Evidence review

This evidence review is based on the 3 randomised controlled trials (RCTs) that provide the best published evidence for fluticasone furoate/vilanterol for treating chronic obstructive pulmonary disease (COPD) and that have been published in full. Dransfield et al. (2013) published results from two 52-week randomised, double-blind, parallel group studies that investigated whether fluticasone furoate/vilanterol in combination (3 different doses of fluticasone furoate) would prevent more exacerbations compared with vilanterol alone.

A second RCT included in this evidence summary (Kerwin et al. 2013) compared 2 strengths of fluticasone furoate/vilanterol with the same strengths of the individual components and placebo in patients with COPD. The 2 co-primary outcomes were the weighted mean FEV1 (forced expired volume in 1 second; 0 to 4 hours post-dose) on day 168, and the change from baseline in trough FEV1 (23–24 hours post-dose) on day 169. An additional study (Martinez et al. 2013) had the same design as that by Kerwin et al. (2013) but for methodological reasons, statistical analysis of the results for fluticasone furoate/vilanterol 100/25 micrograms could not be performed.

A further relevant phase III trial has been completed, but results have not yet been published in full (trial reference NCT01342913). This trial assessed the 24-hour spirometry effect (FEV1) of fluticasone furoate/vilanterol 100/25 micrograms once daily compared with fluticasone propionate/salmeterol 500/50 micrograms (Seretide 500 Accuhaler) twice daily over a 12-week treatment period in people with COPD.

Dransfield et al. (2013)

  • Design: 2 simultaneous, replicate, 52-week, randomised, double-blind, parallel group studies. Each had a 4-week open-label run-in period using combination fluticasone propionate/salmeterol 250/50 micrograms twice daily. The method of allocation described suggests that this was concealed.

  • Population: 1622 adults in study 1 and 1633 adults in study 2. Study 1 involved 167 sites in 15 countries and study 2 involved 183 sites in 15 countries. Participants were 40 years or older (mean 64 years) with COPD (post-bronchodilator FEV1 70% predicted or less, mean range 44.3% to 46.4%, and FEV1/FVC [forced vital capacity] ratio 0.7 or less), a history of at least 1 COPD exacerbation in the previous year that needed systemic or oral corticosteroids, antibiotics or admission to hospital, and a smoking history of 10 or more pack-years.

  • Intervention and comparison: participants were randomised in approximately equal numbers to 4 treatments, taken once daily using the Ellipta inhaler:

    • fluticasone furoate 50 micrograms (emitted dose 44 micrograms) plus vilanterol 25 micrograms (emitted dose 22 micrograms)

    • fluticasone furoate 100 micrograms (emitted dose 92 micrograms) plus vilanterol 25 micrograms (emitted dose 22 micrograms)

    • fluticasone furoate 200 micrograms (emitted dose 184 micrograms) plus vilanterol 25 micrograms (emitted dose 22 micrograms)

    • vilanterol 25 micrograms (emitted dose 22 micrograms).

  • Outcome: the primary efficacy end point was the yearly rate of moderate and severe COPD exacerbations. Moderate exacerbations were defined as worsening symptoms of COPD needing treatment with oral corticosteroids and/or antibiotics. Severe exacerbations were defined as those that needed hospital admission. Secondary and additional end points included the time to first moderate or severe exacerbation, yearly rate of severe exacerbations, the number of night-time awakenings due to symptoms, and dyspnoea score. Specific safety end points included haematological and clinical measurements, incidence of bone fractures and clinically diagnosed pneumonia. The 2 studies were analysed separately and in a predefined pooled analysis, based on the intention-to-treat population. Results of this pooled analysis are summarised in table 2.

Table 2 Summary of the pooled analysis, Dransfield et al. (2013)

Fluticasone furoate/vilanterol once daily (micrograms)

Vilanterol 25 micrograms once daily

50/25

100/25

200/25

Efficacy (ITT population)

n=820

n=806

n=811

n=818

Primary outcome

Moderate and severe exacerbations; LS mean yearly rate

0.93

0.81

0.85

1.11

LS mean yearly RR for moderate and severe exacerbations (95% CI) compared with vilanterol alone

0.8 (0.7 to 1.0), p=0.014

0.7 (0.6 to 0.8), p<0.0001

0.8 (0.7 to 0.9), p=0.0003

Selected secondary and additional outcomes

Time to first moderate or severe exacerbation: HR (95% CI) compared with vilanterol alone

0.9 (0.8 to 1.0), p=0.114

0.8 (0.7 to 0.9), p=0.0002

0.8 (0.7 to 0.9), p=0.0001

Severe exacerbation LS mean yearly rate

0.08

0.09

0.08

0.10

LS mean yearly RR for severe exacerbations (95% CI) compared with vilanterol alone

0.8 (0.6 to 1.2), p=0.313

0.9 (0.6 to 1.4), p=0.695

0.8 (0.5 to 1.2), p=0.280

Night-time awakenings, LS mean difference (95% CI) from vilanterol alone

−0.06 (−0.10 to −0.01), p=0.011

−0.08 (−0.12 to −0.03), p=0.001

−0.07 (−0.12 to −0.03), p=0.002

Dyspnoea scorea: LS mean difference (95% CI) from vilanterol alone

−0.08 (−0.12 to −0.03), p=0.0006

−0.09 (−0.014 to −0.05), p<0.0001

−0.11 (−0.16 to −0.07), p<0.0001

Safety (ITT population)

n=820

n=806

n=811

n=818

Adverse events leading to discontinuation or withdrawalb

6.5% (53/820)

7.7% (62/806)

7.5% (61/811)

5.5% (45/818)

Local corticosteroid effectsb

17.3% (142/820)

15.0% (121/806)

17.3% (140/811)

11.7% (96/818)

Pneumoniab

5.9% (48/820)

6.3% (51/806)

6.8% (55/811)

3.3% (27/818)

Bone disorders (including fractures)b

2.9% (24/820)

3.3% (27/806)

2.6% (21/811)

1.1% (9/818)

Abbreviations: CI, confidence interval; HR, hazard ratio; ITT, Intention-to-treat; LS, least squares; RR, rate ratio.

a Dyspnoea was scored on a scale of −2 to +2, with −2 indicating 'much less than usual' and +2 indicating 'much more than usual'.

b No statistical analysis of safety outcomes was presented.

Kerwin et al. (2013)

  • Design: 24-week double-blind, placebo-controlled RCT. The method of allocation described suggests that this was concealed.

  • Population: 1030 adults in 9 countries aged at 40 years or older (mean 63 years) with COPD (post bronchodilator FEV1 70% predicted or less, mean range 46.9–49.9%, and FEV1/FVC ratio 0.7 or less), a smoking history of at least 10 pack-years, and a score of at least 2 on the Modified Medical Research Council Dyspnoea Scale. No previous history of COPD exacerbations was needed but about a quarter of participants had had at least 1 moderate exacerbation of COPD (needing treatment with oral corticosteroids and/or antibiotics but not hospital admission) and about 7% had had at least 1 severe exacerbation (needing hospital admission) in the year before trial entry.

  • Intervention and comparison: participants were randomised in approximately equal numbers to 5 treatments, taken once daily in the morning using a dry powder inhaler:

    • fluticasone furoate 50 micrograms (emitted dose 44 micrograms) plus vilanterol 25 micrograms (emitted dose 22 micrograms)

    • fluticasone furoate 100 micrograms (emitted dose 92 micrograms) plus vilanterol 25 micrograms (emitted dose 22 micrograms)

    • fluticasone furoate 100 micrograms (emitted dose 92 micrograms)

    • vilanterol 25 micrograms (emitted dose 22 micrograms)

    • placebo.

  • Outcomes: there were 2 co-primary outcomes: weighted mean FEV1 (0–4 hours post-dose) on day 168, and the change from baseline in trough FEV1 (23–24 hours post-dose) on day 169.The primary analysis was based on the intention-to-treat population split into 2 levels pre-specified by the authors to avoid spurious statistically significant findings arising through chance, given the number of possible comparisons:

    • The level 1 analysis consisted of 6 key comparisons of the co-primary end points for fluticasone furoate/vilanterol 100/25 micrograms and vilanterol 25 micrograms. These are summarised in the discussion of the evidence for clinical effectiveness.

    • The authors specified that only if all comparisons reached statistical significance at level 1 would they move on to level 2 analyses, which included comparison with fluticasone furoate/vilanterol 50/25 micrograms. The level 1 analysis did not meet the pre-defined criteria and so no formal statistical testing was performed at level 2.

  • Secondary and additional outcomes included changes in dyspnoea score, night-time awakenings and other symptom-related end points but the statistical hierarchy used in the analysis meant that no statistical significance can be inferred from the results for these outcomes.

Clinical effectiveness

Discussion of the evidence for clinical effectiveness focuses on fluticasone furoate/vilanterol 100/25 micrograms because that is the dose and strength that has been submitted for licensing. In their predefined pooled analysis, Dransfield et al. (2013) found that the mean yearly rate for moderate and severe exacerbations for fluticasone furoate/vilanterol 100/25 micrograms was 0.81 compared with 1.11 for vilanterol 25 micrograms alone. The yearly rate ratio was 0.7 (95% confidence interval [CI] 0.6 to 0.8), a reduction of 30% in relative terms. This was similar in people with a history of frequent exacerbations; defined as at least 2 moderate or severe exacerbations in the previous year (yearly rate ratio 0.7, 95% CI 0.6 to 0.9, p=0.0005). The full NICE guideline on COPD considered a relative reduction in the risk of exacerbations of 20% or more to be clinically important. This dose of fluticasone furoate/vilanterol also increased the time to first moderate or severe exacerbation compared with vilanterol 25 micrograms alone (hazard ratio 0.8, 95% CI 0.7 to 0.9, p=0.0002). However, it was not shown to reduce the mean yearly rate of severe exacerbations (those needing admission to hospital) compared with vilanterol 25 micrograms alone (rate ratio 0.9, 95% CI 0.6 to 1.4, p=0.695).

Compared with vilanterol 25 micrograms alone, fluticasone furoate/vilanterol 100/25 micrograms also produced statistically significant improvements in night-time awakenings (mean difference −0.08, 95% CI −0.12 to −0.03, p=0.0012) and dyspnoea (mean difference −0.09, 95% CI −0.014 to −0.05, p<0.0001, on a scale of −2 to +2, with −2 indicating 'much less than usual' and +2 indicating 'much more than usual').

Kerwin et al. (2013) found that fluticasone furoate/vilanterol 100/25 micrograms was statistically significantly superior to placebo in improving post-dose weighted mean FEV1 (173 ml, 95% CI 123 to 224 ml, p<0.001) and trough FEV1 (115 ml, 95% CI 60 to 169 ml, p<0.001) after 24 weeks' treatment. However, there was no statistically significant difference in trough FEV1 between fluticasone furoate/vilanterol 100/25 micrograms and vilanterol 25 micrograms (48 ml, 95% CI -6.0 to 102ml, p=0.082).

A further study by Martinez et al. (2013) was of a similar design to that by Kerwin et al. (2013) and was intended to provide multiple statistical comparisons between fluticasone furoate/vilanterol 200/25 micrograms and fluticasone furoate/vilanterol 100/25 micrograms compared with their individual components and placebo. The authors used a similar hierarchy of statistical analysis to Kerwin et al. (2013), which involved analysis of comparisons of fluticasone furoate/vilanterol 200/25 micrograms in the first stage. Pre-specified criteria were not met so the second stage, which would have included comparisons of fluticasone furoate/vilanterol 100/25 micrograms, was not conducted.

Safety

Statistical analysis of the safety data from all 3 of the studies was not presented, which therefore limits the conclusions that can be drawn.

In the pooled analysis of the RCT by Dransfield et al. (2013), 7.7% of patients receiving fluticasone furoate/vilanterol 100/25 micrograms experienced an adverse event leading to discontinuation or withdrawal from the study, compared with 5.5% of those receiving vilanterol 25 micrograms. As the NICE guideline on COPD notes, there are particular risks associated with long-term use of inhaled corticosteroids. Local corticosteroid effects, pneumonia and bone disorders (including fractures) were seen more frequently in the fluticasone furoate/vilanterol 100/25 microgram group than the vilanterol group.

There were 6 non-traumatic fractures in the fluticasone furoate/vilanterol 100/25 microgram group (0.74%) compared with 2 in the vilanterol group (0.24%). The incidence of pneumonia in the fluticasone furoate/vilanterol 100/25 microgram group was approximately double that in the vilanterol group (6.3% compared with 3.3%). Moreover, 3.1% (25/806) of patients in the fluticasone furoate/vilanterol 100/25 microgram group had pneumonia that required admission to hospital compared with 0.98% (8/818) of patients in the vilanterol 25 microgram group. There was 1 fatal pneumonia-related adverse event in the fluticasone furoate/vilanterol 100/25 microgram group compared with none in the vilanterol 25 microgram group. The potential risk of pneumonia or other infections of the lower respiratory tract associated with the use of inhaled corticosteroids in patients with COPD has previously been highlighted by the Medicines and Healthcare products Regulatory Agency (MHRA).

In the Kerwin et al. (2013) study, the same percentage of patients receiving fluticasone furoate/vilanterol 100/25 micrograms and placebo had an adverse event leading to discontinuation or withdrawal from the study (9%). Pneumonia occurred in 2% of the fluticasone furoate/vilanterol 100/25 microgram group compared with 1% of the placebo group.

Evidence strengths and limitations

Kerwin et al. (2013) demonstrated that fluticasone furoate/vilanterol 100/25 micrograms has beneficial effects on lung function as measured by trough FEV1 in people with COPD compared with placebo, when used up to 24 weeks. However, the full NICE guideline on COPD considers 100 ml to be the minimum clinically important difference in FEV1. The lower limit of the 95% confidence interval for the difference did not reach this limit.

The NICE guideline on COPD advises that the effectiveness of bronchodilator therapy should not be assessed by lung function alone but should include a variety of other measures, such as improvement in symptoms, activities of daily living, exercise capacity, and rapidity of symptom relief. Dransfield et al. (2013) demonstrated that fluticasone furoate/vilanterol 100/25 micrograms reduces the yearly rate of moderate and severe exacerbations compared with vilanterol 25 micrograms alone and the time to first moderate or severe exacerbation, but not the rate of exacerbations requiring admission to hospital. There were statistically significant benefits from fluticasone furoate/vilanterol 100/25 micrograms on symptomatic measures (night-time awakening and dyspnoea scores) but the differences were small in absolute terms and are of questionable clinical significance. Kerwin et al. (2013) also assessed effects on dyspnoea but observed differences were less than the minimally clinically important difference and the results could not be assessed statistically.

The studies by Dransfield et al. (2013) and Kerwin et al. (2013) also had several methodological limitations that affect their usefulness in assessing the place in therapy of fluticasone furoate plus vilanterol in the NHS. The rationale for the design of the studies by Dransfield et al. (2013), encompassing 2 simultaneous, replicate trials of a common condition with a very large number of centres relative to the number of patients recruited (such that each centre would recruit only a small number of patients) is unclear. The primary end point depended in part on local practice (whether or not to start systemic corticosteroids or antibiotics, or admit to hospital) and this may have varied between sites. The comparator in all the studies was placebo or vilanterol 25 micrograms, and not established combination therapy or monotherapy licensed for use in COPD.

One trial that has not yet been published in full (trial reference NCT01342913) compared fluticasone furoate/vilanterol 100/25 micrograms once daily with fluticasone propionate/salmeterol 500/50 micrograms (Seretide 500 Accuhaler) twice daily. This trial was relatively short term (12 weeks) and it measured a disease-orientated outcome (24-hour spirometric effect [FEV1]), which will therefore limit the conclusions that can be drawn from it.