Characterization of 11 human sarcoma cell strains

Evaluation of cytogenetics, tumorigenicity, metastasis, and production of angiogenic factors




Human sarcomas have a propensity for aggressive local invasion and early pulmonary metastasis. Frequently, deaths are due to uncontrolled pulmonary metastases. The purpose of the current study was to evaluate cytogenetics, tumorigenicity, metastatic potential, and production of angiogenic factors in human sarcoma cell strains. A secondary purpose was to establish low passage cell strains for studying new therapeutic approaches.


The authors established 11 cell strains from human sarcoma surgical specimens and characterized their in vitro tumor properties, including growth in soft agar, expression of angiogenic growth factors (vascular endothelial growth factor [VEGF] and basic-fibroblast growth factor [bFGF]), and cytogenetics.


All of the cell strains remained diploid. All exhibited the ability to grow in soft agar and expressed both VEGF as well as bFGF. In addition, 6 of the 11 established sarcoma cell strains were tumorigenic, 5 of which spontaneously metastasized to the lungs in nude mice. Four of the five cell strains that yielded lung metastases were derived from lung metastases in patients.


The 11 cell strains, which were derived from diverse sarcoma histologies, will provide a model for studying not only metastatic progression but also the in vitro and in vivo efficacy of new therapeutic modalities for human sarcomas. Cancer 2002;95:1569–76. © 2002 American Cancer Society.

DOI 10.1002/cncr.10879

Sarcomas are a rare malignant tumor with a tendency of persistant local invasion and early pulmonary metastasis. Despite the use of aggressive multimodality treatments, including surgery, radiotherapy, and chemotherapy, outcomes remain mediocre; most treatment failures are due to pulmonary metastasis.1, 2 Therefore, a better understanding of human sarcoma tumorigenesis and metastasis may ultimately lead to improvement in therapy. Because freshly isolated human tumor cell strains and tumor cell lines are useful in studying the biology of sarcomas, we established a series of human sarcoma cell strains in culture, and we present their in vitro tumorigenic properties as well as in vivo tumorigenesis and metastasis using the nude mouse xenograft model.

Freshly isolated human sarcoma cells are valuable tools in studying the biology of human sarcoma. Using freshly isolated human sarcoma cells, we found clonal expansion of p53-mutated tumor cells in synovial sarcoma lung metastasis and identified a point mutation at codon 135 within the critical DNA-binding region of p53.3 Reintroduction of the wild-type p53 gene suppressed the tumorigenicity and enhanced the chemosensitivity and radiosensitivity of sarcoma cells bearing that gene.4 Using the freshly isolated human sarcoma cells and their autologous fibroblasts, human peritumoral and lung fibroblasts were found to produce paracrine motility factors for human sarcoma cells. A human lung fibroblast motility-stimulating factor was isolated and identified as an NH2-terminal fragment of human fibronectin; the factor was produced by human lung fibroblasts and stimulated the motility of human sarcoma cells.5

In the current study, we used 11 recently established cell strains derived from human sarcoma surgical specimens. We found that all of the human sarcoma cell strains had the ability to grow in soft agar and expressed vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) Cytogenetically, all 11 cell strains remained diploid. In the nude mouse, which was used as a tumor-bearing host, six of the cell strains were tumorigenic, five of which developed spontaneous lung metastases. Four of the lung-metastatic cell strains were derived from lung metastases in patients. A followup study showed that the in vivo tumorigenic and metastatic potential of sarcoma cells inversely correlated with patient survival.

The aim of the current study was to evaluate cytogenetics, tumorigenicity, metastatic potential, and production of angiogenic factors in human sarcoma cell strains. A second aim was to establish low passage cell strains which may be useful in evaluating new sarcoma therapies.


Sarcoma Cell Strains and Nude Mice

Eleven sarcoma cell strains were established in primary cell culture directly from surgical sarcoma specimens obtained at The University of Texas M.D. Anderson Cancer Center using methods described previously6 (Table 1). Briefly, sterile fragments from the resected tumor were minced in culture medium and then disaggregated by overnight incubation in collagenase (100 U/mL at 37 °C). Cells were cultured in Dulbecco's modified eagle's Medium containing 10% fetal calf serum (FBS, Life Technologies, Basel, Switzerland), penicillin-streptomycin (FBS and bovine pituitary extract (FBS). Once cells became confluent, adherent cells were removed by trypsin treatment (FBS) and replated at a 1:2 or 1:3 ratio with medium. Throughout the establishment of these cell strains, each maintained a typical spindle-shaped morphology. All of the cell strains were passaged fewer than 10 times. Additionally, they were routinely examined with GenProbe (Fisher Scientific, Pittsburgh, PA) and found to be negative for mycoplasma contamination.

Table 1. Designation, Patient Characteristics, and Tumor Characteristics of Human Sarcoma Cell Strains
Cell strain designation (number)Patient age (years)GenderTumor sourceTumor siteTumor histologyGrowth in mice (route)Spontaneous metastases in mice
  1. M: male; F: female; met: metastasis; MFH: malignant fibrous histiocytoma; s.c.: subcutaneous; i.m.: intramuscular.

UCS-2 (8)31MPrimaryThighUnclassified sarcomaNN
MFHm-2 (17)17MMetRetro-peritoneumMyxoid MFHNN
UCS-4 (23)40MMetLungUnclassified bone sarcomaY (s.c., i.m)Y
SYN-1 (24)28FMetLungSynovial sarcomaY (s.c., i.m.)Y
UCS-5 (26)48FMetLungUnclassified sarcomaNN
LEIO-2 (36)66MMetStomachLeiomyo-sarcomaNN
SYNb-1 (43)20MPrimaryForearmSynovial sarcomaNN
MFH-3 (50)46MMetLungMFHY (s.c., i.m.)Y
SYNb-2 (56)20MMetLungSynovial sarcomaY (i.m.)N
EES-1 (58)20FMetLungEwing sarcomaY (i.m.)Y
NFS-2 (60)30FMetRetro-peritoneumNeurofibrosarcomaY (i.m.)Y

Three to four week-old nu/nu athymic nude mice were purchased from Harlan Sprague Dawley (Indianapolis, IN) and housed under specific pathogen-free conditions. The care and use of the animals were carried out in accordance with institutional and National Institutes of Health guidelines.

Northern Blot Analysis

Total RNA was extracted from cell strains in log growth phase using the RNAzol B reagent (Biotecx Laboratories, Houston, TX). Northern blot analysis was performed using Hybond N nylon membranes (0.45 μm pore size; Amersham, Arlington Heights, IL) in an aqueous hybridization solution as described by Ausubel et al.7 The hybridization probes used were VEGF, bFGF, and glyceraldehyde-3-phosphate dehydrogenase cDNA fragments radiolabeled using a random-primed synthesis kit (Life Technologies, Rockville, MD). Finally, blots were washed at high stringency (0.5 x saline-sodium citrate and 0.1% sodium dodecyl sulfate at 68 °C) and exposed to Kodak BioMax film (Eastman Kodak, Rochester, NY).

Soft Agar Colony Formation Assay

The soft agar colony formation assay was performed as described previously.8 Briefly, 0.5 mL of an underlayer consisting of Dulbecco's modified essential (DME) F-12 medium, 10% fetal bovine serum, and 0.7% agarose was plated in 24-well plates. Two thousand cells/mL were suspended in DME/F-12 medium with 10% fetal bovine serum and 0.35% agarose, and 0.5 mL (1,000 cells/well) of the suspension were plated on top of the gelled underlayer. After incubation in 5% CO and 95% air at 37 °C for three weeks, cell colonies were stained with Iodonitrotetrazolium violet (Sigma, St. Louis, MO) and counted under light microscopy.

Chromosome Analysis

Subconfluent sarcoma cell cultures were treated with 15 mL of a solution containing 10 μg/mL colcemid (Life Technologies) for 10 minutes at 37 °C. The cells were harvested after treatment with a solution containing 0.25% trypsin for one minute. Pooled cells were then resuspended in a hypotonic solution containing 0.075 M KCl; 1 mL of fixation solution (75% methanol, 25% glacial acetic acid) was added, mixed via inversion, and centrifuged. Next, the pellet was resuspended in 5 mL of fixation solution, washed several times, and dropped onto slides. The slides were stained with Giemsa stain (Biomedical Specialties, Santa Monica, CA), and metaphase chromosomes were analyzed under light microscopy as described previously.9, 10

In Vivo Tumorigenicity of Sarcoma Cells in Nude Mice

Subconfluent sarcoma cell cultures were harvested by treatment with a solution containing 0.25% trypsin and 1 mM ethylenediaminetetraacetic acid (EDTA), washed twice by centrifugation and resuspension in serum-free Dulbecco's modified essential media (DMEM), and finally resuspended in 1 mL of ice-cold, serum-free DMEM. In vivo tumorigenicity experiments were performed using subcutaneous and intramuscular implantation methods. In the subcutaneous implantation experiments, 2 × 106 sarcoma cells (UCS-2, UCS-5, or MFHm-2) or 4 × 106 sarcoma cells (UCS-4, SYN-1, LElo-2, MFH-3, EES-1, NFS-2, SYNb-1, or SYNb-2) in 0.1 mL of DMEM were injected subcutaneously into the flank of 5–6-week-old nu/nu athymic nude mice using 23-gauge needles. Three mice received injections of each sarcoma cell strain. The viability of the sarcoma cells was confirmed via trypan blue exclusion; the viability of all sarcoma cell strains was above 80%. In the intramuscular implantation experiments, 5 × 106 sarcoma cells in 0.1 mL of Hank's balanced salt solution (Life Technologies) were injected intramuscularly into the hip of 5–to 6-week-old nu/nu athymic nude mice using 23-gauge needles. Six mice received injections of each sarcoma cell strain in the presence of 33% (v/v) Matrigel (Becton Dickinson and Co., Franklin Lakes, NJ), while another six received injections without Matrigel. The viability of the sarcoma cells was confirmed by trypan blue exclusion; the viability for all sarcoma cell strains was above 80%.

The nude mice were housed three per cage and checked periodically for tumor formation. Once a tumor nodule formed, the mouse was evaluated, and the nodule was measured weekly. The mice were sacrificed when their tumor nodules grew beyond 15 mm in the greatest diameter or when they became moribund. The tumor nodule, lungs, heart, liver, kidneys, and spleen of each mouse were fixed in a 10% formaldehyde solution (Fisher Scientific), embedded in paraffin, and processed. Sections 4 mm thick were obtained and stained with hematoxylin and eosin prior to examination under light microscopy.


Expression of Angiogenic Growth Factors

Angiogenesis is necessary for in vivo tumor growth and metastasis.11 Vascular endothelial growth factor and bFGF have been shown to play major roles in tumor angiogenesis.11 Therefore, we examined the expression of VEGF and bFGF mRNA in 11 sarcoma cell strains using northern blot analysis. All 11 sarcoma cell strains expressed VEGF and bFGF (Fig. 1), which is consistent with their intimate involvement in tumor angiogenesis.

Figure 1.

Northern blot analysis of vascular endothelial growth factor (VEGF) and basic–fibroblast growth factor (bFGF) expression in human sarcoma cell strains. GAPDH: glyceraldehyde–3–phosphate dehydrogenase.

Anchorage Independent Growth

Anchorage independent growth is one of the properties of tumor cell lines.8 To test whether sarcoma cells established from surgical specimens retained this tumor-associated property, they were examined using the soft agar colony formation assay. All 11 sarcoma cell strains were able to grow in soft agar (Fig. 2), suggesting that they retained the tumorigenic phenotype for anchorage independent growth.

Figure 2.

Anchorage independent growth of human sarcoma cell strains.

Cytogenetic Analysis

Human sarcomas frequently harbor chromosomal abnormalities. To analyze the cytogenetics, sarcoma cells were subjected to colcemid and light trypsin treatment and dropped onto slides to separate metaphase chromosomes.9, 10 All 11 sarcoma cell strains remained diploid according to conventional metaphase analysis by light microscopy (Fig. 3). However, the absence of cytogenetic abnormalities did not speak to the possible presence of cryptic translocations, deletions, or mutations.

Figure 3.

Conventional cytogenetic analysis of human sarcoma cells (representative karyotype of SYNb-2 cells).

In Vivo Tumorigenic and Metastatic Properties of Sarcoma Cells

Using fresh surgical sarcoma specimens, we established 11 human sarcoma cell strains in culture (Table 1). To examine whether these cells retained their in vivo biologic properties, such as tumorigenicity and propensity to metastasize to the lungs, we studied these properties using the nude mouse xenograft model.

In the subcutaneous tumorigenicity assays, the 11 sarcoma cell strains were injected into the flank of nude mice. Three of 11 sarcoma cell strains formed tumors, with an average latency period of more than 6 months. Specifically, the cell strain MFH-3 formed tumors in both the flank and neck regions, whereas SYN-1 and UCS-4 formed tumors only in the neck region. These three cell strains were originally isolated from pulmonary metastases (Table 1).

In the intramuscular tumorigenicity assays, 11 sarcoma cell strains were injected with or without Matrigel into the hips of nude mice. Six of the cell strains formed tumors, with a latency period of 6–12 months (data not shown). The groups that received Matrigel in their injections generally had shorter latency periods than did those without Matrigel; this shorter latency may have been due to the angiogenic stimulatory effect of Matrigel on tumor xenografts in immunodeficient animals.12 Six of the cell strains (EES-1, MFH-3, NFS-2, SYN-1, SYNb-2, and UCS-4) formed tumors in the flank, and five of these (EES-1, MFH-3, NFS-2, SYN-1, and UCS-4) spontaneously formed metastatic lesions in the lungs (Fig. 4). This metastatic pattern closely resembled that observed in sarcoma patients. Interestingly, the original patient sources of these sarcoma cell strains were either locoregional or pulmonary metastatic lesions. Specifically, the cell strains that formed metastatic lesions in the lungs were mostly derived from pulmonary metastatic lesions (Table 1), suggesting that these human sarcoma cells retained their biologic properties, including tumorigenicity and a propensity for spontaneous lung metastasis.

Figure 4.

Hematoxylin and eosin stained sections of pulmonary metastatic lesions formed by human sarcoma xenografts. A) EES-1. B) MFH-3. C) NFS-2. D) SYN-1. E) UCS-4. All photomicrographs were taken at the same magnification. Scale bar = 50 μm.

Examination of the tissue sections obtained from tumor xenografts grown in nude mice (sarcoma cell strains MFH-3, NFS-2, SYN-1, SYNb-2, and UCS-4) revealed histopathologic features reminiscent of the original tumors (Fig. 5).13 For example, MFH-3 consisted of sheets of spindle cells of various sizes and large polygonal, histiocytic cells having a pronounced nuclear pleomorphism, hyperchromaticity, and abundant mitotic figures, all of which are morphologic features consistent with malignant fibrous histiocytoma.13 Also in the sections obtained from nude mice that received intramuscular injections, infiltration of striated muscle by tumor cells was observed, and the muscle fibers were separated by numerous invading tumor cells (Fig. 5).

Figure 5.

Hematoxylin and eosin stained sections of human sarcoma xenografts. A) MFH-3. B) NFS-2. C) SYN-1. D) SYNb-2. E) UCS-4. F) EES-1. All photomicrographs were taken at the same magnification. Scale bar = 50 μm.

Additionally, the tumorigenic and metastatic potential of human sarcoma cell strains cultured in vitro reflected the invasive and metastatic behaviors of sarcomas growing in patients. It is possible that these experimental manifestations may therefore be useful in predicting the clinical aggressiveness of the tumor and prognosis for sarcoma patients. To test this hypothesis, clinical information on nine soft-tissue sarcomas (STSs) was compared by Kaplan-Meier analysis (data not shown); cell strains UCS-4 and EES-1 were obtained from lung metastases that originated in the bone and therefore were not included in the analysis. Although not statistically significant, patients whose cell strains formed tumors in nude mice tended to have shorter survival durations after tumor resection than did patients in the nontumorigenic group.


There is a lack of suitable human sarcoma models, most likely due to the rarity of this tumor type.1, 2 However, sarcoma models have recently been developed that enable us to better understand these tumors. For example, the human synovial sarcoma cell line HS-SY-II,14 a human clear-cell sarcoma cell line HS-MM,15 and human epithelioid sarcoma cell line ES02048816 have been established and shown to form tumors in nude mice and retain their original morphologic properties both in vitro and in vivo. Furthermore, the mesenchymal chondrosarcoma cell line HS-16, myxoid liposarcoma cell line HS-18, malignant hemangiopericytoma cell line HS-30, and a malignant mesenchymoma cell line have been established and characterized for their chemosensitivity to paclitaxel, methotrexate, vinblastine, and 5-fluorouracil.17 Two of the human synoviosarcoma cell strains in the current study SYNb-1 and SYNb-2—were established from primary and metastasic lesions in the same patient, respectively. Clonal expansion of p53-mutated tumor cells has been found in SYNb-2;3 additionally, the human synoviosarcoma cell line SYN-1 was established, and a human lung fibroblast motility-stimulating factor was isolated due to its ability to stimulate motility of SYN-1 cells.6 Finally, in the current study, five of the human sarcoma cell strains that we established retained the property to metastasize to the lungs in nude mice, which may be useful for studying sarcoma lung metastasis and its treatment.

Tumor-induced angiogenesis is essential for successful tumor growth and metastasis.11 Therefore, we examined the angiogenic growth factors VEGF and bFGF. We found that both factors were expressed in all of the sarcoma cell strains, which is consistent with their important role in tumor angiogenesis. However, production of proangiogenic cytokines, bFGF and VEGF, was not sufficient for tumorigenesis or metastasis and did not correlate with patient survival. The finding that sarcoma cell strains universally express factors that promote angiogenesis is remarkable and should herald the exploration of antiangiogenic therapy for sarcoma. We have found that the use of antiangiogenic therapy enhances the antitumor efficacy of doxorubicin in human xenografts.18

Anchorage independent cell growth is a property of the tumorigenic phenotype and has been used as a marker for transformation in vitro.8 In the current study, we found that all the sarcoma cell strains were able to grow in soft agar, which is consistent with the tumorigenic phenotype. Thus, these cell strains are a valid model for studying tumor cell characteristics in vivo.

In some types of solid tumors, cells remain diploid, and no gross cytogenetic abnormalities can be appreciated using conventional cytogenetic analysis. Only limited, subtle changes, such as inversion or translocation of small chromosomal regions, occur, and they can only be detected using sensitive methods such as G-banding analysis, as seen in the study of a recurrent adenoid cystic carcinoma9 and a primary mucoepidermoid carcinoma of the parotid gland.10 Moreover, in the sarcoma cell strain SYNb-2, the wild-type p53 gene was functionally inactivated by a point mutation at codon 135 within the critical DNA-binding region.3 Thus, the functional inactivation of p53 was not manifested at the cytogenetic level.3 In the current study, no gross cytogenetic abnormalities, such as gain or loss of chromosomes, hyperploidy, or translocation of large chromosomal regions, were found by conventional cytogenetics in the sarcoma cell strains. The sarcoma cells remained diploid, although it is possible that cryptic translocations were present.

Although chromosomal translocations are common in Ewing sarcoma and synovial sarcoma, many of these cytogenetic abnormalities may not be apparent using conventional cytogenetic analyses. However, it is not infrequent for Ewing sarcoma cell lines and short-term cultures to lack the t(11;22) when studied by conventional cytogenetics.19, 20 Moreover, other investigators have shown that Ewing sarcoma cell lines and short term cultures may have normal cytogenetics when assayed by conventional methods. Douglass et al. described 15 Ewing sarcomas that had sufficient tissue to study detailed cytogenetics and found that 2 of 15 Ewing sarcomas had normal metaphases.21 Workman et al. reported that one of four Ewing sarcomas had, normal karyotype.22 Interestingly, not until after 29 passages in nude mice did the karyotype exhibit a del(13). Primitive neural ectodermal tumors (PNET) are similar to Ewing sarcoma and share the frequent t(11;22). Gorman et al.23 and Sorenson et al.24 have both reported PNET cell lines with normal cytogenetics.

The presence of t(X;18) appears to have marked specificity for synovial sarcoma, but normal karyotypes have been reported. Panagopoulos et al.25 reported the karyotypes of 60 cases of synovial sarcoma and found that three of these displayed normal karyotype by conventional cytogenetics but a molecular t(X;18) by reverse transcriptase polymerase chain reaction (RT-PCR). A recent case report described a 26–year–old man with a pelvic tumor that could not be classified.26 Conventional cytogenetic studies were nondiagnostic, but a t(x;18) was detected by fluourescent in situ hybridization. Subsequently, this synovial sarcoma was further characterized as SYT-SSX2 by RT-PCR methods. Others have also reported synovial sarcomas with normal karyotypes by conventional cytogenetics that harbored cryptic translocations.27 This is in no way surprising, since molecular translocation of the Philadelphia chromosome in chronic myelogenous leukemia is well-accepted. Although translocation t(X;18) is not found in 10% of synovial sarcomas, virtually 100% of these tumors have detectable chimeric transcripts by RT-PCR.27

Therefore, it is not impossible for these Ewing sarcoma and synovial sarcoma cell strains to lack chromosomal translocation when evaluated by conventional cytogenetic methods. Although neither the t(11;22) nor the t(X;18) were observed in the current study, it remains possible that there are cryptic translocations of the EWS1 gene or the SYT gene at the molecular level. The molecular studies necessary to elucidate these cryptic translocations are beyond the scope and intent of the current report. However, it is clear from the current study that both the Ewing sarcoma cell strain (EES-1) and the synovial sarcoma cell strains (SYN-1 and SYNb-2) grow in soft agar and are tumorigenic (Table 1) in nude mice. Tumorigenicity would not be expected if the cell strain were composed solely of contaminating fibroblasts. Nonetheless, it remains possible that the karyotypes obtained from short term culture may be from contaminating fibroblasts. Acknowledgment of this limitation does not deter the potential utility of these cell strains in the study of sarcoma.

Human tumor xenografts require an appropriate microenvironment to grow in immunodeficient animals.28 Orthotopic sites that most closely resemble the microenvironments in which human tumors develop are the most likely sites for successful tumor implantation.28 In fact, it has been observed that orthotopically implanted human tumor xenografts have an increased chance of establishment, growth, and metastasis in immunodeficient animals when compared with nonorthotopically implanted xenografts,28 as is the case with human renal cell carcinoma29 and breast carcinoma.30 Additionally, in experiments using the subcutaneous implantation method, only the cell strain MFH-3 consistently developed tumor nodules at the injection site. The cell strains SYN-1 and UCS-4 did form tumors but less consistently than MFH-1. In contrast, in experiments using the orthotopic intramuscular implantation method, most of the sarcoma cell strains (EES-1, MFH-3, NFS-2, SYN-1, SYNb-2, and UCS-4) formed tumor nodules at the injection sites, underscoring the effects of an appropriate microenvironment on the establishment and growth of human sarcoma xenografts in nude mice. In addition, orthotopic intramuscular implantation of the cell strains EES-1, MFH-3, NFS-2, SYN-1, and UCS-4 in the right hip led to the appearance of spontaneous lung metastasis, which is consistent with the observation that orthotopic implantation tends to increase the likelihood of developing visceral metastases in immunodeficient animals.28 It is of importance to note that all sarcoma cell strains that grew in nude mice were isolated from a metastatic lesion. Interestingly, the sarcoma cell strain SYNb-1, which was derived from a primary sarcoma lesion, was nontumorigenic in nude mice, whereas SYNb-2, which was derived from a metastatic lesion in the same patient, was tumorigenic in nude mice. This may have been due to clonal expansion of p53-mutated tumor cells in SYNb-2,3 which implicates inactivation of p53 as a critical event in development of the metastatic phenotype.

Interestingly, analysis of patient survival data showed a trend toward decreased survival duration if their sarcoma cells were tumorigenic in nude mice. The tumorigenicity of human malignancies in immunodeficient animal hosts has also been correlated with poor patient survival in studies of cerebellar primitive neuroectodermal tumor,31 renal cell carcinoma,32 colorectal and nonsmall cell lung tumors,33 and neuroblastoma.34 Although not of statistical significance in the current small study, a larger study may find that the nude mouse tumorigenicity assay may be a useful predictor of prognosis for sarcoma patients.

The current study resulted in the establishment and characterization of 11 sarcoma cell strains, derived from diverse histologic subtypes of soft-tissue and bone sarcoma. These are all early-passage cell strains and have not been subjected to potential selection pressures of prolonged tissue culture. Also, these cell strains comprise tumorigenic and nontumorigenic as well as metastatic and nonmetastatic phenotypes. The current results suggest that evaluation of the tumorigenic and metastatic properties of human sarcomas is possible using low passage cell strains. As new therapeutic agents are developed, such as inhibitors of angiogenesis, tyrosine kinases, retinoid family receptors, matrix metalloproteinases, and cell cycle progression, the understanding and availability of these cell strains will be critical to the development of new models to test these agents. Because of the relative chemoresistance of metastatic sarcoma, development of in vitro models is necessary to further study the use of preoperative and postoperative regimens containing chemotherapy, radiotherapy and novel agents.


The authors thank Dr. R. Morrison for donating the basic–fibroblast growth factor cDNA.