Severe Asthma in Adults

Severe or refractory asthma, although afflicting a small percentage (likely < 10%) of the asthma population, remains a frustrating disease for both patients and the clinicians treating them. Because these patients remain difficult to treat and prone to severe exacerbations, they contribute disproportionately to the overall costs of asthma (1). The introduction of high-potency inhaled corticosteroids had a marked impact on the numbers of patients dependent on oral corticosteroids, but further attempts at therapy have met with mixed success, partially because of the heterogeneity of the disease (2). More research is needed to expand the understanding of the "diseases" that make up severe asthma, and eventually improve treatment. This article discusses a range of topics related to severe asthma, which are organized as follows:

DEFINITIONS

Before in-depth studies of any disease can be undertaken, a recognized definition must be developed. This is especially important in a complex disease such as asthma, which is likely a collection of different phenotypes, rather than a single, specific disease with a unifying pathogenic mechanism. Various clinical definitions have been proposed through national and international guidelines, working groups, and workshops, which incorporate symptoms, lung function, exacerbations, and, in many cases, specific use of high-dose corticosteroids (3-6). In the original European Network description, patients with severe asthma were defined as those who were difficult to control after evaluation and treatment by an asthma specialist for a year or more (5, 7). Perhaps the most comprehensive attempt at a definition was undertaken by an American Thoracic Society-sponsored workshop, the proceedings of which were published in 2000 (6). This definition included one of two major criteria (continuous high-dose inhaled corticosteroids or oral corticosteroids for > 50% of the previous year), with two of seven additional minor criteria required. The minor criteria included aspects of lung function, exacerbations, disease stability, and the use of one or more additional controller medications (Table 1). In conjunction with these criteria, patients must have had adherence and exacerbating factors fully addressed. This definition has recently been incorporated into an NHLBI-sponsored network to study severe asthma. Although these definitions are a start, it is likely that objective investigations of populations with severe asthma, such as those undertaken by the NHLBI, the European Network for Understanding Mechanisms of Severe Asthma (ENFUMOSA), and the follow-up to ENFUMOSA, BioAir, or pharmaceutical-sponsored studies, such as TENOR (The Epidemiology and Natural History: Outcomes and Treatment Regimens, a study of nearly 5,000 "difficult to treat" patients with asthma), will markedly improve our ability to identify, study, and treat patients with severe asthma (7, 8). Of note, severe asthma can be easily confused with "poorly controlled" asthma. Although distinctions are still being drawn, poorly controlled asthma may be a more transient state of the disease, in which the levels of symptoms can be improved when standard approaches to therapy are appropriately used. On the contrary, severe asthma represents disease that is not responsive to standard therapy, thus remaining poorly controlled despite vigorous attempts at management. This article focuses on severe asthma, not asthma that is poorly controlled; however, occasionally, distinctions based on published literature are difficult to confirm.

EPIDEMIOLOGY OF THE DEVELOPMENT AND MAINTENANCE OF SEVERE ASTHMA

Natural History for the Development of Severe Asthma

Surprisingly little is known about the development of severe asthma. Whether severe asthma develops slowly over time because of yet unproven genetic and environmental factors or whether an acute event occurs near the onset of disease that irreversibly alters the structure of the lungs (airways or parenchyma) to promote severe asthma is not clear. Furthermore, it is not certain whether any patient with asthma is at risk for developing severe asthma, or only a poorly defined subset. Currently, longitudinal studies have focused primarily on lung function, specifically FEV^sub 1^. Although decline in FEV^sub 1^ contributes to more severe disease, it is likely that additional factors are required for disease progression. These additional factors may include worsening levels of inflammation hyperresponsiveness, lung compliance, or even levels of symptoms. Longitudinal studies that incorporate all of these elements (beyond FEV^sub 1^ alone) are required to better understand the development of severe asthma. Unfortunately, although limited in usefulness, FEV^sub 1^ is the only outcome measured longitudinally to date to which development of severe asthma can currently be linked. With this limitation in mind, helpful developmental information can still be derived from two large cohorts of subjects with asthma and control subjects from Australia and New Zealand, which have been followed for 17 to 35 years (9, 10). Those data suggest that children with reduced lung function early in life are likely to have reduced lung function in adulthood. However, "progressive decline" in lung function was shown to be modest as compared with the initial loss in this group, and as compared with control groups. This finding is in contrast to the study by Lange and colleagues (11), which reported a more rapid decline in FEV^sub 1^ over time in "all" patients with asthma compared with control subjects (11). In that study, no attempt was made to break subjects with asthma into severity groups. Furthermore, two studies from Europe suggest that late-/adult-onset asthma is associated with a more rapid decline in lung function (12, 13). In a recent study of 80 patients with severe asthma seen at National Jewish, approximately two-thirds had onset in early childhood, with the remaining one-third at the age of 12 years or later. The patients with late-onset disease had a lower FEV^sub 1^ as an adult than the childhood-onset group, despite the fact that the adult-onset group had the disease for substantially fewer years (14). Although this study supports the concept that late-onset asthma is associated with a more rapid decline in lung function (or an early profound loss), larger longitudinal studies are required to verify this.

Risk Factors for the Development of Severe Asthma

Genetics. As with most complex diseases, risk factors for disease development can be divided into genetic and environmental. From a genetic perspective, it has been difficult to implicate one gene; rather, asthma is a complex disease involving multiple genes. Severe asthma is not likely to be different. There are reports of relevant mutations in both the promoter region of the interleukin 4 (IL-4) gene or the coding regions of the IL-4 receptor, some of which have been linked to loss of lung function, others to near-fatal events (15-17). Furthermore, genes related to "non-Th2" factors have also been associated with severity of asthma, including those for transforming growth factor ß^sub 1^ (TGF-ß^sub 1^) and monocyte chemotactic protein 1, both of which can promote fibrotic reactions (18, 19). A small study from the Netherlands suggested that polymorphisms in ADAM-33 were associated with more rapid declines in lung function (20). Additional genetic elements of severe asthma could be related to genetically determined responses to standard medications. Polymorphisms in the ß-2 receptor, particularly those leading to arginine substitutions at position 16 (Arg/Arg), have been associated with worsened outcomes in prospective studies of regularly scheduled short-acting ß-agonist use (21). This polymorphism is present in approximately 15% of the white population and approximately 25% of the African American population (22). Because there is documented evidence of ß-agonist overuse in lower socioeconomic groups, genetically determined differences in responses to these medications in certain racial/ethnic groups compared with the general population could lead to worsened outcomes (23).

Specific genetic abnormalities related to the female sex have yet to be identified in association with severe asthma, but, in adults, severe asthma is more commonly seen in women (7, 8). Theories for this skewing of the disease population toward the female sex include elements related to hormonal changes as well as the influence of obesity (which is more prevalent in women), but definitive relationships have not yet been shown.

Environmental risk factors. Allergen exposure has long been described as an important environmental risk factor for asthma, and severe asthma in particular, with the strongest data for house dust mite, cockroach, and alternaria exposures (24-26). In all these studies, however, there has not been formal distinction between severe asthma and poorly controlled asthma. In the National Jewish dataset of 80 patients with severe asthma, 50 to 80% of the cohort had worsening asthma symptoms (most or all of the time) in response to environmental triggers, whereas more than 80% were atopic. Similar percentages were seen in the control group of subjects with milder asthma (unpublished observations). This could imply that allergic responses must be combined with other factors, including prolonged and/or highdose allergen exposure, to increase disease severity, as opposed to merely decreasing asthma control in the short term. For instance, continuing to own a cat despite being aware of the negative effects on symptoms certainly influences asthma control in the short term, but over the long term could impact asthma severity as well. These data could also suggest that allergic responses are seen at all levels of asthma severity, and do not play a direct role in determining level of severity, but rather contribute to the current level of asthma control. Similarly, subjects with severe asthma in the ENFUMOSA study were less atopic than a control group with milder asthma, further suggesting that severity can be disassociated from atopic/allergic reactions. These combined data suggest that, although the presence of allergic responses may predispose to allergic asthma, they may not be the strongest contributing factors to severity (7).

Cigarette smoking also can contribute to the development of severe asthma, with data to suggest smokers are more symptomatic, have more frequent and severe exacerbations, and have a more rapid decline in lung function than nonsmoking subjects with asthma (11, 27-29). However, severe asthma also develops in lifelong nonsmokers, suggesting an interplay with other factors. Cigarette smoking also limits responses to the most effective therapy for asthma, corticosteroids (30). Aggressive efforts to discontinue tobacco consumption are necessary. Furthermore, sputum eosinophils appear to be diminished in smokers with asthma, as compared with nonsmokers with asthma, but studies of pathologic/biopsy differences are not available (31). Therefore, it is not clear how closely the pathogenesis of smoking-related severe asthma relates to the severe asthma arising in nonsmokers.

Finally, infection could also contribute to severe disease, with respiratory syncytial virus infections implicated in childhood, whereas pathogens like mycoplasma and chlamydia may play a role in adults (13, 32, 33).

Exacerbating factors. A variety of comorbidities have been associated with severe asthma, but, similar to the factors listed previously, their impact on the development or maintenance of severe asthma is less clear. Sinusitis is extremely common in severe asthma, with evidence for some disease in more than 80% of this population. The severity of the sinusitis has been associated with both inflammation and lung function abnormalities (34). Unfortunately, little effective long-term therapy for sinusitis exists to allow determination of whether this is a parallel or a causative process. A recent large-scale epidemiologic study of severe/difficult-to-treat patients with asthma (TENOR) suggested that the body mass index increases with increasing severity of disease (76% of that severe cohort were either overweight or obese) (35). Many of the obese patients may have obstructive sleep apnea as a comorbid condition, which can affect overall lung function; thus, screening and treatment of this syndrome are appropriate. Although weight loss has been reported to improve lung volumes, the relationship to airway obstruction, reactivity, and asthma symptoms is controversial (36, 37).

Although psychologic disease has previously been associated with near-fatal asthma, more extensive studies have not suggested that psychologic disease is more common in the population with severe asthma than in society in general (38-40). However, patients with severe asthma with the most frequent health care use do exhibit more psychopathology than those who use health care less often (41). Even in these instances, the majority of psychologic abnormalities relate to expected increases in anxiety, depression, and lack of trust for health care providers (42). More extreme forms of psychopathology, such as bipolar disorder, personality disorders, and schizophrenia, have not been identified as occurring more commonly in severe asthma.

Another exacerbating factor that has been associated with severe asthma but which, in reality, may be more important in poorly controlled asthma is compliance/adherence to medications. Studies have suggested that in children and adolescents, instability of disease ("poor control") is related to adherence to corticosteroid therapy (43). In addition, prescription refills for inhaled corticosteroids are approximately two to four canisters/ year, suggesting that problems with compliance/adherence cross all severities of asthma, and are not exclusive to severe asthma (44, 45), yet the majority of patients do not have severe disease.

Biologic response to corticosteroids should also be considered as a lack of biologic response may give the physician a false perception of poor compliance/adherence. A recent small study suggested that corticosteroid therapy is clinically effective in only approximately 70% of the general population with asthma, under rigorously controlled study trial conditions (46). A much larger clinical trial in moderate asthma (the Gaining Optimal Asthma Control, or GOAL, study) seems to confirm a similar pattern of response to inhaled and oral corticosteroids (47). In addition, data suggest that adherence to medications increases with increasing severity of disease such that further study is required in the severe population to determine whether there may be even higher percentages of patients who truly do not respond to therapy as compared with those who do not respond because of compliance/adherence issues (48). If compliance/adherence is a concern when the patient is on oral corticosteroids, an early-morning cortisol level demonstrating total or nearly complete suppression can be helpful. Assessing response to treatment with injectable long-acting corticosteroids, such as depomethylprednisolone or triamcinolone, will override the issue of compliance (49, 50).

PHYSIOLOGIC FACTORS, INCLUDING OBSTRUCTION, REACTIVITY, AND RECOIL

Although FEV^sub 1^ has been used by most guidelines to indicate the presence of severe disease, it is clear that the correlation between FEV^sub 1^ and disease symptoms is poor at best (51). Although airflow limitation is a component of the physiologic changes of severe asthma, it is likely only part of the picture. One prevalent theory regarding severe asthma is that it develops because of a progressive long-term increase in airflow limitation, which is often presumed to be irreversible (52). Although this may be true in some patients, others may have severe airflow limitation at presentation, whereas other patients, primarily adults, may develop a more rapid decline in lung function over a 10-year (or less) period of time (13). Finally, patients with "brittle asthma" may have completely normal lung function between episodes. Longitudinal studies to evaluate the rate of decline in lung function in severe asthma as compared with milder forms of the disease would begin to address some of these issues.

Changes in airway reactivity also play a role in the severity of asthma, but the correlations of PC^sub 20^ with disease severity, though present, are poor as well (53). Increased airway reactivity likely relates to stability of airflow-hence, variability in peak flow/FEV^sub 1^. This "instability" may be an important aspect to the symptomatology of a subgroup of patients with severe asthma, in whom continuous airflow limitation plays a more minor role (54).

Because FEV^sub 1^ and airway reactivity alterations do not adequately explain disease severity, it is conceivable that other physiologic factors, such as changes in elastic recoil/collapsibility and/ or small airway physiology, are also important. It has long been observed that the elastic recoil properties of the lung in asthma are not normal (55, 56). Lung compliance has been reported to be increased in patients with moderate persistent asthma, and a recent study suggests that both loss of elastic recoil at total lung capacity and hyperinflation are risk factors for near-fatal asthma (Figure 1) (57, 58). This loss of elastic recoil is not caused by emphysematous changes, as measured by computed tomography scans. Further studies to evaluate the mechanisms leading to this loss of recoil are needed, but loss of alveolar-airway attachments in fatal asthma may play a role (59).

There are also suggestions that the airways (in addition to parenchyma) may be more compliant or collapsible than normal (60). This is somewhat surprising given the longstanding perception of airway wall thickening in asthma, which would seem to decrease compliance. However, in a subset of patients with severe asthma with persistent eosinophilia, the forced vital capacity to slow vital capacity ratio is decreased, which is suggestive of airway collapse on forced expiration (60). These patients with asthma also appear to be at a higher risk for near-fatal events than those with a more normal (1:1) ratio. Studies that evaluate the physiologic relationship between airway collapsibility and loss of elastic recoil are needed.

Studies of small airway physiology are limited in patients with severe asthma. A study of patients with difficult to control compared with stable asthma suggested that, although FEV^sub 1^ and residual volume were similar between the groups, that the patients with difficult-to-control asthma had greater evidence for early closure of small airways, as measured by closing volume (61). However, in addition, air trapping, without marked hyperinflation, has also been reported, with residual volumes over 200% predicted in severe asthma, with only modestly increased thoracic gas volumes (60). Whether this increase in residual volume is reflective of small airway or distal lung disease is not known, although a recent study in nocturnal asthma found a modest correlation of parenchymal inflammation with residual volume (62). Studies that evaluate changes in small airway structure, especially those involving the outer wall/alveolar attachments region, should prove very helpful.

Another physiologically related abnormality in severe asthma is diminished perception of dyspnea (63,64). This lack of appreciation of diminished airflow has been reported in several studies and likely puts patients with already compromised lung function at greater risk for a severe, life-threatening event. Persistent eosinophilic inflammation has been linked to a decreased perception of dyspnea (65). However, other studies have shown that patients with severe asthma with persistent eosinophilic inflammation are actually more symptomatic than patients without eosinophilic inflammation, such that the mechanisms behind this decreased perception require further testing (14).

More than likely, physiologic explanations for asthma severity will require an integration of measures of airflow limitation, airway reactivity, elastic recoil, and perhaps small airway disease as well. This integrated approach will more likely explain why a patient with an FEV^sub 1^ of 70% of predicted might have more severe or difficult-to-control disease than a comparable patient with an FEV, of 55% predicted.

PHENOTYPES OF SEVERE ASTHMA AS DEFINED BY AGE AT ONSET, AND ALLERGIC, INFLAMMATORY, AND TREATMENT-RELATED PARAMETERS

It has long been clear that variability exists in the clinical presentation of severe asthma. Numerous attempts at classifying potential phenotypes of asthma and severe asthma have been proposed, including eosinophilic asthma, noneosinophilic asthma, intrinsic versus extrinsic asthma, brittle versus stable airflow limitation, as well as aspirin-sensitive asthma. Despite the likelihood that these phenotypes overlap, studies have generally tended to focus on one or the other without integrating the potential phenotypes. However, reasonable supporting evidence exists for the presence of at least four (and likely more) general severe asthma phenotypes (Figure 2, Table 2).

Early- versus Late-Onset Severe Asthma

Although considerable focus on asthma epidemiology has been on childhood-onset disease, a large percentage of asthma may develop in late childhood or adulthood (66). Although there has been concern that recall bias may prevent some patients from recognizing disease earlier in life, this does not account for all patients with late-onset disease. In addition, the apparent allergic, pathologic, and physiologic differences between early- and late-onset disease strengthen the case for differences in the phenotypes (14).

As noted, a study of 80 patients with severe asthma evaluated at National Jewish Medical and Research Center found that nearly two-thirds developed disease before the age of 12 and one-third developed asthma after age 12 (late-onset; Table 2) (14). These two groups differed markedly in allergic responses, with 98% of patients with early-onset asthma demonstrating positive allergy skin tests, whereas 76% of patients with late-onset asthma had positive tests (p = 0.007). Support for differences in allergic elements came through questions regarding asthma symptoms in response to seasonal changes, house dust, or furred animals. Approximately 70 to 75% of patients with early-onset asthma answered that symptoms occurred most or all of the time to these stimuli, whereas 40 to 50% of patients with late-onset asthma answered similarly. In addition, both a history of eczema and a family history of asthma were significantly more common in the early-onset group. Furthermore, there were higher percentages of African Americans in the early-onset group as well. Although there was no difference in general asthma symptoms between the two groups, lung function (FEV^sub 1^ and FVC) was generally worse in late-onset severe asthma than in early-onset disease, despite the shorter duration of illness. Last, early-onset asthma had higher numbers of tissue lymphocytes than late-onset disease, whereas late-onset disease had higher numbers of eosinophils. Patients with early-onset asthma would appear to be a remarkably homogeneous group, with strong genetic influences and the presence of allergic responses (similar to extrinsic asthma), whereas patients with late-onset disease are a more heterogeneous group, with evidence for both allergic and nonallergic disease. Certainly, there is overlap with the terms "extrinsic" and "intrinsic" asthma (67).

Phenotyping by Presence or Absence of Airway Eosinophils

Eosinophilic phenotypes. Pathologic (airway and bronchoalveolar lavage) studies of severe asthma suggest that one-half to two-thirds of patients with severe asthma have persistent large airway tissue eosinophils, despite continued high-dose systemic and inhaled steroids. The presence of eosinophils (as measured by sputum, lavage, biopsy, or perhaps exhaled nitric oxide) may represent yet another subtype of severe asthma characterized by a higher level of active symptoms, a lower FEV^sub 1^, and a greater likelihood for exacerbations and near-fatal events, than a subtype without eosinophils (60, 68, 69). Recently, this differentiation by presence or absence of eosinophils has been applied to early- and late-onset severe asthma, with indication for both similarities and differences in the eosinophilic process dependent on age at onset (14). As noted, persistent eosinophilia appears to be more prevalent in late-onset than in early-onset disease, despite similar high-dose corticosteroid use. In early-onset disease, increases in eosinophils are associated with increases in T lymphocytes and mast cells, whereas late-onset disease with eosinophils has little evidence for involvement of other inflammatory cell types (Figures 2A and 2b). Late-onset disease is additionally characterized by higher cysteinyl leukotriene levels than in early disease, even when controlling for the numbers of eosinophils present. Early-onset eosinophilic asthma had the highest numbers of patients with previous near-fatal events.

Despite these differences, there also appear to be similarities in the eosinophilic inflammation in severe asthma that are not differentiated by age at onset. Patients with severe asthma with persistent eosinophils at any age have increased levels of TGF-ß (specifically, TGF-ß2) in tissue and the 15-lipoxygenase enzyme and its product, 15-hydroxyeicosatetraenoic acid (60, 70-72). Structural changes, such as a thicker subepithelial basement membrane (SBM), also appear to be associated with persistence of eosinophilic inflammation without regard to age at onset (14, 60).

Although eosinophils have been associated with Th2 immune processes, whether Th2 inflammation drives lung eosinophilia in severe asthma is not clear. IL-4 or IL-13 appear to be elevated in both atopic and nonatopic forms of mild asthma (73). However, neither IL-4 nor IL-13 has been definitively shown to increase in relation to disease severity or eosinophilic disease. A recent analysis of bronchoalveolar lavage cells or tissue-derived IL-4 or IL-13 mRNA and protein demonstrated lower levels in patients with severe (steroid-treated) asthma than in control subjects with milder asthma, independent of age at onset or eosinophils (74). Indeed, although "allergic" disease is more common in early-onset severe asthma (onset before the age of 12 years), this group is less likely to have persistent tissue eosinophilia than severe asthma with disease onset later in life (i.e., where there is less evidence for atopy/allergy). Similarly, patients in the ENFUMOSA study had a lower percentage of atopy than the control group with milder asthma, but a higher percentage had aspirin sensitivity (7). Although aspirin-sensitive asthma is typically associated with nonatopic disease, it is generally also considered to be a highly eosinophilic process. Therefore, further studies to determine the relationship of Th2 pathways to persistent eosinophilic inflammation are necessary.

Neutrophilic phenotypes. In some (but not all) cases, where eosinophils are absent, there may be an increase in neutrophils. The increase in neutrophils is not always exclusive for the absence of eosinphils, and the two cell types may be concomitantly present in tissue (60, 75). This increase in neutrophils has been seen in sputum, lavage, and biopsy studies from patients with severe/difficult asthma on high doses of inhaled/oral steroids (76, 77). The mechanisms or the clinical implications for this neutrophilic inflammation are not clear. In some cases (likely more common in late-onset disease), it may represent a "pathologically different disease," such as a bronchiolitis obliterans or a variant thereof (78). Interestingly, a recent comparison study of computed tomography scans in patients with severe asthma versus bronchiolitis obliterans showed that changes on computed tomography could not convincingly discriminate bronchiolitis obliterans from asthma (78). Because the clinical definition of asthma is primarily "physiologic," several diseases, including bronchiolitis obliterans, may meet the criteria. However, the pathology of these diseases may vary substantially from that which is classically believed to represent asthma. In other cases, particularly noneosinophilic early-onset disease, neutrophils may be the only "residual" inflammation, with steroids having effectively reduced the eosinophils (79). Finally, steroids themselves have been reported to suppress neutrophil apoptosis, such that the treatment of severe asthma may increase the numbers of neutrophils as well (79).

Whatever the underlying reason for increased neutrophils, their presence is associated with an increase in matrix metalloproteinase 9 (MMP-9) in bronchoalveolar lavage fluid and tissue (SBM), as well as lower lung function (80). The MMP-9 is specifically increased in a high-molecular-weight form associated with the neutrophil granular protein, lipocalin, which is believed to be poorly inhibited by the natural inhibitors of MMPs, tissue inhibitors of MMPs (or TIMPs) (81). In severe asthma, expression of this MMP-9 is also poorly inhibited by corticosteroids both in vivo in bronchoalveolar lavage fluid and in vitro in bronchoalveolar lavage cell supernatants.

Pauci-inflammatory phenotype. In addition to persistent eosinophilia or neutrophilia, there remains a group of patients with severe asthma in whom virtually no inflammation of the classic cellular variety is present on endobronchial biopsy (Figures 2C and 2D) (82). Very little is understood about the pathogenesis of disease in this group. Possibilities for disease pathogenesis in this phenotype include the presence of localized distal lung inflammation, not usually accessible by bronchoscopic studies, or, similar to patients with neutrophilic inflammation, the presence of a totally different, perhaps bronchiolitic disease. In support of this hypothesis is the observation that tissue from late-onset severe asthma without "classic" inflammation has no evidence for SBM thickening commonly seen in most asthma biopsies (Figure 2D) (14). It is also possible that the lungs have been structurally altered to result in persistent clinical symptoms, but that inflammation, in the classic sense, is no longer present. In fact, the studies by Benayoun and colleagues (83) report that the primary differentiating factor between tissue from patients with mild and severe asthma was not inflammation related, but rather was an increase in the amount of smooth muscle. Finally, it is also possible that a type of inflammation exists involving "nonclassic" inflammatory/asthma cells. In a study of severe oral steroid-treated patients with asthma, there was an increase in monocyte/macrophage activation of the nuclear factor-?B pathway (84, 85). Unfortunately, the degree of cellular inflammation in these patients with severe asthma was not reported. Therefore, it is difficult to know whether persistent activation of these or alternative inflammatory or oxidation-related pathways, without increases in classic inflammatory cells, could perpetuate some forms of severe asthma.

Thus, at our current level of understanding, early-onset eosinophilic disease might represent a classic Th2 inflammation poorly responsive to steroids (Figure 2A). Early-onset severe asthma without eosinophils may represent a group where the inflammation has responded to steroids but the disease has not (Figure 2C). Late-onset eosinophilic asthma includes both allergic asthma and perhaps variants of hypereosinophilic syndromes (Figure 2B), whereas late-onset disease without eosinophils shares very few characteristics with the other three groups and may represent one or more separate poorly understood and/or different diseases (Figure 2D).

Poor steroid responsiveness. It is likely that absolute steroid resistance rarely occurs, even in severe asthma (86). Rather, severe asthma generally is characterized by a shift of the dose response to the right, with requirement for much higher levels of steroids to maintain stability. Thus, although steroid unresponsiveness may not be the cause of severe asthma, it is certain to be involved in the overall control of disease.

Studies of the phenotype of poorly steroid responsive (resistant) asthma have traditionally focused on the lymphocyte as the target cell for the resistant pathways. However, the phenotype of poor steroid responsiveness likely encompasses many different underlying causes (Table 3). Phenotypes of severe asthma associated with persistent eosinophilic inflammation, in either earlyor late-onset disease, may represent the more classic version of poor steroid responsiveness, where the inflammation is not responsive to steroids. In phenotypes without persistent eosinophil/lymphocytic inflammation, the mechanisms are less clear. It is possible that, in some cases, the inflammation (as we know it) is responsive to steroids, whereas steroids do not impact remaining poorly understood structural elements driving the severity. It is possible that, in these situations, the damage has already been done, such that the remodeled lung changes do not respond further to steroids. Similarly, a different type of inflammation, as might be seen in neutrophilic-associated disease, may not be as responsive to steroids as an eosinophil-associated process (87). Finally, concomitant smoking may also decrease steroid responsiveness in the subset of smokers with severe asthma (37).

The majority of studies to address steroid unresponsiveness have focused on that subset of patients with persistent eosinophilic/lymphocytic inflammation. Possibilities for poor steroid responsiveness in this group include high levels of proinflammatory mediators sequestering the glucocorticoid receptor, diminished binding of the glucocorticoid receptor to the genome, or increased levels of an alternatively spliced glucocorticoid receptor (glucocorticoid receptor ß), which has no direct transcriptional-related effects (88-90). Abnormalities in the balance of the histone acetylation and deacetylalion pathways in asthma, which contribute to regulation of inflammatory gene transcription, may also play a role (91). A recent study suggested that peripheral blood cells from patients with severe asthma had less histone deacetylation activity in response to steroids than in patients with milder asthma, thereby preventing some of the antiinflammatory responses to steroids from occurring (92). In addition, in patients with persistent eosinophilic inflammation, in the absence of lymphocytes, other, nonlymphocytic eosinophil-specific factors may play a role.

As our understanding of asthma phenotypes evolves, implications for treatment will likely emerge. For instance, moderate to severe asthma with a neutrophilic process may respond differently to corticosteroids than asthma with persistent eosinophils. Green and colleagues (87) described a neutrophil-predominant pattern as being associated with a less robust response to inhaled corticosteroids. In contrast, eosinophilic severe asthma treated with a high dose of intramuscular triamcinolone showed marked improvement in symptoms and FEV^sub 1^ and a reduction in sputum eosinophils (50). A similar study specifically looking at the predictive value of eosinophils regarding treatment has also been reported on severe childhood asthma. This would suggest that the presence of eosinophils or neutrophils could be used to guide whether a higher or lower dose of steroids should be initiated.

THE ROLE OF SMALL AIRWAYS

Physiologic and pathologic data suggest that inflammatory changes exist in the lung periphery as well as in the large airways, where tissue is more easily accessible. Although the relationship to severity of disease remains unclear, autopsy studies have reported both increased inflammation and wall thickness in patients who died of asthma, as opposed to control subjects with milder asthma and normal control subjects (93, 94). Studies of living patients with asthma suggest that distal lung inflammation may be as or more important than proximal inflammation to disease severity (95, 96). The distribution of inflammatory cells may also be different in the distal lung, with mast cells, specifically chymase-positive mast cells, increased in the small airway outer wall and alveolar attachments (70, 94). These chymase-positive mast cells in the small airways were the only cell type strongly (and positively) related to pulmonary function changes in patients with severe asthma (70). Potential explanations for the positive impact could relate to reparative effects from the specific mast cell phenotype, but further work is needed. Other inflammatory and structural changes remain to be investigated.

These changes in the small airways and alveolar attachments could also have considerable implications for current drug therapy of asthma, and severe asthma in particular. Most inhaled medications are unlikely to reach the lung periphery in high amounts, resulting in inadequate drug delivery to the small airways (97).

AIRWAY REMODELING

In the last 5 years, it has been suggested that the apparent progressive loss of lung function in more severe forms of asthma is caused by structural or remodeling changes in the airways and perhaps parenchyma. However, what the precise changes are remains unclear. Numerous structures have been implicated, including the SBM, epithelium, smooth muscle, nerves, and blood vessels. In general, there is a paucity of data on any of these structures in relation to disease severity.

The SBM has been the most extensively studied structure associated with lung remodeling in asthma, but its relationship to disease severity is unclear. Patients with severe asthma with persistent eosinophils consistently have the thickest SBM when compared with normal control subjects, subjects with milder asthma, and those patients with asthma without eosinophils (14, 60). This thickened SBM is associated with high numbers of TGF-ß-positive cells in the submucosa (60, 98, 99). TGF-ß from these cells could activate closely associated fibroblasts to increase collagen production, increasing the thickness of the SBM (100). Furthermore, when an airway fibroblast is stimulated with the combination of TGF-ß and a Th2 cytokine, such as IL-4 or IL-13, a profound increase in the production of eotaxin-1, a potent eosinophil chemoattractant, occurs (101), with an additive effect on production of procollagen I (100). Thus, in eosinophilic forms of severe asthma associated with increased TGF-ß (even in the presence of modest Th2 inflammation), the fibroblast could not only contribute to the fibrotic response but perpetuate the tissue eosinophilia as well. Figure 3 depicts possible interactions between eosinophils and other cell types in severe asthma that may contribute to fibrosis.

Furthermore, factors such as MMP-9, associated with tissue breakdown (as opposed to fibrosis), are also present in the SBM (80). Although the function of the MMP-9 in the SBM is not clear, it could contribute to a continuous, ongoing remodeling of the SBM by enhancing membrane turnover. The two factors, TGF-ß and MMP-9, could work in an opposing fashion, with the former resulting in an increase in collagen deposition, whereas the latter (and perhaps other MMPs) breaks it down. This cycling may also explain why the absolute increase in thickness is small and not likely to explain the increase in airflow limitation. However, the rate of turnover of the SBM may be a sign of disease activity. In addition, it may be a marker for abnormalities in composition, distribution, or quantity of extracellular matrix elements in other regions of the airway or parenchyma that are not as easily identified (102).

Although the epithelium is almost certainly abnormal in asthma, studies in severe asthma are few. In patients with asthma, there is an increase in the goblet-to-ciliated epithelial cell ratio, and mucus plugging of the small and medium airways, seen on autopsies, may contribute to airflow limitation and air trapping. Early studies suggest that two epithelial cell pathways, epidermal growth factor receptor and TGF-ß2, may be altered in asthma and contribute to an inappropriate and inadequate repair process, augmenting goblet cell metaplasia and mucus production (103-105). TGF-ß2 produced by the epithelium appears to contribute to increases in goblet cells and mucus production, and may drive IL-13-induced mucus production (105). In addition to increasing mucus production, IL-13 may also impact ciliary function, contributing to decreased clearance of mucus (106). TGF-ß2 present in the epithelium could also contribute to collagen deposition in the SBM (Figure 3) (105).

The amount (and perhaps phenotypes) of smooth muscle in the airways of patients with severe asthma has also been reported to be increased. Patients dying of status asthmaticus were reported to have increased smooth muscle mass in the airways from the largest airways to nearly the smallest (107). As noted previously, it has been suggested that the major pathologic differentiating factor between mild and severe asthma is the amount of smooth muscle present in large airway biopsies, as well as the hypertrophy of the individual smooth muscle cells (83). However, analysis of the amount of smooth muscle and individual cells in endobronchial biopsies is problematic, and these results should be interpreted with caution. Mechanisms driving changes in smooth muscle are not clear but could involve a host of growth factors, including epidermal growth factor and TGF-ß. A defect in the antiproliferative response of smooth muscle to corticosteroids in patients with asthma, involving the CCAT/enhancer binding protein a (C/EBPagr;) pathway, could also contribute (108).

Relating any of these structural changes to functional changes has been difficult. Airway wall thickness, as measured on computed tomography scans, has been suggested to predict airflow limitation, but the relationship is not strong (52). This may be because other factors, such as loss of elastic recoil/collapsibility, are also important. Although loss of recoil has been suggested to be caused by airway-parenchymal uncoupling, pathologic studies to support this concept have been limited. Elastin has been shown to be abnormal (decreased or disordered) in patients who died of asthma in both the large and small airways (59, 109, 110). Furthermore, proteolytic enzymes that alter elastin composition (MMP-2, MMP-9) have been shown to be increased in several instances in asthma, but perhaps more importantly in cases of status asthmaticus (111, 112). Most important, there are now data suggesting that the airway-parenchymal attachments in the small airways of individuals who died of status are disrupted compared with control subjects, pathologically explaining the physiologic findings (59).

EVALUATION OF SEVERE ASTHMA AND THERAPEUTIC OPTIONS

Evaluation

There are no validated algorithms to substantiate the most useful approach to the evaluation of severe asthma. However, it is rational to suggest that the approach should have at least three components: (1) confirmation that the disease is asthma, (2) evaluation of confounding/exacerbating factors, and (3) evaluation of asthma phenotype (Table 4). Under Category 1 (confirmation), appropriate tests include full pulmonary function tests (with lung volumes and diffusing capacity), chest computed tomography scan, and perhaps most important, a standard methacholine challenge followed immediately by a laryngoscopy to look for paradoxic inspiratory closure of the vocal cords (vocal cord dysfunction) (113, 114). These tests can be used to differentiate subtle changes related to interstitial diseases and mixed obstructive/ restrictive diseases (hypersensitivity pneumonitis and sarcoidosis), as well as to evaluate for allergic bronchopuimonary aspergillosis, emphysema, bronchiolitis obliterans, and vocal cord dysfunction. Under Category 2 (confounding/exacerbating disease), the following tests are helpful: a pH probe to evaluate gastroesophageal reflux disease; a sinus computed tomography to evaluate for surgically amenable disease, nasal/sinus polyps, and fungal disease; an IgE level and allergen skin testing; a sleep study; and, when patients are on systemic steroids, an early-morning cortisol to evaluate compliance/adherence. Finally, testing specific for Category 3 (evaluation of phenotype), likely the least proven, could potentially be highly beneficial. Categorization into early/allergic or late-onset disease can be made when appropriate questions are asked beyond "How long have you had asthma?" Additional helpful questions include those related to history of bronchitis or pneumonias in childhood and inability to keep up with physical activities in childhood, which may suggest that, although a diagnosis was not made, the disease was present. Support for an early-onset or a late-onset allergic phenotype can be made by asking about symptoms related to allergen exposure and performing allergy skin testing. Relationship to onset of disease after a respiratory infection, especially in adults, is also helpful, as are queries regarding aspirin sensitivity, because they may also guide approach to therapy. Evaluation for the presence or absence of eosinophilic inflammation can be done with bronchoscopy, although this should only be done in centers with a bronchoscopist and in those centers experienced in performing these procedures in patients with severe asthma, and with excellent pulmonary pathology support. Alternatively, sputum eosinophils (and even exhaled nitric oxide) can be helpful, as a recent abstract suggested a strong correlation between tissue eosinophils and sputum eosinophil percentages in severe asthma (115).

Treatment

Treatment of severe asthma remains highly problematic. Steroids remain the drug of choice, likely because of their broad and nonspecific effects. If a patient has not been on high-dose, high-potency inhaled corticosteroids, a trial is certainly warranted (2). Whether the addition of an inhaled corticosteroid with smaller particle size and better peripheral distribution will improve inflammation in the distal lung and thereby improve severity is not yet clear. However, it is also possible that a patient has poor steroid responsiveness for any number of reasons, such that only high-dose systemic steroids will prove beneficial to the asthma (but not without side effects; see Table 3). Benefit may be seen in some cases with leukotriene modifiers, especially in the large percentage (20-25%) of patients with severe asthma who may be aspirin sensitive (7, 116). The 5-lipoxygenase inhibitors may be particularly helpful (personal observation in more than 20 patients with severe asthma). Long-acting ß-agonists, which have efficacy in moderate, persistent asthma, may not be as helpful in patients with severe disease (117, 118). Other forms of therapy, such as cyclosporine and methotrexate, have only limited applicability in this population, with cyclosporine also carrying a very large risk for side effects (119).

In the first attempts at treating by phenotype, anti-IgE therapy has been shown to reduce hospitalizations in patients with moderate to severe allergic asthma, but their utility in very severe asthma (or "less allergic") asthma is not clear (120). Similarly, treating by the presence or absence of eosinophil phenotype may also be desirable, because at least two studies have suggested good responses to high-dose antiinflammatory (steroid) therapy in patients with persistent eosinophils (50, 68). Whether specific eosinophil-targeted therapies, such as imitinib or mepolizumab (anti-IL-5), will have efficacy in this group awaits clinical trials. Similarly, the utility of other biologic approaches, such as anti-tumor necrosis factor a or modulation of Th2 cytokine or chemokine-related pathways, either generally or by phenotype, has yet to be determined.

CONCLUSIONS

The basic understanding of the mechanisms behind the development of asthma is only beginning to be elucidated. Development of severe asthma is likely complicated and requires several combined insults, including genetic, structural, and environmental elements, and perhaps also responses to medications. The disease is likely a mix of various syndromes that have differentiating elements, but that also share similarities at a pathophysiologic level, including consistencies in airway obstruction, hyperresponsiveness, and loss of elastic recoil. A better understanding of the immunologic and pathologic phenotypes of severe asthma should enhance our ability to both understand the genetics of these syndromes as well as to improve our approaches to therapy.

Conflict of Interest Statement: S.W. has received $10,000/year for the last 3 years as a consultant and member of the Advisory Board and $4,000/year for the last 2 years as a speaker for Cenentech. She has been on the Advisory Board for Critical Therapeutics and received $1 0,000 for the last 2 years. She is the Principal Investigator of a trial of anti-tumor necrosis factor a in severe asthma for which she received $150,000. She has been a speaker for Merck for 2 years and received $4,000. She is a speaker for IVAX and received $4,000. The National Jewish Medical and Research Center has a $900,000 relationship with Genentech.

Acknowledgment: The author thanks Dr. Silvana Balzar for her elegant photomicrographs, Ms. Lynn Reed for her technical help, and Drs. Balkisson, Katial, and Sutherland for their scholarly advice.

© 2005 American Thoracic Society Provided by ProQuest LLC. All Rights Reserved.