Primary osteosarcoma is the most common primary bone tumor that predominantly develops in adolescents and young adults. Even after the introduction of aggressive perioperative chemotherapy and wide excision of tumors, a substantial number of patients still experience local recurrence or distant metastases after curative resection of their primary tumors. The clinical outcome is extremely poor for these recurrent or metastatic patients (1). Thus, several attempts have been made to identify the prognostic factors for survival for the patients with osteosarcoma and to identify the patients with a high risk of developing recurrence from those patients with a low risk (2,3). There are currently very limited prognostic indicators; these include factors such as age, the tumor site, the histologic type, the tumor volume and an elevated serum alkaline phosphatase (ALP) level (4).

Ezrin is a membrane cytoskeletal linker protein which is produced by the expression of the Vil2 gene and ezin is a member of the ERM (Ezrin-Radixin-Moesin) family (5,6). Ezrin is involved in the interaction between cells or between cell and the extracellular matrix. It is related to the Rho mediated signal transduction pathway and to the Akt mediated apoptotic pathway. Ezrin is currently known to play an important role in the metastasis of various human cancers, including breast cancer, colorectal carcinoma, soft tissue sarcoma and serous ovarian carcinoma (7-10). Furthermore, ezrin has been suggested to be an importantly involved molecule in the metastasis of human osteosarcoma. In a previous pilot study, a high expression of ezrin was found to be associated with the early development of metastases in a mouse model (11). In addition, it has been noted that the disease-free interval is significantly shorter in pediatric patients with a high ezrin expression (11). However, there have been only a few reports regarding the expression of ezrin in osteosarcoma or its relationship with other clinicopathologic parameters (8,12,13).

The aim of this study is to determine the relationship between an ezrin expression and several key clinicopathologic features and to elucidate the potential value of an ezin expression for predicting the clinical outcome of patients with primary osteosarcoma.

Table 1 shows the baseline clinical characteristics. The male-to-female ratio was 1.12. The median age at the time of diagnosis of osteosarcoma was 15.7 years (range: 3.8~64.4 years). The most prevalent primary site of osteosarcoma was the femur (65.7%) and the most common histologic subtype was osteoblastic osteosarcoma (51.4%). Most patients (88.6%) underwent a limb salvage operation and the surgical specimens revealed favorable responses (≥90% necrosis) in 40 patients (57.1%).

Of the 70 tumor specimens, 39 (55.7%) revealed the expression of ezrin, while 31 (44.3%) showed no expression of ezrin. Fig. 1 shows the representative expressions of ezrin based on the staining intensities. In the positive cases, ezrin was expressed in the cytoplasm of the osteosarcoma tumor cells.

With the median follow-up of 59.9 months, among the 70 enrolled patients, 32 patients (45.7%) remained event-free, while 36 patients (51.4%) had recurrent disease and 2 patients (2.9%) died of postoperative infectious complications without any evidence of recurrence. Of the 36 recurred patients, 29 died of progressive disease and 8 survived until the time of analysis. After the curative resection of tumor, 68 patients underwent adjuvant chemotherapy, while the remaining 2 died of postoperative complications.

Table 2 shows the clinicopathologic parameters according to the expression of ezrin. There were no differences in gender, age, the site of tumor, the type of surgery and the status of the resection margin between the patients with or without an expression of ezrin. However, a more prevalent osteoblastic histology and a more elevated initial ALP level were observed in the ezrin positive patients than in the ezrin negative patients (p=0.008 and 0.001, respectively). The proportion of patients who favorably responded to neoadjuvant chemotherapy (≥90% necrosis) was significantly higher for the ezrin positive patients than that of the ezrin negative patient (72.2% vs 45.2%, respectively, p=0.024). In addition, the patients with ezrin positivity showed more frequent tumor recurrence than did those patients with ezrin negativity (64.1% vs 35.5%, respectively, p=0.017).

The 5-year OS was 61.1% and the 5-year EFS was 45.2% for all the patients. A poor histologic response to previous chemotherapy (<90% necrosis) and an expression of ezrin were negative prognostic factors on the univariate analysis (p=0.047 and 0.022, respectively), while gender, the tumor site, the histologic type, the initial ALP level, the type of surgery and the resection margin had no prognostic value (Table 3) (Fig. 2). On the Cox regression multivariate analysis, an ezrin expression showed independent significance for EFS (p=0.025), whereas its significance was marginal for OS (p=0.055) (Table 4).

Osteosarcoma is a very heterogenous disease entity and there are multiple factors that have an influence on the prognosis of osteosarcoma. Some patients have shown a good response to neoadjuvant chemotherapy and they were eventually cured by the subsequent surgery, while others have shown resistance to chemotherapy and the disease rapidly progressed (14). The well known clinical prognostic factors are age, the tumor site, the histologic type, the tumor volume and an elevated serum alkaline phosphatase (ALP) level. Yet the molecular biomarkers of osteosarcoma are not well known and much remains to be investigated. There have been several reports on the expression patterns of molecular markers such as VEGF, ErbB-2, P-glycoprotein and ezrin, and their prognostic value for osteosarcoma (4). In this current study, ezrin was selected for IHC staining because ezrin is known to be involved in the early processes of metastasis and recurrence of osteosarcoma (11-13).

In our study, 55.7% of the osteosarcoma patients showed an expression of ezrin. This result was consistent with the results of previous reports that ranged from 40% to 60% (12,15,16). The patients with an ezrin expression also experienced a greater frequency of osteosarcoma recurrence after curative resection, and they subsequently had poorer EFS and OS than did those patients without an ezrin expression. The median EFS was 34.6 months for the patients with an ezrin expression, while the EFS curve did not reach the median for the patients without an ezrin expression because more than half of the patients without an ezrin expression survived. This negative prognostic value of an ezrin expression is similar with the reports on other solid malignancies (8,9,17).

In the three major previous studies regarding the relationship between the subtype of osteosarcoma and survival, the fibroblastic subtype and the chondroblastic subtype showed better survival than did the conventional osteoblastic subtype (18-20), but no satisfactory explanation was given about this difference in the prognosis. The present study demonstrated that the conventional osteoblastic histology is associated with a frequent expression of ezrin, as 27 out of 36 (75%) patients with an osteoblastic histology expressed ezrin according to IHC staining, while only 6 of 12 (50%) of the patients with an chondroblastic subtype showed a ezrin expression and none of the patients with a fibroblastic subtype showed an ezrin expression. Because our study's sample size was small, we cannot conclude a definite relationship between the histologic subtype and an ezrin expression. Yet we suggest the possibility that the more frequent ezrin expression in the osteoblastic subtype might be related to a bad prognosis for the osteoblastic subtype. An ezrin expression and more aggressive features for metastasis could be associated with a bad prognosis for this subtype.

The concept of neoadjuvant chemotherapy was introduced in the late 1970s and this is now accepted as one of standard treatments for metastasis-free osteosarcoma (21,22). The advantages of neoadjuvant chemotherapy are the early treatment of potentially undetectable micrometastasis and the opportunity to assess the histologic response to neoadjuvant chemotherapy. Although the patients in the previously reported studies that responded favorably to neoadjuvant chemotherapy with ≥90% necrosis had a 70~87% 5-year OS, the patients with a poor response of <90% necrosis had less than 50.0% 5-year OS (2,22,23). However, the response to neoadjuvant chemotherapy is not always a good prognostic factor. A substantial number of good responders still had early recurrence at distant sites after curative resection. In this current study, among the 40 good responders, 10 patients (25.0%) recurred within 1 year after curative resection, and 8 of whom revealed ezrin immunoreactivity in the surgical specimen. In contrast, 10 of the 27 (37.0%) poor responders were able to achieve long term survival of more than 5 years after the surgery. Only 1 of these 10 long-term surviving poor responders showed ezrin immunoreactivity in their surgical specimen. The EFS curves, according to an ezrin expression and chemoresponsiveness, showed similar trends, which is consistent with these result (Fig. 3). The ezrin negative/ chemoresponsive group showed the best 5-year EFS (71.4%), followed by the ezrin negative/chemoresistant group (52.9%), the ezrin positive/chemoresponsive group (38.1%) and the ezrin positive/chemoresistant group (13.6%) (p=0.036). Overall survival according to an ezrin expression and chemoresponsiveness also showed similar trends, but it was not significant (p=0.126). That is to say, good responders with an ezrin expression might have a poorer outcome with early metastasis than the poor responders without an ezrin expression. Therefore, the good responders with an ezrin expression, who were more likely to recur early in spite of a favorable chemotherapy response, need more active tumor surveillance. Because lung and bone were the most prevalent sites of distant metastasis of osteosarcoma after the curative surgery (Table 5), regular follow-ups that include imaging of the lungs and bones could be helpful for the early detection of recurrence in this subset of patients.

The expression of ezrin by IHC staining was found in more than half (55.7%) of the patients with metastasis-free osteosarcoma. Immunoreactivity to ezrin is a negative prognostic factor for OS and EFS. Identifying an ezrin expression might offer a valuable piece of information when treating patients with primary osteosarcoma.