Introduction

Although it affects only 1 in 70 women, ovarian cancer is the most lethal of the gynecologic malignancies¹. The high rate of mortality associated with this disease may be attributable to the late diagnosis common to ovarian cancer, due to the initial disease being virtually asymptomatic, as well as the lack of reliable diagnostic screening methods. Furthermore, successful treatment is limited by the high rate of chemoresistance in recurrent ovarian cancer. Research on the efficacy of novel therapeutics is of the utmost importance for improvement of patient survival, and the development of well-characterized and representative models of the disease is essential for this mandate.

Epithelial ovarian cancer comprises five histological subtypes (serous, mucinous, clear cell, endometrioid, and transitional cell/Brenner tumors)². Tumors that do not show the distinctive histologic features of any of the subtypes are designated as undifferentiated surface epithelial adenocarcinomas. Studies suggest that the developmental pathways for these subtypes, and consequently their gene expression profiles and responsiveness to various therapeutic strategies, are fundamentally different^3,4. Improved understanding of the etiology of the disease and of the pathologic and molecular characteristics of the different subtypes would allow for more strategic and selective treatment regimens for individual patients.

Since the first report in 1969⁵, xenografts of human cancer cells into immune-deficient mice have been useful and widely used tools both to demonstrate the tumorigenicity of cells and to test the efficacy of therapeutic compounds in vivo. The importance of well-characterized cell lines of ovarian cancer is unquestionable, and the original references for common ovarian cancer cell lines provide brief descriptions of the tumor of origin (e.g., OCC1⁶, OV2008⁷, C13^*8, OVCA 429 and OVCA 433⁹, SKOV3¹⁰, and ES-2¹¹); yet generally speaking, the histological characteristics of tumors that develop from cell lines in xenograft models are minimally discussed. There are a few exemplary reports that detail important histological information about both the specimens from which the cell lines were derived and the xenograft tumors that result following heterotransplantation. For example, the characterization of HOC-1, HOC-7, and HEY cells by Buick et al. reported that the HOC lines were derived from well-differentiated serous adenocarcinomas, the HEY cells from a moderately differentiated papillary cystadenocarcinoma¹². Six other cell lines (OV-MZ 1–6) have also been described and are recognized as serous cystadenocarcinomas, based on both the original pathology report and postxenograft analysis¹³. Molthoff et al. in 1991 thoroughly characterized the histology and antigen expression of 10 primary ovarian cancers and three established cell lines. Both A2780 and H134 cells have been reported to form undifferentiated tumors and OVCAR3 cells poorly differentiated serous adenocarcinomas when injected into nude mice¹⁴. A few additional groups have also studied the tumorigenic characteristics of primary ovarian tumors by inoculating either ascites fluid or solid tumor samples directly into nude mice; however, as with the majority of cell line reports, the histology of the resultant tumors was discussed only to the extent that the characteristics of the original tumor were maintained^15,16. Attempts to personalize treatment regimens with relevant targeted therapeutics must be preceded by preclinical drug evaluations in cancer cell lines for which characteristics such as histology and gene expression profiles are well understood.

The difficulty in monitoring intraperitoneal (ip) disease formation and progression in vivo is one major limitation of xenograft models. Visual assessment of tumors is possible with magnetic resonance imaging¹⁷ or with cells that stably express green fluorescent protein¹⁸; however, the cost associated with the equipment necessary for the visualization of tumors through these methods can often be prohibitive. The location of the equipment must also be considered, as the need to transport immunodeficient animals repeatedly from their protective environments to the equipment for regular monitoring of tumor growth can be problematic. Some studies have successfully detected exogenous serum markers such as secreted alkaline phosphatase¹⁹, but the identification of a widely expressed endogenous marker would be of great value in xenograft models. CA125, for example, is a useful, blood-borne antigen easily monitored by relatively noninvasive methods clinically, yet has been used only minimally as a tool in OVCAR3, IGROV3, and SHIN3 xenografts^20,21. Circulating nucleic acids, thought to be shed from all tumor types, are optimistically being considered as a new potential diagnostic tool for cancer^22,23; however, there are only very few examples of their use in the diagnosis and monitoring of xenograft models of cancer^24,25.

An additional limitation of many xenograft models is the relevance of the location of tumor formation (e.g., subcutaneous (sc) vs. ip vs. orthotopic) and their semblance to the human disease. The majority of ovarian cancer xenografts described involve inoculation of the tumor cells ip or sc, and only a few reports describe orthotopic placement of intact primary tumors^26,27,28 or cell lines (A2780²⁹). Rodents have a unique bursal membrane that surrounds the ovary and is continuous with the oviduct, which facilitates orthotopic injection of cancer cells. The ability of the ovarian microenvironment to influence ovarian cancer cell behavior and the accuracy of these models in mimicking the various stages of ovarian cancer warrant further attention. Ovarian cancer metastases frequently appear disseminated throughout the abdomen and this is an important clinical problem as many major organs are vulnerable to adhesion and invasion. As well, the ovary is known to be a site for metastases of many nongenital primary tumors, including cancers of the colon, appendix, breast, small bowel, stomach, pancreas, gall bladder, and urinary bladder³⁰. Unfortunately, there are very few models available to study either ovarian-specific metastases of cancers or the metastatic behavior of ovarian cancer cells. Some clones of common cancer cell lines that are capable of metastasizing to the ovary have been established, such as a B16-F1 line, which preferentially colonizes in the ovary³¹, and MDCC-RPI lymphoma lines that home to either the liver or the ovary³². More models for the investigation of cancer metastases may improve our understanding of this complex process and provide insight into the relevance of targeted therapies based on disease stage.

Herein we report the histological characterization of eight xenograft models of early and advanced stage ovarian cancers, as well as a model of ovarian-specific metastasis demonstrating the ability of ovarian-specific factors to influence cancer cell behavior.

Results

Characterization of intraperitoneal xenograft mouse models of ovarian cancer

Eight of 11 cell lines injected ip into CD-1 nude mice resulted in tumor growth (Table 1). Specifically, ES-2-injected animals formed primarily ascites that was highly cellular and milky in appearance. The cells did not aggregate but existed more as a single-cell suspension or very small cell clumps. There was some ES-2 cell invasion of the body wall and diaphragm and accumulation of the tumor cells on the omentum, which were classified as solid tumors. We observed a similar disease presentation in the OCC1-injected animals, with tumor cells covering and invading both the diaphragm and the body wall, and some small (<0.5 cm³) tumor nodules aggregating in the mesentery and omentum. The health of the OCC1 models was further compromised by the apparently leaky vasculature associated with the OCC1 tumor growth, as the animals presented with severely distended abdomens full of bloody ascites, accompanied by cold, pale body extremities. The phenotype for the A2780 cell lines consisted mainly of large solid tumors. As observed with the OCC1-injected mice, the ascites was almost entirely blood and the dispersed disease within the ascitic fluid formed very distinct aggregates. Injection of HEY cells formed dense solid tumors, and half of the animals presented with relatively acellular ascitic fluid accumulation. Mice inoculated with OVCA 429, OV2008, or SKOV3 shared very similar phenotypes of large (>1.5 cm³) solid tumors adhered loosely to fat in the pelvic region, intestine, and/or omentum. Many smaller tumors (<0.25 cm³) were noted loose in the peritoneum, with a tendency to accumulate in the same three areas, and the ascitic fluid was bloody. C13^*, OVCA 433, and OVCAR3 cells failed to form ip tumors within 3 months; however, one of five C13^*-injected mice developed a small subcutaneous tumor at the site of injection that was histologically analyzed.

Table 1 - Survival time, disease presentation, and immunophenotype of intraperitoneal xenograft models of ovarian cancer cells.

Full table

The longevity of the mice postxenograft served as an indicator of the tumorigenicity of the cells. Median survival time for each group of mice is recorded in Table 1, and a Kaplan–Meier plot (Fig. 1) illustrates the results. Each arm represents the percentage of animals alive at the indicated time point following injection. The four most aggressive cell lines, ES-2, OCC1, A2780-cp, and HEY, resulted in a median survival time of less than 30 days, whereas A2780-s, OVCA 429, OV2008, and SKOV3 cells were less aggressive and formed tumors more slowly, in 2 to 3 months.

Figure 1.

Survival curves for intraperitoneal xenograft mouse models of ovarian cancer. For each of the 11 cell lines 1 10⁷ cells were resuspended in 500 l of PBS and injected intraperitoneally (day 0) into CD-1 nude mice. The survival curve represents the percentage of animals alive at the indicated time point after injection. The number of mice in each arm is 3, except for ES-2 (n = 5), OV2008 (n = 5), and SKOV3 (n = 7). , ES-2; ●, OCC1; , A2780-cp; , HEY; , A2780-s; , OVCA 429; , OV2008; , SKOV3. Animals injected with C13^*, OVCA 433, and OVCAR3 did not form tumors and are omitted.

Full figure and legend (36K)

To analyze the histopathology of the ip xenograft tumors, we performed immunohistochemical staining on all tumor types to determine expression of vimentin (VIM), high-molecular-weight keratin (HMWK), low-molecular-weight keratin (LMWK), epithelial membrane antigen (EMA), and leukocyte common antigen (LCA), five antigens widely used for histological classification of epithelial tumors. We also analyzed some tumor sections for cytoplasmic glycogen content, which is characteristic of clear cell adenocarcinomas, with periodic acid Schiff (PAS). Immunoreactivity for all five antigens is summarized in Table 1. Fig. 2 shows representative fields of view for ES-2, OCC1, OVCA 429, SKOV3, and OV2008, including hematoxylin and eosin (H&E) sections and relevant immunohistochemistry results used for the determination of the tumor types. Specifically, Figs. 2A–2D illustrate the undifferentiated ES-2 tumors, with irregular nuclei, abundant cytoplasm, and virtually no architectural features (2A). Characteristic of an undifferentiated tumor, there was low expression of the epithelial markers EMA (2B) and LMWK (2C), but extensive expression of vimentin, a mesenchymal cytoskeletal protein (2D). OCC1 tumors shown in Figs. 2E and 2F were also undifferentiated, with no distinct structural features or cell characteristics. Additionally, ES-2 and OCC1 tumors were both negative for cytoplasmic glycogen content by PAS and PAS-D (data not shown), a signature characteristic of clear cell adenocarcinomas. OVCA 429 cells did form clear cell adenocarcinomas, with extensive papillae and gland formation (2G), and were highly positive for EMA (2H), LMWK (2I), and cytoplasmic glycogen by PAS (2J). Similar clear cell characteristics in the SKOV3 tumors are illustrated in Fig. 2K. High EMA (2L), LMWK (2M), and PAS positivity (2N) confirmed their identity. The OV2008 tumors exhibited nests of HMWK-positive, eosinophilic squamous cells (2P) scattered throughout an endometrioid tumor (2O).

Figure 2.

Pathological analysis of intraperitoneal xenograft tumors. All tumors were formalin-fixed, paraffin-embedded, sectioned, and stained for analysis and photography. Representative fields of view are shown for each cell type, illustrating the characteristic features used for diagnosis. ES-2 tumors stained for H&E, EMA, LMWK, and VIM are shown in A–D, respectively. H&E-stained sections of an OCC1 tumor are shown in E and F. OVCA 429 tumors stained for H&E, EMA, LMWK, and PAS (diastase-digested negative control in inset) are shown in G–J. The characteristics of the SKOV3 tumors are illustrated by (K) H&E, (L) EMA, (M) LMWK, and (N) PAS (diastase-digested negative control in inset). (O) H&E- and (P) HMWK-stained sections of OV2008 tumors are shown. Original magnifications: C and H, 100; A, B, D, E, G, I, and J, 200; F, K, L, M, N, O, and P, 400.

Full figure and legend (220K)

The diagnoses of the pathology for each of the tumor types are A2780-s, A2780-cp, ES-2, HEY, and OCC1, undifferentiated carcinoma; OVCA 429 and SKOV3, clear cell adenocarcinoma; and OV2008 and C13^*, endometrioid adenocarcinoma with foci of squamous differentiation (Table 1).

CA125 expression and circulating human DNA investigated as potential diagnostic markers

ELISA determined the expression of CA125 by the cell lines under investigation. Analysis of the supernatant of cells cultured for 24 h showed detectable levels only in SKOV3 and OVCAR3 cells (data not shown). The failure of these two cell lines to develop into tumors consistently in nude mice prevented the evaluation of CA125 as a tumor marker for these models.

To investigate DNA shedding and the presence of circulating human DNA in the plasma and serum of mice postxenograft, we collected plasma from nude mice immediately prior to cancer cell injection and 26 days post-ip injection with A2780-cp cells for PCR amplification of Alu repeats. Before injection of the cancer cells, there were not any DNA sequences containing Alu repeats in the mouse plasma; however, 26 days postinjection there were detectable quantities (Fig. 3).

Figure 3.

Circulating human tumor DNA as a diagnostic tool of xenograft models. Plasma was collected from mice prior to A2780-cp ip xenograft and 26 days postxenograft and 1 l was analyzed by PCR for Alu repeats. The + control lane is a simultaneously performed Alu PCR on genomic DNA extracted from A2780-cp cells in culture, and the - control represents a PCR on genomic DNA from a control mouse ear-punch.

Full figure and legend (22K)

A2780-cp ovarian cancer cells metastasize to and proliferate preferentially within the ovary

In follow-up studies with the A2780-cp ip xenograft models, 11 of 88 (12.5%) animals injected presented with severe unilateral or bilateral ovarian disease. Fig. 4 shows both macro- and microscopic pathology of the ovarian tumors. In most cases, the bursal membrane appeared structurally intact and the intrabursal space entirely occupied by tumor. The smoothness of the tumor boundary varied from well circumscribed (4C) to jagged with migrating cells. A representative tumor in Figs. 4E and 4F illustrates the migratory and/or invasive characteristics of the A2780-cp cells, as they disseminate to/from a tumor (4E) and invade the bursal membrane (4F). In this example the ovary remained independent of the tumor within the bursal space; the bursal membrane appeared thickened by malignant human cells, and we identified potentially migrating cells. In most cases, we found normal ovarian structures and follicles within the tumors (4G, 4H). PCR amplification of DNA extracted from these tumors confirmed that the tumors arose from the injected A2780-cp cells and not spontaneously from the mouse ovary (data not shown).

Figure 4.

Mice injected ip with A2780-cp present with ovary-specific metastases. 1 10⁷ A2780-cp cells were injected ip into a total of 88 CD-1 nude mice and animals were monitored until a predetermined clinical endpoint was reached. Upon necropsy, 12.5% (11 of 88) were found to have severe unilateral or bilateral ovarian disease. The reproductive tracts were removed and photographed (representative photos shown in A and B). Subsequently, all tumor samples were formalin-fixed, paraffin-embedded, sectioned, and H&E stained for histological analysis. (C and D) The bursal membrane appeared structurally intact and the intrabursal space appeared entirely occupied by tumor. The smoothness of the tumor boundary varied from well circumscribed (C) to jagged with migrating cells (D). (E and F) In one case the ovary remained independent of the tumor within the bursal space; the bursal membrane appeared thickened by cancer cells and potentially migrating cells were identified. (G and H) In most cases, normal ovarian structures and follicles were found within the tumors. Original magnifications: C and F, 25; D, 400; E, G, and H, 200.

Full figure and legend (82K)

Comparison of the Kaplan–Meier survival curves (Fig. 5) of animals with disseminated peritoneal disease and ovarian tumors by a log-rank test indicated that the ip disease developed significantly more slowly than the ovarian tumors (median survival 37 vs. 52 days; P < 0.05). We compared the mean numbers of Ki67-positive cells from 10 fields of view (1000 magnification) from three tumors of the two anatomic sites [peritoneal (Figs. 6A) and ovarian (Fig. 6B)] by Student's t test. We observed significantly more proliferative cells in the ovarian tumors than in the ip counterparts (Fig. 6C, 59.23 1.95 vs. 42.37 1.73; P < 0.0001). Cancer cell metastasis to the ovary was unique to the A2780-cp cells. Extensive follow-up studies were also conducted with ES-2 (n = 35), HEY (n = 59), and OCC1 (n = 46) and none of these models exhibited an ovarian phenotype.

Figure 5.

Mice with A2780-cp xenografts with spontaneous ovarian metastases have a significantly shorter survival time compared to mice with dispersed ip disease. 1 10⁷ A2780-cp cells were injected ip into a total of 88 CD-1 nude mice (day 0). The survival curves (dashed line, mice with dispersed ip disease; solid line, mice with ovarian metastases) represent the percentage of animals alive at the indicated time point after injection. The group designated as having ovarian metastases had severe ovarian disease upon necropsy and had a significantly shorter survival time (P < 0.05). The number of mice in each arm is peritoneal, n = 77; ovarian, n = 11.

Full figure and legend (30K)

Figure 6.

A2780-cp cell proliferation (Ki67) is higher in ovarian vs peritoneal tumors. All tumors were formalin-fixed, paraffin-embedded, sectioned, and stained for Ki67. Representative fields of view from (A) peritoneal and (B) ovarian A2780-cp tumors are shown (original magnification 1000). Ki67-positive cells were counted in 10 randomly chosen fields of view of three tumors from different animals for each anatomic location. (C) A significant increase in the number of Ki67-positive cells in the ovarian tumors was observed (P < 0.0001).

Full figure and legend (47K)

Characterization of intrabursal xenograft mouse models of ovarian cancer

We injected two cell lines (ES-2 and OVCA 429) under the bursal membrane of CD-1 nude mice and both cell types resulted in tumor growth. The take rates for ES-2 and OVCA 429 cells injected intrabursally (ib) were 86 (n = 7) and 40% (n = 5), respectively. Fig. 7 shows both macro- and microscopic pathology of resultant tumors (ES-2, 7A–7C; OVCA 429, 7D–7F). Figs. 7A and 7D illustrate the severe ovarian disease of the mice following ib injection of the cancer cells. Ovaries were enlarged to greater than 20 the normal volume (normal reproductive tract shown in inset of Fig. 7A). The effects on the normal ovarian tissue ranged from crowding and displacement (7E, OVCA 429) to complete invasion (7B, ES-2). There was no detectable peritoneal disease in either of the OVCA 429 animals that succumbed to ovarian disease or in 3 of 6 (50%) of the ES-2 animals. The other three ES-2-injected mice showed extensive peritoneal dissemination comparable to the phenotype of the ip-injected counterparts.

Figure 7.

Gross and microscopic pathology of intrabursal xenograft tumors. At necropsy, the reproductive tract was removed and photographed (A, ES-2; D, OVCA 429; inset of A shows age-matched, uninjected control). Subsequently, all tumor samples were formalin-fixed, paraffin-embedded, sectioned, and H&E stained for histological analysis. The arrows in (B) (ES-2) and (E) (OVCA 429) show normal ovarian follicles identifiable within the sample. One representative field of view of tumor tissue that illustrates the characteristic features used for diagnosis is shown for each cell type (C, ES-2; F, OVCA 429). Scale bar (A and D), 5 mm. Original magnifications: B, C, and E, 200; F, 400.

Full figure and legend (97K)

Representative H&E sections illustrate the lack of distinct architecture of the undifferentiated ES-2 tumors (Fig. 7C) and the gland formation and areas of clear cell differentiation in the OVCA 429 serous adenocarcinomas (7F). We compared the histologic appearances of ip and ib tumors to determine whether the site of injection affected cell behavior (Fig. 2 vs. Fig. 7). Tumors developing from ip-injected cells appear histologically identical to the tumors resulting from the same cell type injected ib. For example, in both locations OVCA 429 tumors exhibited extensive gland formation and papillae and many cells with clear, glycogen-rich cytoplasm, characteristic of clear cell adenocarcinomas. Similarly, ES-2 tumors, regardless of injection site, were undifferentiated, composed of sheets of cells with irregularly shaped nuclei, prominent nucleoli, and abundant cytoplasm. Furthermore, Ki67 staining was identical in ES-2 tumors of both anatomic locations (ib and ip, data not shown). As well, survival curve comparison of the ip and ib models, unlike in the metastatic A2780-cp models, did not show accelerated disease progression in ib-injected animals for either the ES-2 or the OVCA 429 cells; however, this information is interpreted cautiously due to fewer cells being injected ib. We observed notably less peritoneal disease in the ib- vs. ip-injected mice, suggesting that the malignant cells were at least partially confined to the ovary by the bursal membrane. A potentially important difference noted between the ip and the ib models was a low take rate (40%, n = 5) in the ib-injected OVCA 429 animals, compared to ip (100%, n = 3).

Discussion

This study histologically characterized ip, orthotopic, and metastatic xenograft models of ovarian cancer representative of a cross section of histological subtypes useful in the evaluation of novel therapeutics. Eleven human ovarian cancer cell lines were injected ip into CD-1 nude mice and 8 resulted in tumor growth. The longevity of the mice postxenograft served as an indicator of the tumorigenicity of the cells. The median survival times for the mice suggest that ES-2, OCC1, A2780-cp, and HEY cells are the most tumorigenic of the collection, forming tumors quite rapidly (less than 30 days), whereas A2780-s, OVCA 429, OV2008, and SKOV3 are less aggressive and form tumors more slowly (2 to 3 months). The number of days for the animals to reach a humane endpoint for ES-2 cells was consistent with previous reports and animals succumbed to disease in less than 20 days¹¹. HEY³³ and A2780²¹ models were also consistent in terms of aggressiveness, with both sets of animals reaching endpoint approximately 30 to 40 days postxenograft. This is the first report of survival times of mice inoculated with OCC1, OV2008, or OVCA 429 cells ip. In advanced stage ovarian cancer, the cells tend to accumulate within ascitic fluid in the peritoneum² and all tumorigenic cells in this study of ip xenografts recapitulated this condition very well. The dissemination of the cells throughout the abdomen was similar to human disease, with cell accumulation commonly occurring near the liver and spleen and adherence to the walls of the peritoneal cavity.

Animals injected with C13^*, OVCA 433, and OVCAR3 did not have any visible ip disease after at least 90 days. There is no evidence in the literature demonstrating the tumor-forming capabilities of C13^* cells in xenograft models, only one report of OVCA 433 xenografts³⁴, and multiple accounts of OVCAR3 xenografts including both ip^35,36 and sc models^37,38. However, we are not the first to observe a low take rate for the OVCAR3 cell line³⁹. A second round of xenograft experiments with a new sample of OVCAR3 cells from the American Type Culture Collection ensured that our results were replicable, as the OVCAR3 line still did not establish tumors under the conditions described. The lack of establishment that we have observed for these cells may be attributable to variable immunocompetence of the recipient mice (e.g., a different strain of nude mice, or SCID vs. nude mice) as previously suggested¹⁵. Alternatively, the lack of tumor establishment in these models could be due to the absence of structural support and/or interaction with extracellular matrices for the cells following ip injection. Specifically, the rate of establishment for breast and ovarian cancer xenografts has been shown to be significantly improved by co-injection with Matrigel³⁹. The low take rate observed with SKOV3 cells and the relatively long survival time of the SKOV3-injected mice that did succumb to disease (108 vs.14 to 45 days^40,41) may similarly be explained.

The inability of the C13^* cell line to form tumors in nude mice is curious as it is a chemoresistant line derived by continuous cisplatin exposure of OV2008 cells, which did successfully form tumors in four of the five animals injected. One animal injected ip with C13^* cells developed a small subcutaneous tumor at the site of injection and analysis revealed endometrioid histology similar to that of the OV2008 tumors. This indicates that, although some changes may have occurred during prolonged culture or cisplatin exposure that decreased their tumorigenicity, the histology of the C13^* cells has not deviated from the parental line.

Assessment of the histology of the xenograft tumors revealed that five of the eight models yielded undifferentiated adenocarcinomas (A2780-s, A2780-cp, ES-2, HEY, and OCC1), which are most common clinically¹. OVCA 429, SKOV3, and OV2008-injected mice produced histologically distinct tumors representing the different subsets of ovarian epithelial carcinomas. Specifically, OVCA 429 and SKOV3 formed clear cell adenocarcinomas, and OV2008 formed endometrioid tumors with foci of squamous differentiation. This is the first description of the distinct histology of these three tumor types; previous reports are limited to identification of the sources as simply adenocarcinomas, with mention that xenografts resemble the tumor of origin⁴². ES-2 and OCC1 cells were originally reported to be derived from clear cell carcinomas^11,6; however, in our study both cell types developed into undifferentiated carcinomas and appeared to have lost their clear cell characteristics. A possible explanation for this difference is that although the majority of tumor cells initially may have shown clear cell characteristics, the tumor was likely heterogeneous, and perhaps one cell type (i.e., undifferentiated) was favored by the conditions used to establish the cell line. Similarly, HEY cells were originally derived from a moderately differentiated papillary cystadenocarcinoma¹²; however, in this study inoculation of these cells resulted in tumors lacking any distinctive architectural features and were designated as undifferentiated carcinomas.

An ongoing limitation of xenografts is the ability to monitor disease formation and progression in vivo. Some studies have successfully used cells that stably express green fluorescent protein¹⁸ or exogenous secreted markers such as secreted alkaline phosphatase¹⁹, but the establishment, selection, and characterization of stably expressing clones is a deterrent. The identification of a widely expressed endogenous marker would be of great use in xenograft models. CA125, for example, is extremely valuable clinically, yet has been employed and proven useful in only a few preclinical studies evaluating the effectiveness of therapeutics in OVCAR3, IGROV3, SHIN3, and UCI 101 xenograft models^21,20. Its potential is limited by the fact that very few cell lines express the antigen. We detected CA125 in the supernatant of only 2 of 11 ovarian cancer cell lines tested: SKOV3 and OVCAR3.

Circulating nucleic acids in plasma and serum (CNAPS) of cancer patients have recently been described as a "diagnostic gold mine"⁴³, yet they have been used only minimally in animal models^24,25, and the potential value of CNAPS for monitoring the tumor burden of xenograft cancer models has not been fully explored. This study confirmed the presence of tumor DNA in the mouse circulation postxenograft, which represents a relatively noninvasive method of tumor detection. To distinguish between DNA of mouse and tumor origin, we chose to amplify specifically the Alu repeats shed by the human tumors, but which would be absent from the mouse genome. Reports that CNAPS in patients frequently have mutated oncogenes and/or tumor suppressors identical to those of the primary tumors demonstrate that at least a portion of plasma DNA in patients is tumor-derived^44,45; yet questions remain as to how circulating levels correlate with prognosis and treatment response. This provides an opportunity for future studies to improve our understanding of the effects of successful chemotherapy and cell death on the short- and long-term quantities of circulating tumor DNA in patients, as well as the contribution of host inflammatory cells and blood clotting to DNA concentrations.

A frequent criticism of many xenograft models is the relevance of the location of tumor formation and their ability to mimic the human disease. Recent microarray studies have convincingly demonstrated that cancer cell gene expression profiles vary significantly, depending on the tumor location^46,47,48. It follows that the tumor microenvironment (e.g., the ovary) has the ability to influence cell behavior, including metastatic potential²⁷. Two cell lines (ES-2 and OVCA 429) were injected intrabursally and both cell types resulted in tumor growth. The effects on the normal ovarian tissue ranged from crowding and displacement (OVCA 429) to complete invasion (ES-2), which likely reflects the invasive characteristics of the cell lines. Incidence of peritoneal disease in the ib-injected animals was dramatically reduced, suggesting that cells can be successfully confined to the mouse ovary by the bursal membrane; however, invasion through the membrane can occur. There is a possibility that during the surgical procedure some cells may have become loose within the peritoneum via the injection wound, but based on the lack of abdominal disease in the majority of the ib-injected animals, this seems unlikely. The variable rate of ascites formation may be a result of some of the animals reaching an endpoint (e.g., >5 g weight gain) due to the severe ovarian phenotype prior to the cells escaping via invasion through the bursal membrane. Such orthotopic models of ovarian cancer may be useful in the evaluation of invasive characteristics of cells through examination of the degree of invasion of the ovary and the ability to cross the bursa into the peritoneum. Although this model of early stage disease still does not provide information on disease initiation, it has the potential to provide insight into tumor progression as well as the ovarian signals that are conducive to cancer cell proliferation, migration, and survival.

Our report of the unique ovarian-specific metastasis of A2780-cp cancer cells provides evidence for an ovarian-specific signal that chemoattracts cancer cells and perhaps also modulates other cell behaviors, such as proliferation, migration, and survival. This model will allow for future studies on the effects of the tumor microenvironment on cancer cells and may also provide information on the signals contributing to all forms of metastases, in particular ovarian metastases of nongenital primary tumors (e.g., peritoneal, colon) and the formation of bilateral ovarian tumors. We hypothesize that the cells migrate in a specific manner due to the secretion of chemoattractants by the destination organ, which the literature suggests may include TGF- family members⁴⁹, chemokines (e.g., CXCs)^50,51, and/or neurotransmitters⁵². Although the reason for the variable incidence (12.5%) of this phenotype is unknown, it was recently reported that the differential hormonal milieu throughout the estrous cycle can alter organ-specific metastasis of cancer cells⁵³. Thus, expression of chemoattractants may vary with the phase of estrous, and/or the ovarian microenvironment may change with the estrous cycle to affect cell establishment and survival. For example, the fluctuations in fluid accumulation in the bursal space during the estrous cycle⁵⁴ could alter the ability of xenografted cells to remain in place during drainage. As well, the protein content of the bursal fluid changes throughout the reproductive cycle⁵⁵, and it is possible that the environment was harsher at the time of some of the injections. A better understanding of the effects of the reproductive environment on cancer cells may indicate that the use of synchronized, rather than randomly cycling, female mice would improve consistency with the models. Large-scale follow-up studies in our lab using ES-2, HEY, and OCC1 xenografts (n > 35 for each cell line) have never reproduced an animal with ovarian-specific metastases as seen in the A2780-cp model, suggesting that the response is cell-specific. In this case it seems A2780-cp cells express the protein(s) required to respond to the ovarian signal, but the ES-2, HEY, and OCC1 cells do not.

It was hypothesized that the ovarian microenvironment may modulate other cancer cell behaviors such as histology, proliferation, and/or survival, and this was assessed in both the ovarian metastatic A2780-cp tumors and the ib-injected models. No histological differences were observed between the ib and the ip tumors of ES-2 or OVCA 429 origin, or in the A2780-cp tumors of the ovary, compared with those from other peritoneal sites. However, a low take-rate (40%) was observed in the ib-injected OVCA 429 models compared to the ip models (100%) and perhaps this difference can be similarly explained as the variable incidence of ovarian-specific metastases in the A2780-cp-injected mice. For example, changes in volume of fluid in the bursal space and/or the protein contents of the fluid at the time of injection may have been less amenable to cell survival at the time of some of the OVCA 429 injections. Significantly more Ki67-positive, proliferating cells were observed in the A2780-cp ovarian tumors compared to peritoneal controls, and it follows that the animals with ovarian disease succumbed to disease more quickly. As with the organ-specific metastases, this phenomenon was cell-type dependent and the ovarian microenvironment did not similarly increase the proliferation rate of ES-2 tumors. Due to the difference in the number of cells injected ip and ib, we were unable to assess any divergence in survival times between these models, but the Ki67 results would suggest that the ES-2 cells within the ovary would not have the same negative effect on survival as observed with the metastatic A2780-cp cells.

This study has described the histopathologic details of eight animal models of epithelial ovarian cancer that represent four histologically distinct models of early and late stage disease. These xenografts provide an important cross section of ovarian cancer phenotypes for both in vitro and in vivo assessment of the efficacy of ovarian cancer therapeutics. Evidence is accumulating that the subtypes of epithelial ovarian cancer are distinct in their developmental pathways, their molecular characteristics, and their responsiveness to therapeutic strategies^3,4. Therefore, it is critical that evaluations of new drugs, such as targeted therapies, are conducted on an array of histological tumor models as their expression and activity of molecular targets may vary.

Materials and methods

Cell culture

The following cell lines were used in this study: HEY, OVCA 429, OVCA 433, OCC1, OVCAR3, and SKOV3 human ovarian carcinoma cells from Dr. G. Mills (Houston, TX, USA); A2780-s, A2780-cp, OV2008, and C13^* human ovarian carcinoma cells from Dr. M. Molepo (Ottawa, ON, Canada); and ES-2 from Dr. J. Bell (Ottawa, ON, Canada).

For maintenance, HEY, OVCAR3, and OCC1 cells were cultured in -minimal essential medium (-MEM; Gibco BRL, Burlington, ON, Canada) supplemented with 10% heat-inactivated fetal calf serum (FCS; CanSera, Rexdale, ON, Canada). OVCA 429 and OVCA 433 cells were cultured in -MEM + 10% FCS supplemented with 1% nonessential amino acids solution (NEAA; Gibco BRL). SKOV3 and ES-2 cells were cultured in McCoy's 5A medium plus 2.2 g/L NaHCO₃ and 15% FCS. A2780-s and -cp cells were cultured in DMEM/F12 medium (Gibco BRL) supplemented with 0.84 g/L NaHCO₃, 4.8 g/L Hepes, 1% NEAA, 7.5% FCS, and 7.5% newborn calf serum (Gibco BRL). OV2008 and C13^* cells were cultured in RPMI 1640 medium (Gibco BRL) with 10% FCS.

Animal studies

All animal experiments described in this study were performed according to the Guidelines for the Care and Use of Animals established by the Canadian Council on Animal Care. Female CD-1 nu/nu mice (Charles River Laboratories, Wilmington, MA, USA) ages 6–8 weeks were housed with free access to food and water.

For characterization of ip disease formed by each cell line (HEY, OVCA 429, OVCA 433, OCC1, OVCAR3, SKOV3, A2780-s, A2780-cp, OV2008, C13^*, ES-2), 1 10⁷ cells, resuspended in 500 l of phosphate-buffered saline, were injected with a 25-gauge needle ip into at least three mice and disease progression was monitored based on overall health until a predetermined endpoint was reached. Survival time reflects the time required for the animals to reach any endpoints, including tumor ulceration, weight loss exceeding 15%, weight gain exceeding 5 g, anorexia, and/or diarrhea. Necropsies were performed, and when present, tumor samples were fixed in 10% buffered formalin (VWR, Mississauga, ON, Canada) for 6–24 h, paraffin-embedded, and sectioned at 3 m for immunohistochemical analysis, 5 m for H&E staining, or 7 m for DNA extraction. From a subset of mice, plasma was collected prior to the cancer cell injection and once weekly throughout the experiment via the saphenous vein: blood drawn into heparin-coated capillary tubes was incubated on ice for approximately 30 min, centrifuged for 10 min, and the plasma was transferred to an Eppendorf tube and stored at -20°C.

Intrabursal (ib) disease was characterized for two cell lines (OVCA 429 and ES-2) in at least five mice each. Specifically, mice were anesthetized with avertin (275–350 mg/kg), a small dorsal incision (1–2 cm) was made, ovaries were externalized, and 1 10⁶ cells, resuspended in 10 l of PBS, were injected with a 30-gauge needle under the bursal membrane. Disease progression was monitored daily until a predetermined endpoint was reached. Necropsies were performed and tissues were collected as described above.

Immunohistochemistry

Immunohistochemical staining was performed on all tumor types to determine expression of the mesenchymal cytoskeletal protein VIM; three epithelial markers, HMWK, LMWK, and EMA; and LCA. Specifically, paraffin sections of 3 m were deparaffinized in xylene and rehydrated in decreasing concentrations of ethanol according to standard protocol. Proteolytic enzyme digestion (0.4% pepsin A in Tris-buffered saline (TBS), 15 min at 37°C followed by 5 min of washing in cold running water) was performed for EMA, HMWK, and LMWK only. Endogenous peroxidase activity was then blocked with 3% hydrogen peroxide in TBS for 10 min and samples were subsequently rinsed for 5 min in TBS. All tissue was blocked (4% BSA, 10% sucrose, 1% normal swine serum in TBS) for 20 min at room temperature (RT). Primary antibodies (DAKO, Cytomation, Carpinteria, CA, USA) were diluted in the blocking solution to the following concentrations: vimentin, 1:100; LMWK, 1:10; HMWK, 1:1000; EMA, 1:1000; LCA, 1:500; and Ki67, 1:50. Sections were incubated with the antibodies overnight at 4°C. Following two 5-min washes in TBS, sections were incubated with an anti-mouse/anti-rabbit horseradish peroxidase-labeled polymer (DAKO Cytomation) for 30 min at RT and developed with diaminobenzidine (DAB) as a substrate (0.2% DAB, 0.001% H₂O₂). Slides were counterstained with hematoxylin, dehydrated, and mounted for analysis and photography.

PAS staining for cytoplasmic glycogen content was performed according to standard protocol on OVCA 429 and SKOV3 cells to confirm the diagnosis of clear cell adenocarcinoma. PAS staining was also performed on ES-2 and OCC1 cell tumors to demonstrate the absence of glycogen-containing clear cells. Tissue sections pretreated with 1% diastase for 1 h served as a negative control (PAS-D).

Analysis of CA125 expression by ELISA

Cells, cultured to approximately 70% confluence under standard conditions in 35-mm dishes, were incubated in 1 ml of serum-free medium for 24 h. Medium was collected, any cells were removed by centrifugation, and samples were frozen at -20°C until analysis. CA125 ELISA was performed according to the manufacturer's instructions (Panomics; Redwood City, CA, USA).

DNA extraction and PCR amplification of Alu repeats

Alu repeats were amplified both from 1 l of circulating mouse plasma collected in heparin-coated capillary tubes by saphenous vein bleed and from genomic DNA extracted by the following method from formalin-fixed, paraffin-embedded tumor tissue. Seven-micrometer sections of tumor were put directly into 100 l of 0.05% Tween 20, boiled for 1 min to melt the paraffin, immediately centrifuged at >10,000g for 10 min, and briefly put on ice, and the wax disc on the surface of the sample was then removed with forceps. DNA extraction buffer (250 l; 10 mM Tris/0.1 M EDTA/0.5% SDS/20 g/ml proteinase K) was added to the tissue sample and incubated at 58°C overnight. DNA was precipitated by a standard protocol using saturated NaCl and isopropanol. The DNA pellet was washed once with 75% ethanol and resuspended in 100 l Tris–EDTA (50 mM, pH 6.8).

PCR for Alu sequences was performed as previously described^56,57. Briefly, the 50-l polymerase chain reaction contained 20 pmol of each primer (Alu-sense, 5'-ACGCCTGTAATCCCAGCACTT-3'; Alu-antisense, 5'-TCGCCCAGGCTGGAGTGCA-3'), 100 M dNTPs, 1 mM MgCl₂, 1 PCR Buffer (Invitrogen, Burlington, Canada), and 1 U Taq DNA polymerase (Invitrogen). The protocol had an initial 10-min denaturation at 94°C, followed by 25 cycles of 94°C for 30 s, 58°C for 45 s, and 72°C for 45 s, and a final elongation at 72°C for 10 min. PCR products were separated by electrophoresis on a 1.0% agarose gel and stained with ethidium bromide for visualization.

Survival curves and statistical analysis

Kaplan–Meier survival curves were plotted using GraphPad Prism software (version 4.0; GraphPad Software, San Diego, CA, USA, www.graphpad.com) and were compared using a log-rank test. Significance was inferred at P < 0.05.

References

1. National Cancer Institute, (2003). SEER Cancer Statistics Review 1973–1999, Table I-15. Natl. Cancer Inst: Bethesda, MD.
2. Scully, R. E., Young, R. H. and Clement, P. B. (1998). Atlas of Tumor Pathology.
3. Chatzistamou, I., Schally, A. V., Pafiti, A., Kiaris, H. and Koutselini, H. (2002). Expression of growth hormone-releasing hormone in human primary endometrial carcinomas. Eur. J. Endocrinol. 147: 381–386. | Article | PubMed | ChemPort |
4. Tsukagoshi, S., et al. (2003). Thymidine phosphorylase-mediated angiogenesis regulated by thymidine phosphorylase inhibitor in human ovarian cancer cells in vivo. Int. J. Oncol. 22: 961–967. | PubMed | ChemPort |
5. Rygaard, J. and Povlsen, C. O. (1969). Heterotransplantation of a human malignant tumour to "Nude" mice. Acta Pathol. Microbiol. Scand. 77: 758–760. | PubMed | ChemPort |
6. Wong, W. S., et al. (1990). Establishment and characterization of a new human cell line derived from ovarian clear cell carcinoma. Gynecol. Oncol. 38: 37–45. | Article | PubMed | ChemPort |
7. DiSaia, P. J., Sinkovics, J. G., Rutledge, F. N. and Smith, J. P. (1972). Cell-mediated immunity to human malignant cells: a brief review and further studies with two gynecologic tumors. Am. J. Obstet. Gynecol. 114: 979–989. | PubMed | ChemPort |
8. Andrews, P. A. and Albright, K. D. (1992). Mitochondrial defects in cis-diamminedichloroplatinum(II)-resistant human ovarian carcinoma cells. Cancer Res. 52: 1895–1901. | PubMed | ISI | ChemPort |
9. Berchuck, A., et al. (1992). Regulation of growth of normal ovarian epithelial cells and ovarian cancer cell lines by transforming growth factor-. Am. J. Obstet. Gynecol. 166: 676–684. | PubMed | ISI | ChemPort |
10. Fogh, J. and Trempe, G. (1975). New Human Tumor Cell Lines. 115–159.
11. Lau, D. H., Lewis, A. D., Ehsan, M. N. and Sikic, B. I. (1991). Multifactorial mechanisms associated with broad cross-resistance of ovarian carcinoma cells selected by cyanomorpholino doxorubicin. Cancer Res. 51: 5181–5187. | PubMed | ChemPort |
12. Buick, R. N., Pullano, R. and Trent, J. M. (1985). Comparative properties of five human ovarian adenocarcinoma cell lines. Cancer Res. 45: 3668–3676. | PubMed | ChemPort |
13. Mobus, V., et al. (1992). Morphological, immunohistochemical and biochemical characterization of 6 newly established human ovarian carcinoma cell lines. Int. J. Cancer. 52: 76–84. | PubMed | ChemPort |
14. Molthoff, C. F., Calame, J. J., Pinedo, H. M. and Boven, E. (1991). Human ovarian cancer xenografts in nude mice: characterization and analysis of antigen expression. Int. J. Cancer. 47: 72–79. | PubMed | ChemPort |
15. Elkas, J. C., Baldwin, R. L., Pegram, M., Tseng, Y., Slamon, D. and Karlan, B. Y. (2002). A human ovarian carcinoma murine xenograft model useful for preclinical trials. Gynecol. Oncol. 87: 200–206. | Article | PubMed |
16. Verschraegen, C. F., et al. (2003). Establishment and characterization of cancer cell cultures and xenografts derived from primary or metastatic Mullerian cancers. Clin. Cancer Res. 9: 845–852. | PubMed | ChemPort |
17. Gossmann, A., et al. (2000). Magnetic resonance imaging in an experimental model of human ovarian cancer demonstrating altered microvascular permeability after inhibition of vascular endothelial growth factor. Am. J. Obstet. Gynecol. 183: 956–963. | Article | PubMed | ISI | ChemPort |
18. Chaudhuri, T. R., Mountz, J. M., Rogers, B. E., Partridge, E. E. and Zinn, K. R. (2001). Light-based imaging of green fluorescent protein-positive ovarian cancer xenografts during therapy. Gynecol. Oncol. 82: 581–589. | Article | PubMed | ISI | ChemPort |
19. Nilsson, E. E., Westfall, S. D., McDonald, C., Lison, T., Sadler-Riggleman, I. and Skinner, M. K. (2002). An in vivo mouse reporter gene (human secreted alkaline phosphatase) model to monitor ovarian tumor growth and response to therapeutics. Cancer Chemother. Pharmacol. 49: 93–100. | Article | PubMed | ChemPort |
20. Kraeber-Bodere, F., et al. (1996). Pharmacokinetics and biodistribution of samarium-153-labelled OC125 antibody coupled to CITCDTPA in a xenograft model of ovarian cancer. Eur. J. Nucl. Med. 23: 560–567. | PubMed | ChemPort |
21. Burbridge, M. F., Kraus-Berthier, L., Naze, M., Pierre, A., Atassi, G. and Guilbaud, N. (1999). Biological and pharmacological characterisation of three models of human ovarian carcinoma established in nude mice: use of the CA125 tumour marker to predict antitumour activity. Int. J. Oncol. 15: 1155–1162. | PubMed | ChemPort |
22. Lui, Y. Y. and Dennis, Y. M. (2002). Circulating DNA in plasma and serum: biology, preanalytical issues and diagnostic applications. Clin. Chem. Lab. Med. 40: 962–968. | Article | PubMed | ChemPort |
23. Anker, P., Lyautey, J., Lederrey, C. and Stroun, M. (2001). Circulating nucleic acids in plasma or serum. Clin. Chim. Acta. 313: 143–146. | Article | PubMed | ISI | ChemPort |
24. Trejo-Becerril, C., et al. (2003). Circulating nucleosomes and response to chemotherapy: an in vitro, in vivo and clinical study on cervical cancer patients. Int. J. Cancer. 104: 663–668. | Article | PubMed | ChemPort |
25. Liu, Y., Wu, M. C., Qian, G. X. and Zhang, B. H. (2003). Nested RT-PCR in detection of blood AFPmRNA in animal model of human hepatocellular carcinoma. Hepatobiliary Pancreatic Dis. Int. 2: 414–417. | ChemPort |
26. Orsulic, S., Li, Y., Soslow, R. A., Vitale-Cross, L. A., Gutkind, J. S. and Varmus, H. E. (2002). Induction of ovarian cancer by defined multiple genetic changes in a mouse model system. Cancer Cell 1: 53–62. | Article | PubMed | ISI | ChemPort |
27. Yoshida, Y., Kamitani, N., Sasaki, H., Kusumi, K., Tominaga, T. and Kotsuji, F. (1998). Establishment of a liver metastatic model of human ovarian cancer. Anticancer Res. 18: 327–331. | PubMed | ChemPort |
28. Fu, X. and Hoffman, R. M. (1993). Human ovarian carcinoma metastatic models constructed in nude mice by orthotopic transplantation of histologically-intact patient specimens. Anticancer Res. 13: 283–286. | PubMed | ChemPort |
29. Li, J., et al. (2002). [Monitoring impact of measurement of human urinary -chorionic gonadotropin on orthotopic implantation model of human ovarian carcinoma in athymic nude mice]. Ai Zheng. 21: 949–953. | PubMed |
30. Moore, R. G., Chung, M., Granai, C. O., Gajewski, W. and Steinhoff, M. M. (2004). Incidence of metastasis to the ovaries from nongenital tract primary tumors. Gynecol. Oncol. 93: 87–91. | Article | PubMed |
31. Brunson, K. W. and Nicolson, G. L. (1979). Selection of malignant melanoma variant cell lines for ovary colonization. J. Supramol. Struct. 11: 517–528. | Article | PubMed | ChemPort |
32. Shearman, P. J. and Longenecker, B. M. (1980). Selection for virulence and organ-specific metastasis of herpesvirus-transformed lymphoma cells. Int. J. Cancer. 25: 363–369. | PubMed | ISI | ChemPort |
33. Auzenne, E., et al. (2002). Superior therapeutic profile of poly-L-glutamic acid– paclitaxel copolymer compared with taxol in xenogeneic compartmental models of human ovarian carcinoma. Clin. Cancer Res. 8: 573–581. | PubMed | ChemPort |
34. Spinella, F., Rosano, L., DiCastro, V., Nicotra, M. R., Natali, P. G. and Bagnato, A. (2003). Endothelin-1 decreases gap-junctional intercellular communication by inducing phosphorylation of connexin 43 in human ovarian carcinoma cells. J. Biol. Chem.
35. Manetta, A., Satyaswaroop, P. G., Hamilton, T., Ozols, R. and Mortel, R. (1989). Therapeutic effect of a radiolabeled monoclonal antibody on human ovarian cancer xenograft in nude mice. Gynecol. Oncol. 32: 368–370. | Article | PubMed | ChemPort |
36. Andersson, H., Lindegren, S., Back, T., Jacobsson, L., Leser, G. and Horvath, G. (2000). The curative and palliative potential of the monoclonal antibody MOv18 labelled with 211At in nude mice with intraperitoneally growing ovarian cancer xenografts—a long-term study. Acta Oncol. 39: 741–745. | PubMed | ChemPort |
37. Andersson, H., Lindegren, S., Back, T., Jacobsson, L., Leser, G. and Horvath, G. (1999). Biokinetics of the monoclonal antibodies MOv 18, OV 185 and OV 197 labelled with ¹²⁵I according to the m-MeATE method or the iodogen method in nude mice with ovarian cancer xenografts. Acta Oncol. 38: 323–328. | PubMed | ChemPort |
38. Bauer, J. A., et al. (2002). Effects of interferon beta on transcobalamin II-receptor expression and antitumor activity of nitrosylcobalamin. J. Natl. Cancer Inst. 94: 1010–1019. | PubMed | ChemPort |
39. Mullen, P., Ritchie, A., Langdon, S. P. and Miller, W. R. (1996). Effect of Matrigel on the tumorigenicity of human breast and ovarian carcinoma cell lines. Int. J. Cancer. 67: 816–820. | Article | PubMed | ISI | ChemPort |
40. Calabresi, P., Goulette, F. A. and Darnowski, J. W. (2001). Taurolidine: cytotoxic and mechanistic evaluation of a novel antineoplastic agent. Cancer Res. 61: 6816–6821. | PubMed | ChemPort |
41. Chun, Y. J., et al. (2002). Enhancement of radiation response in human cervical cancer cells in vitro and in vivo by arsenic trioxide (As₂O₃). FEBS Lett. 519: 195–200. | Article | PubMed | ChemPort |
42. Fogh, J., Fogh, J. M. and Orfeo, T. (1977). One hundred and twenty-seven cultured human tumor cell lines producing tumors in nude mice. J. Natl. Cancer Inst. 59: 221–226. | PubMed | ChemPort |
43. Ziegler, A., Zangemeister-Wittke, U. and Stahel, R. A. (2002). Circulating DNA: a new diagnostic gold mine?. Cancer Treat Rev. 28: 255–271. | Article | PubMed | ChemPort |
44. Maire, F., et al. (2002). Differential diagnosis between chronic pancreatitis and pancreatic cancer: value of the detection of KRAS2 mutations in circulating DNA. Br. J. Cancer. 87: 551–554. | Article | PubMed | ISI | ChemPort |
45. Shao, Z. M. and Nguyen, M. (2002). Tumor-specific DNA in plasma of breast cancer patients. Anticancer Drugs. 13: 353–357. | Article | PubMed | ChemPort |
46. Hao, X., et al. (2004). Differential gene and protein expression in primary breast malignancies and their lymph node metastases as revealed by combined cDNA microarray and tissue microarray analysis. Cancer. 100: 1110–1122. | Article | PubMed | ISI | ChemPort |
47. Margalit, O., et al. (2003). Overexpression of a set of genes, including WISP-1, common to pulmonary metastases of both mouse D122 Lewis lung carcinoma and B16-F10.9 melanoma cell lines. Br. J. Cancer. 89: 314–319. | Article | PubMed | ChemPort |
48. Yanagawa, R., et al. (2001). Genome-wide screening of genes showing altered expression in liver metastases of human colorectal cancers by cDNA microarray. Neoplasia. 3: 395–401. | Article | PubMed | ChemPort |
49. Reinholz, M. M., Iturria, S. J., Ingle, J. N. and Roche, P. C. (2002). Differential gene expression of TGF-beta family members and osteopontin in breast tumor tissue: analysis by real-time quantitative PCR. Breast Cancer Res. Treat. 74: 255–269. | Article | PubMed | ISI | ChemPort |
50. Strieter, R. M., Belperio, J. A., Phillips, R. J. and Keane, M. P. (2004). Chemokines: angiogenesis and metastases in lung cancer. Novartis Found. Symp. 256: 173–184. | PubMed | ChemPort |
51. Varney, M. L., Li, A., Dave, B. J., Bucana, C. D., Johansson, S. L. and Singh, R. K. (2003). Expression of CXCR1 and CXCR2 receptors in malignant melanoma with different metastatic potential and their role in interleukin-8 (CXCL-8)-mediated modulation of metastatic phenotype. Clin. Exp. Metastasis. 20: 723–731. | Article | PubMed | ChemPort |
52. Entschladen, F., Drell, T. L., Lang, K., Joseph, J. and Zaenker, K. S. (2004). Tumour-cell migration, invasion, and metastasis: navigation by neurotransmitters. Lancet Oncol. 5: 254–258. | Article | PubMed | ChemPort |
53. Vantyghem, S. A., Postenka, C. O. and Chambers, A. F. (2003). Estrous cycle influences organ-specific metastasis of B16F10 melanoma cells. Cancer Res. 63: 4763–4765. | PubMed | ISI | ChemPort |
54. Shinohara, H., Nakatani, T. and Matsuda, T. (1985). The presence of lymphatic stomata in the ovarian bursa of the golden hamster. Anat. Rec. 213: 44–52. | Article | PubMed | ChemPort |
55. Byrd, J. A., Hayes, T. K., Wright, M. S., Dean, C. E. and Hargis, B. M. (1993). Detection and partial characterization of an anti-steroidogenic peptide from the humoral immune system of the chicken. Life Sci. 52: 1195–1207. | Article | PubMed | ChemPort |
56. Kim, J., Yu, W., Kovalski, K. and Ossowski, L. (1998). Requirement for specific proteases in cancer cell intravasation as revealed by a novel semiquantitative PCR-based assay. Cell 94: 353–362. | Article | PubMed | ISI | ChemPort |
57. Just, L., et al. (2003). Identification of human cells in brain xenografts and in neural co-cultures of rat by in situ hybridisation with Alu probe. J. Neurosci. Methods. 126: 69–77. | Article | PubMed | ChemPort |

Acknowledgements

The authors thank Manon Dubé and the staff at Animal Care and Veterinary Services and Pathology and Laboratory Medicine, University of Ottawa, for technical assistance. This research was supported by the National Cancer Institute of Canada with funds from the Canadian Cancer Society (B.C.V.) and the Betty Irene West doctoral research scholarship from the Canadian Institutes of Health Research (T.J.S.).