Pharmacokinetics of fluticasone furoate, umeclidinium, and vilanterol as a triple therapy in healthy volunteers

Objective: Two single-center, four-way, single-dose, crossover studies assessed the systemic exposure, systemic pharmacodynamics (PD), and safety profile of the closed triple fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) therapy compared with dual therapies. These are the first studies where pharmacokinetic (PK) profile assessment was possible for this inhaled triple fixed-dose combination product. Methods: Healthy volunteers were randomized to receive 4 consecutive inhalations (each administered as a single dose) via a single ELLIPTA® dry powder inhaler: in study 1 (CTT116415/NCT01691547), FF/UMEC/VI at total doses of 400/500/100 μg, FF/UMEC 400/500 μg, UMEC/VI 500/100 μg, or FF/VI 400/100 μg; in study 2 (200587/NCT01894386), FF/UMEC/VI at total doses of 400/500/100 μg or 400/250/100 μg, FF/VI 400/100 μg, or UMEC/VI 250/100 μg. PK and PD parameters and safety were assessed. Results: Of 88 subjects, 95% completed both studies and received all planned treatments. Total systemic exposure was similar for FF, UMEC, and VI when administered as a triple therapy compared with FF/VI and UMEC/VI. No clinically significant systemic PD findings were detected. The incidence of adverse events was low and similar across treatment arms. Conclusions: Systemic exposure to all three components of the closed triple therapy, following single-dose delivery, was similar to that seen with the dual therapies FF/VI and UMEC/VI. The delivered lung dose and safety profile of all three agents, delivered via a single inhaler, are expected to be similar to those of the dual therapies.


Introduction
Patients with chronic obstructive pulmonary disease (COPD) experience increasing airflow obstruction and enhanced chronic inflammation within the respiratory system. Current guidelines recommend an incremental approach to treatment involving the combination of different drug classes as COPD severity progresses [1].
Triple combination therapy is increasingly used in COPD management [2,3]. An inhaled corticosteroid (ICS)/long-acting muscarinic receptor antagonist (LAMA)/long-acting β 2adrenergic receptor agonist (LABA) combination is now a recommended second option for symptomatic COPD patients at high risk of exacerbations (GOLD grade D) [1]. Triple therapy, when administered as a once-daily single-dose combination from one inhaler, offers improved convenience of administration and a reduced burden of polypharmacy for patients with the potential for improved compliance compared to individual agents from separate inhalers.
Fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI), also known as the Closed Triple, is a once-daily ICS/LAMA/LABA therapy in phase III development in patients with COPD, for administration via the singlestep activation ELLIPTA ® (ELLIPTA ® is a trade mark of the GSK group of companies) dry powder inhaler (DPI). Fixed-dose combinations of FF/VI and UMEC/VI, delivered via the ELLIPTA DPI, are approved in the United States (US), European Union, and Canada, among other countries/regions, for the treatment of moderate-severe COPD.
Here, we present findings from the first two single-dose, healthy volunteer studies to assess pharmacokinetics (PK) of FF, UMEC, and VI administered as a closed triple therapy in a single inhaler compared with the same components administered as dual combinations.
For both studies, the primary objective was to assess the systemic exposure of FF, UMEC, and VI following single inhaled doses (4 inhalations) of FF/UMEC/VI, UMEC/ VI, or FF/VI, including (in study 2 only) the assessment of 2 UMEC dose levels (500 and 250 μg). Secondary objectives assessed systemic exposure of FF/UMEC (study 1 only), systemic pharmacodynamics (PD; study 1 only), and urine PK of UMEC (study 1 only).

Pharmacokinetic analysis
In study 1, plasma PK samples were collected predose and at 5, 15, and 30 minutes and 1, 2, 4, 6, 8, 12, 16, 24, 36, and 48 hours postdose, and urine PK samples were taken from 0 to 6, 6 to 10, 10 to 14, 14 to 18, and 18 to 24 hours. Urine was collected in study 1 only. To better characterize the concentration-time profile of VI within the first 15 minutes postdose in study 2, additional plasma PK samples were collected at 3, 7, 10, 12, and 45 minutes and 1.5 hours postdose. Samples were not collected at 16, 36, or 48 hours postdose in study 2. Plasma PK samples were obtained from 5 to 6 mL blood samples. Analysis of plasma and urine samples was performed by York Bioanalytical Solutions (managed by Bioanalytical Science and Toxicokinetics, Drug Metabolism and Pharmacokinetics, GSK). Analysis of FF, UMEC, and VI levels was conducted using validated analytical techniques (online supplement) with 10 pg/mL as the lower limit of quantification for each compound.

Pharmacodynamic analysis (study 1 only)
Electrocardiograms (ECGs) were assessed from screening and predose on the 1 st day of each treatment period, then at 5, 15, and 30 minutes and 1, 2, 4, 12, and 24 hours postdose. Heart rate was assessed from 0 to 4 hours, and blood samples to assess systemic PD of VI (serum potassium, and blood glucose) were also collected predose and at 30 minutes, 1, 2, and 4 hours postdose, then assayed using approved analytical methodology (Supplementary material).

Safety analysis
Adverse events (AEs), serious AEs (SAEs), laboratory values, vital signs, ECGs, and physical examinations were monitored for safety assessment. AEs were coded using the Medical Dictionary for Regulatory Activities.
No formal hypothesis testing was conducted for either study as an estimation approach was adopted. Data were reported using statistical analysis software (SAS, Cary, NC, USA) version 9.1.3. on a UNIX platform. Log e -transformed AUC (0-t) , and C max were analyzed separately using a mixedeffects model, with period and treatment fitted as fixed effects and subject as a random effect. For comparison of the 2 UMEC dose levels, UMEC PK parameters were dose normalized prior to analysis by dividing the respective PK parameter values by the received planned dose. For each comparison of interest, point estimates and corresponding 90% confidence intervals (CIs) were calculated for the ratio (for PK) or difference (for PD) between the means of the test and reference treatments. Although the studies were not conducted as formal bioequivalence studies, treatments were considered comparable if the 90% CIs of the ratio between the triple and dual therapies were within 0.8 -1.25 (standard bioequivalence acceptance limits). Where small differences were seen, these were further assessed for their clinical relevance. Each derived PD endpoint was analyzed using a mixed effects model.
Planned enrollment for each study was 44 subjects, with the target of ≥ 36 subjects completing dosing and critical assessments. See Supplemental material for sample size assumptions.
All subjects who received ≥ 1 dose of study medication were included in the "all subjects" population and used for the PD and safety data. The "PK population" was defined as subjects within the "all subjects" population for whom a PK sample was obtained and analyzed.

Posthoc exploratory analyses
Three posthoc exploratory analyses were performed. The first two analyses, described below, investigated the hypothesis that VI PK may not have been optimally defined in study 1, due to a low number of time points in the sampling schedule. The 3 rd analysis, presented in the Supplemental material, was undertaken to support future development of VI as a monotherapy.

Population pharmacokinetic modeling (study 1)
Posthoc population PK modeling of logtransformed VI concentration-time data was performed using a three-compartment population PK model to investigate the impact of the PK sampling schedule on VI C max . Based on a previous VI population PK modeling analysis (data on file), additional time points were included at 1 minute intervals between 2 and 14 minutes to improve C max characterization. Individual AUC (0-24h) was derived as the ratio of the nominal dose divided by the individual posthoc estimate of apparent clearance from the final population PK model. Individual C max was derived from the predicted concentration-time profile obtained from the model for each subject.

Exploratory pharmacokinetic analysis (study 2)
Posthoc exploratory PK analysis was performed to repeat the PK analysis from study 2, with the following PK time points removed: 3, 7, 10, 12, 45, and 90 minutes. In addition, FF and VI PK parameters were compared between studies 1 and 2 to assess FF/UMEC(500)/VI with FF/VI. PK parameters were analyzed using a mixed-effects model, as described in the original analysis.

Results
Subject disposition and demographic characteristics (Table 1) In study 1, 44 subjects were randomized and included in the "all subjects" population: 93% (n = 41) completed the study and  received all planned treatments; three subjects discontinued study medication due to an SAE (n = 1) or investigator discretion (n = 2). The PK population comprised 98% of subjects (n = 43).
In study 2, 44 subjects were randomized and included in the "all subjects" population: 98% (n = 43) completed the study; 1 subject withdrew due to a protocol violation. All subjects were included in the PK population.
The demographic characteristics were similar for each study.

Pharmacokinetics
PK parameters are reported for FF (Table  2a, b), UMEC (Table 3a, b), and VI (Table  4a, b). Table 5a and Table 5b provide the statistical comparisons of systemic exposure between the triple and dual therapies.

Fluticasone furoate
FF was quantifiable from 3 to 12 minutes postdose, dependent on therapy and study, and in the majority of subjects ≤ 24 hours postdose ( Figure 1A, 1B).
In both studies, similar FF systemic exposure (AUC and C max ) was observed between triple therapy and the FF/VI and FF/ UMEC dual therapies, except for a slightly higher C max in study 1 where the triple vs. FF/VI ratio was 1.23 (90% CI, 1.14 -1.33; Table 5a). This difference was considered to be clinically nonrelevant. In study 2, there were no observed differences in FF systemic exposure (C max and AUC) between the two triple combinations (500 and 250 μg UMEC).
In study 1, similar UMEC systemic exposure (AUC and C max ) was observed between triple therapy and the UMEC/VI and FF/UMEC dual therapies. UMEC urine PK parameters, such as the percent of dose excreted in urine over 24 hours (~2%), urinary half-life (13 -16 hours), and renal clearance (12 L/h), were also similar between the triple and both dual therapies. In study 2, UMEC systemic exposure (C max and AUC) in-  creased in proportion with increasing UMEC dose and was similar between triple and dual therapies for all parameters.

Vilanterol
VI was quantifiable at 5 minutes (study 1) and 3 minutes (study 2) postdose, and very few subjects showed detectable VI at 24 hours postdose following triple, FF/VI, and UMEC/VI therapies. Similar VI systemic exposures, in terms of AUC, were observed between the triple and FF/VI or UMEC/VI therapies (Table 5a). C max was slightly higher for VI when administered as the triple therapy vs. FF/VI (1.46 [90% CI, 1.38 -1.56]) and as the triple therapy vs. UMEC/VI (1.33 [90% CI, 1.25 -1.42]), but this was not considered to be clinically significant (Table 5a).
In study 2, VI systemic exposure (AUC and C max ) was similar between the triple and FF/VI ( Figure 3A), and between the two triple therapies (500 and 250 μg UMEC; Figure 3B). VI systemic exposure was also similar in terms of AUC between the triple and UMEC/VI; however, C max was slightly higher following the triple therapy compared to UMEC/VI (1.21 [90% CI, 1.14 -1.27]; Table  5b). This difference was not considered to be clinically significant.

Pharmacodynamics (study 1 only; Table 6)
There were no differences in heart rate for FF/UMEC/VI compared with either UMEC/VI or FF/VI. Maximum and weighted mean change (0 -4 hours) from baseline heart rate increased for FF/UMEC/VI vs. FF/ UMEC (mean difference ~ 14 and ~ 6 bpm, respectively; data not shown). These differences were not considered to be clinically significant. No clinically relevant difference in weighted mean change from baseline potassium level was observed for any triple vs. dual comparison. Weighted mean changes from baseline in blood glucose levels did not differ for FF/UMEC/VI vs. either UMEC/ VI or FF/VI. A small difference in weighted mean change from baseline in blood glucose was reported with FF/UMEC compared with the triple therapy (~ 0.5 and 0.15 mmol/L, respectively; data not shown). These small increases, which are not considered to be clinically significant, can be attributed to the VI component and are similar to those seen in previous studies with FF/VI (800/100 μg) [4].

Safety
Few AEs were reported (Table 7), and none were deemed by the investigator to be related to study treatment. No notable differences in AE incidence were observed between treatment groups. One SAE of diabetes mellitus was reported (study 1), which, upon further investigation, was established to be a previously undiagnosed case of type 2 diabetes mellitus.

Posthoc exploratory analyses
PK population pharmacokinetic modeling (study 1; Supplementary  Table 1) Posthoc PK data were modeled with fixed input times ranging from 2 to 12 minutes. Mean VI C max was estimated to range from 620 to 690 pg/mL, with maximum C max values observed for input times of 7 -9 minutes.

Exploratory pharmacokinetic analysis (study 2; Supplementary Table 2)
In the posthoc exploratory analysis of study 2 PK data, there was a greater difference in VI C max for the triple vs. dual therapies when the additional time points were removed, compared with the full study 2 data.

Discussion
These were the first studies to assess the PK of a new, closed, triple (FF/UMEC/ VI) fixed-dose combination therapy vs. the dual therapies of FF/VI, UMEC/VI, and FF/ UMEC in healthy volunteers. Overall results from the two studies suggest no clinically relevant differences in systemic exposure of FF, UMEC, or VI when administered as a triple vs. dual therapy.
In study 1, an increase in FF C max was reported with the triple therapy vs. FF/VI. This may reflect the small differences in absorption kinetics and be due to the t max difference for FF between the triple therapy (0.23 hours) and FF/VI (1.0 hours). No difference in FF systemic exposure was observed in study 2 when FF was administered in the triple therapy with 2 different UMEC doses.
VI C max was slightly higher in study 1 following triple therapy compared with either FF/VI or UMEC/VI. Assessment of individual VI time profiles in study 1 suggested that C max may have occurred earlier for the triple combination (~ 5 minutes), whereas C max for FF/VI and UMEC/VI may have occurred later, between the 5 and 15-minute time points. Hence, the C max value recorded for VI administered as a dual therapy may represent an underestimate in study 1. It is therefore possible that the small differences seen between the VI C max of the triple combination and FF/VI and UMEC/VI may represent an apparent increase in the ratios of the respective VI C max comparisons.
The VI PK findings from both studies were therefore investigated in two posthoc exploratory analyses. The outcome of these concluded that the PK differences for VI C max were most likely influenced by the low number of time points ~ 7 -9 minutes in the sampling schedule. Additional time points were therefore included in study 2 to en-  sure more accurate characterization of C max . There was no observed difference in VI systemic exposure (C max and AUC) in study 2 between FF/UMEC/VI and FF/VI; however, VI C max following triple therapy was slightly higher compared with UMEC(250)/VI. This increase was less pronounced than that observed in study 1, and the difference is not considered clinically significant as the overall systemic exposure (in terms of AUC) was similar between the triple and dual UMEC/ VI therapies.
Where differences in C max were observed for VI, these were not associated with any clinically relevant differences in the corresponding pharmacologically-related PD endpoints. This is despite the doses in this study being 4 times the proposed therapeutic dose, which represents a twelve-fold higher systemic exposure in healthy subjects than seen in patients with COPD (BREO ® ELLIPTA ® Prescribing Information, GSK [5]).

Conclusion
Overall results from the two studies suggest no clinically relevant differences in systemic exposure of FF, UMEC, or VI when   a Subjects who experienced at least 1 AE. b In 1 subject, 1 of these AEs was identified as SAE (type 2 diabetes mellitus) resulting in the subject being withdrawn from the study. AE = adverse event; SAE = serious adverse event.

Supplementary information Pharmacodynamic analysis (study 1)
Each derived endpoint was assessed using a mixed-effect model that included treatment and period as fixed effects, period-level and subject-level baseline as continuous covariates, and subject as a random effect.

Statistical analyses: sample size assumptions (studies 1 and 2)
The sample size calculation for both studies was based on the largest available within-subject standard deviation (SD) for C max and AUC across the three components (FF, UMEC, VI) from previous studies. The largest available within-subject SD was 0.33 for study 1 and 0.34 for study 2 (based on study 1 data).
Assuming a target sample size of 36, precision for the comparison of interest was estimated to be within 14% of the observed point estimate for both studies. Calculations were based on a symmetric two-tailed test procedure on the log e -scale and a type I error rate of 10%. Precision is expressed as the half-width of the 90% CI, and the lower and upper bounds of the 90% CI for the ratio would be within 14% of the observed ratio.
Although the studies were an estimation-based approach, and no formal statistical hypotheses were to be tested, a sample size of 36 would provide sufficient power to

Analysis of VI pharmacokinetic parameters (studies 1 and 2)
In order to support the development of VI as a monotherapy, a third posthoc analysis investigated VI systemic exposure across treatment groups for the dual therapies FF/ VI 400/100 μg vs. UMEC/VI 500/100 μg in studies 1 and 2. The concentration-time data were analyzed by noncompartmental methods. C max and either AUC (0-2) (study 1) or AUC (0-6) (study 2) were analyzed separately using a mixed-effects model, with period and treatment as fixed effects and subject as a random effect.