ORIGINAL INVESTIGATION. Microbiologic Determinants of Exacerbation in Chronic Obstructive Pulmonary Disease

Transcription

1 ORIGINAL INVESTIGATION Microbiologic Determinants of Exacerbation in Chronic Obstructive Pulmonary Disease Antoni Rosell, MD; Eduard Monsó, MD; Néstor Soler, MD; Ferràn Torres, MD; Joaquim Angrill, MD; Gerdt Riise, MD; Rafael Zalacaín, MD; Josep Morera, MD; Antoni Torres, MD Background: The culture of bronchial secretions from the lower airway has been reported to be positive for potentially pathogenic microorganisms (PPMs) in patients with stable chronic obstructive pulmonary disease (COPD), but the determinants and effects of this bacterial load in the airway are not established. Methods: To determine the bronchial microbial pattern in COPD and its relationship with exacerbation, we pooled analysis of crude data from studies that used protected specimen brush sampling, with age, sex, smoking, lung function, and microbiologic features of the lower airway as independent variables and exacerbation as the outcome, using logistic regression modeling. Results: Of 337 study participants, 70 were healthy, 181 had stable COPD, and 86 had exacerbated COPD. Differences in the microbial characteristics in the participating laboratories were not statistically significant. A cutoff point of 10 2 colony-forming units (CFU) per milliliter or greater for the identification of abnormal positive culture results for PPMs was defined using the 95th percentile in the pooled analysis of healthy individuals. Bronchial colonization of 10 2 CFU/mL or greater by PPMs was found in 53 patients with stable COPD (29%) and in 46 patients with exacerbated COPD (54%) (P.001, 2 test), with a predominance of Haemophilus influenzae and Pseudomonas aeruginosa. Higher microbial loads were associated with exacerbation and showed a statistically significant dose-response relationship after adjustment for covariates (odds ratio, 3.62; 95% confidence interval, ), but P aeruginosa persisted as a statistically significant risk factor after adjustment for microbial load (odds ratio, 11.12; 95% confidence interval, ). Conclusions: One quarter of the patients with COPD are colonized by PPMs during their stable periods. Exacerbation is associated with the overgrowth of PPMs and with the appearance of P aeruginosa in the lower airway, which is associated with exacerbation symptoms independent of load. Arch Intern Med. 2005;165: Author Affiliations: Servei de Pneumologia, Hospital Germans Trias i Pujol, Badalona, Spain (Drs Rosell, Monsó, and Morera); Institut Clínic de Pneumologia i Cirurgia Toràcica, Hospital Clínic i Provincial, Barcelona, Spain (Drs Soler, Angrill, and A. Torres); Laboratori de Bioestadística i Epidemiologia, Hospital de Sant Pau, Barcelona (Dr F. Torres); Department of Pulmonary Medicine, Sahlqrenska University Hospital, Göteborg, Sweden (Dr Riise); and Servicio de Neumología, Hospital de Cruces, Cruces-Barakaldo, Vizcaya, Spain (Dr Zalacaín). Financial Disclosure: None. THE CULTURE OF BRONCHIAL secretions from the lower airway has been reported to be positive for potentially pathogenic microorganisms (PPMs) in one quarter of the patients with stable chronic obstructive pulmonary disease (COPD) using quantitative cultures of bronchial secretions obtained by the protected specimen brush (PSB) method, 1-3 a highly reliable technique for obtaining uncontaminated specimens during bronchoscopy. 4,5 However, the determinants and effects of this bacterial load in the airways of patients with COPD are not well established, especially in terms of the appearance of exacerbations. 6-8 During the past decade, several studies have used the PSB technique to determine the lower airway microbial load in COPD, focusing on populations with different characteristics and severities. We report a pooled analysis of original data from clinical studies published between January 1, 1993, and December 31, 2002, that performed quantitative cultures of bronchial secretions obtained with a PSB in nonreversible COPD to characterize the microbial pattern of the lower respiratory tract in this obstructive lung disease and to determine the microbiologic characteristics associated with exacerbation. METHODS STUDY POPULATION We identified studies of cultured PSB samples from healthy individuals with normal lung function and from patients with COPD by searching MEDLINE for articles published between January 1, 1993, and December 31, We included in the present analysis all stud- 891

2 ies that used the sampling method described by Wimberley et al 4,5 provided that the authors had reported spirometric values that expressed their results quantitatively, had specified previous treatments, and had included patients not treated with antibiotics during the month before PSB sampling. Six crosssectional studies 2,3,9-12 from 4 laboratories fulfilled these criteria, and their authors agreed to the use of their crude data for the present analysis. Patients with pneumonia, reversibility test results greater than 10%, cystic fibrosis, bronchiectasis identified in the chest radiograph, age younger than 18 years, neoplasia, long-term corticosteroid therapy, or immunosuppression for any other cause where excluded. Patients with COPD who reported increased dyspnea, sputum production, or purulence when the microbiologic sample was obtained were considered to be exacerbated. 13 MICROBIOLOGIC SAMPLING In all studies included in this pooled analysis, microbiologic sampling was performed according to the method described by Wimberley et al. 4,5 In brief, a PSB catheter (Mill-Rose Laboratories, Mentor, Ohio) was inserted through a flexible bronchoscope and advanced to a segmental/subsegmental bronchus. After sampling, the brush was cut and placed in 1 ml of a saline solution. The obtained fluid was serially diluted and cultivated in selective media for microorganisms, which were identified according to standard methods. 14 Only cultures that grew PPMs such as Haemophilus influenzae, Streptoccocus pneumoniae, Moraxella catarrhalis, Pseudomonas aeruginosa, enterobacteria, and Staphylococcus aureus were considered positive for the purposes of the analysis, according to previously defined criteria. 1 Results of quantitative cultures were expressed as colony-forming units (CFU) per milliliter, and they were considered to have a high microbial load when there were 10 4 CFU per milliliter or more. 2 In patients with polymicrobial cultures, the load of the microorganism that attained the highest load was recorded for the sample. LUNG FUNCTION In all the studies that contributed pooled data, forced spirometry was performed using a dry spirometer and according to standard techniques 15 during a stable period of 2 months or longer after any exacerbation. Results were compared with ageand height-adjusted reference values 16 and are given as percentages. For the present study, COPD was defined and categorized according to the Global Initiative for Chronic Obstructive Lung Disease guidelines. 17 STATISTICAL ANALYSIS A statistical software program (SPSS 10.0; SPSS Inc, Chicago, Ill) was used for statistical analysis. Values are expressed as means with SDs or as absolute and relative frequencies. Exacerbation was the main outcome variable, and age, sex, current smoking ( 1 cigarettes per day), lung function (forced expiratory volume in 1 second as a percentage of predicted [FEV 1 %]), and microbiologic features of bronchial secretions were the independent variables. First, the 95th percentile of the microbial load in bronchial secretions was determined from the pooled sample of healthy individuals with normal lung function and was considered the cutoff value for the upper limit of normal. Second, to identify statistically significant interlaboratory differences in the measurement of the bacterial load, clinical characteristics and microbiologic results for the patients studied in the participating laboratories were compared. Logistic regression models to predict the presence of positive cultures were created for the microbiologic results that showed differences in the univariate comparisons, with inclusion of the participating laboratories as dummy variables and age, sex, current smoking, and FEV 1 % as clinical covariates. To determine the effect of lower airway microbiologic characteristics on exacerbation, univariate and multivariate analyses of the pooled data from patients with COPD, stable and exacerbated, were performed using logistic regression modeling, with age, sex, smoking, and lung function as independent variables to define a clinical model with the variables that show a statistically significant association with the outcome. A second model was created with bacterial load (sterile/low: 10 1 CFU/mL, medium: CFU/mL, and high: 10 4 CFU/ ml) as an independent variable, using sterile/low load as the reference, with adjustment for the covariates in the clinical model, to determine the microbial load cutoff value above which there was a statistically significant association with exacerbation and to assess dose-response relationships. Finally, the associations between the different recovered microorganisms and the outcome were assessed, with adjustment for the covariates that showed statistically significant associations with exacerbation in the previous models, to determine the effect of the different recovered PPMs on exacerbation, independent of the clinical characteristics and the microbial load. Multivariate models were adjusted for all covariates that showed an association with the outcome variable (P<.25) in the univariate model, and the most parsimonious model that still explained the data was accepted as final. 18 Results were considered statistically significant at P.05 (2-sided). RESULTS STUDY POPULATION A total of 337 individuals fulfilled the inclusion criteria (70 healthy individuals with normal lung function, 181 patients with stable COPD, and 86 patients with exacerbated COPD). Healthy individuals with normal lung function, patients with stable COPD, and 36 patients with exacerbated COPD (42%) were sampled as outpatients. Fifty patients with exacerbated COPD (58%) were inpatients at the time of microbiologic sampling, and in all these cases samples were obtained within 24 hours of hospital admission and any inpatient treatment. In patients who were not intubated, bronchial secretions were recovered after aerosolization of lidocaine hydrochloride or tetracaine hydrochloride; in patients with exacerbated COPD who required mechanical ventilation (n=45), samples were obtained by inserting the bronchoscope through the orotracheal tube within 24 hours of intubation. Antibiotic drugs had not been used by outpatients before sampling and had not been administered to inpatients before hospital admission. The baseline characteristics of the patients are given in Table 1. MICROBIOLOGIC RESULTS The pooled sample of 70 healthy participants with normal lung function consisted of middle-aged individuals (Table 1). Age, sex, and smoking differences were found between laboratories, but the interlaboratory differences in microbial load were not statistically significant, allowing for performance of the pooled analysis. The cul- 892

3 Table 1. Sociodemographic Characteristics and Lung Function Laboratory Characteristic 1 2 3or4* Overall P Value Healthy Individuals Reference No. 2, , 3, 10, 12 NA Patients, No NA Age, mean (SD), y 59.7 (9.4) 50.2 (10.2) 61.4 (8.2) 55.9 (10.7).001 Sex (male), No. (%) 18 (100) 11 (34) 20 (100) 49 (70).001 Current smoking, No. (%) 13 (72) 19 (59) 2 (10) 34 (49).001 FEV 1 %, mean (SD) 91.1 (20.6) 95.3 (9.6) 88.4 (8.2) 92.3 (13.2).17 PPM culture positive, No. (%) 10 2 CFU/mL 1 (6) 2 (6) 0 3 (4) CFU/mL 0 1 (3) 0 1 (1).55 Patients With Stable COPD Reference No. 2, , 9 2, 3, 9, 10, 12 NA Patients, No NA Age, mean (SD), y 63.5 (9.5) 57.8 (8.6) 66.4 (7.2) 64.5 (8.6).001 Sex (male), No. (%) 60 (97) 12 (67) 101 (100) 173 (96).001 Current smoking, No. (%) 34 (55) 18 (100) 44 (44) 96 (53).001 FEV 1 %, mean (SD) 58.6 (21.6) 63.4 (8.4) 53.3 (13.5) 56.1 (16.7).02 PPM culture positive, No. (%) 10 2 CFU/mL 14 (23) 2 (11) 37 (37) 53 (29) CFU/mL 5 (8) 1 (6) 8 (8) 14 (8).93 Patients With Exacerbated COPD Reference No , 9, 11 NA Patients, No NA Age, mean (SD), y 63.4 (9.1) 67.7 (5.3) 68.8 (7.6) 66.8 (8.2).02 Sex (male), No. (%) 27 (93) 12 (100) 45 (100) 84 (98).13 Current smoking, No. (%) 7 (24) 5 (42) 19 (42) 31 (36).26 FEV 1 %, mean (SD) 43.8 (14.1) 34.2 (7.7) 32.7 (13) 36.6 (13.8).002 PPM culture positive, No. (%) 10 2 CFU/mL 17 (59) 7 (58) 22 (49) 46 (54) CFU/mL 8 (28) 3 (25) 8 (18) 19 (22).59 Abbreviations: CFU, colony-forming units; COPD, chronic obstructive pulmonary disease; FEV 1 %, forced expiratory volume in 1 second as a percentage of predicted; NA, not applicable; PPM, potentially pathogenic microorganism. *Laboratory 3 for healthy individuals and patients with stable COPD and laboratory 4 for patients with exacerbated COPD. 2 Test or 1-way analysis of variance. ture of lower airway secretions in healthy participants was sterile for PPMs in 66 cases (94%), and 1 case each had PPM loads of 10 1,10 2,10 3, and 10 4 CFU/mL. Accordingly, a cutoff value of 10 2 CFU/mL or greater was defined to identify an abnormally high PPM-positive culture, considering the 95th percentile of microbial load in healthy individuals with normal lung function to be the upper limit of normal. In the assessment of the interlaboratory variability of the clinical and microbiologic characteristics, statistically significant differences were found between most clinical variables, but only minimal differences were found in the prevalence of positive results at a load of 10 2 CFU/mL or greater, the difference being statistically significant only in patients with stable COPD (P=.03; 2 test) (Table 1). However, when a logistic regression model was created with positive cultures of 10 2 CFU/mL or greater as the outcome and laboratories as independent variables, with adjustment for the clinical covariates, the association of laboratory with positive cultures at 10 2 CFU/mL or greater was not statistically significant, and only FEV 1 % was significantly associated with the finding of a positive culture at 10 2 CFU/mL or greater (odds ratio [OR], 0.97; 95% confidence interval [CI], ) (Table 2). Accordingly, it was considered that the interlaboratory differences in positive cultures at 10 2 CFU/mL or greater in patients with stable COPD were attributable to differences in lung function, and we proceeded to a pooled analysis of microbiologic data from patients with COPD. Bronchial colonization of 10 2 CFU/mL or greater by PPMs was observed in more than one quarter of the patients with stable COPD, with high microbial loads found in slightly less than 10% of the cases, proportions that were well above those found in healthy individuals (P.001; 2 test) (Table 3). This high prevalence of bronchial colonization in stable COPD was mainly attributable to H influenzae and Streptoccoccus pneumoniae (P=.008 and P=.004, respectively; 2 test), microorganisms that reached prevalences well above 5% (Table 3). Bronchial colonization was found in none of the 10 patients with mild disease, 26 (24%) of 109 patients categorized as having moderate disease, 24 (45%) of 53 with severe disease, and 3 (33%) of 9 with very severe disease. Statistically significant differences were found between these groups (P=.006; 2 test), and a linear trend 893

4 Table 2. Risk Factors for Colonization by Potentially Pathogenic Microorganisms of 10 2 CFU/mL or Greater in 181 Patients With Stable COPD Clinical Variable Noncolonized (n = 128) Colonized (n = 53) Crude OR (95% CI) P Value Adjusted OR (95% CI)* P Value Age, mean (SD), y 63.8 (8.4) 66.3 (8.9) 1.04 ( ).08 NA NA Sex (male), No. (%) 122 (95) 51 (96) 1.25 ( ).79 NA NA Current smoking, No. (%) 68 (53) 28 (53) 0.99 ( ).97 NA NA FEV 1 %, mean (SD) 58.5 (17.2) 50.3 (13.9) 0.97 ( ) ( ).003 Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; FEV 1 %, forced expiratory volume in 1 second as a percentage of predicted; NA, not applicable; OR, odds ratio. *Multivariate analysis adjusted for laboratory and all covariates that showed an association with the dependent variable (P.25) according to the Hosmer-Lemeshow procedure (see the Statistical Analysis subsection for details). Table 3. Potentially Pathogenic Microorganisms Variable Healthy Individuals (n = 70) Patients With Stable COPD (n = 181) P Value* Patients With Exacerbated COPD (n = 86) P Value PPM culture 10 2 CFU/mL, No. (%) 3 (4) 53 (29) (54).001 PPM load, No. (%) 10 2 CFU/mL 1 (1) 5 (3) 4 (5) 10 3 CFU/mL 1 (1) 34 (19) (28) CFU/mL 1 (1) 14 (8) 18 (21) Microorganisms, No. (%) Haemophilus influenzae 3 (4) 31 (17) (30).01 Streptococcus pneumoniae 0 16 (9) (7).60 Moraxella catarrhalis 0 7 (4).10 6 (7).21 Enterobacteria 0 5 (3).19 6 (7).10 Pseudomonas aeruginosa 0 2 (1).99 8 (9).002 Staphylococcus aureus 0 2 (1) Polymicrobial culture 0 9 (6).05 6 (7).57 Abbreviations: CFU, colony-forming units; COPD, chronic obstructive pulmonary disease; PPM, potentially pathogenic microorganism. *Comparison of healthy individuals and patients with stable COPD; 2 test or Fisher exact test as required. Comparison of patients with stable vs exacerbated COPD; 2 test or Fisher exact test as required. confirmed an increase in prevalence associated with impaired lung function (P=.003; 2 test). The prevalence of positive cultures of 10 2 CFU/mL or greater in patients with exacerbated COPD was greater than 50%, and their cultures reached high loads in more than 20% of the cases, showing clear differences with the microbiologic results found in patients with stable COPD (P.001; 2 test) (Table 3). Haemophilus influenzae and P aeruginosa were the microorganisms with the highest prevalences in patients with exacerbated COPD (30% and 9%, respectively), and significant differences were found for the prevalences of these PPMs between patients with stable and exacerbated COPD (P=.01 and P=.002, respectively) (Table 3). EFFECT OF MICROBIAL LOAD AND TYPE ON EXACERBATION SYMPTOMS Table 4 summarizes the results of the analysis that considered exacerbation as the outcome variable and included all patients with COPD (stable and exacerbated). In the univariate clinical models with age, sex, smoking, and FEV 1 % as independent variables, only FEV 1 % was significantly associated with outcome in the multivariate model (OR, 0.92; 95% CI, ). The following models, therefore, were adjusted for this clinical variable. For assessment of the association between the results of bronchial secretion microbiologic features and exacerbation, a univariate model was first created, with microbial load as the independent variable, using the patients with sterile/low loads as the reference, and this model showed statistically significant associations between medium (OR, 2.21; 95% CI, ) and high (OR, 4.34; 95% CI, ) microbial loads and exacerbation, with a dose-response relationship. After adjustment for FEV 1 %, the association was statistically significant only for the positive cultures with a high PPM load (OR, 3.62; 95% CI, ), however, suggesting a major role for PPM overgrowth in the appearance of COPD exacerbation. Univariate models with the different recovered PPMs as independent variables were also created, showing that positive cultures of 10 2 CFU/mL or greater for H influenzae (OR, 2.10; 95% CI, ) and P aeruginosa (OR, 9.18; 95% CI, ) were significantly associated with exacerbation. After adjustment for FEV 1 % and microbial load, however, only the recovery of P aeruginosa from bronchial secretions emerged as significantly associated with exacerbation (OR, 894

5 Table 4. Microbiologic Risk Factors for Exacerbation in 267 Patients With COPD Variable Stable COPD (n = 181) Exacerbated COPD (n = 86) Crude OR (95% CI) P Value Adjusted OR (95%CI) P Value Clinical variables Age, mean (SD), y 64.5 (8.6) 66.8 (8.2) 1.03 ( ).04 NA NA Sex (male), No. (%) 173 (96) 84 (98) 1.94 ( ).41 NA NA Current smoking, No. (%) 96 (53) 31 (36) 0.50 ( ).01 NA NA FEV 1 %, mean (SD) 56.1 (16.7) 36.6 (13.8) 0.92 ( ) ( ).001 Microbial load, No. (%)* Sterile/low 128 (71) 40 (47) NA NA NA NA Medium 39 (22) 27 (31) 2.21 ( ) ( ).22 High 14 (8) 19 (22) 4.34 ( ) ( ).005 PPM, No. (%) Haemophilus influenzae 31 (17) 26 (30) 2.10 ( ) ( ).79 Streptococcus pneumoniae 16 (9) 6 (7) 0.77 ( ).61 NA NA Moraxella catarrhalis 7 (4) 6 (7) 1.86 ( ).28 NA NA Enterobacteria 5 (3) 8 (9) 2.64 ( ) ( ).73 Pseudomonas aeruginosa 2 (1) 8 (9) 9.18 ( ) ( ).04 Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; FEV 1 %, forced expiratory volume in 1 second as a percentage of predicted; NA, not applicable; OR, odds ratio; PPM, potentially pathogenic microorganism. *Adjusted for the statistically significant variables in the adjusted clinical model. Sterile/low: 10 1 colony-forming units (CFU) per milliliter; medium: CFU/mL; and high: 10 4 CFU/mL. Sterile/low was used as the reference. Patients with polymicrobial cultures were excluded. Adjusted for the statistically significant variables in the adjusted clinical and microbial load models. Staphylococcus aureus (n 10) was not included in the analysis ; 95% CI, ), suggesting that the effect of H influenzae on exacerbation depends on the attainment of a high load, but P aeruginosa may determine the appearance of exacerbation independent of load. COMMENT In this analysis of crude data from studies that used the PSB technique to examine the microbiologic characteristics of lower airway secretions in healthy individuals with normal lung function and in patients with COPD, we found that the interlaboratory variability in the prevalence of colonization by PPMs was not statistically significant, allowing us to perform a pooled analysis with a cutoff value of 10 2 CFU/mL or greater to identify an abnormal positive culture result. More than a quarter of the patients with stable COPD had a lower airway colonized by a PPM with a load of 10 2 CFU/mL or greater, and the prevalence of colonization in patients with more severe COPD was higher, but a high microbial load ( 10 4 CFU/mL) was reached in less than one tenth of the cases. The prevalence of positive cultures of 10 2 CFU/mL or greater was much higher in patients with exacerbated COPD, almost a quarter of whom had a high PPM load, and H influenzae and P aeruginosa were the PPMs more frequently recovered. Multivariate analyses confirmed that in addition to lung function impairment, a high PPM load in lower airway secretions was a major determinant of exacerbation in COPD, but the recovery of P aeruginosa was associated with exacerbation independent of the attained load. The present study included only microbiologic data obtained using the PSB, a technique that has proved to provide highly accurate microbiologic diagnoses 5,19-21 and a quantitative repeatability close to 75% in patients with stable COPD when a CFU/mL or greater threshold has been used. 12,22 The analysis of healthy individuals with normal lung function allowed us to define a threshold of 10 2 CFU/mL or greater for identification of an abnormal recovery of PPMs from the lower airway. This cutoff point has been previously suggested for the identification of patients with colonized COPD, 1 and our results confirm that microbial loads that reach this value must be considered abnormal. In this study, PPM colonization of 10 2 CFU/mL or greater was common in patients with stable COPD ( 29%), but it reached a high PPM load in only a few cases (8%). Bronchial colonization was not found in patients with mild disease, but one quarter of the patients with COPD with moderate disease were colonized, and almost half of those with severe disease had positive cultures, with statistically significant differences between these types of patients. In addition, multivariate analysis demonstrated that the high prevalence of bronchial colonization in patients with stable COPD was attributable to H influenzae and Streptococcus pneumoniae, microorganisms that reach prevalences close to or greater than 10%, much higher than the prevalences found in healthy individuals. Our analysis found that bronchial secretion PPM recovery was 10 2 CFU/mL or greater in more than half of the patients with exacerbated COPD, a prevalence that is well above that of positive cultures in patients with stable COPD. Exacerbation was associated with the recovery of a high PPM load in nearly a quarter of the patients, with H influenzae and P aeruginosa being the main microorganisms cultured during the exacerbation periods. This microbial profile is also found in stable bronchiectasis, confirming the microbiologic similarities between exacerbated COPD and that disease. 23 In the analysis performed to determine the risk factors for exacerbation in COPD, we see that in addition to lung function impair- 895

6 ment, the recovery of PPMs from bronchial secretions was associated with exacerbation, and this association showed a dose-response relationship that was statistically significant after adjustment for covariates for the cultures with a high PPM load. This observation is consistent with studies that suggest that the role of lower respiratory tract PPMs in the pathogenesis of COPD exacerbations may be partly mediated by the overgrowth of persistent 24,25 or newly colonizing 26 microorganisms. Accordingly, the association between the recovery of H influenzae and exacerbation in the univariate analysis when the different recovered PPMs were assessed as agents related to exacerbation was not statistically significant when adjusted for lung function impairment and microbial load. However, the relationship found between the recovery of P aeruginosa and exacerbation in the univariate analysis continued to be statistically significant after adjustment for the covariates, an observation that suggests that the growth of P aeruginosa in the lower airway of patients with COPD is associated with exacerbation independent of load, in agreement with previous findings, 26,27 pointing to specific characteristics of P aeruginosa infection in COPD. Microbial colonization in stable COPD seems to be mainly a marker of disease severity because local host defenses keep PPM growth below high loads most of the time. When defense mechanisms are overcome and high PPM loads are attained or P aeruginosa colonizes the bronchial tree, an inflammatory response may develop and manifest as an exacerbation. Half of the patients with exacerbated COPD included in the present analysis had exacerbations severe enough to require mechanical ventilation. Most of these patients had severe COPD, which was related to the recovery of P aeruginosa, as we found and as has been previously reported. 27,31-34 However, more than 40% of the patients with exacerbated COPD included in our study were sampled as outpatients, guaranteeing that enough patients with exacerbated COPD with less severe disease were included to minimize bias from inclusion of advanced COPD cases in the pooled analysis. In the subsample of patients with severe COPD who required mechanical ventilation, the performance of PSB sampling on the first day after hospital admission minimized the effect of the treatments prescribed at admission on the microbiologic results, 35 and we have not pooled data from patients who had undergone long-term treatment with oral corticosteroids. An effect of inhaled corticosteroids on bronchial secretion cultures cannot be ruled out from the obtained data, however, because detailed information about this treatment was not available from the studies included in our analysis. Similarly, we have not looked at other mechanisms that may have an effect on the appearance of exacerbations besides lung function and microbial load and type, such as the presence of intracellular microorganisms in the bronchial mucosa, which have recently shown a relationship with the appearance of acute symptoms in patients with COPD, 36,37 or abnormalities in the specific immunity against the recovered microorganisms, which also seem to increase the risk of recurrent bronchial infections. 38,39 In summary, this pooled analysis of data from studies that used the PSB technique for microbiologic assessment of the lower airway in COPD demonstrated that a PPM load of 10 2 CFU/mL or greater must be considered abnormal and allowed us to estimate that a quarter of the patients with stable COPD are colonized by PPMs, with colonization increasing in prevalence as lung function becomes more impaired. In addition, we confirmed that COPD exacerbations are associated with the recovery of PPMs from the lower airway and depend on the attainment of a high PPM load for most microorganisms, but P aeruginosa is associated with the appearance of exacerbations independent of load. Accepted for Publication: October 19, Correspondence: Eduard Monsó, MD, Servei de Pneumologia, Hospital Germans Trias i Pujol, Carretera de Canyet s/n, Badalona, Catalonia, Spain Funding/Support: This study was funded by Fondo de Investigaciones Sanitarias, Madrid, Spain (Red Respira ISCiii-RTIC-03/11 and Red Gira ISCiii-03/063). Acknowledgment: We thank Adoración Ruiz, RN, for her technical help and M. E. Kerans, PhD, for her revision of the use of language in the manuscript. REFERENCES 1. Cabello H, Torres A, Celis R, et al. Bacterial colonization of distal airways in healthy subjects and chronic lung disease: a bronchoscopic study. Eur Respir J. 1997; 10: Monsó E, Ruiz J, Rosell A, et al. Bacterial infection in chronic obstructive pulmonary disease: a study of stable and exacerbated outpatients using the protected specimen brush. Am J Respir Crit Care Med. 1995;152: Zalacaín R, Sobradillo V, Amilibia J, et al. Predisposing factors to bacterial colonization in chronic obstructive pulmonary disease. Eur Respir J. 1999;13: Wimberley N, Faling LJ, Bartlett JG. A fiberoptic bronchoscopy technique to obtain uncontaminated lower airway secretions for bacterial culture. Am Rev Respir Dis. 1979;119: Wimberley NW, Bass JB, Boyd BW, Kirkpatrick MB, Serio RA, Pollock HM. Use of a bronchoscopic protected brush for the diagnosis of pulmonary infections. Chest. 1982;81: Sethi S. Infectious etiology of acute exacerbations of chronic bronchitis. Chest. 2000;117:380S-385S. 7. Murphy TF, Sethi S. Bacterial infection in chronic obstructive pulmonary disease. Am Rev Respir Dis. 1992;146: Irwin RS, Erickson AD, Pratter MR, et al. Prediction of tracheobronchial colonization in current cigarette smokers with chronic bronchitis. J Infect Dis. 1982; 145: Zalacaín R, Achótegui V, Pascal I, Camino J, Barrón J, Sobradillo V. The protected specimen brush technique in patients with severe chronic obstructive pulmonary disease. Arch Bronconeumol. 1997;33: Riise GC, Larsson P, Jeansson S, Andersson BA. The intrabronchial microbial flora in chronic bronchitis patients: a target for N-acetylcysteine therapy? Eur Respir J. 1994;7: Soler N, Torres A, Ewig S, et al. Bronchial microbial patterns in severe exacerbations of chronic pulmonary disease (COPD) requiring mechanical ventilation. Am J Respir Crit Care Med. 1998;157: Monsó E, Rosell A, Bonet G, et al. Risk factors for lower airway bacterial colonization in chronic bronchitis. Eur Respir J. 1999;13: Anthonisen NR, Manfreda J, Warren CP, Hershfield ES, Harding GKM, Nelson NA. Antibiotic therapy in exacerbations of chronic obstructive pulmonary disease. Ann Intern Med. 1987;106: Balows A, Hausler WJ, Herrmann KL, Isenberg HD, Shadomy HJ. Manual of Clinical Microbiology. 5th ed. Washington, DC: American Society of Microbiology; American Thoracic Society. Standardization of spirometry: 1987 update. Am Rev Respir Dis. 1987;136: Quanjer PH, Tammeling GJ, Cotes JE, Pedersen OF, Peslin R, Yernault JC. Lung volumes and forced ventilatory flows: Report Working Party Standardization of 896

7 Lung Function Tests, European Community for Steel and Coal: official statement of the European Respiratory Society. Eur Respir J Suppl. 1993;16: GOLD Scientific Committee. NHLBI/WHO Global Initiative for Chronic Obstructive Lung Disease workshop report. Available at: Accessed December 15, Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons; Chastre J, Fagon JY, Bornet-Lecso M. Evaluation of bronchoscopic techniques for the diagnosis of nosocomial pneumonia. Am J Respir Crit Care Med. 1995; 152: Marquette CH, Herengt F, Saulnier F, et al. Protected specimen brush in the assessment of ventilator-associated pneumonia. Chest. 1993;103: Timsit JF, Misset B, Francoual S, Goldstein FW, Vaury P, Carlet J. Is protected specimen brush a reproducible method to diagnose ICU-acquired pneumonia? Chest. 1993;104: Ioanas M, Angrill J, Baldo X, et al. Bronchial bacterial colonization in patients with resectable lung carcinoma. Eur Respir J. 2002;19: Angrill J, Agustí C, De Celis R, et al. Bacterial colonisation in patients with bronchiectasis: microbiological pattern and risk factors. Thorax. 2002;57: Wilkinson TM, Patel IS, Wilks M, Donaldson GC, Wedzicha JA. Airway bacterial load and FEV 1 decline in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2003;167: Murphy TF, Brauer AL, Schiffmacher AT, Sethi S. Persistent colonization by Haemophilus influenzae in chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2004;170: Sethi S, Evans N, Grant BJB, Murphy TF. New strains of bacteria and exacerbations of chronic obstructive pulmonary disease. N Engl J Med. 2002;347: Monsó E, Garcia-Aymerich J, Soler N, et al. Bacterial infection in exacerbated COPD with changes in sputum characteristics. Epidemiol Infect. 2003;131: Hill AT, Campbell EJ, Hill SL, Bayley DL, Stockley RA. Association between airway bacterial load and markers of airway inflammation in patients with stable chronic bronchitis. Am J Med. 2000;109: Soler N, Ewig S, Torres A, Filella X, Gonzalez J, Xaubet A. Airway inflammation and bronchial microbial patterns in patients with stable chronic obstructive pulmonary disease. Eur Respir J. 1999;14: Gompertz S, O Brien C, Bayley DL, Hill SL, Stockley RA. Changes in bronchial inflammation during acute exacerbations of chronic bronchitis. Eur Respir J. 2001;17: Groenewegen KH, Wouters EF. Bacterial infections in patients requiring admission for an acute exacerbation of COPD: a 1-year prospective study. Respir Med. 2003;97: Patel IS, Seemugal TA, Wilks M, Lloyd-Owen SJ, Donaldson GC, Wedzicha JA. Relationship between bacterial colonization and the frequency, character, and severity of COPD exacerbations. Thorax. 2002;57: Eller J, Ade A, Schaberg T, Niederman MS, Mauch H, Lode H. Infective exacerbations of chronic bronchitis: relation between bacteriologic etiology and lung function. Chest. 1998;113: Miravitlles M, Espinosa C, Fernandez-Lao E, et al. Relationship between bacterial flora in sputum and functional impairment in patients with acute exacerbations of COPD. Chest. 1999;116: Prats E, Dorca J, Pujol M, et al. Effects of antibiotics on protected specimen brush sampling in ventilator-associated pneumonia. Eur Respir J. 2002;19: Bandi V, Apicella MA, Mason E, et al. Nontypeable Haemophilus influenzae in the lower respiratory tract of patients with chronic bronchitis. Am J Respir Crit Care Med. 2001;164: Foxwell AR, Kyd JM, Cripps AW. Nontypeable Haemophilus influenzae: pathogenesis and prevention. Microbiol Mol Biol Rev. 1998;62: King PT, Hutchinson PE, Johnson PD, Holmes PW, Freezer NJ, Holdsworth SR. Adaptive immunnity to nontypeable Haemophilus influenzae. Am J Respir Crit Care Med. 2003;167: Vitovski S, Dunkin KT, Howard AJ, Sayers JR. Nontypeable Haemophilus influenzae in carriage and disease: a difference in IgA1 protease activity levels. JAMA. 2002;287: