Abstract
Macrolide-resistant Mycoplasma pneumoniae (MR M. pneumoniae) has been isolated from clinical specimens in Japan since 2000. A comparative study was carried out to determine whether or not macrolides are effective in treating patients infected with MR M. pneumoniae. The clinical courses of 11 patients with MR M. pneumoniae infection (MR patients) treated with macrolides were compared with those of 26 patients with macrolide-susceptible M. pneumoniae infection (MS patients). The total febrile days and the number of febrile days during macrolide administration were longer in the MR patients than in the MS patients (median of 8 days versus median of 5 days [P = 0.019] and 3 days versus 1 day [P = 0.002], respectively). In addition, the MR patients were more likely than the MS patients to have had a change of the initially prescribed macrolide to another antimicrobial agent (63.6% versus 3.8%; odds ratio, 43.8; P < 0.001), which might reflect the pediatrician's judgment that the initially prescribed macrolide was not sufficiently effective in these patients. Despite the fact that the febrile period was prolonged in MR patients given macrolides, the fever resolved even when the initial prescription was not changed. These results show that macrolides are certainly less effective in MR patients.
Mycoplasma pneumoniae is a common pathogen causing community-acquired respiratory tract infection mainly in children and young adults. Macrolides are generally considered to be the first-choice agents for treatment of M. pneumoniae infection. Although tetracyclines and fluoroquinolones are effective against M. pneumoniae, these agents are not recommended for children because of their toxicity. Tetracyclines can cause depression of bone growth, permanent gray-brown discoloration of the teeth, and enamel hypoplasia when given during tooth development. Although the clinical importance of fluoroquinolones has not been demonstrated, they produce cartilage erosion in young animals. Thus, these agents should be given only when there is no alternative (15).
As reported by Lucier et al. (9) and Okazaki et al. (14), an A-to-G transition or A-to-C transversion at position 2063 or 2064 of domain V of the M. pneumoniae 23S rRNA gene results in resistance to macrolide antibiotics. We have previously reported the isolation of macrolide-resistant (MR) M. pneumoniae from ca. 20% of clinical specimens collected from pediatric patients in Japan (11). Most of those isolates were highly resistant to 14-membered ring macrolides (MIC, >256 μg/ml) and moderately resistant to 15- and 16-membered ring macrolides.
Even in the cases of patients infected with MR M. pneumoniae, some pediatricians had the impression that there was a good response to macrolide therapy (11). There is a similar debate about the management of infection due to pneumococci. As noted in The Infectious Diseases Society of America (IDSA) guidelines for community-acquired pneumonia management (10), despite the increase of resistant isolates, a corresponding increase has not been seen in the number of clinical treatment failures.
One possible explanation for this is the nonantimicrobial effects of macrolides. It is known that macrolides have beneficial immunomodulating effects (1, 4, 6, 20), and they are clinically effective in hypersecretory conditions such as diffuse panbronchiolitis (7, 8) and cystic fibrosis (16). Thus, macrolides could be clinically effective even in MR M. pneumoniae infections.
It is important to know the clinical significance of MR M. pneumoniae infection, because in vitro susceptibility testing for M. pneumoniae is not available for daily management of patients. If macrolides are clinically effective against MR M. pneumoniae infection, pediatricians do not need to consider the use of tetracyclines or fluoroquinolones, even if the prevalence of MR M. pneumoniae rises in the future. Therefore, we performed a comparative study to determine whether or not MR M. pneumoniae influences the clinical outcome in patients treated with macrolides.
MATERIALS AND METHODS
Study population and sample collection.
Three pediatric clinics in three different areas in Japan participated in this study. Sera and throat swabs or sputa taken from inpatients or outpatients suspected of M. pneumoniae infection were subjected to the laboratory tests.
Isolation.
Isolation and identification of M. pneumoniae was carried out as described in a previous report (11).
PCR detection of M. pneumoniae.
Sputa were obtained from patients, suspended in a small amount of saline, mixed well, and centrifuged at 2,000 rpm for 15 min, and then DNA was extracted from the supernatant with a QIAamp DNA Mini kit (QIAGEN K. K., Tokyo, Japan) according to the manufacturer's instructions. M. pneumoniae DNA was detected by the nested PCR method with primer sets for amplification of the P1 gene as previously described (17). The first primer set was ADH2F (5′-GGC AGT GGC AGT CAA CAA ACC ACG TAT-3′) and ADH2R (5′-GAA CTT AGC GCC AGC AAC TGC CAT-3′). The second primer set was ADH3F (5′-GAA CCG AAG CGG CTT TGA CCG CAT-3′) and ADH3R (5′-GTT GAC CAT GCC TGA GAA CAG TAA-3′).
Serological diagnosis.
Particle agglutination (PA) antibody titers for M. pneumoniae were assayed by using Serodia-MYCO II (Fuji Rebio Ltd., Tokyo, Japan), which is manufactured using artificial gelatin particles, sensitized with cell membrane components of M. pneumoniae, according to the manufacturer's instructions.
Detection of resistance point mutations in domain V of 23S rRNA.
MR M. pneumoniae isolates were screened on the basis of MIC of erythromycin (ERY), and identification of point mutations in domain V of 23S rRNA for ERY-resistant M. pneumoniae was performed according to our previously reported methods (11). For PCR-positive samples of M. pneumoniae DNA, the detection of a point mutation is indicative of a resistant phenotype because there is only a single rRNA operon in the genome (2). Neither plasmids with erm genes to mediate ribosomal modification nor any enzymes that inactivate macrolides have been found in M. pneumoniae. Thus, the prevalence of MR M. pneumoniae detected by the PCR methodology should reflect the true incidence of resistant strains.
Patient extraction for comparison of clinical courses.
Clinical information was collected for the patients from whom M. pneumoniae had been isolated or its DNA detected. Patients who fulfilled the following criteria were extracted: (i) M. pneumoniae infection was laboratory confirmed, (ii) macrolides were prescribed during the illness, and (iii) complete information about prescribed antimicrobial agents and febrile days was available from the medical record. Laboratory-confirmed M. pneumoniae infection was defined as (1-a) isolation of M. pneumoniae from throat swabs or (1-b) detection of M. pneumoniae DNA from the sputum by PCR methods and serologically positive results, i.e., fourfold or greater rise of antibody titer in paired serum samples or titer higher than 1:640 in a single-serum sample by PA assay.
Patients infected with M. pneumoniae showing a point mutation in domain V of the 23S rRNA gene were defined as MR M. pneumoniae-infected patients (MR patients), and those infected with M. pneumoniae without the mutation were defined as macrolide-susceptible M. pneumoniae-infected patients (MS patients). MS patients were selected from the same study population as MR patients, and there were approximately twice as many of them as MR patients.
Measurement of clinical efficacy.
To compare the clinical courses of MR and MS patients, we adopted the number of febrile days as a main outcome measurement. A febrile day was defined as a day during which the body temperature exceeded 38.0°C at least once. Total febrile days and the number of febrile days during macrolide administration were assessed. As these parameters would be affected by the time of commencement of macrolide administration, the number of febrile days before macrolide administration was also assessed. Other clinical symptoms and signs, such as cough and chest roentgenogram findings, were not taken into account in this study on account of the difficulty of objective and unified assessment through a retrospective review of medical records.
The numbers of patients whose prescribed antibiotic was changed were also compared. We speculated that a change in prescribed antimicrobial agent might reflect the pediatrician's clinical decision that the initial therapy had insufficient efficacy based on the general clinical condition of the patients. The pediatricians had no information about the susceptibility of M. pneumoniae at the time of clinical decision-making.
Statistical analysis was performed using SPSS software, version 9.05 for Windows (SPSS, Inc., Chicago). Differences in categorical variables were assessed with the two-tailed Fisher's test, and for the comparison of medians the exact Wilcoxon rank-sum test was used. P values of less than 0.05 were considered to indicate statistical significance.
RESULTS
Prevalence of MR M. pneumoniae.
The prevalence of MR M. pneumoniae among clinical isolates and specimens with positive M. pneumoniae DNA is shown in Table 1. Before 1999, no MR M. pneumoniae was found among 296 clinical isolates. In 2000, however, MR M. pneumoniae appeared in 10% of isolates, and its prevalence rose to 33.3% in 2001. The overall prevalence of MR M. pneumoniae during 2000 to 2004 was 15.9%. All MR M. pneumoniae isolates had a resistance point mutation in domain V of 23S rRNA. A similar trend was seen in specimens with PCR-positive M. pneumoniae. Although the number of positive specimens before 1999 was limited (n = 12), no MR M. pneumoniae was detected. The prevalence of MR M. pneumoniae during 2000 to 2004, based on PCR-positive specimens, was 15.2%.
TABLE 1.
Prevalence of macrolide-resistant M. pneumoniae in Japan
| Year | Isolation method
|
DNA detection method
|
||
|---|---|---|---|---|
| No. of isolates | No. of resistant isolates (%) | No. of specimens with positive M. pneumoniae DNA (no. of examined specimens) | No. of specimens with macrolide resistance mutation in M. pneumoniae DNA (%) | |
| 1999 | 296 | 0 | 12 (630) | 0 |
| 2000 | 10 | 1 (10.0) | 9 (92) | 4 (44.4) |
| 2001 | 6 | 2 (33.3) | 28 (384) | 3 (10.7) |
| 2002 | 12 | 3 (25.0) | 44 (352) | 5 (11.4) |
| 2003 | 54 | 7 (13.0) | 10 (236) | 1 (10.0) |
| 2004 | 6 | 1 (16.7) | 8 (183) | 2 (25.0) |
| 5-year total (2000-2004) | 88 | 14 (15.9) | 99 (1247) | 15 (15.2) |
Comparison of the clinical courses between MR and MS patients.
Eleven MR patients were selected for the analysis according to the criteria given above, and 26 MS patients were used as controls.
The patients' characteristics are summarized in Table 2. All patients were outpatients at the time of onset and had no severe underlying disease that might have influenced the clinical course. MR patients tended to be older and had a lower male/female ratio than MS patients, but the differences lacked statistical significance. Most patients were first prescribed β-lactam antimicrobial agents by primary physicians, followed by prescription of macrolides after attendance at a hospital. The prescribed macrolides differed among MR and MS patients. Significantly more MR patients than MS patients were prescribed 14-membered ring macrolides (72.7% versus 26.9%; P = 0.025). The majority of MS patients (19 out of 26 [73.1%]) were prescribed only 15-membered ring macrolides (azithromycin [AZM]).
TABLE 2.
Characteristics of enrolled patients
| Characteristic | MR patients (n = 11) | MS patients (n = 26) | P |
|---|---|---|---|
| Age (yr) | |||
| Median (range) | 9.0 (0-13) | 5.5 (1-14) | 0.30 |
| Mean | 7.6 | 6.5 | |
| Sex, male/female | 4/7 | 14/12 | 0.33 |
| No. of patients prescribed 14-membered ring macrolides (%) | 8 (72.7) | 7 (26.9) | 0.025 |
The clinical courses in the MR and MS patients are summarized in Table 3. The total febrile days and the number of febrile days during macrolide administration were significantly greater in MR patients than in MS patients (median of 8 days versus 5 days [P = 0.019] and 3 days versus 1 day [P = 0.002], respectively). Febrile periods before macrolide administration, which consist of antimicrobial-free and mostly β-lactam-administered days, showed no statistically significant difference (median of 3 days versus 4 days, P = 0.402).
TABLE 3.
Comparison of clinical courses in MR patients and MS patients
| Characteristic | MR patients (n = 11) | MS patients (n = 26) | P |
|---|---|---|---|
| Febrile days | |||
| Median (range) | 8 (4-19) | 5 (2-9) | 0.019 |
| Mean | 9.3 | 5.5 | |
| Febrile days during macrolide administration | |||
| Median (range) | 3 (1-11) | 1 (1-5) | 0.002 |
| Mean | 4.3 | 1.4 | |
| Febrile days before macrolide administration | |||
| Median (range) | 3 (1-10) | 4 (1-8) | 0.402 |
| Mean | 3.8 | 4.1 | |
| No. of patients with a febrile period exceeding 48 h after macrolide administration (%) | 8 (72.7) | 5 (19.2) | 0.006 |
| No. of patients with a change of prescription after macrolide administration (%) | 7 (63.6) | 1 (3.8) | <0.001 |
The MR patients were more likely to have had the initially prescribed macrolide changed to another antimicrobial agent by their pediatricians (63.6% versus 3.8%; odds ratio, 43.8; P < 0.001). Among seven MR patients whose prescriptions were changed, all but one were changed to minocycline.
The results were similar for patients to whom 14-membered ring macrolides were administered (Table 4). Among these 15 patients (8 MR patients and 7 MS patients), 9 patients were prescribed clarithromycin, while the remaining 6 were prescribed ERY. Presumably due to the fact that the number of febrile days during macrolide administration was greater in MR patients than in MS patients (median of 3.5 days versus 1.0 day, P = 0.004), the initially prescribed macrolide was more frequently changed among MR patients than MS patients (75% versus 0%, P = 0.007). Although there was no statistical significance, there was a prolongation of total febrile days for MR patients (median of 10 days versus 6 days, P = 0.152).
TABLE 4.
Comparison of patients prescribed 14-membered ring macrolides
| Characteristic | MR patients (n = 8) | MS patients (n = 7) | P |
|---|---|---|---|
| Total febrile days | |||
| Median (range) | 10.0 (4-19) | 6.0 (4-9) | 0.152 |
| Mean | 10.4 | 6.6 | |
| Febrile days during macrolide administration | |||
| Median (range) | 3.5 (1-11) | 1.0 (1-2) | 0.004 |
| Mean | 4.9 | 1.1 | |
| Febrile days before macrolide administration | |||
| Median (range) | 3.0 (1-10) | 5.0 (3-8) | 0.152 |
| Mean | 4.0 | 5.4 | |
| No. of patients with a febrile period exceeding 48 h after macrolide administration (%) | 7 (87.5) | 1 (14.3) | 0.01 |
| No. of patients with a change of prescription after macrolide administration (%) | 6 (75.0) | 0 | 0.007 |
When we focused on patients given 15-membered ring macrolides, 2 MR patients and 19 MS patients, the differences were not clear. Although there were only two MR patients in this group, their total febrile days and number of febrile days during macrolide administration were not different from those of MS patients (medians of 4.5 days versus 5.0 days and 1.0 day versus 1.0 day, respectively).
DISCUSSION
There are few reports on the isolation of MR M. pneumoniae from clinical specimens, and most of the isolates were obtained following ERY treatment (13, 19). In our survey, MR M. pneumoniae was not found in any of 296 clinical isolates or 12 M. pneumoniae PCR-positive specimens collected between 1983 and 1999, but it has been found in 15% to 20% of clinical isolates or PCR-positive specimens since 2000. MR M. pneumoniae first appeared in 2000 and rapidly spread throughout Japan (11, 12). Thus, it is important to evaluate the clinical significance of MR M. pneumoniae.
In our study, when patients infected with MR M. pneumoniae were treated with macrolides, the total febrile period was 3 days longer than that of patients with MS M. pneumoniae. Although we did not assess other clinical outcome variables, such as chest roentgenogram findings, a higher frequency of changes in prescription was observed in MR patients than in MS patients. This might reflect the pediatrician's judgment, based on the patient's clinical condition, that the initially prescribed macrolide was not sufficiently effective, even though the pediatricians had no information about the susceptibility of isolates at the time of clinical management. This tendency was also seen in patients who were treated only with 14-membered ring macrolides.
It was difficult to assess the immunomodulatory effects of macrolides in patients with M. pneumoniae infection in this study, because all the patients enrolled were prescribed macrolides according to the inclusion criteria. To evaluate the immunomodulatory effects of macrolides, it will be necessary to compare the clinical outcomes among MR patients treated with and without macrolides. An alternative is to compare the number of febrile days of MR patients with that of patients without antimicrobial agent therapy in the literature. According to review articles, fever might persist for about a week in the natural course of M. pneumoniae infection (3, 18). Kingston et al. (5) evaluated the effect of demethylchlortetracycline in a double-blind study, and the mean duration of fever in the treated group was 2.13 days, while it was 8.14 days in the placebo group. They started to count the number of febrile days not at the point of onset but only after entry into the study. In our study, the mean number of febrile days of MR patients was 9.2, which is similar to that of the placebo group in Kingston's study. This implies that the antimicrobial effect is dominant over immunomodulatory effects in macrolide therapy, at least as far as duration of fever in M. pneumoniae infection is concerned. On the other hand, we did not assess the duration of other symptoms, such as malaise, sore throat, and cough, and it is possible that the immunomodulatory effects of macrolides can shorten these symptoms even in MR M. pneumoniae infection.
A difference of three febrile days in MR patients might not have a great impact in the management of M. pneumoniae infection, because it is often a mild and self-limiting disease, and the fever resolved even when the initially prescribed macrolide was not changed. However, it is reasonable to consider the use of alternative antimicrobial agents, such as minocycline, when macrolides are less effective than expected in patients more than 8 years old with possible M. pneumoniae infection.
The criteria for M. pneumoniae infection used in this study were stringent enough to confirm acute M. pneumoniae infection. This was a retrospective study based on a review of medical records, and patients with incomplete records were excluded. In general, clinical records of patients showing mild illness with M. pneumoniae infection were incomplete, and their clinical evaluation was excluded from this study.
In conclusion, we compared clinical outcomes in 11 MR patients and 26 MS patients given macrolide therapy. The MR patients showed more febrile days (by a median of 2 days) during the initial macrolide therapy than MS patients. On the other hand, no apparent treatment failure or serious illness was reported for MR patients. The influence of the emergence of MR M. pneumoniae on the treatment for M. pneumoniae infection deserves further study.
Acknowledgments
We thank Hitomi Oya of Kanagawa Prefectural Institute of Public Health, Kanagawa, Japan, for her technical assistance and useful discussions.
This work was partly supported by a Grant for Studies on Emerging and Re-emerging Infectious Diseases (H15-Shinko-24) from the Ministry of Health, Labor and Welfare, Japan.
REFERENCES
- 1.Abe, S., H. Nakamura, S. Inoue, H. Takeda, H. Saito, S. Kato, N. Mukaida, K. Matsushima, and H. Tomoike. 2000. Interleukin-8 gene repression by clarithromycin is mediated by the activator protein-1 binding site in human bronchial epithelial cells. Am. J. Respir. Cell Mol. Biol. 22:51-60. [DOI] [PubMed] [Google Scholar]
- 2.Dandekar, T., M. Huynen, J. T. Regula, B. Ueberle, C. U. Zimmermann, M. A. Andrade, T. Doerks, L. Sanchez-Pulido, B. Snel, M. Suyama, Y. P. Yuan, R. Herrmann, and P. Bork. 2000. Re-annotating the Mycoplasma pneumoniae genome sequence: adding value, function and reading frames. Nucleic Acids Res. 28:3278-3288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Denny, F. W., W. A. Clyde, Jr., and W. P. Glezen. 1971. Mycoplasma pneumoniae disease: clinical spectrum, pathophysiology, epidemiology, and control. J. Infect. Dis. 123:74-92. [DOI] [PubMed] [Google Scholar]
- 4.Ichiyama, T., M. Nishikawa, T. Yoshitomi, S. Hasegawa, T. Matsubara, T. Hayashi, and S. Furukawa. 2001. Clarithromycin inhibits NF-κB activation in human peripheral blood mononuclear cells and pulmonary epithelial cells. Antimicrob. Agents Chemother. 45:44-47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kingston, J. R., R. M. Chanock, M. A. Mufson, L. P. Hellman, W. D. James, H. H. Fox, M. A. Manko, and J. Boyers. 1961. Eaton agent pneumonia. JAMA 176:118-123. [DOI] [PubMed] [Google Scholar]
- 6.Kohyama, T., H. Takizawa, S. Kawasaki, N. Akiyama, M. Sato, and K. Ito. 1999. Fourteen-member macrolides inhibit interleukin-8 release by human eosinophils from atopic donors. Antimicrob. Agents Chemother. 43:907-911. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Koyama, H., and D. M. Geddes. 1997. Erythromycin and diffuse panbronchiolitis. Thorax 52:915-918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kudoh, S., A. Azuma, M. Yamamoto, T. Izumi, and M. Ando. 1998. Improvement of survival in patients with diffuse panbronchiolitis treated with low-dose erythromycin. Am. J. Respir. Crit. Care Med. 157:1829-1832. [DOI] [PubMed] [Google Scholar]
- 9.Lucier, T. S., K. Heitzman, S. K. Liu, and P. C. Hu. 1995. Transition mutations in the 23S rRNA of erythromycin-resistant isolates of Mycoplasma pneumoniae. Antimicrob. Agents Chemother. 39:2770-2773. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mandell, L. A., J. G. Bartlett, S. F. Dowell, T. M. File, Jr., D. M. Musher, and C. Whitney. 2003. Update of practice guidelines for the management of community-acquired pneumonia in immunocompetent adults. Clin. Infect. Dis. 37:1405-1433. [Epub ahead of print.] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Matsuoka, M., M. Narita, N. Okazaki, H. Ohya, T. Yamazaki, K. Ouchi, I. Suzuki, T. Andoh, T. Kenri, Y. Sasaki, A. Horino, M. Shintani, Y. Arakawa, and T. Sasaki. 2004. Characterization and molecular analysis of macrolide-resistant Mycoplasma pneumoniae clinical isolates obtained in Japan. Antimicrob. Agents Chemother. 48:4624-4630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Morozumi, M., K. Hasegawa, R. Kobayashi, N. Inoue, S. Iwata, H. Kuroki, N. Kawamura, E. Nakayama, T. Tajima, K. Shimizu, and K. Ubukata. 2005. Emergence of macrolide-resistant Mycoplasma pneumoniae with a 23S rRNA gene mutation. Antimicrob. Agents Chemother. 49:2302-2306. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Niitu, Y., S. Hasegawa, T. Suetake, H. Kubota, S. Komatsu, and M. Horikawa. 1970. Resistance of Mycoplasma pneumoniae to erythromycin and other antibiotics. J. Pediatr. 76:438-443. [DOI] [PubMed] [Google Scholar]
- 14.Okazaki, N., M. Narita, S. Yamada, K. Izumikawa, M. Umetsu, T. Kenri, Y. Sasaki, Y. Arakawa, and T. Sasaki. 2001. Characteristics of macrolide-resistant Mycoplasma pneumoniae strains isolated from patients and induced with erythromycin in vitro. Microbiol. Immunol. 45:617-620. [DOI] [PubMed] [Google Scholar]
- 15.Reese, R. E., and R. F. Betts. 2003. Tetracyclines, p. 1112-1116. In R. F. Betts, S. W. Chapman, and R. L. Penn (ed.), A practical approach to infectious diseases, 5th ed. Lippincott Williams and Wilkins, Philadelphia, Pa.
- 16.Saiman, L., B. C. Marshall, N. Mayer-Hamblett, J. L. Burns, A. L. Quittner, D. A. Cibene, S. Coquillette, A. Y. Fieberg, F. J. Accurso, and P. W. Campbell III. 2003. Azithromycin in patients with cystic fibrosis chronically infected with Pseudomonas aeruginosa: a randomized controlled trial. JAMA 290:1749-1756. [DOI] [PubMed] [Google Scholar]
- 17.Sasaki, T., T. Kenri, N. Okazaki, M. Iseki, R. Yamashita, M. Shintani, Y. Sasaki, and M. Yayoshi. 1996. Epidemiological study of Mycoplasma pneumoniae infections in Japan based on PCR-restriction fragment length polymorphism of the P1 cytadhesin gene. J. Clin. Microbiol. 34:447-449. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Smith, C. B., W. T. Friedewald, and R. M. Chanock. 1967. Shedding of Mycoplasma pneumoniae after tetracycline and erythromycin therapy. N. Engl. J. Med. 276:1172-1175. [DOI] [PubMed] [Google Scholar]
- 19.Stopler, T., C. B. Gerichter, and D. Branski. 1980. Antibiotic-resistant mutants of Mycoplasma pneumoniae. Isr. J. Med. Sci. 16:169-173. [PubMed] [Google Scholar]
- 20.Takizawa, H., M. Desaki, T. Ohtoshi, T. Kikutani, H. Okazaki, M. Sato, N. Akiyama, S. Shoji, K. Hiramatsu, and K. Ito. 1995. Erythromycin suppresses interleukin 6 expression by human bronchial epithelial cells: a potential mechanism of its anti-inflammatory action. Biochem. Biophys. Res. Commun. 210:781-786. [DOI] [PubMed] [Google Scholar]