Real-world Treatment Sequencing in Patients with Metastatic Castration-resistant Prostate Cancer: Results from the Prospective, International, Observational Prostate Cancer Registry

Take Home Message Patient characteristics appear to influence the first-line treatment of metastatic castration-resistant prostate cancer. Patients older and with lower Gleason scores tended to receive abiraterone, those younger and with more advanced disease tended to receive docetaxel, and those with fewer metastases and more favourable performance tended to receive enzalutamide. The benefit to patients from these observations remains unknown.


Introduction
Prostate cancer is the second most frequently occurring cancer and the fifth leading cause of cancer death in men worldwide [1], with an estimated 1.41 million new cases globally in 2020 [2]. Since 2010, several new agents have been approved for metastatic castration-resistant prostate cancer (mCRPC) that have demonstrated an overall survival (OS) benefit in clinical studies. However, robust evidence based on randomised trials for optimal treatment sequences for mCRPC is lacking [3][4][5], with no high-level recommendations in guidelines or expert consensus [6].
Current prostate cancer registries primarily report incidence and mortality data in selected countries and regions [7,8]. They are not designed to enable comprehensive assessment of patient characteristics that can determine optimal treatment allocation, necessitating further registry studies. The Prostate Cancer Registry (PCR; NCT02236637) was initiated to document the characteristics and disease management of patients with mCRPC in routine clinical practice, independent of the treatment used. With >3000 enrolled patients, the PCR is the largest international registry of patients with mCRPC to date, with participation from 199 centres experienced in treating prostate cancer in 16 countries (Austria, Belgium, France, Germany, Israel, Italy, Luxembourg, Poland, Portugal, Russia, Slovenia, Spain, Sweden, Switzerland, Turkey, and the UK). The PCR was designed specifically to record the initiation of a new systemic mCRPC treatment during surveillance within routine clinical practice. Data from the PCR have shown the effectiveness, in parallel, of first-line abiraterone acetate plus prednisone (referred to as abiraterone hereafter), enzalutamide, and docetaxel, including in patients with comorbidities, such as cardiovascular disease or diabetes mellitus, and in patients with visceral metastases [9]. An analysis of PCR data has also recently shown the efficacy and tolerability of abiraterone as first-and second-line treatment for mCRPC in the real-world setting [10]. This analysis of data from the PCR was conducted to increase the understanding of real-world treatment pathways and outcomes in a broad group of patients treated with abiraterone, enzalutamide, or docetaxel as their initial treatment for mCRPC. It uses real-world information on baseline demographics and disease characteristics of patients with mCRPC who received sequential treatment with hormonal therapy and chemotherapy in this setting.

Patients and methods
This was a prospective, noninterventional, multicentre PCR data analysis of men with mCRPC, from June 14, 2013 to July 9, 2018; data from the PCR have been described previously [9]. This analysis was limited to men included in the PCR treated with abiraterone, enzalutamide, or docetaxel as first-line mCRPC therapy who had not received previous systemic mCRPC treatment. Abiraterone and enzalutamide were not routinely available for the treatment of mCRPC in all 16 countries at the start of the registry.
The study timeline is shown in Supplementary Figure 1. For each patient, first-line and subsequent mCRPC treatments were recorded until the PCR ended (for up to 3 yr). The criteria for inclusion of sequences were as follows: (1) first-and second-line treatments were to be used as single agents in addition to androgen deprivation therapy and approved by the European Medicines Agency for use in this patient population (ie, no off-label use) and (2) patient numbers in the sequence were to be adequate for a reliable analysis (!25 patients with a sequence involving two lines of treatment in mCRPC).
Observational data collected for each treatment sequence included baseline patient demographics; disease and health-related quality of life characteristics at baseline and the start of second-line treatment, and their changes from baseline; reasons for starting and stopping mCRPC treatment; and efficacy outcomes.
Disease and health-related quality of life data included Eastern Cooperative Oncology Group (ECOG) performance status; prostate-specific antigen (PSA), lactate dehydrogenase, alkaline phosphatase, and haemoglobin levels; and EuroQol 5-dimension 5-level questionnaire (EQ-5D-5L) pain/discomfort dimension. Efficacy outcomes were time to progression (TTP), measured from the start of first treatment to the start of  Treatment decisions were made at the discretion of the treating   physician, as per routine clinical practice, and only data available from clinical practice were collected. The treating physician could be either an oncologist or a urologist, designation of which was not recorded in the registry. Prior disease history and management data were collected at study inclusion. Clinical data were collected at study inclusion and prospectively every 3 mo during routine follow-up (at least every 3 mo per protocol) over the 3-yr study period, with clinical trial levels of monitoring quality.
The study was conducted according to the applicable regulatory prerequisites. Prior to data collection, all patients provided informed consent in accordance with local requirements. The study was approved by the ethics committees of every country that participated and was conducted in accordance with the Declaration of Helsinki.

Statistical methods
Descriptive statistics were used to present observed data from routine clinical practice by treatment sequence for three first-line treatment groups. Percent data for outcomes were calculated from the number evaluable for each parameter as the denominator. Kaplan-Meier methods were used to analyse time-to-event variables (TTP, PFS2, and OS) for each treatment sequence, including median estimates and corresponding 95% confidence intervals. Missing demographic and baseline data were typical of this type of study and were not expected to substantially influence results. Statistical comparisons of efficacy outcomes were not performed; therefore, missing data imputation methods were not applied (no sensitivity analyses), and subgroups were constructed based on medical criteria (ie, treatment sequence received).

Patients overall
Data from 727 patients in the PCR were eligible for inclusion, with seven treatment sequences analysed

3.2.
Patients who received first-line abiraterone

Reasons for switching treatment
Disease progression (assessed by PSA and/or radiological and/or clinical methods) was the most frequently cited reason for initiating and stopping first-line abiraterone; toxicity was cited as the second most common reason for stopping therapy early (Supplementary Table 1).

Efficacy outcomes
The median TTP, PFS2, and OS (Kaplan-Meier estimates) ranged from 5.8, 15.9, and 27.0 mo, respectively, in the ABI-DOCE sequence to 8.7, 20.1, and 30.8 mo, respectively, in the ABI-ENZA sequence (Figs. 1-3 and Table 3). The median PSA, lactate dehydrogenase, and alkaline phosphatase increased from baseline to the start of second-line therapy in all sequences in the abiraterone group (Table 2).

3.2.4.
Change from baseline to start of second-line therapy in disease and health-related quality of life characteristics The proportion of patients with no change or improvement in ECOG performance status was 63.8% in the ABI-ENZA sequence, 53.4% in the ABI-DOCE sequence, and 43.4% in the ABI-RAD sequence ( Table 2).
The proportion of patients with no change or improvement in EQ-5D-5L pain/discomfort dimension scores ranged from 62.3% in the ABI-DOCE sequence to 50.0% in the ABI-RAD sequence ( Table 2).

3.3.
Patients who received first-line docetaxel

Baseline characteristics
In the first-line docetaxel sequences, the median patient age ranged from 68.5 yr in the DOCE-CABA sequence to 71.0 yr in the DOCE-ENZA sequence, and 18.9-35.3% of patients were aged !75 yr across sequences (  Table 3).

3.3.4.
Change from baseline to start of second-line therapy in disease and health-related quality of life characteristics The proportion of patients with no change or improvement in ECOG performance status from baseline to the start of second-line therapy ranged from 74.3% in the DOCE-ABI sequence to 39.3% in the DOCE-CABA sequence ( Table 2).
The proportion of patients with no change or improvement in EQ-5D-5L pain/discomfort dimension scores from baseline to the start of second-line therapy was 78.9% in the DOCE-ENZA sequence, 73.7% in the DOCE-ABI sequence, and 57.6% in the DOCE-CABA sequence ( Table 2).

Baseline characteristics
The median age of patients at baseline was 70.5 yr, with 31.0% of patients aged !75 yr. The proportion of patients with five or more bone metastases was 25.0% and that with visceral metastases was 8.8% (Table 1). No patients in this group had ECOG performance status !2 ( Table 2).

Reasons for switching treatment
Disease progression was the most frequently cited reason for stopping first-line enzalutamide (Supplementary Table 1). All patients who initiated second-line docetaxel after enzalutamide did so because of disease progression (Supplementary Table 1).

3.4.4.
Change from baseline to start of second-line therapy in disease and health-related quality of life characteristics There was no change or improvement in ECOG performance status from baseline to the start of second-line therapy in 40.9% of patients (Table 2). In the EQ-5D-5L pain/discomfort dimension, 55.5% of patients had no change or improvement from baseline to the start of second-line therapy.

Discussion
The aim of this analysis was to report the use of different treatment sequences in real-world clinical practice. This report evaluates the characteristics and outcomes of the largest international cohort of patients receiving the most frequently prescribed first-and second-line mCRPC treatment in a real-world setting and presents information from patients in 16 countries.
There was a high degree of variation in demographics and disease characteristics between the sequence groups. While it is interesting to note this variation, the nonrandomised nature of this study and variation in availability of treatments in different countries mean that formal comparisons between treatment sequences cannot be made. Overall, abiraterone appeared to be prescribed more commonly as a first-line treatment to older patients and to     those with lower Gleason scores. In general, patients with M1 disease at initial diagnosis with visceral metastasis were more commonly treated with first-line docetaxel than other first-line treatments. Patients with five or more bone metastases and those with visceral metastases appeared less likely to receive first-line enzalutamide. Baseline ECOG performance status scores were lowest in the first-line enzalutamide group. Thus, baseline patient characteristics could play a role in the selection of first-line treatment for mCRPC in the real-world setting. This was suggested in the ABI-RAD sequence, which appeared to be more commonly given to patients with five or more bone metastases, and only one patient with visceral metastases was given this treatment sequence; this is in accordance with the supporting indication for radium-233 in mCRPC [11]. There appeared to be a preference for first-line treatment with docetaxel in patients with visceral metastases and in patients of a younger age, likely because younger patients would be expected to tolerate docetaxel better than older patients [12]. In this context, it is rather surprising that the highest proportion of patients with ECOG performance status !2 was in the DOC-CABA sequence, although this was possibly influenced by reimbursement of the treatment options in different countries.
Efficacy outcomes were strictly observational, and no statistical comparisons were made between the treatment groups, as this was not a randomised study. We observed variation in TTP, PFS2, and OS within the first-line abiraterone group, with the longest duration of these seen when enzalutamide was given as second-line treatment. In patients who received first-line docetaxel, the TTP was similar between sequences; PFS2 and OS appeared shorter in patients who received second-line cabazitaxel, which could be influenced by the slightly higher proportion of patients with visceral metastases or ECOG performance status !2 in this sequence. Treatment groups were small, nonrandomised, and not stratified by patient characteristics; therefore, any differences observed in efficacy outcomes should be interpreted with caution. The efficacy data observed aid the understanding of mCRPC treatments in the real-world setting but cannot indicate an optimal treatment sequence in this patient population.
These results from the prospective analysis of data from the PCR can be placed in context of a small number of largely retrospective studies specifically investigating treatment sequencing outcomes for patients with mCRPC; however, methodological limitations restrict the interpretation of the findings from these studies. A recent open-label, phase II, crossover trial that was limited to a single country and included just 202 patients suggested a clinical benefit for abiraterone followed by enzalutamide relative to an ENZA-ABI sequence [13]. However, this benefit was limited to PSA-based endpoints, and no significant difference in OS was found between treatment sequences [14]. Outcomes with enzalutamide following abiraterone have also been compared between 115 docetaxel-experienced and docetaxel-naive patients in a Canadian study [15]. This study reported that the antitumour activity of ABI-ENZA was limited in patients with mCRPC, notwithstanding prior docetaxel, suggesting that earlier treatment with docetaxel does not lead to efficacy benefits for the ABI-ENZA sequence.

Limitations
This was an observational study, with no randomisation to allow treatment/sequence comparisons. As treatment followed routine clinical practice, duration of exposure varied with treatment cycle, and although some visits were performed according to the study protocol, routine clinical visits also resulted in data collection. These visits may not have occurred at regular intervals for all patients. The staggered commercial availability of the treatments in participating countries and differences in reimbursement and access to high-cost medication present potential for a bias, for example, newer treatments such as abiraterone and enzalutamide may not have been available in some countries at the time of data collection. Additionally, differences in treatment practices and settings (eg, among physicians [oncologists or urologists] and also practitioners in different countries) and timing of enrolment may have influenced the findings. Furthermore, patients' preference of treatment may have been considered by treating physicians; however, this information was not collected. It is not possible to compare the efficacy of the different drug combinations statistically because of differences in demographics and patient characteristics in the treatment groups.

Conclusions
This analysis of real-world baseline characteristics, treatment patterns, and effectiveness outcomes in patients with mCRPC provides data on real-life application of treatment sequencing for mCRPC. The differences we observed in base-line patient characteristics suggest that these play an evident role in the selection of first-line treatment for mCRPC in the real-world setting, but we acknowledge that other factors, such as treatment availability, have an impact. The efficacy data observed provide good insight into real-world treatments for mCRPC, but are limited and cannot guide optimal treatment sequencing in this patient population.
Author contributions: Anders Bjartell had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Wapenaar.
Acquisition of data: Not applicable.  28.1 (20.0-NE) ABI = abiraterone acetate plus prednisone/prednisolone; CABA = cabazitaxel; CI = confidence interval; DOCE = docetaxel; ENZA = enzalutamide; mCRPC = metastatic castration-resistant prostate cancer; NE = not estimable; RAD = radium-223. a Measured from the start date of the first mCRPC treatment to the start date of the second treatment. Time to progression was censored after the start of the second treatment. Patients with no progression at the end of registry are censored. b Measured from the start of first treatment to progression/death after the start of second treatment. Patients with no progression at the end of registry are censored. c Measured from the start date of mCRPC treatment to the date of death (irrespective of cause); survival time of living patients was censored at the last date a patient was known to be alive (for those withdrawn from the study) or the end-of-registry date. E U R O P E A N U R O L O G Y O P E N S C I E N C E 4 5 ( 2 0 2 2 ) 1 2 -2 2