original research azithromycin for bronchial asthma in
TRANSCRIPT
ORIGINAL RESEARCH
Azithromycin for Bronchial Asthma in Adults: AnEffectiveness TrialDavid L. Hahn, MD, MS, Mike Grasmick, PhD, Scott Hetzel, MS, and Steven Yale, MD, andon behalf of the AZMATICS (AZithroMycin-Asthma Trial In Community Settings) Study Group
Background: Macrolides have antimicrobial and anti-inflammatory properties that may be useful in thetreatment of chronic asthma.
Methods: We performed a randomized, placebo-controlled, double-blinded effectiveness trial of 12weekly doses of adjunctive azithromycin, with follow-up to 1 year after randomization, in adults withpersistent asthma. Measurements included overall asthma symptoms, asthma quality of life (AQL), andasthma control. Eligible subjects who declined to participate in randomization were offered enrollmentinto a parallel open-label (OL) azithromycin treatment arm.
Results: Of 304 adult asthma patients screened, 97 (32%) were enrolled: 38 were randomized to azithromy-cin, 37 were randomized to placebo, and 22 opted in as OL subjects. OL subjects had higher rates of severe persis-tent asthma compared with randomized subjects (32% vs 8%, respectively; P � .012). At 1 year, compared with theplacebo arm, subjects randomized to azithromycin were more likely to have an AQL score >1 unit increase com-pared with baseline, but this difference was not statistically significant (36% vs 21% for placebo; P � .335). Com-pared with placebo, OL subjects had significant improvements in overall asthma symptoms from baseline (P �.0196), AQL (P � .0006), and asthma control (P � .0148).
Conclusions: Adults with asthma who were randomized to azithromycin did not show statistically significantimprovement in asthma outcomes, although the study was underpowered to detect clinical improvement in 15%(number needed to treat � 7). Adults with severe persistent asthma who elected OL treatment documented clinicalimprovements in asthma symptoms, AQL, and asthma control that persisted after completion of OL azithromycin(number needed to treat � 2). (J Am Board Fam Med 2012;25:442–459.)
Keywords: Antibiotics; Asthma; Clinical Trials, Randomized; Infectious Diseases; Practice-based Research Networks
There is increasing interest in the therapeutic po-tential of macrolides in chronic asthma. A 2005Cochrane review concluded that there was insuffi-
cient evidence to support or refute the use of mac-rolides in patients with chronic asthma and re-commended further studies.1 Potential macrolidemechanisms of action include nonantimicrobial ef-fects,2 antimicrobial activity targeting the respira-tory pathogens Chlamydia pneumoniae and Myco-plasma pneumoniae, or both.3
This article was externally peer reviewed.Submitted 3 November 2011; revised 8 February 2012;
accepted 13 February 2012.From the Dean Clinic, Madison, WI (DLH); the Depart-
ment of Family Medicine (DLH, MG) and the Department ofBiostatistics (SH), University of Wisconsin, School of Medi-cine and Public Health, Madison; and the Department ofInternal Medicine, Marshfield Clinic, Marshfield, WI (SY).
Funding: Pfizer, Inc., donated identical matching azithromycinand placebo. The Wisconsin Academy of Family Physicians; theAmerican Academy of Family Physicians Foundation, under theauspices of the Joint Grant Awards Program; the Dean Founda-tion for Health Research and Education; and private donors pro-vided financial support for direct costs of AZMATICS.
Prior presentation: Presented in part at the North AmericanPrimary Care Research Group Conference, Seattle, Washing-ton (November 13–17, 2010); and at the European RespiratorySociety Conference, Amsterdam, the Netherlands (September24–28, 2011). The following organizations provided in-kindsupport: the Wisconsin Research and Education Network; theCommunity-Academic Partnerships core of the University of
Wisconsin Institute for Clinical and Translational Research,funded through a National Institutes of Health Clinical andTranslational Science Award, grant no. 1 UL1 RR025011; theWisconsin Network for Health Research; the University ofWisconsin Department of Biostatistics; and the Center forUrban Population Health, Milwaukee, WI.
Conflict of interest: none declared.Disclaimer: The funding sources had no involvement in
study design, data collection, analysis, interpretation, reportwriting or publication. The writing committee had solecontrol and access to all data and accepts full responsibilityfor the content of this report.
Corresponding author: Dr. David L. Hahn, Dean EastClinic, 1821 S. Stoughton Road, Madison, WI 53716(E-mail: [email protected]).
442 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
Current guideline-recommended asthma treat-ments have limited generalizability because of alack of effectiveness trials of representative samplesof asthma patients and because of the systematicexclusion of large numbers of asthma patients fromthe efficacy trials on which the guidelines are based.On average, 19 of 20 people with physician-diag-nosed asthma were excluded from the clinical re-search4,5 and more than half of adults with asthmamay remain poorly controlled, even when they aretreated.6 Thus, it is important to test therapies forasthma in generalizable effectiveness trials.
We therefore conducted a primary care, prac-tice-based effectiveness trial designed to include 3months of adjunctive treatment (in addition tousual care) with the azalide macrolide azithromycin
followed by a 9-month observational period aftertreatment. The goal of our trial, AZithroMycin-Asthma Trial In Community Settings (AZMATICS),was to investigate whether azithromycin has value forpatients with persistent asthma in reducing the severityof their symptoms over time.
MethodsAdults with persistent asthma symptoms as definedby current guidelines7 were randomized to receive12 weekly doses of azithromycin or placebo as ad-junctive therapy between June 2006 and November2009. Self-reported questionnaire data were col-lected for 1 year after randomization. Patients wererecruited and enrolled from community practice-
Table 1. Inclusion, Exclusion, and Outcome Criteria
Criteria
Inclusions • Adults �18 years of age with physician-diagnosed asthma (symptomatic �2 days per week and/or �2 nightsper month or in exacerbation)
• Objective evidence for reversible airway obstruction (�12% and �200 mL change in FEV18 and/or a 25% and
60 L/min change in PEFR9) either spontaneously or after treatment• Asthma for at least 6 months before enrollment
Exclusions • Not English literate or has no email address or Internet access• Macrolide allergy• Pregnant or lactating• �4 weeks of continuous use of macrolides, tetracyclines, or quinolones within 6 months of randomization• Asthma symptoms less than 6 months’ duration• Unstable asthma requiring immediate emergency care• Comorbidities likely to interfere with study assessments or follow-up (eg, cystic fibrosis, obstructive sleep apnea
requiring CPAP, cardiomyopathy, congestive heart failure, terminal cancer, alcohol or other drug abuse, orany other serious medical condition that, in the opinion of the study physician, would seriously interfere withor preclude assessment of study outcomes or completion of study assessments)
• Medical conditions for which macrolide administration may possibly be hazardous (eg, acute or chronichepatitis, cirrhosis, or other liver disease; chronic kidney disease; or history of prolonged cardiac repolarizationand QT interval or torsades de pointes).
• Specified medications for which close monitoring has been recommended in the setting of macrolideadministration (digoxin, theophylline, warfarin, ergotamine or dihydroergotamine, triazolam, carbamazepine,cyclosporine, hexobarbital, or phenytoin)
Outcomes • Asthma symptom scores (0 � none, 1 � mild, 2 � moderate, 3 � severe, 4 � worst ever) within the past 24hours; every 1.5 months
• AQL (Juniper AQL questionnaire)10 within the past 2 weeks; every 3 months• Asthma control (mini-Juniper Asthma Control Questionnaire, without pulmonary function)11,12 within the past
week; every 3 months• Exacerbations (a steroid burst, an unscheduled or emergency visit and/or a hospitalizations for asthma) within
the past 1.5 months; every 1.5 months• Other respiratory illnesses within the past 1.5 months; every 1.5 months• Off-study antibiotic use within the past 1.5 months; every 1.5 months• Adverse events within the past 1.5 months; every 1.5 months• Use of asthma-controller medications (oral or inhaled steroids, LABAs, or antileukotriene agents) within the
past 3 months; every 3 months• Self-reported improvement in asthma (compared with baseline) within the past 3 months; every 3 months
AQL, asthma quality of life; CPAP, continuous positive airway pressure; FEV1, forced expiratory volume in 1 second; LABA,long-acting bronchodilators; PEFR, peak expiratory flow rate.
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 443
based settings throughout North America. Studyclinician members and/or staff of 5 practice-basedresearch networks (PBRNs) and one community-based allergist enrolled patients from their prac-tices for this study. The PBRNs included one na-tionwide network (American Academy of FamilyPhysicians National Research Network) and 4 re-gional networks (Ambulatory Network for Schol-arship and Research, Illinois; Cleveland Ambula-tory Research Network, Ohio; OklahomaPhysicians Resource/Research Network; and theWisconsin Research and Education Network). Fur-ther participant details are provided in the Ac-knowledgments. During this “real-world” study,we encountered eligible patients, most of whomhad severe treatment-resistant or refractoryasthma, who declined randomization in favor of
being treated with azithromycin. Rather than ex-clude these patients entirely, we elected to enrollthem as a parallel observational cohort. Patientswho opted to participate in this “open-label” (OL)arm received a prescription for azithromycin fromtheir personal physician and were followed for thesame outcomes as the randomized arm.
Study EligibilityEligibility criteria can be found in Table 1. Com-munity-based clinicians enrolled and randomizedeligible patients and remained available to assessside effects or severe adverse reactions, but theywere not involved in follow-up data collection. Allsubjects continued to receive usual asthma carefrom their doctor. All study sites received approval
Figure 1. CONSORT diagram. *Unavailable pulmonary function tests (PFTs) as the only disqualification of 67 of170. †Nonqualifying PFTs in 10 of 24.
Assessed for eligibility (n=304)
Excluded from randomized trial (n=228) -Not meeting inclusion criteria* (n=170) (*Unavailable pulmonary function tests as the only disqualification in 67 of 170) -Eligible, declined randomization (n=34) -Other† (n=24) (†Non-qualifying PFTs in 10 of 24)
Analyzed for sx, control and AQL: 3 months (n= 22) 6 months (n= 22) 9 months (n= 23) 12 months (n= 19)
Completed 3-month follow up (n= 22) Completed 6-month follow up (n= 22) Completed 9-month follow up (n= 23) Completed 12-month follow up (n= 19)
Allocated to azithromycin (n=38) Received allocated intervention (n=38)
Allocated to placebo (n=37) Received allocated intervention (n=37)
Allocation
Analysis
Follow-Up
Enrollment
Randomization
Enrolled in open-label cohort (n=22) Received open-label azithromycin (n=22)
Completed 3-month follow up (n= 26) Completed 6-month follow up (n= 27) Completed 9-month follow up (n= 22) Completed 12-month follow up (n= 25)
Completed 3-month follow up (n= 15) Completed 6-month follow up (n= 17) Completed 9-month follow up (n= 15) Completed 12-month follow up (n= 13)
Analyzed for sx, control and AQL: 3 months (n= 26) 6 months (n= 27) 9 months (n= 22) 12 months (n= 25)
Analyzed for sx, control and AQL: 3 months (n= 15) 6 months (n= 17) 9 months (n= 15) 12 months (n= 13)
Open-label Cohort
444 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
from their respective human subject committees,and subjects provided written, informed consent.
Randomization and MaskingAn independent statistician prepared the random-ization codes used for subject assignment to the
azithromycin or placebo study arms. The investi-gators, study subjects, and study site personnelwere blinded to treatment allocation. Study medi-cation was azithromycin 600 mg, 1 tablet daily for3 days followed by 1 tablet weekly for 11 weeks, oridentical matching placebo tablets. Each study site
Table 2. Patient Characteristics
CharacteristicRandomized
Placebo (n�37)Randomized
Azithromycin (n�38)Open-Label
Azithromycin (n�22) P*
Age(years), mean(SD) 47.4 (14.2) 45.7 (15.5) 45.4 (15.2) .621/.745At asthma diagnosis 24 (�1–59) 24 (�1–58) 28 (11–59) .603/.104
Diagnosis at age �18 years 21 (57) 24 (63) 19 (86) .641/.023Male sex 13 (35) 11 (29) 12 (55) .626/.078Smoking status .187/.028
Never 13 (35) 20 (53) 16 (73)Former 19 (51) 12 (32) 6 (27)Current 5 (14) 6 (16) 0 (0)
White race 33 (89) 36 (95) 18 (82) .430/.227Education, median years (range) 15 (10–20) 14 (12–22) 17 (12–25) .550/�.001
�High school graduate 35 (95) 38 (100) 21 (100) .240/1.000Chronic sinusitis 11 (30) 14 (37) 17 (77) .626/�.001AtopyAllergy tested 18 (53) 18 (49) 19 (86) .814/.003
Negative 2 (11) 1 (6) 6 (32) .759/.003Positive for 1–3 positive 4 (22) 3 (17) 8 (42)Positive for �4 12 (67) 14 (78) 5 (26)
Infectious asthma† 6 (16) 17 (46) 13 (59) .011/.024Exacerbations (previous 2 years), n (%) 24 (65) 26 (68) 14 (64) .809/.802
Hospitalized 0 (0) 2 (5) 4 (18) .493/.023Emergency visit 14 (38) 19 (50) 9 (41) .355/1.000Steroid burst 22 (59) 24 (63) 13 (59) .815/1.000
Baseline asthma severity, n (%)Day symptom frequency‡ .675/.009
Mild to moderate 35 (95) 34 (89) 15 (68)Severe 2 (5) 4 (11) 7 (32)
Night symptom frequency‡ 1.000/.046Mild to moderate 33 (89) 33 (87) 15 (68)
Severe 4 (11) 5 (13) 7 (32)Coexisting COPD 8 (22) 5 (13) 2 (9) .375/.509Lung function, mean (SD)FEV1, L (n) 18 19 7
Low§ 2.24 (1.25) 2.33 (1.05) 2.48 (1.19) .812/.688% Change� 42 (47.4) 26 (25.9) 33 (26.8) .214/.969
PEFR, L/min (n with value) 25 25 18Low§ 258 (110) 276 (110) 300 (105) .566/.281% Change� 62 (56) 63 (49) 85 (63) .927/.140
Any controller medication 33 (89) 25 (66) 19 (86) .026/.549Inhaled corticosteroid 30 (81) 24 (63) 18 (82)Long-acting bronchodilator¶ 26 (70) 14 (37) 15 (68)Leukotriene inhibitor 8 (22) 9 (24) 6 (27)Oral prednisone 4 (11) 2 (5) 1 (5)
Continued
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 445
received coded study medication bottles (1:1 allo-cation) in blocks of 6 and was instructed to distrib-ute them (numbered 1 to 6) in numerical ascendingorder to eligible consenting study subjects. After 3weeks of taking study medication, subjects wereasked to guess their allocation.
Open-Label TreatmentInternet trial registration (http://clinicaltrials.gov/show/NCT00266851) and another Internet site(http://asthmastory.com) identified asthma patientswanting to participate in AZMATICS.13 Whenthey learned that they had a 50% chance of receiv-ing a placebo, many of these patients declined to berandomized but remained interested in the use ofazithromycin for treatment of their asthma. Insti-tutional review board approval was obtained to en-ter eligible subjects requesting azithromycin into aparallel, OL observational cohort. OL subjects ob-tained a 12-week prescription for weekly azithro-mycin from their personal physician and weremonitored for the same baseline and outcome datathat were being collected for the randomized co-hort. Because 600-mg azithromycin tablets werenot uniformly available, OL treatment consisted of2 azithromycin tablets (250 mg each), to achieve a500-mg daily dose for 3 days, followed by 3 tabletsto achieve a 750-mg, once-weekly dose for 11 ad-ditional weeks.
Study OutcomesSubjects submitted follow-up data via Internetquestionnaires. During the first 3 months after ran-
domization (the study medication administrationperiod), subjects reported weekly via the Internetwhether they had taken their assigned study med-ication during the previous week. In addition tostudy medication adherence, subjects also reportedweekly on side effects for the first 3 months. Out-come data were recorded every 1.5 months (6weeks) through 12 months. Outcome measures arelisted in Table 1.
Statistical AnalysisAll analyses were by intention to treat, and nosubjects with available data were excluded from anyanalysis. Repeated-measures (RM) analysis of vari-ance (ANOVA)—with intervention and time asfixed effects; subjects as random effects; and age,sex, and ever-smoking as covariates—was con-ducted for each of the continuous variables. Afterfinding significant effects in the RM ANOVA mod-els, we explored the effects with tests at specifictime points. We used Fisher exact test for categor-ical variables, Wilcoxon rank sum tests for contin-uous variables (reported as median and range), andt tests for continuous variables (reported as meanand standard deviation). We controlled for the ef-fects of age, sex, ever-smoking, and asthma control-ler medication using ANOVA for normally distrib-uted continuous outcomes and logistic regressionfor binary outcomes. On the basis of our pilotresults, a total sample size of 58 had 80% power(for � � 0.05) to detect a 0.66-unit (13%) differ-ence in overall asthma symptoms (the primary out-come). AZMATICS used a Data Safety Monitoring
Table 2. Continued
CharacteristicRandomized
Placebo (n�37)Randomized
Azithromycin (n�38)Open-Label
Azithromycin (n�22) P*
Baseline asthma measures, mean (SD)**Overall asthma symptoms 1.48 (0.94) 1.42 (0.77) 2.06 (0.73) .744/.005Asthma quality of life 4.97 (1.28) 4.98 (1.27) 4.12 (1.29) .988/.023Asthma control 1.56 (1.02) 1.75 (0.93) 2.26 (1.35) .424/.090
Values provided as n (%) unless otherwise indicated.*P values comparing randomized placebo versus randomized azithromycin/randomized cohort versus open-label cohort.†History showed first asthma symptoms began after an acute respiratory illness.‡Day: mild � �2 days/week to less than daily; moderate � �1 per day to less than continuous; severe � continuous. Night: mild ��2 per month to �1 per week; moderate � �1 per week to �1 per night; severe � �1 per night.§Before bronchodilator use or lowest spontaneous value.�Based on data after bronchodilator use or highest spontaneous value. Some subjects had either FEV1 or PEFR values, not both.¶All subjects using a long-acting bronchodilator also were using an inhaled corticosteroid.**See Methods for definitions.COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 second; PEFR, peak expiratory flow rate.
446 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
Board that met approximately every 6 months. TheData Safety Monitoring Board did not identify anyreason for early termination.
ResultsScreening and EnrollmentOf 304 adult asthma patients screened, 97 (32%)were enrolled (38 were randomized to azithromy-cin, 37 were randomized to placebo, and 22 electedOL treatment). An additional 67 of 304 screenedpatients (22%) who were otherwise eligible lacked
pulmonary function data and could not completethe screening process (Figure 1).
AdherenceSelf-reported mean adherence to taking azithromy-cin or placebo ranged from 96% to 99%, with nosignificant differences among the 3 study groups(P � .706). Adherence to reporting follow-up dataranged from 63 of 97 participants (65%) at 12weeks to 57 participants (59%) at 48 weeks, with no
Figure 2. Differences from baseline for asthma symptoms, quality of life, and control. A: Symptoms—rating of overall asthmasymptoms for the past 24 hours (negative numbers indicate decreased symptoms and hence improvement). B: Quality of life—Juniper Asthma Quality of Life Questionnaire (AQLQ); positive numbers indicate higher quality of life scores and henceimprovement). C: Control—Juniper Mini-Asthma Control Questionnaire (ACQ; negative numbers indicate better control andhence improvement). See Methods for details. Symbols represent the mean paired differences from baseline. Bars represent95% confidence intervals. *P < 0�05, **P < 0�01, ***P < 0�001 (t tests, placebo versus open label).
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 447
Table 3. Asthma Outcomes*
RandomizedPlacebo
RandomizedAzithromycin
Open-labelAzithromycin
P, Placebo vsRandomizedAzithromycin
P, Placebo vsOpen-Label
Change in overall asthmasymptoms, frombaseline
Week 6 �0.60 (1.07) (n�30) �0.15 (0.83) (n�33) �0.88 (0.81) (n�16) .071/.071† .333/.428†
Week 12 �0.48 (1.16) (n�23) �0.31 ((0.74) (n�32) �1.0 (1.37) (n�16) .551/.580† .223/.723†
Week 18 �0.15 (1.13) (n�27) 0.10 (0.75) (n�31) �0.94 (1.12) (n�16) .344/.178† .034/.161†
Week 24 �0.08 (0.95) (n�25) �0.10 (0.96) (n�30) �1.0 (1.17) (n�17) .939/.599† .012/.013†
Week 30 �0.05 (1.09) (n�22) �0.34 (1.01) (n�29) �0.81 (1.17) (n�16) .322/.280† .048/.870†
Week 36 �0.21 (1.14) (n�24) �0.22 (0.70) (n�27) �1.27 (0.70) (n�15) .959/.942† .001/.017†
Week 42 �0.04 (0.88) (n�23) �0.12 (0.93) (n�25) �1.21 (0.70) (n�14) .770/.858† �.001/.001†
Week 48 �0.10 (1.07) (n�20) �0.07 (0.88) (n�29) �1.07 (0.95) (n�13) .916/.758† .011/.067†
Change in AQL, frombaseline
Week 12 0.50 (0.95) (n�22) 0.67 (1.10) (n�26) 1.54 (1.91) (n�15) .584/.682† .067/.180†
Week 24 0.31 (1.36) (n�22) 0.49 (1.11) (n�27) 1.36 (1.75) (n�17) .618/.382† .049/.151†
Week 36 0.23 (1.02) (n�23) 0.58 (1.04) (n�22) 2.05 (1.40) (n�15) .261/.342† �.001/.001†
Week 48 0.40 (1.33) (n�19) 0.50 (1.10) (n�25) 1.70 (1.42) (n�13) .784/.929† .015/.068†
Change in asthmacontrol, frombaseline
Week 6 �0.37 (0.88) (n�30) �0.46 (0.60) (n�33) �1.24 (1.14) (n�16) .634/.654† .014/.009†
Week 12 �0.41 (1.01) (n�23) �0.40 (0.80) (n�32) �1.38 (1.87) (n�16) .946/.998† .075/.324†
Week 18 �0.40 (1.05) (n�27) �0.28 (0.88) (n�31) �1.18 (1.53) (n�16) .637/.573† .085/.179†
Week 24 �0.37 (1.12) (n�25) �0.34 (1.03) (n�30) �1.35 (1.69) (n�17) .925/.536† .045/.034†
Week 30 �0.25 (1.22) (n�22) �0.56 (0.81) (n�29) �1.04 (1.38) (n�16) .307/.281† .078/.163†
Week 36 �0.39 (1.00) (n�24) �0.39 (0.79) (n�27) �1.63 (1.41) (n�15) 1.000/.852† .007/.015†
Week 42 �0.22 (1.32) (n�23) �0.38 (0.72) (n�25) �1.42 (1.41) (n�14) .604/.817† .016/.068†
Week 48 �0.45 (1.00) (n�20) �0.34 (0.88) (n�29) �1.08 (1.20) (n�13) .692/.525† .132/.379†
AQL improved �1 unit,n/N (%)
Week 12 5/22 (23) 11/26 (42) 9/15 (60) .221/.136‡ .038/.098‡
Week 24 6/22 (27) 8/27 (30) 11/17 (65) 1.000/.745‡ .026/.048‡
Week 36 5/23 (22) 6/22 (27) 12/15 (80) .738/.738‡ .001/.003‡
Week 48 4/19 (21) 9/25 (36) 7/13 (54) .335/.386‡ .072/.116‡
Asthma control improved�1 unit, n/N (%)
Week 6 6/30 (20) 4/33 (12) 9/16 (56) .498 /.421‡ .021/.010‡
Week 12 7/23 (30) 7/32 (22) 11/16 (69) .539/.531‡ .025/.045‡
Week 18 7/27 (26) 4/31 (13) 9/16 (56) .315/.437‡ .059/.054‡
Week 24 6/25 (24) 10/30 (33) 10/17 (59) .556/.144‡ .029/.017‡
Week 30 5/22 (23) 9/29 (31) 8/16 (50) .546/.502‡ .098/.105‡
Week 36 5/24 (21) 6/27 (22) 10/15 (67) 1.000/.875‡ .007/.009‡
Week 42 6/23 (26) 7/25 (28) 8/14 (57) 1.000/.862‡ .085/.152‡
Week 48 5/20 (25) 8/29 (28) 5/13 (38) 1.000/.980‡ .461/.998‡
All values are mean (SD) unless otherwise indicated.*See Methods for definitions.†Univariate (t test)/multivariate (analysis of variance); controlled for age, sex, and ever-smoking at each time point, as well as forcontroller medication use at weeks 12, 24, 36, and 48 (controller data is unavailable for other time points).‡Univariate (Fisher exact test)/multivariate (logistic regression); controlled for age, sex, and ever-smoking at each time point, as wellas for controller medication use at weeks 12, 24, 36, and 48 (controller data is unavailable for other time points).AQL, asthma quality of life.
448 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
significant differences between study groups (P �.122).
Baseline CharacteristicsTable 2 presents the patient characteristics. Thestudy group was mostly white and non-Hispanicwith a high school education or greater and a broadrange of age at asthma onset. The randomizedgroups were well balanced in major baseline char-acteristics (age, sex, smoking status, education, andasthma severity), but the group randomized to azi-thromycin reported more asthma onset after anacute respiratory illness (“infectious asthma”) andless use of controller medication compared with theplacebo group. Compared with the randomizedgroups, the OL cohort had significantly greaterasthma severity at baseline (more hospitalizationsfor asthma, greater day and night symptom fre-quency, worse overall asthma symptoms, and worse
asthma quality of life [AQL]); more adult-onsetasthma; more chronic sinusitis; more allergy testing(but fewer positive tests); and more infectiousasthma. Of the baseline covariates in RM ANOVAs,smoking was significantly associated with worsesymptoms, AQL, and asthma control (P � .05 foreach); older age was significantly associated onlywith worse AQL (P � .05), and sex was not signif-icant in any analysis.
OutcomesRandomized TrialAfter 3 weeks of receiving study medication, 13%of those randomized to placebo and 29% random-ized to azithromycin correctly guessed their alloca-tion; 34% of placebo and 29% of azithromycinsubjects guessed incorrectly, and the remainderwere unsure of their allocation (P � .27). Subjects
Figure 3. Improvement in asthma quality of life (AQL) after azithromycin treatment may be “all or none.” AQLchange scores from baseline to 12 months after enrollment (9 months after treatment completion) are defined asfollows: “AQL change <0” � AQL change worse than baseline; “AQL change 0 < .5” � change between 0 to <.5units; “AQL change 0.5 < 1” � change between 0.5 to <1 (change of 0.5 unit is considered the minimumclinically important change); “AQL change 1 < 2” � change between 1 and 2 (change >1.5 units represents alarge change); “AQL change >2” � change of 2 units or more. The contrasting patterns between placebo andopen-label azithromycin were statistically significant, as noted in the text. The differences between randomizedazithromycin and placebo were not statistically significant, as noted in the text.
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
Randomized to Placebo Randomized to Azithromycin Open-label Azithromycin
AQL change <0AQL change 0<.5AQL change .5<1AQL change 1<2AQL change ≥2
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 449
randomized to azithromycin or placebo had nosignificant differences in overall asthma symptoms,AQL, or asthma control (Figure 2 and Table 3). At1 year, subjects randomized to azithromycin weremore likely than placebo subjects to have an AQLscore �1-unit increase compared with baseline(36% vs 21%). This difference was not statisticallysignificant (P � .335). Compared with subjects ran-domized by PBRN members (n � 69), subjectsrandomized by the community allergist (n � 6) weremore likely to have been skin tested (49% of PBRNsubjects skin tested vs 100% of allergist subjects; P �.024) but otherwise were similar in baseline charac-teristics, including comparable distribution of skin
test results. Removing the allergist-recruited subjectsfrom the RM ANOVA did not alter the outcomeresults (data not shown).
Open-Label CohortComparing the OL and placebo groups, RMANOVA found that the effect of intervention andthe interaction of intervention and time were sig-nificant for symptoms, AQL, and asthma control(P � .05 for each). In univariate analyses, OL groupasthma symptoms were significantly improvedfrom month 4.5 to month 12 compared with pla-cebo and AQL was significantly improved frommonth 6 to month 12, whereas asthma control
Table 4. Side Effects*
Side Effects, n (%) Randomized Placebo Randomized Azithromycin Open-Label Azithromycin P†
Nausea .016/.008/.458None 31 (91) 25 (71) 12 (60)Mild to moderate 1 (3) 9 (26) 6 (30)Severe 2 (6) 1 (3) 2 (10)
Vomiting 1.000/.128/.456None 32 (94) 33 (94) 18 (90)Mild to moderate 0 (0) 1 (3) 2 (10)Severe 2 (6) 1 (3) 0 (0)
Stomach pain .076/.001/.196None 30 (88) 24 (69) 9 (45)Mild to moderate 3 (9) 10 (29) 10 (50)Severe 1 (3) 1 (3) 1 (5)
Diarrhea .106/.002/.196None 29 (85) 23 (66) 9 (45)Mild to moderate 3 (9) 10 (29) 10 (50)Severe 2 (6) 2 (6) 1 (5)
Rash .743/.046/.420None 33 (97) 33 (94) 16 (80)Mild to moderate 0 (0) 2 (6) 3 (15)Severe 1 (3) 1 (3) 1 (5)
Swelling .239/.716/.128None 32 (94) 35 (100) 18 (90)Mild to moderate 1 (3) 0 (0) 2 (10)Severe 1 (3) 0 (0) 0 (0)
Hearing loss .743/1.000/.999None 32 (94) 34 (97) 19 (95)Mild to moderate 1 (3) 1 (3) 1 (5)Severe 1 (3) 0 (0) 0 (0)
Vaginal candidiasis .670/1.000/.999None 20 (91) 21 (84) 7 (88)Mild to moderate 3 (9) 4 (16) 1 (13)Severe 0 (0) 0 (0) 0 (0)
*Worst reported severity during the 12-week treatment period.†Fisher exact tests: randomized placebo versus randomized azithromycin/randomized placebo versus open-label azithromycin/randomized azithromycin versus open-label azithromycin.
450 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
Tabl
e5.
Rand
omiz
edTr
ials
ofSe
cond
-Gen
erat
ion
Mac
rolid
es/A
zalid
esfo
rAs
thm
a:St
udy
Des
igns
Ref
eren
ceA
geG
roup
Sam
plin
gFr
ame
Ast
hma
Seve
rity
Subj
ects
(n)
Des
ign
RX
/Dur
atio
nO
bser
vatio
naf
ter
RX
Shoj
i(19
99)1
7A
dults
Hos
pita
last
hma
clin
icM
ild/m
oder
ate
14Si
ngle
site
;dou
ble-
blin
dcr
oss-
over
(4-w
eek
was
hout
)R
oxith
rom
ycin
,300
mg
daily
orpl
aceb
o/8
wee
ksN
one
Asp
irin
-int
oler
ant
asth
ma
Am
ayas
u(2
000)
18
Adu
ltsN
otst
ated
(hos
pita
las
thm
acl
inic
�s�?
)M
ild/m
oder
ate
17Si
ngle
site
;dou
ble-
blin
dcr
oss-
over
(4-w
eek
was
hout
)C
lari
thro
myc
in,2
00m
gda
ilyor
plac
ebo/
8w
eeks
Non
e
Bla
ck(2
001)
19
18–6
0ye
ars
old
Maj
ority
ofsu
bjec
tsre
crui
ted
from
the
gene
ralp
opul
atio
n;re
crui
tmen
tm
etho
d(s)
not
spec
ified
Mod
erat
e/se
vere
232
Mul
tinat
iona
l;do
uble
-blin
d,pa
ralle
lgr
oups
(Aus
tral
ia,N
ewZ
eala
nd,
Ital
y,A
rgen
tina)
Rox
ithro
myc
in,3
00m
gda
ilyor
plac
ebo/
6w
eeks
24w
eeks
Kra
ft(2
002)
20
You
ngad
ults
Subj
ects
recr
uite
dfr
omth
ege
nera
lpop
ulat
ion
via
adve
rtis
ing
Not
stat
edM
ean
FEV
1%
pred
�69
.335
%w
ere
taki
ngIC
S
52Si
ngle
site
;dou
ble-
blin
dpa
ralle
lgr
oups
All
subj
ects
unde
rwen
tbr
onch
osco
pybe
fore
and
afte
rpr
escr
iptio
n
Cla
rith
rom
ycin
,100
0m
gda
ilyor
plac
ebo/
6w
eeks
Non
e
Kos
tadi
ma
(200
4)2
118
–70
year
sol
dN
otst
ated
(ref
erra
lsp
ecia
lity
sett
ing)
Not
stat
ed,p
roba
bly
mild
63Si
ngle
site
;dou
ble-
blin
d,pa
ralle
lgr
oups
Cla
rith
rom
ycin
,500
or75
0m
gda
ilyor
plac
ebo/
8w
eeks
Non
eM
ean
FEV
1%
pred
�85
%Su
bjec
tsus
ing
albu
tero
l�
2tim
esw
eekl
yw
ere
excl
uded
Hah
n(2
006)
22
�18
year
sol
dC
omm
unity
-bas
edhe
alth
care
sett
ings
Mos
tlym
ild/m
oder
ate
46M
ultis
ite;d
oubl
e-bl
ind,
para
llel
grou
psA
zith
rom
ycin
,600
mg
daily
for
3da
ysth
en60
0m
gw
eekl
yor
plac
ebo/
6w
eeks
12w
eeks
Pia
cent
ini(
2007
)23
Chi
ldre
nIn
patie
ntse
ttin
gN
otst
ated
16Si
ngle
site
;dou
ble-
blin
d,pa
ralle
lgr
oups
Azi
thro
myc
in,1
0m
g/kg
/day
for
3of
7da
ysor
plac
ebo/
8w
eeks
Non
e
Sim
pson
(200
8)2
4A
dults
Spec
ialty
outp
atie
ntcl
inic
Seve
rere
frac
tory
45Si
ngle
site
;dou
ble-
blin
d,pa
ralle
lgr
oups
Cla
rith
rom
ycin
,100
0m
gda
ilyor
plac
ebo/
8w
eeks
4w
eeks
Stra
tified
byhi
gh(�
61%
)/lo
win
duce
dsp
utum
neut
roph
ilpr
opor
tion
Stru
nk(2
008)
25
Chi
ldre
nA
cade
mic
asth
ma
cent
ers
Mod
erat
e/se
vere
55M
ultis
ite;d
oubl
e-bl
ind,
para
llel
grou
psT
his
was
ast
udy
ofm
acro
lide
asa
“ste
roid
-spa
ring
”ag
ent,
not
asan
antim
icro
bial
Azi
thro
myc
in,2
50–5
00m
gda
ilyor
mon
telu
kast
5–10
mg
daily
ofpl
aceb
o/24
wee
ks
6w
eeks Con
tinue
d
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 451
improvements were less consistent (Figure 2). Inmultivariate analyses (controlled for age, sex, ever-smoking, and concurrent controller medicationuse) the OL group reported significant improve-ments in symptoms, AQL, and asthma control thatwere maximal at study month 9 and waned some-what at month 12 (Table 3).
AQL score change of 1 unit or more was aprespecified secondary outcome. AQL � 1-unitimprovement was achieved significantly more oftenin OL than in placebo subjects at the 3-, 6-, and9-month time points. A similar but less consistentpattern was noted for asthma control scores with a�1-unit improvement (Table 3). After adjustmentfor age, sex, ever-smoking, and asthma controllermedication use at 9 months (6 months after treat-ment completion), 80% of OL patients versus 22%of those enrolled in the placebo arm reported AQLscore changes of �1-unit improvement (P � .001;number needed to treat [NNT] � 2) and 67%versus 21% reported asthma control score changesof �1-unit improvement (P � .023; NNT � 3).AQL and asthma control score improvements of�1 unit also were correlated significantly with self-reported asthma improvement at all time points(P � .01 for each).
We performed further exploratory analyses of dif-ferent increments of change in AQL score up to andincluding a change of �2 units (Figure 3). The resultsshowed that changes in AQL after azithromycin(both randomized or OL) assumed U-shaped distri-butions, whereas changes in AQL for placebo wereskewed to the left, suggesting a binary “all or none”response to azithromycin. Finding patient character-istics that predict a treatment response are potentiallyimportant. Other than asthma severity, however, ourdata yielded no indications that patients’ clinical char-acteristics were predictive of an azithromycin treat-ment response. For example, we analyzed OL andplacebo subjects in a logistic model of AQL change of�2 units from baseline as the dependent variable andincluded age, sex, smoking status, and “infectiousasthma” as other possible predictors. In this model,only azithromycin treatment was a significant predic-tor of AQL �2 (P � .026).
ExacerbationsThis study was not powered to detect significantdifferences in exacerbation frequency. During the12-month study period, 51% of subjects reportedone or more asthma exacerbations. There were noTa
ble
5.Co
ntin
ued
Ref
eren
ceA
geG
roup
Sam
plin
gFr
ame
Ast
hma
Seve
rity
Subj
ects
(n)
Des
ign
RX
/Dur
atio
nO
bser
vatio
naf
ter
RX
Suth
erla
nd(2
010)
26
18–6
0ye
ars
old
Aca
dem
icas
thm
ace
nter
sSu
bopt
imal
lyco
ntro
lled
asth
ma
92M
ultis
ite;d
oubl
e-bl
ind,
para
llel
grou
psSt
ratifi
edon
Mpn
orC
pnP
CR
(bro
ncho
scop
icsa
mpl
ing)
The
stud
yw
asun
derp
ower
edto
test
PC
R
case
s
Cla
rith
rom
ycin
,100
0m
gda
ilyor
plac
ebo/
16w
eeks
Non
e
Hah
n(2
012)
,cu
rren
tst
udy
�18
year
sol
dC
omm
unity
-bas
edhe
alth
care
sett
ings
Mild
/mod
erat
e(r
ando
miz
ed)
75ra
ndom
ized
22op
en-l
abel
Mul
tisite
;dou
ble-
blin
d,pa
ralle
lgr
oups
Azi
thro
myc
in,6
00m
gda
ilyfo
r3
days
then
600
mg
wee
kly
orpl
aceb
o/12
wee
ks
36w
eeks
Seve
re(o
pen-
labe
l)
%pr
ed,p
erce
ntof
the
pred
icte
dva
lue;
Cpn
,Chl
amyd
iapn
eum
onia
e;FE
V1,
forc
edex
pira
tory
volu
me
in1
seco
nd;I
CS,
inha
led
cort
icos
tero
id;M
pn,M
ycop
lasm
apn
eum
onia
e;P
CR
,pol
ymer
ase
chai
nre
actio
n.
452 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
significant differences between the 3 study groupsin exacerbation frequency at any time point orcumulatively.
Serious Adverse Events and Side EffectsOne subject allocated to placebo was hospitalizedfor acute coronary syndrome. Another subject
Table 6. Randomized Trials of Second-Generation Macrolides/Azalides for Asthma: Exclusions, Outcomes, andResults
Reference Exclusions* Outcomes Reported Results of Macrolide Treatment
Shoji (1999)17 Smokers Blood eosinophils and ECP Decreased eosinophils/ECPController medication Sputum cell counts and ECP Decreased eosinophils/ECP
(No differences in sputumneutrophils)
Sulpyrine provocation test Not improved(No patient-oriented outcomes
reported)Amayasu (2000)18 Smokers Blood eosinophils and ECP Decreased eosinophils/ECP
Aspirin sensitivity Sputum cell counts and ECP Decreased eosinophils/ECPARI for 6 weeks BHR ImprovedAny asthma controller
medicationPulmonary function Not improvedOverall asthma symptoms Improved
Black (2001)19 FEV1 �50% predicted Pulmonary function(PEF) Improvement at end of prescriptionthat waned after prescription
C. pneumoniae IgG � 1:64 andIgA � 1:16
Pulmonary function(FEV1) Not improved
Smoking �20 pack-years Asthma symptoms Not improvedBronchiectasis AQL Not improvedPrednisone burst in previous
monthRespiratory illness
Kraft (2002)20 Smoking �5 pack years PCR for Mpn or Cpn 31 of 55 were PCR for Mpn and/or Cpn
Any cigarette within 2 years Pulmonary function Improved FEV1 in PCR subjectsubgroup
Any lung comorbidity Lung inflammation Decreased inflammatory cytokinesAny LRTi within 3 months (No patient-oriented outcomes
reported)Kostadima (2004)21 Asthma diagnosis �1 year ago BHR Decreased BHR
Not on ICS Pulmonary function Not improvedRescue inhaler �2 times
weeklySerum free cortisol Not affected
Any smoking history (No patient-oriented outcomesreported)Any other medication
FEV1 � 50% predictedAny ARI or exacerbation
within 4 weeks before orduring the study
Hahn (2006)22 None AQL No improvementRescue medication use No improvementCpn IgG and IgA antibodies Baseline IgA predicted worsening
symptomsOverall asthma symptoms Improved at end of prescription
and persisted after prescriptionPiacentini (2007)23 Oral steroids in the preceding
3 months or during thestudy
Lung function No improvement
Signs of airway infection inthe preceding month orduring the study
BHR ImprovedLung inflammation Reduced induced sputum
neutrophils
Continued
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 453
allocated to placebo discontinued study medica-tion because of side effects. Compared with pla-cebo, subjects taking azithromycin (randomizedand OL combined) reported significantly morenausea (33% vs 9% for placebo), stomach pain(42% vs 12% for placebo), and diarrhea (42% vs15% for placebo). The majority of these sideeffects were mild to moderate in severity and nosubject taking azithromycin (either randomizedor OL) reported discontinuation because of sideeffects. There were no significant differences inside effect frequency or severity when the arm ran-
domized to azithromycin was compared with the co-hort that elected OL azithromycin (Table 4).
DiscussionWe found no significant treatment effect for sub-jects randomized to azithromycin. Our a prioripower calculations were based on overall asthmasymptom results from a previous pilot study thatdid not experience self-exclusion of patients withsevere asthma, and in this study we were unable todemonstrate any statistically significant effects of
Table 6. Continued
Reference Exclusions* Outcomes Reported Results of Macrolide Treatment
Simpson (2008)24 Current smoking Sputum inflammatory markers Decreased airway IL-8 andneutrophils
History of smoking, �5 pack-years
Pulmonary function No improvement
Antihistamine medication BHR No improvementAsthma control No improvementAsthma symptoms Decreased wheezing after
prescriptionAQL Improved (NNT � 6 for �0.5
units improvement)It was unclear whether the AQL
was reported at the end of theprescription or after theprescription
Strunk (2008)25 No controller medication Time to inadequate control aftersteroid step-down
No improvement in asthma control(futility analysis)
FEV1 � 50%pred Recruitment was discontinued early(292 screened, only 55randomized)
�3 hospitalizations in pastyear
Sinus surgery in past yearLung comorbidities
Sutherland (2010)26 Exacerbation within 6 weeks Asthma control No differences in asthma controlARI within 6 weeks Pulmonary function No improvement�2 exacerbations or ARI prior
to entryExhaled nitric oxide No improvement
Smoking BHR ImprovedHistory of smoking, �10
pack-yearsRescue medication use No improvement
Lung comorbidities AQL No improvementHahn (2012), current
studyNone Overall asthma symptoms Randomized: no improvements in
any outcomeAQL Open label: improved overall
asthma symptoms and AQLscore at end of prescription thatpersisted after prescription(improvements maximal at the9-month study point)
ACQ
*Other than for safety and logistics.ACQ, asthma control questionnaire; ARI, acute respiratory illness; AQL, asthma quality of life; BHR, bronchial hyperresponsiveness;Cpn, Chlamydia pneumoniae; ECP, eosinophil cationic protein; FEV1, forced expiratory volume in 1 second; ICS, inhaled cortico-steroid; Ig, immunoglobulin; IL, interleukin; LRTi, lower respiratory tract illness; Mpn, Mycoplasma pneumoniae; NNT, numberneeded to treat; PCR, polymerase chain reaction; PEF, peak expiratory flow.
454 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
treatment on overall asthma symptoms. We founda 15% difference in AQL score of �1-unit im-provement favoring azithromycin (NNT � 7) at 12months that was not statistically significant (Table 3).Because a change in AQL score �1 represents animportant clinical improvement, we advocate fu-ture azithromycin effectiveness trials of patientswith mild to moderate asthma that are adequatelypowered to detect, at a minimum, an NNT of 10 to20 for this chronic, morbid, and expensive condi-tion.
The participants in the OL cohort had moresevere asthma than those randomized to azithro-mycin, and their asthma was often refractory toguideline treatment. The OL cohort demon-strated statistically and clinically significant ben-efits that largely persisted at 12 months com-pared with placebo treatment. Six months aftercompleting azithromycin, the OL cohort experi-enced a 21% improvement in overall asthmasymptoms, a 1.8-unit (26%) improvement inAQL score and a 1.2-unit (20%) improvement inasthma control score. This AQL score changewas more than 3 times greater than the minimalclinically important change (0.5 units) and ex-ceeded the threshold (1.5 units) for a largechange.14 These azithromycin treatment benefitsare greater and of longer duration than thoseachieved by current guideline treatments. Forexample, a recent efficacy trial in nonsmokingadults with uncontrolled asthma achieved lowerbenefits (1) when inhaled steroid dose was dou-bled (0.05 AQL questionnaire units, 0.03 asthmacontrol units); (2) when the long-acting broncho-dilator salmeterol was added (0.28 AQL ques-tionnaire units, 0.31 asthma control units); or (3)when tiotropium was prescribed (0.15 AQL ques-tionnaire units, 0.22 asthma control units).15 InAZMATICS, treating only 2 OL subjects withazithromycin was required to achieve an AQLscore improvement of 1 unit or greater in one ofthem (NNT � 2) at 9 months. Because 80% ofasthma morbidity and health care utilization isexperienced by patients with the most severeforms of the disease,16 our results suggest thatazithromycin therapy may be a promising novelintervention to decrease the burden of morbidityand cost associated with management of severeand uncontrolled adult asthma, and futureblinded, randomized trials are warranted.
Previous StudiesTen randomized controlled trials of second-gener-ation macrolides/azalides (azithromycin, clarithro-mycin, and roxithromycin) for chronic stableasthma have been published (see Tables 5 and 6 formore details).17–26 A 2005 Cochrane review con-cluded that there was insufficient evidence to con-firm or refute the role of macrolide treatment inchronic asthma.1 Since 2005, 3 additional trials ofmacrolides in chronic adult asthma have reportedpatient-oriented outcomes.22,24,26 Hahn et al22 (ourpilot) performed a practice-based effectiveness trialthat included mainly patients with mild to moder-ate asthma who were randomized to 6 weekly dosesof azithromycin or placebo. Outcomes of interestincluded asthma symptoms, AQL, and levels of C.pneumoniae–specific antibodies up to 3 months afterthe completion of treatment. Azithromycin treat-ment had a significant effect on asthma symptomreduction during the treatment phase, which per-sisted through study termination 3 months later.Elevated levels of C. pneumoniae–specific immuno-globulin A antibodies predicted worsening ofasthma symptoms.22 AQL improved by 0.25 units,but the change was not statistically significant.Simpson et al24 performed an efficacy trial in non-smoking adults with severe refractory asthma whowere randomized to 8 weeks of clarithromycin orplacebo and were followed for an additional 4weeks after treatment. Clarithromycin treatmentsignificantly improved AQL at the end of treatment(median AQL questionnaire score at baseline, 5.5;median score at end of treatment, 6.2; P � .01).The improvement was more pronounced insubjects with “noneosinophilic” (“neutrophilic”)asthma and waned 4 weeks after treatment.24
Sutherland et al26 performed an efficacy study thatincluded bronchoscopic polymerase chain reaction(PCR) testing for C. pneumoniae and M. pneumoniaein a highly selected group of mild to moderateasthmatics who were randomized to 16 weeks ofclarithromycin or placebo without observation af-ter treatment. Clarithromycin treatment did notimprove overall asthma control at the end of treat-ment. The PCR-positive subgroup had weak evi-dence (P � .06) of a more rapid achievement inasthma control score �0.5 units (the minimal im-portant clinical difference). This study excludedsubjects older than 60 years, those with severeasthma, smokers, those with coexisting chronic ob-structive pulmonary disease (COPD), and any pa-
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 455
tient with more than 2 exacerbations or respiratorytract infections before study entry (http://www.clinicaltrials.gov/ct/show/NCT00318708?order �2). Each of these exclusion characteristics is associ-ated with biomarkers of C. pneumoniae infec-tion.27–31 These exclusions probably explain thedetection of fewer atypical pathogens than antici-pated (13% PCR positive compared with an ex-pected 50%).26 AZMATICS included at least 53asthma subjects (55%) who would have been ex-cluded from typical asthma efficacy trials such asthat of Sutherland et al26; 15% of AZMATICSsubjects were aged 60 years or older; 24% hadsevere persistent asthma; 11% were current smok-ers; 25% had a history of 10 or more pack-years ofcigarette use; and 15% had coexisting COPD.
AZMATICS’ OL results are consistent withSimpson et al24 for subjects with severe asthma inthat they demonstrate a significant benefit aftermacrolide treatment and are unique among allstudies of macrolide treatment for asthma (see Ta-bles 5 and 6) in that they demonstrate persistentbenefits 1 year after treatment. Outcomes of theAZMATICS study also support findings of Hahnet al22 and Sutherland et al,26 who reported lesseror no benefit associated with macrolide treatmentin subjects with mild to moderate asthma. How-ever, all 3 studies were underpowered to detectclinically significant effects in milder asthma.
Treatment, Side Effects, and Serious AdverseEventsWe chose azithromycin over other macrolides, in-cluding clarithromycin, because it has (1) uniquepharmacodynamic properties that allow weeklydosing to maintain high intracellular drug levels,(2) few drug interactions, and (3) a good safetyprofile that has been demonstrated in more than5000 adults in previous trials.32,33 A recent trial ofdaily azithromycin for 1 year in patients withCOPD actually found a lower incidence of mac-rolide-resistant oral pathogens in the azithromy-cin-treated arm compared with the placebo arm,alleviating concerns about antibiotic resistance.34,35
An additional benefit is that, unlike clarithromycin,azithromycin has no residual taste that might com-promise blinding in some subjects. The cumulativeazithromycin doses in the randomized and OLarms (8400 mg vs 9750 mg, respectively) differedsomewhat, but both treatment regimens resulted inequally effective and prolonged intracellular con-
centrations of azithromycin.36 Side effects weregenerally mild and of similar frequency for bothdoses (Table 4), and no serious adverse events ordiscontinuations because of side effects were attrib-uted to either azithromycin regimen.
Mechanism of ActionOur trial did not directly address mechanism ofaction. AZMATICS was designed to distinguishbetween effects that wane (consistent with directanti-inflammatory mechanisms) or persist (consis-tent with antimicrobial mechanisms) after comple-tion of treatment. We interpret the residual bene-fits found in OL subjects to be most consistent withan antimicrobial mechanism of action.
LimitationsThis trial has several limitations, including the ab-sence of biomarkers for atypical infections and fol-low-up pulmonary function testing (PFT). Lack ofPFT precluded the comparison of objective mea-sures of airway function with the patient-orientedclinical outcomes and decreased our sample size byexcluding enrollment of otherwise-eligible patients(Figure 1). Importantly, AZMATICS was under-powered to detect clinically important improve-ments in subjects with mild to moderate persistentasthma. Our sample was limited to subjects withInternet access, was deficient in minority represen-tation, and did not standardize asthma treatmentacross groups.
Our proportion lost to follow-up exceeded 20%at year 1, which lowered the quality of our trialfrom a level 1 to a level 2 study, according to theStrength of Recommendation Taxonomy, a stan-dard adopted by many primary care publications.37
The placebo control group (with milder asthma)and the OL cohort (with more severe asthma) didnot have comparable asthma prognoses. Becauseasthma prognosis is generally worse for severeasthma, this disparity could have diminished con-trasts (ie, tendency to bias the OL results toward anull effect). By necessity, the OL cohort subjectswere not blinded to azithromycin allocation. If pla-cebo effects are accepted as an explanation onlyduring the time period of medication administra-tion, then placebo effects cannot be invoked as anexplanation for peak benefits that occurred 6months after completing azithromycin treatment.It is possible that other mechanisms, such as denial,cognitive dissonance, or both, could have promoted
456 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
systematic misreporting by OL subjects. A truetreatment effect is supported by the strength andconsistency of the results (including 2 validatedinstruments, the AQL questionnaire and theasthma control questionnaire), by agreement withour blinded pilot results22 and by the results of theprincipal investigator’s (DLH) previous prospec-tive observational cohort study that included pa-tients with confirmed C. pneumoniae infections.38
Although exploratory analyses suggest that smok-ing may have a significant modifying effect,AZMATICS was underpowered for subgroup anal-yses. Future studies should be powered for robustsubgroup analyses, which can be performed only ifthe subgroups are enrolled rather than systemati-cally excluded from enrollment.
The Importance of Effectiveness Studiesof AsthmaA technology assessment commissioned by theNational Asthma Education Panel states that“short-term drug efficacy studies are over-repre-sented in the present literature.”39 Standard ex-clusion criteria in adult asthma efficacy trialsinclude restricted age ranges, pulmonary func-tion parameters, current and previous smoking,and lung comorbidities such as coexistingCOPD.4,5 As a result, 95% of asthma subjectshave been systematically excluded from the trialsused to support guideline recommendations.4,5
Thus, asthma efficacy trials and the guidelinesderived from them may not generalize well to thebroader population of asthma sufferers.
To address some of the limitations of asthmaefficacy studies, AZMATICS was designed as a“pragmatic” or “practical clinical trial” to include adiverse population of study participants, recruit-ment of participants from different practice set-tings, and a range of health outcomes.40,41 Weenrolled subjects from community-based practicesand applied exclusion criteria that were solely basedon safety and logistic considerations (eg, the abilityto complete the study). Furthermore, AZMATICSis, to our knowledge, the first trial to include an OLarm for subjects with severe asthma who otherwisewould have been excluded based on patient prefer-ence. Using this approach, we succeeded in enroll-ing 1 of 3 screened patients; we probably wouldhave been able to enroll closer to 1 in 2 screenedpatients had funding to perform PFT been avail-able (Figure 1). Our actual enrollment (30%) and
potential enrollment (50%) experience exceeds the5% enrollment average for asthma efficacy trials.5
Future larger pragmatic trials may achieve evenhigher enrollment proportions by using physician-diagnosed asthma as the primary eligibility crite-rion and using PFTs as a baseline covariate and anoutcome measure but not as an additional eligibilitycriterion.
ConclusionsThis randomized trial of 12 weekly doses of azi-thromycin failed to demonstrate statistically signif-icant improvements after 1 year in any of thepatient-oriented outcomes that we evaluated. Inter-pretation is complicated because a significant num-ber of eligible asthma subjects—who had greaterthan average severity of the disease—declined to berandomized. Azithromycin treatment was well tol-erated overall, and in the OL group with moresevere, often refractory asthma, there seemed to bepersisting substantive, clinically significant benefitsto asthma symptoms, AQL, and asthma control forat least 6 months in about half of treated subjects.We advocate further effectiveness trials of persis-tent asthma of all severity categories that include anarray of biomarkers to allow for secondary sub-group analyses, the results of which might favorone biological mechanism over another. However,at this time we do not favor approaches that insiston making a microbiologic diagnosis before ran-domization26 or treatment,42 as advocated by oth-ers, because this approach inevitably excludes pa-tients who are unable to undergo or toleratebronchoscopy. Future cost-effectiveness studiesmay help to determine whether making a microbi-ologic diagnosis via bronchoscopy versus treatmentof all severe refractory asthma patients (especiallygiven the NNT suggested by this study) is the morecost-effective approach.
Pending further randomized trials and given therelative safety of azithromycin and the significantdisease burden from severe refractory asthma, pre-scribing prolonged azithromycin therapy to pa-tients with uncontrolled asthma may be consideredby managing clinicians, particularly for patientswho have failed to respond to conventional treat-ment and as an alternative to instituting immuno-modulatory agents.
We gratefully acknowledge the voluntary participation of theData Safety Monitoring Committee (DSMB): Dr. David
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 457
DeMets (chair) and Dr. David Rabago. We also thank Dr. JonTemte for suggesting the use of Internet data collection. Wegratefully acknowledge the voluntary participation of the studysubjects and the following clinicians and staff who enrolledthem: the AAFP National Research Network: Dr. Ed Bujold,Dr. Melissa Devalon, Debbie Graham, Dr. Steve Mattson, Dr.Andrew Pasternak, and Dr. Elisabeth Spector; the AmbulatoryNetwork for Scholarship and Research: Dr. Keith Knepp andDr. Gregg Stoner; the Cleveland Ambulatory Research Net-work: Dr. Sandra Snyder and Dr. Chris Young; the MarshfieldClinic: Dr. Jeremy Bufford, Gloria Cornelius, and Dr. StevenYale; the Oklahoma Physicians Resource/Research Network:Dr. Cheryl Aspey, Eileen Merchen, Katy Duncan-Smith, EmilyTeasdale, and Crystal Turner; and the Wisconsin Research andEducation Network: Dr. Elizabeth Bade, Dr. Jennifer Frank,Dr. Dan Jarzemsky, Dr. Cheri Olson, Therese Pedace, Kather-ine Pronchinske, and Dr. Ayaz Samadini.
References1. Richeldi L, Ferrara G, Fabbri LM, Lasserson TJ,
Gibson PG. Macrolides for chronic asthma. Co-chrane Database Syst Rev 2005;(3):CD002997.
2. Altenburg J, de Graaff CS, van der Werf TS,Boersma WG. Immunomodulatory effects of mac-rolide antibiotics - part 1: biological mechanisms.Respiration 2011;81:67–74.
3. Johnston SL, Martin RJ. Chlamydophila pneumoniaeand Mycoplasma pneumoniae. A role in asthma patho-genesis? Am J Respir Crit Care Med 2005;172:1078–89.
4. Herland K, Akselsen J-P, Skjønsberg OH, BjermerL. How representative are clinical study patientswith asthma or COPD for a larger “real life” pop-ulation of patients with obstructive lung disease?Respir Med 2005;99:11–9.
5. Travers J, Marsh S, Williams M, et al. Externalvalidity of randomised controlled trials in asthma:to whom do the results of the trials apply? Thorax2007;62:219–23.
6. Demoly P, Gueron B, Annunziata K, Adamek L,Walters RD. Update on asthma control in fiveEuropean countries: results of a 2008 survey. EurRespir Rev 2010;19:150–7.
7. Global Initiative for Asthma (GINA) 2007. Global Strat-egy for Asthma Management and Prevention. Updated2007. GINA Executive Committee, 92 pages.
8. American Thoracic Society. Lung function testing:selection of reference values and interpretive strat-egies. Am J Resp Dis 1991;144:1202–18.
9. Dekker FW, Schrier AC, Sterk PJ, Dijkman JH.Validity of peak expiratory flow measurement inassessing reversibility of airflow obstruction. Tho-rax 1992;47:162–6.
10. Juniper EF, Buist AS, Cox FM, Ferrie PJ, King DR.Validation of a standardized version of the AsthmaQuality of Life Questionnaire. Chest 1999;115:1265–70.
11. Juniper EF, O’Byrne PM, Guyatt GH, Ferrie PJ,King DR. Development and validation of a ques-
tionnaire to measure asthma control. Eur Respir J1999;14:902–7.
12. Juniper EF, O’Byrne PM, Roberts JN. Measuringasthma control in group studies: do we need airwaycalibre and rescue beta2-agonist use? Respir Med2001;95:319–23.
13. Hahn DL. An unanticipated effect of clinical trialregistration. BMJ [serial online] Available from:http://www.bmj.com/content/325/7376/1314?tab�responses, 2007. Accessed April 27, 2012.
14. Juniper EF, Guyatt GH, Willan A, Griffeth LE.Determining a minimal important change in a dis-ease-specific quality of life instrument. J Clin Epi-demiol 1994;47:81–7.
15. Peters SP, Kunselman SJ, Icitovic N, et al.Tiotropium bromide step-up therapy for adultswith uncontrolled asthma. N Engl J Med 2010;363:1715–26.
16. Smith DH, Malone DC, Lawson KA, Okamoto LJ,Battista C, Saunders WB. A national estimate of theeconomic cost of asthma. Am J Resp Crit Care Med1997;156:787–93.
17. Shoji T, Yoshida S, Sakamoto H, Hasegawa H,Nakagawa H, Amayasu H. Anti-inflammatory ef-fect of roxithromycin in patients with aspirin-intol-erant asthma. Clin Exp Allergy 1999;29:950–6.
18. Amayasu H, Yoshida S, Ebana S, et al. Clarithro-mycin suppresses bronchial hyperresponsiveness as-sociated with eosinophilic inflammation in patientswith asthma. Ann Allergy Asthma Immunol 2000;84:594–8.
19. Black PN, Blasi F, Jenkins CR, et al. Trial of roxi-thromycin in subjects with asthma and serologicalevidence of infection with Chlamydia pneumoniae.Am J Respir Crit Care Med 2001;164:536–41.
20. Kraft M, Cassell GH, Pak J, Martin RJ. Myco-plasma pneumoniae and Chlamydia pneumoniae inasthma. Effect of clarithromycin. Chest 2002;121:1782–8.
21. Kostadima E, Tsiodras S, Alexopoulos EI, et al.Clarithromycin reduces the severity of bronchialhyperresponsiveness in patients with asthma. EurRespir J 2004;23:714–7.
22. Hahn DL, Plane MB, Mahdi OS, Byrne GI. Sec-ondary outcomes of a pilot randomized trial ofazithromycin treatment for asthma. PLoS Clin Tri-als 2006;1:e11.
23. Piacentini GL, Peroni DG, Bodini A, et al. Azi-thromycin reduces bronchial hyperresponsivenessand neutrophilic airway inflammation in asthmaticchildren: a preliminary report. Allergy Asthma Proc2007;28:194–8.
24. Simpson JL, Powell H, Boyle MJ, Scott RJ, GibsonPG. Clarithromycin targets neutrophilic airway in-flammation in refractory asthma. Am J Respir CritCare Med 2008;177:148–55.
25. Strunk RC, Bacharier LB, Phillips BR, et al. Azi-thromycin or montelukast as inhaled corticoste-
458 JABFM July–August 2012 Vol. 25 No. 4 http://www.jabfm.org
roid-sparing agents in moderate-to-severe child-hood asthma study. J Allergy Clin Immunol 2008;122:1138–44.
26. Sutherland ER, King TS, Icitovic N, et al. A trial ofclarithromycin for the treatment of suboptimallycontrolled asthma. J Allergy Clin Immunol 2010;126:747–53.
27. von Hertzen L, Vasankari T, Liippo K, WahlstromE, Puolakkainen M. Chlamydia pneumoniae andseverity of asthma. Scand J Infect Dis 2002;34:22–7.
28. Hahn DL, Golubjatnikov R. Smoking is a potentialconfounder of the Chlamydia pneumoniae-coro-nary artery disease association. ArteriosclerThromb 1992;12:945–7.
29. Pasternak R, Huhtala H, Karjalainen J. Chlamydo-phila (Chlamydia) pneumoniae serology andasthma in adults: a longitudinal analysis. J All ClinImmunol 2005;116:1123–8.
30. Branden E, Koyi H, Gnarpe H, Tornling G.Chronic Chlamydia pneumoniae infection is a riskfactor for the development of COPD. Respir Med2005;99:20–6.
31. Hahn DL. Infectious asthma: a reemerging clinicalentity? J Fam Pract 1995;41:153–7.
32. Grayston JT, Kronmal RA, Jackson LA, et al. Azi-thromycin for the secondary prevention of coro-nary events. N Engl J Med 2005;352:1637–45.
33. O’Connor CM, Dunne MW, Pfeffer MA, et al.Azithomycin for the secondary prevention of coro-nary heart disease events: the WIZARD Study: arandomized controlled trial. JAMA 2003;290:1459–66.
34. Albert RK, Connett J, Bailey WC, et al. Azithro-mycin for prevention of exacerbations of COPD.N Engl J Med 2011;365:689–98.
35. Hahn DL. Azithromycin for prevention of exacer-bations of COPD. N Engl J Med 2011;365:2235–7.
36. Schentag JJ, Ballow CH. Tissue-directed pharma-cokinetics. Am J Med 1991;91:5S–11S.
37. Ebell MH, Siwek J, Weiss BD, et al. Strength ofrecommendation taxonomy (SORT): a patient-cen-tered approach to grading evidence in the medicalliterature. J Am Board Fam Pract 2004;17:59–67.
38. Hahn DL. Treatment of Chlamydia pneumoniae in-fection in adult asthma: a before-after trial. J FamPract 1995;41:345–51.
39. Aronson N, Lefevre F, Piper M, et al. Managementof chronic asthma. Evidence report/technology as-sessment number 44. AHRQ publication no. 01-E044. Rockville, MD: Agency for Healthcare Re-search and Quality; 2001.
40. Tunis SR, Stryer DB, Clancy CM. Practical clinicaltrials. Increasing the value of clinical research fordecision making in clinical and health policy. JAMA2003;290:1624–32.
41. Thorpe KE, Zwarenstein M, Oxman AD, et al. Apragmatic - explanatory continuum indicator sum-mary (PREVIS): a tool to help trial designers. J ClinEpidemiol 2009;62:464–75.
42. Rollins DR, Beuther DA, Martin RJ. Update oninfection and antibiotics in asthma. Curr AllergyAsthma Rep 2010;10:67–73.
doi: 10.3122/jabfm.2012.04.110309 Azithromycin for Bronchial Asthma in Adults 459