INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary infection and neutrophil-dominated inflammation are the main causes of morbidity and mortality in cystic fibrosis (CF) (1). Recent studies have shown increased concentrations of interleukin 8 (IL-8) and high numbers of neutrophils in bronchoalveolar lavage fluids (BALF) of some infants and young children with CF even in the absence of positive bacterial cultures (2, 3). These findings have raised the question of whether regulation of the inflammatory response is abnormal in the lungs of CF patients (4). We have previously reported that BALF from infected children with CF contained significantly more IL-8 and neutrophils than BALF from infected children with other chronic respiratory conditions (5). Preliminary analysis in that report indicated that the CF inflammatory response was increased relative to control subjects, even if bacterial quantity was taken into account (5). This finding led us to hypothesize that the inflammatory response to bacterial infection is exaggerated in the infant CF airway.

We have continued to collect and analyze BALF from pediatric patients, to further quantify the relationship between bacterial infection and inflammation in CF relative to other conditions. The specific goals of the present study were to improve the validity of comparisons between CF and other patients by controlling for the variables of bacterial quantity and type, by increasing the number of patients studied, and by expressing quantity of inflammation in a way that does not depend on estimation of dilution of lung secretions by lavage fluid.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

All infants and children who underwent clinically indicated bronchoscopies between January 1995 and December 1997 were eligible for participation in this study. Exclusion criteria were the use of inhaled or systemic steroids or nonsteroidal antinflammatory medications during the 2 mo preceding the bronchoscopy. The use of other medications during this time was recorded and included bronchodilators, antireflux therapy, and recent antibiotics in both groups. All subjects were off antibiotics for at least 48 h prior to bronchoscopy and none of the subjects was currently receiving recombinant human deoxyribonuclease (rhDNase). Patients in whom viral or mycobacterial infection was detected in BALF were excluded from the data analysis. Informed consent was obtained from all parents and informed assent was obtained from patients older than 6 yr. The study was approved by the University of North Carolina Committee on the Protection of the Rights of Human Subjects.

Bronchoscopy

Bronchoscopies were performed according to clinical routine procedure as we previously described (5). Children were sedated or placed under general anesthesia, and topical lidocaine was instilled at the level of the larynx and the tracheal bifurcation. To avoid aspiration or contamination of the specimen, patients were placed in a head down Trendelenburg position and suctioning through the bronchoscope was only performed below the vocal cords. The location of bronchoalveolar lavage (BAL) was at the discretion of the bronchoscopist, but was generally done in the lung segment most affected by disease, as evidenced by radiographic changes or by visual appearance at bronchoscopy. The bronchoscope was wedged in a bronchus and two to three 10-ml aliquots of buffered normal saline solution were instilled and immediately aspirated through the bronchoscope. In some smaller infants 5-ml aliquots were used. The total volume of instilled lavage fluid was 1 to 3 ml/kg body weight and the volume return was 53 ± 0.6% for children with CF and 48 ± 0.02% for control subjects (mean ± SEM; p = 0.56). An aliquot of BALF for the research project was taken in a sterile manner. The remainder of the BALF was sent to the hospital laboratory for cytology and cultures, which were ordered by the physician responsible for the patient.

Microbiology

BALF specimens were cultured for gram-positive and gram-negative bacteria on horse blood, colistin-nalidixic acid (CNA) agar, and MacConkey agar. Mannitol-salt agar was used for recovery of Staphylococcus aureus (6) and Pneudomonas cepacia (PC) agar for recovery of Burkholderia cepacia (7) in CF patients. Cultures and stains for viruses (including cytomegalovirus, respiratory syncytial virus, influenza A and B, parainfluenza 1 to 3, and adenovirus), fungi, and mycobacteria were carried out for all CF patients, and for control patients when these infections were clinically suspected. "Oropharyngeal flora" in BALF specimens were defined according to standard protocol for the University of North Carolina (UNC) Hospitals Clinical Microbiology Laboratory. These consist of a mixture of resident flora including Streptococcus viridans, nonhemolytic streptococci, saprophytic Neisseria species, diphtheroids, coagulase-negative Staphylococcus species, micrococci, lactobacilli, hemolytic Haemophilus species, Haemophilus parainfluenzae, Eikenella corrodens, non-Group A beta-hemolytic Streptococcus, Bacillus species, and Stomatococcus species.

Cell Count and Cytokine Levels

The samples were processed immediately in the bronchoscopy laboratory. BALF was gently mixed and the total number of white cells in the native, unfiltered lavage fluid was determined in a hemocytometer. Microscope slides were prepared in a Wescor Cytopro 7620 cytocentrifuge (Logan, UT) by centrifuging 150 µl fluid at 750 rpm for 4 min. To obtain a cell count of approximately 5 × 10⁴ cells per slide, some samples had to be further diluted in normal saline lavage fluid before loading onto the cytocentrifuge. Slides were fixed and stained with modified Wright-Giemsa stain (Fisher Diagnostics, Pittsburgh, PA). Differential cell counts were determined by counting 200 consecutive cells under oil immersion (magnification ×400) and classifying them as neutrophils, macrophages, lymphocytes, eosinophils, or epithelial cells. Differential cell counts were made without knowledge of BALF culture results.

Cell-free supernatant was obtained by centrifuging the remainder of the BALF at 1,500 rpm for 5 min. The supernatant was immediately frozen on dry ice, then stored at 70° C until assayed. Immunoreactive IL-8 was measured with a specific, commercially available ELISA (R&D Systems, Minneapolis, MN) having a working range of 31.2 to 2,000 pg/ml. BALF samples having greater than 2,000 pg/ml IL-8 were diluted and reassayed. Correlation coefficient of standard curves for all IL-8 assays was > 0.98. The technician performing ELISA assays was blinded as to patient diagnosis and BALF culture results.

Statistical Methods

Data are reported as mean ± SEM unless otherwise stated. All data were log-transformed prior to statistical analysis. For ratios of inflammatory markers to bacteria, comparisons between CF and control subjects were made using a 2-tailed Student's t test. Linear regressions were calculated by the method of least squares, and slopes and elevations of the lines for CF and control subjects were compared using Student's t test (8). We also fitted a 3-parameter sigmoidal shaped curve to CF and control data using the following formula: Y = L + (U  L)/[1 + EC₅₀ · exp (X)], where Y = neutrophils or IL-8, X = bacterial colony-forming units (cfu), L = Y at lower plateau, U = Y at upper plateau, and EC₅₀ = bacterial cfu that gives a response halfway between upper and lower plateaus. Statistical significance was defined as p < 0.05 throughout. Statistical computations were carried out on a personal computer using the statistical software packages InStat2 and Prism2 (both from GraphPad, San Diego, CA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

BALF from a total of 59 CF and 59 control children were processed in the study. Indications for bronchoscopy in CF subjects included obtaining sputum cultures in patients unable to expectorate who either had pulmonary exacerbations, or were newly diagnosed and presented with pulmonary symptoms. The indications for control patients to undergo bronchoscopy were one or more of the following complaints (number of patients in parentheses): recurrent or persistent pulmonary infiltrates (24), upper airway obstruction and stridor (20), chronic cough and wheeze (11), obstructive sleep apnea (2), atelectasis (1), and anatomic evaluation for laryngoesophageal cleft (1). Four control and five CF patients were excluded from further analyses because their BALF grew viruses or mycobacteria. This left 54 CF and 55 control BALF for further analyses. The ages and demographics of the excluded patients were similar to those of the other participants. The median age of the CF patients was 1.8 yr (range, 3 wk to 13.3 yr), and for the control patients 1.0 yr (range, 2 wk to 8.3 yr).

BALF

For data analysis, BALF samples were classified as "infected" based on a bacterial count of > 5 × 10⁴ cfu/ml of pathogenic bacteria, or "uninfected" based on a bacterial count of < 1 × 10⁴ cfu/ml of pathogenic bacteria. These values were based on previously published adult BALF studies (9, 10). Thus defined, 28 of the 54 CF and 27 of the 55 control patients were infected. Patients in each group who had bacterial counts between 1 × 10⁴ and 5 × 10⁴ cfu/ml were not included in either category. There were three CF and five control patients in this indeterminate group. Oropharyngeal flora (OPF) were considered as nonpathogenic bacteria for this classification and were not included in bacterial counts.

Results of cell count in BALF are summarized in Table 1. The total number of white cells was significantly higher in CF than control patients, whether infected or uninfected patients were compared. CF patients also had a higher percentage of neutrophils than control patients. The total number of neutrophils was significantly higher in CF patients than in the non-CF infants (p < 0.001). Overall, neutrophil counts in uninfected CF patients were of the same magnitude as those of the infected control patients. The same pattern was observed for IL-8 concentrations, which were significantly higher in the CF patients than in control subjects, both in the presence and absence of infection (p < 0.001).

Microbiology

The pathogens most frequently recovered from BALF of CF patients were P. aeruginosa (n = 9), Haemophilus influenzae (n = 10), and S. aureus (n = 10), either as single pathogens or as mixed infections. Other bacteria in mixed infections included Escherichia coli and Xanthomonas maltophilia and Serratia marcescens in a total of six patients. B. cepacia was not isolated in any sample but Burkholderia gladioli was found in a 2-yr-old CF patient; this patient was described by Barker and coworkers in a previously published case report (11). Culture results in the control group showed H. influenzae in 13 and Moraxella catarrhalis in 10 samples; four of these patients had concomitant infection with both pathogens. Six control patients were infected with Streptococcus pneumoniae as the only pathogen, and one other infant had mixed infection with S. pneumoniae and M. catarrhalis. S. aureus was isolated in only one control patient who had DiGeorge syndrome.

Inflammatory Response to Specific Pathogens

Acquisition of P. aeruginosa is associated with decline in lung function in children with CF (12, 13). Therefore, among CF patients, we analyzed whether there was an increased inflammatory response to P. aeruginosa. Compared with other infected CF patients, P. aeruginosa-positive children were older than those without this pathogen (7.6 ± 1.6 yr versus 1.9 ± 0.4 yr). However, there was not a significant increase in either BALF neutrophils (6,045 ± 1,819 versus 3,190 ± 768 × 10⁶ polymorphonuclear leukocytes [PMN]/ml, p = 0.09) or IL-8 (18,560 ± 6,516 versus 18,769 ± 5,959 pg/ml, p = 0.68) associated with P. aeruginosa compared with other pathogens. Among patients who had single-pathogen infections with either P. aeruginosa, S. aureus, or H. influenzae, there were no statistically significant differences for either neutrophils or IL-8.

Ratios of Inflammatory Markers to Bacteria

To account for bacterial "load" in our analysis of inflammatory responses to bacterial infection, we calculated ratios of inflammatory markers (neutrophils, IL-8) to bacterial counts for each BALF sample. BALF having either no detectable pathogenic bacteria or unquantifiably high numbers of bacteria were excluded from this analysis, as a ratio cannot be quantified for such samples. Figure 1 compares the distributions of these ratios for CF and control subjects. Ratios were calculated both including and excluding OPF, which did not alter the findings. If all quantifiable samples were included, CF samples had much greater values than control samples, both for neutrophils (mean ± SEM, 36 ± 18 [n = 21] versus 2.1 ± 1.5 [n = 27]; p < 0.01) and for IL-8 (0.37 ± 0.16 [n = 22] versus 0.01 ± 0.004 [n = 27]; p < 0.001). If the analysis was limited to BALF samples having more than 5 × 10⁴ cfu/ml bacteria, ratios were still greater in CF samples, although the difference between CF and control samples was not as marked. In this more heavily infected group, the neutrophil to bacteria ratio was 4.9 ± 2.3 for CF (n = 13) versus 2.3 ± 2.0 for control samples (n = 20) (p = 0.02); the IL-8 to bacteria ratio was 0.08 ± 0.07 in CF (n = 13) versus 0.003 ± 0.0008 in control samples (n = 20) (p = 0.01). Inclusion of data from the few BALF samples with positive viral or mycobacterial cultures had no effect on the comparison between CF and control groups (data not shown).

View larger version (17K): [in this window] [in a new window]   Figure 1.   Ratio of (A) neutrophils or (B) IL-8 to bacterial cfu in BALF from children with CF (closed circles) or children with other chronic respiratory conditions (open circles). Data are shown both including and excluding OPF in bacterial totals. Ratios are shown on a logarithmic scale. Each data point represents a separate BALF sample. Horizontal lines indicate mean of log-transformed data. For each comparison, CF was significantly greater than controls (p < 0.01, Student's t test).

Correlation of Bacterial Count with Inflammation

To further define the quantitative relationship between infection and inflammation, BALF inflammatory markers were plotted as a function of bacterial quantity. In this analysis, samples from all subjects were included, regardless of whether they were considered to be "infected" or not. For BALF samples in which both neutrophils and bacterial counts were fully quantified, calculation of linear regression for CF and control neutrophil counts yielded parallel lines whose slopes were both significantly positive (p < 0.05). However, there was highly significant elevation of the line for CF compared with control groups (p < 0.001; Figure 2A).

View larger version (15K): [in this window] [in a new window]   Figure 2.   Log10 neutrophils, plotted as a function of log10 bacterial cfu for BALF from children with CF (closed circles) and other chronic respiratory conditions (open circles). (A) Data for samples with numbers of bacteria accurately quantified between 103 and 108 cfu/ml, with linear regression lines for CF (solid line) and control (dotted line) data. Slopes of the lines were not significantly different, but elevation of CF line was significantly greater than control (p < 0.001). (B) Same data as in (A), with the addition of data for samples having < 103 cfu/ml bacteria (assigned an arbitrary low value for bacterial cfu), and samples having > 106 cfu/ml (but unquantified) bacteria (assigned an arbitrary high value for bacterial cfu). Solid line represents sigmoidal curve fitted to CF data; dotted line represents sigmoidal curve fitted to control data.

For descriptive purposes, in Figure 2B we added neutrophil data for samples with very low numbers of bacteria (< 10³ cfu/ml) or with unquantified but high (> 10⁶ cfu/ml) numbers of bacteria to the data already described. BALF with very low bacterial counts were assigned an arbitrary bacterial count of 200 cfu/ml; BALF with unquantified but very high bacterial counts were assigned an arbitrary count of 2 × 10⁸ cfu/ml. Assuming that concentrations of neutrophils and IL-8 are low but measurable in the absence of infection, and that concentrations of these inflammatory markers reach a maximum at some point, we then fitted a sigmoidal curve to data for CF and for control patients. This also yielded parallel curves, with a shift upward for CF compared with control data.

Performing similar linear regression analysis for IL-8 data again yielded lines with significantly positive slopes, but for IL-8 the slopes differed significantly (p = 0.016), with the lines diverging at lower levels of infection (Figure 3A). Addition of semiquantitative data and fitting of sigmoidal curves (Figure 3B) again suggested that while the data for CF were shifted upward at a given quantity of bacteria, there was a relatively greater elevation of IL-8 in CF compared with control samples in the presence of lower bacterial quantities.

View larger version (14K): [in this window] [in a new window]   Figure 3.   Log10 IL-8, plotted as a function of log10 bacterial cfu for BALF from children with CF (closed circles) and other chronic respiratory conditions (open circles). (A) Data for samples with numbers of bacteria accurately quantified between 103 and 108 cfu/ml, with linear regression lines for CF (solid line) and control (dotted line) data. Slope of the line for CF was significantly less than that for control data (p = 0.016). (B) Same data as in (A), with the addition of data for samples having < 103 cfu/ml bacteria (assigned an arbitrary low value for bacterial cfu), and samples having > 106 cfu/ml (but unquantified) bacteria (assigned an arbitrary high value for bacterial cfu). Solid line represents sigmoidal curve fitted to CF data; dotted line represents sigmoidal curve fitted to control data.

H. influenzae was the only bacterial pathogen that was commonly recovered in BALF from both CF and control patients. Figure 4 shows neutrophils and IL-8 as a function of bacterial quantity for subjects who had H. influenzae as the single bacterial pathogen. There was clearly a tendency for increased inflammation in CF patients, but owing to small sample size, linear regression was not performed. The ratios of neutrophils and IL-8 to quantified H. influenzae were as follows (mean ± SEM): neutrophils/cfu H. influenzae = 7.52 ± 6.23 (CF) versus 0.78 ± 0.57 (control) (p = 0.39); and IL-8/cfu H. influenzae = 0.21 ± 0.2 (CF) versus 0.01 ± 0.01 (control) (p = 0.18). Thus, although CF inflammation was not statistically greater than control for this subset of samples, the ratio means were of similar magnitude to those for the larger, statistically different groups (Figure 1).

View larger version (12K): [in this window] [in a new window]   Figure 4.   Relationship between infection and inflammation for BALF samples having only H. influenzae as a bacterial pathogen. (A) Log10 neutrophils and (B) log10 IL-8 plotted as a function of log10 bacterial cfu for BALF from children with CF (closed circles) and other chronic respiratory conditions (open circles).

Because it is conceivable that aspirated OPF may function as pathogens or stimulate inflammation in CF, we also plotted inflammation as a function of quantity of OPF for BALF that had no other pathogens present (Figure 5). As for H. influenzae, there was a tendency for BALF from CF patients to contain more inflammation at a given amount of OPF than BALF from control subjects, though it was not clear that inflammation increased with greater amounts of OPF for either group. Ratios of inflammatory markers to OPF calculated from these data were significantly greater for CF patients, both for neutrophils (p = 0.013) and IL-8 (p < 0.001).

View larger version (12K): [in this window] [in a new window]   Figure 5.   Relationship between infection and inflammation for BALF samples having no pathogens other than OPF. (A) Log10 neutrophils and (B) log10 IL-8 plotted as a function of log10 bacterial cfu for BALF from children with CF (closed circles) and other chronic respiratory conditions (open circles).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies of inflammation in early CF have classified subjects as infected or uninfected based on number of bacteria in BALF, but have not quantitatively assessed the effect of bacterial load on inflammation (2, 5, 13). In the present study, we have found that BALF neutrophils and IL-8 are significantly increased in young children with CF compared with children with other chronic respiratory conditions, even after adjusting these markers of inflammation for quantity of bacteria present in the BALF. Because we expressed inflammation as a ratio to or function of bacterial quantity for each BALF sample, our comparisons do not depend on estimates of dilution of lung secretions by lavage fluid. Increased inflammation in CF occurred in the presence of both low and high amounts of bacteria, and neither age nor percent lavage return differed significantly between CF and control subjects.

In contrast to the data for neutrophils (Figure 2), comparison of the linear regressions of IL-8 versus bacteria (Figure 3) showed different slopes for the CF and control groups because of proportionally higher IL-8 at lower levels of infection among CF patients. It is possible that at very high levels of infection, airway IL-8 production reaches a maximum which is similar in CF and control subjets, whereas neutrophil recruitment due to other factors continues to be enhanced in CF.

Because acquisition of infection with P. aeruginosa has been associated with decline in lung function in children with CF (12, 13), it may be hypothesized that this pathogen elicits more inflammation than other bacteria, thus promoting the exaggerated inflammation observed in our study. However, we found no difference in concentrations of neutrophils or IL-8 between BALF from P. aeruginosa-positive and P. aeruginosa- negative CF patients. Furthermore, comparison of inflammatory responses to H. influenzae between CF and control subjects suggested that increased inflammation per bacterial quantity exists for CF, even in the presence of a pathogen other than P. aeruginosa. We therefore conclude that the specific bacterial pathogen is unlikely to be the sole determinant of the inflammatory response in the early CF airway.

Viral culture was performed in all CF BALF, but in control subjects only if clinically suspected. It is thus possible that some control subjects could have had undetected viral infection. Because viral infection is proinflammatory, this would have led to an overestimation of inflammatory responses to bacteria among our control subjects, and would not have altered our conclusion that there is greater inflammation in CF patients.

The great majority of our BALF specimens, in both CF and control patients, contained variable quantities of OPF. The presence of these organisms in BALF is presumably due to some combination of aspiration of oral secretions, and contamination of specimens by passage of the bronchoscope through the upper airway. It is not known whether aspirated OPF can function as bacterial pathogens in CF. We therefore also compared CF with control BALF inflammation in the presence of this "pathogen" alone (Figure 5). As for H. influenzae, there was greater inflammation among CF patients with only OPF in BALF compared with control subjects. Although this could reflect increased basal inflammation in uninfected CF patients, an intriguing possibility is that sufficient quantities of aspirated OPF might trigger excessive inflammation in the young CF airway.

It has long been held that chronic influx of activated neutrophils into infected airways is responsible for damage to the CF airway and progressive respiratory insufficiency (14, 15). The beneficial effects of anti-inflammatory therapies on lung function in CF are supportive of this hypothesis (16, 17). The role of inflammation as an independent factor in the early pathogenesis of CF lung disease has gained increased attention recently. Previously published autopsy studies (18) and recent BALF data from Armstrong and coworkers (19) suggest that there is at least a short period during early infancy in which there is little or no infection or inflammation. However, other recent studies have shown that later in infancy and early childhood, a portion of CF patients have inflammation in BALF without detectable infection (2, 3, 5), raising the question of whether control of inflammation is altered in CF. Epithelial cell culture studies have suggested that cystic fibrosis transmembrane regulator (CFTR) dysfunction may be linked to increased IL-8 production via upregulation of P. aeruginosa receptors (20), although in our laboratory IL-8 production in response to the nonspecific stimuli tumor necrosis factor- (TNF-) and respiratory syncytial virus was not increased relative to non-CF control cells (21). Alternatively, evidence has been reported that the anti-inflammatory cytokine interleukin-10 (IL-10) is downregulated in CF epithelia (22).

Our BALF data are also consistent with hypotheses of CF pathogenesis that do not invoke primary defects in control of inflammation. Infection and inflammation are apparently intermittent early in the disease (19), but chronically high levels of bacteria are established in the airway at some point during childhood in CF, regardless of clinical status (1). Thus, although the control subjects in our study all had subacute or chronic respiratory conditions as indications for bronchoscopy, the chronicity of infection itself may have been greater in our CF subjects. Chronic bacterial infection, combined with poor clearance of bacterial breakdown products or exoproducts due to altered secretions, might result in the continued presence in the airway of proinflammatory stimuli other than live bacteria, such as endotoxin. In this case, the inflammatory "abnormality" in CF might actually be secondary to persistence of as yet undefined inflammatory stimuli, rather than due to a primary defect in control of inflammation.

In summary, we have shown that there is an unusually high level of airway inflammation in early CF in relation to numbers of live bacteria recovered. Further mechanistic studies will be necessary to determine whether this represents a primary or secondary phenomenon in CF airway pathogenesis. In either case, our data provide further rationale for anti- inflammatory therapy in young patients with this disease.