Supplementary appendix

This appendix formed part of the original submission. We post it as supplied by the
authors.

Supplement to: Gibson PG, Yang IA, Upham JW et al. Effect of azithromycin on
asthma exacerbations and quality of life in adults with persistent uncontrolled
asthma (AMAZES): a randomised, double-blind, placebo-controlled trial. Lancet 2017;
published online July 4. http://dx.doi.org/10.1016/S0140-6736(17)31281-3.
Supplementary Appendix
Azithromycin reduces exacerbations in adults with persistent symptomatic asthma
(AMAZES): a randomised double-blind placebo-controlled trial.
Authors
Peter G. Gibson, Ian A. Yang, John W. Upham, Paul N. Reynolds, Sandra Hodge, Alan L. James, Christine
Jenkins, Matthew J. Peters, Guy Marks, Melissa Baraket, Heather Powell, Steven L. Taylor, Lex E.X Leong,
Geraint B. Rogers, Jodie L. Simpson

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)1

Table of Contents

Supplementary Methods...3

Limitations...3

Table S1. Reasons for screen failure ...4

Table S2. Other reasons for withdrawal.. 4

Table S3. Effect of treatment on cell counts at end of treatment.....4

Table S4. Cell counts: change from randomisation to end of treatment..5

Table S5. ECG QTC Analysis..5

Tables S6. Sputum AZM resistant pathogen at baseline...5

Tables S7. Sputum AZM resistant pathogen at end of treatment.. ..5

Table S8. Sputum Nose and Throat colonisation at baseline ...6

Table S9. Sputum Nose and Throat colonisation at end of treatment..6

Table S10. Change in pathogen resistance from baseline to end of treatment ...6

Table S11. AMAZES severe exacerbations adjusted for Faiths diversity score in induced sputum...6

Table S12. Sensitivity analysis of quality of life and secondary outcomes..7

Table S13. Sensitivity analysis of quality of life domains...8

References....9

Figure S1. Overlap of peripheral blood eosinophilia and sputum eosinophilia..9

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)2

Supplementary Methods

Microbiome Assessment
DNA extraction
DNA extraction was performed on 100 l induced sputum aliquots. Pelleted sputum containing bacteria were
homogenised via bead-beating with 200 g of silica: zirconium beads (1:1 of 0.1 m and 1.0 m; Biospec
Products, Inc., OK, USA) and a single chrome bead (3.2 mm, Biospec Products, Inc., OK, USA), at 6.5 m/s for
60 sec in a FastPrep-24 Instrument (MP Biomedicals, CA, USA). Homogenate was heated to 90 C for 5 min,
before being cooled on ice for 5 min. Enzymatic lysis was performed using lysozyme (Roche, ThermoFisher
Scientific, Victoria, Australia) and lysostaphin (Sigma-Aldrich, MO, USA) at a final concentration of 2 mg/mL
and 0.1 mg/mL, respectively. Organic material was removed by phenol:chloroform:isoamyl alcohol (25:24:1;
saline buffered at pH8.0; Sigma-Aldrich, MO, USA) phase separation following incubation with Proteinase K
(Fermentas, ThermoFisher Scientific, Victoria, Australia) and sodium dodecyl sulphate (Sigma-Aldrich, MO,
USA) at final concentrations of 1.2 mg/mL and 1.5 %, w/v respectively. DNA was recovered from the aqueous
layer using an EZ-10 Spin column in accordance with manufacturers instructions (Bio Basic, Inc., Ontario,
Canada), following precipitation by the addition of 10 M ammonium acetate and 99% ethanol (Sigma Aldrich,
MO, USA).
16S rRNA gene amplicon sequencing, data processing and diversity measurements
The V1-3 hypervariable region of the bacterial 16S rRNA gene was amplified using modified primers 27F and
519R as previously described.1 Amplicons were generated, cleaned, indexed and sequenced according to the
Illumina MiSeq 16S Metagenomic Sequencing Library Preparation protocol. 16S rRNA sequence data
generated by paired-end amplicon sequencing was processed using previously described bioinformatics
pipeline.1 Spurious operational taxonomic units (OTUs) were removed systematically using previous reports of
common laboratory sequencing contaminants.2 Faiths phylogenetic diversity (a measure of microbiota diversity
that incorporates phylogenetic distance between taxa) was calculated using QIIME (version 1.8.0).

Limitations

The study protocol (CRE website web address) states that the primary analysis was to evaluate the effect of
azithromycin on asthma exacerbations. We also discuss the importance of looking at the effect when patients are
classified by inflammatory subtype, ie eosinophilic vs noneosinophilic asthma(NEA). The rationale for this is
further expanded in the introduction. We consider both analyses as important and necessary in order to
determine the place of this treatment in asthma therapy. Subsequent to the publication of the AZISAST study,
we modified our prestated analysis (ANZCTR) to give priority to the NEA analysis, however still maintained
the need to evaluate the effects of azithromycin as stated in the protocol. The current analyses report the effect
of azithromycin on asthma overall, as stated in the protocol, as well as reporting the effects by inflammatory
subtype.

We used the last observation carried forward (LOCF) for analysis of our co-primary outcome, QoL and
secondary outcomes ,symptoms and spirometry. The use of LOCF may bias results in favour of the treatment
group if data are not missing at random. A sensitivity analysis of the co-primary outcome, QoL, using observed
data did not change the results.

The results of the exacerbation subgroup analyses should also be viewed with caution due to sample size and
repeated analysis.

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)3

Table S1: Reasons for screen failure

Inclusion criteria not met N=108

ACQ<0.75 28 (259%)
No Dr diagnosis of asthma 2 (19%)
Unstable asthma 5 (46%)
Not taking maintenance combination therapy (at least GINA step 3) 4 (37%)
No Variable airflow obstruction 28 (289%)
Inability to attend study visits 5 (46%)
FEV1 <40% predicted 7 (65%)
Ex smoker > 10 pack years & DLCO <70% predicted 7 (65%)
Treatment with macrolide, tetracycline or oral corticosteroid in preceding month 2 (19%)
Hypersensitivity to macrolides 2 (19%)
QTc>0.48s 16 (148%)
Taking medication that will interact with azithromycin in regard to QTc 15 (139%)
Planning pregnancy 1 (09%)
Respiratory disease other than asthma 13 (120%)
Impaired liver/renal function 7 (65%)
Ocular, abdominal, chest or brain surgery in previous 3 months 1 (09%)
Existing ECG abnormalities that may lead to arrhythmias 5 (46%)
Current lung cancer or other blood, lymphatic or solid organ malignancy 1 (09%)
Taking SMART therapy as maintenance 1 (09%)
Hearing insufficiency 2 (19%)
Investigator decision 2 (19%)
SMART: single corticosteroid/long-acting beta-agonist inhaler maintenance and reliever therapy; ACQ: asthma
control questionnaire score

Table S2: Other reasons for withdrawal

Other reason for withdrawal Placebo Azithromycin

N=12 N=14
Personal 6 10
Concern about side effects 2 1
Elevated QTc 0 2
Dr/GP ceased asthma medication 2 0
Investigator decision (multiple health problems) 1 0
Unknowm 1 1

Table S3. Effect of treatment on cell counts at end of treatment. Modified ITT analysis- includes
participants with follow up data only. Data are median (Q1, Q3); *ANCOVA adjusted for baseline
measurement

Placebo Azithromycin P value*

N=113 N=107
Sputum Cell Counts
Total cell count x106/ml 360 (189, 666) 350 (180, 621) 0500
Neutrophil% 330 (1550, 5850) 345 (150, 5725) 0555
Neutrophil x104/mL 10334 (3905, 25869) 11813 (3686, 31568) 0487
Eosinophil% 238 (075, 875) 20 (050, 1325) 0237
Eosinophil x104/mL 1105 (214, 4032) 689 (203, 4638) 0021
Blood Eosinophils N=163 N=168
Blood eosinophils x 109/L 020 (012, 035) 020 (010, 037) 0433

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)4

Table S4. Cell counts: change from randomisation to end of treatment by treatment group.

Placebo Azithromycin P value
N=113 N=107
Total cell count x106/ml -045 (-271, 162) -081 (-270, 125) 0655
Neutrophil % 075 (-150, 1725) -129 (-1875, 1450) 0454
Neutrophil x 104/mL -873 (-11477, 10602) -2176 (-12465, 9036) 0475
Macrophage % 075 (-1325, 160) 325 (-70, 180) 0375
Macrophage x104/mL 476 (-1161, 8262) -2025 (-8444, 8870) 0842
Lymphocyte% 0 (-05, 10) 0 (-10, 050) 0073
Lymphocyte x104/mL 011 (-280, 278) -029 (-308, 288) 0245
Columnar epithelial% 05 (-150, 325) 010 (-250, 30) 0282
Columnar epithelial x104/mL 066 (-896, 1244) 0 (-905, 1006) 0728
Squamous% 027 (-379, 588) 025 (-338, 449) 0582
Squamous x104/mL 312 (-1617, 2055) 151 (-1843, 1249) 0475
Eosinophil% -038 (-488, 163) 0 (-15, 10) 0161
Eosinophil x104/mL -231 (-4932, 513) -038 (-855, 4.84) 0106

Wilcoxon ranksum test

Table S5. AMAZES ECG QTC Analysis

Placebo Azithromycin P value

Baseline N=207 N=213
QTC result 410 (398, 424) 410 (396, 430) 0861
QTC >480# 2 (10%) 0 0242
Visit 3 N=200 N=197
QTC result 413 (398, 426) 417 (400, 431) 0261
QTC >480# 3 (15%) 2 (10%) 10
QTC change from baseline 0 (-11, 105) 2 (-8, 14) 0135
QTC change >30# 13 (65%) 24 (122%) 0051
QTC change >60# 4 (20%) 4 (20%) 10
Visit 10 N=162 N=163
QTC result 4125 (400, 431) 418 (403, 434) 0127
QTC >480# 4 (25%) 3 (18%) 0723
QTC change from baseline 0 (-9, 13) 2 (-6, 18) 0148
QTC change >30# 12 (74%) 23 (141%) 0051
QTC change >60# 4 (25%) 4 (25%) 10
Withdrawn due to QTC, n(%) 1 (05%) 1 (05%) 10

Median (q1, q3), Wilcoxon ranksum test; # n (%),Chi2 or Fishers exact test

Table S6. Sputum azithromycin resistant pathogens at baseline

Placebo Azithromycin
Enteric gram negative rod 0 1
Haemophilus influenzae 0 2
Pseudomonas aeruginosa 5 5
Staphylococcus aureus 2 1
Streptococcus pneumoniae 1 1

Table S7. Sputum azithromycin resistant pathogens at end of treatment (number also resistant at
baseline)

Placebo Azithromycin
Enteric gram negative rod 1 1
Haemophilus influenzae 1 4 (1)
Pseudomonas aeruginosa 4 (2) 4 (1)
Staphylococcus aureus 0 2
Streptococcus pneumoniae 1 (1) 1

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)5

Table S8. Sputum, nose and throat organisms at baseline.

Placebo Azithromycin
*Sputum azithromycin resistant organism 17/58 (293%) 20/56 (357%)

Throat azithromycin resistant organism 21/69 (304%) 30/67 (448%)

Nose azithromycin resistant organism 18/69 (261%) 14/65 (215%)
Numbers are n/N (%) of patients tested, Chi2 ; *Site 1 only; Sites 1 & 5 only

Table S9. Sputum, nose and throat organisms at end of treatment.

Placebo Azithromycin P value
*Sputum azithromycin resistant organism 12/42 (286%) 19/39 (48.7%) 0062

Throat azithromycin resistant organism 17/42 (405%) 18/39 (462%) 0606

Nose azithromycin resistant organism 17/42 (405%) 20/39 (513%) 0329
Numbers are n/N (%) of patients tested, Chi2 test.; *Site 1 only; Sites 1 & 5 only

Table S10. Change in pathogen resistance from baseline to end of treatment.

Placebo Azithromycin
N=11 N=10
Azithromycin resistant pathogens Baseline End of Baseline End of
treatment treatment
Enteric gram negative rod 0 0 1 (100%) 1 (100%)
Haemophilus influenzae 0 0 1 (100%) 3 (300%)
Pseudomonas aeruginosa 3 (273%) 3 (273%) 1 (100%) 2 (200%)
Staphylococcus aureus 0 0 0 1 (100%)
Streptococcus pneumoniae 1 (91%) 1 (91%) 0 0
Total AZM resistant pathogens 4 (364%) 4 (364%) 3 (300%) 6 (600%)
Numbers are N (%) of patients

Table S11. Effect of treatment on annualised asthma exacerbation rates and incidence rate ratio (IRR)
adjusted for Faiths score (whole population) and subgroups: Faiths score or <10, and Faiths score
tertiles+

Exacerbation rate/person-year IRR(95%CI) P value

Placebo Azithromycin
Faiths score (whole population) N=70 N=64
199 106 050 (028, 087) 0015
Faiths score <10 N=6 N=5
497 152 036 (003, 492) 0444
Faiths score 10 N=64 N=59
172 103 052 (035, 076) 0001
Faiths score tertile 1 (0-137) N=23 N=22
271 119 048 (019, 122) 0124
Faiths score tertile 2 (138 1665) N=24 N=21
146 109 075 (027, 206) 0579
Faiths score tertile 3 (1669-234) N=23 N=21
184 091 033 (014, 075) 0009

+
Models with Faiths score tertile and treatment group interaction terms were also conducted with no significant
interaction present

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)6

Table S12. Sensitivity analysis-Effect of treatment on asthma outcomes (observed data only)

Placebo Azithromycin
Co-primary endpoint
Quality of life
Number of patients analysed 163 164
AQLQ mean score end of treatment, (mean, 95%CI) 588 (509 to 6.41) 6.11 (522 to 6.66)
AQLQ mean score end of treatment difference vs placebo (adjusted mean, 95%CI, p value) 035 (012to 058); p=0008
*
Secondary endpoints
Symptoms
Number of patients analysed 166 168
ACQ6 score end of treatment , (mean, 95%CI) 117 (0.67 to 183) 0.83 (0.33 to 167)
ACQ6 score end of treatment difference vs placebo (adjusted mean, 95%CI) * -020 (-045 to 006)
Pre-bronchodilator Spirometry
Number of patients analysed 163 167
FEV1 (L) end of treatment , (mean, 95%CI) 205 (1.61 to 266) 204 (144 to 252)
FEV1 (L) end of treatment difference vs placebo(adjusted mean, 95%CI) * -005 (-013 to 0.03)
Visual Analogue Scales (VAS)
Number of patients analysed 166 166
Nasal symptoms end of treatment, (mean, 95%CI) 2.60 (0.90 to 5.20) 220 (0.90 to 4.80)
Nasal symptoms end of treatment difference vs placebo(adjusted mean, 95%CI) * -064 (-165 to -037)
Breathlessness end of treatment, (mean, 95%CI) 2.55 (0.9 to 4.8) 20 (0.5 to 4.2)
Breathlessness end of treatment difference vs placebo (adjusted mean, 95%CI) * -029 (-101 to 042)
Wheeze end of treatment, (mean, 95%CI) 1.15 (0.1 to 3.0) 0.9 (0 to 3.4)
Wheeze end of treatment difference vs placebo (adjusted mean, 95%CI) * 0.15(-102 to 1.33)
Sputum production end of treatment, (mean, 95%CI) 215 (0.3 to 4.4) 1.2 (0.2 to 3.8)
Sputum production end of treatment difference vs placebo (adjusted mean, 95%CI) * -005 (-0.54 to 0.44)
Cough end of treatment, (mean, 95%CI) 210 (0.3 to 4.5) 1.6 (0.5 to 4.0)
Cough end of treatment difference vs placebo (adjusted mean, 95%CI) * -041 (-116 to 0.35)
ANCOVA adjusted for baseline measurement, phenotype (Non-eosinophilic asthma/Eosinophilic asthma) and phenotype/treatment group interaction; No significant
phenotype/treatment group interaction was present; AQLQ: Asthma Quality of Life Questionnaire; ACQ6: Asthma Control Questionnaire.

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)7

Table S13: Sensitivity analysis- Effect of treatment on asthma related quality of life (AQLQ) domains(observed data only)

Placebo Azithromycin
N=164 N=166
Randomization End of treatment Change from Randomization End of treatment Change from
Randomization Randomization
Quality of life
AQLQ Activity domain 573 (509, 636) 6.09 (532, 664) 018 (-027, 064) 591 (509, 645) 623 (536, 677) 036 (-009, 064)
AQLQ Symptoms domain 525 (458, 575) 567 (475, 6.33) 033 (-017, 092) 525 (467, 5.92) 6.0 (5.0, 658) 058 (0, 125)
AQLQ Emotions domain 560 (460, 640) 6.0 (5.0, 680) 020 (-040, 080) 560 (480, 660) 620 (5.0, 70) 040 (0, 1.0)
AQLQ Environment domain 575 (475, 65) 625 (550, 675) 025 (-025, 10) 575 (5.0, 638) 638 (55, 675) 05 (0, 10)

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)8

References

1 Jervis-Bardy J, Leong L, Marri S, et al. Deriving accurate microbiota profiles from human
samples with low bacterial content through post-sequencing processing of Illumina MiSeq data.
Microbiome. 2015;3:19.
2 Salter S, Cox M, Turek E, et al. Reagent and laboratory contamination can critically impact
sequence-based microbiome analyses. BMC Biology 2014;12:1-12.

Figure S1. Overlap of peripheral blood eosinophilia and sputum eosinophilia . PB EA: peripheral blood
eoainophilic asthma; Sp EA: sputum eosinophilic asthma

Gibson_Lancet _AMAZES Efficacy Supplement(20june17)9