We are indebted to Berlex Laboratories (Seattle, Wash) for providing granulocyte-macrophage–colony-stimulating factor (GM-CSF) for Chilean patients and to Wright Medical Company (Arlington, Tenn) for providing prostheses at a reduced price for Chilean patients. We acknowledge the valuable contributions of Charles B. Pratt, MD (deceased); Fredric A. Hoffer, MD; Carlos Rodriguez-Galindo, MD; Gaston K. Rivera, MD; Karen Wodowski, PNP; Vickie Given, RN; Dana Hawkins, RN, BSN, CCRC; Valerie McPherson, CCRP; and Debbie Poe, CCRP (St. Jude) and of Myriam Campbell, MD; Juan Jose Latorre, MD; Jesus Ortega, MD; and Emma Concha, MD (Chile). We thank Sharon Naron for editorial assistance. We thank Christine Wogan for her substantive editing of the manuscript.
The standard treatment of osteosarcoma includes cisplatin and high-dose methotrexate (HDMTX); both agents exert significant toxicity, and HDMTX requires complex pharmacokinetic monitoring and leucovorin rescue. In the previous OS91 trial, the treatment of localized disease with carboplatin, ifosfamide, doxorubicin, and HDMTX yielded outcomes comparable to those of cisplatin-based regimens and caused less toxicity. To build on this experience, the authors conducted a multi-institutional trial (OS99) that evaluated the efficacy of carboplatin, ifosfamide, and doxorubicin without HDMTX in patients with newly diagnosed, localized, resectable osteosarcoma.
Treatment was comprised of 12 cycles of chemotherapy administered over 35 weeks: 3 cycles of carboplatin (dose targeted to area under the concentration-time curve of 8 mg/mL × min on Day 1) and ifosfamide (at a dose of 2.65 g/m2 daily ×3 days) and 1 cycle of doxorubicin (at a dose of 25 mg/m2 daily ×3 days) before surgical resection, followed by 2 additional cycles of the combination of carboplatin and ifosfamide and 3 cycles each of doxorubicin (25 mg/m2 daily ×2 days) combined with ifosfamide or carboplatin.
A total of 72 eligible patients (median age, 13.4 years) were enrolled between May 1999 and May 2006. Forty of the 66 (60.6%) evaluable patients had good histologic responses (>90% tumor necrosis) to preoperative chemotherapy. The estimated 5-year event-free survival rate was 66.7% ± 7.0% for the OS99 trial compared with 66.0% ± 6.8% for the OS91 trial (P = .98). The estimated 5-year survival rate was 78.9% ± 6.3% for the OS99 trial and 74.5% ± 6.3% for the OS91 trial (P = .40).
Osteosarcoma is the most common primary bone malignancy diagnosed in children and young adolescents. The majority of patients present with localized disease,1, 2 and 60% to 70% of these patients survive with contemporary treatment regimens.3-6 Therapy is comprised of aggressive surgery and multiagent chemotherapy, which usually includes doxorubicin, cisplatin, and high-dose methotrexate (HDMTX). Although cisplatin and HDMTX are very effective, they can cause significant toxicities, including nephrotoxicity, otoxicity, mucositis, hepatotoxicity, pulmonary toxicity, and neurotoxicity. Some MTX toxicities, such as mucositis and nephrotoxicity, are associated with delayed MTX clearance, which can be affected by renal function and other nephrotoxic drugs such as cisplatin.7
Our previous osteosarcoma trial (OS91) evaluated the combination of carboplatin and ifosfamide given as upfront window therapy, plus doxorubicin and HDMTX.8 Although single-agent carboplatin has shown very limited activity against previously untreated metastatic osteosarcoma,9 carboplatin combined with ifosfamide demonstrated substantial antitumor activity in the OS91 trial.8 This activity was not attributable to ifosfamide alone, because the rate of early disease progression was significantly lower than that noted with ifosfamide alone.8, 10 For localized osteosarcoma, the OS91 trial yielded outcomes comparable to those of cisplatin-based regimens and caused less toxicity.8
The necessity for HDMTX in the context of multiagent chemotherapy for osteosarcoma has been questioned, and some postulate that HDMTX toxicity interferes with the dose-intensive delivery of other active agents.7, 11-13 In addition, HDMTX administration requires rigorous pharmacokinetic monitoring and rescue with leucovorin that is dose-adjusted to MTX levels.7, 14 The expertise and technology required for this monitoring are not available at some institutions. Therefore, an effective chemotherapy regimen that does not contain HDMTX would be of benefit, especially for patients with renal dysfunction.
We wished to investigate whether the complexity and potential toxicity of HDMTX administration can be avoided. Thus, the OS99 trial assessed the use of carboplatin, ifosfamide, and doxorubicin without HDMTX for the treatment of patients with localized, resectable osteosarcoma. In addition, because minimal resection of tumor-free bone may improve prosthesis fixation and help preserve growth potential in children, the OS99 trial also explored whether resection of the primary tumor with a 3-cm rather than a 5-cm (as used in the OS91 trial) bone margin can be performed without increasing the rate of local disease recurrence.
MATERIALS AND METHODS
Patients were enrolled at 3 centers: St. Jude Children's Research Hospital, Memphis, Tennessee; Luis Calvo MacKenna Hospital, Santiago, Chile; and Washington University Medical School, St. Louis, Missouri. Eligibility requirements included age <25 years; previously untreated, nonmetastatic, histologically proven, high-grade osteosarcoma or malignant fibrous histiocytoma of bone (patients with parosteal or periosteal osteosarcoma were not eligible) that was resectable by either limb-sparing surgery or amputation; total serum bilirubin <1.5 times that of normal; serum creatinine <2 times that of normal; no evidence of cardiac rhythm disturbance or cardiomyopathy; and signed informed consent provided by the patient, parent, or guardian, as appropriate. The study was approved by the Institutional Review Boards of all participating institutions.
The protocol was comprised of 12 cycles of chemotherapy (1 cycle every 3 weeks over 35 weeks) (Fig. 1). Preoperative chemotherapy was comprised of 3 cycles of carboplatin (at a dose targeted to achieve an area under the concentration-time curve [AUC] of 8 mg/mL × min on Day 1) and ifosfamide (at a dose of 2.65 g/m2 daily on Days 1–3) and 1 cycle of doxorubicin (at a dose of 25 mg/m2 daily for 3 days). Tumor resection at Week 12 was followed by 2 additional cycles of the combination of carboplatin and ifosfamide given as described above and 3 cycles each of ifosfamide (at a dose of 2.65 g/m2 daily on Days 1–3) and doxorubicin (at a dose of 25 mg/m2 daily on Days 1 and 2) and of carboplatin (dose targeted to an AUC of 8 mg/mL × min on Day 1) and doxorubicin (at a dose of 25 mg/m2 daily on Days 1 and 2). Carboplatin dosage was based on the glomerular filtration rate (GFR) as measured by technetium 99m-diethylene triamine penta-acetic acid clearance or 24-hour urine creatinine clearance to ensure consistent systemic exposure among patients.15 The carboplatin dose (infused over 1 hour) was calculated as follows: dose (mg/m2) = 8×[(0.93 × GFR in ml/min/m2) + 15].16, 17 Ifosfamide was infused over 15 to 30 minutes and doxorubicin over 1 hour. Granulocyte–colony-stimulating factor or granulocyte-macrophage–colony-stimulating factor was administered after each cycle. The total cumulative dose of doxorubicin was 375 mg/m2 and that of ifosfamide was 63.6 g/m2.
Standard laboratory tests to assess toxicity, including complete blood counts, serum chemistries, and urinalysis, were obtained at baseline and at regular intervals until at least 4 years after the completion of therapy. The initial staging workup was comprised of plain radiography and magnetic resonance imaging (MRI), including dynamic contrast-enhanced MRI18, 19 of the primary tumor, technetium 99 methyldiphosphonate nuclear bone scanning, and computed tomography (CT) of the chest. These studies were repeated after 3 cycles of carboplatin and ifosfamide (Week 9), before definitive surgery (Week 12), and (without MRI) at the completion of therapy. Chest CT was performed during therapy at Weeks 23 and 32. After the completion of treatment, patients were monitored by plain radiography or CT scans of the chest and by radiography of the primary tumor site every 2 months and by bone scans every 6 months for the first year. Subsequently, patients underwent radiography or CT of the chest every 3 months to 6 months and radiography of the primary tumor site every 6 months to 12 months for at least 4 years after the completion of therapy.
Evaluation of Response
Clinical and radiologic responses were assessed at Weeks 9 and 12 (Fig. 1). Patients were considered to have achieved a response if they became free of pain without the use of analgesics and MRI indicated either decreased (≥50% reduction in the product of the 3 perpendicular tumor diameters) or stable (<50% reduction or <25% increase in the product of the 3 perpendicular tumor diameters) tumor volume. Patients with persistent pain and stable tumor volume on MRI were considered to have stable disease. Patients with significantly increased tumor size (≥25% increase in the product of the 3 perpendicular tumor diameters) on MRI or those with new lesions were considered to have progressive disease. The 4-grade system of Huvos et al20 and Rosen et al21 was used for histologic assessment of the tumor response at the time of surgical resection. Central review of tumor histology and tumor histologic response was performed at St. Jude Children's Research Hospital.
Evaluation of Toxicity
Toxicity was assessed using the National Cancer Institute Common Toxicity Criteria (version 2.0). Patients underwent echocardiography, electrocardiography, pure-tone audiometry, and GFR assessment before the initiation of treatment and serially thereafter.
The OS99 trial was designed as a historical control phase 2 study to compare the histologic response rate induced by 2 preoperative regimens: 3 cycles of carboplatin and ifosfamide plus 1 cycle of doxorubicin in patients treated in the OS99 trial and 3 cycles of carboplatin and ifosfamide in patients treated in the OS91 trial.22, 23 Patients with localized, resectable osteosarcoma treated in the OS91 trial served as the historical control group. The study included an interim analysis for futility after half of the patients were evaluated. The Fisher exact test was used to compare the histologic response rate between the OS91 and OS99 trials and confidence intervals of the response rate were calculated using the Blyth-Still-Casella method. The study also compared the event-free survival (EFS) and survival of patients treated without HDMTX (OS99 trial) with those of patients treated with HDMTX (OS91 trial). EFS and survival were not the primary endpoints of the OS99 trial and therefore the study was not designed to detect a prespecified difference in EFS or survival between the OS99 and OS91 trials. Survival was defined as the time between study enrollment and last follow-up or death from any cause. EFS was defined as the time between study enrollment and disease progression or recurrence, diagnosis of a second malignancy, death from any cause, or last follow-up. Survival and EFS distributions were estimated using the method of Kaplan and Meier.24 The log-rank test was used to compare survival and EFS distributions.
The effect of reduced surgical bone margin was assessed by comparing the cumulative incidence of local treatment failure (local disease progression or recurrence)25 in the OS99 and OS91 trials using the Gray test.26 Competing events were distant failure, second malignancy, or death before local failure. Patients with simultaneous local and distant treatment failure were considered to have local failure for this analysis.
Between May 1999 and May 2006, 72 eligible patients were enrolled on the OS99 trial (Table 1). The median age was 13.4 years (range, 3.2 years-23.0 years). Forty-one patients (57%) were male and 33 (46%) were white. The most common primary tumor site was the femur (n = 46; 64%). Our previous OS91 trial enrolled 47 patients with localized, resectable osteosarcoma8; the median age of the patients was 13.6 years (range, 5.9 years-19.4 years). Patients treated on the OS99 and OS91 trials were similar with respect to age (P = .53), gender (P = .35), and primary tumor site (extremity vs other; P = .38). Data regarding tumor size for the patients on the OS91 trial were not available for comparison.
Table 1. Characteristics of 72 Patients With Localized Resectable Osteosarcoma Enrolled on the OS99 Trial
Median, 13.4 (range, 3.2–23.0)
Primary tumor site
Anatomic tumor extent
Comparison of the histologic response rate of the OS99 trial at Week 12 (after 3 cycles of carboplatin and ifosfamide and 1 cycle of doxorubicin) with that of the OS91 trial at Week 9 (after 3 cycles of carboplatin and ifosfamide) included patients in the OS99 trial who received 4 cycles of chemotherapy before tumor resection (n = 63 patients) plus 3 patients who developed local disease progression before Week 12 and were considered nonresponders (total n = 66 patients). Six patients were considered inevaluable for histologic response: 1 died of bacterial sepsis after 3 cycles of preoperative chemotherapy and 5 underwent delayed tumor resection (after Week 12). Four of these 5 patients received 5 cycles of chemotherapy before surgery because of delayed readiness of the endoprostheses (n = 3 patients) or surgery scheduling problems (n = 1 patient). Tumor resection in the fifth patient was delayed until after the completion of chemotherapy because of underlying insulin-dependent diabetes mellitus, which may increase the risk of delayed wound healing and infection. The histologic grade of the resected tumors in these 5 patients was grade IV in 2 patients, grade III in 2 patients, and grade II in 1 patient.
Forty patients who underwent tumor resection at Week 12 had a histologic response (>90% tumor necrosis): grade III (n = 29) or IV (n = 11). Therefore, the histologic response rate to preoperative chemotherapy (3 cycles of carboplatin and ifosfamide and 1 cycle of doxorubicin) was 61% (40 of 66 patients; 95% confidence interval [95% CI], 48%-72%). In comparison, in the OS91 trial, the histologic response rate to preoperative chemotherapy (3 cycles of carboplatin and ifosfamide) in patients with localized, resectable disease was 51% (95% CI, 36%-66%) (P = .34).
The median duration of follow-up after study enrollment was 5.1 years (range, 2.2 years-9.9 years) for the 58 surviving patients treated in the OS99 trial. All survivors were seen or contacted within 1 year of the analysis. First events included disease recurrence or progression (n = 22 patients), second malignancy (acute myelogenous leukemia; n = 1 patient), and death from sepsis (n = 1 patient). Twelve of the 22 patients with recurrent or progressive disease had died at the time of last follow-up. The 5-year estimates of EFS and survival were 66.7% ± 7.0% and 78.9% ± 6.3%, respectively (Fig. 2).
Thirty-four of the 47 patients with localized, resectable osteosarcoma treated in the OS91 trial survived. The 34 survivors had a median follow-up of 13.1 years (range, 8.2 years-16.4 years), and 77% had been seen or contacted within 2 years of the analysis.
There was no evidence of a difference between the OS91 and OS99 trials with regard to EFS or survival distributions (Fig. 3). The 5-year estimates of EFS were 66.7% ± 7.0% for the OS99 trial and 66.0% ± 6.8% for the OS91 trial (P = .98). The 5-year estimates of survival were 78.9% ± 6.3% for the OS99 trial and 74.5% ± 6.3% for the OS91 trial (P = .40).
Of the 67 patients with extremity tumors who underwent surgery, 59 (88%) underwent limb-salvage procedures and 8 (12%) underwent amputation. The other 3 patients with extremity tumors developed local disease progression (n = 2 patients) or died (n = 1 patient) before surgery. In the 57 patients who underwent limb-salvage surgery and for whom data were available, the median bone resection margin was 4 cm (range, 1 cm-15 cm) and was at least 3 cm in 43 (75%) patients. The closest soft tissue margin in the majority of patients measured 1 mm to 2 mm.
There were 7 local failures as first events among the 72 patients on the OS99 trial. Five of these were disease progression: 3 patients developed local progression demonstrated by imaging before surgery and 2 developed local progression suggested by imaging and confirmed by histologic analysis after surgical resection. There were only 2 (3%) local recurrences: 1 was associated with a distant recurrence 3.5 years after diagnosis and the other occurred 4 years after diagnosis. The bone resection margins in these 2 cases were 3 cm and 5 cm, respectively. The cumulative incidence of local failure in the OS99 and OS91 trials was estimated from the data regarding first events. There was no evidence that local control differed by protocol (Fig. 4). The 5-year estimates of the cumulative incidence of local failure were 10.6% ± 3.9% for the OS99 trial and 12.8 ± 4.9% for the OS91 trial (P = .64).
Protocol chemotherapy was well tolerated and delivered primarily (65% of cycles) in the outpatient setting. Table 2 summarizes the grade 3 or 4 toxicities observed during the 782 cycles of chemotherapy. The most common toxicity was myelosuppression. One patient died of bacterial sepsis after the third cycle of chemotherapy. Another patient, who achieved complete remission of osteosarcoma, had a diagnosis of acute myelogenous leukemia M5 with the t(9;11) at 7 months after the completion of treatment and died 2 days after the diagnosis. There was no grade 3 or 4 serum creatinine toxicity observed or seen; grade 1 or 2 serum creatinine elevation was observed in 6 patients. There was no grade 3 or 4 ototoxicity observed or seen; grade 1 or 2 hearing loss was observed in 3 patients. Grade 3 or 4 stomatitis/pharyngitis occurred in 3 patients. No patients had grade 4 hepatic toxicity; 2 patients had grade 3 hyperbilirubinemia and 6 had a grade 3 increase in hepatic transaminase activity.
Table 2. Grade 3 and 4 Toxicitiesa Observed in >1% of the 782 Chemotherapy Cycles According to Grade
No. of Cycles (% of Total Cycles)
No. of Patients
No. of Cycles (% of Total Cycles)
No. of Patients
According to National Cancer Institute Common Toxicity Criteria (version 2.0).
One additional patient had a grade 5 infection with neutropenia and died of bacterial sepsis.
Ten patients did not receive all planned therapy because of progressive or recurrent disease (n = 7 patients), death from toxicity (n = 1 patient), refusal of treatment (n = 1 patient), or a decision to receive alternative therapy (n = 1 patient). Of the remaining 62 patients, 57 (92%) received all 12 chemotherapy cycles, 4 received 11 cycles, and 1 received 10 cycles.
The results of the current trial demonstrated that the administration of carboplatin, ifosfamide, and doxorubicin produces EFS and survival outcomes comparable to those of HDMTX-containing or cisplatin-containing regimens in patients with localized, resectable osteosarcoma.3-5, 8, 27-29 Protocol chemotherapy was found to be well tolerated and was administered mainly in the outpatient setting. This regimen eliminates the need for MTX pharmacokinetic monitoring and avoids the toxicities associated with HDMTX and cisplatin. The regimen used in the OS99 trial offers a good alternative for patients with localized, resectable disease, particularly those who demonstrate intolerance of HDMTX, and for institutions that cannot provide MTX monitoring.
The histologic response rate to preoperative chemotherapy in the current study (3 cycles of carboplatin and ifosfamide and 1 cycle of doxorubicin; 12 weeks) was 61%. In comparison, the histologic response rate was 43% to preoperative cisplatin, doxorubicin, and HDMTX (10 weeks) and 48% to preoperative ifosfamide, doxorubicin, and HDMTX (10 weeks) in the Intergroup osteosarcoma trial INT0133.3 That trial defined histologic response as residual viable tumor of <5% at the time of definitive surgery,30, 31 whereas the threshold in the OS99 trial was 10%. Although single-agent carboplatin did not demonstrate significant activity against newly diagnosed metastatic osteosarcoma, it did demonstrate substantial activity against osteosarcoma when combined with ifosfamide in 2 consecutive trials at the St. Jude Children's Research Hospital (OS91 and OS99).
The role of HDMTX has been investigated in patients with nonmetastatic osteosarcoma. The first European Osteosarcoma Intergroup (EOI) study, which randomly assigned patients to 18 weeks of treatment with cisplatin and doxorubicin versus cisplatin, doxorubicin, and HDMTX (8 g/m2), found no greater likelihood of disease-free survival or survival in the HDMTX arm.11 The subsequent EOI study also found no difference in progression-free survival or survival between patients randomly assigned to cisplatin and doxorubicin for 18 weeks and those assigned to a multiagent regimen including HDMTX (at a dose of 8-12 g/m2) for 44 weeks.12 However, the cisplatin and doxorubicin regimen appears to produce survival estimates (55% at 5 years) that are inferior to those of cisplatin, doxorubicin, and HDMTX (12 g/m2) (71% at 6 years).3, 12, 28 A third study that used ifosfamide, cisplatin, doxorubicin, and etoposide without MTX demonstrated a 3-year progression-free survival rate of 70%.13 These data suggest that osteosarcoma can be treated without HDMTX but that a 2-drug chemotherapy regimen is not sufficient.
Carboplatin, ifosfamide, and doxorubicin given without HDMTX in the OS99 trial resulted in 5-year EFS and survival estimates of 66.7% ± 7.0% and 78.9% ± 6.3%, respectively, for patients with localized resectable osteosarcoma. These results are comparable to those of regimens containing cisplatin and HDMTX,3, 28 including the most widely accepted chemotherapy regimen for localized osteosarcoma (cisplatin, doxorubicin, and HDMTX) (6-year EFS and survival estimates, 64% and 71%, respectively).28 Outcomes in our OS91 trial, which used ifosfamide and carboplatin rather than cisplatin, were comparable to those of cisplatin-based regimens and reduced the risk of ototoxicity.8 The elimination of HDMTX in the OS99 trial did not appear to negatively affect outcome. It is important to note that all but 2 of our patients had extremity tumors, and such patients have a more favorable prognosis than patients with axial tumors, who were included in very limited numbers in some of the other studies.3, 28 The regimen used in the OS99 trial avoids the toxicity of HDMTX and uses carboplatin, which causes less nephrotoxicity and ototoxicity than cisplatin; however, it includes ifosfamide, which may cause urinary electrolyte wasting, hemorrhagic cystitis, and infertility.8, 32
In Intergroup osteosarcoma trial INT0133,3 677 patients with localized osteosarcoma were randomly assigned to receive cisplatin, doxorubicin, and HDMTX with or without ifosfamide, and with or without muramyl tripeptide (MTP), a synthetic lipophilic glycopeptide that activates monocyte and macrophage antitumor activity. Although an initial analysis did not demonstrate an improvement in EFS with the addition of MTP,3 an updated analysis concluded that the 6-year survival rate had improved from 70% to 78%.28 MTP has been approved in Europe in combination with chemotherapy for the treatment of patients with newly diagnosed nonmetastatic osteosarcoma. Notably, the addition of ifosfamide to the combination of cisplatin, doxorubicin, and HDMTX in the INT0133 trial did not appear to enhance EFS or overall survival.3, 28 The most recent Children's Oncology Group (COG) pilot studies indicated a 2-year EFS rate of 69% for patients with nonmetastatic osteosarcoma without significant benefit from intensification of the doxorubicin dose (cumulative dose, 600 mg/m2) or the addition of high-dose ifosfamide (14 g/m2/cycle) and etoposide for patients with <98% tumor necrosis at the time of surgery.27 These trials suggest that intensification of chemotherapy and the addition of new cytotoxic agents do not improve patient outcome, underscoring the need for novel treatment strategies. A better understanding of the biology of osteosarcoma is critical to identify potential targets for therapy.
Most of our patients underwent limb-salvage surgery for local control. The baseline MRI was used to determine the extent of tumor resection unless there was evidence of disease progression, in which case the most recent MRI was used. The use of a 3-cm bone resection margin in the OS99 trial did not appear to adversely affect the rate of local control. This finding may be attributable to improved surgical and imaging techniques, especially because modern MRI allows for better delineation of the extent of the tumor and its relation to the neurovascular bundle. Our rate of local recurrence (3%) was similar to the 2% to 10% rates cited by others.2, 29, 33-35 A smaller bone resection margin improves the likelihood of preserving the growth plate in growing children and provides a greater area for fixation of the endoprosthesis.
Although the current study was a phase 2 study with a limited number of patients, the results provide evidence that localized osteosarcoma can be treated with carboplatin, ifosfamide, and doxorubicin, as used in the OS99 trial, without compromising EFS or survival. These results have important implications not only for avoiding the toxicity of HDMTX but also for reducing the cost and treatment complexity associated with HDMTX. Although minimizing toxicity and cost is important, improvement in outcome requires further understanding of tumor biology and the development of rational targeted therapies.
Presented in part at the 45th Annual Meeting of the American Society of Clinical Oncology; May 29-June 2, 2009; Orlando, FL.
CONFLICT OF INTEREST DISCLOSURES
Supported by Cancer Center Support Grant CA21765 and Grant CA23099 from the National Institutes of Health and by the American Lebanese Syrian Associated Charities (ALSAC).
Michael D. Neel, MD, is a consultant for Wright Medical Technology (Arlington, Tennessee).