INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Severe, refractory asthmatics, although only a small percentage of the asthma population, utilize a highly disproportionate amount of health care time and dollars (1). They often respond poorly to standard care for asthma, including inhaled and systemic corticosteroids. Several theories have been proposed to explain this clinically severe state, but very little is known regarding the pathology of the process.

One of the first pathologic reports investigating severe, living asthmatics suggested that these refractory patients may demonstrate a different process than that seen in asthmatics with milder disease, not treated with inhaled corticosteroids (ICS). Neutrophils were present in increased quantities in these refractory patients compared with patients with milder disease (2). Because neutrophils are relatively steroid-resistant, it was hypothesized that this increased neutrophilic inflammation might explain the poor response to corticosteroids (CS) (2, 3). Eosinophils were present, but in numbers generally comparable to milder asthmatics. However, considerable heterogeneity was noted in the ranges for both the eosinophil and neutrophil populations. Other explanations for the poor response to CS in these patients have focused on inflammation-induced changes in the binding affinity of the glucocorticoid receptor and alterations in CS suppression of transcription factor binding (4, 5). In contrast to the theory that neutrophils may be important in the pathophysiology of the disease, these changes were thought to be associated with an undertreated, predominantly eosinophil/lymphocyte driven process. Finally, it has been suggested that in severe patients, the disease may be refractory to therapy because the airways are considerably more fibrosed or "remodeled" than in their milder counterparts, primarily on the basis of the prolonged (perhaps "undertreated") course of their disease (6, 7). However, a recent report in very severe asthmatics demonstrated that although the subbasement membrane (SBM) was thicker in asthmatics, it did not correlate with severity of disease (8). Once again, considerable variability existed.

In addition to these pathologic inconsistencies, there are clinical differences in severe asthmatics. Although they all meet the American College of Chest Physicians-American Thoracic Society (ACCP-ATS) definition of asthma, some severe asthmatics have wide variability in airflow over time, whereas others are more chronically severely obstructed (9, 10). Similarly, some may have repeated bouts of respiratory failure, whereas others do not. Finally, some may have a rapidly progressive downhill course, whereas others are severe but stable for many years. Whether these differences are caused by a broad spectrum of a single disease, or by "different" diseases presenting with a single physiologic phenotype is not known.

Therefore, it was hypothesized that at least two different pathologic subtypes of severe asthma would be present in airway tissue: a "classic" eosinophilic process, and a "pathologically unclassic" process, perhaps associated more strongly with neutrophils. It was further hypothesized that the particular type of inflammation present would determine airway structural abnormalities, such as SBM thickening, which together could differentiate the groups both clinically and physiologically.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Severe refractory asthmatics were all patients referred to National Jewish Medical and Research Center for evaluation and treatment of their disease. Asthmatic and normal control subjects were recruited from the established asthma database at National Jewish and from local advertising. For purposes of this study, severe asthma was defined as: (1) FEV₁ of < 70% predicted (adults), < 80% predicted (children) on more than one occasion over the year prior to evaluation; (2) 10 mg/d of prednisone or equivalent for at least 75% of the previous year, while taking standard asthma medications (ICSs, long acting -agonists, theophylline, leukotriene modulators, and short-acting -agonists); and (3) > 2 urgent care visits for asthma in the previous year. Moderate asthma was defined as an FEV₁ of < 80% predicted on 400 to 1,000 µg of ICSs and -agonists, without a history of urgent care visits or systemic steroid use. Mild asthma was defined as an FEV₁ of > 80% predicted, on -agonists alone. Normal control subjects had normal pulmonary function, no bronchodilator response, and no history of any respiratory illness. However, a history of mild allergies (upper airway only) and atopy was allowed.

Study Design

All subjects underwent a routine history and physical examination. All subjects had spirometry testing, pre- and postbronchodilator. All asthmatics filled out an extensive questionnaire to historically evaluate their disease, including duration of disease and numbers of intubations. Additionally, severe asthmatics (only) had an evaluation of lung volumes, airway conductance, and slow vital capacity (SVC) measurements performed in the pulmonary physiology unit of National Jewish. All subjects then underwent bronchoscopic evaluation, including endobronchial biopsy and bronchoalveolar lavage (BAL) to evaluate the degree and type of inflammation in their airways. The study was approved by the National Jewish institutional review board and all subjects gave informed consent. This research was conducted according to the principles of the Declaration of Helsinki.

Bronchoscopy with Endobronchial Biopsy and Lavage

Bronchoscopies were performed as previously described (2). Atropine (0.6 mg), codeine (30 to 60 mg) intramuscularly and nebulized albuterol were given prior to study. Lidocaine (< 400 mg total) was used for local anesthesia in the upper and lower airways. The subjects received oxygen and vital signs were monitored throughout the procedure. Spirometry was performed pre- and postprocedure. The bronchoscope was passed nasally and six endobronchial biopsies were taken from the first or second subcarinae of the right or left lower lobes. The bronchoscope was then repositioned in the opposite lung where BAL was performed in subsegments of the lingula or right middle lobe using four 60-ml aliquots of warmed sterile saline, with sequential instillation and manual aspiration.

Lavage Processing

BAL was processed as previously described (2). The fluid was immediately placed on ice and centrifuged to separate fluid from cells. The fluid was aliquoted and frozen at 70° C for histamine and tryptase analysis.

Tissue Processing

Endobronchial tissue was fixed overnight at 20° C in acetone and embedded in glycol methacrylate resin. Tissue blocks were stored at 20° C until 2-µm sections were cut using a Reichert Ultracut E ultramicrotome (Leica Inc., Deerfield, IL). Tissue sections were stained with antibodies against cell markers: eosinophils (EG2; Pharmacia, Piscataway, NJ), neutrophils (neutrophil elastase; DAKO, Carpinteria, CA), lymphocytes (CD3+, CD4+ [both from Beckton-Dickinson, Bedford, MA] and CD8+ [DAKO]), mast cells-AA1, macrophages- CD68 (both from DAKO), and transforming growth factor- 1,2,3 (TGF-) (Genzyme, Cambridge, MA).

Sections were treated with 0.3% H₂O₂ in 0.05 M TRIS-buffered saline (TBS, pH 7.6) for 30 min to inhibit endogenous peroxidase, and incubated with 1% normal horse or goat serum for 30 min to block potential nonspecific binding sites. The slides were then incubated with the primary antibodies mentioned previously for 2 h at room temperature, followed by incubation with biotinylated horse anti-mouse IgG or goat anti-rabbit IgG for 1 h at room temperature. After rinsing the slides in TBS, 0.03% aminoethylcarbazole (AEC) in 0.03% H₂O₂ was used as substrate to develop a peroxide-dependent red color reaction. Slides were counterstained with Mayer's hematoxylin and covered with Crystalmount (Biomeda Corp., Foster City, CA). Appropriate control slides were similarly treated but with primary antibodies replaced by nonimmune serum or TBS. Positive cells were counted blindly in the submucosa of the biopsy slices and expressed as number of cells/mm².

SBM thickness (in µm) was measured as previously described using antibodies against collagen type I (Biodesign International, Kennebunk, ME) (8). Only collagen I staining is reported as a previous study reported excellent correlations (r = 0.93 to 0.96), between antibodies directed against collagen I or III and Sirius red staining for total collagen (8). Briefly, a National Institutes of Health (NIH) scion image analysis program was used to measure the thickness of SBM. One to three good quality pieces of tissue were included for the analyses. Black-and-white pictures were taken and transferred to the computer via video camera. The stained area for collagen was black and high in intensity compared with the nonstained area (white). The multiple point-to-point measurement method was used at 50-µm intervals for a total of 35 to 60 measurements. The thickness of the SBM was vertically measured at ×200 magnification in areas of good epithelial orientation.

Tryptase and Histamine Assays

-Tryptase levels in BAL fluid were measured by a sandwich enzyme immunoassay (EIA) using the B12 monoclonal antibody for capture and biotin-G5 monoclonal antibody for detection of -tryptase (11, 12), except that BAL fluid was first concentrated 10-fold by lyophilization. The ionic strength of the detector monoclonal antibody: sample:capture monoclonal antibody incubation mixture was adjusted to 0.5 M NaCl as before and the sensitivity was 0.1 ng/ml. Total tryptase concentrations were also measured in concentrated BAL fluid using the B12 monoclonal antibody for capture and biotin-G4 and biotin-G3 monoclonal antibodies for detection as described (13). Although the assay for -tryptase was less sensitive than that for total tryptase, only the data with -tryptase was shown in the current study because this most directly reflects mast cell activation. However, -tryptase concentrations showed a linear relationship to total tryptase levels (r = 0.94), with those of -tryptase being 67% of total tryptase levels, suggesting that most of the tryptase detected in BAL fluid is in the active form (data not shown). Histamine was analyzed by a commercially available kit (Immunetech, Marseilles, France). The sensitivity of the histamine assay was 50 pg/ml.

Subgrouping by Eosinophils

As a subset of the severe asthmatics appeared to have low to moderate concentrations of eosinophils, as compared with normal control subjects, while a second group had nearly absent eosinophils, well within the range seen in the normal control subjects, the severe asthmatics were divided into those who were eosinophil (+) or eosinophil () (Figure 1). Eosinophil () subjects (n = 14) were defined as those falling within two standard deviations of the normal mean (1 ± 4 cells/ mm²) for eosinophil numbers. The remaining severe asthmatics were classified as the eosinophil (+) group (n = 20). This subgrouping was used for all the remaining analyses of other cell types, structural, clinical, and physiologic parameters.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Eosinophil (+) severe asthmatics had significantly higher numbers of EG2+ cells/mm2 tissue than eosinophil () severe asthmatics, but were not higher than milder asthmatics. Eosinophil () severe asthmatics were significantly less than mild asthmatics, and by definition, not significantly different from normal control subjects (overall p < 0.0001). All data expressed as median with interquartile range.

Statistical Analysis

The majority of data were not normally distributed. Therefore, data were anyalyzed using nonparametric analysis. The five groups were initially compared using the Kruskal-Wallis variation of the Wilcoxon signed-rank test for multiple comparisons to develop an "overall" level of statistical difference among the groups. When a significant difference was found among the five groups, a post hoc comparison of each individual group was performed using Dunn's template (14). Correlations were performed using Spearman's rho testing. For comparisons of the two severe asthma groups which required a yes/no response, chi-square analysis was used. A value of p < 0.05 was considered significant. All testing was done using a JMP program (SAS Institute, Cary, NC) for Macintosh computers, developed from a SAS-based system.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Characteristics

Thirty-four severe asthmatics underwent bronchoscopy with BAL and biopsy. Additionally, 11 moderate asthmatics, 10 mild asthmatics, and 11 normal control subjects were studied (Table 1). Because of limitations in amounts of tissue or fluid, not all studies were performed on all subjects (indicated with "n" specific for the given analysis). The combined group of severe asthmatics was on a median dose of prednisone of 30 mg/d (range 10 to 40 mg/d). The severe asthmatics were significantly more obstructed on the basis of FEV₁ and FEV₁/FVC than all the control groups (overall p = 0.001, p < 0.05 between the severe asthmatics and the asthmatic and normal control subjects). The median bronchodilator response was significantly greater in the severe asthmatics as compared with all the asthmatic and normal control subjects (overall p = 0.001, p < 0.05, for all comparisons with severe asthma). Thirteen of the 34 severe asthmatics had one or more episodes of respiratory failure requiring intubation. On the basis of clinical impression, steroid side effects, and A.M. cortisol levels, the severe subjects were deemed to be at least moderately adherent to their steroid regimens. There were no complications with any of the endobronchial biopsies or BAL. Specifically, there were no episodes of prolonged wheezing, coughing, or bleeding. All subjects were able to leave within 4 h of bronchoscopic evaluation.

Evidence for Differences in Cellular Inflammation between the Two Severe Groups

Even after after using the eosinophil numbers to divide the severe asthmatics into two groups, there were no significant differences in eosinophil numbers between the eosinophil (+) severe asthmatics and either of the asthmatic control groups (Figures 1, 2A, and 2C). Eosinophils in the eosinophil (+) severe asthmatics were, however, higher (p < 0.05) than the normal control subjects. The eosinophil () severe group was significantly lower than the mild asthmatics, and by definition, was not different from the normal control subjects. CD3+ (as well as CD4+ and CD8+) lymphocytes and macrophages (CD68) were significantly higher in eosinophil (+) severe asthmatics than in eosinophil () asthmatics (overall p = 0.001, p < 0.05 for differences among the two severe asthmatic groups) (Figure 3 and Table 2). However, there were no differences in lymphocyte or macrophage numbers among either severe asthmatic group compared with the milder asthmatics or normal control subjects. Neutrophils were significantly higher in eosinophil (+) severe asthmatics than in mild asthmatics and normal control subjects (p < 0.05) (overall p = 0.003), but there were no significant differences between the two severe asthmatic groups (Table 2, Figures 2B and 2D). Mast cell numbers were also significantly higher in eosinophil (+) severe asthmatics compared with eosinophil () severe asthmatics (overall p = 0.02, p < 0.05), but were not different from milder asthmatics or normal control subjects (Table 2). Mast cell activity, as assessed by the mast cell specific marker, -tryptase, was significantly increased in the BAL of eosinophil (+) severe asthmatics, compared with moderate asthmatics and normal control subjects (p < 0.05, overall p = 0.004). In contrast to the biopsy mast cell numbers, there was no difference in tryptase concentrations between the two severe asthmatic groups (Figure 4). Histamine concentrations tracked similar to tryptase, but were not significantly different among the groups (data not shown) (p = 0.07).

View larger version (118K): [in this window] [in a new window]   Figure 2.   EG2 (A) and neutrophil elastase (B) staining of tissue from an eosinophil (+) severe asthmatic. EG2 (C ) and neutrophil elastase (D) staining of tissue from an eosinophil () severe asthmatic.

View larger version (17K): [in this window] [in a new window]   Figure 3.   Eosinophil (+) severe asthmatics had significantly higher numbers of CD3+ lymphocytes/mm2 tissue than eosinophil () severe asthmatics. Although the overall p = 0.001, there were no other significant differences between the groups (overall p = 0.004). All data expressed as median with interquartile range.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Eosinophil (+) severe asthmatics had significantly higher levels of the mast cell specific mediator, -tryptase, in BAL fluid than the normal control subjects, or the moderate asthmatics, on ICS. There were no significant differences between the two severe asthmatic groups (overall p = 0.004). All data expressed as median with interquartile range.

Evaluating all groups, CD3+ cells correlated with EG2+ cells with a Spearman's rho of 0.59, p < 0.0001. CD68 cells correlated with EG2+ cells (r = 0.50, p < 0.001) and with neutrophils (r = 0.51, p < 0.001). Mast cells correlated with both EG2+ cells and CD3+ cells (r = 0.45 and 0.43, respectively), but did not correlate with either histamine or tryptase (r < 0.3 for each). -tryptase concentrations in BAL correlated with histamine, r = 0.66 (p < 0.0001).

Evidence for Inflammatory Group-Dependent Influences on Structural Elements (SBM)

The SBM was significantly thicker in the eosinophil (+) severe asthmatics compared with the eosinophil () severe asthmatics (Figure 5). Only the SBM in the eosinophil (+) severe asthmatics was thicker than in the normal control subjects (Figure 6). The SBM thickness was significantly correlated with EG2+ cells (r = 0.50, p < 0.0001). To determine the relationship of the SBM to the growth factor TGF-, cells positive for TGF- were analyzed in the submucosa from tissue biopsy samples. TGF- was significantly increased in eosinophil (+) severe asthmatics, compared with eosinophil () severe asthmatics and normal control subjects (Figure 7). TGF- was significantly correlated with EG2+ cells, neutrophils, and macrophages (r = 0.60, 0.48, 0.70, respectively, all p < 0.0001) but only marginally with SBM thickness (r = 0.30, p < 0.02).

View larger version (21K): [in this window] [in a new window]   Figure 5.   Eosinophil (+) severe asthmatics had a significantly thicker SBM than normal control subjects or eosinophil () severe asthmatics. There were no significant differences between the eosinophil (+) severe asthmatics and the milder asthmatic control subjects (overall p = 0.002). All data expressed as median with interquartile range.

View larger version (49K): [in this window] [in a new window]   Figure 6.   Collagen I antibody staining in tissue from an eosinophil (+) severe asthmatic (A), an eosinophil () severe asthmatic (B), and a normal control subject (C ).

View larger version (23K): [in this window] [in a new window]   Figure 7.   Eosinophil (+) severe asthmatics had significantly higher numbers of TGF-+ cells in the submucosa than normal control subjects or eosinophil () severe asthmatics. There were no significant differences between the eosinophil (+) severe asthmatics and the milder asthmatic groups (overall p = 0.0003). All data expressed as median with interquartile range.

Evidence for Inflammatory Group-Dependent Influences on Physiologic Parameters

There was a marginal difference between the two severe groups in FEV₁ (percent predicted) (p = 0.05), with the eosinophil () group having a slightly lower FEV₁ (Table 3). There was no difference in the percent bronchodilator response. There were also no differences between the groups in residual volume, with both groups demonstrating hyperinflation. In contrast, the FVC/SVC (as a potential indicator of airway collapse during forced expiration), was lower (indicating more collapse), in the eosinophil (+) group than the eosinophil () group (p = 0.03). There were no correlations between eosinophils or SBM thickness and FEV₁ or SVC/FVC. Neutrophil numbers correlated weakly with FEV₁ (r = 0.29, p = 0.019), whereas the percent bronchodilator response correlated weakly with SBM thickness (r = 0.33, p = 0.015).

Evidence for Inflammatory Group-dependent Influences on Clinical Parameters

There were proportionately more females in the eosinophil (+) group than in the eosinophil () group, but this was not significantly different (Table 4). There were no differences in age, duration of disease, or CS use between the two severe asthmatic groups. There were also no differences in peripheral eosinophils, IgE levels, or atopy (data not shown). However, 12 of 20 eosinophil (+) severe asthmatics had one or more intubations, whereas only 1 of 14 eosinophil () asthmatics had a history of a similar episode (chi-square p = 0.004).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Asthma is often described as a heterogenous process. However, little data exist to define or characterize the underlying subgroups from a pathologic (or clinical) perspective. This study is the first to describe two different and distinct pathologic processes, both of which present clinically and physiologically as "severe, refractory asthma." Additionally, these differences in inflammation are associated with structural, physiologic, and clinical differences between the two groups. These data should have clear implications for the treatment of not only severe disease, but also "asthma" in the more general sense.

Heterogeneity of asthma has been inferred from differing degrees of allergic involvement, age at onset, exacerbating factors, and response to treatment. Recent studies with leukotriene-modulating drugs have further highlighted the heterogeneity of the process, with reports suggesting that 50 to 60% of asthmatics treated with these drugs demonstrate a clinically significant response (15, 16). A review of reports showing individual pathologic data points in mild asthma demonstrates a wide range of variability in cell types (including eosinophils) and cytokine/mediator levels (17). However, previous attempts at pathologic differentiation of the most basic of subtypes (allergic versus nonallergic) have led to mixed results (22). It is likely that this heterogeneity is contributing to the lack of strong correlations between physiologic parameters and various inflammatory markers (21, 25).

The results presented here strongly support the concept of pathophysiologic heterogeneity of asthma, to the point of beginning to define two different pathologic phenotypes. The 34 severe asthmatics studied (in toto) represented a superficially homogenous group with severely obstructed physiology, preserved bronchodilator response, atopy, high medication needs (including CS), and ongoing frequent health care utilization. Despite this clinical similarity, two distinct pathologic patterns emerged from these patients: eosinophil (+) severe asthmatics and eosinophil () severe asthmatics. We believe that this differentiation represents the natural extension of our first reported study in severe asthmatics. In that study, there were higher numbers of neutrophils in the severe asthmatics than in the control groups, but the eosinophil numbers were widely variable and not significantly different from the mild asthmatics (2). We believe this differentiation of groups helps to explain some of the original heterogeneity.

The eosinophil (+) group continue to demonstrate high levels of "classic" eosinophilic asthmatic inflammation despite the high doses of CS. Moreover, they have associated evidence for increased numbers of CD3+ T cells, with concomitant increases in CD4+ and CD8+ T cells. Mast cells, both tissue numbers and activity, are also increased in this group, although some overlap may exist with the eosinophil () group. Interestingly, macrophages also appear to be increased in the eosinophil (+) asthmatics, but not the eosinophil () asthmatics. Although different from our original report, we believe these differences in macrophage numbers are explained on the basis of: (1) the use of immunohistochemical stains for macrophages which makes the analysis much easier and more reliable, and (2) the division of the severe asthmatics into the two groups (one with high and one with low levels of macrophages). In general, there appeared to be a greater total inflammatory cell infiltrate in the eosinophil (+) group compared with the eosinophil () group (data not shown).

Which specific cytokines are driving this process in these eosinophil (+) severe patients (e.g., interleukin [IL]-4, IL-5) is unknown, but currently under investigation. Determining the cytokines that remain after high-dose CS should give considerable insights into the mechanism behind the inflammatory process in this group. In contrast, the eosinophil () asthmatics had virtually no evidence for "classic" asthmatic inflammation, despite the remarkably similar clinical and physiologic picture. It is currently impossible to determine whether this group is a completely distinct pathologic entity or whether this group began as an eosinophil-associated disease, only to have the cortiocosteroids eliminate the eosinophils from this group. Whatever the explanation is, the tissue level response appears to be different in the two groups. Although there are very few distinct positive features to this subgroup [generally, they are characterized by the "absence" of features seen in the eosinophil (+) group], there may be some processes that help to explain the clinical and physiologic abnormalities of this group. Despite the lack of eosinophils, there are no differences between the two severe groups in the number of neutrophils or activity of mast cells. (Interestingly, perhaps owing to the lower total number of cells, the percentage of neutrophils, as opposed to the absolute number/mm², was significantly higher in the eosinophil () group compared with the eosinophil (+) group.) However, as the relevance of this is unclear, it is likely that other, as yet poorly defined features are also contributing to the severity of this group.

As noted previously, the two severe groups shared very few inflammatory features. However, they both had persistent elevation in neutrophils despite (or because of) the high doses of CS. The role of the neutrophil in severe asthma remains unclear, but there has been accumulating evidence for its presence, if not its direct involvement, in situations associated with severe asthma. Neutrophils have been shown to be elevated in nocturnal asthmatics, in sputum from exacerbating asthmatics, in bronchial wash from patients intubated for status asthmaticus, and in autopsy reports from asthmatics dying of status asthmaticus (26). As both the eosinophil (+) and eosinophil () groups were markedly obstructed, it is reasonable to suggest that the presence of the neutrophil may have preferentially contributed to the development of that obstruction. The better (albeit "loose") correlation of neutrophil numbers with FEV₁ than seen with eosinophils would further support that concept. Similarly, these data also imply that the eosinophil may not be necessary to induce a physiologic phenotype of asthma which is indistinguishable from that defined by ACCP-ATS criteria 30 years ago (9).

Intriguingly, the eosinophil did appear to be associated with the development of structural abnormalities, specifically the development of SBM thickening, which have previously been associated with asthma. The results evaluating the SBM contrast to our previously reported data which evaluated severe asthmatics as a single group (8). As would be suggested from the very thin SBM seen in the eosinophil () group, there was no association at that level of clinically based characterization between SBM thickness, severity of disease, or even numbers of eosinophils. Now, after differentiating the groups by eosinophils, the thickness of the SBM in subjects with eosinophils is increased in relation to the number of eosinophils and compared with milder asthmatics and normal control subjects (7, 30). Unlike others, we were unable to find a close (although "significant") correlation between TGF-+ cells and SBM thickness (30), despite the relatively good correlation between eosinophils and SBM thickness. However, these results imply that in asthmatics with eosinophil-associated disease, the SBM may serve as a surrogate marker for airway remodeling and effects of treatment.

The results of this study further suggest that the type of inflammation and structural changes present may influence certain clinical and physiologic parameters. There were small differences in the baseline FEV₁ in the two groups, with the eosinophil () group actually more obstructed on the day of the bronchoscopy, in the absence of increased subepithelial fibrosis. In contrast, there were no differences in measures of air trapping (i.e., residual volume and thoracic gas volume), bronchodilator response, or airway conductance. Although only measured in a small group of severe asthmatics (because of the low starting FEV₁), there were no differences in the two groups in provocative concentration of methacholine causing a 20% reduction in FEV₁ (PC₂₀) (data not shown, but all less than 0.5 mg/ml). Interestingly, there was a suggestion of more airway collapse in eosinophil (+) subjects, as the FVC/SVC ratio (as a surrogate marker for airway collapsibility) was lower in these subjects than in the eosinophil () group (overall p = 0.03). This might suggest loss of elastic recoil in these patients, as has been reported by pressure-volume curve analysis in pediatric asthmatics who required mechanical ventilation (31).

In the context of this physiologic difference, the associated finding that the eosinophil (+) subjects had a much higher incidence of respiratory failure and mechanical ventilation than the eosinophil () group (p = 0.0004) may have considerable significance. It is noteworthy that the group which required mechanical ventilation [eosinophil (+) with more evidence for airway collapsibility] also had a significantly thicker SBM than the eosinophil () group. Taken together, these data suggest that although the eosinophil () subjects were at least as obstructed as the eosinophil (+) group, the degree of obstruction alone generally did not lead to acute respiratory compromise requiring intubation and mechanical ventilation. SBM thickening and eosinophils may be more important in driving the acute physiologic changes associated with asthma than they are in driving more chronic airflow limitation.

It should be stressed that the differentiation of the groups by eosinophil numbers was based on a strict, but somewhat arbitrary, criterion (2 standard deviations from the normal mean). There are, in fact, several subjects in the eosinophil (+) group who fall just above that cutoff. It is possible that some of these subjects might be more appropriately classified in the eosinophil () group. However, more work defining the subgroups appropriately is required before a better distinction can be made than the one offered in this report.

Finally, as mentioned initially, these findings have considerable impact on potential treatment options for severe, refractory asthmatics. Multiple "steroid-sparing" studies have been performed in severe asthmatics, with mixed and marginal results (32). It is conceivable that these marginal results are due to similar treatment approaches to different pathologies. Treatment approaches that either take into account the differences in pathology and/or attempt to treat the few similarities between the groups would seem to be indicated in future therapeutic trials in these patients.

In conclusion, these findings strongly suggest that the phenotype of "severe asthma" is composed of at least two distinct pathologic subtypes based on the presence or absence of eosinophils. These pathologic subtypes are associated with distinct structural changes (SBM thickening), which together may lead to differences in the basic physiology and clinical status of these patients, but more work is needed in this regard. Similarly, much remains to be understood regarding the inflammatory and genetic mechanisms contributing to the different pathologies. Future treatment options for these refractory asthmatics should likely keep these differences in mind.