Thrombin induces osteosarcoma growth, a function inhibited by low molecular weight heparin in vitro and in vivo

Procoagulant nature of osteosarcoma




Procoagulant states, leading to activation of the coagulation protease thrombin, are common in cancer and portend a poor clinical outcome. Although procoagulant states in osteosarcoma patients have been described, studies exploring osteosarcoma cells' ability to directly contribute to procoagulant activity have not been reported. This study explores the hypothesis that osteosarcoma can regulate thrombin generation and proliferate in response to thrombin, and that attenuating thrombin generation with anticoagulants can slow tumor growth.


Pathologic analysis of osteosarcoma with adjacent venous thrombus was performed. In vitro proliferation assays, cell-based coagulant activity assays, and quantification of coagulation cofactor expression were performed on human and murine osteosarcoma cell lines with varying aggressiveness. The efficacy of low molecular weight heparin (LMWH) attenuation of tumor-dependent thrombin generation and growth in vitro and in vivo was determined.


Venous thrombi adjacent to osteosarcoma were found to harbor tumor surrounded by fibrin expressing coagulation cofactors, a finding associated with poor clinical outcome. More aggressive osteosarcoma cell lines had greater surface expression of procoagulant factors and generated more thrombin than less aggressive cell lines and were found to proliferate in response to thrombin. Treatment with LMWH reduced in vitro osteosarcoma proliferation and procoagulant activity as well as tumor growth in vivo.


These findings suggest that elements of the coagulation cascade may play a role in and represent a pharmaceutical target to disrupt osteosarcoma growth. They also have broader implications, as they suggest that, to be effective, dosing of anticoagulants must take into account an individual tumor's capacity to generate thrombin. Cancer 2012. © 2011 American Cancer Society.


Cancer initially develops as localized clones of tumor cells; as these cells progress to a rapid growth phase and migrate from the local microenvironment to establish metastatic tumors, a malignant phenotype develops.1 The malignant phenotype imposes morbidity and mortality as the rapidly growing tumor impedes normal physiologic processes or the tumor itself inflicts physiologic changes resulting in pathologic conditions. An increased tendency to activate the clotting cascade, referred to as a procoagulant state, is common in cancer patients and portends a poor clinical outcome.2 The procoagulant activity, mediated by coagulation proteases such as thrombin, is thought to contribute to the malignant phenotype both directly, by stimulating tumor cell proliferation, and indirectly through the development of tumor-associated thromboemboli.3 The etiology of the procoagulant state in cancer patients is likely multifactorial,2 although evidence suggests that transcriptional changes may contribute to hypercoagulability.3 For example, cancers in multiple target organs have been shown to be intrinsically procoagulant, as evidenced by correlations between the level of tissue factor (TF) expression (cell surface catalyst of local coagulation) and increased tumor growth rates.4-7 In this manner, a cancer cell potentially uses a coagulation-dependent autocrine system by activating coagulation through thrombin, which augments the malignant phenotype.

Hypercoagulability in osteosarcoma patients has been described previously.8 The present study explores the hypothesis that osteosarcoma cells acquire the capacity to directly activate the coagulation cascade, resulting in increased tumor proliferation and tumor thrombus formation. We tested this hypothesis using histological analysis of primary human osteosarcoma specimens, in vitro molecular analysis of osteosarcoma cells, and an in vivo model of murine osteosarcoma. If validated, our hypothesis would suggest that anticoagulant therapy directed at decreasing thrombin in the tumor microenvironment might represent a novel and appropriate strategy for attenuating osteosarcoma tumor growth in vivo.


Collection of clinical data was performed in accordance with the guidelines of the institutional review board (#40194). Animal handling and care protocols were approved by the institutional animal care and use committee.

Clinical Data

Patients were identified who were diagnosed with a primary sarcoma of bone and noted at the time of resection to have a deep venous thrombosis. Imaging studies, pathology reports, operative reports, and intraoperative photos (when available) were obtained. Data collection included patient demographics; tumor size, location, and stage; utilization of neoadjuvant and adjuvant chemotherapy; method of surgical resection (including tumor thrombectomy) and any surgical complications; tumor response to neoadjuvant therapy; presence or absence of edema in the affected limb at surgery; and patient outcomes, including subsequent metastasis and overall survival.


Immunohistochemical stains for human TF (1:50; Santa Cruz Biotechnology, Santa Cruz, Calif), fibrin (1:100, antihuman fibrin neotope on beta-chain; American Diagnostica, Greenwich, Conn), and thrombomodulin (TM; 1:100; Santa Cruz Biotechnology) were performed using a standard avidin-biotin-peroxidase–based protocol (Envision+ HRP/DAB System; Dako North America, Carpinteria, Calif).

Immortalized Osteosarcoma Cell Lines and Culture

Human osteosarcoma lines TE85 and 143B were provided by Dr. H. Luu (Department of Orthopedics, University of Chicago, Chicago, Ill) and maintained at 37°C in 5% CO2 in complete Dulbecco Modified Eagle Medium (DMEM; Invitrogen Life Technologies, Carlsbad, Calif) supplemented with 10% fetal calf serum (FCS; HyClone/Thermoscientific, Novato, Calif) and 1.0 mM sodium pyruvate. Murine osteosarcoma K-lines (K12, K7M2, and K7M3) initially derived from a primary osteosarcoma in a Balb/C mouse9 were provided by Dr. Eugenie Kleinerman (Department of Pediatrics, The University of Texas MD Anderson Cancer Center, Houston, Tex) and maintained as outlined above. EA.hy926 human endothelial-like cells were provided by Dr. Edgell (University of North Carolina, Chapel Hill, NC) and maintained in DMEM supplemented with 10% fetal bovine serum (FBS).10-12 MG-63 cells13 and SaOS2 cells14 were provided by RIKEN BioResource Center (Tsukuba, Japan). MG-63 cells were maintained in DMEM with 10% FBS, and SaOS2 cells were maintained in McCoy5A with 15% FBS. MDAMB23115 and MCF716 were provided by Ann Richmond at Vanderbilt University Cancer Biology (Nashville, Tenn) and maintained in DMEM with 10% heat-inactivated FCS and 2 mmol L-glutamine. All cells were maintained in 5.0% antimycotic solution (Invitrogen Life Technologies).

Measures of Cell Viability and Proliferation


For methyl-tetrazolium bromide mitochondrial activity assay (MTT), 143B and TE85 cells were plated in 96-well plates (7.0 × 103 cells/well), and incubated for 4 hours in 100 μL growth media, after which cells were cultured in serum-free medium for an additional 24 hours. Cells were then exposed to 0, 10, or 20 nM thrombin (α-thrombin, specific activity 4134 U/mg; Hematologic Technologies, Essex Junction, Vt) in serum-free media for 24 hours. Subsequently, 20 μL of MTT reagent (CellTiter 96 AQueous 1 Solution Reagent; Promega, Madison, Wis) was added for 1 to 4 hours. Absorbance was measured at 490 nm with a microplate reader (Biotek, Winooski, Vt).


5-Bromo-2′-deoxyuridine (BrdU) incorporation assays were performed as recommended by the manufacturer (Cell Proliferation ELISA, BrdU kit; Roche, Indianapolis, Ind). 143B and TE85 cells were grown and plated as described for the MTT assay. K7M3 cells were plated at 2.3 × 103 cells/well and incubated for 4 hours in 100 μL growth media, changed to serum-free conditions, and incubated for 18 hours. Absorbance was measured at 450 nm.

Measures of Coagulation Activation

Factor Xa generation, activated protein C (APC) generation, and tumor clot time were performed as 2-stage assays modified from previously described methods.17 Cells (2.5 × 104/well) were incubated for 12 hours at 37°C in a 96-well tissue-culture plate (BD Biosciences, Franklin Lakes, NJ) in DMEM/10% FCS. Culture medium was aspirated, and cells were washed and incubated in serum/phenol red-free medium for 30 minutes.

Factor Xa generation assay

Aliquots of Factor VIIa (Diagnostica Stago, Parsippany, NJ) and Factor X (Diagnostica Stago) were added to the wells and incubated at 37°C for 30 minutes in Hepes buffer with 5 mM CaCl2 (Invitrogen Life Technologies). Conditioned supernatant (50 μL) was removed and placed into a separate 96-well plate with 50 μL of Tris buffer (pH8.2) at 37°C for 3 minutes. Fifty microliters of 2 mM S 2765 substrate (Chromogenix Instrumentation Laboratory, Milan, Italy) was added, and absorbance was monitored at 405 nm for 10 minutes.

APC generation assay

Three hundred nanomolar protein-C and 4 nM thrombin (Diagnostica Stago) in Hepes buffer containing 5mM CaCl2 at 37′C was added to cell cultures for 60 minutes. Conditioned supernatant (50 μL) was removed and incubated at 37°C for 5 minutes in a separate 96-well plate with 50 μL of Tris buffer containing 2.5 μL hirudin (American Diagnostica). Fifty microliters of 0.5 mM S 2366 substrate (Chromogenix Instrumentation Laboratory) was added, and absorbance was monitored at 405 nm for 10 minutes.

Tumor clot times

Supernatant from cells exposed to Factor VIIa/X (50 μL) was added to 50 μL of activated partial thromboplastin time reagent (Diagnostica Stago) and 90 μL of Factor VIII-deficient plasma (Hematologic Technologies) in a cuvette, preincubated for 120 seconds, and then initiated with 90 μL of 25mM CaCl2. Times were determined with an ST-4 coagulometer (Diagnostica Stago).18

Osteosarcoma-Initiated Thrombin Generation Assay

The cell lines EA.hy926, K12, K7M2, K7M3 MG63, SaOS2, TE85, 143B MCF7, and MDAMB231 (2.5 × 104/well) were incubated on a clear-bottom, black 96-well plate in DMEM for 6 hours at 37°C, then switched into serum/phenol red-free medium containing 1% bovine serum albumin (BSA; Sigma Aldrich, St. Louis, Mo) for 12 hours. Medium was removed, and cells were washed with 100 μL of Ca2+-free phosphate-buffered saline (PBS) at 37°C for 10 minutes. After washing, Factor VIII-deficient plasma (80 μL), Ca2+-free PBS (12.5 μL), thrombin substrate (Z-Gly-Gly-Arg-AMC•HCI; Bachem, Torrance, Calif; 12.5 μL, final concentration 4.2 mM), and 164mM CaCl2 (10 μL) were added for a total volume of 115 μL/well. Fluorescence was monitored at 360 nM (excitation) and 460 nM (emission) at 30-second intervals for 50 minutes at 37°C. Each assay was performed in triplicate. Data were analyzed using algorithms previously developed,19, 20 correcting for inner-filter effect, hemolysis, lipidema, and residual thrombin-α2–macroglobulin complex formation. This algorithm converts the relative fluorescent units to concentration of thrombin produced.

Quantitative Real Time Polymerase Chain Reaction

Total RNA was isolated from cell lines using RNeasy (Qiagen, Valencia, Calif). Single-strand cDNA was synthesized with SuperScript III reverse transcriptase (Invitrogen Life Technologies) and oligo-dT primers. Real-time reverse-transcriptase polymerase chain reaction (PCR) was performed on the cDNA using iQ SYBR Green supermix (Bio-Rad Laboratories, Hercules, Calif) with the StepOnePlus Real-Time PCR System (Applied Biosystems Life Technologies, Carlsbad, Calif). Primers for human TF (sense,5′-CGGGTGCAGGCATTCCAGAG-3′; antisense,5′-CTCCGTGGGACAGAGAGGAC-3′) and β-actin (sense,5′-ACCCAGATCATGTTTGAGAC-3′; antisense,5′-GTCAGGATCTTCATGAGGTAGT-3′) were synthesized by Integrated DNA Technologies (Coralville, Iowa). The mRNA expression levels were normalized to the mean expression of β-actin.

Measurement of Cell Surface TF, TM Expression

Cells grown in culture were resuspended in PBS containing 1% BSA with antibodies for human and mouse TF, TM, or Phycoerythrin (PE)-isotype (American Diagnostica) for 30 minutes at 4°C. Cells were washed twice, centrifuged at 1500 rpm for 5 minutes in a ST-16R centrifuge (Thermo Fisher Scientific, Waltham, Mass), and incubated for 30 minutes with goat antirabbit immunoglobulin G conjugated with AlexaFluor647 (Invitrogen Life Technologies). Cells were analyzed using FACSC and Diva Software (BD Biosciences).

In Vivo Assessment of Tumor Growth Treated With LMWH

Tumor inoculation

Single-cell suspensions (1 × 105) of murine osteosarcoma (K7M3) cells in 10 μL PBS were injected into the left tibias of a total of 20 (n = 10 control; n = 10 LMWH-treated) 6-week-old BALB/c mice (Jackson Laboratories, Bar Harbor, Me) as previously described.21 Mice were sacrificed after 4 weeks by CO2 inhalation.

Anticoagulant therapy

To assure that dosing of low molecular weight heparin (LMWH; Enoxaparin-Lovenox; Sanofi-Aventis, Bridgewater, NJ) in mice was relevant to levels used in humans, 3 doses of LMWH (5, 10, and 20 mg/kg in saline) were injected subcutaneously twice daily. Control mice were injected with an equivalent volume of saline to assure the volumes of LMWH injected did not cause anticoagulation by hemodilution. Mice were sacrificed 3 hours after the third injection, and plasma was collected in tubes containing 0.109 M sodium citrate. Platelet-free plasma was processed by microcentrifugation at 1500 × g for 15 minutes and centrifugation of supernatant at 13,000 × g for 15 minutes. Anticoagulation was measured with thrombin-generation assay described above, substituting 4 μM phospholipid and 5 pM TF in place of cells for assessment of the rate of coagulation initiation by the mouse plasma and substituting Factor VIII-deficient plasma with platelet-free mouse plasma. Ten micrograms per kilogram LMWH twice daily provided the same level of inhibition of thrombin (53%) as measured by peak height (nM thrombin) as in humans treated with a prophylactic dosage of LMWH (4 μg/mL).

Analysis of tumor growth

Mice were observed daily for signs of tumor growth, including soft-tissue swelling or limping. Weekly radiographic evaluations were performed using a LX-60 Radiography System (Faxitron X Ray, Lincolnshire, Ill) (8 seconds at 35 kV). Assessment for enlargement of the soft tissues surrounding the injected tibia was conducted by analysis of radiographs using Image J (version 1.44u, National Institutes of Health, Bethesda, Md) as previously described.22 Tumor growth was calculated weekly as the percentage increase in size of the injected tibia compared with the paired contralateral control. In addition, magnetic resonance imaging (MRI-9.4T Inova imaging system; Varian, Palo Alto, Calif)23 and histological examination confirmed tumor size (mice that did not develop histological evidence of tumor were excluded from the study). Extent of osteolysis was determined with microcomputed x-ray tomography (μCT40; Scanco Medical AG, Bruttisellen, Switzerland) of the trabecular bone volume within the tibial metaphysis via contiguous cross sections of the metaphyseal region (70 kV, 114 μA, 300 milliseconds integration, 500 projections per 180° rotation, with a 12 μm isotropic voxel size).


One-way analysis of variance with Bonferroni multiple 2-sided comparison test and a 2-sided Student t test were used to analyze in vitro and in vivo animal model experiments. All statistical calculations were determined using Prism version 4.0 (Graph Pad Software, La Jolla, Calif). Type I error probability (α) was considered to be .05.

Sample Size

In a previous study,22 x-ray analysis of the soft tissue masses (at 4 weeks) averaged an intraobserver error (n = 3) of ±6.64% and interobserver error of ±15.84% (n = 3), with a normal distribution and standard deviation of 23%. Five mice per group was determined to detect a 25% reduction in tumor size, with a statistical power (β) of .8 and α = .05 (PS Power and Sample Size Calculations software24). To account for variability of the orthotopic tumor model and early mortality, we used 10 mice/group.


A Surgically Excised Osteosarcoma Expresses Procoagulation and Anticoagulation Components

Clinical findings from a patient with osteosarcoma were responsible for the hypothesis that local activation of the coagulation cascade by osteosarcoma cells leads to thrombin production, elevating the potential for both accelerated tumor growth and thrombus formation. Specifically, a preoperative computed tomography scan of a pelvic osteosarcoma involving the outer table of the ilium revealed a thrombus in the common iliac vein (Fig. 1A). Radiographic and intraoperative exploration disclosed no direct anatomical association between the primary osteosarcoma and the tumor thrombus within the iliac vein. Histopathologic examination indicated that the primary tumor was composed of pleomorphic cells producing osteoid, diagnostic of high-grade osteosarcoma (Fig. 1B). In addition, sections from the resected thrombus showed neoplastic cells morphologically similar to the primary tumor (Fig. 1C), consistent with a macroscopic intravascular tumor thrombus. Immunohistochemical stains for TF (a procoagulant cofactor) and TM (an anticoagulant cofactor) were positive for the primary tumor and intravascular tumor thrombus. Further immunohistochemical analysis revealed that nests of osteosarcoma were often embedded in a matrix of fibrin (cleaved from fibrinogen by thrombin). The existence of an occlusive thrombus harboring tumor adjacent to an osteosarcoma has been observed in 7 additional patients (Table 1). The findings of a tumor-laden thrombus were associated with a poor clinical outcome, as 75% of these patients were deceased within 3 years of diagnosis (average time from diagnosis to death, 13 ± 12 months.)

Figure 1.

Osteosarcoma expression of coagulation cofactors and receptors in vivo is shown. Radiographic/macroscopic (A) and histologic (B, C) evaluation of a pelvic high-grade osteosarcoma (*) and associated intravascular tumor thrombus (arrow) involving the internal iliac vein is shown. The patient was treated with neoadjuvant chemotherapy followed by right internal hemipelvectomy and resection of the iliac vein thrombus (A; macroscopic). Histopathology of the tumor (B) showed residual viable osteosarcoma (hematoxylin & eosin [H&E]) staining positive for tissue factor (TF; CD142), thrombomodulin (TM; CD141), and fibrin. The iliac vein thrombus (C) also demonstrated viable tumor cells within the organized fibrous clot (H&E), as well as TF, thrombomodulin, and fibrin deposition. These findings suggest that tumor-associated thrombi may harbor osteosarcoma cells expressing factors that regulate coagulation. CT, computed tomography; DVT, deep vein thrombosis.

Table 1. Review of Patients Diagnosed With Osteosarcoma and a Tumor-Laden Venous Thrombus
PatientSexAge (yrs)DiagnosisTumor Location (Size)Enneking StageVessel InvolvedOperative Treatment of TumorTreatment of Tumor ThrombusMarginC/RMetsOutcome
  1. Abbreviations: A, adjuvant; C/R, chemotherapy/radiation; F, female; M, male; Mets, metastases; Micro, microscopically positive; N, neoadjuvant; NA, not applicable; Neg, negative; post-op, postoperative; pre-op, preoperative.

1M29Osteogenic sarcomaPelvis (12 × 11 × 15 cm)IIICommon iliac veinInternal hemipelvectomy/partial corpectomy (L5 & S1Common femoral vein thrombectomy & repairMicroN, post-opLung pre-opDeceased 6 months after tumor excision
2F26Pleomorphic sarcoma of bonePelvis (14 × 8 × 15 cm)IIIInternal iliac veinHindquarter amputation/partial internal hemipelvectomyResection of internal iliac veinMicroNLung pre-opDeceased 4 months after tumor excision
3F13Osteogenic sarcomaPelvis (12 × 9 × 8 cm)IIBCommon and internal iliac veinsInternal hemipelvectomyCommon femoral vein thrombectomyNegN, ANoneAlive 5 years post-op
4M13Osteogenic sarcomaProximal femur (18 × 9 × 8 cm)IIBMedial circumflex femoral veinRadical resection/osteoarticular allograftResection of medial femoral circumflex veinNegN, ALung pre-opDeceased 36 months after tumor excision
5M12Osteogenic sarcomaFemoral diaphysis (15 × 5 × 4 cm)IIBRetrograde vein superficial to Hunter canalRadical resection/ intercalary allograftResection of retrograde vein at Hunter canalNegN, ANoneAlive 22 months post-op
6M19Osteogenic sarcomaPelvis (10 × 8 × 8 cm)IIBCommon and internal iliac veinsNoneNoneNANNoneDeceased 6 months after diagnosis
7F58Osteogenic sarcomaMedial clavicle (10 × 6 × 5 cm)IIBInnominate veinRadical resection/ clavicle & chest wallInnominate vein thrombectomy & repairMicroANoneDeceased 18 months after diagnosis
8M59Pleiomorphic sarcoma of boneMedial clavicle (12 × 10 × 8 cm)IIBSubclavian veinRadical resection clavicle & chest wallResection of subclavian veinNegNoneLung pre-opDeceased 8 months after tumor excised

Thrombin Enhances Osteosarcoma Growth In Vitro

The effect of thrombin in the tumor microenvironment was examined by its ability to stimulate cell proliferation. It was determined that thrombin induces proliferation in human osteosarcoma cells with high (143B) and low metastatic potential (TE85) in vitro in a concentration-dependent manner as determined by BrdU incorporation (Fig. 2A) and mitochondrial enzyme (MTT) assays (Fig. 2B). This demonstrates that thrombin in the tumor microenvironment may potentiate the malignant phenotype.

Figure 2.

Thrombin enhances human osteosarcoma proliferation in a concentration-dependent fashion. TE-85 and 143B cell-line proliferation was measured with 5-bromo-2′-deoxyuridine (BrdU) proliferation assay (A) and mitochondrial activity assay (MTT) (B) with thrombin concentrations of 0 nM, 10 nM, and 20 nM for 24 hours. One-way, 2-sided analysis of variance test with Bonferroni comparison test showed significant difference between 20 nM thrombin and control (0 nM thrombin; *P < .05, n = 6)

Capacity of Human and Murine Osteosarcoma Cell Lines to Generate Thrombin Correlates With Malignant Potential

The association between the capacity of tumor cells to generate thrombin and their malignant phenotype in vivo was examined. Malignant human osteosarcoma cells (143B) were compared with parental TE85 cells in vitro (Fig. 3). The 143B cell line has been characterized as more aggressive, being highly tumorigenic and possessing a high metastatic potential when injected into the proximal tibia of athymic mice, than TE85 cells that are reported to be neither tumorigenic nor metastatic.25 A modified thrombin generation assay was used to monitor cell-based contributions to Factor Xa-dependent cleavage of prothrombin to thrombin. The contribution of the intrinsic coagulation pathway was eliminated by conducting the cellular incubations with plasma deficient in Factor VIIIa. As shown in Figure 3A, the more malignant 143B cells mediate a greater degree of thrombin generation than the less aggressive cell line, TE85. By adding an anti-TF antibody to the thrombin generation assay, thrombin generation is dramatically inhibited, indicating the role of TF for activation of the coagulation cascade by osteosarcoma cells (Fig. 3A). This was consistent with the interpretation that increased aggressive properties of osteosarcoma cells, in terms of malignant potential, are associated with a greater ability to support the extrinsic coagulation cascade.

Figure 3.

The extent of tissue factor (TF) expression and thrombin generation correlates with the malignant potential of osteosarcoma cell lines. (A) The more aggressive osteosarcoma cell line (143B) generates more thrombin (P < .05, n = 6) than less clinically aggressive TE85 osteosarcoma cells. Addition of anti-TF antibody significantly inhibits thrombin generation of both cell lines (P < .05 peak height isotype control compared with anti-TF antibody, n = 6), indicating that osteosarcoma-initiated thrombin generation is TF-dependent. 143B cells express roughly 2-fold greater amounts of TF than TE85 cells as measured by fluorescent-activated cell sorting analysis (B, *P < .05) and real time polymerase chain reaction (C, *P < .05). A 2-sided Student t test was performed. MFI, mean fluorescent intensity.

Therefore it was postulated that the enhanced ability of 143B cell line to generate thrombin was because of increased cell surface expression or bioavailability of TF. Consistent with this hypothesis, flow cytometry revealed a >2-fold increase in cell-surface expression of TF in 143B cells compared with TE85 cells (Fig. 3B). Assessment of the area under the curve normalized to cell number (ie, fluorescent-activated cell sorting mean fluorescent intensity) allows for comparison. In addition, significantly more mRNA encoding TF is detected in 143B cells compared with the TE85 cells (Fig. 3C). This suggests that differential expression of TF in these cell lines is because of changes in TF gene transcription. This further supports the hypothesis that osteosarcoma may use thrombin in an autocrine fashion.

Tumor-Mediated Thrombin Generation in Human and Murine Osteosarcoma Relative to Their Malignant Phenotype, Endothelial-Like Cells and Human Breast Cancer

To determine whether the results from the thrombin generation assay in the human osteosarcoma cell line were merely a characteristic of these specific cells, the capacity to initiate thrombin generation was determined in other well-characterized cell lines. The cellular thrombin generation assay provides a sensitive method to detect changes in the rate of cell-mediated thrombin generation, as seen by a comparison of unstimulated human endothelial-like cells (EA.hy926) with those stimulated with tumor necrosis factor α (Fig. 4A). Figure 4B reveals that in addition to the human osteosarcoma cell lines TE85 and 143B, the capacity of cellular-mediated thrombin generation correlates with malignant potential in the murine osteosarcoma K-line. To confirm that the correlation of malignant potential and thrombin generation in human osteosarcoma is not unique to the 143B cell line, we compared MG63 to SaOS2, TE85, and 143B human osteosarcoma cell lines (Fig. 4C). The SaOS2, TE85, and 143B cell lines, exhibiting more malignant potential than MG63, all showed an increase in thrombin generation compared with MG63. To extend the correlation between malignant potential and thrombin generation in tumor cells other than osteosarcoma, the highly invasive human breast cancer cell line MDAMB231 was compared with the less invasive MCF7 cell line26, 27 and also found to have a greater capacity to initiate thrombin generation. These results suggest that there is an association between thrombin generation and malignant potential in tumorigenic cells.

Figure 4.

Tumor-mediated thrombin generation is shown in human and murine osteosarcoma relative to their malignant phenotype, endothelial-like cells, and human breast cancer. (A) Endothelial-like cells, EA.hy926 (2.5 × 104/well) cells stimulated with 10 ng/mL tumor necrosis factor α (TNFα) for 5 hours, enhance their ability by 5.5-fold to generate thrombin (*P < .05 Student t test). (B) The capacity to initiate thrombin generation (2.5 × 104 cells/well) correlates with malignant potential in the murine osteosarcoma K-line with a 3-fold (K7M2) and 6.5-fold (K7M3) increase compared with the parental line (K12; *P < .05 compared with K12 by 1-way analysis of variance [ANOVA] Bonferroni test). (C) Human osteosarcoma cell lines SaOS2 (30% greater proliferation kinetics than MG6335) and TE85 (10% increase in doubling time compared with MG6336) had a 3-fold increase in thrombin generation compared with the less malignant MG63 cell line (derived from an osteogenic sarcoma with low malignant potential10), whereas the highly malignant 143B cell line had a 10-fold increase compared with the MG63 cell line (*P < .05 compared with K12 by 1-way ANOVA Bonferroni test). (D) The highly invasive human breast cancer cell line MDAMB231 compared with the less invasive MCF7 cell line26, 27 had a 23-fold increase in the rate of thrombin generation (*P < .05, Student t test).

As local thrombin generation depends on cellular production of procoagulant and anticoagulant cofactors, human and murine osteosarcoma cells were assayed for both expression and function of TF (procoagulant) and TM (anticoagulant). Table 2 compares the surface expression of TF and TM and coagulation properties of human and murine osteosarcoma cell lines. Findings from the murine osteosarcoma K-cell lines are consistent with the hypothesis that increased TF expression and decreased TM is associated with increased malignant potential. The human osteosarcoma cells, however, revealed increased expression of both TF and TM. Because these independent molecular parameters fail to fully characterize the functional consequences of coexpression of procoagulant and anticoagulant activities at the tumor cell surface, thrombin generation and tumor clot time were measured as a functional evaluation of the cells' endogenous coagulation potential. The data reveal that the capacity to activate prothrombin and generate a fibrin clot is associated with increased metastatic potential in both human and murine osteosarcoma cell lines, regardless of the relative expression and functional activity of the anticoagulant cofactor TM.

Table 2. Comparison of Malignant and Less Malignant Osteosarcoma Cell Lines
SpeciesCellMalignant PotentialTissue Factor Expression RatioThrombomodulin Expression RatioFactor × ActivationActivation of Protein CThrombin GenerationClot Time
  • a

    Denotes P < .05 as compared to the less metastatic osteosarcoma as measured by a 1-way, 2-sided analysis of variance with Bonferroni comparison test.

HumanTE85Low capacity to form tumor or metastasize251.00 ± 0.141.00 ± 0.2410.47 ± 1.261.67 ± 0.03979.60 ± 1.2318.87 ± 0.99
143BHighly tumorigenic and metastatic252.10 ± 0.28a7.50 ± 0.78a24.20 ± 2.38a6.02 ± 0.83a2050.00 ± 1.89a13.06 ± 0.66a
MurineK12Metastasis to lung seen in 33% of mice91.00 ± 0.041.00 ± 0.140.92 ± 0.0265.091 ± 2.3072.53 ± 0.8045.58 ± 2.08
K7M2Metastasis to lung seen in 90% of mice91.72 ± 0.07a0.12 ± 0.02a3.26 ± 0.36a30.88 ± 1.89a165.23 ± 5.26a25.68 ± 0.88a
K7M3Metastasis to lung seen in >90% of mice92.05 ± 0.15a0.08 ± 0.02a5.29 ± 0.15a48.21 ± 1.12a236.70 ± 14.41a22.85 ± 1.57a

Anticoagulation With LWMH Attenuates Osteosarcoma Growth In Vitro and In Vivo

Because thrombin supports osteosarcoma growth in vitro, and more aggressive osteosarcomas have greater intrinsic thrombin generation potential, it was postulated that inhibition of thrombin activity would suppress osteosarcoma growth in vitro and in vivo. LMWH (2.5-10 mg/mL, introduced at the initiation of the assay) that brackets the prophylactic plasma levels for LMWH in humans (4 μg/mL) significantly attenuated the capacity of the K7M3 osteosarcoma cells to initiate thrombin generation in a concentration-dependent fashion (Fig. 5A). In addition, LMWH (10-minute incubation with thrombin followed by 3-hour incubation with osteosarcoma cells) also significantly inhibited thrombin-induced proliferation of murine (Fig. 5A and Table 3) and human (Table 3) osteosarcoma cells.

Figure 5.

Low molecular weight heparin (LMWH) attenuates osteosarcoma-initiated thrombin generation in parallel with its capacity to inhibit osteosarcoma growth in vitro and in vivo. (A) Clinically relevant doses of LMWH significantly decrease thrombin generation in murine osteosarcoma K7M3 cells (solid line: peak height of thrombin generation) in a dose-dependent fashion (*P < .05 compared with no treatment). LMWH significantly inhibits thrombin-stimulated osteosarcoma proliferation (dashed line: 5-bromo-2′-deoxyuridine [BrdU] incorporation). (B) Compared with plasma collected from saline-treated mice, plasma from LMWH-treated mice demonstrates decreased systemic thrombin generation (*P < .05), roughly equivalent to humans treated with twice daily prophylactic dosing of LMWH (data not shown). (C) Longitudinal analysis reveals tumor size is significantly smaller in LMWH-treated mice (n = 6) 3 and 4 weeks postinoculation compared with saline-treated mice (n = 6) as measured by x-ray analysis of tumor invasion into soft tissue (*P < .05). (D) Bone fraction volume of the tibial metaphyseal injection site showed a decrease in tumor-induced osteolysis in the LMWH-treated mice compared with saline mice (P < .05; 1-way, 2-sided analysis of variance with a Bonferroni comparison test).

Table 3. LMWH Attenuates Thrombin-Enhanced Osteosarcoma Cell Proliferation
Cell LineLMWH, μg/mLCell Proliferation (BrdU Absorbance 450 nm)Pa
  • Abbreviations: BrdU, 5-bromo-2′-deoxyuridine; LMWH, low molecular weight heparin.

  • a

    Compared to 20 nm thrombin without LMWH (1-way analysis of variance Bonferroni multiple comparison test).

143B04.90 ± 0.18
53.07 ± 0.29.001
103.17 ± 0.27.001
K7M304.08 ± 0.06
53.10 ± 0.13.001
102.8 ± 0.20.001

To translate to an in vivo model, the dosing strategy of LMWH that provided continuous in vivo inhibition of thrombin was needed. Preliminary studies established that subcutaneous injection of 10 mg/kg LMWH produced a plasma concentration functionally equivalent to the human prophylactic concentration (4 μg/mL) and required dosing twice daily to maintain continuous thrombin inhibition as measured by a plasma-based thrombin generation assay. Mice orthotropically inoculated with K7M3 osteosarcoma cells and treated with LMWH remained anticoagulated at the time of sacrifice, as evidenced by decreased thrombin generation compared with mice injected with saline vehicle control (Fig. 5B).

Consistent with in vitro findings, analyses revealed that tumor growth in mice inoculated with intraosseous K7M3 osteosarcoma cells was attenuated with LMWH treatment compared with mice treated with saline alone (control). Although there was no difference in the number of mice that developed histologic evidence of tumor at the time of sacrifice (6 of 10 mice in each group), longitudinal analysis showed that LMWH-treated mice had significantly reduced tumor expansion into the soft tissue compared with saline-injected mice 3 weeks (110.6% ± 21.4% vs 199.5% ± 67.8%) and 4 weeks (135.1% ± 44.7% vs 227.4% ± 76.3%) postinoculation (Figs. 5C and 6). Intramedullary lytic lesions, as determined by bone fractional volume from microcomputed x-ray tomography, were also significantly smaller in LMWH-treated mice (Fig. 5D; 1.02% ± 0.3% vs 5.44% ± 1.81%). Similar findings were demonstrated by MRI (0.327 ± 0.035 U vs 0.268 ± 0.015 U, n = 1).

Figure 6.

Attenuated osteosarcoma growth in mice treated with low molecular weight heparin (LMWH) anticoagulant therapy. Representative radiographic (x-ray and microcomputed x-ray tomography [μCT]), magnetic resonance imaging (MRI), and histologic (hematoxylin & eosin [H&E]) images are shown from animals treated subcutaneously with twice daily saline (A) or 10 mg/kg LMWH (B) after intratibial inoculation of murine osteosarcoma K7M3 cells; images from mice not inoculated with tumor (C) are shown for comparison. X-ray, histology, and MRI revealed that invasion of tumor into surrounding soft tissue (yellow line and *) in LMWH-treated mice (B) was significantly less than in the saline-treated mice (A). Representative microcomputed x-ray tomography axial images (yellow line indicates area for bone fractional volume measurement) and 3-dimensional reconstructions (*indicates lytic lesion) of the inoculated tibias demonstrate significantly fewer and less extensive osteolytic lesions in LMWH-treated mice (B) compared with saline-injected mice (A).

In summary, analysis of representative sections by x-ray, MRI, microcomputed x-ray tomography, and histopathologic examination revealed a significant reduction in osteosarcoma growth in mice treated with LMWH in a dosing strategy based on thrombin inhibition in LMWH-treated mice (Fig. 6).


Osteosarcoma is the most common primary malignant bone tumor in children and adolescents.28 Despite significant improvements in clinical outcome because of chemotherapy and surgery, patients with metastatic or locally recurrent disease continue to have a poor prognosis. Our findings suggest that osteosarcoma may activate coagulation via cell surface expression of procoagulant factors. The presence of thrombin in the tumor microenvironment may directly stimulate cell proliferation. Consequently, we postulated, and our in vivo data confirm, that blocking local thrombin generation or its effector mechanisms can attenuate disease progression.

Multiple studies of nonsarcomatous tumor growth in animal models demonstrate that thrombin is an essential component of a malignant phenotype. Thrombin is thought to augment a malignant phenotype through direct cellular mechanisms and indirect modulation of the tumor microenvironment through formation of a thrombus (Fig. 7). For example, mice partially deficient in prothrombin have significantly diminished in vivo metastasis of Lewis lung carcinoma.29 Many studies have demonstrated, through PAR-1 and other thrombin effector mechanisms, that thrombin directly promotes tumor cell proliferation, adhesion to platelets, production of endothelial cells, matrix proteins, and vascular endothelial growth factor A (VEGF-A) to support angiogenesis.3 Fibrin appears to be important in this process, as fibrinogen-deficient mice also have diminished in vivo metastasis.30 Fibrin is thought to enhance tumor adhesion to platelets and endothelial cells and/or to provide a matrix to support tumor growth. Although the presence of a macroscopic occlusive tumor-thrombus in our case examples (Fig. 1 and Table 1) is evidence of potential tumor-mediated hypercoagulability, these patients represent the extreme presentation of tumor-laden thrombi and that tumor-mediated hypercoagulability would more commonly lead to a nonocclusive thrombus. Regardless of the mechanism, these results highlight that thrombin may enhance the malignant phenotype through direct and indirect mechanisms; therefore, we postulate that a tumor-laden thrombus represents a means of cancer progression via microemboli.

Figure 7.

Potential mechanisms of enhanced tumor growth as a result of tumor-mediated hypercoagulability are shown. Thrombin, activated through tissue factor (TF) and inhibited through thrombomodulin (TM; A), may contribute to malignant transformation through direct cellular and indirect extracellular mechanisms. Thrombin may directly activate thrombin receptors on the cell surface, leading to cellular proliferation (B), production and release of vascular endothelial growth factor (VEGF; C),3 and enhanced adhesion to matrices, platelets (PLT), or cells by up-regulation of integrins (D).3 Indirectly, thrombin cleaves fibrinogen to form a fibrin clot, which can represent an ideal matrix for tumor growth and enhance tumor adhesion to cells and platelets (E). Low molecular weight heparin (LMWH) may attenuate these events through multiple mechanisms, including 1) by directly inhibiting thrombin or other coagulation proteins by catalyzing antithrombin,3 2) by indirectly inhibiting thrombin formation by stimulating the release of tissue factor pathway inhibitor, an effective inhibitor of TF-mediated coagulation,3 and 3) by inhibiting VEGF-A.3

This study provides the first evidence that osteosarcomas are capable of supporting thrombin generation resulting in direct activation of cellular proliferation and thrombus formation in the microenvironment. We hypothesize that a therapeutic approach directed at inhibiting thrombin would attenuate the malignant phenotype. LMWH, a therapeutic anticoagulant noted for its safety and efficacy in humans, inhibits Factor Xa and thrombin by catalyzing their interactions with antithrombin, as well as inducing systemic release of tissue factor pathway inhibitor from the endothelium.31 LMWH may also affect tumor growth by enhancing or sequestering growth factors. For example, at low doses, LMWH has been found to enhance cellular interactions of VEGF-A, but at high doses promotes inhibition31 (Fig. 7). The efficacy of LMWH as an antitumor agent has been variable. It has been shown to attenuate in vivo growth in several cancers in early,32, 33 but not advanced disease.21 We show that LMWH significantly reduced tumor cell proliferation and thrombin generation in vitro and in vivo. Although we were able to attenuate osteosarcoma growth with LMWH, the treated mice still developed tumor. Our in vitro data indicate that thrombin's capacity to induce proliferation is dose-dependent, as is LMWH's capacity to inhibit thrombin and osteosarcoma proliferation. Our plasma-based thrombin generation reveals that, although anticoagulated, LMWH-treated mice still generate a considerable amount of thrombin. These results may explain the variability of the effectiveness of LMWH in preserving bone mineral density. Thus, an investigation into altering the dosing strategy to inhibit a greater amount of thrombin is needed to determine whether tumor growth can be further attenuated. Also, additional studies are needed to evaluate the ability of LMWH to suppress tumor metastasis in osteosarcoma because of the limitations of our experimental model (detectable pulmonary metastasis does not develop until 6 weeks after inoculation, whereas lameness and signs of morbidity require euthanasia or amputation 3-4 weeks after inoculation34). Nevertheless, the results suggest that a valid hypothesis as to the variable results using LMWH in early versus late stage cancers is that late stage cancers have the capacity to generate a significantly greater amount of thrombin, and therefore would require significantly more anticoagulation. This hypothesis is supported by our cellular thrombin generation assay results comparing low versus high aggressive cancers: osteosarcoma and breast. This implies that the use of anticoagulation as an adjuvant therapy requires dosing strategies specific to a tumor's capacity to generate thrombin.

Our findings suggest that osteosarcoma cells activate coagulation and use the resultant coagulant activity in an autocrine mechanism to promote growth. Thrombin generation induced by the tumor cell can contribute to morbidity by causing thrombosis. Considering that LMWH and other anticoagulants currently used for non-neoplastic indications have been tested for safety and efficacy in humans, the clinical impact of this work provides a rationale for cost-effective and rapid new adjunctive strategies to treat osteosarcoma. Future clinical investigations exploring the effectiveness of anticoagulant and antithrombin receptor therapies as adjuvants for current treatment protocols for osteosarcoma will be required to determine whether the findings in this report translate into clinical practice.


This project was funded by the Brooks Foundation, Caitlin Lovejoy Foundation, and departmental funds.


The authors made no disclosures.