Prognostic Relevance of Increased Angiogenesis in Osteosarcoma

Angiogenesis, the formation of new blood vessels from preexisting ones, is involved in the growth, maintenance, and metastasis of most solid tumors (1). Several reports have demonstrated that neovascularization assessed by intratumoral microvessel density (MVD) correlates with clinicopathological factors and patient prognosis in a variety of tumors (reviewed in ref. 2).

In contrast to carcinomas derived from epithelial cells and hematologic malignancies, few data are available regarding the relevance of angiogenesis and its prognostic impact in osteosarcoma. Osteosarcoma is the most frequent primary malignant bone tumor. The inclusion of cytotoxic polychemotherapy in multimodal treatment strategies has led to dramatic prognostic improvements in patients with osteosarcoma, with survival rates reaching 50% to 80% (3, 4, 5). Recent reports have identified tumor site and size, primary metastases, response to chemotherapy, and surgical remission as independent prognostic factors in osteosarcoma (6). However, the role of angiogenesis in osteosarcoma still remains a matter of debate. Whereas one report by Wang et al. (7) showed evidence for decreased overall survival in osteosarcoma patients with high MVD, Mantadakis et al. (8) failed to demonstrate a correlation between intratumoral neovascularization and long-term outcome in patients with nonmetastatic osteosarcoma. Experimental studies on the role of tumor MVDs revealed a significant correlation between MVD and pulmonary metastasis (9). Increased pretherapeutic levels of vascular endothelial growth factor (VEGF), a well known proangiogenic factor, in patients with osteosarcoma correlated with MVD and metastasis (10). Moreover, VEGF overexpression in osteosarcoma was associated with reduced disease-free and overall survival (11). First experimental studies on the effect of anti-VEGF antibodies in osteosarcoma in a chick embryo chorioallantoic membrane model resulted in growth arrest of tumor xenografts and decreased MVD (12). Furthermore, expression of the VEGF coreceptor neuropilin-2 correlated with increased vascularity and poor prognosis in osteosarcoma patients (13).

In the present study, we examined the clinical significance of intratumoral neovascularization in patients with osteosarcoma before the initiation of polychemotherapy within a multimodal treatment strategy by correlating MVD with relapse-free and overall survival and response to chemotherapy. The aim of the study reported here was to examine the prognostic relevance of angiogenesis in chemotherapy-treated osteosarcoma. We hypothesized that increased angiogenesis, which is a predictor of shortened survival in most cancers, could favor response and survival in a chemotherapy-sensitive tumor such as osteosarcoma.

Paraffin-embedded, pretherapeutic biopsy specimens from 60 patients with non-pretreated osteosarcoma were studied in a retrospective immunohistochemical study. There were 34 males and 26 females with a median age of 17 years (range, 5–51 years) at the time of diagnosis. These patients were treated in the University Hospital of Muenster (Muenster, Germany). Criteria for including patients in the present investigation were sufficient representative pretherapeutic biopsy material for MVD analysis and treatment within a protocol of the Cooperative German-Austrian-Swiss Osteosarcoma Study Group (COSS). Patients were excluded if written consent to the protocol and follow-up were not available.

The 60 patients were enrolled between October 1990 and December 2000 onto a neoadjuvant study of COSS and treated in the University Hospital of Muenster according to COSS protocols (6), which have been approved by the institutional review board. Informed consent was required from all patients or their legal guardians, depending on the patient’s age. Preoperative and postoperative chemotherapy was to be given to all patients according to the COSS protocol active at the time of enrollment. All protocols included high-dose methotrexate (12 g/m²) with leucovorin rescue, doxorubicin (60–90 mg/m²), cisplatin (90–150 mg/m²) and ifos-famide (6–10 g/m²) each per course. Bleomycin, cyclophosphamide, and dactinomycin were used in varying combinations. The scheduled duration of chemotherapy ranged from 24 to 38 weeks. Definitive surgery was scheduled to take place between weeks 9 and 11. Follow-up was started at time of diagnosis and included systemic control for pulmonary metastasis (X-ray) and local control (X-ray, ultrasound, and, if needed, computed tomography or magnetic resonance imaging). A first basic evaluation was performed 4 weeks after termination of treatment. Time points of further controls were scheduled as follows: for local control within the first 4 years, every 3 months and afterward on suspicion; for systemic control in the 1^st and 2^nd years, every 4 to 8 weeks; in the 3^rd and 4^th year, every 8 to 12 weeks; and from the 5^th year to the 10^th year, every 6 months.

The following variables were evaluated for their distribution in the patient cohort and for possible correlations with outcome as described previously (6): patient age, age > 40 years, patient gender, tumor site (osteosarcoma of the extremity versus trunk), primary metastases (at the time of diagnosis), primary or secondary osteosarcoma (patients were assessed as to whether their osteosarcoma represented the first or a subsequent malignancy), histologic subtype, response to chemotherapy [which was assessed according to the six-grade scale of Salzer-Kuntschik et al. (14, 15)], and surgical remission [surgical margins to the lesion according to Enneking et al. (16)]. A good response to chemotherapy was defined as <10% viable tumor cells (response grades, 1–3); a good surgical remission was defined as a radical or wide surgical margin to the lesion.

Serial sections of paraffin-embedded biopsy specimens were processed for immunohistochemical identification of microvascular endothelial cells with an anti-CD31 antibody (clone JC/70A; working dilution, 1:100; Dako, Glostrup, Denmark). Immunohistochemical localization was performed by the alkaline phosphatase/anti-alkaline phosphatase double bridge technique (Dako-APAAP kit; Dako). Before staining, tissue sections were deparaffinized in xylene, rehydrated in a graded EtOH series, and microwaved at 450 W for 7 minutes for antigen retrieval. The primary antibodies were applied overnight at 4°C. Subsequent steps were performed according to the manufacturer’s instructions. The fast red substrate (Dako) was used for revelation of phosphatase activity (10 minutes at room temperature). Sections were counterstained with 0.1% (w/v) hematoxylin.

The degree of angiogenesis was determined by the number of microvessels in defined areas of the biopsy specimen according to the method of Padró et al. (17) and an international consensus report (18). Microvessel counting was simultaneously assessed by two independent, experienced investigators using light microscopy. The investigators were not aware of the diagnosis and clinical characteristics of the patient before performing the microvessel counting. The entire biopsy specimen section was systematically scanned, i.e., field per field, at ×100 magnification to find the areas showing the most intense vascularization. The decision to start counting individual microvessels was based on observing restricted areas within a field at ×100 magnification with an impression of a higher count of CD31 antigen-positive cells and cell clusters relative to adjacent areas of the same field and areas of the previous fields. The magnification was then changed to ×250 or ×400, and the investigators were allowed to reposition the slide until the highest number of microvessels was within the ×400 field. This area was defined as a hot spot after achievement of a consensus between both investigators, thus reducing the interobserver error of microvessel counting (19). Areas of vascularization adjacent to normal tissue were excluded because vascularization is not representative of neoangiogenesis in these areas. In each hot spot, both investigators performed individual microvessel counting in a ×400 field (0.26-mm² field area). In a slight modification of the method described by Weidner et al. (20), any red-stained endothelial cell or endothelial cell cluster, with or without a lumen, that was clearly separated from adjacent microvessels was considered as a single, countable microvessel. In each biopsy sample, microvessels were counted in at least 3 independent hot spots per section (range, 3–4 hot spots per section) and in 2 to 3 sections stained with anti-CD31 antibody. The mean value of all independent readings of the tumor biopsy specimen of a single patient was calculated, and MVD was defined as the mean count of microvessels per 0.26-mm² field area (i.e., ×400 field). The median MVD of the entire group was predetermined to classify patients into two groups with high (>median) and low (≤median) MVD according to an international consensus report (18).

The distributions of the time-to-event variables were estimated using the Kaplan-Meier method, and comparisons were based on the log-rank test. All P values reported are two-sided. Potential prognostic factors for osteosarcoma (6) were tested using the Cox proportional hazards model. A multivariate analysis included the variables MVD [high (>median) versus low (≤median)], primary metastasis (detectable versus not detectable at the time of diagnosis), and response to chemotherapy [good versus bad response according to Salzer-Kuntschik et al. (14, 15)]. A correlation analysis among age, histologic response to preoperative chemotherapy [Salzer-Kuntschik et al. (14, 15)], and MVD was performed by Spearman’s correlation. The factors patient age (>40 years versus <40 years), gender (male versus female), tumor site (trunk versus extremity), primary metastases, primary and secondary osteosarcoma, histologic response to preoperative chemotherapy [good versus bad response according to Salzer-Kuntschik et al. (14, 15)], and surgical remission [good versus bad remission according to Enneking et al. (16)] compared with MVD were analyzed by the Mann-Whitney test. The comparison of MVD with histologic subtypes was analyzed by the Kruskal-Wallis test. All calculations were performed using the SPSS package (SPSS, Inc., Chicago, IL).

Subjects of the present study were 60 patients with osteosarcoma (Table 1). A total of 56 primary tumors were located in an extremity, and 4 were located in the trunk. Of the 60 patients, 50 presented with apparent localized disease. Ten patients presented with distant metastases that were histologically verified or proven by progression. Osteosarcoma arose as a secondary malignancy in 3 of the 60 patients. All of the 60 osteosarcomas were high-grade central osteosarcomas. Histologic subtypes of these 60 osteosarcomas were as follows: 29 osteoblastic, 17 fibroblastic, 6 chondroblastic, 4 teleangiectatic, 3 sclerotic, and 1 small cell.

Information on histologic response to chemotherapy was available for 56 patients, 34 of whom achieved a good response [<10% viable tumor, corresponding to histologic grading scale of Salzer-Kuntschik et al. (14, 15), grades 1–3]. Surgical remission was available for 56 patients. Of these 56 patients, the operative margins to the lesions were, as follows: 49 were wide, 1 radical, 3 marginal, and 3 intralesional.

The median MVD was 52 microvessels per 0.26-mm² field area, with an interquartile range of 31 to 77 microvessels per 0.26-mm² field area (Fig. 1). A representative osteosarcoma tissue specimen is shown in Fig. 2,A for a tumor with low MVD and in Fig. 2,B for a tumor with high MVD. Based on the median MVD of the entire group, we classified patients in two groups: those with high MVD [(MVD > median), ≥53 microvessels per 0.26-mm² field area] and those with low MVD [(MVD ≤ median), ≤52 microvessels per 0.26-mm² field area]. There was no statistically significant difference between the MVD of patients with primary metastases (n = 10) and the MVD of patients without primary metastases at the time of diagnosis (n = 50; medians, 48 and 58 microvessels per 0.26-mm² field area, respectively; P = 0.215). Regarding histologic subtypes of osteosarcoma, subtypes were evenly distributed in the high- and low-MVD groups without statistical difference (Table 2). Moreover, no statistical association of MVD with patient age of >40 years, gender, tumor site, primary or secondary osteosarcoma, and surgical remission was observed (Table 2). Furthermore, patient age did not correlate with MVD (r = −0.163; P = 0.212).

The median follow-up period was 3.5 years (range, 1 month to 11.1 years). Patients with high intratumoral MVD had a significantly longer overall survival than those with low intratumoral MVD (P = 0.0029). Estimated overall survival rates for patients with tumors with high MVD were 89% for 3 years and 84% for 5 and 10 years compared with overall survival rates of 55% for 3 years and 49% for 5 and 10 years for patients with low MVD (Fig. 3,A). The estimated relapse-free survival was also significantly better in patients with tumors with high MVD as compared with patients with tumors with low MVD (P = 0.0064), with 3-year relapse-free survival rates of 78%, 5-year relapse-free survival rates of 73%, and 10-year relapse-free survival rates of 66% for patients with high MVD compared with 3-year relapse-free survival rates of 46% and 5- and 10-year relapse-free survival rates of 41% for patients with low MVD (Fig. 3 B).

From 60 patients with osteosarcoma, a subgroup of 44 patients with primary high-grade central osteosarcoma of the extremities without primary metastases and a good surgical remission was chosen for further subgroup analysis. This group was selected to avoid an imbalance between the two groups (low versus high MVD) concerning known adverse prognostic factors in osteosarcoma (6). The median follow-up for this subgroup was 3.9 years (range, 0.7–11 years). In this subgroup, we also found a significantly better overall survival for patients with high MVD than for those with low MVD (P = 0.034). The 3-, 5-, and 10-year estimated overall survival rates for patients in this subgroup with high MVD were 91% compared with estimated 3-year overall survival rates of 68% and 5- and 10-year survival rates of 58% for patients in this subgroup with low MVD (Fig. 4,A). In this subgroup, estimated relapse-free survival appeared to be prolonged in patients with high MVD osteosarcoma (3-year relapse-free survival rate of 86% and 5- and 10-year relapse-free survival rates of 72%) compared with those with low MVD osteosarcoma (3-, 5-, and 10-year relapse-free survival rate of 52%). However, this was not statistically significant (P = 0.056; Fig. 4 B). After exclusion of patients with chondroblastic osteosarcoma in this subgroup, we found significantly better estimated overall and relapse-free survival for patients with high MVD than for those with low MVD (n = 41; P = 0.026 and P = 0.0361, respectively). This exclusion was performed due to incongruent data on the prognosis of these patients (21, 22, 23, 24). Moreover, another study excluded patients with chondroblastic osteosarcoma because cartilaginous stroma was described to be associated with hypovascularity (8).

Multivariate Cox regression analysis revealed that MVD was an independent prognostic factor for survival.

To study the association of tumor angiogenesis with response to chemotherapy, we evaluated the correlation between the grade of response to chemotherapy assessed by the histologic grading scale of Salzer-Kuntschik et al. (14, 15) and MVD. A good response to chemotherapy was defined as <10% viable tumor (response grades 1–3), whereas a bad response was defined as >10% viable tumor (response grades 4–6). Patients with a good response to chemotherapy had a significantly higher MVD than patients with a bad response (P = 0.005, Mann-Whitney test). Furthermore, we observed an inverse correlation between MVD and the histologic grading scale of Salzer-Kuntschik (Pearson’s correlation coefficient r = −0.364; P = 0.006).

The aim of our present study was to examine the clinical significance of intratumoral neovascularization in patients with osteosarcoma before the initiation of polychemotherapy within the context of a multimodal treatment strategy. Until now, there has been a lack of clinical stratifying categories before initiating neoadjuvant chemotherapy in osteosarcoma to predict the response to chemotherapy. The present investigation demonstrates a significant association of intratumoral MVD with survival in a large group of osteosarcoma patients and a significant correlation between MVD and response to chemotherapy for patients treated within the multimodal therapy concept of COSS. Patients with a high degree of MVD within the tumor had significantly better overall and relapse-free survival than patients with low MVD.

The integration of intensive polychemotherapy into a multimodal treatment strategy has led to dramatic prognostic improvements in young patients with relapse-free survival rates of approximately 50% to 80% compared with <20% after surgical treatment alone (3, 6). This demonstrates that osteosarcoma is a very chemotherapy-sensitive tumor. The extent of tumor necrosis as a parameter of response to neoadjuvant chemotherapy is the strongest predictor of outcome in resectable osteosarcoma (25, 26). Extensive or complete necrosis (grades 1–3) is associated with an excellent long-term outcome, whereas patients with <90% necrosis (grades 4–6) of the primary tumor have worse prognosis (14, 15, 27). The association of the degree of MVD with response to chemotherapy described here is one of the first reports showing that a defined histologic parameter (i.e., MVD) can predict response to neoadjuvant chemotherapy such as in the COSS protocol in osteosarcoma. Moreover, this relationship might explain why patients with a high degree of intratumoral MVD have a significantly longer survival than patients with low MVD. The correlation between response to chemotherapy and MVD may be due in part to improved accessibility of the chemotherapy to proliferating osteosarcoma cells. Tumor microcirculation is an important factor in drug delivery to cancer cells (28). The efficacy of drug delivery is much higher in a tumor with a high degree of microvessels than in a tumor with low MVD, especially in a chemotherapy-sensitive tumor such as osteosarcoma. Moreover, direct antiangiogenic mechanisms of cytotoxic chemotherapy by induction of endothelial cell apoptosis may contribute to the eradication of tumor cells. Indeed, direct toxic effects on endothelial cells as well as real antiangiogenic effects have been described in in vitro and in vivo models for different cytostatic agents [e.g., anthracyclins, Vinca alkaloids, and paclitaxel (29, 30, 31, 32, 33)].

Two other studies estimating angiogenesis in osteosarcoma based on determining intratumoral MVD have been published thus far. One group (7) reported that high vascular density is associated with poor prognosis and an increased tumor proliferation index. In contrast, another group did not find any correlation between MVD and long-term outcome in nonmetastatic osteosarcoma (8). The apparent contrast of these findings to our results might be easily explained by the fact that patients in our study received a standardized treatment with intensive pre- and postoperative chemotherapy according to the COSS protocols. Therefore, increased angiogenesis might even be associated with a poor prognosis in patients who are treated exclusively by surgery or less intensive chemotherapy. Thus, the conclusion seems justified that increased angiogenesis is a predictor of prolonged survival for patients treated within a multimodal concept such as the COSS protocols. In line with this, our study demonstrated a significant correlation between the response to chemotherapy and the degree of MVD, which was not investigated in the studies mentioned. Furthermore, differences between our results and those of the other authors can in part be explained by different study population and size and the choice of endothelial markers used for identifying MVD.

In contrast to the present study, it has been reported that high intratumoral MVD in other cancers is associated with poor prognosis and that MVD was related to invasion and distant metastasis (reviewed in ref. 34). Most of these studies, however, have been conducted in patients with epithelial cancers. Furthermore, most investigations had been performed either in advanced epithelial cancers with poor prognosis and poor response to chemotherapy or in stage I epithelial cancers treated exclusively surgically. The superior chemotherapy sensitivity of osteosarcoma compared with most of these solid tumors studied thus far and the intensive chemotherapy protocol used by the COSS investigators might be the reasons why in our investigation patients with a high MVD have a more favorable outcome than the ones with a low MVD. Moreover, most studies did not use standardized chemotherapy protocols and did not investigate the correlation between response to chemotherapy and MVD. Some of the discrepancies may also be explained by the fact that details of the methodology used to assay MVD can influence its value as a prognostic indicator [e.g., the antibody (CD31, CD34, von Willebrand factor) or whether MVD is assessed at the periphery or the center of the tumor (34)]. Moreover, differences might be explained, in part, by considering that prognosis will also depend on other factors such as expression of oncogenes, adhesion molecules, growth factors, degree of apoptosis, and the mode of metastatic spread in different malignancies.

To avoid selection bias, we performed a subgroup analysis of patients with known favorable patient characteristics (primary high-grade central osteosarcoma of the extremities without primary metastases and a good surgical remission). In this subgroup, high MVD was also significantly associated with a better overall survival. Relapse-free survival seemed to be prolonged in patients with high MVD tumors compared with those with low MVD osteosarcoma. However, this was not statistically significant (P = 0.056). Reasons why the relapse-free survival in this favorable subgroup was not significantly prolonged could be that the sample size in this cohort was too small to detect significant differences. Moreover, this selection could have led to imbalances of other prognostic factors such as the expression of oncogenes or growth factors and comorbidity between both groups.

Inconclusive data exist on the prognosis of patients with different histologic osteosarcoma subtypes. Especially for patients with chondroblastic tumors, it was described that prognosis might be different from that of other histotypes (21, 22, 23, 24). In a previously reported study (8) estimating angiogenesis in osteosarcoma based on determining intratumoral MVD, chondroblastic osteosarcoma was excluded based on the association of cartilaginous stroma with hypovascularity. In our study, an association of MVD with histologic subtype was not found. Statistical evaluation of the above-described subgroup after exclusion of patients with chondroblastic osteosarcoma again showed that patients with a high degree of MVD within the tumor had significantly better overall and relapse-free survival than patients with low MVD. Therefore, our data do not suggest that histologic subtype has any influence on MVD. However, the numbers of different histologic subtypes are too low in the present study to exclude any association between MVD and histologic subtype.

Primary metastases can be detected in about 10% of osteosarcoma patients at the time of diagnosis and are detected mostly in the lung, similar to the present study (16.7%). Metastatic disease is associated with poor survival in most tumors. However, in osteosarcoma, it has been shown that patients with primary metastases have a 5-year survival of up to 30% to 50% after intensive polychemotherapy and complete resection of the metastases (35, 36). In our study, a statistical association between primary metastases (i.e., at time of diagnosis) and MVD was not found, which might be explained by the favorable survival rates of osteosarcoma patients with primary metastases. Moreover, no statistical association of MVD with age, age > 40 years, gender, tumor site, primary or secondary osteosarcoma, or surgical remission was observed. However, due to the small numbers of patients in some groups [age >40 years, tumor site (trunk), secondary osteosarcoma], additional studies in larger groups are necessary.

In conclusion, the present study provides evidence for the prognostic significance of intratumoral MVD in osteosarcoma, demonstrating higher survival rates for patients with higher intratumoral MVD treated according to intensive chemotherapy regimens such as the COSS protocols. Thus, MVD might be a potentially useful prognostic marker in osteosarcoma patients and should be further explored as a potential tool for treatment stratification of osteosarcoma patients.

We are grateful to M. Kevric (Department of Pediatric Hematology and Oncology, University of Muenster) and Dr. A. Heinecke (Department of Biostatistics, University of Muenster), for their support.