Over a decade of careful clinicopathological investigation has characterised the allergen-triggered Th2 response in asthma that leads to eosinophilic airway inflammation. This research has directed drug discovery programmes and we now have effective treatment for most steps in the eosinophilic asthma pathway. This list includes interventions that act at discrete levels such as allergen avoidance, allergen immunotherapy, anti-IgE antibodies, anti-interleukin-5 monoclonal antibodies and leucotriene receptor antagonists, together with corticosteroids that act on multiple levels in the pathway. Despite this significant success in therapeutic discovery, asthma persists. There must be something more to the pathogenesis of asthma. What could it be?

Airway remodelling and non-eosinophilic asthma (NEA) are both topical answers to this question. To date these have been pursued as distinct entities, but the paper by Berry and colleagues1 published in this issue of Thorax (see p 1043) addresses both issues and allows consideration of the interaction and overlap between airway remodelling and inflammatory subtype in asthma.

NEA refers to an asthma subtype where patients exhibit asthma symptoms and abnormal airway physiology (airway hyperresponsiveness (AHR), variable airflow obstruction) in the absence of a significant airway eosinophilia.2 Its importance arises because NEA is common, it seems to have a different pathogenesis from allergen-induced asthma3 and it may be relatively resistant to corticosteroid therapy.4

Airway remodelling in asthma refers to changes in structural components of the airway wall and is believed to result in fixed airflow obstruction, persistent AHR and a poor short-term treatment response. Typical features are fibrosis of the subepithelial basement membrane, mucus gland hyperplasia and changes in airway smooth muscle.

Berry et al investigated the relationship between features of airway remodelling in asthma, where subjects were classified by their underlying inflammatory subtype. When viewed in the context of other studies in chronic cough5 and refractory asthma,6 a consistent pattern begins to emerge.

First, the concept that the eosinophil pathway causes AHR now seems untenable. Several studies clearly show that AHR can occur in the absence of eosinophilic infiltration of the airway mucosa or the airway lumen,1 6 and that eosinophilic bronchitis occurs without accompanying AHR.5 What seems more plausible is that eosinophilic inflammation can modulate the degree of airway responsiveness along a continuum, explaining how steroid treatment and allergen exposure can shift airway responsiveness both within and outside the asthmatic range.7 So what else determines AHR? Remodelled airway smooth muscle looms large here, and the observations of Berry et al1 both confirm and extend recent observations that a mast cell infiltrate into airway smooth muscle is associated with AHR.5 That this is the case in eosinophilic asthma is expected, but what Berry et al show is that, in NEA where eosinophils are absent from both the airway lumen and the airway mucosa, the presence of AHR is accompanied by a mast cell infiltration of the airway smooth muscle.

Second, the concept that a hallmark feature of asthma is increased thickness of the fibrous layer beneath the epithelial basement membrane (subepithelial fibrosis) is also untenable. This feature is not restricted to asthma since it occurs in airway diseases associated with eosinophilic inflammation but not asthma such as allergic rhinitis and cough with eosinophilic bronchitis. In subjects with NEA, the thickness of the subepithelial fibrous tissue was the same as in normal subjects, and the only patients in whom it was increased were those with eosinophilic asthma. Thus, subepithelial fibrosis may be the result of eosinophilic bronchitis, whatever the accompanying clinical phenotype. This is biologically plausible given evidence from animal models and known profibrotic cytokines released by eosinophils.8 That AHR with mast cell myositis is distinct from subepithelial fibrosis indicates a separate regulatory pathway for these two features of airway wall remodelling and suggests that different treatments will be needed for these separate components of airway disease.

Third, the concept that NEA is poorly responsive to corticosteroid treatment is strengthened by the small randomised controlled trial conducted by Berry et al where the response to inhaled corticosteroid was examined with stratification by inflammatory phenotype. The results confirm those of others that eosinophilic asthma is a highly steroid-responsive condition. Importantly, the study provides the first evidence from a randomised controlled trial that NEA is poorly responsive to short-term administration of inhaled corticosteroid. This has important practical implications for the treatment of people with asthma, and provides a pathophysiological explanation for the beneficial effects seen when asthma is managed using sputum cell counts to adjust treatment.9 10 This may also be a useful explanation for the synergy seen when inhaled corticosteroids are combined with long-acting β agonists in asthma. Given the high prevalence of NEA, the combination of inhaled corticosteroids and long-acting β agonists would provide effective treatment for both phenotypes where inhaled steroids treat eosinophilic asthma and long-acting β agonists control the AHR seen in NEA, as well as partially suppressing interleukin-8-mediated neutrophilic inflammation.11

The relative steroid resistance of NEA is still, however, a contentious issue. Uncontrolled studies disagree on whether there is a difference in clinical outcomes after steroid treatment between eosinophilic asthma and NEA. The lack of an effect in uncontrolled studies could be explained by several design flaws that are addressed by using a randomised controlled trial. A larger randomised controlled trial would strengthen the conclusion that NEA is relatively unresponsive to inhaled corticosteroid.

HOW SHOULD NEA BE RECOGNISED?

This is an important question, since the subtype represents the occurrence of asthma symptoms and abnormal physiology in the absence of eosinophilic inflammation. Several variables could confound the diagnosis of NEA, and these relate to causes of transient neutrophilic bronchitis and other diseases associated with neutrophilic bronchitis which include viral infection, bronchiectasis and potentially chronic obstructive pulmonary disease. While these factors will need to be considered in research studies, they may be of lesser importance in clinical practice where the purpose of “inflammometry” is to guide treatment.

Certain asthma triggers, such as rhinovirus infection, can result in a transient neutrophilic bronchitis. This can suppress a sputum eosinophilia, and the sputum eosinophilia returns once the infection has resolved.12 13 Classification of the phenotype during the infection may be erroneous, so the diagnosis of NEA should be made with preferably more than one assessment. Just how many assessments are needed to identify a stable inflammatory subtype requires more work, but will be of immense practical importance. The role of mixed phenotypes (eosinophil and neutrophil) also warrants more attention. These are uncommon in stable asthma but may be more important in complicated asthma.2

WHAT IS THE BEST ASSAY TO RECOGNISE NEA?

At present this seems to be induced sputum. While Berry et al1 show that biopsy or bronchoalveolar lavage could also be used, a similar study by Lemiere et al14 found sputum to be more reliable. The ideal situation would be a marker that can be used to positively identify the eosinophilic and non-eosinophilic subtypes. The fraction of expired nitric oxide (Fe_NO) may be useful in steroid-naïve individuals since Fe_NO levels were normal in NEA and raised in eosinophilic asthma. However, the usefulness of Fe_NO in subtype classification in steroid-treated asthma remains to be determined. The positive identification of NEA requires a marker that is positive in NEA and negative in eosinophilic asthma. Some promising candidates are neutrophil elastase and interleukin-8 which are increased in the neutrophil subtype of NEA.15

In conclusion, the persistence of asthma teaches us that there is much more to learn about the pathogenesis and treatment of this much studied condition. Studying inflammatory subtypes is a useful approach to this problem that is paying dividends both in understanding the mechanisms of persistent asthma and relating these to clinical management.