Radium-223 Fails to Improve Skeletal Outcomes When Added to Cabozantinib in Metastatic Renal Cell Carcinoma

Why This Study Matters

Bone metastases occur in roughly 30% of patients with metastatic renal cell carcinoma and carry substantial skeletal morbidity along with a worse prognosis — historically, skeletal-related event rates in this population ranged from 71% to 85%. Cabozantinib, a multi-target tyrosine kinase inhibitor, already showed a pronounced benefit in mRCC patients with bone involvement in the phase 3 METEOR trial. Separately, radium-223 — an alpha-emitting, bone-seeking radiopharmaceutical — extended survival and delayed skeletal events in metastatic castration-resistant prostate cancer. RADICAL investigators hypothesized that combining the two agents could further reduce skeletal morbidity in mRCC, building on an earlier pilot study that established the safety of radium-223 alongside VEGF-targeted therapy in this setting.

Study Design

RADICAL (Alliance A031801) was a phase 2, open-label, randomized trial with an embedded safety run-in, conducted through the Alliance for Clinical Trials in Oncology. Ninety-eight patients with mRCC and at least one bone metastasis were randomly assigned 1:1 to cabozantinib plus radium-223 (Arm A) or cabozantinib alone (Arm B), stratified by IMDC risk category, use of osteoclast-targeted therapy, prior treatment status, and baseline opioid use. In Arm A, radium-223 was given intravenously at 1.49 μCi/kg on day 1 of each 28-day cycle for up to six cycles, and cabozantinib was started at a reduced dose of 40 mg once daily, escalating to 60 mg from cycle 2 onward if tolerated. In Arm B, cabozantinib was dosed at the standard approved 60 mg once daily. Concurrent osteoclast-targeted therapy (a bisphosphonate or denosumab) was mandated in both arms unless contraindicated.

Patient Population

Eligible patients had mRCC of any histology with at least one bone metastasis not previously irradiated, could be treatment-naïve or previously treated (but not with cabozantinib or radium-223 before), and required a Karnofsky performance status of at least 60%. Patients with an imminent pathologic fracture or spinal cord compression were excluded. Enrollment ran from December 2019 to December 2025. Median age was 63 years; 73.5% of patients were male, and 82.7% had clear cell histology. IMDC risk was favorable in 18.4%, intermediate in 67.3%, and poor in 14.3% of patients. Most patients (87.8%) had received prior systemic therapy, and 62.2% had a prior symptomatic skeletal event. A prespecified interim futility analysis, conducted after 90 patients were enrolled, crossed the futility boundary (stratified hazard ratio 1.24, 95% CI 0.62–2.48), and the Alliance Data and Safety Monitoring Board recommended closing the study to accrual at 98 patients.

Primary Endpoint Results

The primary endpoint was symptomatic skeletal event-free survival (SSE-FS) — time from randomization to the earliest of a symptomatic skeletal event or death from any cause. Adding radium-223 to cabozantinib did not improve SSE-FS.

Subgroup Analyses

Prespecified subgroup analyses of SSE-FS — by osteoclast-targeted therapy use, IMDC risk group, treatment status, and baseline opioid use — did not identify a subgroup with a consistent benefit favoring the combination, though the authors note confidence intervals were wide given the limited sample size within each subgroup. Per-subgroup hazard ratios, confidence intervals, and p-values beyond what is described here were not reported in the source article; we do not estimate figures the source does not state.

Safety Profile

Ninety-five patients were evaluable for adverse events (46 in the combination arm, 49 in the cabozantinib-alone arm). Any-grade adverse events occurred in 97.8% versus 98.0% of patients, and grade 3 or higher adverse events occurred in 69.6% versus 75.5%, respectively. Hematologic toxicity — anemia, neutropenia, thrombocytopenia, and lymphopenia — was more frequent with the combination, consistent with radium-223’s mechanism as a bone-marrow-adjacent radiopharmaceutical. Six on-treatment deaths occurred overall (four in the combination arm, two in the cabozantinib-alone arm); no meaningful excess fracture signal was observed in the combination arm.

Interpretation and Broader Context

The authors describe RADICAL as the first randomized trial specifically designed to test a bone-targeted therapeutic strategy in mRCC with bone metastases, and among the largest prospective studies of a therapeutic radiopharmaceutical in this disease. Despite the negative primary result, several contextual points are relevant for practicing oncologists.

Although the addition of radium-223 to cabozantinib did not improve SSE-FS, the combination demonstrated a manageable safety profile and the results of this trial provide a valuable foundation for the design of future bone-targeted therapeutic strategies in mRCC.
— Discussion, McKay et al., Journal of Clinical Oncology (2026)

The observed skeletal-event rate was markedly lower than the 71%–85% seen in historical cohorts, which the authors attribute to contemporary supportive-care practices — mandated osteoclast-targeted therapy and earlier palliative radiation to high-risk bone lesions — rather than to radium-223 itself. This lower background event rate may have limited the trial’s statistical power to detect an incremental benefit. The authors also flag a notable discordance between the stratified (hazard ratio 1.46) and unstratified (hazard ratio 0.97) analyses of the primary endpoint, cautioning that the stratified model, with four stratification factors applied to a modest sample size, may have been overparameterized — a methodological point relevant to how the primary result should be weighted. The trial also enrolled over an unusually long six-year accrual period (2019–2025), spanning the shift toward frontline immune-checkpoint-inhibitor combinations as standard mRCC therapy — a limitation the authors note may have introduced selection bias over time in who was referred to and enrolled on the study.

Limitations

This trial closed early after an interim futility analysis at 90 of the originally intended enrollment, so its primary result rests on a smaller, less mature dataset than initially planned, with correspondingly wide confidence intervals around every endpoint. The discordance between the stratified and unstratified hazard ratios for the primary endpoint means the point estimate is sensitive to model specification, and the authors caution against over-interpreting either figure in isolation. The six-year accrual window overlapped a period when frontline treatment for mRCC shifted substantially toward immune-checkpoint-inhibitor combinations, which the authors say may have introduced selection bias in the population enrolled in later years relative to earlier ones. Overall survival and time-to-event secondary endpoints also carry limited statistical power given the reduced sample size, and none of the reported comparisons should be read as demonstrating a survival benefit or harm in either direction.