Chondrosarcoma, the second most frequent primary malignant bone tumor, is classified into 3 grades according to histologic criteria of malignancy. However, a low-grade lesion can be difficult to distinguish from a benign enchondroma, whereas some histologically low-grade lesions may carry a poor prognosis. The receptor for advanced glycation endproducts (RAGE) and its ligand, high-mobility group box-1 (HMGB1), was quantified in enchondromas and chondrosarcomas to determine whether these markers were associated with histological malignancy and prognosis.
Enchondromas (n = 20) and typical chondrosarcomas (n = 39) were evaluated for RAGE, endogenous secretory RAGE (esRAGE, a splice variant form), and HMGB1 protein expression by immunohistochemistry including laser confocal microscopy. The content of esRAGE in resected specimens was measured with an enzyme-linked immunosorbent assay. Associations of these molecules with histology and clinical behavior of tumors were analyzed.
Expression of esRAGE and HMGB1 was observed in all specimens. The numbers of cells positive for esRAGE and HMGB1 expression were positively associated with histologic grade. Expression of esRAGE was significantly higher in chondrosarcomas than in enchondromas (P < .001). Tissue esRAGE content was also significantly higher in grade 1 and 2 chondrosarcomas than enchondromas (P = .0255 and P = .008, respectively). High expression of esRAGE in grade 1 chondrosarcoma was associated with subsequent recurrence (P = .0013), lung metastasis (P = .0071), and poor survival (P < .001).
Chondrosarcomas are the second most frequent primary tumor of bone.1–4 On the basis of histologic features, typical chondrosarcomas are divided into grades 1 to 3,3, 5, 6 a distinction long considered to correlate with clinical behavior.3, 7 However, distinguishing a low-grade chondrosarcoma from a benign enchondroma solely from microscopic cytologic features is often impossible.8, 9 Grade 1 chondrosarcomas tend toward slow growth, infrequent metastasis, and an overall favorable prognosis. However, because an occasional histologically low-grade tumor shows metastatic potential associated with poor prognosis, new diagnostic and prognostic markers for chondrosarcomas are needed.
The receptor for advanced glycation endproducts (RAGE) is a multiligand receptor that has an extracellular region, consisting of 1 V-type and 2 C-type immunoglobulin-like domains, a transmembrane region, and a short C-terminal intracellular portion.10, 11 The V-domain of RAGE is critical for binding of various ligands including advanced glycation endproducts (AGE),10–14 S100,15, 16 β-amyloid,17, 18 Mac-1,19 and high-mobility group box-1 (HMGB1),19–23 which have been implicated in diabetic vascular complications, neurodegenerative disorders, proinflammatory reactions, and cancer.18 Among these ligands, HMGB1 has been linked with tumor growth, invasion, and metastasis through interactions with RAGE20, 24; RAGE expression has been examined in gastric,25 colorectal,26 prostatic,27, 28 lung,26 and breast26 cancers.
Recently, Yonekura et al.29 identified a novel splice variant of RAGE, endogenous secretory RAGE (esRAGE), which lacked a transmembrane region and represented a soluble form. This variant is considered to act as a modulating factor opposing the RAGE-associated conditions mentioned above and various diseases, including cancers. Immunohistochemistry using domain-specific antibodies for RAGE or esRAGE showed esRAGE to be distributed in a wide variety of normal human organs and tissues.30 However, examination comparing RAGE to esRAGE expression has not yet been reported in any type of cancer, including bone tumors.
In this study we investigated the expression of RAGE and esRAGE as well as HMGB1 in surgical specimens of enchondroma and chondrosarcomas. We sought associations between expression of these molecules and the clinical behavior of these tumors.
MATERIALS AND METHODS
Resection specimens containing enchondromas of long bones (n = 20) and specimens including typical chondrosarcomas (n = 39, grade 1, 24 cases; grade 2, 13; and grade 3, 2). Of 39 specimens in chondrosarcoma, 30 primary tumors and individual 9 recurrent lesions were obtained from 30 patients. The data from the recurrent tumors were not used for further clinical outcome analyses. Specimens studied were obtained from 50 patients (27 men and 23 women), all treated at the Kanazawa University Hospital from 1988 to 2004. Locations of enchondromas and chondrosarcomas are shown in Table 1.
Table 1. Patient Characteristics
Median age, y (range)
After histologically establishing the diagnosis, all enchondromas were treated by curettage and bone grafting. Chondrosarcomas were treated by en bloc excision. The study was approved by the Ethics Committee for Medical Studies at the Kanazawa University Graduate School of Medical Science. Written informed consent was obtained from each subject or their guardian.
Specimens used had been fixed in 20% formalin and embedded in paraffin. They were retrieved from the surgical pathology files of the Pathology Section of Kanazawa University Hospital, School of Medicine, Kanazawa University, Kanazawa, Japan. For each case 1 representative block of formalin-fixed, paraffin-embedded tumor tissue was selected. All specimens were decalcified. We found that the decalcification step did not influence the IHC for any of the stains. All sections were cut at a 4-μm thickness for IHC. As a control we used an uninvolved epiphyseal cartilage resected for osteosarcoma from a distal femur in an 8-year-old girl, because the epiphyseal chondrocyte has an active growth and differentiation potential, and some primary bone tumors such as osteosarcoma and chondrosarcoma often occur around the growth plate. Immunostaining procedures included a new microwave technique for antigen retrieval as described previously.31 Primary antibodies used in this study included a goat polyclonal antibody against the V domain of RAGE proteins (Chemicon, Temecula, Calif; dilution, 1:800) that recognized both full-length and esRAGE; a rabbit polyclonal antibody against the C-terminal 16 amino acids of esRAGE (dilution, 1:1000), recognizing only esRAGE; a rabbit polyclonal antibody against 20 C-terminal amino acids of RAGE representing the intracellular domain (Santa Cruz Biotech, Santa Cruz, Calif; dilution, 1:100), which recognizes the full-length RAGE29; and goat polyclonal antibody against HMGB1 (Santa Cruz Biotech; dilution, 1:50). Peroxidase-labeled polymers of antibodies raised against rabbit polyclonal antibody (EnVision, DAKO, Carpinteria, Calif) or goat polyclonal antibody (Histone, Simple Stain, Nichirei, Tokyo, Japan) were used as secondary antibodies. After visualization of the reaction product, sections were counterstained with Meyer hematoxylin and then coverslipped for microscopic observation. Areas with apparent brown staining were evaluated for positive immunostaining based on the manufacturer's instructions (Envision System, DAKO). In each case all of the positive and negative cells were counted in 5 nonoverlapping visual fields at ×200 magnification. The labeling indexes (LIs) for esRAGE and HMGB1 were calculated as a percentage of positive cells among the total number of cells counted, at least 250 tumor cells.32
Enzyme-Linked Immunosorbent Assay (ELISA) for esRAGE
We examined frozen tissue samples from 10 cases of enchondroma and 11 cases of chondrosarcoma (5 grade 1 cases, 6 grade 2 cases) from among the cases used for immunohistochemistry. Tissue samples were dissected and homogenized in a lysis buffer (1% Nonidet P-40, 0.5% deoxycholate, 10 mM EDTA, 0.1% sodium dodecyl sulfate, and 1.0 mM phenylmethylsulfonyl fluoride) also containing a protease inhibitor cocktail (Sigma-Aldrich, St. Louis, Mo). The homogenate then was cleared of debris by centrifugation at 12,000 rpm for 30 minutes at 4°C. The supernatant was collected and its total protein concentration was determined (BCA assay kit, Model 550, BioRad Laboratories, Hercules, Calif). Concentrations of esRAGE protein were determined using a human esRAGE (ELISA) system (B-Bridge International, Sunnyvale, Calif).33 Twenty micrograms of protein (100 μL of homogenate) were used for this assay. Content of esRAGE protein in tumor tissue is reported as picograms per gram of protein.
The procedure for tissue section treatment was the same as for immunohistochemistry. Primary antibodies used included goat polyclonal antibody against the V domain of RAGE (dilution, 1:800), rabbit polyclonal antibody specific for esRAGE (dilution, 1:1000), and goat polyclonal antibody against HMGB1 (dilution, 1:50). Secondary antibodies were chicken antirabbit IgG labeled with Alexa Fluor 488 (A-21441, Molecular Probes, Eugene, Ore; dilution, 1:200) for esRAGE and donkey antigoat IgG labeled with Alexa Fluor 556 (A-11056, Molecular Probes; dilution, 1:200) for the V domain of RAGE and HMGB1. A Zeiss confocal microscope (objective lens power, ×63) and BioRad 1024 software were used to produce images.
Analyses were performed using Stat View (v. 5.0; SAS Institute, Cary, NC). LIs of esRAGE and HMGB1 obtained by immunostaining and concentrations of esRAGE protein according to ELISA were compared with histologic tumor grades of enchondroma and chondrosarcomas by Fisher exact tests. Correlations between the 2 variables (esRAGE LI and HMGB1 LI) were analyzed using the Pearson correlation coefficient. Associations of esRAGE and HMGB1 LI with subsequent tumor recurrence and distant metastasis were analyzed by Fisher exact tests. Univariate analysis of time to death as a result of chondrosarcomas was carried out using a product-limit procedure (Kaplan-Meier method) and a log-rank test, based on esRAGE and HMGB1 LI.
Immunohistochemistry for RAGE, esRAGE, and HMGB1
We detected expression of esRAGE (Fig. 1A-E) and HMGB1 (Fig. 1F-J) in all specimens examined in this study, including normal epiphyseal cartilage, enchondroma, and chondrosarcoma. In an epiphyseal cartilage, the expressions of esRAGE and HMGB1 were observed in the chondrocytes from the proliferating through the hypertrophic zones to the calcifying zone, but very weak signals in the resting zone. Contrary to our expectation, reaction product staining intensity was very weak in enchondroma and chondrosarcoma when specimens were stained with antibody against the intracellular domain of RAGE (data not shown). Confocal microscopic distribution of the goat polyclonal antibody against the V domain of RAGE proteins, which recognized both RAGE and esRAGE proteins, was compared with distribution of the esRAGE-specific antibody, indicating that total RAGE signals overlapped with esRAGE signals (Fig. 2A-C). A diffuse cytoplasmic esRAGE staining pattern was observed in all tumor cells, but strong staining occupying the entire cytoplasm was present only in chondrosarcomas. The esRAGE LI was 15.7% in normal epiphyseal cartilage (control), 33.5% ± 11.4% in enchondroma, 67.7% ± 12.1% in grade 1 chondrosarcoma, 79.0% ± 7.7% in grade 2, and 93.6% ± 2.2% in grade 3 (mean ± SD; Fig. 1K). The esRAGE LI was found to increase as tumor grade advanced. Significant differences in esRAGE LI were seen not only between grade 1 chondrosarcomas and enchondromas (P < .0001) but also between grade 1 and grade 2 chondrosarcomas (P = .0039; Fig. 1K). Faint cytoplasmic staining for esRAGE was encountered in scattered chondrocytes of the normal control cartilage.
HMGB1 LI was 6.0% in normal epiphyseal cartilage, 42.2% ± 19.0% in enchondroma, 61.8% ± 19.5% in grade 1 chondrosarcoma, 75.9% ± 11.1% in grade 2, and 77.9% ± 7.7% in grade 3 (Fig. 1L). HMGB1 LI also was found to increase as tumor grade advanced. HMGB1 LI was significantly higher in grade 1 chondrosarcoma than in enchondroma (P = .0006), and a significant difference in HMGB1 LI was noted between grade 1 and grade 2 chondrosarcomas (P = .024; Fig. 1L). A significant positive correlation was found between HMGB1 and esRAGE expression in enchondromas and chondrosarcomas (y = 0.532x + 27.283, r = 0.547, P < .0001; Fig. 1M).
HMGB1-positive cells were also stained by esRAGE-specific antibody. In contrast, signals for esRAGE and HMGB1 were not colocalized in chondrosarcoma cells examined by confocal microscopy (Fig. 2D-F). A cultured chondrosarcoma cell line, H-EMC-SS, showed diffuse cytoplasmic staining by esRAGE- and HMGB1-specific antibody, but these signals did not necessarily merge (data not shown).
Quantitative Evaluation of esRAGE Protein by ELISA
To quantitatively evaluate the esRAGE expression in surgical specimens, an esRAGE ELISA was performed in extracts from enchondroma (n = 10), and from grade 1 (n = 5) and grade 2 (n = 6) chondrosarcomas. Concentrations of esRAGE protein in enchondroma were significantly lower than in grade 1 or grade 2 chondrosarcomas (P = .0255 and P = .008, respectively) (Fig. 3). These findings support observations made by esRAGE immunostaining.
Correlation Between esRAGE Expression and Clinical Outcome of Chondrosarcomas
A significant association was present between esRAGE LI and subsequent tumor recurrence in grade 1 chondrosarcoma (P = .0013; Fig. 4A). Mean esRAGE LI in chondrosarcomas with metastasis was significantly higher than in those without metastasis (P = .0071; Fig. 4D); this difference was most evident for grade 1 chondrosarcomas (P = .0033; Fig. 4C). A significant difference in survival time was evident between high (≥70%) and low (< 70%) esRAGE LI in patients with chondrosarcomas (P = .0017; Fig. 5A). HMGB1 LI was not significantly associated with clinical features or prognosis in patients with chondrosarcoma (Figs. 4A-D, 5B).
RAGE has been associated with prognosis in several types of cancers, such as those of lung,26 breast,26 and prostate,27, 28 as well as melanoma.26 However, RAGE expression and its clinical importance in bone tumors had not yet been analyzed. In this study we selectively detected protein expression of esRAGE and RAGE as well the expression of their ligand, HMGB1, in human enchondroma and chondrosarcoma specimens. Recently, esRAGE has been identified as a novel splice variant of RAGE mRNA coding for a soluble form of RAGE.29 Considered a modulating factor in cancer phenotypes, esRAGE can trap ligands outside of cells without initiating signal transduction within cells. When Cheng et al.30 performed a systemic analysis of esRAGE expression in various human tissues and organs, esRAGE expressions showed patterns distinctive for individual organs including brain, thyroid, salivary gland, lung, pancreas, and others. We hypothesized that a balance between RAGE and esRAGE expression might be a factor modifying RAGE-associated malignant phenotypes.
The relative expression of esRAGE in chondrosarcoma was associated with histologic grade, recurrence, lung metastasis, and clinical outcome; yet RAGE expression was very weak in both enchondromas and chondrosarcomas. Confocal microscopic analysis comparing reactive to a pan-RAGE antibody and to an esRAGE-specific antibody indicated that the form of RAGE predominantly expressed in bone tumors was esRAGE as opposed to membrane-bound RAGE in bone tumors. The relative expression of HMGB1 in chondrosarcomas was associated with histologic grade but not with lung metastasis or clinical outcome. Colocalization of esRAGE and HMGB1 was not seen, suggesting that their actions were distinct and that esRAGE could not take part in ligand-initiated signal transduction. Mechanisms causing esRAGE and HMGB1 up-regulation in chondrosarcoma remain to be investigated.
We next focused on esRAGE expression as a diagnostic marker and as a predictor of clinical features including prognosis. Histopathologically, chondrosarcomas are divided into 3 grades based on cellularity, atypia, and pleomorphism.3–5 The multiple subtypes of chondrosarcoma based on histologic grades have been considered predictive of clinical outcome.3, 7, 34, 35 However, difficulty is often encountered even in distinguishing a low-grade chondrosarcoma from an enchondroma.8, 9 Conversely, even though, grade 1 chondrosarcomas have been considered to have a favorable prognosis, occasional grade 1 lesions develop metastases resulting in a poor survival outcome. Accordingly, many markers and tests have been examined for the ability to distinguish low-grade chondrosarcoma from enchondroma and to predict clinical outcome in chondrosarcoma. Methods have involved DNA content,36, 37 molecular markers (MIB-1,38, 39 p53,40 or hTERT41), cytogenetics,42 and morphometry.43 Previously, Tsuchiya et al.44 reported histologic and clinical findings in 6 cases of borderline chondrosarcoma in which they examined the expression of types I, II, III, V, and VI collagen immunohistochemically. The absence of tumor lobule rims containing collagen types I and V was useful in distinguishing borderline chondrosarcoma from enchondroma. In the present study, 2 grade 1 chondrosarcomas (in the radius of a 58-year-old man, Case 1; in the femur of a 25-year-old woman, Case 2) appeared to be borderline chondrosarcomas. LIs for esRAGE and HMGB1 were, respectively, 58.2% and 61.5% in Case 1 and 64.1% and 69.8% in Case 2. The patients were treated by curettage and bone grafting based on a frozen-section diagnosis of enchondroma. Fortunately, no recurrence or metastasis occurred subsequently in either case. Further investigation is required concerning expression of esRAGE in very low-grade chondrosarcomas such as borderline44 or grade 1/2 lesions.45 Distinguishing an enchondroma from low-grade chondrosarcoma is among the most important aspects of diagnosis of chondrosarcoma, both for planning the surgical procedure and for predicting outcome.
In this study, esRAGE protein was detected in enchondroma and chondrosarcoma by IHC and ELISA. The esRAGE LI rose as the histologic grade of the chondrosarcoma advanced. A few esRAGE-immunopositive cells were observed in normal epiphyseal cartilage used as a control. Staining patterns of esRAGE included a punctuate paranuclear pattern and diffuse cytoplasmic staining. Various intensities of diffuse cytoplasmic staining were observed in enchondroma and chondrosarcoma cells. Strong staining occupying the entire cytoplasm area was detected only in chondrosarcomas. Accordingly, esRAGE LI should be a useful marker for distinguishing low-grade from high-grade tumors. Further, esRAGE LI correlated with tumor recurrence and lung metastasis in grade 1 chondrosarcomas (P = .0013 and .0033). Generally, grade 1 chondrosarcoma is considered to grow slowly, metastasize infrequently, and carry a favorable prognosis.8 However, in our series 4 of the 17 patients with grade 1 chondrosarcoma developed lung metastasis (25%), underscoring that even when the histologic diagnosis is grade 1, a few cases might have potential for progression. All 4 of these cases showed an esRAGE LI of at least 70%. Analysis of cumulative survival rates based on esRAGE LI demonstrated that patients with a high esRAGE of at least 70% may have a poor prognosis. Although further studies are required to understand the functional roles of esRAGE and mechanisms of its induction in chondrosarcoma, esRAGE expression appears to be a promising marker in predicting clinical outcome.
In conclusion, we first found that esRAGE indeed was expressed in tumors of cartilage, and that relative expression of this protein was associated with histologic tumor grade and cumulative survival rate. The esRAGE labeling index could be a useful complement to routine histologic grading as a predictor of lung metastasis and survival outcome. Our data therefore supported esRAGE as a novel diagnostic and prognostic marker in chondrosarcoma. Further large-scale prospective studies are warranted to validate this marker as a routine diagnostic and prognostic tool in the assessment of cartilage neoplasms.