Article Text

Download PDFPDF

Stability of inflammatory phenotypes in asthma
  1. Ruth H Green,
  2. Ian Pavord
  1. Department of Respiratory Medicine, Glenfield Hospital, Leicester, UK
  1. Correspondence to Dr Ruth H Green, Department of Respiratory Medicine, Glenfield Hospital, Leicester LE3 9QP, UK;{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

While asthma has long been recognised as a heterogeneous disease, recent interest has concentrated on the identification of phenotypes based on the pattern of inflammation in the airways. The application of induced sputum as a non-invasive ‘inflammometer’ has facilitated this process, resulting in the recognition of apparently distinct ‘eosinophilic’ and ‘non-eosinophilic’ phenotypes. The characterisation of patients in this way appears attractive since the response to treatment, particularly with inhaled corticosteroids, has been shown to differ according to the pattern and extent of inflammation. This has contributed to the concept of a ‘holy grail’ of individualised therapy based on phenotypic expression and a flurry of studies aiming to further explain and refine the phenotypic diversity seen in both adults and children with asthma. A number of questions remain, however, and one important one raised by Fleming et al1 is whether there are differences in the nature and significance of airway inflammation between adults and children with asthma.

Adult studies using induced sputum have consistently identified distinct eosinophilic and non-eosinophilic asthma subgroups. While the use of inhaled corticosteroids, which effectively suppress sputum eosinophilia, is a significant confounder, normal sputum eosinophil counts have been reported in up to 25% of adult patients with untreated symptomatic asthma2 and for over 50% of adult patients treated with high doses of inhaled corticosteroids.3 Simpson and colleagues have suggested that airway inflammation in adult asthma could be further categorised into four inflammatory subtypes, namely, neutrophilic asthma (neutrophils >61%), eosinophilic asthma (eosinophils >3%), mixed granulocytic asthma (neutrophils and eosinophils both increased) and paucigranulocytic asthma where neutrophils and eosinophils are both within the normal range.4 In populations of patients with stable adult asthma, the majority treated with inhaled corticosteroids, paucigranulocytic asthma appeared to be the most common inflammatory phenotype followed by neutrophilic inflammation.5 ,6 Non-eosinophilic asthma has also been reported in children with asthma.7 ,8 Paucigranulocytic asthma was the predominant finding in children with stable asthma, but in contrast with adults eosinophilic inflammation was more likely and neutrophilic inflammation uncommon.8 In adults studied during the stable phase, clinical features are similar across the inflammatory phenotypes although sputum eosinophilia appears to predict a greater likelihood of asthma exacerbation9 and non-eosinophilic patients may be more likely to be female subjects and non-atopic than the remaining group.2 Findings in children differ in that the presence of eosinophilic inflammation appears to predict more severe persistent asthma with impaired lung function and increased AHR.8 ,10 Differences in inflammatory phenotypes have also been reported between adults and children presenting with an acute severe exacerbation of asthma with adults being much more likely to have neutrophilic or paucigranulocytic sputum whereas in children the common finding is of eosinophilic or mixed granulocytic inflammation.5

Adults and children with asthma may also differ in the way their inflammatory profile predicts treatment response, particularly response to corticosteroids. In adults, there is a body of evidence to show that non-eosinophilic asthma is associated with an attenuated response to corticosteroids both in the short and long term.11–13 Whether the same is true in paediatric asthma is not clear and there is some evidence to suggest the contrary. Children with difficult asthma given systemic corticosteroids in the form of either oral prednisolone or intramuscular triamcinalone demonstrated similar improvements in FEV1 irrespective of whether they had eosinophilic or non-eosinophilic sputum before treatment.14 Furthermore, in adults, the presence of sputum eosinophilia almost invariably predicts a response to intramuscular triamcinalone (where corticosteroid adherence can be assured)15 but much higher rates of corticosteroid resistance have been reported in children.16

One problem with most studies that aim to characterise inflammatory phenotypes in asthma is that they confine their analyses to cross-sectional data measured at a single interval thus assuming phenotypic stability, a potentially significant limitation given that asthma, by definition, is a variable disease. The study by Fleming et al challenges this assumption by examining the hypotheses that inflammation would be found more frequently in children with severe compared with mild or moderate asthma and that sputum inflammatory phenotypes would be stable in children with asthma. Their findings are notable in that raised levels of inflammatory cells were common across the range of asthma severity but that significant phenotypic variability was seen with 63% of children demonstrating a change in inflammatory phenotype on repeated assessment. This variability does not appear to be due to changes in inhaled corticosteroid treatment, since changes in phenotype were not associated with changes in doses of inhaled corticosteroids although variable adherence in treatment cannot be fully excluded. Fleming et al discusses other possible mechanisms for this phenotypic instability including variations in allergen or viral exposure over time.

So, do these findings provide evidence for yet more differences between adult and paediatric asthma or do they go further and challenge our understanding of the concept of an inflammatory phenotype in asthma altogether? Does the suggestion of phenotypic instability draw into question the utility of inflammometry in individualised asthma management? The available evidence suggests that adult asthma is associated with greater phenotypic stability than that reported by Fleming and colleagues in children. Early studies demonstrated that induced sputum differential cell counts are highly repeatable in the short term in adults with stable asthma with 95% of repeated sputum eosinophil measures lying within a twofold range of the original measurement when samples were taken 6 days apart.17 Simpson et al showed that the absence of a sputum eosinophilia was a consistent finding 4 weeks and 5 months after it was first demonstrated18 and we identified a subgroup of patients with predominantly non-eosinophilic sputum on repeated observations made over a period of 12 months.9 Jayaram et al showed that the pattern of sputum inflammation was similar at baseline and during exacerbations in adults with asthma studied longitudinally for 2 years, indicating that patients with non-eoinsophilic asthma were far less likely to have eosinophilic exacerbations.19 Finally, in a prospective double-blind placebo controlled trial of inhaled corticosteroids in non-eosinophilic asthma patients had a bronchoscopy at baseline and then underwent repeated induced sputum six times over 6 months. None of the 11 patients studied demonstrated an airway eosinophilia at any point and at bronchoscopy all had normal basement membrane thickness.13 This supports the suggestion that the non-eosinophilic phenotype is stable in adults since increased basement membrane thickness has been shown to be a long term marker of eosinophilic airway inflammation.20 The fact that inflammometry using induced sputum has been shown to be a successful strategy to prevent asthma exacerbations in adults9 ,19 but not in children21 may also support the theory that the stability or significance of inflammation in the two groups differ, although there are other potential explanations including a failure to optimally suppress eosinophilic inflammation in the paediatric study.21 Nevertheless it is possible that, given the apparent variability in inflammation over time in children, a management strategy using inflammometry to guide asthma treatment which included more frequent measurements of airway inflammation would yield improved results in a paediatric population.

To conclude, phenotypic analysis using induced sputum does still appear to have value, not least as an inflammometer to guide corticosteroid treatment in adults with refractory disease but question marks remain, particularly in children. Clearly asthma is a complex disease, and attempts to classify it on the basis of a single dimension such as inflammation represent a gross oversimplification. Even those studies which have gone further analysing multiple aspects of the disease using mathematical modelling techniques22 ,23 have not as yet included the dimension of time. Doing so adds yet another layer of complexity but a failure to include longitudinal changes in inflammation and other variables is likely to lead to inaccurate results. As proposed recently by Anderson,24 perhaps we should now target our energies on the search of ‘endotypes’—stable subgroups defined by unique and specific genetic or molecular characteristics rather than ‘phenotypes’ which, defined by biomarkers of disease activity, lead to uncertainty with time and changes in therapy.


View Abstract


  • Linked article 201064.

  • Competing interests None.

  • Provenance and peer review Commissioned; internally peer reviewed.

Linked Articles