Primary treatment of chondroblastoma with percutaneous radio-frequency heat ablation: report of three cases.

Chondroblastomas are benign cartilaginous lesions; however, intervention is necessary to stop progression and alleviate pain. The authors evaluated three patients in whom minimally invasive percutaneous radio-frequency heat ablation was used to treat pathologically proven chondroblastoma to determine whether this treatment demonstrated long-term success. The authors found that this approach may be an effective alternative to surgical intervention in some cases.

Management of infected bulk allografts with antibiotic-impregnated polymethylmethacrylate spacers.

Twenty-seven patients developed infection following bulk allograft transplantation and were treated with resection of the allograft, placement of an antibiotic-impregnated polymethylmethacrylate (PMMA) spacer, and intravenous antibiotics. Overall, the infection was eradicated in 14 (52%) of 27 patients. Of the 11 patients who did not undergo allograft reimplantation, 5 underwent amputations for persistent infection in the presence of the spacer and 4 had a retained spacer at most recent follow-up. No significant correlation was noted between successful eradication of the infection and age, sex, diagnosis, adjuvant therapy, acute or chronic infection, number of operative procedures, type of allograft procedure, duration of antibiotics, or type of organism. Although deep infection of allograft transplantations continues to result in a high rate of failure, this method of management results in successful limb salvage in almost half of the patients.

Presurgical window of carboplatin and surgery and multidrug chemotherapy for the treatment of newly diagnosed metastatic or unresectable osteosarcoma: Pediatric Oncology Group Trial.

PURPOSE: Relapse remains a significant problem in patients with metastatic osteosarcoma. The response to carboplatin of patients with newly diagnosed metastatic or unresectable osteosarcoma was assessed in an upfront phase II window, which was followed-up by surgery and intensive multiagent chemotherapy. PATIENTS AND METHODS: Thirty-seven patients, ages 3 to 23 years with histologically confirmed diagnoses of osteosarcoma, were treated between January 1992 and November 1994 with carboplatin 1,000 mg/m2 per dose administered as a 48-hour continuous infusion. Two courses were administered in 3-week intervals, depending on marrow recovery. After radiographic reevaluation, patients underwent surgical removal of tumor (if feasible) and then 40 weeks of chemotherapy with high-dose methotrexate, ifosfamide, doxorubicin, and cisplatin. RESULTS: One of the 37 evaluable patients demonstrated a partial response to carboplatin; there were no complete responses. Patients were additionally analyzed by the response of pulmonary metastases to therapy and the extent of tumor necrosis of the primary lesion. By these criteria, 8 of 37 (22%) of patients showed a response at one or more sites, whereas 20 of 37 (54%) had unequivocal disease progression. Severe myelosuppression was the major toxicity. The projected 3-year event-free and overall survival rates were 23.9% and 31.9%, respectively. Only 1 of 17 patients with unresectable disease or distant bone metastases remains alive, in contrast to 6 of 17 patients with the lung as their only metastatic site and two of three patients with resected regional bone metastases. CONCLUSIONS: Continuous-infusion carboplatin demonstrated limited activity as an upfront agent in patients with metastatic osteosarcoma at diagnosis, even at doses that result in severe and prolonged myelosuppression. Patients with isolated pulmonary metastases or resectable bone metastases have a longer median survival time and greater chance of long-term survival than do patients with unresectable bone disease, for whom the prognosis remains dismal.

Intra-articular lipoma with osteochondroid metaplasia in the knee joint.

We describe a case of lipoma with osteochondroid metaplasia in the knee joint. Although the location of the lesion and radiographic findings were unusual, computed tomography and magnetic resonance imaging were useful in characterizing adipose, cartilaginous and osseous tissue components within the lesion.

Relationship between histologic grade and cytofluorometric cellular DNA and RNA content in primary bone tumors.

The diagnosis and grading of bone tumors remains a challenging problem. We studied the relationship between histologic grade and cytofluorometric cellular DNA and RNA content in 108 primary bone tumors. The data included DNA ploidy, mean DNA content (MDC), S-phase fraction (SPF), mean RNA content (MRC) and RNA/DNA ratio (RDR; MRC/MDC) which represents the RNA content normalized for the DNA content. Benign tumors had a diploid stem line with low MDC (mean; 1.04), low SPF (0.9), high MRC (2.41) and high RDR (2.31). Giant cell tumors of bone, which are locally aggressive benign tumors, showed diploidy with relatively higher MDC (1.07, p < 0.01) and SPF (2.6, p < 0.01) and lower MRC (1.81, p < 0.01) and RDR (1.69, p < 0.01). Similar results were obtained in low-grade sarcomas. In high-grade sarcomas, the data depended on the histologic findings. Pleomorphic sarcomas such as osteosarcomas revealed aneuploidy with remarkably higher MDC (1.70 in osteosarcomas, p < 0.01) and SPF (6.5, p < 0.01), but lower RDR (1.70, p < 0.01). In contrast, small cell sarcomas, such as Ewing's sarcomas, showed diploidy with low MDC (1.11 in Ewing's sarcomas, N.S.) and SPF (2.5, p < 0.01) and extremely low RDR (1.34, p < 0.01). The RDR value was higher in well-differentiated tumors than in primitive tumors, rendering it useful in grading bone tumors with a diploid stem line. By combining the RDR value with the MDC value, 96% of diploid sarcomas could be distinguished from benign tumors. These results indicate that cellular DNA and RNA content analysis may be of value in assessing the malignant potential of diploid as well as aneuploid bone sarcomas.

The effect of adjuvant chemotherapy on osteoarticular allografts.

Two hundred lower extremity osteoarticular allografts (in 200 patients) performed for aggressive or malignant bone tumors between 1976 and 1997 included 124 grafts of the distal femur, 46 of the proximal tibia, and 30 of the proximal femur. Seventy-four patients did not receive chemotherapy, and 126 received either adjuvant or neoadjuvant therapy. The diagnoses, mean ages, and length of followup were different for the two groups because most of the patients in the chemotherapy group had osteosarcoma, whereas the largest number in the control group had chondrosarcoma or parosteal osteosarcoma. The extent of the surgery was essentially the same for both patient groups, as is reflected by a low recurrence rate (7% for the control and 6% for the chemotherapy group). A statistical comparison of the various parameters showed that the infection, fracture, and amputation rates were the same, but the nonunion rate was markedly increased in the patients who received chemotherapy (32% versus 12%). Cox regression and Kaplan-Meier studies showed that chemotherapy had a significant effect on outcome, with the success rates for the two groups being quite different (72% versus 56%). The results for the distal femur showed a greater effect than for either the proximal tibia or the proximal femur. Analysis of these data suggest the distal femur is perhaps the most prone to healing problems, possibly based in part on the extent of the surgery. A final study supports the concept that the results improved in later years, suggesting a modification or application of the drugs used, better selection of patients, and improvements in surgical technique.

Primary osteosarcoma of the distal femur in two consecutive brothers.

The following report describes two consecutive brothers from a nonimmigrant family, with no identifiable predisposing factors, who presented with osteosarcomas of their distal femurs, one at the age of 18 years and the other at the age of 21 years. Until a cost-effective program is developed to screen for osteosarcoma, a detailed family history should be obtained from every new patient with osteosarcoma and parents should be urged to schedule early evaluations of siblings with complaints of painful extremities. Increased frequency of cytogenetic studies to screen for genetic abnormalities in patients with osteosarcoma is recommended to help elucidate the cause of osteosarcoma.

Allograft fractures revisited.

A retrospective review of patients with allograft fractures was done at the authors' institution. Between 1974 and 1998, 185 of 1046 (17.7%) structural allografts fractured in 183 patients at a mean of 3.2 years after transplantation. Initial allograft fixation included internal fixation with plates and screws in 181 patients. Patients with grafts that were longer than the average length (15.5 cm) tended to have worse results. Adjuvant therapy had no effect on fracture rate. Seventy-three patients with fractures had other allograft complications. Infection and nonunion with allograft fracture significantly worsened the outcome. The incidence of fracture in the patients with osteoarticular and arthrodesis transplants was significantly higher than those patients who had intercalary and composite reconstructions. Treatment of the allograft fractures included open reduction and internal fixation in 41 patients, reconstruction with a new allograft in 38, allograft-prosthesis composite in five, oncologic prosthesis in 19, amputation in 15, arthroscopic removal of loose bodies in three, resurfacing of fractured osteoarticular allograft surfaces in 39, allograft removal and cement spacer placement in 15. Twenty patients did not receive treatment. Eight of the fractures in patients who were not treated healed spontaneously. Outcomes were judged as excellent in nine patients (4.9%), good in 72 patients (38.9%), fair in 17 patients (9.2%), and in 85 patients (45.9%) the allograft reconstruction failed.

Factors affecting nonunion of the allograft-host junction.

Nonunion of allograft-host junction after bone transplantation is not uncommon, and its treatment frequently is problematic. To improve the understanding of these nonunions, a retrospective review was performed of 163 nonunions in 945 patients who underwent allograft transplantation (17.3%) for various benign and malignant tumors at the authors' institution between 1974 and 1997. Of these 945 patients, 558 did not receive adjuvant therapy. Chemotherapy was administered to 354 patients and only 33 patients received radiation therapy alone. Seventy-one patients had radiation treatment and chemotherapy. Of the 163 patients who had nonunion develop at the allograft-host junction, there were 269 reoperations performed on the involved extremity. In 108 patients, treatment was successful resulting in union of the allograft-host junction. Forty-nine patients did not respond to multiple surgical treatment attempts. The greater the number of surgical procedures, the worse the outcome. The rate of nonunions increased to 27% for the patients who received chemotherapy as compared with 11% for the patients who did not receive chemotherapy. The order of allografts from highest rate of nonunion to lowest was as follows: alloarthrodesis, intercalary, osteoarticular, and alloprosthesis. Infection and fracture rates were higher in the patients with nonunions as compared with the patients without nonunions.

Osteochondromas of the distal aspect of the tibia or fibula. Natural history and treatment.

BACKGROUND: There is little information on the natural history or treatment of osteochondromas arising from the distal aspect of either the tibia or the fibula. It is believed that there is a risk of deformation of the ankle if these exostoses are left untreated or if the physis or neurovascular structures are injured during operative intervention. METHODS: We reviewed the records of twenty-three patients who had been treated for osteochondroma of the distal aspect of the tibia or fibula between 1980 and 1996. Four of the patients had hereditary multiple cartilaginous exostoses. There were seventeen male and six female patients, and the average age at the time of presentation was sixteen years (range, eight to forty-eight years). RESULTS: Preoperative radiographs showed evidence of plastic deformation of the fibula in eleven patients who had a large osteochondroma. Four patients elected not to have an operation. The tumor was excised in nineteen patients. Postoperatively, all nineteen patients had a Musculoskeletal Tumor Society score of 100 percent for function of the lower extremity with pain-free symmetrical and unrestricted motion of the ankle at the latest follow-up examination. Partial remodeling of the tibia and fibula gradually diminished the asymmetry of the ankles in all nineteen operatively managed patients; however, the remodeling was most complete in the younger patients. Pronation deformities of the ankle did not change after excision of the tumor. Complications of operative treatment included four recurrences (only three of which were symptomatic), one sural neuroma, one superficial wound infection, and one instance of growth arrest of the distal aspects of the tibia and fibula. CONCLUSIONS: Osteochondromas of the distal and lateral aspects of the tibia were more often symptomatic than those of the distal aspect of the fibula; they most commonly occurred in the second decade of life with ankle pain, a palpable mass, and unrestricted ankle motion. Untreated or partially excised lesions in skeletally immature patients may become larger and cause plastic deformation of the tibia and fibula and a pronation deformity of the ankle. Ideally, operative intervention should be delayed until skeletal maturity, but, in symptomatic patients, partial excision preserving the physis may be necessary for the relief of symptoms and the prevention of progressive ankle deformity. However, partial excision is associated with a high rate of recurrence, so a close follow-up is required. Skeletally mature patients who are symptomatic may require excision of the tumor.

Parosteal osteosarcoma of the posterior aspect of the distal part of the femur. Oncological and functional results following a new resection technique.

BACKGROUND: Parosteal osteosarcoma is a low-grade malignant bone tumor that arises from the surface of the metaphysis of long bones. Parosteal osteosarcoma is usually well differentiated and displays a low propensity to metastasize. Wide resection of a parosteal osteosarcoma has been shown to provide a relatively risk-free method of preventing local recurrence. We propose a new method of resection of parosteal osteosarcomas located in the popliteal paraosseous space of the distal part of the femur. This method involves resection of the mass through separate medial and lateral incisions, which allows for wide margins yet limits the amount of dissection of the soft tissues and the neurovascular bundle. METHODS: Six patients with parosteal osteosarcoma located on the posterior aspect of the distal part of the femur underwent resection of the lesion and reconstruction with a posterior hemicortical allograft through dual medial and lateral incisions. The patients were evaluated with regard to pain, postoperative function, union of the allograft (osteosynthesis), and the prevalence of local recurrence. RESULTS: The average time until the last follow-up assessment was 4.3 years. No metastases developed, and there were no local recurrences. All patients were free of disease at the last follow-up evaluation. Postoperatively, the average range of motion of the knee was 0 to 122 degrees. Five of the six patients were free of pain at the time of the latest follow-up. Five of the six patients returned to their preoperative active functional status. CONCLUSIONS: We recommend resection of a parosteal osteosarcoma located on the posterior surface of the femur through separate medial and lateral incisions. This approach provides minimal dissection of the neurovascular bundle but ample exposure for reconstruction with a hemicortical allograft.

Chondroblastoma of bone.

BACKGROUND: Chondroblastoma of bone is a rare lesion, and few large series have been reported. The purpose of this paper is to report forty-seven cases treated by one group of surgeons and to identify factors associated with more aggressive tumor behavior. METHODS: Seventy-three patients with chondroblastoma of bone were treated between 1977 and 1998. We were able to obtain historical data, imaging studies, histological findings, and adequate personal or telephone follow-up to determine the outcome for forty-seven patients. RESULTS: The lesions were distributed widely in the skeleton, but most were in the epiphyses or apophyses of the long bones, especially the proximal part of the tibia (eleven tumors) and the proximal part of the humerus (ten tumors). The principal presenting symptoms were pain and limitation of movement. The treatment consisted of a variety of procedures, but the majority of the patients had intralesional curettage and packing with allograft or autograft bone chips or polymethylmethacrylate. Most of the patients had an excellent functional result, although in three osteoarthritis developed in the adjacent joint. Seven patients (15 percent) had a local recurrence; three of them had a second recurrence and one, a third recurrence. One patient died of widespread metastases, and another who had metastases to multiple sites was alive and disease-free after aggressive treatment of the metastatic lesions. CONCLUSIONS: While the size of the lesion, the age and gender of the patient, the status of the growth plate, and an aneurysmal-bone-cyst component to the tumor had no significant effect on the recurrence rate, lesions around the hip (the proximal part of the femur, the greater trochanter, and the pelvis) accounted for the majority (five) of the seven recurrent tumors and one of the two metastatic lesions.

Intrinsic resistance to chemotherapeutic agents in murine osteosarcoma cells.

BACKGROUND: There are two general categories of drug resistance: acquired and intrinsic. The mechanisms involved in acquired drug resistance have been extensively studied, and several mechanisms have been described. However, the mechanisms responsible for intrinsic drug resistance have not been elucidated, to our knowledge. The purpose of the present study was to investigate the cytological and biochemical differences between acquired and intrinsic drug resistance in osteosarcoma cells. METHODS: We previously isolated a clonal cell line (MOS/ADR1) to study acquired resistance in osteosarcoma by exposure of parental murine osteosarcoma cells (MOS) to doxorubicin. In the present study, we cloned a new, intrinsically resistant cell line (MOS/IR1) by single-cell culture of MOS cells and we investigated the differences in cell phenotype and the mechanisms of resistance in both of these resistant clones. RESULTS: The MOS/ADR1 and MOS/IR1 cells were sevenfold and fivefold more resistant to doxorubicin than the parental murine osteosarcoma cells. Morphologically, the MOS/ADR1 cell line was composed of polygonal cells, whereas the MOS/IR1 cell line consisted of plump spindle cells with long cytoplasmic processes. The MOS/IR1 cells showed a much lower level of alkaline phosphatase activity than did the MOS/ ADR1 and MOS cells. There were no substantial differences in the cellular DNA content or the doubling time among these three lines. Overexpression of the P-glycoprotein involved in the function of an energy-dependent drug-efflux pump was detected in the MOS/ADR1 cells but not in the MOS/ IR1 cells. After the cells were incubated with doxorubicin for one hour, the two resistant lines had less accumulation of the drug than did the parent line (p < 0.05). The addition of a P-glycoprotein antagonist, verapamil, or the depletion of cellular adenosine triphosphate resulted in a marked increase in the accumulation of doxorubicin in the MOS/ADR1 cells (p < 0.05) but not in the MOS/ IR1 cells. The MOS/ADR1 cells were found to exhibit cross-resistance only to substrates for P-glycoprotein (such as doxorubicin, vincristine, and etoposide), whereas the MOS/IR1 cells were resistant to all of the drugs studied (including cisplatin and methotrexate). The degree of drug resistance in the MOS/IR1 cells was found to be associated with the molecular weight of the drugs (p < 0.05). Permeabilization of the plasma membrane by saponin increased both the accumulation of doxorubicin (p < 0.05) and the cytotoxic activity of this drug in all lines, but the effects were most pronounced in the MOS/IR1 cells. CONCLUSIONS: Taken together, this data suggests that reduced drug accumulation in the MOS/IR1 cells may be due to the effect of decreased permeability of the plasma membrane on the transport of drugs from the extracellular environment into the cytosol of the cell and that this may be the mechanism responsible for intrinsic resistance to multiple drugs in the MOS/IR1 cell line. CLINICAL RELEVANCE: Current drug treatment for human osteosarcoma may include multiple chemotherapeutic agents, such as doxorubicin, cisplatin, and methotrexate. These drugs exhibit different cytotoxic actions and, thus, the mechanisms of resistance to individual drugs vary. Clinical resistance to multidrug chemotherapy may be observed in tumors that recur after repetitive chemotherapy and in previously untreated tumors. In the former group, a tumor cell may express multidrug resistance by combining several different mechanisms due to its exposure to various drugs. In the latter group, however, this is not likely. Decreased intracellular drug accumulation due to reduced permeability of the plasma membrane, found in the MOS/IR1 cells, is one possible mechanism and may explain the intrinsic resistance to multidrug chemotherapy for the treatment of osteosarcoma. Further study regarding the resistance mechanism in the MOS/IR1 cells may help to overcome the intrinsic drug resistance in oste

Ewing sarcoma of the rib: results of an intergroup study with analysis of outcome by timing of resection.

OBJECTIVE: We sought to establish the outcome and optimal therapeutic sequence for patients with nonmetastatic Ewing sarcoma/primitive neuroectodermal tumor of the chest wall. METHODS: Patients 30 years of age or younger with nonmetastatic Ewing sarcoma/primitive neuroectodermal tumor of the bone were randomly assigned to receive vincristine, doxorubicin, cyclophosphamide, and dactinomycin or those drugs alternating with ifosfamide and etoposide. Local control was obtained with an operation, radiotherapy, or both. RESULTS: Fifty-three (13.4%) of 393 patients had primary tumors of the chest wall (all rib). Event-free survival at 5 years was 57% for the chest wall compared with 61% for other sites (P >.2). Ifosfamide and etoposide improved outcome in the overall group (5-year event-free survival, 68% vs 54%; P =.002), and a similar trend occurred in chest wall lesions (5-year event-free survival, 64% vs 51%). Patients with chest wall lesions had more attempts at initial surgical resection (30%) than those with other primary tumor sites (8%, P <.01). The attempt at initial resection for chest wall lesions did not correlate with size. Initial resections at other sites were restricted to smaller tumors. Initial resection resulted in negative pathologic margins in 6 of 16 patients, whereas the delayed resection resulted in negative margins in 17 of 24 patients (P =.05). Although there was no difference in survival by timing of the operation in rib lesions, a higher percentage of patients with initial surgical resection received radiation than those with resection after initial chemotherapy (P =. 13). CONCLUSIONS: Although rib primary tumors are significantly larger than tumors found in other sites, their outcome is similar. We favor delayed resection whenever possible to minimize the number of patients requiring radiation therapy.

Complications of irradiated allografts in orthopaedic tumor surgery.

Massive structural allografts used for replacement of bone defects after removal of bone tumors have several complications, including fracture, infection, and nonunion. To decrease the rate of infection, irradiation of selected allografts before their implantation was performed. This study evaluated the complications in patients with these irradiated grafts. Twenty-four patients were identified who had received allografts from 1987 through 1991 that were irradiated before implantation. The dosage of radiation was between 10 kGy and 30 kGy. The mean length of followup of the patients was 5 years (range, 2-9 years). These grafts were compared with a control group of grafts that were not irradiated but were implanted during the same time and used for similar diagnostic problems with defects of similar size. The outcomes of the groups differed significantly only in the incidence of allograft fracture. These findings indicate that high-dose irradiation to bone allografts is associated with a higher rate of fracture than are similar reconstructions using nonirradiated allografts.

Intraosseus teratoma of the iliac bone.

We present a 4-year-old child with a large iliac bone mass incidentally discovered in a plain abdominal radiograph. The pathological examination revealed a benign mature teratoma. To the best of our knowledge, the occurrence of intraosseous mature teratoma has not been previously reported. The child had had an immature teratoma of the neck discovered in a fetal ultrasound, and resected on day 6 of life. The neck teratoma recurred twice, at 16 months and at 3.5 years of age. In these two recurrences the lesion appeared progressively more mature. At the time of discovery of the iliac bone teratoma there was no evidence of residual neck disease. The radiological and pathological characteristics, differential diagnosis, and clinical course are discussed.

Polyploidization induced by acridine orange in mouse osteosarcoma cells.

This study was undertaken to clarify the in vitro effect of acridine orange (AO) on the cell kinetics of mouse osteosarcoma cells, as well as the mechanism of cell growth inhibition induced by AO. A mouse osteosarcoma cell line (MOS), established from a radiation-induced mouse osteosarcoma, was cultured under exposure to 0.05, 0.5, 5, and 50 micrograms/ml of AO, either continuously or for 10 minutes. The cell kinetic analysis was performed using the following parameters: tumor cell growth by trypan blue exclusion test, mitotic activity, DNA synthetic activity by BrdU labeling and DNA ploidy by cytofluorometry. The results showed that continuous exposure to 5 and 50 micrograms/ml of AO or 10 minute exposure to 50 micrograms/ml of AO quickly killed the tumor cells within 12 hours, whereas continuous exposure to 0.5 microgram/ml of AO or 10 minute exposure to 5 micrograms/ml of AO gradually inhibited tumor cell growth. Under the latter conditions, mitotic activity was rapidly and completely inhibited within 48 hours but DNA synthetic activity was not completely inhibited even after 96 hours. DNA ploidy analysis demonstrated that most of the tumor cells arrested at the S-G2 phase after 12 hours, followed by G2 phase arrest after 24 hours and progressive DNA synthesis to a higher DNA ploidy class after 48 to 96 hours. We therefore concluded that a high concentration of AO has a strong cytocidal effect due to cytotoxicity whilst a moderate concentration of AO induces progressive and synchronous polyploidization by mitotic inhibition without DNA damage in MOS cells. We presume that this in vitro effect on MOS cells may be caused by protein synthetic inhibition after transfer RNA inactivation caused by AO binding.

P-glycoprotein levels predict poor outcome in patients with osteosarcoma.

To evaluate the relationship between the expression of P-glycoprotein by osteosarcomas and the rate of metastasis and death, a retrospective review of 172 patients who were diagnosed with osteosarcoma between 1987 and 1992 was performed. Forty patients had P-glycoprotein levels available. The majority of the osteosarcomas were Stage II-B (33 patients), with the remaining seven being Stage III. Tumor sites included 25 femurs, seven humeri, five tibias, and one each of pelvis, radius, and fibula. The patients with Stage III disease at presentation were treated differently from the time of diagnosis and therefore, these seven patients with Stage III osteosarcoma were excluded from additional analyses. The expression of P-glycoprotein by cultured tumor cells from biopsy specimens was determined using immunofluorescent microscopy. In the 33 patients with Stage IIB osteosarcoma with detectable P-glycoprotein, 67% (10 of 15) had metastases develop as compared with 28% (five of 18) of patients with undetectable P-glycoprotein. Similarly, 53% (eight of 15) of patients with tumors expressing P-glycoprotein died of disease compared with 11% (two of 18) with no detectable P-glycoprotein. Expression of P-glycoprotein by tumor cells seems to be associated with an estimated ninefold increase in the odds of death and a fivefold increase in the odds of metastases in patients with Stage IIB osteosarcoma. Kaplan-Meier survivorship analysis revealed that patients with detectable P-glycoprotein fared worse in terms of survival time and metastasis-free survival. Adjusting for covariates in the Cox proportional hazards model, expression of P-glycoprotein and its level were significantly predictive of time to death in patients with Stage IIB osteosarcoma.