The primary objective of this BA/BE study was to determine the BE of the RS-DAT formulation of niraparib/AA with respect to single-agent combination (SAC) at steady state under modified fasting conditions in patients with mCRPC (periods 2 and 3). The secondary objective was to determine the relative BA (rBA) of the LS-DAT formulation of niraparib/AA with respect to SAC after a single-dose administration under modified fasting conditions in patients with mCRPC (period 1).

This open-label randomized multicenter study in patients with mCRPC was conducted at 14 sites in the USA and Europe between 10 December 2020 and 15 October 2021. The genotoxic potential of niraparib precluded conduct of this study in healthy participants. For evaluation of the BE of the RS-DAT with respect to SAC, we used a randomized two-way crossover, multiple-dose design followed by a multiple-dose extension phase to ensure therapeutic intent in patients with mCRPC. A similar BE study design for the evaluation of LS-DAT was not considered acceptable for ethical reasons: the dosing regimen (niraparib 100 mg/AA 1000 mg) has not been established as an effective therapeutic regimen. Thus, a single-dose run-in period was used to evaluate the LS-DAT bioavailability relative to SAC. The resulting study design included a 21-day screening phase, a PK assessment phase, an extension phase, and a follow-up phase (Fig. 1). Data only from the PK assessment phase were used for PK, pharmacodynamic, and safety assessments presented herein.

Study design with treatment sequences for randomization. Period 1: niraparib 100 mg and AA 1000 mg. Period 2 and 3: niraparib 200 mg and AA 1000 mg. During repeated dosing (periods 2, 3, and extension), all patients received niraparib and AA once daily in combination with prednisone/prednisolone 5 mg twice daily. AA abiraterone acetate, D day, LS-DAT lower-strength dual-action tablet (niraparib 100 mg/AA 1000 mg), RS-DAT regular-strength dual-action tablet (niraparib 200 mg/AA 1000 mg), SAC single-agent combination, seq sequence

Following confirmation of eligibility during the screening phase, patients were randomly assigned before first administration of study treatment in period 1 to one of the four treatment sequences based on a computer-generated randomization schedule that was prepared before the start of the study. The niraparib 200 mg/AA 1000 mg dose was administered orally as two RS-DATs, which each contained niraparib 100 mg and AA 500 mg, and the niraparib 100 mg/AA 1000 mg dose was administered orally as two LS-DATs, which each contained niraparib 50 mg and AA 500 mg. Study drugs were administered under modified fasting conditions, consistent with clinical recommendations [21], defined as study treatment intake on an empty stomach only (intake 1 h before or ≥ 2 h after a meal). On intensive PK sample collection days −7, 11, and 22, patients were required to fast overnight and were allowed to eat a standard breakfast at exactly 1 h after dosing. Strong CYP3A4 inducers and p-glycoprotein inhibitors/inducers were prohibited during the PK assessment phase on the basis of interaction with niraparib and/or abiraterone.

The PK assessment phase comprised three periods: period 1, single dose of niraparib 100 mg/AA 1000 mg on study day −7 as either LS-DAT or SAC; period 2, daily dose of niraparib 200 mg/AA 1000 mg from study days 1–11 as either RS-DAT or SAC; and period 3, daily dose of niraparib 200 mg/AA 1000 mg from study days 12–22 as either SAC or RS-DAT (whichever was not taken in period 2). This scenario resulted in four possible treatment sequences to which patients may have been randomized. In the extension phase, both niraparib and AA as a SAC or AA alone were continued (study day 23 to discontinuation). From period 2 onward, and throughout the study, niraparib/AA was given in combination with prednisone/prednisolone 5 mg twice daily (once in the morning and once in the evening with food, except on study days 11 and 22 when it was taken with lunch and dinner). Patients received study treatment until disease progression, withdrawal of consent, loss to follow-up, lack of clinical benefit in the opinion of the investigator, start of subsequent anticancer therapy, or until the sponsor ended the study.

The study protocol and amendments were reviewed by an independent ethics committee or institutional review board and were conducted in accordance with the ethical principles that have their origin in the Declaration of Helsinki and that are consistent with Good Clinical Practices and applicable regulatory requirements. Patients or their legally acceptable representatives provided written consent to participate in the study.

The study population consisted of patients with mCRPC who, in the opinion of the investigator, may derive benefit from study treatment. Other inclusion criteria included age ≥ 18 years, Eastern Cooperative Oncology Group Performance Status scale score of ≤ 1, and acceptable organ function as defined per study protocol. Patients who received prior therapy with apalutamide or enzalutamide were required to have had ≥ 8-week or ≥ 6-week wash-out, respectively, before first dose of study treatment. Patients who previously had progressed on AA alone or AA in combination with a PARPi, or who previously discontinued treatment with AA or PARPi due to related toxicity, were excluded. Other exclusion criteria included history or current diagnosis of myelodysplastic syndromes/acute myeloid leukemia, ongoing serious systemic infection, disorders affecting gastrointestinal absorption, clinically significant cardiovascular disease, uncontrolled hypertension, moderate or severe hepatic impairment, active hepatitis B or C infection, and inadequately controlled human immunodeficiency virus infection. All patients were tested at baseline for presence of an HRR gene alteration using archived tumor material and blood testing. Results were communicated to investigators before the end of the PK assessment phase to guide continuation of therapy with the combination of niraparib and AA or AA alone during the extension phase. The planned total sample size, based on statistical assumptions to power BE based on a crossover study design, was approximately 120 patients. The study was not powered to support BE with parallel design (LS-DAT). Additional patients could be enrolled to replace patients who withdrew during the PK assessment phase.

Venous blood samples were collected for measurement of niraparib and abiraterone plasma concentrations. PK samples were collected up to 3 days post-dose (72 h post-dose) during period 1 and under steady state conditions from pre-dose up to 24 h post-dose on study days 11 and 22 during periods 2 and 3. Plasma samples were analyzed for niraparib and abiraterone concentrations using a validated, specific, and sensitive liquid chromatography mass spectrometry/mass spectrometry method. For niraparib, separation was done with a C18 1.7 µm × 2.1 mm × 50 mm column, and detection was done on an API-5500 [25]. For abiraterone, separation was done with a C18 column 1.7 µm × 2.1 mm × 50 mm column, and detection was done on an API-4000 [26]. Plasma concentrations of niraparib and abiraterone were determined using a lower limit of quantification of 5.00 ng/mL and 0.200 ng/mL, respectively.

Primary PK parameters for LS-DAT evaluation were maximum observed analyte concentration (Cmax) and area under the plasma concentration–time curve (AUC) from time 0 to 72 h post-dosing (AUC0–72h) as calculated by a linear up-log down method using Phoenix™ WinNonlin® (version 8.1; Certara LP, USA). Primary PK parameters for RS-DAT evaluation were Cmax at steady state (Cmax,ss) and AUC from time 0–24 h at steady state (AUC0–24h,ss). For both evaluations, if ≥ 1 PK parameter of interest was not estimable for a given patient in ≥ 1 treatment periods, the patient’s data were not included in statistical analyses of that PK parameter.

Descriptive statistics, including arithmetic mean, standard deviation (SD), coefficient of variation (CV), geometric mean (GM), median, minimum, and maximum were calculated for plasma concentrations for each study treatment at each sampling time and for all PK parameters using Phoenix™ WinNonlin® (version 8.1; Certara LP, USA). The PK-evaluable population was composed of all patients who completed ≥ 1 PK period and had sufficient concentration data to accurately estimate ≥ 1 PK parameter. The BE evaluable population was composed of all patients who completed both periods 2 and 3 with sufficient PK sample collection to accurately estimate ≥ 1 PK parameter and without events deemed to affect PK.

A descriptive PK analysis and a primary statistical analysis to determine the BE of the RS-DAT with respect to SAC were performed on log-transformed PK parameters data (Cmax,ss, AUC0–24h,ss, and observed trough analyte concentration at steady state [Ctrough,ss]) for niraparib and abiraterone from the PK and BE evaluable populations, respectively, from periods 2 and 3. A linear mixed-effect model that included treatment, period, and sequence as fixed effects, and patient within sequence as a random effect, was used to estimate the least squares mean and intrapatient variance. Using these parameters, the point estimate and 90% confidence intervals (CIs) for the difference in means on a log scale between test and reference were constructed. Limits of the CIs were retransformed using antilogarithms to obtain 90% CIs for the GM ratios (GMRs) of Cmax,ss and AUC0–24h,ss between the RS-DAT and SAC for niraparib and abiraterone. BE between the RS-DAT versus SAC was concluded if the 90% CIs for the GMRs of RS-DAT over SAC for the primary PK parameters of both compounds fell simultaneously between 80% and 125%.

A descriptive PK analysis and the rBA assessment of the LS-DAT versus SAC were performed on PK parameters data for niraparib and abiraterone from the PK evaluable population from period 1. An analysis of variance (ANOVA) model with treatment as a fixed effect was applied to construct 90% CIs for the GMRs of primary PK parameters between the LS-DAT and SAC for niraparib and abiraterone.

To further assess the rBA of abiraterone in the LS-DAT versus SAC within the same patients and to improve precision of the estimates, a paired analysis using abiraterone PK from treatment sequences 3 and 4 was performed. Specifically, since abiraterone PK at the 1000 mg dose is linear and stationary, Cmax,ss of the LS-DAT was obtained from the corresponding single-dose Cmax (observed in period 1) via nonparametric superposition and by applying accumulation factors (from single dose to steady state) derived from the abiraterone pre-final population PK (PPK) model (described in more detail below).

Each patient in the analysis received both the LS-DAT and SAC; therefore, this analysis was a paired comparison for Cmax,ss (Cmax,ss for LS-DAT extrapolated from single-dose Cmax observed in period 1 versus Cmax,ss for SAC from periods 2 and 3) and AUC0–24h,ss (AUC0–∞ from period 1 used as AUC0–24,ss for the LS-DAT versus AUC0–24h,ss for SAC from periods 2 and 3). A linear mixed-effects model that included treatment as a fixed effect and patient as a random effect was applied to construct 90% CIs for the GMRs of Cmax,ss and AUC0–∞ for the LS-DAT and AUC0–24h,ss for SAC between the LS-DAT and SAC for abiraterone.

Because BA assessment of the LS-DAT was conducted as a parallel-group design and was not intended to formally demonstrate BE between LS-DAT and SAC, a preplanned clinical trial simulation was conducted to evaluate the BE of LS-DAT versus SAC using a crossover design. First, the characteristics of niraparib and abiraterone PPK, based on available PK data of the RS-DAT, the LS-DAT, and SAC, were determined [27]. For model-based BE assessment of the LS-DAT versus SAC, PPK simulations of 1000 replicates of the BA/BE study design in periods 2 and 3 (i.e., two-way steady-state crossover PK assessment phase) were performed using R version 3.4.1 with a sample size of 120, which was approximately the sample size at final analysis of the BA/BE study.

The niraparib PPK model included the LS-DAT as a covariate in the PPK absorption parameters [27]. Based on the totality of available PK data for abiraterone, no formulation differences in absorption between the LS-DAT and SAC could be demonstrated for abiraterone. Hence, the final abiraterone PPK model did not contain any effect of the LS-DAT on absorption parameters. For this reason, generating BE simulations of LS-DAT versus SAC for abiraterone using the final PPK model would have yielded a 100% probability of meeting BE criteria. To generate a more conservative estimate of the probability of meeting BE criteria, a pre-final abiraterone PPK model [27] was used for the model-based BE assessment of the LS-DAT. This pre-final model included effects of the LS-DAT on first-order absorption rate constant (estimated as −20% for the LS-DAT versus SAC, with 100% relative standard error [RSE]) and duration of zero-order drug release (estimated as −34% for the LS-DAT versus SAC, with 89% RSE). As such, for the purpose of BE simulations, a difference in these two absorption parameters between the LS-DAT and SAC was still assumed (while accounting for their large RSE), even in the absence of a statistically significant difference between formulations.

Using the PPK models [27], individual steady-state (study days 11 and 22) PK profiles were generated from individual PPK parameters, which were obtained by randomly sampling from the uncertainty in the estimation of structural and covariate parameters, the distribution of covariates in the BA/BE study, the interindividual variability, and the residual unexplained variability. Day 11 and 22 individual exposure parameters AUC0–24h,ss and Cmax,ss for the LS-DAT and/or SAC for both niraparib and abiraterone were calculated from the simulated data using noncompartmental analysis, and BE was evaluated as described above for the RS-DAT versus SAC. The probability of demonstrating BE for the LS-DAT versus SAC was calculated as the proportion of simulated clinical trial replicates in which BE criteria (90% CI of estimated GMR within the 80%–125% range) were simultaneously met for both AUC0–24h,ss and Cmax,ss for both niraparib and abiraterone. Additionally, the GMR point estimates and lower and upper 90% CI bounds across the simulated replicates for both AUC0–24h,ss and Cmax,ss for both niraparib and abiraterone were presented graphically.

A similar model-based assessment was additionally conducted to assess the probability of demonstrating BE under a parallel, single-dose design. The same methodology described above was applied, except for the following. A total of 60 patients were simulated for each formulation (the LS-DAT and SAC), which is similar to the sample size for the LS-DAT and SAC in the BA/BE study. Single doses were simulated over the same nominal sampling time grid as defined for niraparib 100 mg/AA 1000 mg given as LS-DAT or SAC in the BA/BE study (i.e., PK sampling up to 72 h). The individual exposure parameters AUC0–72h and Cmax were analyzed with a log linear mixed-effects model with treatment (LS-DAT vs SAC) as a fixed effect. Point estimates and 90% CIs of the estimated GMR for the parallel comparison of LS-DAT versus SAC were then obtained for each of the 1000 simulated study replicates, and the proportion of simulated trials where BE criteria were simultaneously met for both AUC0–24h,ss and Cmax,ss for both niraparib and abiraterone was calculated. The results then were compared with the results from the rBA assessment of the LS-DAT versus SAC on the basis of period 1 data described above.

All patients were closely monitored at regular intervals during the conduct of the study by monitoring adverse events (AEs), vital signs (blood pressure, heart rate, body temperature), electrocardiograms, and laboratory safety. Special attention was paid to the development of niraparib-related and AA-related events, such as hematological toxicity, hypertension, hypokalemia, and liver toxicity, for which specific management guidelines were provided in the study protocol.