Phase I Study of Niraparib in Combination with Radium-223 for the Treatment of Metastatic Castrate-Resistant Prostate Cancer

An estimated 250,000 men were diagnosed with prostate cancer in 2021, which remains the second most common cause of cancer death for men in the United States (1). Prostate cancer cells are dependent on androgen stimulation and the androgen axis for growth and androgen deprivation therapy is the cornerstone of treatment (2). Despite androgen deprivation, progression of disease occurs over time, in part, through mechanisms that re-activate the androgen receptor (AR) independent of androgen stimulation, leading to castrate-resistant prostate cancer (3). There are currently several different therapies for mCRPC that are approved which include: taxane chemotherapy, androgen axis inhibitors such has enzalutamide, apalutamide and abiraterone, sipuleucel-T, Radium-223, and in select populations, PARP inhibitors Olaparib or Rucaparib and Pembrolizumab, an anti-PD1 antibody (4–11).

Radium-223 is an alpha particle-emitting, radioactive agent that is FDA approved for treatment of patients with mCRPC and symptomatic bone metastases (12). In comparison to gamma radiation, alpha particle radiation produces higher rates of single-strand DNA breaks that are more frequently converted to double-strand breaks (13). The agent targets calcium hydroxyapatite in bone and accumulates in regions of osteoblastic activity, emitting alpha particles that have low penetrance depth and high linear energy transfer, making its effect localized to the cortical bone and minimizing bone marrow toxicity (14, 15). This localized effect and minimal systemic toxicity creates the potential for Radium-223 to be combined safely with other agents.

PARP-1 is a nuclear enzyme that recruits proteins that impact the DNA damage repair (DDR) pathways. Inhibitors of PARP-1 have demonstrated clinical activity in mCRPC when concurrent mutations in genes that code for DNA repair proteins are identified (10, 11, 16). PARP may also play an important role in prostate cancer through interaction with the AR (17–20). Niraparib is an orally available, highly selective PARP-1/PARP-2 inhibitor that has shown activity in tumors with DNA repair deficiencies and is currently approved, at a dose of 200 to 300 mg daily, for clinical use in ovarian cancer (21–23). Niraparib has also been found to be effective in treating mCRPC with known DNA repair gene defects (24, 25). The benefit of PARP inhibitors may also extend to patients with mCRPC without deficiencies in DDR pathways, as evidenced by the phase III results from the PROpel study, which demonstrated an improvement in PFS for patients without mutations in homologous recombination repair (HRR) genes (26).

We hypothesize that for patients with mCRPC, regardless of status of DNA repair mutational status, PARP inhibitors may prove effective when given in combination with agents that stimulate DNA damage. Given that Radium-223 creates high rates of DNA single-strand breaks as a mechanism for cell death, and PARP is primarily responsible for repair of such breaks (27) there is the potential for synergy between these agents when treating mCRPC. In addition, the potential for niraparib to further suppress the AR pathway, suggests another potential mechanism for synergy when combined with Radium-223. We believe these interactions could extend benefit to patients without somatic or germline DNA repair mutations.

We conducted a phase IB study to determine the MTD and recommended phase II dosing (RP2D) of niraparib combined with standard dosing of Radium-223 in the treatment of patients with metastatic mCRPC in men with and without prior chemotherapy exposure. Additional correlative studies were conducted to evaluate oncogenic genetic mutations at baseline and gene expression and pathway profile changes induced by therapy.

This was a multicenter, phase IB dose escalation trial (ClinicalTrials.gov ID NCT03076203) that was approved by the institutional review boards at all participating institutions. Patients were provided written informed consent prior to starting on trial and enrollment followed international standards of good clinical practice. There were three dose levels of niraparib evaluated along with standard of care dosing of Radium-223 in two cohorts (chemotherapy naïve and prior chemotherapy) of men with mCRPC. The primary objective was to determine the MTD and RP2D of niraparib combined with Radium-223. The MTD was defined as the highest dose level at which the probability of a dose-limiting toxicity (DLT) was less than 20%.

Because the toxicity of radium-223 is often delayed, a time-to-event continuous reassessment method (TITE-CRM; ref. 28) design was used to identify the MTD based on toxicities observed over 12 weeks of treatment (three cycles of Radium-223). A DLT was defined as any treatment-related Grade ≥3 nonhematologic clinical that required medical intervention, or persisted for ≥7 days; any grade 4 hematologic toxicity with the exception of neutropenia Grade 4 lasting for <7 days and not associated with fever >38.5°C and/or infection; a dose interruption for a non-DLT laboratory abnormality lasting ≥14 days or a dose interruption for nonhematologic AE leading to <80% of an intended dose of niraparib being administered. Patients were followed from time of enrollment until progression of disease or death. The trial was conducted under the auspices of the Prostate Cancer Clinical Trials Consortium.

Eligible patients had histologic or cytologic diagnosis of adenocarcinoma of the prostate without neuroendocrine or small cell features, bone metastases detected by conventional bone scans and evidence of progressive disease after receiving at least 1 line of AR-targeted therapy in the hormone sensitive or castrate-resistant setting. Each patient was required to have serum testosterone levels in the castrate range (less than 50 ng/dL) and this was measured at time of screening. Patients were ineligible if they had: previously received a PARP inhibitor, more than one prior line of AR-targeted therapy or more than one chemotherapy agent. Patients who were previously identified to be carriers of a pathogenic germline or somatic mutation in BRCA-1 or BRCA-2 were excluded; however, additional testing to identify BRCA or other DDR pathway mutations was not conducted prior to enrollment. In addition, patients with bulky visceral disease (defined as >4 cm), brain or leptomeningeal disease or impending spinal cord compression were excluded.

Patients were assigned to one of the three different treatment arms based using TITE-CRM (28) design, which allowed for three different dose levels of niraparib to be tested simultaneously with Radium-223. Subjects were given niraparib at doses of 100, 200, or 300 mg oral daily in combination with standard dosing of Radium-223, 55 kBq per kg body weight, given at 4-week intervals for a total of six injections. Following six cycles, subjects were to continue on niraparib alone until objective progression, intolerance of therapy, or withdrawal of consent. Enrollment was stratified by prior chemotherapy use and we aimed to enroll up to 30 patients total, with no more than 15 patients in each arm.

The TITE-CRM model was specifically chosen to improve the time to completion of the study, given that toxicities to Radium-223 are often delayed by several weeks. In the TITE-CRM, dose levels were assigned to each newly enrolled subject based on DLT data from subjects already enrolled on trial, which were weighted to account for the proportion of the observation period that each enrolled subject had been observed. Dose interruptions and reductions of niraparib were allowed for management of AEs per study protocol. The dose of Radium-223 was not adjusted, both agents were held in the event of a DLT and Radium-223 was not continued as a single agent in the event of niraparib discontinuation. Patients were removed from the study if they had an AE on the lowest dose of niraparib (100 mg) or if the subject required a dose interruption of more than 28 days. The study treatment also stopped for any patient who had clinical disease progression, intercurrent illness that made further treatment unsafe, developed myelodysplastic syndrome (MDS) or acute myeloid leukemia (AML), or withdrew of the consent.

The primary outcome of this phase IB study was determination of the MTD and RP2D of niraparib combined with standard dosing of Radium-223 in the treatment of patients with mCRPC. Secondary endpoints included proportion of subjects with 50% prostate-specific antigen (PSA) reduction at 12 weeks, radiographic progression-free survival (rPFS) at 6 months and long-term safety and tolerability of treatment combination. AEs were defined by the NCI's Common Terminology Criteria for Adverse Events version 4.0. Subjects were re-assessed for any AEs weekly for the first 4 weeks, then bi-weekly for the next 8 weeks and finally monthly until disease progression or death. Any individual who was withdrawn from the study due to an AE was followed until event had stabilized or resolved. Disease status (response or progression) was assessed every 12 weeks and were determined using a combination of the revised RECIST v1.1 criteria and the guideline for prostate cancer endpoints developed by the Prostate Cancer Clinical Trials Working Group 3 (29, 30).

The TITE-CRM statistical design was used, assuming prior probabilities of toxicity for the three doses of 0.04, 0.08, and 0.12, with a target probability of toxicity for the optimal dose of 0.20, and a DLT evaluation time of 84 days. Calculation of the next dose to be assigned and the final toxicity probability estimates were performed using the SAS macro developed by the University of Michigan Comprehensive Cancer Center Biostatistics Unit (31). Patients were stratified within group by prior chemotherapy exposure and analyses were performed separately by strata. rPFS was estimated using the Kaplan–Meier method. Summary statistics were used to assess PSA change and to describe translational results. Time was calculated from day of first treatment until incidence of toxicity, progression, or death.

Gene expression changes in whole blood samples at baseline and during treatment (cycle 1, day 15 and cycle 3, day 15) were evaluated in 18 of 30 patients. RNA from whole blood, collected from participants, was isolated using the Qiagen miRNA Isolation Kit (32) and was analyzed using the PanCancer Immune Pathways and PanCancer Driver Pathways panels, as described previously (33). The PanCaner Driver and Immune Panels included 770 genes each from 24 different driver pathways and 13 cancer associated canonical pathways, respectively.

In addition, plasma samples were collected at baseline in 15 patients and analyzed for circulating tumor DNA (ctDNA) using the Guardant 360 Liquid Biopsy platform (34). Each sample was assessed for different somatic and germline mutations including SNVs, insertions, deletions, gene fusions, and copy-number variants (CNV). For the 15 other participants, there were not viable samples to run such analysis.

Data from this study is available through the Prostate Cancer Clinical Trials Consortium by emailing [email protected].

From May 2018 to January 2020 30 patients were enrolled at six different clinical sites in the United States and patient demographics are included in Table 1. Two patients were withdrawn from the study early. One patient withdrew shortly after randomization and never received a dose of either trial medication. One patient in the prior chemotherapy cohort at dose level 2 (Niraparib 200 mg) was removed from the study due prolonged thrombocytopenia. The remainder of the patients continued therapy until progression of disease or death. All patients were included in treatment analysis. The median age of patients was 69 years. Twenty-five patients were Caucasian, 5 patients identified as black or African American. For patients who had previously received chemotherapy, the median time from completion of chemotherapy to treatment start date was 86 weeks (range 21–193 weeks).

DLTs were evaluated in all three treatment groups and stratified by prior chemotherapy exposure (Table 2). There were no DLT events for patients who received the 100 mg dose of niraparib and no patients were removed from the study due to a treatment related adverse event. In the cohort that had never received chemotherapy, 7 patients received a 200 mg dose of niraparib and there was 1 DLT event, leading to an estimated probability of DLT at this dose level of 0.156 with a 95% confidence interval of 0.042 to 0.385. There were two DLT events in the 5 patients who received a 300 mg dose of niraparib, leading to an estimated probability of DLT for this cohort of 0.214. Because the probability of DLT was greater than 0.2 in the 300 mg cohort, the 200 mg dose of niraparib was confirmed as the MTD and RP2D in chemotherapy naïve patients. There were three DLT events in 5 patients who had previously received chemotherapy for the 200 mg treatment group, leading to an unacceptable DLT probability and a recommendation of 100 mg niraparib as the MTD and RP2D for patients with prior chemotherapy exposure.

There were 297 recorded adverse events that occurred during the trial, 153 of which were deemed attributable to study medications (Table 3). Most commonly, low-grade disturbances to the gastrointestinal system occurred: with nausea (60%), diarrhea (30%), and constipation (27%) observed most frequently. Constitutional complaints such as fatigue (53%) and decreased appetite (27%) were also frequently observed. Significant hematologic toxicities were encountered infrequently, with only 3 patients with grade 3 or higher neutropenia. The average time to neutrophil recovery was 5.5 days, whereas the average time to platelet recovery was 18 days. There were no deaths considered related to study drug. One patient died secondary to widespread progression of disease and this event was not considered to be related to either study medication.

PSA data were recorded for 28 of 30 patients. Of these, 9 patients had a PSA decline, of whom 3 had a PSA decline of >50% (Supplementary Fig. S1). For five participants, the PSA assessment was made prior to the 12th week, given that progressive disease had already occurred.

The median rPFS for all patients included in analysis was 7.1 months (CI, 2.7–9.3 months) with an estimated 6-month rPFS of 51% (CI, 31–67%). For patients who were chemotherapy naïve (Fig. 1A), the median rPFS was 2.9 months (CI, 2.5–10.8 months) and for patients with prior chemotherapy exposure the median rPFS was 8.0 months (CI, 2.7–10.3 months). The estimated 6 month rPFS for these groups was 47% (CI, 21–69%) and 54% (CI, 23–77%). rPFS distributions were similar by dose overall (Fig. 1B) and in chemotherapy naïve (Fig. 1C) and prior chemotherapy cohorts (Fig. 1D).

We performed gene expression profiling in total RNA from blood obtained from patients at the stated time points, with the hypothesis that potential differences in gene expression would be observed between patients experiencing longer duration on treatment as a surrogate marker for response. Indeed, several intriguing changes in gene expression profiles between patients based on treatment duration (median 21 weeks duration for the 18 patients that had viable RNA available for analysis) were detected. At baseline, paired box protein 5 (PAX5) and CD-19 expression were higher in the group that was on treatment for longer than 21 weeks in comparison with patients that stayed on treatment for less than 21 weeks (Fig. 2A). By cycle 3 day 15, expression of both PAX5 (linear fold change = −5.08) and CD19 (linear fold change = −3.82) was more attenuated in the longer treatment duration group in comparison with the shorter treatment duration group.

In contrast, gene expression patterns for IL-1R2, Arginase 1 (ARG1), and FMS like tyrosine kinase 3 (FLT3) increased in the group treated for less than 21 weeks in comparison with the group treated for longer than 21 weeks (Fig. 2B). Specifically, by cycle 3 day 15, ARG1 (linear fold change = 3.32), FLT3 (linear fold change = 3.32), and IL-1R2 (linear fold change = 2.80) had all markedly increased in patients who experienced a relatively faster time to progression. These changes in gene expression are intriguing and need further investigation in future studies.

ctDNA was detected in 14 of 15 samples collected at baseline. Of these, seven had AR amplification and four instances of point mutations within the gene were detected (Table 4). Somatic mutations in TP53 and PIK3CA were detected in 7 and 4 patients respectively as well. Deletions in RB1 and BRCA2 were also detected via copy-number variations in 7 and 4 patients, respectively. There were six mutations found in BRCA2, two of which were germline SNVs, and one mutation found in ATM and CDK12 genes, respectively. Notably, the median time to progression for 7 patients with HRR variants was only 21 weeks; however, the median time to progression for the 2 patients with germline BRCA2 mutations was 43 weeks, significantly longer than the median duration for the entire population.

This multicenter phase IB trial was the first to assess the combination of a PARP inhibitor with Radium-223 in patients with mCRPC without known BRCA mutations. The combination of 200 mg of niraparib with standard dosing of Radium-223 was found to be tolerable in a subset of patients who were chemotherapy naïve, whereas a dose of 100 mg niraparib was tolerable for patients who had experienced prior treatment with taxane therapy. Currently, the combination of olaparib with Radium-223 is under investigation and investigators used a 3+3 trial design to evaluate 12 patients at two different dose levels of olaparib, determining 200 mg twice daily as their R2PD (35). Importantly, our study demonstrates the feasibility and utility of the TITE-CRM statistical design when investigating therapeutics that have a delayed toxicity profile like radio-ligands. We were able to safely evaluate 30 patients over three different dose levels simultaneously, which accelerated trial accrual without exposing participants to excess risk. This trial can serve as a framework for the development of other combined modality therapies using radio-ligands, which may prove pivotal with the recent approval of Lu 177 vipivotide tetraxetan for patients with mCRPC (36).

By downregulating several DNA repair pathways, PARP inhibitors have gained interest as radiosensitizers in many different cancer models and recent studies have demonstrated safety of PARP inhibitors with other radiation modalities (37–39). Our study adds to these findings as we were able to show tolerability of 200 mg of niraparib with standard of care dosing of Radium-223 in patients without prior chemotherapy exposure. The majority of adverse events related to therapy were either grade 1 or 2, with nausea, diarrhea or constipation and fatigue reported as the most common symptoms. Despite potential overlapping toxicities to the bone marrow, we did not observe significant issues with impaired hematopoiesis and there were few grade 3 events of cytopenia. In all but one instance, the cytopenia resolved after holding niraparib. There were no instances of grade 3 infections and there were no deaths attributable to study medications.

Currently, niraparib is approved as a monotherapy for treatment of ovarian cancer at 300 mg daily and studies are ongoing in biomarker selected patients with metastatic prostate cancer at 200 mg daily in combination with abiraterone acetate 1,000 mg. In regards to combination with radiotherapy, studies evaluating Olaparib in lung and glioblastoma have concluded that, when combined with radiation, the MTD of Olaparib needs to be reduced from the recommend monotherapy dose to limit hematologic toxicity (37, 40). In addition, trials are currently enrolling that are testing reduced doses, 100 or 200 mg, of Niraparib in combination with radiotherapy for high-risk early-stage prostate cancer (41, 42). Recently, de Haan and colleagues were able to show that a 25 mg dose of Olaparib was able to effectively reduce baseline poly ADP-ribose (PAR) levels along with radiation induced PARylation in lung tumor tissue and peripheral blood mononuclear cells, concluding that a reduced dose of a PARP inhibitor still has biologic activity in combination with radiotherapy (37). Taken together, our findings that using either 100 or 200 mg doses of Niraparib in combination with radiotherapy are not likely to drastically reduce the clinical activity of the PARP inhibitor.

Our translational studies demonstrate that using the NanoString platform to perform gene expression profiling on whole blood samples was feasible. Although one limitation to testing whole blood is the uncertainty of the source of RNA, whether it is tumor-derived versus from normal cells (43).

We observed differences in PAX5 and CD19 expression when comparing patients who had a response to therapy to those without a response. At baseline, RNA transcripts were higher for both of these targets in responders compared with nonresponders and a statistically significant reduction in mRNA level was observed with treatment in those who remained on study for longer than 21 weeks. PAX5 is an oncogene that encodes a potent transcription factor leading to B-cell development and aberrant expression has been observed frequently with development of aggressive B-cell malignancies (44). CD19, a cell surface marker expressed on all B lymphocytes, is critical for cell function and CD19 expression is in part regulated by PAX5 (45). In addition, significant PAX5 expression has also been demonstrated in nonlymphoid malignancies, including breast, small cell lung cancer, and pediatric tumors and is believed to play a role in cell survival and metastatic potential (44). One downstream target of PAX5 is c-MET, a tyrosine kinase receptor for which mutations and aberrant expression are known promote oncogenesis (46).

B cells in the tumor microenvironment can promote tumorigenesis and increase metastatic potential by causing antibody mediated T-cell suppression, promoting lymphangiogenesis and producing immunomodulatory cytokines that suppress the antitumor immune response (47). In prostate cancer, it has been observed that B-cell infiltration is higher in high-grade tumors and in tumors that eventually recurred or progressed in comparison with low or intermediate grade tumors (48). In addition, infiltrating B cells were shown to support castrate resistant cell growth by secreting lymphotoxin, which lead to cell survival (49).

ARG1 expression was similar at baseline when comparing patients who responded to therapy to those who did not; however, in the patients who did not respond to therapy, by cycle 3 day 15, ARG1 expression was nearly four times higher. These suggest that expression of ARG1 could possibly serve as a protective mechanism, limiting the effect of treatment. ARG1 expression by tumor-associated macrophages (TAM) has been demonstrated as a mechanism to suppress the effect of infiltrating T cells by limiting the amount of L-arginine available for T-cell function (50). Interestingly, we observed expansion of FLT3 transcripts mirrored ARG1 in patients who had relatively faster time to progression. FLT3 critically drives dendritic cell maturation and activity in the tumor microenvironment. This may either lead to an amplified immune response, through increased antigen presentation to infiltrating cytotoxic T cells, or diminished local immune response, by expansion of T regulatory cells (Treg; ref. 51).

With these hypothesis-generating observations in whole blood RNA expression, a speculative mechanism for the effect of niraparib with Radium-223 may be through the immune response within the tumor microenvironment. In patients who stayed on treatment for longer than 21 weeks we observed changes indicative of decreased B-cell function which may allow for increased immune activity within the microenvironment. Conversely, resistance to therapy may be mediated through decreased local T-cell activity, by increased ARG1 expression by TAMs and FLT3 expression driving Treg expansion. Further investigation is needed to explore this phenomenon, but if the mechanism is valid, one could speculate that adding therapies to improve T-cell-mediated immune response could augment this therapeutic approach. Notably, recent phase III trials have reported improved outcomes in mCRPC when combining PARP inhibitors with abiraterone acetate plus prednisone in those with germline/somatic homologous recombination defects, with potentially a less robust benefit in unselected patients (25, 52). Trials are also evaluating the combination of PARP inhibitors with other androgen inhibitors, immune checkpoint inhibitors, and Lu-PSMA-617 (53–55). Thus, the optimal combination partner remains to be determined.

Our exploratory analysis of baseline ctDNA discovered 4 patients with somatic BRCA variants and 2 patients that had germline BRCA2 mutations. These 2 patients were in the 300 and 100 mg cohorts and had dramatically longer times to progression of 38 and 49 weeks, respectively. In addition, the patient in the 100 mg cohort had previously received chemotherapy. These outcomes are concordant with prior observations of favorable responses to Radium-223 in patients with known BRCA mutations (56–58), in addition to the apparent effect of PARP inhibitors in this population. Given the known efficacy of PARP inhibitors in patients with germline/somatic BRCA mutations, we intentionally excluded patients with prior record of these alterations at trial enrollment. However, this expected finding highlights the impact of ctDNA analysis and further demonstrates the importance of screening patients for molecular biomarkers before choosing therapies.

This study was not powered nor designed to evaluated efficacy, but we were able to collect data on PSA change and rPFS to help describe trends. PSA increased in the majority of patients on study; however, rapid PSA response is not a hallmark of Radium-223 therapy, as Radium-223 can cause an early PSA flare in patients who are responding to therapy (59, 60). Nearly 30% of patients had a PSA decline, however only 10% had a PSA decline of >50% at week 12. Three of the 9 patients with a PSA decline were found to have BRCA mutations on ctDNA; 5 of the remaining 6 patients did not have blood samples evaluated for ctDNA. In regards to rPFS, there did not appear to be a statistically significant difference between patients who had prior chemotherapy exposure and those who were chemotherapy naïve, with wide overlapping confidence intervals ranging from 2.5 to 10.8 months.

In conclusion, niraparib in combination with Radium-223 was demonstrated to be safe in combination for the treatment of mCRPC for patients who have previously progressed on AR therapies. We intend to continue investigation with a prospective phase II trial that will better evaluate efficacy endpoints and assess potential long-term toxicities and will serve as a companion to the ongoing trial of olaparib with Radium-223 in unselected patients with mCRPC (35).

B. Leiby reports grants from NCI during the conduct of the study as well as personal fees from Bayer Healthcare outside the submitted work. G. Sonpavde reports personal fees from Janssen during the conduct of the study as well as grants from Sanofi, Helsinn, and Jazz; grants and personal fees from AstraZeneca, Gilead, and EMD Serono/Pfizer; personal fees and nonfinancial support from Lucence; personal fees from Seagen/Astellas, Merck, Bicycle Therapeutics, G1 Therapeutics, Eli Lilly/Loxo, Immunovaccine Technology (IMV), Elsevier Practiceupdate, Mereo, Vial, and Suba Therapeutics; other support from Myriad; and nonfinancial support from BMS outside the submitted work. A.D. Choudhury reports grants and personal fees from Bayer and personal fees from Janssen during the conduct of the study as well as personal fees from Pfizer, Astellas, Eli Lilly, AstraZeneca, Clovis, Dendreon, and Blue Earth outside the submitted work. C. Sweeney reports grants and personal fees from Janssen, Astellas, Pfizer, Bayer, Lilly, AstraZeneca, and CellCentric outside the submitted work. D. Einstein reports grants from Bristol-Myers Squibb, Puma Biotechnology, Cardiff Oncology, and MiNK Therapeutics and other support from Foundation Medicine outside the submitted work. R. Szmulewitz reports personal fees from Janssen and Astra Zeneca and grants and personal fees from Astellas outside the submitted work. O. Sartor reports grants and personal fees from Advanced Accelerator Applications, AstraZeneca, Bayer, Constellation, Janssen, Merck, Progenics, and Tenebio; personal fees from Astellas, Blue Earth Diagnostics, Inc., Bavarian Nordic, Bristol Myers Squibb, Clarity Pharmaceuticals, Clovis, EMD Serono, Dendreon, Fusion, Isotopen Technologien Meunchen, Moyvant, Myriad, Noria Therapeutics, Noxopharm, Novartis, Northstar, POINT Biopharma, Pfizer, Sanofi, Telix, and Theragnostics; and grants from Amgen, Endocyte, Invitae, and Lantheus outside the submitted work. K. Knudsen reports grants from NCI and Prostate Cancer Foundation during the conduct of the study as well as the following relationships over the past 3 years: research support (Novartis, CellCentric) and consultancy/advisory (CellCentric, Sanofi, Celgene, Janssen, Genentech). E.S.-H. Yang reports personal fees from Clovis, Bayer, and AstraZeneca and grants from Eli Lilly and PUMA outside the submitted work. W.K. Kelly reports other support from Janssen and Bayer during the conduct of the study as well as nonfinancial support from Janssen and Bayer outside the submitted work. No disclosures were reported by the other authors.

Z. Quinn: Formal analysis, writing–original draft, writing–review and editing. B. Leiby: Data curation, formal analysis, methodology. G. Sonpavde: Conceptualization, investigation. A.D. Choudhury: Conceptualization, supervision, investigation. C. Sweeney: Investigation. D. Einstein: Investigation, writing–review and editing. R. Szmulewitz: Conceptualization. O. Sartor: Conceptualization, formal analysis, investigation, writing–review and editing. K. Knudsen: Conceptualization, methodology, writing–review and editing. E.S.-H. Yang: Conceptualization, formal analysis, investigation. W.K. Kelly: Conceptualization, formal analysis, supervision, investigation, writing–review and editing.

The authors would like to thank Ms. Deborah Della Manna and Dr. Jianqing Zhang for performing the NanoString experiments. In addition, we thank Jake Vinson, Garrett Abrams, Sarah Wise, and other members of the Prostate Cancer Clinical Trials Consortium for their ongoing support.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.