Presented in part at the 38th Annual Meeting of the American Society of Clinical Oncology, May 18– 21, 2002.
The outcome of patients with metastatic osteosarcoma treated in two consecutive trials from 1986 to 1997 was analyzed to evaluate the efficacy of carboplatin-based multiagent chemotherapy and to identify prognostic factors. The initial study (OS-86) used ifosfamide, cisplatin, doxorubicin, and high-dose methotrexate, and the subsequent study (OS-91) used the same agents at similar doses, but carboplatin was substituted for cisplatin.
Twelve patients (median age, 15.1 yrs) were treated in OS-86 for osteosarcoma metastatic to the lung only (11 patients) or bone only (1 patient), and 17 patients (median age, 15.1 yrs) were treated in OS-91 for osteosarcoma metastatic to the lung only (12 patients), bone only (2 patients), lung and bone (2 patients), or other site (1 patient).
Patients with metastatic disease enrolled in OS-86 and those with metastatic disease enrolled in OS-91 did not differ in terms of demographic features, histologic subtype, site of primary tumor, or site of metastases. There was a difference in survival according to treatment protocol (P = 0.054). All survivors (four of whom were enrolled in OS-86 and one of whom was enrolled in OS-91) had lung metastases only. Five-year survival estimates for patients with lung metastases only were 45.5 ± 13.7% (OS-86) and 8.3 ± 5.6% (OS-91) (P = 0.084). Unilateral lung metastases (P = 0.006), no more than three lung nodules (P = 0.014), and surgical remission (P = 0.001) were associated with improved survival probability.
Osteosarcoma is the most common primary bone malignancy in children and young adolescents. Approximately 10–20% of patients have metastases at the time of diagnosis.1, 2 The most common sites of metastasis are the lung and bone.1–3 The presence or absence of metastasis at diagnosis is the only widely accepted prognostic factor available at diagnosis.2 With the use of contemporary treatment regimens, approximately 70% of patients with localized osteosarcoma of the extremity survive, but the outcome for patients with metastatic disease is much worse.1, 3, 4
Between March 1986 and July 1997, we conducted two consecutive trials (OS-86 and OS-91) for the treatment of newly diagnosed osteosarcoma. Treatment consisted of neoadjuvant chemotherapy followed by definitive surgery and adjuvant chemotherapy. OS-86 evaluated the activity of ifosfamide alone given as up-front window therapy and included high-dose methotrexate (HDMTX), cisplatin, doxorubicin, and additional ifosfamide.5 OS-91 evaluated the activity of the combination of carboplatin (a second-generation cisplatin analog) and ifosfamide given as up-front window therapy and included the same agents as OS-86 with the exception of cisplatin, which was replaced by carboplatin. Although cisplatin is one of the most effective agents against osteosarcoma,6–12 it causes irreversible nephrotoxicity and ototoxicity13–19 and is highly emetogenic.20 In OS-91, carboplatin was used because its spectrum of antitumor activity was similar to that of cisplatin in preclinical studies and because it demonstrated promising activity in adult clinical trials21–23 and less permanent nephrotoxicity and ototoxicity.24–29 The 5-year survival estimates for patients with localized osteosarcoma treated in OS-86 and OS-91 were similar (68.4 ± 7.4% and 73.1 ± 6.1%, respectively; P = 0.84).5 In this report, we present the results of both trials for patients with metastatic disease and the results of our analysis of the patients' outcomes in relation to clinical characteristics, surgical remission, and treatment protocol.
MATERIALS AND METHODS
We analyzed prospectively collected clinical data of patients with metastatic osteosarcoma who were treated in OS-86 and OS-91 and retrospectively reviewed the computed tomography (CT) scans of the chest at diagnosis and recorded the number of pulmonary nodules and type of lung involvement (unilateral or bilateral).
Eligibility criteria for the two trials were similar. Patients with previously untreated metastatic osteosarcoma (regardless of resectability of the primary tumor) and good performance status (Eastern Cooperative Oncology Group [ECOG] scores 0–2) were eligible. Biopsy of the primary tumor was required for diagnosis; biopsy of metastases was not required for assignment of disease stage. Other requirements included adequate bone marrow function and normal renal, cardiac, and hepatic function. OS-86 and OS-91 were approved by the Institutional Review Board at St. Jude Children's Research Hospital. Informed consent was obtained from patients, parents, or guardians as appropriate.
A complete medical history was obtained and a physical examination was performed before a patient enrolled in a study. The initial staging workup was comprised of plain radiography, CT, and magnetic resonance imaging (MRI) (dynamic contrast-enhanced MRI30, 31 for patients enrolled in OS-91 only) of the primary tumor; technetium 99 methyldiphosphonate (99Tc-MDP) nuclear bone scanning; and plain radiography and CT of the chest. Laboratory tests included complete blood counts, comprehensive chemistry panel, and urinalysis. Other studies included echocardiography, electrocardiography, and pure-tone audiometry. Evaluation of response after two cycles of preoperative chemotherapy and immediately before definitive surgery was accomplished by plain radiography, CT, and MRI of the primary tumor; CT of the chest; and 99Tc-MDP nuclear bone scanning (only immediately before definitive surgery for patients enrolled in OS-91). Tumor status was reevaluated at the completion of therapy. Follow-up evaluation to detect recurrence was comprised of plain radiography of the chest every 6–8 weeks, CT of the chest every 4 months, and bone scanning every 6 months for the first 2 years. After this point, patients were evaluated by clinical assessment and plain radiography of the chest unless symptoms suggested tumor recurrence.
Treatment schemas are shown in Figure 1. In the OS-86 protocol, patients received 2 cycles of ifosfamide therapy (at a dose of 1.6 g/m2 daily for 5 days with uroprotection provided by mesna), and response was assessed at Week 6. Unless tumor progression was evident, a third cycle of ifosfamide therapy was administered and followed by HDMTX therapy (3 treatments at a dose of 12 g/m2 per dose with leucovorin rescue) and 1 cycle of doxorubicin therapy (at a dose of 30 mg/m2 daily for 3 days). After clinical and radiologic assessment, patients with resectable disease underwent ablative surgery of the primary tumor at Week 13. After resection of the primary tumor, patients received 4 cycles of cisplatin (at a dose of 100 mg/m2 on Day 1) and doxorubicin (at a dose of 37.5 mg/m2 daily on Days 2 and 3), 2 additional cycles of ifosfamide therapy, and 6 additional HDMTX treatments. Patients whose disease progressed during the up-front window of ifosfamide therapy remained in the study and received the remainder of the planned chemotherapy without ifosfamide.
In the OS-91 protocol, patients received 2 cycles of carboplatin (at a dose of 560 mg/m2 on Day 1) and ifosfamide therapy (at a dose of 2.65 g/m2 daily for 3 days with uroprotection provided by mesna), and response was assessed at Week 6. Unless tumor progression was evident, a third cycle of carboplatin and ifosfamide therapy was administered. After clinical and radiologic assessment, patients with resectable disease underwent ablative surgery of the primary tumor at Week 9. After resection of the primary tumor, patients received doxorubicin (a total of 5 treatments at a dose of 75 mg/m2 per dose), HDMTX (9 treatments at a dose of 12 g/m2 per dose with leucovorin rescue), and 2 additional cycles of carboplatin and ifosfamide therapy. Patients whose disease progressed during the up-front window of carboplatin and ifosfamide therapy could remain in the study and receive cisplatin (at a dose of 100 mg/m2 per dose) in place of carboplatin and ifosfamide cycles. The total cumulative chemotherapy doses for both protocols is shown in Table 1.
Table 1. Total Cumulative Chemotherapy Doses
Cumulative dose during OS-86
Cumulative dose during OS-91
Thoracotomy to resect pulmonary lesions was usually performed after resection of the primary tumor unless the pulmonary lesions were unresectable. At the time of local control (Week 13 in OS-86 or Week 9 in OS-91), patients whose primary tumors were deemed unresectable by aggressive surgery including limb-salvage surgery or major amputation were removed from the study and received individualized local therapy, usually with irradiation. All surgeries on extremity tumors were performed by orthopedic oncologists.
Clinical and radiologic responses were assessed at Week 6 (Fig. 1). Patients were considered to have a clinical response if they became pain-free without the use of analgesics and had either increased peripheral rimming calcification (healing mineralization)32 or decreased tumor size. Patients with no significant radiologic change in tumor were considered to have stable disease. Patients whose tumors growth was measurable or patients in whom new lesions developed were considered to have progressive disease. The four-grade system of Huvos et al.12, 33 was used for histologic assessment of the primary tumor at resection.
Associations between categoric variables were examined by using the Fisher exact test. Age at diagnosis was compared between groups by using the exact Wilcoxon rank-sum test. Survival was defined as the time from the date of diagnosis to the date of last follow-up or death due to any cause. Event-free survival (EFS) was defined as the time from the date of diagnosis to the date of disease progression or recurrence, second malignancy, death due to any cause, or last follow-up. Survival and EFS distributions were estimated using the method of Kaplan and Meier34; standard errors were calculated using the method of Peto and Pike.35 All survivors had been seen or contacted within 1 year of the date of the analyses. Outcome was analyzed in relation to demographic features, histologic type of osteosarcoma, primary tumor site, location and extent of metastasis, histologic response, and treatment protocol. The exact log-rank test was used to compare survival distributions.36
Twelve (24%) of the 50 eligible patients enrolled in OS-86 and 17 (25%) of the 69 patients enrolled in OS-91 had metastatic disease at the time of diagnosis (P = 1.0). Comparison of patients who had metastatic disease and were enrolled in OS-86 with those who had metastatic disease and were enrolled in OS-91 revealed no significant differences with regard to age, gender, race, tumor histology (osteoblastic osteosarcoma vs. other types), primary tumor site (long bone vs. flat bone), or metastasis site (lung only vs. other sites) (Table 2). Eleven (92%) of 12 patients enrolled in OS-86 had metastases to the lung only and 1 patient had metastasis to bone. Twelve (71%) of the 17 patients enrolled in OS-91 had metastasis to the lung only, 4 patients had metastasis to either bone alone or bone and lung, and 1 patient had a drop metastatic lesion at the costophrenic angle.
Table 2. Characteristics of 29 Pediatric Patients with Metastatic Osteosarcoma Treated in the OS-86 and OS-91 Protocols
Comparison of white versus other races.
Comparison of osteoblastic versus other histologic types of osteosarcoma.
Comparison of lung metastases only versus metastases at other sites.
Comparison of unilateral lung metastases versus bilateral lung metastases.
In the subgroup of 23 patients with lung metastases only, the median number of lung nodules was 5 (range, 1 to > 15 lung nodules). Ten patients (43%) had ≤ 3 nodules, and 8 patients (35%) had unilateral lung involvement. There were no differences noted between patients enrolled in OS-86 and those enrolled in OS-91 with regard to the type of lung involvement (unilateral or bilateral) or the number of pulmonary nodules (three or fewer nodules or more than three nodules) (P ≥ 0.40).
Table 3 shows data for the clinical and radiologic response at Week 6 and the histologic response of the primary tumors. All six patients who had disease progression during the up-front window of ifosfamide therapy in OS-86 continued to receive treatment in the study. Two of the three patients who had disease progression during the up-front window of carboplatin and ifosfamide therapy on OS-91 were removed from the study after one course of carboplatin/ifosfamide. The third patient was removed from OS-91 after hip disarticulation and treatment with one cycle of doxorubicin therapy without cisplatin and with two doses of HDTMX.
Table 3. Tumor Responses among 29 Pediatric Patients with Metastatic Osteosarcoma Treated in the OS-86 and OS-91 Protocols
No. of patients
No. of patients
Data for the 19 patients who were assessable for histologic response.
Data regarding histologic response of the primary tumor was available for 18 of the 29 patients (Table 3); 1 additional patient who was enrolled in OS-91 and underwent hip disarticulation because of disease progression after 1 course of carboplatin and ifosfamide therapy was assigned a Grade I histologic response. Histologic response data were unavailable for four patients enrolled in OS-86 because the primary tumor was not resected (because of distant disease progression [one patient] or the patient's refusal of surgery [one patient]) or because of up-front amputation (two patients). The same type of data were unavailable for six patients enrolled in OS-91 because the primary tumor was not resected because of unresectable disease (three patients) or disease progression at the local site, distant site, or both (three patients). In total, 21 patients underwent surgery for local control (8 above-knee amputations, 3 hip disarticulations, 1 forequarter amputation, 8 limb-salvage procedures, and 1 rib and chest wall resection).
Of the 29 patients with metastatic osteosarcoma, only 5 patients survived (17%). At the time of last follow-up, 4 patients enrolled in OS-86 remained alive at 12.2 years, 13.4 years, 14.0 years, and 16.5 years after diagnosis, whereas only 1 patient enrolled in OS-91 was alive (9.0 years after diagnosis). The overall 5-year survival and EFS estimates for the 29 patients were 24.1 ± 7.4% and 6.9 ± 3.8%, respectively (Fig. 2).
We investigated potential prognostic factors (Table 4). There were no differences noted in survival estimates for the 29 patients when the comparisons were based on gender, race, site of the primary tumor (long bone vs. flat bone), tumor histology (osteoblastic osteosarcoma vs. other types), or histologic response (Rosen Grade I/II vs. III/IV) (P ≥ 0.138). Although the difference in survival distributions by site of metastasis (lung only vs. other sites) did not reach statistical significance (P = 0.096), the five patients who survived had lung metastasis only and all patients with bone metastasis died. There was some evidence that the survival estimate for patients enrolled in OS-86 differed from that for patients enrolled in OS-91 (Fig. 3, P = 0.054). Three of the six patients whose disease progressed during the up-front window of ifosfamide therapy (OS-86) survived, but none of the three patients whose disease progressed during the up-front window of carboplatin and ifosfamide (OS-91) survived.
Table 4. Factors Prognostic of Survival for 29 Pediatric Patients with Metastatic Osteosarcoma
No. of patients
No. alive/no. dead
5-Yr Survival estimate (1 SE) %
SE: standard error.
Two patients in whom complete remission in the lung was achieved after preoperative chemotherapy were excluded from this analysis.
Analysis of data from 19 patients who were assessable for histologic response.
All patients with metastatic osteosarcoma (n = 29)
In the subgroup of patients with lung metastases only, a strong association was observed between the type of lung involvement and the number of pulmonary nodules. Patients with unilateral disease had fewer pulmonary nodules than did those with bilateral disease (P = 0.006). Patients with unilateral metastases had a better survival probability than did patients with bilateral metastases (Fig. 4A) (P = 0.006). In addition, patients with three or fewer lung nodules fared better than did patients with more than three nodules (Fig. 4B) (P = 0.014). Patients treated in OS-86 had better survival estimate than patients treated in OS-91, but the difference did not achieve statistical significance (Fig. 5) (P = 0.084). Because of the small sample size, multiple regression analysis was not possible. However, results of exploratory analyses of patient survival in relation to either treatment protocol and the type of lung involvement or treatment protocol and the number of pulmonary nodules are provided (Table 4).
Of 25 patients with lung metastases, 2 also had bone metastases and did not undergo surgical resection of their lung metastases. Of the 23 patients with lung metastases only, 13 underwent resection of pulmonary metastases (7 patients underwent unilateral thoracotomy and 6 underwent staged bilateral thoracotomies, with a median time of first thoracotomy of 5.9 mos after diagnosis [range, 0.5–14.0 mos after diagnosis]). Pulmonary metastases in 2 patients were resected within 3 weeks after diagnosis, and pulmonary metastases in the remaining 11 patients were resected at a median time of 3.7 months (range, 1.8–11.4 mos) after surgery for local control. The median time between the staged bilateral thoracotomies was 2.3 months (range, 0.3–7.2 mos). Five of the 13 patients whose pulmonary metastases were resected were treated in OS-91, whereas the other 8 patients (including 5 of the 6 patients whose disease progressed during the up-front window of ifosfamide therapy) were treated in OS-86.
Of the 29 patients with metastatic osteosarcoma, 2 patients experienced remission of disease in the lung after preoperative ifosfamide or carboplatin and ifosfamide therapy; these 2 patients underwent resection of local tumor, but not of lung metastases, and both patients survived. Macroscopically complete surgical remission of disease at all sites was achieved in 10 of the remaining 27 patients; microscopic residual tumor was left at the resection margin of the primary tumor in 2 of the 10 patients. Macroscopically complete surgical remission was not achieved in 17 patients because of unresectable primary tumor (patients) or metastases (patients), disease progression (7 patients), and the patient's refusal of therapy (2 patients). Surgical remission positively affected survival estimates (P = 0.001) (Fig. 6); the 5-year survival estimates were 40.0 ± 13.9% for patients in whom surgical remission was achieved (10 patients) and 5.9 ± 4.0% for patients in whom surgical remission was not achieved (17 patients). Repeat analysis of data for the 21 patients with lung metastases only (excluding the 2 patients whose disease in the lungs entered remission after preoperative chemotherapy) also revealed a significant difference in survival between patients in whom surgical remission was achieved (10 patients, with 5-year survival estimates of 40.0 ± 13.9%) and those in whom such remission was not achieved (11 patients, with 5-year survival estimates 0 ± 0%) (P = 0.005).
The outcome of patients with metastatic osteosarcoma treated in two consecutive trials at St. Jude was poor; the 5-year overall survival and EFS estimates were 24.1 ± 7.4% and 6.9 ± 3.8%, respectively. Although the outcome of patients with localized disease treated with a carboplatin-based regimen in OS-91 was similar to that of patients treated with cisplatin-based regimens, the outcome of patients with metastatic disease treated in OS-91 was extremely poor. A Brazilian study that used a carboplatin-based regimen to treat newly diagnosed osteosarcoma obtained similar results; the 3-year survival and EFS estimates were 71% and 65%, respectively, for patients with nonmetastatic osteosarcoma and 17% and 14%, respectively, for those with metastatic disease.37 In preclinical studies using osteosarcoma cell lines, carboplatin demonstrated cytotoxicity equivalent to that of cisplatin when the concentrations of carboplatin were two to eight times greater than those of cisplatin,38 and carboplatin exhibited less antitumor activity than did cisplatin against osteosarcoma xenografts (unpublished data). Carboplatin (at a dose of 1000 mg/m2 per dose administered as a 48-hr continuous infusion) alone demonstrated limited activity in patients with metastatic osteosarcoma at diagnosis.39 However, when carboplatin was combined with ifosfamide in the OS-91 trial, the histologic response rate was 56%, which translated to a good outcome for patients with localized disease.5 This response rate cannot be attributed to ifosfamide alone because the rate of disease progression seen with ifosfamide alone during OS-86 was higher than that observed with the combination of carboplatin and ifosfamide during OS-91.5 The use of carboplatin in place of cisplatin for localized or low-risk osteosarcoma is advantageous in terms of alleviating long-term toxicity.5, 14, 18, 20, 40
The 5-year survival estimate for patients with metastatic osteosarcoma treated in OS-91 (11.8% ± 6.4%) and that of the subgroup of patients with lung metastases only (8.3% ± 5.6%) appear to be lower than the survival estimates for patients treated with cisplatin-based regimens at St. Jude or other institutions.41–43 In 1992, Marina et al.41 reported a 4-year survival estimate of 30% for patients with osteosarcoma metastatic to the lungs treated with cisplatin-based therapy and aggressive attempts at resection of pulmonary disease. In addition, the Pediatric Oncology Group reported a 5-year EFS estimate of 46.7% and a 5-year survival estimate of 53.3% for patients with metastatic osteosarcoma treated with an up-front window of ifosfamide therapy and subsequent multiagent chemotherapy comprised of cisplatin, ifosfamide, HDMTX, and doxorubicin.42 The addition of etoposide to high-dose ifosfamide therapy produced a response rate of 59% and a 2-year survival estimate of 55% for patients with metastatic osteosarcoma treated with a cisplatin-based multiagent chemotherapy regimen.43 Together these results suggest that cisplatin is more effective than carboplatin in treating osteosarcoma; although cisplatin is associated with more toxicity, its use in treating high-risk osteosarcoma is warranted.
The dismal outcome for the patients with bone metastases in the current study is consistent with the results of other studies.1, 3, 42 Patients with isolated pulmonary metastases or resectable bone metastases have a greater long-term survival probability than patients with unresectable bone disease.39 The outcome of patients with metastatic disease may be improved by treatment with intensive chemotherapy coupled with aggressive resection of resectable bone lesions and the potential use of samarium 153 ethylene diamine tetramethylene phosphonate (153Sm-EDTMP), a novel bone-seeking radiopharmaceutical agent that provides therapeutic irradiation to osteoblastic bone metastases.43–45 Preliminary results of a Phase II/III Pediatric Oncology Group study of high-dose ifosfamide and etoposide demonstrated an exceptionally good 2-year progression-free survival rate (58%) for 12 patients with bone metastases (with or without lung metastases); however, the validation of these results in a larger study and longer follow-up are needed.43
Similar to the current study, others have shown that unilateral metastases and a low number of pulmonary nodules are associated with improved outcome.3, 42, 46 Although the cut-off number of nodules in the Pediatric Oncology Group study (26 patients with lung metastases only) was eight,42 our cut-off number of three is in agreement with that of the Japanese Musculoskeletal Oncology Group, who conducted a trial of 280 patients with metastatic osteosarcoma in the lung.46 Therefore, we believe that patients with metastatic osteosarcoma can be assigned to stratified risk groups on the basis of the presence or absence of bone metastases, type of lung involvement (unilateral or bilateral), and number of lung nodules (≤ 3 or > 3). These criteria have been used in an ongoing trial within the Children's Oncology Group to select patients with newly diagnosed high-risk osteosarcoma for investigational therapy with multiagent chemotherapy and trastuzumab (Herceptin®; Genentech, South San Francisco, CA), a recombinant humanized monoclonal antibody against the transmembrane tyrosine kinase HER-2.
A good histologic response was observed in 58% of patients with metastatic osteosarcoma treated in OS-86 and OS-91. We were unable to demonstrate any difference in outcome based on histologic response. This finding is similar to that for patients with localized disease treated in OS-91,5 but is inconsistent with findings of other trials of neoadjuvant chemotherapy for osteosarcoma. Our trials were not designed to test this endpoint, and the power to detect differences was limited. In addition, data about histologic responses were not available for a subset of patients. Histologic response of the primary tumor was significantly associated with better survival in a study of 156 patients whose metastatic osteosarcoma in the lung was identified at diagnosis or after the completion of therapy.46
Disease in the lung of two of our five long-term survivors entered remission after preoperative chemotherapy, and both patients underwent surgery only for local control. Apart from these two patients, surgical removal of all sites of disease was crucial for survival. This finding underscores the importance of aggressive surgical resection in addition to effective chemotherapy for the successful treatment of osteosarcoma. Resection of metastases should be considered for all patients except those in whom lung metastases resolve after preoperative chemotherapy. Similarly, resection of all tumor sites identified at presentation is essential for survival, as shown by the Memorial Sloan-Kettering Cancer Center and the Cooperative Osteosarcoma Study Group studies.1, 3 In addition, the Japanese Musculoskeletal Oncology Group study found that resection of metastases is the most effective treatment of detectable osteosarcoma in the lung.46
In general, the outcome of patients with metastatic osteosarcoma is poor. Aggressive surgical resection is essential, and the use of cisplatin is warranted. Development of compounds with at least equivalent antitumor activity and lower toxicity than cisplatin will be beneficial; such potential agents include liposomal cisplatin and oxaliplatin.47 Efforts to improve survival probability should focus on identifying new active agents and on better understanding the biology of osteosarcoma to identify new prognostic factors and therapeutic targets.
The authors thank Julia Cay Jones, Ph.D., E.L.S., for editing the article.