Andreas Doll and Marta Gonzalez contributed equally to this work
An orthotopic endometrial cancer mouse model demonstrates a role for RUNX1 in distant metastasis
Version of Record online: 3 FEB 2009
Copyright © 2009 UICC
International Journal of Cancer
Volume 125, Issue 2, pages 257–263, 15 July 2009
How to Cite
Doll, A., Gonzalez, M., Abal, M., Llaurado, M., Rigau, M., Colas, E., Monge, M., Xercavins, J., Capella, G., Diaz, B., Gil-Moreno, A., Alameda, F. and Reventos, J. (2009), An orthotopic endometrial cancer mouse model demonstrates a role for RUNX1 in distant metastasis. Int. J. Cancer, 125: 257–263. doi: 10.1002/ijc.24330
- Issue online: 12 MAY 2009
- Version of Record online: 3 FEB 2009
- Accepted manuscript online: 3 FEB 2009 12:00AM EST
- Manuscript Accepted: 19 DEC 2008
- Manuscript Received: 19 AUG 2008
- Marató TV3. Grant Number: 050431
- Ministerio de Educación y Ciencia. Grant Number: SAF 2008-03996
- Fondo de Investigaciones Sanitarias. Grant Numbers: 99/0897, 01/10076, 02/0733, 05/00240
- Departament de Universitats, de Recerca i de Societat de la Informacio de la Generalitat de Catalunya. Grant Number: 2005SGR00553
- Instituto de Salud Carlos III. Grant Number: RTIC RD06/0020/0058
- Institut Catala de la Salut
- endometrial cancer;
- animal model;
- orthotopic implantation;
Endometrial carcinoma is the most common malignancy of the female genital tract in industrialized countries. Metastasis is the major cause of endometrial cancer deaths. Therefore, there is a vital need for clinically relevant in vivo models allowing the elucidation of the molecular and cellular mechanisms underlying metastatic behavior. In this study, we describe an innovative experimental orthotopic model of human endometrial carcinoma. Implantation in the bifurcation of the uterine horns resulted in tumors integrated into the myometrial compartment, which can be used and further exploited for the study of in vivo angiogenesis, myometrial invasion, and the metastatic capacity of endometrial cancer cells. This orthotopic model also represents a suitable tool to analyze how tumorigenesis and distant metastasis of endometrial cancer might be influenced by gene alteration, by modulating its expression in the original cancer cell line. One of the candidate genes implicated in endometrial cancer is the transcription factor RUNX1. The over-expression of RUNX1 in the endometrial cancer cell line HEC1A and the transplantation of these cells to the uterus of nude mice were associated specifically with distant metastasis in the lung. RUNX1 plays a role in the establishment of metastases in endometrial cancer. Translated to the clinics, these models would be equivalent to an advanced undifferentiated carcinoma with node affectation (stage IIIC) and distant metastasis (stage IVB). These patients would be candidates for adjuvant therapy, not efficient until today, and therefore, our models are actually suitable for the design and evaluation of experimental therapies. © 2009 UICC
Endometrial carcinoma is the most frequently diagnosed malignancy of the female genital tract in industrialized countries. Its incidence is estimated at 15–20 of 100,000 women per year. Overall, 90% of the cases are sporadic whereas the remaining 10% arise from a genetic background. Approximately, 80% of the cases are diagnosed at the FIGO stage I, because of early clinical signs (peri- and post-menopausal metrorrhagia). They are usually well to moderately differentiated endometrioid adenocarcinomas, which are confined to the uterine corpus at diagnosis, and thus, most can be cured. Notwithstanding the excellent prognosis, myometrial infiltration and particularly distant metastasis are the most devastating points in EC; approximately 25% of the patients who have undergone surgical staging are found to have extra-uterine disease.1, 2 Since endometrial carcinoma spreads primarily to the pelvic and para-aortic lymph nodes, as well as to adnexa and pelvic viscera, distant metastases via the hematogenous route have a low incidence. Even so, pulmonary metastases, which usually occur via the hematogenous route, have been reported to be one of the most common distant metastasis stemming from endometrial carcinoma. However, its incidence ranges from 2.3 to 4.6%.3, 4 The precise molecular events that occur during the development, progression/invasion and formation of metastasis in endometrial carcinoma are largely uncharacterized and are still relatively poorly understood.5–8
Experimental models based on human endometrial adenocarcinoma cells are widely used; the existing animal-derived models mainly use subcutaneous xenografts, injecting tumor cells into the subcutis of nude rodents.9 Although there is evidence that these models can predict clinical efficacy, they still have significant limitations, which make the development of more complex models desirable. The subcutaneous tumor model, for example, often causes central tumor necrosis because of the low blood supply in the subcutaneous tissue. They normally show solitary tumor growth without invasion, because of the lack of surrounding visceral organs and, also, due to the unusual tissue compartment (the subcutis) where tumor growth occurs. Finally, and most importantly, they show a lack of metastatic behavior, which is most closely linked to clinical outcome and the most important cause of death from cancer.
A step toward the development of more complex models was first described in 1984 by Sordat and Wang10 in colon cancer. Later Fidler et al.11, 12 observed that the orthotopic injection of suspensions of human tumor cells in the corresponding organs of nude mice can enhance their metastatic capability. Similar results have also been achieved with the orthotopic implantation of cell lines from human lung cancer, pancreatic cancer, bladder cancer, ovarian cancer, colon cancer and stomach cancer.13 However, recent reports have indicated that cell suspensions used for orthotopic implantation may not express the full metastatic potential of the original tumor.14, 15 To address this problem, surgical orthotopic implantation of histologically intact tissue is currently being widely used, maintaining a three-dimensional tissue architecture, cell-to-cell interactions and subsequent tumor angiogenesis, which all seem to be important to metastatic behavior.14–16 The biological characteristics of these models, such as improved tumor take and invasive and metastatic properties are now well established.
Although an animal model of endometrial cancer, such as the subcutaneous implantation of endometrial cancer cells in a xenograft model, has been used in experimental oncology, there has been no report of an appropriate orthotopic endometrial cancer model. A surgical orthotopic implantation animal model of endometrial cancer, which reflects invasion and metastatic capacity, must still be established. The conditions that cause myometrial invasion, cellular dissemination to vessels, and finally, distant metastasis, as well as how these processes are influenced by gene expression, remain to be elucidated.
One of the candidate genes implicated in endometrial cancer and metastasis is the transcription factor RUNX1. It belongs to the RUNX family of transcription factors encoded by three genes, RUNX1, RUNX2 and RUNX3.17, 18 The RUNX genes are closely related to each other and are essential for hematopoiesis, osteogenesis and neurogenesis, but they are also important for other developmental processes.19 They bind DNA as heterodimers with the non-DNA binding protein core binding factor (CBF)β, creating a RUNX/CBFβ complex, termed CBF.20 The RUNX proteins bind to the consensus DNA site (TGT/CGGT) via the runt homology domain (rhd), and they function as both transcriptional activators and repressors, depending upon promoter and cellular context.21, 22 RUNX1 is of special interest in EEC, because myometrial affectation, as the initial event in tumor invasion, and distant dissemination determine an increase in the rate of recurrence after a first surgical treatment and a decrease in survival at the 5-year follow-up.8, 23
This article reports an orthotopic endometrial cancer model with node metastases in nude mice using a human endometrial cancer cell line, since the interpretation of experiments using subcutaneously implanted endometrial tumors is limited, due to different tissue-specific factors. By modulating RUNX1 expression, in addition to addressing in vivo invasion and the metastatic capacity of endometrial cancer cells, we investigated how these processes are influenced by gene expression.
Material and methods
Cell lines, constructs and stable transfection
Human endometrial carcinoma 1A (HEC1A) cell line was cultured as described.7 pEGFP-C2 vector was purchased from BD Biosciences (Franklin Lakes, NJ). Plasmid construction consisted of the full-length human RUNX1 being removed with EcoRI from a pSV-hRUNX1 construct; it was then inserted at the EcoRI position at the multiple cloning site of the pEGFP-C2 vector. The correct orientation and reading frame were confirmed by sequencing.
Stable HEC1A cell lines were either transfected with the pEGFP-C2 vector alone or the hRUNX1 containing pEGFP-C2 vector. Stably expressing cells were selected by incubation with 500 μg/ml of geneticin G-418 sulphate (GIBCO, Invitrogen, Carlsbad, CA).
Animals and tumor implantation procedure
Athymic Swiss Webster mice, 5-week-old female (Charles River Laboratories, Wilmington, MA), were used for this study. The animals were housed in individually ventilated cage units and were maintained under pathogen-free conditions. Food and water were provided ad libitum. Animal care and experiments were carried out in accordance with the guidelines of the Spanish Council on Animal Care and the Institutional Review Board.
Subcutaneous xenografts were established by the injection of 2 × 106 HEC1A (H), HEC1A/GFP (HG) and HEC1A/GFP/RUNX1 (HGR) tumor cells in 200 μl PSB into the subscapular region of 12 nude mice (for each cell line). The animals were killed at 6 weeks. Primary tumors were subcutaneously transplanted to other nude mice for a second passage, to amplify tumor tissue.
For orthotopic implantation, 24 nude mice were anesthetized with intraperitoneal 1.25% sodium pentobarbital (45 mg/kg), and the lower abdomen and back were swabbed with 70% alcohol. A longitudinal incision of the lower abdomen of about 15–20 mm in length was made (medial laparotomy). The retroperitoneum was then opened and the uterus visualized and individualized.
One tissue block of approximately 1 mm3 was implanted onto the posterior face of the uterus making a pocket and fixed with a 5–0 surgical suture to isolate tumor tissue from the rest of the organs in the abdominal cavity. The organs were reintroduced into the abdominal cavity and the retroperitoneum and skin were then closed with a 5–0 surgical suture.
Six weeks later, the animals were killed by cervical dislocation for necropsy. Then, for all of the animals, the peritoneal cavity was opened and examined macroscopically. Samples of the lungs, brains and tumors were snap-frozen in liquid nitrogen for RNA extraction. The remaining tissue was formalin-fixed and processed for routine histological examination, hematoxylin–eosin (H&E) staining and immunohistochemistry (IHC).
Immunohistochemical staining was performed on 24 hr formalin fixed paraffin-embedded 4 μm sections using routine procedures. All stainings were semiautomated and performed on a TechMate 500 plus (DAKO, Glostrup, Denmark) by using the DAKO Envision+ Detection Kit, as recommended by the manufacturer.
Primary monoclonal antibodies, clones, and dilutions were as follows: β-catenin (βcatenin-1), 1:200; ERα-receptor (1D5), 1:400; PR-receptor (PgR636), 1:400; HER2/NEU (A0485), 1:350; PTEN (6H2.1), 1:100; Ki-67 antigen (MIB-1), 1:100; p53 (DO-7), 1:75; Factor VIII (rabbit polyclonal antibody), 1:320, RUNX1 1:50. All antibodies were from DAKO, except RUNX1 (Santa Cruz).
The deparaffinized tissue sections were treated with heat before the IHC staining procedure for epitope retrieval by immersion of tissue sections in citrate buffer pH 6.0 in an autoclave at 121°C for 1 min.
Detection of disseminated tumor cells by RT-PCR
The extraction of total RNA from the frozen tumor samples, weighing approximately 20 mg, was performed using the Rneasy Mini kit (Qiagen, Hilden, Germany), and 1 μg of the total RNA was subjected to one-step RT-PCR (Qiagen), according to the manufacturer's protocol.
PCR detection of tumoral cells was based on the amplification of the β2-microglobulin gene (β2-m) as a marker for cells of human origin. Polymerase chain reaction was performed using forward primer β2-m: CCATCCGACATTGAAGTTGA and reverse primer β2-m: TGGAGCAACCTGCTCAGATA (annealing temperature 57°C, 35 cycles). PCR detection was normalized with GAPDH, forward primer GAPDH: 5′-CGTCTTCACCACCATGGAGA-3′ and reverse primer GAPDH: 5′-CGGCCATCACGCCACAGTTT-3′ (annealing temperature 61°C, 27 cycles). All primers were obtained from Thermo Electron Corporation (Waltham, MA).
Orthotopic tumor grafting generated a clinically relevant endometrial cancer model
The generation of an orthotopic model for endometrial cancer, by surgical implantation in the uterus of a subcutaneous tumor, is described in Figure 1. Primary tumors were derived from the injection of 2 × 106 HEC1A tumor cells, derived from an adenocarcinoma of the endometrium (Fig. 1a), in 200 μl PSB into the flank region of nude mice (Fig. 1b). The tumor volume doubling time was approximately 7 days and, similar to the original HEC1A cell line, the HEC1A xenografts were poorly differentiated, generally lacking glandular differentiation (data not shown).
As can be observed macroscopically, the subcutaneous tumors grew encapsulated and, although superficial vascularization demonstrated a response of the host to the tumor, no evident interaction with the environment could be described (Fig. 1c). To overcome this limitation and to try to develop a more clinically relevant model, we implanted the endometrial tumor into the uterus context. To this purpose, nude mice were anesthetized with intraperitoneal 1.25% sodium pentobarbital (45 mg/kg), and a longitudinal incision of the lower abdomen of about 15–20 mm in length was made (medial laparotomy). The retroperitoneum was then opened and the uterus visualized and individualized. One tissue block from the subcutaneous tumor of approximately 1 mm3 was implanted onto the posterior face of the uterus making a pocket (Figs. 1d,1e) and fixed with a 5–0 surgical suture to isolate tumor tissue from the rest of the organs in the abdominal cavity (Fig. 1f). The organs were reintroduced into the abdominal cavity, and the retroperitoneum and skin were then closed with a 5–0 surgical suture. Six weeks later, the animals were killed by cervical dislocation for necropsy, and the peritoneal cavity was opened and examined macroscopically. The grafting of approximately 1 mm3 of HEC1A subcutaneous xenograft into the posterior face of the uterus of nude mice produced tumor take rates of 96% (n = 24). As can be observed, orthotopic tumors grew in the area of implantation in the uterus, without affecting other organs or the peritoneum (Fig. 1g). Of interest, the macroscopic examination indicated a close direct contact with the host tissue and suggests a contribution from the microenvironment to the development of the tumor (Figs. 1h–1i).
Furthermore, we speculated that in this model the hypothetical dissemination of tumor cells arose through the same mechanisms that human endometrial cancer cells make use of to metastasize (i.e., myometrial invasion, angiogenesis). To confirm this hypothesis, tumor tissue was formalin-fixed and processed for histological examination by H&E staining and immunohistochemistry. First, the grafts showed poorly differentiated features and the local invasion of the hosts' uterine tissue (see Figs. 2a–2c). The histology of the mice uteri showed both myometrial infiltration (Figs. 2a,2b) and vascular invasion (Fig. 2c). The latter was confirmed by staining the endothelial cells lining the lumen of vessels with anti-Factor VIII (Fig. 2d).
Second, none of the tumors expressed estrogen receptor alpha and progesterone receptor, whereas a nuclear staining pattern in the circumventing myometrium of the mice was observed (Fig. 3; upper left panel). An adjacent section stained positively with Ki-67, showing a nuclear staining pattern in the tumor, though negative in the uterus (Fig. 3; upper middle panel). Tumoral cells showed p53 nuclear staining. Normal endometrial cells did not show immunoreactivity for p53 (Fig. 3; upper right panel). Furthermore, the orthotopic grafts intensely expressed β-catenin (Fig. 3; lower left panel) and HER2/neu (Fig. 3; lower middle panel) with a focal labeling of the cell membranes and also positive staining in the normal endometrium (data not shown). At last, PTEN immunoreactivity was shown in the tumor and normal endometrium (Fig. 3; lower right panel), in accordance with the described expression levels of PTEN in the HEC1A cells that originated the tumors.
All of this data points to an orthotopic endometrial cancer model showing the characteristics of an aggressive tumor with high levels of ki67 and p53 and low levels of hormone receptors.
Orthotopic xenografts developed metastases to paraaortic lymph nodes
Once demonstrated that the orthotopic endometrial cancer model behaved as human carcinomas in myometrial invasion and vascular dissemination, we analyzed whether the pattern of metastasis also reproduced the one usually present in the clinics, i.e., lymph node affectation. Orthotopically implanted mice developed metastases to para-aortic nodes in 75% of the cases (9 out of 12 grafts). Lymph node lesions had histological and cytological features of the original tumor cells, including a solid growth pattern, moderately pale to clear cytoplasm, and oval nuclei and nucleoli (Fig. 4).
These results confirmed this orthotopic model to be a clinically relevant model for endometrial cancer.
Over-expression of RUNX1 resulted in orthotopic xenografts developing distant micrometastasis
Once we characterized the orthotopic animal model for endometrial cancer as suitable for the study of molecular and cellular events related to the early steps of invasion, tumor dissemination and metastasis, we sought to evaluate the contribution of RUNX1 over-expression to this process. We first observed that the orthotopic tumors, which originated from the HEC1A cells that over-expressed RUNX1, behaved as those that originated from the untransfected HEC1A cells or from cells transfected with the empty vector, in myometrial infiltration, vascular invasion, or for the pattern of expression of the different immunohistochemistry markers described above (data not shown).
We then evaluated the promotion of distant metastasis. To detect the presence of disseminated tumor cells in a sensitive and reproducible manner, which could eventually be established as micrometastases in the dissected tissues, we analyzed the expression of the human specific β2-microglobulin and GFP (data not shown) as a marker for maintenance of the RUNX1 overexpression by RT-PCR (see “Material and methods”). Only the cDNA coding for these genes from the human tumor cells implanted in the host tissues, and not the homologous genes of murine origin, were amplified (see tumor and brain in Fig. 5a, as human positive and murine negative controls, respectively).
We analyzed the lung as the main receptor of metastasis in endometrial cancer at the experimental end point of 6 weeks, and the results of the RT-PCR detection of tumor cells are shown in Fig. 5. All mice orthotopically bearing tumors from the RUNX1 over-expressing HEC1A cells were considered positive for the presence of disseminated tumor cells in the lungs, whereas only two out of eight of the HEC1A mice and one out of eight of the HEC1A-GFP mice were found to be positive in the lungs. The brains and tumors were tested for the presence of disseminated tumor cells as negative tissues and positive β2-microglobulin controls, respectively (Fig. 5a).
These results were confirmed by immunohistochemistry. Tumors originating from the RUNX1 over-expressing HEC1A cells developed micrometastases to the lungs (Fig. 5b), whereas control mice showed normal lung histology (data not shown). Lung lesions caused by RUNX1 had histological and cytological features of the original tumor cells, including a solid growth pattern, moderately pale to clear cytoplasm, and oval nuclei and nucleoli (Fig. 5b). Similar to the histological results, immunohistochemistry with human-specific RUNX1, Ki-67 and p53 detected specific expression in the lung lesions. Ki-67 showed an intense positive staining in intravascular tumor cells (Fig. 5c). A positive staining of p53 was detected in a group of intravascular tumor cells and was positive in metastatic foci (Fig. 5c). IHC analyses confirmed weak expression of HER2/neu and ßcatenin in the metastases in the lungs (data not shown). Estrogen receptor α and progesterone receptor were not expressed in any of the metastases of the engrafted animals (data not shown).
Subcutaneous tumor implantation has been a standard method for establishing animal models of human cancer.24 Although such models have helped to understand the nature of tumors and have lead to some therapeutic approaches to human cancer, many problems remain unsolved. One major problem is that tumors derived from patients and later implanted subcutaneously into an immunodeficient animal no longer behave as they once did in the human patient. Although a tumor can sometimes grow subcutaneously, it is often encapsulated and usually fails to metastasize either regionally or to distant sites.12 In our laboratory, we generated subcutaneous tumors by injection of the endometrial cancer cell line HEC1A. As shown in Fig. 1, subcutaneous tumors were indeed encapsulated and no dialogue with the environment could be expected. An alternative animal model system that can be potentially applicable for drug testing and gene-target validation, not only in endometriosis, but also in other various types of neoplastic disease, has been recently described.25 This animal model system involves the transplantation of relevant primary culture cells or cell lines beneath the kidney capsule of immunodeficient mice and has demonstrated the capacity for tissue regeneration and reconstruction.
To develop an animal model that more closely mimics the clinics, we performed an orthotopic implantation of the subcutaneous tumors. In theory, tumors that develop orthotopically (from the Latin for “correct place”) will more closely resemble human tumors, because the nearby blood vessel system and supporting tissues will better mirror a tumor's microenvironment, realistically influencing its growth. Orthotopic implantation has been used to develop rodent models of metastatic human cancer.11, 26 In the first generation of these models, cell line suspensions or disaggregated tumor cells were injected into the organ of the mouse, which corresponded to the organ from which the human tumor was derived. This method allowed metastasis to occur in certain cases.27, 28 However, the cell line suspensions and disaggregated cells used for orthotopic implantation did not have the three-dimensional tissue structure of human tumor tissue. Lack of tissue architecture may lead to changes in the biological behavior of a tumor and could be the reason for greatly reduced metastatic rates, perhaps because of subsequent insufficient tumor angiogenesis after the implantation of a cell suspension. In this light, we have developed surgically orthotopic implant models, utilizing intact tissue, such as that obtained directly from surgery or from human tumors growing in nude mice. As shown for other models of human cancer in nude mice,29–33 we were able to demonstrate patient-like tumor behavior of the orthotopic endometrial cancer model in myometrial invasion and vascular dissemination, and the development of metastasis to clinically-relevant organs.
Moreover, the immunohistochemical characterization of the orthotopic model indicated that this model was correspondent to an advanced carcinoma with low levels of hormone receptors and high levels of Ki-67 and p53 markers. Regarding p53, mutations are considered late events of tumorigenesis, since they are predominantly found in poorly differentiated tumors. Four regions of the p53 gene were selected for analysis based on their high rate of evolutionary conservation: exon 5 and exon 6, exon 7 and exon 8. Direct sequencing of the corresponding PCR bands revealed an A-to-G change in nucleotide 15360 of the TP53 gene, which resulted in a substitution of glutamine for arginine (data not shown). Although we do not know whether this particular mutation compromises the functionality of the p53 protein, this could be a reason for the nuclear accumulation of p53. In this model, the high Ki-67 expression levels (>99% of the tumor cells) and the accumulation of p53 suggest an eventual uncontrolled cell division and, therefore, an accumulation of DNA damage and mutations associated with advanced carcinoma.
Our orthotopic model developed metastases to para-aortic nodes (PAN) in 75% of the cases, reinforcing the assumption that this model represents an advanced stage of endometrial cancer and demonstrating again the adequacy of this model system to the clinics. Likewise, distant metastasis in intraperitoneally injected models arose from the mechanical spread of the implanted tumor material, rather than a true metastasis. Tumors grown subcutaneously in nude mice rarely metastasize; so, both models do not reflect optimal conditions for preclinical screenings for antimetastatic and anti-angiogenic agents. In this study, we could also characterize a role for RUNX1 as an inducer of distant metastasis in endometrial cancer. The mammalian RUNX genes (RUNX1–3) are transcription factors that play essential, lineage-specific roles in development. A growing body of evidence implicates these genes as mutational targets in cancer, where, in different contexts, individual family members have been reported to act as tumor suppressors, dominant oncogenes, or mediators of metastasis. RUNX1 has been previously described as being up-regulated in endometrioid endometrial carcinoma and purposed to play a role in endometrial tumorigenesis.8 Here, in the context of this model, we describe the influence of the over-expression of RUNX1 as a candidate gene for invasion and metastasis.
The rhd mediates protein-protein interactions with a variety of partners, including CBFβ and members of the Ets transcription factor family.20, 34 However, CBF/AML proteins are relatively weak transcriptional activators in isolation, and they potently enhance transcription rates in cooperation with several transcription factors (e.g., Ets-1, cMyb, CCAAT/enhancer protein) via cooperative DNA binding or interactions with coactivators.35 Interestingly, we have previously described the Ets family member ERM/ETV5, specifically upregulated in endometrial cancer associated with an initial switch to myometrial infiltration, acting through MMP-2 gelatinolytic activity to confer invasive capabilities.6, 7, 36 Further work must be performed to fully characterize the mechanisms underlying the promotion of distant metastasis by RUNX1 modulation.
It must be pointed out that the poor survival rate of patients with early endometrial cancer has been related to the extra-pelvic spread of the cancer and is significantly higher in more advanced stages.37 Therefore, further advances in controlling local and distant metastatic disease are needed. These models could be used to gain further insight into and to develop new therapeutic treatments for human endometrial cancer and lymph-node and lung metastasis, in particular. Effective therapies will need to target not only the tumor (the seed), but also its microenvironment (the soil). Therefore, orthotopic models will be necessary research tools, since they better reproduce the microenvironment needed to study these tumors and their metastasis.12 An example has been recently published for a combination of antivascular therapy with docetaxel in endometrial cancer.27
Collectively, in nude mice these orthotopically implanted tumors could be good candidates for the simulation of treatment for primary tumors, clinical progression and the metastasis of endometrial cancer. Translated to the clinics, these models would be equivalent to an advanced undifferentiated carcinoma with node affectation and distant metastasis. These patients would be candidates for adjuvant therapy, not efficient until today; thus, our model would actually be suitable for the design and evaluation of experimental therapies. It must be mentioned that not only the adequacy of these models to the clinics in appropriate tumor environment, myometrial invasion procedure, angiogenesis or tumor cell dissemination make it a good preclinical model system. The reproducibility of the model, the high rates of tumor uptake and the timing for tumor progression, as well as the appearance of node and distant metastasis also account for experimental therapeutic assays. Finally, we also present evidence of the utility of these models for the evaluation of how the process of metastasis is influenced by gene-expression. This will ideally improve the clinical treatment of endometrial cancer by pharmaceuticals, since invasion and metastasis are the major causes of death from cancer.38