Supplementary Appendix

Transcription

1 Supplementary Appendix This appendix has been provided by the authors to give readers additional information about their work. Supplement to: O Byrne PM, FitzGerald JM, Bateman ED, et al. Inhaled combined budesonide formoterol as needed in mild asthma. N Engl J Med 2018;378: DOI: /NEJMoa

2 Inhaled Combined Budesonide Formoterol as Needed in Mild Asthma Authors and affiliations: Paul M. O Byrne 1, M.B., J. Mark FitzGerald 2, M.D., Eric D. Bateman 3, M.D., Peter J. Barnes 4, M.D., Nanshan Zhong 5, Ph.D., Christina Keen 6, M.D., Carin Jorup 6, M.D., Rosa Lamarca 7, Ph.D., Stefan Ivanov 6, Ph.D., and Helen K. Reddel 8, M.B., B.S., Ph.D. 1 Firestone Institute for Respiratory Health, St Joseph s Healthcare and Department of Medicine, Michael G. DeGroote School of Medicine, McMaster University, Hamilton, ON, Canada; 2 Institute for Heart and Lung Health, University of British Columbia, Vancouver, BC, Canada; 3 Department of Medicine, University of Cape Town, Cape Town, South Africa; 4 Airway Disease Section, National Heart and Lung Institute, Imperial College, London, United Kingdom; 5 State Key Laboratory of Respiratory Diseases, First Affiliated Hospital, Guangzhou Medical University, Guangzhou, China; 6 AstraZeneca R&D, Gothenburg, Sweden; 7 AstraZeneca, Barcelona, Spain; 8 Woolcock Institute of Medical Research, University of Sydney, Sydney, Australia 1

3 Supplementary Appendix Table of contents 1. Eligibility Criteria Inclusion Criteria Exclusion Criteria Stratification Technique ediary-derived Well-Controlled Asthma Weeks Sample Size Estimate Statistical Models Primary Efficacy Variable Secondary Variables Figures and Tables Figure S1. Patient Disposition (CONSORT Flow Diagram) Figure S2. ediary-derived Well-Controlled Asthma Weeks by Overall Study Population and Pre-Study Treatment Subpopulations Figure S3. Well-Controlled Asthma Weeks by Treatment Group, Overall and for Individual Components Figure S4. Annualized Exacerbation Rate Figure S5. Asthma-Related Discontinuations Figure S6. Time to Additional Glucocorticoids for Asthma Figure S7. Asthma Control Questionnaire 5-Item Version Score Responders Based on Minimal Important Difference Figure S8. Least-Squares Mean Change from Baseline (95% CI) in Overall Asthma Quality of Life Questionnaire Score (Standardized Version) Table S1. Oversight Authorities Table S2. Additional Baseline Demographics and Clinical Characteristics Table S3. Well-Controlled Asthma Week Supportive Analysis: Individual Components Table S4. Pre-Specified Analysis of ediary-derived Well-Controlled Asthma Weeks Removing the As-Needed Medication Component and Other Individual Components from the Model Table S5. Post Hoc Analysis of Modified ediary-derived Well-Controlled Asthma Weeks, Discounting the First Two As-Needed Inhalations Per Day Table S6. Inhaled Glucocorticoid Dose: Mean Daily Metered Dose During the Randomized Treatment Period

4 Table S7. ACQ-5 Score Change from Baseline (Full Analysis Set) Table S8. Change in Pre-Bronchodilator FEV 1 (ml) from Baseline (Full Analysis Set) Table S9. Summary of Adverse Events (AE) by Treatment Group (Safety Analysis Set) Table S10. Summary of Patient Deaths Table S11. Mean Change from Baseline in Morning and Evening Peak Expiratory Flow Table S12. Mean Change from Baseline in As-Needed Inhalations by Time of Day Table S13. Mean Change from Baseline in Daily Total Asthma Symptom Score Table S14. Mean Change from Baseline in Percentage of Nights with Awakenings Due to Asthma 39 Table S15. Mean Change from Baseline in Percentage of Symptom-Free Days Table S16. Mean Change from Baseline in Percentage of As-Needed Medication-Free Days Table S17. Mean Change from Baseline in Percentage of Asthma Control Days Table S18. Percentage of Inhaled Glucocorticoid Controller Days Table S19. Number of Poorly Controlled Asthma Weeks Per Patient During the Randomized Treatment Period Table S20. Number of Patients with High As-Needed Medication Usage, by Treatment Group Table S21. List of Study Sites

5 1. Eligibility Criteria 1.1 Inclusion Criteria For inclusion in the study patients should fulfill the following criteria: 1. Provision of informed consent prior to any study-specific procedures. For under-age patients, signed informed consent from both the patient and the patient s parent/legal guardian is required 2. Outpatients of either gender aged 12 years at Visit 1 3. Diagnosis of asthma according to Global Initiative for Asthma (GINA) criteria based on symptoms with a documented history of at least 6 months prior to Visit 1. Lung function and reversibility tests performed as part of Visit 2 and 3 can be used as a confirmation of asthma diagnosis according to GINA criteria if there is no measure of lung function available before Visit 1 4. Patients who are in need of GINA (2012) Step 2 treatment: - uncontrolled on inhaled short-acting bronchodilator(s) as needed (short-acting β 2 -agonist [SABA] and/or short-acting anticholinergic agent) as judged by the investigator for the last 30 days before Visit 2, or - controlled on mono-maintenance therapy - with low stable dose inhaled glucocorticoid ( 400 μg budesonide per day or corresponding inhaled dose of other agent) or leukotriene receptor antagonist (LTRA) - in addition to as needed use of inhaled short-acting bronchodilator(s) (SABA and/or shortacting anticholinergic agent), as judged by the investigator for the last 30 days prior to Visit 2 5. Based on lung function tests (forced expiratory volume in 1 second [FEV 1 ] and forced vital capacity assessed by spirometry) at Visit 2, patients pre-treated with: - an inhaled short-acting bronchodilator only should have pre-bronchodilator FEV 1 60% predicted and post-bronchodilator FEV 1 80% predicted according to the European Respiratory Society (ERS) guidelines 1 - low-dose inhaled glucocorticoid or LTRA medication in addition to inhaled short-acting bronchodilator(s) should have pre-bronchodilator FEV 1 80% predicted according to the ERS guidelines 6. Reversible airway obstruction according to a reversibility test performed at Visit 2 defined as an increase in FEV 1 12% and 200 ml relative to baseline, after inhalation of 1 mg terbutaline Turbuhaler. The test can be repeated at Visit 3 in case the patients fail at Visit 2. If patients fail at both occasions, they can still be included if they have a documented historical reversibility within the last 12 months prior to Visit 3, with an increase in FEV 1 12% and 200 ml relative to baseline after administration of a rapid-acting β 2 -agonist 1 Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnic reference values for spirometry for the 3-95-yr age range: the global lung function 2012 equations. Eur Respir J 2012;40:

6 For randomization at Visit 3, patients should fulfill the following criteria: 7. Use of terbutaline Turbuhaler as needed due to asthma symptoms on at least 3 separate days during the last week of the run-in period 8. Ability to use Turbuhaler correctly and to complete the ediary correctly. Morning and evening data must be recorded for at least 8 days (any 8) of the last 10 days of the run-in period 5

7 1.2 Exclusion Criteria Patients should not enter the study if any of the following exclusion criteria are fulfilled: 1. Involvement in the planning and/or conduct of the study (applies to both AstraZeneca staff and/or staff at the study site) 2. Previous randomization in the present study 3. Participation in another clinical study with a non-biologic investigational product or new formulation of a marketed non-biologic drug during the last 30 days prior to Visit 1 4. Participation in another clinical trial with any marketed or investigational biologic drug within 4 months or 5 half-lives whichever is longer, prior to Visit 1 5. Any asthma worsening requiring change in asthma treatment other than inhaled short-acting bronchodilator(s) (SABA and/or short-acting anticholinergic agent) within 30 days prior to Visit 1 6. Use of oral, rectal, or parenteral glucocorticoid within 30 days and/or depot parenteral glucocorticoid within 12 weeks prior to Visit 1 7. Use of any β-blocking agent including eye-drops 8. Known or suspected hypersensitivity to study drugs or excipient 9. Smoker (current or previous) with a smoking history of 10 pack-years 10. Medical history of life-threatening asthma including intubation and intensive care unit admission 11. Any significant disease or disorder (e.g., cardiovascular, pulmonary other than asthma, gastrointestinal, hepatic, renal, neurological, musculoskeletal, endocrine, metabolic, malignant, psychiatric, major physical impairment) which, in the opinion of the investigator, may either put the patient at risk because of participation in the study, or may influence the results of the study, or the patient s ability to participate in the study 12. Any clinically relevant abnormal findings in physical examination and/or vital signs at Visit 2, which, in the opinion of the investigator, may put the patient at risk if participating in the study 13. Pregnancy, breast-feeding, or planned pregnancy during the study. Fertile women not using acceptable contraceptive measures, as judged by the investigator 14. Planned hospitalization during the study 15. Suspected poor capability, as judged by the investigator, of following instructions of the study 6

8 For randomization at Visit 3, patients should not fulfill any of the following criteria: 16. Use of 6 terbutaline Turbuhaler as needed inhalations per day, for a certain number of days depending on the actual length of run-in: for 2 days out of 14 days; for 3 days out of days; for 4 days out of 22 or more days of run-in 17. Any asthma worsening requiring change in asthma treatment other than inhaled short-acting bronchodilator(s) (SABA and/or short-acting anticholinergic agent) from Visit 1 until Visit 2 and/or requiring any asthma treatment other than run-in study medication from Visit 2 until randomization 1.3 Stratification Technique Country and pre-study treatment groups were used as stratification factors ensuring approximately equal number of patients per treatment group within each country and equal number of patients per treatment group for each pre-study treatment group, respectively. 7

9 2. ediary-derived Well-Controlled Asthma Weeks (ewcaw) For any week, the composite ewcaw endpoint was achieved when: (A) Two or more of the following criteria were fulfilled: and no more than 2 days with a daily asthma symptom score >1; no more than 2 days of as-needed medication use, up to a maximum of four occasions per week (multiple occasions per day were regarded as separate occasions); morning peak expiratory flow (PEF) 80% predicted every day; (B) Both of the following criteria were fulfilled: no night-time awakenings due to asthma; and no additional inhaled and/or systemic glucocorticoid treatment due to asthma (since use of additional glucocorticoid treatment implied that, otherwise, on the patient s randomized treatment, their asthma would not have been well controlled). For each patient and study week, the binary variable ewcaw was derived. In addition, the percentage of patients with ewcaw was also derived for each week. A week was evaluable for control status (either well-controlled asthma week or not-well-controlled asthma week ) if there were at least 5 days of complete ediary data or, for patients with <5 days of ediary data, if it was shown unequivocally that they were not asthma controlled, such as reporting any night-time awakening. In order for a week to be eligible to be considered for an ediary-derived wellcontrolled asthma week (ewcaw), the ediary had to be completed on at least 5 days in a week; this condition was not required for the evaluation of a not well-controlled asthma week. A missing asthma control week was defined when it could not be unequivocally determined whether the week was wellcontrolled or not. 8

10 3. Sample Size Estimate A total of 3750 patients (625 patients/treatment group/pre-study treatment group) were required to give >95% power to detect superiority for ewcaw of as-needed budesonide/formoterol (BUD/FORM) compared with as-needed terbutaline and 90% power to establish non-inferiority for ewcaw of asneeded BUD/FORM compared with budesonide maintenance plus as-needed terbutaline, with a predefined non-inferiority limit of 0.8, i.e. if the lower 95% confidence interval (CI) of the odds ratio (OR) for as-needed BUD/FORM versus budesonide maintenance plus as-needed terbutaline was 0.8. For the two pre-study treatment subgroups, assuming that equal numbers of patients would be recruited to each, this sample size also gave 80% power to detect a difference for ewcaw between asneeded BUD/FORM and as-needed terbutaline, and 80% power to establish non-inferiority for ewcaw of as-needed BUD/FORM compared with budesonide maintenance plus as-needed terbutaline, with a pre-defined non-inferiority limit of

11 4. Statistical Models 4.1 Primary Efficacy Variable The primary variable, ewcaw, was analyzed by a repeated measures logistic regression model with treatment, pre-study treatment, and region as fixed effects, and study week as a categorical time variable. It used an exchangeable correlation structure. Odds ratios averaged over the 52-week period and their corresponding 95% CIs were derived from the model. The primary treatment comparison was as-needed BUD/FORM versus as-needed terbutaline (superiority test, primary objective), and the secondary comparison was as-needed BUD/FORM versus budesonide maintenance (non-inferiority test, secondary objective). A sequential testing approach was implemented: the secondary comparison was to be evaluated only if the primary comparison test was statistically significant (P<0.05). To declare noninferiority for the secondary comparison, the lower 95% 2-sided CI of the OR (as-needed BUD/FORM versus budesonide maintenance) had to be 0.8 (non-inferiority margin). The repeated measures logistic regression model was also used when analyzing each component of the definition of ewcaw separately, and when assessing the effect of removing one component of the ewcaw. The treatment effect was investigated in the two subgroups as defined by pre-study treatment to assess the consistency of the treatment effect across subgroups. Sensitivity analyses A key component of ewcaw is the use of as-needed medication, which is confounded with the different treatment regimens. In particular, the use of budesonide maintenance treatment may result in a higher proportion of ewcaw, due to less as-needed medication use, even if overall asthma control is truly similar to as-needed BUD/FORM treatment (since all BUD/FORM use is counted as as-needed medication). In addition, guidelines have conventionally included both symptoms and reliever use in the assessment of asthma symptom control, because higher SABA use, independent of symptom frequency, is associated with a higher risk of exacerbations, indicating worse overall asthma control. 2 By contrast, the study hypothesis was that use of as-needed BUD/FORM would reduce exacerbation risk. The protocol therefore pre-specified a modified ewcaw analysis in which the as-needed component was omitted from ewcaw. A sensitivity analysis was also conducted investigating the effect of changing the threshold for the asneeded component to less stringent criteria (no more than 5 days of as-needed medication use, up to a maximum of 14 occasions per week). A post hoc analysis was also performed, in which the first two doses of as-needed medication per day were discounted for as-needed BUD/FORM and as-needed terbutaline treatment groups, as if these had been taken as maintenance doses. 2 Global Strategy for Asthma Management and Prevention (Accessed 19/01/2018, at 10

12 Since analysis of ewcaw relied on patient ediary data (for which some elements of the diary may have been unavailable), sensitivity analyses were also performed to assess the effect of missing diary observations on ewcaw based on different assumptions of the missing data mechanisms. 4.2 Secondary Variables No adjustment for multiple comparisons for the following secondary variables were performed. Asthma exacerbations Moderate or severe and severe asthma exacerbation rates were analyzed by a negative binomial regression model with randomized treatment, pre-study treatment, region, and number of severe exacerbations in the 12 months prior to screening (0 or 1) as factors. Time to first moderate or severe and time to first severe exacerbation were analyzed by a Cox proportional hazards model with randomized treatment, pre-study treatment, region, and number of severe exacerbations 12 months prior to screening (0 or 1) as factors. The probability of having an exacerbation by treatment group was estimated using the Kaplan-Meier product limit method. Additional glucocorticoid use Time to administration of additional glucocorticoids was analyzed by a Cox proportional hazards model with randomized treatment, pre-study treatment, and region as factors. Kaplan-Meier curves were also generated. Glucocorticoid load The mean daily inhaled glucocorticoid load (expressed as metered dose for consistency between treatment arms) and the number of days with systemic glucocorticoids due to asthma were presented descriptively by randomized treatment. Discontinuation due to study-specific asthma-related events Time to study discontinuation due to study-specific asthma-related events was analyzed by a Cox proportional hazards model with randomized treatment, pre-study treatment, and region as factors. Study-specific asthma-related discontinuation criteria included: a severe asthma exacerbation with duration for >3 weeks; two severe asthma exacerbations during 3 months; or three severe asthma exacerbations in total during the study. Lung function measurements Average change from baseline in pre-bronchodilator and post-bronchodilator FEV 1 was analyzed using a mixed-model for repeated measures including the following covariates: randomized treatment group, pre-study treatment, region, baseline FEV 1, visit, and visit by randomized treatment group interaction. Variance estimation was based on an unstructured covariance matrix. 11

13 Patient-reported outcomes Asthma Control Questionnaire 5-item version (ACQ-5) and Asthma Quality of Life Questionnaire (standardized version) were analyzed using the same mixed-model for repeated measures as used for FEV 1, replacing the corresponding baseline variable. ediary variables Change from baseline in PEF, averaged over the randomized treatment period, was analyzed using analysis of covariance (ANCOVA) models (with morning and evening PEF analyzed using separate models). The ANCOVA models included randomized treatment, pre-study treatment, and region as categorical factors, and the mean PEF value during run-in as a continuous covariate. The change from baseline in asthma symptom score (day, night, and total) and number of as needed medication occasions (day, night, and total), percentage of night-time awakening(s) due to asthma symptoms, percentage of symptom-free days, percentage of asthma-control days, percentage of as needed -free days, and percentage of controller use days were analyzed using ANCOVA models following the same method. Poorly controlled asthma weeks were analyzed using summary statistics by randomized treatment group. A poorly controlled asthma week was defined by the fulfillment of one of the following conditions: two or more consecutive days with awakenings due to asthma on both nights; a recorded use of as-needed medication for symptom relief of at least three occasions per day, for at least two consecutive days; additional systemic glucocorticoid treatment required for severe exacerbation. Summary statistics for ediary adherence were collected by time of day (morning/evening), and descriptive statistics were presented by each randomized treatment group, in total, and also with morning and evening adherence presented separately. Maintenance medication adherence and as needed medication use Summary statistics of maintenance medication adherence, high use of as-needed medication (>8 and >12 inhalations in a day), as needed medication use, and change from baseline in as needed medication were produced. Change from baseline in as needed medication use was also analyzed using ANCOVA models. Day, night, and total as needed use were analyzed using separate models, following the same modeling method. The ANCOVA models included randomized treatment, pre-study treatment, and region as categorical factors, and baseline as-needed medication use as a continuous covariate. Safety variables Safety variables were presented descriptively by actual given treatment and pre-study treatment; no formal statistical hypothesis testing was conducted. 12

14 5. Figures and Tables Figure S1. Patient Disposition (CONSORT Flow Diagram) BUD/FORM, budesonide/formoterol. 13

15 Figure S2. ediary-derived Well-Controlled Asthma Weeks (ewcaw) by Overall Study Population and Pre-Study Treatment Subpopulations Subgroup 1 = uncontrolled on as-needed inhaled short-acting bronchodilator(s). Subgroup 2 = well-controlled on maintenance low-dose inhaled glucocorticoid or LTRA plus as-needed short-acting bronchodilator(s). BUD/FORM, budesonide/formoterol; CI, confidence interval: LTRA, leukotriene receptor antagonist; OR, odds ratio. 14

16 Figure S3. Well-Controlled Asthma Weeks by Treatment Group, Overall and for Individual Components As-needed terbutaline 15

17 As-needed budesonide/formoterol Budesonide maintenance PEF, peak expiratory flow. 16

18 Figure S4. Annualized Exacerbation Rate BUD/FORM, budesonide/formoterol; CI, confidence interval; RR, rate ratio. 17

19 Figure S5. Asthma-Related Discontinuations As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Number (%) of patients with event 21 (1.6) 4 (0.3) 6 (0.5) Hazard ratio (2-sided 95% CI) 0.18 (0.06 to 0.52) (0.19 to 2.35) P-value* Cox regression model with randomized treatment, pre-study treatment, and region as covariates. * P-values not controlled for multiplicity. Study-specific asthma related discontinuation criteria: A severe asthma exacerbation with duration for >3 weeks; two severe asthma exacerbations during 3 months; or three severe asthma exacerbations in total during the study. BUD/FORM, budesonide/formoterol; CI, confidence interval. 18

20 Figure S6. Time to Additional Glucocorticoids for Asthma As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Number (%) of patients with event 345 (27.0) 164 (12.8) 187 (14.6) Hazard ratio (2-sided 95% CI) 0.41 (0.34 to 0.50) (0.70 to 1.07) P-value* < Cox regression model with randomized treatment, pre-study treatment, and region as covariates. * P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval. 19

21 Figure S7. Asthma Control Questionnaire 5-Item Version (ACQ-5) Score Responders Based on Minimal Important Difference Patient improved/worsened responses are based on a minimal important difference of 0.5. Improved: Change in ACQ-5 score 0.5 points. No clinically important change: Change in ACQ-5 score > 0.5 and <0.5 points. Worsened: Change in ACQ-5 score 0.5 points. ACQ-5 includes five items about symptoms based on a 7-day recall period: frequency of night-time awakening, severity of asthma symptoms upon awakening, extent of activity limitations, frequency of shortness of breath, and frequency of wheezing). Each question has a 7-point scale response from 0 (no impairment) to 6 (maximum impairment), the theoretical range for the total score goes from 0 (well controlled) to 6 (extremely poorly controlled). Minimally clinically important difference for ACQ-5 score is defined as The shortened ACQ-5 version can be used without loss in validity or change in interpretation as compared to the original ACQ 7-item version. BUD/FORM, budesonide/formoterol. 3 Juniper EF, Guyatt GH, Willan A, Griffith LE. Determining a minimal important change in a disease-specific Quality of Life Questionnaire. J Clin Epidemiol 1994;47:

22 Figure S8. Least-Squares Mean Change from Baseline (95% CI) in Overall Asthma Quality of Life Questionnaire (AQLQ) Score (Standardized Version) AQLQ includes 32 items all assessed on a 7-point Likert scale from 1 to 7. Domain scores as well as the overall scores are calculated from the unweighted arithmetic means of the individual item scores. AQLQ overall score ranges from 1 to 7, with higher scores indicating better quality of life. A difference of 0.5 in the overall score is defined as minimally clinically important difference. 4 BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation. 4 Juniper EF, et al. Measurement properties and interpretation of three shortened versions of the asthma control questionnaire. Respir Med 2005;99:

23 Table S1. Oversight Authorities Australia: Department of Health and Ageing Therapeutic Goods Administration Australia: Human Research Ethics Committee Brazil: National Committee of Ethics in Research Brazil: National Health Surveillance Agency Bulgaria: Bulgarian Executive Drug Agency Bulgaria: Ethics Committee for Multicentre Studies Canada: Health Canada Chile: Instituto de Salud Pública China: State Food and Drug Administration Hungary: National Institute of Pharmacy Hungary: Research Ethics Medical Committee Korea: Ministry of Food and Drug Safety Mexico: Comité Bioético para la Investigación Clínica Mexico: Comité de Ética en Investigación de la Facultad de Medicina y Hospital Universitario Mexico: Federal Commission for Protection Against Health Risks Peru: Instituto Nacional de Salud Philippines: Bureau of Food and Drugs Poland: Komisja Bioetyczna Przy Okręgowej Izbie Lekarskiej W Białymstoku Poland: Office for Registration of Medicinal Products, Medical Devices and Biocidal Products Romania: National Medicines and Medical Devices Agency Russia: Ethics Committee of the Ministry of Health of the Russian Federation Russia: Ministry of Health of the Russian Federation South Africa: Medicines Control Council Ukraine: Central Ethics Committee of the Ministry of Health of Ukraine 22

24 Table S1. Oversight Authorities (cont d) Ukraine: State Pharmacological Center Ministry of Health United Kingdom: Medicines and Healthcare Products Regulatory Agency Vietnam: Ministry of Health 23

25 Table S2. Additional Baseline* Demographics and Clinical Characteristics As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Total (N=3836) Age (years) Mean (SD) 40.0 (16.3) 39.8 (16.9) 39.0 (16.7) 39.6 (16.6) Range Age, years [n (%)] 12 to < (11.3) 161 (12.6) 173 (13.5) 478 (12.5) 18 to < (82.5) 1019 (79.8) 1021 (79.6) 3094 (80.7) (6.2) 97 (7.6) 88 (6.9) 264 (6.9) Female sex, n (%) 771 (60.4) 777 (60.8) 797 (62.2) 2345 (61.1) Race, n (%) White 726 (56.9) 721 (56.5) 728 (56.8) 2175 (56.7) Asian 336 (26.3) 340 (26.6) 347 (27.1) 1023 (26.7) Black or African American 33 (2.6) 24 (1.9) 24 (1.9) 81 (2.1) Other 182 (14.3) 192 (15.0) 183 (14.3) 557 (14.5) Region, n (%) EU 400 (31.3) 396 (31.0) 397 (31.0) 1193 (31.1) East Asia 321 (25.1) 320 (25.1) 325 (25.4) 966 (25.2) Latin America 257 (20.1) 259 (20.3) 257 (20.0) 773 (20.2) Rest of the World 299 (23.4) 302 (23.6) 303 (23.6) 904 (23.6) Smoking status, n (%) Current 40 (3.1) 34 (2.7) 30 (2.3) 104 (2.7) Former 136 (10.6) 139 (10.9) 120 (9.4) 395 (10.3) Time since asthma diagnosis, years (median [range]) 6.3 ( ) 6.5 ( ) 6.3 ( ) 6.4 ( ) ACQ-5 score at: Study entry, mean (SD) 1.52 (0.96) 1.57 (0.97) 1.53 (0.97) 1.54 (0.97) Baseline, mean (SD) 1.54 (0.95) 1.61 (0.97) 1.55 (0.96) 1.57 (0.96) ACQ-5 score 1.5 at: Study entry, n (%) 549 (47.3) 601 (51.2) 568 (48.3) 1718 (48.9) Baseline, n (%) 602 (47.9) 649 (51.6) 596 (47.4) 1847 (49.0) 24

26 Table S2. Additional Baseline* Demographics and Clinical Characteristics (cont d) As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Total (N=3836) AQLQ, mean (SD) 5.25 (0.99) 5.20 (1.01) 5.27 (1.01) 5.24 (1.00) Pre-bronchodilator FEV 1 % predicted, mean (SD) (14.08) (14.24) (13.91) (14.07) Pre-bronchodilator FEV 1, L Mean (SD) 2.57 (0.74) 2.56 (0.74) 2.57 (0.73) 2.57 (0.74) Post-bronchodilator FEV 1 % predicted, mean (SD) (13.53) (14.02) (13.43) (13.66) Post-bronchodilator FEV 1, L Mean (SD) 2.91 (0.79) 2.92 (0.81) 2.92 (0.79) 2.92 (0.80) Morning PEF 80% predicted every morning, n (%) 362 (28.4) 340 (26.6) 376 (29.3) 1078 (28.1) Morning PEF ǁ Mean (SD) (114.1) (110.2) (109.8) (111.4) Mean <80% predicted, n (%) 638 (50.0) 648 (50.8) 635 (49.5) 1921 (50.1) Reversibility (%), mean (SD) 14.4 (11.5) 14.9 (11.3) 14.6 (11.6) 14.6 (11.5) Pre-study treatment, n (%) Uncontrolled** on BD Controlled** on inhaled glucocorticoid or LRTA 565 (44.2) 712 (55.8) 565 (44.2) 712 (55.8) 576 (44.9) 706 (55.1) 1706 (44.5) 2130 (55.5) Severe exacerbation in previous 12 months, n (%) 256 (20.0) 257 (20.1) 241 (18.8) 754 (19.7) * Baseline is defined as the assessment at Visit 3 (i.e. the point at which randomization took place). Study entry is defined as the assessment at pre-run-in visit (i.e. Visit 1 or Visit 2). These calculations for study entry were performed post hoc. Where morning PEF was 80% of predicted normal on every day of the last 10 days of the run-in period (these data provided for comparison with data in Figure S2). ǁ Baseline morning PEF defined as mean of the measurements taken during the last 10 days of the run-in period. These data are for reversibility as measured at study entry (Visit 1 or 2). If not confirmed, reversibility could also be measured at the baseline visit (Visit 3). Alternatively, a documented positive historical reversibility test within the last 12 months prior to randomization was acceptable to meet the inclusion criterion for reversibility. ** Level of asthma control on pre-study treatment was physician-assessed. There were no statistically significant differences between groups in demographic or clinical characteristics. AQLQ, Asthma Quality of Life Questionnaire (standardized version); ACQ-5, Asthma Control Questionnaire (5-item version); BD, short-acting bronchodilator; BUD/FORM, budesonide/formoterol; EU, European Union FEV 1, forced expiratory volume in 1 second; LTRA, leukotriene receptor antagonist; PEF, peak expiratory flow; SD, standard deviation. 25

27 Table S3. Well-Controlled Asthma Week Supportive Analysis: Individual Components Criterion Group Mean percentage of weeks that the criterion was met per patient Odds ratio (95% CI) P-value No more than 2 days of as-needed medication use, up to a maximum of 4 occasions per week No more than 2 days with a daily asthma symptom score >1 As-needed terbutaline (0.91 to 1.16) As-needed BUD/FORM 65.0 Budesonide maintenance (0.59 to 0.76) <0.001 As-needed terbutaline (1.03 to 1.36) As-needed BUD/FORM 46.1 Budesonide maintenance (0.75 to 1.00) As-needed terbutaline (1.09 to 1.46) Morning PEF 80% predicted every day As-needed BUD/FORM 29.5 Budesonide maintenance (0.70 to 0.93) As-needed terbutaline (1.11 to 1.41) <0.001 No night-time awakenings As-needed BUD/FORM 68.4 Budesonide maintenance (0.64 to 0.83) <0.001 No additional inhaled and/or systemic steroids As-needed terbutaline (2.15 to 4.70) <0.001 As-needed BUD/FORM 98.0 Budesonide maintenance (0.64 to 1.62) Repeated measures logistic regression model with week, randomised treatment, pre-study treatment and region as fixed effects. The structure of the correlation matrix is exchangeable. An odds ratio greater than 1 favours as-needed budesonide/formoterol. BUD/FORM, budesonide/formoterol; CI; confidence interval; NA, not applicable; PEF, peak expiratory flow. 26

28 Table S4. Pre-Specified Analysis of ediary-derived Well-Controlled Asthma Weeks (ewcaw) Removing the As-Needed Medication Component and Other Individual Components from the Model Treatment arm Mean percentage of ewcaw per patient, after removal of the selected component Odds ratio 95% CI P-value No more than 2 days of as-needed medication use, up to a maximum of 4 occasions As-needed terbutaline As-needed BUD/FORM Budesonide maintenance to to 0.88 < NA Additional supporting analyses No more than 2 days with a daily asthma score >1 As-needed terbutaline As-needed BUD/FORM Budesonide maintenance to to NA Morning PEF 80% predicted every day As-needed terbutaline As-needed BUD/FORM Budesonide maintenance to to NA No night-time awakenings due to asthma As-needed terbutaline As-needed BUD/FORM Budesonide maintenance to to NA No additional inhaled and/or systemic glucocorticoid As-needed terbutaline As-needed BUD/FORM Budesonide maintenance to to 0.73 As stated in the study protocol, the primary interest in this pre-specified analysis was to understand the extent to which the primary endpoint of ewcaw was driven by as-needed medication use. A key component of the primary endpoint is the use of as-needed medication that is confounded with the different treatment regimens. In particular, the use of budesonide maintenance treatment may result in a higher proportion of ewcaw, due to less as-needed medication use, even if overall asthma control is truly similar to as-needed BUD/FORM (since all BUD/FORM use is counted as as-needed medication). A pre-specified analysis removed the as-needed component. Additional supportive analyses assessed the effect of removing other components of ewcaw. Repeated measures logistic regression model with week, randomized treatment, pre-study treatment and region as fixed effects. The structure of the correlation matrix is NA 27

29 exchangeable. An odds ratio greater than 1 favors as-needed BUD/FORM. BUD/FORM, budesonide/formoterol; CI; confidence interval; NA, not applicable; PEF, peak expiratory flow. 28

30 Table S5. Post Hoc Analysis of Modified ediary-derived Well-Controlled Asthma Weeks (ewcaw), Discounting the First Two As-Needed Inhalations Per Day Treatment group Mean percentage of ewcaw per patient, after discounting As-needed terbutaline 0.5 mg (N=1271) Comparison between as-needed BUD/FORM and control group As-needed BUD/FORM 200/6 µg (N=1269) Twice-daily budesonide 200 µg (N=1279) Odds ratio % CI 1.24 to to 1.19 This analysis discounted the first two as-needed inhalations per day for both as-needed BUD/FORM and as-needed terbutaline treatment groups. BUD/FORM, budesonide/formoterol; CI, confidence interval; NA, not applicable. 29

31 Table S6. Inhaled Glucocorticoid Dose: Mean Daily Metered Dose During the Randomized Treatment Period IP: Mean daily inhaled glucocorticoid metered dose (µg) Total inhaled glucocorticoid*: Mean daily inhaled glucocorticoid metered dose (µg) As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Mean (SD) NA 92.9 (102.2) (89.2) Median (95% CI) NA 56.9 (50.0 to 64.6) (332.1 to 346.7) Range NA Mean (SD) 22.6 (68.6) 99.1 (109.7) (97.8) Median (95% CI) 0 (0, 0) 60.2 (53.7 to 67.0) (338.0 to 349.6) Range *Including open-label glucocorticoid prescribed for moderate or severe exacerbations or for long-term poor asthma control. BUD/FORM, budesonide/formoterol; CI, confidence interval; IP, investigational product; NA, not applicable; SD, standard deviation. 30

32 Table S7. ACQ-5 Score Change from Baseline (Full Analysis Set)* Treatment group As-needed terbutaline 0.5 mg (N=1225) As-needed BUD/FORM 200/6 µg (N=1241) Twice-daily budesonide 200 µg (N=1237) Baseline ACQ-5, mean (SD) 1.54 (0.95) 1.61 (0.97) 1.55 (0.96) Mean change from baseline (95% CI) (-0.21 to -0.14) (-0.36 to -0.29) (-0.51 to -0.44) Comparison between as-needed BUD/FORM and control group Estimate for difference % CI to to Mixed model for repeated measures analysis with baseline ACQ-5, randomized treatment, pre-study treatment, region, visit, and randomized treatment-by-visit interaction term. Missing data for ACQ-5 are dealt with by the direct likelihood approach using a mixed-model for repeated measures, no imputation is to be performed. 5 ACQ-5, Asthma Control Questionnaire (5-item version); BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation. 5 Beunckens C, Molenberghs G, Kenward MG. Direct likelihood analysis versus simple forms of imputation for missing data in randomized clinical trials. Clin Trials 2005;2:

33 Table S8. Change in Pre-Bronchodilator FEV 1 (ml) from Baseline (Full Analysis Set)* Treatment group Baseline FEV 1 pre-bronchodilator, mean L (SD) As-needed terbutaline 0.5 mg (N=1243) As-needed BUD/FORM 200/6 µg (N=1261) Twice-daily budesonide 200 µg (N=1261) (0.739) (0.738) (0.728) Mean change from baseline, ml (95% CI) 11.2 (-6.4 to 28.9) 65.0 (47.6 to 82.4) (101.9 to 136.7) Comparison between as-needed BUD/FORM and control group Estimate for difference, ml % CI 29.1 to to P-value < <0.001 * Mixed model for repeated measures analysis with baseline FEV 1, randomized treatment, pre-study treatment, region, visit, and randomized treatment-by-visit interaction term. P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; FEV 1, forced expiratory volume in 1 second; SD, standard deviation. 32

34 Table S9. Summary of Adverse Events (AE) by Treatment Group (Safety Analysis Set) All patients, n (%) As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Patients with at least one AE 545 (42.7) 485 (38.0) 512 (39.9) Most common AEs (occurring in 2% of patients) Upper respiratory tract infection 76 (6.0) 71 (5.6) 93 (7.3) Viral upper respiratory tract infection 79 (6.2) 75 (5.9) 84 (6.6) Asthma 109 (8.5) 37 (2.9) 57 (4.4) Pharyngitis 34 (2.7) 33 (2.6) 48 (3.7) Bronchitis 41 (3.2) 33 (2.6) 37 (2.9) Headache 25 (2.0) 23 (1.8) 29 (2.3) Allergic rhinitis 28 (2.2) 28 (2.2) 19 (1.5) Patients with at least one serious AE (including outcome = death) 50 (3.9) 38 (3.0) 37 (2.9) Patients with any AE leading to discontinuation 37 (2.9) 10 (0.8) 15 (1.2) Asthma-related discontinuations* 21 (1.6) 4 (0.3) 6 (0.5) Total number of deaths 0 (0.0) 0 (0.0) 2 (0.2) * Study-specific asthma-related discontinuation criteria: A severe asthma exacerbation with duration for >3 weeks; two severe asthma exacerbations during 3 months; or three severe asthma exacerbations in total during the study. 33

35 Upper gastrointestinal hemorrhage (n=1) and brain neoplasm (n=1). BUD/FORM, budesonide/formoterol. 34

36 Table S10. Summary of Patient Deaths Treatment group Budesonide maintenance Budesonide maintenance Sex Age (years) * Episode term as reported by the investigator F 50 Upper gastrointestinal bleed M 36 Polycystic tumor of brain (unknown) Adverse event (preferred term) Upper gastrointestinal hemorrhage Time from start of treatment to onset of adverse event (days) Time from last dose to death (days) Time from start of treatment to death (days) Reasonable possibility adverse event caused by study medication No Brain neoplasm No * Age at study entry. Neither of these deaths was adjudicated by the blinded independent adjudication panel as being asthma related. Upper gastrointestinal hemorrhage: The study monitoring team was informed on May 11, 2016 by the patient s partner that the subject expired on May 3, He claimed that the patient did not complain of any symptoms prior to hospitalization. She suddenly complained of hematemesis prior to confinement. The patient did not take any other medication apart from the study drug. Brain neoplasm: On October 8, 2015 the patient was admitted to Neurosurgery Unit with a 3-week history of balance and walking disorder, weakening of left lower limb, and falls. Patient reported a tumor of deep structures of the brain in an imaging examination approx. 20 years earlier, when they were disqualified from neurosurgical treatment, plus cerebral palsy with leftsided hemiparesis. Patient was being prepared for surgery; but this was postponed due to inflammation in lungs (suspicion of pneumonia from October 15, 2015). The patient s condition significantly deteriorated before emergency surgery on October 18 and 19, Tumorous cysts were endoscopically recanalized and evacuated, and a solid tumor was partially removed. The patient s general condition deteriorated, leading to mechanical ventilation from October 28, 2015 due to cardiorespiratory failure and was in very severe condition, requiring transfusion of blood derivatives. On November 1, 2015 features of left-sided pneumonia with exudate in the left pleural cavity were seen on X-ray. Laboratory tests showed pancytopenia, increasing renal and inflammatory parameters, E. coli in blood and ESBL Klebsiella discharge from the respiratory tract. On November 2, 2015 cardiac arrest occurred, after resuscitation heart rate was restored, blood derivatives were transfused. On the same day, in the evening, resuscitation was performed due to cardiac arrest. The patient was declared dead at Primary cause of death was stated as neoplasm of CNS. Hospitalization occurred in another clinic and no additional information or documentation is available. Relative onset days are calculated in relation to the first day of randomized study medication. Day 1 is the first day of dosing and day -1 is the day before the first dose. A time value of 1 indicates that the event occurred on the same day as the start of treatment. If result is 1 then this indicates death occurred on day of last dose. If result is 1 then this indicates death occurred on treatment start date. ǁ As assessed by the investigator. F, Female; M, Male. 35

37 Table S11. Mean Change from Baseline in Morning and Evening Peak Expiratory Flow (PEF) As-needed terbutaline 0.5 mg As-needed BUD/FORM 200/6 µg Twice-daily budesonide 200 µg Morning PEF (L/min) n (baseline) Baseline, mean (SD) (114.1) (110.2) (109.8) n (treatment period) Mean change from baseline (95%CI) ( to ) (-6.86 to -1.09) 6.01 (3.13 to 8.88) Comparisons between as-needed BUD/FORM and comparator groups* Estimate for difference (95% CI) (7.89 to 16.00) P< ( to -5.93) P<0.001 Evening PEF (L/min) n (baseline) Baseline, mean (SD) (116.9) (113.0) (112.8) n (treatment period) Mean change from baseline (95%CI) ( to ) ( to -8.39) (-7.77 to -2.17) Comparisons between as-needed BUD/FORM and comparator groups* Estimate for difference (95% CI) (6.99 to 14.90) P<0.001 * P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation ( to -2.29) P=

38 Table S12. Mean Change from Baseline in As-Needed Inhalations by Time of Day As-needed terbutaline 0.5 mg As-needed BUD/FORM 200/6 µg Twice-daily budesonide 200 µg Total n (baseline) Baseline, mean (SD) 1.45 (0.84) 1.42 (0.83) 1.46 (0.92) n (treatment period) Mean change from baseline (95% CI) (-0.83 to -0.77) (-0.99 to -0.93) (-1.07 to -1.00) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (-0.20 to -0.11) 0.08 (0.03 to 0.12) Day n (baseline) Baseline, mean (SD) 0.73 (0.52) 0.72 (0.52) 0.74 (0.55) n (treatment period) Mean change from baseline (95% CI) (-0.40 to -0.37) (-0.48 to -0.44) (-0.52 to -0.48) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (-0.10 to -0.05) 0.04 (0.01 to 0.06) Night n (baseline) Baseline, mean (SD) 0.72 (0.55) 0.70 (0.56) 0.73 (0.59) n (treatment period) Mean change from baseline (95% CI) (-0.43 to -0.40) (-0.52 to -0.48) (-0.55 to -0.52) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (-0.10 to -0.06) 0.04 (0.01 to 0.06) BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation. 37

39 Table S13. Mean Change from Baseline in Daily Total Asthma Symptom Score* As-needed terbutaline 0.5 mg As-needed BUD/FORM 200/6 µg Twice-daily budesonide 200 µg Total n (baseline) Baseline, mean (SD) 1.58 (1.12) 1.56 (1.06) 1.52 (1.06) n (treatment period) Mean change from baseline (95% CI) (-0.15 to -0.07) (-0.27 to -0.19) (-0.36 to -0.28) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (-0.18 to -0.06) P< (0.03 to 0.15) P=0.004 Day n (baseline) Baseline, mean (SD) 0.83 (0.57) 0.81 (0.54) 0.80 (0.55) n (treatment period) Mean change from baseline (95% CI) (-0.11 to -0.07) (-0.16 to -0.12) (-0.20 to -0.16) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (-0.08 to -0.02) P= (0.01 to 0.07) P=0.005 Night n (baseline) Baseline, mean (SD) 0.75 (0.60) 0.74 (0.56) 0.72 (0.57) n (treatment period) Mean change from baseline (95% CI) (-0.05 to 0.00) (-0.12 to -0.07) (-0.16 to -0.12) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (-0.10 to -0.04) P< (0.01 to 0.08) P=0.005 * Asthma symptom score is assessed on a 4-point scale from 0 to 3, with higher values indicating more severe asthma symptoms. Total asthma symptom score is the sum of the day and night scores. P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation. 38

40 Table S14. Mean Change from Baseline in Percentage of Nights with Awakenings Due to Asthma As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) n (baseline) Baseline, mean (SD) 23.5 (30.6) 22.7 (30.4) 22.5 (30.5) n (treatment period) Mean change from baseline (95% CI) (-5.63 to -3.47) (-8.86 to -6.69) ( to -9.06) Comparisons between as-needed BUD/FORM and comparator groups* Estimate for difference (95% CI)* (-4.75 to -1.70) P<0.001 * P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation (0.84 to 3.88) P=

41 Table S15. Mean Change from Baseline in Percentage of Symptom-Free Days As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) n (baseline) Baseline, mean (SD) 24.5 (32.3) 24.0 (31.8) 24.9 (32.2) n (treatment period) Mean change from baseline (95% CI) 1.49 (0.20 to 2.78) 4.19 (2.90 to 5.48) 7.10 (5.82 to 8.39) Comparisons between as-needed BUD/FORM and comparator groups* Estimate for difference (95% CI) 2.70 (0.89 to 4.51) P=0.003 * P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation (-4.72 to -1.10) P=

42 Table S16. Mean Change from Baseline in Percentage of As-Needed Medication-Free Days As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) n (baseline) Baseline, mean (SD) 24.8 (23.4) 26.0 (23.7) 25.5 (23.9) n (treatment period) Mean change from baseline (95% CI) (42.51 to 45.38) (42.53 to 45.40) (49.68 to 52.54) Comparisons between as-needed BUD/FORM and comparator groups* Estimate for difference (95% CI) 0.02 (-1.99 to 2.04) P=0.982 * P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; SD, standard deviation (-9.16 to -5.14) P<

43 Table S17. Mean Change from Baseline in Percentage of Asthma Control Days* As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) n (baseline) Baseline, mean (SD) 8.5 (15.8) 9.3 (16.9) 9.1 (16.8) n (treatment period) Mean change from baseline (95% CI) (11.52 to 14.26) (12.09 to 14.83) (17.35 to 20.08) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) 0.57 (-1.36 to 2.50) P= (-7.18 to -3.33) P<0.001 * An asthma control day was defined by the fulfillment of all of the following criteria: A day and night with no asthma symptoms (i.e. asthma symptom score=0); a night with no awakenings due to asthma symptoms; a day and night with no use of as-needed medication. P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval; SD standard deviation. 42

44 Table S18. Percentage of Inhaled Glucocorticoid Controller Days* As-needed terbutaline 0.5 mg (N=1277) As-needed BUD/FORM 200/6 µg (N=1277) Twice-daily budesonide 200 µg (N=1282) Total N=1273 N=1276 N=1281 Mean change from baseline (95% CI) 5.78 (4.56 to 7.01) (29.68 to 32.13) (84.54 to 86.98) Comparisons between as-needed BUD/FORM and comparator groups Estimate for difference (95% CI) (23.40 to 26.85) P< ( to ) P<0.001 * Inhaled controller use days during the randomized treatment period were calculated as the cumulative days when any controller medication (containing an inhaled glucocorticoid) was taken, including maintenance (budesonide maintenance group) and as-needed medication (as-needed BUD/FORM group) and additional prescribed inhaled glucocorticoids for asthma exacerbations and/or long-term poor asthma control (all treatment groups). The percentage of controller use days was calculated as the number of controller use days divided by the number of days in the randomized treatment period. P-values not controlled for multiplicity. BUD/FORM, budesonide/formoterol; CI, confidence interval. 43