World-wide oligometastatic prostate cancer meta-analysis leveraging individual patient data from randomized trials (WOLVERINE): a systematic review and meta-analysis from the X-MET collaboration

Abstract

Background:

Increasing evidence suggests a spectrum of metastatic disease, on one end of which exists oligometastatic disease. Metastasis-directed therapy (MDT) is increasingly used to treat oligometastatic disease despite the lack of level 1 evidence. We undertook a systematic review and individual patient meta-analysis of all randomized trials to evaluate the effectiveness of MDT for oligometastatic prostate cancer.

Methods:

We systematically searched Embase, PubMed, CENTRAL, MEDLINE, and ClinicalTrials.gov to identify randomized trials of MDT enrolling patients with oligometastatic prostate cancer. Initial database search occurred from inception to Nov 3, 2023 and was updated May 4, 2025. Identified studies were evaluated utilizing the Cochrane Collaboration’s risk-of-bias (RoB) assessment (Ver. 2.0). Individual patient data were provided by investigators and endpoints were harmonized across trials. Co-primary endpoints were progression-free survival (PFS), defined as biochemical or radiographic progression or death; and overall survival (OS). Secondary endpoints were radiographic PFS (rPFS) and castration-resistance–free survival (CRFS). To compare MDT plus standard of care (SOC) vs SOC we conducted a trial-level analysis employing a random effects model and a patient-level analysis utilizing patient-level data stratified by trial (Prospero: CRD42023479078).

Findings:

Of 2,975 studies identified for screening, 7 phase 2 studies randomizing 574 men with oligometastatic prostate cancer were included. Six trials randomizing patients to MDT+SOC (n=248) versus SOC (n=224) were used to evaluate the effectiveness of MDT. In both trial- and patient-level analyses, MDT was associated with improved PFS (trial-level hazard ratio [HR]=0.44, 95% CI 0.35–0.56, p<0.001; patient-level HR=0.45, 95% CI 0.35–0.57, p<0.001), rPFS (trial-level HR=0.60, 95% CI 0.42–0.85, p=0.004; patient-level HR=0.59, 95% CI 0.46–0.76, p<0.001), and CRFS (trial-level HR=0.58, 95% CI 0.37–0.92, p=0.019; patient-level HR=0.58, 95% CI 0.37–0.91, p=0.017). The association between MDT and OS exhibited a HR of 0.63 (95% CI 0.39–1.00; p=0.051) in trial-level and 0.64 (95% CI 0.40–1.01; p=0.057) in patient-level analyses.

Interpretation:

WOLVERINE demonstrated a benefit with MDT for oligometastatic prostate cancer in PFS, rPFS, and CRFS. There is some potential for benefit in OS.

Funding:

Philanthropic gift and National Cancer Institute.

Introduction

Oligometastatic disease refers to the presence of limited (up to 5) metastases.¹ The application of definitive local treatment to control these metastases, known as metastasis-directed therapy (MDT), has emerged as a promising option for patients with oligometastatic disease from several types of solid tumors.^1-5 Most of the randomized prospective data supporting MDT for oligometastatic disease is on prostate cancer, yet to date no phase 3 trials have been reported for this or any other histologies.^6-9

Despite the lack of level 1 evidence, MDT has become standard clinical practice for oligometastatic prostate cancer, primarily based on findings from randomized phase 2 trials.⁶ However, these phase 2 studies vary substantially in design, use of hormone therapy (HT), imaging used for initial staging (prostate-specific membrane antigen [PSMA] positron emission tomography [PET] vs. conventional), and clinical stages included (M1a-b vs. M1c and metachronous vs. synchronous). Completed phase 2 studies have demonstrated promising benefit in various clinical endpoints (e.g., prostate-specific antigen [PSA] response, progression-free survival [PFS]), but their signal-finding design rendered them underpowered for investigating longer-term endpoints.^6-10 Thus, few studies have shown benefit in longer-term endpoints that have broader clinical implications, such as radiographic progression-free survival (rPFS) and castration resistance-free survival (CRFS).^6-10

We hypothesized that MDT improves not only PFS but also the longer-term endpoints rPFS, CRFS, and overall survival (OS) in oligometastatic prostate cancer. We thus undertook the WOLVERINE systematic analysis to comprehensively identify all randomized trials investigating oligometastatic prostate cancer, including trials enrolling patients with across several histologies. Individual patient data were then leveraged from the 'XRT for Metastatic Disease Individual patient Meta Analysis' (X-MET) collaboration for a meta-analysis investigating the benefit of MDT + standard of care (SOC) versus SOC. We further explored whether the effect of MDT differs among patient and treatment subgroups and investigated the prognosis of subgroups after receiving MDT.

Methods

Trial Search and Selection

We conducted a literature review according to PRISMA guidelines to identify randomized clinical trials that enrolled patients with oligometastatic prostate cancer, including trials enrolling only prostate cancer as well as those enrolling other tumor histologies. The literature search was conducted from database inception to Nov 3, 2023 and updated on May 4, 2025. Databases searched were ClinicalTrials.gov, MEDLINE, CENTRAL, PubMed, and Embase from dates of their creation until May 4, 2025. An additional hand search of ClincialTrials.gov and relevant radiation oncology, urology, and medical oncology conferences was conducted to identify additional studies. Search terms and strategy used are presented in Supplementary Table S1. The systematic review was performed in COVIDENCE with two investigators (CT and ADS) performing independent screens. Disagreements were resolved via adjudication with a third investigator (PO). Inclusion criteria for study selection were: 1) prospective trial, 2) randomized design, 3) enrolled prostate cancer patients, 4) enrolled patients with oligometastatic (up to 5 metastasis) disease, 5) recorded sufficient data to determine the primary outcomes for this meta-analysis (PFS and OS), and 6) published and thus able to provide individual patient data. All studies meeting inclusion criteria were evaluated independently by two authors utilizing the Cochrane Collaboration’s risk-of-bias (RoB) assessment 2 (Ver. 2.0) for randomized controlled trials.¹¹

Individual patient data from selected studies were provided by trial investigators after institutional approval. Patients provided informed consent to participate in their respective trials. The meta-analysis plan was approved by the institutional review board of The University of Texas MD Anderson Cancer Center and was registered with PROSPERO (CRD42023479078). Data from X-MET is not publicly available.

Data Analysis

We used a structured data collection form to collect individual patient data from eligible studies, including PSA at enrollment, patient outcomes, age, baseline staging imaging used, number of metastases, baseline castration-resistant prostate cancer (CRPC) versus castration-sensitive prostate cancer (CSPC), and organ systems involved. We confirmed the quality and completeness of the data for all patients and generated summative baseline and outcomes outputs for each trial to compare with the original publications.

The prespecified co-primary endpoints were PFS and OS; prespecified secondary endpoints were rPFS and CRFS. All primary and secondary endpoints were harmonized across trials. A PFS event was a composite of the first occurrence of biochemical failure (per individual trial definitions), radiographic (RECIST 1.1) progression, or death from any cause, and rPFS event was a composite of the first occurrence of radiographic progression or death. CRFS analysis was limited to patients with baseline CSPC and was a composite of time until castration resistance (i.e., progression with serum testosterone <50 ng/dL) or death. OS was the occurrence of death. All time-to-event analyses were initiated from the date of trial enrollment.

For primary and secondary efficacy analyses, we limited our analysis to trials randomizing patients between MDT + standard of care (SOC) versus SOC, where SOC could be either HT or observation. As prespecified, two parallel analyses were conducted. In the trial-level analysis, a two-step meta-analysis was conducted in which individual trial outcomes were generated and the aggregated effect of MDT+SOC versus SOC was analyzed via a random effects model. In the patient-level analysis, Cox proportional hazards regression models analyzed individual patient data stratified by trial. To estimate the absolute benefit of MDT+SOC versus SOC, 3-year restricted mean survival time (RMST) benefit was estimated for both arms. A 3-year truncation time was chosen as it was the maximum whole number of years that allowed inclusion of all arms of all trials across all endpoints. Publication bias was visually assessed by using funnel plots and the degree of asymmetry evaluated with Egger’s test. Inter-trial heterogeneity and sampling error were evaluated by using Cochran’s Q, I², and τ² metrics.

Two sets of exploratory analyses were conducted. To assess whether the effects of MDT varied between patient and treatment subgroups (such as castration-resistance status or the presence of an untreated primary tumor), a univariable unstratified patient-level Cox regression was conducted to estimate the effect of MDT+SOC vs SOC within each subgroup for the four harmonized primary and secondary outcomes. This analysis was limited to trials comparing MDT+SOC versus SOC. In this analysis, trial-level stratification produced uninterpretable large confidence intervals given the collinearity of inclusion criteria for individual trials with the subgroups.

A second exploratory analysis was conducted to assess the prognosis of patient subgroups after receiving MDT using patient-level data. The prognosis among subgroups was compared with Cox regression analysis. This analysis was stratified by trial and repeated for all harmonized primary and secondary endpoints. The prognosis analysis included all trials regardless of randomization but was limited to patients that received MDT on trial.

For Cox regression analyses, the Schoenfeld residual test was used to confirm the proportionality of hazards. For the primary and secondary MDT efficacy analyses, comparisons were in the intention-to-treat population. All statistical analysis was performed using R version 4.4.1 software, using 2-sided statistical testing at the 0.05 significance level.

Role of the funding source

Support was provided by a philanthropic gift to CT. Institutional support was provided by the Cancer Center Support Grant to MD Anderson Cancer Center (grant P30 CA016672). The funders had no role in data gathering, interpretation, or manuscript writing.

Results

The systematic review identified 2,973 publications, with an additional 3 studies identified by manual searching. From these 2,976 studies, 931 were identified by COVIDENCE to be duplicates (Figure 1). A further 1,675 studies were excluded based on title and abstract. This led to 369 publications investigating oligometastatic solid tumors, of which 177 were excluded for not enrolling patients with prostate cancer. The remaining 192 studies underwent full text assessment from which 134 studies were excluded because they described the same clinical trial (n=111), used a different intervention (n=4), used a different study design (n=12), evaluated a different population (n=6), or were withdrawn before activation (n=1). Of the remaining 59 studies describing randomized trials enrolling men with prostate cancer, 46 were not included in the meta-analysis because they described trials in which only the protocol was provided, without a completed primary analysis. Another 2 studies were excluded because they examined several tumor histologies and could not provide prostate-specific demographic data or data on prostate-specific endpoints including biochemical progression. Five studies presented primary analyses of protocols in abstract form only. The study team contacted the Principal Investigators of these five studies to inquire whether individual patient data would be available for this meta-analysis; however, only one (EXTEND-continuous HT basket) was available at that time. Individual patient data could not be obtained from the remaining 4 abstracts and thus were excluded. Details of the 52 randomized studies that were not included in the meta-analysis are provided in Supplementary Table S2 (prostate-specific trials) and Supplementary Table S3 (multi-histology trials).

Figure 1:

PRISMA flow diagram detailing systemic review and trial selection for meta-analysis

Thus the meta-analysis included 7 studies, 6 of which randomized patients with oligometastatic prostate cancer to MDT +SOC versus SOC and were included in the primary and secondary efficacy analyses.^3,7-9,12-14 The EXTEND intermittent-HT basket and EXTEND continuous-HT basket were conducted under the EXTEND basket trial protocol (NCT03599765) but were independently powered, randomized, and reported and thus were considered separate studies in this meta-analysis. The seventh study (RADIOSA) randomized patients to MDT + observation versus MDT + intermittent HT and was included in the exploratory prognostic analysis.¹³ Overall, these 7 studies included 574 patients (Table 1). Most patients in these 7 studies had castration-sensitive disease (n=375, 65%), except for 3 studies that enrolled either a subset of CRPC (EXTEND intermittent- and continuous-HT baskets) or entirely CRPC (ARTO) (Supplementary Table S4). The prostate primary had received previous definitive local therapy before enrollment in 491 patients (85.5%); prior primary-tumor treatment was required in all studies except the EXTEND intermittent- and continuous-HT baskets and ARTO.

Table 1:

Trials included in the meta-analysis

	Trial name (NCT No.#)	Primary inclusion criteria	Enrollment period	SOC HT / observation	LN counting method	Definition of biochemical progression	Primary endpoint
	Prostate-specific trials (Randomized MDT + SOC vs. SOC)
6 Trials included in primary and secondary MDT +SOC vs. SOC analysis	STOMP (NCT01558427)	Prior treated primary, CSPC, ≤ 3 metastases, and staged with choline-PET CT (n=62)	2012–2015	Observation	Individual LNs were counted separately	≥ 2 ng/mL and ≥ 25% above nadir if PSA was ≥ 2ng/mL at baseline ≥ 25% above nadir if PSA was < 2 ng/mL at baseline	ADT-free survival
	ORIOLE (NCT02680587)	Prior treated primary, CSPC, ≤ 3 metastases, and staged with conventional imaging (n=54)	2015–2018	Observation	Each LN station was counted separately	PSA increase ≥ 2 ng/mL and ≥ 25% above nadir	Disease progression at 6 months
	EXTEND intermittent HT basket (NCT03599765)	CRPC or CSPC, ≤ 5 metastases, and staged with conventional imaging, fluciclovine PET-CT, or PSMA PET-CT (n=84)	2018–2020	Intermittent ADT ± 2nd-generation ARPI	Each LN station was counted separately	PSA increase ≥ 2 ng/mL and ≥ 25% above nadir	Progression-free survival
	ARTO (NCT03449719)	CRPC, ≤ 3 bone or nodal metastases, and staged with conventional imaging, flucicolvine PET-CT, PSMA PET-CT, or choline-PET-CT (n=157)	2019–2022	ADT + abiraterone acetate	Individual LNs were counted separately	PSA increase ≥ 2 ng/mL and ≥ 25% above nadir	Biochemical response rate (PSA decrease ≥ 50% from baseline) at 6 months
	EXTEND continuous HT basket (NCT03599765)	CRPC or CSPC, ≤ 5 metastases, and staged with conventional imaging, flucicolvine PET-CT, or PSMA PET-CT (n=96)	2018–2022	Continuous ADT ± 2nd-generation ARPI	Each LN station was counted separately	PSA increase ≥ 2 ng/mL and ≥ 25% above nadir	Progression-free survival
	Multi-histology trials (Randomized MDT + SOC vs. SOC)
	SABR-COMET (NCT01446744)	Prior treated primary, CSPC, ≤ 5 metastases, and staged with conventional imaging (n=16, prostate only)	2012–2016	Observation, continuous, and intermittent ADT	Individual LNs were counted separately	PSA increase ≥ 2ng/mL and ≥ 25% above nadir	Overall survival
	Prostate-specific trials (Randomized MDT + observation vs. MDT + ADT)
	RADIOSA (NCT03940235)	Prior treated primary, CSPC, ≤ 3 metastases limited to lymph node and bone, and staged with choline PET-CT, PSMA PET-CT, or whole-body MRI (N=102)	2019–2023	Observation vs. 6 months ADT	Individual LNs were counted separately	PSA increase ≥ 20% above baseline in MDT + observation arm PSA increase to ≥ 1 ng/mL in MDT + ADT arm	Clinical progression-free survival

Conventional imaging consisted of CT body imaging, MRI body imaging, and/or bone scan.

Abbreviations: LN, lymph node; CSPC, castration-sensitive prostate cancer; CRPC, castrate-resistant prostate cancer; PET-CT, positron emission tomography-computed tomography; SOC, standard of care; ADT, androgen deprivation therapy; PSMA, prostate specific membrane antigen; PSA, prostate-specific antigen; ARPI, androgen receptor pathway inhibitor.

From these 7 randomized studies, 6 trials randomized patients to MDT + SOC versus SOC and were included in the primary and secondary MDT efficacy analyses as well as the exploratory MDT within-subgroup analysis. These 6 studies included 472 patients with a median follow-up time of 40.7 months (IQR: 25.6–53.7). COMET-SABR was the only multi-histology study to be included, from which 16 men with prostate cancer were included in the meta-analysis. COMET-SABR investigators provided information on biochemical progression, which allowed harmonization for the primary endpoint of PFS, but information on the secondary endpoint of castration resistance was not available. Patient demographics, disease, and treatment characteristics of the 6 studies randomizing patients to MDT +SOC versus SOC stratified by randomization are presented in Table 2. Overall, patients were well-balanced between arms with both arms exhibiting a median PSA of 1.9. The patients in the SOC arm were slightly older (median [range] 71 [50-90] vs. 68 [48-88] years) and more received a second-generation androgen receptor pathway inhibitor (ARPI) (59.8% [n=134] vs. 50.4%[n=125]).

Table 2:

Baseline patient characteristics in the six trials that randomized patients between MDT+SOC versus SOC

Characteristics	SOC (n=224)	MDT+SOC (n=248)
Age, years, median (min-max)	71 (50, 90)	68 (48, 88)
PSA at enrollment, ng/mL, median (min-max)	1.9 (0, 127.8)	1.9 (0, 117)
Time from diagnosis or definitive prior primary treatment to enrollment, months, median (min-max)*	44 (0,232)	48 (0,313)
Baseline castration sensitivity, n (%)
Castration resistant	120 (53.57%)	153 (61.69%)
Castration sensitive	104 (46.43%)	95 (38.31%)
ADT, n (%)
No	50 (22.32%)	69 (27.82%)
Yes	174 (77.68%)	177 (71.37%)
Unknown	0 (0.00%)	2 (0.81%)
2nd generation ARPI, n (%)
No	90 (40.18%)	123 (49.6%)
Yes	134 (59.82%)	125 (50.4%)
Primary previously treated, n (%)
No	37 (16.52%)	42 (16.94%)
Yes	185 (82.59%)	204 (82.26%)
Surgery	132 (71.35%)	162 (79.41%)
Radiation	50 (27.03%)	42 (20.59%)
Interventional radiology ablation	3 (1.62%)	0
Unknown	2 (0.89%)	2 (0.81%)
Baseline imaging, n (%)
PSMA PET-CT	28 (12.50%)	33 (13.31%)
Fluciclovine or choline PET-CT	117 (52.23%)	105 (42.34%)
Conventional	79 (35.27%)	110 (44.35%)
Number of metastatic lesions, n (%)
1	79 (35.27%)	100 (40.32%)
2	66 (29.46%)	73 (29.44%)
3	62 (27.68%)	58 (23.39%)
4	12 (5.36%)	9 (3.63%)
5	5 (2.23%)	8 (3.23%)
Lymph metastasis, n (%)
No	83 (37.05%)	105 (42.34%)
Yes	141 (62.95%)	143 (57.66%)
Bone metastasis, n (%)
No	104 (46.43%)	104 (41.94%)
Yes	120 (53.57%)	144 (58.06%)
Visceral metastasis, n (%)
No	219 (97.77%)	244 (98.39%)
Yes	5 (2.23%)	4 (1.61%)

Abbreviations: MDT, metastasis-directed therapy; SOC, standard of care; PSA, prostate-specific antigen; ADT, androgen deprivation therapy; ARPI, androgen receptor pathway inhibitor; PSMA, prostate specific membrane antigen; PET-CT, positron emission tomography-computed tomography.

^*Time from diagnosis to enrollment was available for COMET-SABR, ARTO, ORIOLE, and STOMP while time from primary treatment was available from EXTEND intermittent hormone therapy and EXTEND continuous hormone therapy baskets.

Trial- and patient-level analyses evaluating the efficacy of MDT were conducted from the 6 trials randomizing patients to MDT+SOC versus SOC: EXTEND intermittent-HT basket, EXTEND continuous-HT basket, STOMP, ORIOLE, ARTO, and COMET-SABR trials (prostate subgroup). In the trial-level analysis, MDT was associated with improved PFS in all trials individually and across all trials in aggregate (hazard ratio [HR]=0.44, 95% confidence interval [CI] 0.35–0.56, p<0.001) (Figure 2a). For rPFS, MDT was also associated with improvement in two trials individually (ARTO and SABR-COMET) and across all trials in aggregate (HR=0.60, 95% CI 0.42–0.85, p=0.004) (Figure 2b). The association between MDT and OS exhibited HR<1 for each individual trial. Across trials the aggregated association exhibited a HR=0.63 (95% CI 0.39–1.00, p=0.051) (Figure 2c). Among patients with CSPC (n=257), MDT was associated with CRFS in only the EXTEND-intermittent HT basket and was associated with CRFS across all trials in aggregate (HR=0.58, 95% CI 0.37–0.92, p=0.019) (Figure 2d). A sensitivity analysis showing 1-year RMST is presented in Supplementary Table S5 Apart from rPFS which revealed moderate inter-trial heterogeneity (I²=50%; Figure 2b), metrics of inter-trial heterogeneity and sampling error (Figure 2) as well as funnel plots evaluating publication bias (Supplementary Figure S1) did not reveal substantial effects. Egger’s test should be interpreted with caution given the small number of studies included. Summary or risk-of-bias assessment is present in Supplementary Figure S2. There were “some concerns” noted for most study stemming from the non-blinded randomization across studies. A sensitivity analysis excluding trials in which SOC was observation in all or part of the patient population: STOMP, ORIOLE, and COMET-SABR identified results similar to the primary analysis. In this sensitivity analysis MDT was associated with improvements in PFS (HR=0.46, 95% CI 0.34-0.63, P<0.001), rPFS (HR=0.56, 95% CI 0.36-0.89, P=0.014), and CRFS (HR=0.46, 95% CI 0.23-0.91, P=0.025). For OS, HR=0.65 (95 CI 0.37-1.13, P=0.13) (Supplementary Figure S3).

Figure 2:

Trial-level (two-step) meta-analyses investigating association of metastasis-directed therapy (MDT) with standard of care (SOC) versus SOC alone with (A) progression free survival (PFS), (B) radiographic progression free survival (rPFS) and (C) overall survival. The association between MDT and (D) castration resistance-free survival (CRFS) was evaluated in patients with castration-sensitive metastatic disease only. 3-year absolute benefit and restricted mean survival time (RMST) benefit is shown for each endpoint. Instances where either of these metrics were not estimable (NE) are indicated. Abbreviations: cADT, continuous androgen deprivation therapy; iADT, intermittent androgen deprivation therapy.

Results from the trial-level analyses were replicated in the patient-level analyses, which revealed similar associations of MDT with PFS (HR=0.45, 95% CI 0.35–0.57, p<0.001, Figure 3a), rPFS (HR=0.59, 95% CI 0.46–0.76, p<0.001), OS (HR=0.64, 95% CI 0.40–1.01, p=0.057, Figure 3c), and CRFS (HR=0.58, 95% CI 0.37–0.91, p=0.017, Figure 3d). Similar results were observed in a sensitivity analysis excluding patients on trials in which SOC was observation (Supplementary Figure S4).

Figure 3:

Kaplan Meier curves showing patient-level analyses investigating association of metastasis directed therapy (MDT) with standard of care (SOC) (blue) versus SOC (red) with (A) progression-free survival (PFS), (B) radiographic progression-free survival (rPFS), and (C) overall survival (OS). The association between MDT and (D) castration resistance-free survival (CRFS) was evaluated in patients with castration-sensitive metastatic disease only

An exploratory analysis of MDT within subgroups was conducted in the 6 trials randomizing patients to MDT+SOC versus SOC. Given the importance of castration sensitivity, the analysis was conducted in the full cohort and separately in the CSPC and CRPC subgroups (Supplementary Figures S5-8). For PFS, MDT was associated with improved PFS across most patient subgroups, and no significant interactions were observed (Supplementary Figure S5). The benefit of MDT was similar among CSPC prostate cancer patients irrespective of SOC: observation (HR=0.42, 95% CI 0.28-0.63), ADT (HR=0.40, 95% CI 0.22-0.70), and ADT with a 2^nd generation ARPI (HR=0.60, 95% CI 0.30-1.20). For rPFS and CRFS, MDT was associated with improvements across most subgroups (Supplementary Figure S6 and S8); however, significant interactions were noted between MDT and baseline PSA (median PSA on HT: 0.12 ng/ml and off HT: 5 ng/ml) within the CSPC cohort for both endpoints (all p<0.05). MDT was associated with improved rPFS and CRFS in patients with lower-than-median baseline PSA but not in patients with higher-than-median PSA. Given the relative paucity of OS events, the 95% CI intervals for the association between MDT and OS crossed 1 for most patient subgroups. No significant interaction terms were observed for OS (Supplementary Figure S7).

The exploratory prognosis analysis was conducted in all seven trials but was limited to patients who received MDT (n=350). The prognostic association of baseline clinical and treatment-related factors were analyzed for the overall, CPSC, and CRPC cohorts (Supplementary Tables S6-S9). Patients with bone metastases only exhibited worse PFS than patients with lymph node (LN) metastases only in the entire cohort (HR=1.45, 95% CI 1.05-2.00) and CSPC subgroup (HR=1.61, 95% CI 1.13-2.17). Within the CSPC subgroup, HT use and lower baseline PSA (median PSA on HT: 0.6 ng/ml and off HT: 2.8 ng/ml) were consistently associated with improved prognosis for several endpoints. Use of androgen deprivation therapy (ADT) was associated with improved PFS (HR=0.45, 95% CI 0.29–0.72) and rPFS (HR=0.43, 95% CI 0.25–0.76) over observation as was the combined ADT and second-generation androgen receptor pathway inhibitor (ARPI) (PFS: HR=0.38, 95% CI 0.16–0.88; and rPFS: HR=0.38, 95% CI 0.15–0.97) (Supplementary Table S6 and S7). Higher-than-median baseline PSA was associated with worse PFS, rPFS, and CRFS within the CSPC subgroup and with PFS and rPFS in the CRPC subgroups (Supplementary Table S6, S7, and S9).

The frequency of worst adverse events experienced by a patient was similar among patients receiving MDT+SOC versus SOC in the 6 trials that randomized between these treatments (Supplementary Table S10). No grade 5 toxicities were observed in either arm and grade ≥2 adverse events were observed in 34 of 224 (15.18%) of patients receiving SOC and 42 of 248 (16.94%) receiving MDT+SOC.

Discussion

The WOLVERINE individual patient meta-analysis demonstrated a robust association between MDT and improved PFS both within individual trials and across all trials in aggregate. MDT was also associated with improvement in the longer-term endpoints of rPFS and CRPC, which has been rarely demonstrated in prospective randomized trials. MDT was not associated with OS; however, the estimated overall effect size of 0.63 was in favor of MDT+SOC, with p values of 0.051 for trial-level and 0.057 for patient-level analyses. These associations were consistent across two prespecified methods: a two-step trial-level analysis using a mixed effect model and a patient-level analysis using a trial-stratified Cox proportional hazards regression model. These findings are even more impressive given the fact that all trials allowed patient crossover from the SOC arm to receive MDT after progression, thus diluting the ability to identify signals in later endpoints. Finally, exploratory analyses, which did not adjust for multiple testing and are intended to be hypothesis-generating, identified a significant interaction between MDT and baseline PSA: specifically, patients who had lower baseline PSA exhibited a strongly favorable association between receipt of MDT and longer rPFS and CRFS, whereas patients with higher baseline PSA did not.

Individual randomized trials have demonstrated improvements in PFS but generally have not shown improvement in later endpoints such as rPFS or CRFS. To the best of our knowledge, only two analyses combining individual patient data from separate randomized trials have been conducted, both aggregating data from two trials. Deek et al. combined patients from the ORIOLE and STOMP trials and demonstrated improvements with MDT for PFS but no benefit for rPFS, CRFS, or OS.¹⁰ The EXTEND basket protocol included a prespecified analysis combining the prostate intermittent-HT and continuous-HT baskets and found improvement in PFS in addition to rPFS and CRFS for the combined analysis.¹⁴ A limited number of meta-analyses restricted to trial-level data have been conducted with similar outcomes.^15,16 Miszczyk et al. conducted a trial-level meta-analysis of single-arm and randomized oligometastatic prostate cancer trials.¹⁵ Their calculated 2-year PFS of 42%–46% and OS of 97% were consistent with outcomes from WOLVERINE. In addition to clinical trials, the UK National Health Service (NHS) registry is a large observational registry that enrolled 1422 patients with prostate cancer (n=406) as the largest single histologic subgroup. Although follow-up for that registry is limited (median 13 months), patients with oligometastatic prostate cancer had excellent 2-year OS (94.6%) and promising new-metastasis–free survival (40.0%).¹⁷

Although several prognostic factors have been established for metastatic prostate cancer, it is unclear whether those factors also modify outcomes after MDT in oligometastatic prostate cancer. In our exploratory analysis of MDT within subgroups, a consistent benefit from MDT for PFS was apparent across subgroups. For the later endpoints of rPFS and CRFS, higher baseline PSA modified the effect of MDT, in that patients with higher baseline PSA derived less benefit (interaction p values <0.05). One explanation is that higher PSA may reflect greater microscopic disease burden. This is further supported by our finding that among patients who received MDT, a higher baseline PSA was also an adverse prognostic marker for PFS, rPFS and CRFS.

In addition to baseline PSA, several notable subgroups did not exhibit a significant interaction with MDT in the current analysis. Disease and therapy factors underlying these subgroups are active areas of investigation for ongoing randomized trials. These factors include higher number of metastases, use of systemic therapy, baseline CRPC status, and the presence of a previously untreated primary. Regarding number of metastases, several ongoing randomized studies are investigating the benefit of MDT for patients with increasing numbers of metastases. These include SABR-COMET 10 (NCT05717166) and SABR-SYNC (NCT05717166), which are enrolling patients with up to 10 metastases, whereas ARREST II (NCT05508464) is enrolling patients with polymetastatic disease (>10 metastases). Although the current analysis identifies similar benefit for MDT in patients with low (1-2) and higher (3-5) numbers of metastases and metastasis number was not prognostic for outcomes after MDT, the number of patients with 3-5 metastases was low (n=85). More trials will be needed to define the upper limit of metastases in patient populations that derive benefit from MDT.

Whether the benefit of MDT is enhanced by HT is an active area of investigation. Although some ongoing and planned phase III trials (e.g. PSMA-DC [NCT05939414]) are using MDT + observation as the control arm, most incorporate ADT with or without 2^nd-generation ARPI into control arms (e.g., STAMPEDE 2 [NCT06320067], INDICATE [NCT04423211], and OLIGO-PRESTO [NCT04115007]). The question of whether HT enhances the benefit of MDT is the subject of several ongoing trials that are randomly assigning patients with oligometastatic CSPC to MDT ± HT, including ADOPT (NCT04302454), DART (NCT04641078), and PROMETHEAN (NCT05053152). The only reported randomized trial investigating this question is the RADIOSA trial (included in WOLVERINE), which randomized CSPC patients to MDT vs. MDT and 6 months ADT.¹³ RADIOSA found that the addition of ADT improved clinical PFS (similar to rPFS) and PFS; however, later endpoints such as CRPC or OS were not improved in the primary analysis. Although HT use did not modify the effect of MDT+SOC vs. SOC in the current analysis, HT use was associated with improved prognosis for the PFS and rPFS endpoints among patients who received MDT.

Treatment of oligometastatic CRPC remains an active area of investigation, because completed randomized trials generally have enrolled patients with CSPC. Although CRPC status is a clear indication of lack of sensitivity to HT, no studies have demonstrated that CRPC status modifies the benefit of MDT. Two trials enrolling exclusively patients with CRPC have been reported to date: ARTO, included in WOLVERINE, and GROUQ-PCS9, reported as an abstract in GU ASCO 2025, randomized patients with oligometastatic CRPC to ADT + 2^nd-generation ARPI versus MDT + ADT + 2^nd-generation ARPI and demonstrated a benefit from MDT (ARTO: PFS HR=0.33, 95% CI 0.16–0.67; and GROUQ-PCS9: bPFS HR=0.56, 95% CI 0.31–0.99).^12,18 These results are similar to the benefit from MDT observed for the CRPC (PFS HR=0.45, 95% CI 0.30–0.66) and CSPC (PFS HR=0.51, 95% CI 0.38–0.68) subgroups in WOLVERINE, which did not identify an interaction between castration-sensitivity status and MDT. Ongoing studies such as DECREASE (NCT04319783) and PEACE 8 (NCT06276465) are further investigating the benefit of MDT for patients with oligometastatic CRPC.

Another area in which data are limited is the application of MDT for patients with de novo metastatic prostate cancer, as most completed studies to date have excluded patients with untreated primary disease. The benefit of MDT in WOLVERINE seemed to be similar in patients with and without prior primary treatment, and no prognostic association was found between an untreated primary and patient outcomes after MDT. However, the number of patients who received MDT and had a previously untreated primary was relatively low (n=42). The treatment of patients with de novo disease is being actively investigated in several trials, including STAMPEDE 2 (NCT06320067) comparison M, TERPS (NCT05223803), and the multi-histology SABR-SYNC (NCT05717166).

Several limitations of the current study deserve mention. First, not all trials defined PFS and biochemical PFS in the same way. Across trials, the most common definition of PFS was a composite of biochemical and radiographic progression, given that the occurrence of either endpoint usually necessitates a change in systemic therapy management. WOLVERINE also applied a composite PFS definition after careful data harmonization across studies. For transparency, definitions of biochemical progression in individual trials are shown in Table 1. Second, given the longevity expected for men with oligometastatic prostate cancer the number of OS events was low in the current analysis. Third, given the widespread availability of salvage therapies, signals for later endpoints such as OS are difficult to interpret. Fourth, SOC regimes utilized were heterogeneous reflecting the era in which each trial was run. Observation and single agent ADT arguably do not reflect current practice for metastatic prostate cancer. Fifth, initial clinical staging on trial was subject to variability and counting systems for lymph node nodes were heterogeneous. For transparency, lymph node counting strategy in individual trials are shown in Table 1. As PSMA PET represents the most sensitive form of imaging and the most common imaging modality currently used in several countries including the United States, caution should be taken in extrapolating results from patients staged with other imaging modalities. Sixth, additional endpoints such as patient reported quality-of-life and disease specific survival, which were not available here, deserve evaluation. Finally, there were several efficacy endpoints and multiple comparisons were conducted for exploratory subgroup analyses without adjustment for multiple testing. Thus, the results of our exploratory analyses should be considered hypothesis-generating.

To better address the important questions raised by the subgroup analyses, future focused meta-analyses from X-MET will incorporate ongoing trials to explore these subgroups. Additional biomarkers capable of stratifying this patient population should also be developed such as genomic alterations, PSMA-avid tumor volume, and circulating tumor DNA. In addition to these limitations, several strengths deserve mention. Our systematic review was comprehensive and included not only prostate cancer-specific trials, but also multi-histology trials that included patients with oligometastatic prostate cancer. To the best of our knowledge, the WOLVERINE meta-analysis included all available published trials randomizing men with oligometastatic prostate cancer to MDT+SOC versus SOC, regardless of other criteria. Finally, this meta-analysis represents the first individual patient meta-analysis investigating MDT for oligometastatic prostate cancer that includes more than 2 clinical trials.

Given the lack of phase 3 data, MDT should not be considered standard treatment. In context with the current evidence, the WOLVERINE individual patient-level results provide the strongest evidence to date that MDT improves PFS, rPFS, and CRFS, findings that were consistent across individual studies and most patient and treatment subgroups. MDT also exhibited a favorable association with OS, with p values for this association of 0.051 for trial-level analyses and 0.057 for patient-level analyses. Exploratory subgroup analyses identified an interaction between MDT and PSA in which patients with low PSA showed a stronger association between MDT and improved outcomes relative to those with higher PSA. Notable subgroups that did not modify MDT benefit include number of metastases, systemic therapy use, baseline CRPC status, and previously untreated primary, all of which are being explored in ongoing randomized trials and will be evaluated in future analyses from the X-MET collaboration as randomized data becomes available.

Research in Context.

Evidence before this study

Despite the lack of level 1 evidence, metastasis-directed therapy (MDT) is becoming widely used for the treatment of oligometastatic prostate cancer. Available evidence is limited to phase 2 randomized trials that demonstrated benefit from MDT + standard of care (SOC) over SOC for the signal-finding endpoints (e.g., progression-free survival [PFS]) for which they were designed. However, evidence of improvement in later endpoints such as radiographic PFS (rPFS), castration-resistance–free survival (CRFS), and overall survival (OS) is lacking. We conducted a systematic review to identify all randomized trials investigating MDT for oligometastatic prostate cancer patients from Embase, PubMed, CENTRAL, MEDLINE, and ClinicalTrials.gov databases, our search included trials evaluating multiple tumors types. General search terms were: (oligometa* OR oligo-metasta* OR oligorecurrent) AND cancer AND (surgery OR radiation OR ablation) AND (randomized controlled trial OR controlled clinical trial). This review was conducted Nov 3, 2023 and updated May 4, 2025. Seven randomized trials enrolling men with oligometastatic prostate cancer were identified, of which 6 randomized patients to MDT + SOC versus SOC. Two additional trials, RAVENS (which randomized to MDT versus MDT + Radium 223) and PEACE-V STORM (which randomized men with only lymph node oligometastases to MDT versus MDT + elective local therapy to the uninvolved lymph nodal chain) were published after our review.

Added value of this study

Our individual patient meta-analysis included 7 trials randomizing patients with oligometastatic prostate cancer. Six of these studies compared MDT + SOC versus SOC and were included in the primary and secondary analysis examining the efficacy of MDT. This study represents the first individual patient meta-analysis employing more than 2 trials, demonstrating that MDT was significantly associated with benefit in PFS, rPFS, and CRFS. The association between MDT and OS exhibited a notable positive effect without meeting significance criteria. Furthermore, the benefit of MDT was consistent across most studies and in relevant patient and treatment subgroups.

Implications of all the available evidence

To the best of our knowledge, WOLVERINE includes all available clinical trials randomizing men with oligometastatic prostate cancer to MDT + SOC versus SOC during our search window. Our findings thus offer the highest-level evidence to date supporting the use of MDT in men with oligometastatic prostate cancer. Exploratory subgroup analyses offered additional insights into the value of MDT within relevant patient and treatment subgroups.