Flow cytometric characterization of cell populations in bronchoalveolar lavage and bronchial brushings from patients with chronic obstructive pulmonary disease

Authors

  • Sandra J. Hodge,

    Corresponding author
    1. Department of Thoracic Medicine, Royal Adelaide Hospital, and Lung Research Laboratory, Hanson Institute, Adelaide, Australia
    • Lung Research, Hanson Institute, Frome Road, Adelaide, South Australia 5001, Australia
    Search for more papers by this author
  • Greg L. Hodge,

    1. Department of Thoracic Medicine, Royal Adelaide Hospital, and Lung Research Laboratory, Hanson Institute, Adelaide, Australia
    2. Haematology Department, Women's and Children's Hospital, North Adelaide, Australia
    Search for more papers by this author
  • Mark Holmes,

    1. Department of Thoracic Medicine, Royal Adelaide Hospital, and Lung Research Laboratory, Hanson Institute, Adelaide, Australia
    Search for more papers by this author
  • Paul N. Reynolds

    1. Department of Thoracic Medicine, Royal Adelaide Hospital, and Lung Research Laboratory, Hanson Institute, Adelaide, Australia
    Search for more papers by this author

Abstract

Background

Chronic obstructive pulmonary disease (COPD) is a serious, chronic inflammatory disease of the airway associated with cigarette smoking. Leucocytes are involved in the inflammatory process in the airways in COPD. There is a need for accurate characterization of cellular populations in bronchoalveolar lavage (BAL) due to variation in the predominant cell types reported, which were investigated mostly with manual counting techniques.

Methods

Bronchial brushings and BAL were obtained from human subjects undergoing fiber optic bronchoscopy. Flow cytometry was applied to identify various cell types. Quenching of autofluorescence of BAL-derived alveolar macrophages was achieved with η-octyl β-D-galactopyranoside and crystal violet. Comparisons of cell counts obtained with flow cytometric and manual counting methods were performed.

Results

Correlation analysis showed that manual cell counting methods overestimated the percentage of macrophages when compared with flow cytometric methods (R2 = 0.54). There was also a small tendency by manual counting to underestimate the percentage of lymphocytes and neutrophils. Using flow cytometry, the percentage and absolute numbers of alveolar macrophages and lymphocytes in BAL were not significantly different between patients with COPD and control subjects. The percentage and absolute numbers of neutrophils were higher in BAL from patients with moderate to severe COPD.

Conclusions

This novel flow cytometric assay for identification of various cell types from heterogenous samples of BAL and bronchial brushing will allow further investigation of cell characteristics, such as cytokine production and receptor expression, and an accurate evaluation of apoptosis for different cell types and provide a rationale for urgently required effective treatments for COPD. © 2004 Wiley-Liss, Inc.

Chronic obstructive pulmonary disease (COPD) arises as a result of noxious injury to the lungs, most commonly due to cigarette smoking. Despite its high prevalence and the financial burden to the international community, the pathogenesis of COPD and therefore the rationale for therapy are poorly understood, and currently there are no satisfactory treatments for this important disease. New insights into the basis of COPD, which should lead to more effective treatments, are thus urgently required. Lung damage in COPD involves proximal and distal airways and alveoli. Several studies have shown that neutrophils, lymphocytes, and alveolar macrophages are involved in the inflammatory process in the airways in COPD (1–7). There is a correlation between the numbers of these cells in the lungs and the severity of COPD (2). However, there is variation in the predominant cell types reported, possibly resulting from a difference in inflammatory cell type numbers in various parts of the lung (8, 9). Bronchoalveolar lavage (BAL) is a useful, relatively noninvasive technique that samples the inflammatory milieu of the most distal airways and alveoli. Airway brushings can more conveniently sample medium-size airways and yields more than 90% of airway epithelial cells. Because many studies in COPD have been performed on bronchial biopsies of more proximal airways, there is a need for accurate characterization and clarification of cellular populations in BAL from the more distal airway lumina and alveoli in COPD. Current techniques to enumerate cell types in BAL rely on manual differential counting methods that do not allow for analysis of additional characteristics of infiltrating cells, such as cytokine production and receptor expression, and an accurate evaluation of apoptosis for different cell types.

Previous studies have applied flow cytometry to distinguish lymphocytes and their subsets (10, 11). However, these studies did not investigate the clinically important alveolar macrophage and neutrophil components. One study applied flow cytometry to successfully distinguish granulocytes from macrophages in BAL by virtue of CD15 expression (12). However, these studies did not address the problem of the greatly increased autofluorescence properties of BAL-derived alveolar macrophages, especially those obtained from smokers. These properties hinder analysis by flow cytometry due to spectral overlap between alveolar macrophage autofluorescence and fluorochrome emission spectra. Several dyes have been used to quench autofluorescence of BAL-derived alveolar macrophages (13–16). Further, permeabilization of the cell membrane with η-octyl β-D-galactopyranoside has been reported to increase the efficiency of quenching with crystal violet (16). We applied flow cytometric techniques to identify various cell types in BAL. To establish optimal quenching methods, a range of quenching agents was investigated. Comparisons of cell counts obtained with flow cytometric and manual counting methods are also described.

For these studies, samples of bronchial brushings and BAL were obtained directly from human subjects undergoing fiber optic bronchoscopy.

MATERIALS AND METHODS

Subjects

Patients undergoing fiber optic bronchoscopies for diagnostic purposes were invited to participate in the study, and fully informed consent was obtained. Exclusion criteria were based on guidelines of the European Respiratory Society (17). The study protocol was approved by the research ethics committee of the Royal Adelaide Hospital according to the guidelines of the Declaration of Helsinki. Patients underwent pulmonary function tests as part of their routine clinical assessment.

BAL and bronchial brushings were collected from 16 patients with COPD. For 10 nonsmokers undergoing bronchoscopy for clinically indicated reasons and with no history of COPD, asthma, or allergy, specimens were obtained and used as controls (Table 1). The diagnosis of COPD was established with criteria according to the European Respiratory Society (17). Eight patients with COPD were categorized as having mild COPD (forced expiratory volume in 1 s ≥ 70% predicted, with clinical correlation) and eight as having as moderate to severe COPD (forced expiratory volume in 1 s < 70% predicted) There was no exacerbation of disease for 6 weeks before involvement in the study.

Table 1. Demographic Characteristics of the Population Studied*
CharacteristicsControl groupCOPD (total)Mild COPDModerate/severe COPD
  • *

    Results are expressed as mean ± standard deviation. COPD, chronic obstructive pulmonary disease; DLCO, diffusing capacity for carbon monoxide; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity.

Subjects (n)101688
Age (years)63 ± 1766 ± 969 ± 963 ± 8
Smoking (packs/year)067 ± 2656 ± 2083 ± 26
FEV1 (%predicted)94.7 ± 10.468.1 ± 21.182.5 ± 6.754 ± 18.5
FVC (%predicted)97.8 ± 9.782.5 ± 18.894.0 ± 17.076 ± 13.7
FEV1 (%FVC)81.7 ± 15.871 ± 13.774.0 ± 5.064 ± 19.2
DLCO97.6 ± 2.367 ± 22.469.0 ± 38.364 ± 4.3

Bronchoscopic Procedure

Brushings and BAL were obtained from airways on the right side; if there was other pathology in this area, then samples were obtained from the left side. BAL was performed as previously described (18). Particular attention was given to use a minimal amount of lidocaine HCL (Xylocaine; 100 mg) in the airways in view of the adverse effect it has on cell viability (19). Bronchial brushings were obtained by positioning the tip of the bronchoscope in a subsegmental airway and then gently advancing a standard cytology brush (Fuginon Inc., Wayne, NJ, USA) into four to six medium-size airways. Airway epithelial cells (AECs) were obtained with several gentle passages of the brush into each airway to avoid bleeding. Cells were deposited by washing the brush in 10 ml of RPMI 1640 medium (Gibco Berlin, Germany) in 10-ml conical polypropylene tubes (Johns Professional Products, Sydney, Australia) that were kept on ice. Cells were processed within 1 h of collection.

Preparation of Ex Vivo Samples

For each collection from an individual patient, BAL specimens 2 and 3 were pooled. Cells were pelleted by centrifuging at 500 g for 5 min. Supernatant was discarded and cells were resuspended to a cell count of 4 × 105 cells/ml with RPMI 1640 medium.

Brushing-derived AECs were pelleted by centrifuging at 500 g for 5 min, the supernatant was discarded, and cells were resuspended to a cell count of 4 × 105 cells/ml with RPMI 1640 medium.

Identification of Cell Types by Manual Counting

Before processing for flow cytometry, morphology of cells derived from BAL and bronchial brushings was assessed on cytospin preparations, and cells were counted. Manual differential cell counts were carried out by a diagnostic laboratory (Histopathology Department, IMVS, Adelaide, Australia). Total cell counts were done with a hemocytometer (American Optical, New York). Cytospin preparations were made by using 200 μl of sample (82 g for 10 min; Shandon Cytospin III, Shandon, Cheshire, UK) and stained by a Giemsa technique carried out on an automated staining machine (Shandon Veristat Southern Products, Astmore, UK). Differential counts were performed by counting 500 cells on each slide.

Identification of Cell Types by Flow Cytometry

Reagents.

Phycoerythrin (PE)–conjugated monoclonal antibodies to CD45 (BD Biosciences, San Jose, CA, USA), fluorescein isothiocyanate (FITC)–conjugated monoclonal antibodies to CD33 (macrophage/monocyte marker; Pharmingen, San Diego, CA, USA), and epithelial cell antigen (ECA; Dako, Glostrup, Denmark) were used. PE cyanocobalamin (PE-CY5)–conjugated monoclonal antibodies to CD14 and CD3 (Pharmingen) were included.

Cell surface staining with monoclonal antibodies.

For surface staining, 200-μl aliquots of BAL and 100-μl aliquots of bronchial brushings were added to labeled fluorescence-activated cell sorting tubes. To block Fc receptors and decrease nonspecific binding, 20 μl of normal human immunoglobulin (60 g/l; Intragam, Commonwealth Serum Laboratories, Sydney, Australia) was added to each tube for 20 min at room temperature. As previously described (20), after an additional incubation for 20 min in the dark, with directly conjugated monoclonal antibodies to surface markers of interest, cells were washed with 0.5% bovine serum albumin in Isoton II (Coulter Immunotech, Hialeah, FL, USA; hereafter referred to as wash buffer) and centrifuged at 1,500g for 90 s, and the supernatant was discarded. Twenty microliters of wash buffer was added, and events were acquired immediately with a FACScalibur flow cytometer (BD Biosciences) and analyzed with CellQuest software (BD Biosciences). Ten thousand events were collected from bronchial brushings. Because brushing-derived cells comprised more than 98% of AECs, these cells henceforth are referred to as AECs. For BAL, 50,000 events were collected.

Identification of bronchial brushing-derived AECs.

AECs were stained as described above with fluorescent-conjugated monoclonal antibodies to identify AECs (FITC-conjugated ECA) and contaminating leucocytes (PE-conjugated CD45).

A region (R1) was drawn to exclude debris and red blood cells based on known forward (FSC) and side (SSC) scatter characteristics. All subsequent analysis was carried out on cells gated in R1 (Fig. 1a). Leucocyte contamination of bronchial brushing was excluded from another region (R2) based on bright staining with CD45 and negative staining with ECA (Fig. 1b). All subsequent analysis was carried out on cells gated in R1 and R2.

Figure 1.

Identification of brushing-derived AECs by flow cytometry. a: Debris and red blood cells were excluded from R1 based on low FSC and SSC. All subsequent analyses were carried out on cells from R1. b: AECs gated in R2 based on bright staining with ECA and dim staining with CD45 (to exclude ECA-CD45+ leucocytes). All subsequent analyses were carried out on cells from R1 and R2.

Identification and differential counting of BAL-derived cells.

BAL-derived cells were stained as described above by using fluorescent-conjugated monoclonal antibodies to identify the various cell types: tube 1 contained ECA FITC/CD45 PE (to quantify AEC), tube 2 contained CD 33 FITC/CD45 PE (to quantify alveolar macrophages, lymphocytes, and neutrophils), and tube 3 contained CD 33 FITC/CD45 PE/CD14 PE-Cy5 (to verify results).

Details of flow cytometric analysis to identify the various cell types are shown in Figure 2. A quenching strategy to decrease autofluorescence of alveolar macrophages was also used (Figure 3).

Figure 2.

Identification and differential counting of BAL-derived cells using flow cytometry. a: FSC versus SSC characteristics of BAL. b: R1 was drawn to define leucocytes identified based on bright staining with CD45 (debris and red blood cells were excluded). All subsequent analyses were carried out on cells from R1. R2 defines lymphocytes based on low SSC and bright staining with CD45. c: R3 defines alveolar macrophages based on high SSC and dim staining with ECA. R4 defines airway epithelial cells based on bright staining with ECA and low SSC. d–f: Identification of neutrophils and confirmation of macrophage gating. Neutrophils were considered to be any cells not gated in the previous regions. These cells demonstrated dim staining with CD33 and CD14, bright staining with CD45, and lower SSC characteristics than did macrophages. These gating strategies were confirmed by examining stained preparations of sorted cell populations. A small population of cells considered to be blood monocytes were gated in R5 (CD45 bright, medium SSC) and excluded from all subsequent analysis. Arrows indicate a subpopulation of neutrophils considered apoptotic (we previously found that apoptotic neutrophils lose some CD45 expression and therefore stain CD45 dim with low FSC and SSC characteristics) (18).

Figure 3.

Autofluorescence of alveolar macrophages: effect of quenching. Representative dot plots of alveolar macrophages before and after quenching with η-octyl β-D-galactopyranoside and with then 0.2% crystal violet plus phosphate buffered saline (pH 7.4). a: ECA-FITC versus CD45-PE showing marked autofluorescence of macrophages before quenching. b: The same preparation after quenching. Note the well-differentiated positive staining with CD45 and the dim staining with ECA. c: CD14-FITC versus CD45-PE showing marked autofluorescence of macrophages before quenching. d: After quenching, note the well-differentiated positive staining of macrophages with CD45 and CD14.

The gated populations were confirmed by sorting the cell populations from the BAL sample and preparing a stained preparation of these cells on a microscope slide followed by Giemsa staining as described above (data not shown).

Autofluorescence of Alveolar Macrophages: Effect of Quenching

To establish optimal quenching methods for autofluorescence of alveolar macrophages, a range of quenching agents was investigated (Table 2). BAL-derived cells were stained for surface antigens as described above, resuspended in 100 μl of quenching agent for 1 min, washed, and then acquired and analyzed by flow cytometry. For some experiments, permeabilization of the cell membrane was carried out with a commercial reagent, FACSPerm (BD Biosciences). Cells were stained for surface antigens, washed, and then left at room temperature in the presence of 500 μl of FACSPerm for 10 min before washing and quenching. Optimal quenching was determined by observation of the decrease in autofluorescence and nonspecific binding in dot plots of fluorescence channel 1 (FL1) versus fluorescence channel 2 (FL2), ECA (FITC) versus CD45 (PE; Fig. 3a and 3b), and CD14 (FITC) versus CD45 (PE; Fig. 3c and 3d).

Table 2. Quenching of Autofluorescence of BAL-Derived Alveolar Macrophages: Agents Investigated*
Agent investigatedDecrease in MFI (%)
  • *

    Representative of three separate experiments showing percentage of decrease in MFI in the presence of various agents versus no quenching. BAL, bronchoalveolar lavage; CV, crystal violet; MFI, mean fluorescence intensity; PBS, phosphate buffered saline.

No quenching0
0.2% CV/PBS, pH 7.487.1
0.1% η-Octyl β-D-galactopyranoside/PBS76.9
Galactopyranoside + CV102.2
Facsperm5.2
Facsperm + CV75.4
Facsperm + galactopyranoside + CV76.9
Nonfat milk (1% in PBS)30.2
Nonfat milk + CV62.2
Nonfat milk + galactopyranoside + CV64.9
Tween (0.1% in PBS)10.9
Tween + CV78.0
Tween + galactopyranoside + CV80.1

Crystal violet was determined to be the most successful quenching agent, with cell permeabilization by FACSPerm or 0.1% η-octyl β-D-galactopyranoside marginally improving the effects (Table 2). Hence, crystal violet was routinely applied for analysis of BAL.

Statistical Analysis

Correlations between flow cytometry and manual counting differential techniques were performed with Spearman's rank correlation. Bland-Altman plots were used to analyze the degree of variation between the two techniques. The nonparametric Mann-Whitney test was used to analyze the data. This analysis was performed with SPSS software. P < 0.05 was considered statistically significant.

RESULTS

Identification of Brushing-Derived AECs

For bronchial brushings, manual differential cell counting and flow cytometry showed 95% to 97% of the cells to be of an epithelial type (data not shown). The AECs demonstrated well-preserved nuclear and cytoplasmic architectures.

Correlation Between Manual and Flow Cytometric Differential Cell Counts in BAL

The correlation between lymphocyte and neutrophil counts using manual counting or flow cytometric techniques was satisfactory (R2 = 0.75 for lymphocytes and R2 = 0.8 for neutrophils; Fig. 4). Bland-Altman analysis demonstrated a small but significant tendency by cytospins to underestimate the percentage of lymphocytes by 4% (95% confidence interval, −4.03 to −4.19) when compared with flow cytometric methods. The tendency for cytospins to underestimate the percentage of neutrophils was minimal (0.3%; 95% confidence interval, −0.38 to −0.23) when compared with flow cytometric methods. The correlation between alveolar macrophage counts with the two methods was less satisfactory (R2 = 0.5; Fig. 4). Bland-Altman analysis demonstrated a significant tendency by cytospins to overestimate the percentage of macrophages by 5.9% (95% confidence interval, 4.0 to 7.8) when compared with flow cytometric methods. AEC contamination was less than 2% for both methods.

Figure 4.

Correlation between manual and flow cytometric differential cell counts. Analysis of BAL-derived cells using flow cytometry (x axis) and manual counting of Papanicolaou-stained cytospin preparations (y axis) demonstrated lymphocytes (a), alveolar macrophages (b), and neutrophils (c). Note the good correlation between the manual and flow cytometric counting methods for lymphocytes (R2 = 0.75) and neutrophils (R2 = 0.80) but a less satisfactory correlation between methods for alveolar macrophages (R2 = 0.54; n = 25).

Identification and Differential Counting of BAL-Derived Cells

For BAL, leucocyte cell types ranged from 0% to 55% (lymphocytes), from 47% to 97% (alveolar macrophages), and from 0% to 38% (neutrophils). AEC contamination was less than 2%. The percentage of alveolar macrophages or lymphocytes was not significantly higher in BAL from patients with COPD than from control subjects (Fig. 5). The percentage of neutrophils was higher in BAL from patients with moderate to severe COPD and those with mild COPD compared with control subjects, although the difference was significant only for patients with moderate to severe COPD (p = 0.04; Fig. 5). Absolute numbers of lymphocytes were not significantly different between patients with COPD and controls (23 × 105 ± 18 for all patients with COPD vs. 22 × 105 ± 18 for controls, p = 0.5; 18 × 105 ± 18 for patients with mild COPD vs. 22 × 105 ± 18 for controls, p = 0.3; 29 × 105 ± 19 for patients with moderate to severe COPD, p = 0.2). Similarly, absolute numbers of macrophages were not significantly different between patients with COPD and controls (207 × 105 ± 174 for all patients with COPD vs. 125 × 105 ± 50 for controls, p = 0.08; 186 × 105 ± 144 for patients with mild COPD vs. 126 × 105 ± 50 for controls, p = 0.1; 228 × 105 ± 210 for patients with moderate to severe COPD vs. 126 × 105 ± 50 for controls, p = 0.1). As observed for percentage of expression, neutrophil numbers were increased in patients with COPD but although reached statistical significance only in patients with moderate to severe COPD (19 × 105 ± 22 for all patients with COPD vs. 6 × 105 ± 12 for controls, p = 0.051; 16 × 105 ± 21 for patients with mild COPD vs. 6 × 105 ± 12 for controls, p = 0.1; 19 × 105 ± 23 × 105 for patients with moderate to severe COPD, p = 0.03).

Figure 5.

Differential cell counting of BAL-derived leucocytes from all patients with COPD (n = 16), those with mild COPD (n = 8), those with moderate to severe COPD (n = 8), and control subjects (n = 10) demonstrated lymphocytes (a), alveolar macrophages (b), and neutrophils (c). Results are expressed as mean ± two standard deviations. Note the significantly increased numbers of neutrophils in BAL from patients with moderate to severe COPD. *Statistically significant at P ≤ 0.05.

DISCUSSION

In this study we discussed the development of a flow cytometric method for accurate differential counting of cells from heterogeneous BAL and brushing-derived cell populations and the application of this technique to the important lung disease, COPD. BAL-derived alveolar macrophages have increased autofluorescence properties that hinder analysis by flow cytometry due to spectral overlap between alveolar macrophage autofluorescence and fluorochrome emission spectra. Crystal violet was determined to be the most successful quenching agent, with cell permeabilization by FACSPerm or 0.1% η-octyl β-D-galactopyranoside marginally improving the effects. Crystal violet is a dye that readily enters cells and that, being violet, absorbs green light. This allows nonradiative energy transfer to occur between the crystal violet and the green autofluorescence material. Because crystal violet is nonfluorescent, the transferred energy is lost in a nonradiative manner. This method enabled cell type identification in BAL from smokers and allowed further staining for surface markers and cytokines currently being investigated for COPD and control groups.

Differential cell counts of lymphocytes, neutrophils, and alveolar macrophages by manual counting and flow cytometry were compared. Unlike asthma, COPD is associated with few eosinophils in the airway (2). Therefore, eosinophils were not identified and counted. Although there was good correlation between methods for lymphocyte and neutrophil counts, the correlation for alveolar macrophage counts was less satisfactory. Manual counting as opposed to flow cytometry overestimated macrophage numbers. Stained preparations of BAL showed numerous cells with apoptotic changes, making cell type identification difficult. It is possible that some necrotic neutrophils or lymphocytes may have been classed as macrophages on examination of stained BAL due to the loss of an obvious “segmented nuclear” appearance of some apoptotic neutrophils and loss of cell structure, as previously reported (21). Manual counting marginally underestimated the percentage of lymphocytes compared with flow cytometry. It is also possible that manual counting of 500 cells underestimated cells present in small numbers (vs. flow cytometric count of 50,000 events). These findings are consistent with those of Barry et al. (12) who found a small but significant tendency by cytospins to underestimate the percentage of lymphocytes and overestimate the percentage of macrophages in BAL when compared with flow cytometry.

Consistent with reports by others (4, 22) we found no significant change in the percentage of alveolar macrophages or absolute alveolar macrophage numbers in the groups tested. These results are in contrast to reports of increased macrophage numbers in bronchial biopsy in COPD (2, 23).

Similarly, as found by others (4, 22), there were no significant changes in the percentages of lymphocytes or absolute numbers of lymphocytes in patients with COPD in the present study.

COPD is associated with increased neutrophil numbers in BAL (22). Our results support these findings; percentages and absolute numbers of neutrophils were higher in BAL from patients with moderate to severe COPD and those with mild COPD compared with control subjects. However, the difference was significant only for patients with moderate to severe COPD. In contrast to these findings in BAL, studies of bronchial biopsies in COPD have reported no significant changes in neutrophil numbers (23). Thus, it appears that the inflammatory processes present in the distal airway lumen and alveoli (investigated using BAL) do not reflect those in the airway wall (investigated using biopsy), with a higher percentage of neutrophils being present in the BAL and a predominance of alveolar macrophages and lymphocytes in the airway wall. This is supported by studies that showed no correlation between inflammatory cell numbers in BAL and biopsy (8, 9). A possible explanation for the discrepancy is the rapid migration of neutrophils across the tissue into the airway lumen, thus making increased neutrophil numbers undetectable by tissue analysis but detectable by BAL. It is also possible that cytokines, including granulocyte-macrophage colony-stimulating factor and transforming growth factor β, released by neighboring AECs and alveolar macrophages into the airway lumen, augment the prolonged survival of neutrophils in COPD (24). We previously reported upregulation of transforming growth factor β and increased apoptosis of peripheral blood-derived T-cells in COPD (25), and we are currently using the flow cytometric techniques described in this report as a basis to further investigate production of these cytokines in the airways.

In conclusion, we have described a novel and readily applicable flow cytometric assay for identification of various cell types from heterogenous samples of BAL and bronchial brushing. We have shown that this method is more accurate than manual differential cell counting and, hence, provides more valid results when used in the investigation of COPD and other inflammatory lung diseases. The pathogenesis of COPD and therefore the rationale for therapy are poorly understood, and currently there are no satisfactory treatments for this important disease. These methods also allow further investigation of the characteristics of infiltrating cells, such as cytokine production and receptor expression, and an accurate evaluation of apoptosis for different cell types. Greater insights into the inflammatory processes in COPD should help to provide a rationale for urgently required effective treatments for COPD.

Acknowledgements

S.J.H. is supported by an Australian Lung Foundation/Boehringer Ingelheim Chronic Airflow Limitation Fellowship.

Ancillary