Prostate cancer (PCa) is one of the most common cancers worldwide. According to the Surveillance Epidemiology and End Results Program, it is estimated that in 2022 as many as 34,500 people will die of PCa (1, 2). Androgen deprivation therapy (ADT) remains the key systemic therapy for patients with metastatic prostate cancer. Despite the initial sensitivity to ADT, PCa transforms into the uncurable castration-resistant stage of the disease (3). Currently registered treatment options for PCa patients consist of chemotherapy (docetaxel (DXL), cabazitaxel) and second-generation antiandrogens (abiraterone acetate (ABI), enzalutamide (ENZA), apalutamide, darolutamide), radiopharmaceutical therapy (Radium-223), immunotherapy (sipuleucel-T, pembrolizumab) and PARP inhibitors (olaparib, rucaparib) (4, 5). The drug selection depends on the castration status, tumor stage and genetic mutation status. According to National Comprehensive Cancer Network, oncological guidelines ABI and ENZA are the first-choice treatment options for patients who progressed after prior DXL chemotherapy and were not treated with novel hormone therapy (category 1) (6).

Advances in cancer pathobiology and understanding the mechanisms contributing to cancer progression allow for designing antiandrogen-targeted therapy. ABI and ENZA changed the treatment landscape for patients with PCa (7, 8). Several differences between these drugs include mechanisms of action, dosing method and pattern of side effects. ABI inhibits 17α-hydroxylase/C17.20-lyase (CYP17), an enzyme involved in the biosynthesis of androgens (8). ENZA inhibits the androgen receptor signaling pathway affecting androgen binding to androgen receptors, translocation of androgen receptors to the nucleus and interaction with DNA (7). Both drugs are used once daily, ABI needs additional steroid supplementation. ABI and ENZA are registered treatment options for chemotherapy-naive and chemotherapy-pretreated PCa patients with metastases. COU-AA-301 and AFFIRM were the phase III randomized clinical trials which showed that ABI and ENZA increase overall survival (OS) in PCa patients previously treated with chemotherapy (7, 8). There is a lack of established predictive factors and head-to-head comparative studies, which may facilitate clinicians’ choice between these two drugs. The current study aimed to compare the oncological outcome measures and evaluate prognostic factors affecting survival in real-life populations of patients treated with ABI and ENZA in Poland.

This retrospective analysis was approved by the local bioethics committee in Olsztyn (5/21/VII.) and was conducted in accordance with the Declaration of Helsinki. The analyzed data were obtained from the Polish National Health Fund Drug Program database. The study population consisted of patients with metastatic castration-resistant prostate cancer (mCRPC) receiving 1,000 mg ABI with 10 mg prednisone or 160 mg ENZA once a day who progressed after chemotherapy with DXL and were qualified for this treatment in seven comprehensive cancer centers in Poland (Olsztyn, Otwock, Cracow, Brzozów, Wrocław, Siedlce, Przemysl) between 2014-2021 (ABI) and 2018-2021 (ENZA). The inclusion and exclusion criteria were based on the Polish National Health Fund Department, which reimbursed the treatment. The criteria are mostly equivalent to those from clinical trials.

Inclusion criteria included: age over 18 years, a pathological diagnosis of prostate adenocarcinoma with radiologic evidence of metastases, and a serum testosterone level of 50 ng/dL or less (≤1.7 nmol/L) due to ADT (surgical or pharmacological). All qualified patients received DXL before ABI or ENZA treatment. The disease progression after or during chemotherapy was defined as biochemical progression if a patient had two consecutive increases in the prostate-specific antigen (PSA) concentration (from the lowest PSA level reached during or after DXL) or radiological progression (radiographic evidence of disease progression in bone or soft tissue). All patients had an Eastern Cooperative Oncology Group (ECOG) performance status score of 0 or 1. Patients could not be qualified if they had significant hepatic dysfunction, unstable or uncontrolled cardiovascular disorders, or a history of prior abiraterone acetate, enzalutamide or ketoconazole therapy. All the patients’ characteristics were registered before the initiation of ABI/ENZA treatment.

The primary endpoints of the analysis were:

Disease progression* was defined according to the Polish drug program:

The analyzed clinicopathological data included:

The statistically significant variables in univariate analysis were chosen for multivariate analysis.

Statistical analyses were performed using Stata^® Software ver. 14.1 (StataCorp LLC). Nominal parameters were presented as a percentage frequency. The study used the χ2 test (categorical variables), the independent t-test (continuous, normally distributed variables) and the Mann-Whitney U (non-normally distributed variables) for comparisons between the groups. The r-Pearson and Spearman correlations were used for assessing the association between continuous variables. Survival curves and Cox proportional hazard model (univariate and multivariate) were used to determine the predictors for longer TTF/OS during ABI/ENZA treatment. A level of p<0.05 was recognized as statistically significant.

Real-life studies allow for a better understanding of disease courses in a specific population of patients and facilitate the selection of a drug, which is especially important for practicing oncologists. The choice of the drug by clinicians may be based on patient comorbidities, expected side effects, patient preferences and cost-effectiveness. The multicenter retrospective analysis in the current study is the first study to describe the outcomes for mCRPC patients treated with ABI and ENZA in Poland. The study was conducted on all subsequent patients from these centers who met the criteria for participation in the Polish drug program, which meets the criteria of a retrospective case-control study (evidence level IIIE). There is a lack of randomized, comparative phase III trials in patients treated with ENZA and ABI in mCRPC, which justifies the current real-life clinical data analysis.

COU-AA-301 and AFFIRM were registered trials that determined the role of ABI and ENZA in mCRPC patients pretreated with DXL chemotherapy (7, 8). A total of 800 patients were treated with ENZA, and 300 were treated with a placebo in AFFIRM. The final analysis revealed that the median of OS was 18.4 vs. 13.6 months, respectively (p < 0.001, HR: 0.63; 95% CI: 0.53–0.75). The trial also met the secondary endpoints, including the proportion of patients with a reduction in the PSA level by 50% or more (54% vs. 2%, p<0.001), the time to PSA progression (8.3 vs. 3.0 mo.; p<0.001), radiographic progression-free survival (rPFS) (8.3 vs. 2.9 mo.; p<0.001). The analysis of HR for death showed the superiority of ENZA over placebo in all patient subgroups, including age (65< vs. ≥65 years), baseline ECOG, type of progression at entry study, visceral disease and PSA level at baseline (7). In the COU-AA-301 trial, 797 patients were treated with ABI and prednisone, and 398 received a placebo. The median OS for the study group was 15.8 months (95% CI: 14.8-17) vs. 11.2 months (10.4-13.1; HR 0.74, 95% CI: 0.64-0.86; p<0·0001). The trial also met the secondary endpoints: median time to PSA progression (8.5 vs. 6.6 mo. p<0·0001), median rPFS (5.6 vs. 3.6 mo. p<0.0001), and proportion of patients who had a PSA response (29.5% vs. 5.5%; p<0.0001). The analysis of HR also confirmed the superiority of the study drug over the placebo in subgroups, including age, ECOG and type of progression (8).

In the current study, the median OS was 15.7 months in the ABI group (95% CI: 13.5 - 17.4) and 26.1 months (95% CI: 17.8-29.4) in the ENZA group. The difference between drugs was statistically significant. TTF was also longer in the ENZA group (14.2; 95% CI: 11.3-17.6 mo.) than the ABI group (7.6; 95% CI: 6.64-8.85; p<0.001). The 50% PSA decline was higher in the ENZA group than in the ABI group (87.5 vs. 56%; p<0.001), which was higher than in the registered trials (56 vs. 29.5%). The 50% PSA decline was observed more frequently in the ENZA group (87.5%) than in the ABI group (56%; p<0.001). The result of OS in the current study and the registered trial was comparable to the ABI group (15.7 vs. 15.8 months) but differed from the ENZA group (26.1 vs. 18.4 months). However, the length of treatment was longer in the ENZA group (p<0.001), TTF and rPFS are not comparable oncological outcome measures.

The observational study of real-life clinical data was analyzed in many countries, and results vary between the different populations of patients. The main differences between the studies include the number of patients, the line of treatment (pre or post-chemotherapy) and the following lines of treatment after progression.

The most extensive observational study was published by Schoen et al. and presented the results of treatment on 5,822 US veterans. It shows that the OS was longer in patients treated with ENZA than ABI (24.2 vs 22.1 months) in pre- and post-chemotherapy treated patients. Importantly, patients with cardiovascular diseases also had better survival (9). Also, a direct comparison of drugs in 10,308 chemotherapy-naive patients with CRPC based on the 2014-2018 French population study showed that patients treated with ENZA had a better OS than ABI (34.2 vs. 31.7 mo.) (10). Li et al. also recently published the results of a retrospective cohort population-based study in the unselected Taiwanese population, which showed that treatment with ENZA (n=118) was associated with better OS than treatment with ABI (n=1046), although without differences in TTF (11). In another unselected Australian population of 250 patients (53% of ABI; 38% of ENZA post-chemo), patients treated with ENZA had a greater PSA response (70.3% vs. 39.5%). The OS was longer in the ENZA group, 29 months (95% CI: 21.3-36.7) vs. 7 months (95% CI: 0-18.5%; p=0.002) in the ABI group. The authors did not find any factors significantly associated with OS, including the Gleason scale (12). Interesting data come from Chowdhury et al., who analyzed the oncological outcomes among patients from 16 countries with mCRPC treated in the first line of therapy with ABI, ENZA or chemotherapy. The OS for ABI and ENZA was the same (27.1 months) (13).

Data from selected post-DXL-treated patients show similar results. Contrary to the pre-DXL setting, in post-DXL, the OS in the ENZA (26 ± 7 (12.3-39.7)) was longer than the ABI (13 ± 1.6 (9.8-16.2); 0.021) cohort of patients. Moreover, the rPFS was longer in the ENZA (11± 5.1 (1.1-20.9) than the ABI (pre-and post-DXL) group. PSA ≥50% decline occurred more frequently in the ENZA group than in the ABI group (p=0.02). The authors concluded that the good prognostic factors for rPFS were ENZA treatment, age ≥75 years and PSA ≥50% decline at 12 weeks of treatment. The authors did not find any prognostic factors for OS (14).

Another Taiwanese study that indirectly compared the ENZA (n=13) and ABI (n=63) in post-DXL chemotherapy showed that the OS from second-line hormone treatment was 30.2 months in the ABI group and 16.2 months in the ENZA group, although statistical significance was not shown. The same study shows no difference between PSA and PFS responses. PSA 50% response was seen in 48.4% in ABI and 69.2% in the ENZA group (p=0.171); the median PFS was 7.3 months (95% CI: 4.79-9.80) in ABI and 9.5 months (95%CI: 5.743-13.257) in ENZA (p of log rank=0.766) (15). In Austrian populations of patients, the OS for ABI was 14 months (mean: 15.8 ± 0.9 months), and for ENZA it was 19 months (mean: 17.2 ± 1.4 months). A randomized phase II cross-over study confirmed that ENZA is associated with better biochemical response, however, without changes in time to PSA progression (16). Hu et al. found a difference in OS between ENZA and ABI (17.9 vs. 15.4 mo.; p=0.8224) (17). Other studies show different OS in post-DXL treated with ABI. In the Marret et al. trial, the OS was 13.4 months, although only 58 patients were analyzed (18).

Finally, Wei et al. performed a meta-analysis of 5,199 patients treated with ABI and ENZA in randomized clinical trials. Contrary to rPFS and time to PSA progression, which were significantly better in the ENZA group, the mOS did not vary significantly between ABI and ENZA (HR=1.03, 95% CI: 0.854-1.242) (19). The same results regarding OS came from Bianchi et al. analysis (20). Another indirect analysis of AFFIRM and COU-AA-301 trials favors ENZA in terms of time to PSA progression, PSA response and radiological PFS, although without difference in OS (21). Chung et al. also studied the effectiveness of both drugs in sequential treatment. The sequence ABI-ENZA is better regarding oncological outcomes than ENZA-ABI (22). In Poland, sequential treatment is not reimbursed.

The current study attempted to identify the prognostic factors associated with ABI and ENZA therapy. Although the univariate analysis revealed that ENZA treatment, ECOG 0, BMI ≥25 kg/m², total DXL dose ≥750 mg, ≥50% PSA decline, and PSA nadir during or after DXL treatment <17.35 ng/mL are factors associated with better survival, the multivariate analysis shows that ENZA treatment and total DXL dose ≥750 mg and PSA nadir <17.35 ng/mL during or after DXL treatment were independent prognostic factors for longer OS. Early PSA response is a good independent prognostic factor in next-generation androgen receptor inhibitors (23, 24). The ≥50% drop in PSA from baseline within the three months of treatment correlated with better OS and PFS (23). However, although it was found that DXL dose has an impact on the prognosis of patients with PCa, treatment with an early DXL was not found to be a prognostic factor. Patients with hormone-sensitive PCa were not included in clinical trials, and data in such a population of patients are limited and not well explored. Additionally, it was studied if BMI affects prognosis during antiandrogen treatment. The majority of patients (74.4%) were overweight or obese. In univariate analysis, patients with BMI ≥ 25 kg/m² had longer TTF and OS, although this lacked statistical significance in multivariate analysis. Other studies suggest that obesity may play a protective role associated with increased survival (25–27).

In COU-AA-301, the authors of the study found that patients who received DXL for more than three months had better OS than those treated for less than three months (8). Chi et al. published a risk model for predicting OS in chemotherapy-pretreated patients treated with ABI, including ECOG 2, presence of liver metastases and time from ADT to the start of ABI ≤36 months (28). The multivariate analysis of the hazard ratio for death showed that ECOG 0, mean pain score on Brief Pain Inventory-Short Form <4, PSA progression at study entry, and no visceral disease at screening were associated with better survival (7). Patients with age > 75, Charlson comorbidity scores > 2, presence of symptoms, time from prostate cancer diagnosis < 3 years and time from last chemotherapy < 6 months had lower survival (17). Multivariate analysis revealed that PSA response, Gleason score ≥8 and PSA-doubling time <2 months correlated with OS. Patients with visceral metastases had worse oncological outcomes in terms of OS (2.8 vs. 18; p=0.0007) and PFS (2.8 vs. 6.8; p=0.0088) (29). Another study showed that low levels of miR-21 are an unfavorable prognostic factor in PCa patients (30).

In the current study, it was shown that ENZA treatment may be related to more favorable oncological outcomes than ABI treatment in the Polish population of patients. There was no difference in the efficacy of ABI and ENZA on time to treatment failure in docetaxel-naive and docetaxel-pretreated prostate cancer patients. A 50% decline in PSA is an indicator of longer TTF and OS. Due to the non-randomized and retrospective nature of the analysis, the current results require prospective validation. Although many different observational and metanalyses show that treatment with enzalutamide has a better treatment outcome, there is no single explanation for the results. The main difference between drugs includes different action mechanisms because abiraterone inhibits 17α-hydroxylase/C17,20-lyase (CYP17), and enzalutamide has a threefold mechanism of action. First, it is a potent, competitive binder of androgens at the level of the androgen receptor (AR), and it prevents the translocation of the AR from the cytoplasm to the nucleus. Within the nucleus, it inhibits AR binding to chromosomal DNA, which prevents further transcription of tumor genes. Therefore, compared to abiraterone, enzalutamide may act more selectively and comprehensively on the AR signaling pathway in prostate cancer cells (31). The indirect analysis of drugs has several limitations which may also affect the results. One of the major disadvantages of real-life data trials is the heterogenicity of the groups and follow-up time in the study. In the current cohort of patients, the differences included the ECOG scale, GS and location of metastases. The population of patients treated with ENZA was in better performance status (ECOG 0: 32.3 vs. 16.3%), had a higher percent of bone-only limited metastases (58.2 vs. 43.8%), received the higher cumulative dose of DXL (62.5 vs. 48.7%), and more patients had PSA nadir of <17.35 during or after DXL (69.3% vs 43.1%) but had less differentiated tumors (72.6 vs. 51.4%) which may interfere with the results. ABI was introduced earlier than ENZA in Poland, which limits the adjustment of drug selection. Patients included in our study were not treated during the same period (ABI 2014-2021, ENZA 2018-2021); however, the subgroup analysis of patients treated during the same time supports the superiority of ENZA in terms of TTP and OS. The unbalanced population of patients were also an issue in similar studies. The current study limitations include an unbalanced study population, the retrospective nature of the study and the lack of some clinical aspects like pain score or quality of life (similar to other published studies).

The current study confirmed the clinical activity of ABI and ENZA in the Polish population of patients. The presented analysis suggests that treatment with ENZA may be related to more favorable outcomes than treatment with ABI.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by local bioethics committee in Olsztyn (5/21/VII.). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

Conceptualization, DS and MW; methodology, DS and MW; statistical analysis, MW; writing—original draft preparation, DS, data collection: all the authors, supervision: IS. All authors contributed to the article and approved the submitted version.