies could detect nascent tumors or monitor small changes in tumor status. Similarly, surface proteins expressed by TECs, such as TNFRSF21, may circumvent existing biodistribution barriers of tumors and function as useful targets for cancer antibody-based therapy or molecular imaging. Importantly, some of the TVMs identified here were not expressed by reproductive tissues that exhibit increased angiogenesis, such as placenta, corpus luteum, or proliferative endometrium. This is a critical feature to enhance specificity in reproductive-age women requiring surveillance because of a hereditary risk of ovarian cancer.

Some of the TVMs identified in ovarian cancer were expressed in a variety of other tumors. Thus, diagnostic or therapeutic tools developed on the basis of such TVMs may be applicable to several tumor types. The use of large gene expression array databases can significantly accelerate mapping of these markers in different tumor types to select disease targets for further clinical validation. Importantly, we found that overexpression of select TVMs in purified vascular cells was significantly predictive of poor clinical outcome. Because some of these TVMs are expressed in other tumor types, their prognostic potential may be applicable to other tumors. These data confirm our hypothesis that tumor vascular proteins can provide novel biomarkers, which could aid in early tumor detection or prognosis. Additional studies will be required to determine the clinical relevance of the biomarkers identified.

	AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment: N/A Leadership: N/A Consultant: Ronald J. Buckanovich, Aisling Capital Stock: N/A Honoraria: N/A Research Funds: N/A Testimony: N/A Other: N/A

	AUTHOR CONTRIBUTIONS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Conception and design: Ronald J. Buckanovich

Financial support: Ronald J. Buckanovich, George Coukos

Provision of study materials or patients: Dionysios Katsaros, Raphael Sandaltzopoulos

Collection and assembly of data: Ronald J. Buckanovich, Dimitra Sasaroli, Anne O'Brien-Jenkins, Phyllis A. Gimotty

Data analysis and interpretation: Ronald J. Buckanovich, Dimitra Sasaroli, Jeffrey Botbyl, Rachel Hammond, Phyllis A. Gimotty

Manuscript writing: Ronald J. Buckanovich, Lance A. Liotta, Phyllis A. Gimotty, George Coukos

Final approval of manuscript: Ronald J. Buckanovich, Dimitra Sasaroli, Phyllis A. Gimotty, George Coukos

	Appendix

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 

	ACKNOWLEDGMENTS

 
We thank Steven M. Albelda, MD, and David Elder, MD (University of Pennsylvania), for providing fresh thoracic tumor and melanoma specimens, respectively. We thank John Tobias for assistance with Genespring data analysis.

	NOTES

 
Supported by National Institutes of Health (NIH) Grant No. R01 CA098951; National Cancer Institute (NCI) Ovarian Cancer Specialized Program of Research Excellence (SPORE) Grant No. P50-CA083638; US Army Medical Research and Materiel Command Grant No. OC-050314; and the Marcia and Philip Rothblum Foundation. R.J.B. was supported by NIH/National Institute of Child Health and Human Development Grant No. K12-HD43459 and the Ovarian Cancer Research Fund. The laser-capture microdissection facility was supported by a generous grant by the Fannie Rippel Foundation.

R.J.B and D.S. contributed equally to this work.

Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
1. Bild AH, Yao G, Chang JT, et al: Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature 439:353-357, 2006[CrossRef][Medline]

2. Tsuda H, Ito YM, Ohashi Y, et al: Identification of overexpression and amplification of ABCF2 in clear cell ovarian adenocarcinomas by cDNA microarray analyses. Clin Cancer Res 11:6880-6888, 2005[Abstract/Free Full Text]

3. Ruoslahti E: Specialization of tumour vasculature. Nat Rev Cancer 2:83-90, 2002[CrossRef][Medline]

4. St Croix B, Rago C, Velculescu V, et al: Genes expressed in human tumor endothelium. Science 289:1197-1202, 2000[Abstract/Free Full Text]

5. Carson-Walter EB, Watkins DN, Nanda A, et al: Cell surface tumor endothelial markers are conserved in mice and humans. Cancer Res 61:6649-6655, 2001[Abstract/Free Full Text]

6. Folkman J: Role of angiogenesis in tumor growth and metastasis. Semin Oncol 29:15-18, 2002[Medline]

7. Kaplan RN, Riba RD, Zacharoulis S, et al: VEGFR1-positive haematopoietic bone marrow progenitors initiate the pre-metastatic niche. Nature 438:820-827, 2005[CrossRef][Medline]

8. Hazelton DA, Hamilton TC: Vascular endothelial growth factor in ovarian cancer. Curr Oncol Rep 1:59-63, 1999[Medline]

9. Bamberger ES, Perrett CW: Angiogenesis in epithelian ovarian cancer. Mol Pathol 55:348-359, 2002[Abstract/Free Full Text]

10. Ozols RF: Systemic therapy for ovarian cancer: Current status and new treatments. Semin Oncol 33:S3-S11, 2006[Medline]

11. de Gramont A, Van Cutsem E: Investigating the potential of bevacizumab in other indications: Metastatic renal cell, non-small cell lung, pancreatic and breast cancer. Oncology 69:46-56, 2005 (suppl)

12. Monk BJ, Choi DC, Pugmire G, et al: Activity of bevacizumab (rhuMAB VEGF) in advanced refractory epithelial ovarian cancer. Gynecol Oncol 96:902-905, 2005[CrossRef][Medline]

13. Bidus MA, Webb JC, Seidman JD, et al: Sustained response to bevacizumab in refractory well-differentiated ovarian neoplasms. Gynecol Oncol 102:5-7, 2006[CrossRef][Medline]

14. Ueda M, Terai Y, Kanda K, et al: Tumor angiogenesis and molecular target therapy in ovarian carcinomas. Hum Cell 18:1-16, 2005[Medline]

15. Buckanovich RJ, Sasaroli D, O'Brien-Jenkins A, et al: Use of immuno-LCM to identify the in situ expression profile of cellular constituents of the tumor microenvironment. Cancer Biol Ther 5:635-642, 2006[Medline]

16. Emmert-Buck MR, Bonner RF, Smith PD, et al: Laser capture microdissection. Science 274:998-1001, 1996[Abstract/Free Full Text]

17. Fend F, Emmert-Buck MR, Chuaqui R, et al: Immuno-LCM: Laser capture microdissection of immunostained frozen sections for mRNA analysis. Am J Pathol 154:61-66, 1999[Abstract/Free Full Text]

18. Murakami H, Liotta L, Star RA: IF-LCM: Laser capture microdissection of immunofluorescently defined cells for mRNA analysis rapid communication. Kidney Int 58:1346-1353, 2000[CrossRef][Medline]

19. Coukos G, Benencia F, Buckanovich R, et al: The role of dendritic cell precursors in tumour vasculogenesis. Br J Cancer 92:1182-1187, 2005[CrossRef][Medline]

20. Conejo-Garcia JR, Buckanovich RJ, Benencia F, et al: Vascular leukocytes contribute to tumor vascularization. Blood 105:679-681, 2005[Abstract/Free Full Text]

21. Michailidis G, Shedden K: The application of rule-based methods to class prediction problems in genomics. J Comput Biol 10:689-698, 2003[CrossRef][Medline]

22. Irizarry RA, Hobbs B, Collin F, et al: Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4:249-264, 2003[Abstract]

23. Madden SL, Cook BP, Nacht M, et al: Vascular gene expression in nonneoplastic and malignant brain. Am J Pathol 165:601-608, 2004[Abstract/Free Full Text]

24. Allinen M, Beroukhim R, Cai L, et al: Molecular characterization of the tumor microenvironment in breast cancer. Cancer Cell 6:17-32, 2004[CrossRef][Medline]

Submitted August 21, 2006; accepted December 5, 2006.

This article has been cited by other articles:

				R. Bhati, C. Patterson, C. A. Livasy, C. Fan, D. Ketelsen, Z. Hu, E. Reynolds, C. Tanner, D. T. Moore, F. Gabrielli, et al. Molecular Characterization of Human Breast Tumor Vascular Cells Am. J. Pathol., May 1, 2008; 172(5): 1381 - 1390. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Buckanovich, R. J. Articles by Coukos, G. Search for Related Content PubMed PubMed Citation Articles by Buckanovich, R. J. Articles by Coukos, G.