Introduction

As prostate cancer (PC) advances, bone metastasis is common, ultimately occurring in 80–90% of patients [1, 2]. Bone metastasis is associated with reduced quality of life (QoL), and with increased disability, risk of death, and treatment costs [2,3,4,5,6,7,8,9,10,11]. Therefore, it is essential to manage bone metastasis effectively as part of metastatic castration-resistant prostate cancer (mCRPC) treatment.

Radium-223 (Ra-223) is the first targeted alpha therapy that prolongs overall survival (OS) in patients with bone mCRPC [12]. In a randomized phase III study in patients with CRPC and symptomatic bone metastases (ALSYMPCA), Ra-223 significantly prolonged OS compared with placebo (median: 14.9 versus 11.3 months; hazard ratio: 0.70, p < 0.001). Ra-223 also significantly prolonged time to first symptomatic skeletal event and improved QoL compared with placebo [12].

Ra-223 was approved in Japan for mCRPC treatment in March 2016 and, to date, has been used in several thousand patients. Because understanding of the safety and usefulness of an approved agent in clinical settings is essential, post-marketing surveillance (PMS) has been conducted in Japan since June 2016, in accordance with Japanese regulations. Primary PMS study results, the safety and effectiveness of Ra-223 in the total study population (n = 296), have already been reported, with no new safety concerns identified during the 6-month observation period [13].

The characteristics of patients with CRPC are heterogeneous and, thus, to select the most appropriate treatment, it is necessary to consider the treatment history and characteristics of individual patients. Regarding safety, chemotherapy is generally known to be associated with bone marrow toxicity [14, 15], and hematological treatment-emergent adverse events (TEAEs) are common with Ra-223 treatment [12]. Therefore, the safety and usefulness of Ra-223 based on the status of previous chemotherapy use is of specific interest. Indeed, prior chemotherapy use in non-Japanese patients was shown to lead to an increased risk of hematologic toxicity under Ra-223 treatment in ALSYMPCA [16, 17] and observational [18] studies. This trend is thought to be consistent in Japanese patients, but corroborative data from Japanese patients are lacking. Additionally, we have to consider that patient characteristics often differ due to differences in healthcare systems, reimbursement status, and recommended therapeutic strategies among countries.

It is also essential to assess agents in the context of current real-world treatment strategies. Clinical trials, such as ALSYMPCA and a Japanese phase II study [19], did not include patients who had received prior or concomitant treatment with second-generation androgen-receptor axis-targeted agents (ARATs). Because, nowadays, ARATs are used commonly for PC treatment, it is important to reassess the real-world outcomes of Ra-223 in patients who are treated sequentially with various agents. In particular, data in patients without concomitant use of ARATs are meaningful for evaluating the effects of Ra-223 itself.

In order to obtain further information about the optimal patient characteristics for Ra-223 treatment, and the safety and effectiveness of Ra-223 itself, we performed exploratory subgroup analyses of data from the Japanese PMS study in patients with and without prior chemotherapy, and in patients who did not receive concomitant second-generation ARATs. Also, because the occurrence of hematotoxicity is a safety interest, we investigated patient baseline characteristics contributing to the occurrence of hematotoxicity with Ra-223 treatment.

Patients and methods

Study design and patients

This was a non-interventional, prospective, multicenter, single-cohort PMS study performed in Japan (ClinicalTrials.gov NCT02803437). Patients with CRPC and bone metastasis, who were selected for treatment with Ra-223 according to the participating physicians’ routine clinical practice, were eligible for inclusion. The target sample size was 300 patients, and the enrollment period was set at 18 months. The observation period for the primary study was 6 months, from the first administration of Ra-223 to 30 days after the last administration. Ra-223 was administered at a dose of 55 kBq/kg every 4 weeks for up to six cycles.

The institutional review board at each participating center approved the study, which was conducted in accordance with the Declaration of Helsinki and Japanese regulations on Good Post-marketing Study Practice. According to Japanese regulations, written informed consent from patients was not required.

Endpoints

The primary endpoints were TEAEs and drug-related TEAEs. TEAEs were defined as events which occurred following the first administration of Ra-223 until 30 days after the last administration. An important focus was hematological TEAEs, including anemia, leukopenia, neutropenia, thrombocytopenia, pancytopenia, and other hematological events. The occurrence of skeletal-related events (SREs) was also assessed; SREs included pathological bone fractures, spinal cord compression, events requiring external-beam radiotherapy (EBRT) to relieve skeletal symptoms, events requiring orthopedic surgical intervention, and hypercalcemia. Secondary endpoints included laboratory findings, such as total alkaline phosphatase (t-ALP) and prostate-specific antigen (PSA) levels from baseline.

Statistical analysis

Statistical analyses were generally exploratory and descriptive in nature. For the analysis based on chemotherapy history, patients were divided into those who had received chemotherapy (docetaxel, cabazitaxel, and other chemotherapy) before the first administration of Ra-223 (defined as prior-chemo group) and those who had not (defined as no prior-chemo group). For the prior-chemo group, results are also shown for the subgroup of patients who had received two lines of prior chemotherapy (defined as two lines of prior-chemo group). In the subgroup analysis of patients who received Ra-223 without concomitant ARATs, only patients with treatment periods for enzalutamide or abiraterone that did not overlap with those for Ra-223 were included (defined as without concomitant ARATs group). Because one patient had received docetaxel overlapping with the Ra-223 treatment period, the patient was excluded from the without concomitant ARATs group. Also, in this manuscript, "(second-generation) ARAT" always refers to abiraterone or enzalutamide. Univariate and multivariate analysis with logistic regression analysis (forward–backward stepwise selection method, significance level of 0.05) for the occurrence of hematotoxicity was performed using baseline patient characteristics and prior treatments as individual risk factors. In this analysis, hematotoxicity was defined as the occurrence of hematological laboratory abnormalities (decline) of ≥ CTCAE grade 2; patients with grade 3/4 abnormalities were too few for meaningful analysis. In addition, to investigate the risk factors for hematotoxicity objectively, hematological laboratory abnormalities, not TEAEs, were used. Corresponding baseline hematological laboratory values were always added into the model of multivariate analysis regardless of the univariate analysis results (e.g. a variable of baseline hemoglobin was always added into the model for investigating the occurrence of decreased hemoglobin), given that these factors were expected to correlate with respective hematotoxicities. Log transformation was performed for baseline variables with heavily-skewed distributions (e.g., t-ALP and PSA). Other statistical analysis details are described in Supplementary Methods.

Results

Baseline characteristics and prior and concomitant therapies

Overall, 334 patients were enrolled into the PMS study between June 2016 and November 2017. In November 2019, data for 299 patients were available, and 296 patients were included in the safety and effectiveness analysis sets; three patients who did not receive Ra-223 were excluded. The median observation period was 5.6 (range 1.0–10.6) months.

Baseline characteristics and prior/concomitant therapies are presented in Table 1 according to prior chemotherapy status. Among the 296 patients, 170 (57%) were in the no prior-chemo group, while 126 (43%) were in the prior-chemo group and, of these, 33 (26% of the prior-chemo group, 11% of the overall population) had received two lines of chemotherapy.

Table 1 Patient characteristics, previous and concomitant therapy, according to the status of prior chemotherapy
Full size table

The proportion of patients with WHO’s cancer pain ladder score of 0 was numerically smaller (60% vs 75%) and the median PSA value (33.8 vs 17.3 ng/mL) was higher in the prior-chemo group than the no prior-chemo group. The prior-chemo group tended to have more bone metastases; the proportions of patients with extent of disease (EOD) 1 were 21% and 34%, and, for EOD3, were 37% and 29% in the prior-chemo and no prior-chemo groups, respectively. The prior-chemo group was more likely to have previously received ARATs (78% vs 65%) and EBRT (26% vs 14%).

The two lines of prior-chemo group tended to have more advanced disease, indicated by a smaller proportion having PS 0 and WHO’s cancer pain ladder 0, and a higher median PSA value, even compared with the overall prior-chemo group.

Among the overall population (n = 296), 201 (68%) patients were treated with Ra-223 without receiving concomitant ARATs. Overall, the baseline characteristics and pattern of prior or concomitant therapies in the without concomitant ARATs group were similar to those of the overall population (Supplementary Table 1).

Ra-223 exposure

In the overall population, 69% of patients completed six cycles of Ra-223 treatment, and 75% completed five or six cycles. More patients in the no prior-chemo group completed six cycles of Ra-223 (75%) compared to the prior-chemo group (60%). Patients in the two lines of prior-chemo group were notably less likely to complete six cycles of Ra-223 (39%). In the without concomitant ARATs group, 65% of patients completed six cycles of Ra-223.

Safety

TEAEs and drug-related TEAEs

Among the overall population, TEAEs and drug-related TEAEs occurred in 49% and 26% of patients, respectively; the majority of events were grade 1 or 2 in severity.

TEAEs and drug-related TEAEs in the no prior-chemo group and prior-chemo group are summarized in Table 2. The incidences of TEAEs and drug-related TEAEs were 47% vs 53% and 25% vs 29% in the no prior-chemo group and prior-chemo group, and for ≥ grade 3 TEAEs/drug-related TEAEs were 15% vs 23% and 5% vs 9%, respectively. The incidences of these events tended to be even higher in the two lines of prior-chemo group.

Table 2 Treatment-emergent adverse events (TEAEs) and drug-related TEAEs
Full size table

Among drug-related hematological TEAEs, the most common event was anemia (13% of the overall population) (Supplementary Table 2); this trend was similar across groups (Table 2). The incidence of hematological drug-related TEAEs of any grade was similar in the no prior-chemo group and the prior-chemo group (17% vs 18%), and was numerically higher in the two lines of prior-chemo group (24%). Among hematological drug-related TEAEs, the incidences of serious events were 1%, 5%, and 6%, and for events ≥ grade 3 were 4%, 7%, and 12% in the no prior-chemo, prior-chemo, and two lines of prior-chemo groups, respectively.

The incidences of TEAEs and drug-related TEAEs in the without concomitant ARATs group were similar to those for the overall population (Supplementary Table 2).

Hematological laboratory abnormalities

Hematological laboratory abnormalities that occurred during Ra-223 treatment are summarized in Table 3. The incidence of ≥ grade 3 hemoglobin decreased was numerically higher in the prior-chemo group (14%) compared with the no prior-chemo group (7%), tending to be even higher in the two lines of prior-chemo group (25%). This trend was also similar for ≥ grade 3 platelet count decreased; the incidences were 1%, 5%, and 13% in the no prior-chemo, prior-chemo, and two lines of prior-chemo groups, respectively. The incidences of ≥ grade 3 neutrophil count decreased were low across all groups (from 0 to 2%).

Table 3 Hematological laboratory abnormalities that occurred during Ra-223 treatment, by worst CTCAE gradea
Full size table

The overall incidence of hematological laboratory abnormalities in the without concomitant ARATs group was similar to that in the overall population.

Related baseline characteristics for ≥ grade 2 hematological laboratory abnormalities

Patient baseline characteristics related to the occurrence of ≥ grade 2 hematological abnormalities are shown in Table 4. Multivariate analysis showed that, low hemoglobin high PSA, and high t-ALP at baseline were significantly related to the occurrence of ≥ grade 2 hemoglobin decrease. With respect to the occurrence of ≥ grade 2 platelet count decrease, low hemoglobin, low platelet count, high LDH, and two lines of prior chemotherapy were significantly related. Regarding the occurrence of decreased neutrophil count, because the availability of the model was low (0.635) due to missing data, this result was considered as a reference only.

Table 4 Univariate and multivariate analyses for the incidence of ≥ grade 2 hematological laboratory abnormalities (logistic regression analysis)
Full size table

Skeletal-related events and fractures

A drug-related SRE occurred in one patient in the overall population (0.3%); the patient was in the prior-chemo group (0.8%) and also belonged to the without concomitant ARATs group (0.5%). The SRE was an event requiring EBRT.

During the observation period, a total of five fractures were reported, all of which were judged unrelated to Ra-223 by investigators; fractures included femoral (n = 2), rib (n = 1), femoral neck (n = 1), and pathological bone (n = 1). Only one of the five patients had received concomitant bone-modifying agents.

Effectiveness

Percent changes in t-ALP and PSA over 24 weeks

Percent changes from baseline in t-ALP and PSA over 24 weeks are summarized in Fig. 1, Supplementary Fig. 1, and Supplementary Table 3.

Fig. 1
figure 1

Percent changes in (a) t-ALP and (b) PSA from baseline over 24 weeks by prior chemotherapy status

Full size image

t-ALP decreased in most patients across all groups during the observation period and the transition of changes was not evidently different across groups. t-ALP reductions were observed from week 4 after the first administration of Ra-223 and were sustained throughout the observation period.

Median percent changes in PSA from baseline were shown to increase in all groups, although the increases were relatively smaller in the no prior-chemo group compared to those in the prior-chemo group; the increases tended to be larger in patients who had two lines of prior chemotherapy. The transition of changes in PSA over 24 weeks in the without concomitant ARATs group was similar to that for the overall population.

t-ALP and PSA changes at Week 12 by category

When categorized by extent of change at week 12, any decline in t-ALP was observed in 63–82% of patients across groups (Supplementary Table 3).

At week 12, any decline in PSA was observed in 34%, 23%, and 12% of patients in the no prior-chemo, prior-chemo, and two lines of prior-chemo groups, respectively. The proportions of patients with any decline in PSA at week 12 were similar in the without concomitant ARATs group and overall population.

Discussion

This PMS study is the largest prospective observational study of Ra-223 conducted in Japan. The primary results for the overall population (n = 296) have been published previously [13]. Considering that Ra-223 has been used for patients with various backgrounds in Japanese clinical settings, we performed exploratory analyses to investigate its safety and effectiveness in subgroups by status of prior chemotherapy or concurrent use of ARATs.

In this study, 43% of patients had received chemotherapy before Ra-223 treatment. In the prior-chemo group, some baseline patient characteristics suggesting more advanced PC were observed compared to the no prior-chemo group, including high PSA values and more bone metastases. Additionally, other treatments, including ARATs and EBRT, were used more frequently before Ra-223 in the prior-chemo group, indicating that this population was heavily treated. Overall, Ra-223 was well tolerated in all groups. However, the incidences of TEAEs/drug-related TEAEs with serious or ≥ grade 3 events, and ≥ grade 3 hematological TEAEs, were numerically higher in the prior-chemo group than the no prior-chemo group. Notably, baseline characteristics aligned with advanced PC and a more frequent incidence of TEAEs were evident in the two lines of prior-chemo group, suggesting that more adverse events were presumably due to more advanced disease stage in these patients.

In multivariate analysis, the selected significant factors for hematological abnormalities were low hematological laboratory values at baseline or baseline characteristics suggesting advanced primary disease. While some selected factors were different from those in ALSYMPCA [17], the trend was not contradictory. It is important to understand that differences in era, patient population, and study design may affect the results; for example, cabazitaxel had not yet been approved at initiation of ALSYMPCA. A history of docetaxel use was a significant factor for thrombocytopenia in ALSYMPCA [17], and prior treatment with two lines of chemotherapy was a selected factor for decreased platelet count in the current study. In line with the fact that the incidence of overall TEAEs was especially higher in the two lines of prior-chemo group compared to other groups, this highlights the importance of monitoring patients and paying attention to the occurrence of adverse events particularly in Japanese patients with a history of intensive chemotherapy.

While the decline in t-ALP from baseline was observed in many patients across all groups, the median of percent change in PSA during the Ra-223 treatment period was relatively higher in the prior-chemo group. Notably, the proportion of patients who completed six cycles of Ra-223 was smaller in the prior-chemo group, especially in the two lines of prior-chemo group. Given that some patients might discontinue Ra-223 when PSA increases are unbearable, PSA dynamics might have influenced, partially at least, the Ra-223 completion rate. Unfortunately, because detailed information about disease progression, as a reason for discontinuation, was not obtained in this study, the profile of disease progression was unclear. However, as recommended by guidelines [20], the assessment of treatment effect and decisions to change treatment should be performed comprehensively.

Consensus about the best CRPC treatment sequence has not yet been reached [21, 22], and the best sequence with Ra-223 and other agents also remains unclear. From the sub-analysis of ALSYMPCA, it was shown that, compared with placebo, the Ra-223 survival benefit was sustained regardless of prior chemotherapy [16], and that Ra-223 treatment did not affect subsequent chemotherapy [23]. Also, considering these reports, current subgroup results indicating tendencies of fewer TEAEs and a higher Ra-223 completion rate in the no prior-chemo group might partially suggest that Ra-223 treatment before chemotherapy may be beneficial to patients; however, to reach decisive conclusions, findings regarding the effect of each treatment sequence on overall survival are awaited.

We also investigated the safety and effectiveness of Ra-223 in patients without concomitant use of second-generation ARATs. The fact that most outcomes in this subgroup were similar to those in the overall population provides encouragement that the safety and effectiveness of Ra-223 itself in current real-world settings is sustained without concomitant life-prolonging CRPC treatment, and in patients receiving various sequential CRPC treatments.

There are some limitations to the results, including small sample size and single-cohort study design. Subgroup analysis results from a single cohort should be interpreted cautiously, because they are often attributed to differences in baseline characteristics. However, our analysis reflects real-world patients and should help to develop treatment strategies in clinical practice without influence from patient selection or interventions. Also, although the effect of treatment on QoL is an important aspect for patients, endpoints related to QoL were not included in the protocol of this study because PMS is a survey to mainly investigate the drug’s safety. Prospective investigations about QoL, as well as quality adjusted life-year, in Japanese patients treated with Ra-223 would be expected in the future. Follow-up for this study, up to 3 years, is ongoing and future findings, including prognosis and fracture information, will further strengthen the findings from this exploratory analysis.