Study design
This was a randomized, double-blinded, dose-ranging, placebo-controlled, parallel group study. The study consisted of a 28-day screening period, a 29-day treatment period (with study visits on days 1, 15, and 29), and a follow-up visit approximately 14 days after completing treatment. For patients on dialysis TIW, the day 1, 15, and 29 study visits could not occur on the first dialysis session following the longest intradialytic interval.
The dose-response relationship for once-daily dosing of daprodustat in the hemodialysis population was established in a 24-week phase 2 study (
ClinicalTrials.gov identifier NCT01977482) [
9]. In that study, the hemoglobin response (change from baseline) was determined at 4 weeks over the dose range of 4 mg to 12 mg; therefore, in order to make a direct comparison between the once-daily and TIW regimens, a 29-day treatment period was utilized.
The study was initiated on February 17, 2016 and completed on January 25, 2017. Twenty-nine centers in 5 countries randomized at least one participant: 8 centers in the Russian Federation, 7 each in Spain and the United States, 4 in Canada, and 3 in Germany.
Study participants were stratified by low or high prior dose of rhEPO (or its analogs) based on the average weekly dose during the 12 weeks prior to randomization (day 1). Low dose was defined as < 100 IU/kg/week epoetin or < 0.5 μg/kg/week darbepoetin. High dose was defined as ≥100 IU/kg/week epoetin or ≥ 0.5 μg/kg/week darbepoetin. Participants were randomized in equal proportions to a daprodustat dose or placebo within each stratum without using prior rhEPO dose to determine the daprodustat dose.
Participants discontinued their current rhEPO therapy prior to day 1 and were randomized by a computer-generated randomization schedule using the Prism randomization system in a 1:1:1:1:1 ratio to receive oral daprodustat 10, 15, 25, or 30 mg or placebo TIW on a dialysis day. Evidence with epoetin alfa (Epogen), has shown that increases in dosing interval can be achieved by a proportionate increase in the dose administered at each interval [
11]. As the daprodustat dose response with hemoglobin was generally linear between 4 and 24 mg once daily, it was anticipated that a similar time-proportional efficacy relationship would exist for daprodustat and the management of hemoglobin levels would remain as the dosing was extended from once every day to three times weekly. Based on the hypothesis that the total weekly TIW dose that achieves the same total weekly once-daily dose would have the same Hgb effect, the TIW dose would be equivalent to ~ 2.3 times the once-daily dose (7 doses/ week × 1 week/ 3 doses = 2.3 multiplier), doses of 10, 15, 25, and 30 mg TIW were investigated for 29 days to most closely match the 4, 6, 10, and 12 mg once-daily doses that were evaluated in the study by Meadowcroft et al. [
9]. Studying multiple doses allowed a robust analysis of a range of TIW doses to determine the “equivalent” ratio to bridge to once-daily dose levels. The dose-hemoglobin response relationship was generally linear in the once-daily dose range of interest (4 to 24 mg) but it was not known if this was maintained with the higher doses needed for TIW dosing.
Each participant was assigned a randomization number by the interactive voice and web response system, which was managed by Pharmaceutical Product Development, LLC, Wilmington, NC, USA. Once a randomization number was assigned, it could not be reassigned. Study participants, investigators, and site and sponsor staff were all blinded to the treatment assignment. To maintain the blind, each participant took one tablet of study treatment from each of three bottles of study treatment, on each dosing day. The combination of tablets provided the appropriate dose of study treatment, and included a combination of tablet strengths, active and placebo, to correspond to the appropriate dose. An internal Safety Review Team (SRT), which included GlaxoSmithKline personnel involved in the conduct of the study, monitored blinded safety data in-stream during the study.
Study population
ndividuals who were at least 18 years old and had stable baseline hemoglobin values between 9.0 g/dL and 11.5 g/dL based on average values obtained on day − 28, day − 14, and day 1 were eligible for the study. Study participants were also required to be on a stable hemodialysis regimen of 3 to 5 times weekly for at least 90 days, with a single-pool Kt/Vurea of ≥1.2. Additional eligibility criteria included that participants be on a stable dose of rhEPO or its analogs, with the total weekly dose varying by no more than 50%, during the 4 weeks prior to day − 28. A protocol amendment on May 11, 2016 updated the eligibility criteria to include the complete required timeframe for participants to be on a stable dose of rhEPO or its analogs up to 4 weeks prior to day − 28 through day 1. Those participants who were receiving maintenance oral or intravenous (IV) iron supplementation were also required to be on a stable dose from 4 weeks prior to day − 28 through day 29.
Patients were ineligible for the study if they were being treated with ≥360 IU/kg/week IV or ≥ 250 IU/kg/week subcutaneous (SC) epoetin, ≥1.8 μg/kg/week IV or SC darbepoetin, or ≥ 2.2 μg/kg/week methoxy polyethylene glycol-epoetin beta within the 8 weeks prior to randomization. Patients with low values of vitamin B12 (< 200 pg/mL [148 pmol/L]), folate (< 2.0 ng/mL [4.5 nmol/L]), ferritin (< 100 ng/mL [225 pmol/L]), or transferrin saturation (TSAT) (< 20%) were also ineligible. Additional exclusion criteria included a planned change in dialysis modality or renal transplant within the study time period; significant cardiovascular, hematologic, hepatic, or chronic inflammatory disease; active gastrointestinal (GI) bleeding or GI bleeding within 8 weeks prior to day − 28 through day 1; QT interval corrected for heart rate using Bazett’s formula (QTcB) > 500 msec or QTcB > 530 msec in patients with bundle branch block; and malignancy within 2 years of screening, currently receiving treatment for cancer, or a known complex kidney cyst (ie, Bosniak Category II F, III, or IV) > 3 cm.
Withdrawal criteria
Study medication was permanently discontinued if participants met the hemoglobin stopping criteria, based on HemoCue measured hemoglobin values as follows: a confirmed hemoglobin < 7.5 g/dL, a confirmed hemoglobin increase ≥1.0 g/dL over the previous 2 weeks, or a confirmed hemoglobin decrease ≥2.0 g/dL over the previous 2 weeks based on point of care hemoglobin measurements. For confirmed hemoglobin ≥13.0 g/dL, a decision regarding withdrawal was made in consultation with the GlaxoSmithKline Medical Monitor. Central lab hemoglobin values were also analyzed retrospectively to identify how many participants met each of the hemoglobin stopping criteria.
Participants were also permanently discontinued from study treatment if they received a blood transfusion or kidney transplant, became or intended to become pregnant, had active GI bleeding, were diagnosed with cancer (with the exception of localized squamous cell and basal cell carcinoma), or missed 2 consecutive dialysis sessions. Liver chemistry was also monitored during the study.
Participants who discontinued study treatment completed an early withdrawal assessment and follow-up visit 14 ± 3 days after the last dose of study treatment.
Study analyses
No formal sample size or power calculations were performed. The focus of this study was on characterizing the hemoglobin-TIW dose-response relationship and using the results to estimate the dose conversion ratio between once-daily and TIW doses that would produce the same hemoglobin response, not on formal hypothesis testing. A sample size of approximately 90 randomized participants was determined based on feasibility (18 per treatment group).
The primary study endpoint was change in hemoglobin from baseline at day 29. Baseline hemoglobin was defined as the average of the hemoglobin measured at the day − 14 and day 1 visits. This change from the planned analysis, which defined baseline hemoglobin as the day 1 hemoglobin value, was made in order to reflect the overall pretreatment stable hemoglobin value and was prespecified in the Reporting and Analysis Plan, which was finalized before database freeze and unblinding.
The dose-response relationship between daprodustat TIW and hemoglobin at day 29 was characterized using a three-parameter E
max model fitted using Bayesian statistical methods. E
max was assumed to be an appropriate model based on the biology of daprodustat and once-daily dose-response modeling completed to date for daprodustat. Prior distributions for each of the model parameters (E0, E
max, ED50, and variance) used in this trial were informed by the once-daily dose-response from a previous phase 2 study (
ClinicalTrials.gov identifier NCT01977482) [
9] This modeling provided confidence in setting prior distributions for E0, however, there was greater uncertainty in the estimation of E
max and ED50, which was explored in sensitivity analyses with alternative (weaker) prior distributions, including a Bayesian three-parameter E
max model based on the completer population, Bayesian three-parameter E
max model using sensitivity (weaker, less informative) priors, and linear dose-response model based on the intention-to-treat (ITT) population.
Once the dose-response model was fitted satisfactorily, the posterior distribution of the model parameters (E0, Emax, ED50, and variance) was summarized through means, standard deviation (SD), median, Monte Carlo standard error (MCSE), MCSE/SD and equal-sided 95% credibility intervals. The predicted value of hemoglobin change from baseline at day 29 at a series of doses (range of possible doses from 0 to 50 was sectioned off into a sequence of 0.5 increments; ie, dose levels of 0, 0.5, 1.0, 1.5, 2.0….50) was also monitored and the median and 95% credibility values obtained at each dose.
The observed change in hemoglobin from baseline at day 29 was used as the response variable in the Emax model. Participants who had a day 15 hemoglobin measurement, but a missing day 29 hemoglobin measurement, were included with a day 29 value imputed using 2 × (day 15 hemoglobin – baseline hemoglobin). Participants who prematurely discontinued dosing prior to day 15 were nonevaluable for the dose-response analysis.
The dose conversion ratio of the once-daily dose to the corresponding TIW dose that produced an equivalent hemoglobin response was estimated based on the dose-hemoglobin response relationship for the once-daily regimen derived from a previous study [
9] and for the TIW regimen derived from the current trial. TIW dosing and once-daily dosing were both estimated by Bayesian E
max model.
The observed change in hemoglobin from baseline to day 29 was analyzed by a repeated measures model using the PROC MIXED procedure. Model effects included baseline hemoglobin, treatment, time, and treatment-by-time interactions. The variance-covariance structures for repeated measures within the same subject were unstructured. For day 29, least squares mean differences were calculated as estimates of the difference between the daprodustat cohorts and placebo, together with 95% confidence intervals (CIs).
Additional secondary endpoints included observed change from baseline in hepcidin, ferritin, transferrin, total iron, unsaturated iron binding capacity (UIBC), total iron binding capacity (TIBC), hematocrit, red blood cell (RBC) count, reticulocyte count, and reticulocyte hemoglobin (CHr), which were summarized by visit (day 1, 15, and 29) and treatment group. Additionally, maximum observed change from baseline in plasma EPO and maximum observed percent change from baseline in VEGF were assessed and summarized by treatment group. Descriptive PK summaries of daprodustat and major metabolites were also summarized by daprodustat dose level. In general, secondary analyses were descriptive and estimation based.
Safety assessments included incidence and severity of AEs and serious AEs (SAEs), reasons for discontinuation of investigational product, discontinuations for safety-related reasons, and absolute values and changes from baseline over time in laboratory parameters, electrocardiograms (ECGs), and vital signs. Dose-dependent AE relationship was not assessed in this study. The investigator or site staff was responsible for detecting, documenting, and reporting AEs/SAEs. AEs were collected from the start of study treatment (day 1) and until the follow-up visit. SAEs were recorded from the time of subject consent to participate in the study up to and including any follow-up contact.