Approximately one-third of patients who receive primary localized treatment for prostate cancer experience biochemical recurrence within 10 years (1, 2). This recurrence is characterized by a rise in prostate-specific antigen (PSA) levels and is associated with a higher risk of worse clinical outcomes (2). To determine the best treatment approach for patients with biochemical recurrence, risk classification systems have been developed, as described in the American Society of Clinical Oncology Clinical Practice Guidelines (2021) (3) and prostate cancer guidelines from the European Association of Urology-European Association of Nuclear Medicine-European Society for Radiotherapy and Oncology-European Society of Urogenital Radiology-International Society of Geriatric Oncology (EAU-EANM-ESTRO-ESUR-SIOG) (4). According to these guidelines, patients are classified as having high-risk biochemical recurrence after radical prostatectomy if they have a PSA doubling time of <10–12 months or a Gleason score of 8–10, whereas high-risk biochemical recurrence after radiotherapy (RT) requires an interval from primary therapy to biochemical failure of <18 months or an initial biopsy Gleason score of 8–10 (3, 4). By contrast, guidelines from the American Urological Association/American Society for Radiation Oncology/Society of Urologic Oncology (AUA/ASTRO/SUO) (2024) define patients at high risk of developing metastasis with biochemical recurrence after local treatment as those with a PSA doubling time of <12 months, a shorter time to biochemical recurrence, or a higher grade or stage of disease (5, 6).

Before the EMBARK trial, specific treatment guidelines for patients with high-risk biochemically recurrent (BCR) non-metastatic hormone-sensitive prostate cancer (nmHSPC, also known as non-metastatic castration-sensitive prostate cancer [nmCSPC]) were limited and variable (1, 3, 7). Patients with nmHSPC have no detectable metastases according to conventional imaging, but if left untreated, many will eventually develop metastatic disease that can be seen on conventional imaging (8). While some guidelines recommend observation, others recommend continuous androgen-deprivation therapy (ADT), and yet others recommend intermittent ADT (i.e., an ADT regimen that includes periods of treatment suspension between ADT cycles in the absence of signs of biochemical or clinical progression) (9, 10). Although previous studies had demonstrated that enzalutamide in combination with ADT improved survival in patients with mHSPC (11, 12), little research had been conducted into the effects of this combination therapy in patients with nmHSPC. To address this gap, EMBARK—a phase 3, three-arm randomized trial that included 1068 patients—was conducted to evaluate the use of enzalutamide for high-risk BCR nmHSPC (13–15). The results of EMBARK demonstrated that enzalutamide with or without ADT was associated with improved oncological outcomes, with no new observed safety signals or decrease in quality of life (13, 16).

Based on the findings of EMBARK, enzalutamide is now the only androgen receptor pathway inhibitor approved by the United States Food and Drug Administration (FDA) (17) and European Medicines Agency (18) for the treatment of high-risk BCR nmHSPC, and it is specifically named as a preferred systemic treatment option for high-risk BCR nmHSPC by the National Comprehensive Cancer Network^® (NCCN) (10) and the European Association of Urology (9). However, other novel therapies may also have potential benefits in the treatment of BCR nmHSPC. Therefore, the objective of this network meta-analysis (NMA) was to provide indirect evidence of the relative efficacy of enzalutamide versus current and emerging systemic therapies for the management of patients with high-risk BCR nmHSPC whose disease has progressed after definitive therapy.

We conducted a systematic literature review (SLR) to form the evidence base for the NMA. Searches were conducted on April 13, 2022, and October 3, 2023. Articles and conference abstracts were obtained by searching the Embase^®, MEDLINE^®, and Cochrane^® databases. Search strategies are presented in Supplementary Tables 1A–E . We also conducted manual searches on October 3, 2023, of conference records, trial registries, and reference lists of existing systematic literature reviews. A protocol was not prepared for this review. This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.

A total of 3560 citations were identified in the SLR, from which 16 publications or presentations based on 10 eligible trials were included in the NMA ( Figure 1 ). The characteristics of the 10 trials are presented in Table 2 , and trial design details are presented in Supplementary Table 2 .

In terms of heterogeneity, baseline patient characteristics and duration of follow-up were generally comparable between studies ( Table 2 ). Patients’ median age ranged from 64–74 years. Median serum PSA was lowest among patients in the Morris 2021 study (ADT alone: 0.7; ADT + docetaxel: 0.8). Follow-up periods ranged from 4.9 months to 64.4 months. The availability of each outcome is presented in Supplementary Table 3 , and networks of evidence are presented in Supplementary Figures 1A–F . Outcome definitions were generally consistent across trials. In all trials, OS was defined as the time from randomization to death from any cause. For time to PSA progression, variations were noted in PSA progression thresholds across the included trials. Time to castration resistance was defined as “three increases in the PSA level at least 1 month apart or evidence of new clinical disease while the patient was receiving ADT and the testosterone was at castrate levels” in two studies (14, 25) and as “a rise in PSA while on ADT” in one study (26). For the proportion of patients with PSA < 0.2 ng/ml, EMBARK assessed serum PSA at 36 weeks (14, 15, 22, 23), while Autio et al. (2021) assessed serum PSA at 32 weeks (24). Finally, safety events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) versions 3.0 (34), 4.0 (35), or 4.03, with the exception of the SWOG-JPR7 study, where the methods for assessing adverse events (AEs) were not specified. For the purposes of this NMA, AEs in the trial by Morris et al. (27) that were not specified as “treatment emergent” but were presented by treatment arm were assumed to be treatment related, as no “unexpected” grade ≥3 AEs were reported.

In the evaluation of the proportional hazards assumption for the EMBARK trial, no pattern was identified in the Schoenfeld residual plot over time, indicating no evidence against the null hypothesis of proportional hazards for OS and MFS. Similarly, no violations of the proportional hazards assumption were identified for time to PSA progression or time to castration resistance. The results of the proportional hazards evaluation are presented in Supplementary Figure 2 .

The results of the risk-of-bias assessment are presented in Supplementary Table 4 . Most studies were deemed to be low risk or to have some concerns.

Patients with high-risk BCR have several different treatment options. To systematically compare treatments, we conducted, to our knowledge, the first NMA focusing on the effectiveness of enzalutamide for the treatment of high-risk BCR nmHSPC. Our analysis included 16 citations based on 10 phase 2/3 trials. We assessed the efficacy of multiple treatments on several clinically important endpoints, including OS, MFS, and time to PSA progression. Finally, to explore the impact of trial heterogeneity on our conclusions, we conducted a wide range of sensitivity analyses and found consistent results across all scenarios. We found that enzalutamide combination therapy demonstrated superiority over ADT alone and enzalutamide monotherapy for OS and over all comparators for MFS (vs enzalutamide monotherapy, ADT alone, and ADT + docetaxel) and time to castration resistance (vs ADT alone). It also demonstrated superiority to all comparators for time to PSA progression (vs enzalutamide monotherapy, abiraterone monotherapy, ADT + abiraterone, ADT + docetaxel, and ADT alone, as well as ADT + apalutamide and ADT + abiraterone + apalutamide in sensitivity analysis). Importantly, for oncological outcomes, no treatment was superior to enzalutamide combination therapy. Likewise, enzalutamide monotherapy demonstrated superiority to ADT alone and ADT + docetaxel in terms of MFS and time to PSA progression. These benefits must be balanced against the higher rates of grade ≥3 TRAEs with enzalutamide (with or without ADT) relative to other options except ADT + docetaxel, which was associated with more grade ≥3 TRAEs than enzalutamide. Together, the findings support the use of enzalutamide with or without ADT as a new standard of care for patients with high-risk BCR.

The benefits identified for enzalutamide in this analysis are aligned with treatment recommendations already included in current clinical guidelines. Specifically, the NCCN guidelines list enzalutamide as a preferred treatment option for patients with high-risk BCR nmHSPC (10). Similarly, the 2024 EAU guidelines contain a “strong” recommendation for providers to offer enzalutamide with or without ADT to patients with high-risk BCR nmHSPC (9). Further, our analysis is consistent with previous safety signals (36). In terms of the rate of grade ≥3 TRAEs, enzalutamide with and without ADT demonstrated superiority over ADT + docetaxel. The greater toxicity of docetaxel identified in our analysis, combined with a lack of greater clinical benefit, supports current recommendations for the use of enzalutamide in this patient population. Moreover, recent evidence from a Canadian study suggests that, compared with ADT alone, ADT + enzalutamide is cost-effective at established willingness to pay thresholds, making it a preferred treatment option for patients with high-risk BCR nmHSPC (37).

This study has some limitations. Survival data from several of the trials included in the NMA, including the EMBARK trial, are relatively immature, as the median survival time had not been reached for all survival outcomes at the time of data cut-off for each study; this could potentially lead to implausible estimates of survival benefit. However, previous trials that reported immature OS data for non-metastatic castration-resistant prostate cancer (nmCRPC) (PROSPER, SPARTAN, and ARAMIS) and metastatic hormone-sensitive prostate cancer (mHSPC) (ARCHES and TITAN) found that data maturity strengthened the results, and eventually, statistically significant and clinically meaningful benefits were achieved (38–42). In other indications for the use of enzalutamide in prostate cancer (nmCRPC, mHSPC, and mCRPC), a consistent effect was observed for enzalutamide at all evaluated disease stages, lending credibility to our findings (11, 12, 43–45). Within the trials included in this NMA, EMBARK had the longest follow-up period of all eligible trials that reported OS and included data from over 1000 patients. Of 271 OS events required across the treatment groups in EMBARK to achieve the protocol-defined power needed for this outcome, nearly half (n = 130, 48%) had occurred by the cut-off date, which supports these potential survival benefits.

Although both fixed- and random-effects models were run, the latter were limited by the small number of studies informing each treatment comparison in our analysis, resulting in less precise estimates. Thus, the fixed-effects models were used for inference, which does not account for between-study heterogeneity, and may, therefore, underestimate uncertainty in effect estimates As potential sources of clinical heterogeneity were identified in the evidence base, a series of sensitivity analyses were conducted to explore the impact of removing studies that deviated from the other studies in terms of baseline characteristics. After assessing the impact of excluding these studies, the overall results remained consistent, which lends credibility to our findings. In addition, the sparse evidence networks in this NMA mean that credible intervals were wide, leading to high uncertainty, particularly for the proportion of patients with undetectable PSA at 36 weeks and the rate of grade ≥3 TRAEs. Specifically, treatment-emergent and unspecified grade ≥3 AEs in the Morris study (27) were assumed to be TRAEs, which may have resulted in overestimation. However, as no unexpected grade ≥3 AEs were reported in the included trial, we do not expect this assumption to have substantially impacted the results. Additionally, treatment comparisons for TRAEs were not adjusted for time on treatment, which may have impacted our findings. Finally, we did not assess the exact make-up of TRAEs. For example, although enzalutamide combination showed superiority over enzalutamide monotherapy in multiple oncological outcomes with similar safety, there are differences in the side effect profiles of the two that may favor use of one regimen over another (i.e., better preserved sexual function with monotherapy). As such, the choice to use a particular regimen should be based on shared decision-making after evaluating efficacy, grade ≥3 TRAEs, and any specific side effects.

We evaluated the feasibility of conducting subgroup analyses, but the small number of studies reporting subgroup results meant that this was not viable. However, based on the results of the studies that conducted and reported the results of subgroup analyses (e.g., EMBARK), no groups were identified that would not be expected to benefit from treatment with enzalutamide (13).

Finally, a limitation of meta-analyses in general is that the results of more recent studies that were published or analyzed after the initial literature review was conducted will not be captured. In this study, available data on apalutamide was preliminary, or included only in sensitivity analyses, and was only available for time to PSA progression. Although enzalutamide with and without ADT demonstrated superiority over apalutamide in a sensitivity analysis, future analyses should incorporate more recently available apalutamide data to confirm this finding. Similarly, although the SLR search strategy followed best practice guidelines, there remains a risk that not all relevant studies were captured.

Overall, this novel NMA provides up-to-date evidence on the relative efficacy and safety profile of interventions for the treatment of high-risk BCR nmHSPC, demonstrating that enzalutamide with or without ADT provides considerable oncological benefit in high-risk BCR nmHSPC, albeit with a higher risk of TRAEs compared to ADT alone. Future research should prioritize updating this NMA to incorporate more mature OS data, as well as more recent data from the treatments evaluated in the NMA. Further research is also needed to identify predictive biomarkers that may help to identify tumors that are more sensitive to treatment with ARPIs compared with other mechanisms of action (e.g., chemotherapy).