Prospective Evaluation of Bone Metabolic Markers as Surrogate Markers of Response to Radium-223 Therapy in Metastatic Castration-resistant Prostate Cancer

This article is featured in Highlights of This Issue, p. 2081

Over the past decade, the treatment landscape for metastatic castration–resistant prostate cancer (mCRPC) has evolved to include cytotoxic chemotherapy (cabazitaxel, docetaxel), androgen axis inhibitors (enzalutamide, abiraterone, and apalutamide), bone-targeted therapy (radium-223), and a prostate cancer therapeutic vaccine (Sipuleucel-T). Bony metastases occur in up to 90% of patients with mCRPC and are associated with significant pain, limited mobility, and increased risk for fractures (1, 2). To date, radium-223, a calcium-mimetic radiopharmaceutical of alpha radiation, is the only bone-targeted agent associated with improved survival outcomes approved for mCRPC (3). Radium-223 complexes with bone mineral hydroxyapatite at areas of increased bone turnover thereby concentrating at areas of bone metastases (4). Radium-223 was approved on the basis of the ALSYMPCA trial that reported a reduction in the risk of death and symptomatic skeletal events with treatment (5, 6). The recommended treatment regimen is six doses of radium-223 at a 4-week interval. Currently, response to radium-223 cannot be accurately assessed or monitored by imaging, clinical symptoms, laboratory values, or biomarkers. Surrogate markers of treatment outcomes with radium-223 are needed to improve tailoring of treatment duration according to response. Identification of real-time markers of response would aid in determining which patients should stop treatment early due to a lack of response.

Bone-metabolic markers (BMMs) have shown promise as predictive biomarkers of response for bone-targeted therapies in mCRPC. The SWOG 0421 phase III trial of docetaxel with or without atrasentan (a bone-targeting agent) for mCRPC, reported that at week 9 of treatment, increasing levels of BMMs N-telopeptide (NTP), pyridinoline (PYR), and alkaline phosphatase were associated with inferior overall survival (OS). In addition, patients with high baseline levels of BMMs had poor prognosis, but also derived survival benefit from treatment with atrasentan (7). Finally, continuous decline in BMMs during treatment was associated with improved survival in a pooled retrospective analysis of three clinical trials in patients with metastatic prostate cancer (8).

In the present report, results from a prospective, randomized trial of radium-223 and enzalutamide (RE) versus enzalutamide monotherapy are presented. The underlying hypothesis was that treatment with radium-223 will lead to a decline in BMMs, and associate with improved treatment outcomes. Thus, BMMs may emerge as surrogate biological markers to guide the duration of radium-223 treatment.

Patients with progressive mCRPC were recruited to a phase II clinical trial of RE versus enzalutamide alone. Enzalutamide was used as a comparator in lieu of placebo due to the progressive nature of patient's disease. Standard dose of enzalutamide (160 mg orally daily) and standard dose of radium-223, that is, 55 kBq/kg IV every 4 weeks for six doses were used. The study had an initial nonrandomized single arm, feasibility cohort (RE, n = 8 patients) followed by a randomized (2:1) group comparing RE versus enzalutamide alone (n = 39 evaluable patients; Supplementary Fig. S1). The design included a feasibility and safety cohort to ensure there was no significant immediate toxicities observed with the combination, which was novel at the time of conceptualization of the study. This study was conducted in accordance with the Declaration of Helsinki, approved by the Institutional Review Board of the University of Utah (IRB 68770), and is registered with clinicaltrials.gov under accession NCT02199197. Written informed consent was obtained from all patients.

Patients with progressive mCRPC who had prior docetaxel treatment or were ineligible or refused docetaxel, and who were candidates for treatment with RE or enzalutamide were eligible for inclusion. Patients had to be ≥18 years of age with a life expectancy ≥6 months, and documented mCRPC as defined by disease progression on continuous androgen-deprivation therapy (ADT) with a castrate level of serum testosterone (≤50 ng/dL). Metastatic disease had to be evidenced by baseline imaging studies (bone scan and/or CT scan, or MRI of the abdomen and pelvis) within 28 days of registration. Key eligibility criteria also included histologic confirmation of adenocarcinoma, ECOG performance status ≤2, disease progression on prior therapy by PCWG2 criteria, no evidence of visceral metastasis, and adequate liver, kidney and bone marrow function at baseline. Patients were excluded if previously treated with enzalutamide or radium-223.

A comprehensive panel of BMMs, including N-terminal propeptide of type 1 collagen (P1NP), NTP, bone alkaline phosphatase (BAP), C-telopeptide (CTP), and PYR were measured at baseline and monthly for 6 months. Baseline imaging included CT chest, abdomen and pelvis, and a nuclear medicine bone scan. Imaging was repeated after 3 months, and 6 months on treatment, and/or at the time of PSA or clinical disease progression if occurred earlier. Subjects remained on study treatment for up to 6 months. Patients could continue enzalutamide after completion of the 6-month study duration. A total of 2 years long-term follow-up was planned to assess for OS and potential long-term toxicity.

Conduct of this clinical trial was approved by the local institutional review board.

The study had two co-primary endpoints. First, was comparison of the change in serum NTP levels, from baseline to after 6 months on treatment (or at disease progression whichever occurred first), between the two treatment arms. The second co-primary endpoint of safety compared the incidence of grade ≥3 hematologic adverse events (AE) in all patients treated on the combination arm to the incidence reported in the ALSYMPCA trial (5). Prespecified secondary endpoints included PSA progression-free survival (PFS); PSA response rate reported as a 50% (PSA50) and 90% (PSA90) reduction from baseline; radiographic PFS (rPFS); radiographic disease control rate (rDCR, defined as complete response, partial response or stable disease); OS (defined as time from treatment initiation to death from any cause); BAP PFS; changes to BMMs (P1NP, NTP, BAP, CTP, and PYR); and AEs of special interest. Only patients enrolled in the randomized portion of the trial were evaluated for the primary endpoint comparing change in NTP levels. All patients in the combination arm were evaluated for the primary safety endpoint, and all patients from both arms were evaluated for the secondary endpoints. Radiographic assessments were made by independent radiologist review, un-blinded to study arm allocation. A description of arm allocation to the different study endpoints is shown in Supplementary Fig. S1.

The change in log₂ NTP was compared between treatment arms using an ANCOVA model with the final value as response, treatment group as predictor, and initial log₂ NTP as an adjustment variable. On the basis of the data from the SWOG 0421 study (7), the estimated baseline mean and standard deviation of log₂ NTP were 3.95 and 1.14 nmol/L, respectively. We further assumed the Pearson correlation of baseline and follow-up NTP was 0.50. With these assumptions and with 27 evaluable patients in the RE arm, and 12 evaluable patients in the enzalutamide arm, there was 85% power to detect a 2.2-fold difference between treatment arms at the two-sided 0.05 significance level.

For the co-primary endpoint of safety, an overall proportion of subjects with grade 3 or higher hematologic AEs, significantly higher than those reported for the radium-223 arm in the ALSYMPCA trial (5), would be evidence of unacceptable toxicity. For the comparison of 4/35 events in the current study versus 128/600 events in the ALSYMPCA trial (5), a maximum likelihood test from a Poisson model was used. With a safety population of 35 evaluable patients, ≥13 events of grade ≥3 hematologic AEs would indicate an unacceptable rate of hematologic AEs. An exact binomial test was performed comparing the proportion of neutropenia, anemia, or thrombocytopenia subjects with grade ≥3 events in the study to the proportions of neutropenia, anemia, or thrombocytopenia subjects in the ALSYMPCA trial (5). Exact 90% confidence intervals were calculated for AEs and serious AEs. With 35 evaluable patients an exact 90% confidence interval for an AE will extend no more than 17.5% from the observed proportion, and there is a 90% probability that an AE with a probability of 6.4% would be observed at least once. For the secondary endpoint of safety, Fisher exact test two-sided was used to compare AE between the combination and monotherapy arms.

Interim analysis of grade ≥3 hematologic toxicity in the first 10 patients recruited to the combination arm (radium-223 + enzalutamide) was separately evaluated as required by the sponsor (Bayer pharmaceutical). The null hypotheses were based on proportions reported in the ALSYMPCA trial (5). The purpose of the interim analysis was to formally review safety data and assess whether continued enrollment of subjects was acceptable. The safety data in this interim analysis were reviewed by the institutional data safety review committee. A proportion of subjects with grade 3 or higher hematologic AEs, significantly more than those reported for the radium-223 arm of the phase 3 ALSYMPCA (5) trial would be evidence of unacceptable toxicity. In the radium-223 arm of the phase III ALSYMPCA trial (5) 12.7% (76/600) subjects experience grade ≥3 anemia, 39/600 (6.5%) experienced grade ≥3 thrombocytopenia, and 13/600 (2.2%) experienced grade ≥3 neutropenia.

The Wilcoxon rank-sum test with continuity correction, two sided was used to compare response rates between the combination and monotherapy arms.

Univariate associations between changes in bone markers (log₂ baseline minus log₂ final value) and outcomes in all patients (N = 47) was performed. Ordinal outcomes, radiographic response (progressive disease, stable disease, partial response and complete response) and PSA response (<30%, 30%–49%, 50%–89%, ≥90%), were examined in the context of proportional odds logistic regression models, binary outcomes were assessed in the context of logistic regression models, and time to event outcomes were assessed in the context of Cox proportional hazards models. For the ordinal outcomes (radiographic and PSA response), the odds ratios represent the increase in odds that a patient's response is better per unit best response in marker. For example, a patient with a 1-U greater best response in N-Telopeptide, Cross-Linked than another patient (corresponding to a twice greater decline because of the log₂ scale), we would estimate to have 6.11-fold increased odds of having a better response.

All statistical analysis were performed using the R-statistical package v.3.5.2 (https://www.r-project.org/).

Between July 2014 and November 2017, a total of 49 patients were enrolled in this study. Baseline demographics and clinical characteristics for eight patients treated with the combination RE therapy in the initial nonrandomized single-arm feasibility cohort, followed by 41 patients in the randomized (2:1) to either the combination or monotherapy arms respectively are described in Table 1. At time of data cutoff, the median duration of follow-up was 26.0 and 21.0 months in the RE and enzalutamide arms, respectively. Only 2 of 49 patients (one in each arm) refused and did not receive a concomitant bone strengthening therapy (zoledronic acid or denosumab) at enrollment and throughout the study. Twenty-one out of 35 (60%) patients in the combination arm and nine out of 14 (64%) patients treated with enzalutamide had prior progression on abiraterone. Twelve of 35 (34%) of RE patients and three of 12 (25%) enzalutamide patients had prior progression on docetaxel. Of the patients randomized to enzalutamide, four were subsequently treated with radium-223, three are still on enzalutamide, three received chemotherapy, and two patients chose to pursue hospice because of rapidly progressing disease.

A statistically significant relative change to NTP ratios between arms (0.64; 95% CI, 0.51–0.81; P = 0.00048) was observed and favored the combination RE versus enzalutamide arm (Table 2). The estimated NTP relative changes in each arm are 0.84 (95% CI, 0.75–0.95) and 1.34 (95% CI, 0.99–1.82) in the RE and enzalutamide arms, respectively (Table 3). This reflects a relative 39% decrease in RE versus enzalutamide arm. Overall, BMMs decreased with the RE therapy compared with enzalutamide alone, with the exception of PYR that had a similar change in both treatment arms (Table 2). A decreased ratio in all BMMs between treatment arms was statistically significant with exception of CTP and PYR (Table 3).

There was evidence that declines in bone markers is associated with better outcomes, and conversely increases in bone markers may be associated with worse outcomes (Table 4). Notably, declines in NTP showed statistically significant relationships with improved radiographic response, PSA response, radiographic DCR, and radiographic PFS. BAP, CTP, P1NP, and PYR showed similar trends, but with less consistent statistical significance. See Supplementary Table S2 for timing to nadir of BMMs.

At the time of data cutoff, median PSA PFS was 9.86 months (95% CI, 4.67–22.5) in the RE arm, and 3.34 months (95% CI, 2.66 to not estimable, NE) in the enzalutamide arm (HR, 0.79; 95% CI, 0.36–1.75; P = 0.60; Fig. 1A). Median rPFS was 11.31 months (95% CI, 9.07–NE) in the RE arm and 5.62 months (95% CI, 2.76–NE) in the enzalutamide arm (HR, 0.54; 95% CI, 0.22–1.32; P = 0.20; Fig. 1B). Median OS was 26.0 months (95% CI, 17.6–NE) in the RE arm and 21.0 months (95% CI, 16.6–NE) in the enzalutamide arm (HR, 0.54; 95% CI, 0.22–1.32; P = 0.40; Fig. 1C). A statistically significant difference in median BAP PFS was noted with the median not reached in the RE arm and 4.34 months (95% CI, 2.76–NE) in the enzalutamide arm (HR, 0.12; 95% CI, 0.03–0.50; P = 0.0007; Fig. 1D).

Improvements in PSA30, 50, and 90 response rates were observed in patients in the RE arm compared with the enzalutamide arm. The categorical PSA response significantly favored the RE arm compared with the enzalutamide arm (P = 0.024; Table 5). In the RE arm, radiographic disease control rate (rDCR) was 89% (31/35 patients), with 28 patients had stable disease (80%), four had progressive disease (11%), three had a partial response (9%), and no complete responses (P = 0.058; Table 5). In contrast, in the enzalutamide arm, rDCR was 67% (8/12 patients), with eight stable disease (67%), four progressive disease (33%), and no partial or complete responses. As of data cutoff, two patients in the enzalutamide arm had not yet experienced radiographic or symptomatic disease progression compared with nine patients receiving RE.

A prespecified interim analysis of safety (mandated by the study sponsor, Bayer Pharma) was conducted in the first 10 patients that completed treatment on the combination arm or discontinued study treatment early (Supplementary Table S1). Enrollment was continued after no safety signals were detected.

There was no significant difference in observed grade ≥3 hematologic toxicities in patients treated with RE in our study compared with the ALSYMPCA trial (P = 0.91, Table 6; ref. 5). When comparing toxicity profiles between treatment arms, the overall incidence of AEs was higher in the RE arm compared with enzalutamide; however, grade≥3 AEs were similar in both arms (Table 7). No pathological fractures were observed in any patient enrolled in this trial at the time of data cutoff, that is, a median follow-up of 26.0 months in the RE combination arm and 21.0 months in the enzalutamide arm. The cytopenias in the RE patients were of expected frequency and duration compared with radium-223 alone in the ALSYMPCA trial. No patients required growth factor support or a bone marrow biopsy due to severe or prolonged cytopenias.

Currently there are no laboratory or radiographic markers available to differentiate between responders and non-responders to radium-223. BMMs have shown some promise for real-time monitoring of response to radium-223. The results reported here support this contention, because a significant decrease in BMM levels was associated with a corresponding response to treatment. In an exploratory analysis from the ALSYMPCA trial, Sartor and colleagues (9) reported that decrease in total alkaline phosphatase from baseline to 12 weeks of radium-223 was associated with longer OS compared with patients with no decline (17.8 vs. 10.4 months, P < 0.0001). In SWOG 0421, elevated baseline levels and increasing levels of BMMs (NTP, PYR, BAP, and P1NP) after 9 weeks of atrasentan therapy were associated with worse survival (P < 0.001; ref. 7). In a phase II trial of 53 men with mCRPC randomized to either radium-223 and docetaxel or docetaxel alone, alkaline phosphatase and P1NP levels declined more rapidly in the combination arm compared with docetaxel alone (10). In a phase II trial of radium-223 in combination with either sorafenib or pazopanib for patients with mRCC, NTP, CTP, BAP, and P1NP declined significantly between baseline and weeks 8 and 16 of treatment (11).

This study is the first to report on the correlation of BMM profiles with response outcomes and efficacy from a prospective, randomized trial in mCRPC. Furthermore, our findings are consistent with the preliminary safety results from the phase III clinical trial (PEACE-3, NCT02194842), which showed that RE is safe when used with bone-strengthening agents. The only data currently available from PEACE-3 are the frequency of skeletal fractures, which indicated an absolute increased risk of 33% with RE versus 13% with enzalutamide alone (12). This difference in risk was abolished when patients received a bone-strengthening agent at least 6 weeks before first treatment with RE (12). These safety results are in contrast with the phase III clinical trial (ERA223) of radium-223 and abiraterone acetate (AA), which was unblinded early following an increased rate of skeletal fractures (29%) and deaths in the combination arm over placebo (13).

In this study, the phase II clinical trial of RE versus enzalutamide alone met its primary efficacy endpoint of a significant decline in serum NTP levels with the combination therapy versus enzalutamide monotherapy. In addition, BAP and P1NP, declined significantly more in the combination arm compared with enzalutamide monotherapy. Furthermore, the changes in the BMMs directly correlated with treatment outcomes (Table 4). Patients in the combination arm demonstrated significantly improved outcomes in PSA response rates and rDCR compared with enzalutamide alone. Our findings are in contrast with the phase III ERA223 trial that found no improvement in OS (HR, 1.195; P = 0.1280) or radiographic PFS (HR, 1.152; 95% CI, 0.96–1.383) with the combination of radium-223 plus abiraterone over abiraterone alone. This may be due to differences between bone-strengthening therapy use between the trials or differences between abiraterone and enzalutamide when used in combination with radium-223. The ongoing PEACE-3 trial will provide definitive evidence of clinical efficacy and safety of the combination of radium-223 plus enzalutamide in this population. RE was safe in patients with mCRPC, as the incidence of grade ≥3 hematologic AEs were similar to the ALSYMPCA trial. Furthermore, non-hematologic AEs were similar in both arms.

This study is limited by its small sample size and was not powered to compare survival outcomes and efficacy of RE versus enzalutamide alone. Also, this trial did not stipulate the therapy used after disease progression. Differences in subsequent therapy may have influenced the results on OS in this study. Addition of a third arm using radium-223 monotherapy would have better defined the impact of radium-223 alone on BMMs. However, this arm would not have allowed for correlation of BMMs with PSA PFS, radiographic PFS, and objective responses, as monotherapy with radium-223 has never been shown to impact any of these efficacy endpoints. Nevertheless, findings of improved PSA and radiographic response are encouraging.

To conclude, combination therapy with RE is associated with a significant decline in BMM levels compared with enzalutamide alone, and correlated with improved outcomes. Moreover, the combination is safe and feasible when used with concurrent bone-strengthening agents. Following independent validation, BMMs may emerge as surrogate markers to inform and optimize treatment duration with radum-223. Furthermore, these data may position BMMs as biomarkers of response to therapy in upcoming trials of novel radio-isotopes such as next-generation alpha and beta emitters in the setting of metastatic castration-resistant prostate cancer.

N. Agarwal is a paid consultant for Astellas, AstraZeneca, Bristol-Myers Squibb, Bayer, Clovis, Eisai, Exelixis, EMD Serono, Eli Lilly, Foundation Medicine, Genentech, Janssen, Merck, Novartis, Nektar, Pfizer, Pharmacyclics. R.H. Nussenzveig is a paid consultant for Tempus Laboratories. J. Hoffman reports receiving commercial research grants via his university from Blue Earth Diagnostics and GE Healthcare. S. Gupta reports receiving commercial research grants from Bristol-Myers Squibb, Rexahn, Incyte, LSK, Five Prime, Mirati, QED, Debiopharm, Merck, Pfizer, AstraZeneca, MedImmune, Clovis, and Immunocore, and has immediate family members who hold ownership interest in Salarius Pharmaceuticals. B. Haaland is a paid consultant for Prometic Life Sciences. B.L. Maughan is a paid advisory board member for Exelixis, Astellas, Bayer Oncology, Bristol-Myers Squibb, Jannsen Oncology, Tempus, and Peloton Therapeutics, and reports receiving commercial research grants via his institution from Bavarian-Nordic, Clovis, and Bristol-Myers Squibb. No potential conflicts of interest were disclosed by the other authors

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH http://dx.doi.org/10.13039/100000002.

Conception and design: N. Agarwal

Development of methodology: N. Agarwal, R.H. Nussenzveig, A.W. Hahn, J.M. Hoffman, K. Boucher, B. Haaland, B.L. Maughan

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Agarwal, J.M. Hoffman, K. Morton, S. Gupta, J. Batten, J. Thorley, B.L. Maughan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Agarwal, R.H. Nussenzveig, A.W. Hahn, J.M. Hoffman, K. Morton, S. Gupta, J. Batten, J. Thorley, J. Hawks, V.S. Santos, G. Nachaegari, X. Wang, K. Boucher, B. Haaland, B.L. Maughan

Writing, review, and/or revision of the manuscript: N. Agarwal, R.H. Nussenzveig, A.W. Hahn, J.M. Hoffman, K. Morton, S. Gupta, J. Batten, J. Thorley, J. Hawks, V.S. Santos, G. Nachaegari, X. Wang, K. Boucher, B. Haaland, B.L. Maughan

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): N. Agarwal, R.H. Nussenzveig, J. Hawks, G. Nachaegari, B.L. Maughan

Study supervision: N. Agarwal, B.L. Maughan

We thank the patients who participated in this clinical trial and their families for ongoing support. Research reported in this publication was supported by the NCI of the NIH under Award Number P30CA042014 and 3P30CA042014-25S2 (to N. Agarwal). This study was approved by the Institutional Review Board of the University of Utah (IRB 68770) and is registered with clinicaltrials.gov under accession number NCT02199197.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.