Asthma

Bronchiectasis

Chronic obstructive pulmonary disease

Bronchiolitis obliterans (including post transplantation)

Use of macrolides in other respiratory conditions

Safety issues

Cardiac effects

Ototoxicity

Other side effects

Antimicrobial resistance

Establishment of guideline development group

The GDG was convened in June 2016, with the first meeting taking place in October 2016. The full GDG met three times during the development of the guideline and kept in close contact by teleconference/WebEx and email throughout the process.

Methodology

This is the first BTS guideline to use the GRADE approach as part of the process of guideline development. Previous guidelines have used the SIGN methodology. BTS has made this change reflecting common practice in guideline development internationally across all medical specialities. The advantages of the GRADE approach are described in detail in the GRADE handbook and the BTS GRADE guideline production manual. GRADE specifies an approach to framing questions, choosing outcomes of interest and rating their importance, evaluating the evidence, and incorporating evidence with considerations of values and preferences of patients and society to arrive at recommendations.1 The methodology used to write the guideline adheres strictly to the criteria as set by the AGREE II collaboration, which is available online (www.agreetrust.org/resource-centre/agree-ii/). The British Thoracic Society Standards of Care Committee guideline production manual is available online (https://www.brit-thoracic.org.uk/quality-improvement/guidelines/).

Summary of key questions, outcomes and literature search

Clinical questions were formulated in the PICO (Patient, Intervention, Comparison and Outcome) format (see table 1). The key questions are summarised below.

1. Should long-term, low-dose macrolides be used in the treatment of adults with asthma?

2. Should long-term, low-dose macrolides be used in the treatment of adults with bronchiectasis?

3. Should long-term, low-dose macrolides be used in the treatment of adults with chronic obstructive pulmonary disease?

4. Should long-term, low-dose macrolides be used in the treatment of adults with bronchiolitis obliterans?

5. Should long-term, low-dose macrolides be used in the treatment of adults with chronic cough?

6. Should long-term, low-dose macrolides be used in the treatment of adults with organising pneumonia?

7. Should long-term, low-dose macrolides be used in the treatment of adults with diffuse panbronchiolitis?

The following patient-centred outcomes were identified by the group when the scope of the guideline was agreed:

• Quality-of-life measures.

• Symptom improvement/symptom score.

• Exacerbation rates.

• Hospital admission rate.

• Disease progression and changes in lung function tests.

• Mortality.

• Exercise capacity/tolerance.

• Sputum volume/colour/character and microbiological resistance/dysbiosis.

• Drug monitoring/side effects/toxicity.

The PICO framework was used to define the scope of the guideline and formed the basis of the literature search. The initial search was completed in February 2017 by York University. Systematic electronic database searches were conducted in order to identify all papers which might potentially be included in the guideline. For each question, the following databases were searched: Cochrane Database of Systematic Reviews (CDSR), Database of Abstracts of Reviews of Effects (DARE), Health Technology Assessment Database (HTA), Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and MEDLINE In-Process, EMBASE and PubMed. The search strategy is available for review in online supplementary appendix 1.

Appraisal of the literature

The literature search was run in 2016. The search was limited to papers published in English. The search identified a total of 5225 abstracts; after initial screening, this was reduced to 1152.

Criteria formulated for initial screening of the abstracts were

• Whether the study addressed the clinical question.

• Whether the appropriate study type was used to produce the best evidence to answer the clinical question.

• Abstract was in English.

• Studies in which macrolide therapy was used to treat acute respiratory tract infection were rejected.

• Abstracts were not rejected on the basis of the journal of publication, or the country in which the research was performed.

The full list of abstracts was retained and is kept in an archive. The literature search was repeated in December 2017 to capture additional published evidence while the guideline was in development prior to finalising the draft document. Additional references were included from critical appraisal of the literature and review of existing evidence base as appropriate.

Letters, conference papers and news articles were then removed—518 abstracts were rejected at this stage. IDR and DS read the remaining abstracts, marked those considered relevant to the scope of the guideline and allocated each relevant abstract to a clinical question(s). In total, 634 abstracts were allocated to clinical question(s).

The literature search was updated in late 2017 and identified 138 additional abstracts that were reviewed and 14 papers were selected to be critically appraised.

GDG members were allocated to work on individual questions in small groups.

Each abstract was read and at least two members agreed whether the paper was relevant, possibly relevant or irrelevant to the particular clinical question. Papers were excluded at the title/abstract sift if the following applied:

• If the paper did not address the clinical question and at least one of the outcomes concerned.

• If it was a case series of less than 20 patients—however, this was not an absolute cut-off. Professional judgement was applied such that some smaller case series were considered if evidence was sparse, and indeed some case reports of more than 20 patients were excluded if there was higher quality evidence available.

• If the language of the full paper was not English.

Full papers were obtained for all relevant, or possibly relevant, abstracts. Each paper was read in full and critically appraised by at least two members of the GDG to confirm relevance to the clinical question and the presence of at least one outcome of interest.

Each paper was then appraised by outcome(s), to generate a best estimate of the effect on each outcome and an index of the uncertainty associated with that estimate where possible. An evidence profile entry was completed for each outcome which included grading of the quality of the evidence. The type of evidence available for each outcome varied from systematic reviews through to case series; for each outcome, the highest quality evidence available was included. GRADEPro was used to generate the evidence profiles, published online on the BTS website where they are available for review (see online supplementary appendix 2). GRADEPro (https://gradepro.org/) is an easy-to-use, all-in-one web-based electronic guideline development tool to support guideline development using the GRADE process and methodology. In addition, it provides an electronic platform to present evidence tables, considered judgement and assessment of the evidence base.

The GRADE approach to rating the quality of evidence begins with the study design (table 2) and then, through a process of considered judgement, applies five reasons to possibly rate down the quality of evidence and three reasons to possibly rate up the quality (table 3).

The GDG reviewed each clinical question during the regular meetings and consensus was reached.

In assessing the evidence the guideline development group combined low and very low evidence into one category (Low) as the body of evidence was limited.

Development of recommendations

Having generated evidence profiles for each of the clinical questions the GDG as a whole then considered the importance of each of the outcomes for each clinical question and proceeded to grade the overall body of evidence for critical and important outcomes.

The GDG went on to decide on the direction and strength of recommendations considering the quality of the evidence, the balance of desirable and undesirable outcomes and the values and preferences of patients and others. GRADE specifies two categories of strength of a recommendation as shown in table 4.

Good practice points (GPPs) were developed by consensus in areas where there was no quality evidence but the GDG felt that some guidance based on the clinical experience of the GDG might be helpful to the reader. These are indicated as shown below.

From the outset, it was acknowledged that there would be little high-quality evidence for some of the clinical questions identified. In this instance, low-grade evidence was considered, along with expert opinion via consensus at the meetings.

Cost-effectiveness was not considered in detail as in-depth economic analysis of recommendations falls outside of the scope of the BTS guideline production process. However, the GDG were asked to be mindful of any potential economic barriers to the implementation of recommendations and GPPs.

Research recommendations were also highlighted and are listed in PICO format in an online supplementary appendix 3.

Drafting the guideline

The GDG corresponded regularly by email and meetings of the full group were also held in the period between July 2016 and early 2019. The guideline was discussed at an open session at the BTS Winter Meeting in December 2018. A revised draft guideline document was circulated to all the relevant stakeholders for consultation in December 2018 followed by a period of online consultation. The BTS SOCC reviewed the draft guideline in March 2018 and March 2019.

Review of the guideline

This guideline topic was chosen to inform practice in a growing clinical area, but also to act as a pilot for the use of GRADE methodology for all future BTS guidelines. The GDG recognise that the topic area would include a range of diseases with differing levels of evidence and thus act as a useful learning exercise for the introduction of GRADE methodology. It is not proposed to update this guideline as a distinct entity—the intention is for macrolide use to be encompassed in future disease-specific guideline updates.

Declarations of interests

BTS Declarations of Interest forms have been completed by all members for each year they were part of the GDG. Details of these forms can be obtained from BTS Head Office. “Declarations of Interests” was a standing item at each GDG meeting.

Stakeholders

Stakeholders were identified at the start of the process. All stakeholder organisations were notified when the guideline was available for public consultation and a list is published in online supplementary appendix 2.

Other available guidance

The GDG were aware of parallel workstreams within the BTS (eg, the BTS guideline for bronchiectasis in adults2) and in other organisations (eg, NICE) either complete or in progress at the time this guideline began development. Some of those in progress have subsequently been published and overlap with this guideline. The majority of these guidelines are disease specific rather than focusing on a single drug group. The pool of evidence available for these guidelines is the same, but the sample drawn from that pool is often slightly different. In addition, the methodology applied and the resources available to individual organisations are also variable. It is thus not surprising that the recommendations sometimes differ between contemporaneous documents. We have included a list of recently published guidelines and Cochrane reviews which overlap with this document.

BTS Guideline for Bronchiectasis in Adults 20192

NICE CF diagnosis and management Oct 20173

NICE Asthma diagnosis, monitoring and management Nov 20174

NICE Chronic obstructive pulmonary disease (acute exacerbation): antimicrobial prescribing Dec 20185

NICE Bronchiectasis (acute exacerbation): antimicrobial prescribing Dec 20186

NICE COPD guideline 20187

GOLD guideline COPD 20198

ERS/ATS COPD Exacerbations 20179

Cochrane Prophylactic antibiotic therapy for COPD 201810

Cochrane Prolonged antibiotics for non-cystic fibrosis bronchiectasis in children and adults 201711

Cochrane Antibiotics for exacerbations of asthma 201812

Cochrane Macrolide antibiotics for bronchiectasis 201813

Cochrane Macrolides for chronic asthma 201514

GINA 2018 Global Strategy for Asthma Management and Prevention15

Cochrane Adverse events in people taking macrolide antibiotics vs placebo for any indication 201916

Within each of sections 4, 5 and 6, a summary of other guideline recommendations for that disease has been included.

Introduction

The majority of asthma treatments aim to reduce airway inflammation with subsequent symptomatic improvement. Pathologically, asthma is characterised by airway inflammation and bronchial hyper-responsiveness. Airway inflammation can be predominantly neutrophilic or eosinophilic in nature.15 Despite an increase in therapeutic options and clear guidance on stepwise management, asthma symptoms can be difficult to manage and exacerbations remain a predominant feature for many. Adherence to treatment can be challenging. Newer therapeutic options, including biologics, can be expensive, require frequent monitoring and are limited to specific clinical phenotypes.15 25

There has been significant interest in the potential role of oral macrolide therapy as an option in asthma. It is relatively inexpensive, easy to adhere to and has a physiologically relevant mechanism of action which may reduce airway inflammation in asthma. There is also evidence suggesting increasing asthma severity may be linked to chronic colonisation with Chlamydiae and Mycoplasma pneumoniae. It has therefore been postulated that benefit from macrolide therapy may also be through its antibacterial effects on these and other lung organisms.26

As a consequence, there have been a number of studies over several decades examining the response to macrolide therapy in asthma which vary considerably in design and outcome measures. The first study examining the role of macrolides in asthma was their use as a steroid-sparing agent in 1970.17 Studies since have examined their role in reducing airway inflammation and bronchial hyper-responsiveness, impact on symptoms and quality of life and most recently on their ability to reduce exacerbation frequency.

Evidence base

Examining the evidence base for macrolide therapy in asthma is particularly challenging because of the significant heterogeneity in study designs, outcome measures and study populations, as well as the heterogeneity of the asthma population itself. A Cochrane review of macrolide therapy in asthma identified significant publication bias, poor study design and lack of consistency in outcome measures as the key barriers to making recommendations based on the available evidence.14 Considering the size of the clinical population and burden of disease in this population,15 there is a relative paucity of high-quality evidence regarding macrolide use in asthma. There are relatively few studies with small study populations which are often underpowered to detect appropriate changes in the outcome measures used. The study populations are also varied limiting the applicability of their results to the wider asthma population and the ability to determine where macrolide therapy should sit within an asthma treatment regime.

Despite this, across multiple outcome measures statistical significance was achieved in many studies, but the magnitude of the changes seen were very small. The clinical significance of these changes varies from unknown to unlikely to be of any clinical benefit. This is reflected in the recommendations made in this guideline. No evidence of the impact of macrolide therapy on mortality, exercise capacity, disease progression or sputum production in people with asthma was found therefore no recommendations in regard to these outcomes can be made in this guideline. In all cases, macrolides should be considered after current therapy has been optimised and adherence established.

Other recent guidance

Recommendations on macrolide therapy are not included in current NICE/BTS/SIGN asthma guidelines.4 25 NICE guidance suggests referral to an asthma specialist and consideration of the addition of a trial of ‘an additional drug’ (which could include a macrolide) if asthma is uncontrolled in adults (aged 17 and over) on a moderate maintenance ICS dose with a LABA (either as MART or a fixed-dose regimen).

Quality of life (QOL)

QOL was the most consistently assessed outcome measure in studies of macrolide therapy in asthma. All but one study27 used the well-validated Asthma Quality of Life Questionnaire (AQLQ) score. This score was first validated in 1992 and has a minimal important difference (MID) of 0.5.28 This is critical to appraising this body of evidence, as in similarity with the symptomatic improvements seen, after treatment with macrolides there is a consistent small but statistically insignificant (p>0.05) improvement in QOL seen across the majority of studies.14 27 29–36 In two studies comparing macrolide therapy with placebo,33 34 this change reached the MID of the AQLQ. The first used 8 weeks’ treatment with clarithromycin34 and the second was the open-label treatment arm of a study of 48 weeks’ treatment with azithromycin.33

It is therefore clear that treatment with macrolides may result in measurable improvements in QOL for people with asthma, but the clinical impact of these changes remains unknown and may be very small.

Symptoms

Thirteen studies used symptom scoring as an outcome in response to macrolide therapy.14 27 29–31 34–41 There was considerable variation in the scoring systems used, from the validated Asthma Control score (ACQ) to simplified unvalidated scoring systems, created specifically for an individual study. Overall, 10 of the studies demonstrated a reduction in symptoms. Three studies29 31 33 demonstrated an increase in symptoms, of which only one used a validated and reproducible symptom score. Although the majority of studies demonstrated an improvement, the actual changes were minimal, unlikely to be of clinical significance and reached statistical significance in only three cases.37 38 41 None of the studies using the ACQ score demonstrated a change which would meet the ACQ minimal important difference (MID) (0.5).42

Further work is needed to see if these improvements may be of greater significance over more prolonged treatment periods, and whether symptomatic improvement is related to the impact of macrolide therapy on exacerbation rate.

Exacerbations

Exacerbations of asthma are defined as an increase in symptoms, accompanied by a reduction in peak flow rate (PEFR), of graded severity dependent on the degree of clinical and PEFR deterioration.4 25 Exacerbation rate is the definitive outcome measure in regard to the clinical efficacy of macrolides in other respiratory diseases, and the major purported driver for their use in cystic fibrosis, bronchiectasis and COPD.14 Two studies30 40 have been specifically designed to address this question in asthma. In other studies, exacerbations have been measured as a secondary outcome, with differing definitions between studies and insufficient statistical power to detect a reduction in exacerbation frequency. A meta-analysis of these studies showed exacerbations requiring steroids in 19.6% (31/158) of those treated with macrolides and 24.2% (32/132) of those not receiving macrolide treatment.14 However, this reduction was not sufficient to reach statistical significance (p>0.05).

Of the two studies30 40 specifically designed to determine a reduction in exacerbation frequency in response to azithromycin, the AMAZES trial demonstrated a significant reduction in exacerbations (p<0.0001),40 but the AZIZAST trial did not (p=0.68).30 This may reflect the duration of treatment (AMAZES=48 weeks; AZISAST=6 months), dose effect (AMAZES=500 mg thrice weekly; AZISAST=250 mg thrice weekly) or differing study populations. AMAZES recruited a larger population (n=420) on high-dose inhaled steroids with an average of one exacerbation in the previous year, whereas the AZISAST population was smaller (n=109) with higher inhaled steroid doses and two exacerbations in the previous year (see table 5). Both studies also had slightly differing definitions of an exacerbation. These studies are the highest quality evidence found for the purposes of this guideline and were suitably powered to detect a change in exacerbation frequency.

The AZISAST study demonstrated no significant difference between exacerbation rates with 48% (26/54) of the placebo group and 47% (26/55) of the azithromycin group experiencing an exacerbation (relative risk 0.98, 95% CI 0.68 to 1.43, p=0.68).30 In AMAZES, 61% (127/203) of the placebo group experienced an exacerbation compared with 44% (94/213) of the azithromycin group giving an incidence rate ratio of 0.59 (95% CI 0.47 to 0.74, p<0.0001)40 (see table 5).

Both studies assessed whether the predominant type of lung inflammation impacted the effectiveness of macrolide therapy. In a predefined subgroup analysis of the AZISAST population, those with non-eosinophilic asthma showed a significant reduction in exacerbations in response to azithromycin treatment, with exacerbations occurring in 33% (9/27) of the azithromycin group compared with 62% (18/29) of the placebo group (relative risk 0.54, 95% CI 0.29 to 0.98, p=0.037).30 In AMAZES, both those with eosinophilic asthma and non-eosinophilic asthma demonstrated a reduction in exacerbations, with a slightly greater reduction seen in the eosinophilic group. The incidence rate ratio in the eosinophilic group was 0.52 (95% CI 0.29 to 0.94, p=0.030) and 0.66 (95% CI 0.47 to 0.93, p=0.019) in the non-eosinophilic group.40 This difference in outcomes may be partly explained by the differing definition of phenotypes between the two studies. In AZISAST, non-eosinophilic asthma was defined by a fraction of exhaled nitric oxide (FeNO) lower than the upper limit of normal and a blood eosinophilia ≤200/µL.30 In AMAZES, non-eosinophilic asthma was defined by a baseline sputum eosinophil count <3% or blood eosinophil count <300/μL if sputum unavailable.40

Overall, this evidence suggests that macrolide therapy may vary in its effectiveness between differing asthma phenotypes and that clinical benefit may be determined by more than the predominant type of inflammation. Further work is needed to investigate the impact of macrolides on exacerbation rate in differing asthma phenotypes and populations. However, it appears from AMAZES40 that there is evidence of a reduction in exacerbations, but studies may need to run in larger populations for a longer duration to assess this effect successfully.

Steroid dose reduction

The original studies of macrolide use in asthma had the aim of reducing steroid dosage,17 but this has not been the stated outcome of a study using macrolide therapy since the 1990s. With the advent of newer biologic therapies and the focus on clinical phenotyping, there has been a move away from using therapies as steroid-sparing agents, towards a stratified approach to asthma therapy.

The three studies using steroid reduction as an outcome all used troleandomycin for between 2 and 52 weeks.43–45 All showed a reduction in steroid dosage, which has been confirmed in two Cochrane Reviews.14 39 However, these studies also included paediatric populations. Troleandomycin is also not widely used and is not available in the UK.

Lung function/PEFR

FEV₁ remains a key outcome measure in trials of airways diseases. However, in asthma, FEV₁ can show considerable variation and is used as a secondary outcome measure in many studies. Peak flow was used frequently as a primary and secondary outcome, but with considerable variation in timing, rigour of performance and units of measurement.

Overall FEV₁ improved in response to oral macrolide therapy, but to a small degree which may not result in any clinical improvement.14 31 35 37 39 46 This is illustrated by the results of four meta-analyses which individually compared FEV₁ responses across 15 studies using a range of macrolides. The Cochrane meta-analysis demonstrated a 0.08 L (95% CI 0.02 to 0.14 (p=0.0097)) improvement in FEV₁ (47); the Tong et al meta-analysis showed a 0.11 L (95% CI 0.06 to 0.16 (p<0.001)) improvement36; and Reiter et al showed no effect on FEV₁ (standardised mean difference (SMD) 0.05 (95% CI −0.14 to 0.25 (p=0.60)).41 FEV₁ (% predicted) was only analysed in the Tong et al meta-analysis, showing a SMD of 0.27% (95% CI 0.05 to 0.59 (p=0.10)).36 The fourth meta-analysis was confined to studies examining troleandomycin, showing a 0.06 L (95% CI −0.8 to 0.92 (p=0.88)) change in FEV₁.39

Morning PEFR was the most commonly measured parameter across studies. A positive change in morning PEFR was seen in several studies, of a relatively smaller magnitude than that seen in FEV₁.14 30 35 36 41 The Cochrane and Tong et al meta-analyses showed a 2.22 L/min (95% CI 9.73 to 14.17 (p=0.72))14 improvement and a SMD 0.25 (95% CI 0.1 to 0.39 (p+0.001))36 in morning PEFR, respectively. A third meta-analysis by Reiter et al also showed a 6.7 L/min (95% CI 1.35, 12.06 (p=0.014)) improvement, but this analysis included paediatric study populations.41 The change in evening PEFR is more mixed across studies, with some showing a reduction35 rather than improvement, although with a smaller evidence base.14 29 30 With the known variability of PEFR in asthma, again it is difficult to judge the clinical importance of these changes.

Inflammation

Studies examining the impact of oral macrolide therapy on inflammation in asthma have used a variety of outcome measures in both blood and sputum. In keeping with common clinical phenotypes, the impact of macrolides on both neutrophilic and eosinophilic inflammation has been assessed. Two meta-analyses14 examined the effect of macrolide therapy on sputum and blood eosinophil counts measured in three separate RCTs after treatment with roxithromycin, clarithromycin and azithromycin. Consistent reductions in blood eosinophils were shown (mean difference (MD) −33.5 (95% CI −30.9 to −36.11)×10⁴/mL (p<0.00001)), but only the study examining clarithromycin (MD −74×10⁴/mL) demonstrated a reduction in sputum eosinophils.37

Five studies examined the impact of roxithromycin,46 clarithromycin37 and azithromycin31 on sputum neutrophils. One study demonstrated an increase in sputum neutrophils (MD 19.2 (95% CI −24.2 to 62.6), p>0.05) in response to azithromycin,47 but pooled results of the other studies, as part of a meta-analysis, demonstrated an overall reduction of −0.25% (95% CI −0.62 to 0.13, p=0.20).36

Emerging biomarkers in the form of matrix metalloproteinase (MMP-9), neutrophil elastase and eosinophilic cationic protein (ECP) have also been examined. Clarithromycin resulted in a reduction in sputum MMP-9 and neutrophil elastase after 8 weeks of treatment.34 Serum and sputum ECP were both reduced after treatment with clarithromycin and roxithromycin.37 46

FeNO is increasingly used as a biomarker of inflammatory activity in asthma and a measure of response to therapy. Three randomised trials used FeNO as an outcome measure all demonstrating a small and non-significant reduction in response to treatment with 12 weeks of azithromycin (−1.94 (95% CI −5.97 to 2.10) ppb (p=0.34)),31 26 weeks of azithromycin (MD −1.6 ppb (p=0.52))30 and 16 weeks of clarithromycin (−4.6 (SE ±4.2) ppb (p=0.5)), respectively.35

While there is considerable variation in study populations, treatment regimens and outcome measures, the overall picture is that macrolides do reduce airway inflammation in asthma. Further work is needed to clarify if this translates into clinically relevant improvements for patients and to determine which inflammatory biomarker may be of most use in determining response to treatment.

Bronchial hyper-responsiveness

Bronchial hyper-responsiveness has been used as an outcome measure in a number of studies of macrolide therapy in asthma31 37 46 to provide proof of physiological effect and potential explanation for improvement in other clinical outcome measures. The studies confirm that macrolide therapy does result in improvements in bronchial hyper-responsiveness measured by methacholine challenge test (PC20/PD20). Two studies were specifically designed to assess bronchial hyper-responsiveness as a primary outcome.37 48 When pooled, they demonstrated a 1.99 (95% CI 0.46,3.52 (p=0.011)) SMD improvement in methacholine response. A second meta-analysis of five studies demonstrated an improvement of 0.9 (95% CI 0.5,1.75 (p=0.04)).36 Other studies demonstrated varying magnitudes in the degree of improvement seen, likely related to differing macrolides, treatment dosages, duration of treatment, study populations and measurement differences.31 35 37 46 48

While these findings support a physiological mechanism of action of macrolides through a reduction in airway inflammation and bronchial hyper-responsiveness, there seems to be a lack of clear correlation between the degree of improvement seen and any corresponding improvement in clinical outcomes. This requires further investigation in future trials.

Microbiology

Five studies specifically reported the impact of oral macrolide therapy on microbiological outcomes. Black et al conducted an RCT of 6 weeks of roxithromycin versus placebo in 232 people with asthma and raised IgG/IgA antibody titres to Chlamydia pneumoniae. 29 This demonstrated a reduction in IgG antibodies, but not IgA in response to treatment; but with no clinically significant changes in other outcome measures. Hahn et al demonstrated that high IgA but not IgG antibodies to C. pneumoniae were significantly associated with more severe asthma symptoms at the end of follow-up.33 Azithromycin resulted in improvements in IgA antibodies regardless of whether IgA antibody levels were high or low at the start of treatment.33

Sutherland et al attempted to conduct a PCR-stratified RCT based on PCR positivity for C. pneumoniae or Mycoplasma pneumoniae.35 Due to lower than anticipated numbers of PCR-positive participants, the two arms could not be successfully matched in number, but comparison of the two groups showed no significant differences in outcome between PCR positive (n=12) and PCR negative (n=80) after treatment with 16 weeks of clarithromycin.35

The AMAZES study monitored a subgroup of study participants for emergence of antimicrobial resistance to azithromycin within serial sputum samples over the 48-week study period.40 At the study end, resistance to macrolides was found in 19/39 (48.7%) receiving azithromycin and 12/42 (28.6%) receiving placebo.40 AZISAST performed a similar subgroup analysis in 46 participants demonstrating an increase in macrolide-resistant streptococci from 47.8% to 87% participants in the azithromycin group and a reduction from 39.1% to 35% of participants in the placebo group.30 Interestingly, the percentage of macrolide-resistant streptococci reduced from 73.8% to 45.9% in the 4-week washout period at the end of the study.30 Further analysis of the oropharyngeal microbiome of a subgroup from AZISAST demonstrated a fivefold increase in Streptococcus salivarius and corresponding fivefold decrease in Leptotrichia wadei during azithromycin treatment. However, azithromycin had little impact on the rest of the microbiota. In keeping with the culture-based microbiology, the microbiome had already returned to pre-treatment levels in 50% participants by the end of the 4-week washout period.49

This suggests that chronic macrolide therapy may increase antimicrobial resistance to macrolides. This result has yet to be reproduced in other large studies and the clinical impact of this remains unknown. It is also unclear whether this increase in resistance is temporary or fluctuating. However, it does suggest that monitoring for antimicrobial resistance may be of benefit in those receiving chronic macrolide therapy. Use of breaks in chronic therapy, if the desired clinical outcome is achieved, may be considered to reduce resistance, as is the practice in some bronchiectasis services. Further work is needed to examine this relationship in more depth and determine the clinical impact of increasing resistance rates.

Safety and adverse events

Reporting of adverse events in the majority of studies was of low quality with no specific assessment for the well-documented adverse effects of macrolide therapy in the majority of studies. However, overall there was little evidence of significant reactions or side effects secondary to macrolide therapy in asthma. The most commonly reported side effects were gastrointestinal, including nausea and abdominal pain. In the meta-analysis performed by Reiter et al, a significantly increased risk of nausea was seen in those receiving macrolide therapy.41 Black et al also showed that 12.4% (13/105) of the roxithromycin group developed nausea compared with 4.5% (5/112) of the placebo group.29

The AMAZES and AZISAST trials specifically assessed and reported adverse events as part of the planned study protocols.30 40 In AMAZES, the overall serious adverse event (SAE) rate in the azithromycin group was 7.5% (16/213) compared with 12.8% (26/203) in the placebo group.40 Gastrointestinal side effects were more common in those on azithromycin compared with placebo—diarrhoea (33.8% (72/213) vs 19.2% (39/203)) and abdominal pain (17.8% (38/213) vs 14.8% (30/203)).40 QTc prolongation was not seen more commonly in the azithromycin group with 0.5% (1 participant) of each group developing ECG changes during the study.40 There was also no difference in the rates of tinnitus (0.9% (2/213) vs 1% (2/203)) or hearing loss (2.8% (6/213) vs 3.4% (7/203)) seen.40

In AZISAST, the SAE rate was identical in both placebo and azithromycin groups (11%).30 Discontinuation of treatment was more common in the placebo group (9% (5/54 vs 4% (2/55)). Diarrhoea, nausea and abdominal pain were also more common in the placebo group. Two participants in the azithromycin group developed abnormal liver function tests, but the severity of this was not reported. No participants reported any change in their hearing.30

Overall, it appears that macrolides are well tolerated in people with asthma, although the quality of evidence is low and inclusion of formal assessment for adverse events should be included in future studies. There is a slightly higher incidence of gastrointestinal side effects which did not lead to cessation of treatment. Warning of the possibility of these side effects developing should be given to patients on initiation of treatment. There is insufficient evidence on the frequency of other adverse events specifically QTc prolongation, liver function abnormalities and audiological changes to make a formal recommendation on screening for these events. In general, participants with evidence of QTc prolongation or hearing loss were excluded from study populations further limiting the applicability of this evidence to the wider asthma population.

Outcome weighting

In the creation of this guideline, a set of key outcome measures in relation to macrolide therapy in management of respiratory disease were determined. However, this guideline’s scope covers a spectrum of respiratory disease and the importance of individual outcome measures within each disease area varies. In regards to asthma, the GDG gave importance to those outcomes of most clinical relevance to respiratory physicians managing people with asthma and our recommendations reflect this. The group only made recommendations on these outcomes where the quality of evidence and magnitude of change was deemed sufficient to be of clinical utility. Therefore, recommendations are made in regard to exacerbation reduction, but not in relation to symptoms or quality of life as the changes seen in these outcomes were too small to be clinically important.

The body of evidence in regard to other outcomes, which may be of scientific or physiological relevance, is outlined in the evidence summary below, but no recommendations have been made in areas where the outcome is not used as a clinical endpoint—for example, bronchial hyper-responsiveness.

Evidence summary

Oral macrolide therapy improves quality of life in people with asthma, but changes are small and their clinical significance is uncertain. (Moderate)

Treatment with macrolides results in an improvement in asthma symptoms in some patients although this is small in magnitude and may not be applicable to all patients with asthma. (Low)

Oral macrolide therapy can reduce exacerbations of asthma in adults (50–70 years) with ongoing symptoms despite >80% adherence to high-dose inhaled steroids (>800 µg/day) and at least one exacerbation requiring oral steroids in the past year. (Low)

Azithromycin given thrice weekly at 500 mg over 48 weeks results in a reduction in asthma exacerbations. (Low)

In some individuals, oral macrolide therapy may result in a reduction in oral steroid dose, but this is not a consistent finding. (Low)

Oral macrolide therapy can result in a small improvement in lung function and PEFR in people with asthma. (Low)

Oral macrolide therapy may reduce airway inflammation in asthma. (Low)

Oral macrolide therapy may reduce bronchial hyper-responsiveness in asthma. (Low)

Two studies demonstrated that treatment with oral macrolides (azithromycin) may result in an increase in bacteria within sputum resistant to macrolides, but currently there is no evidence of this adversely affecting clinical outcomes. (Low)

Oral macrolide therapy may result in an increase in gastrointestinal side effects including abdominal pain, nausea and diarrhoea. (Low)

There is insufficient evidence to make any recommendation on the impact of oral macrolide therapy on mortality, exercise capacity, disease progression or sputum characteristics in people with asthma.

Introduction

Bronchiectasis is a chronic respiratory condition in which abnormally dilated bronchi are radiologically displayed in patients with a relevant clinical syndrome. These patients suffer from multiple symptoms, including a chronic productive cough, recurrent exacerbations associated with bronchial infection, dyspnoea and fatigue. While patients with cystic fibrosis can have both the radiological changes and the clinical syndrome, the term ‘bronchiectasis’ generally refers to those who do not have cystic fibrosis and the condition is often referred to as non-cystic fibrosis bronchiectasis. Even after the exclusion of cystic fibrosis, a wide range of aetiologies are attributed to bronchiectasis, though a large proportion of cases are described as idiopathic or post-infective.50–54 Recent UK data report prevalence rates of 566/100 000 in adult women and 486/100 000 in adult men.53 Patients with bronchiectasis bear a significant burden of morbidity and mortality.53 55

The basic concept for the management of bronchiectasis is to attempt to break the vicious recurrent cycle of chronic bacterial infection, inflammation, impaired mucociliary clearance and structural lung disease.56 With the goal of breaking this cycle and the extrapolation of management strategies from cystic fibrosis, it is unsurprising that macrolides have been used to try to improve outcomes in bronchiectasis. As well as having broad antimicrobial effects, anti-inflammatory effects are likely. Studies in patients with bronchiectasis have suggested reductions in Th17 cell responses57 and evidence of reduced airway inflammation via measuring markers such as IL8, neutrophil elastase and matrix metalloproteinase-9.58 A further attractive mechanism, particularly in a condition where P. aeruginosa is an important pathogen, is the inhibition of quorum sensing (a form of chemical signalling between bacteria).59

Evidence base

Davies and Wilson described early experiences of using azithromycin in 200460 and subsequent published studies of varying quality, using different regimens, have added to the available evidence. The main body of high-level evidence is based on three RCTs61–63 (see table 6). These studies were performed in the Netherlands, Australia and New Zealand and were all placebo controlled. The macrolides used were azithromycin in the Netherlands and New Zealand studies, while the Australian study used erythromycin. Audit data suggest that azithromycin is the preferred macrolide in the UK at present.64 The two studies which used azithromycin had different dosing regimens (250 mg daily and 500 mg three times a week). In addition, other studies of lower quality evidence have used other regimens such as azithromycin 250 mg three times a week,60 65 66 azithromycin 500 mg twice weekly,67 erythromycin 500 mg twice a week,68 clarithromycin 500 mg once daily57 and roxithromycin 150 mg a day.58 The lack of consistency in the macrolide regimen used, and the absence of head-to-head comparisons, make the optimum macrolide and dosing unclear.2 Also, because none of the studies were longer than a year, the longer-term benefits and risks have not been established. Subsequent to the three RCTs which provide high-quality evidence, meta-analyses have been produced.69–71 While this provides pooled data, some lower quality evidence influence the output.66 70 A further meta analysis of individual patient data (IPD) from the three main RCTs was published after the completion of literature searches and is referred to below because of the potential importance of its findings.72

Other recent guidance

The BTS bronchiectasis guidelines cover options beyond macrolides when considering long term antibiotic therapy in stable disease.2 They recommend macrolides as first choice for patients without P. aeruginosa with inhaled gentamicin as a second-line alternative. For patients with P. aeruginosa, the BTS guideline recommends inhaled colistin with macrolides as an alternative for those intolerant of inhaled antibiotics. Interestingly, the IPD meta-analysis by Chalmers et al suggests that macrolides are effective in patients with and without P. aeruginosa. 72 In addition, the BTS guideline recommends a starting dose of azithromycin of 250 mg three times a week to minimise side effects. While accepting that side effects are common with macrolides, we note that dropouts from studies due to side effects are rare and the strongest evidence of efficacy lies with higher dosing regimes. We suggest a pragmatic starting regime of azithromycin 500 mg three times a week or 250 mg daily in patients with bronchiectasis. However, if there is a patient history of drug intolerances, it would be sensible to start at the lower dose of 250 mg three times a week.

A NICE guideline was published in 2018 addressing acute exacerbations in bronchiectasis.6 With regards to preventing exacerbations, the document recommends the consideration of prophylactic antibiotics in people with repeated exacerbations. The guidance is less specific on which antimicrobials to use and what a significant exacerbation rate is. However, both the NICE and BTS documents advocate specialist input.

Quality of life

QOL data were collected from the three main RCTs and consequently there is high-level evidence available. While pooled data are also available, there are limitations of these data due to the influence of data of lower quality.66 The studies with the highest quality of evidence used the SGRQ to assess quality of life and any effect seen does not appear to be large. The BAT study did find a significant difference between treatment and placebo, with a reduction of −12.18 points after 1 year of treatment and a reduction of −4.12 in the placebo group (p=0.046).62 A change of 4 points is considered clinically significant. In the treatment group, 64% achieved this at 1 year while 46% did in the placebo group. While the other two main RCTs did not show a statistically significant improvement in SGRQ, the results from the BAT study should not be viewed as being inconsistent. In the EMBRACE study, 6 months of treatment failed to demonstrate a statistically significant difference though there was a trend towards benefit (−5.17 vs −1.92, 95% CI −7.21 to 0.72).61 Similar findings were also seen in the BLESS study with 48 weeks of therapy.63 This lack of clear effect may be due to a lack of sensitivity of the SGRQ for this type of therapy. Other studies have reported significant improvements in quality of life; however, the open-label nature or lack of a placebo group means they are of considerably lower quality of evidence.60 66 Meta-analysis data did suggest a benefit with a change of −5.39 in the SGRQ; however, these data were influenced by a large change in a study of lower quality.69 70

On the basis of the current high-quality evidence, it appears that to obtain a significant improvement in quality of life, a long course of therapy may be required. On the basis of the BAT trial, this may need to be for a year. It is unclear what the impact of therapy beyond 1 year is. As the only high-quality evidence that suggests an improvement in quality of life is the BAT trial, the dosing regimen of azithromycin 250 mg once a day carries the best evidence for achieving this outcome.

Symptoms

The prospective collection of symptom scores in RCTs has allowed the accumulation of some high-level evidence with regards to symptoms. This has been described in various ways including the symptom component of the SGRQ, the Leicester Cough Questionnaire and the lower respiratory tract infections—visual analogue scale (LRTI-VAS). The EMBRACE study demonstrated an improvement in the symptom component of the SGRQ with a difference in change from baseline of –6.7 (95% CI –13.37 to –0.04).61 The BLESS study suggested a similar benefit, with a treatment effect of −5.3 reported; however, this did not reach statistical significance (95% CI −12.6 to 2.1).63 There was also no statistically significant change found in the Leicester Cough Questionnaire (LCQ). The BAT study used the LRTI-VAS and described a larger benefit in symptom improvement in those receiving treatment than those receiving placebo (p=0.047).62 While not unanimous, there does appear to be some evidence of benefit with regards to symptoms, and inconsistencies may in part be due to differences and weaknesses in the scoring systems.

Exacerbations

Due to the three main RCTs using exacerbations as a primary endpoint and the subsequent meta-analyses, there is high level evidence which is in favour of using long-term macrolides to reduce exacerbations in bronchiectasis.61–63 69–71 As in other respiratory conditions, there is variability in the interpretation of what exactly constitutes an exacerbation. A consensus definition has been published; however, this was subsequent to the important studies in this field.73

The previously mentioned three principal RCTs described benefit for this outcome when macrolides were used for between 6 and 12 months.61–63 The BLESS study described an incidence rate ratio of 0.57 (95% CI 0.42 to 0.77) when compared with no intervention.63 The BAT study reported an absolute risk reduction of 33.5% (95% CI 14.1% to 52.9%).62 Of interest in this study, there was an improvement in median number of exacerbation from 5 to 2 in the placebo group. Over a 6-month period, the EMBRACE study identified a rate ratio of 0.38 between the treatment group and the placebo group (95% CI 0.26 to 0.54).61 It is important to note that the EMBRACE study continued for a further 6 months after stopping the intervention and these benefits were maintained. However, it should also be noted that this study provided two definitions of ‘exacerbations’ and the results were not consistent, and that there was a higher baseline exacerbation rate in the control group. The meta-analysis data are consistent with an improvement in exacerbation rates.69–71

For macrolides to be used to reduce exacerbations, there needs to be a relevant exacerbation rate at baseline to consider when implementing this therapy. There were different entry criteria for the main trials with regards to baseline exacerbation rates. While entry into the EMBRACE study required just a single exacerbation for study recruitment, the mean exacerbation rate in the preceding year was over three (3.34 in treatment group and 3.93 in control group). In the BLESS and the BAT studies, two and three exacerbations were required, respectively, in the preceding year. In the BLESS study, approximately a third had five or more exacerbations in the preceding year, but it is unclear what the mean or median baseline exacerbation rate was. In the BAT study, the median baseline exacerbation rate was 4 (and 5 in the control group). In light of the baseline exacerbation rates in these trials, and three or more exacerbations being a marker of disease severity,74 a threshold of three or more exacerbations per year seems appropriate as a guide. It should be acknowledged that the aforementioned IPD meta-analysis that has followed the literature search for this guideline performed multiple subgroup analyses including for different exacerbation rates. One of these subgroups was those with a preceding exacerbation rate of less than this threshold (ie, 1 or 2 per year). While this group was small (37 receiving macrolides vs 36 receiving placebo), there appeared to be a benefit with regards to exacerbations, but a trend towards decreased QOL. It is conceivable that the benefits of macrolides may be outweighed by side effects in those with lower exacerbation rates. This subgroup require further study.

While out of keeping with the BTS Bronchiectasis Guideline and some UK respiratory clinicians’ current practice, the regimens used in the three main RCTs (azithromycin 250 mg daily, azithromycin 500 mg three times a week and erythromycin ethylsuccinate 400 mg twice a day) have the greatest supportive evidence of benefit. The BTS bronchiectasis guideline suggest a lower starting dose (azithromycin 250 mg three times a week) noting the dose-related nature of side effects. Withdrawals due to side effects in the three main studies were, however, rare. It would seem sensible to start at the higher dose (azithromycin 500 mg three times a week or 250 mg daily) unless there is a history of previous drug intolerances.2 As well as variable dosing regimens, variable durations of therapy have also been studied. The highest quality studies used macrolide therapy for 6 or 12 months. While it is conceivable that it continues to provide benefit, there is no evidence either in favour or against the prolongation of therapy beyond 12 months.

Exercise capacity

Exercise capacity was measured by a 6 min walk test in both the BLESS and the EMBRACE studies.61 63 This provides some high-level evidence via objective measures with blinded subjects and investigators. In neither study was there statistical evidence of benefit or harm from therapy. In addition, the activity component of the SRGQ was reported in the EMBRACE study61 and no impact on this outcome was seen. On the basis of this evidence from two studies, there is no indication that macrolides influence exercise capacity or tolerance.

Sputum characteristics

Sputum production is a common occurrence in many patients with bronchiectasis and is considered a significant issue by patients.75 There is, however, no high-quality evidence with regards to the effects of macrolide therapy on sputum. There is moderate level evidence of reduced sputum weight and sputum volume.63 70 Estimates of the effect of therapy may be a reduction of approximately 11 mL per day (95% CI −12.70 to −8.83) based on pooled data.70 The BLESS study reported a treatment effect of a reduction of 4.3 g per day.63 It is unclear whether this scale of benefit is important to a patient but could be considered in discussion alongside assessment of other factors such as exacerbation rate when considering therapy.

Safety and adverse events

The evidence base for adverse events is quite variable within the bronchiectasis literature. While QTc prolongation is a potential concern with long-term macrolide use, there is very little evidence with regards to this adverse effect. In the BLESS study, QTc was monitored and one patient was withdrawn due to concerns with regards to prolongation,63 however pre-therapy and post-therapy ECGs suggested that this may not have been a drug-related effect. The EMBRACE and BAT studies did not address this potential adverse event.61 62 An adverse impact on hearing is a further concern with macrolides; however, there is also limited evidence for this potential side effect. The BLESS and EMBRACE studies did not assess it, while the BAT study just performed a post-study questionnaire (and did not detect an effect).61–63

There is, however, considerably more evidence in these cohorts for gastrointestinal side effects. The meta-analysis pooled data show high-quality evidence for diarrhoea being an adverse event associated with long-term macrolide use.69 71 The frequency of patients suffering diarrhoea was 19.3%–20.6% in the treatment groups compared with 4.1%–4.5% in the placebo groups. There was also high-level evidence for abdominal pain/discomfort when reported in the BAT study (18.6% vs 2.5%, RR 7.44) and meta-analysis data (OR 6.97).62 71 Despite these gastrointestinal side effects, meta-analysis data assessing for withdrawal from studies due to side effects did not reveal any differences between those on macrolide or placebo (OR 1.18, 95% CI 0.33 to 4.19).69

While not a patient-centred outcome, high-quality evidence from the BLESS study reports increased micro-organism resistance.63 Further moderate level evidence was also added from the BAT study, though not the EMBRACE trial where routine testing was not performed.61 62 In the BLESS study, there was a significant increase of macrolide resistance in commensal oropharyngeal streptococci with erythromycin use (27.7% vs 0.04%, p<0.001).63 The BAT study provided a significant amount towards the meta-analysis data and in the pathogens that were tested for macrolide sensitivity, 88% became resistant compared with 26% in the placebo group.62 69 At present, there is no evidence that the resistance seen either does or does not have an impact on clinical outcomes.

Outcome weighting

The available data allow assessment of the pre-specified outcomes with varying levels of evidence. Certain outcomes were excluded from our final assessment and recommendations. These included lung function, hospital admission rates, disease progression and death. Lung function was excluded despite available data, as the guideline group did not feel it is a useful measure when taken in the context of the current evidence. The trials were of too short a period for a meaningful change in lung function to be elicited despite statistically significant outcomes. For example, meta-analysis data from five studies described a statistically significant greater improvement of FEV₁ of 0.02 L (mean weighted difference).70 This was not considered a clinically meaningful change for an individual patient. Hospital admission rate data were available; however, due to such low incidence rates for this outcome in the three RCTs which reported it, no recommendations are made for this outcome.61–63 The outcomes of disease progression and death are not commented on due to a lack of data. The outcomes of QOL, symptoms, exacerbation rates, exercise capacity, sputum characteristics and adverse events were all considered ‘important’ in the GRADE process.

Evidence summary

There is evidence of an improvement in QOL as measured by SGRQ when azithromycin 250 mg daily is used for 1 year (High). Other high-quality studies with other dosing regimens showed a trend towards benefit but not a statistically significant benefit. Meta-analysis data reported a benefit but there are concerns regarding the inclusion of a study of lower quality (Low).

There is evidence from high-quality studies of improvement in symptom scores; however, inconsistency exists which may be due to different scoring systems being used. (Moderate)

Long-term macrolide treatment reduces exacerbations in bronchiectasis. (High)

When using macrolides to reduce exacerbation rates, the dosing regimens with the greatest supportive evidence are azithromycin 250 mg daily, azithromycin 500 mg three times a week and erythromycin ethylsuccinate 400 mg twice a day. (Moderate)

Studies with other dosing regimens, including azithromycin 250 mg three times a week (as pragmatically suggested in the BTS Bronchiectasis Guideline), have also reported a reduction in exacerbation, but have a lower evidence base. (Low)

The studies with the greatest evidence for reducing exacerbations used therapy for a minimum of 6 months. (High) The impact beyond 12 months is unknown.

There is evidence for a reduction in exacerbations over 12 months when therapy is used for 6 months and then not for the subsequent 6 months. (Moderate) It is unknown what the impact is of subsequently recommencing.

Long-term macrolide therapy is not associated with improved exercise capacity. (High)

Long-term macrolide therapy may reduce sputum volume and weight. (Moderate)

Long-term macrolide therapy is associated with diarrhoea and with abdominal pain. (High)

Long-term macrolide usage can result in increased antimicrobial resistance. (High) It is unknown if this has a clinical impact.

Introduction

COPD is a progressive, inflammatory disease of the airways, characterised by ongoing development of non-reversible airflow limitation and acute episodes of exacerbation. Acute exacerbations are described as episodes of acute worsening of respiratory symptoms that necessitate a change in regular medication.76 The natural disease course, health-related QOL, hospital admissions and mortality associated with COPD are influenced by acute exacerbations.77 In addition, acute exacerbations of COPD (AECOPD) are estimated to account for between 50% and 75% of the total costs for COPD treatment.78 Treatment interventions effective in preventing acute exacerbations are therefore of significant clinical benefit.

The mechanisms for COPD disease progression are complex and involve inflammatory and immune responses. Neutrophils, IL-17 and IL-23 are considered to play an active role in disease progression while CD8-positive T cells, neutrophils and macrophages mediate chronic inflammatory responses.79 Acute exacerbations of COPD are postulated to be caused by bacteria or bacteria in combination with viral infection in up to 50% of cases.80 Macrolide antibiotics possess both anti-inflammatory and antibacterial effects and are therefore an important potential treatment strategy to prevent AECOPD.

The potential mechanisms of action and clinical implication of long-term macrolide antibiotic therapy in COPD have been described and reviewed in several studies.81 Table 7 summarises the potential effects of macrolides on disease modulators in COPD.

Evidence base

To confirm the hypothesis that macrolide antibiotics are an effective treatment strategy in COPD to prevent acute exacerbations and possibly modify disease progression, several studies have been conducted yielding varied results. The evidence base for effectiveness of macrolide antibiotics in this context is compromised by the significant heterogeneity in populations investigated, intervention (type, dose and duration of macrolide), definition of outcomes measured and lack of phenotyping COPD to identify subgroups most likely to respond.

Six meta-analyses were identified in addition to the published ERS/ATS Guideline on Prevention of COPD exacerbations.82 Nine RCTs were reviewed and included in the guideline. The COLUMBUS trial published in 201483 and the small RCT by Simpson et al 84 were included in data analysis of one of the meta-analysis and the ERS/ATS guideline.82 Table 8 illustrates the matrix of evidence reviewed, all RCTs and meta-analyses published since 2000 with evidence included in meta-analysis.

Other recent guidance

Symptoms and QOL

Health status measurement is a standardised and objective means of quantifying the impact of disease on patients’ daily life, health and well-being. Health status questionnaires usually address emotional and psychological effects of the illness as well as the physical. The St George’s Respiratory Questionnaire (SGRQ) is validated to measure health impairment in patients with COPD. It is in two parts; Part 1 produces the Symptoms score, and Part 2 the Activity and Impacts scores, which results in a total score.91 92

Macrolide therapy (azithromycin) improved health-related QOL in four studies (1323 patients) which used the SGRQ.83 84 93 94 There was significant heterogeneity among the studies (I²=97.10%).

The Berkhof et al study94 assessed QOL in patients with COPD, chronic cough and at least one AECOPD in the previous year. Patients with COPD and chronic cough is a heterogeneous group of patients, but from this single study, patients with more severe cough reported better QOL (LCQ) on macrolide treatment. Three other studies did not include LCQ in the disease-specific impact on QOL and the impact of cough could therefore not be pooled and analysed for these studies.83 84 93 The pooled assessment for long-term azithromycin suggests its use improved QOL as measured by the SGRQ total score (MD=−2.12, 95% CI −3.44 to −0.79, p=0.002, I²=0%). Although statistically significant, the results did not meet clinical significant outcome of 4 units as minimum clinically important difference (MCID).95

Pooled analysis for the Albert et al 2011 and Uzun et al 2014 studies on QOL (SGRQ) in 491 patients on azithromycin and 498 patients on placebo demonstrated improved total scores (MD 2.18 lower, 95% CI 1.53 lower to 2.82 lower), and in symptoms (MD 3.36 lower, 95% CI 2.42 lower to 4.29 lower), activity (MD 1.82 lower, 95% CI 1.03 lower to 2.62 lower) and impacts (mean difference 2.04 lower, 95% CI 1.28 lower to 2.81 lower).83 93 The Albert study had a significant 12-month dropout rate in both intervention (444 completed SGRQ at 12 months of 556 enrolled) and control (453 completed SGRQ at 12 months of 555 enrolled patients) groups, and this may have affected the results.93

Banerjee reported improved health status (SGRQ) at 12 weeks on clarithromycin measured in total, symptom, and impact scores.96 Erythromycin 125 mg once daily for 6 months did not demonstrate significant changes in quality of life (SGRQ) at 3 and 6 months.97

Exacerbations

Nine RCTs were identified that reported the effect of long-term macrolide antibiotics on AECOPD. The study entry criteria varied in terms of the number of exacerbations in patients in the year prior to the trial, the severity of COPD and the definition of the severity of an acute exacerbation.

Hospitalisation for AECOPD

Five studies involving 1424 patients reported on the number of hospitalisations due to acute exacerbations of COPD as secondary outcomes.83 93 94 98 101 Pooled data, as reported by Ni et al, did not demonstrate a significant reduction in hospital admissions in patients with AECOPD on macrolide therapy in comparison with the control group (RR 0.89, 95% CI 0.64 to 1.24, p=0.50).100 Albert et al found a non-significant trend towards reduction in the rate of hospital admissions using macrolides,93 whereas Uzun and colleagues found no significant delay in time to first hospitalisation in patients on long-term macrolide therapy.83

The published evidence does not support the use of long-term macrolides in patients with COPD to reduce severity of exacerbations or requirement for hospitalisation due to exacerbation. The studies were underpowered and definition of severity of exacerbations not comparable with pool data.

Lung function tests and disease progression

The natural history of COPD is characterised by progressive airflow limitation and historically, FEV₁ decline has been considered the single most important marker of disease progression.104 However, data obtained in two different retrospective cohorts have shown that FEV₁ decline is not invariably progressive and FEV₁ is only weakly correlated with patient-related outcomes.105 106

Randomised trials assessing the impact of long-term macrolide treatment in COPD have not reported on disease progression as a secondary outcome. Rate of decline in FEV₁ as single measure is not accurate in the measurement of disease progression, or in assessing the effect of therapeutic interventions in altering the progression of the disease process.107

Lung function was reported in four studies. Pooled analysis was not possible due to the varied ways in which results were reported. Banerjee et al reported no difference in the intervention and placebo groups after 12 weeks of clarithromycin in spirometry measurements or shuttle walk distance.96 Simpson et al reported no difference in FEV₁ reported as percentage predicted with 12 weeks of azithromycin 250 mg a day (30 patients, 15 in each arm).84 No difference in FEV₁ after 12 months’ treatment with erythromycinstudy was withdrawn due to 250 mg or placebo was demonstrated by Seemungal and colleagues in patients with COPD.101 No significant changes were shown between groups in post-bronchodilator FVC, FEV₁ or 6 min walk test in patients on azithromycin 500 mg three times a week after 12 months83

There is no evidence that long-term macrolide therapy results in improvements, or lessened decline, of lung function parameters in patients with COPD.

Safety and adverse events

Long-term treatment with macrolides has been associated with several adverse effects.108 The most common side effects relate to the gastrointestinal tract by stimulating gastrointestinal motility through motilin-like activity. Symptoms including anorexia, nausea, vomiting, diarrhoea and abdominal pain were reported in patients with COPD in the studies reviewed. Cessation of treatment due to gastrointestinal side effects were reported in the studies published by Banerjee et al, Suzuki et al, Seemungal et al, He et al and Blasi et al.96–98 101 102 Uzun published side-effect profiles, indicating that diarrhoea was the only event that was higher statistically significantly more frequent in the azithromycin group compared with placebo.83

Albert et al reported hearing impairment in 142 (25%) patients with COPD treated with azithromycin 250 mg a day and in 110 (20%) patients receiving placebo (p=0.04).93 A small, but significant difference was seen at 3-month review in audiology measurements (in all four sound frequencies) in the treatment arm.93 Repeat audiology testing demonstrated improvement regardless if azithromycin was stopped or not. The authors postulate too stringent criteria for measurable deficit and over reporting of hearing impairment as explanation. None of the other studies reported on hearing impairment as an outcome.

Erythromycin and clarithromycin are both associated with ototoxicity that includes vertigo, hearing loss, deafness and tinnitus.109 A single patient receiving long-term erythromycin in the Seemungal et al study was withdrawn due to significant tinnitus.101

The recognised side-effect profile associated with macrolide therapy includes allergic reactions, skin eruptions, hepatotoxicity, cardiac arrhythmias and QTc prolongation.110–112 These side effects were not reported as adverse events in the studies reviewed based on the trial protocols and exclusion criteria (designed to exclude high-risk patients specifically with regards to cardiac toxicity). Three studies including 212 patients reported four cardiovascular events in the treatment arm and two in the control arm (p=0.43).83 94 97 Albert and colleagues reported the death rate due to cardiovascular events in both the azithromycin and placebo arm as 0.2%.

Pooled analysis of all nine RCTs reviewed demonstrated higher adverse events reported in the intervention group, with gastrointestinal side effects the most commonly reported (OR 1.55, 95% CI 1.00 to 2.39, p=0.05)84 93 94 96–98 101 102 Albert and colleagues reported only 11/558 patients in the treatment arm on azithromycin that had to stop the treatment due to GI side effects and 6/559 patients in the control group stopped placebo for the same reason.93

Antimicrobial resistance

Macrolide antibiotics have direct antibacterial activity against Gram-positive, Gram-negative and atypical bacteria such as Legionella spp, Mycoplasma and Chlamydia spp. Macrolides inhibit RNA-dependent protein synthesis by reversibly binding to the P-site on the 50S subunit of bacterial ribosomes and inhibit transpeptidation or translocation of nascent peptides.81

The severity and rate of AECOPD have been linked to both C. pneumoniae and Haemophilus influenzae.113–117 The antimicrobial activity of macrolide antibiotics with possible reduction in bacterial load may explain in part the effect on exacerbation rate seen in clinical trials in patients with COPD. The acquisition of macrolide resistance with long-term low-dose use of macrolide antibiotics remain uncertain in patients with COPD.

Seven studies monitored and reported on macrolide antibiotic susceptibility changes during treatment period.83 93 94 96–98 101

Albert et al found only 15% of patients could expectorate sputum after a 3-month period on macrolide therapy.93 Investigators used sputum in patients who could expectorate and nasopharyngeal swabs to report on colonisation and resistance. This study subsequently reported Staphylococcus aureus as the most common organism isolated (treatments group 60 (10.7%) vs placebo group 71 (12.7%)) likely due to nasopharyngeal sampling. Moraxella spp and S. pneumoniae were both isolated and did not show significant difference between treatment and placebo arms in the study. During the 12-month study period, significantly more patients in the placebo group became colonised (172 patients) versus 66 patients in the treatment arm on azithromycin 250 mg a day (p<0.001). Newly colonised patients in the treatment group demonstrated significantly higher resistance to macrolide antibiotics 81% compared with 41% in the placebo group (p<0.001).

Seemungal et al reported in 109 patients receiving erythromycin 250 mg twice a day over a 12-month period that only one patient developed resistance to S. pneumoniae. H. influenzae was isolated in 22 of the 109 patients; all of the isolates were found to be resistant to erythromycin.101

At baseline, Uzun et al obtained 22 sputum samples in the treatment group and 20 sputum samples in the placebo group. After a 12-month study period, 51 sputum samples in the treatment group and 57 sputum samples in the placebo group were analysed. This study also reported fewer new colonised patients in the treatment arm, 4 patients versus 12 patients (p=0.044). Macrolide resistance was reported in 6% (3 patients) receiving azithromycin 500 mg three times a week and in 24% (11 patients) receiving placebo for the 12-month period (p=0.036).83

Banerjee et al and He et al did not observe significant differences in colonisation rate or resistance to macrolide antibiotics, but these studies were shorter in duration, 3 and 6 months, respectively.96 97

In a single-centre, single-blind randomised placebo controlled trial, Brill et al demonstrated, after 3 months of azithromycin 250 mg three times a week, no difference in airway inflammation or bacterial load (0.42log₁₀cfu/mL) in sputum when compared with placebo (0.08 CI 0.38 to 0.54).118

Mortality

In studies where mortality was specifically reported, no significant difference was observed between the control and macrolide groups (RR=0.9, 95% CI 0.48 to 1.69).83 93 102

It should be noted that the included studies all report mortality as secondary outcome measure, and longer-term potential impact on mortality (through for example reduced exacerbations) may not manifest in the 12-month follow-up period which was typical for these studies.106 119 120 There is therefore insufficient evidence to indicate a reduction in mortality with the use of long-term macrolides, but no evidence of increased mortality in relation to treatment.

Outcome weighting

The guideline group identified rate of AECOPD, time to first exacerbation, hospitalisation, serious adverse events and mortality as critical outcomes. Important outcomes identified include QOL, disease progression, lung function tests, airway inflammation and acquisition of macrolide resistance.

Evidence summary

Long-term macrolide treatment showed statistically significant improvement in QOL in patients with COPD as measured by SGRQ, but this does not reach the MCID—4 unit change. (Moderate)

Long-term macrolide antibiotics are effective in reducing the acute exacerbation rate in patients with COPD with high exacerbation rates (ie, more than three exacerbations per year, prolonged exacerbations or exacerbations resulting in hospitalisation). (Moderate)

Treatment courses of 12 months demonstrated the biggest effect size in reduction of exacerbation rate. (Low)

The number of hospitalisations was not significantly reduced in patients receiving long-term macrolide therapy. (Moderate)

No significant mortality benefit has been shown in 12-month follow-up studies using macrolide therapy, and there is insufficient evidence to demonstrate the effect of long-term macrolide therapy on the mortality in patients with COPD. (Low)

There is no evidence to suggest that long-term macrolide therapy impacts on disease progression or spirometry measurements and exercise capacity measurements. (Low)

Insufficient evidence is available on airway colonisation and acquisition of macrolide resistance in patients with COPD receiving long-term, low-dose macrolide therapy. (Moderate)

Introduction

Bronchiolitis obliterans is a condition characterised by subepithelial inflammatory and fibrotic narrowing of small airways within the lung. It occurs following a variety of insults to the lung such as infections or inhaled toxins. It also occurs in the context of systemic diseases such as rheumatoid arthritis and other connective tissue diseases. It can occur as a complication of transplantation of lung or haematopoietic stem cells (HSCT). The histological correlate is obliterative bronchiolitis or constrictive bronchiolitis.121

In the context of lung transplantation, bronchiolitis obliterans causing lung function impairment without histological confirmation has been called bronchiolitis obliterans syndrome (BOS). A standard definition of BOS was first developed in 1993122 based on changes in FEV₁ following transplantation. The definition has undergone subsequent revisions in 2002123 and in 2014124 when the over-arching term of chronic lung allograft dysfunction (CLAD) with subdivisions of restrictive CLAD (restrictive allograft syndrome, RAS) and obstructive CLAD (BOS) were proposed. BOS is common after lung transplantation occurring in over 50% of those surviving more than 5 years. It is less common after HSCT, between 2% and 14% in the first 5 years post transplant. Long-term macrolides have been used both to prevent BOS and to treat BOS when it occurs. Several observational studies have noted a positive response to azithromycin in a proportion of patients with BOS following lung transplantation, especially those with neutrophilic bronchoalveolar lavage results. This has been labelled neutrophilic reversible allograph dysfunction or azithromycin responsive allograph dysfunction distinguishing it from the non-reversible CLAD.

Evidence base

There is a significant body of literature relating to bronchiolitis obliterans occurring in the context of lung or HSCT. In the lung transplantation field, the majority of published studies are case series. There is marked variation between studies in patient groups, concomitant immunosuppressive regimens, choice of macrolide and dosing regimens, and choice of outcome measures making meaningful systematic analysis of the data impossible. In patients with bronchiolitis obliterans complicating HSCT, there are fewer studies, predominantly case series, also with a high degree of variation in populations, background immunosuppressive regimens, macrolide interventions (often in conjunction with other novel therapeutic agents) and outcome measures precluding meaningful systematic analysis.

Macrolides in BOS/CLAD following lung transplantation

An International Society for Heart and Lung Transplantation/American Thoracic Society (ATS)/European Respiratory Society (ERS) clinical practice guideline on the diagnosis and management of BOS was published in 2014 following the GRADE approach to develop specific clinical recommendations including the use of azithromycin.124 This guideline identified 10 studies (one observational study and nine case series) that described the effects of azithromycin on the lung function of lung transplant patients with BOS. It also identified two observational studies that described the effect of azithromycin on mortality in lung transplant patients with BOS. The guideline recommended a trial of azithromycin (250 mg orally daily for 5 days and then thrice weekly for a minimum of 3 months) for lung transplant recipients who developed a decline in FEV₁ consistent with the onset of BOS. The recommendation was conditional.

There have been two RCTs investigating the role of macrolides in BOS following lung transplantation. Vos et al examined the effect of azithromycin versus placebo for the prevention of BOS following lung transplantation.125 Corris et al studied the effect of azithromycin versus placebo in the treatment of patients who developed BOS following lung transplantation.126 In the Vos et al single-centre paper, patients who had undergone lung transplantation were randomised to receive either azithromycin, 250 mg three times a week (40 patients) or placebo (43 patients) for 2 years in a fully blinded study. Primary end points were freedom from BOS and survival at 2 years. If patients developed BOS on placebo, they were switched to open-label ‘rescue’ treatment with azithromycin. Patients receiving prophylactic azithromycin were less likely to develop BOS than those on placebo (12.5% vs 44.2%, p=0.0017). Overall survival was comparable between groups.125 This group of investigators have subsequently published a post hoc analysis of the same patients who were treated for a further year applying the newer CLAD classification system over an extended 7-year period of observation.127 The treated patients were less likely to develop CLAD at any stage compared with the placebo patients (28% vs 51%, p=0.043). There were also benefits in CLAD-free survival, long-term pulmonary function and functional exercise capacity, but graft loss (re-transplantation and mortality) was similar in both groups (53% vs 40%, p=0.27).127 In the Corris et al paper, 46 patients with BOS following lung transplantation were randomised to receive either azithromycin, 250 mg on alternate days (23 patients) or placebo (23 patients) over a 12-week study period in double-blind fashion. The primary end point was change in FEV₁ at 12 weeks. Patients who had a rapid and severe deterioration in lung function were withdrawn from the study and received open-labelled azithromycin. On an intention-to-treat (ITT) analysis, there was no significant difference between treatments in the primary outcome (0.035 L, p=0.6). For study completers (16 azithromycin and 17 placebo), there was a significant difference of 0.278 L (p≤0.001). In addition, 9/23 ITT patients in the azithromycin group had a ≥10% gain in FEV₁ from baseline while no patient in the placebo group had ≥10% gain in FEV₁ while on placebo (p=0.002).126 This latter observation, previously noted in other studies,128 suggests a subgroup of BOS patients who exhibit a response to azithromycin and is one of the drivers for recommending the re-classification of CLAD following lung transplantation noted above.

Evidence summary

The evidence for macrolides in BOS following lung transplantation covers both prophylactic use to prevent BOS and treatment following the occurrence of BOS. In both areas, there are single RCTs and a number of case series. The overall quality of the evidence is at best modest; however, BOS is a devastating complication of lung transplantation, so any intervention that offers the chance of prevention, reversal or stabilisation is welcome. Long-term macrolide use is a low-risk intervention. On this basis, we can make two recommendations. (Low)

Macrolides in BOS following HSCT

There are a number of case series and one RCT examining the role of macrolides (sometimes in combination with other emerging therapies) in BOS following HSCT. Khalid et al reported prospectively from Saudi Arabia on a series of eight HSCT patients with BOS diagnosed on the basis of lung function and high-resolution CT scanning who were treated with azithromycin for 12 weeks. All patients showed improvement in FEV₁ and all but one in FVC following treatment. The changes were statistically (FEV₁, p=0.0052; FVC, p=0.0067) and clinically (FEV₁ 21.6%, FVC 20.6%) significant.129 Norman et al described a retrospective, single-centre series of eight HSCT patients in North America with newly diagnosed BOS who were treated with a combination of fluticasone (inhaled), azithromycin and montelukast (the FAM regime) in addition to a more rapid taper in systemic corticosteroids than was normal practice. The intention was to reduce systemic corticosteroid exposure in these patients. Comparison with historical controls (HSCT patients with BOS treated with much higher overall doses of systemic steroids) suggested equivalent efficacy in maintaining lung function.130 Jo et al from South Korea, in a retrospective study, described the effect of prophylactic azithromycin in reducing the incidence of BOS in a series of 100 HSCT patients compared with over 1000 similar patients who did not receive prophylactic azithromycin. The incidence of BOS in those who received azithromycin was 12% compared with 6.4% in those who did not. Subsequent multivariate analysis did not suggest that prophylactic azithromycin was associated with the development of BOS.131 In a multicentre open-labelled prospective study, Williams et al examined the effect of the FAM regimen on new-onset BOS in HSCT patients. The primary end point was treatment failure by 3 months defined as an absolute decline in % predicted FEV₁ of > or =10%. Thirty-six patients from 10 centres were studied. Two patients (6%) had treatment failure compared with 40% of historical controls.132 The single-centre RCT from Lam et al in Honk Kong compared the effect of 12 weeks’ treatment with daily azithromycin, 250 mg, compared with placebo in 22 HSCT patients with BOS. Primary outcomes were changes in QOL (SGRQ) and lung function (FEV₁, FVC and FEF 25–75) measured monthly and 1 month after completion of treatment. The authors reported no change in lung function or SGRQ although the QOL data were analysed in an unusual fashion.133

Evidence summary

The literature for macrolides in BOS following HSCT is limited. There is only one RCT which has significant methodological limitations. The majority of case series are small and half are retrospective. There are differing diagnostic criteria for BOS and in some studies multiple drugs including azithromycin are used. Several authors comment on the difficulty of performing studies in this patient group where the condition is a rare complication of HSCT (making it difficult to gather high patient numbers) with devastating consequences (making it difficult to justify and recruit to a placebo limb). This situation is discussed in detail in an editorial.134 There is insufficient evidence for us to make a recommendation.

Macrolides for other causes of bronchiolitis obliterans

There are no published studies looking at macrolide treatment for other causes of bronchiolitis obliterans. There is a single case report of bronchiolitis obliterans following dust exposure on 9/11 with response to azithromycin135 and one recent review article suggested a possible role for macrolides in the treatment of bronchiolitis obliterans associated with rheumatoid arthritis.136

Cough

Chronic cough, defined as a cough lasting more than 8 weeks,137 is a very common symptom, with an estimated prevalence in total of 9.6% of the population, with variations according to geographical area from as high as 12.7% in Europe to 4.4% in Asia.138 It has been reported as the the most common symptom for which patients seek medical attention.139 It occurs most commonly in women in the fifth and sixth decades of life, and has been associated with significant impacts on physical and psychological health.140 141 The most common causes of chronic cough remain cough variant asthma, postnasal drip and gastro-oesophageal reflux disease, with around 40% remaining of ‘unknown cause’ after investigation.141

When an underlying cause of chronic cough is not identified, or when cough is resistant to prolonged treatment of the underlying cause, a number of treatments have been assessed to attempt to reduce cough frequency, reduce symptoms and improve QOL. These include randomised trials demonstrating the utility of low-dose opioids142 and gabapentin.143 Amitriptyline has been trialled (non-randomised) in cough with some success.144

Macrolides have been used in an attempt to reduce the symptoms or frequency of chronic cough. The rationale for their potential use in chronic cough includes impact on neutrophilic airways inflammation which is raised in some patients with chronic cough,145 and has been shown to be decreased with use of macrolides,34 101 and the potential anti-inflammatory effects of macrolides, independent of their antimicrobial actions.146

Evidence base

Organising pneumonia

Organising pneumonia (OP) is a clinicopathological syndrome associated with characteristic patterns on lung imaging and biopsy which represents an aberrant healing response to injury within the small airways and alveolar spaces. CT scanning typically shows patchy consolidation with air bronchograms, typically in subpleural locations, which may appear migratory on serial imaging. Ground glass change, reversed halo (atoll) sign, reticulation or nodules may also be seen. Histological features of OP include fibroblasts and inflammatory cells embedded in extracellular matrix within the small airways and alveoli, often forming polypoid masses known as Masson bodies. OP can occur in a variety of conditions including infection, connective tissue disease, malignancy, following radiotherapy, drug reactions and immunodeficiency. If no cause can be found on exhaustive testing the term cryptogenic organising pneumonia (COP) (previously called bronchiolitis obliterans organising pneumonia) is used. COP is a rare disease (annual incidence ~1:100 000) that presents with breathlessness and cough over weeks or months, often accompanied by fever, myalgia and elevated blood inflammatory markers. Lung function tests typically show restriction and impaired gas transfer.

The prognosis of COP is good, particularly when consolidation is the primary pattern on CT. The natural history is spontaneous remission in many untreated cases.149 In non-remitting or progressive COP, oral corticosteroids have been used often in reported case series.150 There have been no placebo-controlled trials of steroid therapy, but is believed that there is an autoimmune aetiology in a proportion of cases and that steroids speed resolution. In steroid-responsive cases, relapse may occur after stopping therapy.

Diffuse panbronchiolitis

Diffuse panbronchiolitis (DPB) was defined as a specific clinicopathological entity in the 1960s, being characterised histologically by multiple micronodular lesions consisting of chronic inflammatory cells infiltrating bronchiolar walls.163 164 It has almost exclusively been reported in East Asian countries, notably Japan, but also others such as China and South Korea.163 Relatively few cases have been diagnosed in Caucasian and other populations.163 165 166

DPB was one of the first respiratory conditions for which it was identified that a regular macrolide, erythromycin, at a lower dose than usually used in therapy, could be associated with notably improved outcomes for sufferers. This was discovered in the early 1980s, although there has been a paucity of published rigorous and well-designed placebo-controlled trials subsequently to test the hypothesis, with further supporting evidence largely in the form of (usually small) case series and observational studies.163 167 The largest published study was retrospective, involving nearly 500 Japanese patients with DPB.168 These patients were divided into three cohorts on the basis of likely date of diagnosis, which were then compared with respect to outcome. There was a statistically significant survival advantage (p<0.0001) to being in the third cohort and thereby diagnosed between 1985 and 1990. This was ascribed as being directly related to when the beneficial role of low-dose erythromycin in DPB had first been reported.168 To provide support for this, the records of this third cohort of 87 patients were analysed in more detail—and there was a significantly higher survival rate at 5 years of those who were known to have received erythromycin compared with those who had not (p=0.01).168 Moreover, those who had not received erythromycin had a similar 5-year survival rate to those in the first cohort, who had been diagnosed between 1970 and 1979, and therefore before the first reports of benefit with such a strategy.163 168 DPB case series studies have also shown the effect of low-dose macrolide therapy in controlling clinical symptoms, such as by reducing sputum production, and objectively improving lung function.167

The benefits of using low-dose macrolide therapy in the setting of DPB have been endorsed by clinical guidelines developed in the Far East.163 However, due to the current rarity of this condition in the UK, the GDG felt that it was not warranted for BTS to conduct a comprehensive evidence-based review and develop specific recommendations regarding the role of macrolides in managing DPB.

Gastrointestinal effects

Gastrointestinal (GI) side effects of macrolides have been recognised since the introduction of erythromycin in 1952. Although initially thought to be due to alterations in gut flora, it is now clear that macrolides act through interactions with motilin receptors in the gut.169 170 In humans, macrolides potentiate gastric and small bowel motility,171–175 increase lower oesophageal sphincter pressure,176 177 and influence colonic transit and gall bladder function.178

The prokinetic effects of macrolides on the GI tract has led to their therapeutic use in conditions characterised by reduced GI motility including diabetic gastroparesis,179 anorexia nervosa,180 colonic pseudo-obstruction, postoperative ileus181 and in critical care patients.182

GI symptoms are the most commonly reported side effects of macrolides occurring in up to 70% of patients taking erythromycin,183 less commonly with clarithromycin and azithromycin. Symptoms include nausea, vomiting, abdominal pain, diarrhoea and anorexia. Commenting on 11 studies of long-term, low-dose macrolide treatment in various chronic respiratory diseases in nearly 1000 children and adults, Altenburg et al 23 found that mild to moderate GI complaints had been reported which hardly ever caused study drug discontinuation.

Thus, although GI symptoms are common and predictable, they are rarely sufficiently troublesome to necessitate discontinuation of therapy. Dose reduction may improve tolerability.

Cardiac effects

Macrolide antibiotics, along with a wide range of other pharmacological agents, have the potential to interfere with conduction in cardiac tissue and hence to cause dangerous cardiac arrhythmias and sudden cardiac death.184 Macrolides can both directly prolong the QT interval185 and also inhibit the metabolism of other proarrhythmogenic drugs by acting on cytochrome P450 in the liver.

In recent years, there have been reports of an increased risk of cardiovascular death in patients treated with short, oral antibacterial courses of azithromycin110 and of no increased risk of death in a younger group of similarly treated patients.186 Discussion of these seemingly contradictory findings commented that the group of patients with increased risk exhibited a high level of baseline risk and comorbidity.187 Factors increasing the risk of malignant arrhythmias with macrolides are well recognised and include age >80, female gender, heart disease, use of other QT prolonging medication, reduced drug elimination, bradycardia, prolonged QT interval before therapy and genetic predisposition.188 These events are rare; 85 deaths for each 1 million courses prescribed in the increased risk group described above,110 so it is not surprising that such events do not appear in the literature of long-term, low-dose macrolide use.

A helpful review article for the American COPD Clinical Research Network describes a pragmatic approach to risk stratification when considering long-term, low-dose macrolide therapy.189 The authors combined history taking targeting evidence of heart failure, episodes of hypokalaemia, a family history of long QT syndrome, or use of other medications known to prolong the QT interval together with a pre-treatment ECG and an ECG 1 month after starting therapy. With this approach in a COPD population entering a macrolide trial, 7.1% were excluded on history and 1.6% because of a prolonged baseline QTc. A further 0.8% were withdrawn after developing QTc prolongation with treatment or on placebo. Using such a strategy, exclusions in asthma and bronchiectasis populations are likely to be fewer.

Ototoxicity

Macrolide ototoxicity is well recognised usually causing a dose-dependent, reversible, sensorineural hearing loss (SNHL).23 The only systematic review of this topic was published recently190 including data from three prospective studies. In these studies including 93 patients, 14 (15%) developed SNHL confirmed by audiometry following exposure to macrolide antibiotics. All but one recovered on cessation of the drug. Earlier studies of long-term macrolide use for the treatment of mycobacterial infections have also reported reversibility of ototoxicity following either withdrawal191 192 or dose reduction193 of the macrolide. Albert reported hearing decrements in 25% of a COPD population (vs 20% in the placebo group) with normal hearing at the outset of 1 year of low-dose azithromycin.93 Hearing subsequently improved in 34% who discontinued treatment, but also in 32% who continued due to protocol errors. Similar proportions of placebo-associated hearing loss also improved with or without drug withdrawal. The authors of this study felt in discussion that their criteria for defining hearing loss were too stringent and that the incidence of hearing decrements were overestimated.

Overall, the literature suggests that for low-dose, long-term macrolide use the incidence of ototoxicity is rare and almost always reversible. It would seem sensible to assess potential recipients for hearing impairment or difficulties with balance prior to initiating therapy.

Other side effects

A low rate (1%–5%) of asymptomatic elevation of serum aminotransaminase levels is known to occur with any of the four orally absorbed macrolide antibiotics. The elevation is generally mild to moderate in degree and rarely requires dose modification or discontinuation. More seriously, a cholestatic picture can occur which carries a higher risk of permanent liver damage.194 195 Monitoring liver function at the start of therapy seems a sensible precaution.

Prescribing off-label

It is important to be aware that none of the macrolide antibiotics have a product license in the UK for the indication of long-term, low-dose usage as immunomodulatory agents. Their prescription in this context is thus off label. Clinicians should make patients aware of this and follow GMC guidance in this area which states that “decisions should be made in collaboration with the patient by discussing the options with them and ensuring that they have sufficient information about the medicine to allow them to make an informed decision”. An example of an information leaflet used to support such discussions is shown in online appendix 4.

Good practice points

• The risks associated with increasing antimicrobial resistance should be discussed with patients prior to starting low dose macrolide therapy. Patients should understand the risk that there may not be an effective antibiotic for them, or someone else, when needed in the future.

• Prior to initiating low-dose macrolide monotherapy, patients should be asked if they have a history of previous or current NTM infection or disease. Current NTM infection should be managed with reference to BTS guidance and precludes low-dose macrolide monotherapy. Successfully treated NTM disease should not preclude low-dose macrolide monotherapy.

• If there is any clinical suspicion of possible NTM disease, patients should be screened via examination of sputum samples prior to starting therapy. If positive for recognised potential pathogenic species, low-dose macrolide prophylaxis is contraindicated.