Statin use and survival in patients with metastatic castration-resistant prostate cancer treated with abiraterone or enzalutamide after docetaxel failure: the international retrospective observational STABEN study

Background Statins may potentiate the effects of anti-hormonal agents for metastatic castration-resistant prostate cancer (mCRPC) through further disruption of essential steroidogenic processes. We investigated the effects of statin use on clinical outcomes in patients with mCRPC receiving abiraterone or enzalutamide. Materials and methods This was a retrospective multicenter study including patients that received abiraterone or enzalutamide for mCRPC. The effect of concurrent statin use on outcomes was evaluated. The associations of statins with early (≤12 weeks) prostate-specific antigen (PSA) declines (> 30%), cancer-specific survival and overall survival (OS) were evaluated after controlling for known prognostic factors. Results Five hundred and ninety-eight patients treated with second-line abiraterone or enzalutamide after docetaxel for mCRPC were included. A total of 199 men (33.3%) received statins during abiraterone/enzalutamide treatment. Median OS was 20.8 months (95% CI = 18.3–23.2) for patients who received statins, versus 12.9 months (95% CI = 11.4–14.6) for patients who did not receive statins (P < 0.001). After adjusting for age, alkaline phosphatase, PSA, neutrophil-to-lymphocytes ratio, Charlson comorbidity score, Gleason score, visceral disease, hemoglobin, opiate use and abiraterone versus enzalutamide treatment, the use of statin therapy was associated with a 53% reduction in the overall risk of death (hazard ratio [HR] = 0.47; 95% CI = 0.35–0.63; P < 0.001). Statin use was also associated with a 63% increased odds of a > 30% PSA decline within the first 12 weeks of treatment (OR = 1.63; 95% CI = 1.03–2.60; P = 0.039). Conclusions In this retrospective cohort, statin use was significantly associated with both prolonged OS and cancer-specific survival and increased early > 30% PSA declines. Prospective validation is warranted.


INTRODUCTION
In developed countries, prostate cancer is the most prevalent malignancy in men, with 142,000 patients dying each year, and an 8.8% cumulative lifetime incidence [1]. Statins are a therapeutic class of medications that are commonly prescribed to lower circulating cholesterol levels through inhibition of 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase [2], and have an established role in primary and secondary cardiovascular prevention [3]. Over the past decade, a preponderance of evidence from numerous studies, mostly conducted in patients with hormonesensitive disease, has shown that statin use in prostate cancer patients is associated with longer cancer-specific and overall survival (OS) [4]. The putative mechanism for this observed improvement in survival is that statins may impair prostate cancer growth via multiple cholesterol-and non-cholesterol-mediated effects [4]. In a recently published study of a large, registry-based cohort, which included >30,000 prostate cancer patients [5], statin use was predictive of improved cancer-specific and OS, after adjusting for stage, Gleason score and primary treatment at diagnosis. Conversely, there is little evidence regarding the effects of statins among patients with castration-resistant prostate cancer (CRPC), and the potential synergism with active systemic treatments (e.g., abiraterone and enzalutamide).
Abiraterone works by inhibiting residual adrenal and intra-tumoral androgen synthesis via CYP17A blockade [6], while enzalutamide acts by inhibiting binding of testosterone to the androgen receptor (AR) as well as by blocking androgen-mediated change and nuclear translocation of AR [7]. In one small retrospective study, statin use was significantly associated with longer OS and early PSA declines in men who received abiraterone [8]. In contrast, this OS advantage has not been consistently observed in other studies [9,10]. Furthermore, there is prospective evidence from a phase III trial suggesting that statins may be discontinued in the palliative care setting with no detrimental effect on survival [11].
In view of the potential additive effect of statins with novel hormonal agents and of the unknown value of continuing versus discontinuing statin therapy in patients with metastatic CRPC (mCRPC), a multi-center retrospective study was conducted to further explore the effects of statin use on PSA response and survival outcomes during second-line (post-docetaxel) treatment with abiraterone or enzalutamide, after adjusting for multiple known predictive factors in the second-line setting [12].

Patients' characteristics and outcomes
Six hundred and forty-two patients were initially included in this dataset. Of these, 44 patients were excluded because statin use could not be ascertained. Baseline characteristics and outcomes are presented for the remaining 598 patients in Table 1A-1D. Notably, > 50% of patients came from one treatment center (BCCA) and an additional 21% of patients came from a second center (Federico II Napoli). Median age of the population was 72 years (range, 42-96). Most of the study patients received abiraterone. Median duration of second-line treatment with abiraterone or enzalutamide was 8.3 months (range, 0.4-47. 5

Statin use
Approximately one-third of the evaluable study population (199 of 598 patients) received statins during treatment, with 107 patients receiving atorvastatin (18% of patients). Importantly, statin use was documented by the local investigator using prescription data in almost 91% of cases. Only eleven patients were reported to have started statin after abiraterone or enzalutamide or to have interrupted statins before suspending abiraterone/enzalutamide treatment (2% of

Association of statins with OS and cancer-related survival
Median OS was significantly improved for mCPRC patients who received concomitant statins, when compared to patients not treated with statins (20.8 versus 12.9 months; hazard ratio [HR] = 0.57, 95% CI = 0.46-0.71, P < 0.001) ( Figure 1). Table 2A summarizes the results of univariable and multivariable models for OS. In the multivariable model, statin use remained strongly associated with OS with a 53% reduction in the risk of death. This association was similar in subgroup analyses and in the landmark analyses. Among the study patients who had died (n = 513), over 91% of the deaths were attributable to prostate cancer, and thus   (Table 2B).
No statistically significant treatment effects were observed between enzalutamide versus abiraterone, nor were treatment differences observed based on type (atorvastatin versus other) or dose of statin.

Association of statin use and cardiovascular or cerebrovascular events
Thirty-three study patients experienced a cardiovascular or cerebrovascular event during the time period analyzed. Timing of events was not consistently reported, and therefore time-to-event analyses could not be performed. Among the 199 patients prescribed statins, 12 (6.0%) experienced a cardiovascular event, and 10 (5.0%) experienced a cerebrovascular event. In contrast, among the 399 patients not prescribed statin therapy, 8 (2.0%) experienced a cardiovascular event, and 3 (0.8%) experienced a cerebrovascular event. After adjusting for other factors in a multivariable model, concomitant statin use remained a significant predictive factor of increased risk of cardiovascular or cerebrovascular events (OR = 3.24, 95% CI = 1.15-9.17, p-value = 0.03) ( Table 4).

DISCUSSION
Although statin use has been associated with reduced cancer-related mortality in a variety of malignancies [13], the potential synergism of statins with anti-cancer medications has been prospectively investigated only in a few clinical trials. Data from the recently published phase III double-blind, placebo-controlled LUNGSTAR trial failed to detect an OS or progression-free survival (PFS) benefit when pravastatin was added to first-line standard chemotherapy in patients with small-cell lung cancer [14]. Similarly, no benefit in overall survival associated with the use of statins added to chemotherapy was reported in two additional phase III trials conducted in advanced gastric [15] and colorectal [16] cancer patients, respectively.   Biologically, statins can potentiate the efficacy of anti-androgen treatments, such as abiraterone and enzalutamide, in mCRPC through a number of potential mechanisms, including: inhibition of intra-tumoral de novo steroid biosynthesis [17], inhibition of biosynthesis of isoprenoids [18], as well as inhibition of the organic anionic transporters (e.g., SLCO2B1) [19] that are responsible for adrenal androgen dehydroepiandrosterone (DHEA) influx into cancer cells [20].
In one translational study, Harshman et al. [21] showed that statins impaired DHEA influx through competitive inhibition of the SLCO2B1 transporter both in both androgen-dependent (LNCaP) and partially androgen-dependent (22RV1) prostate cancer cell lines. This was supported by their retrospective clinical study of 926 patients, treated with androgen deprivation, which demonstrated that patients who received statin therapy experienced longer median time to progression, when compared to patients not treated with a statin (27.5 versus 17.4 months; P < 0.001). Because abiraterone is also a SLCO2B1 substrate, the same research group [10] hypothesized that statin use could be a negative predictive factor for patients taking abiraterone. However, their retrospective study of 224 abiraterone-treated patients demonstrated that statin use trended toward longer treatment duration (14.2 versus 9.2 months; HR: 0.79, 95% CI, 0.57-1.09, P = 0.14). Despite lack of validation in an independent cohort of 270 abiraterone-treated patients [10], the authors concluded that concomitant stain use did not negatively impact survival. In our previous retrospective observational study (n = 187 mCRPC patients from 10 participating centers who received abiraterone), statin use was associated with longer OS in univariate (HR = 0.51, 95% CI = 0.37-0.72, P < 0.001) and multivariate analyses (HR = 0.40, 95% CI = 0.27-0.59, P < 0.001). Statin use was also significantly associated with early PSA declines (>50% declines at week 12 in statin users versus non-users: 72.1% vs. 38.5; P < 0.001). This study was limited by several factors, including the relatively small sample size, the lack of information about statin type and statin treatment duration, comorbidities, cardiovascular events, and prostate cancer-specific survival. To overcome these limitations, we designed a retrospective observational study to be conducted in an international setting that could better define concomitant treatment with statins. One of the purposes of the STABEN trial was to assess whether the potential advantage associated with statin use could be related to their known cardiovascular and cerebrovascular protective effect, of particular potential importance in an elderly population receiving abiraterone -an agent with known cardiovascular toxicity [22]. In the present retrospective study, multivariable models that included known prognostic factors in prostate cancer (e.g., baseline PSA levels, hemoglobin levels, Gleason score, alkaline phosphatase and LDH levels [23], visceral involvement [24] and neutrophil to lymphocyte ratio [25]) revealed that statin use was associated with a 53% reduction in the risk of all-cause mortality, and a 57% reduction in the risk of prostate cancer-specific mortality. It also appeared that statin co-administration increased the odds of having an early >30% PSA decrease, which is consistent with our previously reported findings and adds strength to the hypothesis of a potential synergism with abiraterone/ enzalutamide. Notably, the positive effect of statins on survival did not appear to be influenced by the known protective statin effect against vascular events. While the observed incidence of cardiovascular events reported in this study are consistent with previously reported rates of grade 2 or higher abiraterone-associated cardiovascular events [22], mCRPC patients from this study who were prescribed statin therapy appeared to be at an approximately 4-fold greater risk of experiencing a vascular event. Although such analyses did not account for time-toevent, competing risks, or a history of pre-existing cardiac conditions, this finding could be explained by the observation that patients prescribed statins often present with a greater number of co-morbidities and therefore a greater cardio-and cerebro-vascular risk, when compared to non-statin users [26]. Furthermore, the protective effect of statins was maintained after correcting for Charlson comorbidity index. Although statin consumption was modeled by using a binary variable, it must be noted that only a few patients were not prescribed statins throughout the entire abiraterone or enzalutamide treatment duration, which does not make useful to model statin exposure as a time-dependent variable. Finally, the novelty of the STABEN study also relies in the increased survival in mCRPC patients receiving concomitant enzalutamide and statins vs. enzalutamide alone, which is consistent with the multiple putative pharmacodynamic interactions of statins with anti-androgen receptor agents.
Despite its larger sample size versus published series [8][9][10], this study still suffers from the limitations that apply to retrospective studies, including the lack of data on some key factors such as LDH and time from castration-sensitive disease, as well as the non-systematic selection of participating centers.

CONCLUSIONS
In the large retrospective, observational STABEN study, we found a positive association of statin use with overall-and cancer specific-survival in patients receiving abiraterone or enzalutamide in the secondline setting after docetaxel failure. Statin use was documented by using high-quality prescription data in most patients. The positive association found in our patient cohort with survival was reported both in abiraterone-and enzalutamide-treated men and was consistent with early >30% PSA declines. Analyzed together with previous epidemiology and biological findings, the STABEN results may serve as the basis to design prospective clinical trials assessing the value of adding statins to abiraterone or enzalutamide in mCRPC patients. Optimizing statin use in patients with advanced prostate cancer represents a compelling clinical opportunity to improve survival via the addition of a safe and inexpensive drug.

Inclusion criteria
Medical records were reviewed at eight participating centers for patients with diagnosed mCRPC who were treated with second-line abiraterone or enzalutamide between January 2011 and January 2016. Histologicallyconfirmed prostate cancer and previous docetaxel-based treatment were required for inclusion in this study. Castration-resistance was determined per Prostate Cancer Clinical Trials Working Group 2 (PCWG2) criteria [22]. Patients who received at least one 28-day cycle of abiraterone or enzalutamide in the second-line setting were regarded as eligible for this study. Patient data including medical and prostate cancer history, demographic, and baseline characteristics were retrieved starting at the time of abiraterone or enzalutamide initiation. Data collected regarding statin use included: type and dose of statin prescribed, source of the data (claims versus prescription data), and dates of statin use initiation and discontinuation.

Data analysis
Summary statistics were used to describe patient outcomes. Time-to-event outcomes were calculated from the first date of treatment with abiraterone or enzalutamide.
The primary objective of this study was to determine whether concomitant statin therapy was predictive of OS improvement for mCRPC patients treated with secondline abiraterone or enzalutamide. The secondary objective of the study was to determine whether concomitant statin therapy was predictive of early (≤ 12 weeks) >30% PSA declines. The Kaplan-Meier method was used to estimate differences in survival between mCRPC patients treated who did and did not receive statin therapy, while Cox proportional hazards regression was used to investigate prognostic factors of overall survival. Logistic regression was used to investigate predictive factors of early >30% PSA declines. Using Cox proportional hazards, multivariable models were constructed to examine the effects of concomitant statins after adjusting for all other potential sources of variation. However, there were large numbers of missing data for some factors. Thus, a priori, it was decided to include only those factors which had <30% missing data and were significant on univariate analysis, or those factors with <15% missing data overall. The impact of statins was then assessed after adjusting for factors included in the multivariable model. Supportive analyses were performed by including only those treated with abiraterone (~80% of the cohort), only those treated with enzalutamide, by performing a cancer-specific survival analysis and by performing a landmark analysis using 3-months as the landmark time. For the purposes of the landmark analysis, any patient who was not prescribed statin therapy at the time of abiraterone or enzalutamide initiation, experienced interruption of statin therapy, or received less than 3 months of statin therapy, was deemed to not have received statins. Data modifications were performed for statistical purposes. Specifically, a logarithmic transformation was used on covariates which were highly non-normal. Duration from prostate cancer diagnosis to detection of metastases, and duration of prostate cancer diagnosis to determination of castrationresistant disease were dichotomized. All analyses included site as a stratification factor. All tests were two-sided and a p-value of 0.05 or less was considered statistically significant. No p-value adjustments were performed due to multiple testing; however, inferences were performed understanding that multiple analyses were performed.