- Split View
-
Views
-
Cite
Cite
G. W. Amsden, Anti-inflammatory effects of macrolides—an underappreciated benefit in the treatment of community-acquired respiratory tract infections and chronic inflammatory pulmonary conditions?, Journal of Antimicrobial Chemotherapy, Volume 55, Issue 1, January 2005, Pages 10–21, https://doi.org/10.1093/jac/dkh519
- Share Icon Share
Abstract
Background: It has been recognized for more than 20 years that the macrolides have immunomodulatory effects that are beneficial for those suffering from chronic pulmonary inflammatory syndromes, such as diffuse panbronchiolitis, cystic fibrosis, asthma and bronchiectasis. The macrolides have consistently been associated with decreased length of stay and mortality when used alone or in combination with β-lactam antibiotics. This effect can be demonstrated against combinations consisting of β-lactams and other antibiotics active against ‘atypical chest pathogens’ when treating community-acquired pneumonia (CAP) in hospitalized patients. As such, it appears that the macrolides' effects in CAP patients are more than just antibacterial in nature.
Aims of this review: This review aims: to give the reader information on the background areas described, as well as related areas; to review the CAP benefits with macrolides and how they may be related to the immunomodulatory properties they demonstrate, albeit in a shorter period of time than previously demonstrated with chronic pulmonary disorders; to use ex vivo data to support these extrapolations.
Literature search: A literature search using Medline was conducted from 1966 onwards, searching for articles with relevant key words such as macrolide, diffuse panbronchiolitis, community-acquired pneumonia, biofilm, immunomodulation, cystic fibrosis, erythromycin, clarithromycin, roxithromycin and azithromycin, bronchiectasis and asthma. When appropriate, additional references were found from the bibliographies of identified papers of interest. Any relevant scientific conference proceedings or medical texts were checked when necessary.
Conclusions: (1) Research into macrolide immunomodulation for chronic pulmonary disorders demonstrates consistent positive effects, although of types other than seen with diffuse panbronchiolitis. These effects, together with their inhibitory activity on biofilms, have the potential to make them a useful option. (2) The benefits for CAP are consistent, and higher when a macrolide is given with another atypical agent than if the other atypical agent is given alone, suggesting a non-antibacterial benefit. (3) Recent research of the immunomodulatory properties of azithromycin imply that azithromycin may have a previously unknown short-term biphasic effect on inflammation modulation: enhancement of host defence mechanisms shortly after initial administration followed by curtailment of local infection/inflammation in the following period. (4) Additional in vivo research is needed prior to developing any firm conclusions.
Macrolide benefits beyond antimicrobial activity: early observations
Although macrolides have been used predominantly for their antimicrobial activity over the past 50 years, a lesser known application, which has been modest for ∼2–3 decades, is becoming more mainstream, in both research and use. In the early 1980s, it was discovered that chronic treatment with erythromycin resulted in dramatically improved 5 year survival rates in patients suffering from the chronic inflammatory pulmonary disease, diffuse panbronchiolitis (DPB). Prior to this discovery, the 5 year survival rate of DPB patients was ∼63%, which decreased to ∼8% when the disease progressed to a point where patients became colonized with Pseudomonas aeruginosa.1 Once the utility of erythromycin in the treatment of DPB became known, it resulted in its becoming common practice to place patients on it daily, long-term. Five year survival rates subsequently rose to 92%.1–5 These advances occurred in patients where the disease had improved without eliminating bacteria, including cases where P. aeruginosa had been involved. This fact, and the low sputum concentrations achieved with the low-dose, long-term therapies, encouraged many investigators at the time to focus their research on erythromycin's anti-inflammatory effects—as the mechanism behind its success in DPB.1 In the following review, the immunomodulatory properties of macrolides in chronic pulmonary inflammatory disorders will be reviewed. In addition, there will be discussions on how these results provide the scientific basis for the macrolides in acute inflammatory conditions, including pulmonary infections such as community-acquired pneumonia (CAP).
Immunomodulation by macrolides of chronic inflammatory pulmonary disorders
DPB
DPB is a chronic, obstructive pulmonary disease that is found almost exclusively in Japan. At presentation, patients have dyspnoea upon exertion, a productive cough, wheezing, weight loss and potentially sputum that is colonized with one or more pathogens. They also typically have a history of several years of chronic sinusitis. Prior to discovering the benefits of chronic low-dose erythromycin therapy (which was recently reviewed in this journal in detail)6 the 5 year mortality from DPB—once a patient became colonized with P. aeruginosa—was >90%. The non-antibacterial benefits from 14- and 15-membered macrolides on an inflamed DPB pulmonary system appear to be multifaceted. First, animal studies demonstrated that chronic erythromycin dosing resulted in decreased sputum/mucus production in a dose-dependent fashion. This occurred secondary to the binding of the macrolide to the epithelial cell chloride channels—which in turn blocked the channels—plus the inhibition of water secretion that moves with the chloride ions across the cell membrane.7,8 This blockade results in a decrease of hypersecretion and has been demonstrated to occur in humans during chronic lower and upper airway inflammatory conditions when clarithromycin is administered as a long- (8 weeks) and short-term (7 days) regimen.9–11
There is no doubt that decreased secretions would result in a better quality of life (QOL) for patients with these chronic respiratory diseases, especially those with high secretion output volumes. However, the ability to decrease the ceaseless inflammatory processes that perpetuate the ongoing pulmonary damage would also be advantageous. This has been demonstrated for macrolides during investigations into their mechanism of action in DPB patients, as detailed in Table 112,13 and a recent review article.6 As noted from the results of the two DPB studies described in Table 1 as well as the 17 others described in the recent review, no matter which macrolide (erythromycin, clarithromycin, roxithromycin and azithromycin) was utilized for DPB treatment the impact they had on the pulmonary inflammatory process was very consistent. When administered chronically in low doses, the macrolides are able to suppress the overabundance of neutrophils (PMNs) present in the lungs. This is effected by reducing their chemotaxis to the lungs by diminishing the responsible cytokines [i.e. interleukin (IL)-8]. As such, months after the start of macrolide therapy, a sputum differential would be more likely to be consistent with that of a healthy volunteer or of a patient with non-inflammatory pulmonary disease, in whom there is always a higher percentage of alveolar macrophages (AMs) than of PMNs. The suppression of the acute inflammatory process is likely to result in other secondary benefits, such as improved pulmonary function and decreased incidence of acute exacerbations, both of which would be a significant clinical improvement for DPB patients. However, there is inconsistence as to when these effects occur upon starting macrolide therapy: the studies in Table 1 and the review note highly variable times to onset of clinical benefit—anywhere from a few months to 16 months.6 Although it would be advantageous to identify the source of this variance, this is less pressing, as long as the population consistently benefits from the macrolides some time after commencing treatment.
Other inflammatory pulmonary disorders
Inspired by the encouraging results of the DPB studies, other pulmonary inflammatory conditions, such as asthma, bronchiectasis, bronchiolitis obliterans syndrome and cystic fibrosis (see Table 1 for summary of the most relevant studies) have undergone chronic macrolide therapy trials. The design and outcomes of these trials have varied significantly from the DPB trials that inspired them.
Asthma
As an example, the study investigating the impact of clarithromycin on mild to moderate bronchial asthma not only had no healthy volunteers as a control, but had a defined dosing period of 8 weeks rather than ‘dosing to effect’, as had been the method in the DPB studies.14 Whereas the lack of healthy volunteers is unlikely to be important, the discontinuation of dosing at 8 weeks may have been shortsighted, since it took between 5–16 months in the DPB studies for clinical end-points to be reached. Nevertheless, there were significant decreases in: serum and sputum eosinophil counts; eosinophilic cationic protein; patient symptom score (based on incidence of asthma attacks, amount of disability during attacks and amount of nocturnal asthma) (P < 0.05); and increases in PC20. The fact that there were no changes in pulmonary function tests may be merely a function of the mild to moderate severity of the asthma of patients in the study, as opposed to a more severe obstructive form, in which there may be more likelihood of change. The authors' conclusions were that beyond clarithromycin's antibacterial effects, there were also anti-asthmatic effects not related to bronchodilation. There was also an eosinopenic effect, through a mechanism that is probably related to eosinophil cytokine expression.14 However, one thing cannot be ignored: like the DPB trials, this study has too small a sample size. This means that more research is necessary before this treatment modality can be accepted widely.
Bronchiectasis
Studies investigating prolonged administration of low-dose macrolides in the treatment of bronchiectasis had contrasting designs as compared to the DPB studies, as well as amongst each other. This subsequently resulted in the outcomes contrasting to those of the DPB studies, as well as each other.15,16 Whereas the first was a study of bronchietatic children with increased airway responsiveness (AR), the latter study was of adult patients with idiopathic bronchiectasis. Whereas the DPB patients were dosed with macrolides until clinical response, the children were dosed with roxithromycin for 12 weeks and the adults with erythromycin for 8 weeks. Although the DPB patients underwent baseline and follow-up BALs (where BAL stands for bronchoalveolar lavage) to collect respiratory secretion samples for white blood cell (WBC) differentials and cytokine measurements, both bronchiectasis studies used either expectorated or induced sputum samples. Even though expectorated sputum in this case should be reflective of the diseased pulmonary area, the BAL would have been definitive proof and should have been an option in case the patient was not productive or serially produced contaminated samples. The fact that there was no change in pathogen, WBC or cytokine densities in the adult study leads one to wonder whether their results would have differed if more disease-specific specimens—taken by means of BAL—had been analysed. Although the non-DPB studies have been investigating standard pulmonary function test (PFT) results before and after initiating treatment to discover potential differences, results have been very mixed, as was seen with these two studies, which had contrasting outcomes. However, in addition to PFTs, airway responsiveness testing was conducted in these more recent trials. As such, even though, as with the paediatric trial, the PFT changes may be insignificant, there may be significant AR decreases. These were measured by high-dose methacholine challenge tests before and after treatment. The maximal response and provocative cumulative dose of methacholine that produced a 20% fall in FEV1 were used as the two indices of AR.15 The data are strong and promising and should lead to additional research in this population with this class of drugs. However, these studies also suffer from being too small.
Bronchiolitis obliterans syndrome
In an open label pilot study, azithromycin 250 mg was administered three times a week to six lung transplant patients with stage 1 or greater bronchiolitis obliterans syndrome (BOS) to see if pulmonary function improved or not.17 After an average of ∼14 weeks, patients had a mean increase of 17% (P ≤ 0.05) in FEV1, as compared with baseline, together with an absolute FEV1 increase of 0.50 L on average. The authors concluded that there was a potential role for macrolides as part of the maintenance therapy in lung transplant patients with BOS.17 The ability to discover a significant change with so few patients would seem unlikely, especially with PFTs over such a relatively short period of time. In this case, it is apparent that the macrolide was producing the effect, rather than other concurrent medication, because azithromycin was the only new medication. For a regular medication to have caused an improvement in PFTs, there would have to have been an increase in dose or exposure to it, such as would happen in a drug interaction. However, since azithromycin lacks any significant drug interactions—including against the anti-rejection agents the patients were receiving—it seems highly unlikely the improvement was due to anything except azithromycin. Unfortunately, as a pilot study, markers of pulmonary inflammation were not conducted at baseline and repeated throughout to try to correlate the change in PFTs with a decrease in pulmonary inflammation or other physiological change(s). However, these six patients provide a stimulus for additional work with this diagnosis.
Cystic fibrosis
As seen in Table 1 and the recent review mentioned earlier,6 larger trials have been conducted in cystic fibrosis (CF). To date, three relatively large trials have been conducted: one in children (n=41),18 one in adults (n=60)19 and one in both age groups (n=185).20 Each of these trials utilized azithromycin, as this was most likely to minimize any drug–drug interactions with the numerous concurrent drugs administered in CF. As noted in Table 1, all three of the studies had the same basic outcomes: decreased incidence of antibiotic requiring exacerbations and improvement or stabilization of PFTs while on azithromycin, as compared with worsening of both while on placebo. As the mixed age group study demonstrated, daily dosing with azithromycin did not appear to be necessary regardless of age. The study showed that by only giving azithromycin three times a week the results were the same as the other CF trials.18–20 Further study is required to find out whether an even smaller regimen would obtain the same results.
Chronic airway inflammation, P. aeruginosa and biofilms
What most of the above chronic inflammatory pulmonary conditions have in common is that the patients suffer from frequent infectious exacerbations. Initial infections may be caused by repeatedly acquired common community-acquired respiratory pathogens, but the subsequent bronchiolar damage becomes increasingly worse. As a result, patients become persistently infected, or colonized with mucoid strains of P. aeruginosa.6,21 This colonization leads to chronic inflammation and further pulmonary deterioration.21,22 Mucoid strains hyperproduce alginate and exist as a biofilm, coating both natural and introduced foreign material airway surfaces. Within these biofilms, the P. aeruginosa isolates are immune to any antibiotic treatment as well as to the destructive effects of the immune system. Once formed, the clinical progress of these isolates takes two forms. First, it can cause an acute invasive pulmonary infection when bacteria are released from the biofilm when the conditions are favourable to them. In this instance, whereas systemic antibiotics will cause a temporary improvement of clinical symptoms due to lysis of these ‘floating bacteria’, eradication is impossible due to the live bacteria that remain within the biofilm.23 The other concern is that the alginate produced by biofilm bacteria acts as an antigen in the anti-alginate antibody reaction occurring in the peripheral area of the airways. As the biofilms are chronically present, the long-term existence of the antigen in the surrounding peripheral airways results in massive lymphocyte infiltration, along with granulomatous development around the small airways. As the amount of excess antigen grows, the concentration of immune complexes containing alginate increases in the serum, resulting in the worsening of the patient's clinical symptoms. This is a result of increasing concentrations of immune complexes. This increase results in the level of immune complexes that settle in lung tissue increasing, thereby subsequently stimulating PMN chemotaxis, which consequently causes local damage.23
To date, although the 14- and 15-membered macrolides have demonstrated a variety of benefits, as demonstrated in the sections above and in Table 1, they also provide added benefits once pulmonary conditions such as DPB, cystic fibrosis or bronchiectasis progress to a stage when infection/colonization with these mucoid strains of P. aeruginosa occur. Probably the most commonly identified benefit to date is the ability of the 14- and 15-membered macrolides to break down, as well as prevent further development of the biofilms protecting the mucoid P. aeruginosa strains. This ability affects isolates that attach to either the natural airways of the patient or the synthetic material used for endotracheal/tracheostomy tubes, and occurs at subminimal inhibitory concentrations of the macrolides.24–29 Once exposed to these low macrolide concentrations, there is a decrease in the amount of alginate and hexose in the biofilm and also most likely the overall amount of polysaccharides. This decrease leads to an eradication of the membranous structures of the biofilm, thereby letting in concurrently administered anti-pseudomonal antibiotics, as well as acute reactant phagocytes, such as neutrophils, to eradicate the once shielded pathogens.24–29 In addition, although acutely absent of any inherent anti-pseudomonal activity, as P. aeruginosa accumulates the macrolides intracellularly with continued dosing, there is a resultant inhibition of protein synthesis. As the exposure and inhibition goes on, there is eventual bactericidal activity by the macrolides against the P. aeruginosa, even though the extracellular concentrations were below the MIC for the isolates of the macrolides.30 The other main effect that macrolides have on P. aeruginosa, especially mucoid strains, is an attenuation of various virulence factors that are essential for its pathogenesis. Azithromycin, for example, has been demonstrated to inhibit the quorum-sensing circuitry of P. aeruginosa by decreasing the production of autoinducers, which subsequently decreases virulence factors (autoinducer 3-oxo-C12-HSL is suppressed which leads to decreased IL-8 production) and pulmonary morbidity.31–33 Although all of the macrolides share these anti-virulence-factor properties, it has been demonstrated that azithromycin has the highest potency of the 14- and 15-membered macrolides and would therefore be the most effective option for those seeking this effect.32 By using the macrolides in cystic fibrosis, DPB or bronchiectasis patients who are infected/colonized with P. aeruginosa it is obvious from these data that it may be possible to not only potentially eradicate them when previously it was not possible to, but also, alter them so that they cause much less damage to the pulmonary tree. Evidence to date that these in vitro and other data are clinically applicable to humans is indirectly available in the DPB–macrolide literature. The fact that chronic low-dose macrolides have improved the 5 year survival rate from <10% once patients were colonized with P. aeruginosa to >90% demonstrates—at least for this condition—that these in vitro and murine data are most likely to be duplicated clinically as well. If this is true for DPB then it is probably true for other chronic pulmonary inflammatory diseases.
Other types of mucoid disease
Chronic diffuse sclerosing osteomyelitis of the mandible (DSOM) is a rare disease that is typified by symptoms such as repetitive mandibular pain, swelling and trismus, and radiology of the area demonstrates sclerotic changes with partial osteolysis.34 Because conventional antibiotics are ineffective, treatment is typically surgical. However, a large percentage of cases are intractable, as post-operative acute exacerbation is recurrent and symptoms persist for years thereafter. As chronic osteomyelitis, including DSOM, is considered a biofilm disease, a study of nine DSOM patients (two considered intractable) was conducted in which they were started on chronic regimens of oral roxithromycin 300 mg per day in two divided doses. Patients were evaluated every 2 weeks clinically to see if treatment should continue or not—with clinical symptoms of pain, trismus and X-ray findings graded on a 0–2 scale (0 being no improvement). Patients were treated for 68 days–66 months, with 7/9 (78%) cases resolving within 12 months with no identified causative pathogen. X-rays demonstrated resolution of osteolytic changes, but a persistence or even a greater predominance of osteosclerosis. The authors stated that long-term roxithromycin may be a useful treatment for DSOM, that it should be trialled prior to surgery being attempted and that the optimal duration of treatment should be based on X-ray interpretation when there is a disappearance of osteolytic changes along with an amelioration of clinical symptoms.34 Although this was a non-controlled trial involving a small number of patients, it should carry greater weight in terms of recommendations due to the rarity of the disease and the inability to conduct large-scale trials.
Implications of short-term macrolide therapy for CAP
Discussions concerning the pathogenesis, treatment and outcomes of chronic inflammatory pulmonary syndromes and CAP do not appear to go hand-in-hand. However, if one considers the findings of the DPB studies and the inflammatory response that occurs in conjunction with the development of any type of acute infectious process in an immunocompetent host, the findings are not as disparate as one would assume. In response to an acute infection, the type of immune cells that are attracted are similar to those in DPB patients, with neutrophils being recognized as the acute reactant cells that influx to the area in disproportionate numbers, compared with other types of white blood cells. These cells, and other cells native to the infected region (e.g. alveolar macrophages), or cells that have been attracted to this area, express a number of cytokines that either mobilize a coordinated immune/inflammatory response by attracting additional other types of cells, or induce production of essential reactant cells, or activate/enhance the cytolytic abilities of existing immune cells. It seems appropriate that the short-term impact of macrolides on an immune response should be characterized as a potential source of the positive clinical outcomes associated with CAP, due to the similarities in cytokines, immune cells involved and potential damage that can occur from the inflammatory reaction.
Clinical evidence of macrolide activities beyond antimicrobial activity
Despite steadily increasing incidences of mainly pneumococcal resistance, macrolides continue to be recommended for first-line empirical therapy (i.e. where atypical pathogens are a possibility) of even serious hospitalized patients with CAP, in combination with an advanced-generation cephalosporin.35–37 Even though the recommendation to use higher doses of β-lactams—to overcome possible pneumococcal resistance to these agents—has been repeated many times, the baseline doses of the macrolides that were approved for these types of infections are still the ones recommended for use with the β-lactams.35–39 This is true not only in regions with either a relatively low incidence of pneumococcal macrolide resistance or low-level efflux based resistance (MICs1–32 mg/L), but also in areas with both a high incidence of resistance and of high-level methylase resistance (MICs > 32 mg/L), such as in Southern Europe and Asia.35,40 Although it is true that macrolides would not continue to be recommended if they did not maintain their efficaciousness, either alone or in combination with a β-lactam,41,42 there is a unique finding in trials of macrolide-containing therapies for all severities of inpatient CAP that also promote interest in their continued use. Repeatedly throughout the literature (see Table 2), macrolide-containing regimens have been associated with lower mortality rates and shorter lengths of hospital stay than other types of mono- or combination therapies.43–52 As is seen in Table 2, even in a report that was published for the purpose of bringing notice to what was felt were clinical failures due to macrolide resistance, the bacteraemic patients who ‘failed’—who were on macrolides—had a 0% mortality rate, as compared with the 18% rate of the case patients who had bacteraemia but were not receiving a macrolide at the time.50 Thus, even though the patients may have had a breakthrough bacteraemia while on macrolide-containing therapy, they had a much greater chance of survival than those who were not. Even though this paper set out to demonstrate the adverse clinical impact of resistance, ‘patients’ when given the choice would prefer the bacteraemia while being on a macrolide, rather than becoming a mortality statistic when receiving a non-macrolide-containing regimen.50 Although the temptation would be to state that this benefit was the result of the atypical coverage that the macrolides add, these benefits were noticeable even when compared against other therapies that have atypical coverage, such as those with fluoroquinolones.47,49 In fact, compared with fluoroquinolone-containing regimens, macrolide-containing regimens demonstrate their benefits either much more rapidly or to a greater degree. In one commonly referenced study, the significant decrease in mortality, as compared with monotherapy with a non-pseudomonal third-generation cephalosporin, occurred on days 2 and 3 for regimens containing a macrolide plus either a second- or third-generation cephalosporin, respectively.48 In contrast, in the same study a fluoroquinolone monotherapy regimen did not manifest its significant decrease in mortality until day 7 of therapy.48 Although this difference was not statistically significant, it does appear clinically significant. In a separate study, as the dual-therapy treatment groups consisted of all of the monotherapy drugs (i.e. β-lactams and fluoroquinolones) plus a macrolide, the fact that overall mortality associated with the macrolide-containing dual therapies was 40%–70% lower than that of patients who received β-lactam or fluoroquinolone monotherapy (P < 0.05) has more meaning to it than the authors allow.49 As mortality benefits were more pronounced than the atypical coverage of fluoroquinolone monotherapy, the added benefit(s) must be secondary to other effects that are not obvious or taken into account when discussing the anti-infective treatment of CAP and are in addition to the anti-infective properties of the macrolides.
When discussing the benefits described in this section, it is important to remember that the sources for these findings are all retrospective or non-interventional prospective analyses. Although it would be preferable to have prospective, well-designed, double-blinded, randomized trials to document these benefits, it is unlikely that they will be conducted or reported in the near future. Despite the fact that the credibility of the results of the studies described above and in Table 2 may be questioned by some, credibility in fact is strengthened by the consistently documented positive effects throughout all reports. Although nothing may be a proper substitute for prospective research findings, the consistent findings in the reports provides evidence of sufficient stature and quality/quantity as to be relevant in discussions of treatment guidelines by international experts.35–39
Source of CAP benefits? Immunomodulatory pharmacodynamic aspects of macrolides
As demonstrated in the first half of this review, it is already common knowledge that the long-term macrolides have multiple types of positive immunomodulatory effects in patients suffering from chronic pulmonary inflammatory disorders. As evidenced in Table 1, these benefits manifest themselves in as little as 6–8 weeks or they can take several months. In contrast to chronic pulmonary disorders, the treatment of CAP requires instant beneficial effect. As demonstrated repeatedly in the literature, the crucial period of time for patients admitted to hospital with moderate to severe CAP is the first 8 h.48,53,54 This therapeutic window is when patients must be given the first dose(s) of their empirical initial antibiotic regimen, which will result in shorter length of stay and lower mortality.48,53,54 Not only is there a need to start therapy quickly for a positive outcome, but the actual positive changes that the antibiotics—in concert with the patients' immune systems—produce during the patients' recovery from CAP occur in a matter of a few to several days, rather than months, as seen with the chronic diseases previously described. For example, an average immunocompetent, non-elderly CAP patient who receives proper antibiotic therapy will most likely experience resolution of their fever in 2.5 days, normalization of white cell count by day 4, their cough after 7.9 days, and ‘crackles’ on auscultation by day 8.55 For elderly patients or those with significant co-morbidities these numbers may shift slightly and take a little longer to achieve across the board, though still far short of the changes one would look for in a DPB patient. As X-rays will still pick up the remaining inflammation and healing processes that are clearing/repairing the infected pulmonary area, these may remain positive for up to 4–10 weeks after initial presentation.56
In order to identify what the macrolide was inducing in the short-term in relation to a CAP regimen of a macrolide, the first studies have been conducted ex vivo using healthy volunteers. In one study, 12 healthy volunteers were administered the standard European adult CAP 3 day regimen of oral azithromycin (500 mg/day) and had blood samples drawn both prior to therapy and 2.5 and 24 h and 28 days after the end of therapy to measure drug effects on various neutrophil functions and circulating inflammatory mediators.57 Healthy volunteers were chosen for this initial study to assure that the harvested neutrophils were neither primed nor defective secondary to an inflammatory condition. Ex vivo analyses of the subjects' neutrophils demonstrated an initial neutrophil degranulating effect of azithromycin, which was reflected by rapid decreases in azurophilic granule enzyme activities in the cells and corresponding increases in the serum. The oxidative burst response to a particulate stimulus was also acutely enhanced. These actions both occurred when serum and neutrophil azithromycin concentrations were generally higher, being maximal at the 24 h time point and gradually decreasing over the subsequent 27 days to a point where they were less than those at baseline. In addition, a continuous fall in chemokine (IL-8 and human growth related oncogene-α) and IL-6 serum concentrations, within the non-pathological range, accompanied a down-regulation of the oxidative burst and an increase in neutrophil apoptosis up to 28 days after the end of dosing. Neutrophils harvested at the 28 day time-point still contained measurable azithromycin concentrations, which indicated that transfer of the drug from other types of cells (e.g. fibroblasts) to rapidly replaced neutrophils had occurred. The authors concluded that the neutrophil degranulation and oxygen burst in response to particulate matter, which occurs immediately following initiation of azithromycin dosing, may enhance endogenous host defence mechanisms to complement the direct antibacterial activity of the drug itself. In the weeks that follow the completion of the azithromycin regimen, sufficient drug from other secondary or tertiary physiological compartments is available and continually transferred to local or circulating neutrophils, and is then able to inhibit neutrophil activity, including enhancement of neutrophil apoptosis. This more prolonged process would lead to a curtailment of local inflammation and subsequently help to clear tissues from potentially damaging mediators, subsequent to the resolution of the infection.57 Although it is unclear whether these latter effects are secondary to intracellular azithromycin, the release of anti-inflammatory cytokines, such as IL-1ra as was demonstrated in CAP patients,58 or a combination of the two, an in vitro study59 provides support for the above described study57 and actually sheds light on how the introduction of a common community-acquired respiratory tract infection (RTI) pathogen modifies, yet supports, the results demonstrated in healthy volunteers.
The aim of the study was to determine the effects of azithromycin on neutrophil apoptosis, oxidative function and IL-8 production in the presence or absence of Streptococcus pneumoniae (its lysate), in comparison with penicillin, erythromycin, dexamethasone or phosphate-buffered saline.59 Neutrophils utilized were collected from healthy volunteers. The results of the study showed that the level of apoptosis noted after 1 h of exposure to azithromycin was similar to the level of apoptosis demonstrated after the neutrophils were incubated for 3 h with tumour necrosis factor (TNF)α. However, if S. pneumoniae lysate was present, azithromycin was unable to induce neutrophil apoptosis and this finding was verified via an experiment with a system that detects late-stage apoptosis. After 6 h of incubation, even though azithromycin-induced apoptosis could be verified after 1 min when the lysate was not present, it was again prevented if the bacterial lysate was co-incubated. Penicillin, dexamethasone and erythromycin had no effect on apoptosis. After 1 h of incubation, the testing for oxidative function did not show a change in function when neutrophils were incubated with any of the drugs, whether the bacterial lysate was present or not. Although in vitro exposure of the neutrophils to the S. pneumoniae lysate did indeed induce IL-8 synthesis, penicillin, erythromycin and azithromycin did not affect its production, whether the bacterial lysate was present or not. In contrast, dexamethasone significantly inhibited IL-8 production.59 The authors concluded that azithromycin-induced neutrophil apoptosis may be detected in the absence of any effect on neutrophil function, and the pro-apoptotic properties of azithromycin are inhibited in the presence of S. pneumoniae. Although at face value the results seem to conflict with the results of the first study57 and would indicate that there is only benefit in healthy volunteers, and that those with pneumococcal CAP would only benefit from azithromycin's antibacterial properties, the results actually support the benefits that macrolides can produce in CAP patients. When comparing the two studies it is important to remember that they are not very different and are both representative of the acute period of an infection as well as the phase when the body is ‘cleaning up’ the infection, and the host's response to it. As the two studies demonstrate, if a patient presented with pneumococcal CAP, the ability of the body to attract more neutrophils to the site will probably not be affected to a point that would impair it, due to no or minimal acute suppression of IL-8. This is in stark contrast to the highly immunosuppressive effects noted with the corticosteroid, dexamethasone. Once the neutrophils reach the infection site, their ability to react to the bacterial stimulants is not adversely affected by azithromycin and actually may induce a reaction that would enhance the bactericidal effects of the neutrophils. Once the bacteria have been killed and cleared, azithromycin is able to induce apoptosis, regardless of whether there is any residual IL-8 present or not, and thereby allow the body to clear these apoptotic neutrophils without them spilling their pro-inflammatory products in the process. This latter ability minimizes, if not suppresses, any further inflammation that may cause ongoing local and/or systemic damage. The fact that azithromycin was able to induce apoptosis at concentrations even lower than erythromycin, suggests that, due to its prolonged tissue half-life, azithromycin will continue to induce local apoptosis and minimize any further inflammation once the bacteria are cleared for the large proportion of the time that it may take to free the leftover inflammation from a case of CAP.57,59 Whether the beneficial biphasic effects demonstrated in these studies with azithromycin are equally present if other macrolides are used is unclear. Despite the differences noted in the latter study between azithromycin and erythromycin,59 the benefits noted in RTIs have been described with a variety of macrolides, and the benefits of one versus the other has yet to be discerned from RTI study results. In fact, as demonstrated in the literature, as long as the drug is present locally, erythromycin and roxithromycin demonstrate similar induction of neutrophil apoptosis, and in vitro investigations with clarithromycin have also noted this biphasic effect with clarithromycin and human THP-1 monocytes.60,61
A small number of studies is hardly sufficient to lead to any definitive conclusions. However, the findings to date, especially from the azithromycin study, which was dosed analogously to a common CAP regimen, are intriguing and leave one with the ability to begin to formulate extrapolations of their in vivo/ex vivo immune response effects and the improved outcomes described earlier. The enhancement of the antibacterial effects of neutrophils immediately upon the start of an azithromycin regimen may be the mechanism by which the use of a macrolide as monotherapy, or in combination with a β-lactam or fluoroquinolone, consistently appears to result in more rapid resolution of the signs/symptoms of CAP, thereby allowing for earlier hospital discharges and decreased mortality rates. Although the anti-inflammatory effects of macrolides have been established for over two decades, measurements of these effects have been typically conducted weeks to months after the start of therapy. The fact that these anti-inflammatory effects begin to manifest at such an early time-point after the start of a treatment regimen also theoretically correlates with the improved outcomes noted in CAP patients. Not only does the patient receive the benefit of a period of antibacterial enhancement, but also a relatively rapid quashing of the inflammatory response, which in most cases will help prevent any perpetuation of the inflammatory cascade and subsequent tissue damage that may result in slower resolution of the infectious process.
However, these benefits will remain only educated extrapolations until research is undertaken to try to document this biphasic effect of macrolides in the treatment of CAP in in vivo systems, such as animal models or even humans. Although a cold comment, an immunocompetent animal model of CAP, or any infection for that matter, would be ideal to demonstrate this benefit, as animals could be sacrificed serially to document the biphasic effect. Once infected and the infection has proceeded to a pre-determined pathogen load, sampled animals could be sacrificed to measure baseline infection site cell differentials, cytokine/chemokine contents and the viability of the acute reactant phagocytes. The remaining animals could then be dosed with a macrolide (± another antibiotic) and then serially sacrificed for repeat measurements of the same parameters, during which time an infection would be considered to be in its acute phase. This process could be repeated later, at which time the host would be considered to be in a convalescent phase, with the site cleansed of the inflammatory milieu of the infection. Although it would be ideal to do this in humans, the serial site sampling that would be necessary would limit the types of patients appropriate for such a study. To assure maximal collection techniques the best human model may be intubated pneumonia patients whose deep pulmonary secretions could be obtained for testing, as the infection devolves secondary to antibiotic treatment.
Short-term immunomodulation in non-infectious conditions
Further supportive conclusions may be drawn from studies of non-infectious conditions. It is well known that inflammation is not unique to infection, as a number of medical conditions and insults to the body can cause significant inflammation. When this occurs, the pathways that are activated are the same as those activated for CAP or for acute exacerbations of chronic inflammatory syndromes. As such, treatments—like macrolides—that are effective against those syndromes are probably effective against the non-infectious conditions as well. As an example, a study of 54 women undergoing mastectomy received either clarithromycin 500 mg twice daily from the day prior to surgery to 3 days after plus standard pain management, or, just standard pain management to assess whether clarithromycin had a beneficial impact on the local and/or systemic inflammatory response from the surgery.62 Although compared with the control group there was no difference in the incidence of post-operative infections, clarithromycin-treated patients had significant suppression of the febrile response, tachycardia, tachypnoea and post-operative increase in monocyte counts. In addition, those treated with clarithromycin experienced significantly less intense discomfort (average of 3.2 versus 4.2 on a 0–10 pain scale, P < 0.05), which lasted for a significantly shorter duration (average of 2.63 days versus 3.98 days, P < 0.005) than that experienced by control patients. Although there was no significant alteration in cytokine concentrations (IL-6, TNFα), the interval changes in both C-reactive protein and erythrocyte sedimentation rate were statistically lower for the clarithromycin-treated patients. The authors concluded that a brief course of clarithromycin modulated the acute inflammatory response associated with the surgery and produced a superior overall outcome in the patients who received it.62 The results of this study raise interesting questions. Should macrolides be included in routine surgical antibiotic prophylaxis, not only for their antimicrobial but also for their anti-inflammatory properties? Such characteristics may help patients—even following major surgery—experience a better recovery in terms of requiring less pain medication and for a shorter period. Also, if the macrolide is able to decrease inflammation at the surgical site, there may be other areas of interest to study: the pain issue—already investigated—and whether less scar tissue would be formed at surgical sites. The ability to suppress scar tissue production or over-production might decrease post-operative complications and be more aesthetically pleasing. And although an argument could be made for using the current macrolides for these studies, it might be desirable to develop related compounds without antimicrobial properties, thus avoiding promotion of resistance. Hopefully, future research will have the answers to these and future questions.
Conclusions
The macrolides have long been associated with anti-inflammatory benefits in patients with chronic pulmonary inflammatory disorders. The findings in DPB patients are very reproducible. The data obtained from cystic fibrosis, asthma and bronchiectasis, however, although positive, have been highly variable to date and more research needs to be conducted to understand the reason(s) for this variability.
The markers or clinical tests that demonstrate the positive impact of macrolides on the clinical progress of patients need to be identified for future trials as well as for clinical practice.
The macrolides can decrease the overall severity of these diseases. They not only enable the administration of systemic antibiotics but also support the immune system in their attack on mucoid pathogens, once ‘invulnerable’, due to their anti-biofilm and anti-quorum sensing abilities. These abilities provide added survival benefits, as compared with the past when colonization with these types of pathogens led to much higher mortality rates than for patients not colonized with them.
Recent work demonstrates that the decreased mortality and length of stay seen with macrolide-based CAP treatment regimens may also be due to short-term immunomodulatory effects of the macrolides. It has been demonstrated that a 3 day course of azithromycin appears to have a biphasic effect, where it enhances the immune system's response to bacteria initially. After that it shuts it down, once the bacteria have been eradicated, thereby eliminating the inflammatory response as quickly as possible. However, before this mechanism can be referred to as ‘gospel’, it needs to be replicated in at least an animal model of CAP.
A greater representation of the macrolides' short-term benefits is recent in vivo evidence of their ability to decrease pain medication needs, and the length of time pain medications are needed after mastectomy, a major surgery. With the focus on pain management clinically these days, the development of a macrolide without antibacterial properties—which could be used in combination with pain medications in this manner—may be a new way to manage post-operative pain syndromes.
It is becoming increasingly obvious that beyond the antibacterial properties of the macrolides, the 14- and 15-membered macrolides potentially have a very large number of both short- and long-term immunomodulatory uses. Since they provide the majority of the same benefits as a corticosteroid, but do not cause immunosuppression, it is possible that they may become a favoured agent for patients who require systemic corticosteroids chronically.
For some potential indications, such as post-operative pain, it may be preferable to have a macrolide derivative that has been stripped of its antibacterial properties. Many of the disease states discussed within this paper, both long- and short-term, would benefit from such a macrolide.
Transparency declarations
G.W.A. is a consultant, researcher and speaker for Pliva dd and Pfizer, Inc. and has done antimicrobial research for Abbott, Bayer, GlaxoSmithKline and Bristol-Myers Squibb. Partial support for this paper was provided by Pliva dd. Remaining support was not related to industry sources.
Author . | Disorder . | Treatment . | Outcome parameters . | Results . |
---|---|---|---|---|
Kadota et al.12 | DPB n=14 HV n=5 | DPB only—200 mg 3×/day erythromycin till response (6–12 months), no steroids/other antibiotics | DPB baseline BAL diff 71% PMN 15% AM; HV 91% AM, 1.6% PMN; pre-NCA- DPB 51%, HV 28% (P < 0.01) | DPB F/U BAL 54% AM, 28% PMN; DPB NCA 30%- sig. corr. with ↓ in PMN count P < 0.05 |
Sakito et al.13 | DPB n=19 pulm. sarcoid n=17 HV n=7 | DPB only—600 mg/day erythromycin ×16 months or 150 mg/day roxithromycin ×4.4 months; treated till response; no steroids/other antibiotics | baseline BAL PMN (62% versus 1.3% versus 0.8%), IL-1β (40 versus 20 versus 23a), IL-8 (419 versus 21 versus 5.5a), TNFα (19 versus 2a versus undetect) | sig. improve in pulm fxn; 80% pts mod/marked clinically improved; F/U BAL PMN 22% (P=0.0001); IL-1β 22a (P < 0.015); IL-8 92a (P < 0.046); TNFα 1.1a |
Amayasu et al.14 | asthma n=17 | 200 mg clarithromycin or match placebo 2×/day ×8 weeks | AR by PC20, FVC, FEV1, sputum+serum eosinophil counts, ECP | FVC+FEV1 NC; mean log clari PC20 2.96 versus 2.60, P < 0.01; serum/ sputum eos. sig. ↓ 70–80% on clari P < 0.01 as did ECP by 75% P < 0.01 |
Koh et al.15 | bronchiectasis n=25 (children) | 4 mg/kg roxithromycin or placebo 2×/day ×12 weeks | AR by PD20 and maximal response; FEV1; sputum purulenceb; leucocyte scoreb | FEV1 NC; roxith sig. ↓ purul. (2.54 to 1.39 P < 0.01) and leuc. (2.23 to 1.31 P < 0.01) by week 6; PD20 sig.↑ from baseline (169 versus 87 P < 0.01); max. resp. sig.↓(32.5 versus 40.9% P < 0.01) |
Tsang et al.16 | bronchiectasis n=21 | 500 mg erythromycin or placebo 2×/day ×8 weeks | PFTs; sputum tests—24 h volume, WBC+pathogen densities, IL-1α, TNFα, LTB4 | NC in sputum WBC/pathogen densities or cytokine conc.; 30%↓24 h sputum volume, 13%↑FEV1, 6%↑ FVC P < 0.05 with eryth |
Equi et al.18 | CF n=41 (children) | 250–500 mg/day azithromycin and placebo ×6 months, crossover after 2 month w/o | FEV1; sputum cultures, sputum IL-8+NE conc.; exercise test, QOL, Abx use, pulmonary excerbation rate | FEV1↑on azith P=0.031; NC sputum cultures, sputum IL-8+NE conc., exercise test, QOL; azith arm 17/41 pts use 24 less courses Abx (P=0.005) |
Wolter et al.19 | CF n=60 | 250 mg/day azithromycin or placebo ×3 months | FEV1+ FVC% pred., iv Abx needed for exacerbations, CRP and QOL azithromycin versus placebo | FEV1+FVC% pred. stable in azith versus↓(P=0.047, P=0.001) for placebo; azith needed < iv Abx (0.37 versus 1.13 P=0.016); azith ↓ CRP 10 to 5.4 mg/mL P < 0.001; azith sig.↑ QOL P=0.035 |
Saiman et al.20 | CF n=185, age ≥6 years, infected with P. aeruginosa for ≥1 year and FEV1% pred. of ≥30% | 250 mg ( < 40 kg) or 500 mg (≥40 kg) azithromycin or placebo thrice weekly ×168 days | FEV1% predicted, QOL, weight gain, incidence of pulmonary exacerbations | FEV% pred. azith↑4.4%, placebo↓1.8% (P=0.001); azith pts had sig. lower risk of exac. (HR 0.65; 95%CI 0.44–0.95, P=0.03) and↑wgt sig. mean 0.7 kg P=0.02 |
Author . | Disorder . | Treatment . | Outcome parameters . | Results . |
---|---|---|---|---|
Kadota et al.12 | DPB n=14 HV n=5 | DPB only—200 mg 3×/day erythromycin till response (6–12 months), no steroids/other antibiotics | DPB baseline BAL diff 71% PMN 15% AM; HV 91% AM, 1.6% PMN; pre-NCA- DPB 51%, HV 28% (P < 0.01) | DPB F/U BAL 54% AM, 28% PMN; DPB NCA 30%- sig. corr. with ↓ in PMN count P < 0.05 |
Sakito et al.13 | DPB n=19 pulm. sarcoid n=17 HV n=7 | DPB only—600 mg/day erythromycin ×16 months or 150 mg/day roxithromycin ×4.4 months; treated till response; no steroids/other antibiotics | baseline BAL PMN (62% versus 1.3% versus 0.8%), IL-1β (40 versus 20 versus 23a), IL-8 (419 versus 21 versus 5.5a), TNFα (19 versus 2a versus undetect) | sig. improve in pulm fxn; 80% pts mod/marked clinically improved; F/U BAL PMN 22% (P=0.0001); IL-1β 22a (P < 0.015); IL-8 92a (P < 0.046); TNFα 1.1a |
Amayasu et al.14 | asthma n=17 | 200 mg clarithromycin or match placebo 2×/day ×8 weeks | AR by PC20, FVC, FEV1, sputum+serum eosinophil counts, ECP | FVC+FEV1 NC; mean log clari PC20 2.96 versus 2.60, P < 0.01; serum/ sputum eos. sig. ↓ 70–80% on clari P < 0.01 as did ECP by 75% P < 0.01 |
Koh et al.15 | bronchiectasis n=25 (children) | 4 mg/kg roxithromycin or placebo 2×/day ×12 weeks | AR by PD20 and maximal response; FEV1; sputum purulenceb; leucocyte scoreb | FEV1 NC; roxith sig. ↓ purul. (2.54 to 1.39 P < 0.01) and leuc. (2.23 to 1.31 P < 0.01) by week 6; PD20 sig.↑ from baseline (169 versus 87 P < 0.01); max. resp. sig.↓(32.5 versus 40.9% P < 0.01) |
Tsang et al.16 | bronchiectasis n=21 | 500 mg erythromycin or placebo 2×/day ×8 weeks | PFTs; sputum tests—24 h volume, WBC+pathogen densities, IL-1α, TNFα, LTB4 | NC in sputum WBC/pathogen densities or cytokine conc.; 30%↓24 h sputum volume, 13%↑FEV1, 6%↑ FVC P < 0.05 with eryth |
Equi et al.18 | CF n=41 (children) | 250–500 mg/day azithromycin and placebo ×6 months, crossover after 2 month w/o | FEV1; sputum cultures, sputum IL-8+NE conc.; exercise test, QOL, Abx use, pulmonary excerbation rate | FEV1↑on azith P=0.031; NC sputum cultures, sputum IL-8+NE conc., exercise test, QOL; azith arm 17/41 pts use 24 less courses Abx (P=0.005) |
Wolter et al.19 | CF n=60 | 250 mg/day azithromycin or placebo ×3 months | FEV1+ FVC% pred., iv Abx needed for exacerbations, CRP and QOL azithromycin versus placebo | FEV1+FVC% pred. stable in azith versus↓(P=0.047, P=0.001) for placebo; azith needed < iv Abx (0.37 versus 1.13 P=0.016); azith ↓ CRP 10 to 5.4 mg/mL P < 0.001; azith sig.↑ QOL P=0.035 |
Saiman et al.20 | CF n=185, age ≥6 years, infected with P. aeruginosa for ≥1 year and FEV1% pred. of ≥30% | 250 mg ( < 40 kg) or 500 mg (≥40 kg) azithromycin or placebo thrice weekly ×168 days | FEV1% predicted, QOL, weight gain, incidence of pulmonary exacerbations | FEV% pred. azith↑4.4%, placebo↓1.8% (P=0.001); azith pts had sig. lower risk of exac. (HR 0.65; 95%CI 0.44–0.95, P=0.03) and↑wgt sig. mean 0.7 kg P=0.02 |
DPB, diffuse panbronchiolitis; HV, healthy volunteer; BAL, bronchoalveolar lavage; PMN, polymorphonuclear leucocytes (neutrophils); AM, alveolar macrophage; NCA, neutrophil chemotactic activity; F/U, follow-up; AR, airway responsiveness; PC20, methacholine concentration needed to produce a 20% decrease in forced expiratory volume in 1 second (FEV1); FVC, forced vital capacity; ECP, eosinophilic cationic protein; NC, no significant change; PD20, same as PC20; PFTs, pulmonary function tests; CF, cystic fibrosis; w/o, washout; NE, neutrophil elastase; Abx, antibiotics; QOL, quality of life; iv, intravenous; CRP, C-reactive protein; HR, hazard ratio; CI, confidence interval; pts, patients; WBC, white blood cells; LTB4, leukotriene B4; sig., significant/significantly; leuc., leucocyte score; max. resp., maximal response.
Units are pg/mL.
Both measures are scored from 1 (meaning least severe) to 3 (meaning most severe).
Author . | Disorder . | Treatment . | Outcome parameters . | Results . |
---|---|---|---|---|
Kadota et al.12 | DPB n=14 HV n=5 | DPB only—200 mg 3×/day erythromycin till response (6–12 months), no steroids/other antibiotics | DPB baseline BAL diff 71% PMN 15% AM; HV 91% AM, 1.6% PMN; pre-NCA- DPB 51%, HV 28% (P < 0.01) | DPB F/U BAL 54% AM, 28% PMN; DPB NCA 30%- sig. corr. with ↓ in PMN count P < 0.05 |
Sakito et al.13 | DPB n=19 pulm. sarcoid n=17 HV n=7 | DPB only—600 mg/day erythromycin ×16 months or 150 mg/day roxithromycin ×4.4 months; treated till response; no steroids/other antibiotics | baseline BAL PMN (62% versus 1.3% versus 0.8%), IL-1β (40 versus 20 versus 23a), IL-8 (419 versus 21 versus 5.5a), TNFα (19 versus 2a versus undetect) | sig. improve in pulm fxn; 80% pts mod/marked clinically improved; F/U BAL PMN 22% (P=0.0001); IL-1β 22a (P < 0.015); IL-8 92a (P < 0.046); TNFα 1.1a |
Amayasu et al.14 | asthma n=17 | 200 mg clarithromycin or match placebo 2×/day ×8 weeks | AR by PC20, FVC, FEV1, sputum+serum eosinophil counts, ECP | FVC+FEV1 NC; mean log clari PC20 2.96 versus 2.60, P < 0.01; serum/ sputum eos. sig. ↓ 70–80% on clari P < 0.01 as did ECP by 75% P < 0.01 |
Koh et al.15 | bronchiectasis n=25 (children) | 4 mg/kg roxithromycin or placebo 2×/day ×12 weeks | AR by PD20 and maximal response; FEV1; sputum purulenceb; leucocyte scoreb | FEV1 NC; roxith sig. ↓ purul. (2.54 to 1.39 P < 0.01) and leuc. (2.23 to 1.31 P < 0.01) by week 6; PD20 sig.↑ from baseline (169 versus 87 P < 0.01); max. resp. sig.↓(32.5 versus 40.9% P < 0.01) |
Tsang et al.16 | bronchiectasis n=21 | 500 mg erythromycin or placebo 2×/day ×8 weeks | PFTs; sputum tests—24 h volume, WBC+pathogen densities, IL-1α, TNFα, LTB4 | NC in sputum WBC/pathogen densities or cytokine conc.; 30%↓24 h sputum volume, 13%↑FEV1, 6%↑ FVC P < 0.05 with eryth |
Equi et al.18 | CF n=41 (children) | 250–500 mg/day azithromycin and placebo ×6 months, crossover after 2 month w/o | FEV1; sputum cultures, sputum IL-8+NE conc.; exercise test, QOL, Abx use, pulmonary excerbation rate | FEV1↑on azith P=0.031; NC sputum cultures, sputum IL-8+NE conc., exercise test, QOL; azith arm 17/41 pts use 24 less courses Abx (P=0.005) |
Wolter et al.19 | CF n=60 | 250 mg/day azithromycin or placebo ×3 months | FEV1+ FVC% pred., iv Abx needed for exacerbations, CRP and QOL azithromycin versus placebo | FEV1+FVC% pred. stable in azith versus↓(P=0.047, P=0.001) for placebo; azith needed < iv Abx (0.37 versus 1.13 P=0.016); azith ↓ CRP 10 to 5.4 mg/mL P < 0.001; azith sig.↑ QOL P=0.035 |
Saiman et al.20 | CF n=185, age ≥6 years, infected with P. aeruginosa for ≥1 year and FEV1% pred. of ≥30% | 250 mg ( < 40 kg) or 500 mg (≥40 kg) azithromycin or placebo thrice weekly ×168 days | FEV1% predicted, QOL, weight gain, incidence of pulmonary exacerbations | FEV% pred. azith↑4.4%, placebo↓1.8% (P=0.001); azith pts had sig. lower risk of exac. (HR 0.65; 95%CI 0.44–0.95, P=0.03) and↑wgt sig. mean 0.7 kg P=0.02 |
Author . | Disorder . | Treatment . | Outcome parameters . | Results . |
---|---|---|---|---|
Kadota et al.12 | DPB n=14 HV n=5 | DPB only—200 mg 3×/day erythromycin till response (6–12 months), no steroids/other antibiotics | DPB baseline BAL diff 71% PMN 15% AM; HV 91% AM, 1.6% PMN; pre-NCA- DPB 51%, HV 28% (P < 0.01) | DPB F/U BAL 54% AM, 28% PMN; DPB NCA 30%- sig. corr. with ↓ in PMN count P < 0.05 |
Sakito et al.13 | DPB n=19 pulm. sarcoid n=17 HV n=7 | DPB only—600 mg/day erythromycin ×16 months or 150 mg/day roxithromycin ×4.4 months; treated till response; no steroids/other antibiotics | baseline BAL PMN (62% versus 1.3% versus 0.8%), IL-1β (40 versus 20 versus 23a), IL-8 (419 versus 21 versus 5.5a), TNFα (19 versus 2a versus undetect) | sig. improve in pulm fxn; 80% pts mod/marked clinically improved; F/U BAL PMN 22% (P=0.0001); IL-1β 22a (P < 0.015); IL-8 92a (P < 0.046); TNFα 1.1a |
Amayasu et al.14 | asthma n=17 | 200 mg clarithromycin or match placebo 2×/day ×8 weeks | AR by PC20, FVC, FEV1, sputum+serum eosinophil counts, ECP | FVC+FEV1 NC; mean log clari PC20 2.96 versus 2.60, P < 0.01; serum/ sputum eos. sig. ↓ 70–80% on clari P < 0.01 as did ECP by 75% P < 0.01 |
Koh et al.15 | bronchiectasis n=25 (children) | 4 mg/kg roxithromycin or placebo 2×/day ×12 weeks | AR by PD20 and maximal response; FEV1; sputum purulenceb; leucocyte scoreb | FEV1 NC; roxith sig. ↓ purul. (2.54 to 1.39 P < 0.01) and leuc. (2.23 to 1.31 P < 0.01) by week 6; PD20 sig.↑ from baseline (169 versus 87 P < 0.01); max. resp. sig.↓(32.5 versus 40.9% P < 0.01) |
Tsang et al.16 | bronchiectasis n=21 | 500 mg erythromycin or placebo 2×/day ×8 weeks | PFTs; sputum tests—24 h volume, WBC+pathogen densities, IL-1α, TNFα, LTB4 | NC in sputum WBC/pathogen densities or cytokine conc.; 30%↓24 h sputum volume, 13%↑FEV1, 6%↑ FVC P < 0.05 with eryth |
Equi et al.18 | CF n=41 (children) | 250–500 mg/day azithromycin and placebo ×6 months, crossover after 2 month w/o | FEV1; sputum cultures, sputum IL-8+NE conc.; exercise test, QOL, Abx use, pulmonary excerbation rate | FEV1↑on azith P=0.031; NC sputum cultures, sputum IL-8+NE conc., exercise test, QOL; azith arm 17/41 pts use 24 less courses Abx (P=0.005) |
Wolter et al.19 | CF n=60 | 250 mg/day azithromycin or placebo ×3 months | FEV1+ FVC% pred., iv Abx needed for exacerbations, CRP and QOL azithromycin versus placebo | FEV1+FVC% pred. stable in azith versus↓(P=0.047, P=0.001) for placebo; azith needed < iv Abx (0.37 versus 1.13 P=0.016); azith ↓ CRP 10 to 5.4 mg/mL P < 0.001; azith sig.↑ QOL P=0.035 |
Saiman et al.20 | CF n=185, age ≥6 years, infected with P. aeruginosa for ≥1 year and FEV1% pred. of ≥30% | 250 mg ( < 40 kg) or 500 mg (≥40 kg) azithromycin or placebo thrice weekly ×168 days | FEV1% predicted, QOL, weight gain, incidence of pulmonary exacerbations | FEV% pred. azith↑4.4%, placebo↓1.8% (P=0.001); azith pts had sig. lower risk of exac. (HR 0.65; 95%CI 0.44–0.95, P=0.03) and↑wgt sig. mean 0.7 kg P=0.02 |
DPB, diffuse panbronchiolitis; HV, healthy volunteer; BAL, bronchoalveolar lavage; PMN, polymorphonuclear leucocytes (neutrophils); AM, alveolar macrophage; NCA, neutrophil chemotactic activity; F/U, follow-up; AR, airway responsiveness; PC20, methacholine concentration needed to produce a 20% decrease in forced expiratory volume in 1 second (FEV1); FVC, forced vital capacity; ECP, eosinophilic cationic protein; NC, no significant change; PD20, same as PC20; PFTs, pulmonary function tests; CF, cystic fibrosis; w/o, washout; NE, neutrophil elastase; Abx, antibiotics; QOL, quality of life; iv, intravenous; CRP, C-reactive protein; HR, hazard ratio; CI, confidence interval; pts, patients; WBC, white blood cells; LTB4, leukotriene B4; sig., significant/significantly; leuc., leucocyte score; max. resp., maximal response.
Units are pg/mL.
Both measures are scored from 1 (meaning least severe) to 3 (meaning most severe).
Reference/study type . | No. of pts and type . | Treatment . | Outcome measure . | Results . |
---|---|---|---|---|
Gleason et al.48 Chart review | 12 945 hospitalized | initial empirical Abx regimens | 30 day mortality assoc with initial Abx regimen and when received first dose | 14.9% np3gc versus 8.4% 2gc+m P < 0.05 on day 2 of hosp. versus 9.1% np3gc+m P < 0.05 on day 3 of hosp. versus 10.6% FQ P < 0.05 on day 7 of hosp.;↓mortality with np3gc+m when first dose w/in 8 h of pt admit (HR 0.73; 95%CI 0.55–0.97) |
Martinez et al.45 Retrospective | 409 bacteraemic pneumococcal pneumonia; most with signif. co-morbid. | 238 pts got β-lactam+macrolide; 171 pts got β-lactam monotherapy | see what correlated with mortality | shock P < 0.0001; admit to ICU P < 0.0001; pneumococcus resistant to penicillin and eryth P=0.02; Abx therapy with other than β-lactam +macrolide P=0.001 (greater risk than having resistant isolate) |
Lonks et al.50 Matched case control chart review | 86 case pts, 141 control pts with breakthrough bacteraemic pneumococcal pneumonia | pts on mostly β-lactam, macrolide, or a combination of the two | 19/86 case pts had I or R isolates drawn while on macrolide 0/141 control pts; but what about mortality? | of case pts 0/19 that got bacteraemia while on macrolide died; 12/67 (18%) case pts got bacteraemia not on macrolide died (P=0.06–probably due to low n) |
Stahl et al.43 Non-interventional prospective | 100 enrolled admitted CAP pts, 76 evaluable | within 24 h 68 got β-lactam, 12 got macrolide (11 with β-lactam) and rest other agents in groups of 1–2 per drug | mean length of stay (LOS) for evaluable group was 4.9 days | signif. ↓ LOS if got macrolide in 1st 24 h of admit (2.75 versus 5.3 days, P=0.01); only existed for early Abx use as if within 48 h (4.05 versus 5.2 days, P=0.19) |
Trowbridge et al.51 Retrospective | 450 enrolled, 376 evaluable mild to moderate hospitalized CAP, PSI Class III on average | np3gc+m (eryth or azith); FQ mono; azith mono; 2gc or np3gc mono; ββi mono | ANCOVA to compare LOS with antibiotic regimen, PSI score and categorical factors for antibiotic regimen and therapy site | when adjusted for PSI and site, azith and FQ monos and np3gc+azith had 1.7–2.7 days↓LOS than cephalosporin mono.; those that stayed on initial regimen this grew to 2.3–5.2 days–also np3gc+eryth versus pip-tazo mono =5.1 versus 9.2 days P < 0.05; initial empirical regimen having atypical coverage had 1.8–3.3 day↓LOS (P < 0.001) |
Lentino & Krasnicka52 Retrospective | n=94 hospitalized CAP, both groups had PSI score Class IV on average | group I – empirical initial other than mono with azith (n=50, 25 got ceftriaxone+macrolide); group II—mono azith (n=44) | treatment duration and LOS by empirical initial treatment group | minus ICU pts alive at 90 days treatment duration Group I versus Group II (10.8 versus 8 days, P=0.026); LOS 6.5 versus 4.4 days, P=0.012; whole population- treatment duration 11.4 versus 8.1 days, P=0.0007; LOS 9.7 versus 4.6 days, P < 0.0001; azith mono assoc. with $2400 per 2 day less LOS per pt episode |
Brown et al.49 Analysis of hospital claims database | Of 188,627 pts put in database 44,814 met criteria; hospitalized CAP pts Fine PSI Class A-D [= I–IV]; too much variability for Fine V (E) | mono with cef, oceph, FQ, macs or pcns or dual therapy of one of the β-lactams or FQ with a mac | impact of choice of initial empirical antibiotic therapy on 30 day mortality, total hospital costs and hospital LOS | mono pts—mac pts lowest mortality (2.2 versus ≥5%, P < 0.001) but were in lower risk classes; cef had↓LOS (mean 4.99 days versus 5.82–6.71 for other β-lactams and FQ P < 0.005) and assoc. cost ($7200 versus >$8200); dual therapy groups–cef+mac ↓ LOS versus all else (mean 4.98 versus 5.6–6.5 days, P < 0.005) and cost ($7900 versus $9700–11 300, P < 0.005); mac dual therapy 40%–70% ↓ mortality than pts got β-lactam or FQ mono (P < 0.05) |
Reference/study type . | No. of pts and type . | Treatment . | Outcome measure . | Results . |
---|---|---|---|---|
Gleason et al.48 Chart review | 12 945 hospitalized | initial empirical Abx regimens | 30 day mortality assoc with initial Abx regimen and when received first dose | 14.9% np3gc versus 8.4% 2gc+m P < 0.05 on day 2 of hosp. versus 9.1% np3gc+m P < 0.05 on day 3 of hosp. versus 10.6% FQ P < 0.05 on day 7 of hosp.;↓mortality with np3gc+m when first dose w/in 8 h of pt admit (HR 0.73; 95%CI 0.55–0.97) |
Martinez et al.45 Retrospective | 409 bacteraemic pneumococcal pneumonia; most with signif. co-morbid. | 238 pts got β-lactam+macrolide; 171 pts got β-lactam monotherapy | see what correlated with mortality | shock P < 0.0001; admit to ICU P < 0.0001; pneumococcus resistant to penicillin and eryth P=0.02; Abx therapy with other than β-lactam +macrolide P=0.001 (greater risk than having resistant isolate) |
Lonks et al.50 Matched case control chart review | 86 case pts, 141 control pts with breakthrough bacteraemic pneumococcal pneumonia | pts on mostly β-lactam, macrolide, or a combination of the two | 19/86 case pts had I or R isolates drawn while on macrolide 0/141 control pts; but what about mortality? | of case pts 0/19 that got bacteraemia while on macrolide died; 12/67 (18%) case pts got bacteraemia not on macrolide died (P=0.06–probably due to low n) |
Stahl et al.43 Non-interventional prospective | 100 enrolled admitted CAP pts, 76 evaluable | within 24 h 68 got β-lactam, 12 got macrolide (11 with β-lactam) and rest other agents in groups of 1–2 per drug | mean length of stay (LOS) for evaluable group was 4.9 days | signif. ↓ LOS if got macrolide in 1st 24 h of admit (2.75 versus 5.3 days, P=0.01); only existed for early Abx use as if within 48 h (4.05 versus 5.2 days, P=0.19) |
Trowbridge et al.51 Retrospective | 450 enrolled, 376 evaluable mild to moderate hospitalized CAP, PSI Class III on average | np3gc+m (eryth or azith); FQ mono; azith mono; 2gc or np3gc mono; ββi mono | ANCOVA to compare LOS with antibiotic regimen, PSI score and categorical factors for antibiotic regimen and therapy site | when adjusted for PSI and site, azith and FQ monos and np3gc+azith had 1.7–2.7 days↓LOS than cephalosporin mono.; those that stayed on initial regimen this grew to 2.3–5.2 days–also np3gc+eryth versus pip-tazo mono =5.1 versus 9.2 days P < 0.05; initial empirical regimen having atypical coverage had 1.8–3.3 day↓LOS (P < 0.001) |
Lentino & Krasnicka52 Retrospective | n=94 hospitalized CAP, both groups had PSI score Class IV on average | group I – empirical initial other than mono with azith (n=50, 25 got ceftriaxone+macrolide); group II—mono azith (n=44) | treatment duration and LOS by empirical initial treatment group | minus ICU pts alive at 90 days treatment duration Group I versus Group II (10.8 versus 8 days, P=0.026); LOS 6.5 versus 4.4 days, P=0.012; whole population- treatment duration 11.4 versus 8.1 days, P=0.0007; LOS 9.7 versus 4.6 days, P < 0.0001; azith mono assoc. with $2400 per 2 day less LOS per pt episode |
Brown et al.49 Analysis of hospital claims database | Of 188,627 pts put in database 44,814 met criteria; hospitalized CAP pts Fine PSI Class A-D [= I–IV]; too much variability for Fine V (E) | mono with cef, oceph, FQ, macs or pcns or dual therapy of one of the β-lactams or FQ with a mac | impact of choice of initial empirical antibiotic therapy on 30 day mortality, total hospital costs and hospital LOS | mono pts—mac pts lowest mortality (2.2 versus ≥5%, P < 0.001) but were in lower risk classes; cef had↓LOS (mean 4.99 days versus 5.82–6.71 for other β-lactams and FQ P < 0.005) and assoc. cost ($7200 versus >$8200); dual therapy groups–cef+mac ↓ LOS versus all else (mean 4.98 versus 5.6–6.5 days, P < 0.005) and cost ($7900 versus $9700–11 300, P < 0.005); mac dual therapy 40%–70% ↓ mortality than pts got β-lactam or FQ mono (P < 0.05) |
Abx, antibiotics; np3gc, non-pseudomonal third-generation cephalosporin; 2gc + m, second-generation cephalosporin plus a macrolide; np3gc + m, non-pseudomonal third-generation cephalosporin plus a macrolide; FQ, fluoroquinolone; hosp., hospitalization; w/in, within; HR, hazard ratio; CI, confidence interval; I, intermediately susceptible; R, resistant; mono, monotherapy; azith, azithromycin; ββi, β-lactam/β-lactamase inhibitor; ANCOVA, analysis of covariance; cef, ceftriaxone; oceph, other cephalosporins; macs, macrolides; pcns, penicillins; PSI, Pneumonia Severity Index; pts, patients.
Reference/study type . | No. of pts and type . | Treatment . | Outcome measure . | Results . |
---|---|---|---|---|
Gleason et al.48 Chart review | 12 945 hospitalized | initial empirical Abx regimens | 30 day mortality assoc with initial Abx regimen and when received first dose | 14.9% np3gc versus 8.4% 2gc+m P < 0.05 on day 2 of hosp. versus 9.1% np3gc+m P < 0.05 on day 3 of hosp. versus 10.6% FQ P < 0.05 on day 7 of hosp.;↓mortality with np3gc+m when first dose w/in 8 h of pt admit (HR 0.73; 95%CI 0.55–0.97) |
Martinez et al.45 Retrospective | 409 bacteraemic pneumococcal pneumonia; most with signif. co-morbid. | 238 pts got β-lactam+macrolide; 171 pts got β-lactam monotherapy | see what correlated with mortality | shock P < 0.0001; admit to ICU P < 0.0001; pneumococcus resistant to penicillin and eryth P=0.02; Abx therapy with other than β-lactam +macrolide P=0.001 (greater risk than having resistant isolate) |
Lonks et al.50 Matched case control chart review | 86 case pts, 141 control pts with breakthrough bacteraemic pneumococcal pneumonia | pts on mostly β-lactam, macrolide, or a combination of the two | 19/86 case pts had I or R isolates drawn while on macrolide 0/141 control pts; but what about mortality? | of case pts 0/19 that got bacteraemia while on macrolide died; 12/67 (18%) case pts got bacteraemia not on macrolide died (P=0.06–probably due to low n) |
Stahl et al.43 Non-interventional prospective | 100 enrolled admitted CAP pts, 76 evaluable | within 24 h 68 got β-lactam, 12 got macrolide (11 with β-lactam) and rest other agents in groups of 1–2 per drug | mean length of stay (LOS) for evaluable group was 4.9 days | signif. ↓ LOS if got macrolide in 1st 24 h of admit (2.75 versus 5.3 days, P=0.01); only existed for early Abx use as if within 48 h (4.05 versus 5.2 days, P=0.19) |
Trowbridge et al.51 Retrospective | 450 enrolled, 376 evaluable mild to moderate hospitalized CAP, PSI Class III on average | np3gc+m (eryth or azith); FQ mono; azith mono; 2gc or np3gc mono; ββi mono | ANCOVA to compare LOS with antibiotic regimen, PSI score and categorical factors for antibiotic regimen and therapy site | when adjusted for PSI and site, azith and FQ monos and np3gc+azith had 1.7–2.7 days↓LOS than cephalosporin mono.; those that stayed on initial regimen this grew to 2.3–5.2 days–also np3gc+eryth versus pip-tazo mono =5.1 versus 9.2 days P < 0.05; initial empirical regimen having atypical coverage had 1.8–3.3 day↓LOS (P < 0.001) |
Lentino & Krasnicka52 Retrospective | n=94 hospitalized CAP, both groups had PSI score Class IV on average | group I – empirical initial other than mono with azith (n=50, 25 got ceftriaxone+macrolide); group II—mono azith (n=44) | treatment duration and LOS by empirical initial treatment group | minus ICU pts alive at 90 days treatment duration Group I versus Group II (10.8 versus 8 days, P=0.026); LOS 6.5 versus 4.4 days, P=0.012; whole population- treatment duration 11.4 versus 8.1 days, P=0.0007; LOS 9.7 versus 4.6 days, P < 0.0001; azith mono assoc. with $2400 per 2 day less LOS per pt episode |
Brown et al.49 Analysis of hospital claims database | Of 188,627 pts put in database 44,814 met criteria; hospitalized CAP pts Fine PSI Class A-D [= I–IV]; too much variability for Fine V (E) | mono with cef, oceph, FQ, macs or pcns or dual therapy of one of the β-lactams or FQ with a mac | impact of choice of initial empirical antibiotic therapy on 30 day mortality, total hospital costs and hospital LOS | mono pts—mac pts lowest mortality (2.2 versus ≥5%, P < 0.001) but were in lower risk classes; cef had↓LOS (mean 4.99 days versus 5.82–6.71 for other β-lactams and FQ P < 0.005) and assoc. cost ($7200 versus >$8200); dual therapy groups–cef+mac ↓ LOS versus all else (mean 4.98 versus 5.6–6.5 days, P < 0.005) and cost ($7900 versus $9700–11 300, P < 0.005); mac dual therapy 40%–70% ↓ mortality than pts got β-lactam or FQ mono (P < 0.05) |
Reference/study type . | No. of pts and type . | Treatment . | Outcome measure . | Results . |
---|---|---|---|---|
Gleason et al.48 Chart review | 12 945 hospitalized | initial empirical Abx regimens | 30 day mortality assoc with initial Abx regimen and when received first dose | 14.9% np3gc versus 8.4% 2gc+m P < 0.05 on day 2 of hosp. versus 9.1% np3gc+m P < 0.05 on day 3 of hosp. versus 10.6% FQ P < 0.05 on day 7 of hosp.;↓mortality with np3gc+m when first dose w/in 8 h of pt admit (HR 0.73; 95%CI 0.55–0.97) |
Martinez et al.45 Retrospective | 409 bacteraemic pneumococcal pneumonia; most with signif. co-morbid. | 238 pts got β-lactam+macrolide; 171 pts got β-lactam monotherapy | see what correlated with mortality | shock P < 0.0001; admit to ICU P < 0.0001; pneumococcus resistant to penicillin and eryth P=0.02; Abx therapy with other than β-lactam +macrolide P=0.001 (greater risk than having resistant isolate) |
Lonks et al.50 Matched case control chart review | 86 case pts, 141 control pts with breakthrough bacteraemic pneumococcal pneumonia | pts on mostly β-lactam, macrolide, or a combination of the two | 19/86 case pts had I or R isolates drawn while on macrolide 0/141 control pts; but what about mortality? | of case pts 0/19 that got bacteraemia while on macrolide died; 12/67 (18%) case pts got bacteraemia not on macrolide died (P=0.06–probably due to low n) |
Stahl et al.43 Non-interventional prospective | 100 enrolled admitted CAP pts, 76 evaluable | within 24 h 68 got β-lactam, 12 got macrolide (11 with β-lactam) and rest other agents in groups of 1–2 per drug | mean length of stay (LOS) for evaluable group was 4.9 days | signif. ↓ LOS if got macrolide in 1st 24 h of admit (2.75 versus 5.3 days, P=0.01); only existed for early Abx use as if within 48 h (4.05 versus 5.2 days, P=0.19) |
Trowbridge et al.51 Retrospective | 450 enrolled, 376 evaluable mild to moderate hospitalized CAP, PSI Class III on average | np3gc+m (eryth or azith); FQ mono; azith mono; 2gc or np3gc mono; ββi mono | ANCOVA to compare LOS with antibiotic regimen, PSI score and categorical factors for antibiotic regimen and therapy site | when adjusted for PSI and site, azith and FQ monos and np3gc+azith had 1.7–2.7 days↓LOS than cephalosporin mono.; those that stayed on initial regimen this grew to 2.3–5.2 days–also np3gc+eryth versus pip-tazo mono =5.1 versus 9.2 days P < 0.05; initial empirical regimen having atypical coverage had 1.8–3.3 day↓LOS (P < 0.001) |
Lentino & Krasnicka52 Retrospective | n=94 hospitalized CAP, both groups had PSI score Class IV on average | group I – empirical initial other than mono with azith (n=50, 25 got ceftriaxone+macrolide); group II—mono azith (n=44) | treatment duration and LOS by empirical initial treatment group | minus ICU pts alive at 90 days treatment duration Group I versus Group II (10.8 versus 8 days, P=0.026); LOS 6.5 versus 4.4 days, P=0.012; whole population- treatment duration 11.4 versus 8.1 days, P=0.0007; LOS 9.7 versus 4.6 days, P < 0.0001; azith mono assoc. with $2400 per 2 day less LOS per pt episode |
Brown et al.49 Analysis of hospital claims database | Of 188,627 pts put in database 44,814 met criteria; hospitalized CAP pts Fine PSI Class A-D [= I–IV]; too much variability for Fine V (E) | mono with cef, oceph, FQ, macs or pcns or dual therapy of one of the β-lactams or FQ with a mac | impact of choice of initial empirical antibiotic therapy on 30 day mortality, total hospital costs and hospital LOS | mono pts—mac pts lowest mortality (2.2 versus ≥5%, P < 0.001) but were in lower risk classes; cef had↓LOS (mean 4.99 days versus 5.82–6.71 for other β-lactams and FQ P < 0.005) and assoc. cost ($7200 versus >$8200); dual therapy groups–cef+mac ↓ LOS versus all else (mean 4.98 versus 5.6–6.5 days, P < 0.005) and cost ($7900 versus $9700–11 300, P < 0.005); mac dual therapy 40%–70% ↓ mortality than pts got β-lactam or FQ mono (P < 0.05) |
Abx, antibiotics; np3gc, non-pseudomonal third-generation cephalosporin; 2gc + m, second-generation cephalosporin plus a macrolide; np3gc + m, non-pseudomonal third-generation cephalosporin plus a macrolide; FQ, fluoroquinolone; hosp., hospitalization; w/in, within; HR, hazard ratio; CI, confidence interval; I, intermediately susceptible; R, resistant; mono, monotherapy; azith, azithromycin; ββi, β-lactam/β-lactamase inhibitor; ANCOVA, analysis of covariance; cef, ceftriaxone; oceph, other cephalosporins; macs, macrolides; pcns, penicillins; PSI, Pneumonia Severity Index; pts, patients.
References
Kudoh, S. (
Kudoh, S. & Kimura, H. (
Kudoh, S., Azuma, A., Yamamoto, M. et al. (
Kudoh, S., Uetake, T., Hagiwara, K. et al. (
Yamamoto, M., Kondo, A., Tamura, M. et al. (
Schultz, M. J. (
Goswami, S. K., Kivity, S. & Marom, Z. (
Tamaoki, J., Isono, K., Sakai, N. et al. (
Tamaoki, J., Takeyama, K., Tagaya, E. et al. (
Tagaya, E., Tamaoki, J., Kondo, M. et al. (
Rubin, J.-I, Druce, H., Ramirez, O. E. et al. (
Kadota, J.-I., Sakito, O., Kohno, S. et al. (
Sakito, O., Kadota, J.-I., Kohno, S. et al. (
Amayasu, H., Yoshida, S., Ebana, S. et al. (
Koh, Y. Y., Lee, M. H., Sun, Y. H. et al. (
Tsang, K. W. T., Ho, P.-I., Chan, K.-n. et al. (
Gerhardt, S. G., McDyer, J. F., Girgis, R. E. et al. (
Equi, A., Balfour-Lynn, I. M., Bush, A. et al. (
Wolter, J., Seeney, S., Bell, S. et al. (
Saiman, L., Marshall, B. C., Mayer-Hamblett, N. et al. (
Howe, R. A. & Spencer, R. C. (
Kobayashi, H. (
Kobayashi, H. (
Yasuda, H., Ajiki, Y., Koga, T. et al. (
Vranes, J. (
Ichimiya, T., Yamasaki, T. & Nasu, M. (
Bui, K. Q., Banevicius, M. A., Nightingale, C. H. et al. (
Takeoka, K., Ichimiya, T., Yamasaki, T. et al. (
Yanagihara, K., Tomono, K., Sawai, T. et al. (
Tateda, K., Ishii, Y., Matsumoto, T. et al. (
Tateda, K., Comte, R., Pechere, J.-C. et al. (
Molinari, G., Guzmán, C. A., Pesce, A. et al. (
Nguyen, T., Louie, S. G., Beringer, P. M. et al. (
Yoshii, I., Nishimura, H., Yoshikawa, T. et al. (
Heffelfinger, J. D., Dowell, S. F., Jorgensen, J. H. et al. (
Macfarlane, J., Boswell, T., Douglas, G. et al. (
Memish, Z. A., Shibl, A. M., Ahmed, Q. A. A. et al. (
Pallares, R., Linares, J., Vadillo, M. et al. (
Mandell, L. A., Bartlett, J. G., Dowell, S. F. et al. (
Jung, S., Song, J., Oh, W. et al. (
Amsden, G. W., Baird, I. M., Simon, S. et al. (
Lode, H., File, T. M., Jr, Mandell, L. et al. (
Stahl, J. E., Barza, M. D., Desfardin, J. et al. (
Mufson, M. A. & Stanek, R. J. (
Martinez, J. A., Horcajada, J. P., Almela, M. et al. (
Weiss, K., Cortes, L., Beaupre, A., et al. (
Gupta, A. K., Rai, S., Farag, B. et al. (
Gleason, P. P., Meehan, T. P., Fine, J. M. et al. (
Brown, R. B., Iannini, P., Gross, P. et al. (
Lonks, J. R., Garau, J., Gomez, L. et al. (
Trowbridge, J. F., Artymowicz, R. J., Lee, C. E. et al. (
Lentino, J. R. & Krasnicka, B. (
Meehan, T. P., Fine, M. J., Krumholz, H. M. et al. (
Battleman, D. S., Callahan, M. & Thaler, H. T. (
Lehtomaki, K. (
Jay, S. J., Johnson, W. G., Jr & Pierce, W. K. (
Čulić, O., Eraković, V., Čepelak, I. et al. (
Kolling, U. K., Hansen, F., Braun, J. et al. (
Koch, C. C., Esteban, D. J., Chin, A. C. et al. (
Inamura, K., Ohta, N., Fukase, S. et al. (
Ives, T. J., Schwab, U. E., Ward, E. S. et al. (