Pneumonia Risk with Inhaled Fluticasone Furoate and Vilanterol in COPD Patients with

Moderate Airflow Limitation: the SUMMIT trial

Courtney Crim1, Peter M.A. Calverley2, Julie A. Anderson3, Andrew P. Holmes4, Sally Kilbride3,

Fernando J. Martinez5, Robert D. Brook6, David E. Newby7, Julie C. Yates1 , Bartolomé R. Celli8,

and Jørgen Vestbo,9 on behalf of the SUMMIT investigators

1GlaxoSmithKline, Research Triangle Park, USA, 2University Hospital Aintree, Liverpool, UK;

3GSK R&D, Stockley Park, UK; 4Veramed Limited, Twickenham, UK; 5NewYork-Presbyterian

Hospital/Weill Cornell Medical Center, New York, USA; 6University of Michigan Health

System, Ann Arbor, USA; 7University of Edinburgh, Edinburgh, UK; 8Brigham and Women’s

Hospital, Harvard Medical School, Boston, USA,: 9University of Manchester University Hospital

and South Manchester NHS Foundation Trust, Manchester, UK .

Correspondence and requests for reprints should be sent to Courtney Crim, Respiratory

Therapeutic Area Unit, GlaxoSmithKline, Five Moore Dr., 5.3317.3A, Research Triangle Park,

NC 27709-3398. E-mail: Courtney.c.crim@gsk.com. Telephone: 1-919-483-3765. Facsimile: 1-

919-483-4300.

Author Contributions: Conception and design: CC, JV, FJM, JAA, RDB, PMAC, DEN, JCY,

BRC; analysis and interpretation: CC, JV, FJM, JAA, RDB, PMAC, DEN, JCY, BRC, SK, APH;

drafting the manuscript for important intellectual content: All authors.

Sources of support: This study was funded by GlaxoSmithKline, NCT01313676 (GSK study

number 113782): Study to Understand Mortality and Morbidity In COPD Trial (SUMMIT).

Author disclaimer: Employees of the sponsor were involved in the conception, design and

conduct of the study, and in data analysis and interpretation. All authors, including authors

employed by the sponsor, participated in the development of the manuscript, and had access to

the data from the study. The decision to submit for publication was that of the authors alone.

Running head: Pneumonia risk with inhaled steroids in moderate COPD

Descriptor number (selected from ATS subject category list): 9.14 – COPD: pharmacological

treatment

Key words: Pneumonia; pulmonary disease, chronic obstructive; corticosteroids

Total word count: 3419 words

Data supplement statement: This article has a data supplement, which is accessible from this

issue's table of contents online at www.atsjournals.org

Abstract

Rationale: Pneumonia risk with inhaled corticosteroid use in chronic obstructive pulmonary

disease (COPD) has not been thoroughly assessed in patients with moderate airflow limitation.

Objectives: To determine the incidence of pneumonia and risk factors in COPD patients with

moderate airflow limitation who had, or were at high risk for cardiovascular disease.

Methods: In the Study to Understand Mortality and MorbidITy in COPD (SUMMIT), 16,590

subjects with moderate airflow limitation (50% ≤ FEV1 ≤ 70% predicted) and heightened

cardiovascular risk were randomized double-blind 1:1:1:1 to inhaled once-daily vilanterol 25 µg

(VI), fluticasone furoate 100 µg (FF), vilanterol 25 µg combined with 100 µg fluticasone furoate

(FF/VI), or matched placebo. In a pre-specified analysis, we assessed investigator-reported

adverse pneumonia events, and independently-adjudicated fatal events.

Measurements and Main Results: The safety population comprised 16,568 subjects who

actually received study medication. There were 1017 pneumonia events reported from 842

subjects. For placebo, FF, VI, and FF/VI, reported pneumonia incidence was 5%, 5%, 4% and

6%, respectively. When adjusted for time on treatment, event rates were similar in the placebo,

FF and FF/VI containing arms (3.84, 4.24 and 3.95/100 treatment years, respectively) but lower

in the VI group (2.77/100 treatment years). Risk factors for pneumonia risk included: greater

degree of airflow limitation (i.e. FEV1 <60% predicted), prior exacerbation history, and BMI <

25kg/m2.

Conclusions: In contrast to previous studies in patients with severe disease, increased

pneumonia risk with inhaled corticosteroid use was not evident in COPD subjects with moderate

airflow limitation and heightened cardiovascular risk.

According to the World Health Organization in 2014, lower respiratory tract infections

and chronic obstructive pulmonary disease (COPD) represented the third and fourth leading

causes of death worldwide (1). Furthermore, COPD itself is a known risk factor for pneumonia

(2, 3). We and others have previously reported that the chronic use of inhaled corticosteroid

(ICS)-containing regimens further increase adverse events of pneumonia in COPD patients (4-7)

and both the United States Food and Drug Administration and European Medicines Agency

require labeling of such products to highlight this risk (8-10).

Mechanisms to explain this increased pneumonia risk in COPD patients are unclear but

are undoubtedly multifactorial. In addition to impaired mucociliary clearance of inhaled

pathogens, potential mechanisms are likely to be in part related to recognized defects in both

innate and acquired immunity (11-17). A history of COPD exacerbations is not only an

additional risk factor for pneumonia, possibly related to the acquisition of new strains of bacteria

(18), but such exacerbation-prone individuals might also have impaired innate immunity. For

example, Berenson and colleagues reported that compared with those without prior recent

exacerbations, alveolar macrophages recovered from exacerbation-prone COPD subjects elicited

a lower cytokine (e.g. interleukin (IL)-8, tumor necrosis factor (TNF)-) response in the

presence of H. influenzae, M. catarrhalis and S. pneumoniae when induced with Toll-like

receptor (TLR)-2 and TLR4 ligands (13). Furthermore, although the induction of TLR4 on

alveolar macrophages did not differ between exacerbation-prone and non-exacerbators following

exposure to any of these bacteria, TLR2 expression following exposure to M. catarrhalis and S.

pneumoniae was reduced (13).

Severe airflow limitation appears to increase pneumonia risk in COPD patients (4, 7).

However, in our report with the ICS/long-acting beta-2-agonist (LABA) regimen of fluticasone

furoate/vilanterol (FF/VI) in an exacerbating COPD population, we were unable to demonstrate

an increased pneumonia risk in the subgroup of subjects with an FEV1 ≥50% of predicted (7).

The Study to Understand Mortality and MorbidITy in COPD (SUMMIT) investigated whether

FF/VI could improve survival in patients with moderate COPD who had, or were at high risk for,

cardiovascular disease (CVD) (19, 20). The risk of pneumonia in these patients with COPD and

moderate airflow limitation receiving FF, VI, or the combination was assessed as a safety

endpoint in SUMMIT. In the pre-specified analysis reported here, we investigated the hypothesis

that in this population, the incidence of pneumonia with an FF-containing treatment regimen

would not be increased compared with placebo.

Methods

Study Design and Subjects

This was a pre-specified analysis of the SUMMIT trial, for which the design and primary

result have been previously published (19, 20). Briefly, in this prospective, double-blind,

parallel-group, placebo-controlled, event-driven trial, subjects were randomized to receive one of

the following (once daily) from the ELLIPTA™ dry powder inhaler (GSK, UK; ELLIPTA is a

trademark of the GSK group of companies): FF/VI 100/25 μg, FF 100 μg, VI 25 μg, or placebo.

All treatment groups were allowed to continue short-acting bronchodilators and/or theophylline;

use of inhaled corticosteroids and long-acting bronchodilators was discontinued at least 48 hours

before study entry. Participants were current or former smokers aged 40–80 years, with a

diagnosis of COPD and a post-bronchodilator FEV1 ≥50 and ≤70% of the predicted value, a post-

bronchodilator FEV1/forced vital capacity ≤0.70, and ≥2 on the modified Medical Research

Council dyspnea scale.

At screening, eligible subjects were required to have a history of CVD (coronary artery

disease, peripheral arterial disease, prior stroke or myocardial infarction, or diabetes mellitus

with target organ disease) or increased cardiovascular risk (≥60 years and receiving medications

for ≥2 of the following: hypercholesterolemia, hypertension, diabetes mellitus, or peripheral

vascular disease). For the assessment of the primary endpoint of all-cause mortality for the

comparison of FF/VI 100/25 μg with placebo, this event-driven study concluded at a common

end date when approximately 1000 deaths were predicted to have occurred. Secondary objectives

evaluated the effect of FF/VI compared with placebo on the rate of FEV1 decline, and the effect

of FF/VI compared with placebo on a cardiovascular composite endpoint comprised of on-

treatment CV death, myocardial infarction, stroke, unstable angina and TIA. COPD

exacerbations were collected as an efficacy endpoint. The risk of pneumonia in the active

treatment arms was assessed as a safety endpoint of special interest and included on-treatment

events reported after the common end date until closure of the site.

Pneumonia Adverse Events (AEs) and Serious AEs

There was no a priori definition of pneumonia, nor were chest x-rays, sputum cultures or

laboratory evaluations required to confirm the clinical diagnosis. Pneumonia as an adverse event

(AE) or serious AE (SAE) was reported by the investigator using available clinical information

and coded using the Medical Dictionary for Regulatory Activities (MedDRA® Version 18.0;

International Federation of Pharmaceutical Manufacturers and Associations (IFPMA), Geneva,

Switzerland). All MedDRA preferred terms that could relate to pneumonia (see Table E1 in the

online supplement) were counted to provide a more complete assessment of all physician-

reported pneumonias (defined as pneumonia AE of special interest, AESI).

All deaths were categorized by an independent clinical endpoints committee (CEC), blind

to treatment allocation, based on additional source documents (e.g. medical records, chest x-rays,

autopsy reports, death certificates) regardless of the investigator reported fatal SAE term (21).

Fatal pneumonia events reported by the investigator were those present at the time of death; fatal

pneumonia events adjudicated by the CEC were those deemed to be the actual cause of death

regardless of the reported AE term. The CEC adjudicated deaths therefore provided a consistent

classification of deaths across the study.

Statistical Analysis

Subjects were included in the analysis if they took at least one dose of study medication

and analyzed according to the treatment they took for the majority of the treatment period. Apart

from 1 subject, this was the treatment to which they were randomized. Pneumonia and COPD

exacerbations were included in the analysis if they started between treatment start and the day

after treatment stop date, inclusive. Deaths were considered on-treatment if they occurred

between treatment start and seven days after treatment stop date, inclusive. The number and

proportion of subjects who were reported as having any of the above grouped pneumonia terms

as an AE or SAE was summarised by treatment group. Due to differential treatment exposure

between the treatment groups, the rate of pneumonia events per 100 treatment-years was

calculated by dividing the number of AEs by the number of years subjects were exposed to study

treatment, then multiplying by 100.

The time to first pneumonia and time to first composite of pneumonia or

moderate/severe COPD exacerbation was compared between treatment groups using Kaplan–

Meier estimates and the Cox Proportional Hazards model (PH) including covariates of age and

gender; for the analysis of the composite endpoint a covariate of previous exacerbations

(exacerbations in the year prior to the study as 0, 1, ≥2) was also included. This was repeated for

time to serious pneumonia and the composite of serious pneumonia or severe COPD

exacerbation. Kaplan–Meier cumulative incidence curves were also produced. A similar Cox PH

analysis was performed post-hoc for the on-treatment deaths adjudicated as pneumonia

accounting for the non-pneumonia deaths as competing risks. The numbers of subjects who died

within 30 days of their last on-treatment pneumonia were summarized by treatment group.

To identify risk factors for pneumonia, a Cox PH model for time to first pneumonia was

fitted, using only subjects in the placebo arm, including covariates of smoking status (current vs

former), smoking pack-years (quintiles), age group (<55, ≥55 to <65, ≥65 to <75 and ≥75 years),

baseline % predicted FEV1 (<60% and ≥ 60%), gender, exacerbations in the year prior to the

study (0, 1, ≥2), body mass index (BMI) categories (<25 vs ≥25 kg/m2), modified Medical

Research Council (mMRC) dyspnoea score (1-2 vs 3-4), region, diabetes history and congestive

heart failure history. All covariates were fitted in the model, then a backwards selection

procedure was used to identify a final model, using criterion of p≥0.10. A plot was produced

showing the effect of risk factors selected in the final model. To calculate the

p-values for treatment interactions, a Cox PH model was fitted with

covariates of age, gender, subgroup and subgroup by treatment for each

baseline variable. P-values ≤ 0.05 were considered nominally significant.

Results

Of the 23,835 subjects recruited into the study, 16,590 were randomized into one of the

four treatment arms. However, 22 of the randomized subjects never took investigational product;

therefore, the safety population consisted of 16,568 subjects. The demographic and baseline

characteristics of this safety population are shown in Table 1. Subjects were mostly male with a

mean age of 65 years, a mean smoking history of approximately 41 pack-years and a mean (SD)

predicted post-bronchodilator FEV1 of 60% (6%). Just under one-half the subjects were current

smokers at screening.

Mean exposure to study medication was 1.7 years and was greater in all active treatment

groups compared with placebo, being greatest in the group that took FF/VI. The increase in mean

exposure over the placebo group was 4%, 5%, and 6% in the FF, VI and FF/VI treatment groups,

respectively.

Adverse Event Reports of Pneumonia

Of the 16,568 randomized subjects who took study medication, 1017 events of

pneumonia were reported in 842 subjects during the double-blind treatment period (Table 2). Of

the subjects in the placebo group, 5.2% had a pneumonia event compared with 5.5%, 3.9% and

5.7% of subjects in the FF, VI and FF/VI groups respectively. When adjusted for time on

treatment, the rates of these events were similar in the placebo, FF and FF/VI treatment groups

(3.84, 4.24 and 3.95/100 treatment years, respectively) but lower in the VI group (2.77/100

treatment years). There appears to be no difference between the FF containing arms and placebo,

regardless of whether patients had an exacerbation in the year prior to randomization (see Table

E2 in the online supplement). There was one event of non-pulmonary tuberculosis in the FF arm

that was coded by the MedDRA dictionary as tuberculosis, hence, included as an AE/SAE

pneumonia event. Removal of this one event would not likely change the results.

Kaplan-Meier probability estimates from the analysis of the time to first pneumonia

revealed that the risk was similar in the FF-containing and placebo arms but lower in VI arm

(Figure 1a). Hazard ratios from the analysis of time to first AE pneumonia of special interest

showed that compared with placebo the risk was similar with both FF (HR 1.035, 95% CI 0.859,

1.247; p=0.716) and FF/VI groups (HR 1.038, 95% CI 0.863, 1.249; p=0.693), but lower with

VI; 0.722 (95% CI 0.589, 0.886; p=0.002) (Table 3).

Serious Adverse Event Reports of Pneumonia

Serious pneumonia represented a subset of all reported pneumonia adverse events; 592

SAE pneumonia events occurred in 517 subjects. The incidence of SAE pneumonia was similar

across all treatment groups when expressed as a percentage. However, as with AE pneumonia

events, Kaplan-Meier probability estimates from the analysis of the time to first SAE pneumonia

revealed that the risk was similar in the FF-containing and placebo arms but lower in the VI arm

(Figure 1b). Hazard ratios from the analysis of time to first SAE pneumonia of special interest

also showed that compared with placebo the risk was similar with both FF and FF/VI groups, but

lower with VI (HR 0.778, 95% CI 0.600, 1.008; p=0.057) (Table 3a).

Exacerbations and Pneumonia Events

Inhaled corticosteroids are recommended as a treatment adjunct to decrease the risk and

rate of COPD exacerbations. However, antibiotics used to treat infective moderate exacerbations

and those necessitating hospitalization are similar to regimens used for treatment of pneumonia.

Compared with placebo, all active therapies reduced the rate of moderate and severe

exacerbations (i.e., 10% reduction with VI, 12% with FF and 29% reduction with the

combination), and specifically those that necessitated hospital admission (20). As a result of the

concern that ICS use may increase pneumonia risk, we also assessed for any relative increase in

pneumonia with exacerbation reduction by combining these events as a composite of pneumonia

or moderate/severe COPD exacerbation.

The risk of a moderate or severe COPD exacerbation or AE pneumonia of special interest

was reduced with FF/VI by 19.6% compared with placebo (HR 0.804; 95% CI 0.742, 0.872;

p<0.001), 17.7 % compared with FF and 12.6% compared with VI; only an 8.0% reduction was

observed with VI compared with placebo (Table 4). Similarly, compared with placebo, the risk

of a severe COPD exacerbation or serious pneumonia was reduced by 20.2% and 16.1% with

FF/VI and VI, respectively (Table 4). For these more serious events, no difference was observed

between the combination regimen and the individual components (Table 4).

Pneumonia Deaths

Investigator-reported pneumonia events present at the time of death occurred in <1% of

the subjects, whether while on treatment or after discontinuation of study drug. When expressed

as either incidence or exposure-adjusted for rate, the FF-containing regimens appeared to be

associated with more investigator-reported on-treatment fatal events compared with placebo; the

VI group revealed approximately one half the placebo numbers (Table 2). Conversely, there was

less disparity across treatment groups when deaths within 30 days of the last on-treatment

pneumonia were assessed (n=19, 22, 15, and 20 for placebo, FF, VI and FF/VI, respectively,

Table 5). Moreover, when pneumonia as the actual cause of death was adjudicated by the

Clinical Endpoints Committee, differences between the FF-containing treatment arms and

placebo were less evident; the pneumonia deaths in the VI group remained less than placebo by

one-third for both events and exposure-adjusted rate (Table 6).

Risk Factors for Pneumonia

The backwards selection procedure with the Cox PH Model for the time to first

pneumonia for the subjects randomized to placebo eliminated smoking pack-years, congestive

heart failure history and mMRC dyspnoea score (p values all > 0.1). A prior exacerbation

history was clearly associated with an increased pneumonia risk, as was BMI <25 kg/m2 and a

greater impairment of lung function, i.e. FEV1 50-60% predicted; increasing age trended towards

an association with an increased risk, as was being a male or a former smoker. Of the subjects

randomized to placebo, the mean ± SD pack years for current smokers was 42 ± 23 vs 40 ± 26

for former smokers; the median pack years for current and former smokers was 40 vs 35,

respectively. Diabetes mellitus also trended towards being identified as being an important risk

factor (Figure 2). There was no evidence of an interaction of any of these risk factors with

treatment; specifically, there was no additional risk in patients randomized to the ICS-containing

arms (all p values >0.14).

Discussion

The most recent Global Initiative for Chronic Obstructive Lung Diseases strategy

document recommends the use of a long-acting anticholinergic or β2-agonist, either alone or in

combination, for the management of symptomatic patients with stable COPD and either

moderate airflow limitation or no more than one moderate exacerbation per year (i.e. GOLD B)

(22). The use of ICS is not recommended in this patient group, in part due to the limited evidence

of efficacy and the perceived increased risk of pneumonia (22), even in those with moderate

airflow limitation. SUMMIT is the largest study conducted solely in COPD patients with

moderate airflow limitation. With over 8,000 subjects randomized to an FF-containing treatment

arm, when compared with subjects receiving placebo in addition to short-acting bronchodilators,

these data demonstrate no increased pneumonia risk in COPD patients with moderate airflow

limitation. Just as important, when adjusted for duration of treatment exposure, there was no

increase in pneumonia events leading to or prolonging hospitalization, or as the cause of death

when compared with placebo. The lack of an increase in pneumonia mortality observed in this

trial is in concert with that of a recent review of published randomized control trials and

observational studies of patients with COPD using ICS (23).

Prior studies that demonstrated an increased pneumonia risk with ICS were performed in

patients with COPD and severe airflow obstruction (4, 5, 7, 24-25), and severe airflow limitation

is an identified risk factor (4). In the previous replicate 1-year studies in patients with COPD, an

exacerbation history and mean post-bronchodilator FEV1 of 45.5% of predicted, we

demonstrated an approximate doubling of the pneumonia risk with FF/VI 100/25 compared with

VI 25 mcg (7). However, we were unable to confirm that this risk remained elevated in the

subgroup of patients with moderate airflow limitation (HR 1.9; 95% CI 0.7, 5.4), due to the small

number of patients in this subgroup (7). The much larger sample size from SUMMIT afforded a

better opportunity to assess this risk in these less severely obstructed patients.

Data from the SUMMIT trial confirmed that the risk of pneumonia is lower in patients

with COPD and moderate airflow limitation compared with those with more severe disease. As a

comparison, the exposure adjusted rates for AE and SAE pneumonia were higher in the TORCH

trial (mean FEV1 44% predicted): 5.2 and 3.0/100 treatment years versus 8.8 and 5.5/100

treatment years in the placebo and ICS/LABA arms, respectively (4). In the previously reported

replicate 1-year exacerbation studies with FF/VI (mean FEV1 45% predicted), exposure adjusted

rates for AE and SAE pneumonia were 4.2 and 1.2/100 treatment years versus 8.6 and 4.3/100

treatment years, in the VI and FF/VI arms respectively (7).

In the TORCH trial severity of airflow limitation and exacerbation history were identified

risk factors independent of treatment (4). These remained risk factors in SUMMIT as illustrated

by the findings in patients randomized to placebo; a 41% increase in pneumonia risk in those

with an FEV1 of at least 50% but less than 60% predicted compared with COPD patients whose

FEV1 was 60% predicted or greater; and a 51% and 90% increased pneumonia risk for those

patients with 1 or ≥2 exacerbations, respectively, in the year prior to randomization, compared

with patients without such history. There was also no interaction of these risk factors with

treatment.

Pneumonia as an adverse event occurred at a significantly lower rate in the vilanterol

treatment arm compared with the placebo and FF-containing regimens. This observation raises

the issue about the most appropriate comparator for evaluating pneumonia risk in patients with

COPD receiving ICS-containing regimens. According to International Conference of

Harmonisation (ICH) guidelines, the most reliable comparator for assessing adverse events

associated with an active treatment is placebo, since it maximizes the ability to distinguish

adverse drug effects from the underlying disease or intercurrent illness and broadly represent the

background rate in the population being studied (26). Nevertheless, the difference between VI

and placebo was unexpected and we have no clear explanation for its occurrence. Mechanisms

by which a LABA could decrease pneumonia risk are speculative at best. For example,

efferocytosis of apoptotic cells has been shown to impair phagocytosis and killing by alveolar

macrophages of bacteria such as Escherichia coli, Klebsiella pneumoniae, and S. pneumoniae

(14). The collectin surfactant proteins SP-A and SP-D function as components of the innate

immune system by recognizing pathogen-associated molecular patterns (PAMPs) on

microorganisms, thereby enhancing their opsonization and killing by phagocytes (27). Secretion

of SP-A is augmented by β2-adrenergic agents (28, 29) and binding of SP-A or SP-D to signal

inhibitory protein (SIRP) on macrophages suppresses efferocytosis, acting through inhibitory

cytoplasmic tyrosine phosphatases and other downstream molecules (30); the net result being

enhanced immune function and potentially, decreased pneumonia risk. However, this hypothesis

and other potential mechanisms were not explored in SUMMIT and would require additional

studies. We believe it would be inappropriate to conclude from these data that vilanterol is

protective and FF negates this beneficial effect as there was no appreciable difference between

FF/VI and FF in the time to first pneumonia event or rates. Also, we are uncertain whether this

apparent decrease in pneumonia risk is unique to vilanterol or could be expected with molecules

within the entire LABA class. However, it is noteworthy that in similarly designed studies, the

rate of serious adverse events of pneumonia observed with vilanterol was approximately one-

third of that reported with salmeterol (7, 24, 25).

An important limitation to this analysis is that SUMMIT was not designed as a

pneumonia trial. It is recognized that the clinical presentation of pneumonia may overlap with

that of an exacerbation (31), and radiographic confirmation of investigator-reported pneumonia

events was not required in this study. However, from a previous trial in patients with more

severely obstructed COPD, where radiographic imaging was required for suspected pneumonia

cases and exacerbations, the radiographically defined events were always lower than those

reported by the investigator, and the proportionate increase with FF/VI compared with VI in

investigator-reported cases paralleled that of cases confirmed with imaging modalities (7).

Moreover, when we combined exacerbation and pneumonia events in this trial, the treatment

benefit afforded by the FF-containing arms was not diminished, suggesting that we did not

underestimate pneumonia cases.

In conclusion, ICS-containing regimens are efficacious in the maintenance treatment of

patients with COPD, particularly for those with an exacerbation history, to delay the time to first

and decrease the rate of exacerbations. Concerns have been appropriately raised regarding the

potential risk for pneumonia when such therapies are prescribed. However, for patients with

moderate airflow limitation, this risk does not appear to be enhanced. These data therefore

suggest that the benefit: risk profile of an ICS-containing regimen may be greater for patients

with moderate airflow limitation compared with those with more severe disease.

References

1. World Health Organization. The top 10 causes of death. The 10 leading causes of

death in the world, 2000 and 2012;[accessed 2016 March 24]. Available from:

http://www.who.int/mediacentre/factsheets/fs310/en.

2. Jackson ML, Nuezil KM, Thompson WW, Shay DK, Yu O, Hanson CA, Jackson LA.

The burden of community-acquired pneumonia in seniors: Results of a population-

based study. Clin Inf Dis 2004; 39:1642–1650.

3. Lange P, Vestbo J, Nyboe J. Risk factors for death and hospitalisation from

pneumonia. A prospective study of a general population. Eur Respir J 1995; 8:1694–

1698.

4. Crim C, Calverley PMA, Anderson JA, Celli B, Ferguson GT, Jenkins C, Jones PW,

Willits LR, Yates JC, Vestbo J. Pneumonia risk in COPD patients receiving inhaled

corticosteroids alone or in combination: TORCH study results. Eur Respir J 2009;

34:641–647.

5. Sharafkhaneh A, Southard JG, Goldman M, Urinyak T, Martin UJ. Effect of

budesonide/formoterol pMDI on COPD exacerbations: A double-blind randomized

study. Respir Med 2012; 106:257–268.

6. Kew KM, Seniukovich A. Inhaled steroids and risk of pneumonia for chronic

obstructive pulmonary disease. Cochrane Database Syst Rev 2014; 3:CD010115.

7. Crim C, Dransfield MT, Bourbeau J, Jones PW, Hanania NA, Mahler DA, Vestbo J,

Wachtel A, Martinez FJ, Barnhart F, Lettis S, Calverley PMA. Pneumonia risk with

inhaled fluticasone furoate and vilanterol compared with vilanterol alone in patients

with COPD. Ann Am Thorac Soc 2015; 12: 27-34.

8. ADVAIR® DISKUS® [package insert]. Research Triangle Park, NC: GlaxoSmithKline; 2016

9. BREO®  ELLIPTA®  [package insert].  Research Triangle Park, NC:  GlaxoSmithKline; 2015.

10. EMA completes review of inhaled corticosteroids for chronic obstructive pulmonary

disease; [accessed 2016 May 6]. Available from:

http://www.ema.europa.eu/ema/index.jsp?curl=pages/news_and_events/news/

2016/04/news_detail_002521.jsp&mid=WC0b01ac058004d5c1.

11. Droemann D, Goldmann T, Tiedje T, Zabel P, Dalhoff K, Schaaf B. Toll-like

receptor 2 expression is decreased on alveolar macrophages in cigarette smokers and

COPD patients. Respir Res 2005; 6:68.

12. MacRedmond RE, Greene CM, Dorscheid DR, McElvaney NG, O’Neill SJ.

Epithelial expression of TLR4 is modulated in COPD and by steroids, salmeterol and

cigarette smoke. Respir Res 2007; 8:84.

13. Berenson CS, Kruzel RL, Eberhardt E, Sethi S. Phagocytic dysfunction of human

alveolar macrophages and severity of chronic obstructive pulmonary disease. J Infect

Dis 2013; 208:2036-45.

14. Medeiros AI, Serezani CH, Lee SP, Peters-Golden M. Efferocytosis impairs

pulmonary macrophage and lung antibacterial function via PGE2/EP2 signaling. J Exp

Med 2009; 206:61-68.

15. Janssen WJ, McPhillips KA, Dickinson MG, Linderman DJ, Morimoto K, Xiao YQ,

Oldham KM, Vandivier RW, Henson PM, Gardai SJ. Surfactant proteins A and D

suppress alveolar macrophage phagocytosis via interaction with SIRP. Am J Respir

Crit Care Med 2008; 178:158-67.

16. McCubbrey AL, Sonstein J, Ames TM, Fereeman CM, Curtis JL. Glucocorticoids

relieve collectin-driven suppression of apoptotic cell uptake in murine alveolar

macrophages through downregulation of SIRP. J Immunol 2012; 189:112-119.

17. Stolberg VR, McCubbrey AL, Freeman CM, Brown JP, Crudgington SW, Taitano

SH, Saxton BL, Mancuso P, Curtis JL. Glucocorticoid-augmented efferocytosis

inhibits pulmonary pneumococcal clearance in mice by reducing alveolar macrophage

bactericidal function. J Immunol 2015 doi:10.4049/jimmunol.1402217.

18. Sethi S, Murphy TF. Infection in the pathogenesis and course of chronic obstructive

pulmonary disease. N Engl J Med 2008; 359:2355-65.

19. Vestbo J, Anderson J, Brook RD, Calverley PMA, Celli BR, Crim C, Haumann B,

Martinez F, Yates J, Newby DE. The study to understand mortality and morbidity in

COPD (SUMMIT) study protocol. Eur Respir J 2013; 41:1017-22.

20. Vestbo J, Anderson J, Brook RD, Calverley PMA, Celli BR, Crim C, Martinez F, Yates J,

Newby DE. Fluticasone furoate and vilanterol and survival in chronic obstructive

pulmonary disease with heightened cardiovascular risk (SUMMIT): a double-blind

randomised controlled trial. Lancet 2016; 387:1817-26.

21. McGarvey LP, Matthias J, Anderson JA, Zvarich M, Wise RA. Ascertainment of

cause-specific mortality in COPD: operations of the TORCH clinical endpoints

committee. Thorax 2007; 62:411-15.

22. From the Global Strategy for the Diagnosis, Management and Prevention of COPD,

Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2016. Available

from: http://www.goldcopd.org/.

23. Festic E, Scanlon P.D. Incident pneumonia and mortality in patients with chronic

obstructive pulmonary disease. Am J Respir Crit Care Med 2015; 191:141-48.

24. Ferguson GT, Anzueto A, Fei R, Emmett A, Knobil K, Kalberg C. Effect of

fluticasone propionate/salmeterol (250/50 µg) or salmeterol (50 µg) on COPD

exacerbations. Respir Med 2008; 102:1099–1108.

25. Anzueto A, Ferguson GT, Feldman G, Chinsky K, Seibert A, Emmett A, Knobil K,

O’Dell D, Kalberg C, Crater G. Effect of fluticasone propionate/salmeterol (250/50)

on COPD exacerbations and impact on patient outcomes. COPD 2009; 6:320–329.

26. ICH Harmonised Tripartite Guideline. Choice of control group and related issues in

clinical trials. E10; Step 4 Version, July 2000; [accessed 2016 June 9]. Available

from: http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/

Efficacy/E10/Step4/E10_Guideline.pdf

27. LeVine AM, Whitsett JA. Pulmonary collectins and innate host defense of the lung.

Microbes Infect 2001; 3:161-66.

28. Xu X, McCormick-Shannon K, Voelker DR, Mason RJ. KGF increases SP-A and SP-

D mRNA levels and secretion in cultured rat alveolar type II cells. Am J Respir Cell

Mol Biol 1998; 18:168-78.

29. Gobran LI, Rooney SA. Pulmonary surfactant secretion in briefly cultured mouse

type II cells. Am J Physiol Lung Cell Mol Physiol 2004; 286:L331-36.

30. Janssen WJ, McPhillips KA, Dickinson MG, Linderman DJ, Morimoto K, Xiao YQ,

Oldham KM, Vandivier RW, Henson PM, Gardai SJ. Surfactant proteins A and D

suppress alveolar macrophage phagocytosis via interaction with SIRP. Am J Respir

Crit Care Med 2008; 178:158-67.

31. Calverley PMA, Stockley RA, Seemungal TAR, Hagan G, Willits LR, Riley JH,

Wedzicha JA. Reported pneumonia in patients with COPD. Findings from the

INSPIRE study. Chest 2011;139:505–512.

Table 1 – Baseline Characteristics

Placebo

n=4131

Fluticasone

furoate

n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

Age (years) 65 ± 8 65 ± 8 65 ± 8 65 ± 8

Female 1050 (25%) 1098 (26%) 1071 (26%) 1019 (25%)

Race White

Asian

Other

3348 (81%)

680 (16%)

103 (2%)

3380 (81%)

683 (16%)

94 (2%)

3360 (81%)

680 (16%)

100 (2%)

3350 (81%)

680 (16%)

110 (3%)

Body mass index (kg/m2) 28 ± 6 28 ± 6 28 ± 6 28 ± 6

Current Smokers 1949 (47%) 1952 (47%) 1941 (47%) 1875 (45%)

Smoking Pack Years 41 ± 25 41 ± 24 41 ± 24 40 ± 24

%predicted FEV1 60 ± 6 60 ± 6 60 ± 6 60 ± 6

Exacerbation

History

0

1

≥ 2

2454 (59%)

1049 (25%)

628 (15%)

2552 (61%)

997 (24%)

608 (15%)

2510 (61%)

994 (24%)

636 (15%)

2532 (61%)

1007 (24%)

601 (15%)

mMRC 1

2

3

0

3051 (74%)

1040 (25%)

1 (<1%)

3123 (75%)

983 (24%)

1 (<1%)

3027 (73%)

1071 (26%)

1 (<1%)

3057 (74%)

1020 (25%)

Placebo

n=4131

Fluticasone

furoate

n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

4 39 (<1%) 50 (1%) 40 (<1%) 60 (1%)

Region USA

Asia

Europe 1

Europe 2

RoW

655 (16%)

666 (16%)

1330 (32%)

1158 (28%)

322 (8%)

658 (16%)

676 (16%)

1341 (32%)

1157 (28%)

325 (8%)

659 (16%)

673 (16%)

1332 (32%)

1155 (28%)

321 (8%)

659 (16%)

671 (16%)

1331 (32%)

1156 (28%)

323 (8%)

Diabetes Mellitus History 1244 (30%) 1260 (30%) 1235 (30%) 1281 (31%)

Congestive Heart Failure

History

877 (21%) 849 (20%) 886 (21%) 861 (21%)

Data are mean ± Standard deviation, or n (%). mMRC = modified Medical Research Council dyspnea score.

RoW = Rest of World

Table 2 – Summary of Investigator-reported On-treatment Pneumonia

Placebo

n=4131

Fluticasone

furoate

n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

IP Exposure (patient-years)

6614 6889 6955 7038

Subjects reporting pneumonia related AEs, n (%)

214 (5.2%) 228 (5.5%) 163 (3.9%) 237 (5.7%)

Pneumonia-related AE event rate/100 pt-yrs (number of events)

3.84 (254) 4.24 (292) 2.77 (193) 3.95 (278)

Subjects reporting pneumonia related SAEs, n (%)

127 (3.1%) 146 (3.5%) 104 (2.5%) 140 (3.4%)

Pneumonia-related SAE event rate/100 pt-yrs (number of events)

2.22 (147) 2.51 (173) 1.64 (114) 2.24 (158)

Subjects with pneumonia-related fatal SAEs, n (%)

10 (0.2%) 17 (0.4%) 5 (0.1%) 16 (0.4%)

Pneumonia-related fatal SAE event rate/100 pt-yrs (number of events)

0.15 (10) 0.25 (17) 0.07 (5) 0.23 (16)

* Deaths were considered on-treatment if they occurred between treatment start and seven days after treatment stop date, inclusive.

Note: AEs and SAEs (including fatal) were events reported by the investigator

Table 3 – Analysis of Time to First On-treatment Pneumonia

Placebo

n=4131

Fluticasone

furoate n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

Pneumonia AE

Status at Treatment End Had Event

214 (5.2%) 228 (5.5%) 163 (3.9%) 237 (5.7%)

Active vs. Placebo Hazard Ratio 1.035 0.722 1.038 95% CI (0.859, 1.247) (0.589, 0.886) (0.863, 1.249)

p-value 0.716 0.002 0.693

Combination vs. Component Hazard Ratio

1.003 1.437

95% CI (0.836, 1.203) (1.177, 1.754)

p-value 0.977 <0.001

Pneumonia SAE

Status at Treatment End Had Event

127 (3.1%) 146 (3.5%) 104 (2.5%) 140 (3.4%)

Active vs. Placebo Hazard Ratio 1.117 0.778 1.022 95% CI (0.880, 1.416) (0.600, 1.008) (0.804, 1.300)

p-value 0.364 0.057 0.858

Combination vs. Component Hazard Ratio

0.916 1.315

95% CI (0.726, 1.155) (1.020, 1.695)

p-value 0.456 0.035

Table 4 – Analysis of Time to First On-treatment Pneumonia or Exacerbation Composite

Placebo

n=4131

Fluticasone

furoate n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

Pneumonia AE or Moderate/Severe ExacerbationsStatus at Treatment End Had Event

1243 (30.1%)

1262 (30.4%) 1210 (29.2%) 1109 (26.8%)

Active vs. Placebo Hazard Ratio 0.977 0.920 0.804 95% CI (0904, 1.057) (0.850, 0.996) (0.742, 0.872)

p-value 0.563 0.039 <0.001

FF/VI vs. Component Hazard Ratio

0.823 0.874

95% CI (0.759, 0.892) (0.806, 0.949)

p-value <0.001 0.001

Pneumonia SAE or Severe ExacerbationStatus at Treatment End Had Event

372 (9.0%)

343 (8.3%) 328 (7.9%) 319 (7.7%)

Active vs. Placebo Hazard Ratio 0.896 0.839 0.798 95% CI (0.774, 1.038) (0.723, 0.974) (0.687, 0.927)

p-value 0.142 0.021 0.003

FF/VI vs. Component Hazard Ratio

0.890 0.951

95% CI (0.764, 1.037) (0.815, 1.109)

p-value 0.136 0.520

Table 5 – Summary of number of deaths within 30 days of last on-treatment pneumonia onset.

Placebo

n=4131

Fluticasone

furoate n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

Had on-treatment pneumonia AEs, n (%)

214 (5.2%)

228 (5.5%) 163 (3.9%) 237 (5.7%)

Died within 30 days of last on-treatment pneumonia onset, n (%)*

19 (8.9%) 22 (9.6%) 15 (9.2%) 20 (8.4%)

Pneumonia AEs are considered on-treatment if they occurred between treatment start and the day after treatment stop date, inclusive.

*percentages calculated from the number of patients with a pneumonia

Table 6 – Analysis of Time to On-treatment Death Adjudicated as Pneumonia

Placebo

n=4131

Fluticasone

furoate n=4157

Vilanterol

n=4140

Combination

therapy

n=4140

Number of Patients with On-treatment Pneumonia Death (%)

9 (0.2%) 10 (0.2%) 6 (0.1%) 13 (0.3%)

On-treatment adjudicated pneumonia deaths* event rate/100 pt-yrs

0.14 0.15 0.09 0.18

Active vs. Placebo Hazard Ratio

1.079 0.632 1.330

95% CI (0.438, 2.657) (0.225, 1.777) (0.567, 3.117)

p-value 0.868 0.384 0.512

Combination vs. Component Hazard Ratio

1.232 2.105

95% CI (0.539, 2.818) (0.800, 5.535)

p-value 0.621 0.132

Figure 1a

Figure 1b

Figure 2 Risk factors for Pneumonia in Placebo Arm

Supplemental Table

Table E2. Summary of on-treatment pneumonia AEs split by exacerbation history in the 1-year

prior to study entry

Placebo Fluticason

e furoate

Vilanterol Combination

therapy

0 exacerbations in prior year

n=2454 n=2552 n=2510 n=2532

IP Exposure (patient-years)

3825 4103 4114 4209

Subjects reporting pneumonia related AEs, n (%)

110 (4%) 147 (6%) 88 (4%) 136 (5%)

Pneumonia-related AE event rate/100 pt-yrs (number of events)

3.40 (130) 4.68 (192) 2.41 (99) 3.82 (161)

Subjects reporting pneumonia related SAEs, n (%)

62 (3%) 98 (4%) 49 (2%) 71 (3%)

Pneumonia-related SAE event rate/100 pt-yrs (number of events)

1.78 (68) 2.88 (118) 1.29 (53) 1.90 (80)

≥1 exacerbations in prior year

n=1677 n=1605 n=1630 n=1608

IP Exposure (patient-years)

2789 2787 2841 2829

Subjects reporting pneumonia related AEs, n (%)

104 (6%) 81 (5%) 75 (5%) 101 (6%)

Pneumonia-related AE event

4.45 (124) 3.59 (100) 3.31 (94) 4.14 (117)

rate/100 pt-yrs (number of events)Subjects reporting pneumonia related SAEs, n (%)

65 (4%) 48 (3%) 55 (3%) 69 (4%)

Pneumonia-related SAE event rate/100 pt-yrs (number of events)

2.83 (79) 1.97 (55) 2.15 (61) 2.76 (78)