The Immunomodulatory Effects of Macrolides—A Systematic Review of the Underlying Mechanisms

Background The mechanisms underlying the non-antimicrobial immunomodulatory properties of macrolides are not well understood. Objectives To systematically review the evidence for the immunomodulatory properties of macrolides in humans and to describe the underlying mechanism and extent of their influence on the innate and adaptive immune system. Methods A systematic literature search was done in MEDLINE using the OVID interface from 1946 to December 2016 according to the preferred reporting items for systematic reviews and meta-analysis (PRISMA). Original articles investigating the influence of four macrolides (azithromycin, clarithromycin, erythromycin, and roxithromycin) on immunological markers in humans were included. Results We identified 22 randomized, controlled trials, 16 prospective cohort studies, and 8 case–control studies investigating 47 different immunological markers (186 measurements) in 1,834 participants. The most frequently reported outcomes were a decrease in the number of neutrophils, and the concentrations of neutrophil elastase, interleukin (IL)-8, IL-6, IL-1beta, tumor necrosis factor (TNF)-alpha, eosinophilic cationic protein, and matrix metalloproteinase 9. Inhibition of neutrophil function was reported more frequently than eosinophil function. A decrease in T helper (Th) 2 cells cytokines (IL-4, IL-5, IL-6) was reported more frequently than a decrease in Th1 cytokines (IL-2, INF-gamma). Conclusion Macrolides influence a broad range of immunological mechanisms resulting in immunomodulatory effects. To optimize the treatment of chronic inflammatory diseases by macrolides, further studies are necessary, particularly comparing different macrolides and dose effect relationships.

The antimicrobial activity of macrolides results from inhibition of bacterial protein synthesis through reversible binding to the peptide exit tunnel of ribosomes (4).
In addition to their antibiotic activity, macrolides have immunomodulatory properties, which were first described soon after their introduction in the 1950s (3,(5)(6)(7). The concept of using macrolides primarily for their immunomodulatory activities was introduced in the 1970s (8). The seminal study that distinguished between macrolides' antimicrobial and their immunomodulatory effects was in adults with diffuse panbronchiolitis (DPB) in whom treatment with ERM dramatically improved survival independent of bacterial colonization (9). These results encouraged further research on the use of macrolides for the treatment of other chronic inflammatory conditions (10)(11)(12)(13)(14).
The mechanisms underlying the non-antimicrobial effects of macrolides are less well understood. Aside from ribosomalmediated inhibition of pathogen virulence factor production, a number of other mechanisms have been proposed, including action on host immunity.
The objective of this review was to systematically summarize studies which investigated immunomodulatory properties of macrolides in humans and to describe the underlying mechanism and extent of their influence on the innate and adaptive immune system.

metHODS
This review was done according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) (15). A literature review was done in December 2016 searching MEDLINE using the OVID interface from 1946 to 2016 using the search terms: (macrolide OR azithromycin OR clarithromycin OR erythromycin OR roxithromycin) AND (anti-inflammatory OR immunomodulatory OR immunolides) without any language limitations or limitation of study design (Figure 1). Only studies in humans, in which the participants received one of the four mentioned macrolides and which investigated immunological markers involved in inflammation were included. Studies reporting clinical endpoints only or studies in which macrolides were investigated for their antimicrobial activity were excluded. References were hand-searched for additional publications. Search results were independently screened by one reviewer, and checked by a second reviewer. Potentially eligible full-text articles were assessed according to our inclusion and exclusion criteria. The following variables were extracted from the included studies: year of study, country, study design, number of participants, age of participants, underlying disease, type, dose and duration of macrolide use, type of samples collected, and measured immune markers. Changes were classified as being significant when the p-value was ≤0.05.

immunomodulatory Properties in Different Diseases
In the following, the immunomodulatory properties of macrolides are summarized and categorized by the disease in which they were investigated ( Table 1).

Blepharitis
Blepharitis is a common chronic inflammation of the eye lid leading to dry, itchy, and erythematous eyes. Anterior blepharitis is often associated with bacterial infections, while posterior blepharitis is linked to dysfunction of Meibomian glands. Many studies report clinical improvement in patients with blepharitis treated with topical AZM, due to a decrease in secretions and plugging of the Meibomian glands but did not investigate the underlying immunological mechanisms. The one study which did investigate immunological markes shows that concentrations of IL-1beta, IL-8, and MMP-9 in conjunctival cells of patients with blepharitis are higher than in healthy controls (16). Concentrations of these cytokines decrease with local AZM treatment, but return to pre-treatment levels after discontinuation (16).

Periodontitis
Periodontitis is an inflammatory process of the gums with a complex pathogenesis including microorganisms as well as neutrophils, macrophages and fibroblasts. One key immunological mechanism underlying the pathogensis of periodontitis has been described as a TNF-alpha-induced increase in vascular endothelial growth factor (VEGF) leading to an aberrant angiogenesis (61). Both AZM and RXM decreased TNF-alpha and VEGF concentrations as well as other cytokines including IL-1beta, IL-8, and transforming growth factor beta (TGF-beta) in gingival crevicular fluid (17,18). Since oral bacteria play an important role in periodontitis, however, some of the some of the benefits of macrolides may be attributable to antimicrobial rather than to immunomodulatory effects.

chronic Rhinosinusitis and Nasal Polyposis
Chronic rhinosinusitis (CRS) with nasal polyps is characterized by a T helper (Th) 2 cells-dominated inflammation with upregulation of IL-4, IL-5, and IL-13 and an increase in eosinophil count, ECP, and immunoglobulin E. CRS without nasal polyps is characterized by Th1-dominated inflammation with upregulation of IL-2, TGF-beta, and IFN-gamma. Studies in patients with CRS treated with CAM and RXM show a significant reduction in macrophage, neutrophil, and eosinophil counts and concentrations of neutrophil elastase, ECP, CC-chemokine ligand-5 (CCL-5), IL-1beta, IL-6, IL-8, interferon (IFN)-gamma, TNF-alpha, myeloperoxidase (MPO), and alpha-macroglobulin in nasal secretions (19-23, 25, 26, 62). One of the postulated mechanisms by which macrolides inhibit the development of nasal polyps is through their anti-oxidative effects inhibiting the TGF-beta-induced production of reactive oxygen species (24). However, the immunomodulatory mechanisms differ in allergic and non-allergic nasal polyposis patients. While CAM reduces IL-6 and CCL-5 in all patients, it reduces IL-1beta and IL-6 only in patients with allergic CRS and TNF-alpha and ECP only in patients with non-allergic CRS (19,20). asthma Asthma is characterized by chronic airway inflammation, reversible airway obstruction, and airway hyper-responsiveness. In eosinophilic asthma, eosinophils, mast cells, and Th2mediated inflammation play an important role. Concentrations of IL-4, IL-5, IL-6, IL-9, IL-10, IL-13, vascular cell adhesion molecule-1, CC chemokines, and granulocyte macrophage colony-stimulating factor (GM-CSF) are elevated. In severe asthma, in addition to eosinophils, increased neutrophils and

Bronchiectasis
Bronchiectasis is characterized by permanent enlargement of bronchi and cytokines play an important role in the pathogenesis. In BAL samples of patients with bronchiectasis, elevated concentrations of IL-1beta and IL-8, as well as Th17-cytokines     (IL-17A and IL-23), are found. In this setting, CAM and RXM lead to a decrease in total cell and neutrophil counts, concentrations of neutrophil elastase, IL-8, and MMP-9 in sputum or BAL of patients with bronchiectasis (38,39). Interestingly, in BAL samples, these drugs significantly increase macrophage counts (38). Furthermore, macrolides lead to a decrease in peripheral blood Th17 cells and IL-17 concentrations (37).

chronic Obstructive Pulmonary Disease
Chronic obstructive pulmonary disease (COPD) is characterized by chronic inflammation of lung parenchyma and peripheral airways with an increase in alveolar macrophages, neutrophils, T cells (predominantly Th1-, and Th17-cells), and innate lymphoid cells. These cells, as well as structural cells, such as epithelial cells, endothelial cells, and fibroblasts, secrete a variety of pro-inflammatory cytokines. Although most patients with COPD have a predominantly neutrophilic inflammation, some also have elevated eosinophil counts in sputum. Oxidative stress plays a key role in COPD, and can result in activation of the pro-inflammatory transcription factor nuclear factor (NF)-kappaB. Moreover, COPD is associated with increased apoptosis and defective phagocytosis in the airways. In patients with COPD, IL-1beta, IL-4, IL-8, and TNF-alpha concentrations in blood are elevated, while IL-10 concentrations are lower compared to healthy adults. In patients with COPD, AZM leads to a decrease in white blood cell and platelet counts and concentrations of CRP, IL-8, E-selectin, and lactoferrin in blood (40). By contrast, macrolides increase neutrophil oxidative burst and neutrophil glutathione peroxidase activity in blood (40). In the sputum of COPD patients, CAM and ERM lead to a significant decrease in total cell and neutrophil count and inhibit neutrophil chemotaxis and decrease concentrations of neutrophil elastase (41,42).

Diffuse Panbronchiolitis
Diffuse panbronchiolitis (DPB) is a chronic distal airway inflammation characterized by diffuse micronodular pulmonary lesions mostly consisting of neutrophils. Neutrophils and epithelial cells produce IL-8, which is an important chemotactic factor to attract more neutrophils. The neutrophil count in BAL samples of patients with DPB correlates to the concentrations of IL-1beta and IL-8 (48). ERM reduces IL-1beta concentrations in BAL samples of patients with DPB which leads to a subsequent reduction of IL-8 concentrations and a decrease in neutrophil count and neutrophil chemotactic activity (44)(45)(46)(47)(48)63). Furthermore, ERM treatment also results in a decrease in lymphocyte count, IL-2, interferon-gamma, and to increase in CD4/CD8 ratio, IL-4, IL-5, IL-13 in BAL samples of patients with DPB (43).

cystic Fibrosis
In patients with cystic fibrosis (CF), chronic airway inflammation results from cytokines secreted by epithelial and immune cells, which leads to neutrophil influx into airways. The release of neutrophil proteases, including neutrophil elastase, contributes to the development of bronchiectasis. Sustained inflammation is mainly due to an increase in the transcription of NF-kappaB activity, which leads to an increase in IL-8 production. These immunological mechanisms are influenced by AZM and CAM, which in CF-patients lead to a decrease in neutrophil count, concentrations of neutrophil elastase, IL-4, IL-8, TNF-alpha, and INF-gamma, and to an increase in numbers of macrophages in BAL samples or in sputum (51)(52)(53). In CF-patients macrolides also lead to a decrease in neutrophil count, concentrations of IL-4, IL-8, TNF-alpha, MPO, high-sensitivity C reactive protein, serum amyloid A, and calprotectin in blood (50,53).

DiScUSSiON
Macrolides are important therapeutic options in the treatment of many chronic inflammatory diseases because of their immunomodulatory effects. To understand the mechanisms underlying these effects, we reviewed all human studies that analyzed the influence of macrolides on immunological markers. The non-antimicrobial effects of macrolides are extensive and range from changes in cell counts and function, up-and downregulation of cytokine production to expression of adhesion molecules. The most frequently and consistently reported immunomodulatory effect of macrolides is a reduced neutrophilic inflammation. Reduced numbers of neutrophils and inhibition of neutrophilic function lead to lower concentrations of neutrophil elastase and IL-8, and ultimately to a decrease in tissue injury. Furthermore, macrolides also reduce IL-1beta concentrations, another key mediator of the inflammatory response that is most abundantly produced by monocytes and macrophages. Evidence from animal and in vitro studies show that the inhibition of the key pro-inflammatory cytokines IL-8 and IL-1beta results from macrolides' ability to alter intracellular signaling, particularly through the inhibition of NF-kappaB activation and expression of activator protein-1 (64)(65)(66). Notably, this effect has been observed in the absence of an infectious agent.
On the basis of these observed in vitro immunological effects of macrolides, patients with diseases mediated by neutrophilic inflammation such as periodontitis, severe asthma, DPB, bronchiectasis, COPD, and CF should benefit from treatment with this class of antibiotics. Indeed, clinically beneficial effects have been shown in randomized controlled studies in patients with COPD and CF with improved symptom scores, respiratory function and decreased frequency of exacerbations (67)(68)(69). For DPB, bronchiectasis and asthma, however, there is an absence of randomized controlled studies showing clinical beneficial effects of macrolides (70)(71)(72).
Macrolides are more commonly and consistently reported to inhibit neutrophilic than eosinophilic function. This is consistent with clinical studies that show patients with eosinophil-driven chronic inflammatory diseases associated with increased IgE (such as CRS or atopic asthma) have significantly lower improvement rates with macrolide treatment than those with normal serum IgE (26,62,73). Although the effect of macrolides on eosinophils has been less commonly investigated, a few studies report decreased eosinophil counts, and concentration of ECP (a ribonuclease secreted by eosinophils responsible for local cytotoxic effect). This suggests that there may be a role for the use of macrolides in allergic chronic inflammatory diseases (43,74,75). The possible influence of macrolides on eosinophilic inflammation is further supported by the finding that Th2 cytokines, such as IL-4 and IL-5, are more frequently reduced than Th1 cytokines, such as IL-2 and INF-gamma (19,20,22,32,42,43,53,55,56). The stronger effect of macrolides on Th2 compared with Th1 responses is further supported by evidence from animal and in vitro studies (74,75). However, some of the anti-inflammatory effects might also be explained through their antibiotic effect on (undiagnosed) pathogens which trigger and sustain inflammation.
It is likely that immunomodulatory effects vary between different macrolides. Although some studies included more than one macrolide, none of the human studies directly compared different macrolides. Interestingly, AZM was less frequently associated with changes in measured immunological markers compared to the other macrolides. However, most of these studies were either in healthy volunteers or AZM was administered for only a few days (56)(57)(58)76). By contrast, clinical studies in patients with CF suggest that AZM, but not CAM, leads to an improvement in respiratory function and reduction in pulmonary exacerbations (69,77). In vitro studies comparing the immunomodulatory effects of different macrolides suggest that CAM has less immunomodulatory activity compared to other macrolides. For example, RXM, but not CAM or ERM, was shown to decreased chemotaxis of Th1 and Th2 cells (78). Similarly, CAM had a significantly weaker effect on reducing IL-6 production by human macrophages than ERM (79). Furthermore, another study showed that AZM, but not CAM or RXM, inhibits IL-1alpha and IL-1beta production (80).
Immunomodulatory effects of macrolides have been described with the recommended dose for antimicrobial treatment. Macrolides have excellent tissue penetration compared to other classes of antibiotics resulting in tissue concentrations generally exceeding serum concentrations (except for RXM). For the immunomodulatory effects macrolides' ability to accumulate in neutrophils and macrophages is particularly important. Concentrations in macrophages have been shown to be 400-to 800-fold higher compared to serum for CAM and AZM and 5-to 100-fold higher in tissue compared to serum for ERM, CAM and AZM (81)(82)(83)(84)(85). This drug accumulation in immune cells may result in immunomodulatory effects occurring at lower doses and lasting longer compared to the antimicrobial effects. The relationship between macrolide dose and immunomodulatory effect is, therefore, an interesting avenue for future research.
The main limitation of this review is the heterogeneity of study populations, underlying diseases, type of macrolide and methods used to assess the immunomodulatory effect. A further limitation is selection and reporting bias and based on study types other biases including carry-over effect in cross-over trials and recall bias in case-control studies.
In summary, there is substantial evidence that macrolides exhibit immunomodulatory effects through inhibition of neutrophilic inflammation and macrophage activation. However, there is considerable heterogeneity between studies and in the immunological markers measured. Further studies will help delineate the exact mechanisms underlying the immunomodulatory properties of macrolides and the relative activity of different macrolides. This will enable the optimal use of this class of antibiotics in the treatment of chronic inflammatory diseases.

aUtHOR cONtRiBUtiONS
PZ and NC designed the study. PZ drafted the initial manuscript and approved the final manuscript as submitted. PZ, VZ, and NR did the risk of bias analysis. VZ, NC, and NR critically reviewed and revised the manuscript, and approved the final manuscript as submitted.