Zoledronate in combination with chemotherapy and surgery to treat osteosarcoma (OS2006): a randomised, multicentre, open-label, phase 3 trial

Articles

Zoledronate in combination with chemotherapy and surgery to treat osteosarcoma (OS2006): a randomised, multicentre, open-label, phase 3 trial Sophie Piperno-Neumann, Marie-Cécile Le Deley, Françoise Rédini, Hélène Pacquement, Perrine Marec-Bérard, Philippe Petit, Hervé Brisse, Cyril Lervat, Jean-Claude Gentet, Natacha Entz-Werlé, Antoine Italiano, Nadège Corradini, Emmanuelle Bompas, Nicolas Penel, Marie-Dominique Tabone, Anne Gomez-Brouchet, Jean-Marc Guinebretière, Eric Mascard, François Gouin, Aurélie Chevance, Naïma Bonnet, Jean-Yves Blay, Laurence Brugières, on behalf of the Sarcoma Group of UNICANCER, the French Society of Pediatric Oncology (SFCE), and the French Sarcoma Group (GSF-GETO)

Summary Background Based on preclinical data for the antitumour effect of zoledronate in osteosarcoma, we assessed whether zoledronate combined with chemotherapy and surgery improved event-free survival in children and adults with osteosarcoma. Methods In this randomised, multicentre, open-label, phase 3 trial (OS2006), patients aged between 5 years and 50 years with newly diagnosed high-grade osteosarcoma were randomly assigned to receive standard chemotherapy with or without ten zoledronate intravenous infusions (four preoperative and six postoperative). Adults older than 25 years received 4 mg zoledronate per infusion, patients aged 18–25 years received 0·05 mg/kg for the first two infusions and 4 mg for the remaining eight infusions, and younger patients received 0·05 mg/kg per infusion. Chemotherapy comprised high-dose methotrexate based chemotherapy in patients younger than 18 years, and doxorubicin, ifosfamide, and cisplatin in adults older than 25 years; patients aged 18–25 years were treated with either regime at the discretion of the treating centre. Balanced randomisation between the two groups was done centrally with online randomisation software, based on a minimisation algorithm taking into account centre, age, combined with chemotherapy regimen, and risk group (resectable primary and no metastasis vs other). Patients and investigators were not masked to treatment assignment, but the endpoint adjudication committee members who reviewed suspected early progressions were masked to group allocation. The primary endpoint was event-free survival, estimated from the randomisation to the time of first failure (local or distant relapse, progression, death) or to the last follow-up visit for the patients in first complete remission, analysed on a modified intention-to-treat population, which excluded patients found not to have a malignant tumour after central review. Three interim analyses were planned. This trial is registered with ClinicalTrials.gov, number NCT00470223. Findings Between April 23, 2007, and March 11, 2014, 318 patients, median age 15·5 years (range 5·8–50·9), were enrolled from 40 French centres; of whom 158 were assigned to the control group (chemotherapy alone) and 160 to the zoledronate group, including 55 (17%) patients with definite metastases. The trial was stopped for futility after the second interim analysis. With a median follow-up of 3·9 years (IQR 2·7–5·1), 125 events occurred (55 in the control group and 70 in the with zoledronate group). Event-free survival at 3 years for all 315 randomly assigned patients was 60·3% (95% CI 64·5–65·9); 3-year event-free survival was 63·4% (55·2–70·9) for the control group and 57·1% (48·8–65·0) for the zoledronate group. The risk of failure was not reduced and was even marginally higher in the zoledronate group than in the control group (hazard ratio [HR] 1·36 [95% CI 0·95–1·96]; p=0·094). No major increase in severe toxic effects of grade 3 or higher associated with zoledronate, barring expected hypocalcaemia (45 [29%] of 153 participants in the zoledronate group vs ten [6%] of 155 participants in the control group; p<0·0001) and hypophosphataemia (61 [40%] of 151 in the zoledronate group vs 26 [17%] of 156 in the control group; p<0·0001). No significant difference in orthopaedic complications was noted between the two groups (27 in the control group and 29 in the zoledronate group). Two treatment-related deaths were reported (one from cardiomyopathy in the control group and one from multiorgan failure in the zoledronate group before the first zoledronate infusion). Interpretation From the results observed in this study, we do not recommend zoledronate in osteosarcoma patients. Further biological studies are required to understand the discordance between the results of OS2006 trial and preclinical data. Funding French National Cancer Institute (INCa), Novartis, Chugai, Ligue Nationale contre le Cancer, Fédération Enfants et Santé, Société Française des Cancers et Leucémies de l’Enfant.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

Lancet Oncol 2016 Published Online June 17, 2016 http://dx.doi.org/10.1016/ S1470-2045(16)30096-1 Medical Oncology Department, Institut Curie, Paris, France (S Piperno-Neumann MD); Paris-Saclay University, Paris-Sud University, CESP, INSERM, Villejuif, France (M-C Le Deley PhD); UMR 957, INSERM, Université de Nantes, France (F Rédini PhD); Pediatric Oncology Department, Institut Curie, Paris, France (H Pacquement MD); Pediatric Oncology Department, Centre Léon Bérard, Lyon, France (P Marec-Bérard MD); Department of Radiology, CHU La Timone, Marseille, France (P Petit MD); Department of Radiology, Institut Curie, Paris, France (H Brisse MD); Pediatric Oncology Department, Centre Oscar Lambret, Lille, France (C Lervat MD); Pediatric Oncology Department, CHU La Timone, Marseille, France (J-C Gentet MD); Pediatric Oncology Department, CHU Hautepierre, Strasbourg, France (Prof N Entz-Werlé MD); Medical Oncology Department, Institut Bergonié, Bordeaux, France (A Italiano MD); Pediatric Oncology Department, Hôpital Mère-enfant, Nantes, France (N Corradini MD); Medical Oncology Department, Institut de Cancérologie de l’Ouest, Saint Herblain, France (E Bompas MD); Medical Oncology Department, Centre Oscar Lambret, Lille, France (N Penel MD); Pediatric Oncology Department, Hôpital Trousseau, Paris, France (M-D Tabone MD); Pathology Department, CHU Toulouse, France (Prof A Gomez-Brouchet MD);

1

Articles

Pathology Department, Institut Curie, Saint-Cloud, France (J-M Guinebretière MD); Pediatric Orthopedic Surgery Department, Hôpital Necker Enfants Malades, Paris, France (E Mascard MD); CHU Hôtel-Dieu, and INSERM UI957, Nantes, France (Prof F Gouin MD); Biostatistics Unit, Gustave Roussy, Villejuif, France (A Chevance MSc, M-C Le Deley); Unicancer, Paris, France (N Bonnet PhD); Medical Oncology Department Centre Léon Bérard, and Claude Bernard University, Lyon, France (Prof J-Y Blay MD); and Department of Children and Adolescents Oncology, Gustave Roussy, Villejuif, France (L Brugières MD) Correspondence to: Dr Sophie Piperno-Neumann, Medical Oncology Department, Institut Curie, 75248 Paris Cedex 05, France sophie.piperno-neumann@ curie.fr

See Online for appendix For the protocol see http://www. unicancer.fr/protocolesarcome-09

2

Research in context Evidence before this study Despite advances in the treatment of osteosarcoma combining limb-sparing surgery and multiagent chemotherapy, new strategies are needed to improve survival in this rare tumour. We searched PubMed for studies published in English before January, 2015, using the terms “osteosarcoma” and “bisphosphonate”. More than 300 studies were identified, mostly focusing on preclinical data, and describing beneficial effects of zoledronate in osteosarcoma cell lines or animal models. Only two clinical trials have been published so far. In a small single arm study published by Meyers et al in 2011, concurrent treatment of pamidronate with chemotherapy was safe in 40 patients with osteosarcoma. The second study assessed the maximum tolerated dose and the feasibility of zoledronate when combined with chemotherapy in 24 patients with metastatic osteosarcoma (Goldsby et al, 2013). In both reports, the sample size was too small for any conclusion about

the effect of pamidronate or zoledronate to be made on the efficacy of treatment for osteosarcoma. Added value of the study Our study is the first randomised trial aimed to assess the efficacy of zoledronate combined with chemotherapy in children and adult patients with previously untreated osteosarcoma. After the inclusion of 318 patients, the combination of zoledronate with chemotherapy and surgery did not improve event-free survival; moreover, adding zoledronate to chemotherapy might increase the risk of failure in patients with osteosarcoma. Implications of all the available evidence Further biological studies are required to understand the discordance between the results of our trial and preclinical data. The use of zoledronate in osteosarcoma patients is not recommended.

Introduction

Methods

The evolution of systemic treatments for osteosarcoma over the past two decades has been disappointing. Survival has not improved despite several clinical trials conducted worldwide.1,2 The survival of patients with metastases at diagnosis or after a relapse remains poor and requires new therapeutic approaches.3 Among the several new drugs associated with preclinical activity, bisphosphonates seem to be one of the most promising.4 Bisphosphonates inhibit osteoclastic bone resorption and are widely used to reduce cancer therapy-induced bone loss and skeletal-related events in patients with bone metastatic breast cancer.5 In addition to their effects on bone, preclinical evidence strongly suggests that bisphosphonates exert direct anticancer activity by inhibiting angiogenesis, invasion, tumour cell adhesion, and by stimulating immunity.6 Zoledronate, a bisphosphonate approved for use in adults with bone metastases from solid tumours, has been shown to exert a direct antitumour effect on osteosarcoma cell lines4,7 and to reduce primary tumour growth, suppress lung metastases, and prolong survival in murine models using osteosarcoma cells injected intravenously or transplanted intraosseously.8–11 Moreover, in preclinical studies, zoledronate combined with ifosfamide enhanced tumour regression and tissue repair.9 These data provided the rationale for the clinical investigation of zoledronate in patients with osteosarcoma. We aimed to determine whether zoledronate combined with chemotherapy improves event-free survival in patients with newly diagnosed osteosarcoma compared with chemotherapy alone. We report here the updated results of the second interim analysis, after the Independent Data Monitoring Committee (IDMC) recommended stopping the trial early, in the interest of patient care, and publishing the results.

Study design and participants The OS2006 trial is a multicentre, national open-label, parallel group, phase 3, randomised controlled trial for patients with localised or metastatic high-grade osteosarcoma. Key eligibility criteria were newly diagnosed, biopsy-proven, high-grade (malignant) osteosarcoma, confirmed by central review; age between 5 years and 50 years; normal haematological, renal, cardiac, and hepatic function; and no previous treatment with chemotherapy or radiotherapy. Patients with small cell osteosarcoma, multiple metastases unamenable to complete resection, maxillary osteonecrosis, osteosarcoma of the jaw, recent severe dental events, glomerular filtration rate below 70 mL/min per 1·73 m², bilirubin dosage higher than twice the upper limit of normal value; or shortening fraction below 28% or ventricular ejection fraction below 50% were excluded. There was no restriction according to performance status in the eligibility criteria. Written informed consent was obtained from all patients or their parents or guardians if patients were under 18 years of age before enrolment. Pathological confirmation of primary metastases was not recommended. The protocol was approved by an independent ethics committee and the appropriate institutional review boards. The study was done in 40 French centres in accordance with the ethical principles of the Declaration of Helsinki and with Good Clinical Practice guidelines. The trial was designed jointly by the senior academic authors from the French Society of Paediatric Oncology, the French Sarcoma Group, and the sponsor (UNICANCER). Data were analysed by the biostatisticians at Gustave Roussy. The protocol synopsis is available in the appendix (p 1); the full trial protocol is available online.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

Articles

Randomisation and masking Patients were randomly assigned (1:1) to receive combination chemotherapy alone (control group) or combination chemotherapy plus zoledronate (zoledronate group). After the participant’s eligibility was established, informed consent had been obtained, and stratification factors had been defined, the participant was enrolled in the study, based on a randomisation form received by fax. Randomisation was done centrally by staff of the Gustave Roussy Biostatistics Unit (Villejuif, France), using online centralised randomisation software (TENALEA version 2.2, Netherlands Cancer Institute, Amsterdam, Netherlands), ensuring the concealment of the next patient allocation. Randomisation was based on a minimisation algorithm taking into account: (1) the age group (<18 years vs 18–25 years vs >25 years) combined with the backbone chemotherapy regimen (high-dose methotrexate based chemotherapy [HD-MTX] vs doxorubicin, cisplatin, and ifosfamide [API-AI]) leading to four categories (<18 years HD-MTX, 18–25 years HD-MTX, 18–25 years API-AI, and >25 years API-AI), (2) the risk group (resectable primary and no definite metastasis vs other), and (3) the study site, with a random factor set at 0·8. Patients and investigators were not masked to treatment assignment, but the endpoint adjudication committee members who reviewed suspected early progressions were masked to group allocation.

Procedures Patients in both treatment groups received preoperative chemotherapy for 13 weeks, based either on high-dose methotrexate combined with etoposide-ifosfamide HD-MTX regimen (<18 years) M M M EI

Surgery

M M EI

M M

12 g/m² 75 mg/m² 3 g/m² 37·5 mg/m² 120 g/m²

API

AI

AI

API

Weeks 1 2 3 4 5 6 7 8 9 10 11 12 13 14 API: Doxorubicin Cisplatin Ifosfamide AI: Doxorubicin Ifosfamide PI: Cisplatin Ifosfamide EI: Etoposide Ifosfamide

60 mg/m² 100 mg/m² 3 g/m² 60 mg/m² 3 g/m² 100 mg/m² 3 g/m² 75 mg/m² 3 g/m²

Day 1 Day 1 Days 2–3 Day 1 Days 1–2 Day 1 Days 1–2 Days 1–4 Days 1–4

M M M I

M M M

M AP

M AP

M AP

M AP

M AP

GR/M–/R: Good histological response in patients with a localised resectable primary tumour PR/M+/U: Patients with either poor response to chemotherapy or metastases at diagnosis or unresectable primary Surgery

API

M M M EI

PR/M+/U 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Day 1 Days 1–4 Days 1–4 Days 1–2 Day 2

API-AI regimen (>25 years)

M M M EI

GR/M–/R 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Weeks 1 2 3 4 5 6 7 8 9 10 11 12 13 14 M: Methotrexate EI: Etoposide Ifosfamide AP: Doxorubicin Cisplatin

according to the protocol used in the previous French OS94 trial (HD-MTX)12 or doxorubicin, cisplatin, and ifosfamide (API-AI)13 according to their age (figure 1). In the HD-MTX based regimen, preoperative chemotherapy combined seven courses of methotrexate (12 g/m²) and two courses of etoposide (300 mg/m²) and ifosfamide (12g/m²) in all patients, and postoperative chemotherapy 12 courses of methotrexate, two courses of etoposide and ifosfamide, and one course of ifosfamide (12 g/m²) in patients with good histological response and no metastasis and five cycles of methotrexate and doxorubicin 75 mg/m² and cisplatin 120 mg/m² in patients with poor histological response, initial metastases, unresectable primary, or both. In the API-AI regimen, preoperative chemotherapy combined three courses of API (doxorubicin 60 mg/m² with ifosfamide 6 g/m² and cisplatin 100 mg/m²) and two courses of AI (doxorubicin 60 mg/m² with ifosfamide 6 g/m²) and post-operative chemotherapy two courses of AI alternating with two courses of PI (cisplatin 100 mg/m² and ifosfamide 6 g/m²) in patients with good histological response and no metastases, and five cycles of etoposide and ifosfamide in patients with poor histological response metastases, unresectable primary, or both. In addition to the above chemotherapy regimen, patients assigned to the zoledronate group also received ten monthly intravenous infusions of zoledronate (four preoperative and six postoperative), at a dose of 4 mg per infusion in patients older than 25 years; 0·05 mg/kg per infusion for the first two courses, then 4 mg per infusion for the remaining eight courses in patients aged 18–25 years; 0·05 mg/kg

AI

PI

AI

PI

GR/M–/R 1 2 3 4 5 6 7 8 9 10 11 EI

EI

EI

EI

EI

PR/M+/U 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Figure 1: Backbone chemotherapy scheme Patients younger than 18 years received the HD-MTX regimen, patients older than 25 years received the API-AI regimen, and patients aged 18–25 years could receive HD-MTX or API-AI according to the decision of the treating centre at the beginning of the study. HD-MTX=high-dose methotrexate based chemotherapy. API-AI=doxorubicin, cisplatin, and ifosfamide.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

3

Articles

per infusion (maximum 4 mg) in patients younger than 18 years. 25-OH vitamin D3 and calcium supplementation was recommended in both groups. The follow-up schedule was similar for the control and zoledronate groups, including clinical assessment, adverse event monitoring, and laboratory monitoring of haematological, renal, and hepatic function. Calcaemia was monitored at 24 h and 72 h then once a week after the first zoledronate infusion, and for the subsequent infusions, before each course and 48 h later. Zoledronate dose could be reduced (–33%) if patients had hypocalcaemia (defined as less than 1·6 mmol/L). Patients stopped treatment on completion of treatment (including ten infusions of zoledronate in the zoledronate group), or in case of disease progression or relapse, inacceptable toxicity, patient request, or physician decision. Tumour response was assessed during treatment at weeks 7 and 14 with radiography and MRI of the primary site, and a chest radiograph (or a thoracic CT scan for patients with initial lung nodules) to determine resectability and to exclude progression. After completion of chemotherapy, a chest radiograph was done every 2–3 months for 3 years, then every 4 months for 2 years and yearly thereafter. Yearly radiographic assessment of the primary site was recommended. These investigations were completed if necessary by CT scan, MRI, bone scans, and histology. Central review of the primary endpoint was not done, except for all suspected progressions reported before surgery that were reviewed by an endpoint adjudication committee, comprising two radiologists, two clinicians, and one statistician, who were blinded to the treatment allocation. Adverse events were assessed after each chemotherapy course and zoledronate infusion, using a list of 25 selected items, and graded according to the Common Terminology Criteria for Adverse Events (version 3.0), including creatinine and calcaemia concentrations. A free-text area was available to document other adverse events. Grade 4 haematological toxicities and grade 3 or higher of all extrahaematological toxicities were considered as severe toxicities. Fractures of any grade have been classified as severe adverse events due to their putative oncological effect.

Outcomes The primary endpoint was event-free survival defined as the time from random assignment until a first event (progression, relapse, secondary malignancy, or death from any cause), or to the last follow-up visit for patients in first complete remission. The secondary endpoints reported in this report are overall survival, defined as the time to death from any cause; the histological response to preoperative treatment classified using modified Huvos grading for which a good response corresponds to less than 10% viable cells; and acute toxicity considering the maximum grade 4

during preoperative and postoperative chemotherapy for each type of toxicity. Long-term toxicities, including effect of zoledronate on patient growth and incidence of osteoporosis, and quality of life, are also secondary objectives and will be reported in a forthcoming paper with a longer follow-up.

Statistical analysis Assuming a baseline 3-year event-free survival of 55% with chemotherapy alone, 169 event-free survival events were required to achieve an 80% power to detect a difference of 13% from 55% to 68% (alternative hypothesis: hazard ratio [HR] 0·645 in favour of the zoledronate group) with a twosided α test set at 0·05. The planned sample size was 470 patients considering interim analyses and assuming that the proportion of secondary exclusions (proportion of patients found not to have a malignant tumour after central review) would be 2%. Three interim analyses were done with a Lan and DeMets α-spending function based on the O’Brien-Fleming group sequential boundary function.14 These analyses were to be disclosed to the IDMC. The primary efficacy analysis of event-free survival was done on the modified intention-to-treat population: only patients for whom the diagnosis of a malignant bone tumour was rejected after the central pathology review were excluded. Event-free survival was estimated with the Kaplan-Meier method. The HR of events associated with the treatment effect was estimated in a Cox model stratified according to prespecified subgroups of preoperative chemotherapy regimen (HD-MTX or API-AI), age (<18 years, 18–25 years, >25 years), risk group (non-metastatic and resectable primary vs other). Post-hoc sensitivity analyses were done: (1) excluding patients allocated to the zoledronate group who did not receive any zoledronate infusion (per-protocol analysis), (2) excluding patients with metastasis other than lung metastasis, (3) after adjustment on the tumour size (largest diameter <10 cm vs ≥10 cm) on the intention-to-treat population, (4) after adjustment on the tumour size on the per-protocol population, and (5) after adjustment on the tumour size and excluding patients with metastasis other than lung metastasis. The heterogeneity of the effect of zoledronate according to stratification variables (preplanned analysis) and tumour size (exploratory analysis) was assessed in multivariable models including interaction terms, and shown in a forest plot. Overall survival was estimated with the Kaplan-Meier method and HRs estimated in a Cox model, on the modified intention-to-treat population, using the same approach as the analysis of event-free survival. The proportion of patients with good histological response was estimated among patients who underwent surgery and compared between randomised groups using the χ² test. Safety analyses were done on the per-protocol population, excluding patients allocated to the zoledronate group who did not receive any zoledronate infusion. For each toxicity term, the proportion of patients who had a severe toxicity was compared between randomised groups using a χ² test.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

Articles

Estimates are provided with 95% CIs or SEs. All tests are two-sided. Data were analysed with SAS version 9.3. This trial is registered with ClinicalTrials.gov, number NCT00470223.

Role of the funding source This study was mainly funded by a national research grant (Programme Hospitalier de Recherche Clinique) from the French Cancer Institute, and additional grants from Novartis-France, Chugai, Ligue Nationale Contre le Cancer, Fédération Enfants et Santé, Société Française des Cancers et Leucémies de l’Enfant. Novartis and Chugai had no role in the study design, data collection, data analysis, or data interpretation. Unicancer as a non-profit organisation, was the sponsor of the study and was in charge of the collection and the monitoring of the data, oversaw the interpretation of the data, but had no role in the study design and analysis. Before completion of the trial, M-CLD and AC had access to the raw data. After completion of the trial and data analysis, an initial draft of the report was prepared jointly by a writing committee of three academic authors (SP-N, M-CLD, and LB) who had full access to all the data and had final responsibility for the decision to submit for publication.

Six (4%) of the 159 patients randomly assigned to receive zoledronate did not start zoledronate (one due to death before the first injection, three for medical decision (dental contraindication in two and renal failure after the first chemotherapy course in one), and two for patient decision, and 55 (35%) patients received less than eight zoledronate infusions. The reasons for early termination of treatment with zoledronate were: disease progression or relapse (n=8), patient decision (n=8),

522 patients assessed for eligibility

89 patients did not meet eligibility criteria*

24 were not included because of suspension of randomisation†

409 patients potentially eligible

91 patients were excluded 84 patient decision 7 for miscellaneous reasons‡

Results Between April 23, 2007, and March 11, 2014, 522 patients from 40 French centres were assessed for eligibility; 89 patients did not meet eligibility criteria and 24 were not included because of temporary stop of randomisation. Among the 409 patients potentially eligible, 318 were included in the randomised trial: 158 were assigned to the control group and 160 to the zoledronate group (figure 2). After the second interim analysis, accrual was prematurely stopped for futility because the estimated likelihood of showing an event-free survival benefit if the trial had continued was practically null (conditional power to show a benefit <0·0001). The current analysis is based on the updated database, into which entry was frozen on Aug 27, 2015. Table 1 shows the baseline demographic and clinical characteristics of the 315 patients included in the main analysis based on the modified intention-to-treat dataset (156 in the control group and 159 in the zoledronate group) after excluding three cases diagnosed as benign lesions after central review. The median age was 15·5 years (range 5·8–50·9). Overall, 214 (68%) of 315 patients had localised disease, 46 (15%) had possible metastases, and 55 (17%) definite metastases. The baseline characteristics were well balanced between the two groups, apart from a higher proportion of large tumours (≥10 cm) in the zoledronate group (92 [59%] of 155) than in the control group (72 [47%] of 153). Median follow-up was 3·9 years (IQR 2·7–5·1), and was similar in both groups (4·0 years [2·7–5·1] in the control group and 3·8 years [2·7–5·0] in the zoledronate group). Only three patients in the zoledronate group were lost to follow-up during the first 3 years and none in the control group.

318 patients included in randomised trial

158 assigned to control group

160 assigned to the zoledronate group

2 excluded due to no malignant tumour found after central review

1 excluded due to no malignant tumour found after central review

156 with a malignant bone tumour

159 with a malignant bone tumour

156 received assigned intervention

153 received assigned intervention 6 did not receive zoledronate§

0 lost to follow-up <3 years

156 included in the modified intention-to-treat analysis 156 included in the per-protocol analysis

3 lost to follow-up <3 years

159 included in the modified intention-to-treat analysis 153 included in the per-protocol analysis

Figure 2: Trial profile *89 patients did not meet eligibility criteria: contraindication to planned treatment (n=46), including dental contraindication (n=34), osteosarcoma of the jaw (n=8), primary resected tumour (n=7), unresectable metastases (n=7), previous chemotherapy (n=7), low-grade osteosarcoma (n=4), age younger than 5 years or older than 50 years (n=5), follow-up not possible (n=3), no social insurance (n=1), or extraosseous osteosarcoma (n=1). †Accrual was temporarily suspended for all patients between May and September, 2007, after the first patient had grade 4 hypocalcaemia, and between November, 2012, and April, 2013, for patients younger than 15 years for safety concerns regarding tooth growth after preclinical studies in mice. ‡Seven patients did not enter the trial for miscellaneous reasons: delayed randomisation from start of chemotherapy (n=2), familial issues (n=2), genetic abnormality (n=1), intellectual disability (n=1), and primary treatment in a non-participating centre (n=1). §Six patients allocated to the zoledronate group did not start zoledronate: death before the first injection for one patient, a medical decision for three patients (dental contraindication in two, renal failure after the first chemotherapy course in one), and patient decision for two patients.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

5

Articles

Control group (n=156)

Zoledronate group (n=159)

Sex Male

92 (59%)

Female

64 (41%)

72 (45%)

15·7 (6·4–48·7)

15·4 (5·8–50·9)

Age (years)

87 (55%)

Age and chemotherapy regimen* <18 years, HD-MTX

107 (69%)

110 (69%)

18–25 years, HD-MTX

18 (12%)

18 (11%)

18–25 years, API-AI

12 (8%)

11 (7%)

>25 years, API-AI

19 (12%)

20 (13%)

Limb

143 (92%)

147 (92%)

Axial

13 (8%)

12 (8%)

Largest diameter <10 cm

81 (53%)

63 (41%)

Largest diameter ≥10 cm

72 (47%)

92 (59%)

Primary tumour site

Primary tumour size

Data missing

3

4

Histological subtype Conventional

143 (92%)

147 (92%)

Telangiectasic

8 (5%)

4 (3%)

High-grade surface

0

2 (1%)

Small cell

0

1 (1%)

Other†

5 (3%)

4 (3%)

Primary tumour a-priori resectable No

2 (1%)

4 (3%)

Yes

154 (99%)

155 (97%)

No

140 (90%)

148 (93%)

Yes

16 (10%)

11 (7%)

106 (68%)

108 (68%)

Pathological fracture at diagnosis

Initial staging Localised disease Possible metastases

24 (15%)

22 (14%)

Definite metastases

26 (17%)

29 (18%)

113 (72%)

111 (70%)

Lung metastases‡ No Possible

19 (12%)

23 (14%)

Definite

24 (15%)

25 (16%)

142 (91%)

145 (91%)

Other metastases No Possible

9 (6%)

3 (2%)

Definite

5 (3%)

11 (7%)

129 (83%)

127 (80%)

27 (17%)

32 (20%)

Risk group Standard risk—resectable primary and no definite metastasis High risk—unresectable primary or definite metastasis

Data are n (%) or median (range). *Chemotherapy regimens were HD-MTX regimen containing high-dose methotrexate based chemotherapy after surgery or API-AI consisting of doxorubicin-cisplatin-ifosfamide and doxorubicin-ifosfamide sequential regimen. †Other histological types were giant cell osteosarcoma (n=3), juxtacortical osteosarcoma (n=1), well-differentiated osteosarcoma (n=1), fibroma-chondromyxoid bone sarcoma (n=1), pleomorphic bone sarcoma (n=1), and unclassifiable bone sarcoma (n=2). ‡Lung metastases were considered definite if there were at least a lesion 10 mm or greater, two or more lesions 5 mm or greater, or five or more lesions whatever the size.

Table 1: Baseline demographic and clinical characteristics

6

omission with no specified reason (n=11), clinician’s decision due to a contraindication (n=2), toxicity (n=2), and temporary or definitive discontinuation of the trial as recommended by the sponsor (n=24). The dose of zoledronate was reduced by more than 10% of the protocol dose, as allowed per protocol, in 83 (7%) of 1166 injections (34 patients), mostly because of hypocalcaemia. The dose intensity (p=0·96) and duration of preoperative chemotherapy (p=0·33) were very similar in both groups, as well as the time interval between surgery and start of postoperative chemotherapy (p=0·93; appendix p 23). A good histological response was noted in 194 (65%) of the 300 patients who underwent surgery (12 patients did not undergo surgery of the primary tumour [unresectable primary in seven patients, preoperative progression in three, early death in one patient, and consent withdrawal in one patient]; in three additional cases, histological response was unknown), with no statistical difference between treatment groups (98 [65%] of 150 in the control group vs 96 [64%] of 150 in the zoledronate group; p=0·81). As of data cutoff on Jan 1, 2015, 125 events were reported (55 in the control group and 70 in the zoledronate group): nine local progressions or relapses (four in the control group and five in the zoledronate group), 95 metastases (43 in the control group and 52 in the zoledronate group), 19 local and metastatic lesions (seven in the control group and 12 in the zoledronate group), and two treatment-related deaths (one from cardiomyopathy in the control group and one from multiorgan failure in the zoledronate group before the first zoledronate infusion). Event-free survival at 3 years for all 315 randomly assigned patients was 60·3% (95% CI 64·5–65·9). The treatment effect of zoledronate was estimated as an HR of 1·36 (95% CI 0·95–1·96; p=0·094). Event-free survival at 3 years was 63·4% (95% CI 55·2–70·9) for the control group and 57·1% (48·8–65·0) for the zoledronate group (figure 3A). The treatment effect estimate was relatively stable in the post-hoc sensitivity analyses, that excluded 16 patients (five in the control group and 11 in the zoledronate group) with metastases other than lung (HR 1·3 [95% CI 0·89–1·90]; p=0·17, post-hoc analyses are detailed in the appendix p 24). No significant heterogeneity of the treatment effect was noted according to age (p=0·70), chemotherapy regimen (p=0·97), risk group (p=0·78), or tumour size (p=0·07; figure 3B). 74 deaths were reported at the time of analysis (cutoff, Jan 1, 2015; 32 in the control group and 42 in the zoledronate group. Two deaths were thought to be treatment related, one in each group of treatment. 3-year overall survival was 84·4% (77·3–89·6) in the control group and 73·4% (65·2–80·2) in the zoledronate group (figure 3C; 1·61 [0·995–2·61]; p=0·052). 3-year overall survival for all patients was 78·9% (73·6–83·4).

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

Articles

A 100 Event-free survival (%)

Based on the per-protocol dataset of patients who had received the assigned treatment, including 309 patients (156 in the control group and 153 in the zoledronate group), no significant increase in acute toxicity was noted during treatment in the zoledronate group than in the control group, except for a large excess of hypocalcaemia and hypophosphataemia (grade 3–4: 45 (29%) of 153 patients in the zoledronate group vs ten [6%] of 155 patients in the control group; p<0·0001) and hypophosphatemia (grade 3–4: 61 [40%] of 151 in the zoledronate group vs 26 [17%] of 156 in the control group; p<0·0001), and a slight increase of thrombocytopenia (figure 4; table 2). The proportion of patients who had a fracture during chemotherapy or wound-healing complications after surgery was similar between both groups (table 2).

80 60 40 20

Control Zoledronate

0 0

B

2

3

4

5

37 46

20 27

Years from randomisation

Number at risk Zoledronate 159 Control 156

126 132

81 95

62 72

Events (n)/patients (n)

HR (95% CI)

Zoledronate Control group group

Discussion Chemotherapy regimen HD-MTX

54/128

40/125

1·36 (0·90–2·06)

API-AI

16/31

15/31

1·38 (0·65–2·92)

<18

47/110

34/107

1·49 (0·96–2·32)

18–25

14/29

12/30

1·32 (0·54–3·20)

9/20

9/19

0·95 (0·37–2·43)

Age class (years)

>25 Risk group Standard

49/128

37/129

1·44 (0·94–2·20)

High

21/31

18/27

1·19 (0·59–2·39)

<10 cm

27/63

19/81

1·99 (1·09–3·66)

≥10 cm

41/92

35/72

0·97 (0·61–1·56)

Overall

70/159

55/156

Primary tumour size

1·36 (0·95–1·96)

0·1

1·0

12·0 Favours control

Favours zoledronate

C 100 Overall survival (%)

The results of this trial show that combining zoledronate to chemotherapy failed to improve event-free survival, percentage of good histological response, and overall survival in patients with osteosarcoma. To our knowledge, this study is the first randomised trial aimed to assess the efficacy of zoledronate combined with chemotherapy in patients with previously untreated osteosarcoma. Overall, the outcome of this large cohort of patients, including patients with an axial tumour or a metastatic disease, is slightly better than anticipated at the design stage (baseline 3-year event-free survival 55%), and is within the range of the results of the recent international EURAMOS-1 trial.2,15 The screening failure rate among potentially eligible patients (91 [22%] of 409) is comparable to other osteosarcoma trials.2 However, the chemotherapy backbone is different from the MAP chemotherapy regimen (methotrexate, doxorubicin, and cisplatin) accepted as a standard by most osteosarcoma study groups,2,15 which restricts the external validity of our findings. Furthermore, the wide eligibility criteria in this relatively small randomised trial might complicate the interpretation of the results; however, in our study, relative treatment effect of zoledronate seems rather homogeneous across the different risk groups. In view of the rarity of metastatic osteosarcoma, sufficiently sized randomised phase 3 trials are not feasible in metastatic patients only, and therefore such patients were included in our trial. The toxicity noted with these chemotherapy regimens was as expected, with no major increase of toxicity in patients receiving zoledronate, except for reversible hypocalcaemia and hypophosphataemia. We did not note any case of jaw osteonecrosis in this population of young patients. In view of preclinical observations,16 zoledronate could affect bone healing and bone ingrowth (important factors for reconstruction). In this trial, we did not note any significant difference in terms of orthopaedic complications (ie, fracture, wound-healing complications, or other complications leading to delayed postoperative

1

80 60 40 20

Control Zoledronate

0 0 Number at risk Zoledronate 159 Control 156

1

2

3

4

5

53 60

26 34

Years from randomisation 143 146

108 123

78 94

Figure 3: Event-free survival (A, B) and overall survival (C) (A) Kaplan-Meier estimates of event-free survival by treatment group. The shadowed bands represent the 95% Hall-Wellner confidence bands. At the time of this analysis (cutoff date Jan 1, 2015), 125 events were reported: 55 in the control group and 70 in the zoledronate group. (B) Forest plot of event-free survival according to subgroups. The HR of events estimated in a Cox proportional hazard model, stratified by age category combined with the type of chemotherapy and risk group: (1) chemotherapy regimen: HD-MTX=regimen containing high-dose methotrexate combined with etoposide-ifosfamide; API-AI=doxorubicin-cisplatin-ifosfamide; (2) risk group (standard risk, resectable primary and no definite metastasis vs high risk, unresectable primary or definite metastasis; and (3) primary tumour size (largest diameter <10 cm vs largest diameter ≥10 cm; seven missing data). (C) Kaplan-Meier estimates of overall survival by treatment group. The shadowed bands represent the 95% Hall-Wellner confidence bands. HR=hazard ratio.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

7

Articles

A

B

Hypocalcaemia* Hypophosphataemia† Peripheral neuropathy Dermatology/skin toxicity Cardiac toxicity Mucositis Renal toxicity‡ Thrombopenia Miscellaneous Gastrointestinal toxicity Anaemia Metabolic/laboratory toxicity Fever Central neurological toxicity Headache Mood alteration Febrile neutropenia Neutropenia Infection without neutropenia AST or ALT elevation Auditory/ear toxicity Wound-healing complication Weight loss Creatinine elevation Constitutional symptoms Bilirubin elevation§ 100

75

50

25

0

Control group (%; n=156) All grades

Severe toxicity

25

50

75

100

Zoledronate group (%; n=153) All grades

0·1

1

10

Relative risk (95% CI)

Severe toxicity

Figure 4: Adverse events (A) The proportion of patients who had an adverse event, any grade (light red for control group and light blue for zoledronate group), and a severe adverse event (dark red for control group and dark blue for zoledronate group) according to the randomisation group. Grade 4 haematological toxicities and grade 3 or higher of all extrahaematological toxicities were considered as severe toxicities as were fractures of any grade. (B) The relative risk of a severe adverse event in the zoledronate group compared with the control group, with 95% CIs. Adverse events were assessed after each chemotherapy course using a list of 25 selected items from the National Cancer Institute-Common Terminology Criteria for Adverse Events version 3.0; a free-text area was available to document other adverse events. Adverse event types for which there were less than five patients with this adverse event have been pooled in the category Miscellaneous. For each adverse event type, the analysis is based on the maximum grade observed over the whole treatment duration. The adverse event types are ordered by decreasing value of the relative risk. AST=aspartate aminotransferase. ALT=alanine aminotransferase. *Data for one participant missing from the control group. †Data for two participants missing from the zoledronate group. ‡Data for one participant missing from the zoledronate group. §Data for one participant missing from the zoledronate group.

chemotherapy) between the two groups. However, the potential long-term local and distant effects of zoledronate warrant a prolonged follow-up, including assessment of patient growth and bone mineral density, which is planned in our trial. Although the number of patients included in the trial is lower than initially planned, the absence of a positive effect of zoledronate on event-free survival in our patients is incontestable. The conditional power was practically null even under our optimistic hypothesis that the 3-year event-free survival would increase from 55% to 68%. Furthermore, the treatment effect was quite stable in the sensitivity analyses, and we did not note any significant heterogeneity of the treatment effect across the studied subgroups. Event-free survival was not improved, and some excess of events or deaths was observed in the zoledronate group with tests for differences being statistically 8

non-significant. Both progressions and relapses at the primary site and metastatic events were slightly increased in the zoledronate group. So far, no deleterious survival effect of zoledronate has been shown in studies done in adults with cancer, but rather beneficial effects. A large individual patient data meta-analysis showed a definite survival benefit with bisphosphonates in high-risk post-menopausal patients with early breast cancer.17 A survival advantage was also reported with zoledronate in various adult malignancies with spread to the bone.18–20 Zoledronate seemed to be a good candidate in the treatment of osteosarcoma, because several preclinical studies have suggested that it exerts pleiotropic antitumour effects (eg, antiproliferative, anti-angiogenic, and immunomodulatory effects) against osteosarcoma cells in vitro and in animal models. Two previous uncontrolled studies showed the feasibility of combining

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

Articles

Control group (n=156) Grade 1

Grade 2

Zoledronate group (n=153) Grade 3

Grade 4

Grade 1

Grade 2

Grade 3

Grade 4

Haematological Anaemia

6 (4%)

35 (22%)

81 (52%)

34 (22%)

7 (5%)

26 (17%)

82 (54%)

38 (25%)

Neutropenia

1 (1%)

1 (1%)

1 (1%)

150 (96%)

1 (1%)

2 (1%)

6 (4%)

142 (93%)

35 (22%)

16 (10%)

23 (15%)

67 (43%)

24 (16%)

12 (8%)

20 (13%)

87 (57%) 19 (12%)

Thrombocytopenia Metabolic/laboratory Hypocalcaemia*

80 (52%)

31 (20%)

8 (5%)

2 (1%)

23 (15%)

84 (55%)

26 (17%)

9 (6%)

12 (8%)

24 (15%)

2 (1%)

8 (5%)

41 (27%)

53 (35%)

8 (5%)

AST or ALT elevation

15 (10%)

5 (3%)

65 (42%)

56 (36%)

16 (10%)

14 (9%)

65 (42%)

47 (31%)

Bilirubin elevation‡

39 (25%)

25 (16%)

9 (6%)

2 (1%)

30 (20%)

25 (16%)

7 (5%)

Hypophosphataemia†

0

Gastrointestinal Mucositis Other

36 (23%)

44 (28%)

32 (21%)

5 (3%)

31 (20%)

39 (25%)

44 (29%)

6 (4%)

8 (5%)

101 (65%)

45 (29%)

1 (1%)

8 (5%)

91 (59%)

51 (33%)

1 (1%)

1 (1%)

6 (4%)

119 (76%)

6 (4%)

13 (8%)

45 (29%)

51 (33%)

Infection Febrile neutropenia Infection without neutropenia

0

2 (1%)

116 (76%)

3 (2%)

19 (12%)

0

36 (24%)

46 (30%)

2 (1%)

Constitutional symptoms Fever

46 (29%)

34 (22%)

4 (3%)

1 (1%)

54 (35%)

56 (37%)

5 (3%)

0

Weight loss

55 (35%)

54 (35%)

9 (6%)

0

59 (39%)

57 (37%)

7 (5%)

0

Other

30 (19%)

51 (33%)

14 (9%)

1 (1%)

39 (25%)

59 (39%)

10 (7%)

48 (31%)

10 (6%)

2 (1%)

2 (1%)

40 (26%)

14 (9%)

3 (2%)

0

5 (3%)

5 (3%)

3 (2%)

0

2 (1%)

3 (2%)

1 (1%)

3 (2%) 0

1 (1%)

Renal Creatinine elevation Other Neurological Headache

28 (18%)

28 (18%)

3 (2%)

0

37 (24%)

27 (18%)

3 (2%)

Mood alteration

9 (6%)

17 (11%)

3 (2%)

0

9 (6%)

20 (13%)

2 (1%)

1 (1%)

CNS

7 (4%)

23 (15%)

11 (7%)

1 (1%)

8 (5%)

30 (20%)

9 (6%)

3 (2%)

24 (15%)

19 (12%)

3 (2%)

16 (10%)

19 (12%)

5 (3%)

0

Auditory/ear toxicity

15 (10%)

16 (10%)

6 (4%)

1 (1%)

11 (7%)

28 (18%)

6 (4%)

0

Cardiac toxicity

32 (21%)

7 (4%)

0

2 (1%)

22 (14%)

6 (4%)

4 (3%)

0

Dermatology skin toxicity

63 (40%)

23 (15%)

7 (4%)

0

49 (32%)

36 (24%)

9 (6%)

1 (1%)

5 (3%)

2 (1%)

5 (3%)

0

6 (4%)

3 (2%)

4 (3%)

0

Peripheral neuropathy

0

Other

Wound-healing complication Fracture§¶

1 (1%)

9 (6%)

4 (3%)

1 (1%)

6 (4%)

5 (4%)

5 (4%)

0

Miscellaneous||

38 (24%)

52 (33%)

10 (6%)

1 (1%)

58 (38%)

34 (22%)

13 (8%)

0

Adverse events were assessed after each chemotherapy course using a list of 25 selected items from the Common Terminlogy Criteria for Adverse Events version 3.0, including creatinine and calcaemia levels. A free-text area was available to document other adverse events, which are documented here in the category Other. Two treatment-related deaths occurred (one from cardiomyopathy in the control group and one from multiorgan failure in the zoledronate group before the first zoledronate infusion). For each adverse event type, the analysis is based on the maximum grade observed over the whole treatment duration, in patients of the per-protocol population. AST=aspartate aminotransferase. ALT=alanine aminotransferase. *Grade cannot be determined for one. †Grade cannot be determined for two. ‡Grade cannot be determined for two. §Patients with a pathological fracture at diagnosis were excluded to estimate the proportion of patients experiencing a fracture during chemotherapy. ¶Grade cannot be determined for 23 (16 in the control group and seven in the zoledronate group). ||All adverse event types for which the total number of severe toxicities was below five (allergic reaction, cystitis, endocrine toxicity, genitourinary toxicity, haemorrhage, musculoskeletal toxicity, ocular/visual toxicity, pulmonary toxicity, reproductive function/sexual disorder, upper respiratory tract toxicity, vascular toxicity) have been pooled in the category Miscellaneous.

Table 2: Distribution of toxicity grades for each type of adverse event according to treatment group (per-protocol dataset)

bisphosphonates with chemotherapy in osteosarcoma,21,22 but the small number of patients included and the design of these trials precluded any conclusions about efficacy. Several hypotheses could explain the absence of a benefit with zoledronate combined with chemotherapy in our study. First, the slight imbalance in the primary tumour size between the treatment groups could lead to a residual confounding bias in the main analysis. However, results were very similar in our sensitivity

analysis that adjusted for primary tumour size. Nevertheless, some other confounding factors might not be equally distributed between randomised groups due to chance in this relatively small sized trial. Second, addition of zoledronate to chemotherapy might have increased toxicity and consequently reduced the dose intensity, but our findings do not support this hypothesis. Third, the dose of zoledronate used might be inadequate, especially in patients younger than 18 years.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

9

Articles

Clinical experience of zoledronate in children with cancer was restricted 9 years ago when the study was designed and the optimal dose in the pediatric oncology setting had not been established. The protocol dose of zoledronate (0·05 mg/kg) used in our trial was based on the experience in children treated for bone disorders23 and was lower than the recommended dose in adult patients with cancer (4 mg). Since the beginning of the trial, two paediatric dose-finding trials have been done in combination with chemotherapy: the maximum tolerated dose of zoledronate was defined at 4 mg/m² in combination with low-dose cyclophosphamide in neuroblastoma24 and at 2·3 mg/m² combined with methotrexate, doxorubicin, and cisplatin in metastatic osteosarcoma.22 These doses are higher than the one chosen for patients younger than 18 years in our trial. However, we treated older patients at higher doses and did not note significant heterogeneity of the treatment effect according to age category. Furthermore, a large meta-analysis individual patient data from randomised trials assessing biphosphonates in patients with breast cancer over 20 years17 has likewise shown that the benefit of bisphosphonate treatment in patients with high-risk early breast cancers seemed to be independent of the type of bisphosphonate, dose, and duration of treatment. Fourth, the absence of a benefit with zoledronate could also be related to hormonal factors. The survival benefit with bisphophonates in patients with breast cancer was higher in post-menopausal women than in pre-menopausal women.25 In a preclinical breast cancer model, the activity of zoledronate was linked to the endocrine status of mice: the growth of MDA-231 breast cancer cells in bone declined exclusively in ovariectomised mice, showing a link between the menopausal status and the antitumour effects of zoledronate.26 Conversely, in our trial, in which the mean age at diagnosis was 15 years, most patients were treated during the peak of pubertal bone development during which steroid sex hormones are known to play a major part. Fifth, a preclinical study recently suggested that osteoclasts might inhibit the migration of osteosarcoma cells.27 The zoledronate-induced reduction of osteoclasts might therefore increase the risk of lung metastases in osteosarcoma.27,28 Whereas in most animal models zoledronate was shown to reduce the risk of lung metastases in osteosarcoma,8,10,11 a few studies have reported the absence of efficacy28 or even increased zoledronate-induced metastatic spread.27 Sixth, the effect of zoledronate on the immune system has thus far mostly thought to be antitumourigenic via the activation of Vγ9Vδ2 T cells.29 However, the effect of zoledronate treatment on immunological parameters such as NK-cell expansion, or macrophage depletion or polarisation in the context of the bone microenvironment has yet to be addressed.30 This immunological effect could have a negative influence on the specific setting of 10

osteosarcoma in which, unlike other tumour types, a high number of tumour-infiltrating macrophages is associated with improved survival.31 Seventh, a further hypothesis to explain the absence of a benefit with zoledronate combined with chemotherapy in our study is the potential upregulation of RANK expression, known to promote osteosarcoma pathogenesis, by osteosarcoma cells following long-term zoledronate treatment.32 To understand the discordance between the results of this trial and preclinical data suggesting the efficacy of bisphosphonates in osteosarcoma, several ongoing translational studies are associated with the OS2006 clinical trial, aiming to assess the effect of zoledronate on parameters such as RANK expression, the immune response, and macrophage polarisation. In conclusion, from the results observed in this study, zoledronate cannot be recommended in patients with osteosarcoma. Contributors SP-N, M-CLD, FR, HP, PM-B, J-CG, NE-W, NC, NP, M-DT, EM, FG, J-YB, and LB designed the trial. S-PN, HP, PM-B, PP, HB, CL, J-CG, NE-W, AI, NC, EB, NP, M-DT, AG-B, J-MG, EM, FG, J-YB, and LB provided study material or patients. M-CLD, AC, and NB were responsible for collection and data assembly. M-CLD and AC did the statistical analysis. All authors were associated with the interpretation of the results. SP-N, M-CLD, and LB wrote the first draft of the report. All authors contributed to subsequent drafts and made the decision to submit the report for publication. Declaration of interests M-CLD has received grants from TAKEDA outside of the submitted work. FG has received personal fees from Amgen, grants from Novartis France, grants from Takeda, and is a founding shareholder for Atlantera outside of the submitted work, and has a patent for a surgical device (femoral stem) with royalties paid. LB has received grants and personal fees from MILLENIUM outside of the submitted work. All other authors declare no competing interests. References 1 Isakoff MS, Bielack SS, Meltzer P, Gorlick R. Osteosarcoma: current treatment and a collaborative pathway to success. J Clin Oncol 2015; 33: 3029–35. 2 Bielack SS, Smeland S, Whelan JS, et al. 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. J Clin Oncol 2015; 33: 2279–87. 3 Bielack SS, Kempf-Bielack B, Delling G, et al. Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols. J Clin Oncol 2002; 20: 776–90. 4 Ohba T, Cates JMM, Cole HA, et al. Pleiotropic effects of bisphosphonates on osteosarcoma. Bone 2014; 63: 110–20. 5 Van Poznak CH, Von Roenn JH, Temin S. American society of clinical oncology clinical practice guideline update: recommendations on the role of bone-modifying agents in metastatic breast cancer. J Oncol Pract 2011; 7: 117–21. 6 Yuasa T, Kimura S, Ashihara E, Habuchi T, Maekawa T. Zoledronic acid—a multiplicity of anti-cancer action. Curr Med Chem 2007; 14: 2126–35. 7 Horie N, Murata H, Kimura S, et al. Combined effects of a third-generation bisphosphonate, zoledronic acid with other anticancer agents against murine osteosarcoma. Br J Cancer 2007; 96: 255–61. 8 Ory B, Heymann M-F, Kamijo A, Gouin F, Heymann D, Redini F. Zoledronic acid suppresses lung metastases and prolongs overall survival of osteosarcoma-bearing mice. Cancer 2005; 104: 2522–29. 9 Heymann D, Ory B, Blanchard F, et al. Enhanced tumor regression and tissue repair when zoledronic acid is combined with ifosfamide in rat osteosarcoma. Bone 2005; 37: 74–86.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

Articles

10 11

12

13

14 15

16 17

18

19

20

21

22

Dass CR, Choong PFM. Zoledronic acid inhibits osteosarcoma growth in an orthotopic model. Mol Cancer Ther 2007; 6: 3263–70. Koto K, Horie N, Kimura S, et al. Clinically relevant dose of zoledronic acid inhibits spontaneous lung metastasis in a murine osteosarcoma model. Cancer Lett 2009; 274: 271–78. Le Deley M-C, Guinebretière J-M, Gentet J-C, et al. SFOP OS94: a randomised trial comparing preoperative high-dose methotrexate plus doxorubicin to high-dose methotrexate plus etoposide and ifosfamide in osteosarcoma patients. Eur J Cancer 2007; 43: 752–61. Assi H, Missenard G, Terrier P, et al. Intensive induction chemotherapy without methotrexate in adult patients with localized osteosarcoma: results of the Institut Gustave-Roussy phase II trial. Curr Oncol 2010; 17: 23–31. Lan KG, DeMets DL. Discrete sequential boundaries for clinical trials. Biometrika 1983; 70: 659–63. Smeland S, Whelan J, Bielack S. Event-free survival and overall survival in 2,253 patients with osteosarcoma registered to EURAMOS-1. Proc Am Soc Clin Oncol 2015; 33 (supp): abstr 10512. Matos MA, Tannuri U, Guarniero R. The effect of zoledronate during bone healing. J Orthop Traumatol 2010; 11: 7–12. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG), Coleman R, Powles T, et al. Adjuvant bisphosphonate treatment in early breast cancer: meta-analyses of individual patient data from randomised trials. Lancet 2015; 386: 1353–61. Zarogoulidis K, Boutsikou E, Zarogoulidis P, et al. The impact of zoledronic acid therapy in survival of lung cancer patients with bone metastasis. Int J Cancer 2009; 125: 1705–09. Wang X, Yan X, Li Y. A meta-analysis of the antitumor effect and safety of bisphosphonates in the treatment of multiple myeloma. Int J Clin Exp Med 2015; 8: 6743–54. Ahn SG, Kim SH, Lee HM, Lee SA, Jeong J. Survival benefit of zoledronic acid in postmenopausal breast cancer patients receiving aromatase inhibitors. J Breast Cancer 2014; 17: 350–55. Meyers PA, Healey JH, Chou AJ, et al. Addition of pamidronate to chemotherapy for the treatment of osteosarcoma. Cancer 2011; 117: 1736–44. Goldsby RE, Fan TM, Villaluna D, et al. Feasibility and dose discovery analysis of zoledronic acid with concurrent chemotherapy in the treatment of newly diagnosed metastatic osteosarcoma: a report from the Children’s Oncology Group. Eur J Cancer 2013; 49: 2384–91.

23

24

25

26

27

28

29 30

31

32

Högler W, Yap F, Little D, Ambler G, McQuade M, Cowell CT. Short-term safety assessment in the use of intravenous zoledronic acid in children. J Pediatr 2004; 145: 701–04. Russell HV, Groshen SG, Ara T, et al. A phase I study of zoledronic acid and low-dose cyclophosphamide in recurrent/refractory neuroblastoma: a new approaches to neuroblastoma therapy (NANT) study. Pediatr Blood Cancer 2011; 57: 275–82. Coleman R, Cameron D, Dodwell D, et al. Adjuvant zoledronic acid in patients with early breast cancer: final efficacy analysis of the AZURE (BIG 01/04) randomised open-label phase 3 trial. Lancet Oncol 2014; 15: 997–1006. Ottewell PD, Wang N, Brown HK, et al. Zoledronic acid has differential antitumor activity in the pre- and postmenopausal bone microenvironment in vivo. Clin Cancer Res 2014; 20: 2922–32. Endo-Munoz L, Cumming A, Rickwood D, et al. Loss of osteoclasts contributes to development of osteosarcoma pulmonary metastases. Cancer Res 2010; 70: 7063–72. Labrinidis A, Hay S, Liapis V, Findlay DM, Evdokiou A. Zoledronic acid protects against osteosarcoma-induced bone destruction but lacks efficacy against pulmonary metastases in a syngeneic rat model. Int J Cancer 2010; 127: 345–54. Kunzmann V, Bauer E, Wilhelm M. Gamma/delta T-cell stimulation by pamidronate. N Engl J Med 1999; 340: 737–38. Junankar S, Shay G, Jurczyluk J, et al. Real-time intravital imaging establishes tumor-associated macrophages as the extraskeletal target of bisphosphonate action in cancer. Cancer Discov 2015; 5: 35–42. Buddingh EP, Kuijjer ML, Duim RAJ, et al. Tumor-infiltrating macrophages are associated with metastasis suppression in high-grade osteosarcoma: a rationale for treatment with macrophage activating agents. Clin Cancer Res 2011; 17: 2110–19. Abe T, Sato T, Kokabu S, Hori N, Shimamura Y, Sato T, Yoda T. Zoledronic acid increases the circulating soluble RANKL level in mice, with a further increase in lymphocyte-derived soluble RANKL in zoledronic acid- and glucocorticoid-treated mice stimulated with bacterial lipopolysaccharide. Cytokine 2016; 83: 1–7.

www.thelancet.com/oncology Published online June 17, 2016 http://dx.doi.org/10.1016/S1470-2045(16)30096-1

11