• osteosarcoma;
  • microarray analysis;
  • cancer-testis antigens;
  • neoplasm;
  • metastasis;
  • prognosis


  1. Top of page
  2. Abstract


From 30% to 40% patients with osteosarcoma eventually experience medical failure; and few biomarkers of prognostic significance have been established. High-throughput methods like gene microarray analysis can help to identify molecular biomarkers that are useful for diagnosing osteosarcoma and targeting its treatment.


Oligonucleotide microarrays were used to compare expression profiles of osteosarcoma cell lines and osteoblasts. Differentially expressed genes were confirmed by real-time polymerase chain reaction (PCR) analysis. Corresponding proteins were evaluated by flow cytometry and Western blot analysis in osteosarcoma cell lines and by immunohistochemistry in osteosarcoma tissues. The association between staining intensity and clinical outcome was analyzed further.


Cancer-testis antigens, including melanoma antigen family A (MAGEA), chondrosarcoma-associated gene family, member 2 (CSAG2), and preferentially expressed antigen in melanoma (PRAME), were increased significantly in all osteosarcoma cell lines that were analyzed. Real-time PCR examinations indicated that cancer-testis antigen expression was frequent and coordinated in patients with osteosarcoma. The expression of MAGEA was confirmed by Western blot and flow cytometry analyses in osteosarcoma cell lines. Furthermore, immunohistochemical staining analysis suggested that MAGEA expression may be used to predict distant metastasis and poor survival. The adjusted relative risk for lung metastasis was 2.79 (95% confidence interval, 1.12-6.93; P = .028) for MAGEA-positive patients. Five-year survival rates for patients with and without MAGEA expression were 39.6% ± 8.4% and 80% ± 8.9%, respectively (log-rank test; P = .01).


The combined use of an oligonucleotide microarray, a clinical database, and a tissue bank was useful for identifying molecular tumor markers. The frequent expression of MAGEA and other cancer-testis antigens in osteosarcoma indicates that they may be useful as diagnostic markers and targets of immunotherapy that warrant further investigation. Cancer 2012. © 2011 American Cancer Society.


  1. Top of page
  2. Abstract

Osteosarcoma is the most common primary malignant bone tumor in both children and adolescents, accounting for 56% of malignant bone tumors that occur in the first 2 decades of life.1 In the last 3 decades, remarkable progress has been made in the treatment of osteosarcoma as a result of combined chemotherapy and advances in surgical and diagnostic imaging techniques. However, despite these advances, 30% to 40% patients still eventually die of their disease.2, 3 Moreover, rare definitive prognostic markers have been identified in patients with nonmetastatic osteosarcoma. The histopathologic response to preoperational chemotherapy, which was recognized previously as a dependable and reproducible prognostic indicator, currently is controversial.4 Therefore, a better understanding of the molecular biology and specific identifying biomarkers that lead to tumor initiation and progression is critical to the effective diagnosis and treatment of osteosarcoma.

High-density oligonucleotide microarrays (OMAs)5 have been introduced to synchronously analyze the expression levels of thousands of genes,6 overcoming limitations inherent in the analyses of single genes. Over the last decade, gene expression profiling by microarray has identified key genes and cellular signaling pathways that are involved in osteosarcoma development and metastasis, indicators of canine osteosarcoma survival, and markers of human osteosarcoma chemoresistance.7-10 Conventional comparisons of the genetic expression profiles of osteosarcoma and normal osteoblasts have been performed,11 but the correlations with clinical parameters and patient prognosis were not evaluated.

The objective of the current study was to identify potential osteosarcoma biomarkers of both prognostic significance and utility as therapeutic targets by screening differentially expressed genes between tumor cell lines and normal osteoblasts with high-density OMAs. The microarray that was used for these studies (U133 plus 2.0; Affymetrix, Santa Clara, Calif) incorporated 47,000 probes that accounted for a large fraction of the expressed human genome. Over expressed genes were confirmed by real-time reverse transcriptase-polymerase chain reaction (RT-PCR) and examined in a panel of osteosarcoma specimens. Associations between the expression of selectively up-regulated genes and clinical patient outcomes were analyzed.


  1. Top of page
  2. Abstract

Cell Lines, Patients, and Tissue Samples

The human osteogenic sarcoma cell lines U-2OS, Saos-2, and MG63 were obtained from the American Type Culture Collection (Manassas, Va). The IOR/OS9 cell line, which was established from the bone metastasis of a high-grade osteosarcoma,11 was a generous gift from Dr. M. Serra (Rizzoli Institute, Bologna, Italy). The Zos and Zos-M cell lines were established in our department from the primary tumor and skip metastasis, respectively, of a patient with high-grade osteosarcoma.12 All of these osteosarcoma cell lines were maintained and propagated as described previously.12 Human osteoblasts were cultured by the sequential collagenase/trypsin digestion of bone graft as previously described.13, 14 The cells from the third generation were harvested for use upon confluence.

The medical records of patients with osteosarcoma who underwent surgery at the First Affiliated Hospital of Sun Yat-sen University between January 2003 and December 2007 were reviewed. Two independent pathologists confirmed the diagnosis of osteosarcoma. Patients who remained in active clinical treatment or who were followed by regular telephone contact were included in this study group. Fresh-frozen tumor specimens were prepared by immersing the tumor tissue in liquid nitrogen immediately after surgical resection followed by storage at −80°C in the tissue bank of the Department of Musculoskeletal Oncology. Only tumor samples that were evaluated by pathologists and composed of >80% tumor cells were used for the real-time RT-PCR study. Paraffin-embedded tissue blocks containing osteosarcomas that were excised at the time of surgery were prepared routinely and stored in the tissue bank of the Department of Pathology. In addition, a clinical database was established that included patient, tumor, and prognostic characteristics. Consent from patients or their guardians was obtained and approved by the Sun Yat-sen University Ethics Committee.

Expression Profile Analysis

Expression profiling of 4 osteosarcoma cell lines, including U-2OS, Saos-2, Zos, and Zos-M, and osteoblasts derived from 2 normal individuals were conducted as previously described.15 All cells were cultured until confluent, at which time they were serum starved for 18 hours. Media was changed, and cells were cultured for an additional 8 hours; then, total RNA was harvested with Trizol (Invitrogen, Carlsbad, Calif). RNA degradation was tested by denaturing gel electrophoresis. The RNA was reverse transcribed (Invitrogen) into combinational DNA (cDNA) that was then transcribed in vitro into biotinylated cRNA. The target cRNA was then fragmented and hybridized to a human U133 plus 2.0 microarray (Affymetrix) following the manufacturer's recommended protocol. Hybridization of cRNA to the Affymetrix human U133 plus 2.0 chips, signal amplification, and data collection were performed using an Affymetrix fluidics station and chip reader. Chip files were then analyzed using the Affymetrix version 5.0 (MAS5) comparison analysis software. Criteria indicated by Affymetrix were used to determine robust changes in gene expression. Briefly, transcripts were defined as up-regulated in osteosarcoma only when identified as “present” by the Affymetrix detection algorithm and as “significantly increased” as determined by the Affymetrix change algorithm, with a change in P value of < .01. The fold change between tumor cells and osteoblasts had to be at least 3 to identify a transcript as altered.

Quantitative Real-Time Reverse Transcriptase-Polymerase Chain Reaction

Total RNA was used as a template to synthesize double-stranded cDNA with oligo primers in conjunction with Superscript II reverse transcriptase (Invitrogen). Relative levels of RNA were measured by quantitative real-time PCR using the ABI 7500 Sequence Detection System default settings. For all genes, amplification was conducted with SYBR Green Master Mix (Applied Biosystems, Foster City, Calif). Primer sequences designed with Primer3 ( [Accessed May 6th, 2009]) and blasted on ( [Accessed May 6th, 2009] )for individual genes are listed in Table 1. To avoid DNA contamination, each primer was designed to span introns. Cycle threshold values were defined as the cycle in which fluorescence intensity reached the geometric phase of amplification. Samples were performed in triplicate and averaged. Values for individual genes of interest were then normalized to the value of the housekeeping gene β-actin, allowing the calculation of fold differences in gene expression using the ABI software.

Table 1. Primer Sequence and Product Length of Cancer-Testis Antigens
GenePrimer SequenceaFragment Size, bp
  • Abbreviations: bp, base pairs; CSAG2, chondrosarcoma-associated gene family, member 2; MAGEA, melanoma antigen family A; PRAME, preferentially expressed antigen in melanoma.

  • a

    The primers for cancer-testis antigens gene were designed with the Primer3 (; [Accessed May 6th, 2009]) and the sequences were blasted (; [Accessed May 6th, 2009]).


Flow Cytometry

The 6 human osteosarcoma cell lines Zos, Zos-M, U-2OS, Saos-2, MG-63, and IOR/OS9, were used for flow cytometry analysis. Cells were harvested and suspended at 2 × 106 cells/100 μL with staining buffer containing 1% fetal bovine serum (Invitrogen). Phycoerythrin (PE)-conjugated monoclonal antibody against human melanoma antigen family A (MAGEA) (6C1; Santa Cruz Biotechnology, Santa Cruz, Calif), which is reactive with MAGE-A1, MAGE-A3, MAGE-A4, and MAGE-A6, were added to the cell suspension at the manufacturer's suggested concentration and incubated at 4°C in the dark for 60 minutes. The labeled cells were fixed in 100% methanol on ice for 5 minutes. Flow cytometry analyses were performed in triplicate using a Cytomics FC 500 machine (Beckman Coulter, Brea, Calif). Data were analyzed using CXP Analysis 2.0 software (Beckman Coulter). The data presented are the mean values of 3 independent experiments.

Western Blot Analysis

Western blot analysis was performed by using standard methods. Briefly, 25 to 50 μg of protein extracted from cultured cells was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto nitrocellulose membranes. Membranes were then blocked and immunoblotted with antibody against human MAGEA (6C1; Santa Cruz Biotechnology). Horseradish peroxidase-conjugated secondary antibodies were detected by enhanced chemiluminescence substrate (Amersham Biosciences, Piscataway, NJ). IRdye680-conjugated and IRdye800-conjugated secondary antibodies (Molecular Probes, Eugene, Ore) were detected using Odyssey Imaging (LICOR Biosciences, Lincoln, Neb).

Immunohistochemical Staining and Evaluation

Paraffin-embedded osteosarcoma sections that were 4 mm thick were placed on silane-coated slides (DAKO, Glostrup, Denmark) for immunohistochemical analysis. The samples were dewaxed in xylene and rehydrated in increasing concentrations of alcohol. Antigen retrieval was performed by boiling in ethylene diamine tetracetic acid buffer, pH 8.0, for 10 minutes in a microwave oven. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 5 minutes. Sections were then incubated with a monoclonal antibody specific to MAGEA (6C1; Santa Cruz Biotechnology) diluted in antibody-dilution buffer (DAKO) at 1:100 overnight. The slides were then processed using Dako Envision+ horseradish peroxidase (DakoCytomation, Glostrup, Denmark) according to the manufacturer's protocol, counterstained briefly with Mayer hematoxylin (Amber Scientific, Belmont, Wash), and coverslipped. Testes specimens were used as positive controls, and negative controls were prepared by omitting the primary antibody. The staining intensity of MAGEA for each slide was evaluated and scored by 2 independent pathologists: A score of “++” was used to indicate strong positive staining, “+” was used to indicate weak positive staining, and “−” was used to indicate no staining.

Statistical Analysis

The differences in expression of dysregulated genes between the 2 groups were compared using chi-square tests with Yates correction. Survival curves were calculated using the Kaplan-Meier method and were compared using the log-rank test for univariate analysis. Prognostic factors were analyzed using a Cox proportional hazards model with adjustment for potential confounding factors. Categorical variables were compared using the Fisher exact test. Differences were considered significant if the P value was < .05.


  1. Top of page
  2. Abstract

Detection of Cancer-Testis Genes Overexpressed in Osteosarcoma by Oligonucleotide Microarrays

A comparison of the expression profiles of osteosarcoma cell lines and normal human osteoblasts identified 174 genes that were altered consistently in all 4 osteosarcoma cell lines. Twenty-eight genes were up-regulated, and 146 genes were down-regulated.

Among the up-regulated genes (Table 2), a notable feature was the frequency of tumor antigens represented by 6 MAGEAs (MAGEA1, MAGEA2, MAGEA3, MAGEA5, MAGEA6, and MAGEA12); chondrosarcoma-associated gene family, member 2 (CSAG2); preferentially expressed antigen in melanoma (PRAME); and tumor protein D52 (TPD52), which constituted 30% of the up-regulated list. These transcripts also displayed the greatest fold increases. With the exception of TPD52, all of the genes belong to the CT-antigen family, which led us to focus on the expression of these antigens in osteosarcomas. In addition to the CT-antigens that were increased consistently in all 4 osteosarcoma cell lines, other members of the MAGE family that were up-regulated in osteosarcoma cell lines and identified by OMA are summarized in Table 3. Expression levels of these transcripts varied in different osteosarcoma cell lines.

Table 2. Consistently Up-Regulated Genes Detected by Oligonucleotide Microarray in Osteosarcoma Cell Lines
Classification/ Probe IDGene SymbolLog2 (Fold Increase)
  1. Abbreviations: ASXL1, additional sex combs like 1; CBX4, chromobox homolog 4; CHMP4C, chromatin modifying protein 4C; CSAG2, chondrosarcoma-associated gene family, member 2; CXCL14, chemokine (C-X-C motif) ligand 14; DKFZp686A01247, LIM and calponin homology domains 1; DNM3, dynamin 3; GART, phosphoribosylglycinamide formyltransferase, phosphoribosylglycinamide synthase, phosphoribosylaminoimidazole synthetase; HADH2, 3,hydroxyacyl-coenzyme A dehydrogenase type II; ID, identification; KIF21A, kinesin family member 21A; MAGEA, melanoma antigen family A; LPHN1, latrophilin 1; MCAM, melanoma cell adhesion molecule; MRPL13, mitochondrial ribosomal protein L13; NDUFB9, nicotinamide adenine dinucleotide dehydrogenase (ubiquinone) 1 beta subcomplex, 9, 22 kDa; PARD6B, par-6 partitioning defective 6 homolog beta; PRAME, preferentially expressed antigen in melanoma; PSMD4, proteasome (prosome, macropain) 26S subunit, nonadenosine triphosphatase, 4; RBM38, RNA binding motif protein 38; STAU2, staufen, RNA binding protein, homolog 2; TPD52, tumor protein D52; WNTB, WntB wingless-type MMTV integration site family; ZNF323, zinc finger protein 323.

  2. aExpression profiling of 4 osteosarcoma cell lines, including U-2OS, Saos-2, Zos, Zos-M, and 2 normal human osteoblasts, were compared using an oligonucleotide microarray (human U133 plus 2.0 microarray; Affymetrix, Santa Clara, Calif). Twenty-eight genes that were consistently up-regulated in all 4 osteosarcoma cell lines compared with normal osteoblasts were classified according their functions.

Tumor antigen  
Cell motility  
 Cell adhesion  
Transcription and translation
Signal transduction  
Table 3. Expression of Cancer-Testis Antigen Genes in Osteosarcoma Cell Lines Detected by Oligonucleotide Microarraya
   Fold Increaseb
Gene SymbolProbe IDChromosomal LocationZos CellsZos-M CellsSaos-2 CellsU-2OS Cells
  • Abbreviations: ChrX, chromosome X; CSAG2, chondrosarcoma-associated gene family member 2; ID, identification; MAGEA, melanoma antigen family A; PRAME, preferentially expressed antigen in melanoma.

  • a

    The fold increase in cancer-testis antigen gene expression in 4 osteosarcoma cell lines is compared with the expression in normal human osteoblasts detected by oligonucleotide microarray analysis.

  • b

    Data shown are the fold increases in osteosarcoma cells over human osteoblasts (log2[fold increase]).


Cancer-Testis Antigen Gene Expression in Osteosarcoma Cell Lines and Tissues

To confirm the OMA results, expression levels of MAGEA1, MAGEA2, MAGEA3, MAGEA12, PRAME, and CSAG2 were examined using real-time RT-PCR in the osteosarcoma cell lines MG-63 and IOR/OS9 in addition to the 4 osteosarcoma cell lines that were used for OMA analysis. The results are provided in Table 4 and are expressed as the fold increase over cultured osteoblasts. For the Zos, Zos-M, U-2OS, and Saos-2 cell lines, the results were similar to the OMA results; high expression levels of the CT-antigen genes were observed in the MG-63 and IOR/OS9 cell lines.

Table 4. Expression of Cancer-Testis Antigen Genes Confirmed in Osteosarcoma Cell Lines by Quantitative Polymerase Chain Reaction Analysisa
 Fold Increaseb
Gene SymbolZos CellsZos-M CellsSaos-2 CellsU-2OS CellsIOR/OS9 CellsMG63 Cells
  • Abbreviations: CSAG2, chondrosarcoma-associated gene family member 2; MAGEA, melanoma antigen family A; PRAME, preferentially expressed antigen in melanoma.

  • a

    The up-regulation some cancer-testis antigens was confirmed by real-time reverse transcriptase-polymerase chain reaction analysis in 4 osteosarcoma cell lines, and the analysis included 2 other osteosarcoma cell lines, IOR/OS9 and MG63. The fold increase in each cell line was compared with that in normal human osteoblasts.

  • b

    Data shown are the fold increase in osteosarcoma cells over human osteoblasts (log2[fold increase]).


Expression levels of the CT-antigen genes were tested further in 28 fresh-frozen osteosarcoma specimens by using real-time RT-PCR. The results were normalized to the ratio amount expressed in normal human osteoblasts. A ratio <2-fold was defined as negative. Each specimen expressed at least 2 of the 6 tumor antigens, and 82% of the specimens expressed at least 4 of the 6 CT antigens. Overall, 42.86% of the specimens expressed all 6 CT antigens. Figure 1 indicates that, among the 6 CT antigens, MAGEA1 and MAGEA3 were detected in all specimens followed by PRAME (96.4% ± 3.6%), MAGEA2 (82.1% ± 6.2%), CSAG2 (78.6% ± 3.6%), and MAGEA12 (46.4% ± 7.1%). Furthermore, the expression level of each CT antigen was parallel to the expression in the other antigens. Table 5 indicates that all 6 antigens were statistically significantly coexpressed in osteosarcoma tissues according to Pearson correlation analysis; tumors that were positive for 1 CT antigen often had simultaneous expression of other CT antigens.

thumbnail image

Figure 1. The percentage of cancer-testis (CT) antigen expression was detected by real-time polymerase chain reaction analysis in fresh-frozen osteosarcoma tissues from 28 patients. Melanoma antigen family A member 1 (MAGEA1) and MAGEA3 were detected in all patients followed by preferentially expressed antigen in melanoma (PARAME), MAGEA2, chondrosarcoma-associated gene family, member 2 (CSAG2), and MAGEA12, which could be detected in only approximately 50% of patients.

Download figure to PowerPoint

Table 5. Coexpression of Cancer-Testis Antigen Genes in Osteosarcoma Tissue Detected by Quantitative Polymerase Chain Reaction Analysis
CT1CT2No. of Specimens Tested for Both CT1 and CT2Correlation CoefficientaP
  • Abbreviations: CT antigen, cancer-testis antigen; CSAG2, chondrosarcoma-associated gene family member 2; MAGEA, melanoma antigen family A; PRAME, preferentially expressed antigen in melanoma.

  • a

    Pearson correlation analysis was used to determine the correlation coefficient for the expression level of each CT antigen compared with the others CT antigens in 28 fresh-frozen osteosarcoma tissues using real-time reverse transcriptase-polymerase chain reaction. The results indicate that the coexpression of all 6 antigens was statistically significant in osteosarcoma tissues.


Cancer-Testis Antigen Expression in Osteosarcoma Cell Lines

To find out whether increased MAGEA in messenger RNA (mRNA) could be detected in the protein of corresponding osteosarcoma cell lines, a Western blot analysis was carried out. Figure 2a indicates that MAGEA was detected in the osteosarcoma cell lines U-2OS, Saos-2, Zos, Zos-M, MG-63, and IOR/OS9 and was absent in normal human osteoblasts.

thumbnail image

Figure 2. (a) Melanoma antigen family A (MAGEA) expression was detected in 6 osteosarcoma cell lines at different levels by Western blot analysis but was absent in normal human osteoblasts (Hobs). (b) The ratio of MAGEA-positive cells was examined in Hobs and in 6 osteosarcoma cell lines using flow cytometry. IOR/OS9 cells had the highest ratio of MAGEA-positive cells followed by Zos cells, Zos-M cells, U-2OS cells, SAOS-2 cells, and MG63 cells, and the trend was consistent with results detected by Western blot analysis.

Download figure to PowerPoint

To analyze the percentage of MAGEA-positive cells in osteosarcoma cell lines, flow cytometry was performed in 6 osteosarcoma cell lines, as illustrated in Figure 2b. The mean (±standard error) ratios of MAGEA-staining positive cells were 22% ± 3.2% in Saos-2 cells, 30% ± 2.8% in U-2OS cells, 43% ± 5.8% in Zos cells, 26% ± 3.6% in Zos-M cells, 11.8% ± 2.3% in MG-63 cells, and 63% ± 8.6% in IOR/OS9 cells, all of which had the same trend revealed by Western blot analysis.

Cancer-Testis Antigen Gene Expression in Patients With Osteosarcoma and its Correlation With Clinical Outcome

The characteristics of the patients with osteosarcoma in the current study are listed in Table 6. The average age was 19.8 years (range, 6-52 years). All the patients were followed closely for a mean of 48 months (range, 36-62 months). Eighty-six specimens, including 80 specimens of localized lesions from 64 patients and 6 specimens of lung metastases, were examined for MAGEA expression by immunohistochemistry. Figure 3b-d illustrates that MAGEA staining in osteosarcoma cells was located mainly in the cytoplasm. It is noteworthy that most of the MAGEA-positive cells, especially the strongly positive cells, gathered in the periphery of tumor tissues at the site of actively proliferating cell foci (Fig. 3c,d). In total, 42 of 64 patients (65.6%) patients had positive MAGEA staining in localized osteosarcoma tissue; whereas 6 different patients' osteosarcoma tissues from lung metastases also were MAGEA positive. In addition, MAGEA staining intensity was much stronger in lung metastasis than in localized osteosarcoma (strong positive staining, 83.3% vs 40.6%; P = .04) (Fig. 3e).

Table 6. Clinical Characteristics of Patients, Tumors, and Melanoma Antigen Family A Staining
  MAGEA Staining 
CharacteristicTotal No. of PatientsPositiveNegativeP
  1. Abbreviations: MAGEA, melanoma antigen family A; NA, not available.

  2. aClinical characteristics and MAGEA expression in 64 patients with osteosarcoma are presented. The results indicate that MAGEA expression was not associated significantly with age, sex, tumor site, Enneking stage, or histologic classification (Fisher exact chi-square test).

Age, y    
 Other sites752.054
Enneking stage    
thumbnail image

Figure 3. The expression of melanoma antigen family A (MAGEA) in osteosarcoma (OS) tissue was detected by immunohistochemistry and was located mainly in the cytoplasm. MAGEA expression was (a) detected in testis as a positive control, (b) weakly positive in some localized osteosarcoma, (c) strongly positive in other localized osteosarcoma, and (d) strongly positive in lung metastasis of osteosarcoma (d). (c,d) More intensive staining of MAGEA-positive cells is observed around the periphery of the tumor. (e) MAGEA staining intensity distribution is illustrated in a localized osteosarcoma and a lung metastasis from osteosarcoma.

Download figure to PowerPoint

Next, we analyzed the association of MAGEA expression with clinical parameters analyzed. MAGEA expression was not associated with age, sex, tumor site, or histopathology type. The ratio of MAGEA expression was relatively higher in patients who had Enneking stage III disease than in patients who had Enneking stage II disease, but no statistical significance was determined (75% vs 64.3%; P = .274).

To explore whether MAGEA expression could predict the prognosis of patients with osteosarcoma, survival analyses were conducted by excluding the patients who presented with distant metastases at diagnosis. The results indicate that the estimated 5-year survival for patients with and without MAGEA expression was 39.6% ± 8.4% and 80% ± 8.9%, respectively (P = .01) (Fig. 4a). Among the patients who had positive MAGEA staining, the estimated 5-year survival of those with strong and weak MAGEA staining was 25.6% ± 9.8% and 60% ± 12.6%, respectively (P = .026) (Fig. 4b), indicating that MAGEA expression predicts an unfavorable prognosis for patients with osteosarcoma. Further analysis performed between MAGEA expression and tumor progression events suggested that MAGEA expression is predictive of a higher risk of distant metastasis but not local recurrence (Table 7). The risk of distant metastases for patients with MAGEA staining was 2 times greater than the risk for those without MAGEA staining (relative risk, 2.79; 95% confidence interval, 1.12-6.93; P = .028) when adjusted for age and sex (Fig. 4c).

thumbnail image

Figure 4. These Kaplan-Meier curves illustrate a survival analysis of 56 patients with nonmetastatic osteosarcoma. (a) The clinical outcome of patients who lacked melanoma antigen family A (MAGEA) expression was much better than the outcome of patients who were positive for MAGEA expression (5-year survival rate: 80% ± 8.9% vs 39.6% ± 8.4%, respectively; P = .01), and (b) the clinical outcome of patients with strong positive MAGEA expression had a lower survival rate than patients who had weak MAGEA expression (5-year survival rate: 25.6% ± 9.8% vs 60% ± 12.6%, respectively; P = .026). (c) Patients with MAGEA expression had a much higher risk of distant metastasis than those without MAGEA expression (relative risk, 2.79; 95% confidence interval; 1.12-6.93; P = .028).

Download figure to PowerPoint

Table 7. Adjusted Relative Risk of Melanoma Antigen Family A Expression for Recurrence, Metastasis, and Survival in 56 Patients Without Metastasis at the Time of Diagnosisa
  MAGEA Staining  
VariableTotal No. of PatientsPositiveNegativeRR (95% CI)P
  • Abbreviations: CI, confidence interval; MAGEA, melanoma antigen family A; RR, relative risk.

  • a

    The associations between MAGEA expression and prognosis, including recurrence, metastasis, and survival, were analyzed in 56 patients with nonmetastatic osteosarcoma using a Cox proportional hazard model that was adjusted for potential confounding factors. The results indicate that higher MAGEA expression is associated with a greater risk of distant metastasis and mortality.

  • b

    Eleven patients, including 10 with MAGEA expression who died without exact confirmation of the date metastasis was diagnosed, were excluded from the analysis.

 Yes7340.45 (0.10-2.05).301
 Yes191542.79 (1.12-6.93).028
 Yes252143.75 (1.28-11.01).016


  1. Top of page
  2. Abstract

An intriguing finding of the current study is the high expression levels of CT antigens in osteosarcoma cell lines, including the MAGEA family, PRAME, and CSAG2. CT antigens are encoded by a unique group of genes that are expressed predominantly in human testicular germ cells and have little or no expression in somatic adult tissues but become aberrantly activated in various malignancies, including melanoma, ovarian cancer, lung cancer, and other various cancers.16 Because of their tumor-specific expression and their ability to elicit autologous T-cell responses, CT antigens are considered to be promising targets for tumor immunotherapy. Several CT antigens, including NY-ESO-1, MAGE-A3, and MAGE-A4, have been or are being studied as target antigens in vaccine clinical trials for various types of tumors, including nonsmall cell lung cancer, ovarian cancer, gastrointestinal carcinoma, and urothelial carcinoma.7, 9-13 In osteosarcoma, several members from the MAGE-A family, namely L antigen (LAGE) and G antigen (GAGE), are expressed, as reported previously,17, 18 but their roles in osteosarcoma have yet to be fully explored because of limited data. In the current study, the expression of CT antigens in osteosarcoma was identified with a high-throughput tool, OMA, and was confirmed at the mRNA and protein levels. Furthermore, the prognostic value of MAGEA was evaluated in 56 patients with nonmetastatic osteosarcoma. The results demonstrated that MAGEA was associated with a high risk of metastasis and poor survival. These data are consistent with findings in melanoma and lung cancer.19, 20

The examination of CT-antigen expression in osteosarcoma reveals several characteristics. First, consistent with what has been observed in other solid tumors, the prevalent expression of CT antigens is present in osteosarcoma, as confirmed by mRNA transcripts that were detected in all specimens examined. Among the CT antigens that we tested, MAGEA1 and MAGEA3 were the most common CT antigens, whereas MAGEA12 was the least common. Second, the coordinated expression of CT antigens was observed in osteosarcoma, not only with CT antigens on the X chromosome (X-CTAs), such as the family members of MAGE-A, CSAG2, and MAGE-A, but also with X-CTAs and non-X-CTAs, such as PRAME, MAGE-A, and CSAG2, which are reminiscent of nonsmall cell lung cancer and esophageal cancer.20, 21 Finally, different tumors and cell lines appear to have specific patterns of CT antigen expression. For example, increased expression of MAGEA8 and MAGEA11 was observed in U-2OS cells but not in Zos or Zos-M cells. Conversely, expression of MAGEA9 and MAGEA10 was elevated in Zos and Zos-M cells but not in U-2OS cells. Epigenetic deregulations may control the frequent and coordinated expression of CT antigens,22-25 but elucidating the detailed mechanisms responsible for the expression pattern in osteosarcoma will require further investigation.

Another novel finding of the current study is that PRAME and CSAG2 are expressed in osteosarcoma. PRAME was first cloned by Ikeda et al from a patient with recurrent melanoma,26 and it is correlated with an advanced tumor stage and good prognosis/survival in melanoma, neuroblastoma, and breast cancer.26-28 cDNA microarray analyses identify higher PRAME expression in anthracycline-containing, chemotherapy-resistant diffuse large B-cell lymphoma, and it is linked to a poor prognosis.29 Functional investigation revealed that PRAME could interfere with retinoic acid receptor (RAR) signaling in solid tumors, inhibiting myeloid differentiation in normal hematopoietic and leukemic progenitor cells.30, 31 Osteosarcoma is considered to be a differentiation disorder of mesenchymal stem cells.5, 32 In the current study, we observed that several members of the wingless (WNT), fibroblast growth factor (FGF), and bone morphogenic protein (BMP) pathways that are instrumental for osteoblast differentiation have been dysregulated in osteosarcoma; and, as such, the mechanism by which PRAME affects the dedifferentiation of osteosarcoma still must be elucidated. The first report of CSAG2 overexpression was in a paclitaxel-resistant ovarian carcinoma cell line and was predictive of a poor clinical outcome in ovarian cancer.33 Currently, paclitaxel is used in the treatment of osteosarcoma; thus, the detection of CSAG2 in osteosarcoma may contribute to individualized treatment with paclitaxel.

The expression of MAGEA correlates with distant metastasis and predicts a poor clinical outcome in patients with osteosarcoma. The immunohistochemical staining patterns of MAGEA coincide with these statistical analyses, indicating that it is more than just a biomarker. Advanced-stage tumors tend to have higher expression of MAGEA, as observed in melanoma and neuroblastoma, and more intense expression is observed in lung metastases than in the primary tumor. Even in the same tumor, more MAGEA-positive cells were observed at the periphery of the tumor, which is the proliferatively active region, indicating that MAGEA may play instrumental roles in the progression of this disease. Although the functions of CT antigens in osteosarcoma have not been elucidated, cumulative evidence indicates that CT antigens play important roles in tumor proliferation, drug resistance, invasion, and metastasis. It has been demonstrated that MAGEA2 protein strongly down-regulates p53 transactivation function, and an association between MAGEA expression levels and resistance to etoposide treatment was demonstrated in melanoma biopsy cell lines with wild-type p53.34 Multiple MAGE proteins, including MAGEA3, reportedly inhibited the function of p53 by forming complexes with Kap-1, a known corepressor of p53, or by blocking its interaction with chromatin.35 In addition, small interfering RNA suppression of these MAGE genes induces apoptosis, causing increased p53 expression in vitro. In pituitary tumors, down-regulation of MAGEA3 resulted in p53 transcriptional induction and p21 accumulation.34, 36, 37 MAGEA11 activates a hypoxic response and promotes tumor angiogenesis by inhibiting the hypoxia-inducible factor prolyl hydroxylase 2 (PHD2) and stabilizing hypoxia-inducible factor.38 MAGE-C1/CT7 and MAGE-A3 are essential for the survival of clonogenic myeloma precursors, and knock-down of these proteins improved the response of myeloma cells to conventional therapies.39 MAGEA proteins also are present in tumorigenic mesenchymal stem cells and melanoma stem cells.40, 41 These studies suggest that CT antigens are not merely bystanders but are promoters of tumorigenic transformation that potentially may be important in the genesis and maintenance of tumor stem cells. Thus, CT-antigen–based immunotherapy may selectively target the tumor stem cell population.

In conclusion, CT antigens represent ideal targets for therapeutic intervention because of their restricted expression in tumors. The frequent expression of specific CT antigens detected in osteosarcoma indicates that osteosarcoma should be added to the list of growing CT-rich tumors. Furthermore, identification of MAGEA proteins as predictors of tumor progression and poor clinical outcome raises the interesting possibility that MAGEA may serve as a novel biomarker for diagnosis and as a target of immunotherapy. However, limitations of the current study group and an absence of antibodies specific for MAGEA subfamilies prevented us from identifying MAGEA members that are crucial for osteosarcoma oncogenesis. Clearly, their functions in osteosarcoma must be elucidated further.


  1. Top of page
  2. Abstract

This study was supported by the National Natural Science Foundation of China (serial no. 30872967, no. 81072193, and no. 81001194) and the Guangdong Natural Science Fund (serial no. 9151008901000096).


The authors made no disclosures.


  1. Top of page
  2. Abstract