SEARCH

SEARCH BY CITATION

Keywords:

  • osteosarcoma;
  • CXC chemokines;
  • biomarkers;
  • antibody microarray;
  • cell proliferation

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

BACKGROUND:

Osteosarcoma is the most common malignant bone tumor in children. Despite the advent of chemotherapy, the survival of osteosarcoma patients has not been significantly improved recently. Chemokines are a group of signaling molecules that have been implicated in tumorigenesis and metastasis.

METHODS:

The authors used an antibody microarray to identify chemokines that were elevated in the plasma samples of osteosarcoma patients. The results were validated using enzyme-linked immunosorbent assays on an independent set of samples. The tumor expressions of 3 chemokines were examined in 2 sets of osteosarcoma tissue arrays. The authors also evaluated the proliferative effect of the chemokines in 4 osteosarcoma cell lines.

RESULTS:

The authors found that the plasma levels of CXCL4, CXCL6, and CXCL12 in the osteosarcoma patients were significantly higher than those in the controls, and the results were validated by an independent osteosarcoma cohort (P < .05). However, CXCL4 (100%) and CXCL6 (91%) were frequently expressed in osteosarcoma, whereas CXCL12 was only expressed in 4%. Survival analysis further showed that higher circulating levels of CXCL4 and CXCL6, but not CXCL12, were associated with a poorer outcome of osteosarcoma patients. Addition of exogenous chemokines significantly promoted the growth of different osteosarcoma cells (P < .05).

CONCLUSIONS:

The results demonstrate that CXCL4 and CXCL6 are frequently expressed in osteosarcoma, and that the plasma levels of these 2 chemokines are associated with patient outcomes. Further study of these circulating chemokines may provide a promising approach for prognostication of osteosarcoma. Targeting these chemokines or their receptors may also lead to a novel therapeutic invention. Cancer 2011. © 2010 American Cancer Society.

Osteosarcoma (OS) is a primary malignant tumor of bone arising from primitive bone-forming mesenchymal cells.1 OS is the most common malignant bone tumor in children and adolescents, with an annual incidence rate of 5.6 per million, and accounts for 60% of pediatric bone tumors.2 OS is a devastating disease, characterized by a high local aggressiveness and tendency to metastasize to the lungs and distant bones. Despite recent advances in multimodality treatments consisting of adjuvant chemotherapy and surgical resection, pulmonary metastasis occurs in approximately 40% to 50% of patients. In such cases, the overall 5-year survival rate is only 28%.3, 4 Previous studies have found specific populations that are at higher risks of developing OS, such as patients harboring certain gene mutations (TP53,5RB,6 or RECQL47). Identification of tumor-derived factors that are associated with OS will likely improve the diagnosis and treatment of this deadly disease.

Cytokines are secreted or membrane-bound proteins that are released in response to a diverse range of cellular stresses, including inflammation and malignancy. These proteins regulate the growth, differentiation, and activation of immune cells in response to specific cellular stimuli. Chemokines are a family of cytokine-like proteins that selectively attract and activate different cell types, including immune cells. By interacting with G-protein–coupled receptors, chemokines can affect various cellular processes, such as cytoskeleton rearrangement, directional cell migration, and cell adhesion.8, 9 Some chemokines can promote proliferation, angiogenesis, metastasis, or immunosuppression in cancer, whereas other chemokines inhibit tumor-mediated angiogenesis and promote antitumor immune responses.8-11 CXC chemokines contain 2 highly conserved cysteine residues separated by a nonconserved amino-acid residue (cysteine-X-cysteine sequence) at the N-terminus.12 The CXC chemokines are small proteins with a functional ability of activating and directing chemotaxis of different cells including leukocytes.13 In OS, addition of a CXCR4 inhibitor or antagonist can significantly inhibit the development of lung metastasis in mouse models.14, 15 In this study, we tested if other cytokines and/or chemokines were also associated with OS.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Patients and Samples

The sample and patient information used in this study is listed in Tables 1 and 2. The human plasma samples were collected at initial diagnosis from 33 OS patients who were enrolled from 3 collaborating institutions, namely Texas Children's Hospital (Houston, Tex), Cook Children's Medical Center (Fort Worth, Tex), and the University of Oklahoma Health Sciences Center (Oklahoma City, Okla). Patients were 7 to 22 years of age at diagnosis. Twenty-one plasma samples from anonymized hospitalized pediatric patients with noncancerous diseases (well child checkup, flu, constipation, gastroenteritis, and febrile seizure) were collected from Texas Children's Hospital and used as noncancerous disease controls. All patients gave consent to institutional review board-approved protocols. All plasma samples were collected in ethylenediaminetetraacetic acid-containing tubes at room temperature and immediately centrifuged at 1000 rpm for 10 minutes. The plasma supernatant was stored at −80°C until use. Five human plasma samples from healthy 18-year-old donors were used as healthy controls in the enzyme-linked immunosorbent assay (ELISA) validation (Equitech-Bio, Kerrville, Tex).

Table 1. Peripheral Blood Samples Used in the Study
Sample SetGroupNumberUse
  1. ELISA indicates enzyme-linked immunosorbent assay.

DiscoveryOsteosarcoma (OS)33Identification of chemokines using RayBio Antibody Array and ELISAs of 3 CXC chemokines
Disease control (D-CTL)21
Healthy controls (N-CTL)5ELISAs of 3 CXC chemokines
ValidationOsteosarcoma (OS)51ELISAs of 3 CXC chemokines
Healthy controls (N-CTL)11
Table 2. Clinical Characteristics of OS Patients
CharacteristicsAll PatientsTraining SetValidation Set
  • OS indicates osteosarcoma; NA, information is not available.

  • a

    The other primary sites of OS in the discovery set include pelvis and palate.

  • b

    The other primary sites of OS in the validation set include shoulder, pelvis, and nose base.

Total number843351
Age at diagnosis, median y (range)13 (4-22)13 (7-22)13 (4-22)
 ≤1019 (23%)5 (15%)14 (27%)
 11-1758 (69%)25 (76%)33 (65%)
 ≥187 (8%)3 (9%)4 (8%)
Sex
 Male45 (54%)17 (52%)28 (55%)
 Female39 (46%)16 (48%)23 (45%)
Primary site
 Extremities79 (94%)31 (94%)48 (94%)
 Others5 (6%)2 (6%)a3 (6%)b
Metastasis at diagnosis
 Yes18 (21%)8 (24%)10 (19%)
 No66 (79%)25 (76%)41 (80%)
Survival
 Died of disease10 (12%)10 (30%)0 (0%)
 Alive21 (25%)21 (64%)0 (0%)
 NA53 (63%)2 (6%)51 (100%)

Serum samples used in an independent test set were collected by the Children's Oncology Group under the protocol P9581 from multiple institutions. All 51 serum samples used in this study were collected at the time of diagnosis from OS patients who were 4 to 22 years of age (Tables 1 and 2). Eleven serum samples from 18-year-old healthy donors (Equitech-Bio) were also used as healthy controls in the validation phase. The serum and plasma samples were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and the quality and protein patterns between the 2 sets of samples were very similar (data not shown).

RayBio Antibody Array Analysis

Biotin Label-Based Human Antibody Array I (RayBiotech, Norcross, Ga) was used to detect 507 human proteins in plasma, as in the manufacturer's manual. In this study, 33 OS and 21 noncancerous disease control plasma samples were pooled separately and hybridized with the array. The spot signals were detected using a ScanArray Express Microarray Scanner (PerkinElmer, Waltham, Mass). Duplicated experiments were performed for each sample. The median minus background intensities of all the experiments were log-2 transformed and quantile-normalized. The intra-array coefficients of variation ranged from 9.8% to 13.9%. The correlation coefficients of the spot intensities in the replicate experiments of the control and the tumor samples were 0.94 and 0.93, respectively (data not shown), suggesting that the data generated from the array platform were reproducible. Intensity ratios between OS and noncancerous disease controls of each experiment were used to identify candidate biomarkers. The P values were calculated using 2-sample t test. In the antibody study, the candidate biomarkers were identified with a P value <.005 and fold change >1.4 to adjust for multiple testing.

ELISA Verification and Validation

The CXCL4 concentrations were determined using the IMUCLONE Platelet Factor 4 ELISA kit (America Diagnostica, Stamford, Conn). The CXCL6 and CXCL12 concentrations were determined using Human GCP-2/CXCL6 and CXCL12/SDF-1 alpha Quantikine Immunoassay kits (R&D Systems, Minneapolis, Minn), respectively.

Statistical Analysis

The intensity values were analyzed using 2-sample t test. A P value <.05 was considered to be significant. The survival analysis was performed using the Kaplan-Meier estimation, and the significance was calculated by the log-rank test using SPSS (Chicago, Ill) statistical software. The cutoffs for the survival analysis were based on the second or third quartile of the chemokine concentration data from the OS patients. The quartile was chosen based on manual inspection to identify the best separation of the survival groups.

Immunohistochemistry

Two sets of paraffin-embedded OS tissue arrays were used in this study. One array contains 49 OS and 28 chondrosarcoma cases (Biomax, Rockville, Md). The other OS tissue array contains 64 OS and 10 rhabdomyosarcoma (RMS) cases (Children's Oncology Group). The staining was performed using the VECTASTAIN ABC system (Vector Laboratories, Burlingame, Calif). Each of the mouse antihuman monoclonal primary antibodies for CXCL4 (Abcam, Cambridge, Mass), CXCL6, and CXCL12 (R&D Systems) was diluted to 10 μg/mL. Immunostaining score was calculated based on the sum of the proportion score of positive cells and the intensity score of the staining by the pathologist (J.H.). The proportion score of positive cells was classified as 0 when <1% of stained cells were observed within the tumor, 1 when 1% to 25% of the tumor cells were positive, 2 when 25% to 50% of the tumor cells were positive, 3 when 50% to 75% of the tumor cells were positive, and 4 when >75% of the tumor cells were positive. The intensity score was classified as 0, 1, 2, and 3 for trace staining (background or minimal), weak staining, moderate staining, and strong staining, respectively. The results of staining in tumor cells were subdivided into 3 groups as follows: negative if scored 0, weak if scored 1 to 4, and strong if scored 5 to 7.

Proliferation Assay

OS cell lines MG-63, U2-OS, SaOS-2, and SJSA were purchased from American Type Culture Collection (Manassas, Va). Cells were cultured at 37°C in Dulbecco modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, Calif). Cell proliferation was measured by Cell Counting Kit-8 (Dojindo Molecular Technologies, Rockville, Md). For each OS cell line, 6 replicates of 5000 cells were seeded into 96-well flat-bottomed plates in DMEM without fetal bovine serum. Each of the recombinant human CXCL4, CXCL6, and CXCL12 (R&D Systems) samples was reconstituted in phosphate-buffered saline containing 0.2% human serum albumin. After overnight culture, cells were treated with 1 μg/mL of CXCL4, CXCL6, or CXCL12 for 24 hours. 0.2% of human serum albumin in phosphate-buffered saline was used as a control. Then 10 μL of CCK-8 in 100 μL of the medium was added to each well for 3 hours. The absorbance of each of the reactions was measured at the wavelength of 450 nm.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Identification of Circulating Biomarkers for OS Using Antibody Arrays

To identify OS-associated chemokines or cytokines, we analyzed a pooled OS sample and a pooled noncancerous disease control sample on antibody arrays (Table 1). By measuring the signals of the 507 proteins on the array, we identified 20 differentially abundant proteins in the plasma of OS patients when compared with those in noncancerous disease controls. Among these 20 proteins, 8 were higher and 12 were lower in the OS patients (Table 3). Interestingly, 3 of the 8 elevated proteins were members of the CXC chemokine family. They were CXCL4 or platelet factor 4 (OS/noncancerous disease controls = 1.7), CXCL6 or granulocyte chemotactic protein-2 (OS/noncancerous disease controls = 2.1), and CXCL12 or stromal cell derived factor-1 (OS/noncancerous disease controls = 1.4) (Fig. 1A). Ten other CXC chemokines detected by the array (CXCL1, CXCL5, CXCL7, CXCL8, CXCL9, CXCL10, CXCL11, CXCL13, CXCL14, and CXCL16) did not show any differential abundance between the OS patients and the noncancerous disease controls.

thumbnail image

Figure 1. Elevation of the 3 CXC chemokines in the peripheral blood samples of the osteosarcoma (OS) patients is shown. (A) Enzyme-linked immunosorbent assays (ELISAs) of the 3 chemokines in plasma samples from the discovery phase are shown. (B) ELISAs of the 3 chemokines in the serum samples from the validation phase are shown. N-CTL indicates healthy control; D-CTL, disease control. *P < .05, **P < .01.

Download figure to PowerPoint

Table 3. Differentially Abundant Proteins Identified in the RayBio Antibody Arrays
No.Protein NameOS/D-CTL RatioP
  • a

    The 3 CXC chemokines were selected for the validation study. P values were calculated using 2-sample t test.

Increased in OS
 1CXCL6/GCP-2a2.1.000006
 2TGF-beta RIII1.8.00002
 3CXCL4/PF4a1.7.0006
 4MIF1.5.0002
 5PARC/CCL181.5.0004
 6IFN-alpha/beta R11.4.0006
 7CXCL12/SDF-1a1.4.001
 8M-CSF1.4.001
Decreased in OS
 1Smad−4.00008
 2NRG2−2.1.00003
 3MMP−1.8.00004
 4MUSK−1.7.0002
 5NRG3−1.6.000006
 6Frizzled-3−1.5.0009
 7MIP2−1.5.00003
 8IL-9−1.5.00006
 9FGF-19−1.5.0002
 10GFR alpha-2−1.5.0002
 11CD163−1.4.0002
 12Dtk−1.4.0001

Validation of the 3 CXC Chemokines

To confirm the antibody array results, we used ELISAs to measure the concentrations of the 3 chemokines in each of the plasma samples. We also included 5 plasma samples from 18-year-old healthy donors to determine the baseline levels of these 3 chemokines in normal and healthy individuals. The ELISA results showed that higher concentrations of the 3 chemokines were significantly associated with OS patients relative to both the noncancerous disease controls and healthy controls (P < .05; Fig. 1A). For instance, CXCL4 in the OS samples was 2.8-fold and 69.8-fold higher than in the noncancerous disease controls and healthy controls, respectively (Table 4). We also observed that the levels of CXCL4 and CXCL6, but not CXCL12, in the noncancerous disease controls were higher than in the healthy controls, suggesting that these 2 chemokines may also react to other noncancerous conditions. Nonetheless, the 2 chemokines in the OS samples were significantly higher than in the noncancerous disease controls samples (P < .05, Table 4).

Table 4. Elevated Levels of the 3 CXC Chemokines in the OS Peripheral Blood Samples Used in the Discovery and Validation Experiments
ProteinRatio/PaAntibody Array, OS/D-CTLELISA of the Discovery SetELISA of the Validation Set, OS/N-CTL
OS/D-CTLD-CTL/N-CTLOS/N-CTL
  • OS indicates osteosarcoma; D-CTL, disease control; ELISA, enzyme-linked immunosorbent assay; N-CTL, normal control.

  • a

    P values were calculated using 2-sample Student t test.

CXCL4Ratio1.72.825.169.862.5
P 7 × 10−50.040.0032 × 10−7
CXCL6Ratio2.11.8814.413.6
P 0.0360.1390.0114 × 10−8
CXCL12Ratio1.41.40.91.21.5
P 5 × 10−60.160.035 × 10−6

Next, we tested if we could validate the chemokine results in an independent test set of OS patients obtained from the Children's Oncology Group. A comparison between 51 OS and 11 healthy controls serum samples showed that all 3 chemokines were significantly higher in the OS patients relative to the controls (P < .05, Fig. 1B). For instance, CXCL4 was 62.5-fold higher in the OS samples when compared with the healthy controls. Although the equivalent of noncancerous disease controls samples was not available for the validation study, the fold changes of the 3 chemokines between OS and healthy controls in the discovery and validation sets were very consistent (Table 4).

Expression of the 3 Chemokines in OS Tissues

We tested if these circulating chemokines play a direct role in the tumor by examining their expressions in OS tissues on 2 sets of OS tissue arrays. The first OS tissue array (Biomax) contains 49 OS and 28 chondrosarcoma cases (Fig. 2). The second OS tissue array (Children's Oncology Group) contains 64 OS and 10 rhadomyosarcoma (RMS) cases (Fig. 2). For CXCL4, 82% and 88% of OS cases had a strong CXCL4 expression in the Biomax and Children's Oncology Group arrays, respectively (Table 5). Most of the strong expression cases were because of high proportions of positive cells (proportion score of positive cells = 3-4) with a medium to high expression of CXCL4 (intensity score = 2-3; Table 6). We also observed a high expression of CXCL4 in all 10 RMS cases tested; however, only 25% of the chondrosarcoma cases highly expressed CXCL4. For CXCL6, 80% and 100% of OS cases showed a positive staining (score = 1-7) in the Biomax and Children's Oncology Group arrays, respectively. All the RMS cases but only 32% of chondrosarcoma cases showed a positive staining of CXCL6. It is also interesting to note that a large proportion (82%) of OS cases in the Children's Oncology Group array exhibited a high expression of CXCL6 (score = 5-7). In contrast to the other 2 chemokines, CXCL12 was not expressed in all RMS and chondrosarcoma cases, and only a few OS cases showed a positive staining (Table 5).

thumbnail image

Figure 2. Representative immunohistochemistry (IHC) results of the 3 CXC chemokines on the 2 sets of osteosarcoma tissue microarrays (original magnification, ×400) are shown. (A) IHC of CXCL4: A1, OS (score = 7); A2, chondrosarcoma (score = 2); A3, OS (score = 6); A4, rhabdomyosarcoma (score = 7). (B) IHC of CXCL6: B1, OS (score = 7); B2, chondrosarcoma (score = 4); B3, OS (score = 7); B4, rhabdomyosarcoma (score = 7). (C) IHC of CXCL12: C1, OS (score = 5); C2, chondrosarcoma (score = 0); C3, OS (score = 4); C4, rhabdomyosarcoma (score = 0).

Download figure to PowerPoint

Table 5. Summary of the IHC Scores in the IHC Results
ProteinIHC scoreBiomax Tissue ArrayCOG Tissue Array
OSCSOSRMS
  1. IHC indicates immunohistochemistry; OS, osteosarcoma; CS, chondrosarcoma; COG, Children's Oncology Group; RMS, rhabdomyosarcoma.

  2. The numbers in the table represent the total number of tumor cores that had the specific IHC scores. IHC score: 0, no evidence of expression; 1-4, weak expression; 5-7, strong expression.

CXCL400 (0%)11 (39%)0 (0%)0 (0%)
1-49 (18%)10 (36%)8 (12%)0 (0%)
5-740 (82%)7 (25%)56 (88%)10 (100%)
CXCL6010 (20%)19 (68%)0 (0%)0 (0%)
1-422 (45%)5 (18%)12 (18%)1 (10%)
5-717 (35%)4 (14%)52 (82%)9 (90%)
CXCL12046 (94%)28 (100%)63 (98%)10 (100%)
1-40 (0%)0 (0%)1 (2%)0 (0%)
5-73 (6%)0 (0%)0 (0%)0 (0%)
Table 6. The Intensity Score and Proportion Score of Positive Cells in the Tumors With High IHC Scores of CXCL4
IHC ScoreBiomax Tissue ArrayCOG Tissue Array
TotalIntensity ScoreProportion ScoreOSCSOSRMS
  1. IHC indicates immunohistochemistry; OS, osteosarcoma; CS, chondrosarcoma; COG, Children's Oncology Group; RMS, rhabdomyosarcoma.

  2. The description of the scores is listed in Materials and Methods.

5141 (2%)1 (3%)11 (17%)1 (10%)
2315 (31%)3 (11%)10 (16%)1 (10%)
320 (0%)2 (7%)0 (0%)0 (0%)
62410 (20%)0 (0%)30 (47%)6 (60%)
331 (2%)0 (0%)0 (0%)1 (10%)
73413 (27%)1 (3%)5 (8%)1 (10%)
High IHC score cases40 (82%)7 (25%)56 (88%)10 (100%)

The 3 Chemokines Promote Proliferation of OS Cells In Vitro

We further tested if the 3 chemokines could stimulate the proliferation of 4 commonly used OS cell lines (U2-OS, SaOS-2, MG-63, SJSA). The results showed that all 3 chemokines significantly promoted the cell proliferation in all 4 OS cell lines (P < .05, Fig. 3). These results suggest that a high level of these chemokines in the local tumor environment or circulation may promote the tumor growth in the patients.

thumbnail image

Figure 3. Exogenous CXCL4, CXCL6, and CXCL12 significantly stimulated the proliferation of 4 osteosarcoma cell lines (A, U2-OS; B, SaOS-2; C, MG63; and D, SJSA) under a serum-deprived condition. Y axis is the optical density (OD). CTL indicates control. *P<.05

Download figure to PowerPoint

CXCL4 and CXCL6 Were Associated With the Patient Outcome

Our expression and functional results argue that these chemokines may play a role in the prognosis of OS patients. We therefore correlated the plasma levels of these 3 chemokines with the overall survival of the OS patients in the discovery set, where we had clinical follow-up information. We found that the higher levels of both CXCL4 (P = .007) and CXCL6 (P = .03), but not CXCL12 (P = .282), were significantly associated with a poor outcome (Fig. 4). These results were consistent with our findings that only CXCL4 and CXCL6 are expressed frequently and strongly in OS tissues. We then tested if the correlation of the poor outcome was because of an association with the metastatic disease at diagnosis in these patients. However, none of the 3 chemokines significantly correlated with metastasis.

thumbnail image

Figure 4. Survival analysis of the 3 chemokines in OS patients is shown. Low and high denote low and high chemokine levels, respectively, in the patients. Higher levels of (A) CXCL4 and (B) CXCL6, but not (C) CXCL12, were significantly associated with a poor outcome (log-rank test, P<.05).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

In this study, we identified 20 proteins that were differentially abundant in the OS plasma samples. Some of these proteins have been reported to correlate with cancer, such as TGF-beta RIII16 and Smad4.17 Among the 8 elevated proteins in OS, 3 of them (CXCL4, CXCL6, and CXCL12) belong to the CXC chemokine family. By using ELISA, we validated that the higher levels of the 3 chemokines were associated with OS in the independent test set. These findings suggest that these 3 CXC chemokines may be useful for early diagnosis of OS. This is particularly important for children with a known predisposition to OS, such as patients harboring germline mutations in the TP5318 or RB,19 or Rothmund-Thomson Syndrome patients with RECQL5 mutations.20 A blood-based assay of high-risk individuals may allow more timely and periodic monitoring of the disease and save lives. However, further studies of the circulating chemokines in patients with other sarcomas or bone diseases will be needed to test if the chemokines are specific to OS.

Previous studies have reported that CXCL4 is elevated in a variety of gynecologic malignancies21, 22 and prostate cancer23; however, its role in OS is largely unknown. The primary function of CXCL4 is involved in platelet activation.24 In our study, CXCL4 was consistently increased in the OS patients of both the discovery and validation sets (60-fold higher). The serum samples used in the validation phase suggest that the higher level of this chemokines in OS patients was not because of platelet contamination of the plasma samples. In fact, the elevation of CXCL4 is consistent with the enhanced platelet activation observed in cancer patients that contributes to the increased occurrence of thromboembolic events.25 Increased platelets may also provide protection of metastatic cells in the circulation.26, 27

Interestingly, CXCL6 was first discovered from the culture medium of the OS cell line MG-63 after stimulation with interleukin-1β.28 Our results showed that CXCL6 was expressed in OS. However, a strong CXCL6 expression was observed in a higher proportion of OS samples in the Children's Oncology Group array (82%) than those from the Biomax array (35%). This may reflect the heterogeneity of the OS cases used in different arrays, or that the expression of CXCL6 in OS is inherently more variable. We have also observed that the tumor expression patterns of CXCL4 and CXCL6 share some similarities. The expressions of the 2 chemokines in OS are moderately correlated (r = 0.37 in the Children's Oncology Group array and r = 0.56 in the Biomax array), which is consistent with the finding that both CXCL4 and CXCL6 genes are located at the 4q12-q13, whereas CXCL12 is located at 10q11.1.29

CXCL12 is secreted by stromal cells from a variety of tissues, such as bone marrow, lung, and liver.30 The chemotactic effect of CXCL12 is mediated by interaction with its receptor CXCR4. The involvement of the CXCR4/CXCL12 axis has been implicated in tumor progression of a variety of cancers and cancer stem cells.31, 32 In OS, the migration of OS cells expressing CXCR4 receptor follows the CXCL12 gradient, and the adhesion of OS cells to endothelial and bone marrow stromal cells is promoted by a CXCL12 treatment. By using mouse models, the CXCR4/CXCL12 axis was proved to be involved in the metastatic process of OS cells, as development of lung metastasis was significantly decreased in the OS mouse models by the administration of a CXCR4 inhibitor or antagonist.15, 33 An earlier study has found that CXCR4 is highly expressed in metastatic OS,34 and lung expresses CXCL12.35 These findings are consistent with our results that most OS cases did not express CXCL12. The increased level of CXCL12 in the peripheral blood of OS patients may be derived from other host tissues, such as lung or metastasized OS cells.

Our results suggest that these chemokines may play a direct functional role in OS tumorigenesis. Platelet CXCL4 has been found to be elevated in the early tumor growth of liposarcoma, breast cancer, and OS.36 CXCL6 can induce expression of proliferating cell nuclear antigen in small cell lung cancer cell lines in vitro,37 and melanoma cell growth in vivo.38 In addition, CXCL12 induces proliferation in some tumor cell lines, such as ovarian carcinoma39 and nonsmall cell lung cancer.40 To test if the CXC chemokines had a general proliferative effect on tumor cells, we repeated the proliferation assay using 3 different types of non-OS cell lines available in our laboratory. They were TC7 (Ewing), HepG (liver hepatocellular carcinoma), and DAOY (medulloblastoma). Our results showed that the 3 chemokines significantly increased the proliferation of TC7 cells only, having no effects on the HepG and DAOY cells (data not shown). These results suggest that the 3 CXC chemokines only promote growth for certain types of tumors. A recent study has shown that many chemokine receptors are expressed in OS, such as CXCR3, CXCR4, and CXCR5.41 However, CXCR1, the cognate receptor of CXCL6, was not reported. The receptor for CXCL4 in the tumor cell is still not clear. Peptide-based or small molecules that can inhibit the functions of the chemokines or receptors may provide an alternative and effective therapy for OS patients who have a higher level of these chemokines.42 For instance, inhibitor molecules of CXCR4, the receptor of CXCL12, have been already used in 4 clinical trials for various cancers (ClinicalTrials.gov database). Therefore, the development of inhibitors for other chemokines in OS is feasible. Because there is no targeted therapy for OS other than chemotherapy, our findings provide a foundation for future studies of these chemokines and their receptors, which may lead to a novel therapeutic approach to stop tumor growth in OS.

Our results showed that both CXCL4 and CXCL6 significantly correlated with patient outcomes; however, no associations between these 2 chemokines and the metastatic status at diagnosis were found. The high expression of these 2 chemokines in the tumor and plasma of OS patients may exert a strong proliferative effect and promote tumor progression and, hence, a poorer outcome of the patients. In contrast, plasma CXCL12 is likely to be expressed by other parts of the body, and the difference between the plasma levels of the protein in OS patients and healthy controls was relatively small when compared with the other 2 chemokines (Table 4); thus, the plasma level of CXCL12 did not significantly correlate with the patient outcomes. Despite the promising results in this study, we understand that our results were based on a relatively small group of patients, and thus they need to be interpreted with caution. A larger patient cohort is needed to validate the use of CXCL4 and CXCL6 as prognostic factors in OS. We have collaborated with the Children's Oncology Group to use the samples collected from the European and North American OS Study to validate these findings.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

We thank William Meyer at University of Oklahoma Health Sciences Center for contributing osteosarcoma plasma for this study; Carolyn Pena for assisting with the article's preparation; and Mark Krailo, Don Barkauskas, Susan Conway, and Chand Khanna from the Children's Oncology Group for their help in retrieving clinical information of the osteosarcoma cases.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES

Supported by funding from the National Cancer Institute (1U01CA088126-01 and 1U01CA114757-01) and the Gillson Longenbaugh Foundation (C.C.L.), as well as the Wendy Will Case Cancer Foundation, the Sarcoma Foundation of America, and the Fleming and Davenport Award of Texas Medical Center (T.-K.M.). This research is also supported by the Chair's Grant U10 CA98543 and the Human Specimen Banking Grant U24 CA114766 of the Children's Oncology Group from the National Cancer Institute, National Institutes of Health, Bethesda, Maryland. Additional support for this research is provided by an endowment from the WWWW (QuadW) Foundation to the Children's Oncology Group.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. CONFLICT OF INTEREST DISCLOSURES
  8. REFERENCES