Introduction

Gene expression profiling to classify human tumors into functional and clinically relevant subgroups has come of age since this method was first shown to segregate diffuse large B-cell lymphoma,² breast cancer,^39,71 and melanoma⁵ into defined subsets with distinct biological behaviors. The microarray platforms themselves have evolved, providing ever-increasing coverage of the human transcriptome with more robust annotation. Massive parallel sequencing technologies and full RNA sequencing now provide even greater detail by including all the variants from every RNA species.⁵⁷

Yet, there are limitations that preclude immediate and widespread clinical application for these technologies. Surprisingly, cost is not a major limitation. Computational power and instrument costs (in shared resource environments) have reduced the cost of array-based analyses and RNAseq to << $500 per sample (in U.S. currency, 2010 value). Technical limitations are similarly trivial, assuming robust quality assurance and quality control protocols are in place. Rather, the major obstacles to clinical implementation are the magnitude of the datasets generated by these technologies and the paucity of individuals who are trained and capable of handling them. Thus, gene expression profiling has found a niche in accelerating basic research and as a tool to define predictive markers and therapeutic targets for a variety of cancers.²⁴

The application of gene expression profiles in companion animals has lagged predictably behind that of humans. Eleven studies evaluating gene expression profiles in canine tumors or tumor cell lines were indexed in PubMed (August 2010),^{18,27,38,44,45,50,55,59,63,65,75} compared to > 20,000 for humans. This is likely a reflection of economics (and possibly journal selection in PubMed), and does not take into account dozens of ongoing studies applying microarray technologies to canine tumors. Still, it offers a series of opportunities to ask questions selectively, to use published data as a means to add value to new datasets, and as a framework to use appropriate technologies to ask highly relevant questions.

Our group and many collaborators have completed gene expression profiling for > 100 distinct tumors using the Affymetrix canine_2.0 platform. We have defined ways to integrate those data into newer, probably more efficient platforms, and so our current studies use Agilent microarrays and RNAseq almost exclusively. Nonetheless, we have learned valuable lessons from our Affymetrix microarray data, and this review will highlight some of those data as they address specific applications.

Gene Expression Profiling and Molecular Classification of Tumors: An Example Using Canine Lymphoma¹⁶

The diagnosis of canine lymphoma or canine non-Hodgkin lymphoma (archaically also called canine lymphosarcoma) encompasses a heterogeneous group of diseases with a broad range of clinical behavior, from relatively indolent to rapidly lethal. The major shared feature among these diseases is their etiology arising from malignant transformation of a lymphoreticular cell. The modified WHO classification system to stratify canine lymphoma was first published in 2002,⁶⁸ and has since been revised and refined; yet, it remains to be universally adopted by the veterinary pathology community. This is due, at least partly, to the absence of definitive studies verifying that these represent distinct molecular subtypes with unique biological behavior and prognostic significance. Resistance also arises both from pathologists who are uncertain about payoffs for the additional training and time required to apply this classification to every lymphoma they review, as well as from clinicians who do not believe the use of this classification justifies the costs (and potential risk) of obtaining a tissue biopsy for every dog with lymphoma. These topics will be discussed at this meeting.

A history and current update of the modified WHO classification were published in the 6th edition of Schalm's Veterinary Hematology.^49,69,70 This classification includes approximately 30 subtypes, but recent data confirm our experience that the six most common forms of diffuse large B-cell lymphoma (DLBCL), marginal zone (B-cell) lymphoma (MZL), Burkitt or Burkitt-like lymphoma (BL), T-zone lymphoma (TZL), lymphoblastic T-cell lymphoma/leukemia (LBT), and peripheral T cell lymphoma-not otherwise specified (PTCL), account for > 90% of all canine lymphomas.^15,36,41 These tumors have consistent morphological and molecular characteristics, and T-cell tumors can be readily grouped according to aggressive (LBT, PTCL) and indolent (TZL) biological behavior. However, the cellular features that distinguish among B-cell tumors are subtle, and an association between subtype and clinical outcomes has been more elusive.

To date, genome-wide molecular features of canine lymphoma have not been interrogated in a sufficiently large sample size to refine the molecular classification, predictive value, or target discovery for this disease. As a prelude to this, we examined gene expression profiles of 36 primary tumors collected prospectively, and representing the six common subtypes defined above.¹⁶ Our results show that gene expression profiles clearly distinguish canine lymphomas based on their immunological ontogeny (i.e., derivation from B-cell or T-cell cell lineages). Furthermore, biological behavior was clearly associated with distinct patterns of gene expression in T-cell malignancies. Conversely, gene expression differences among the three subtypes of B-cell tumors were much more subtle, although there were differences in the architecture of the affected lymph nodes, differences were apparent when expression data were analyzed by gene set enrichment. Supervised analysis identified genes that are expressed differentially in dogs with B-cell lymphoma that show a robust response well to CHOP-based chemotherapy (survival > 15 months) vs. those that do not (survival < 14 months), although the expression of these genes does not seem to be regulated coordinately. Current work is focused on adapting this information to laboratory tests that are feasible and can provide prognostic and predictive information to help guide therapy. In summary, our data indicate that molecular profiling can help stratify complex neoplastic diseases into classifications that can help predict biological behavior. Ongoing and future work will narrow down most robust criteria to define each subtype and predict response to therapy using conventional and cost effective laboratory methodology in order to translate these findings to the clinical setting.

Gene Expression Profiling to Improve Prediction in Tumors with Clinically Heterogeneous Behavior: An Example Using Canine Osteosarcoma^48,62

Osteosarcoma is a heterogeneous and chaotic disease that has confounded accurate molecular classification, prognosis, and prediction. This is an aggressive disease with a median survival of < 1 year (with current standard of care); however, up to 20% of dogs with osteosarcoma respond to therapy and survive > 2 years.³⁵ This could be due to vigilance and early detection, but it more likely reflects variation in the biological behavior of this disease, mediated by tumor-intrinsic or tumor-autonomous properties. The histological classification of canine osteosarcoma is based on the type of matrix produced by tumors (osteoid, chondroid, or collagen), but this is not predictive of biological behavior or response to therapy. Other pathological and clinical features are similarly unreliable predictors of outcome.

This disease is no less of a conundrum when it occurs in people. The most commonly used prognostic factor in human patients with osteosarcoma is percent necrosis after neoadjuvant therapy, but recent studies indicate the correlation between necrosis and response to therapy is as low as 55% (not significantly better than "flipping a coin").⁷

We took advantage of the inherently reduced genetic heterogeneity present in dogs to reveal orthologous molecular subtypes of osteosarcoma. Using a cohort of 26 dogs with osteosarcoma, we identified a strong differential gene signature that segregated tumor samples into two groups with differential survival distributions, and which consisted of inversely correlated expression of genes associated with 'microenvironment-interaction' and 'cell cycle'. We next explored the reproducibility and the evolutionary conservation of this signature,⁴⁸ applying the restricted osteosarcoma gene vectors to an independent cohort of dogs with osteosarcoma³⁸ or to five independent cohorts of humans with osteosarcoma^7,25,34,38. This approach confirmed (in dogs) and identified (in humans) the existence of previously uncharacterized, molecularly distinct subtypes of osteosarcoma that are prognostically significant. Furthermore, when this profile was combined with analysis of microRNA expression and DNA copy number aberrations,⁶² a clear picture arose suggesting three etiologically distinct, and highly evolutionarily conserved pathways lead to osteosarcoma in people and in dogs.

This illustrates how the narrower genetic diversity of dogs can be utilized to stratify complex diseases, to obtain molecular characterizations that may enhance prognosis and prediction, and potentially to identify clinically relevant therapeutic targets.

Gene Expression Profiling to Identify Tumor Ontogeny and the Contribution of Heritable Factors: An Example Using Canine Hemangiosarcoma^58,59

Malignant soft tissue sarcomas that arise from or resemble constituents of blood vessels represent an understudied and poorly understood group of incurable tumors. In humans, these tumors include angiosarcomas (hemangiosarcomas and lymphangiosarcomas), Kaposi sarcomas, hemangioendotheliomas, and hemangiopericytomas, some of which are associated with medical or occupational exposures to ionizing radiation, viruses, and a variety of industrial and agricultural chemical agents. Their study is complicated by their infrequent occurrence, but their clinical significance is magnified because angiosarcomas in particular are associated with more frequent metastasis and greater patient morbidity and mortality than other soft tissue sarcomas.^10,26,52

Other species also develop hemangiosarcomas. From a comparative perspective, hemangiosarcomas occur rarely in mice as a spontaneous disease, but the incidence is significantly increased in the B6C3F1 hybrid strain after exposure to various classes of pharmaceuticals, making these tumors a factor in risk assessment for drug development.¹⁰ Dogs are the only species where idiopathic (spontaneous) hemangiosarcoma occurs commonly. This disease has been estimated to account for up to 7% of malignant canine tumors,⁶⁶ which would roughly translate into > 50,000 diagnoses per year in the United States. Regardless of species, treatment options for angiosarcoma and hemangiosarcoma are limited, and outcomes are generally unrewarding.^8,19,21 The standard of care in both humans and dogs includes surgery and adjuvant chemotherapy. The median and 5-year survival rates for human patients with angiosarcoma are reported to be approximately 2 to 2.5 years and 30%, respectively.²⁶ In dogs, the prognosis is equally grave: even though 10–15% of dogs with this disease survive 12 months or longer, most die within 3-months of their diagnosis.⁶⁶ Despite anecdotal success using immunotherapy, as well as novel chemotherapy and antiangiogenic strategies to treat canine hemangiosarcoma,^{3,9,29,46,64,67} the past 30 years have brought no improvements in survival for dogs with this disease²⁰.

The lack of effective treatments for humans and dogs with angiosarcoma and hemangiosarcoma is largely due to our incomplete understanding of the factors that promote the survival, growth, and metastases of these malignancies. Inflammation, hypoxia, and angiogenesis all might contribute to the pathogenesis of idiopathic hemangiosarcoma, or of hemangiosarcoma associated with exposure to non-genotoxic agents in each of the target species. The link between inflammation and cancer is becoming clearer, with macrophages and macrophage-derived cytokines playing a central role in modulating the tumor microenvironment to facilitate both tumor survival and metastasis.^31,32,76 Macrophage activation and local tissue hypoxia are central components of the proposed mechanism of action that drives hemangiosarcoma in rodents exposed to a diverse array of compounds such as 2-butoxyethanol, peroxisome proliferator–activated receptor (PPAR) agonists and pregabalin.¹⁰ Parallels have been drawn between canine hemangiosarcoma cells and neoangiogenic endothelial cells in tumors.^1,14 Vessel formation in hemangiosarcoma resembles the morphology of imbalanced, chaotic growth and maturation of neoangiogenic vessels seen in cancer, which is at least partly driven by pro-angiogenic factors such as vascular endothelial growth factor-A (VEGF).^14,40 In fact, hemangiosarcoma cells elaborate growth factors that promote angiogenesis, including not only VEGF, but also platelet-derived growth factor-β (PDGFβ), and basic fibroblast growth factor (bFGF) in vitro.^1,14,20,61 Signaling by each of these growth factors is partly dependent on activation of the phosphoinositide 3-kinase (PI3K) pathway, providing a possible connection between the processes of inflammation, hypoxia, and angiogenesis in the pathogenesis of hemangiosarcoma.²² In this regard, mutations of the PI3K antagonist, PTEN, are common in canine hemangiosarcoma; however, they are restricted to the C-terminal domain and do not affect the phosphorylation of Akt that occurs downstream from PI3K signaling.¹¹ While it is possible that mutations in the C-terminal domain reduce the stability of PTEN⁷³ or increase motility, and hence a cell's invasive potential,^30,37 the precise effects of these mutations in canine hemangiosarcoma remain unclear. The genetic basis of abnormal patterns of growth and signaling requires further characterization.

Mutational events have been documented in sporadic angiosarcomas of humans and hemangiosarcomas of mice and dogs, including cancer-associated genes such as PTEN, Ras, VHL, p53, and connexin.^{12,17,23,33,47,51,60,74} In the case of canine hemangiosarcoma, PTEN mutations did not fully explain the increased levels of VEGF or other growth factors,^11,14 prompting additional assessment of potential roles for mutations that inactivate specific oncogenes or tumor suppressor genes that can lead to elevated VEGF production. Yet, another possibility is that non-malignant cells are responsible for VEGF production in canine hemangiosarcoma,⁵⁶ especially since co-existence of tumor cells with inflammatory cells is a common feature of this disease, and in some cases, the inflammatory cells may provide the principal source of VEGF¹¹. In this scenario, VEGF-producing inflammatory cells could be reactive leukocytes incited by pathologic effects of tumor (e.g., tissue destruction), or macrophages and myeloid cells that are intrinsic components of the tumor microenvironment.^31,56 A third possibility is that hemangiosarcomas originate from a multipotent bone marrow progenitor that can differentiate along the myeloid lineage,^28,77 and these cells thus could reflect the ontogeny of the malignant cells and their plasticity to differentiate into multiple cell types.

While most tumors arise sporadically due to an accumulation of key somatic mutations, it can be presumed that individual genetic backgrounds contribute to the risk, phenotypes and biological behavior of cancer. Yet, until recently, experimental evidence for this presumption, as well as for the magnitude of its effect was lacking for any naturally occurring, non-heritable tumor in any species. Studies exploring how race and ethnicity influenced gene expression and disease susceptibility in humans found few differences, and none especially compelling.^13,54,72 Dogs provide a useful surrogate for human ethnic groups. While they retain individual (outbred) traits, the derivation and maintenance of unique breeds has led to restricted gene pools. These restricted gene pools can be used to study heritable contributions to cancer susceptibility in animals that develop tumors spontaneously and share the human environment, but with the benefit of less ''noise'' from other phenotypic variation. In the case of hemangiosarcoma, there is an apparent predilection for certain breeds such as German Shepherd Dogs, Boxers, and Golden Retrievers.^{4,6,42,43,53,66} Given the strong association between breed and risk, we predicted that gene expression profiles in tumors such as hemangiosarcoma also would reflect features uniquely associated with the breed, and that breed-related gene expression profiles would uncover biologically and therapeutically significant pathways that would inform etiology and identify therapeutic targets.

We used an isolated in vitro system to examine gene expression in canine hemangiosarcoma, comparing data from a cohort of cell lines derived from malignant tumors as well as control endothelial cell lines derived from non-malignant proliferative lesions of the dog spleen (benign hematomas associated with nodular hyperplasia). Our hypotheses were that canine hemangiosarcoma would show characteristic gene expression profiles that would be informative for etiology and progression, and that hemangiosarcomas of Golden Retrievers would be distinguishable from histologically similar hemangiosarcomas of dogs from other breeds (non-Golden Retrievers) based on the overexpression or underexpression of genes preferentially concentrated in one or a few metabolic pathways.

Our results show that genes involved in inflammation, angiogenesis, adhesion, invasion, metabolism, cell cycle, signaling, and patterning can distinguish hemangiosarcoma cells from non-malignant endothelial cells.⁵⁸ This signature did not simply reflect a cancer-associated angiogenic phenotype, as it also distinguished hemangiosarcoma from non-endothelial, moderately to highly angiogenic bone marrow-derived tumors (lymphoma, leukemia, osteosarcoma). Moreover, our results uncovered unique gene sets, also largely confined to these biological processes, which were uniquely enriched in hemangiosarcoma from a single dog breed (sharing a common genetic background).⁵⁹

Our data show that inflammation and angiogenesis are important processes in the pathogenesis of vascular tumors, but a definitive ontogeny of the cells that give rise to these tumors remains to be established. The data do not yet distinguish whether functional or ontogenetic plasticity creates this phenotype, although they suggest that cells which give rise to hemangiosarcoma modulate their microenvironment to promote tumor growth and survival. Similarly, these data suggest that the traits leading to hemangiosarcoma are modulated by a spectrum of heritable traits which may result in greater risk of developing the disease itself or perhaps developing a subtype of the disease that shows more rapid progression. We propose that creative studies of naturally occurring canine cancer offer opportunities to integrate molecular data to stratify disease into pathologically relevant and clinically significant subtypes. Used judiciously, these data may assist in the identification of targets to develop effective strategies for prevention, control, and treatment.

Acknowledgments

The author wishes to acknowledge the valuable contributions of every member of his laboratory and his collaborators' laboratories without whom this work would not have been possible, as well as the funding agencies that have supported this work.

Disclosures

Dr. Modiano holds an equity interest in - and serves as a consultant for - ApopLogic Pharmaceuticals, Inc., the developer of Fasaret, a product that has been used in treatment of canine osteosarcoma. These relationships have been reviewed and managed by the University of Minnesota in accordance with its conflict of interest policies

References

1.  Akhtar N, Padilla M, Dickerson E, Steinberg H, Breen M, Auerbach R, Helfand S. Interleukin-12 inhibits tumor growth in a novel angiogenesis canine hemangiosarcoma xenograft model. Neoplasia 2004;6:106–116.

2.  Alizadeh AA, Eisen MB, Davis RE, Ma C, Lossos IS, Rosenwald A, Boldrick JC, Sabet H, Tran T, Yu X, Powell JI, Yang L, Marti GE, Moore T, Hudson J, Lu L, Lewis DB, Tibshirani R, Sherlock G, Chan WC, Greiner TC, Weisenburger DD, Armitage JO, Warnke R, Staudt LM, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 2000;403:503–511.

3.  Angiogenesis_Foundation: Antiangiogenic therapy for canine cancers. http://www.angio.org/pets_and_wildlife/pet/canine_cancer/therapy.html

4.  Appleby EC, Hayward AH, Douce G. German shepherds and splenic tumors. Vet Rec 1978;102:449.

5.  Bittner M, Meltzer P, Chen Y, Jiang Y, Seftor E, Hendrix M, Radmacher M, Simon R, Yakhini Z, Ben-Dor A, Sampas N, Dougherty E, Wang E, Marincola F, Gooden C, Lueders J, Glatfelter A, Pollock P, Carpten J, Gillanders E, Leja D, Dietrich K, Beaudry C, Berens M, Alberts D, Sondak V, Hayward N, Trent J. Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature 2000;406:536–540.

6.  Brown NO, Patnaik AK, MacEwen EG. Canine hemangiosarcoma: retrospective analysis of 104 cases. J Am Vet Med Assoc 1985;186:56–58.

7.  Cleton-Jansen AM, Anninga JK, Briaire-de Bruijn IH, Romeo S, Oosting J, Egeler RM, Gelderblom H, Taminiau AH, Hogendoorn PC. Profiling of high-grade central osteosarcoma and its putative progenitor cells identifies tumourigenic pathways. Br J Cancer 2009;101:1909–1918.

8.  Clifford CA, Hughes D, Beal MW, Mackin AJ, Henry CJ, Shofer FS, Sorenmo KU. Plasma vascular endothelial growth factor concentrations in healthy dogs and dogs with hemangiosarcoma. J Vet Intern Med 2001;15:131–135.

9.  Cohen LA, Powers B, Amin S, Desai D. Treatment of canine haemangiosarcoma with suberoylanilide hydroxamic acid, a histone deacetylase inhibitor. Vet Comp Oncol 2004;2:243–248.

10. Cohen SM, Storer RD, Criswell KA, Doerrer NG, Dellarco VL, Pegg DG, Wojcinski ZW, Malarkey DE, Jacobs AC, Klaunig JE, Swenberg JA, Cook JC. Hemangiosarcoma in rodents: mode-of-action evaluation and human relevance. Toxicol Sci 2009;111:4–18.

11. Dickerson EB, Thomas R, Fosmire SP, Lamerato-Kozicki AR, Bianco SR, Wojcieszyn JW, Breen M, Helfand SC, Modiano JF. Mutations of phosphatase and tensin homolog deleted from chromosome 10 in canine hemangiosarcoma. Vet Pathol 2005;42:618–632.

12. Duddy SK, Gorospe SM, Bleavins MR, de la Iglesia FA. Spontaneous and thiazolidinedione-induced B6C3F1 mouse hemangiosarcomas exhibit low ras oncogene mutation frequencies. Toxicol Appl Pharmacol 1999;160:133–140.

13. Ferguson SE, Olshen AB, Levine DA, Viale A, Barakat RR, Boyd J. Molecular profiling of endometrial cancers from African-American and Caucasian women. Gynecol Oncol 2006;101:209–213.

14. Fosmire SP, Dickerson EB, Scott AM, Bianco SR, Pettengill MJ, Meylemans H, Padilla M, Frazer-Abel AA, Akhtar N, Getzy DM, Wojcieszyn J, Breen M, Helfand SC, Modiano JF. Canine malignant hemangiosarcoma as a model of primitive angiogenic endothelium. Lab Invest 2004;84:562–572.

15. Fosmire SP, Thomas R, Jubala CM, Wojcieszyn JW, Valli VE, Getzy DM, Smith TL, Gardner LA, Ritt MG, Bell JS, Freeman KP, Greenfield BE, Lana SE, Kisseberth WC, Helfand SC, Cutter GR, Breen M, Modiano JF. Inactivation of the p16 cyclin-dependent kinase inhibitor in high-grade canine non-Hodgkin's T-cell lymphoma. Vet Pathol 2007;44:467–478.

16. Frantz AM, Sarver AL, Phang TL, Valli VEO, Karimpour-Fard A, Scott MC, Ito D, Hunter LE, O'Brien TD, Modiano JF. Diagnostic biomarkers for distinct lymphoma subtypes identified by gene expression profiling. In preparation

17. Froment O, Boivin S, Barbin A, Bancel B, Trepo C, Marion MJ. Mutagenesis of ras proto-oncogenes in rat liver tumors induced by vinyl chloride. Cancer Res 1994;54:5340–5345.

18. Gama A, Alves A, Schmitt F. Identification of molecular phenotypes in canine mammary carcinomas with clinical implications: application of the human classification. Virchows Arch 2008;453:123–132.

19. Hammond TN, Pesillo-Crosby SA. Prevalence of hemangiosarcoma in anemic dogs with a splenic mass and hemoperitoneum requiring a transfusion: 71 cases (2003–2005). J Am Vet Med Assoc 2008;232:553–558.

20. Helfand SC. Canine hemangiosarcoma: a tumor of contemporary interest. Cancer Ther 2008;6:457–462.

21. Hillers KR, Lana SE, Fuller CR, LaRue SM. Effects of palliative radiation therapy on nonsplenic hemangiosarcoma in dogs. J Am Anim Hosp Assoc 2007;43:187–192.

22. Hirsch E, Ciraolo E, Ghigo A, Costa C. Taming the PI3K team to hold inflammation and cancer at bay. Pharmacol Ther 2008;118:192–205.

23. Hong HL, Ton TV, Devereux TR, Moomaw C, Clayton N, Chan P, Dunnick JK, Sills RC. Chemical-specific alterations in ras, p53, and beta-catenin genes in hemangiosarcomas from B6C3F1 mice exposed to o-nitrotoluene or riddelliine for 2 years. Toxicol Appl Pharmacol 2003;191:227–234.

24. Huang R, Wallqvist A, Thanki N, Covell DG. Linking pathway gene expressions to the growth inhibition response from the National Cancer Institute's anticancer screen and drug mechanism of action. Pharmacogenomics J 2005;5:381–399.

25. Kobayashi E, Masuda M, Nakayama R, Ichikawa H, Satow R, Shitashige M, Honda K, Yamaguchi U, Shoji A, Tochigi N, Morioka H, Toyama Y, Hirohashi S, Kawai A, Yamada T. Reduced arginosuccinate synthetase is a predictive biomarker for the development of pulmonary metastasis in patients with osteosarcoma. Mol Cancer Ther 2010;9:535–544.

26. Koch M, Nielsen GP, Yoon SS. Malignant tumors of blood vessels: angiosarcomas, hemangioendotheliomas, and hemangioperictyomas. J Surg Oncol 2008;97:321–329.

27. Krol M, Pawlowski KM, Skierski J, Turowski P, Majewska A, Polanska J, Ugorski M, Morty RE, Motyl T. Transcriptomic "portraits" of canine mammary cancer cell lines with various phenotypes. J Appl Genet 2010;51:169–183.

28. Lamerato-Kozicki AR, Helm KM, Jubala CM, Cutter GR, Modiano JF. Canine hemangiosarcoma originates from hematopoietic precursors with potential for endothelial differentiation. Exp Hematol 2006;34:870–878.

29. Lana S, U'Ren L, Plaza S, Elmslie R, Gustafson D, Morley P, Dow S. Continuous low-dose oral chemotherapy for adjuvant therapy of splenic hemangiosarcoma in dogs. J Vet Intern Med 2007;21:764–769.

30. Leslie NR, Yang X, Downes CP, Weijer CJ. PtdIns(3,4,5)P(3)-dependent and -independent roles for PTEN in the control of cell migration. Curr Biol 2007;17:115–125.

31. Lin WW, Karin M. A cytokine-mediated link between innate immunity, inflammation, and cancer. J Clin Invest 2007;117:1175–1183.

32. Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation. Nature 2008;454:436–444.

33. Marion MJ, Boivin-Angele S. Vinyl chloride-specific mutations in humans and animals. IARC Sci Publ 1999;150:315–324.

34. Mintz MB, Sowers R, Brown KM, Hilmer SC, Mazza B, Huvos AG, Meyers PA, Lafleur B, McDonough WS, Henry MM, Ramsey KE, Antonescu CR, Chen W, Healey JH, Daluski A, Berens ME, Macdonald TJ, Gorlick R, Stephan DA. An expression signature classifies chemotherapy-resistant pediatric osteosarcoma. Cancer Res 2005;65:1748–1754.

35. Modiano JF, Breen M, Lana SE, Ehrhart N, Fosmire SP, Thomas R, Jubala CM, Lamerato-Kozicki AR, Ehrhart EJ, Schaack J, Duke RC, Cutter GC, Bellgrau D. Naturally occurring translational models for development of cancer therapy. Gene Ther Mol Biol 2005;10:31–40.

36. Modiano JF, Breen M, Valli VE, Wojcieszyn JW, Cutter GR. Predictive value of p16 or Rb inactivation in a model of naturally occurring canine non-Hodgkin's lymphoma. Leukemia 2007;21:184–187.

37. Okumura K, Zhao M, Depinho RA, Furnari FB, Cavenee WK. Cellular transformation by the MSP58 oncogene is inhibited by its physical interaction with the PTEN tumor suppressor. Proc Natl Acad Sci USA 2005;102:2703–2706.

38. Paoloni M, Davis S, Lana S, Withrow S, Sangiorgi L, Picci P, Hewitt S, Triche T, Meltzer P, Khanna C. Canine tumor cross-species genomics uncovers targets linked to osteosarcoma progression. BMC Genomics 2009;10:625.

39. Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jeffrey SS, Rees CA, Pollack JR, Ross DT, Johnsen H, Akslen LA, Fluge O, Pergamenschikov A, Williams C, Zhu SX, Lonning PE, Borresen-Dale AL, Brown PO, Botstein D. Molecular portraits of human breast tumours. Nature 2000;406:747–752.

40. Pettersson A, Nagy JA, Brown LF, Sundberg C, Morgan E, Jungles S, Carter R, Krieger JE, Manseau EJ, Harvey VS, Eckelhoefer IA, Feng D, Dvorak AM, Mulligan RC, Dvorak HF. Heterogeneity of the angiogenic response induced in different normal adult tissues by vascular permeability factor/vascular endothelial growth factor. Lab Invest 2000;80:99–115.

41. Ponce F, Marchal T, Magnol JP, Turinelli V, Ledieu D, Bonnefont C, Pastor M, Delignette ML, Fournel-Fleury C. A morphological study of 608 cases of canine malignant lymphoma in France with a focus on comparative similarities between canine and human lymphoma morphology. Vet Pathol 2010;47:414–433.

42. Priester WA, McKay FW. The occurrence of tumors in domestic animals. Natl Cancer Inst Monogr 1980;54:1–210.

43. Prymak C, McKee LJ, Goldschmidt MH, Glickman LT. Epidemiologic, clinical, pathologic, and prognostic characteristics of splenic hemangiosarcoma and splenic hematoma in dogs: 217 cases (1985). J Am Vet Med Assoc 1988;193:706–712.

44. Rao NA, van Wolferen ME, Gracanin A, Bhatti SF, Krol M, Holstege FC, Mol JA. Gene expression profiles of progestin-induced canine mammary hyperplasia and spontaneous mammary tumors. J Physiol Pharmacol 2009;60(Suppl 1):73–84.

45. Rao NA, van Wolferen ME, van den Ham R, van Leenen D, Groot Koerkamp MJ, Holstege FC, Mol JA. cDNA microarray profiles of canine mammary tumour cell lines reveal deregulated pathways pertaining to their phenotype. Anim Genet 2008;39:333–345.

46. Rusk A, McKeegan E, Haviv F, Majest S, Henkin J, Khanna C. Preclinical evaluation of antiangiogenic thrombospondin-1 peptide mimetics, ABT-526 and ABT-510, in companion dogs with naturally occurring cancers. Clin Cancer Res 2006;12:7444–7455.

47. Saito T, Barbin A, Omori Y, Yamasaki H. Connexin 37 mutations in rat hepatic angiosarcomas induced by vinyl chloride. Cancer Res 1997;57:375–377.

48. Scott MC, Sarver AL, Gavin K, Thayanithy V, Getzy DM, Newman R, Cutter GC, Lindblad-Toh K, Kisseberth WC, Hunter LE, Subramanian S, Breen M, Modiano JF. Identification of distinct molecular subtypes of osteosarcoma using an innovative canine-human comparative approach. Bone 2011;49:356–367.

49. Seelig DM, Avery PR, Kisseberth WC, Modiano JF. T-cell lymphoproliferative diseases. In: Schalm's Veterinary Hematology. Weiss DJ and Wardrop KJ (eds.), 6th ed, pp. 525–539. Wiley-Blackwell, Hoboken, NJ, 2010

50. Selvarajah GT, Kirpensteijn J, van Wolferen ME, Rao NA, Fieten H, Mol JA. Gene expression profiling of canine osteosarcoma reveals genes associated with short and long survival times. Mol Cancer 2009;8:72.

51. Sherr CJ. Principles of tumor suppression. Cell 2004;116:235–246.

52. Skubitz KM, D'Adamo DR. Sarcoma. Mayo Clin Proc 2007;82:1409–1432.

53. Spangler WL, Culbertson MR. Prevalence, type, and importance of splenic diseases in dogs: 1,480 cases (1985–1989). J Am Vet Med Assoc 1992;200:829–834.

54. Spielman RS, Bastone LA, Burdick JT, Morley M, Ewens WJ, Cheung VG. Common genetic variants account for differences in gene expression among ethnic groups. Nat Genet 2007;39:226–231.

55. Starkey MP, Murphy S. Using lymph node fine needle aspirates for gene expression profiling of canine lymphoma. Vet Comp Oncol 2010;8:56–71.

56. Stockmann C, Doedens A, Weidemann A, Zhang N, Takeda N, Greenberg JI, Cheresh DA, Johnson RS. Deletion of vascular endothelial growth factor in myeloid cells accelerates tumorigenesis. Nature 2008;456:814–818.

57. Swanton C, Caldas C. Molecular classification of solid tumours: towards pathway-driven therapeutics. Br J Cancer 2009;100:1517–1522.

58. Tamburini BA, Phang TL, Fosmire SP, Trapp S, Slansky JE, Sharkey LC, Cutter GR, Bellgrau D, Gemmill RM, Hunter LE, Modiano JF. Gene expression profiling identifies inflammation and angiogenesis as distinguishing features of canine hemangiosarcoma. BMC Cancer 2010;10:619.

59. Tamburini BA, Trapp S, Phang TL, Schappa JT, Hunter LE, Modiano JF. Gene expression profiles of sporadic canine hemangiosarcoma are uniquely associated with breed. PLoS ONE 2009;4:e5549.

60. Tate G, Suzuki T, Mitsuya T. Mutation of the PTEN gene in a human hepatic angiosarcoma. Cancer Genet Cytogenet 2007;178:160–162.

61. Thamm DH, Dickerson EB, Akhtar N, Lewis R, Auerbach R, Helfand SC, MacEwen EG. Biological and molecular characterization of a canine hemangiosarcoma-derived cell line. Res Vet Sci 2006;81:76–86.

62. Thayanithy V, Sarver AL, Kartha RV, Park C, Scott MC, Angstadt AC, Breen M, Steer CJ, Modiano JF, Subramanian S. Perturbation in 14q32 miRNAs-MYC-miR-17~92 gene network contributes to osteosarcoma and is associated with survival outcome. Submitted

63. Thomson SA, Kennerly E, Olby N, Mickelson JR, Hoffmann DE, Dickinson PJ, Gibson G, Breen M. Microarray analysis of differentially expressed genes of primary tumors in the canine central nervous system. Vet Pathol 2005;42:550–558.

64. U'Ren LW, Biller BJ, Elmslie RE, Thamm DH, Dow SW. Evaluation of a novel tumor vaccine in dogs with hemangiosarcoma. J Vet Intern Med 2007;21:113–120.

65. Uva P, Aurisicchio L, Watters J, Loboda A, Kulkarni A, Castle J, Palombo F, Viti V, Mesiti G, Zappulli V, Marconato L, Abramo F, Ciliberto G, Lahm A, La Monica N, de Rinaldis E. Comparative expression pathway analysis of human and canine mammary tumors. BMC Genomics 2009;10:135.

66. Vail DM, MacEwen EG. Spontaneously occurring tumors of companion animals as models for human cancer. Cancer Invest 2000;18:781–792.

67. Vail DM, MacEwen EG, Kurzman ID, Dubielzig RR, Helfand SC, Kisseberth WC, London CA, Obradovich JE, Madewell BR, Rodriguez CO Jr, et al. Liposome-encapsulated muramyl tripeptide phosphatidylethanolamine adjuvant immunotherapy for splenic hemangiosarcoma in the dog: a randomized multi-institutional clinical trial. Clin Cancer Res 1995;1:1165–1170.

68. Valli VE, Jacobs RM, Parodi AL, Vernau W, Moore PF. Classification of hematopoietic tumors of domestic animals, Second Series ed. AFIP- American Registry of Pathology, Washington, DC, 2002

69. Valli VE. B-cell Tumors. In: Schalm's Veterinary Hematology. Weiss DJ and Wardrop KJ (eds.), 6th ed, pp. 491–510. Wiley-Blackwell, Hoboken, NJ, 2010

70. Valli VE, Vernau W. Classification of leukemia and lymphoma. In: Schalm's Veterinary Hematology. Weiss DJ and Wardrop KJ (eds.), 6th ed, pp. 451–454. Wiley-Blackwell, Hoboken, NJ, 2010

71. van't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M, Peterse HL, van der Kooy K, Marton MJ, Witteveen AT, Schreiber GJ, Kerkhoven RM, Roberts C, Linsley PS, Bernards R, Friend SH. Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002;415:530–536.

72. Wallace TA, Prueitt RL, Yi M, Howe TM, Gillespie JW, Yfantis HG, Stephens RM, Caporaso NE, Loffredo CA, Ambs S. Tumor immunobiological differences in prostate cancer between African-American and European-American men. Cancer Res 2008;68:927–936.

73. Wang X, Shi Y, Wang J, Huang G, Jiang X. Crucial role of the C-terminus of PTEN in antagonizing NEDD4-1-mediated PTEN ubiquitination and degradation. Biochem J 2008;414:221–229.

74. Weihrauch M, Bader M, Lehnert G, Koch B, Wittekind C, Wrbitzky R, Tannapfel A. Mutation analysis of K-ras-2 in liver angiosarcoma and adjacent nonneoplastic liver tissue from patients occupationally exposed to vinyl chloride. Environ Mol Mutagen 2002;40:36–40.

75. Wensman H, Goransson H, Leuchowius KJ, Stromberg S, Ponten F, Isaksson A, Rutteman GR, Heldin NE, Pejler G, Hellmen E. Extensive expression of craniofacial related homeobox genes in canine mammary sarcomas. Breast Cancer Res Treat 2009;118:333–343.

76. Wu Y, Zhou BP. Inflammation: a driving force speeds cancer metastasis. Cell Cycle 2009;8:3267–3273.

77. Yoder MC, Mead LE, Prater D, Krier TR, Mroueh KN, Li F, Krasich R, Temm CJ, Prchal JT, Ingram DA. Redefining endothelial progenitor cells via clonal analysis and hematopoietic stem/progenitor cell principals. Blood 2007;109:1801–1809.