Methotrexate, Doxorubicin, and Cisplatin (MAP) Plus Maintenance Pegylated Interferon Alfa-2b Versus MAP Alone in Patients With Resectable High-Grade Osteosarcoma and Good Histologic Response to Preoperative MAP: First Results of the EURAMOS-1 Good Response Randomized Controlled Trial

Purpose EURAMOS-1, an international randomized controlled trial, investigated maintenance therapy with pegylated interferon alfa-2b (IFN-α-2b) in patients whose osteosarcoma showed good histologic response (good response) to induction chemotherapy. Patients and Methods At diagnosis, patients age ≤ 40 years with resectable high-grade osteosarcoma were registered. Eligibility after surgery for good response random assignment included ≥ two cycles of preoperative MAP (methotrexate, doxorubicin, and cisplatin), macroscopically complete surgery of primary tumor, < 10% viable tumor, and no disease progression. These patients were randomly assigned to four additional cycles MAP with or without IFN-α-2b (0.5 to 1.0 μg/kg per week subcutaneously, after chemotherapy until 2 years postregistration). Outcome measures were event-free survival (EFS; primary) and overall survival and toxicity (secondary). Results Good response was reported in 1,041 of 2,260 registered patients; 716 consented to random assignment (MAP, n = 359; MAP plus IFN-α-2b, n = 357), with baseline characteristics balanced by arm. A total of 271 of 357 started IFN-α-2b; 105 stopped early, and 38 continued to receive treatment at data freeze. Refusal and toxicity were the main reasons for never starting IFN-α-2b and for stopping prematurely, respectively. Median IFN-α-2b duration, if started, was 67 weeks. A total of 133 of 268 patients who started IFN-α-2b and provided toxicity information reported grade ≥ 3 toxicity during IFN-α-2b treatment. With median follow-up of 44 months, 3-year EFS for all 716 randomly assigned patients was 76% (95% CI, 72% to 79%); 174 EFS events were reported (MAP, n = 93; MAP plus IFN-α-2b, n = 81). Hazard ratio was 0.83 (95% CI, 0.61 to 1.12; P = .214) from an adjusted Cox model. Conclusion At the preplanned analysis time, MAP plus IFN-α-2b was not statistically different from MAP alone. A considerable proportion of patients never started IFN-α-2b or stopped prematurely. Long-term follow-up for events and survival continues.


INTRODUCTION
Osteosarcoma is the most frequent primary sarcoma of bone, primarily diagnosed in adolescents and young adults; however, it is rare overall, with only two to three affected individuals per million person-years. Most recent regimens have included several weeks of preoperative chemotherapy, followed by surgery and several months postoperative chemotherapy. 1 Reported outcomes have been similar internationally and have shown little improvement over previous decades. 2,3 Histologic response to preoperative chemotherapy is an important prognostic factor. A good histologic response is usually classified as Ͻ 10% viable tumor in the resected specimen. Good responders have had better 5-year survival than poor responders (75% to 80% v 45% to 55%). 1,4 Four international osteosarcoma groups with a history of successfully conducted clinical trials 5-9 formed the European and American Osteosarcoma Study Group (EURAMOS) 10,11 : the Children's Oncology Group (COG), Cooperative Osteosarcoma Study Group (COSS), European Osteosarcoma Intergroup (EOI), and Scandinavian Sarcoma Group (SSG). The EURAMOS-1 trial established largescale multinational cooperation in clinical trials for osteosarcoma. 11 MAP (methotrexate, doxorubicin, and cisplatin) chemotherapy was accepted as standard. 8,12 Intensified salvage chemotherapy was investigated in poor responders. On the basis of preclinical and clinical evidence, 13 we decided to investigate the value of maintenance treatment with interferon alfa (IFN-␣) in good responders.
Maintenance treatment is well established in acute lymphoblastic leukemia 14 and is being investigated in sarcomas. 15 IFN-␣ has antiproliferative, differentiation-inducing, apoptotic, and antiangiogenic properties, and its clinical activity has been demonstrated in several cancers, including as postchemotherapy maintenance. 13,[16][17][18][19] It has been associated with activity against osteosarcomas (some of which have expressed IFN-␣/␤ receptor 20 ) in vitro, in animal models, and in patients with metastatic disease. 13 Most notably, single-institution treatment of 89 consecutive patients with semipurified leukocyte IFN-␣ as the only adjuvant treatment after surgery resulted in 10-year metastasis-free and sarcoma-specific survival rates of 39% and 43%, respectively. 21 On the basis of this rationale, we aimed to test IFN-␣ as maintenance treatment in osteosarcoma. 13 The objective of our random assignment was to examine whether addition of a pegylated formulation of interferon alfa-2b (IFN-␣-2b) as maintenance therapy after postoperative MAP would improve outcomes, with event-free survival (EFS) as the primary outcome measure.

Setting
EURAMOS-1 was an open-label phase III randomized controlled trial (RCT) for patients with localized or metastatic high-grade osteosarcoma considered suitable for complete surgical resection. Eligibility for registration has been described previously. 10,11 Key criteria were localized or metastatic highgrade osteosarcoma of an extremity or the axial skeleton (with exception of craniofacial sites), with all disease sites potentially amenable to complete surgical resection, and age Յ 40 years.
All patients received induction MAP followed by surgery of the primary. Thereafter, patients age Ն 5 years who had completed two cycles of induction MAP, had undergone macroscopically complete resection of their primary tumor, had Ͻ 10% viable tumor on histologic response assessment, and had no evidence of disease progression were eligible for the good response random assignment. Histologic response assessment was conducted locally before random assignment and later confirmed by a trial reference pathologist. Random assignment had to be performed Ͻ 35 days after surgery. Patients age Ͻ 5 years at potential random assignment were excluded from random assignment because of reports of neurologic complications in young children receiving IFN-␣ for other diseases. 22 Participants and/or their legal guardians, as appropriate, provided written informed consent to registration and random assignment. Regulatory and ethics approvals were obtained according to national requirements.

Trial Treatments and Procedures
Induction MAP (weeks 1 to 10) comprised two 5-week cycles of doxorubicin 75 mg/m 2 of body-surface area, cisplatin 120 mg/m 2 , and methotrexate 12 g/m 2 , followed by surgery of the primary in week 11. Doxorubicin and cisplatin were administered in weeks 1 and 6 and methotrexate in weeks 4 ,5,9,and 10 (Fig 1,treatment scheme;Data Supplement). Up to two additional doses of methotrexate were permitted preoperatively if surgery had to be delayed. The protocol (Data Supplement) contained detailed guidance on mandatory tests and requirements for each treatment cycle, supportive care, and dose adjustments. If present, primary metastases were to be surgically removed in weeks 11 to 20.
After histologic assessment of the resected tumor, consenting patients were randomly assigned in a one-to-one ratio to four postoperative cycles of MAP (weeks 12 to 29; cisplatin omitted in last two cycles) or to the same regimen followed by maintenance pegylated IFN-␣-2b (Fig 1). Treatment allocation was performed using concealed permuted blocks with three stratification factors: trial group (COG, COSS, EOI, or SSG), location of tumor (proximal femur or proximal humerus v other limb v axial skeleton), and presence of metastases (no v yes or possible). Lung metastases, detected by spiral computed tomography scanning, were considered certain if there were three or more lesions Ն 5 mm in maximum diameter or a single lesion Ն 1 cm. Scans of patients registering metastatic disease with fewer or smaller lesions were classified as possible metastatic disease. Patients were randomly assigned centrally through the Medical Research Council Clinical Trials Unit (COSS, EOI, and SSG) or COG.
Subcutaneous IFN-␣-2b was planned weekly from week 30 to 104 at 0.5 g/kg per week (maximum, 50 g) for 4 weeks and increased to 1.0 g/kg per week (maximum, 100 g) thereafter if no flu-like symptoms worse than Common Toxicity Criteria for Adverse Events (version 3.0) 23 grade 2 or other toxicities worse than grade 1 were experienced.

Assessments
During MAP treatment, clinical and toxicity assessments were performed before each drug administration. During IFN-␣-2b, patients were monitored twice per week for 8 weeks and once or twice per month thereafter. Adverse events were graded according to the Common Toxicity Criteria for Adverse Events (version 3.0) 23 and reported centrally as the maximum grade during pre-and postoperative chemotherapy and maximum grade per 3-month period during IFN-␣-2b. Toxicity was assessed in each patient until trial treatment was stopped. Late toxicity throughout follow-up was collected at COSS, EOI, and SSG.
All patients were assessed for local and distant recurrence at predefined intervals by physical examination and radiography of the chest and primary site. Radiographically detected relapse was also imaged by computed tomography, magnetic resonance imaging, and/or bone scans and, if appropriate, confirmed by histology. Patients were observed regularly for Ն 5 years after treatment (Data Supplement).

Statistical Analyses
The primary outcome measure was EFS, defined as time from random assignment until a first event (local recurrence, new metastatic disease, progression of primary metastatic disease, secondary malignancy, or death) or censoring at last contact. Secondary outcome measures included: overall survival (OS; time from random assignment until death resulting from any cause or last contact), short-and long-term toxicities, and quality of life, which will be the topic of separate analyses.
To detect absolute improvements of 10% from 70% to 80% in 3-year EFS (hazard ratio [HR], 0.63 in favor of IFN-␣-2b) with two-sided 5% significance level and 80% power required Ն 147 EFS events. 24 The same applied to an improvement in 5-year OS from 70% to 80%, requiring Ն 147 deaths in the longer term. The initial plan to register 1,400 patients (to randomly assign 1,260 [good responders, n ϭ 567; poor responders, n ϭ 693]) was revised to approximately 2,000 patients because of a lower randomization rate and relatively fewer poor responders than anticipated. 11 Interim data were reviewed annually by an independent data monitoring committee and could have been reported early if either P Յ .001 for EFS 25,26 or severe IFN safety issues were identified.
A prespecified subgroup of patients with localized disease comprised those without definitive metastases at registration. To detect a 10% improvement from 75% to 85% in 3-year EFS and 5-year OS (HR, 0.56) with two-sided 5% significance and 80% power required 98 events.
The primary analysis used intention-to-treat principles. The Kaplan-Meier method was used to estimate survival functions, log-rank tests for differences between survival curves, and Cox models (adjusted for stratification factors) to estimate treatment effects, with suitability checked by tests for proportionality of hazards. All comparisons were expressed relative to control, with HR Ͻ 1 favoring IFN-␣-2b. Consistency of treatment effect was examined using the interaction test ( 2 test for heterogeneity) in subgroups defined posthoc: sex, age, site of disease, location on bone, lung metastases, nonlung metastases, and histologic subtype. Median follow-up was calculated using reverse censoring on death.
In a prespecified exploratory analysis, EFS was computed from 23 weeks after starting postoperative chemotherapy, excluding patients who experienced progression before the expected start of IFN-␣-2b. IFN-␣-2b dose was summarized only for patients who could have completed and reported completing IFN-␣-2b by the data freeze (patients registered before November 15, 2010). Analyses were performed using Stata software (versions 12.1 and 13.1; Stata, College Station, TX).
Median follow-up was 44 months (IQR, 28 to 58) for MAP and 44 months (IQR, 29 to 58) for MAP plus IFN-␣-2b. Twenty patients (6%) in each arm were permanently lost to follow-up. For patients last reported as alive, 94% were seen Ͻ 14 months before data freeze.

Toxicity
The toxicity of preoperative chemotherapy has previously been reported. 11 During postoperative MAP, toxicity was mostly hematologic and did not differ by arm (Data Supplement). One patient died as a result of toxicity (cardiomyopathy); worst toxicity was grade 4 for 628 (88%) and grade 3 for 59 (8%) of 716 patients.
From routinely collected long-term toxicity data, seven (4%) of 193 patients receiving MAP and eight (4%) of 199 patients receiving MAP plus IFN-␣-2b reported grade 3 to 4 LVSD (Data Supplement). One additional grade 4 LVSD was reported as a serious adverse event during follow-up.

Exploratory Analyses
Exploratory subgroup analyses found no evidence of heterogeneity in treatment effect (Fig 3C; Data Supplement). An exploratory EFS analysis (Fig 3D) separated patients allocated to MAP plus IFN-␣-2b who started IFN-␣-2b from those who did not start and compared them with patients allocated to MAP. Patients who did not start their allocated IFN-␣-2b seemed to do worse than patients not allocated to IFN-␣-2b. The exploratory analysis of EFS computed from 23 weeks after start of postoperative chemotherapy included 702 patients who had not previously experienced progression. HR was 0.83 (95% CI, 0.61 to 1.12), similar to the overall EFS estimate.

DISCUSSION
We investigated maintenance pegylated IFN-␣-2b for patients whose resectable osteosarcomas showed good histologic response to MAP

EURAMOS-1: MAP Plus IFN-␣-2b Versus MAP-First Results
www.jco.org induction chemotherapy. The point estimate of treatment effect showed improved EFS and OS. However, neither was statistically significant, and the CIs were consistent with no effect. No change in practice is indicated by these data.
We were able to ask this question, as well as a parallel question concerning chemotherapy intensification in patients whose osteosarcomas had poor histologic response, only because of the cooperative efforts of four multi-institutional groups. 10,11 This will provide a framework for future trials.
With an age range up to 40 years and inclusion of patients with resectable axial and/or primary metastatic disease, our study had broader eligibility than many others. However, all patients had their primary tumors resected, and all of these had shown a good response to chemotherapy. The observed 3-year EFS of 76% for the 716 randomly assigned patients meeting our eligibility criteria is in the range of those previously observed for good responders. 4,9,12,28 Approximately four fifths of first events were exclusively metastatic, and there was no suggestion of an altered distribution of type of event by treatment arm.
Toxicity observed during preoperative MAP was as expected 7 and did not differ by allocation. Death related to toxicity during postoperative MAP was limited to one case of cardiomyopathy. Nevertheless, most patients reported grade 4 toxicities, mostly hematologic, attesting to the treatment burden of osteosarcoma chemotherapy. As expected, 13,29 toxicities observed during IFN-␣-2b were mainly grade 1 to 2. However, grade 3 and 4 toxicities were reported for one half of patients who started IFN-␣-2b, mostly hematologic. Several patients developed signs of cardiac failure during IFN-␣-2b. Although we cannot exclude a contribution from IFN-␣-2b to this complication, we note these patients had previously received doxorubicin 450 mg/m 2 and that a similar number of control-arm patients also developed LVSD. Given the high cumulative anthracycline dose, the overall incidence of severe clinical cardiac toxicity in this mainly adolescent population receiving a high cumulative anthracycline dose by continuous infusion, rather than as a bolus, does not seem excessive. 30 The point estimates of the HR favored IFN-␣-2b maintenance for both EFS and OS, but the CIs were consistent with no effect. The observed effect size for EFS (HR, 0.83; 95% CI, 0.61 to 1.12) was similar to that reported for another biologic agent, liposomal muramyl tripeptide phosphatidylethanolamine (HR, 0.80; 95% CI, 0.62 to 1.0), 8 but smaller than our 0.63 target.
The interpretation of our findings is limited, because approximately one quarter of patients allocated to IFN-␣-2b never started it. Furthermore, not all patients continued IFN-␣-2b after having started; only 128 of 357 patients reported completing the planned protocol treatment. These issues of initiation and adherence arose even though neither the dose nor schedule of IFN-␣-2b nor the duration of treatment was unusual, 16,18,19,31 and a pegylated preparation was expected to result in fewer adverse effects. 13,16 One may speculate why the attrition rate for IFN-␣-2b was high. Patient choice was the most common reason for non-random assignment and for never starting IFN-␣-2b among allocated patients and a common reason for its premature termination. We assume that both previous exposure to 29 protocol weeks of chemotherapy and awareness of a favorable prognosis for good responders affected compliance. A recent RCT of IFN-␣ maintenance for relapsed lymphoma faced similar abandonment problems. Those researchers concluded that it was not clear whether the absence of a demonstrable advantage reflected a lack of intrinsic activity or indicated the inability to administer an adequate dose of IFN to patients for sustained periods. 32 At first glance, our observations might call for an as-treated analysis comparing those who initiated IFN-␣-2b against the control arm. However, patients allocated to IFN-␣-2b who never started the drug fared worse than patients never allocated to receive IFN-␣-2b in the first place, for reasons that are currently obscure.
Would a treatment effect have become more obvious if the chosen IFN-␣-2b dose had been higher or the treatment period longer? Even in melanoma, where many RCTs of IFN have been performed, evidence supporting a specific IFN dose, duration, or formulation and identification of subsets of patients beyond those with detectable residual disease most likely to benefit remain debatable issues, with no RCT showing additional benefit for treatment extending beyond 12 to 18 months. 19,33 For osteosarcoma, such evidence is completely absent. The timing of IFN-␣-2b therapy is similarly uncertain. Although IFN-␣ may enhance the sensitivity of osteosarcomas to selected chemotherapeutic agents, 34 there are no data demonstrating that IFN-␣-2b can be safely administered concurrently with MAP and no data indicating that it would be more efficacious.
Was the good responder cohort, with its relatively low recurrence risk, ideal to observe effects of IFN-␣-2b? Good responders generally have a lower burden of micrometastatic residual disease (because of chemosensitivity of their osteosarcomas) than poor responders, and IFN may work best in such a context of minimal residual disease. This is exemplified by adjuvant data from melanoma, where IFN activity was confined to a subpopulation with microscopic nodal disease. 19,35 In conclusion, our collaborative group completed a large prospective RCT in a rare condition within a reasonable timespan. Although the point estimates for EFS and OS favored the intervention-maintenance with pegylated IFN-␣-2b-the CIs of the HRs included 1, and we conclude no difference; the observed effect size for EFS was smaller than targeted. A considerable proportion of patients allocated to IFN-␣-2b never started or did not complete treatment with the drug, which complicates interpretation of the efficacy data. Reported toxicity in patients who started IFN-␣-2b did not seem excessive.
Although we have reached the target number of EFS events, ongoing follow-up of patients is crucial and will permit the planned analysis of OS. The current EFS results, reported at the protocoldefined analytic end point, do not support the routine use of IFN-␣-2b maintenance after standard chemotherapy for osteosarcoma.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Disclosures provided by the authors are available with this article at www.jco.org.