INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Research on endobronchial biopsies (pre- and post-therapy) has been widely used to assess the effects of provocation and antiasthma therapy on airway inflammation (1). Measured changes in numbers of inflammatory cells between biopsies (obtained after various time intervals) have generally been interpreted as pro- or antiinflammatory effects of an intervention. However, these changes may be due to other factors, including measurement error (attributable to the measurement instrument) and/or the natural variability of airway inflammation over time. In the interpretation and design of intervention biopsy studies, it is worthwhile to consider the changes in biopsy parameters that may occur owing to the variability of airway inflammation over time before ascribing observed changes to an intervention. In the setting of asthma research, it is particularly important to distinguish between changes that can be properly attributed to an intervention and those that may be due to natural variability, because biologic variability is the defining characteristic of bronchial asthma (5). Type 2 errors might readily occur in asthma intervention studies that employ unreliable biomarkers such as asthma outcome measures, because changes in a measure may be falsely explained by an effect of an intervention, rather than by biologic variability (6). We therefore wanted to assess the reproducibility of repeat measures of bronchial mucosal inflammatory cell numbers in subjects with stable asthma after short (2-wk) and long (8-wk) sample intervals.

Airway inflammation in clinically stable asthma has been shown to exhibit considerable biologic variability (7). Few data exist, however, on the effect that the duration of biopsy interval has on the reproducibility of repeat measures of inflammatory cell numbers. Such data are fundamental to the interpretation of biopsy studies, because the usefulness of employing airway inflammatory cell numbers as markers of disease activity depends, at least in part, on their reproducibility over time. In a previous study of the immunopathology of asthma we have demonstrated that, at a single time point, there is little variability in repeat measures of inflammatory cell numbers within the same tissue section (11). In addition, the variability in repeat measures of inflammatory cell numbers in different sections of the same biopsy was consistently less than 10% (11). Greater intrasubject variations occur between measures from biopsies from different areas of lung (e.g., proximal and distal airways) (11). Other workers, using similar methods, have also demonstrated that the variability of repeat measures, both within a single tissue section and within different sections of a single biopsy, is small (9, 10). However, the variability of repeat measures of inflammatory cell numbers in different biopsies consistently outweighs within-biopsy variability (10). To date, the effect of the length of sampling interval on the reproducibility of measures of airway inflammation is unknown. We therefore designed the current study of patients with stable asthma in order to compare the effects of short and long interval duration on the reproducibility (and by implication reliability) of repeat measures of airway inflammation in serial bronchial biopsies.

The intraclass correlation coefficient (ICC) is a dimensionless statistic that describes the reproducibility of repeat measures in the same population (12). The value of ICC is bounded by 0 and 1. An ICC value of 1 for repeat measures indicates perfect reproducibility, whereas a value of 0 is interpreted as reproducibility that is no better or worse than expected by chance. In a stable population, repeat measures with an ICC value in excess of 0.6 are thought to be clinically useful. Repeat measures with an ICC value of less than 0.6 are probably not. In this article the ICC is employed to quantify between-biopsy reproducibility of repeat measures in the same subjects after short and long sample intervals. By quantifying the reproducibility of physiologic and inflammatory cell measures over time, this study is a unique description of the variability of airway inflammation in stable asthma.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Nine subjects with atopic asthma subjects (American Thoracic Society criteria [5] for asthma) were recruited (age, 18-40 yr; seven males) (see Table 1). None had suffered an exacerbation of asthma or taken corticosteroid therapy during the previous 3 mo. Baseline FEV₁ for all subjects was greater than 40% predicted and all subjects showed at least a 250-ml (and 12.5%) increase in FEV₁ after inhaled salbutamol. All subjects demonstrated bronchial hyperreactivity, evidenced by a histamine PC₂₀ (provocative concentraton of an agonist causing a 20% fall in FEV₁) of < 8 mg/ml, within 4 wk of entry (13). Subjects were excluded if they were pregnant or lactating, had a psychiatric illness, or other concurrent clinical condition. Supplementary oral or inhaled corticosteroids, theophyllines, cromolyns, and leukotriene antagonists were not allowed. Rescue inhalers were confined to salbutamol by metered dose inhaler. The Ethics Committee of James Connolly Memorial Hospital (Dublin, Ireland) approved the study protocol. Full written informed consent was obtained from each subject. The nature of the study demanded that great care be taken to provide an easily accessible and prompt medical backup for the subjects enrolled. A team of senior clinicians was on call on a 24-h/7-d basis to provide continuous medical backup. Investigators were available at all times for unscheduled visits to hospital and to ensure early withdrawals from the study in the event of deteriorating asthma or intercurrent illness.

Study Design

Each subject underwent fiberoptic bronchoscopy and endobronchial biopsy on three occasions: at baseline, after 2 wk, and after 8 wk (14). During a 1-wk run-in period and throughout the study period, subjects used a placebo inhaler (four puffs twice daily by metered dose inhaler attached to an aerosol chamber [Volumatic spacer; Allen & Hanburys, Greenford, UK]) and recorded peak expiratory flow rates (PEFRs) (morning and evening) using a mini-Wright peak flow meter. Subjects were instructed to contact the principal investigator (and were withdrawn) if PEFRs declined by more than 20% compared with the baseline mean PEFR determined during week 1.

Bronchodilator medication and caffeine-containing drinks were withheld for at least 12 h before each study visit and subjects always attended the laboratory at 9:00 A.M. Physical examination and routine blood and urine tests were performed before each bronchoscopy. Bronchodilator responses were always studied 4 h before bronchoscopy: FEV₁ was measured before and 20 min after salbutamol (200 µg) was administered via a Volumatic spacer device. The bronchodilator response was expressed as the percentage change in FEV₁ (postbronchodilator) from baseline.

The use of bronchodilator medication and adverse events were checked and recorded at each clinic visit. FVC and FEV₁, bronchodilator response, FEV₁/FVC, and FEF_25-75% were recorded using a Gould 2400 computerized spirometer. One day before to each bronchoscopy, histamine bronchial provocation challenges were performed according to a standardized technique (13). A complete dose- response curve for inhaled histamine was recorded, using doubling concentrations starting with 0.03 mg/ml and ending with 16 mg/ml of histamine dissolved in normal saline. During histamine challenges FEV₁ was measured 30 and 90 s after each dose. The test was terminated if FEV₁ decreased by more than 20% from the postsaline value or if a concentration of 16 mg/ml was reached. Histamine PC₂₀ was determined by linear interpolation of the concentration FEV₁ response curve. Subjects were not allowed salbutamol for 12 h before bronchial challenge. Allergy tests were performed at the screening visit. Subjects were considered atopic if they gave a history of bronchospasm after allergen exposure and exhibited a positive skin prick test (> 3-mm-diameter wheal reaction) to at least one of a panel of common allergens (house dust mite, grass pollen, cat, dog, and birch pollen).

Fiberoptic Bronchoscopy

Each subject underwent flexible fiberoptic bronchoscopy (Olympus, Norwood, MA) on three occasions (Days 0, 14, and 56) (14). Topical anesthesia was obtained by administration of 1% lignocaine to the vocal cords. Subjects were sedated with intravenous propofol (15) and continuous oxygen was administered by nasal cannulae. Oxygen saturation and electrocardiogram (ECG) were monitored throughout the procedure. Up to three endobronchial biopsies were taken through the bronchoscope with spiked cup forceps, always from the second-generation right upper lobe bronchus. Biopsies were immediately placed on sterile phosphate-buffered saline (PBS)-moistened gauze, embedded in O.C.T. medium (Miles, Elkhart, IN) and snap frozen in cooled (in a liquid nitrogen bath) isopentane. The samples were stored in liquid nitrogen until further analysis. Cryostat sections (6 µm thick) of the endobronchial biopsies were cut at 25° C, air dried for 1 h, then fixed in chloroform-acetone (1:1, vol/vol), wrapped in cling film, and stored at 20° C until use. Representative sections of all samples were stained with 0.1% toluidine blue to reveal appearance and tissue integrity. Immunohistological techniques were employed to identify cell types. The indirect immunoperoxidase technique (1, 2, 11) was used to identify T cells (using a cocktail of monoclonal antibodies (MAbs) to CD2, CD3, CD7, and CD8), memory T cells (MAb to CD45Ro), macrophage-monocytes (MAb to CD68), total eosinophils (MAb to EG1), and "activated" eosinophils (MAb to EG2).

Quantification of Immunohistology

All bronchial biopsies were stored in liquid nitrogen and subsequently processed in random time order, in order to avoid any measurement error being attributed to a technical or learning effect. An investigator with no knowledge of the identity, therapy, or timing of biopsies (L.W.P.) quantified the numbers of cells. The presence, distribution, and number of each cell type were assessed using a "Solitaire" (Seescan, Cambridge, UK) computerized image analysis system. Three areas of epithelium and subepithelial connective tissue to a depth of 10 to 12 cells were measured in each section. Total areas of 12 × 10⁴ µm were quantified on duplicate sections for each cell type in samples for each biopsy. Areas of each high-power field measured were determined with the image analyzer by drawing frames around the area to be quantified. These frames were designed to avoid damaged areas and areas of muscle. Numbers of positive cells within framed areas were point counted and all results were then reduced to cells per unit area by dividing the number of cells documented in each field by the area of section in square micrometers calculated by the computer, as previously described (11). Average numbers of cells per unit area (for three areas) were calculated for each cell type at each time point.

Statistics

To quantify the repeatability of measures on the same group of subjects, the intraclass coefficient (ICC) was employed (12). The ICC represents an estimate of the average correlation between all possible ordering of pairs of measures. For each sample interval (two sets of data; e.g., data at 2 wk compared with baseline) the total sum of squares (SS_T) and the sum of squares between subjects (SS_B) were derived with a one-way analysis of variance table. The ICC was calculated as follows: ICC = [(2 × SS_B)  SS_T]/SS_T. ICCs were calculated for both immunopathologic and physiologic parameters. To determine if any significant within-group changes occurred during the course of the study, the significance of within-group changes (after 2 and 8 wk) was estimated using two-way ANOVA.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The subjects in this study formed part of a large biopsy controlled trial of inhaled corticosteroids in asthma (1). Nine subjects with stable atopic asthma were randomized to receive placebo. All subjects completed the protocol, each with three endobronchial biopsies. There were no dropouts. The results of computerized spirometry, histamine provocation challenges, and bronchial mucosal cell numbers are contained in Table 2. These data confirm airway physiology consistent with mild asthma. Taken as a group, no significant change in airway physiology or inflammation was observed after 2 or 8 wk of therapy with inhaled placebo (two-way ANOVA).

The intraclass correlation coefficients for repeat physiologic measures are presented in Figure 1. Postbronchodilator FEV₁ was the most reproducible index of spirometry: the repeat measures ICC was 0.95 at 2 wk, compared with a 2-wk ICC of 0.89 for FEF_25-75%, and ICC of 0.7 for bronchodilator response. After 8 wk the repeat measures ICCs for these indices were 0.75, 0.68, and 0.56, respectively. Repeat measures of PC₂₀ were highly reproducible at both 2-wk (ICC equal to 0.71) and 8-wk (ICC equal to 0.88) intervals.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Intraclass coefficients (ICCs) of reproducibility for repeat measures of FEV1 (solid squares), FEF25-75% (crosses), bronchodilator response (open squares), and PC20 (solid triangles) after 2- and 8-wk intervals. Parameters with an ICC greater than 0.6 are generally considered clinically useful.

Intraclass correlation coefficients for repeat measures of immunologic indices are presented in Figure 2. Repeat measures of T cell (ICC equal to 0.81), CD45Ro⁺ cell (ICC equal to 0.70), and EG2⁺ cell (ICC equal to 0.67) numbers appeared highly reproducible after a 2-wk sample interval. After 8 wk, the ICCs for repeat measures of these parameters were equal to 0.41, 0.51, and 0.73, respectively. Only repeat measures of EG2⁺ cell numbers remained highly reproducible (ICC equal to 0.73) after an 8-wk interval. Repeat measures of CD68⁺ and EG1⁺ cells were less reproducible after both a 2-wk interval (ICC equal to 0.23 and 0.40, respectively) and after an 8-wk interval (ICC equal to 0.07 and 0.26, respectively) (Figure 2).

View larger version (14K): [in this window] [in a new window]   Figure 2.   Intraclass coefficients (ICCs) of reproducibility for repeat measures of T cells (solid diamonds), CD45Ro+ cells (solid squares), EG2+ cells (crosses), EG1+ cells (open squares), and CD68+ cells (solid triangles) after 2- and 8-wk intervals. Parameters with an ICC greater than 0.6 are generally considered clinically useful.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Intervention biopsy studies of the effects of antiasthma therapy on airway inflammation have provided valuable information about disease mechanisms in asthma (1, 3). The design of biopsy studies in asthma is an area of considerable research because antiinflammatory agents and modes of their administration are currently being developed and investigated by such methodology (1, 16). However, the design of biopsy studies in subjects with asthma is fraught with logistic, technical, and ethical problems (17). The situation is further complicated by uncertainty regarding the optimal time interval between biopsies. Until now, no clear data have existed on the optimal duration of bronchial biopsy studies and time intervals between biopsies are chosen on an arbitrary basis (sampling intervals of 2, 4, 6, 8, and 12 wk have been employed). The current study demonstrates that the reproducibility of repeat measures of airway inflammation in stable asthma (on inhaled -agonist and inhaled placebo therapy only) is significantly greater after a 2-wk interval rather than after a 2-mo interval. Repeat measures of most inflammatory cells are less reproducible and are probably clinically unreliable after an 8-wk biopsy interval. The current data suggest that, in studies of asthmatic airway inflammation, shorter intervals (of 2 wk or less) between sampling may minimize type 2 errors due to the variability of inflammatory cell counts. In addition, this study has determined that repeat measures of some parameters, such as T cells, memory T cells (CD45Ro⁺) and EG2⁺ cells are highly reproducible (after a 2-wk interval). These parameters should provide the most reliable information about the severity of airway inflammation in stable asthma. Notably, repeat measures of airway EG2⁺ cells appear particularly reproducible (18) after short and long sampling intervals. This biopsy study demonstrates that repeat measures of most airway inflammatory cell numbers (the "gold standard" markers of airway inflammation) are unreliable after an 8-wk sampling interval. The implication is that surrogate markers of airway inflammation (serum and sputum markers) need to be carefully validated in terms of reliability, especially over long measurement intervals.

In contrast to previous studies of the variability of inflammatory parameters in asthma (9, 10), this study has examined endobronchial biopsies at three time points. We demonstrate that repeat measures of some airway inflammatory cell numbers are highly reproducible after 2 wk, and are likely to be reliable indices of airway inflammation over this period. More importantly, the data demonstrate, for the first time, that repeat measures of most parameters of airway inflammation become less reproducible with length of biopsy interval. The reproducibility of repeat measures of airway inflammatory cells after 2 wk compared with after 2 mo demonstrates that most repeat measures are less repeatable after longer biopsy intervals and therefore are probably not clinically useful (12). The implication is that intervention studies with longer biopsy intervals could lead to type 2 errors because observed differences in some inflammatory cell numbers might be falsely attributed to the effect of an intervention rather than to biologic variability. One can hypothesize that in order to estimate the correct interval between biopsies, there would appear to be a trade-off between longer biopsy intervals, which allow greater time for the biologic effect of an intervention, and shorter sampling intervals, which help to minimize the effect of biologic variability. Such factors should be carefully considered when designing or assessing studies that use biopsies as outcome measures.

This study analyzes immunopathologic data obtained from serial biopsies of subjects with clinically stable asthma and who received inhaled -agonist and inhaled placebo therapy. The data demonstrate that repeat measures of some lymphocyte (T cells and CD45Ro⁺ [memory] T cells) and eosinophil (EG2⁺) subsets are highly reproducible and may provide the most useful, reliable information when the effects of therapy, allergen, or infection on the inflammatory cell profile of the asthmatic airway are studied. Repeat measures of other inflammatory cells (CD68⁺ cells and EG1⁺ cells) were less reproducible in the setting of clinically stable asthma and are probably unreliable indices. Given the variability of repeat measures of airway inflammation in this study of subjects with mild asthma, one can only speculate that the variability of inflammation over time is far greater in severe or clinically unstable asthma.

Another important issue in the interpretation of bronchial biopsy studies is the potential effect of tissue injury, occurring during the act of fiberoptic bronchoscopy and endobronchial biopsy, on future measures of airway inflammation. In theory, the healing process that occurs after biopsy might significantly alter measures of inflammatory cells in the bronchial mucosa at a subsequent biopsy. These effects might be important if the same general anatomic site is being biopsied on more than one occasion. The results of this study are reassuring in this regard. Our results demonstrate that repeat measures of inflammatory cell numbers are more reproducible after a 2-wk interval than after an 8-wk interval. This suggests that the variability of airway inflammation over time probably outweighs the acute effects of bronchoscopic biopsy. We cannot exclude the possibility that the act of bronchoscopic biopsy might lead to acute changes in CD68⁺ and EG1⁺ cells, thus accounting for their poor reproducibility on repeat biopsy 2 wk later. We do not know how much time is needed to recover from bronchoscopic biopsy, but the current study suggests that recovery from endobronchial biopsy almost certainly occurs within 2 wk.

Since the introduction of fiberoptic bronchoscopy, there has been considerable interest in the use of repeat bronchial biopsy studies to assess disease mechanisms in asthma. However, in contrast to the wealth of data that supports the use of highly reproducible physiologic measures (e.g., FEV₁, PC₂₀) as reliable indices of airflow limitation and disease severity (20, 21), few data exist on the reliability of repeat measures of airway inflammation (17). This article describes the reproducibility of repeat measures of airway inflammatory cell numbers in a group of patients with stable asthma and who are not receiving antiinflammatory therapy (inhaled placebo and -agonist therapy only). The lack of significant (within-group) changes in measures of bronchodilator responses, and spirometry confirms that these subjects had stable asthma throughout the study (1). Measures of peak flows and peak flow variability were stable throughout and no significant clinical exacerbation was observed or treated during the course of the study. Our results are consistent with others who have demonstrated that repeat measures of some physiologic indices (e.g., FEV₁) are highly reproducible (20). Oldham and Cole, for instance, have demonstrated an ICC of 0.99 for repeat FEV₁ measures taken on the same day and an ICC of 0.88 for repeat measures after a 2-wk interval (22). The data in this study are in agreement with conventional wisdom: repeat measures of FEV₁ and PC₂₀ are highly reproducible and the current data support their use as reliable parameters of asthmatic physiology (20, 21).

To assess the reproducibility of repeat measures of airway inflammation in asthma (and, in contrast to previous studies, of the variability of airway inflammation in asthma [9, 10]), we examined serial bronchial biopsies from subjects with mild/ moderate asthma who received no antiinflammatory therapy (i.e., those who took placebo inhalers) for the duration of the study. This study was designed for two reasons to examine multiple bronchial biopsies obtained from patients with asthma and not taking inhaled steroids. First, it is now well established that corticosteroids have profound effects on airway inflammation in asthma (19). Steroid therapy has previously been shown to have significant effects on airway physiology and numbers of inflammatory cells in the bronchial mucosa in asthma (1, 16, 19); therefore changes in repeat measures of these parameters in subjects receiving corticosteroid therapy could be attributable to the effects of therapy. Second, corticosteroid therapy is likely not only to downregulate airway inflammation but also to dampen its biologic variability (just as it appears to dampen the variability of airway physiology). For both these reasons care was taken to ensure that only patients with clinically stable asthma and not receiving corticosteroid therapy (inhaled placebo therapy only) were examined. Evidence that the participants had clinically stable asthma is revealed by the lack of spirometric or peak flow deterioration during the study. Moreover, there was no significant (within-group) change in airway physiology or immunopathology after 2 or 8 wk, as assessed by two-way ANOVA (Table 2). Because the subjects in this study did not receive antiinflammatory therapy (which might downregulate inflammatory cell numbers) the current data provide a unique description of the natural variability of airway inflammation in clinically stable asthma.