Changing patterns in long-acting bronchodilator trials in chronic obstructive pulmonary disease

Abstract

Chronic obstructive pulmonary disease (COPD) is a major cause of morbidity and mortality worldwide. Developments in the understanding of COPD have led to standard guidelines for diagnosis, treatment, and spirometry assessments, which have in turn influenced trial designs and inclusion criteria. Substantial clinical evidence has been gained from clinical trials and supports a positive approach to COPD management. However, there appear to be changing trends in recent trials. Large bronchodilator studies have reported lower improvements in trough forced expiratory volume in 1 second (FEV₁) values versus placebo than were observed in earlier studies, while the rate of FEV₁ decline seems to be lower in more recent trials. In addition, recent evidence has called into question the usefulness of bronchodilator reversibility testing as a trial inclusion criterion. Baseline patient populations and use of concomitant medications have also changed over recent years due to increased treatment options. The impact of these many variables on clinical trial results is explored, with a particular focus on changes in inclusion criteria and patient baseline demographics.

Introduction

Chronic obstructive pulmonary disease (COPD) is a major public health concern and is currently the fourth leading cause of death in the United States.¹ COPD is a smoking-related lung disease that progresses over several years with increasing respiratory symptoms (eg, dyspnea, coughing, and sputum production) and systemic effects (eg, weight loss, skeletal muscle dysfunction, and increased risk of cardiovascular disease).¹

In previous decades, treatment options for COPD were limited. However, the past 10–15 years have seen a large increase in clinical trials examining different treatments for this disease and substantial developments in COPD management, for example, long-acting β₂-agonists, long-acting muscarinic antagonists, and their combination with inhaled corticosteroids (ICS). Bronchodilators are the mainstay of pharmacologic treatment of COPD, and the most widely used are β₂-agonists (eg, salbutamol, terbutaline, formoterol, and salmeterol), anticholinergics (eg, ipratropium and tiotropium), and methylxanthines (eg, theophylline). Bronchodilators in combination with each other and with ICS have also been investigated.

Understanding of the clinical phenotypes of COPD is evolving, and there is increasing clinical evidence to guide COPD management. However, several unanswered questions remain due to the variability in clinical trial procedures and patient populations. This review of trials of long-acting bronchodilators examines trends observed in recent COPD trials and explores the factors that may have impacted on the results, with a particular focus on changes in patients’ lung function over time.

Literature analysis

A PubMed literature search (restricted to English literature; no date restriction) using the terms COPD (MeSH) and salmeterol or formoterol or tiotropium was carried out on May 14, 2009, and yielded 223 articles (search limits were: clinical trial, meta-analysis, randomized controlled trial, and humans). To minimize problems due to small sample sizes or short study durations, only full articles reporting randomized, placebo-controlled clinical trials of at least 500 patients and duration of at least 6 months were selected (18 articles) for detailed assessment.

Recent trends in bronchodilator clinical trial results

Forced expiratory volume in 1 second (FEV₁) is the principal measure of lung function used in the assessment of COPD. A minimum clinically important difference (MCID) has not been defined, although improvement of about 100–120 mL in trough FEV₁ has been suggested as a possible benchmark for bronchodilators and anti-inflammatory agents, but this is not based on quantitative data.^2,3 Regulators may consider a change in FEV₁ of 5%–10% from baseline values to be clinically meaningful, and that a <3% change from baseline is not clinically important.³ Assuming a baseline FEV₁ of 1.1 L, this would equate to an increase of 55–110 mL, and not <33 mL. Further consideration of a clinically meaningful change is needed, particularly since recent results from large clinical studies evaluating a range of bronchodilators have reported lower improvements in trough FEV₁ values for monotherapies than those reported in earlier studies.

For example, several studies have reported results of the long-acting anticholinergic, tiotropium. The recent 4-year Understanding Potential Long-term Impacts on Function with Tiotropium (UPLIFT) trial (N = 5992) investigated tiotropium versus placebo in patients with COPD, and throughout the study, trough FEV₁ was significantly improved versus placebo by 87–103 mL (Table 1).⁴ This contrasts with a number of previous studies that reported improvements in trough FEV₁ values with tiotropium over placebo that ranged from 100 to 150 mL (Table 1 and Figure 1).^5–11

Table 1.

Summary of bronchodilator trials in COPD

Reference, trial name (duration), and treatment doses	Inclusion criteria		Other medications		Treatment arm (n)	Baseline demographics					Results: mean change in FEV1 vs placebo, mL
	Pretreatment FEV1, % predicted	COPD definition	Prestudy medications reported	On-study medications allowed		Female, %	Mean age, years	Current/ex-smoker, %	Smoking history, pack-years	Mean FEV1, % predicted	Peak (postdose)	Trough (predose)
Salmeterol
Mahler et al13 (6 months), salmeterol 50 μg bid, fluticasone 500 μg bid	<65	ATS (1995)	ICS	Only theophylline and prn albuterol	Salmeterol + fluticasone (165)	38	62	46/54	55a	41	231†	159†
					Salmeterol (160)	36	64	46/54	53a	40	191†	92†
					Fluticasone (168)	39	64	46/54	54a	41	101†	105†
					Placebo (181)	25	64	54/46	60a	41	–	–
Hanania et al14 (6 months), salmeterol 50 μg bid, fluticasone 250 μg bid	<65	ATS (1995)	ICS	Only theophylline and prn albuterol	Salmeterol + fluticasone (178)	39	63	43/57	53a	41	214†	161†
					Salmeterol (177)	42	64	51/49	57a	42	140†	92†
					Fluticasone (183)	34	63	48/52	60a	42	89†	112†
					Placebo (185)	32	65	47/53	56a	42	–	–
Calverley et al12 TRISTAN (1 year), salmeterol 50 μg bid, fluticasone 500 μg bid	25–70	Clinical diagnosis	ICS, LABA	Yes, except LABA and ICS	Salmeterol + fluticasone (358)	25	63	52/48	42	45	76***	132***
					Salmeterol (372)	30	63	51/49	44	44	28	59***
					Fluticasone (374)	30	64	53/47	42	45	46†	38*
					Placebo (361)	25	63	47/53	43	44	–	–
Stockley et al21 (1 year), Salmeterol 50 μg bid	<70	ERS (1995)	ICS, LABA, A, xanthines	Yes, except oxygen	Salmeterol (316)	23	62	47/53	38–41b	46	120***	NR
					Placebo (318)	24	62	46/54	40	46	–	–
Calverley et al20 TORCH (3 years), salmeterol 50 μg bid, fluticasone 500 μg bid	<60	Clinical diagnosis	ICS, LABA	Yes, except LABA and CS	Salmeterol + fluticasone (1533)	25	65	43/57	47	44	92c,**	NR
					Salmeterol (1521)	24	65	43/57	49	44	42c,**	–
					Fluticasone (1534)	25	65	43/57	49	44	47c,**	–
					Placebo (1524)	24	65	43/57	49	44	–	–
Formoterol
Calverley et al53 (1 year), Budesonide 320 μg bid + formoterol 9 μg bid, budesonide 400 μg bid, Formoterol 9 μg bid	≤50	GOLD (2001) stages III and IV	ICS, SABA, LABA, A, xanthines, β2-agonist	Oral CS; antibiotics; parenteral steroids and/or nebulized treatment (single dose), terbutaline	Budesonide + formoterol (254)	22	64	33/67	39	36	14%**	NR
					Budesonide (257)	26	64	39/61	39	36	2%	NR
					Formoterol (255)	25	63	36/64	38	36	8%**	NR
					Placebo (256)	25	65	30/70	39	36	–	–
Szafranski et al54 (1 year), budesonide 320 μg bid + formoterol 9 μg bid, budesonide 400 μg bid, formoterol 9 μg bid	≤50	GOLD (2001) stages IIB and III	ICS, SABA, LABA, A, xanthines, β2-agonist	Only prn terbutaline	Budesonide + formoterol (208)	24	64	30/70	44	36	15%**	NR
					Budesonide (198)	20	64	36/64	44	37	5%*	NR
					Formoterol (201)	24	63	38/62	45	36	14%**	NR
					Placebo (205)	17	65	34/66	45	36	–	–
Campbell et al22 (6 months), formoterol 9 μg bid + terbutaline 0.5 mg prn, formoterol 9 μg bid + formoterol 4.5 μg bid prn, placebo bid + terbutaline 0.5 mg prn	40–70	Clinical diagnosis	ICS, A, xanthines	Yes, except disodium cromoglycate, ephedrine, antihistamines, β-blockers, bronchodilators	Formoterol + terbutaline prn (215)	39	60	54/46	37	53	92d,*	NR
					Formoterol + formoterol prn (225)	29	60	56/44	38	54	163d,**	NR
					Placebo + terbutaline prn (217)	27	60	55/45	37	54	–	–
Tashkin et al15 SHINE (6 months), budesonide 320 μg/formoterol 9 μg bid, budesonide 160 μg/formoterol 9 μg bid, budesonide 320 μg bid + formoterol 9 μg bid, budesonide 320 μg bid, formoterol 9 μg bid	≤50	Clinical diagnosis	ICS, SAMA, LAMA, SABA, LABA, xanthines, ICS/LABA combo, SABA	Yes, except LAMA, LABA, SABA, oral β2-agonist, antileukotrienes, xanthines	Budesonide/formoterol 320/9 (277)	32	63	44/56	40a	39	170e,**	80**
					Budesonide/formoterol 160/9 (281)	36	64	45/55	40a	40	160e,**	50*
					Budesonide + formoterol (287)	26	64	42/58	42a	39	NR	NR
					Budesonide (275)	32	63	43/57	41a	40	0	0
					Formoterol (284)	34	64	42/58	40a	40	140e,**	40†
					Placebo (300)	31	63	40/60	40a	41	–	–
Tiotropium
Casaburi et al5 (1 year), tiotropium 18 μg qd	≤65	ATS (1995)	ICS, oral CS, A, β2-agonist, theophylline	Yes, except A and LABA	Tiotropium (550)	33	65	100f	63	39	140–220*	120–150*
					Placebo (371)	37	65	100f	59	38	–	–
Donohue et al6,g (6 months), tiotropium 18 μg qd, salmeterol 50 μg bid	≤60	Clinical diagnosis	ICS, oral CS, A, inhaled/oral β2-agonist, theophylline	Yes, except inhaled A and LABA	Tiotropium (209)	26	65	100f	47	41	244***	137***
					Salmeterol (213)	25	65	100f	48	39	161***	85***
					Placebo (201)	25	66	100f	46	41	–	–
Brusasco et al7,g (6 months), tiotropium 18 μg qd, salmeterol 50 μg bid	≤65	Clinical diagnosis	NR	NR	Tiotropium (402)	23	64	100f	44	39	NR	120*
					Salmeterol (405)	25	64	100f	45	38	NR	90*
					Placebo (400)	24	65	100f	42	39	–	–
Niewoehner et al8 (6 months), tiotropium 18 μg qd	≤60	Clinical diagnosis	ICS, oral CS, ipratropium, inhaled β2-agonist, theophylline, oxygen, antileukotrienes	Yes, except A	Tiotropium (914)	2	68	29/71	67	36	170**	100**
					Placebo (915)	1	68	30/70	69	36	–	–
Dusser et al9 (1 year), tiotropium 18 μg qd	30–65	Clinical diagnosis	ICS, OC, IV/IM CS, oral β2-agonist, inhaled SABA, inhaled LABA, SAMA, oxygen, antileukotrienes, xanthines	Yes, except inhaled A, oral/inhaled LABA, and theophylline	Tiotropium (500)	11	65	27/73	NR	48	NR	120***
					Placebo (510)	13	65	24/76	NR	48	–	–
Chan et al10 (1 year), tiotropium 18 μg qd	≤65	Clinical diagnosis	ICS, oral CS, A, oral β2-agonist, inhaled SABA, inhaled LABA, theophylline, oxygen, antileukotrienes	Yes, except inhaled A and oral β2-agonist	Tiotropium (608)	41	67	32/68	50	39	NR	100***
					Placebo (305)	39	67	30/70	51	39	–	–
Tashkin et al4 UPLIFT (4 years), tiotropium 18 μg qd	≤70	Clinical diagnosis	ICS, oral CS, SABA, LABA, SAMA, LAMA, theophylline, oxygen, mucolytic agent, antileukotrienes	Yes, except inhaled A	Tiotropium (2986)	25	65	29/71	49	40	47–65h,**	87–103***
					Placebo (3006)	26	65	30/70	48	39	–	–
Vogelmeier et al23 (6 months), tiotropium 18 μg qd + formoterol 10 μg bid, tiotropium 18 μg qd, formoterol 10 μg bid	<70	GOLD (2001)	NR	ICS and salbutamol	Tiotropium + formoterol (207)	21	63	100f	38	50	>120i,**	NR
					Tiotropium (221)	21	63	100f	39	52	>120i,**	NR
					Formoterol (210)	24	62	100f	35	52	>120i,**	NR
					Placebo (209)	22	63	100f	40	51	–	–
Tonnel et al11 (9 months), tiotropium 18 μg qd	20–70	ATS (1995)	NR	Yes, except oral or inhaled LABA, SAMA, β-blockers, antileukotrienes	Tiotropium (266)	13	65	24/76	44	47	NR	100***
					Placebo (288)	15	64	30/70	43	46	–	–

Notes:

^aMedian value (other values are mean or do not specify median or mean);

^bPack-years for ex-smokers;

^cPostbronchodilator, no time given;

^d30 min to 2 h post study drug;

^e1 h postbronchodilator;

^fAll patients were current or exsmokers, but the proportion of current smokers was not reported;

^gBrusasco et al⁷ reported combined results of two 6-month studies, including one previously published by Donohue et al,⁶ but not referenced by Brusasco et al. An erratum to correct this was subsequently published (Thorax. 2005;60:105);

^h90-min post study drug and ipratropium administration and 30 min after albuterol administration;

ⁱ2-h postdose.

^†P < 0.05 vs placebo;

^*P < 0.01 vs placebo;

^**P < 0.001 vs placebo;

^***P < 0.0001 vs placebo.

Abbreviations: A, anticholinergic; bid, twice daily; CS, corticosteroid; FEV₁, forced expiratory volume in 1 second; ICS, inhaled corticosteroid; LABA, long-acting β₂-agonist; LAMA, long-acting muscarinic antagonist; NR, not reported; OC, oral corticosteroid; SABA, short-acting β₂-agonist; SAMA, short-acting muscarinic antagonist; TORCH, Towards a Revolution in COPD Health; TRISTAN, TRial of Inhaled STeroids ANd long-acting β₂-agonists; prn, pro-re-nata (as-needed); qd, once daily.

Figure 1.

Treatment effect with tiotropium (left-hand axis) and baseline FEV₁ (right-hand axis) against study publication date (Donohue et al,⁶ Casaburi et al (treatment effect results reported as a range),⁵ Brusasco et al,⁷ Niewoehner et al,⁸ Dusser et al,⁹ Chan et al,¹⁰ Tonnel et al,¹¹ and Tashkin et al⁴ (treatment effect results reported as a range).

Abbreviation: FEV₁, forced expiratory volume in 1 second.

The long-acting β₂-agonist, salmeterol, has demonstrated varying improvements in trough FEV₁ compared with placebo of 59–92 mL (Table 1).^6,7,12–14 A recent study of another long-acting β₂-agonist, formoterol, showed improved trough FEV₁ compared with placebo (40 mL), but most of the formoterol studies were not powered on trough FEV₁ (Table 1).¹⁵ Since our literature search was carried out, a large, 12-month, randomized, double-blind study of 1964 patients has been reported, in which budesonide/formoterol demonstrated improvements in trough FEV₁ compared with placebo (~100–110 mL; P < 0.001).¹⁶ Salmeterol in combination with the ICS, fluticasone, has demonstrated improvements in trough FEV₁ compared with placebo ranging from 132 to 161 mL,^12–14 and the combination of formoterol and budesonide improved trough FEV₁ compared with placebo by a range of 50–80 mL (Table 1).¹⁵ Short-acting β₂-agonists (eg, salbutamol) and anticholinergics (eg, ipratropium) have a 4–6 h duration of action and, therefore, have a lower effect on trough FEV₁ compared with longer-acting agents and were not included in the literature search.¹⁷

A new long-acting antimuscarinic agent, aclidinium bromide, is also under development and two large, 12-month, randomized, double-blind studies of 1647 patients in total have been reported since our literature search was carried out.^18,19 When given once daily, 200 μg inhaled aclidinium bromide demonstrated improvements in trough FEV₁ compared with placebo of 59–67 mL (P < 0.001) at week 28.

Variable results in peak (ie, postdose) FEV₁ have also been observed for individual agents across different trials (Table 1). Peak FEV₁ compared with placebo ranged from 28 to 191 mL^{6,12–14,20,21} for salmeterol monotherapy, 76–231 mL^6,12–14,20 for salmeterol + fluticasone, 92–140 mL^15,22 for formoterol, 160–170 mL¹⁵ for formoterol + budesonide, and 47–244 mL for tiotropium.^4–6,8,23

There are no immediately identifiable reasons for the differences in measured effect size for the same agent in different studies. It may simply be due to random sampling from the worldwide population of COPD patients; however, it may be useful to explore other possible mechanisms, which are evaluated in this review.

Observed changes to COPD clinical trial procedures, inclusion criteria, and baseline demographics

Guidelines for COPD diagnosis and definition

Concerted efforts to define COPD have led to the development of guidelines for diagnosis, such as the Global Initiative for Chronic Obstructive Lung Disease (GOLD) and American Thoracic Society (ATS)/European Respiratory Society (ERS) guidelines, which define COPD as a preventable and treatable disease characterized by airflow limitation that is not fully reversible.^1,24 Compared with older definitions,^25,26 this represents a paradigm shift toward positive thinking in COPD management. Changes in the understanding of the disease and evolving guidelines for its management will impact on clinical trial designs and procedures, and trial results, in turn, influence clinical guidelines.

Lung function assessments

Alongside guidelines for the diagnosis of COPD, the standardization of spirometry has also improved over recent years. Current spirometry guidelines were issued by the ATS/ERS in 2005,²⁷ which updated previous guidelines published in 1995 and 1987.^28,29 Recommendations specify acceptability and reproducibility criteria for forced vital capacity (FVC) and FEV₁ assessments. A minimum of three acceptable assessments should have repeatability within 150 mL (or within 100 mL for those with an FVC ≤1.0 L),²⁷ which has shifted from the more lenient 1995 criteria of repeatability within 200 mL.²⁸ Once these criteria are met, ATS/ERS guidelines state that the largest FVC and FEV₁ values should be selected,²⁷ and although these are widely used, others recommend using mean values, such as the mean of the best three of five acceptable assessments.²

Spirometric tests performed in different clinics are often subject to variability due to technical and personal factors such as differences in the use and validation of equipment, as well as differences in measurement procedures, interpretation, and quality control.^2,27 To help overcome such limitations, large trials such as UPLIFT have used centralized quality assurance of spirometry data.

Effect of clinical trial on patient baseline demographics

COPD is a heterogeneous disease, and differences in the patient inclusion criteria can substantially affect trial results and their translation into clinical practice guidelines.

Disease staging of COPD has been classified based on spirometry,^1,24 and the relative proportions of patients with moderate, severe, and very severe COPD within trials may have an impact on results. For example, trough FEV₁, the usual primary outcome measure in trials of long-term bronchodilators, is lower in patients with more severe disease and thus the magnitude of improvement in such patients would be expected to be lower.³⁰ This has been shown in two studies of fluticasone propionate/salmeterol (250/25 μg) in different patient populations. In a study of patients with COPD who had a mean baseline FEV₁ of 42% predicted, fluticasone propionate/salmeterol increased trough FEV₁ by 165 mL from baseline to end point,¹⁴ whereas in a second study of patients with more severe COPD (mean baseline FEV₁ of 33% predicted), fluticasone propionate/salmeterol decreased trough FEV₁ by 12 mL from baseline to end point.³¹ Subgroup analyses from the Towards a Revolution in COPD Health (TORCH) and UPLIFT trials have demonstrated improvements in lung function outcomes in patients with moderate (GOLD stage II) COPD.^32,33 The TORCH analysis demonstrated improvements across different subgroups of GOLD stages, although lung function improvements decreased with increasing disease severity.³³ Similar results have been observed in a subanalysis of the TRial of Inhaled STeroids ANd long-acting β2 agonists (TRISTAN) study.³⁴

Smoking is a key factor in COPD development and, until recently, smoking cessation was the only intervention that has been prospectively shown to slow the rate of lung function decline in COPD.³⁵ The proportion of patients who are current smokers and patients’ smoking history are, therefore, key baseline characteristics to consider. In a comparison of different trials, rates of FEV₁ decline in different trials appear to be related to the proportion of current smokers at baseline (Figure 2). Indeed, a range of factors appears to influence FEV₁ decline, including age, gender, and baseline FEV₁, and could influence the rates observed in different studies.³⁶ In the TORCH study, a slower rate of FEV₁ decline in absolute milliliters per year was observed with patients ≥65 years of age, with females, and those with a baseline FEV₁ <30% predicted.³⁶ However, when the rate of FEV₁ decline was expressed as a percentage change per year, this relationship was preserved with patients ≥65 years of age, but not with females and those with a baseline FEV₁ <30% predicted. Moreover, a subanalysis of TRISTAN study data reported that the improvements in trough FEV₁ with salmeterol/fluticasone versus placebo were equivalent in women (by 152 mL; 95% confidence interval 95, 208) and men (by 127 mL; 95% confidence interval 94, 159; P = 0.455 for the gender interaction).³⁷ A recent analysis of the Framingham Offspring Cohort, which included 5124 male and female participants, showed that changes in lung function from adolescence to old age differ between healthy males and females.³⁸ This analysis of the natural history of chronic airflow obstruction also confirmed the deleterious effects of smoking, as the rate of decline in FEV₁ was increased in smokers compared with never-smokers. Quitting smoking earlier is better, as participants who quit after the age of 40 years showed no significant difference in FEV₁ decline compared with continuous smokers.

Figure 2.

Rate of FEV₁ decline in the placebo arm of long-term COPD studies.⁴,³⁶,⁵⁵–⁵⁷

Abbreviations: COPD, chronic obstructive pulmonary disease; FEV₁, forced expiratory volume in 1 second.

Effect of clinical trial dose selection

There has also been a change in the focus of new drug applications to regulatory agencies, which may have affected outcomes of more recent studies compared with older studies. For example, the US Food and Drug Administration (FDA) routinely aims to identify the dose with the most favorable risk–benefit profile due to concerns about safety. This is, in part, a consequence of safety concerns about long-acting β₂-agonists in asthma: while there is a dose-response for FEV₁ efficacy with formoterol, there is an escalating safety concern with higher doses.³⁹ Recent pivotal Phase III studies have included lower doses along with the Phase IIa-identified dose; therefore, lower improvement in trough FEV₁ and other end points may be seen compared with earlier studies. For example, Phase III trials of nebulized arformoterol investigated a 15 μg twice-daily dose, as well as 25 μg twice daily and 50 μg daily.^40,41 While the changes in trough or predose FEV₁ were similar with the three doses, the mean percentage change in FEV₁ AUC_(0–12) was 12.7%, 13.9%, and 18.9% with 15 μg twice daily, 25 μg twice daily, and 50 μg daily, respectively, compared with 2.7% with placebo and 9.8% with salmeterol. The lowest arformoterol dose (15 μg twice daily) was chosen by FDA for license in the US.⁴⁰

It has also been suggested that the rate of decline in lung function could be affected by bias from regression to the mean caused by missing data, such as in the TORCH study with salmeterol.⁴² This is due to the fact that no long-term COPD trials are designed to have a full intent-to-treat analysis of lung function decline. However, it has been argued that lung function decline is not influenced by regression to the mean in randomized, placebo-controlled trials such as TORCH, since any regression to the mean should affect all groups equally.⁴³

Concomitant medications

Another recent key change in COPD trials inclusion criteria relates to concomitant medications, which are now in wide-spread use. In contrast to earlier studies when treatment options were limited, patients at baseline will now often already have received a variety of short- and long-acting bronchodilators and ICS. These will have already provided benefit to the patients, and their increased use over recent years will have shifted the baseline characteristics of patients entering trials. In the TORCH study, 8%–9% and 18%–22% of patients had received prior medication with an inhaled long-acting β₂-agonist and an ICS, respectively, and 27%–29% of patients had received a combination of the two.³⁶ In the UPLIFT trial, 60% and 61% of patients were receiving prior medication with an inhaled long-acting β₂-agonist and an ICS, respectively.⁴

In many studies, however, patients are still required to stop the therapy they had been receiving during the run-in period prior to randomization, particularly if it is a member of the same class as the study medication. Withdrawal of maintenance treatment for COPD during a washout period may have two important effects. First, withdrawal of the drug may lead to worsening symptoms – so patients who do not meet stability criteria are not randomized to the study therapies. This would lead to selective exclusion of patients who may respond to the class of agents under test. Second, if the patients are randomized and receive placebo, there is evidence that those who have been withdrawn from either an ICS or a long-acting β₂-agonist during the run-in period are more likely to have an exacerbation during the trial.⁴⁴

Concomitant medications are now excluded from use during the trial period less frequently than in the past, in part because of ethical issues; patients in both study drug and placebo arms are now likely to receive one or more respiratory medication other than the study drug. The UPLIFT study allowed all respiratory therapeutics, with the exception of another inhaled anticholinergic agent, in both the tiotropium and placebo arms.⁴ During the trial, 72% and 74% of patients reported having taken an inhaled long-acting β₂-agonist and an ICS, respectively, while 46% reported taking a fixed combination of the two.⁴ The study authors speculate that the concurrent medical care received during the study may have contributed to the generally lower rates of FEV₁ decline across both the tiotropium and the placebo groups (which averaged 30 mL/year before bronchodilation and 41 mL after bronchodilation in the two groups). Post hoc analysis of a relatively small subgroup of patients in the UPLIFT trial who were not receiving maintenance therapy at baseline (13.5% of the total population) suggests that there may be an influence of concomitant therapy on outcome.⁴⁵ Another recent study has shown that there may be very clear benefits on FEV₁ of adding different classes of agent together in COPD, since the addition of budesonide + formoterol to tiotropium showed a significant (P < 0.001) increase in predose (ie, trough FEV₁) of 65 mL and postdose (peak) of 131 mL compared with tiotropium alone.⁴⁶ It, therefore, remains a reasonable and testable hypothesis that there may be additive effects of pharmacological therapy on FEV₁.

It is clear that prestudy and permitted concomitant therapies may have a significant impact on measured treatment effects due to selective recruitment, effects of concomitant therapy during the study, and events that occur during the trial, such as exacerbations.

Bronchodilator reversibility

Substantial progress in understanding the pathophysiology of COPD has been made in recent years. A key concern has been the differential diagnosis of COPD from chronic asthma. As asthma is associated with variable airway caliber, improvement in lung function (eg, FEV₁) after bronchodilator treatment (termed bronchodilator reversibility) was proposed as a method of distinguishing between the two diseases. Bronchodilator reversibility for an individual patient is commonly defined as a postbronchodilator increase in FEV₁ ≥12% and ≥200 mL from baseline (ATS²⁴ and GOLD⁴⁷ criteria) or as a ≥9% change in predicted FEV₁ (ERS criteria⁴⁸).⁴⁹ This approach has led to a common perception that COPD is irreversible, despite evidence that bronchodilator reversibility testing was not sensitive or specific enough to differentiate asthma from COPD using spirometry alone.⁵⁰

Previously, European patients, and some patients in the US, were selected for COPD clinical studies based on a lack of bronchodilator reversibility by assessing how FEV₁ changed following a single dose of bronchodilator (acute changes). Clinical studies then investigated reversibility as the primary efficacy end point, in terms of changes in prebronchodilator FEV₁ over time (chronic changes). Therefore, it is perhaps unsurprising that studies show a small response in terms of FEV₁ change.⁵¹ Inclusion based on a lack of bronchodilator reversibility may act as a self-fulfilling prophecy for the trial outcomes.

In two studies that investigated salmeterol and fluticasone propionate, each as monotherapy and also in combination, patient randomization was stratified by reversibility to albuterol and investigative site. The results showed that, in both studies, patients with reversibility (>12% predicted FEV₁) had a greater response to therapy than patients with nonreversibility (with <4% predicted FEV₁) (Figure 3).^13,14 Another recent study in patients with reversibility and nonreversibility has shown that although both groups showed improved lung function with fluticasone propionate/salmeterol, the response was greater with patients with reversibility.⁵² Patients with a history of asthma (including childhood asthma) were excluded, and inclusion criteria required diagnosis of COPD with smoking history ≥10 pack years, significant airway obstruction, and medication use indicative of COPD. Thus, the authors concluded that the greater responsiveness for patients with reversibility could be attributed to greater bronchodilator reversibility of their COPD and not asthma. The current evidence suggests that strict inclusion criteria based on bronchodilator reversibility are not necessary for future trials. Instead, patients with asthma could be differentiated from those with COPD on the basis of history (eg, nonsmoking) and normalization of lung function values.

Figure 3.

Response to inhaled corticosteroid (fluticasone propionate) + long-acting β₂-agonist (salmeterol) in reversible and nonreversible patients in COPD.¹³,¹⁴

Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; FEV₁, forced expiratory volume in 1 second; ICS, inhaled corticosteroid.

Conclusions

Significant advances in the understanding of COPD and increases in available treatment options have been made in recent years. Clinical evidence supports a positive approach to COPD management, and concerted efforts have produced standard guidelines for diagnosis and assessment. These guidelines influence clinical trial designs and protocols, the results of which feed back into clinical guidelines in a continuous process. This review of data from recent trials in COPD has some unexpected results, for example, the results from the recent large UPLIFT trial of tiotropium reported lower trough FEV₁ outcomes than most previous studies with this drug. Current clinical trial results may not be directly comparable to earlier studies due to substantial changes in the availability of concomitant medications and in the baseline patient populations over time, including use of prior medications, smoking status and history, and disease severity. The impact of such factors on trial outcomes should be carefully evaluated, and consideration should be given as to whether long-acting bronchodilators can achieve a trough FEV₁ value >60–100 mL in future studies. This will require formal study with a full evaluation of indicative values for the MCID of trough and peak FEV₁ changes.