Clinical and immunoregulatory effects of roxithromycin therapy for chronic respiratory tract infection

Clinical and immunoregulatory effects of roxithromycin therapy for chronic respiratory tract infection. H. Nakamura, S. Fujishima, T. Inoue, Y. Ohkubo, K. Soejima, Y. Waki, M. Mori, T. Urano, F. Sakamaki, S. Tasaka, A. Ishizaka, M. Kanazawa, K. Yamaguchi. #ERS Journals Ltd 1999. ABSTRACT: The clinical and immunoregulatory effects of long-term macrolide antibiotic therapy for patients with chronic lower respiratory tract infections (CLRTI) were investigated. Clinical parameters and neutrophil chemotactic mediators in the epithelial lining fluid (ELF) of CLRTI patients (n=10) were examined before and after 3 months oral administration of roxithromycin (RXM). The in vitro effects of RXM were also examined on the release of these mediators from alveolar macrophages (AM) and neutrophils. Arterial oxygen tension (p<0.05), vital capacity (VC) (p<0.001), %VC (p<0.05) and forced expiratory volume in one second (p<0.01) were improved after RXM treatment, but airway bacteria were not eradicated. Among the mediators, the levels of interleukin (IL)-8, neutrophil elastase (NE) and leukotriene B4 (LTB4) were higher in ELF than in plasma of CLRTI patients and they decreased after RXM treatment (n=7, p<0.05 for each). RXM concentrations were significantly increased in the bronchoalveolar lavage cells of the treated patients. In in vitro experiments, RXM showed inhibitory effects on IL-8 release from AM and neutrophils. In conclusion, interleukin-8, neutrophil elastase and leukotriene B4 contribute to the neutrophilic inflammation in the airways of chronic lower respiratory tract infection patients and the clinical effects of roxithromycin may, in part, be attributable to the suppression of excess release of the chemotactic mediators from inflammatory cells. Eur Respir J 1999; 13: 1371±1379. Dept of Medicine, School of Medicine, Keio University, Tokyo, Japan.

The clinical effectiveness of long-term macrolide antibiotic therapy has recently been reported in patients with diffuse panbronchiolitis (DPB) [1±3]. Before the introduction of macrolide therapy, DPB was a chronically progressive disease which resulted in chronic respiratory failure. DPB is characterized by chronic inflammation of the respiratory bronchioles and parabronchial and luminal infiltration of inflammatory cells, and is associated with continuous airway infection especially Pseudomonas aeruginosa [1,2]. This disease is prevalent mainly in Japan, but was recently reported in the USA [4]. Similar airway inflammation, namely chronic lower respiratory tract infections (CLRTI), is also observed in patients with bronchiectasis (BE). Previous studies have suggested that this effect cannot be ascribed to its antibacterial action [3,5] but rather, its immunoregulatory function, such as inhibition of neutrophil chemotaxis and superoxide production in vitro [5,6]. An in vivo study suggested that macrolide treatment induced a decrease in the neutrophil chemotactic activities of bronchoalveolar lavage (BAL) fluid from patients with DPB [2]. However, the precise mechanisms responsible for this clinical efficacy remain unclear.
Neutrophil accumulation is a defining feature of airway inflammation in patients with CLRTI including DPB and BE [1,2]. Although neutrophils are important effector cells against bacterial infections, neutrophil-derived oxygen metabolites and proteolytic enzymes could have harmful effects on lung tissues and may exacerbate clinical symptoms [1,7]. Various mediators including interleukin (IL)-8, IL-1b, leukotriene (LT)B 4 and neutrophil elastase (NE) may be involved in the recruitment and activation of neutrophils in patients with airway infections including CLRTI and cystic fibrosis [7±10]. Previous reports have indicated complex interactions between these mediators both in vivo and in vitro. For example, IL-8 is induced by IL-1b in various cells [11±13]. LTB 4 upregulates neutrophil-derived IL-8 [14]. IL-8 correlates with NE in the airways of CLRTI patients [7], and NE from patients with cystic fibrosis induces IL-8 gene expression in epithelial cells [15]. Thus, it is reasonable to speculate that the combination of these mediators synergistically induces airway inflammation. However, precise roles of these mediators in the pathogenesis of CLRTI have not been fully elucidated. Therefore, the neutrophil chemotactic mediators IL-8, IL-1b, tumour necrosis factor (TNF)-a, LTB 4 , NE and C5a were investigated in CLRTI patients.
In this study, it was hypothesized that macrolide antibiotics decrease the release of neutrophil chemotactic mediators from inflammatory cells, which results in clinical improvement. To prove this hypothesis, roxithromycin (RXM), a new macrolide with enhanced stability in gastric juice and a prolonged half-life [16], was administered to patients with CLRTI including DPB and BE, and the clinical symptoms, chest radiographic findings, pulmonary functions and sputum culture were examined. The concentrations of IL-8, IL-1b, TNF-a, LTB 4 , NE and C5a were measured in BAL fluid from the CLRTI patients before and after the treatment with RXM. Intracellular RXM concentrations in BAL cells, plasma and BAL supernatant were then measured, and the effects of RXM on the release of these chemotactic mediators from alveolar macrophages (AM) and neutrophils in culture investigated.

Subjects
Ten patients with CLRTI, who had complained of chronic purulent sputum for at least 6 months, and whose chest radiographs suggested bronchiolitis and/or BE were studied. Seven patients were diagnosed as DPB on the basis of the clinical diagnostic criteria published by the Japanese Ministry of Health and Welfare [2]. The remaining three patients were identified as having BE by computed tomography (CT) scan, which revealed bronchial dilatation without diffuse granular shadows. All patients had accompanying chronic sinusitis. Their average age (meanSEM) was 51.85.5 yrs, and none of the patients had ever smoked. No patient had prior treatment with glucocorticoids or long-term macrolides. Three control subjects were asymptomatic and had normal chest radiographic findings. Their average age was 45.36.7 yrs. Two were nonsmokers and the other was a smoker. Informed consent was obtained from all of the patients and control subjects.

Treatment and examination schedule
CLRTI patients were treated with RXM 150 mg b.i.d. for 3 months. No other antibiotics or glucocorticoids were administered during the period. Clinical symptoms were recorded by a pulmonologist in the outpatient clinic every 2 weeks. Clinical improvement after the 3-month treatment period was evaluated by three other pulmonologists who were not involved in this study. Based on physical examination and subjective findings, clinical symptoms were classified according to the criteria described in table 1. BAL, blood tests, pulmonary function tests, chest radiography and sputum culture were performed before and after the treatment. Improvement of the chest radiographic findings was assessed by three pulmonologists in a blinded manner. Briefly, three or four chest radiographs of each CLRTI patient, which were taken at different time points and randomly arranged, were evaluated without identification of the patient name and the examination date. The chest radiographic findings were classified according to the criteria described in table 1. Chest CT scans were also performed in all of the CLRTI patients to confirm the diagnosis. BE with peribronchial infiltration was demonstrated in all CLRTI patients when BE was defined as having dilated bronchi greater than the accompanying artery on the CT scans. Diffuse granular shadows with a centrilobular pattern were observed in DPB patients on the CT scans.

Bronchoalveolar lavage
BAL was performed using a previously described standard protocol [17,18]. In brief, a bronchoscope was wedged into a segment of the right middle lobe or lingula in which inflammatory findings were observed on a chest radiograph or CT scan. Sterile normal saline at room temperature was instilled through the bronchoscope in 50-mL volumes to a total of 150 mL. The recovered BAL fluid was filtered through sterile gauze, and the cells were counted. Cytospin preparations were made for cell differential counts. The fluid was centrifuged at 2006g for 5 min, and the supernatant was collected. No significant adverse effects were observed after the procedure.

Quantification of neutrophil chemotactic mediators
Concentrations of neutrophil chemotactic mediators in BAL fluid and plasma of CLRTI patients and control subjects were determined as follows. IL-8 (Genentech Inc., San Francisco, CA, USA), TNF-a (Otsuka, Tokushima, Japan) and NE (E. Merck, Darmstadt, Germany) levels were measured with a sandwich enzyme-linked immunosorbent assay (ELISA). LTB 4 (Amersham, Buckinghamshire, UK), IL-1b (Amersham) and C5a (Medgenix Diagnostics S.A., Brussels, Belgium) levels were measured using a radioimmunoassay (RIA). Epithelial lining fluid (ELF) concentrations of these mediators were calculated using a BAL fluid to plasma ratio of urea nitrogen [19].

Quantification of roxithromycin concentrations
RXM concentrations in plasma, BAL supernatant and BAL cells were determined by disc method using Micrococcus luteus American Type Culture Collection (ATCC; Rockville, MD, USA) 9341 [20]. BAL was performed 16 h after the final ingestion of RXM. Blood was drawn 12 h after the intake. BAL cells were frozen at -708C for 1 h and thawed at 378C for 1 h. These procedures were repeated three times, and intracellular RXM was extracted [21]. Intracellular RXM concentrations of BAL cells were calculated using 2,491 mm 3 as the volume of a macrophage [22]. Neutrophil volume was not considered for this calculation because the total volume in BAL cells is much smaller and its life span is much shorter than a macrophage.
In vitro cytokine production and release from alveolar macrophages and neutrophils BAL cells containing >90% AM from four control subjects were used to examine the potential of AM to produce and release cytokines. The cells were divided into four groups and each group was preincubated with RPMI-1640 containing either 0, 0.5, 5 or 50 mg . mL -1 of RXM for 30 min. Then, 1 mg . mL -1 of lipopolysaccharide (LPS; Escherichia coli 055: B5; Difco, Detroit, MI, USA) was added and the cells were cultured for 4 h. To quantify cellassociated IL-8, the cells were lysed, and cell-associated IL-8, and extracellular IL-8 and TNF-a were measured by ELISA, and extracellular IL-1b was measured by RIA, as described above. Peripheral blood neutrophils were separated by a density gradient centrifugation from seven healthy volunteers. Neutrophil purity was >98%. The preincubation and culture procedures were the same as those for AM except that extracellular TNF-a and IL-1b were not measured.

Bronchoalveolar lavage cells and chemotactic mediators in plasma and bronchoalveolar lavage fluid of untreated patients
BAL cell counts and differentials before and after RXM treatment are shown in table 5. BAL was performed in all 10 CLRTI patients before treatment and in seven patients after RXM treatment. Three patients did not agree to the second bronchoscopy. The percentage of neutrophils was increased in CLRTI patients compared with that in the control subjects before treatment (p<0.05). After treatment with RXM, total cell, lymphocyte, and neutrophil counts were significantly decreased (p<0.05, for each by Wilcoxon test). The decrease in total cell and neutrophil counts was also significant in DPB (n=5) (p<0.05 for both). The percentage of macrophages increased and that of neutrophils decreased after treatment in DPB (p<0.05 for both) as well as in all CLRTI patients. The percentage of neutrophils was higher in untreated DPB (n=5) than in control subjects (p<0.0001). Total cell counts, neutrophil counts and percentage before treatment in BE (n=3) were 97.357.9 610 6 cells . mL ELF -1 , 69.3 60.2610 6 cells . mL ELF -1 , and 46.422.3%, respectively.
Pretreatment levels of neutrophil chemotactic mediators in ELF and plasma are shown in table 6. IL-8 concentrations in the ELF of CLRTI patients were higher than those of control subjects (p<0.05). IL-8, IL-1b, NE and LTB 4 concentrations in the ELF from CLRTI patients were increased compared with plasma concentrations (n=8, p<0.01, p<0.01, p<0.05 and p<0.05, respectively). The mediator concentrations in ELF were compared between DPB (n=7) and BE (n=3). IL-8 levels were lower in DPB than in BE (34.18.5 versus 88.219.4 ng . mL -1 , p<0.05) and were higher in both DPB and BE than in

Roxithromycin concentrations in plasma, epithelial lining fluid and bronchoalveolar lavage cells
After three months of administration, RXM concentrations in plasma, BAL supernatants and BAL cells were measured in CLRTI patients. As shown in table 7, intracellular RXM concentrations in AM were markedly increased compared with those in plasma and ELF (p< 0.001), but ELF concentrations were similar to those in plasma.

Effects of roxithromycin on interleukin-8, tumour necrosis factor-a and interleukin-1b release from alveolar macrophages
The low concentration of RXM (0.5 mg . mL -1 ) increased extracellular and total IL-8 from AM, while the high concentration of RXM (50 mg . mL -1 ) decreased the release of IL-8 and increased the amount of intracellular IL-8 ( fig.  2a). Figure 2b indicated decreased release of IL-8 from neutrophils at the high RXM concentration. These observations suggested that the high concentration of RXM inhibited the release, but not the production, of IL-8 in AM and neutrophils. RXM (50 mg . mL -1 ) decreased extracellular release of TNF-a from AM ( fig. 3a), while there were no significant effects of RXM on the release of IL-1b ( fig. 3b).

Discussion
RXM significantly improved the clinical symptoms of CLRTI patients despite its failure to eliminate airway bacteria. It has been demonstrated in this study that various chemotactic mediators including IL-8, NE, and C5a contribute to the airway inflammation of CLRTI patients. The clinical improvement after the RXM treatment was associated with a decrease in IL-8, NE, and LTB 4 concentrations in ELF and the resulting attenuation of neutrophilic inflammation. A marked RXM accumulation in BAL cells was demonstrated in vivo and the inhibitory effects of RXM on the release of pro-inflammatory cytokines were observed in vitro. These findings support the hypothesis that long-term administration of RXM leads to decreased chemotactic mediator release from inflammatory cells in the airways of CLRTI patients, which may in part account for the clinical efficacy of RXM.
IL-8 was considered to play a primary role in neutrophilic inflammation in the airways of CLRTI patients on the basis of the comparison between the patients and control subjects, before and after the treatment, and the relationship to the ELF neutrophil counts. Although the major IL-8 producing cells in the airways of CLRTI patients were undetermined, AM, neutrophils and epithelial cells can produce IL-8 [13,18,23]. NE contributes to the neutrophilic inflammation as it correlated with the ELF neutrophil counts and decreased after treatment with RXM. NE levels also paralleled IL-8 levels. Previous reports suggest that NE is released from neutrophils activated by IL-8 [24] and induces IL-8 expression in epithelial cells [15]. It was also reported that activated neutrophils produce LTB 4 [25], and that LTB 4 upregulates neutrophil-derived IL-8 [14]. LTB 4 levels in ELF from CLRTI patients were increased compared with those in plasma, but did not significantly correlate with neutrophil counts or IL-8 in the present study. These results suggest that neutrophils may not be a primary source of LTB 4 in the airways of CLRTI patients. Since LTB 4 decreased after treatment with RXM, the effects of RXM may be at least partly mediated by a decrease in LTB 4 levels. IL-1b correlated with IL-8 and NE levels in ELF but not with neutrophil counts. The ELF IL-1b levels were not higher in CLRTI patients compared to control subjects, and did not fall after RXM treatment. C5a concentrations correlated well with neutrophil counts and NE levels in ELF, but the ELF C5a levels in CLRTI patients were similar to those in control subjects. Furthermore, they did not decrease after RXM treatment, and did not correlate with IL-8. Therefore, RXM effects may not be mediated by a decrease in C5a levels. It is unlikely that TNF-a plays important roles in the pathogenesis of CLRTI as its ELF levels were much lower than other mediators and did not change after treatment with RXM.
The effects of RXM on clinical symptoms, pulmonary functions and chest radiographic findings in CLRTI patients were remarkable. Clinical symptoms and chest radiographic findings were improved in all CLRTI patients studied based on the author's scoring criteria, but the percentage of improvement varied in each patient. There was no significant difference in the improvement of these scores between DPB and BE. Further investigations will be required to elucidate the factors determining the response of CLRTI patients to macrolide antibiotic therapy, including chest radiographic findings, infected bacteria, pulmonary function, etc. However, these results suggested that improvement in clinical and chest radiographic findings paralleled the decreased neutrophil accumulation in the airspaces, but was not always accompanied by a decrease in the number of airway bacteria. Although clinical findings paralleled decreases in the amount of bacteria in 30% of patients, they were improved despite an increased or unchanged number of bacteria in 50% of patients. The uncontrolled production of neutrophil chemotactic mediators, including IL-8, by inflammatory cells may play major roles in recruiting neutrophils to the airways and may exacerbate respiratory symptoms, especially in DPB patients. Decreased ELF levels of IL-8, NE and LTB 4 after the RXM treatment suggest that RXM could inhibit the excess release of these mediators from inflammatory cells in vivo, which was supported by the inhibitory effects of RXM on the release of IL-8 from AM and neutrophils in vitro. Although the in vitro RXM effects were significant only at a high concentration, RXM was markedly concentrated in AM after 3 months of administration. In addition, the in vivo results demonstrated that a decrease in IL-8 levels can lead to a decrease in other chemotactic mediators including NE and IL-1b, which may efficiently attenuate neutrophilic inflammation in the airways. Therefore, RXM may actually inhibit the release of IL-8 and improve neutrophilic inflammation in vivo. Previous reports also suggest that macrolide antibiotics can accumulate within phagocytes to higher levels than the other antibiotics, and that this process is more pronounced for RXM than for erythromycin (EM) [26±29]. These results may partly explain the unique effects of macrolide antibiotics on CLRTI patients, and suggest a potential superior action of RXM over EM.
In addition to the inhibitory effects on IL-8 release, various immunoregulatory effects have been reported for macrolide antibiotics. These effects include inhibition of neutrophil chemotaxis [5] and superoxide production [6], resolution of biofilms surrounding bacteria [30], activation of cilial movement [31] and a decrease in mucus secretion [32]. The clinical efficacy of macrolides may be a consequence of all these actions. Since the attenuation of chemotaxis and activation of activated neutrophils by macrolides is distinct among these mechanisms, macrolides may be applicable to other neutrophil-mediated pulmonary disorders, such as cystic fibrosis, adult respiratory distress syndrome and subacutely exacerbating idiopathic pulmonary fibrosis [4,5,18,33].
RXM has a longer plasma half-life than EM [16]. Sufficient plasma concentrations can be obtained when a 150 mg tablet is administered twice a day, which is equivalent  before and after treatment with roxithromycin (RXM). The levels were significantly decreased after treatment. b) Neutrophil elastase (NE) concentrations in ELF before and after treatment with RXM, which were significantly decreased after treatment. c) Leukotriene B 4 (LTB 4 ) concentrations in ELF before and after treatment with RXM, which were significantly decreased after treatment. *: p<0.05. to 1,200 mg daily of EM [34]. In contrast, low-dose EM (200±600 mg daily) has been regularly administered to CLRTI patients [1±3] and the major reason for the low dose is to avoid the adverse effects of EM. Although a standard dose (300 mg . day -1 ) of RXM was given for 3 months in this study, no patient showed any signs or symptoms of adverse effects. The clinical signs and symptoms of the treated patients were markedly improved within 3 months, and the clinical efficacy was observed within 4 weeks after commencement of RXM treatment in most patients. However, it is unclear how long this improvement will continue after the discontinuation of RXM. The optimal duration of the treatment has not been determined, but oral macrolide therapy is often continued over several years in Japan. The duration of the treatment is currently optimized according to the clinical course of an individual patient. The authors reduced the RXM dose to 150 mg . day -1 after the three month treatment with 300 mg of RXM. If the clinical efficacy is due to the intracellularly accumulated RXM in AM, neutrophilic inflam-mation could recur after the discontinuation of RXM. Longer observation is necessary to establish a standard protocol for the treatment of CLRTI patients with macrolide antibiotics. In summary, high levels of the chemotactic mediators interleukin-8, neutrophil elastase and leukotriene B 4 exist in the airway of chronic lower respiratory tract infection patients. Treatment with roxithromycin resulted in a remarkable clinical improvement, which is associated with decreased levels of these mediators. In combination with the results of in vitro experiments, it is speculated that the clinical efficacy may be mediated by the inhibitory effects of intracellularly accumulated roxithromycin on the excess release of chemotactic mediators, especially interleukin-8, from inflammatory cells.