A Multicohort Open-label Phase II Trial of Bipolar Androgen Therapy in Men with Metastatic Castration-resistant Prostate Cancer (RESTORE): A Comparison of Post-abiraterone Versus Post-enzalutamide Cohorts

Abstract

Background:

Cyclic high-dose testosterone injections, also known as bipolar androgen therapy (BAT), is a novel treatment strategy for patients with metastatic castration-resistant prostate cancer (mCRPC). BAT has shown clinical activity in prior studies enrolling men with mCRPC and may potentially restore sensitivity to prior androgen receptor (AR)-targeted agents.

Objective:

To evaluate the clinical activity of BAT in patients progressing on AR-targeted therapy as well as responses to abiraterone or enzalutamide upon rechallenge after BAT.

Design, setting, and participants:

RESTORE is a multicohort phase II study enrolling asymptomatic mCRPC patients after abiraterone or enzalutamide at Johns Hopkins Hospital (NCT02090114). Participants (29 after abiraterone and 30 after enzalutamide) received 400 mg testosterone cypionate intramuscularly every 28 days, with ongoing luteinizing hormone-releasing hormone agonist/antagonist treatment (ie, BAT). Following progression on BAT, patients were rechallenged with their most recent AR-targeted therapy.

Outcome measurements and statistical analysis:

Coprimary endpoints were >50% decline in PSA from baseline (PSA₅₀) responses to BAT and following AR-targeted therapy rechallenge. Outcomes in the post-abiraterone cohort are presented, as well as updated results from the post-enzalutamide cohort and an exploratory AR-V7 analysis.

Results and limitations:

No statistically significant difference in PSA₅₀ response rates to BAT was observed (30% [post-enzalutamide cohort] vs 17% [post-abiraterone cohort], p = 0.4). However, PSA₅₀ responses to AR-targeted therapy rechallenge were higher in the post-enzalutamide cohort (68% vs 16%, p = 0.001). The median time from enrollment to progression following rechallenge with AR-targeted therapy (ie, progression-free survival 2; PFS2) was longer in the post-enzalutamide versus post-abiraterone patients (12.8 vs 8.1 mo, p = 0.04). Outcomes were worse in patients with detectable AR-V7 in circulating tumor cells (median PFS2: 10.3 vs 7.1 mo, p = 0.005).

Conclusions:

BAT shows clinical activity in mCRPC patients and may be more effective at resensitizing to enzalutamide versus abiraterone.

Patient summary:

BAT is well tolerated in metastatic castration-resistant prostate cancer patients. The type of prior AR-targeted therapy might affect response to BAT as well as AR-therapy rechallenge. BAT followed by AR-targeted therapy rechallenge did not improve outcomes in AR-V7–positive patients.

1. Introduction

In preclinical models, paradoxical inhibition of prostate cancer cell growth has been observed following treatment with high concentrations of androgens [1,2]. Supraphysiologic levels of testosterone have been shown to inhibit cell cycle progression, induce double-strand DNA breaks, promote genomic rearrangements, and downregulate androgen receptor (AR) splice variants (eg, AR-V7) [3–9]. Several clinical trials in metastatic castration-resistant prostate cancer (mCRPC) patients investigated the use of bipolar androgen therapy (BAT), in which serum testosterone levels are rapidly driven to a supraphysiologic range followed by a return to near-castrate levels over a 28-d cycle. Several studies have confirmed the clinical benefit of BAT along with a favorable safety profile in men with mCRPC [10–12].

The RESTORE trial enrolled men with mCRPC who have most recently progressed on enzalutamide (post-enzalutamide cohort) or abiraterone acetate (post-abiraterone cohort) as their last therapy. All patients were initially treated with BAT as a single agent. At the time of clinical or radiographic progression on BAT, patients were rechallenged with the AR-targeted agent that they had received most recently. We previously reported the preliminary results from the post-enzalutamide cohort [11]. Here, we report data from the post-abiraterone cohort for the first time. In addition, we compare the efficacy of BAT and AR-targeted therapy rechallenge across the post-enzalutamide and post-abiraterone cohorts.

2. Patients and methods

2.1. Study cohort, design, and outcome measures

RESTORE is an Institutional Review Board–approved, multicohort, open-label phase II trial that was conducted at Johns Hopkins Hospital in Baltimore, MD, USA. Eligible patients were 18 yr or older with mCRPC defined according to Prostate Cancer Working Group 2 (PCWG2) guidelines. For the abiraterone cohort, all patients had prostate-specific antigen (PSA) and/or radiographic progression on abiraterone as their last treatment prior to enrollment. Up to two lines of novel AR-targeted therapy were allowed. No prior chemotherapy for mCRPC was permitted (taxane-based chemotherapy for metastatic hormone-sensitive prostate cancer was allowed). Serum testosterone levels ≤50 ng/dl and adequate bone marrow, renal, and liver function were required at the time of screening. Patients were excluded if they had five or more sites of visceral disease, were on opioid analgesics due to tumor-related pain, or at risk for urinary obstruction or cord compression.

All patients received intramuscular injections of 400 mg testosterone cypionate on day 1 of a 28-d cycle. Patients were maintained on a luteinizing hormone-releasing hormone agonist/antagonist for the duration of the study in order to suppress endogenous testosterone production. During rechallenge, abiraterone was administered orally at 1000 mg daily with 5 mg of prednisone twice per day. The trial schematic is provided in Supplementary Fig. 1. Patients were evaluated with safety laboratory values including PSA every 4 wk. Restaging imaging was obtained every 12 wk. BAT was discontinued for any of the following reasons: (1) radiographic progression, (2) ≤25% rise in PSA from baseline for patients who did not have a decrease in PSA below baseline during BAT, or (3) as per the treating physician’s discretion for patients who experienced a decrease in PSA from baseline. A 28-d washout period from the last testosterone injection was required before rechallenging with abiraterone. For patients rechallenged with abiraterone, treatment was discontinued for radiographic, PSA, or clinical progression. Methods describing the post-enzalutamide cohort have previously been published [11]. PSA progression was defined by PCWG2 criteria [13]. Clinical or radiographic progression was defined by RECIST 1.1 (soft tissue lesions) [14] and PCWG2 (clinical and bone lesions). Objective responses for patients with measurable disease were defined using RECIST 1.1 criteria and assessed in an unblinded fashion by the investigators.

Peripheral blood samples for AR-V7 testing were collected prior to the start of BAT, after three cycles of BAT, and after three cycles of AR-targeted therapy rechallenge. Circulating tumor cells (CTCs) were isolated using the Qiagen Adnatest protocol (Germantown, MD, USA). CTC positivity, AR full length, and AR-V7 copy number were determined using reverse transcriptase polymerase chain reaction as previously described [15].

The coprimary endpoints for this study were the >50% decline in PSA from baseline (PSA₅₀) response to BAT and following abiraterone rechallenge. Secondary endpoints of the study included PSA progression-free survival (PFS) on BAT and abiraterone (time from initiation of BAT or abiraterone to the time of measured increase in PSA of at least 2 ng/dl and 25% from nadir values, confirmed with a second measurement at least 4 wk later), investigator-assessed radiographic progression-free survival (rPFS) on BAT (time from initiation of BAT to the time of radiographic progression defined by PCWG2 and RECIST1.1), investigator-assessed clinical or radiographic progression-free survival (crPFS) on BAT and abiraterone (time from initiation of BAT or abiraterone to the time of unequivocal clinical progression or radiographic progression), and safety/tolerability of BAT. AR-V7 analyses were exploratory as were comparative analyses across the abiraterone and enzalutamide cohorts.

2.2. Statistical analysis

The null hypothesis for each coprimary endpoint was a PSA₅₀ response rate of 5% for the post-abiraterone cohort. The alternative hypothesis was a PSA₅₀ response rate of 20% to BAT and 25% to abiraterone rechallenge. Enrollment of 30 patients was planned, but capped at 29 due to slow accrual. Thirty evaluable patients provided 83% and 90% power to reject the null hypothesis for BAT and abiraterone rechallenge, respectively. The overall type I error was 0.1, allocated equally between coprimary endpoints (0.05). The same calculation was used for the post-enzalutamide cohort [11].

A Kruskal-Wallis test was used to compare the distribution of age and PSA across cohorts, while a chi-square test was used to compare Gleason, race, prior therapy, and AR-V7 status. PSA₅₀ response rates to BAT and abiraterone rechallenge were compared with the null hypothesis of 5% using a one-sided exact binomial test, and the response rates were compared between the two cohorts using a Fisher’s exact test. Unadjusted Kaplan-Meier survival curves and log-rank tests were used to compare PFS (ie, crPFS and PFS2) between the post-enzalutamide and post-abiraterone cohorts. PFS2 was defined as the time from enrollment to crPFS following rechallenge with the AR-targeted therapy. Patients who had clinical or radiographic progression on BAT and not rechallenged with abiraterone/enzalutamide were considered to have had an event (ie, not censored). All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA), and statistical significance was set at p < 0.05.

3. Results

Between 10/6/2014 and 1/8/2018, 29 mCRPC patients were enrolled in the post-abiraterone cohort. Demographic, clinical, and pathologic characteristics are shown in Table 1. A comparison with the post-enzalutamide cohort is also provided in Table 1. No statistically significant differences in patient characteristics were observed. One patient in the post-abiraterone cohort received chemotherapy for metastatic hormone-sensitive prostate cancer. One patient in each cohort withdrew consent after one cycle of BAT.

Table 1 –

Baseline patient demographic and clinical characteristics by prior novel AR-targeted agent.

Baseline characteristics	Post-abiraterone (n = 29)	Post-enzalutamide (n = 30)	p value
Age (yr)			0.3
Median (range)	71 (49–85)	74 (50–89)
Race, N (%)			>0.9
White	26 (90)	27 (90)
Nonwhite	3 (10)	3 (10)
Gleason sum at diagnosis, N (%)			0.28
≤7	13 (45)	9 (30)
≥8	16 (55)	20 (67)
Unknown	0 (0)	1 (3)
Baseline PSA (ng/mL)			0.25
Median (range)	27.7 (2.8–367.9)	39.8 (3.4–245.3)
Prior use of other AR-targeted druga, N (%)			0.14
Yes	9 (31)	15 (50)
No	20 (69)	15 (50)
AR-V7 + circulating tumor cells, N (%)	N = 23	N = 29	0.7
Yes	5 (22)	5 (17)
No	18 (78)	24 (83)

AR = androgen receptor; N = number of patients; PSA = prostate-specific antigen.

^aPrior abiraterone for post-enzalutamide patients and prior enzalutamide for post-abiraterone patients.

The median number of BAT treatments in the post-abiraterone cohort was 5.0 (range 1–25). The PSA₅₀ response rate to BAT in the post-abiraterone cohort was 17% (N = 5/29, 95% confidence interval [CI]: 5.8–36%, p = 0.014) and 11 of 29 patients (38%) achieved some degree of PSA reduction (Fig. 1A). Nineteen of 29 mCRPC patients (66%) were rechallenged with abiraterone following BAT, resulting in a PSA₅₀ response rate of 16% (N = 3/19, 95% CI: 3.8–40%, p = 0.067; Fig. 1B). In patients with measurable disease, BAT induced an objective response rate in 29% (N = 2/7) of patients (Fig. 1C). The median rPFS and crPFS on BAT were estimated to be 5.0 (95% CI: 3.3–9.3) and 4.3 (95% CI: 3.3–5.3) mo, respectively (Fig. 1D and 1E). In patients who were rechallenged (N = 19), the median crPFS on abiraterone was 4.0 mo (95% CI: 3.0–4.0 mo; Fig. 1F). At the time of analysis, one patient remained on BAT for 25 mo and another patient withdrew consent after 4 wk on BAT. All patients who received abiraterone had progressed and were off study.

Fig. 1 –

Clinical benefit of BAT and abiraterone rechallenge in the post-abiraterone cohort. (A) Waterfall plot of best PSA response to BAT and (B) abiraterone rechallenge following BAT. A >50% decline in PSA from baseline (PSA₅₀) from baseline is shown via the dotted line. (C) Best percentage change in target lesion sum to BAT is shown for patients with measureable disease. Two patients achieved an objective response as denoted by a dotted line. (D) Radiographic progression-free survival (rPFS) on BAT is shown as a Kaplan-Meier curve (solid line) bracketed by a dotted line (95% confidence interval [CI]): median rPFS: 5.0 mo (95% CI: 3.3–9.3 mo). Kaplan-Meier curves demonstrated clinical or radiographic progression-free survival (crPFS) for (E) BAT (median crPFS: 4.3 mo [95% CI: 3.3–5.3 mo]) and (F) abiraterone rechallenge following BAT (median crPFS: 4.0 mo [95% CI: 3.0–4.0 mo]). BAT = bipolar androgen therapy; ORR = objective response rate; PFS = progression-free survival; PSA = prostate-specific antigen. * Values truncated at 100% (see Supplementary Table 1 for additional details).

We subsequently conducted a post hoc analysis comparing the post-abiraterone versus post-enzalutamide cohorts. BAT treatment resulted in no statistically significant difference in PSA₅₀ (post-enzalutamide vs post-abiraterone cohorts: 30% vs 17%, p = 0.4) or objective (50% vs 29%, p = 0.4) response rates between cohorts (Table 2). Rechallenge with enzalutamide was associated with a higher PSA₅₀ response rate than abiraterone rechallenge (68% vs 16%, p = 0.001).

Table 2 –

PSA₅₀ and objective response rates in post-abiraterone and post-enzalutamide mCRPC patients treated with BAT.

	Post-abiraterone (N = 29)	Post-enzalutamide (N = 30)	p value
BAT PSA50 RR	17% (N = 5/29)	30%a (N = 9/30)	0.4
BAT objective RR	29% (N = 2/7)	50%a (N = 6/12)	0.4
Rechallenge PSA50 RR	16% (N = 3/19)	68% (N = 15/22)	0.001

BAT = bipolar androgen therapy; mCRPC = metastatic castration-resistant prostate cancer; N = number of patients; PSA = prostate-specific antigen; PSA50 = > 50% decline in PSA from baseline; RR = response rate.

^aOne (N = 1) complete response to BAT observed in the post-enzalutamide cohort.

For each patient, the best PSA response to BAT was paired with the best response to abiraterone rechallenge in a waterfall plot (Fig. 2A). Only two patients who were primary refractory to BAT (ie, PSA levels did not decrease below baseline) achieved a PSA₅₀ response to abiraterone. In contrast, all patients in the post-enzalutamide cohort who were refractory to BAT and rechallenged with enzalutamide attained a PSA₅₀ response (Fig. 2B). Next, we compared the clinical outcomes of BAT and subsequent rechallenge between the post-abiraterone and post-enzalutamide cohorts using crPFS to assess benefit. No statistically significant difference in median crPFS on BAT was observed between cohorts—6.5 (post-enzalutamide cohort) versus 4.3 mo (post-abiraterone cohort, p = 0.4; Fig. 2C). For the patients who were rechallenged with AR-targeted therapy, enzalutamide retreatment resulted in longer median crPFS than abiraterone (6.0 vs 4.0 mo, p = 0.005; Fig. 2D). Last, we compared the median PFS2 between cohorts (ie, total duration of benefit; Fig. 2E). The cohort that received BAT followed by enzalutamide had longer median PFS2 than the cohort that received BAT followed by abiraterone (12.8 vs 8.1 mo, p = 0.04).

Fig. 2 –

BAT and AR-targeted therapy rechallenge: post hoc comparison of post-enzalutamide and post-abiraterone cohorts. (A) Best PSA response to BAT (blue) and abiraterone rechallenge following BAT (red), paired for each patient in the post-abiraterone cohort. One patient was not evaluable for PSA response and is not included. (B) Best PSA response to BAT (blue) and enzalutamide rechallenge following BAT (orange), paired for each patient in the post-enzalutamide cohort. One patient was not evaluable for PSA response and is not included. PSA₅₀ response cutoff was denoted by the dashed line. Clinical or radiographic PFS (crPFS) on (C) BAT and (D) AR-targeted therapy rechallenge, estimated with Kaplan-Meier curves, in the post-abiraterone (red) and post-enzalutamide (blue) cohorts. (E) Kaplan-Meier estimates of PFS2 in the post-abiraterone (red) and post-enzalutamide (blue) patients. AR = androgen receptor; BAT = bipolar androgen therapy; CI = confidence interval; PFS = progression-free survival; PSA = prostate-specific antigen; PSA₅₀ = >50% decline in PSA from baseline. * Values truncated at 100% (see Supplementary Table 1 for additional details).

We also explored the relationship between AR-V7 status and clinical outcome. In a combined analysis of both cohorts, 42 patients (n = 42/52, 81%) had either no detectable AR-V7 mRNA in CTCs or no CTCs detected in a peripheral blood sample; these cases were designated as AR-V7 negative. CTCs from 10 mCRPC patients had detectable AR-V7 splice variants (N = 10/52, 19%) prior to the start of BAT. No significant difference in the distribution of AR-V7–positive patients was observed between cohorts (Table 1). The PSA₅₀ response rates to BAT were comparable irrespective of the AR-V7 status (AR-V7 negative: 26% vs AR-V7 positive: 20%, p = 1; Table 3). Upon AR-targeted therapy rechallenge, AR-V7–negative patients had a numerically higher (N = 16/32, 50%) PSA₅₀ response rate than AR-V7–positive patients (N = 1/8, 13%, p = 0.11). We also examined the change in AR-V7 status following treatment on BAT and subsequent AR-targeted therapy rechallenge. After 3 mo of BAT, one of 10 patients remained AR-V7 positive (Table 4). When rechallenged with an AR-targeted agent, five of six patients tested reverted back to being AR-V7 positive.

Table 3 –

PSA₅₀ response rates in mCRPC patients treated with BAT and AR-targeted therapy rechallenge stratified by AR-V7 status.

	AR-V7 negative (N = 42)	AR-V7 positive (N = 10)	p value
BAT PSA50 RR	26% (N = 11/42)	20% (N = 2/10)	1
Rechallenge PSA50 RR	50% (N = 16/32)	13% (N = 1/8)	0.11

AR = androgen receptor; BAT = bipolar androgen therapy; mCRPC = metastatic castration-resistant prostate cancer; N = number of patients; PSA = prostate-specific antigen; PSA50 = > 50% decline in PSA from baseline; RR = response rate.

Table 4 –

Change in AR-V7 status during course of treatment for AR-V7–positive cohort.

	Pretreatment	BAT	Rechallenge
AR-V7+	100% (10/10)	10% (1/10)	83% (5/6)
AR-V7−	0% (0/10)	90% (9/10)a	17% (1/6)

AR = androgen receptor; BAT = bipolar androgen therapy. AR-V7–positive patients were followed with serial blood samples during the course of treatment. AR-V7 status was assessed prior to BAT (pretreatment), after three cycles of BAT (BAT), and after three cycles on AR-targeted therapy (rechallenge).

^aTwo of nine patients had no circulating tumor cells.

With respect to clinical outcomes, no statistically significant difference was observed in median crPFS in AR-V7–negative versus AR-V7–positive patients treated with BAT (6.1 vs 4.3 mo, p = 0.13; Fig. 3A). Upon AR-targeted therapy rechallenge, AR-V7–negative patients had improved median crPFS compared with AR-V7–positive patients (4.5 vs 3.0 mo, p = 0.001; Fig. 3B). Similarly, median PFS2 was longer in AR-V7–negative (10.3 mo) than in AR-V7–positive (7.1 mo) patients (p = 0.005; Fig. 3C). When patients without detectable CTCs were excluded from the analysis, the median PFS2 continued to be longer in the AR-V7–negative versus AR-V7–positive cohort (9.0 vs 7.1 mo, p = 0.048; Supplementary Fig. 2).

Fig. 3 –

Effect of AR-V7 status on clinical outcomes of BAT and AR-targeted therapy rechallenge. Clinical or radiographic PFS (crPFS) on (A) BAT and (B) AR-targeted therapy rechallenge (combined abiraterone and enzalutamide patients), estimated with Kaplan-Meier curves, in AR-V7–negative (red) and AR-V7–positive (blue) subgroups. (C) Kaplan-Meier estimates of PFS2 in AR-V7–negative (red) and AR-V7–positive (blue) subgroups. AR = androgen receptor; BAT = bipolar androgen therapy; CI = confidence interval; PFS = progression-free survival.

The adverse events following BAT for the post-enzalutamide cohort have been reported previously [11]. In the post-abiraterone cohort, the two most common adverse events during treatment with BAT were musculoskeletal pain and peripheral edema (Table 5). The majority of adverse events were low grade (ie, Common Terminology Criteria for Adverse Events [CTCAE] v.4.03 grade ≤2). Two patients experienced grade 3 toxicities (musculoskeletal pain and alkaline phosphatase elevation), which resolved with treatment delay. One patient experienced grade 3 malignant bowel obstruction at the time of progression on BAT and came off study. One patient developed grade 4 disseminated intravascular coagulation concurrent with radiographic progression on BAT and did not receive abiraterone rechallenge. No patients were removed from the study for drug toxicity to BAT. There were no treatment-related deaths.

Table 5 –

Adverse events on BAT in mCRPC patients in the post-abiraterone cohort.

	Any grade	Grade 3	Grade 4
MSK pain	8 (27.6%)	1 (3.4%)	0
Edema	8 (27.6%)	0	0
Breast tenderness	5 (17.2%)	0	0
Anorexia	4 (13.8%)	0	0
Headache	3 (10.3%)	0	0
Hot flashes	3 (10.3%)	0	0
Elevated Alk Phos	2 (6.9%)	1 (3.4%)	0
Fatigue	2 (6.9%)	0	0
Insomnia	2 (6.9%)	0	0
Nausea	2 (6.9%)	0	0
Pain at injection site	2 (6.9%)	0	0
Urinary obstruction	2 (6.9%)	0	0
DIC	1 (3.4%)	0	1 (3.4%)
Bowel obstruction	1 (3.4%)	1 (3.4%)	0

Alk Phos = alkaline phosphatase; BAT = bipolar androgen therapy; DIC = disseminated intravascular coagulation; mCRPC = metastatic castration-resistant prostate cancer; MSK = musculoskeletal.

Grade 1–2 events occurring in more than one patient and all grade 3 or 4 events are presented.

4. Discussion

In this study, we showed that BAT could induce PSA₅₀ responses in mCRPC patients who had recently progressed on abiraterone, with an acceptable side effect profile. BAT was also able to restore sensitivity to abiraterone rechallenge in a subset of patients. Based on our statistical plan for the coprimary endpoints, this study met its primary endpoint for PSA₅₀ response rate to BAT but not for PSA₅₀ response rate to abiraterone rechallenge. The reason for the limited PSA₅₀ responses to subsequent abiraterone rechallenge may be due to a lack of statistical power. However, it is also possible that BAT may not be an effective “resensitizer” to abiraterone, and this endpoint would not have been met irrespective of underaccrual.

To better assess the benefit of BAT in a post-enzalutamide versus post-abiraterone patient population, we performed a post hoc analysis between these two cohorts. No statistically significant difference in PSA₅₀ response rate, objective response rate, or median crPFS to BAT was observed between the post-enzalutamide and post-abiraterone cohorts. It is likely that these results are under-powered to make definitive conclusions regarding differences in BAT efficacy based on last prior therapy, since the multicohort design of RESTORE did not account for comparisons across cohorts. Despite the small sample size, enzalutamide rechallenge resulted in a higher PSA₅₀ response rate and longer median crPFS than abiraterone rechallenge following BAT. These findings suggest that the strongest benefit of BAT may be in restoration of sensitivity to AR antagonists rather than as a primary therapy. The mechanism by which post-enzalutamide patients would have a more favorable outcome is unclear and may be related to directly inhibiting AR (rather than indirectly via CYP17 inhibition). Further study is needed to determine the role of prior AR-targeted therapies and taxane chemotherapies.

Finally, we explored the effect of BAT in mCRPC patients with detectable AR-V7 mRNA in CTCs. Of the patients in this study, 19% (N = 10/52) tested positive for the AR-V7 splice variant, which is consistent with prior estimates in this population [16]. Interestingly, we observed that BAT led to the conversion from AR-V7 positive to negative, but rechallenge with AR-targeted therapies still resulted in poor PSA₅₀ response rates followed by restoration of AR-V7 expression. Although the number of AR-V7–positive patients in this cohort is low, the median PFS2 of BAT followed by AR-targeted therapy rechallenge was worse in the AR-V7–positive group. Inferior outcomes in AR-V7–positive patients were also observed when patients with no detectable CTCs were excluded from the analysis. Since BAT suppressed AR-V7 splice variant expression, the poor clinical outcomes in the AR-V7–positive group suggest that the presence of this splice variant is prognostic and not predictive of benefit to BAT or subsequent AR-targeted therapy under these treatment conditions.

This study has several limitations. First, the trial was not designed for comparison across cohorts, resulting in a lack of statistical power. Second, within each cohort, patients may have received one or both AR-targeted agents. Although the number of patients treated with both abiraterone and enzalutamide was not statistically different between cohorts, this may have confounded our findings. Third, the endpoints of crPFS and PFS2 in the RESTORE study are not surrogates for overall survival. Fourth, a number of patients were treated with BAT that did not subsequently receive AR-targeted therapy rechallenge at the discretion of the treating physician and patient. This may have resulted in abiraterone rechallenge not meeting the primary endpoint and may also have affected PFS estimations. Last, this study represents the experience of a single institution and may not reflect the findings in different patient populations. Despite these weaknesses, our findings in the post-abiraterone and post-enzalutamide cohorts merit additional studies of BAT in mCRPC patients.

5. Conclusions

BAT shows clinical activity in mCRPC patients and may be more effective at resensitizing to enzalutamide vs. abiraterone.

This paper was presented earlier at the Prostate Cancer Foundation Scientific Retreat 2019—poster presentation.