Marginal improvement in survival among patients diagnosed with metastatic prostate cancer in the second‐line antiandrogen therapy era

Abstract Since 2004, multiple blockbuster drugs have been approved for men with metastatic prostate cancer. Nevertheless, it has been reported that no improvement in survival was observed between 2004 and 2009. Herein, we have analyzed the SEER database to assess the survival outcome of metastatic prostate cancer patients since 2000. The results demonstrated that there was an improvement in both overall and prostate cancer‐specific survival for 4 months among men diagnosed with metastatic prostate cancer from 2010 to 2016 when compared to those in the pre‐2010 period. Interestingly, this survival benefit was limited to patients with bone and visceral metastasis (M1b and M1c stages). Collectively, our observation suggests that despite the new treatment agents such as second‐line antiandrogen therapies introduced in the modern era, the improvement in survival of metastatic prostate cancer patients has been surprisingly small.

States. On presentation, 76% are localized or confined to the prostate, 13% have spread to regional lymph nodes, 6% have metastasized, and 5% are unknown or unstaged. 2 This distribution is significant in which survival rates widely vary depending on stage with 5-year relative survival rates for localized, regional, and metastatic prostate cancer being approximately 100%, 100%, and 31%, respectively. 3 In men diagnosed with metastatic disease, androgen deprivation therapy (ADT) is the initial standard of care. 4 However, prostate cancer cells eventually become resistant to ADT and castration-resistant prostate cancer (CRPC) emerges. In men with CRPC, broadly available treatment options are chemotherapy, immunotherapy, and second-line antiandrogen therapy (SAT). 5 In 2004, the USFDA approved docetaxel, the first agent shown to improve survival rates in patients with CRPC. 6 Since then, several drugs shown to improve survival rates have been approved including cabazitaxel and sipuleucel-T in 2010 and second-line antiandrogen therapies (SAT) abiraterone and enzalutamide in 2011 and 2012, respectively. 7 More recently, additional SAT agents, apalutamide and darolutamide, as well as the PARP inhibitor olaparib have been added to the armamentarium against CRPC. [8][9][10] Despite the approval of docetaxel in 2004, Wu and colleagues reported that overall and prostate cancer-specific survival did not improve in patients with metastatic disease in the docetaxel era (years 2004-2009). 11 However, given the recent advent of several blockbuster drugs in the SAT class for CRPC since 2010 combined with the implementation of more intense therapy supported by clinical trials, 12,13 we hypothesized that there would be significant improvements in survival among patients diagnosed with metastatic prostate cancer from the pre-docetaxel to SAT era (years 2000-2016).

| Data Sources
The study sample was composed of patients at least 18 years old with distant prostate cancer from the Surveillance, Epidemiology, and End Results (SEER) Program, which publishes epidemiologic data on the incidence and survival rates of cancer in the United States. Using SEER*Stat statistical software, patient information was extracted from the Incidence -SEER 18

| Study variables
This study primarily examined the following study variables: age, race/ethnicity, year of diagnosis, treatment with prostatectomy and/or radiotherapy or no local therapy, and metastatic subclass. Metastatic subclass was based on the American Joint Committee on Cancer (AJCC) Stage 3 rd edition for patients diagnosed in 2000-2003, adjusted AJCC 6 th edition for those in 2004-2015, and derived SEER Combined Stage M for those in 2016.

| Statistical analysis
The primary study outcomes were overall survival (OS) and prostate cancer-specific survival (PCSS) based on the diagnostic time period and metastatic subclass when available. Secondary study outcomes investigated changes across years in the distribution of age, race/ethnicity, and treatment with prostatectomy and/or radiotherapy or no local therapy.
The following three time periods were examined: January 2000-December 2003 (era 1), January 2004-December 2009 (era 2), and January 2010-December 2016 (era 3). Overall and prostate cancer-specific survival were estimated using the Kaplan-Meier product limit method stratified by three time periods and their differences were analyzed using the log-rank test and univariate and multivariable Cox proportional hazards models were applied to assess the risk of death and prostate cancer-specific death with time periods. Hazard ratios (HR) and 95% confidence intervals were reported. In multivariable model, age, treatment, and race/ethnicity were analyzed. Because stage is not available for era 1, the stage was adjusted in comparing eras 2 and 3. Changes in the distribution of age, race/ ethnicity, and treatment across time periods were analyzed using Pearson's chi-square test. All statistical analyses were performed using Stata/SE 15.0 with a p-value of less than or equal to 0.05 considered statistically significant.

| Overall survival of patients with metastatic prostate cancer shows modest improvements between 2000 and 2016
The study sample consisted of 41,149 patients at least 18 years old diagnosed with distant prostate cancer between 2000 and 2016. We excluded 392 patients from era 1 (years 2000-2003), 159 from era 2 (years 2004-2009), and 192 from era 3 (years 2010-2016) for being M0 due to the likelihood of stage misattribution. Median follow-up for era 1, 2, and 3 for OS was 179, 117, and 36 months, respectively.
Patient characteristics among the three eras are shown in Table 1. Among the variables examined, men in the most recent SAT era (era 3, 2010-2016) were younger with a median age of 71 versus 73 and 72 in era 1 and era 2, respectively (p < 0.001). The racial/ethnic distribution of patients with metastatic prostate cancer also changed with the number of non-Hispanic white and black patients decreasing from 64.59% in 2000-2003 to 63.13% in 2010-2016 and 19.89% to 17.87%, respectively, and the number of Hispanic, non-Hispanic Asian/ Pacific Islander, and non-Hispanic American Indians/ Alaskan Natives increasing from 9.63% to 11.91%, 5.22% to 5.78%, and 0.43% to 0.60% during that same time period, respectively (p < 0.001). In addition, a stage migration was observed between eras 2 and 3 with a significant shift from M1c to M1b and M1a (p < 0.001). Breakdown of metastatic stage and serum prostate-specific antigen levels was not available for era 1. Notably, there was no significant change in the utilization of local treatment (p = 0.376).
The differences in OS across the three time periods were statistically significant. Median OS was 24 months in both eras 1 and 2, while OS increased by 4 months to 28 months in era 3 (p < 0.0001). Kaplan-Meier curve for 3-year OS is shown in Figure 1 (p < 0.0001, log-rank test). Using era 1 as the referent in Cox proportional hazard model (

| Prostate cancer-specific survival (PCSS) of patients with metastatic prostate cancer also shows modest improvements between 2000 and 2016
As with OS, differences in cause-specific survivals across the three time periods also were statistically significant ( Table 1). The Kaplan-Meier curve for 3-year PCSS is shown in Figure 2 (p < 0.0001, log-rank test). The Cox proportional hazard model demonstrated that PCSS also did not change from era 1 to era 2 (

M1b and M1c patients
While the stage is not provided in era 1, the data were available for 12492 and 18819 men in era 2 and 3, respectively (Table 4). When 3-year OS was stratified by stage between era 2 and 3, we found that patients with M1a prostate cancer did not benefit ( Figure 3, p = 0.4051, log-rank test). However, men with M1b and M1c demonstrated improvements in 3-year OS ( Figure 3, p < 0.0001 and p = 0.0017, respectively, log-rank test). Patients with M1b prostate cancer saw an increase in median OS from 26 months in era 2 to 29 months in era 3, while those with M1c prostate cancer had a more modest increase in median OS from 18 to 20 months across the same time periods (Table 5) (Table 5).

| DISCUSSION
In the present study, we have investigated survival in the SAT era in men diagnosed with metastatic prostate cancer using the SEER database. Our study corroborated the findings of a previous report that found no improvements in OS and PCSS in patients with metastatic prostate cancer between 2000-2003 and 2004-2009 despite the approval of docetaxel in 2004. 11 We did find, however, a modest but statistically significant increase in OS and PCSS of Overall, our observations suggest that there has been only a marginal improvement in treating metastatic prostate cancer in the modern SAT era.
Although the first agent shown to be effective in treating metastatic CRPC was docetaxel in 2004, 6 no survival change was noted in this study until the 2010-2016 era. These findings suggest that the approval of additional drugs since 2010 may have contributed to an improved survival not found between 2000-2003 and 2004-2009. For example, first approved in 2010, cabazitaxel was shown to improve the median survival from 12.7 months in patients given the standard mitoxantrone to 15.1 months in patients given cabazitaxel. 15 More importantly, treatment with cabazitaxel plus prednisone improved overall survival in patients whose disease had progressed during or after docetaxel-based therapy. Likewise, additional It should be noted that there are alternative explanations for our observation concerning the recent improvement in survival among men diagnosed with metastatic prostate cancer. For example, it is entirely possible that the increased survival in the modern era for M1 prostate cancer patients may be due to increased acceptance and utilization of docetaxel since its approval in 2004. In its phase 3 study, TAX 327, the median survival for patients with metastatic prostate cancer improved from 16.5 months in the mitoxantrone group to 18.9 months in patients given docetaxel every 3 weeks and 17.4 months in patients given weekly docetaxel. 6 However, given the relatively rapid progression of M1 disease, such delayed benefit of docetaxel is unlikely. A more probable explanation is that docetaxel may not have been embraced readily by treating physicians and patients. Indeed, a recent study has suggested that nearly half of men diagnosed with metastatic prostate cancer are not treated or treated with androgen deprivation therapy only. 20 Despite numerous advances in treatment including second-line anti-androgens (abiraterone and enzalutamide) between 2010 and 2016, it is noteworthy that median OS improved by mere 4 months. Given that such gain is over two decades, the magnitude of the improvement in outcomes is surprisingly small. This finding may be explained in part by previous studies, which report shared resistance between drugs such as abiraterone and enzalutamide. Possible mechanisms include the presence of androgen-receptor splice variant 7 messenger RNA (AR-V7) and induction of glucocorticoid receptor expression. 21,22 Another study found that patients receiving abiraterone before docetaxel were less likely to respond to docetaxel and achieve a PSA response than patients who had not received abiraterone, suggesting cross-resistance between the two drugs. 23 Collectively, these published data suggest that the effect of these drugs in prolonging OS and PCSS among patients with metastatic prostate cancer is not additive. In this regard, more intense initial therapies have been endorsed recently, 12,13 and new second-line antiandrogens and PARP inhibitors are now available. [8][9][10] As such, analysis in 3-5 years is necessary to assess whether these newer agents are additive or have overlapping resistance mechanisms with the current standard of care. Additionally, previous studies have reported that the U.S. Preventive Services Task Force's recommendation against PSA screening in 2012 may have contributed to a 25% increase in the rate of newly diagnosed metastatic prostate cancer from 2004 to 2014. 24 This increase may have worsened some patients' staging at presentation and further curtailed the survival improvement in mPCa during the SAT era. Notwithstanding, more investigations should be focused on identifying and targeting novel pathways that are outside the conventional androgen signaling pathways to make a major improvement in outcomes in men with metastatic prostate cancer.
When factors associated with improved survival were assessed, local therapy status was associated with the best outcome (adjusted HR 0.3741, 95% CI 0.3277-0.4272, p < 0.0001). This observation is consistent with previous reports that confirmed the association between local therapy (surgery or radiation) and longer survival. 25 More recently, subgroup analyses of prospective studies suggested the benefit of local radiotherapy in men with a low volume metastatic prostate cancer. 26,27 Currently, large-scale prospective studies such as SIMCAP (NCT03456843) and SWOG 1802 (NCT03678025) are underway to further assess the role of surgery and/or radiation in men who present with M1 prostate cancer. 28 Notably, among racial/ethnic groups, Non-Hispanic Black patients with metastatic prostate cancer had decreased OS and PCCS compared to Non-Hispanic White patients (adjusted HR 1.0651, 95% CI 1.0288-1.1027, p = 0.0004 and adjusted HR 1.0652, 95% CI 1.0287-1.1029, p = 0.0004, respectively). Given the recent finding that the survival disparity between White and Black patients was no longer observed after the U.S. Preventive Services Task Force recommended against PSA screening in May 2012, 29 the current observation suggests that there may be different rates of treatment among Black patients relative to White patients with M1 prostate cancer. Indeed, it has been shown that after adjusting for risk factors such as age, PSA, and sites of metastases, Black patients with metastatic castration-resistant prostate cancer who were treated in phase III clinical trials with docetaxel and prednisone (DP) or treatments including DP actually reported increased overall survival relative to White patients. 30 These observations collectively suggest that fewer Black patients with metastatic prostate cancer may be receiving effective treatment regimens compared to White patients. Further analysis is underway to test this hypothesis.
Further stratification of the data by M stage found that the median OS and PCSS increased only in patients with M1b or M1c prostate cancer. Such observation may be a statistical artifact as the sample size of M1a was relatively small (627 in era 2 and 1165 in era 3). Alternatively, a more detailed analysis of additional databases combined with basic laboratory studies is necessary to test this concept.
The strength of our current investigation is that the real-world data with a very large sample size have been analyzed. While previous studies have examined survival improvements in OS and PCSS of mPCa patients over the last few decades, none have conducted survival analyses stratified by M stage and reported a survival benefit limited to patients with bone and visceral metastasis to our knowledge. Notwithstanding, the limitation is that the study is inherently retrospective. Another limitation is that we do not know what if any second-line treatments these men received. Indeed, many men in this cohort may not have access to these typically expensive medications. Finally, the adoption of new treatments cannot be performed using SEER only. As such, our study should be considered a hypothesis-generating study.

| CONCLUSIONS
Since 2000-2003, there has been a modest improvement in survival in patients who are diagnosed with metastatic prostate cancer, with increases in median OS and PCSS of 4 months, respectively. Interestingly, the benefit in survival outcome was observed only in M1b and M1c prostate cancer patients.

ACKNOWLEDGMENTS
We thank the National Cancer Institute for the use of their Surveillance, Epidemiology, and End Results Program (SEER) database.