Ethnic sensitivity assessment of fluticasone furoate/vilanterol in East Asian asthma patients from randomized double-blind multicentre Phase IIb/III trials

Efficacy data are presented from three Phase III studies that included asthma patients in Japan: HZA106827 (Clinicaltrials.gov registration number NCT01165138) [4]; HZA106829 (NCT01134042) [5]; HZA106837 (NCT01086384) [13] (Additional File 1). FF/VI 100/25 μg OD and FF 100 μg OD were investigated in studies HZA106827 (placebo-controlled) and HZA106837. FF/VI 200/25 μg OD, FF 200 μg OD, and fluticasone propionate (FP) 500 μg twice daily (BD) were investigated in study HZA106829. FF/VI and placebo were delivered via the ELLIPTA DPI and FP was delivered via the DISKUS® (DISKUS® is a trademark of the GSK group of companies) DPI. A range of efficacy endpoints were assessed in the individual studies, but only trough FEV₁ measured at the end of a dosing interval (the primary endpoint in studies HZA106827 and HZA106829 and a secondary endpoint in study HZA106837) was analyzed in the Japan subgroup, as pre-specified in the analysis plan. Efficacy data were analyzed for patients recruited from Japan (Japan cohort), all patients that were not recruited from Japan (Not-Japan cohort), and all patients studied (Overall population). Patients were assigned to these cohorts according to the country in which they were recruited and not according to race. Ease of use of the ELLIPTA inhaler was also assessed by questionnaire in study HZA106827.

Change from baseline in trough FEV₁ at Week 12 was pooled for each treatment arm from studies HZA106827 and HZA106837. Statistical analyses for treatment comparisons at Week 12 were conducted for FF/VI 100/25 μg versus placebo, FF/VI 100/25 μg versus FF 100 μg, and FF 100 μg versus placebo. In the individual studies included in this analysis, sensitivity analyses (using repeated measures analyses) supported the primary analysis using last observation carried forward (LOCF) data for imputation of missing data. Consequently, LOCF data were used to perform the current statistical analysis using an analysis of covariance (ANCOVA) model. Effects due to geographic region, baseline FEV₁, gender, age, treatment group, and study were modeled. An interaction term for treatment by geographic region (Japan/Not-Japan) was also included in the model. A similar modeling process was used to analyze the results of the FP 500 μg BD-controlled study HZA106829, in which FF/VI 200/25 μg and FF 200 μg were assessed at Weeks 12 (post-hoc analysis) and 24, using last observation carried forward data.

Withdrawals due to lack of efficacy were summarized for each cohort in the efficacy dataset.

Safety data were pooled from five clinical studies: the three studies included in the efficacy assessment and an additional two placebo-controlled FF Phase IIb studies that included East Asian patients in Korea: FFA109685 (NCT00603278) [14]; FFA109687 (NCT00603382) [15] (Additional File 2). These two studies were included in the integrated safety analysis to increase the number of East Asian patients assessed and thereby increase confidence in the results. There are similarities in intrinsic and extrinsic ethnic factors in the populations of Japan and Korea [16‐18], supporting the pooling of patient results in these two East Asian populations, which has increased the number of East Asian patients assessed and thereby increased confidence in the conclusions that can be drawn.

All five studies assessed a comprehensive range of safety endpoints, which are presented in the relevant publications; however, for brevity we only assessed key safety data in the Japan subgroup, as pre-specified in the analysis plan. Safety data were compared between treatment arms for patients recruited from Japan and Korea (Japan + Korea cohort), patients recruited from all other countries (Not-Japan + Korea cohort), and all patients studied (Overall population). The safety results are considered for OD placebo, FF/VI 100/25 μg, FF/VI 200/25 μg, FF 100 μg, and FF 200 μg from studies HZA106827, HZA106837, HZA106829, FFA109685, and FFA109687. Results for the doses of FF monotherapy that were not progressed to Phase III for FF/VI from these five studies are not presented. FP 100 μg BD, FP 250 μg BD, and FP 500 μg BD were also included in the analysis of urinary cortisol data.

The incidence of all adverse events (AEs), treatment-related AEs, and serious AEs (SAEs), including exacerbations, was summarized for patients in Japan + Korea, Not-Japan + Korea, and the Overall population of the pooled data set. Electrocardiogram (ECG) results including QT corrected using Frederica’s correction (QTcF) were also assessed in a subset of patients from HZA106827 and HZA106829. AEs related to known ICS or LABA class effects were a focus, including lower respiratory tract infections (bronchitis and pneumonia), cardiovascular effects (heart rate and ECG), and effects on glucose and potassium. An Asthma Composite Endpoint was also derived to assess independently adjudicated asthma-related hospitalizations, intubations, and deaths in studies HZA106837, HZA106827, and HZA106829.

The urinary cortisol excretion over 24 h was measured at baseline and at the end of the treatment period in a subset of patients from studies FFA109685, FFA109687, HZA106827, and HZA106829 (Urinary Cortisol population). For treatment comparisons, the urinary cortisol excretion ratio (end of treatment/baseline) was analyzed using an ANCOVA model of the log-transformed urinary cortisol excretion with effects due to geographic region, log baseline urinary cortisol, gender, age, treatment group, and study. For the analysis by geographic region and for estimation of regional treatment differences, an interaction term for treatment by geographic region (Japan + Korea/Not-Japan + Korea) was also included in the model.

Withdrawals due to AEs were summarized for each cohort in the safety dataset.

The intent-to-treat populations in the three efficacy studies comprised 3,214 patients recruited in 11 countries (overall population); including 148 patients (5 %) recruited from Japan (Additional File 1). Eighty-five percent of patients in the Overall population and 70 % of the Japan cohort completed the respective study. A higher proportion of patients recruited from Japan (14 %; N = 21/148) than Not-Japan (3 %; N = 104/3,066) withdrew due to lack of efficacy. In the Japan cohort, a greater proportion of patients withdraw due to lack of efficacy with placebo (42 %) than with FF/VI (7–9 %). A similar pattern was observed in the Not-Japan group (placebo 13 %, FF/VI 1–3 %).

Demographic and baseline characteristics for the Japan and Not-Japan patients are presented in Table 1. All patients recruited from Japan were of East Asian ancestry and patients from the Not-Japan cohort were principally White/Caucasian (81 %), but included small proportions of patients of a range of geographic ancestries (Additional File 1). Mean baseline pre-bronchodilator FEV₁ was lower in patients recruited from Japan than Not-Japan, consistent with the literature and the lower average height of the Japan patients [19, 20]. Furthermore, patients recruited from Japan had greater mean % predicted FEV₁ values than the Not-Japan patients.

At Week 12, increases from baseline in trough FEV₁ were observed for FF/VI 100/25 μg, FF/VI 200/25 μg, and all active treatments in the Japan and Not-Japan patients (Tables 2 and 3, and Fig. 1). The least-squares mean change from baseline in the placebo group showed a slight decrease in the Japan cohort (−0.022 L) compared with an increase in the Not-Japan cohort (0.133 L). Consequently, greater improvements from baseline for FF/VI 100/25 μg OD versus placebo were observed in patients recruited from Japan than Not-Japan, although the variability was greater in the Japan population due to the smaller number of subjects (Table 2). Adjusted treatment differences for FF/VI 100/25 μg versus placebo, FF 100 μg versus placebo, and FF/VI 100/25 μg versus FF 100 μg were directionally the same in the Japan and Not-Japan populations (Table 2, Additional File 4). Furthermore, there was no evidence of a statistically significant difference in trough FEV₁ treatment effect between Japan and Not-Japan patients at Week 12 (P = 0.403).

In the separate FP 500 μg BD-controlled study that assessed FF/VI 200/25 μg and FF 200 μg at Weeks 12 and 24, the least-squares mean changes from baseline in trough FEV₁ in all treatment arms were consistent in the Japan and Not-Japan patients (Table 3). The results in the Japan subset should be interpreted with caution due to the low number of patients in each treatment arm. Nevertheless in the Japan patients, the least-squares mean changes from baseline in trough FEV₁ at Week 24 were numerically greater with OD FF/VI 200/25 μg (0.355 L) than with OD FF 200 μg (0.086 L), or BD FP 500 μg (0.110 L). This pattern was consistent with the results in the Not-Japan patients (FF/VI 200/25 μg: 0.396 L; FF 200 μg: 0.206 L; FP 500 μg: 0.191 L).

In the Japanese asthma patients, the ELLIPTA DPI was considered easy or very easy to use by the majority of subjects (77 %), which is consistent with results for the Overall patient population (Additional Files 5 and 6).

The safety analysis included 188 patients (5 %) recruited from Japan and Korea (Japan N = 148, Korea N = 40) included in the five studies (Overall Safety population N = 4,028 recruited from 21 countries (Additional Files 1 and 2). Demographic and baseline characteristics of the Japan + Korea patients in the Safety population (data not shown) were consistent with the Efficacy population (Table 1). Studies were completed by 139 (74 %) patients in the Japan + Korea cohort and 3,266 (85 %) patients in the Not-Japan + Korea cohort. The number of patients who withdrew due to an AE with FF/VI 100/25 μg, FF/VI 200/25 μg, and placebo in Japan + Korea was 3 (6 %), 1 (7 %), and 0 patients, respectively, and for Not-Japan + Korea was 14 (1 %), 6 (3 %), and 1 (<1 %), respectively.

Exposure to treatment varied across treatment arms as a consequence of the different durations of the studies integrated (Additional File 7) and should be considered when interpreting the incidence of AEs in each treatment group. For example, in both Japan + Korea and Not-Japan + Korea patients, there was a >5-fold longer mean duration of exposure in the FF/VI 100/25 μg treatment group relative to the placebo group.

On-treatment AEs (Table 4) were reported by a greater proportion of Japan + Korea patients (placebo: 41 %; FF/VI: 64–79 %; FF: 69–73 %) than Not-Japan + Korea patients (placebo: 23 %; FF/VI: 45–57 %; FF: 36–54 %). The most frequently reported on-treatment AEs in all populations were nasopharyngitis, headache, bronchitis, and upper respiratory tract infections (Additional File 8). A greater proportion of Japan + Korea patients reported nasopharyngitis (placebo: 21 %; FF/VI: 49–50 %; FF: 23–29 %) than Not-Japan + Korea patients (placebo: 5 %; FF/VI: 10–13 %; FF: 8–10 %). A lower proportion of Japan + Korea patients reported headache (placebo: 3 %; FF/VI: 0–6 %; FF: 7–9 %) than Not-Japan + Korea patients (placebo: 6 %; FF/VI: 6–17 %; FF: 7–15 %). For the pooled dataset, the incidences of AEs deemed to be treatment related by the investigator, as well as fatal SAEs and non-fatal SAEs, were all low and generally reported for similar proportions of patients recruited from Japan + Korea and Not-Japan + Korea, with no trends across active treatment arms (Table 4). The most commonly reported treatment-related AEs in Japan + Korea were the well described local ICS effects of candidiasis and dysphonia, which were also among the most common treatment-related AEs in the Not-Japan + Korea cohort. For other potential class-related effects, the incidence of AEs related to blood glucose or potassium was low in the patients recruited from Japan + Korea and Not-Japan + Korea. Changes in heart rate ≥6 bpm relative to placebo were uncommon. Changes from baseline in the maximum post-baseline QTcF were consistent across all treatment arms, including placebo, in the Japan + Korea and Not-Japan + Korea patients. The incidence of ECG abnormalities of potential clinical importance did not increase for any post-dose assessment relative to pre-dose assessments for any treatment arm in any population.

There were two SAEs (both subarachnoid hemorrhage) in Japan + Korea patients (FF/VI 100/25 μg N = 1; FF 100 μg N = 1), which were not considered treatment related. In the Overall population, the most frequent SAE was asthma exacerbation (FF/VI 100/25 μg N = 11; FF 100 μg N = 9; FF 200 μg N = 1). For the Asthma Composite Endpoint, no events (asthma-related hospitalizations, intubations, and deaths) were reported for patients recruited from Japan + Korea, and the incidence was <1 % in each treatment arm in Not-Japan + Korea patients. The incidence of respiratory-related SAEs and pneumonia-related events was low in both the Japan + Korea and Not-Japan + Korea cohorts.

The urinary cortisol population comprised 1,385 patients (88 [6 %] recruited from Japan + Korea). Geometric mean ratios of 24 h urinary cortisol excretion at the end of treatment relative to baseline were generally close to one (Fig. 2). Treatment comparisons for the Japan + Korea and Not-Japan + Korea cohorts were performed for FF/VI 100/25 μg versus FF 100 μg, FF/VI 200/25 μg versus FF 200 μg, and FF 100 μg or FF 200 μg versus placebo, and were not statistically significant (the 95 % confidence intervals encompassed one; Table 5). In the Overall population, only FF 200 μg versus placebo was associated with statistically lower cortisol ratios (16 % reduction in urine cortisol excretion [ratio of end of treatment to baseline]), which is unlikely to be clinically relevant (Table 5). No statistically significant differences in the treatment comparisons were observed between the Japan + Korea and the Not-Japan + Korea cohorts (Additional File 9).

Further details of each of the 5 studies included in this analysis, and a summary of the efficacy and safety results obtained for each study, are provided in the Clinicaltrials.gov registry (https://​clinicaltrials.​gov) and the GlaxoSmithKline Clinical Study Register:

HZA106827 – http://​www.​gsk-clinicalstudyreg​ister.​com/​study/​106827#ps

HZA106829 – https://​clinicaltrials.​gov/​ct2/​show/​NCT01134042?​term=​HZA106829&​rank=​1

HZA106837 – https://​clinicaltrials.​gov/​ct2/​show/​NCT01086384?​term=​106837&​rank=​1

FFA109685 – https://​clinicaltrials.​gov/​ct2/​show/​NCT00603278?​term=​FFA109685&​rank=​1

FFA109687 – https://​clinicaltrials.​gov/​ct2/​show/​NCT00603382?​term=​FFA109687&​rank=​1

The studies are also fully published in peer-reviewed journals (Bleecker et al. J Allergy Clin Immunol Pract. 2014;2:553–61. O'Byrne et al. Eur Respir J. 2014;43:773–82. Bateman et al. Thorax. 2014;69:312–9. Bleecker et al. Ann Allergy Asthma Immunol. 2012;109:353–8. Bateman et al. Respir Med. 2012;106:642–50).