Immunohistochemical studies of cellular markers can aid in establishing or confirming a histopathological diagnosis, forming a prognosis, and guiding therapeutic decisions1-10. For example, immunohistochemistry is commonly used to determine human epidermal growth factor receptor-2 (HER2) status, which is associated with a more aggressive phenotype in breast cancer, and to identify patients who are most likely to benefit from adjuvant trastuzumab treatment10-12.
Nitric oxide is a biomolecule that regulates neural transmission, vasodilation, and immune function. It is a product of the conversion of L-arginine to L-citrulline by nitric oxide synthase (NOS). There are three isoforms of NOS: NOS1, NOS2, and NOS313.
Nitric oxide is a highly reactive free radical and a short-lived molecule. It reacts with superoxide anions to yield peroxynitrite, which causes oxidative damage and modifies DNA14,15. In addition, peroxynitrite increases cyclooxygenase (COX) activity16. Nitric oxide also oxidizes to nitrogen dioxide, which induces DNA damage by nitrosative deamination and DNA strand breakage and inhibits DNA repair, leading to gene mutations and cancer14,15,17-21. The accumulation of nitrotyrosine, a stable product of the addition of a nitro group to the benzene ring of tyrosine by peroxynitrite in cells, indicates the formation of peroxynitrite and its interaction with tyrosine.
Certain neoplasms, including tumors of the gynecological, nervous, vascular, and lymphatic systems and of the breast, lung, prostate, stomach, pancreas, bladder, and colon, have altered expression of nitric oxide1-4,22-30.
Nitric oxide has both protumor and antitumor effects. Nitric oxide production in tumor-infiltrating macrophages31, inhibition of platelet aggregation32,33, and induction of apoptosis in nitric oxide expressors34,35 may potentiate nitric oxide antitumor activity. In contrast, its protumor effects are associated with the expression of NOS2 in endothelial cells of tumor vessels and in cancer cells. These findings suggest that cancer cells use this isoform to regulate tumor vascularization by inducing vascular endothelial growth factor (VEGF), thereby increasing tumor blood flow and vascular permeability25,36-38.
Cyclooxygenase is an enzyme that regulates the synthesis of prostaglandins. There are two isoforms: COX-1 and COX-2. The latter is not detected in most normal tissues, but it is highly inducible by inflammatory and mitogenic stimuli. COX-2 has been implicated in tumorigenesis5,39-41; it regulates angiogenesis by inducing the production of proangiogenic factors, such as VEGF and proangiogenic prostaglandins, and stimulates endothelial migration and tube formation6,41-44.
Neovascularization in neoplasms can influence tumor prognosis, and a microvessel study can be performed with use of CD34 or L-selectin, which is a transmembrane cell surface glycoprotein on endothelial cells45,46.
The purposes of this study were to analyze the association of NOS, COX-2, and nitrotyrosine with microvessels in chondrosarcomas; the association of NOS, COX-2, nitrotyrosine, and CD34 with histological grade; and their association with the prognosis of chondrosarcomas.
According to the A.C. Camargo Hospital (São Paulo, Brazil) Registry, 170 patients with a diagnosis of chondrosarcoma were managed between 1953 and 2004. A retrospective analysis of the cases of these patients was performed. Primary and secondary bone and soft-tissue chondrosarcomas were included in this study. Tissue samples were preferably collected from surgical resections, which were available for ninety-four (93.1%) of the 101 patients included in the study. Biopsy specimens were obtained from seven patients (6.9%) for whom tissues from surgical resections were unavailable. Metastatic lesions were excluded.
Paraffin blocks, which were fixed in formalin, were obtained from the Anatomic Pathology Department files, and new sections were sliced and stained with hematoxylin and eosin. The cases of all 170 patients were reviewed by two pathologists (I.W.C. and F.A.S.) to confirm the histological diagnosis; the cases were also reviewed by three orthopaedic oncologists (S.A.N., W.T.C., and L.A.A.). Thirty-seven cases (21.7%), for which there was no consensus on clinical, imaging, and histological features, were excluded.
The tumors were classified and graded according to the World Health Organization Classification of Tumors47. The histological grade was assigned according to the highest grade that was observed in each tumor. As a result, 101 patients with chondrosarcoma were included. Fifty-seven patients (56.4%) were male, and forty-four (43.6%) were female. The patients had a mean age of 39.8 years (range, eleven to seventy-three years). Sixty chondrosarcomas (59.4%) were in an axial location; thirty-two (31.7%), in the lower limb; and nine (8.9%), in the upper limb.
Surgical margins were considered on the basis of histological examination. Intralesional surgery, marginal excision, marginal amputation, and contaminated procedures (when the tumor compartment was violated) were classified as positive margins. Uncontaminated wide excision, wide amputation, radical resection, and radical amputation were classified as negative margins.
Five normal cartilage tissue samples were obtained from patients without chondrosarcoma who underwent an amputation for vascular problems, to evaluate NOS1, NOS2, NOS3, nitrotyrosine, and COX-2 expression in normal cartilage.
Data on other comorbidities were not collected.
This study was approved by our Institutional Ethical Committee.
Tissue Microarray Construction
To construct a tissue microarray, a representative area was selected from each of the 101 chondrosarcoma specimens (Table I). This technique allows multiplex histological analysis on the same slide, and it accelerates the analysis of multiple specimens. Tissue microarray is not appropriate when the goal is to evaluate the pattern of vascularization in peritumoral and intratumoral areas. It is necessary to review additional slides to evaluate vascular patterns.
Formalin-fixed, paraffin-embedded biopsy, or surgical resection specimens were confirmed to contain chondrosarcoma cells and were sampled with use of a Tissue Microarrayer (Beecher Instruments, Silver Spring, Maryland). Tissue cores from each specimen were punched and arrayed in duplicate on a recipient paraffin block. Each 1.0-mm-diameter core was spaced 0.2 mm apart from another. Tissue microarray sections were mounted on coated slides for subsequent ultraviolet cross-linking (Instrumedics, Hackensack, New Jersey); the slides were dipped in a layer of paraffin to prevent oxidation and were kept in a freezer at —20°C.
Immunohistochemistry
Immunohistochemistry was performed as previously described48. The staining for NOS1, NOS2, NOS3, COX-2, and nitrotyrosine was performed on two slides of two different tissue microarray sections, representing fourfold redundancy for each case. The second slides were twenty-five sections deeper than the first, resulting in at least 250 µm of distance between the two sections and guaranteeing different cell samples for each section level. We stained high-grade sarcoma areas in the dedifferentiated chondrosarcoma cases. Immunohistochemistry for CD34 was performed in traditional whole sections to evaluate the pattern of vascularization in peritumoral and intratumoral areas. Immunohistochemistry of normal cartilage tissue was performed with use of traditional whole sections.
Tissue microarray slides were deparaffinized for thirty minutes at 60°C with xylene rinses and dehydrated in a series of alcohol and water rinses. The slides were prepared with use of the avidin-biotin immunoperoxidase method. Endogenous peroxidase activity was blocked with 3% H2O2 in absolute methyl alcohol for forty minutes after pressure cooker pretreatment in citrate buffer (pH 6.0) at 98°C for fifteen minutes. The slides were also blocked for avidin-biotin (Biotin-Blocking System; Dako, Carpinteria, California) and protein (Protein Block, Serum-Free; Dako) for twenty minutes each.
Immunostaining was performed at room temperature with NOS1 (clone N31030, 1:50 dilution; Transduction, Lexington, Kentucky), NOS2 (clone N32020, 1:25 dilution; Transduction), NOS3 (clone N30020, 1:100 dilution; Transduction), nitrotyrosine (1:1000 dilution; Dako), COX-2 (clone 4H12, 1:3000 dilution; Novocastra, Benton Lane, Newcastle upon Tyne, United Kingdom), and CD34 (QBEnd 10, 1:50 dilution; Dako) antibodies. The tissues were exposed to biotinylated mouse immunoglobulin (Ig) G, incubated with the streptavidin-biotin peroxidase method (LSAB kit; Dako) and 3,3'-diaminobenzidine, and counterstained with Harris hematoxylin. Appropriate quality control and quality assurance steps, including positive and negative tissue controls with each assay, were implemented. Negative controls were established by applying all primary antibodies to known negative tissue samples.
Evaluation of Immunohistochemistry
Stains were evaluated independently by two observers (I.W.C. and F.A.S.). The immunostainings with antibodies against NOS, COX-2, and nitrotyrosine were evaluated with use of a combined score, multiplying the intensity and frequency of the cells that were stained. Staining intensity was divided into four groups, with 0 indicating negative; 1, weak; 2, moderate; and 3, strong. The percentage of immunostained tumor cells was ranked into four groups, with 1 indicating <10%; 2, 10% to 50%; 3, 51% to 90%; and 4, >90%. The final scores were calculated with arithmetic means of the slides. The scores for NOS and nitrotyrosine were divided into two groups—negative (0 to 1.9) and positive (2 to 12)—as were the scores for COX-2—negative (0 to 4.9) and positive (5 to 12). For CD34 staining, the cells were divided into two groups: intratumoral and peritumoral microvessels.
Statistical Methods
Statistical analyses were performed with use of the SPSS for Windows program (version 12.0; SPSS, Chicago, Illinois).
This study analyzed associations between NOS and nitrotyrosine; between NOS, COX-2, and nitrotyrosine and CD34; and between markers and histological grade. The associations were analyzed by chi-square or Fisher exact tests, when appropriate.
Analyses of local and overall survival rates were performed with use of the Kaplan-Meier method, and the curves were compared by the log-rank test. The Kaplan-Meier method handles incomplete follow-up and proportionally introduces short observations on further analysis. We did not exclude patients with a short duration of follow-up, to analyze whether NOS is an independent prognostic factor.
Local disease-free survival was defined as the period between the date of surgery and the date of local recurrence or most recent follow-up (for patients without local recurrence). Overall survival was defined as the period between the date of surgery and the date of death by any cause or the last information on a patient (for patients who were alive at the last follow-up visit).
Multivariable survival analysis was performed with use of Cox proportional hazards regression analysis.
Statistical significance was defined as a p value of 0.05.
For statistical analysis, because of histological, clinical, and prognostic features, clear cell chondrosarcoma was included in the grade-I group; and mesenchymal and dedifferentiated chondrosarcomas were included in the grade-III group47,49.
Source of Funding
This study was sponsored by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP).
The follow-up period ranged from three weeks to 36.4 years (mean, 7.25 years; median, 4.7 years). At the end of the study, fifty-one patients (50.5%) were alive without recurrence of chondrosarcoma, three (3%) were alive with chondrosarcoma (local and/or distant disease), thirty (29.7%) had died because of the chondrosarcoma or treatment complications (including death within thirty days after tumor resection), seven (6.9%) had died because of other causes, and ten (9.9%) had not returned for follow-up.
All normal cartilage samples were negative for NOS1, NOS2, NOS3, nitrotyrosine, and COX-2.
In the analysis of the chondrosarcoma cells, NOS1 was absent in thirty-four (37%) and present in fifty-eight (63%) of the specimens from patients for whom data were available. NOS2 was absent in sixty-three (67%) and present in thirty-one specimens (33%); NOS3, in twenty-two (23.4%) and seventy-two specimens (76.6%), respectively; nitrotyrosine, in eighteen (24.3%) and fifty-six specimens (75.7%); and COX-2, in fifty-six (60.2%) and thirty-seven specimens (39.8%). CD34 was peritumoral in seventy-four specimens (73.3%) and intratumoral in twenty-seven specimens (26.7%).
Nitric oxide participates in nitrotyrosine production, and NOS is the enzyme that catalyzes nitric oxide production. The presence of NOS and nitrotyrosine in the same tissue provides evidence that NOS is catalytically active; thus, we analyzed NOS and nitrotyrosine expression in chondrosarcoma cases. Our findings suggest that all three NOS isoforms are active because of their association with nitrotyrosine (Table II).
The expression of NOS1, NOS2, NOS3, nitrotyrosine, and COX-2 has been associated with angiogenesis; thus, we analyzed these markers with regard to CD34 expression. Our findings suggest that they are related to angiogenesis because of their association with CD34 expression (Table III).
We found that nitrotyrosine, COX-2, and CD34 were related to the histological grade of the chondrosarcoma. Of the nitrotyrosine-positive cases, 12.5% were found in grade-I lesions; 55.4%, in grade-II lesions; and 32.1%, in grade-III lesions (p = 0.022). Of the COX-2-positive cases, 8.1% were found in grade-I; 51.4%, in grade-II; and 40.5%, in grade-III lesions (p = 0.014). Of the cases with intratumoral expression of CD34, 14.8% were found in grade-I; 44.4%, in grade-II; and 40.7%, in grade-III lesions (p = 0.028). But nitrotyrosine, COX-2, and CD34 were not related with overall prognosis (p = 0.064, p = 0.143, and p = 0.581, respectively). With the numbers studied, the expression of NOS1, NOS2, and NOS3 did not show any significant association with histological grade (p = 0.348, p = 0.075, and p = 0.610, respectively) (Table IV).
The five-year local disease-free survival rate of patients with chondrosarcoma was 70.4%. Considering clinical, surgical, and histopathological features, positive surgical margins (mean survival, 71.62%; 95% confidence interval, 48.45% to 94.79%) and NOS2-positivity (mean, 69.19%; 95% confidence interval, 46.65% to 91.72%) were related to a lower rate of local disease-free survival (p = 0.015 and p = 0.038, respectively).
The five-year overall survival rate was 66.7%. Among clinical, surgical, and histopathological variables, the presence or absence of metastasis at admission, the Musculoskeletal Tumor Society (MSTS) stage (according to the Enneking system50), expression of NOS1 (mean survival, 74.01%; 95% confidence interval, 60.23% to 87.79%), and expression of NOS2 (mean, 54.08%; 95% confidence interval, 36.89% to 71.26%) were associated with a lower overall survival rate (p < 0.001, p < 0.001, p = 0.007, and p < 0.001, respectively) (Fig. 1).
On multivariable analysis, we analyzed surgical margins and expression of NOS2 because they showed a relationship with local disease-free survival, and they demonstrated independent prognostic impacts on local disease-free survival (Table V).
Finally, we analyzed NOS1, NOS2, and the presence or absence of metastasis at admission because they were associated with overall survival. As a result, positive NOS2 expression and the presence or absence of metastasis at admission were independent variables that were associated with lower overall survival (Table VI).
Because positive expressions of NOS1 and NOS2 were related to a lower overall survival rate, we studied their combined effects. Our analysis of NOS1 and NOS2 expression demonstrated that they were related to a lower overall survival rate when both were present (Table VI and Fig. 2).
The prognosis for patients with chondrosarcoma varies and depends on tumor aggressiveness51. The challenge is to develop a histological grade and diagnosis based on the results of a biopsy and to plan a satisfactory surgical regimen52.
This study showed that positive surgical margins were related to a worse rate of local disease-free survival but not to a worse overall survival rate. The relationship between positive surgical margins and local recurrence has been observed by Gitelis et al.53, Lee et al.54, and Rizzo et al.51, all of whom showed a correlation with survival rate. According to Lee et al.54, patients who were managed with a wide resection had significantly higher survival rates than those who were managed with a marginal or intralesional resection. Gitelis et al.53 found that adequately treated patients had lower recurrence rates and better survival statistics. Rizzo et al.51 demonstrated that local recurrence and risk of death that was caused by disease increased with inadequate resections.
Histological grade is related to metastatic disease-free survival and overall survival, according to Gitelis et al.53, and it is an independent variable that is associated with metastatic disease-free survival and overall survival, according to Lee et al.54; those studies grouped patients with grade-II and grade-III chondrosarcoma together. Reith et al.55 observed a relationship between overall survival and histological grade, placing patients with grade-I chondrosarcoma in the same group as patients with grade-II lesions.
Grade is generally accepted as a predictor of survival in chondrosarcoma; however, in our study, the three histological groupings had no relationship with survival rates. This result could have been affected by the low number of patients and by the fact that the groups were analyzed separately; in addition, there were few patients with grade-I or III chondrosarcoma. Our oncological hospital receives a high proportion of complex and previously treated patients. This referral pattern may have had an influence on the high proportion of grade-II chondrosarcomas. Survival rates could be affected depending on the group in which the patients with grade-II disease are placed.
Di Cesare et al. reported that Swarm rat chondrosarcoma cells and cultured human chondrosarcoma cells express NOS256. Our study confirms that human chondrosarcoma cells express NOS2 as well as NOS1 and NOS3. Because there was nitrotyrosine expression that was associated with NOS1, NOS2, and NOS3, we suggest that NOS was catalytically active, leading to nitric oxide and peroxynitrite synthesis.
Tsujii et al.41 and Uefuji et al.7 showed that COX-2 and NOS2 correlate with angiogenesis. Hara and Okayasu observed an association between COX-2 and microvessel density in astrogliosis and between NOS2 and microvessels in high-grade astrocytomas6. A study of rat solid tumors showed that NOS2 expression enhanced vascular permeability and increased blood flow57. Furumatsu et al. demonstrated that a human chondrosarcoma cell line released VEGF, which stimulated the proliferation and migration of endothelial cells58. Our study confirms that COX-2 and NOS2 are related to microvessels (with use of CD34). Furthermore, NOS1, NOS3, and nitrotyrosine are also associated with angiogenesis (with use of CD34).
We found that nitrotyrosine, COX-2, and CD34 overexpression was linked to higher histological grades in chondrosarcomas. Our study confirms the association between angiogenesis and histological grade, as described by Ayala et al.59, who found a direct relationship between histological grade and pericartilage vessels.
Ekmekcioglu et al. observed a correlation between NOS2 and nitrotyrosine expression within the same tumor4. The detection of NOS and nitrotyrosine immunoreactivity in the same tissue provides evidence that NOS is catalytically active4,20,25,28,35,60,61. According to this model, and because nitric oxide is a short-lived molecule, we determined the presence of nitric oxide indirectly by measuring its coproduct (nitrotyrosine) and the enzymes (NOS) that are related to its production. Our findings suggest that NOS is active because of the association of NOS1, NOS2, and NOS3 with nitrotyrosine.
Although nitrotyrosine, COX-2, and CD34 showed a relationship with histological grade in chondrosarcomas, they did not show associations to local or overall survival rates in our study.
Studies have shown that overexpression of NOS and COX-2 is associated with a poor prognosis. Endo et al. demonstrated that the presence of COX-2 portended a poor prognosis in chondrosarcomas8; Hara and Okayasu demonstrated an association between COX-2 overexpression and a poor outcome in astrocytic gliomas6. Ekmekcioglu et al. concluded that the presence of NOS2 and nitrotyrosine during the posttreatment phase in melanomas predicted a poor rate of survival4. Thomsen et al. showed that NOS is expressed in human breast tumors, in which its presence correlates with tumor grade2. Cobbs et al. observed high levels of NOS in astrocytic tumors, where the highest levels were found in higher-grade tumors1. Although Rahman et al. did not identify any impact of NOS2 and COX-2 expression on disease-free survival in patients with hepatocellular carcinoma, patients with hepatitis C virus-positive hepatocellular carcinoma who were negative for both NOS2 and COX-2 had a significant survival advantage over other patients, who had variable expression of NOS2 and COX-262.
Nevertheless, some studies have not found correlations between NOS or COX-2 and prognosis. Kong et al. reported that COX-2 and NOS2 expressions were not related to prognosis in patients with pancreatic cancer63. Dickens et al. studied COX-2 expression in pediatric sarcomas, but COX-2 staining intensity did not show a significant correlation with known prognostic variables, such as histological classification or stage39.
Our study showed that NOS2 is an independent prognostic variable and that its expression is related to lower local disease-free survival; further, NOS1 and NOS2 expression is a dependent variable and is associated with worse overall survival.
According to Ayala et al., pericartilage vessels support tumor growth, whereas intracartilage vessels regulate the acquisition of metastatic potential by cartilage tumors59. We did not, however, find any association between microvessels and survival rates. These results support the finding of Sulh et al. that other factors not directly related to angiogenesis can be involved in prognosis64.
In conclusion, chondrosarcomas may express NOS1, NOS2, NOS3, nitrotyrosine, COX-2, and CD34. NOS1, NOS2, NOS3, nitrotyrosine, and COX-2 are associated with angiogenesis. Expression of nitrotyrosine, COX-2, and CD34 is related to higher histological grades in chondrosarcomas. Positive surgical margins and expression of NOS2 are independent factors that are associated with lower rates of local disease-free survival. NOS2 expression is an independent factor that is linked to worse rates of overall survival; expression of NOS1 or NOS2 is related to lower rates of overall survival, and because the rate is worse when both are present, we propose that they have synergistic effects in chondrosarcomas.