Enzalutamide (Xtandi) plus radium-223 dichloride (Xofigo) demonstrated a delayed but substantial survival benefit in patients with metastatic castration-resistant prostate cancer (mCRPC), with a statistically significant improvement in restricted mean survival time (RMST) at 72 months (P = .01) and an 80% reduction in the risk of death among long-term survivors, according to a reconstructed individual patient data (rIPD) analysis of the phase 3 EORTC 1333/PEACE-3 trial (NCT02194842).1
Results during the 2026 American Urological Association Annual Meeting revealed that the survival advantage for the combination regimen was most pronounced beyond 60 months of follow-up. Piecewise Cox regression for the 60 to 72 month interval demonstrated an 80% reduction in the risk of death for long-term survivors (HR, 0.20; 95% CI, 0.05-0.77; P = .02). Landmark analysis from 60 months confirmed this finding with the same HR (95% CI, 0.05-0.77; P = .02). Furthermore, RMST analysis at 72 months showed a significant benefit of 6.25 months (95% CI, 1.56-10.95; P = .01) in favor of the combination arm, with an RMST of 43.23 months vs 36.98 months for enzalutamide alone.
Despite this long-term benefit, piecewise Cox regression revealed no survival advantage for the combination (HR, 1.18; 95% CI, 0.74–1.84; P = .48) during the initial 0 to 18 month active treatment period for radium-223. Following completion of active radium-223 treatment, the survival curves began to diverge in favor of the combination arm, reflecting an accumulating biological impact. Visualization of the time-varying HR demonstrated this continuous shift from elevated early risk (HR > 1) during active treatment, through a transition period of declining HR, to deep and statistically significant benefit in the late phase (HR < 0.5).
"[These data indicate that] a patient needs sufficient life expectancy to benefit from the delayed effect [of enzalutamide plus radium-223], because time-dependent analysis reveals that [the regimen] produces the delayed but meaningful survival benefit in mCRPC." - Soichiro Yoshida, MD, the Department of Urology, Tokyo Medical and Dental University in Japan.
Topline Takeaways
- The combination of enzalutamide plus radium-223 produced no survival advantage during the active treatment phase but conferred a statistically significant 6.25-month survival advantage at 72 months.
- For patients who survived beyond 60 months, the combination was associated with an 80% reduction in the risk of death compared with enzalutamide alone.
- Non-proportional hazards analysis identified a significant violation of the proportional hazards assumption, underscoring that standard Cox modeling may substantially underestimate the long-term benefit of combination therapy in this setting.
What is the clinical context for this analysis?
The PEACE-3 trial previously demonstrated that enzalutamide plus radium-223 improved radiographic progression-free survival over enzalutamide alone and suggested an overall survival (OS) benefit, with a 31% reduction in the risk of death (HR, 0.69) and median OS of 42.3 months (95% CI, 36.8-49.1) in the combination arm vs 35.0 months (95% CI, 28.8-38.9) in the enzalutamide-alone arm (HR, 0.69; 95% CI, 0.52–0.90; P = .0031).2
However, the original trial analysis also revealed an unusual early survival curve crossing, with six excess deaths at 12 months in the combination arm, raising questions about whether the proportional hazards assumption underlying standard Cox regression was appropriate and whether the true magnitude and timing of benefit were being accurately captured, Chen and colleagues explained.1 To address these questions, investigators conducted further analysis using rIPD and comprehensive non-proportional hazards (NPH) methods to characterize the time-dependent nature of the treatment effect.
What was the analytical approach used in this analysis?
Investigators digitized the OS Kaplan-Meier curve from the published PEACE-3 trial data and performed IPD reconstruction followed by quantitative consistency validation to confirm that reconstructed population characteristics remained consistent with those of the original trial (consistency test: z = 0.04; P = .97). The validated rIPD then served as the source for a series of time-dependent analyses, including RMST analyses across time points from 18 to 73 months, piecewise Cox regression to characterize the HR within prespecified intervals, and landmark analyses from 18, 36, and 48 months. Formal evaluation of the proportional hazards assumption was performed using the Schoenfeld residual test.
Proportional hazards assumption testing revealed a significant violation (χ² = 5.01; P = .03), confirming that the treatment effect was not constant over time and that time-dependent, non-proportional hazards models were necessary for subsequent analyses. Based on this finding, the observation period was divided into 3 clinically meaningful phases: the active treatment phase (0-18 months), the transition period (18-60 months), and the extended follow-up phase (60-72 months).
What happened during the active treatment phase?
RMST analysis at 18 months confirmed the absence of benefit, with an RMST difference of −0.36 months (P = .20) favoring enzalutamide alone numerically. The RMST at 18 months was 16.73 months in the enzalutamide plus radium-223 arm vs 17.09 months in the enzalutamide-alone arm.
The early mortality imbalance in the combination arm was attributed to fractures and cardiovascular toxicities. Notably, mandatory bone-protecting agent use was instituted mid-trial, which the investigators noted significantly reduced fracture risk and may have contributed to the long-term survival benefit observed in subsequent phases.
How did this survival benefit evolve during the transition period?
During the 18 to 60 month transition period, piecewise Cox regression showed a directional trend toward benefit that strengthened progressively over time, though individual interval estimates did not reach statistical significance, as shown below:
- 18 to 24 months: HR, 0.79 (95% CI, 0.41-1.51; P = .48)
- 24 to 36 months: HR, 0.69 (95% CI, 0.38-1.27; P = .24)
- 36 to 48 months: HR, 0.70 (95% CI, 0.37-1.23; P = .27)
- 48 to 60 months: HR, 0.55 (95% CI, 0.23-1.34; P = .19)
Landmark analyses from key time points similarly reflected this progressive deepening of benefit, with HRs of 0.58 from 18 months, 0.54 from 36 months, and 0.45 from 48 months. RMST analysis corroborated the trajectory: the RMST difference at 60 months reached 4.34 months (95% CI, 0.49-8.19; P = .03) in favor of the combination, with an RMST of 39.80 months in the enzalutamide plus radium-223 arm vs 35.46 months in the enzalutamide-alone arm.
What are the proposed mechanisms underlying the delayed benefit?
Chen and his fellow investigators offered several mechanistic explanations for the delayed onset and progressive deepening of the survival advantage. Radium-223 is not a rapid cytoreductive therapy; rather, it delivers targeted alpha-particle radiation to bone metastases through its calcium-mimetic properties, and its biological impact accumulates and becomes clinically visible only months after treatment completion. Additionally, dual targeting of the androgen receptor pathway and the bone microenvironment may disrupt multiple resistance mechanisms over time.
Disclosures: Yoshida did not report any disclosures.
References
- Chen W, Yoshida S, Yajima S, et al. Delayed survival benefit of enzalutamide plus radium-223 in metastatic castration-resistant prostate cancer: a reconstructed individual patient data analysis. Presented at: 2026 American Urological Association Annual Meeting; May 15-18, 2026; Washington, DC. Abstract 26-7131.
- Tombal B, Saad F, Thiery-Vuillemin A, et al. Enzalutamide with or without radium-223 in metastatic castration-resistant prostate cancer: final analysis of the EORTC 1333/PEACE-3 randomized phase III trial. Ann Oncol. 2025;36(9):1058–1067. doi:10.1016/j.annonc.2025.05.010
