INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: sputum; children; cystic fibrosis; neutrophils; interleukin-8; neutrophil elastase

Cystic fibrosis (CF) is characterized by recurrent airway infections with selected bacterial pathogens and neutrophil-dominated airway inflammation (1). Inflammation begins at an early age (2) even in the absence of concomitant infection, and persists and progresses throughout life, ultimately leading to lung destruction (6). Quantitative measures of infection and inflammation are therefore important in staging disease and evaluating new treatments. In addition, the relationship between infection and inflammation in CF remains unclear (1, 6). Each of the current techniques used to define the microbiology and inflammatory response in the CF airway has important limitations. Expectorated sputum provides an accurate measure of infection and inflammation in the lower airways (7, 8) but many children with CF are unable to expectorate sputum. Fiberoptic bronchoscopy and bronchoalveolar lavage (BAL) is invasive, risky, and costly. Serial BALs are particularly difficult to perform. Furthermore, BAL generally samples only one or two segments of the lung, possibly limiting detection of infection. Oropharyngeal cultures, commonly used in young children with CF who are not capable of expectorating sputum, do not reliably predict the presence of lower airway pathogens (9), lack sensitivity for identifying Pseudomonas aeruginosa and Staphylococcus aureus (10), and provide no information about inflammation.

In older children who cannot spontaneously expectorate sputum, sputum induction can be a helpful diagnostic tool. Sputum induction is performed by inhaling aerosolized hypertonic saline (HS) (11). Sputum induction is useful for the cytologic diagnosis of malignancy (12) and for the detection of Pneumocystis carinii (13) and tuberculosis (14). Sputum induction is also a valid and acceptable method for sampling the airways of subjects with asthma to analyze both cellular and biochemical markers of inflammation (15). Yet this diagnostic tool has received very little attention in CF. One systematic study found that sputum induction is safe and acceptable in nonexpectorating children with CF (22). Another group collected induced sputum samples from those patients with CF who could not expectorate, yet did not directly compare these results with those obtained from spontaneously expectorated specimens (23). Recently, sputum induction was compared with expectorated sputum and bronchoalveolar lavage in 10 adults with CF (24). In their subjects, induced sputum yielded similar total cell counts and differentials, and similar detection rates of CF-related bacterial pathogens.

Before applying sputum induction to nonexpectorating children with CF, this technique should confirm previously demonstrated abnormalities between children with CF and healthy children and should yield measures of inflammation and infection that are similar to spontaneously expectorated sputum. In this study, our objectives were to evaluate the presence and degree of airway inflammation and infection, as assessed in induced sputum, in children with CF and healthy children and to directly compare induced sputum and spontaneously expectorated sputum in describing inflammation and infection in the CF airway. To control for salivary contamination, we separately collected and analyzed saliva samples for the same measures of inflammation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Retrospective Determination of Sputum Expectoration in Children with CF

We retrospectively evaluated spontaneous sputum production in our population of patients with CF. At the annual visits, efforts were made by a respiratory therapist to have the patients expectorate sputum. We then determined the proportion of patients between 6 and 12 yr of age who successfully expectorated sputum.

Prospective Study of Sputum Induction

Subjects and controls. Between August and October 2000, we prospectively recruited and studied 20 subjects with CF. All of these patients had a proven diagnosis of CF as evidenced by a sweat chloride > 60 mmol/L. Subjects were enrolled and studied following a routine outpatient clinic visit. Inclusion criteria for the patients with CF included age 7 yr and older, ability to perform reproducible spirometry, and stable pulmonary disease, as defined by both clinical impression and having had no hospitalizations or changes in antibiotic regimen within 1 mo prior to being studied. Subjects with a baseline FEV₁ of less than 50% predicted or who had received any hypertonic saline or investigational agents within 1 mo of the study were excluded. We enrolled both sputum producers (capable of spontaneously expectorating sputum by history) (n = 7) and nonproducers (n = 13). Also, 11 healthy children were recruited as a control population and studied. These healthy subjects had no history of asthma, an FEV₁ > 80% predicted, no respiratory symptoms within 2 wk of being studied, and were nonsmokers. The clinical characteristics of the study subjects are shown in the online data supplement (Table E1).

The protocol was approved by the Institutional Human Subjects Review Board at the University of Colorado Health Sciences Center, and informed consent was obtained from each of the subjects and/or their caretakers.

Study design. Prior to sputum induction, an oropharyngeal culture was obtained in all except one healthy control subject (patient refused). Salivary specimens were collected by direct expectoration into a specimen container. From the CF producers, a spontaneously expectorated sputum specimen was collected following three huff coughs. All subjects were pretreated with albuterol (180 µg by metered-dose inhaler with a spacer device) prior to the induction procedure.

Sputum induction and saliva collection. Sputum was induced by administering 3% hypertonic saline (HS) for six 2-min sessions, a method adapted from Wong and Fahy (25). The nebulized HS was delivered by an Ultraneb 99 ultrasonic nebulizer (DeVilbiss, Somerset, PA), the reservoir of which was filled with 20 ml of 3% HS every 4 min during the 12-min induction. After each 2-min inhalation period, the subjects were directed to expectorate all saliva into a "saliva" container, rinse their mouths with water and spit into a disposable cup, perform forced expiratory techniques and directed coughing, and then expectorate sputum into a sterile "induced sputum" specimen container. In the event that induction was discontinued prior to 12 min, either because of symptoms or fall in pulmonary function, sputum was analyzed as long as greater than 0.5 g was collected. The sputum, saliva, and oropharyngeal specimens were then transported on ice to the laboratory for immediate processing within 20 min. We examined separate saliva samples from six control subjects, nine CF nonproducers, and five CF producers for indices of inflammation.

Safety monitoring. During the sputum induction, subjects were monitored by chest auscultation at 0, 2, 6, and 10 min into the procedure, by spirometry at 0, 4, 8, and 12 min, and by continual oxygen saturation (Sa_O2). If the FEV₁ (L) fell by > 10% but < 20% of the initial postbronchodilator value or if the patient so requested, another two puffs of albuterol (180 µg) was administered to the patients, and the induction was continued as long as the FEV₁ improved to within 10% of the initial postbronchodilator value. If the FEV₁ fell by > 20% or if troublesome symptoms occurred (wheezing on auscultation, intolerability claimed by the child, or drop in Sa_O2 to < 90%), the sputum induction was discontinued.

Sputum and saliva processing. The weights of the sputum and saliva samples were separately determined. An equal volume of sterile dithiothreitol (DTT) (Sputolysin; Calbiochem-Novabiochem Corp., San Diego, CA), freshly diluted to 10% by the addition of sterile saline, was added to both the sputum and saliva. This step was performed under a Bio-safety hood using sterile technique. The samples were then incubated in a shaking water bath at 37° C for 5-10 min, and gently mixed using a transfer pipette at 5-min intervals. This typically resulted in "gross" homogenization of the sputum samples and complete homogenization of the saliva samples. One milliliter of the sputum mixture was removed for quantitative culture for CF pathogens. The weight of the remaining sputum mixture was measured, and a further three times the volume of both DTT and phosphate-buffered saline (Dulbecco's; Gibco BRL, Grand Island, NY) were added. The mixture was incubated once again in the 37° C shaking water bath for another 5-10 min to ensure complete homogenization. Ten microliters of the homogenized sputum and saliva samples, mixed with Trypan Blue stain, was used to calculate total cell counts, using a standard hemacytometer. A further 0.25-0.50 ml of both samples was used to prepare cytospin slides for differential cell counts. After staining the slides with Wright stain, two laboratory technologists, blinded to the subject's history, counted 300 cells. Cell differentials, inclusive and exclusive of squamous cells, were calculated. Only samples that weighed greater than 0.5 g and had greater than 20% nonsquamous cells were considered adequate and included for further analysis. The differential cell count could not be determined in an induced sputum specimen from one of the CF producers because the cytospin was of poor quality.

The remaining homogenized sputum and saliva samples were centrifuged at 500 × g for 10 min at 4° C. The supernatants were aspirated and centrifuged a second time at 4,000 × g for 20 min. The supernatants from this second spin were divided into two aliquots. Approximately two-thirds were treated with the protease inhibitors, phenylmethylsulfonylfluoride (PMSF) (Sigma Diagnostics, St. Louis, MO) and ethylenediamenetetraacetic acid (EDTA) (Sigma Diagnostics), in order to minimize proteolytic activity. The remaining one-third was left untreated. The supernatants were frozen at 70° C for later analysis.

Biochemical assays. Interleukin-8 (IL-8), albumin, total protein, and urea were measured in the thawed sputum and saliva supernatants treated with protease inhibitors. Neutrophil elastase activity was measured in the untreated supernatants. IL-8 was measured by a commercially available ELISA kit according to the manufacturer's recommended protocol (Quantikine, R&D Systems, Minneapolis, MN). Neutrophil elastase activity, total protein, and urea concentrations were quantified by spectrophotometric assays (Sigma Diagnostics) (4). Albumin levels in sputum and saliva samples were determined by a competitive radioimmunoassay, according to the package insert (Diagnostic Products Corporation, Los Angeles, CA). The lower limits of detection for these assays were IL-8, 31 pg/ml; neutrophil elastase, 0.5 µg/ml; albumin, 2.5 µg/ml; total protein, 1.0 mg/dl; urea, 0.1 mg/dl.

Microbiology. Quantitative microbiology on the sputum specimens was performed according to a consensus conference on CF microbiology (26). The oropharyngeal specimens were graded from 0 (no growth) to 4+ (heavy growth).

Statistical Analysis

Comparisons between the CF groups and control subjects were made for each variable (FEV₁, sputum weights, sputum inflammatory markers, and sputum bacterial colony counts); relationships between sputum inflammatory measures were assessed using a standard statistical package (StatView, SAS Institute Inc., Cary, NC). Sputum and saliva samples were analyzed separately. A log₁₀ transformation was used for data with a nonnormal distribution (cell counts, differentials, and each of the biochemical indices). Analysis of variance (ANOVA) was used to assess differences between the groups, using a Fisher's least significant difference procedure for multiple comparisons. Agreement between measures in induced and spontaneously expectorated sputum specimens was determined by calculating Pearson correlation coefficients (r) and constructing Bland-Altman plots (27). Results are presented as the mean ± SEM unless otherwise noted. Significance was assumed for p values less than or equal to 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Retrospective Determination of Sputum Expectoration in Children with CF

The percent of patients with CF who can successfully expectorate sputum at varying ages is shown in Figure 1.

View larger version (12K): [in this window] [in a new window]   Figure 1.   Percent of children with CF who are capable of expectorating sputum at different ages. The number of children included in the analysis is shown in parentheses.

Prospective Study of Sputum Induction

Safety. The subjects with CF experienced a statistically significant decline in FEV₁ during the procedure (CF nonproducers, mean fall in FEV₁, 5.9 ± 1.3%; CF producers, 7.0 ± 3.2%) as compared with the healthy subjects (0.7 ± 1.0%; p = 0.02 for each comparison). In six (46%) CF nonproducers and three (43%) CF producers, the drop in FEV₁ during the procedure exceeded 10%, while in one (14%) CF producer, it exceeded 20%. One healthy subject experienced a greater than 20% fall in FEV₁. Two subjects with CF whose FEV₁ were less than 80% predicted experienced a greater than 10% fall in FEV₁ during the procedure. Oxygen saturations fell in the patients with CF (CF nonproducers, mean fall in O₂, 1.3 ± 0.3%; CF producers, 2.1 ± 0.4%) but this was not significant as compared with the control subjects (1.0 ± 0.4%, p = 0.62 and p = 0.08, respectively). The lowest documented saturation during the induction procedure was 90%, in a CF producer. The procedure was discontinued prior to completion in seven subjects (one control, three CF nonproducers, three CF producers) either because of symptoms (n = 6) or a > 20% fall in FEV₁ (n = 1). Furthermore, eight (62%) CF nonproducers, four (57%) CF producers, and one (9%) control subject received additional albuterol during the procedure either because of audible wheezes or because the patients requested the albuterol based on a subjective feeling of chest tightness. No subject developed bronchoconstriction that was refractory to additional albuterol or needed ongoing monitoring beyond 30 min after the completion of the procedure.

Adequacy of samples. Seven of the 11 healthy control subjects (64%), 12 of 13 (92%) of the CF nonproducers, and all of the CF producers expectorated adequate induced sputum samples (defined as containing at least 20% nonsquamous cells). Two of seven spontaneously expectorated samples from the CF producers were not of sufficient volume to perform all of the indicated biochemical measurements and quantitative bacterial cultures.

Induced sputum weights. The amount of induced sputum collected from CF nonproducers (4.8 ± 1.4 g) was similar to that collected from control subjects (4.4 ± 0.8 g, p = 0.49) and from CF producers (8.3 ± 2.9 g, p = 0.08) (see online data supplement, Figure E1).

Sputum and saliva cell counts and differentials. The total cell counts, absolute neutrophil counts, and percent neutrophils were significantly higher in each of the CF groups as compared with the control group (Figure 2; and online data supplement, Table E2). However, the mean percentage of squamous epithelial cells in sputum from healthy subjects was significantly higher than in sputum from both CF groups, possibly contributing to the difference in neutrophil percentages between the groups. Therefore, we calculated "corrected" sputum cell counts and differentials, by subtracting squamous cell numbers from the entire sputum sample. The corrected total cell counts, neutrophil counts, and percent neutrophils remained significantly increased in subjects with CF as compared with healthy subjects.

View larger version (17K): [in this window] [in a new window]   Figure 2.   Total cell counts (A) and neutrophil counts (B) in sputum from healthy control subjects and children with CF. Bars indicate mean values for each group. Open circles: sputum culture negative; closed circles: sputum culture positive. The total cell counts and neutrophil counts were significantly higher in the CF groups (p values as shown).

In salivary samples, the mean total cell counts were significantly higher in the CF producers versus healthy subjects (CF producers: 14.3 ± 3.3 × 10 ⁵ cells/ml; control subjects: 5.5 ± 0.9 × 10⁵ cells/ml, p = 0.03) while they were not significantly different in the CF nonproducers (8.6 ± 3.3 × 10⁵ cells/ml, p = 0.61). Saliva from each of the groups demonstrated an abundance of squamous epithelial cells, in terms of percentages (CF nonproducers, 84.1 ± 2.9; CF producers, 90.2 ± 3.0; control subjects, 91.8 ± 3.3). Importantly, the mean total cell counts of the induced sputum from each of the groups were at least three times higher than the counts in their matched saliva samples.

Biochemical indices of inflammation in sputum and saliva. The degree of inflammation in the CF airway was demonstrated strikingly by the elevated levels of IL-8 and neutrophil elastase activity detected in induced sputum samples from the patients with CF (Figure 3). Also, there were strong positive correlations between induced sputum IL-8 concentrations and neutrophil counts (r = 0.80, p < 0.001) and between elastase activity and neutrophil counts (r = 0.84, p < 0.001) when all 20 patients with CF were compared.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Concentrations of IL-8 (A) and neutrophil elastase activity (B) in sputum from healthy control subjects and both CF groups. Bars indicate median values for each group. Dotted lines represent limits of detections of the assays. Open circles: sputum culture negative; closed circles: sputum culture positive. IL-8 and elastase activity were significantly higher in the CF groups (p values as shown).

In saliva, mean IL-8 concentrations were significantly higher in the CF groups as compared with the control subjects (online data supplement, Figure E2). The IL-8 levels in induced sputum were at least 15-fold higher than in the matched saliva samples from both CF groups (CF nonproducers, p < 0.001; CF producers, p = 0.005) but not in the paired sputum and saliva samples from healthy subjects (p = 0.56). Neutrophil elastase, which was undetectable in saliva in all of the control subjects and CF nonproducers, was present at low levels in only two CF producers.

Indices of normalization. We also analyzed substances that have been proposed as potential normalizing factors. The mean sputum concentrations of urea (CF nonproducers: 7.8 ± 0.8 mg/dl, p = 0.67; CF producers: 9.2 ± 0.8, p = 0.67; control subjects: 10.2 ± 3.6) and albumin (CF nonproducers: 257.1 ± 49.8 µg/ml, p = 0.10; CF producers: 244.4 ± 69.0, p = 0.10; control subjects: 127.4 ± 52.5) were not significantly different between the CF and control groups, but the mean concentration of total protein (CF nonproducers: 390.1 ± 58.6 mg/dl, p = 0.03; CF producers: 466.3 ± 133.2, p = 0.04; control subjects: 214.7 ± 55.5) was significantly higher in the CF groups. Neutrophil counts remained elevated in the CF groups as compared with control children when normalized to all three of these substances, whereas total cell counts remained elevated in the CF groups when normalized to urea. IL-8 remained elevated in the CF groups as compared with control children when normalized to urea, total protein, or albumin (p < 0.001 in each case). Also, neutrophil elastase was higher in the CF groups when normalized to all three of these substances (p < 0.03 in each case).

In addition, the salivary levels of total protein were significantly higher in the subjects with CF as compared with the control subjects (Figure E2), whereas urea and albumin levels were not significantly different. The total protein levels in induced sputum were twice the levels measured in the salivary samples in each of the groups.

Microbiology. Figure 4 demonstrates that quantitative bacterial counts in induced sputum from CF nonproducers were significantly higher than in the healthy subjects. CF pathogens were isolated from induced sputum specimens in 3 of 7 healthy control subjects (two S. aureus, one H. influenzae), in 7 of 12 CF nonproducers (five S. aureus, four H. influenzae, one P. fluorescens, multiple organisms in three cases), and 5 of 7 CF producers (three S. aureus, one H. influenzae, three P. aeruginosa, multiple organisms in two cases). Concordance of bacteriologic findings between oropharyngeal swabs and induced sputa is presented in Table 1.

View larger version (17K): [in this window] [in a new window]   Figure 4.   Quantitative bacterial colony counts in sputum from healthy control subjects and both CF groups. These counts are inclusive of the following CF bacterial pathogens: Haemophilus influenzae, Staphylococcus aureus, Pseudomonas aeruginosa, Pseudomonas fluorescens. Bars indicate median values for each group. Bacterial counts were significantly higher in the CF groups (p values as shown).

Comparisons of induced and spontaneous sputum. From the CF producers, the mean weight of induced sputum (8.3 ± 2.9 g) was over 15-fold higher than the amount spontaneously expectorated (0.5 ± 0.1 g, p < 0.001). Induced sputum tended to have less squamous cell contamination than did spontaneous sputum. The total cell counts and IL-8 levels in the induced sputum samples from CF producers were significantly correlated with the values in the spontaneously expectorated specimens (Figure 5). The absolute neutrophil counts (r = 0.88, p = 0.02) and neutrophil elastase activity (r = 0.94, p = 0.006) in induced sputum were also comparable to values in spontaneously expectorated samples. Finally, the same bacterial pathogens, in similar colony counts, were isolated from both the induced and expectorated sputum samples from the CF producers.

View larger version (24K): [in this window] [in a new window]   Figure 5.   Correlations between total cell counts (A, B) and IL-8 (C, D) in induced sputum (IS) and matched expectorated sputum (ES) specimens from CF producers. In A and C, each subject's induced sputum value is plotted against the respective expectorated sputum value. The correlation coefficient and significance are indicated. In B and D, the difference between induced and spontaneous sputum cell counts and IL-8 concentrations is plotted against the mean of the two values. SD is standard deviation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As we have shown in a retrospective evaluation of sputum production in our population of patients with CF, young children with CF often cannot produce sputum. This highlights the need for a noninvasive technique to sample the airway. Such a technique must confirm previously demonstrated abnormalities between children with CF and healthy children. In the present study, we found that total cell counts, neutrophil counts, IL-8 concentrations, and neutrophil elastase activity were higher in the induced sputum from patients with CF than in that from healthy subjects. These measures of inflammation were comparable between both the induced and expectorated sputum samples. The nonexpectorating children with CF also demonstrated increased infection compared with healthy subjects, as measured by quantitative bacterial counts in induced sputum. Furthermore, the induced sputum from the CF producers was remarkably similar to expectorated sputum in both detection and quantification of CF-related bacterial pathogens. The only major difference between the induced and spontaneous sputa was the weight of the specimens collected.

Importantly, the cell counts and levels of IL-8 and elastase we measured in our induced sputum samples are similar to those previously reported in other studies of patients with CF that used either expectorated sputum (23, 28) or bronchoalveolar lavage fluid (2, 33). In fact, the IL-8 levels in our clinically stable patients with CF were comparable to or slightly higher than levels in acutely ill patients with CF in these published studies. We found that the salivary IL-8 levels were one-fifteenth the levels in the induced sputum specimens and were, therefore, unlikely to significantly affect the sputum levels. Furthermore, the higher levels of sputum total protein we found have been reported in studies of subjects with CF and other airway diseases (36, 37). Additionally, albumin has been reported to be elevated in sputum from subjects with CF and other airways diseases (36, 37). In this study, although the sputum albumin levels in the CF groups were almost twice that of the healthy subjects, the lack of statistical significance was most likely due to the small number of patients studied. Our findings of elevated total protein in saliva of patients with CF compared with healthy subjects, and of similar albumin levels in both groups, are consistent with previous salivary studies (38, 39). However, Mandel and coworkers (38) reported increased salivary urea in their patients with CF.

Despite the significant degree of inflammation found in the nonexpectorating children with CF, some of them lacked evidence of corresponding infection in their sputum cultures. Similarly, BAL studies in children with CF with stable, mild lung disease revealed significant evidence of inflammation, as demonstrated by elevated neutrophil counts and increased cytokine and free neutrophil elastase activity, either in the presence (2) or absence (5) of infection. The finding of inflammation without infection may be explained in several ways. One possibility is that a robust inflammatory response has removed the bacteria from the airways. Another explanation is that once lung disease is established, the chronic inflammatory response persists while infection occurs intermittently. S. aureus and H. influenzae were recovered more frequently in the CF nonproducers, whereas P. aeruginosa was isolated more frequently in the CF producers.

Sputum induction has a number of potential advantages compared with the current techniques for sampling the CF airway. This noninvasive method can provide quantitative measures of both airway inflammation and infection. It provides the ability to perform repeated samplings in order to monitor airway inflammation, infection, and efficacy of different treatment strategies over time. As compared with oropharyngeal cultures, inducing sputum might be a better way to identify true intrapulmonary pathogens, especially in children who cannot produce sputum. In comparison to expectorated sputum, the larger sample volumes of induced sputum may allow for more extensive measurements of inflammatory markers. Our Bland-Altman plots suggest that there is little bias in the measurement of inflammatory markers such as cell counts and IL-8 in induced sputum when compared with expectorated sputum. In addition, the degree of agreement between induced and expectorated sputum indicates that induced sputum measurements may be useful as outcome measures in clinical trials.

Although sputum induction has several potential applications in clinical practice and research, it does have a few drawbacks. Safety remains an important concern. Our patients with CF experienced a small but significant decrease in FEV₁ during the induction procedure, similar to the children studied in the other systematic sputum induction study in CF (22). Pretreatment with 180 µg of albuterol does not prevent a fall in FEV₁ in all subjects with CF undergoing sputum induction, a finding corroborated in a safety study conducted in patients with asthma (25). As compared with patients with asthma, however, we observed that the decrease in FEV₁ in some of the subjects with CF may have been the result of mucus plugging rather than of bronchoconstriction. For this reason, in addition to premedicating all subjects with albuterol, pulmonary function (FEV₁ or peak flows) needs to be closely monitored throughout the procedure. Since many of our patients were nonexpectorating children with mild lung disease, we would be cautious in using HS in those patients with CF with moderate to severe lung disease. Time is another consideration. The total sputum induction time generally lasts for 45-60 min. Trained personnel are required, as a respiratory therapist generally performs the procedure and laboratory technologists are needed to process the samples.

Potentially important methodological differences in sputum induction and processing protocols exist. Although we used the same concentration of HS (3%) over a fixed time period, other groups have used increasing saline concentrations over varying time periods. Though the concentration of HS does not appear to influence cellular or biochemical markers recovered, there is evidence that different lung compartments are sampled at different time points during the procedure (40). In regard to sputum processing, we processed the entire sample, without selecting out the mucus plugs. We found that total cell counts, neutrophil counts, and levels of IL-8 and neutrophil elastase in saliva are much lower than in sputum, and, therefore, are unlikely to significantly confound analysis of sputum. Processing the entire sputum sample in this manner does not account for the varying dilution of induced sputum samples with saliva and saline either between or within subjects. However, similar dilution problems arise in the measurement of cells and proteins in BAL fluid. Furthermore, both the whole sample approach and the mucus plug technique are reproducible and can reliably provide differential cell counts and measurements of soluble inflammatory mediators (17, 41, 42). In addition, the optimal way to homogenize the sputum for analysis has yet to be determined, especially in regard to sputum from patients with CF. Dithiothreitol (DTT) is the most frequently used mucolytic agent for homogenizing sputum, especially in the studies evaluating airway inflammation in patients with asthma. In sputum from subjects with asthma, DTT treatment resulted in higher total cell counts as compared with phosphate-buffered saline (PBS), whereas proportions of inflammatory cells and levels of IL-8 were similar between the specimens processed with DTT and PBS (43). However, CF sputum is more purulent and viscous than sputum from patients with asthma. Interestingly, one study reported that total cell counts and neutrophil counts are significantly increased when using an enzyme mixture, which includes deoxyribonuclease, for dispersal of CF sputum compared with DTT (44). Finally, we processed our sputum in order to evaluate both markers of inflammation and measures of infection. All previous asthma studies have processed sputum only for indices of inflammation. To preserve cellular integrity while trying to maximize recovery of pathogens from the mucus during homogenization, we employed a shaking water bath and manual mixing of the sample, rather than vortexing.

In this study, we did not address a few points important to the validation of sputum induction. Reproducibility will need to be examined. Also, induced sputum should be obtained under different conditions, such as during and after pulmonary exacerbations, to see how measures of inflammation and infection change. As previously mentioned, induced sputum was compared with BAL in a group of adult subjects with CF and resulted in similar measures of inflammation and infection (24). A similar study is needed in children with CF who are undergoing bronchoscopy and BAL for clinical reasons. Finally, it is possible that HS has a stimulatory effect on airway epithelial cells and might cause a rapid influx of inflammatory cells, especially neutrophils, during the course of the induction procedure. In this study, we have shown that the majority of cellular and biochemical markers of inflammation measured were remarkably similar in the paired induced and expectorated sputum samples. This suggests that the inhalation of hypertonic saline does not significantly alter the cells or biochemical indices measured. Similar colony counts of CF-related bacterial pathogens were recovered from the induced and expectorated samples from the CF producers, suggesting that induced sputum may be useful for quantifying bacterial load in the CF airway.

Although it has been shown previously that total protein concentration is elevated in CF saliva when compared with control subjects (38), this is the first study to show that IL-8 is also elevated in CF saliva. There are several possibilities for this finding. IL-8 may be elevated in saliva because of the communication between infected and inflamed sinuses and the oropharynx. Another likelihood is that clearance of sputum through the day leaves residual IL-8 in the oropharynx. Further, chronic exposure to oropharyngeal pathogens may increase IL-8 production by salivary glands. A fourth possibility is that IL-8 production is constitutively upregulated in the CF salivary gland.

In summary, sputum induction appears to be a relatively safe, noninvasive means of obtaining airway secretions from subjects with CF, especially from those who do not normally produce sputum. Airway inflammation and infection are significantly increased in both nonexpectorating and expectorating children with CF as compared with healthy children. Also, induced sputum samples appear to be comparable to spontaneously expectorated samples in describing both inflammation and infection in the CF airway.