Respiratory Medicine 138 (2018) 129–136

Contents lists available at ScienceDirect

Respiratory Medicine
journal homepage: www.elsevier.com/locate/rmed

Clinical Trial Paper

Long-term azithromycin therapy to reduce acute exacerbations in patients T

with severe chronic obstructive pulmonary disease
Nafiseh Naderia,b,c, Deborah Assayagd, Seyed-Mohammad-Yousof Mostafavi-Pour-Manshadia,b,c,
Zeina Kaddahae, Alexandre Jouberta, Isabelle Ouelleta, Isabelle Drouina, Pei Zhi Lib,
Jean Bourbeaua,b,c,∗
a
Montreal Chest Institute, McGill University Health Centre, Montréal, Québec, Canada
b
Respiratory Epidemiology and Clinical Research Unit, Research Institute of McGill University Health Center, McGill University, Montréal, Québec, Canada
c
Department of Medicine, Division of Experimental Medicine, McGill University, Montreal, Québec, Canada
d
Department of Medicine, McGill University, Montréal, Québec, Canada
e
Charles Lemoyne Hospital, Longueuil, Québec, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: Rationale: According to clinical trials, azithromycin taken daily for 1 year, decreased exacerbations of chronic
Chronic obstructive pulmonary disease obstructive pulmonary disease (COPD).
COPD Objectives: Effectiveness evaluation of long-term azithromycin to reduce exacerbations in severe COPD patient
Long term azithromycin therapy on optimal therapy in real-life practice.
Exacerbations
Methods: We conducted a retrospective observational study of severe COPD patients who were prescribed azi-
Macrolide
thromycin (PA)(250 mg, at least 3 times weekly for at least 6 months). Comparison group included severe COPD
patients not prescribed azithromycin (NPA). Data were extracted from clinical chart review.
Main results: Study included 126 PA and 69 NPA patients. They had severe airflow obstruction, mostly em-
physema and one-third bronchiectasis. A predominant feature in the PA group was respiratory tract colonization
with Pseudomonas aeruginosa. The mean number of exacerbations per patient per year in the PA group was
3.2 ± 2.1 before initiating azithromycin, and 2.3 ± 1.6 during following year on therapy (p < 0.001).
Patients in the NPA group had 1.7 ± 1.3 and 2.5 ± 1.7 exacerbations during first and second follow-up year
respectively (p < 0.001). Exacerbation changes from pre to post differed between groups (p < 0.001).
Decrease in emergency visits and hospital admissions was significant in PA group. Exacerbation reductions and
patient proportions having ≥2 exacerbations extended to the second year of treatment.
Conclusion: These data showed that long-term azithromycin reduces exacerbation numbers in severe COPD
patients, and benefits persist beyond one year. Desirable effects are more likely to outweigh the risks and adverse
events in patients colonized with Pseudomonas aeruginosa.

1. Introduction impact on health status [7–9].About a third of patients with moderate

to severe COPD experience two or more exacerbations per year [10,11].
Chronic obstructive pulmonary disease (COPD) is currently the There seems to be a specific phenotype of patients who are at increased
fourth leading cause of death and will become the third leading cause of risk, where the most important predictor of future exacerbations is the
death by 2030 according to the World Health Organization [1,2]. In number of exacerbations in the previous year [10]. Prevention of ex-
spite of a remarkable decrease in mortality from cardiovascular disease acerbations should be one of the principal management goals in these
in the past three decades, mortality due to COPD has almost doubled patients [2,12].
[2,3].Acute exacerbations are a common complication of COPD a major Prevention of acute exacerbations of COPD can be improved,
cause of increased utilization of health care resources such as emer- especially for those patients having a high burden of exacerbations
gency department visits and hospital admissions, and the most common despite being on optimal therapy [12]. Current therapies such as long-
cause of death [4–6]. Evidence suggests that recurrent exacerbations acting bronchodilators and/or inhaled corticosteroids, smoking cessa-
are associated with faster decline in lung function and a negative tion, pulmonary rehabilitation and influenza immunization are

∗
Corresponding author. Center for Innovative Medicine, McGill University Health Centre, 1001 Decarie Blvd., Room C047371.5, Montreal, QC H4A 3J1, Canada.
E-mail address: [email protected] (J. Bourbeau).

https://doi.org/10.1016/j.rmed.2018.03.035
Received 19 June 2017; Received in revised form 5 December 2017; Accepted 30 March 2018
Available online 05 April 2018
0954-6111/ © 2018 Elsevier Ltd. All rights reserved.
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

recommended to prevent exacerbations [11,13]. Even though these 2.2. Patient population
interventions have shown a reduction in exacerbations, there is a need
for further pharmacological therapies in patients who continue to have All patients who had an outpatient visit at the Montreal Chest
recurrent exacerbations despite optimal care. Recently, an interest in Institute (MCI) from July 2010 to December 2016 with a diagnosis of
using prophylactic antibiotics in COPD to prevent exacerbations has COPD had their chart reviewed. Patients of the non exposed and those
emerged, with an emphasis on the use of macrolide antibiotics. Mac- of the exposed groups were selected according to their exposure to
rolides have been used in several chronic respiratory conditions for azithromycin before any information was retrieved on the outcomes
their anti-inflammatory and immunomodulatory effects. Low-dose under study. The prescribed azithromycin group included patients with
macrolide therapy has been reported to decrease the number of ex- severe or very severe COPD and documented exacerbations, who were
acerbations in a large number of chronic inflammatory pulmonary prescribed regular azithromycin for prevention of exacerbations. The
diseases [14–17]. Macrolides interfere with biofilm development [18] severity of COPD was defined as per the GOLD spirometry criteria, i.e.,
and bacterial adherence of pathogens [19,20] such as Pseudomonas FEV1/FVC ratio ≤0.70 and FEV1 of < 50% predicted (GOLD 3 and 4)
aeruginosa. Administration of macrolides may be helpful in patients [12]. They had to have been prescribed a dose of 250 mg, at least 3
who have been colonized by Pseudomonas aeruginosa [20]. Macrolides times per week, for a minimum of 6 months. Short courses of macrolide
also have prokinetic activity on the gastrointestinal tract, leading to a antibiotics to treat acute exacerbations were disregarded. Patients could
reduction in gastroesophageal reflux and microaspiration, common in not receive azithromycin if they had prolonged QT on baseline elec-
chronic pulmonary conditions [18,20–22]. Macrolides have been trocardiogram (ECG), concurrent lung cancer, tuberculosis and
shown to reduce the number of exacerbations in patients with cystic bronchiectasis as the primary diagnosis. The non-prescribed azi-
fibrosis (CF), especially in those colonized with Pseudomonas aeruginosa thromycin group included patients with severe COPD (GOLD 3 or 4)
[23–26]. and documented recurrent exacerbations within one year who did not
In patients with COPD who have had at least one exacerbation, two receive azithromycin. Identifying non-prescribed patients as compar-
recent large trials have shown that regular macrolides reduce the time ison group was possible as some pulmonologists at the time were re-
to subsequent exacerbation and the frequency of exacerbations at one luctant to use macrolide regularly for prevention of exacerbations.
year [27,28]. Use of macrolides has been recommended in the Global
Initiative for Chronic Obstructive Lung Disease (GOLD 2017) as third- 2.3. Data collection and extraction of information on exacerbations
line therapy for patients with recurrent exacerbations (GOLD D) [12].
Treatment with azithromycin is also associated with an increased in- We used a standardized data collection form to record study in-
cidence of bacterial resistance (Evidence A) and hearing test impair- formation. Two physicians reviewed each inpatient and outpatient
ments (Evidence B) [12]. Macrolides will unlikely be used in every medical charts. Baseline clinical information was extracted, including
patient with COPD known for exacerbations considering those un- demographic data, baseline respiratory medications, comorbid condi-
desirable effects and the small increase in cardiovascular adverse events tions, pulmonary function tests, dyspnea (using the Modified Medical
[29]. Many questions remain unanswered in practice making clinical Research Council scale from 1 to 5), and lower respiratory tract mi-
decisions a challenge. It is still unknown if patients with severe disease crobial colonization (sputum culture results). The presence of any em-
and those colonized with Pseudomonas aeruginosa would benefit from physema or bronchiectasis on computed tomography (CT) scan of the
treatment with macrolides. In addition, benefits beyond one year of chest was noted, based on the official radiologist's report.
treatment have not been studied. Data were extracted one year prior to initiation of azithromycin for
The principal objective of this study was to determine whether long- the exposed patients, and until December 2016, or until the patient
term azithromycin is associated with a decrease in exacerbations in stopped taking azithromycin. In the non-exposed group, information
patients with severe COPD known for recurrent exacerbations while on was extracted for 2 consecutive years to allow comparison with the
optimal inhaled therapy. Secondary objectives were to evaluate: i) the exposed group. With respect to patients in the non-exposed group, we
effect of azithromycin on health service use, specifically emergency took advantage of a period of time when many respirologists were still
medical visits and hospitalizations; ii) the sustained effect on exacer- not prescribing azithromycin for COPD patients with frequent exacer-
bations beyond 1 year and; iii) the safety of azithromycin in real life bations. The follow up period of the patients in the non exposed group
clinical practice. had to be limited to 24 months; beyond 24 months there are patients
who could have being initiated on azithromycin. For this reason, the
period of comparison between patients of the non exposed and exposed
2. Material and methods group was limited to 2 years. In the exposed group, we were able to
follow patients beyond 24 months and to report on the long term
2.1. Study design benefit of azithromycin. The follow-up after taking azithromycin was
32 ± 22 (mean ± SD) months.
This retrospective observational study was performed using data The following information was extracted for each exacerbation:
from chart review of patients with COPD followed by pulmonary phy- severity of exacerbation, duration of antibiotic and/or corticosteroid
sicians working at a single specialized respiratory hospital, the course, emergency visit, hospitalization and duration of hospital ad-
Montreal Chest Institute (MCI), McGill University Health Centre, mission (if applicable). Exacerbations were defined as event-based,
Montreal, Quebec, Canada. Patients with severe COPD are followed in a meaning an increase in respiratory symptoms from baseline requiring
multidisciplinary clinical program that includes a pulmonologist and medication change (antibiotic and/or systemic corticosteroids) and
nurses acting as case managers. Patients are strongly encouraged to contact with the health care provider. These contacts may have been a
contact the team when they have an increase in respiratory symptoms, phone call to a COPD nurse case manager, a physician visit, an un-
and to seek care at the centre's walk-in clinic or emergency department. scheduled visit to the clinic or emergency department visit, or a hospital
Every visit and phone interventions are systematically recorded using admission. Exacerbations were considered moderate if patients re-
standardized forms that are included in the patients' medical records. quired a prescription of systemic corticosteroids, a course of antibiotics,
Information gathered includes symptoms, use of action plan, initiation or both. They were deemed as severe if patients required hospitaliza-
of antibiotics or corticosteroids, emergency visits, and hospital admis- tion.
sions. Ethics approval for this project was obtained from the local in- Adverse effects known to be associated with azithromycin were
stitutional review board. recorded, including QT prolongation on electrocardiogram, and gas-
trointestinal dysfunction. Hearing tests were not done routinely and this

130
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

information was not available from chart review. Repeat sputum cul- Table 1
tures following treatment with azithromycin were noted when avail- Clinical characteristics of patients on treatment (azithromycin group) and those
able. Death was identified from the chart and reviewed with the not on azithromycin (control group).
treating physician and the COPD case managers. Variable Azithromycin Control Group P-value*
Group
2.4. Statistical analysis
N = 126 N = 69

Descriptive data were reported using means and standard deviations Age, years 67.8 ± 9.3 70.8 ± 9.4 0.035
for continuous variables or counts and proportions for categorical
variables. Statistically significant differences for exacerbations and Male sex, n (%) 73 (59.8) 33 (47.8) 0.109
healthcare utilization between the year pre-treatment and the year
Smoking status, n (%)
post-treatment were then compared using GEE model with Poisson Ex-smokers 98 (79.0) 49 (71.0) 0.21
distribution for count data or binary distribution as appropriate. Current smokers 26 (21.0) 20 (29.0) 0.21
Time*group interaction term was added into the model to estimate Cigarette smoking pack- 48.3 ± 17.2 52.1 ± 19.6 0.265
whether the average change in the outcome from pre to post treatment years

differed between two groups. We also performed a logistic regression MRC dyspnea Scale, n (%)
model to estimate the P values for the three responders exacerbations, 2 0 (0.0) 2 (3.8) 0.155
emergency room (ER) visits, and hospitalizations between azithromycin 3 24 (29.6) 19 (35.8) 0.451
and control groups, adjusted for baseline age, sex, current smokers, and 4 or 5 57 (70.4) 32 (60.4) 0.231
FEV1%predicted. GEE model with repeated statement was also per-
Chronic bronchitis, n (%) 15 (41.7) 12 (48.0) 0.624
formed to compare number of exacerbations, ER visits, and hospitali-
zations among pre-treatment year and post-treatment years in azi- Exacerbation previous year 3.2 ± 2.1 1.7 ± 1.3 < 0.001
thromycin group who completed 2-year follow-up. Analysis were
performed using SAS version 9.4. Respiratory medication, n (%)
LABA/LAMA/ICS 124 (98.4) 69 (100.0) 0.293
Theophylline 13 (10.4) 9 (13.0) 0.578
3. Results Leukotriene receptor 10 (8.0) 2 (2.9) 0.219
antagonist
Clinical characteristics of patients on treatment (azithromycin Systemic corticosteroids 10 (8.0) 2 (2.9) 0.219
Long-term oxygen therapy 33 (26.6) 9 (13.0) 0.029
group) and those not on azithromycin (comparison group) are sum-
marized in Table 1. There were 126 patients in the treatment group and FEV1 (L) 0.9 ± 0.4 1.0 ± 0.4 0.042
69 in the comparison group. Patients prescribed azithromycin had more FEV1, % predicted 34.8 ± 14.1 39.9 ± 13.7 0.006
severe COPD with lower FEV1, higher dyspnea score (MRC 4-5/5) and FVC (L) 2.1 ± 0.8 2.2 ± 0.8 0.463
more frequent exacerbations at baseline. Patients from both groups FEV1/FVC (%) 42.9 ± 12.5 46.1 ± 11.9 0.062

were already on triple inhaled therapy but more patients were on long- Comorbid conditions, n (%)
term oxygen therapy in the azithromycin group. CT scan of the chest Asthma 15 (12.0) 4 (5.8) 0.211
showed that emphysema was the underlying primary lung condition in Sleep disordered 23 (18.4) 11 (15.9) 0.666
most patients, while bronchiectasis was present in one-third of the Osteoporosis 19 (15.2) 13 (18.8) 0.513
Coronary artery disease 20 (16.1) 10 (14.5) 0.764
patients from either group. The patients in the azithromycin group had
Atrial fibrillation 15 (12.0) 11 (15.9) 0.44
more frequent colonization of the respiratory tract with Pseudomonas Hypertension 46 (36.8) 30 (43.5) 0.362
aeruginosa compared to those in the comparison group. Diabetes 16 (12.8) 11 (15.9) 0.545
Table 2 shows within-group changes in acute exacerbations and Pulmonary hypertension 5 (4.0) 8 (11.6) 0.043
health service use in patients on treatment (azithromycin group) and
Chest CT scan, n (%) n = 106 n = 53
those not on azithromycin (comparison group), and between-group Bronchiectasis 29 (27.4) 15 (28.3) 0.9
differences in those outcomes. The mean number of exacerbations de- Emphysema 99 (93.4) 44 (83.0) 0.04
creased significantly in the year after initiation of azithromycin (3.2 vs
2.3 exacerbation per year, p < 0.001), whereas in the comparison Colonization, n (%) n = 111 n = 59
Pseudomonas aeruginosa 69 (62.2) 23 (39.0) 0.004
group, the mean number of exacerbations increased between year one
Stenotrophomonas maltophilia 14 (12.6) 5 (8.5) 0.415
and two (1.7 vs 2.7 exacerbations per year, p < 0.001). There was a Othersa 36 (32.4) 39 (66.1) < 0.001
significant within-group reduction in the number of emergency room Normal flora 25 (22.5) 11 (18.6) 0.556
visits, hospital admissions, and days in hospital in the treatment group,
Number of exacerbations in the 3.2 ± 2.1 1.7 ± 1.3 < 0.001
but not in the comparison group. In addition, patients on treatment had
previous 1-year n (%)
a significant reduction in exacerbations, ER visits, hospital admissions 1 exacerbation 16 (12.7) 19 (27.5) 0.009
and duration of hospital stay compared to the non-treated patients. 2 + exacerbations 104 (82.5) 36 (52.2) < 0.001
Fig. 1 shows a higher proportion of responders, i.e., having a reduction
in exacerbations, ER visits or hospital admissions, between azi- Values were mean ± SD or n (%).
thromycin compared to the comparison group. MRC: Medical Research Council; LABA: Long Acting Beta Agonist; LAMA: Long-
Table 3 shows changes in exacerbations in the azithromycin and the Acting Muscarinic Antagonist; ICS: Inhaled Corticosteroid; FEV1: Forced
Expiratory Volume in 1 s; FVC: Forced Vital Capacity; CT: Computerized
comparison groups according to patient and disease characteristics.
Tomography.
Although the reduction in exacerbations appeared to be more modest in
*P-value was obtained by performing T-test (normal distribution) or Wilcoxon
active smokers compared to former smokers, both benefited from two-sample test (non-normal distribution) or Chi-square analysis for category
treatment with azithromycin. The subgroups of patients with re- variables.
spiratory colonization with Pseudomonas aeruginosa, and those with and a
Others incude: Escherichia coli, Enterobacter, Acinetobacter, Staphylococcus
without bronchiectasis on CT scan also had a significant reduction in aureus, Streptococcus pneumoniae, Moraxella catarrhalis, Serratia marcescens,
the number of exacerbations after one year of treatment with azi- Candida.
thromycin.
Figs. 2 and 3 show the number of exacerbations and proportion of

131
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

Table 2
Changes in exacerbations and health service use within treatment and comparison groups and between groups.
Azithromycin Group Non Azithromycin Group Between group
comparison
(N = 126) (N = 69)

0–12 months 12–24 months P-Value‡ 0 -12 months 12-24 months P-Value‡ P-Value§

PreRxa Post Rxb

Number of exacerbations, mean ± SD 3.2 ± 2.1 2.3 ± 1.6 < 0.001 1.7 ± 1.3 2.5 ± 1.7 < 0.001 < 0.001
Patients with exacerbations, n (%)
1 exacerbation 16 (12.7) 37 (29.4) 0.001 19 (27.5) 13 (18.8) 0.226 0.003
2 + exacerbations 104(82.5) 78 (61.9) < 0.001 36 (52.2) 50 (72.5) 0.014 < 0.001
Health care utilization
ER visit/patient/year, mean ± SD 1.5 ± 1.7 1.0 ± 1.2 0.007 1.1 ± 1.1 1.3 ± 1.5 0.926 0.031
Hospitalization/patient/year, mean ± SD 1.4 ± 1.7 0.9 ± 1.1 0.003 0.8 ± 1.0 1.0 ± 1.5 0.763 0.014
Hospital stay for the patients who have hospitalizations, mean ± SD 9.1 ± 10.0 6.5 ± 4.7 0.039 8.7 ± 8.9 10.4 ± 9.8 0.274 0.005

‡P-value was obtained by performing GEE model within the group.

§P-value was obtained by performing GEE model between the azithromycin and control group. For the two groups comparison (between group comparison), we used
GEE model with repeated statement to test time*group interaction term and estimate whether the average change in the outcome from pre to post differed in the two
groups.
a
Pre Rx refers to the year before the patient has been prescribed Azithromycin.
b
Post Rx the year while the patient has been taken Azithromycin.

patients having ≥2 exacerbations, and ER and hospitalizations among 4. Discussion

patients on azithromycin treatment who completed 2-year follow-up.
There was a significant reduction of exacerbations, ER visits and hos- This study of real life practice has shown that patients with severe
pitalizations beyond one year of follow-up. and very severe COPD and frequent exacerbations on optimal inhaled
There were few documented adverse effects attributable to azi- pharmacotherapy benefit from the addition of regular azithromycin to
thromycin, and the medication rarely had to be discontinued. Seven reduce the number of acute exacerbations. This is in contrast to a
patients with COPD (5.6%) discontinued azithromycin, of whom only 2 control group of patients who did not receive long-term azithromycin
discontinued due to QT prolongation (1.6%), and 3 because of gastro- who experienced increased yearly exacerbations during the same time
intestinal symptoms such as diarrhea and/or tenesmus (2.4%). Only 2 period. The probability of being a “responder”, i.e., having a reduction
patients (1.6%) discontinued azithromycin because their physician did in exacerbations and health service use in patients in the azithromycin
not consider the treatment beneficial. Post-treatment sputum culture group, was significantly higher compared to those in the control group.
was collected primarily in patients with Pseudomonas aeruginosa re- The benefit of azithromycin treatment was demonstrated in smokers
spiratory colonization; 60% of the patients remained colonized after 2 and former smokers, in patients colonized with Pseudomonas aeruginosa,
years or more being on azithromycin. The rate of death respiratory and in those with and without bronchiectasis. We have also shown that
cause during the 2 or more years of follow-up was 1.5% (2 patients) and the benefits in reducing exacerbations, emergency visits, and hospital
1.4% (1 patient) in the azithromycin and control groups, respectively. admissions were prolonged beyond one year of treatment. Finally, very

Fig. 1. The proportion of responders between azithromycin

and non-prescribed azithromycin groups. The definition for
responder is patients with the number of exacerbations
during 12–24 months PostRx (post-treatment) being less
than 0–12 months PreRx (pre-treatment). Compared to non-
azithromycin group, responders were all significantly higher
for azithromycin group using logistics model adjusted for
baseline age, sex, current smokers, and FEV1%predicted. ER:
emergency room.

132
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

Table 3
Changes in exacerbations in the azithromycin and the control groups according to patient and disease characteristics.
Patient and disease characteristics Number of exacerbation

Azithromycin Group Non Azithromycin Group Two-group comparison P-Value*

N mean ± SD P-Value* N mean ± SD P-Value*

Smoking status
Ex-smokers
0–12 months PreRxa 98 3.2 ± 2.2 < 0.001 49 1.6 ± 1.4 0.006 < 0.001
12–24 months PostRxb 98 2.2 ± 1.7 49 2.5 ± 1.6
Current smokers
0–12 months PreRx 26 3.3 ± 1.7 0.022 20 1.8 ± 1.1 0.082 0.010
12–24 months PostRx 26 2.6 ± 1.4 20 2.6 ± 1.9
Sputum culture
Pseudomonas aeruginosa
0–12 months PreRx 60 2.7 ± 1.7 0.236 21 1.5 ± 1.2 0.006 0.002
12–24 months PostRx 60 2.3 ± 1.7 21 2.9 ± 2.3
Stenotrophomonas
0–12 months PreRx 5 6.2 ± 3.2 0.018 3 2.0 ± 2.0 0.346 0.039
12–24 months PostRx 5 1.4 ± 1.7 3 4.0 ± 2.6
Pseudomonas or Stenotrophomonas
0–12 months PreRx 74 3.0 ± 2.1 0.035 26 1.6 ± 1.4 0.011 0.001
12–24 months PostRx 74 2.3 ± 1.7 26 3.0 ± 2.3
Any othersc(not Pseudomonas or Stenotrophomonas)
0–12 months PreRx 20 3.3 ± 2.4 0.054 22 1.7 ± 1.1 0.255 0.032
12–24 months PostRx 20 2.3 ± 1.1 22 2.1 ± 1.3
CT scan
Bronchiectasis
0–12 months PreRx 29 3.3 ± 2.2 0.036 15 2.1 ± 1.5 0.085 0.007
12–24 months PostRx 29 2.4 ± 1.6 15 3.2 ± 2.2
No bronchiectasis
0–12 months PreRx 100 3.3 ± 2.1 < 0.001 50 1.8 ± 1.3 0.008 < 0.001
12–24 months PostRx 100 2.3 ± 1.6 50 2.7 ± 1.7

*P-value was obtained by performing GEE model.

For the two groups comparison, we used GEE model with repeated statement to test time*group interaction term and estimate.
Whether the average change in the outcome from pre to post differed in the two groups.
a
PreRx refes to the year before the patient has been prescribed azithromycin.
b
Post Rx reers to the year while the patient has been taken azithromycin.
c
Others incude: Escherichia coli, Enterobacter, Acinetobacter, Staphylococcus aureus, Streptococcus pneumoniae, Moraxella catarrhalis, Serratia marcescens, Candida.

few patients had to discontinue azithromycin treatment because of particular, an increased incidence of microbial resistance.
adverse effects. Few studies reflecting real life clinical practice have been published
Seemungal and colleagues [28] performed a randomized double- where macrolide was initiated in patients with COPD and recurrent
blind control trial evaluating erythromycin for a year versus placebo in exacerbations. One Japanese study [31] performed a retrospective
109 patients with COPD. Subjects receiving erythromycin had sig- analysis of 123 patients with COPD in seven academic hospitals, 45
nificantly fewer exacerbations (81 per year) compared to the subjects were treated with macrolides (clarithromycin or erythromycin), com-
receiving placebo (125 per year) [28]. However, patients having pared to a control group of 78 patients [31]. There were no difference
moderate to severe COPD (mean FEV1 50% predicted) were not re- in mean frequency of yearly exacerbations between the two groups
ceiving what would be considered optimal medical therapy these days, (0.51 exacerbation per year in the treatment group versus 0.63 in the
the majority being on inhaled corticosteroids, 60% on a long-acting control group, p = 0.54). However, there were less patients with > 1.5
beta2-agonist, and less than 40% on the long-acting muscarinic an- exacerbations per year receiving macrolides compared to control pa-
tagonist. More recently, Albert and colleagues [27] performed a mul- tients (4.4% versus 15.4%. p = 0.01). Again in this study, most patients
tisite randomized control trial evaluating on year of azithromycin were mild to moderate (GOLD 1 or 2, 76% in the macrolide group and
(250 mg daily) versus placebo in patients with COPD. This study 59% in the controls). This group with fewer acute exacerbations are
showed the number of exacerbations requiring both antibiotics and more likely to benefit from inhaled medication, and unlikely to require
steroids was 0.76 times fewer in the azithromycin group than the pla- additional treatment with azithromycin.
cebo group (95% CI, 0.63–0.91; P = 0.002). Han and colleagues [30] In our study, a predominant feature of the azithromycin group was
reported the results of a secondary analysis of this cohort, taking into the colonization of the respiratory tract with Pseudomonas aeruginosa,
consideration potentially relevant confounders such as the use of var- representing two thirds of the patients. They had a significant reduction
ious concomitant therapies, clinical characteristics, disease severity, in number of exacerbations on azithromycin treatment. This is in
and the type of COPD exacerbations. They showed that azithromycin keeping with the BLESS study [32], in which patients with non-cystic
had greater efficacy in older patients, lower GOLD stage and those not fibrosis bronchiectasis colonized with Pseudomonas aeruginosa had a
actively smoking. This trial was not designed to include patients with significant reduction in total exacerbations. In these patients, it has
severe or very severe COPD and recurrent exacerbations despite op- been demonstrated that long-term macrolides change the composition
timal inhaled therapy. The patients in our study had more severe dis- of the respiratory microbiota. In patients without Pseudomonas aerugi-
ease, higher recurrent rate of exacerbations, and more were colonized nosa airway infection, macrolides did not significantly reduce exacer-
with Pseudomonas aeruginosa. Many clinicians may be reluctant to use bations and promoted displacement of Haemophilus influenza by more
long-term macrolides in mild to moderate COPD considering that macrolide-tolerant pathogens including Pseudomonas aeruginosa. These
benefits do not clearly outweigh the risks and adverse events, in findings argue for a cautious approach to chronic macrolide use in

133
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

Fig. 3. The number of ER visits (3a) and hospital admissions (3b) among pa-
tients of the azithromycin group who completed 2-year follow-up (n = 68). GEE
model with poisson distribution and repeated statement was performed to
compare ER visits/hospitalizations among the three-time points. Figure3a,
compared to PreRx (pre-treatment) 1 year, there are significant differences for
Fig. 2. The number of ≥ 2 exacerbations (2a) and proportion of ≥ 2 exacer- PostRx (post-treatment) 1 year and PostRx 2-year, with a P value 0.030
bations (2b) among patients of the azithromycin group who completed 2-year and < 0.001, respectively, for ER visit. Also, there are significant differences
follow-up (n = 68). Fig. 2a, GEE model with Poisson distribution and repeated between PostRx 1-year and PostRx 2-year, with a p = 0.019 for ER visit. Figure
statement was performed to compare the number of ≥ 2 exacerbations among 3b, compared to PreRx (pre-treatment) 1 year, there are significant differences
the three-time points; compared to PreRx (pre-treatment) 1 year, there are for PostRx (post-treatment) 1 year and PostRx 2-year and P value 0.014
significant differences for PostRx (post-treatment) 1 year and PostRx 2-year, and < 0.001, respectively and 0.022 for comparison between postRx 1 year
with a P value 0.012 and < 0.001, respectively. Also, there are significant and postRx 2 year for hospitalization. ER; emergency room.
differences between PostRx 1-year and PostRx 2-year, with a p = 0.009. Fig. 2b,
GEE model with binary distribution and repeated statement was performed to
compare the proportion of ≥ 2 exacerbation among the three-time points; of the study design, we had access to both exposure and outcomes at the
compared to PreRx 1 year, there are significant differences for PostRx 1 year time of doing the chart review. Although reviewers of the medical chart
and PostRx 2-year, with a P value 0.028 and < 0.001, respectively. NS: Non- were instructed to select patients based on their exposure, they were
significant. not blind to the patient outcomes. Patients in the comparison group had
less severe disease and were therefore different than the treatment
patients without Pseudomonas aeruginosa airway colonization [33]. It group. However, following a comparison group of non-exposed patients
has also been demonstrated that inhibition of Pseudomonas aeruginosa with severe COPD for the same period of time and tracking their ex-
quorum sensing genes within the airways of patients with bronch- acerbations allowed us to study the natural clinical evolution of these
iectasis receiving long-term low-dose macrolides, without a reduction patients compared to the treatment group. Furthermore, the differences
in bacterial load, represents a potential mechanism of therapeutic im- of more frequent exacerbations in the exposed compared to the non-
pact beyond a classical antimicrobial or anti-inflammatory pathway exposed group with less exacerbations would likely bias our results
[32]. toward the null hypothesis. The fact that we found a significant be-
tween-group difference in exacerbations and health care utilization
suggests that azithromycin is truly beneficial. Our study was not de-
4.1. Strengths and limitations signed or powered to accurately assess adverse events and the risk of
long-term use of azithromycin in a COPD population. Adverse effects
One of the strengths of the present study was that therapy with were rarely a cause of discontinuing the macrolide treatment. A careful
azithromycin was initiated in a well-defined patient population with review of cardiac history and a baseline electrocardiogram was done
severe or very severe COPD and recurrent exacerbations despite op- before initiation of therapy and patient with long QT were not con-
timal medical therapy (combined long-acting bronchodilators and in- sidered for treatment with azithromycin, which likely helps reduce
haled corticosteroids). Another strength of this study was the length of potential harm. Ototoxicity was not clinically reported and it was not
treatment with azithromycin therapy beyond 1 year, as most studies systematically assessed in the study.
previously published stop intervention at one year. Finally, we have
shown that in patients colonized with Pseudomonas aeruginosa, the po-
tential benefits of azithromycin outweighed the risks and adverse 4.2. Conclusions
events, most particularly the increased incidence of microbial re-
sistance. In conclusion, our study showed that regular use of azithromycin in
This study has potential limitations. Due to the retrospective nature patients with severe and very severe COPD and recurrent acute

134
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

exacerbations despite optimal therapy (triple inhaled therapy) has the W.J. Fulkerson, P. Kussin, P. Bellamy, L. Goldman, W.A. Knaus, Outcomes following
potential to reduce the risk of exacerbations, emergency visits and acute exacerbation of severe chronic obstructive lung disease. The SUPPORT in-
vestigators (Study to Understand Prognoses and Preferences for Outcomes and Risks
hospital admissions, and that these benefits are sustained beyond one of Treatments), Am. J. Respir. Crit. Care Med. 154 (1996) 959–967.
year. Benefits are more likely to outweigh the risks and adverse events [7] G.C. Donaldson, T.A. Seemungal, A. Bhowmik, J.A. Wedzicha, Relationship be-
in patients with COPD who are colonized with Pseudomonas aeruginosa. tween exacerbation frequency and lung function decline in chronic obstructive
pulmonary disease, Thorax 57 (2002) 847–852.
However, it is necessary to carefully select patients since long-term use [8] R.E. Kanner, N.R. Anthonisen, J.E. Connett, Lung Health Study Research G. Lower
of macrolides can have deleterious effects. Further studies are needed to respiratory illnesses promote FEV(1) decline in current smokers but not ex-smokers
determine if intermittent therapy is just as effective as daily azi- with mild chronic obstructive pulmonary disease: results from the lung health
study, Am. J. Respir. Crit. Care Med. 164 (2001) 358–364.
thromycin in preventing COPD exacerbations. Our study, in a real-life [9] M. Miravitlles, M. Ferrer, A. Pont, R. Zalacain, J.L. Alvarez-Sala, F. Masa, H. Verea,
practice showed that regular use of azithromycin in patients with severe C. Murio, F. Ros, R. Vidal, I.S. Group, Effect of exacerbations on quality of life in
and very severe COPD and recurrent acute exacerbations despite op- patients with chronic obstructive pulmonary disease: a 2 year follow up study,
Thorax 59 (2004) 387–395.
timal therapy (triple inhaled therapy) has the potential to reduce the
[10] J.R. Hurst, J. Vestbo, A. Anzueto, N. Locantore, H. Mullerova, R. Tal-Singer,
risk of exacerbations, emergency visits and hospital admissions, and B. Miller, D.A. Lomas, A. Agusti, W. Macnee, P. Calverley, S. Rennard, E.F. Wouters,
these benefits are sustained beyond one year. Benefits have been shown J.A. Wedzicha, Evaluation of CLtIPSEI. Susceptibility to exacerbation in chronic
in sub groups of patients, smokers and ex-smokers, those with and obstructive pulmonary disease, N. Engl. J. Med. 363 (2010) 1128–1138.
[11] G.D. James, I. Petersen, I. Nazareth, J.A. Wedzicha, G.C. Donaldson, Use of long-
without bronchiectasis and patients colonized with Pseudomonas term antibiotic treatment in COPD patients in the UK: a retrospective cohort study,
Aeruginosa. Prim. Care Respir. J. : J. Gen. Pract. Airways Group 22 (2013) 271–277.
The present study originality was that therapy with azithromycin [12] From the Global Strategy for the Diagnosis, Management and Prevention of COPD,
Global Initiative for Chronic Obstructive Lung Disease (GOLD), (2017) Available
was initiated in a well-defined patient population with severe or very from: http://goldcopd.org.
severe COPD and recurrent exacerbations despite optimal medical [13] J.A. Wedzicha, T.A. Seemungal, COPD exacerbations: defining their cause and
therapy (combined long-acting bronchodilators and inhaled corticos- prevention, Lancet 370 (2007) 786–796.
[14] F.F. Berkhof, N.E. Doornewaard-ten Hertog, S.M. Uil, H.A. Kerstjens, J.W. van den
teroids). Furthermore, the length of treatment with azithromycin Berg, Azithromycin and cough-specific health status in patients with chronic ob-
therapy showed benefit beyond 1 year; most studies previously pub- structive pulmonary disease and chronic cough: a randomised controlled trial,
lished were at one year. Finally, we have shown that patients colonized Respir. Res. 14 (2013) 125.
[15] P.L. Paggiaro, P. Chanez, O. Holz, P.W. Ind, R. Djukanovic, P. Maestrelli, P.J. Sterk,
with Pseudomonas aeruginosa would be a population to target con- Sputum induction, Eur. Respir. J. Suppl. 37 (2002) 3s–8s.
sidering that the potential benefits of azithromycin outweights the risks [16] D.J. Serisier, M.L. Martin, M.A. McGuckin, R. Lourie, A.C. Chen, B. Brain, S. Biga,
and adverse events, most particularly the increased incidence of mi- S. Schlebusch, P. Dash, S.D. Bowler, Effect of long-term, low-dose erythromycin on
pulmonary exacerbations among patients with non-cystic fibrosis bronchiectasis:
crobial resistance.
the BLESS randomized controlled trial, Jama 309 (2013) 1260–1267.
[17] E. Shafuddin, G.D. Mills, M.D. Holmes, P.J. Poole, P.R. Mullins, P.N. Black, A
Author contributions double-blind, randomised, placebo-controlled study of roxithromycin and doxycy-
cline combination, roxithromycin alone, or matching placebo for 12 weeks in adults
with frequent exacerbations of chronic obstructive pulmonary disease, J. Negat.
NN, DA, SMYMPM and ZK have done the chart review and data Results Biomed. 14 (2015) 15.
extraction. NN, DA and SMYMPM have reviewed the study protocol, [18] K. Suresh Babu, J. Kastelik, J.B. Morjaria, Role of long term antibiotics in chronic
participated in the analysis, and participated in the redaction and re- respiratory diseases, Respir. Med. 107 (2013) 800–815.
[19] D.G. Peckham, Macrolide antibiotics and cystic fibrosis, Thorax 57 (2002) 189–190.
viewed the manuscript. PZL has done the analysis and reviewed the [20] K. Terada, S. Muro, S. Sato, T. Ohara, A. Haruna, S. Marumo, D. Kinose, E. Ogawa,
manuscript. AJ, IO and ID have assisted in the chart review and re- Y. Hoshino, A. Niimi, T. Terada, M. Mishima, Impact of gastro-oesophageal reflux
viewed the manuscript. JB has written the protocol, planned and par- disease symptoms on COPD exacerbation, Thorax 63 (2008) 951–955.
[21] K. Blondeau, L.J. Dupont, V. Mertens, G. Verleden, A. Malfroot, Y. Vandenplas,
ticipated in the analysis, written and reviewed the manuscript. B. Hauser, D. Sifrim, Gastro-oesophageal reflux and aspiration of gastric contents in
No source of support in the form of grants, honorariums or gifts. adult patients with cystic fibrosis, Gut 57 (2008) 1049–1055.
The views expressed in this article do not communicate an official [22] V. Mertens, K. Blondeau, A. Pauwels, R. Farre, B. Vanaudenaerde, R. Vos,
G. Verleden, D.E. Van Raemdonck, L.J. Dupont, D. Sifrim, Azithromycin reduces
position of the McGill University Health Centre or McGill University.
gastroesophageal reflux and aspiration in lung transplant recipients, Dig. Dis. Sci.
54 (2009) 972–979.
Acknowledgments [23] A. Equi, I.M. Balfour-Lynn, A. Bush, M. Rosenthal, Long term azithromycin in
children with cystic fibrosis: a randomised, placebo-controlled crossover trial,
Lancet 360 (2002) 978–984.
The authors wish to thank the patients who kindly participate in the [24] L. Saiman, B.C. Marshall, N. Mayer-Hamblett, J.L. Burns, A.L. Quittner, D.A. Cibene,
study, and to acknowledge the help of all staff and students involved in S. Coquillette, A.Y. Fieberg, F.J. Accurso, P.W. Campbell 3rd, G. Macrolide Study,
the study, especially Rita Abi Maroun Anthony Karzon, and Louise Azithromycin in patients with cystic fibrosis chronically infected with Pseudomonas
aeruginosa: a randomized controlled trial, Jama 290 (2003) 1749–1756.
Auclair. [25] M.J. Schultz, Macrolide activities beyond their antimicrobial effects: macrolides in
diffuse panbronchiolitis and cystic fibrosis, J. Antimicrob. Chemother. 54 (2004)
References 21–28.
[26] J. Wolter, S. Seeney, S. Bell, S. Bowler, P. Masel, J. McCormack, Effect of long term
treatment with azithromycin on disease parameters in cystic fibrosis: a randomised
[1] C.D. Mathers, D. Loncar, Projections of global mortality and burden of disease from trial, Thorax 57 (2002) 212–216.
2002 to 2030, PLoS Med. 3 (2006) e442. [27] R.K. Albert, J. Connett, W.C. Bailey, R. Casaburi, J.A. Cooper Jr., G.J. Criner,
[2] S. Simoens, G. Laekeman, M. Decramer, Preventing COPD exacerbations with J.L. Curtis, M.T. Dransfield, M.K. Han, S.C. Lazarus, B. Make, N. Marchetti,
macrolides: a review and budget impact analysis, Respir. Med. 107 (2013) F.J. Martinez, N.E. Madinger, C. McEvoy, D.E. Niewoehner, J. Porsasz, C.S. Price,
637–648. J. Reilly, P.D. Scanlon, F.C. Sciurba, S.M. Scharf, G.R. Washko, P.G. Woodruff,
[3] A. Jemal, E. Ward, Y. Hao, M. Thun, Trends in the leading causes of death in the N.R. Anthonisen, C.C.R. Network, Azithromycin for prevention of exacerbations of
United States, 1970-2002, Jama 294 (2005) 1255–1259. COPD, N. Engl. J. Med. 365 (2011) 689–698.
[4] J. Bourbeau, M. Julien, F. Maltais, M. Rouleau, A. Beaupre, R. Begin, P. Renzi, [28] T.A. Seemungal, T.M. Wilkinson, J.R. Hurst, W.R. Perera, R.J. Sapsford,
D. Nault, E. Borycki, K. Schwartzman, R. Singh, J.P. Collet, Chronic Obstructive J.A. Wedzicha, Long-term erythromycin therapy is associated with decreased
Pulmonary Disease axis of the Respiratory Network Fonds de la Recherche en Sante chronic obstructive pulmonary disease exacerbations, Am. J. Respir. Crit. Care Med.
du Q. Reduction of hospital utilization in patients with chronic obstructive pul- 178 (2008) 1139–1147.
monary disease: a disease-specific self-management intervention, Arch. Intern. Med. [29] W.A. Ray, K.T. Murray, K. Hall, P.G. Arbogast, C.M. Stein, Azithromycin and the
163 (2003) 585–591. risk of cardiovascular death, N. Engl. J. Med. 366 (2012) 1881–1890.
[5] J.P. Collet, P. Shapiro, P. Ernst, T. Renzi, T. Ducruet, A. Robinson, Effects of an [30] M.K. Han, N. Tayob, S. Murray, M.T. Dransfield, G. Washko, P.D. Scanlon,
immunostimulating agent on acute exacerbations and hospitalizations in patients G.J. Criner, R. Casaburi, J. Connett, S.C. Lazarus, R. Albert, P. Woodruff,
with chronic obstructive pulmonary disease. The PARI-IS Study Steering Committee F.J. Martinez, Predictors of chronic obstructive pulmonary disease exacerbation
and Research Group. Prevention of Acute Respiratory Infection by an reduction in response to daily azithromycin therapy, Am. J. Respir. Crit. Care Med.
Immunostimulant, Am. J. Respir. Crit. Care Med. 156 (1997) 1719–1724. 189 (2014) 1503–1508.
[6] A.F. Connors Jr., N.V. Dawson, C. Thomas, F.E. Harrell Jr., N. Desbiens, [31] M. Yamaya, A. Azuma, H. Tanaka, H. Takizawa, K. Chida, Y. Taguchi, K. Mikasa,

135
N. Naderi et al. Respiratory Medicine 138 (2018) 129–136

J. Kadota, S. Kudoh, Inhibitory effects of macrolide antibiotics on exacerbations and the bronchiectasis and low-dose erythromycin study trial, Ann. Am. Thorac. Soc. 13
hospitalization in chronic obstructive pulmonary disease in Japan: a retrospective (2016) 1697–1703.
multicenter analysis, J. Am. Geriatr. Soc. 56 (2008) 1358–1360. [33] G.B. Rogers, K.D. Bruce, M.L. Martin, L.D. Burr, D.J. Serisier, The effect of long-term
[32] L.D. Burr, G.B. Rogers, A.C. Chen, B.R. Hamilton, G.F. Pool, S.L. Taylor, D. Venter, macrolide treatment on respiratory microbiota composition in non-cystic fibrosis
S.D. Bowler, S. Biga, M.A. McGuckin, Macrolide treatment inhibits Pseudomonas bronchiectasis: an analysis from the randomised, double-blind, placebo-controlled
aeruginosa quorum sensing in non-cystic fibrosis bronchiectasis. An analysis from BLESS trial, Lancet Respir. Med. 2 (2014) 988–996.

136