Structural changes reported in the airways of asthmatics include epithelial fragility, goblet cell hyperplasia, enlarged submucosal mucus glands, angiogenesis, increased matrix deposition in the airway wall, increased airway smooth muscle mass, wall thickening and abnormalities in elastin. Genetic influences, as well as fetal and early life exposures, may contribute to structural changes such as subepithelial fibrosis from an early age. Other structural alterations are related to duration of disease and/or long-term uncontrolled inflammation. The increase in smooth muscle mass in both large and small airways probably occurs via multiple mechanisms, and there are probably changes in the phenotype of smooth muscle cells, some showing enhanced synthetic capacity, others enhanced proliferation or contractility. Fixed airflow limitation is probably due to remodelling, whereas the importance of structural changes to the phenomenon of airways hyperresponsiveness may be dependent on the specific clinical phenotype of asthma evaluated. Reduced compliance of the airway wall secondary to enhanced matrix deposition may protect against airway narrowing. Conversely, in severe asthma, disruption of alveolar attachments and adventitial thickening may augment airway narrowing. The encroachment upon luminal area by submucosal thickening may be disadvantageous by increasing the risk of airway closure in the presence of the intraluminal cellular and mucus exudate associated with asthma exacerbations. Structural changes may increase airway narrowing by alteration of smooth muscle dynamics through limitation of the ability of the smooth muscle to periodically lengthen.
- airway smooth muscle
- lung function
Airway remodelling is a term that indicates changes in the composition, quantity and organization of the cellular and molecular components of the airway wall, alterations considered secondary to chronic injury and repair of the airway epithelial–mesenchymal trophic unit. The epithelial–mesenchymal trophic unit is a concept put forward by developmental lung biologists  to signify that the health of the airway wall depends upon maintenance of normal signalling between the epithelium and the underlying mesenchyme: a basement membrane surrounded by layers of fibroblasts, ECM (extracellular matrix) and smooth muscle cells, interspersed by vessels, neural elements and an immune network. It is proposed that disruption of this trophic state leads to airway remodelling.
Sustained interest in airway remodelling has led to a large body of literature, some of which is speculative in that any observed histological change is called remodelling, without regard to physiology or clinical phenotype. Conceptually, if airway remodelling is driven by chronic inflammatory processes and, in turn, if remodelling is of critical importance for the phenomenon of AHR (airways hyperresponsiveness) (Figure 1), then it has important implications for treatment decisions. This review will first evaluate recent literature documenting histological observations, and then briefly discuss some mechanistic insights, before considering the consequences of remodelling in more depth.
THE EVIDENCE BASE FOR STRUCTURAL CHANGES IN AIRWAYS IN ASTHMA
Figure 2 shows an example of a remodelled airway in severe asthma.
Injury to the epithelium is a common finding in asthma, even in mild disease [2,3]. Epithelial damage is associated with AHR , and in some reports has been observed in almost all subjects with persistent asthma . Extensive denudation of the epithelium has been reported in severe and fatal asthma  and is seen in patients with newly diagnosed asthma , suggesting that mucosal inflammation is an early event. A recent report indicated less epithelial change in severe asthma compared with mild or moderate disease, and attributed this observation to epithelial reconstitution secondary to more treatment of severe disease . Epithelial shedding, as detected by induced sputum or BAL (bronchoalveolar lavage), occurs at a rate 4-fold of that of normal subjects . Many of these cells are ciliated, and seem to be shed preferentially . Although epithelial disruption can be attributed to post mortem or biopsy artefacts , the increased expression of CD44+ cells  and epithelial growth factor receptor  in areas of damage, as markers of repair, indicates that epithelial change does occur in vivo.
Changes in the mucus-secreting structures
A consistent association between asthma and goblet cell hyperplasia has been shown. Neither the distribution of mucus expression nor the degree of goblet cell hyperplasia, however, correlates with asthma severity [14,15]. In both severe  and fatal [15,16] asthma, increased submucosal mucus gland area in large airways has been documented. There is no information on the relationship between either the degree of goblet cell hyperplasia or mucus gland area and airways responsiveness.
An increase in the thickness of the layer immediately beneath the true basement membrane, termed here subepithelial fibrosis, has been an observation considered invariable in asthma and an early change from the normal state [17,18]. The thickness of the subepithelial layer increases with severity of the disease , but is not clearly related to duration of disease . It is probably not a change specific to asthma, since a recent report indicates that in severe COPD (chronic obstructive pulmonary disease) , but not mild-moderate COPD , subepithelial fibrosis also occurs. It has also been documented in perennial rhinitis [18,23] and eosinophilic bronchitis , which occurs without AHR. Proteoglycan and other matrix changes throughout the airway wall are well documented [7,25,26] in asthma. The principal ECM chondroitin/dermatan sulphate proteoglycans include members of two gene families; the large aggregating proteoglycans or lecticans, of which versican is the principal example, and the small leucine-rich proteoglycans, principally decorin and biglycan , but also including lumican. In asthmatic airways, multiple investigators (for review, see ) have indicated that (i) the sub-basement layer shows increased levels of collagens 1, 3 and 5 as well as fibronectin and tenasin, (ii) the submucosa shows increased levels of collagens 1, 3 and 5, fibronectin, versican and hyaluronan, and decreased elastin, and (iii) the muscle layer shows increased levels of collagen 1, fibronectin, versican and hyaluronan. Other abnormalities of airway matrix deposition in asthma include an increase in specific isoforms of the structural glycoprotein laminin that are associated with tissue injury  and the overexpression of lumican and decorin . Cartilage degeneration has been documented in more severe asthma, with evidence of loss of the ECM surrounding the chondrocytes [26,30].
Abnormalities in elastin have been reported in asthmatic airways. Bousquet and co-workers  reported electron microscopic evidence of fragmented elastin. Mauad and co-workers  reported decreased elastin in the immediate subepithelial layer, with increased but fragmented elastin at a deeper submucosal level. Changes were more apparent in central airways. Mauad et al.  later documented decreased alveolar attachments in non-respiratory bronchioles in severe asthma. Carroll and co-workers  reported that the submucosal network of elastic fibres, which exists in a collagen and myofibroblast matrix to form discrete LB (longitudinal bundles), was increased 2-fold in both large and small airways in cases of fatal asthma. They put forward that the LB may affect airway function by altering the mechanical properties of the airway wall or by changing the folding behaviour of the airway mucosa. Supporting this idea, it was noted that the number of mucosal folds was related to the number of LB in asthmatics and non-asthmatics, but was not different between groups.
Activated fibroblasts or myofibroblasts may be involved in the deposition of ECM and formation of airway fibrosis in asthma. Collagen and myofibroblasts are increased in LB in asthma , and the number of fibroblasts in the subepithelial compartment has been reported to be increased in severe asthma compared with milder disease and controls, correlating with collagen deposition .
Increased smooth muscle mass
Increases in smooth muscle mass in asthma have been shown by many investigators, due to both increased myocyte mass and increased bundle size secondary to increased matrix around cells. Multiple mechanisms could be responsible for the increase in smooth muscle noted in asthma (Table 1). Estimates of the increase in smooth muscle mass in non-fatal asthma range from 50–200% and in fatal asthma from 200–400%. However, these estimates combine studies in which the increased matrix within smooth muscle bundles has not been clearly delineated from the increase in cellular components. When this distinction has been made , individuals with long-standing asthma had a 4-fold increase in smooth muscle myocyte mass in small airways compared with age-matched controls, whereas younger individuals with a shorter duration had only a 2-fold increase (Figure 3).
Several investigators have attempted to distinguish between hypertrophy and hyperplasia to explain the increased myocyte mass. Ebina and co-workers  using a morphometric technique showed some individuals with asthma had central airway hyperplasia only, with no small airway changes, whereas others had diffuse large and small airway hypertrophy and only minimal central hyperplasia. A clinical correlate with these two subtypes was not elucidated. Woodruff  evaluated bronchial biopsies from large airways in mild-to-moderate asthma using precise morphometric techniques and reported no evidence for hypertrophy, but clear evidence of more cells being present within smooth muscle bundles. In contrast, Benayoun and co-workers  showed no evidence of hyperplasia in asthmatic bronchial biopsies, but a progressive increase in cell width as asthma became more severe. There was no evidence of increased markers of smooth muscle cell proliferation, the latter a finding in common with several published  and unpublished reports; also noted was a decrease in the distance between the epithelium and the smooth muscle layer.
There is evidence for angiogenesis and vascular remodelling in many inflammatory diseases, including asthma [37,38]. Although some reports suggest that it is the size and not the density of vessels that is increased in asthma, or that vessel changes are only proportional to the increased thickness of the wall [22,39], the weight of evidence favours greater vessel density with vessels of same size on average. The density of vessels positively correlates with the number of mast cells . Substantial overlap exists between normal and asthma, however, and some of the conflicting literature may arise because bronchial biopsy studies evaluate the bronchial circulation in large airways, whereas the peripheral airways evaluated via autopsy or resection are more under the influence of the pulmonary circulation.
Thickening of the airway wall
Multiple studies have documented thickening of the airway wall in asthma. Changes in both inner and outer wall area have been noted. Some changes are probably transient, related to oedema, cell infiltration and the effect of acute mechanical strain induced by a fatal acute attack. Others changes are probably more permanent, related to matrix deposition and increased muscle mass, as discussed above. Wall thickening increases with both disease severity [22,40] and duration of asthma . In the latter report, although the inner wall area was also increased, the adventitia was the predominant site of wall thickening in small airways, being 3-fold greater than controls in disease of long duration.
Changes in small versus large airways
The small airways are typically defined as those less than 2 mm internal diameter. In COPD, there is clear evidence for the small airways being the site of airflow obstruction . In asthma, similar evidence exists , although the matter is more controversial with some evidence for airway closure occurring in medium-sized airways .
The components of the peripheral changes in asthma have been detailed by many investigators [35,40,44–54]. Carroll and co-workers , comparing lungs from patients with fatal and non-fatal asthma with lungs from normal subjects, demonstrated that the relative changes in the airway wall components varied with the size of the airway examined. The cartilaginous airways from non-fatal asthmatic samples were similar to control lung tissue in total airway wall thickness and area of smooth muscle. However, considering airways with a Pi (internal perimeter) <10 mm, the inner wall area (excluding adventitia) was greater in the non-fatal asthmatic lung than in the controls. In addition, airways with Pi between 10 and 18 mm demonstrated increased inner wall thickness in the tissue from patients with fatal compared with those with non-fatal asthma. Carroll et al.  showed similar degrees of inflammation in all airway sizes in asthma, with more eosinophils in fatal disease. Carroll and co-workers  also showed that the elastin LB were increased in both large and small airways in asthma (see above). Haley and co-workers  quantified, in autopsy samples of severe asthma, the cell density of leucocytes and eosinophils in airways greater or less than 3 mm Pi and showed more adventitial eosinophils and leucocyte infiltration in the smaller airways. This abnormality was not present in severe cystic fibrosis, suggesting biological significance specific to the pathogenesis of severe asthma. These patients had been hospitalized for a period of time, as opposed to most studies in fatal asthma, so there may be confounding factors related to mechanical ventilation etc. Synek and co-workers  reported that large airways in fatal compared with non-fatal asthma had less T-lymphocytes and more eosinophils, whereas the peripheral inflammation was similar in both asthma groups. Faul and co-workers  noted relatively greater inflammation in proximal airways in fatal asthma and based on these findings, as well as relatively normal BAL findings in asthma, made the argument that inflammation is more central. In contrast, Hamid and co-workers  showed evidence for inflammation in both large and small airways in non-fatal asthma, with more activated eosinophils in the small airways, albeit predominantly in the adventitia. T-cell numbers were equally distributed between large and small airways, but these data have the potential for being confounded by concurrent smoking in these subjects. Balzar and co-workers  showed distal airways had more inflammatory cells than proximal airways in severe asthma, with the density of the chymase-positive mast cells in small airways correlating with better lung function , suggestive of a reparative function for these cells.
Do different asthma phenotypes show pathological differences?
There has also been substantial recent interest in phenotypic differences identified by induced sputum studies. Some investigators  have reported 30–50% of asthmatics in specialist clinics have persistently neutrophilic disease, raising the possibility of underlying differences in structural features. The results of some pathological studies of airway structure in asthma of different origins, for example occupational asthma induced by cedar dust compared with atopic asthma, do not suggest major differences . However, other biopsy studies of atopic compared with non-atopic asthma do suggest a more neutrophil-predominant inflammatory pattern in non-atopic asthma, associated with less deposition of laminin and tenascin and less epithelial disruption in the non-atopic group . As discussed above, different patterns of smooth muscle change have been reported in one study. Wenzel and co-workers , and others , have reported a subset of patients with severe asthma to have neutrophil predominant disease, although the extent of differences in structural features between groups is unclear. Wenzel and co-workers  do, however, report a subgroup with late onset asthma without subepithelial fibrosis on bronchial biopsies . However, given the problems with an exact definition of asthma, as well as prior treatment effects, the results of some biopsy studies in specialist clinics may not reflect the pathology of the vast number of the asthmatic population. The likelihood is that a considerable degree of so-called phenotype heterogeneity is artefactual and we submit, as do others , that it is premature to accept such groupings.
Although neuronal plasticity, i.e. change in the phenotype of the neural elements within the airway wall, is potentially a very important topic within the general theme of this review, the available literature is sparse and conflicting and will not be evaluated further here.
POTENTIAL MECHANISMS OF STRUCTURAL CHANGES
Space constraints do not allow a full discussion of potential mechanisms in detail. Genetic variants influencing asthma susceptibility or severity that may be involved in structural changes include mucin genes, ADAM33, TGFβ (transforming growth factor β), IL (interleukin)-13 and the mast-cell-associated factors FcεRI-β, PGD2 (prostaglandin D2) receptor and prostanoid DP synthase. For example, investigations of Dutch populations have shown ADAM33 polymorphisms are associated with substantial lung function decline in normal [64a] and asthmatic populations , as well as AHR in asthmatics . Evidence supports a role in remodelling for circulating inflammatory cells such as eosinophils [66–68], lymphocytes [69,70] and neutrophils . Direct interactions between elements of the epithelial–mesenchymal trophic unit not involving circulating cells may also lead to structural change [72–74]. Finally, local (mechanical stretch) and systemic factors (eosinophil products) may interact to produce remodelling changes .
Increased smooth muscle may be primarily through reduction in apoptosis, with a minor component of proliferation . Rates of apoptosis of human airway smooth muscle cells have been shown to be under the influence of cytokines, with members of the GP130 (IL-6) family inhibiting apoptosis . Hypertrophy of human airway smooth muscle cells can be induced in vitro by endothelin, GP130 cytokines, such as CT-1 (cardiotrophin 1) and oncostatin M, cell-cycle inhibitors, such as SV40, and PI3K (phosphoinositide 3-kinase) activators [77–79]. CT-1 can induce hypertrophy in the tissue milieu as well , along with increased matrix. Circulating CD36+ fibrocytes  may be recruited to the airway wall and differentiate into smooth muscle cells, thus increasing smooth muscle bulk.
It has been hypothesized that smooth muscle remodelling and other structural changes associated with asthma begin with the disruption of normal developmental processes following exposure to allergens and other airway insults such as pollutants during intrauterine and early postnatal life when the lungs are undergoing active growth and differentiation. The development of increased smooth muscle mass following postnatal exposure of infant Rhesus monkeys to allergen lends support to this idea . The gene ADAM33 is highly expressed in the mesenchyme of the developing bronchial tree , and perhaps the primary influence of genetic variants such as this is to ‘set up’ the airways, possibly via altered airway branching patterns, to react in different ways to exposures in later life.
FUNCTIONAL CONSEQUENCES OF STRUCTURAL CHANGES
The role of structural change in determining airflow obstruction, AHR and the pathogenesis of a severe episode of airway obstruction may be more important in more severe disease or older asthmatics with longer duration of disease, given that some structural changes in asthma are prominent only in more severe disease or asthma of a longer duration . Functional data in support of this view, as well as other perspectives, are provided in the following sections.
HRCT (high-resolution computerized axial tomography) correlations of wall thickening and outcomes
HRCT shows promise in evaluating structure–function relationships in asthma. After careful evaluation and standardization of the technique , HRCT has revealed an increase in the ratio of bronchial wall thickness to bronchial diameter in large airways in asthma. Kasahara and co-workers  reported that, after 2 weeks of oral steroids to minimize reversible changes in airway wall thickness, wall thickness as assessed by HRCT correlated with subepithelial fibrosis as assessed by biopsy. In addition, increased wall thickness was associated with increased airflow limitation. Interestingly, increased wall areas have been detected by HRCT without a decrease in luminal area, in distinction to morphometric differences in airway calibre in some severe asthmatics compared with controls (Figure 4) . This variance may be due to edge-averaging artefacts in wall measurements when using HRCT. Wall thickening can persist in the absence of clinical symptoms and shows an improved correlation with airflow limitation after treatment with inhaled steroids , suggesting that part of the HRCT-detected thickness is due to labile changes such as oedema or cell infiltration and part is due to more persistent structural change. HRCT wall thickness correlates with asthma duration [87,88], as well as conventional indices of asthma severity and asthma control . Paradoxically, although HRCT wall thickness increases with clinical severity indices, recently, in response to bronchoprovocation, wall thickness has been shown to inversely correlate with a pulmonary function measure of the severity of airway narrowing known as reactivity . This finding needs confirmation, but may be related to decreased wall compliance (see below).
AHR including the deep breath response
An exaggerated response to stimuli causing airway narrowing is a cardinal feature of asthma. Longitudinal studies [90–92] suggest that AHR is best regarded as an independent risk factor for the development of asthma and not an outcome itself. Genetic influences have been detected, e.g. ADAM33; in addition, the risk of persistent AHR is related in longitudinal studies to total IgE in children  and eosinophilia in adults , revealing the probable partial dependence of AHR on inflammatory events. Gronke and co-workers  show in adult asthma of varying disease duration that, in long-standing disease (>16 years), AHR is associated with impaired lung function, whereas in earlier disease it is more associated with markers of airway inflammation.
The airway response to contractile stimuli in vivo is a complex outcome, the net result of a large number of factors opposing or enhancing airway narrowing [95,96], including dynamic contractile and non-contractile elements. In considering how remodelling may relate to AHR, the primary site of airflow resistance needs to be kept in consideration. In reality, the primary site of airflow resistance and hyperresponsiveness in asthma remains controversial, as discussed above. We note that peripheral resistance is abnormal even in mild asthma [97,98], as well as moderate and severe asthma . Furthermore, the larger part of total resistance increase in asthma was apportioned to the peripheral airways by direct measurements by Takishima and co-workers .
Relationship between responsiveness and remodelling
Earlier studies of the consequences of remodelling for airways responsiveness suggested that remodelling alone could account for most of the observed changes seen in asthma without the need for increased smooth muscle contractility . The conclusion reached was that the increase in smooth muscle mass was the most important change, and that the small airways changes were more important than large airway changes. Recent biomathematical modelling based on mice models of allergic AHR re-emphasize the importance of remodelling of noncontractile elements in determining AHR; for example, increased epithelial thickness was key in one model  and airway liquid surface forces and luminal material in another . However, contractile structures are influential in other murine studies: smooth muscle velocity of shortening explains the differences in airways responsiveness between inbred mouse strains . Airway smooth muscle shortening is very unlikely to be uniform, based on induced bronchoconstriction studies using HRCT  or inhaled radioactive materials , as well as observations on asthmatic airways postmortem (T. R. Bai, unpublished work). Computer modelling [105,106] suggests that airway wall remodelling serves to predispose the lung to a more heterogeneous pattern of peripheral airway constriction and this, in turn, is a crucial determinant of increased responsiveness. Moreover, extensive remodelling increased the likelihood of random airway closure or near closure in this model. Given that the thickness of the subepithelial fibrotic layer is reported to correlate with wall thickness [107,108], the finding that subepithelial fibrosis can occur without AHR [23,24] would seem to cast doubt on the hypothesis that remodelling alone is a sufficient cause of persistent AHR. However, it is clear that measurement of subepithelial thickness is only a partial estimate of remodelling, as discussed above.
Another interesting concept by which structural changes could influence airways responsiveness is through modification of smooth muscle dynamics, the phenomenon of length adaptation or mechanical plasticity [109,110]. By reducing the forces periodically stretching (and lengthening) the muscle, wall thickening, decreased elastic recoil and increased wall matrix could allow the muscle to chronically adapt to a shorter length, resulting in increased shortening and luminal closure upon stimulation of the airway.
The difference between asthmatic and normal individuals in the effect of deep inhalation on bronchoconstriction – deep inhalation often substantially reverses induced bronchoconstriction in normals, whereas it often has much less effect on spontaneous or induced bronchoconstriction in asthmatics – may also be due to abnormal dynamic aspects of airway smooth muscle contraction secondary to structural change.
In support of this hypothesis, Slats and co-workers  noted that after allergen challenge the degree of versican deposition in bronchial biopsies, a marker of altered mechanical wall properties, was associated with a change in the M/P ratio (maximal to partial flow ratio) in asthma. The M/P ratio is a marker of abnormal airway behaviour after deep inhalation. AHR also correlates with the degree of proteoglycan deposition both in vitro  and in vivo . Thus the deposition of proteoglycans and glycosaminoglycans may contribute to altered mechanical properties, for example limiting smooth muscle shortening, and may also promote airway obstruction acutely through the hydroscopic properties of matrix components such as hyaluronan.
The weight of evidence suggests that airway segments isolated from asthmatic tissues exhibit normal sensitivity to constrictor agonists when studied during isometric contraction, but the increased muscle mass within asthmatic airways might lead to more total force generation . Furthermore all the studies performed to date have not controlled for the phenomenon of length adaptation. In distinction to these findings, Ma and co-workers  have shown increased absolute shortening potential and velocity of shortening, as well as increased MLCK (myosin light chain kinase) gene expression, in asthmatic cells isolated from biopsies. Ex vivo studies of airway segments from hyperresponsive animals with structural changes induced by chronic allergen challenge have, paradoxically, sometimes shown decreased contractility or maximal shortening compared with controls [115,116]. Consistent with this, in an explant model in which smooth muscle hypertrophy and increased matrix was induced by cytokines, we showed decreased contractile responsiveness . These in vitro studies provide rationale for some of the in vivo finding. Thus Milanase and co-workers  report that in asthma, based on physiological data obtained during bronchoprovocation in asthmatics, the degree of subepithelial fibrosis was positively correlated with sensitivity to a bronchoconstrictor agent, whereas reactivity (severity of airway narrowing) was negatively correlated. These data would be consistent with some of the in vitro findings above as well as those of Niimi and co-workers  discussed above, showing the complexity of the effect of remodelling on airway function; some features contributing to AHR, some opposing it.
The increased matrix component around smooth muscle cells and in the submucosa may lead to persistence of airway inflammation and hence chronic AHR through greater retention of cytokines and chemokines, thus attracting immune cells . Closer approximation of the contractile machinery to the epithelial layer may also augment responsiveness by deceasing the distance inflammatory and other luminal stimuli must travel to induce contraction. In severe asthma, Benayoun and co-workers  have observed this with measurement of a decreased epithelial–muscle distance (67 μm compared with 135 μm in controls). The possibility that the thickened sub-basement layer alters water or heat flux across the epithelial layer and thus contributes to, for example, exercise-induced airway narrowing by increasing surface osmolarity, requires further investigation.
Decreased airway distensability in asthma was reported by Wilson and co-workers in 1993 . Airway distensability can be measured by nitrogen washout techniques, as change in anatomic dead space after inhalation of a bronchodilator , or by estimating airway mechanics through measurement of pleural pressure and airflow . A weak inverse correlation (decreased distensability) with increasing subepithelial fibrosis has been reported. Thus Ward et al.  concluded that fibrosis leads to decreased compliance. Brackel and co-workers  also showed abnormal (decreased) airway compliance in asthma and provided evidence that compliance was reduced more in the lower lobe compared with the trachea: some evidence that the structural change may be greater in medium-sized compared with large airways. The cartilage changes documented in several reports could be expected to increase the collapsibility of central airways and thus augment airflow limitation. It is conceivable that this happens in a subset of severe asthmatics not yet studied in airway compliance experiments. Taken together, however, the data suggest that the stiffening effect of increased collagen and soluble matrix components must out-weigh the effect of elastin and cartilage abnormalities, which would be expected to increase distensability.
Consequences of loss of alveolar attachments or decreased recoil
Structural changes such as loss of alveolar attachments in severe asthma, as shown by Mauad and co-workers , by reducing the tethering or preload imposed by these elastic attachments on smooth muscle can have profound effects on both AHR and the deep breath response. Colebatch et al.  many years ago detailed the catastrophic airway narrowing that could occur if loss of recoil from emphysema coincided with asthma. Gelb and co-workers  and Sterk and co-workers  in patients with severe asthma and without emphysema demonstrated a similar propensity in vivo in some individuals. Gelb et al.  documented abnormalities in elastic recoil and identified these changes as risk factors for near-fatal attack, as well as persistent reduction in FEV1 (forced expiratory volume in 1 s). Furthermore, the loss of lung elastic recoil in asthmatic patients was associated with increased age, longer duration of disease and progressive expiratory airflow limitation.
Consequences of mucus exudate
The intraluminal contents in asthma are a mixture of mucus, inflammatory and epithelial cells, and plasma exudate . Mucus gland hypertrophy and goblet cell hyperplasia, clearly documented in asthma, may have major consequences for airflow . Modelling studies have shown mucus exudate can also contribute to AHR in stable asthma (T. R. Bai, unpublished work), and mucus is seen in small airways even in stable mild disease . The consequences of the increase in airway vascularity documented in asthma on AHR and airflow limitation are uncertain. Some experiments do not suggest a major component of AHR is due to vascular leak, even in the presence of enhanced permeability . However, a recent report  re-emphasizes the adverse consequences of airways plasma protein leak on AHR and the effect of changes in surface liquid forces in enhancing airway closure may have been underestimated.
Irreversible airflow limitation
Unlike COPD, asthma is not a disease normally associated with either irreversible airflow limitation or an accelerated decline in FEV1. Treatment–resistance impairment in airflow at diagnosis is usually inferred as structural change in the presymptomatic phase and may or may not be progressive. An accelerated longitudinal decline in airflow may reflect ongoing structural changes. There are, however, problems in interpreting measures of fixed airflow obstruction in many of the reports in that most airflow measurements are made without a maximal corticosteroid course (beyond the usual 0.6 mg/kg of body weight of prednisone for 2 weeks), such as intramuscular triamcinolone [24,64], to eliminate reversible causes of airflow limitation in those with relative corticosteroid resistance. Even with such treatment, the airways of some steroid-resistant patients may still contain labile causes of airflow reduction. The robust measures FEV1 or FEV1/forced VC (vital capacity) ratio are the usual outcomes in long-term studies, but may not adequately reflect the function of peripheral airways that may be subject to remodelling changes independent of large airways. To date, longitudinal studies have not been able to adequately address this possibility . Although it has been established that AHR predicts decline in lung function in adult life [91,127], the physiological parameters AHR and FEV1 are interrelated; reduced FEV1 is a predictor of increased AHR through simple geometry, and it is thus difficult to assess the contribution of AHR independently of airflow measures. A review of the literature and clinical experience, suggests that the majority of individuals with the classic asthma phenotype will not develop clinically relevant airflow limitation, with perhaps the exception of, on average, some decrease in life expectancy (see below).
A concern in childhood asthma is that the disease adversely impacts on the growth of a child's airways such that maximal lung growth is not achieved. However, a lack of clinically significant airflow obstruction in the majority of subjects, in terms of predicted ongoing symptoms such as regular exertional breathlessness, rather than episodic asthma-related respiratory symptoms, is evident from longitudinal population-based cohorts of subjects with asthma beginning in childhood and studied for 3 or 4 decades, such as the cohorts in Dunedin, New Zealand , Melbourne, Australia  and The Netherlands . Rasmussen and co-workers  analysed lung function in a longitudinal cohort and reported that a low post-bronchodilator FEV1/VC ratio was found in 7.4% of asthmatics at 18 years of age, 6.4% at 26 years of age and in 4.6% at both assessments. Lung function was low throughout childhood in those with a consistently low post-bronchodilator FEV1/VC ratio at both ages. This subgroup also showed a greater decline in the prebronchodilator FEV1/VC ratio from 9–26 years of age compared with those with normal ratios at both ages (males, −12% compared with −6%; females, −10.5% compared with −5.5%). In the population overall, asthma, male sex, AHR and low lung function in childhood were each independently associated with a low post-bronchodilator FEV1/VC ratio, which, in turn, was associated with an accelerated decline in lung function and decreased reversibility . The authors concluded that these data suggest that airway remodelling in asthma, as manifested by impaired lung function, begins in childhood and continues into adult life. Gilliland and co-workers  showed a significant effect of early onset asthma on FEV1 at 15 years of age, being 84 ml less in females and 150 ml less in males. The CAMP study of childhood asthma [132,133] showed a relationship between asthma duration and loss of lung function: prebronchodilator FEV1 declined 0.91% predicted/year of asthma duration. Asthma duration was also associated with enhanced responsiveness. Similar findings were reported by Agertoft and Pedersen  in a cohort of children with asthma who did not receive inhaled budesonide over a 1–7-year period. In this group of children, a predicted decline in FEV1% of 1.2–3.1% per year was noted. A later report by the CAMP investigators  shows that 25% of the cohort showed a decline in lung function over time, but that the majority of this was exhibited by those with early onset disease (onset <7 years of age)]. However, Grol et al.  did not show a relationship between duration and decline in a Dutch cohort. A later analysis  in the same cohort showed subjects with a lower level of lung function in childhood and a smaller subsequent growth of lung function were likely to have persistent asthma at 32–42 years of age. The authors concluded that these early life deficits in lung function may be the result of various factors, including in utero development and early childhood exposures, but may also reflect airway remodelling shortly after the start of the disease. However, several birth cohort analyses have suggested airflow limitation is not present at birth but develops thereafter in those susceptible to asthma [136,137], although AHR (which may indicate remodelling) was a risk factor at 1 month of age for subsequent asthma  in one study.
The available studies are summarized in Table 2. These data mix those treated with anti-inflammatory drugs with those untreated. Lower lung function in young adults with diagnosed or undiagnosed asthma compared with healthy control subjects is seen at enrolment in several studies [138,139]. Apostol and co-workers  report that FEV1 was 230 ml (6.4%) less than expected at 18 years of age in non-smoking asthmatics. James and co-workers  report at enrolment FEV1 was 7.4% less in asthmatic females and 8.7% in asthmatic males, although values for smokers are not differentiated from non-smokers. Lange and co-workers  also showed reduced lung function at enrolment compared with controls, but again values in non-smokers are not segregated from smokers. An age-related accelerated decline in FEV1 in adult asthmatics was shown in this large population survey: between 40 and 59 years of ages, female asthmatics had an excess decline of 13 ml/year, whereas in males this was 9 ml/year. Apostol and co-workers  report an excess decline of FEV1 of 1.6% between the 20 and 40 years of age in never-smoked asthmatics, not subdivided by gender. James and co-workers  showed that rural Australian non-smoking asthmatics had a greater decline in FEV1 over two to three decades (females, 3.78 ml/year; males, 3.69 ml/year). In this cohort, at 60 years of age, a non-smoking adult asthmatic would be expected to have an FEV1 380 ml lower than normals, with a female asthmatic 200 ml less. In another cohort of adult ‘mild-to-moderate asthmatics’, 23% exhibited an abnormal rate of decline and progressed to ‘clinically important’ lung function impairment at 10 years follow-up . Adult subjects with asthma do not always have a more rapid rate of decline, but can still show lower lung function at enrolment ; presumably reflecting a phase of rapid decline just prior to being assessed or perhaps low lung function from birth. The former theory is supported by the findings of Ulrik and Lange  who, in the Copenhagen City Heart Study, found a higher rate of decline in FEV1 in subjects with new asthma. Sherrill and co-workers  have argued that the influential data of Ulrik and Lange  is confounded by a lack of discrimination between asthma and COPD, although this is probably not a justifiable criticism in life-long non-smokers. A subgroup (16%) of a Dutch adult asthma cohort, with a median age of onset of asthma of 4 years, had an FEV1 <80% predicted 26 years later . Those with less AHR initially were more likely to be irreversible at follow-up, possibly a reflection of decreased wall compliance. In other words, advanced remodelling may identify a risk factor for decline at the same time as it identifies those with less exaggerated airway narrowing .
ten Brinke and co-workers  showed adult onset asthma was associated with more persistent airflow obstruction. In a cross-sectional analysis of mild, moderate or severe asthmatics compared with COPD patients and normals, Benayoun et al.  reported that fibroblast numbers and smooth muscle cell size were associated with more airflow obstruction. Bumbacea and co-workers  evaluated a spectrum of asthmatics and showed reduced lung function was associated with older age, longer duration, HRCT evidence of wall thickening and dilatation, and higher indices of airway inflammation.
Exacerbations and lung function decline
It has been proposed that exacerbations represent periods of enhanced remodelling, in part because of evidence of enhanced expression of MMPs (matrix metalloproteinases) associated with fibrosis during exacerbations . Hence frequency of exacerbations should be associated with rate of decline, but clearly this idea requires further study. A review of the available literature, two studies in childhood, do not show a clear relation [133,146], suggesting that, although low initial lung function at diagnosis of asthma is in itself a predictor of exacerbations, there is no evidence of accelerated decline in the group with low lung function compared with those without exacerbations. In other reports, there is insufficient information to distinguish whether it is chronic severity that dictates decline or exacerbation frequency.
FEV1 is a powerful predictor of general, pulmonary and cardiovascular mortality; thus in reality the major effect of the substantial decrement in predicted FEV1, at e.g. 60 years of age, in non-smoking asthma cohorts may be the key outcome to consider. Lange et al.  showed in the Copenhagen City Heart Study a doubling of all- cause mortality, mainly due to pulmonary causes, in life-long non-smokers with asthma after 17 years follow-up. Similarly Hansen and co-workers  showed that lower maximal lung function, after reversibility with bronchodilator or steroids, was associated with increased all-cause mortality over 10 years.
Attempts to reverse or prevent structural changes
Animal studies  have suggested that structural changes induced in the airways of allergen-challenged rodents can be prevented by anti-inflammatory treatment administered before challenge, but not afterwards. Furthermore, assuming that substantial structural change occurs before diagnosis (see above), treatment may never be early enough. Nevertheless, treatment-related decreases in subepithelial thickness have been considered responsible for decreases in AHR [149,150]. Thus treatment with high dose of ICS (inhaled corticosteroids) for 1 year lead to substantial and possibly progressive (ongoing) improvements in AHR . Ward and co-workers  attributed two-thirds of the improvement in AHR to a reduction in remodelling and one-third to a reduction in inflammation. Other, shorter studies of the influence of ICS on remodelling and AHR include that by Boulet et al.  showing that 8 weeks of high-dose ICS did not change type 1 and type 3 collagen deposition in either early or established asthma, but improved AHR to a similar degree. In contrast, Chetta et al.  reported a dose-dependent decrease in subepithelial thickness and vascular area after 6 weeks of treatment with 1000 μg/day fluticasone, but not 200 μg/day. The subepithelial thickness reduction was similar to that reported by Ward and co-workers , who had treated for a much longer period.
The very large START study  showed a benefit of early low-dose ICS in terms of FEV1 decline. In 7000 newly diagnosed mild asthmatics (3000 between 5 and 15 years of age), over 3 years in the placebo group there were significant declines in post-bronchodilator FEV1 of 2.3% in 6–10 year olds and 3.6% in adults; this decline was reduced by 22% by ICS in children and 42% in adults. Haahtela and co-workers  showed a 2 year delay in introduction of ICS in new onset adult asthma was associated with less improved airflow, and Overbeek et al.  showed less improvement in AHR (but not airflow) with a 2.5 year delay in initiation of ICS. It is also important to note that the effect of treatment, being reversible within 6 months in children ( and references within), may reflect an effect of ICS on inflammation, not persistent structural changes. Thus such data cannot be used as conclusive evidence of reduction in remodelling, as the effect may simply be a transient genomic or non-genomic benefit of ICS.
There is now a substantial body of evidence documenting structural changes in the airways of asthmatics. Some structural changes in asthmatic airways, for example subepithelial fibrosis, occur early, whereas others are related to duration of disease and/or long-term uncontrolled inflammation. Genetic influences, as well as fetal and early life exposures, may contribute to structural changes from an early age. Decreased wall compliance and fixed airflow limitation are probably due to remodelling, but the incidence of irreversibility due to structural change may have been overestimated because of submaximal therapy preceding lung function assessment. Nevertheless, with acceptable therapy in terms of long-term safety (e.g. stable dosages of ICS <1000 μg/day beclomethasone dipropionate or equivalent), a subpopulation with fixed airflow obstruction will be observed. There is some evidence that this risk could be reduced by earlier use of ICS [152,153], but the necessity of such intervention is still open to question. Structural changes may be of more importance for the phenomenon of AHR; however, this is probably dependent on the specific clinical phenotype of asthma evaluated. Furthermore, reduced compliance of the airway wall secondary to enhanced matrix deposition may protect against airway narrowing. Conversely, in severe asthma, disruption of alveolar attachments and adventitial thickening may augment airway narrowing and the encroachment upon luminal area may be disadvantageous by increasing the risk of airway closure in the presence of the intraluminal cellular and mucus exudate associated with asthma exacerbations. The contribution of structural changes to altered airway smooth muscle dynamics (length adaptation or mechanical plasticity), increasing airway narrowing by limiting the ability of the smooth muscle to periodically lengthen, requires further study.
We acknowledge the contribution of numerous co-workers, particularly Dr James Hogg, Dr Peter Pare, Dr Nick ten Hacken and Dr Chun Seow, to the concepts contained within this review. We also thank Professor Dirkje Postma for comments on the manuscript. This work is supported by Canadian Institutes of Health Research grant #42537 to T.R.B., and was completed during the tenure of a UBC Killam Fellowship, awarded to T.R.B., at the Groningen Research Institute for Asthma and COPD, University of Groningen, Groningen, The Netherlands.
Abbreviations: AHR, airways hyperresponsiveness; BAL, bronchoalveolar lavage; COPD, chronic obstructive pulmonary disease; CT-1, cardiotrophin 1; ECM, extracellular matrix; FEV1, forced expiratory volume in 1 s; HRCT, high-resolution computerized axial tomography; ICS, inhaled corticosteroids; IL, interleukin; LB, longitudinal bundles; Pi, internal perimeter; VC, vital capacity
- The Biochemical Society