A Phase 2 Trial of RANKL Antibody, Denosumab, in Two Cohorts of Patients with Recurrent / Refractory Osteosarcoma, A Report from the Children’s Oncology Group (COG)

Abstract

Background:

Mouse models demonstrate a role for RANK and its ligand, RANKL, in osteosarcoma. The primary objective of this single arm open label phase 2 trial was to determine if denosumab, a RANKL monoclonal antibody, improved disease control in recurrent osteosarcoma relative to benchmarks derived from historical COG clinical trial data.

Patients and Methods:

Skeletally mature patients 11–49 years old with measurable disease were eligible for cohort 1 or those with complete surgical resection of all sites of disease were eligible for cohort 2. Patients received denosumab 120 mg subcutaneously every 4 weeks with calcium and vitamin D supplementation. The primary endpoints were RECIST response and remaining event-free for 4 months for cohort 1 and remaining event-free for 12 months for cohort 2. Toxicity, pharmacokinetics, and pharmacodynamic effects were additional endpoints.

Results:

Fifteen patients in cohort 1 and 38 in cohort 2 were eligible and evaluable for the primary endpoint. One of 15 cohort 1 patients remained event-free at 4 months. There were no objective responses. Ten of 38 cohort 2 patients were event free at 12 months. The pre-defined efficacy criteria were not met in either cohort. The most common ≥ grade 3 adverse events were hypocalcemia and hypophosphatemia (8% and 11% respectively). At steady state, mean serum denosumab trough concentrations were 23.7–31 ug/mL in cycle 2–7. Serum c-telopeptides and urine n-telopeptides decreased on treatment.

Conclusions:

Denosumab was well tolerated with anticipated side effects, and PK and PD parameters but had insufficient activity for further development in osteosarcoma.

Introduction

In contrast to other pediatric cancers, therapy and outcomes in osteosarcoma have not changed significantly for the past 4 decades¹. Metastatic recurrence or progression of disease primarily occurs in the lungs and remains the primary reason for treatment failure. Beyond resection of all sites of recurrent disease, there is no standard of care therapy for patients who develop recurrent osteosarcoma; outcomes for these patients remain dismal with an event-free survival of 20% or less². Identification of novel therapeutic agents remains a high priority and may be achieved, in part, by conducting phase 2 trials of novel agents in children with recurrent osteosarcoma to quickly identify agents of potential activity in this deadly disease.

Osteosarcoma is unique in that tumor stroma consists of a calcified matrix that persists even with significant tumor necrosis. Because of this, objective response rate is not a good surrogate marker for progression-free or overall survival.^3,4 By analyzing outcomes from previously completed pediatric cooperative groups (Pediatric Oncology Group, Children’s Cancer Group, Children’s Oncology Group) phase 2 trials enrolling patients with recurrent/refractory osteosarcoma, the Children’s Oncology Group (COG) Bone Tumor Committee developed a clinical trial design with osteosarcoma specific endpoints.² This design accounts for the two cohorts of patients with recurrent osteosarcoma typically encountered in clinical practice: 1) those with unresectable, measurable disease (per Response Evaluation Criteria In Solid Tumors, RECIST, criteria); and 2) those with resectable metastases. In patients with unresectable disease, we utilize a single arm, phase 2 trial with dual endpoints of disease control at 4 months defined as being event free at 4 months or confirmed objective response. This is compared to the historical experience of 96 patients enrolled on phase 2 trials conducted by the COG where the Event Free Survival at 4 months was 12% [95% CI: 6.0%–19%]. Similarly, in patients with resectable disease limited to pulmonary parenchyma, 12-month EFS is compared to a historical benchmark.

Denosumab is a fully human monoclonal antibody to the receptor activator of nuclearfactor-κB ligand (RANKL). RANKL and its receptor RANK have the physiologic function of regulating bone turnover. In response to RANKL-RANK binding, osteoclast precursors differentiate and become activated resulting in bone resorption.⁵ Denosumab has been approved by the Food and Drug Administration for prevention of skeletal-related events in adults with solid tumor bone metastases (Prolia), for the treatment of hypercalcemia of malignancy refractory to bisphosphonate therapy, and for adults and skeletally mature adolescents with unresectable giant cell tumor of bone (Xgeva), a tumor with high RANKL expression. Osteosarcoma expresses both RANK and RANKL and, in studies of human tumors and in clinically relevant models, RANK-RANKL interaction modulates gene expression.^6,7,8,9 In a genetically engineered mouse model (GEMM) of osteosarcoma in which expression of the SV40 T/t antigen from the osteocalcin promoter results in highly penetrant osteosarcoma, heterozygous deletion of Prkar1a accelerates development of osteosarcoma via increased RANKL gene expression. In this GEMM, inhibition of RANK-RANKL binding arrests tumor progression, decreases development of pulmonary metastases and improves survival.^10,11 In addition, inhibiting RANK-RANKL interaction in murine osteosarcoma orthotopic models consistently decreased primary tumor burden and variably reduced pulmonary metastatic burden.^12,13,14

Given these data and the paucity of novel available agents in osteosarcoma, we designed a phase 2 trial using denosumab in two cohorts of patients with recurrent/refractory osteosarcoma: cohort 1 consists of those with measurable disease burden and cohort 2 consists of those who had achieved complete surgical resection. The primary objective of this single arm open label phase 2 trial was to determine if denosumab improved disease control in recurrent osteosarcoma relative to benchmarks derived from historical COG clinical trial data. The primary endpoint was response or remaining event-free for 4 months for cohort 1 and event-free rate at 12 months for cohort 2. Secondary objectives included evaluation of denosumab tolerability, pharmacokinetics and pharmacodynamics in this patient population. In addition, the association between RANK/RANKL protein expression with immunohistochemistry (IHC) and disease control was evaluated.

Material and methods

Eligibility

Eligible patients included patients ≥ 11 and < 50 years of age with recurrent or refractory osteosarcoma. Cohort 1 consisted of those with measurable disease as defined by RECIST 1.1; cohort 2 consisted of those who underwent complete resection of all sites of metastatic disease within 30 days of study enrollment. Additionally, cohort 2 patients with lung-only resected metastatic disease had to have refused participation in AOST1421, a clinical trial for this patient population that accrued before and then concurrent with this study.¹⁵ Female patients required a bone age equal to or greater than 12 years; male patients required a bone age equal to or greater than 14 years. Other eligibility criteria included adequate renal, cardiac, and liver functions along with adequate bone marrow function. Adequate performance status (corresponding to ECOG scores of 0, 1, 2) was required. Patients were excluded for prior use of denosumab, mithramycin, strontium-89, samarium-153, or rhenium. Patients receiving bisphosphonates were also excluded. Patients were also excluded if they had disorders of abnormal bone metabolism, Paget’s disease, hypocalcemia uncorrectable by oral calcium supplementation, vitamin D < 20 ng/mL, osteonecrosis of the jaw, or significant dental conditions. Pregnant or breast-feeding women were also excluded.

This trial was approved by the National Cancer Institute (NCI) Pediatric Central Institutional Review Board (CIRB), and by local regulatory boards at all sites not utilizing NCI Pediatric CIRB and was conducted in accordance with the U.S. Common Rule. Informed consent documentation was signed by all patients or their parent/legal guardian, and assent was obtained according to local institutional guidelines prior to enrollment. The study was registered on ClinicalTrials.gov as study NCT02470091. The trial was open starting November 2015 and it completed accrual in July of 2017. Data current to September 30, 2022 were used in this analysis.

Treatment and Response Assessment

Denosumab was administered at a dose of 120 mg subcutaneously every 4 weeks (28 days). Two additional loading doses were given on Days 8 and 15 of Cycle 1 only (Supplemental Figure 1). In the absence of progressive disease or unacceptable toxicity, denosumab was continued for up to 24 months in total or 26 cycles, whichever occurred first. While on protocol therapy, calcium (500 mg daily or greater) and vitamin D (400 IU or greater) supplementation were required unless the patient developed hypercalcemia. Calcium and phosphorus levels were monitored weekly during cycle 1, on day 1 of every subsequent cycle and monthly for 1 year after completing protocol therapy.

Toxicity was evaluated using Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. Dose-limiting toxicities were any ≥ Grade 3 event not considered attributable to osteosarcoma except hypocalcemia ≥ Grade 3 corrected with medical therapy, hypophosphatemia ≥ Grade 3 corrected with medical therapy, ≥ Grade 3 cardiac toxicity in a subject who previously received doxorubicin and ≥ Grade 3 ototoxicity in a subject who previously received cisplatin. Osteonecrosis of the jaw (ONJ) and fractures (excluding pathologic fracture at the site of an osteosarcoma bony metastasis) were considered DLTs and required cessation of therapy. Dose delays were permitted for those with grade 3 or 4 adverse events related to denosumab. Denosumab was resumed when the event resolved to grade 1 or returned to baseline. If toxicity prevented the start of a cycle by 2 or more weeks or if there was grade 3 or 4 allergic or recurrence of grade 2 allergic reaction, then the patient was removed from protocol therapy.

Patients were evaluated for response or recurrence of disease using RECIST 1.1 prior to every odd numbered cycle.

Pharmacokinetics (PK) and Pharmacodynamics (PD)

The first 15 evaluable patients < 18 years of age enrolled on Cohort 1 were required to have serial blood samples for the assessment of denosumab serum concentrations. Pharmacokinetic sampling was optional for all other patients. Samples were obtained on days 1, 8, 15, and 22 of cycle 1. PK samples were collected on day 1 of cycles 2–4, 6 and 7. An additional sample was collected on day 15 of cycle 6 if patients gave consent. Serum was isolated and analyzed for denosumab using a validated enzyme-linked immunosorbent assay (ELISA).

PD effects were measured by urinary N-terminal telopeptides normalized to urinary creatinine ratio (uNTx/Cr), and serum C-terminal telopeptides (sCTx). PD sampling was required for all patients and analyzed according to the treating institution’s laboratory procedures. Serum and urine samples for PD were collected on days 1, 8, 15, and 22 of cycle 1 and on day 1 of cycles 2, 3, 4, and 7.

RANK/RANKL Expression in Pre-treatment Tumor Specimens

Archival tumor specimen (formalin fixed paraffin embedded) from a previous biopsy/resection (most recent metastatic site was preferred) was required for evaluation of RANK/RANKL expression measured by immunohistochemistry (IHC). IHC was performed by NeoGenomics using previously described methods and antibodies.¹⁶ RANK expression was assessed for the tumor cell membrane (RANKm) and cytoplasm (RANKc) while RANKL was assessed for cytoplasmic staining only. The proportion of cells with 0, 1+, 2+ or 3+ staining was recorded. Patients were classified as having positive RANK/RANKL expression if they had > 0% of cells with 1–3+ staining for RANKm, RANKc or RANK-L. Patient were classified as negative RANK/RANKL expression if there was 0% of cells with 1–3+ staining for RANKm, RANKc and RANKL. Patients were classified as unknown for RANK/RANKL expression if there was no sample submitted or none of the submitted samples were adequate for testing. If samples were available from multiple procedures, the most recently obtained specimen was used in the analysis of RANK/RANKL IHC.

Statistical Considerations

Any eligible patient who received at least one dose of denosumab was considered evaluable for response and disease control. For cohort 1, each patient was evaluated for two outcome measures: (1) 4-month disease control success (DCS₄); and (2) RECIST response (CR or PR). A 2-stage design was utilized. For cohort 2, one group of 39 patients was enrolled. The outcome measure for this cohort was 12-month disease control success (DCR₁₂). The statistical properties of evaluation rules for the two cohorts and for monitoring for excessive DLT are presented in an appendix available online. EFS was defined as time from study enrollment until disease progression or death whichever occurred first. Overall survival (OS) was defined as time from study entry until death from any cause. Patients who did not experience an EFS event or death were censored at the date of last follow-up. EFS and OS as a function of time since study enrollment were estimated according to the method of Kaplan and Meier.

The maximum grade of each adverse event in the NCI CTCAE version 5 that that was experienced by each individual while on protocol therapy was determined. These maximum grades were tabulated according to the cohort of enrollment.

All eligible patients who received at least one dose of denosumab and for whom at least one PK or PD sample was submitted were considered evaluable for PK or PD, respectively. PK and PD data were analyzed with descriptive statistics.

Additional details on statistical approach are provided in the Appendix.

Results

Patient Characteristics

A total of 56 patients were enrolled. Two patients were later considered ineligible. One patient did not receive treatment prior to being withdrawn from protocol therapy and was considered inevaluable. Fifty-three patients were eligible and evaluable for efficacy and toxicity. Baseline characteristics for these 53 evaluable patients are shown in Table 1. Median age in cohort 1 was 15 years and in cohort 2 was18 years. Of all 53 patients, 34 (64%) were < 18 years old. In both cohorts, the majority of patients were male (60%) and 70% had lung only recurrent disease. In cohort 2, 53% of patients had one prior recurrence. Of all 53 patients 34 (64%) were < 18 years old and 42 patients consented to PK sample collection. Representativeness of the study population is discussed in Supplemental Table 1.

Table 1.

Baseline characteristics of 53 patients with recurrent osteosarcoma

	AOST1321 Recurrent Osteosarcoma Cohort
	Cohort 1: Measurable (N=15)	Cohort 2: Resected (N=38)	Total (N=53)
Age in Years at Enrollment, Median (Range)	15.0 (11.0, 17.0)	17.5 (12.0, 36.0)	16.0 (11.0, 36.0)
Age at Enrollment (years), n (%)
11–17	15 (100%)	19 (50%)	34 (64%)
18+	0 (0%)	19 (50%)	19 (36%)
Sex, n (%)
Female	5 (33%)	16 (42%)	21 (40%)
Male	10 (67%)	22 (58%)	32 (60%)
Previous Number of Recurrences, n (%)
0 (refractory)		2 (5%)	2 (5%)
1 recurrence		20 (53%)	20 (53%)
>1 recurrence		16 (42%)	16 (42%)
Missing	15*	0	15
Sites of Initial Recurrence, n (%)
1 Lung only	9 (60%)	28 (74%)	37 (70%)
2 Lung & Bone	4 (27%)	3 (8%)	7 (13%)
3 Bone only	2 (13%)	7 (18%)	9 (17%)
Metastatic Disease at Initial Diagnosis, n (%)
No	8 (53%)	26 (68%)	34 (64%)
Yes	7 (47%)	12 (32%)	19 (36%)
Interval from Diagnosis to Initial Recurrence (months), Median (Range)	6.4 (2.3, 48.8)	19.4 (1.6, 106.1)	15.4 (1.6, 106.1)

^*Not collected for this cohort

Disease Control and Response

In Cohort 1, only 1 of 15 evaluable patients had a disease control success by remaining progression-free for 4 months. There were no confirmed objective responses. Because none of the 15 patients experienced RECIST CR or PR and there were two or fewer patients who experienced DCS₄, enrollment to this cohort was stopped after stage 1 with the conclusion that denosumab did not demonstrate sufficient activity to warrant further investigation in this cohort of patients. The estimate and 95% confidence interval (CI) of the probability of DCS₄ are 6.7% and 0.17%–32%, respectively. The estimate of the probability of response and 95% CI are 0% and 0%–18%, respectively.

In Cohort 2, 10 of the 38 evaluable patients did not experience an EFS event prior to month 12. The remaining 28 evaluable individuals enrolled to Cohort 2 experienced an EFS event prior to month 12. As 15 or fewer of the 38 evaluable patients in this cohort demonstrated disease control at 12 months, denosumab did not demonstrated sufficient activity for further development in this cohort of patients.

Figure 1 shows event free survival (EFS) and overall survival estimates. EFS of patients in cohort 1 with measurable disease at 4-months was 7% (95% CI 0.4% to 26%). EFS in cohort 2 with resectable disease at 12-months was 26% (95% CI 14% to 41%).

Figure 1.

EFS (A) and OS (B) from study enrollment for cohorts 1 and 2

Pharmacokinetics and pharmacodynamics

Forty-two patients consented to PK sample collection (15 patients from cohort 1, 27 patients from cohort 2) during cycles 1–7 (Supplemental Table 2). Two patients had no serum trough concentration data for PK analysis. Following weekly loading doses on days 8 and 16, the serum trough concentration reached a plateau at day 22 after the loading doses on days 8 and 15 with mean concentration of 31.6 ug/mL (SD 13 ug/mL, Figure 2). Cycle 2 through cycle 7 mean trough concentrations ranged from 23.7 to 31.2 ug/mL. There was a difference in the mean trough concentration at cycle 2, day 1 between males and females (25.2 ug/mL and 38.7 ug/mL, respectively, p=0.0001; Figure 2).

Figure 2.

Trough denosumab concentrations after administration of denosumab 120 mg monthly with two loading doses in cohort 1 (A) and 2 (B) and mean serum trough concentrations at cycle 2, Day 1 for all patients showing higher concentrations with greater interpatient variability in females (C).

Baseline cycle 1, day 1 mean urine n-telopeptide to creatinine ratios (nMol BCE/mmol creatinine) were 322 (range: 55 to 1220) and 222 (range: 20 to 1703) in cohorts 1 and 2 respectively. Mean baseline serum c-telopeptide was 1,713 pg/uL (range: 237 to 7700) and 1012 (range: 213 to 2081) in cohorts 1 and 2 respectively. These bone resorption markers decreased over time (Supplemental Figure 2).

Immunohistochemical staining of RANK/RANKL

30 patients (7 in cohort 1 and 23 patients in cohort 2) had sufficient tumor material submitted for RANK/RANKL IHC. All tumor samples were from the time of relapse/progression. Of these 30 patients, 12 patients (40%) had positive RANK/RANKL expression (2 in cohort 1 and 10 in cohort 2), and 18 (60%) had negative expression (5 in cohort 1 and 13 in cohort 2). Six patients (20%) had tumor cell membranous RANK (RANKm) expression and these patients also had positive RANKc and RANKL tumor cell expression (Figure 3). Three patients had only RANKL expression and one patient had only RANKc expression. RANK/RANKL positivity was not significantly associated with disease control, or risk for EFS-event (Figure 3). In a post hoc analysis in the resected disease patients enrolled on cohort 2, there was not a statistically significant difference in overall survival in patients with or without RANK/RANKL expression (p-value 0.109).

Figure 3.

RANKL and RANK protein expression by IHC in osteosarcoma samples. Relationship between RANKL, RANK membranous and RANK cytoplasmic expression by IHC (A) and between RANK/RANKL positivity and EFS (B) and OS (C).

Treatment Tolerability

Denosumab was generally well-tolerated with no unanticipated toxicities (Table 2). Most adverse events in cohort 1 were attributed to disease progression. Hypocalcemia and hypophosphatemia were the most common ≥ Grade 3 toxicities with definite, possible, or probable attribution to denosumab occurring in 4 (7.5%) and 9 (17%) of 53 evaluable patients, respectively (Supplemental Table 3). Hypocalcemia and hypophosphatemia most often occurred during cycles 1 and 2. One (1) incident of DLT manifested as grade 3 hypophosphatemia probably related to denosumab was reported during cycle 1 out of 378 cycles delivered to patients. The point estimate of the per-cycle probability of DLT was 0.57%. There was one episode of grade 3 hypercalcemia with associated renal dysfunction 7 months after discontinuing denosumab. There were no episodes of osteonecrosis of the jaw or other skeletal abnormalities noted on protocol therapy.

Table 2.

Adverse events

System Organ Class	AE Term
		Cohort 1: Measurable (N=15)		Cohort 2: Resected (N=38)
		Grade 1–2	Grade 3–5	Grade 1–2	Grade 3–5
		N (%)	N (%)	N (%)	N (%)
Blood/Lymphatic	Anemia		1 (7)		1 (3)
Cardiac	Sinus tachycardia	1(7)
Gastrointestinal	Nausea		1 (7)
	Vomiting	1(7)	1 (7)
General/Administration	Fever	2(13)
	Non-cardiac chest pain		1 (7)
Infections/Infestations	Lung infection		1 (7)
Investigations	Lymphocyte count decreased		1 (7)
	Weight gain				1 (3)
	Weight loss		1 (7)
Metabolism/Nutrition	Anorexia		1 (7)
	Hypercalcemia				1 (3)
	Hyperglycemia		1 (7)
	Hypocalcemia		3 (20)		1 (3)
	Hypokalemia				1 (3)
	Hypophosphatemia		2 (13)		9 (24)
	Obesity				1 (3)
Neoplasms	Neoplasms benign, malignant - Other, specify: Progressive Disease		11 (7)
Nervous	Nervous system disorders - Other, specify: Spinal Cord Compression		1 (7)
Respiratory/Thoracic/Mediastinal	Dyspnea		3 (20)
	Hypoxia		1 (7)
	Pleural effusion	1(7)	1 (7)		1 (3)
	Pleuritic pain		1 (7)
	Pulmonary edema	1(7)
	Respiratory failure		11 (7)
Vascular	Hematoma			1 (3)
	Thromboembolic event	1(7)

¹Two instances of grade 5 AE unrelated to denosumab were reported in two patients, each during cycle 1. Both patients were removed from protocol therapy because of progressive disease during cycle 1.

Discussion

The RANK/RANKL pathway is important in maintaining normal bone homeostasis and response to stress.¹⁷ Recent studies in various models suggested that this pathway may play a role in osteosarcoma pathogenesis and that inhibiting the RANK-RANKL interaction would suppress osteosarcoma growth and possibly metastases.^{10,11,12,13,14} Treating patients with relapsed/refractory osteosarcoma with denosumab, targeting the osteosarcoma-stroma interaction by inhibiting RANK/RANKL, was therefore hypothesized to improve EFS in this patient population. Unfortunately, in this COG phase 2 trial of single agent denosumab in patients with measurable or fully resected recurrent/refractory osteosarcoma, no objective responses were seen and an insufficient number of patients reached the protocol-defined disease control endpoint to suggest denosumab has clinically relevant activity in this patient population. Nevertheless, the relatively rapid accrual of 53 patients with this rare cancer in less than 2 years, allowing us to assess Denosumab activity in both measurable and resected disease, is a success.

Preclinical data for denosumab demonstrated that the RANK/RANKL axis was potentially important in osteosarcoma. Osteosarcoma cell lines and tumor samples from humans and canines were reported express both RANK and RANKL.^6,7,8 Several xenograft and genetically engineered mouse models have shown that inhibition of RANK/RANKL axis resulted in decreased numbers of metastatic lung nodules, decreased tumor volumes, and inhibiting RANK-RANKL interaction improved survival in GEMMs.^{10,11,12,13,14} Several reasons might account for the discrepancy between preclinical findings and clinical trial results. RANK/RANKL expression in human osteosarcoma may not be as common as previously reported or may have inter- and intra-patient heterogeneity. While several studies showed RANK expression in 60–70% of human and 70% of canine osteosarcoma tumor samples^6–8,9, our study showed only 20% of tumors had RANK expression on the cell membrane (RANKm) and more recent studies showed lower rates of RANK and RANKL expression.¹⁸ Even in patients with RANKm expression, the proportion of tumor cells with RANKm expression was at most 30%, indicating intra-tumor heterogeneity with regards to RANKm expression. In addition, amongst the tumors with RANKm, 50% had only weak or moderate (1+ or 2+) expression. Thus, future preclinical investigation and clinical trials in osteosarcoma should address inter- and intra-patient heterogeneity. For this reason, it is very important for tumor material to be collected on a majority of patients enrolled on osteosarcoma clinical trials. In order for this to occur, sample acquisition, processing and storage will need to take into account future research needs. In osteosarcoma, receptor expression has been an unreliable predictor of response to monoclonal antibodies. For example, monoclonal antibodies against IGF1R have not shown robust activity in this disease despite receptor expression¹⁹. Hence, binding of a monoclonal antibody to its target (in this case denosumab to RANKL) may be insufficient in osteosarcoma and either combination with other modalities such as chemotherapy or alternative targeting approaches such as antibody-drug conjugates may be needed. One additional factor could be a lack of understanding of the degree of treatment effect needed in murine osteosarcoma models in order for preclinical results to translate into measurable treatment effect in humans with recurrent disease.

Although not achieving statistical significance, in patients enrolled in cohort 2 (resectable) with tumors expressing RANK/RANKL there was a trend towards better overall survival. It is possible that RANK/RANKL expression is a biomarker for better post-relapse outcomes in patients who have disease that is amenable to resection. Given small patient numbers, this finding needs to be validated in larger patient populations.

Based on the PK and PD data from participants in this trial, it is unlikely that inadequate drug exposure is responsible for the negative trial result. Overall, mean serum denosumab trough concentrations were similar to that reported in adults with giant cell tumor of bone from months 3 to 13 receiving the same dose and schedule (19.9–27.5ug/mL). In addition, markers of bone turnover decreased as expected and as observed in patients with giant cell tumor receiving the same dose and schedule of denosumab.²¹

In general, toxicities were mild and expected. Hypercalcemia with associated renal dysfunction was noted in one patient at 7 months after completion of therapy; this patient received two years of denosumab treatment as part of cohort 2 of this study. This and other published reports of hypercalcemia in younger patients following cessation of denosumab^20,21 emphasize the need for patient education regarding symptoms of hypercalcemia, monitoring following cessation of therapy and prompt testing for symptomatic patients. Previous reports of denosumab highlighted an increased risk of osteonecrosis of the jaw and atypical femoral fractures. These particular issues were not noted in our patient population.

In conclusion, denosumab appears to be well-tolerated even in this patient population who received prior treatment with intensive multi-agent chemotherapy. However, inhibition of the RANK/RANKL by denosumab did not appear to improve the event free survival for patients with recurrent/refractory osteosarcoma who have either measurable disease or fully resected disease. In osteosarcoma, traditional endpoints indicating therapeutic success may not be reliable. Use of unconventional clinically relevant endpoints (such as EFS at 4 months used in cohort 1 patients, and the use of EFS at 12 months in cohort 2 patients) may aid in the rapid assessment of novel agents. This and other similarly designed phase 2 studies of novel agents in recurrent/refractory osteosarcoma can accrue patients rapidly and inform whether clinical activity warrants further evaluation in subsequent clinical trials.^15,22,23 However, the challenge remains lack of activity of this and other recent similarly studied novel agents underscoring the need to identify, in preclinical studies, relevant biomarkers and combination strategies.

Translational Relevance.

A phase 2 trial of RANKL (receptor activator of nuclear factor-κB) antibody denosumab in patients age 11–49 years with recurrent osteosarcoma was conducted based on preclinical activity from mouse models. This was a single-arm phase 2 trial employing osteosarcoma-specific endpoints and leveraging benchmarks for event-free survival derived from historical Children’s Oncology Group (COG) trials. Denosumab pharmacokinetics and pharmacodynamics were as expected. Despite preclinical rationale, insufficient disease control was observed in the trial to warrant further investigation of denosumab in osteosarcoma. Forty percent of tumors expressed RANK and/or RANKL, with only 20% expressing RANK on the cell membrane. Disease control was not associated with the presence of absence of RANK/RANKL expression. This study provides insight into bone turnover and RANK/RANKL expression in osteosarcoma. Further work is needed to better understand the role of RANK/RANKL signaling in human osteosarcoma.

Data Availability

The data analyzed in this study are available in summary form in the results reporting section of ClinicalTrials.gov as study NCT02470091. An individual-level de-identified dataset containing the variables analyzed in the primary results paper is available upon request. Requests should be sent to: datarequest@childrensoncologygroup.org.