Prostate cancer (PCa) is the second most common cancer in men globally, with a rapidly increasing incidence, particularly in China. Androgen deprivation therapy (ADT) is a first-line treatment for aggressive PCa due to its androgen dependence. However, resistance mechanisms like androgen receptor (AR) amplification, AR splice variants, and AR point mutations lead to the development of castration-resistant prostate cancer (CRPC) within 2-3 years of ADT. Second-generation androgen receptor inhibitors (ARIs) have been developed to address this resistance. This network meta-analysis aimed to indirectly compare the therapeutic efficacy and adverse events of the conventional ARI bicalutamide with three second-generation ARIs (enzalutamide, apalutamide, and darolutamide) in treating CRPC patients.

The introduction cites several studies highlighting the mechanisms of resistance to ADT in prostate cancer, emphasizing the role of AR alterations in CRPC progression. The need for second-generation ARIs to overcome this resistance is established, setting the stage for the meta-analysis.

This network meta-analysis followed the PRISMA statement and was registered on PROSPERO. PubMed, Embase, and the Cochrane Library were searched for RCTs published up to October 1, 2022, evaluating bicalutamide and second-generation ARIs in CRPC patients without prior chemotherapy, all receiving background ADT. Inclusion criteria focused on RCTs assessing PFS, PSA-PFS, OS, PSA response rate, metastasis-free survival (MFS), and various adverse events (AEs), including overall AEs, Grade 3+ AEs, serious AEs (SAEs), AE-related drug withdrawals, AE-related mortality, fatigue, and hypertension. Risk of bias was assessed using the Cochrane Handbook tool. Bayesian framework using the Gemtc package in R (version 4.1.2) was used for data analysis. Hazard ratios (HRs) with 95% confidence intervals (CIs) were used for dichotomous data, and mean differences (MDs) with 95% CIs for continuous data. Subgroup analyses were conducted for non-metastatic (nmCRPC) and metastatic (mCRPC) CRPC.

Seven RCTs involving 6,993 patients were included. Second-generation ARIs (enzalutamide, apalutamide, and darolutamide) demonstrated statistically significant improvements in PFS, PSA-PFS, and OS compared to bicalutamide and placebo. Apalutamide showed the most benefit in terms of PSA-PFS. However, there were inconsistencies among the second-generation ARIs; darolutamide appeared slightly less effective than enzalutamide and apalutamide in PFS and PSA-PFS. Regarding AEs, while all ARIs increased the risk of overall AEs compared to placebo, apalutamide showed a higher incidence of Grade 3+ AEs, AE-related drug withdrawals, and AE-related mortality, though not statistically significant compared to others. Enzalutamide showed a significantly higher rate of hypertension and fatigue. Subgroup analyses showed similar trends in both nmCRPC and mCRPC groups, with enzalutamide demonstrating the greatest benefit in mCRPC.

The findings support the superiority of second-generation ARIs over bicalutamide in treating CRPC. The lack of complete equivalence among the second-generation ARIs highlights the need for careful patient selection based on individual risk profiles and potential side effect considerations. The observed differences, while not always statistically significant, suggest potential variations in efficacy and safety profiles that warrant further investigation. Discrepancies between preclinical findings (higher AR inhibition by darolutamide) and clinical outcomes may be due to differences in patient populations across trials (e.g., inclusion/exclusion criteria regarding lymph node size).

This network meta-analysis demonstrates the superiority of second-generation ARIs over bicalutamide for CRPC. However, subtle but notable differences exist among the second-generation ARIs, particularly regarding efficacy and adverse events. Future research should focus on direct comparative trials and explore biomarkers to predict patient response and tailor treatment strategies.

The study acknowledges limitations such as potential for outcome bias due to unclear risk of bias in some original studies, insufficient follow-up time limiting the assessment of OS, and the inability to test consistency due to the absence of a complete 'ring' of interventions in the network meta-analysis.