Efficacy of Once-Daily Indacaterol Relative to Alternative Bronchodilators in COPD: A Patient-Level Mixed Treatment Comparison

Transcription

1 VALUE IN HEALTH 15 (2012) Available online at journal homepage: Comparative Effectiveness Research/Health Technology Assessment Efficacy of Once-Daily Relative to Alternative Bronchodilators in COPD: A Patient-Level Mixed Treatment Comparison Shannon Cope, MSc 1, Gorana Capkun-Niggli, PhD 2, Rupert Gale, MA 3, Cheryl Lassen, MD 3, Roger Owen, PhD 3, Mario J.N.M. Ouwens, PhD 4, Gert Bergman, PhD 4, Jeroen P. Jansen, PhD 1, * 1 MAPI Consultancy, Boston, MA, USA; 2 Novartis Pharma AG, Basel, Switzerland; 3 Novartis Horsham Research Centre, Horsham, UK; 4 MAPI Consultancy, Houten, The Netherlands A B S T R A C T Objective: was evaluated versus placebo, formoterol, and salmeterol in randomized controlled trials. No direct comparisons, however, are available for indacaterol with formoterol or indacaterol with salmeterol. trial evidence was synthesized to provide coherent estimates of indacaterol and indacaterol relative to formoterol, salmeterol, and tiotropium. Methods: Four randomized controlled trials were combined with Bayesian mixed comparisons by using individual patientlevel data. End points of interest were trough forced expiratory volume in 1 second (FEV 1 ), St. George s Respiratory Questionnaire (SGRQ) total score and response ( 4 points), and Transition Dyspnea Index total score and response ( 1 point). Results: demonstrated a higher FEV 1 than did formoterol at 12 weeks and 6 months (0.10 L difference; 95% credible interval [CrI] ), as did indacaterol versus salmeterol (0.06 L difference at 12 weeks; CrI ; 0.06 L at 6 months; CrI ). Regarding SGRQ, indacaterol demonstrated a comparable proportion of responders versus formoterol, as did indacaterol versus salmeterol. In comparison to tiotropium, indacaterol demonstrated a greater proportion of responders (odds ratio 1.52 at 12 weeks; CrI ). For Transition Dyspnea Index, indacaterol and formoterol showed a similar response. was more efficacious than salmeterol (odds ratio 1.65 at 12 weeks; CrI ). Overall, indacaterol showed the greatest efficacy for SGRQ and indacaterol for FEV 1 and Transition Dyspnea Index. Conclusion: is expected to be comparable to formoterol, salmeterol, and tiotropium, providing higher FEV 1 than formoterol and salmeterol and greater improvement in the SGRQ total score than tiotropium. provided comparable improvement in dyspnea, while indacaterol demonstrated the greatest response overall. Keywords: chronic obstructive pulmonary disease, indacaterol, individual patient data, mixed comparison. Copyright 2012, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. Introduction Chronic obstructive pulmonary disease (COPD) is a progressive lung disease resulting in gradual loss of lung function and symptoms of chronic cough, sputum production, breathlessness, and impaired health status. Clinical practice guidelines recommend long-acting bronchodilators be used for patients with moderate or more severe disease if symptoms are not relieved despite the use of short-acting bronchodilators [1]. Such bronchodilators include the twice-daily (BID) long-acting 2 -agonists salmeterol and formoterol, once-daily () longacting 2 -agonist indacaterol, and the anticholinergic tiotropium [1]. is a novel inhaled long-acting 2 -agonist indicated for maintenance bronchodilator of airflow obstruction in adult patients with COPD. The recommended dose in the European Union is one 150- g capsule, using the Onbrez Breezhaler inhaler, although the dose may be increased on medical advice to a maximum dose of one 300- g capsule [2]. In an extensive phase III clinical trial program, indacaterol was compared with formoterol, open-label tiotropium, salmeterol, and placebo in four pivotal trials: INdacaterol: Value in COPD: Longer Term Validation of Efficacy and Safety () [3], INdacaterol [versus tiotropium] to Help Achieve New COPD Excellence () [4], INdacaterol efficacy evaluation using doses with COPD patients 2 (INLIGHT-2) [5], and INdacaterol efficacy evaluation using doses with COPD patients 1 (INLIGHT-1) [6]. No single randomized controlled trial (RCT) simultaneously compared all s of interest, and a direct headto-head comparison was not available for the comparisons of indacaterol with formoterol and indacaterol with salmeterol. The main objective of this study was to evaluate the relative efficacy of indacaterol compared with formoterol 12 g BID as well as indacaterol compared with salmeterol 50 * Address correspondence to: Jeroen P. Jansen, MAPI Consultancy, 180 Canal Street, Boston, MA 02114, USA. jjansen@mapigroup.com /$36.00 see front matter Copyright 2012, International Society for Pharmacoeconomics and Outcomes Research (ISPOR). Published by Elsevier Inc. doi: /j.jval

2 VALUE IN HEALTH 15 (2012) µg 18 µg 300 µg were consistent across studies: patients were at least 40 years old with a diagnosis of COPD, a forced expiratory volume in 1 second (FEV 1 ) of 30% or more and less than 80% of predicted, FEV 1 /forced vital capacity of less than 70%, and a smoking history of at least 20 pack-years. Patients were excluded if they had been hospitalized within 6 weeks prior to the trial or during the run-in period. During the trial period, patients using fixed-dose combinations of 2 -agonists and inhaled corticosteroids were switched to equivalent inhaled corticosteroid monotherapy (at a dose and regimen to remain consistent throughout the study). Rescue medication with salbutamol was also permitted as needed. 50 µg BID g BID by synthesizing the results of the four RCTs. In doing so, this study also generated efficacy estimates of indacaterol 150 and relative to tiotropium 18 g, salmeterol 50 g BID, formoterol 12 g BID, and placebo. For this mixed comparison (MTC) [7,8], individual patient-level data (IPD) were used to allow for adjustment of possible confounding bias. Methods INLIGHT2 INLIGHT1 Placebo 12 µg BID Fig. 1 Overview of the evidence network available using the four key trials. BID, twice-daily;, INdacaterol [versus tiotropium] to Help Achieve New COPD Excellence; INLIGHT-1, INdacaterol efficacy evaluation using doses with COPD patients 1; INLIGHT-2, INdacaterol efficacy evaluation using doses with COPD patients 2;, INdacaterol: Value in COPD: Longer Term Validation of Efficacy and Safety;, once-daily. Evidence base The IPD available from the four pivotal indacaterol RCTs formed the evidence network as presented in Figure 1 ( [3], [4], INLIGHT-2 [5], and INLIGHT-1 [6]), which represented the only studies for which IPD were available. The study evaluated indacaterol 300 and 600 g compared with placebo and formoterol 12 g BID over 52 weeks. assessed indacaterol 150 and compared with placebo and open-label tiotropium 18 g over 26 weeks. INLIGHT-2 compared indacaterol with placebo as well as with salmeterol 50 g BID over 26 weeks, and INLIGHT-1 evaluated indacaterol compared with placebo over 12 weeks. Data for indacaterol 600 g from the study were included in the analysis but results are not presented because the dose is not licensed. The four RCTs included had similar study designs, with all the s administered double-blind except for the tiotropium arm in the study, which was open-label. Based on the Cochrane Grades of Recommendation, Assessment, Development and Evaluation system, the studies were considered of high quality, with the exception of the study, which was considered of moderate quality given the open-label administration of tiotropium [7]. The studies included between 416 patients (INLIGHT-1) and 1732 patients (). The inclusion criteria Outcomes of interest The outcomes of interest were evaluated at 12 weeks and 6 months: FEV 1 at 24 hours postdose ( trough ; mean of the values assessed at 23 hours 10 minutes and 23 hours 45 minutes following the previous morning dose), health status as assessed by the St. George s Respiratory Questionnaire (SGRQ) total score, the proportion of patients with a clinically relevant response in SGRQ ( 4-point decrease from baseline) [10], dyspnea as assessed by the Transition Dyspnea Index (TDI) total score, and the proportion of patients with a clinically relevant response in TDI (score of 1 point) [11]. FEV 1 was selected because it represented the primary outcome in all four trials. The remaining outcomes were chosen because they reflect patient-relevant secondary outcomes measured by validated instruments for which data were available across the four RCTs. Analysis For each of the end points of interest, IPD from the RCTs were used to perform a Bayesian MTC [8,9,12 16]. Analyses within the Bayesian framework involve data, a likelihood distribution, a model with parameters, and a prior distribution [17]. The model for a patient-level MTC relates the IPD from each of the studies to the basic parameters reflecting the relative effects of each intervention compared with placebo. Based on these basic parameters, the relative efficacy estimates between all the interventions of interest can be obtained according to functional relations between the basic parameters of each intervention [8,12,13]. Inconsistency in an MTC occurs when the distribution of effect modifiers is not evenly distributed across comparisons [12 15]. To minimize inconsistency and possible confounding bias, by covariate interactions were incorporated in the models [18]. Covariates potentially causing bias were selected on the basis of clinical expertise and evaluation whether these covariates were indeed effect modifiers of any of the s under study in the individual trials. In addition to and study effects, the following covariates and by covariate interactions were included in the model: baseline value of outcome, proportion of current smokers, reversibility to shortacting 2 -agonist, and reversibility to short-acting anticholinergic. For trough FEV 1, SGRQ total score, and TDI total score, linear models with normal likelihood distributions were used. A logistic model with a Bernoulli likelihood distribution was used to analyze the proportion of patients with a clinically relevant response in SGRQ and TDI. In the Appendix found at doi: /j.jval , details of the models are presented. To avoid influencing the results of the analysis based on prior beliefs, noninformative prior distributions were used for the model parameters to be estimated (normal distributions with mean 0 and a variance of 100). The deviance information criterion, which provides a measure of model fit penalizes model complexity, was used to select fixed-effects models over random-effects models [19,20]. Sensitivity analyses were performed regarding the noninformative prior distributions for the parameters to be estimated, and results were not affected. (Because a fixed-effects

3 526 VALUE IN HEALTH 15 (2012) model was deemed appropriate, no heterogeneity parameter needed to be estimated. Typically, only a heterogeneity parameter is sensitive to the definition of the corresponding noninformative prior.). All models were analyzed by using Markov Chain Monte Carlo techniques with WinBUGS 1.4.1, which was programmed by using R (Version 2.8.1). For each analysis, the Gelman-Rubin statistic (as modified by Brooks and Gelman [21]) was visually inspected on the basis of a graphical plot of the starting iteration and range illustrating the approximate point of convergence. The posterior distribution for the relative efficacy of indacaterol 150 and compared with the alternatives was summarized (difference in FEV 1 /SGRQ/TDI total scores or odds ratio for the proportion of patients with clinically relevant response in SGRQ and TDI) with the median as a measure of the point estimate and the 2.5th and 97.5th percentile to reflect the 95% credible interval. Ninety-five percent credible intervals represent the 95% probability the true underlying effect lies in the interval specified. The probability each was the best is presented in addition to the probability indacaterol 150 and were the alternatives. In the online supplemental material to this article found at doi: /j.jval , the model parameter estimates are provided. Results Study and patient characteristics Table 1 provides information on the patient characteristics for the four included RCTs. Across the studies, the patients were predominantly Caucasian (range from 76% and 94%), mostly male (range from 51% to 82%), and with an average age range between 63 and 64 years. The baseline lung function was comparable across the studies, with the average FEV 1 percentage predicted ranging from 53% to 56% and the average FEV 1 /forced vital capacity percentage ranging from 51% to 54%. The duration of COPD was also similar across the studies, with the average ranging from 6.4 to 7.4 years. Across the studies, there was some variation in the proportion of patients with severe or very severe COPD (as defined by GOLD guidelines), ranging from 38% to 46%. Reversibility to short-acting 2 -agonists as a percentage of prebronchodilator FEV 1 ranged from 11% to 17%. Around half of the patients were current smokers (range from 40% to 53%), with an average smoking history of between 40 and 61 pack-years. Finally, the proportion of patients using concomitant inhaled corticosteroids varied across the studies from 29% to 56%. Mixed comparison Table 2 presents the individual study summary statistics based on the IPD sets for the outcomes of interest, and Tables 3 to 5 present the corresponding results obtained with the MTC. Forced expiratory volume in 1 second All interventions compared were more efficacious than placebo regarding FEV 1 (Table 3). resulted in a higher FEV 1 at 12-weeks and 6-months follow-up compared with formoterol 12 g and salmeterol 50 g. Furthermore, indacaterol was at least as efficacious as tiotropium 18 g. Similar comparative estimates were observed for indacaterol, supporting higher FEV 1 at 12 weeks and 6 months compared with formoterol and salmeterol, as well as comparable estimates to tiotropium. is expected to be the most efficacious in terms of FEV 1. Table 1 Overview of the baseline patient characteristics in the four randomized controlled trials. On ICS Reversibility (Very) post-saba severe FEV 1 / FVC % FEV 1 % predicted FEV 1 mean (L)* Pack-years Current smokers Duration COPD (y) Treatment Males Mean age (y) [3] 80% 63.9 (8.6) 7.4 (7.1) 48.6 (41.5) 42% 1.48 (0.45) 53 (14) 51.1 (10.72) 12% (13) 46% 56% 12 g BID 80% 63.6 (8.5) 7.3 (6.2) 49.0 (60.8) 41% 1.50 (0.47) 53 (14) 51.3 (10.52) 12% (13) 44% 51% Placebo 82% 63.2 (8.3) 7.0 (6.1) 53.3 (69.9) 40% 1.52 (0.50) 53 (14) 52.1 (10.56) 13% (13) 45% 52% [4] 62% 63.4 (9.4) 7.0 (7.2) 48.3 (23.4) 45% 1.52 (0.50) 56 (15) 53.0 (9.97) 16% (15) 38% 38% 63% 63.3 (9.3) 6.7 (6.6) 50.8 (27.7) 45% 1.53 (0.52) 56 (15) 52.6 (10.07) 15% (15) 38% 37% 18 g 65% 64.0 (8.8) 6.8 (6.8) 50.0 (25.1) 45% 1.45 (0.50) 54 (16) 52.7 (10.14) 16% (18) 43% 35% Placebo 61% 63.6 (8.9) 6.6 (6.3) 49.7 (23.9) 46% 1.51 (0.49) 56 (14) 53.4 (10.11) 16% (18) 40% 40% INLIGHT-2 [5] 72% 63.2 (8.7) 6.5 (5.7) 39.6 (17.0) 46% 1.48 (0.49) 54 (14) 53.5 (10.0) 12% (15) 42% 45% 50 g BID 75% 63.4 (9.2) 6.4 (5.7) 40.0 (16.7) 46% 1.48 (0.49) 53 (14) 52.2 (9.88) 11% (14) 43% 46% Placebo 77% 63.9 (8.6) 6.6 (5.8) 41.0 (18.9) 45% 1.46 (0.47) 53 (14) 52.7 (11.03) 13% (16) 44% 40% INLIGHT-1 [6] 51% 62.9 (9.9) 6.6 (6.9) 53.5 (26.8) 51% 1.50 (0.53) 54 (13) 53.5 (9.84) 16% (17) 40% 29% Placebo 54% 63.2 (9.6) 7.3 (5.6) 60.5 (54.1) 53% 1.50 (0.51) 56 (14) 53.5 (10.36) 17% (19) 38% 34% Note. Values are mean (SD) unless indicated otherwise. BID, twice daily; COPD, chronic obstructive pulmonary disease; FEV 1, forced expiratory volume in 1 s; FVC, forced vital capacity; ICS, inhaled corticosteroids;, INdacaterol [versus tiotropium] to Help Achieve New COPD Excellence; INLIGHT-1, INdacaterol efficacy evaluation using doses with COPD patients 1; INLIGHT-2, INdacaterol efficacy evaluation using doses with COPD patients 2;, INdacaterol: Value in COPD: Longer Term Validation of Efficacy and Safety;, once daily; SABA, short-acting 2 -agonist. * Postbronchodilator FEV 1 measured 30 min after salbutamol 400 mg inhalation (, INLIGHT-1/2) or albuterol 360 mg inhalation (). Reversibility was calculated as the difference between the prebronchodilator and postbronchodilator values of FEV 1 (in liters) as a percentage of the prebronchodilator value.

4 Table 2 Individual study results for each study at baseline, 12 wk, and 6 mo by outcome. Outcome Placebo 12 g BID 18 g 50 g BID N Mean (SD) N Mean (SD) N Mean (SD) N Mean (SD) N Mean (SD) N Mean (SD) FEV 1 Bsl (0.46) (0.43) (0.43) 12 wk (0.49) (0.48) (0.48) 6 mo (0.48) (0.49) (0.47) Bsl (0.48) (0.50) (0.48) (0.48) 12 wk (0.47) (0.54) (0.52) (0.52) 6 mo (0.48) (0.54) (0.59) (0.51) INLIGHT-2 Bsl (0.49) (0.48) (0.48) 12 wk (0.49) (0.58) (0.53) 6 mo (0.46) (0.57) (0.53) INLIGHT-1 Bsl (0.59) (0.60) 12 wk (0.55) (0.63) 6mo SGRQ total score Bsl (17.8) (17.09) (17.33) 12 wk (18.48) (17.88) (18.38) 6 mo (18.15) (18.49) (18.36) Bsl (17.29) (19.12) (18.68) (18.11) 12 wk (18.31) (19.62) (18.91) (18.40) 6 mo (17.68) (18.98) (19.01) (19.12) INLIGHT-2 Bsl (17.75) (18.66) (18.47) 12 wk (19.55) (19.42) (18.49) 6 mo (19.21) (20.20) (18.16) INLIGHT-1 Bsl (18.86) (18.90) 12 wk (19.17) (19.74) 6mo TDI total score Bsl (2.21) (2.12) (2.06) 12 wk (3.15) (3.07) (2.96) 6 mo (3.07) (3.10) (3.11) (continued on next page) VALUE IN HEALTH 15 (2012)

5 Table 2 (continued) Outcome Placebo 12 g BID 18 g 50 g BID N Mean (SD) N Mean (SD) N Mean (SD) N Mean (SD) N Mean (SD) N Mean (SD) Bsl (2.37) (2.38) (2.32) (2.23) 12 wk (3.09) (3.27) (3.12) (3.05) 6 mo (3.19) (3.24) (3.17) (3.25) INLIGHT-2 Bsl (2.01) (2.10) (2.21) 12 wk (2.98) (3.28) (2.97) 6 mo (3.46) (3.60) (3.31) N % N % N % N % N % N % % SGRQ 4 points* 12 wk mo wk mo INLIGHT-2 12 wk mo INLIGHT-1 12 wk mo % TDI 1 point 12 wk mo wk mo INLIGHT-2 12 wk mo VALUE IN HEALTH 15 (2012) Notes. Minor differences in the outcomes compared with the study publications are present because of missing data in the covariate values for this analysis. N is the sample size of data included in analysis. Mean (SD) presented for continuous outcomes and proportion (%) of patients with event presented for dichotomous outcomes. BID, twice daily; Bsl, baseline; FEV 1, forced expiratory volume in 1 s;, INdacaterol [versus tiotropium] to Help Achieve New COPD Excellence; INLIGHT-1, INdacaterol efficacy evaluation using doses with COPD patients 1; INLIGHT-2, INdacaterol efficacy evaluation using doses with COPD patients 2;, INdacaterol: Value in COPD: Longer Term Validation of Efficacy and Safety;, once daily; SGRQ, St. George s Respiratory Questionnaire; TDI, Transition Dyspnea Index. * SGRQ clinically relevant improvement 4-point improvement. TDI clinically relevant improvement 1-point improvement.

6 VALUE IN HEALTH 15 (2012) Table 3 Results of patient-level mixed comparison for trough FEV 1. Treatment Comparator Difference in FEV 1 at 12 wk Difference in FEV 1 at6mo Mean is Mean is 18 g 50 g 12 g Placebo 0.13 ( ) 99% 0% 0.15 ( ) 99% 7% Placebo 0.11 ( ) 99% 0% 0.11 ( ) 99% 0% Placebo 0.06 ( ) 99% 0% 0.06 ( ) 99% 0% Placebo 0.16 ( ) 99% 38% 0.16 ( ) 99% 27% Placebo 0.17 ( ) 99% 62% 0.17 ( ) 99% 66% 18 g 0.03 ( ) 97% 0.01 ( 0.02 to 0.05) 77% 50 g 0.05 ( ) 99% 0.05 ( ) 99% 12 g 0.10 ( ) 99% 0.10 ( ) 99% 18 g 0.04 ( ) 98% 0.02 ( 0.01 to 0.06) 90% 50 g 0.06 ( ) 99% 0.06 ( ) 99% 12 g 0.10 ( ) 99% 0.11 ( ) 99% CrI, credible interval; FEV 1, forced expiratory volume in 1 s. St.George s respiratory questionnaire, indacaterol, and formoterol were more efficacious than placebo for both the SGRQ total score and SGRQ clinically relevant response (Table 4). was more efficacious than placebo in terms of the total score only at 12 weeks. was more efficacious than placebo in terms of the SGRQ total score at both 12 weeks and 6 months. For SGRQ clinically relevant response, a significant difference with salmeterol over placebo was observed only at 6 months. and indacaterol were more efficacious than tiotropium 18 g in terms of total score. Relative to formoterol 12 g, a greater improvement in the SGRQ total score was observed with indacaterol and indacaterol, but uncertainty in the estimates suggest the finding is ambiguous. and indacaterol showed similar improvements in the SGRQ total score as did salmeterol. At 12 weeks, a greater proportion of patients experienced a clinically relevant response with indacaterol compared with salmeterol. The effect with indacaterol was less pronounced. Overall, indacaterol was the most efficacious in terms of SGRQ. Transition dyspnea index All the interventions were more efficacious than placebo regarding the TDI total score and response (Table 5). In terms of TDI at 12 weeks, indacaterol was at least as efficacious as salmeterol, formoterol, and tiotropium. At 6 months, the differences favoring indacaterol were smaller than at 12 weeks. was more efficacious than tiotropium, salmeterol, and formoterol at 12 weeks for both total score and proportion of patients with a clinically relevant response. At 6 months, indacaterol was more efficacious than tiotropium and formoterol and comparable to salmeterol in terms of the TDI score, but showed a greater proportion of patients with a clinically relevant response relative to tiotropium and salmeterol. Overall, indacaterol was the most efficacious in terms of TDI. Discussion The main aim of this study was to assess the relative efficacy of indacaterol compared with formoterol and indacaterol compared with salmeterol in patients with moderate to severe COPD. Based on IPD from four RCTs, indacaterol 150 and are more efficacious than formoterol and salmeterol regarding FEV 1 and at least as efficacious as tiotropium. Patients receiving indacaterol experienced greater improvements in the SGRQ total score in comparison to tiotropium, which corresponded to a higher proportion of patients with a clinically relevant improvement for indacaterol 150 g. In comparison to formoterol and salmeterol, SGRQ results were comparable. The clinically relevant response in dyspnea with indacaterol was at least as good as with the other interventions, with indacaterol showing an even greater improvement than indacaterol compared with the alternatives. Although each of the included RCTs provides evidence for the relative efficacy of indacaterol versus an active, none of the studies included both doses of indacaterol compared with all the available interventions. However, all four RCTs form an evidence network given the presence of common s across trials (see Fig. 1). As a result, it is possible to synthesize the results of the four RCTs by means of a Bayesian MTC [12 15]. This framework provides the relative effectiveness for the competing interventions as well as the probability of being the better, which naturally supports decision making and is intuitive for decision-makers [15]. The internal validity of an MTC is contingent upon the extent of confounding bias due to similarity and consistency violations [12 15,18]. Overall, the RCTs were pivotal studies of the indacaterol clinical development program and were considered of high quality, with the exception of the study, in which the open-label

7 530 VALUE IN HEALTH 15 (2012) Table 4 Results of patient-level mixed comparison for SGRQ. Treatment Comparator Difference in SGRQ total score at 12 wk Difference in SGRQ total score at 6 mo Mean is Mean is 18 g 50 g 12 g Placebo 1.62 ( 3.17 to 0.07) 98% 0% 1.29 ( 3.03 to 0.48) 92% 0% Placebo 3.43 ( 5.33 to 1.51) 99% 21% 3.51 ( 5.66 to 1.36) 99% 24% Placebo 2.59 ( 4.29 to 0.89) 99% 2% 2.59 ( 4.53 to 0.65) 99% 4% Placebo 4.23 ( 5.36 to 3.07) 99% 72% 4.21 ( 5.67 to 2.76) 99% 62% Placebo 3.30 ( 4.47 to 2.11) 99% 5% 3.43 ( 4.79 to 2.05) 99% 10% 18 g 2.62 ( 4.23 to 1.06) 99% 2.93 ( 4.72 to 1.13) 99% 50 g 0.81 ( 2.74 to 1.11) 80% 0.70 ( 2.82 to 1.44) 74% 12 g 1.65 ( 3.57 to 0.28) 95% 1.62 ( 3.90 to 0.62) 92% 18 g 1.68 ( 3.29 to 0.07) 98% 2.14 ( 3.93 to 0.36) 99% 50 g 0.13 ( 2.01 to 2.31) 46% 0.09 ( 2.33 to 2.50) 47% 12 g 0.71 ( 2.40 to 0.95) 80% 0.83 ( 2.72 to 1.05) 81% Treatment Comparator SGRQ response at 12 wk SGRQ response at 6 mo OR OR 18 g 50 g 12 g Placebo 1.19 ( ) 91% 0% 1.17 ( ) 87% 0% Placebo 1.32 ( ) 96% 2% 1.61 ( ) 100% 21% Placebo 1.46 ( ) 99% 7% 1.39 ( ) 99% 4% Placebo 1.81 ( ) 99% 89% 1.82 ( ) 99% 69% Placebo 1.49 ( ) 99% 2% 1.54 ( ) 99% 6% 99% 99% 18 g 1.52 ( ) 99% 1.55 ( ) 99% 50 g 1.37 ( ) 98% 1.13 ( ) 77% 12 g 1.24 ( ) 90% 1.31 ( ) 93% 18 g 1.25 ( ) 94% 1.32 ( ) 97% 50 g 1.13 ( ) 75% 0.96 ( ) 40% 12 g 1.02 ( ) 55% 1.11 ( ) 75% CrI, credible interval; OR, odds ratio; SGRQ, St. George s Respiratory Questionnaire. evaluation of tiotropium reflects a potential limitation of the evidence base. Despite this limitation, Donohue et al. [4] reported the effect of tiotropium compared with placebo was similar to previous results where tiotropium was blinded for trough FEV 1 [22 24]. With an MTC, randomization holds only within a trial and not across trials. As a result, there is the risk patients who were assigned the different trials are not comparable. If the distribution of relative -effect modifiers is not similar across trials comparing different interventions in the network of studies, the similarity and consistency assumptions in an MTC are violated and results will be biased [12 15,18]. This bias can be limited by adjusting for these differences by incorporating

8 VALUE IN HEALTH 15 (2012) Table 5 Results of patient-level mixed comparison for TDI. Treatment Comparator Difference in TDI total score at 12 wk Difference in TDI total score at 6 mo Mean is Mean is 18 g 50 g 12 g Placebo 0.81 ( ) 99% 1% 0.79 ( ) 99% 1% Placebo 0.62 ( ) 99% 0% 0.90 ( ) 99% 10% Placebo 0.79 ( ) 99% 0% 0.73 ( ) 99% 0% Placebo 1.17 ( ) 99% 27% 0.95 ( ) 99% 4% Placebo 1.30 ( ) 99% 72% 1.28 ( ) 99% 85% 18 g 0.35 ( 0.06 to 0.78) 95% 0.17 ( 0.28 to 0.61) 77% 50 g 0.55 ( ) 99% 0.05 ( 0.46 to 0.58) 58% 12 g 0.38 ( 0.12 to 0.86) 93% 0.22 ( 0.33 to 0.76) 79% 18 g 0.48 ( ) 99% 0.50 ( ) 99% 50 g 0.68 ( ) 99% 0.38 ( 0.21 to 0.98) 90% 12 g 0.50 ( ) 99% 0.55 ( ) 99% Treatment Comparator TDI response at 12 wk TDI response at 6 mo OR is OR is 18 g 50 g 12 g Placebo 1.70 ( ) 99% 0% 1.41 ( ) 99% 0% Placebo 1.58 ( ) 99% 0% 1.54 ( ) 99% 1% Placebo 1.67 ( ) 99% 0% 1.78 ( ) 99% 2% Placebo 2.06 ( ) 99% 4% 1.68 ( ) 99% 0% Placebo 2.60 ( ) 99% 96% 2.36 ( ) 99% 97% 18 g 1.22 ( ) 92% 1.20 ( ) 89% 50 g 1.31 ( ) 96% 1.10 ( ) 72% 12 g 1.24 ( ) 90% 0.95 ( ) 38% 18 g 1.53 ( ) 99% 1.68 ( ) 99% 50 g 1.65 ( ) 99% 1.53 ( ) 99% 12 g 1.56 ( ) 55% 1.33 ( ) 97% CrI, credible interval; OR, odds ratio; TDI, Transition Dyspnea Index. by covariate interactions in the statistical models used [18]. Often MTCs are performed on the basis of aggregate-level data (AD), and the adjustment for confounding is attempted with reported average values for the patient covariates. A challenge with meta-regression models using AD is the association between a patient-level covariate and the relative effect of the studied interventions at the study level may not reflect the individual-level effect modification of covariate [16,37 42]. As such, it is possible even after adjustment for imbalances across trials, MTC results can still be biased. In the current analysis, an MTC with adjustment for differences in effect modifiers across comparisons based on patient-level data was performed. Given the advantage of IPD over AD to adjust for similarity and consistency violations in an MTC, the former might be considered the gold standard for the synthesis of a network of RCTs. (Of course, there is still the risk of residual confounding bias due to

9 532 VALUE IN HEALTH 15 (2012) imbalance in unmeasured effect modifiers across comparisons; accordingly, an RCT with all interventions of interest included is less prone to confounding bias.) Although we presented the results based on the MTC meta-regression models, it is important to note when we also ran models without covariate adjustment, the relative efficacy estimates were comparable to those obtained for the models with covariate adjustment. The current MTC included four key studies; it does not consider the complete body of published RCT evidence regarding formoterol, salmeterol, and formoterol, which can be considered a limitation. However, IPD were not available to the authors for the other studies. A literature search identified close to 40 RCTs evaluating formoterol, salmeterol, or tiotropium. Overall, these studies were relatively similar in terms of study characteristics as compared with the indacaterol studies. Most of the published studies were placebo-controlled trials and were double-blind and double dummy. The study in the current analysis, however, was open label [4]. The studies in the literature were predominantly multicenter trials in Europe and North America, whereas the indacaterol trials and INSIGHT-2 also included some study centers in South America, Africa, and Asia [3,5]. The patients included in the indacaterol studies seem representative of those included in other studies. All studies required patients to have a clinical diagnosis of COPD, to be 40 years of age or older, and without a history of exacerbations or hospitalizations in the past 4 to 6 weeks. These selection criteria were consistent with the indacaterol studies. However, patients in the other studies typically had an FEV 1 predicted of less than 60% to 70%, while the indacaterol studies included patients with an FEV 1 predicted of less than 80%. Furthermore, most studies required patients had at least 10 pack-years of smoking history, whereas the indacaterol studies required at least 20 pack-years of smoking history. In terms of background, the studies allowed patients to use a short-acting -agonist as needed for rescue medication. In general, the four indacaterol studies for which the authors had access to IPD seem relatively comparable to other COPD trials. As such, an MTC including all the evidence will probably not differ systematically from the results obtained with the current analysis. An analysis combining results of these four IPD studies with other AD RCT evidence is required to assess this hypothesis given the large number of published studies for COPD. Of course, adjustment for confounding bias due to consistency violations is compromised in such an elaborate analysis [15]. The patient population included in the different trials determines the external validity of the MTC. The RCTs included patients from Europe, North America, South America, and Australasia, implying the findings of the current analysis can be considered relevant for patients with moderate to severe COPD in these regions. This does not imply, however, effects are the same for all patients across or within these regions. Hence, a limitation of the current analysis is country effects were not taken into consideration. Such information would help specify the regions where current findings are most pertinent. The outcomes in this study are considered relevant to s for COPD. FEV 1 was the primary end point in all the studies and is also required from a regulatory perspective. Lung function and symptoms are the worst in the early morning and therefore affect patient functionality and daily activities [25,26]. Dyspnea has been reported as the most disabling symptom [27], while SGRQ represents a key patient-reported outcome provides direct insight into the overall health status of patients. The improvements in trough FEV 1 associated with indacaterol ( L relative to placebo) can be considered clinically relevant according to the threshold of 0.12 L prespecified in the RCTs [28]. In comparison to formoterol, the difference of 0.10 L and 0.11 L associated with indacaterol 150 and, respectively, can also be considered clinically important based on the range proposed by Cazzola et al. ( L) [28]. In the study, however, formoterol had less bronchodilation effect compared with earlier studies, which may have been related to the higher FEV 1 reversibility of patients (17%) in the previous trials [29,30]. 150 and also provided additional efficacy over salmeterol in terms of FEV 1 (0.05 L and 0.06 L, respectively), which is known to be an effective bronchodilator, and could be compared with the improvement associated with tiotropium versus BID 2 -agonists ( L) [24,31]. Therefore, indacaterol is expected to be at least as efficacious as tiotropium in terms of bronchodilation, which represents the gold standard bronchodilator in clinical practice in many countries [4]. Moreover, the benefits in lung function associated with indacaterol may allow greater activity, thereby improving the overall health status [4]. Although the FEV 1 improvements were maintained from 12 weeks to 6 months for indacaterol, the benefits in terms of TDI decreased slightly at 6 months. Some minor differences in efficacy were observed between the two doses of indacaterol, with a slight advantage for indacaterol in terms of FEV 1 and TDI. SGRQ results appeared to favor indacaterol over indacaterol, but it should be noted the difference was not clinically meaningful and when the two doses were included in the same underling RCT there was no significant difference. Two of the most recent MTCs published in COPD [32,33] were based on AD identified through systematic reviews performed in 2007 and focused on the proportion of patients experiencing an exacerbation. Authors acknowledge the limitations of this outcome, which is not adjusted for medication or trial duration, and identify the need for IPD to evaluate the rate of exacerbations per year. Analyses for the rate of exacerbation per patient-year were explored, but results should be interpreted with caution because the trial duration was limited and the population was not recruited on the basis of exacerbations. Results indicated all active s were comparable and demonstrated improvements versus placebo, although the rate of exacerbations in the placebo group was lower compared with in the current literature. Longer-term trials may allow for IPD MTCs on the rate of exacerbations per patient-year are ongoing. Although the current MTC focused on lung function, dyspnea, and overall health status, identification of the best or most appropriate cannot be made on the basis of efficacy end points alone. To inform health care decision making for clinical guidelines and reimbursement policies, the efficacy findings must be interpreted in light of the safety profile of the compared interventions and convenience. Compared with the BID dosing required for salmeterol and formoterol, the regimen for indacaterol may improve adherence in clinical practice [34], which has been reported to range from rates as low as 10% to 40% for COPD medication [34 36]. In conclusion, based on an IPD MTC of four RCTs, indacaterol is expected to be comparable to formoterol, salmeterol, and tiotropium. is expected to show greater improvement regarding FEV 1 than formoterol and salmeterol and a greater improvement in health status than tiotropium as measured by the SGRQ total score. The improvement in dyspnea with indacaterol was comparable to with other interventions, with indacaterol showing the greatest response of all the s. Source of financial Support: This study was funded by Novartis. Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at doi: /j.jval or, if a hard copy of article, at issues (select volume, issue, and article).

10 VALUE IN HEALTH 15 (2012) REFERENCES [1] Global Initiative for Chronic Obstructive Lung Disease (GOLD). Global strategy for the diagnosis, management and prevention of COPD Available from: GOLD_Report_2011_Feb21.pdf. [Accessed February 1, 2011]. [2] European Medicines Agency. Onbrez Breezhaler indacaterol. Doc. Ref.: EMA/625202/ Available from: docs/en_gb/document_library/epar_-_summary_for_the_public/ human/001114/wc pdf. [Accessed February 1, 2011]. [3] Dahl R, Fan C, Chung K, Buhl R. Efficacy of a new once-daily longacting inhaled 2 -agonist indacaterol versus twice-daily formoterol in COPD. Thorax 2010;65: [4] Donohue JF, Fogarty C, Lotvall J, et al. Once-daily bronchodilators for chronic obstructive pulmonary disease: indacaterol versus tiotropium. Am J Respir Crit Care Med 2010;182: [5] Kornmann O, Dahl R, Centanni S, et al. Once-daily indacaterol vs twice-daily salmeterol for COPD: a placebo-controlled comparison. Eur Respir J 2011;37: [6] Feldman G, Siler T, Prasad N, et al. Efficacy and safety of indacaterol once-daily in COPD: a double-blind, randomised, 12-week study. BMC Pulm Med 2010;10:11. [7] Schünemann HJ, Oxman AD, Vist GE, et al. Interpreting results and drawing conclusions. In: Higgins JPT, Green S, eds., Cochrane Handbook for Systematic Reviews of Interventions Version (updated March 2011). The Cochrane Collaboration, Available from: [Accessed December 1, 2011]. [8] Lu G, Ades AE. Combination of direct and indirect evidence in mixed comparisons. Stat Med 2004;23: [9] Jansen JP, Crawford B, Bergman G, Stam W. Bayesian meta-analysis of multiple comparisons: an introduction to mixed comparisons. Value Health 2008;11: [10] Jones PW. St. George s Respiratory Questionnaire: MCID. COPD 2005;2: [11] Witek TJ Jr., Mahler DA. Minimal important difference of the transition dyspnoea index in a multinational clinical trial. Eur Respir J 2003;21: [12] Jansen JP, Fleurence R, Devine B, et al. Interpreting indirect comparisons & network meta-analysis for health care decision-making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices Part 1. Value Health 2011;14: [13] Ades AE. A chain of evidence with mixed comparisons: models for multi-parameter synthesis and consistency of evidence. Stat Med 2003;22: [14] Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple s: combining direct and indirect evidence. BMJ 2005;331: [15] Hoaglin DC, Hawkins N, Jansen JP, et al. Conducting indirect comparison and network meta-analysis studies: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices Part 2. Value Health 2011;14: [16] Sutton AJ, Abrams KR, Jones DR, et al. Methods for Meta-Analysis in Medical Research. Chichester, UK: Wiley, [17] Spiegelhalter D, Abrams K, Myles J. Bayesian Approaches to Clinical Trials and Health Care Evaluation. Chichester, UK: Wiley, [18] Cooper NJ, Sutton AJ, Morris D, et al. Addressing between-study heterogeneity and inconsistency in mixed comparisons: application to stroke prevention s in individuals with nonrheumatic atrial fibrillation. Stat Med 2009;28: [19] Dempster AP. The direct use of likelihood for significance testing. Stat Comput 1997;7: [20] Spiegelhalter DJ, Best NG, Carlin BP, van der Linde A. Bayesian measures of model complexity and fit. J R Stat Soc Series B Stat Methodol 2002;64: [21] Brooks SP, Gelman A. General methods for monitoring convergence of iterative simulations. J Comput Graph Stat 1998;7: [22] Casaburi R, Briggs DD, Donohue JF, et al. The spirometric efficacy of once-daily dosing with tiotropium in stable COPD: a 13-week multicenter trial. Chest 2000;118: [23] Casaburi R, Mahler DA, Jones PW, et al. A long-term evaluation of once-daily inhaled tiotropium in chronic obstructive pulmonary disease. Eur Respir J 2002;19: [24] Donohue JF, van Noord JA, Bateman ED, et al. A 6-month, placebocontrolled study comparing lung function and health status changes in COPD patients treated with tiotropium or salmeterol. Chest 2002; 122: [25] Calverley PM, Lee A, Towse L, et al. Effect of tiotropium bromide on circadian variation in airflow limitation in chronic obstructive pulmonary disease. Thorax 2003;58: [26] Partridge MR, Karlsson N, Small IR. Patient insight into the impact of chronic obstructive pulmonary disease in the morning: an internet survey. Curr Med Res Opin 2009;25: [27] Celli BR, Cote CG, Marin JM, et al. The body-mass index, airflow obstruction, dyspnea, and exercise capacity index in chronic obstructive pulmonary disease. N Engl J Med 2004;350: [28] Cazzola M, MacNee W, Martinez FJ, et al. Outcomes for COPD pharmacological trials: from lung function to biomarkers. Eur Respir J 2008;31: [29] Dahl R, Greefhorst LA, Nowak D, et al. Inhaled formoterol dry powder versus ipratropium bromide in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001;164: [30] Rossi A, Kristufek P, Levine B, et al. Comparison of the efficacy, tolerability, and safety of formoterol dry powder and oral, slow-release theophylline in the of COPD. Chest 2002;121: [31] van Noord JA, Aumann JL, Janssens E, et al. Comparison of tiotropium once daily, formoterol twice daily and both combined once daily in patients with COPD. Eur Respir J 2005;26: [32] Baker WL, Baker EL, Coleman CI. Pharmacologic s for chronic obstructive pulmonary disease: a mixed- comparison meta-analysis. Pharmacotherapy 2009;29: [33] Puhan MA, Bachmann LM, Kleijnen J, et al. Inhaled drugs to reduce exacerbations in patients with chronic obstructive pulmonary disease: a network meta-analysis. BMC Med 2009;7:2. [34] Bourbeau J, Bartlett SJ. Patient adherence in COPD. Thorax 2008;63: [35] Campbell LM. Once-daily inhaled corticosteroids in mild to moderate asthma: improving acceptance of. Drugs 1999;58(Suppl. 4): [36] Tamura G, Ohta K. Adherence to by patients with asthma or COPD: comparison between inhaled drugs and transdermal patch. Respir Med 2007;101: [37] Greenland S. A review of multilevel theory for ecologic analyses. Stat Med 2002;21: [38] Riley RD, Lambert PC, Staessen JA, et al. Meta-analysis of continuous outcomes combining individual patient data and aggregate data. Stat Med 2008;27: [39] Jackson C, Best N, Richardson S. Improving ecological inference using individual-level data. Stat Med 2006;25: [40] Jackson C, Best N, Richardson S. Hierarchical related regression for combining aggregate and individual data in studies of socio-economic disease risk factors. J Royal Stat Soc (A) 2008;171: [41] Begg MD, Parides MK. Separation of individual-level and cluster-level covariate effects in regression analysis of correlated data. Stat Med 2003;22: [42] Neuhaus JM, Kalbfleisch JD. Between- and within-cluster covariate effects in the analysis of clustered data. Biometrics 1998;54: