Comparison of Two Manufacturing Processes of Daprodustat for Bioequivalence and Dissolution in Healthy Volunteers: A Randomized Crossover Study

Daprodustat, an orally bioavailable hypoxia‐inducible factor–prolyl hydroxylase enzyme inhibitor, has recently completed phase 3 clinical development for treating anemia of chronic kidney disease. Part A of this 2‐part, randomized, double‐blind, single‐dose, cross‐over study (NCT04640311) compared pharmacokinetic properties of a single oral dose of daprodustat 4 mg tablets manufactured via twin‐screw wet granulation (process 1) to 2 sets of 4 mg tablets manufactured via high‐shear wet granulation (process 2), to assess the impact of different dissolution profiles on pharmacokinetics. Part B assessed the bioequivalence of daprodustat tablets manufactured via process 1 with tablets manufactured via process 2 at 5 different dose strengths (1, 2, 4, 6, and 8 mg). In part A, mean plasma concentrations of daprodustat were comparable over a 24‐hour period despite differences in manufacturing processes and dissolution profiles. In part B, the 90% confidence intervals of the ratios of the least squared means for area under the concentration‐time curve and maximum observed plasma concentration fell within the 0.8–1.25 bioequivalence range for all doses, except for maximum observed plasma concentration at 8 mg. A prespecified sensitivity analysis jointly assessing all doses showed bioequivalence for all doses tested. No new safety concerns for daprodustat were identified.

Patients with chronic kidney disease (CKD) often become anemic, a complication that occurs, in part, due to relative erythropoietin (EPO) deficiency and iron absorption and utilization impairment. 1,2 Daprodustat is an orally bioavailable hypoxia-inducible factorprolyl hydroxylase enzyme inhibitor that has completed global phase 3 development for treating anemia in patients with CKD and is approved for clinical use in Japan. [3][4][5][6][7][8][9][10][11][12] Daprodustat effectively raises hemoglobin concentrations with lower EPO levels than those observed when other exogenously administered EPOstimulating agents are administered. 13,14 Clinical data have shown that daprodustat administered orally is rapidly absorbed, with time to maximum concentration (t max ) of 1-4 hours, and exhibits dose-proportional increases in exposure. [15][16][17] The pharmacokinetic (PK) profile of daprodustat is similar in the healthy adult and CKD populations as the maxi-mum observed drug concentration (C max ), area under concentration-time curve (AUC), and t max have been shown to be comparable in both groups. 18 Coadministration of a single 100 mg daprodustat dose with a highfat, high-calorie meal resulted in an 11% and 31% decrease in daprodustat AUC and C max , respectively, and Daprodustat from process 1 Daprodustat from process 2 Process 1, twin-screw granulation; process 2, high-shear wet granulation. a Sequence followed for all dose levels (1, 2, 4, 6, and 8 mg). a 1-hour mean delay of t max relative to fasted administration while dosing 4 mg of daprodustat 30 minutes after a CKD meal (500-700 kcal, composed of 12-16 g of protein, 1-2 g of salt, and up to 500 mg of potassium) resulted in a 9% and 11% decrease in daprodustat AUC and C max , respectively, and a 1-hour mean delay of t max relative to fasted administration. 19,20 Daprodustat is extensively metabolized by cytochrome P450 2C8, is highly absorbed across the gastrointestinal tract following oral administration, and is mainly cleared through hepatobiliary and fecal routes. 16,20 As daprodustat is a cytochrome P450 2C8 substrate, a significant interaction was observed with gemfibrozil, and a lesser interaction was observed with trimethoprim. 20,21 Additionally, daprodustat has been shown to be noninferior to erythropoiesis-stimulating agents with regard to safety and efficacy. 10,12,13 The PK profile of daprodustat tablets manufactured via the twin-screw granulation process (process 1) has been previously described, 16,21 and tablets using this method have been administered in the phase 3 trials with intent to commercialize upon completion of the trials. To enable manufacturing flexibility and increase supply capacity, tablets produced via a high-shear wet granulation process (process 2) have also been developed. The objectives of this 2-part study were to compare the PK profile of daprodustat tablets manufactured via process 1 to 2 alternative daprodustat tablets manufactured via process 2, each having different (slower) dissolution profiles (part A), and to establish whether the tablets via process 2 are bioequivalent to tablets made via process 1 (part B).

Study Design
Part A of this 2-part study was a randomized, doubleblind, single-dose, 3-period, crossover study in healthy individuals to compare the PK profiles of two 4 mg daprodustat tablets with successively slower dissolution profiles manufactured via process 2 relative to the PK profile of a single dose of the reference 4 mg daprodustat tablet manufactured via process 1. Enrolled participants were randomly assigned in a 1:1:1 ratio to one of three 4 mg daprodustat treatment sequences over 3 periods (Table 1). Study participants, site investigators, and the study sponsor were all blinded, and daprodustat was dispensed by qualified site personnel who were not involved in the conduct of the study. Participants were screened within 30 days prior to the first dose of daprodustat, and a follow-up visit occurred 7 days after the last dose administered. Participants in part A refrained from any food or drink, except water, for at least 10 hours before dosing and 4 hours after dosing with daprodustat. There was a minimum 7-day washout period between each treatment period. The study duration was ≈8 weeks for each participant in part A.
Part B was a randomized, double-blind, singledose, 2-period, crossover study in healthy individuals to determine if the daprodustat tablets made by the two different manufacturing processes (process 1 and process 2) with similar dissolution profiles were bioequivalent. For each of 5 dose strengths (1, 2, 4, 6, and 8 mg), participants were randomly assigned 1:1 to one of two treatment sequences, with each sequence consisting of 2 periods ( Table 1). As with part A, participants, investigators, and staff were blinded; qualified  personnel dispensing daprodustat were unblinded; and participants fasted for 10 hours before and 4 hours after daprodustat administration. Dose strengths were selected on the basis of clinical doses administered in the daprodustat clinical program using tablets made via process 1. As with part A, the evaluation of each dose strength included a screening visit, with a minimum 7-day washout between periods, and a posttreatment follow-up visit 7 days after the final dose of daprodustat. The study took place over a minimum of 7 weeks for each participant for part B.
Both parts of the study were conducted in accordance with Good Clinical Practice Guidelines and the 2013 Declaration of Helsinki, and approved by the local institutional review boards (IRBs) (IntegReview IRB, Austin, TX; Aspire IRB, Santee, CA; WCG IRB, Lenexa, KS; and Advarra, Columbia, MD). All participants provided written informed consent to participate in the study.

Study Population
Participants in the study were healthy volunteers between 18 and 50 years of age (inclusive) with body weight ≥45 kg and body mass index between 19.0 and 31.0 kg/m 2 (inclusive).
Exclusion criteria included a history of malignancy within two years or receiving treatment for cancer at screening (except for localized squamous or basal cell carcinoma of the skin treated ≥12 weeks prior to enrollment), necessary use of prescription or nonprescription drugs including vitamins or herbal and dietary supplements within 7 days or 5 half-lives prior to the first dose of daprodustat, regular use of known drugs of abuse, average weekly alcohol intake of >14 units (1 unit = 8 g of alcohol), or a positive laboratory confirmation for coronavirus disease 2019 . A participant could be enrolled in either part A or B of the study, but not both.

Blood Sample Collection and Analysis
Whole blood samples of ≈5 mL were collected for measurement of plasma daprodustat concentrations. Samples were collected before dosing and 0.5, 1, 2, 2.5, 3, 4, 6, 8, 12, and 24 hours after dosing and placed into dipotassium ethylenediaminetetraacetic acid collection tubes, inverted to mix with anticoagulant and centrifuged, and the supernatant plasma was transferred to Nunc tubes and shipped frozen to PPD, Inc. (Middleton, WI) for analysis.
Plasma samples for daprodustat were analyzed by PPD using a validated analytical method based on solid-phase extraction, followed by high-performance liquid chromatography-tandem mass spectrometric detection analysis using an electrospray interface operated in negative ion mode. A 100-μL aliquot of ethylenediaminetetraacetic acid plasma was extracted using solidphase extraction (Evolute ABN; Biotage, Uppsala, Sweden). The lower limit of quantification and higher limit of quantification were 0.100 and 100 ng/mL, respectively. The precision of the daprodustat assay was ≤15%, as indicated by the percent coefficient of variation (%CV); the accuracy of the assay was within ±15% of the actual value for daprodustat.

Pharmacokinetic Analysis
Plasma daprodustat data were collected as previously described 17,19 and calculated via standard noncompartmental analysis with WinNonlin 6.3 or higher (Certara, Princeton, NJ). All calculations were based on actual sampling times. AUC from time 0 to the time of the last quantifiable concentration (AUC 0-t ) and C max for daprodustat were calculated for parts A and B. AUC from time 0 extrapolated to infinity, t max , terminal halflife, apparent clearance, and apparent volume of distribution were also calculated for both parts of the study.

Statistical Analysis
Assuming a 15% dropout rate, it was estimated that 30 participants, 10 per treatment sequence, were needed for treatment comparisons in part A. Although there was no formal statistical hypothesis for part A, with 30 participants in a 3 × 3 crossover design, a withinparticipant %CV (%CVw) of 25% (based on C max having the highest CV), and assuming an observed ratio of 1.0, it was determined that the 90% confidence interval (CI) of the ratio for C max between process 2 dissolution 1 and 2 and process 1 should lie within the range of 0.80-1.25. Assuming a true ratio of 1.0, %CVw of 25% (based on C max having the highest CV), power of 97%, and a 10% dropout rate, it was estimated that 190 participants (38 participants per group and 19 per treatment sequence) were needed for part B. With these assumptions taken into consideration, it was determined that the 90% CI of the ratio for C max between manufacturing processes should lie within the range of 0.80-1.25. AUC 0-t and C max of daprodustat in both parts of the study were assessed separately using a mixed-effect model described by the following equation: where β0 was the intercept, γ i was the random participant effect for ith participant, τ j was the tablet manufacturing effect (with j accounting for process 1 and both dissolutions in part A, and both processes in part B), π k was the period effect (with k accounting for all periods within each study part), and εijkl was the random error.
A prespecified sensitivity analysis to jointly assess bioequivalence across all doses was performed in part B. The model incorporated all 5 doses within 1 model, and the specification was as follows: where β0 was the intercept of reference, β1 was the main effect for dose for reference (slope), β2 was the difference of intercept between the 2 formulations, β3 was the difference of slope between the 2 formulations, F was the manufacturing process, Si was the random effect for subject I, and εj was the random error.

Safety and Tolerability Measures
The safety assessments included monitoring of adverse events (AEs), adverse events of special interest (AESIs), and serious AEs throughout the dosing and followup periods. AESIs were defined for daprodustat on the basis of data from nonclinical and clinical studies, current information about hypoxia-inducible factorassociated pathophysiology, and identified risks for erythropoiesis-stimulating agents. They include thrombosis and/or tissue ischemia, death, myocardial infarction, stroke, heart failure, cardiomyopathy, pulmonary artery hypertension, cancer-related mortality, tumor progression or recurrence, esophageal and gastric erosions, proliferative retinopathy, macular edema, choroidal neovascularization, rheumatoid arthritis exacerbation, and worsening of hypertension. Physical examinations and electrocardiograms were performed at screening, and clinical laboratory tests and vital signs were obtained at screening before dosing and at follow-up.

Participant Disposition
Part A of the study enrolled 52 healthy volunteers. Thirty-three completed the study, 2 withdrew due to receiving positive COVID-19 tests, and 15 had to be replaced due to inclement weather. In part B, 207 participants were enrolled, and 199 completed the study. Eight participants withdrew (2 due to AEs, 3 due to physician decision, and 3 due to withdrawal by the participant). The demographic characteristics of the study populations for both study parts are shown in Table S1.

Pharmacokinetics
Part A. Upon oral administration of the 4 mg tablets, the mean plasma concentrations of daprodustat over a 24-hour time period were comparable in all 3 treatment groups ( Figure 1 and Table 2). In concordance with this observation, all 3 groups showed comparable values of AUC 0-t , C max , t max , and apparent volume of distribution, as shown in Table 2.
Statistical analyses of daprodustat plasma PK parameters for part A are summarized in Table 3.
Part B. Following single oral administration, daprodustat plasma concentrations over a 24-hour time period were comparable between the process 1 and process 2 treatment groups across all dosage strengths ( Figure 2 and Table 4). In concordance with this observation, the AUC 0-t values for tablets made via process 1 and process 2 were comparable for all dose levels. The C max and t max values for all dose levels made via process 1 and process 2 were also comparable.
Statistical analyses of daprodustat plasma PK parameters for part B are summarized in Table 5. The 90% CIs of the ratios of geometric least squares mean of AUC 0-t for process 2 compared to process 1 were within the 0.8-1.25 limit for indicating bioequivalence for all doses. The 90% CIs of the ratios of geometric least squares mean of C max for process 2 compared to process 1 for the 1, 2, 4, and the 6 mg doses were within the limit for indicating bioequivalence. The 8 g dose was not within the limit for indicating bioequivalence as the lower bound of 0.77 was outside of the 0.8 lower threshold.
The results of the prespecified sensitivity analysis performed for Part B are summarized in Table 6. The 90% CI of the ratios of geometric least squares mean for both AUC 0-t and C max across all doses, including the 8 mg dose, were within the 0.8-1.25 limit in the sensitivity analysis, supportive of bioequivalence.     AUC 0-t , AUC from time 0 to the time of the last quantifiable concentration; CI, confidence interval; C max , maximum concentration; Geo LSM, geometric least squares mean; P1, process 1; P2D1, process 2 dissolution 1; P2D2, process 2 dissolution 2; PK, pharmacokinetic. Process 1, twin-screw granulation; process 2, high-shear wet granulation. a %CVb was a post hoc analysis.

Safety
Treatment-emergent adverse events (TEAEs) are summarized in Table S2. In part A, the most frequently reported TEAE was positive COVID-19 test, reported by 2 participants (4%). All other TEAEs were reported by 1 participant each (2%). All TEAEs reported in part A were characterized as mild. In part B, the most frequently reported TEAE was headache, reported by 7 participants overall. All AEs reported in part B were described as either mild or moderate. There were no deaths reported during the study. There were no serious AEs during part A, and only 1 reported in part B (a participant sustained a spontaneous abortion 21 days after administration of daprodustat). There were no AESIs in part A; however, increased blood pressure was reported in 2 participants in part B (7 and 15 days after the final dose of daprodustat was administered). Investigators determined both events to be unrelated to daprodustat.
Two participants in part A discontinued the study due to positive COVID-19 tests. In part B, 1 participant discontinued due to elevated aspartate aminotransferase and alanine aminotransferase and 1 participant discontinued due to a gout flare. None of the AEs that led to discontinuation were determined to be related to daprodustat by investigators.   %CVb, between-participant coefficient of variation; %CVw, within-participant coefficient of variation; AUC, area under the concentration-time curve; CI, confidence interval; C max , maximum concentration; Geo LSM, geometric least squares mean; PK, pharmacokinetic. Process 1, twin-screw granulation; process 2, high-shear wet granulation. a %CVb was a post hoc analysis.

Discussion
To understand the clinical relevance of tablet dissolution, part A of this study compared 4 mg daprodustat tablets manufactured via process 2 with two successively slower dissolution profiles relative to daprodustat manufactured via process 1. Part B, a larger study, sought to establish bioequivalence between daprodustat tablets at 5 different dose strengths made by these two manufacturing processes.
In part A, exposure of daprodustat 4 mg was comparable for process 1 and process 2 for both dissolution profiles (Table 2). Thus, the rate of dissolution did not appear to impact overall exposure.
In part B, the 90% CIs of the ratios of least square means comparing AUC 0-t for the two manufacturing processes fell within the 0.8-1.25 limit at all dose levels assessed (1, 2, 4, 6, and 8 mg). Although the 90% CI for the ratios comparing C max for the 8 mg tablet were not within the 0.8-1.25 limit in the primary analysis, a post hoc analysis showed the variability between participants observed at 8 mg was the highest of all doses (%CVb, 38.9). Additionally, the study was powered for 25% variability within participants, which was exceeded by the 2, 4, 6, and 8 mg doses (C max %CVw, 28.8, 30.9, 26.7, and 27.6, respectively). In the prespecified sensitivity analysis, the 90% CIs of the ratios of least squares means comparing AUC 0-t and C max for process 1 and process 2 fell within the 0.8-1.25 limit for bioequivalence for all dose levels, including the 8 mg tablet. The results of the prespecified sensitivity analysis, which incorporated all doses, treatment groups, and interactions between doses and treatment groups in a mixed-effect model, indicated that the bioequivalence standards were not met due to the high variability observed in the 8 mg dose group. It should be noted that the high variability observed in C max for 8 mg is not expected to significantly affect the efficacy of daprodustat given that individualized treatment regimens are required. 4 No new safety concerns were identified for daprodustat with any of the dose levels or with the new manufacturing process.

Conclusion
Results from part A showed that dissolution profile differences had no relevant impact on exposure of orally administered daprodustat. Part B demonstrated that bioequivalence was established in all doses for AUC, and all doses except 8 mg for C max . Bioequivalence was further demonstrated in all doses for C max between process 1 and process 2 in the prespecified sensitivity analysis, which jointly assessed all doses tested.