Osteosarcoma is the most common malignant bone tumor in children and is the third most common cancer among adolescents . It is highly metastatic, and > 80% of patients develop pulmonary metastases after surgical removal of the primary tumor as monotherapy . Despite the addition of chemotherapy, up to 40% of patients who present with localized disease will develop pulmonary metastases . The 5-year survival for children with metastatic osteosarcoma is extremely low and the long-term survival for these patients has not improved during the last 30 years . This high risk of metastasis in patients with osteosarcoma is believed to result from subclinical micrometastatic disease present in the lungs before diagnosis. Novel therapies aimed at preventing microscopic tumor foci from blossoming into overt metastatic disease are urgently needed to improve survival.
To understand the complex nature of metastatic progression in osteosarcoma, we need preclinical models that accurately represent the human disease state. Two of the first successfully established osteosarcoma metastatic cell lines, K12 and K7, were derived from an osteosarcoma tumor that spontaneously developed in a mouse. The K7M2 cell line, which was developed from pulmonary metastases of K7 cells in the immunocompetent inbred laboratory mouse strain, BALB/c , has been used to study the biology of metastasis and has been used in vivo to screen novel osteosarcoma therapies .
Because osteosarcoma is thought to arise from bone marrow-derived mesenchymal stem cells , models in which tumor cells are injected orthotopically into the bone marrow cavity of the femur or tibia often are used [8, 12, 15, 17]. This technique permits osteosarcoma tumor cells to interact with their native microenvironment, and several papers have reported the development of spontaneous pulmonary metastases after intratibial injection of osteosarcoma cell lines [5, 8, 12, 15, 17]. However, there is substantial variation in reported efficiency of bone tumor establishment, tumor volumes, and metastatic efficiency across studies using intratibial injection models [5, 8, 14, 17]. The volume used for cell injection varies, and leakage from the bone marrow cavity into the paratibial subcutaneous space has been described [8, 11].
The purpose of our study was to characterize this widely used murine model of metastatic osteosarcoma and determine whether it is an appropriate model with which to study spontaneous pulmonary metastasis. We aimed to generate a reliable method of intratibial injection by determining the maximal volume of injected cells that will remain within the tibia, determining the incidence of primary tumor and metastatic formation, determining the kinetics of pulmonary metastatic seeding and outgrowth, and determining the contribution of the primary tumor to subsequent development of metastasis in this model.
Materials and Methods
Cell Lines and Tissue Culture
The K7M2 murine osteosarcoma cell line was purchased from the American Type Culture Collection (ATCC, Rockville, MD, USA). Cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco, Grand Island, NY, USA) supplemented with 10% heat-inactivated fetal calf serum (FetalClone II; HyClone Laboratories, Logan, UT, USA), 2 mM L-glutamine (Gibco), and 1% penicillin/streptomycin (HyClone Laboratories). Vented tissue culture flasks (Nunc Cell Culture Systems, Rochester, NY, USA) were maintained at 37° C in a humidified atmosphere of 5% carbon dioxide in air, and cells were split when they reached approximately 80% to 90% confluence.
All procedures involving animals were approved by, and in accordance with, the ethical standards of the Institutional Animal Care and Use Committee of The Feinstein Institute for Medical Research (Manhasset, NY, USA). Female BALB/c mice were purchased from Taconic Biosciences (Germantown, NY, USA) at 4 to 5 weeks of age, mean weight 16.7 g (± 1.2 g), and were housed under specific pathogen-free conditions with five mice per cage in a 12-hour light/dark cycle with ad libitum access to food and water. All animal manipulations and surgeries occurred in the morning (8:00 AM to 12:00 PM). We assumed that primary tumor formation would be close to 100% with a maximum of 100% and a minimum > 90%. We also assumed that the presence of pulmonary metastasis would be > 75% in mice with primary tumors and would increase with time . SDs for each time point of measurement or surgical group were estimated to be 15%. We determined that five to six mice per group would be sufficient to measure a difference in means of 25% with 90% power. To determine primary tumor formation and incidence of metastasis, 30 mice were used. For the kinetics of metastatic seeding and growth, and the impact of surgical removal of the injected limb, five to six mice were used per group (Fig. 1).
Establishment of a Reliable Volume for Intratibial Injection
We harvested low-passage (P3-10) K7M2 cells using 0.05% trypsin with 0.2 g/L EDTA in phosphate-buffered saline (PBS; HyClone Laboratories). After harvesting, the cells were washed with PBS and brought to a concentration of 1 x 108 cells/mL in PBS. Cell viability was assessed using trypan blue exclusion (Gibco). We continued the experiments if cell viability was > 90%. Cells were kept at 4° C in preparation for implantation. Mice were anesthetized with buprenorphine (2 mg/kg, subcutaneous) and inhaled 3% isoflurane. The left hind limb was prepared with a 70% ethanol swab. The lateral malleolus, medial malleolus, and lower half of the left tibia were grasped and the leg bent in a combination of flexion and lateral rotation to expose the knee as previously described . Using a 28-G ½-inch needle and a drilling motion, the needle was inserted through the patellar ligament and into the anterior intercondylar area of the tibia. The needle was withdrawn and either a separate 28-G ½-inch needle or a Hamilton syringe (Hamilton Company, Reno, NV, USA) filled with cell suspension was used to slowly inject 1 x 106 cells into the previously drilled tibia tract. Postoperatively, the mice were kept on a heating pad until they recovered from anesthesia, and they received an additional 2-mg/kg dose of buprenorphine 12 hours later. Postoperative monitoring of mice, including collection of body weight and assessment of pain, occurred daily for 3 days after injection and twice weekly thereafter. At 4 weeks post-injection, many of the mice met euthanasia criteria, with large tumors, weight loss, and general ill health, and therefore all studies were terminated no later than 4 weeks after implantation, consistent with a prior report .
To determine the maximum volume of cell suspension that could be injected into the tibia without evidence of leakage, mice were injected with 100, 50, 25, and 10 μL of 30 mg/kg Evan’s blue dye, respectively (n = 8, two mice per group). Anesthetized mice were euthanized immediately after injection and their tibias were dissected free of the overlying skin and soft tissue to evaluate for dye leakage.
Determination of the Incidence of Primary Tumor Formation and Metastatic Development
Mice (n = 30) were injected with 10 μL of cell suspension containing 1 x 106 cells as described previously and observed for 4 weeks. Of note, three mice died on intratibial injection and were excluded from further analysis. Mice were euthanized after 4 weeks and the incidence of primary tumor formation and pulmonary metastasis was recorded.
Determination of the Kinetics of Pulmonary Metastatic Seeding and Outgrowth
To determine the kinetics of metastatic development, mice underwent intratibial injection as described previously and were euthanized at 1, 2, 3, or 4 weeks after injection (n = 24 mice, five to six per group). Mice were randomized into groups based on cage (five mice per cage). One mouse died and one mouse had an inaccurate injection into the soft tissue and were excluded from analysis. Lungs were harvested for analysis, number of metastatic foci was counted, and metastatic burden was quantified in a blinded fashion (described subsequently).
Determination of the Contribution of the Primary Tumor to Subsequent Pulmonary Metastasis
To determine the contribution of the primary tumor to development of metastasis, mice were injected with tumor cells as described previously and amputation of the injected hind limb was performed at 30 minutes, 4 hours, and 24 hours after tumor cell inoculation (n = 21 mice, five to six per group). Control mice consisted of nonamputated, tumor-bearing mice. All mice were euthanized 4 weeks after tumor implantation, their lungs were harvested for analysis, and metastatic burden was calculated in a blinded fashion (see subsequent description). Metastatic burden in mice that underwent amputation was compared with control mice with primary tumors left in situ for 4 weeks.
To perform the amputation, a circumferential skin incision was made proximal to the knee and the femoral neurovascular bundle was identified, ligated with 4-0 Vicryl® suture (Ethicon US, LLC, Cincinnati, OH, USA), and divided. The femur was transected with scissors and the muscle approximated over the femoral stump using 4-0 Vicryl suture. The skin was closed with metal wound clips. Postoperatively, the mice were kept on a heating pad until they recovered from anesthesia; they received an additional 2-mg/kg dose of buprenorphine 12 hours later. Postoperative monitoring occurred twice daily for 3 days and mice were monitored three times weekly thereafter with collection of weight and assessment of pain.
Assessment of Primary Tumors and Pulmonary Metastasis
Mice were euthanized by CO2 inhalation. Primary tumors were excised en bloc along with the surrounding involved bone, including the articulation between the femur and the tibia. The femur was transected with scissors at the proximal border of the tumor; the tibia was transected at the distal border and fixed with 10% neutral-buffered formalin. Bone was decalcified with EDTA/sucrose decalcifying solution (Newcomer Supply, Middleton, WI, USA) before embedding in paraffin. Lungs were dissected, fixed with 10% neutral-buffered formalin, and embedded in paraffin. Incidence of lung metastases was detected by gross observation (presence of nodules on the surface of the intact lung) and histopathologic assessment (presence of histologic metastases, counting the number of metastatic foci, and calculating metastatic burden as described subsequently). Paraffin-embedded specimens were serially cut in longitudinal, 5-µm thick sections and stained by routine hematoxylin and eosin methods. Four sections of each tissue were examined per mouse and averaged. To assess pulmonary metastases, each slide was examined microscopically under low power (0.63-20 x) on a Zeiss Axiovert 200M inverted microscope (Carl Zeiss AG, Oberkochen, Germany) with AxioVision software (Version 126.96.36.199; Carl Zeiss AG) to visualize approximately 50% of the lung section per microscopic field. An experienced pediatric pathologist (ME) assisted with all of the microscopic assessments of the pulmonary metastases and was blinded to information regarding which study group each animal was part of or the presence or absence of a primary tumor. Micrometastases were defined as ≥ 4 cells clustered together. The total number of metastases per lung section was counted, and the average was calculated. Metastatic burden was calculated by measuring the average surface area of metastatic lesions in the lung divided by the total surface area of the lung per section using ImageJ software (Version 1.47; National Institutes of Health, Bethesda, MD, USA). Two independent researchers utilized this method to measure metastatic burden in a blinded fashion with high concordance (r2 = 0.98; 95% confidence interval [CI], 0.95-0.99).
Differences in metastatic burden and the number of metastatic foci between groups were compared using Kruskal-Wallis and one-way analysis of variance, respectively, with multiple comparisons corrected for using the Tukey multiple comparisons test. A difference between groups of p < 0.05 was considered significant.
A Reliable and Reproducible Volume for Intratibial Injection
After injecting increasing volumes of Evan’s blue dye, we determined that the maximum volume that can be instilled in the tibial bone marrow cavity without evident dye spillage is 10 μL. When more than this volume was injected, there was clearly observable leakage of fluid into the adjacent soft tissue.
Efficacy of Primary Tumor Formation and Pulmonary Metastasis After Intratibial Injection
To determine the efficiency of primary tumor development, 30 mice were injected with K7M2 cells. Three mice (10%) died immediately after tumor cell injection and were excluded from further analysis. In all, 77% (21 of 27) of mice developed palpable tumors within 3 weeks of injection, and 25 of 27 (93%) had palpable tumors by 4 weeks. Primary tumors involved the bone marrow cavity in 23 of 27 (85%) mice with tumor invading into the bone and surrounding soft tissue (Fig. 2A). Replacement of normal bone marrow tissue with malignant cells is evident on higher magnification (Fig. 2B). While initially establishing our injection technique, 4 of 27 (15%) of the injected mice had suboptimal injections and developed subcutaneous tumors with no evidence of bone marrow involvement on histologic examination.
When assessing metastatic incidence, 21 of 23 (91%) mice with confirmed intratibial tumors had numerous grossly detectable and coalescent metastatic lesions by 4 weeks, often resulting in consolidation of the lung (Fig. 3A). Moreover, 100% of mice had metastatic lesions detected histologically (Fig. 3B) with tumor replacing 38% (± 18%) of the area of the lung. Gross metastases were only detectable in the lung and pleura without liver or spleen involvement. Notably, the four mice with inaccurate injections without bone marrow involvement did not develop metastatic disease by 4 weeks.
Kinetics of Metastatic Seeding and Metastatic Burden After Intratibial Injection
To determine the timing of metastatic development, mice were euthanized at 1, 2, and 3 weeks and compared with control mice that had been euthanized at 4 weeks. After week 1, 100% of mice had histologically detectable micrometastatic disease in the lungs (Fig. 4 A-B). Metastatic burden increased in the lungs over time (Fig. 4C) with values of 0.88% ± 0.58% at week 1, 6.6% ± 5.3% at week 2, 16.1% ± 12.5% at week 3, and 40.3% ± 14.83% at week 4 and a mean difference from week 1 to week 4 of -39.38 (p < 0.001; 95% CI, -57.39 to -21.37). Although the number of metastatic foci seemingly increased over time, especially between the first and second weeks (Fig. 4D), the differences did not reach statistical significance (week 1 to week 2, 36.4 ± 33.6 versus 77.6 ± 22.2, mean difference -41.2, p = 0.142, 95% CI, -92.42 to 10.02; week 2 to week 3, 77.6 ± 22.2 versus 56.67 ± 31.41, mean difference 20.93, p = 0.631, 95% CI, -29.1 to 69.98; and week 3 to week 4, 56.67 ± 31.41 versus 49.33 ± 26.6, mean difference 7.33, p = 0.97, 95% CI, -39.43 to 54.09). Notably, neither metastatic burden nor number of foci correlated with primary tumor weight (data not shown).
Effect of Surgical Removal of the Injected Limb on Subsequent Metastatic Development
To more definitively determine the source of the metastatic cells (primary tumor versus initial injection versus both), we amputated the tumor-bearing limb at 30 minutes, 4 hours, and 24 hours post-injection and compared the development of metastasis at 4 weeks post-injection with mice that did not undergo amputation. Both control and amputated mice had gross pulmonary lesions, regardless of the timing of amputation (not shown). There was no difference in metastatic burden (Fig. 5) with values for the control group of 40.9% ± 15.3% versus 30-minute amputation of 48.9% ± 6.1% (mean difference -7.96, p = 0.819; 95% CI, -33.9 to 18.0) versus 4-hour amputation of 51.8% ± 17.8% (mean difference -10.9, p = 0.641; 95% CI, -36.79 to 15.08) and versus 24-hour amputation of 48.9% ± 17.9% (mean difference -1.1, p = 0.999; 95% CI, -27.01 to 15.08). Furthermore, histologic examination of the lungs of the three mice that died immediately after intratibial inoculation in the first experiment revealed large tumor emboli in the pulmonary vessels (Fig. 6). Thus, these results indicate that intratibial injection results in the immediate arrival of tumor cells in the lung, independent of the formation of a primary tumor.
Establishing accurate preclinical models is paramount to the study of tumor biology and metastasis. Thorough characterization of the available tumor models and how they represent the different stages of human primary tumor growth and metastasis is important to guide the use of these models in the development of novel therapeutics. Targeting the growth of indolent lung metastases represents an unmet therapeutic need in the treatment of patients with osteosarcoma. The intratibial orthotopic injection model of osteosarcoma has been widely described as a model of spontaneous metastasis and has been utilized to study drugs aimed at preventing metastatic progression. In the present study, we demonstrate that intratibial injection of the commonly used osteosarcoma cell line K7M2 into syngeneic BALB/c mice results in direct seeding of the lung vasculature and, therefore, represents an experimental rather than a spontaneous model of metastasis.
We demonstrate that intratibial injection results in increasing metastatic burden in the lungs over the course of 4 weeks; however, the number of metastatic foci did not change dramatically over the course of the experiment. The number of foci initially was quite variable between mice, most likely reflecting differences in the time between arrival of tumor cells in the lung and subsequent invasion and proliferation to generate visible micrometastases of at least four cells. This presumably stabilized by 2 weeks. There are two possible explanations for the absence of an increase at later times. Either no additional seeding occurs after initial arrival of tumor cells in the lung at the time of injection or the increase in metastatic foci was masked by a coalescence of the enlarging lesions. Coalescence could also explain the apparent drop in number of foci at weeks 3 and 4. Additionally, primary tumor size did not correlate with metastatic burden. Most notably, further investigation demonstrated that amputation of the tumor-bearing limb performed 30 minutes after intratibial injection had no effect on the extent of pulmonary metastases compared with controls when examined at 4 weeks. These findings are in contrast to two papers reporting decreased incidence of metastasis after hind limb amputation [12, 17]. This apparent disparity could be the result of differences in injection technique. Our study represents a true intratibial injection, whereby tumor cells are implanted directly into the bone marrow cavity; however, many other studies that refer to intratibial injection are often injected in proximity to the tibia but not in the medullary space or use large volumes of injected cells (up to 200 μL), resulting in leakage of most of the cells out of the bone marrow cavity and into the soft tissue of the limb [3, 8, 9, 11]. In our experience, subcutaneous tumors generated from this cell line fail to metastasize within 4 weeks, consistent with another report of a much slower rate of metastasis in subcutaneously implanted tumors .
The metastatic seeding we observed appears to be embolic in nature, because mice euthanized at 1 week had multiple intravascular metastatic cells. Additionally, the mice that died immediately after intratibial injection were found to have large pulmonary emboli consisting of a mass of tumor cells. Because tumor cells are injected directly into the highly vascularized, closed space of the bone marrow cavity, we propose that these cells directly enter the venous system of the hind limb, which drains into the inferior vena cava and into the pulmonary system. Bruland et al.  reported that the presence of osteosarcoma cells in bone marrow aspirates of initially nonmetastatic, localized patients correlates with decreased survival and increased incidence of metastasis. The results of our study support the contention that the bone marrow provides an anatomic structure with its loose, fenestrated vasculature that facilitates efficient metastatic spread to the systemic vasculature and provides insight into the mechanism of osteosarcoma metastatic progression with bone marrow involvement of the primary tumor.
This study has several limitations. Although the data suggest that spontaneous metastases coming from the primary tumor do not contribute to total metastatic burden in a substantial way, our study is not powered to definitively conclude this. However, it is likely that the high percentage of mice with pulmonary tumors was the result of the injection procedure itself and not the usual metastatic cascade of a tumor in the medullary bone cavity. Additionally, three mice died immediately after intratibial injection as a result of embolization of large groups of tumor cells. Although our data do not definitively conclude that all mice experience tumor embolism at the time of injection, our results from amputating the primary tumor 30 minutes after injection suggest that single cells are embolizing at least within that timeframe. Euthanasia of mice immediately after injection and histologic analysis of the lungs for the presence of solitary K7M2 cells would be an alternative way to demonstrate early seeding. Live tracking of tumor cells or detection of circulating tumor cells in the bloodstream would be additional methods for the demonstration of rapid dissemination after intratibial injection.
Another limitation is that the K7M2 tumor cells utilized for this study were generated for their ability to metastasize, which is different than what is observed in spontaneous human tumors. It is possible that use of the parental mouse tumor line, another murine osteosarcoma line, or a rat model that has a larger tibial diameter would have yielded different results. We used the K7M2 cell line because it is commonly used for modeling spontaneous osteosarcoma metastasis.
A substantial portion of the pulmonary metastases that forms after intratibial injection of the K7M2 cell line is the result of embolism of tumor cells at the time of injection. Therefore, it is not an appropriate model for examining the entire metastatic cascade, which includes local invasion, intravasation into the vasculature, survival in the circulation, and finally arrest, extravasation, and growth at the metastatic site . Orthotopic implantation of osteosarcoma tumor fragments could be a more adequate model to study these early metastatic processes with the limitation that many reports state that metastatic development occurs over a longer period of time [4, 6, 8]. Addition of Matrigel® (BD Biosciences, San Jose, CA, USA) to the injected cell suspension, to replace the more commonly used PBS, could prevent the embolism of cells to some degree, and this could be pursued in further studies. Furthermore, the use of a tumor implantation model is inherently limited because it involves the use of a needle to inject tumor cells, exposing the entire tract of tissue that the needle traverses to tumor cells. Genetically engineered models that spontaneously develop tumors in situ would be the most accurate representation of tumor development and metastasis; however, they often require long experimental timeframes.
Despite these limitations, the intratibial injection model offers a reliable method of establishing pulmonary metastases for further study. This model may provide an opportunity to examine the later stages of metastasis such as tumor cell extravasation, survival, and outgrowth in the lung microenvironment. Additionally, intratibial injection offers the ability to concurrently study orthotopic primary tumor engraftment and growth in the bone microenvironment. Because patients with osteosarcoma often present with micrometastatic lesions and then undergo primary tumor removal, we propose that this model provides an opportunity to study the relationship of preexisting pulmonary lesions in the presence of a primary tumor and may be used to examine the efficacy of treatments designed to halt the progression of metastatic outgrowth after surgery to remove the primary tumor. An improved understanding of the models used to study osteosarcoma and their utility in evaluating different aspects of osteosarcoma pathophysiology will provide opportunities to develop more successful therapeutics and improve the survival of patients with osteosarcoma.
1. Abarrategi A, Tornin J, Lucia MC, Hamilton A, Enrique MC, Rodrigo JP, Gonzalez MV, Baldini N, Javier GC, Rodriguez R. Osteosarcoma: cells-of-origin, cancer stem cells, and targeted therapies. Stem Cells Int. 2016;2016:3631764.
2. Bruland ØS, Høifødt H, Sæter G, Smeland S, Fodstad Ø. Hematogenous micrometastases in osteosarcoma patients. Clin Cancer Res. 2005;11:4666–4673.
3. Campbell JP, Merkel AR, Masood-Campbell SK, Elefteriou F, Sterling JA. Models of bone metastasis. J Vis Exp. 2012;67:e4260.
4. Chaffee BK, Allen MJ. A clinically relevant mouse model of canine osteosarcoma with spontaneous metastasis. In Vivo. 2013;27:599–604.
5. Garimella R, Eskew J, Bhamidi P, Vielhauer G, Hong Y, Anderson HC, Tawfik O, Rowe P. Biological characterization of preclinical Bioluminescent Osteosarcoma Orthotopic Mouse (BOOM) model: a multi-modality approach. J Bone Oncol. 2013;2:11–21.
6. Goldstein SD, Hayashi M, Albert CM, Jackson KW, Loeb DM. An orthotopic xenograft model with survival hindlimb amputation allows investigation of the effect of tumor microenvironment on sarcoma metastasis. Clin Exp Metastasis. 2015;32:703–715.
7. Khanna C, Fan TM, Gorlick R, Helman LJ, Kleinerman ES, Adamson PC, Houghton PJ, Tap WD, Welch DR, Steeg PS, Merlino G, Sorensen PH, Meltzer P, Kirsch DG, Janeway KA, Weigel B, Randall L, Withrow SJ, Paoloni M, Kaplan R, Teicher BA, Seibel NL, Smith M, Uren A, Patel SR, Trent J, Savage SA, Mirabello L, Reinke D, Barkaukas DA, Krailo M, Bernstein M. Toward a drug development path that targets metastatic progression in osteosarcoma. Clin Cancer Res. 2014;20:4200–4209.
8. Khanna C, Prehn J, Yeung C, Caylor J, Tsokos M, Helman L. An orthotopic model of murine osteosarcoma with clonally related variants differing in pulmonary metastatic potential. Clin Exp Metastasis. 2000;18:261–271.
9. Luu HH, Kang Q, Park JK, Si W, Luo Q, Jiang W, Yin H, Montag AG, Simon MA, Peabody TD, Haydon RC, Rinker-Schaeffer CW, He T-C. An orthotopic model of human osteosarcoma growth and spontaneous pulmonary metastasis. Clin Exp Metastasis. 2005;22:319–329.
10. Potter M. History of the BALB/c family. In: Potter M, ed. The BALB/c Mouse: Genetics and Immunology. Berlin, Heidelberg, Germany: Springer; 1985:1–5.
11. Sasaki H, Iyer SV, Sasaki K, Tawfik OW, Iwakuma T. An improved intrafemoral injection with minimized leakage as an orthotopic mouse model of osteosarcoma. Anal Biochem. 2015;486:70–74.
12. Sottnik JL, Duval DL, Ehrhart EJ, Thamm DH. An orthotopic, postsurgical model of luciferase transfected murine osteosarcoma with spontaneous metastasis. Clin Exp Metastasis. 2010;27:151–160.
13. Stiller CA, Craft AW, Corazziari I. Survival of children with bone sarcoma in Europe since 1978. Eur J Cancer. 2017;37:760–766.
14. Thudi NK, Martin CK, Nadella MVP, Fernandez SA, Werbeck JL, Pinzone JJ, Rosol TJ. Zoledronic acid decreased osteolysis but not bone metastasis in a nude mouse model of canine prostate cancer with mixed bone lesions. Prostate. 2008;68:1116–1125.
15. Uluçkan Ö, Segaliny A, Botter S, Santiago JM, Mutsaers AJ. Preclinical mouse models of osteosarcoma. Bonekey Rep. 2015;4:1–6.
16. Valastyan S, Weinberg RA. Tumor metastasis: molecular insights and evolving paradigms. Cell. 2011;147:275–292.
© 2018 Lippincott Williams & Wilkins LWW
17. Yuan J, Ossendorf C, Szatkowski JP, Bronk JT, Maran A, Yaszemski M, Bolander ME, Sarkar G, Fuchs B. Osteoblastic and osteolytic human osteosarcomas can be studied with a new xenograft mouse model producing spontaneous metastases. Cancer Invest. 2009;27:435–442.