Asthma and chronic obstructive pulmonary disease (COPD) are both common, heterogeneous and usually distinct airway diseases that sometimes overlap in a particular patient. This entity, the so-called asthma–COPD overlap (ACO)1 has attracted attention and triggered debate in recent years as evidenced by the proliferation of reviews and editorials dedicated to the topic.2–7 GOLD and GINA define ACO as persistent airflow limitation with several features usually associated with asthma and other usually associated with COPD, and specifies that ACO encompasses different phenotypes that are likely clinical expressions of distinct underlying mechanisms.8 Although this approach is intuitive, it is also quite imprecise because it does not take into account the relevance of each criterion for the diagnosis of ACO and it may not be useful in daily clinical practice.

ACO is difficult to define due to the lack of understanding of the underlying inflammatory mechanisms and, nowadays, it is an umbrella term that encompasses patients with COPD and eosinophilic inflammation, and smoking asthmatics with irreversible airway obstruction. In this climate of uncertainty, it is not surprising that the diagnosis of this entity remains elusive, with no defined clinical or functional criteria universally accepted. We have recently shown that ACO's manifestations are somewhere in between COPD and asthma, since these patients showed analogous demographic and inflammatory characteristics to those with asthma and functional impairment and the presence of comorbidities similar to those included in the COPD group.9

In this context, the Spanish Society of Pneumology and Thoracic Surgery (SEPAR) recently developed a new strategy for the identification of ACO by incorporating the new evidence generated over the last years.10 The objective was to provide a simple and clear guidance for clinicians to help them in the identification of this entity among patients with chronic obstructive airways disease (COAD). This algorithm first requires the diagnosis of COPD based on current guidelines.11 Once the diagnosis of COPD is established, the patient can be labelled as having ACO if a current and objectively established diagnosis of asthma is present. If this is not the case, the presence of one or two markers of Th2 inflammation – a very positive bronchodilator response (>400ml and >15% in FEV1) and/or a significant blood eosinophilia (>300cells/μl) – also enables the diagnosis of ACO.10

The aim of this study was to differentially characterize patients diagnosed with ACO according to the new SEPAR's algorithm from patients with other COAD categories (asthma and COPD patients who do not meet criteria for ACO). The characterization was made according to clinical, functional and inflammatory features and, as a secondary end-point, we assessed these variables’ – in solo or in combination – capacity to distinguish ACO from asthma and COPD patients without ACO.

• Chronic obstructive airway disease (COAD): patients with a FEV1/FVC post-bronchodilator <70%, regardless if it comes from COPD or asthma.

• Asthma was diagnosed according to international guidelines,8 and was classified as non-smoking asthmatics (NSA): asthma patients -non-smokers or ex-smokers- with smoking history of <20 pack-years; or smoking asthmatics (SA): asthma patients with smoking history of >20 pack-years.

• COPD was diagnosed – according to international recommendations12 – by the presence of post-bronchodilator FEV1/FVC<0.70 in patients with smoking history of >10 pack-years in the absence of a clinical suspicion for asthma. COPD with eosinophilia (COPD-e) was considered in COPD patients with eosinophil count >200eosinophils/μl in blood. We selected this threshold to recruit the patients because below this cut-off patients are unlikely to have sputum eosinophilia, according to published evidence.13,14

• ACO. Patients classified as ACO according to SEPAR-ACO algorithm10 (Fig. 2).

  
  

  Fig. 2.

  SEPAR-ACO algorithm for ACO diagnosis.

  (0.1MB).

More detailed information about data management, pulmonary function tests, fractional exhaled nitric oxide (FeNO) measurement and sputum induction can be found elsewhere.9 Periostin blood levels were measured using a streptavidin-HRP ELISA commercial kit (DuoSet® ELISA, R&D Systems, Minneapolis, MN, USA), according to the manufacturer's instructions.

Categorical variables were summarized by means of group counts and percentages, normally distributed continuous variables by the distribution mean and Standard Deviation and non-normally distributed continuous variables by means of the median and the inter-quartile range.

One-way ANOVA was used to compare normally distributed continuous variables between the 3 clinical categories. Non-parametric Kruskal–Wallis ANOVA was used for continuous parameters without a normal distribution, while Chi-square test was applied to compare group proportions. Student's t tests, Mann–Whitney U test and Chi-square with the appropriate Bonferroni correction were used for pair-wise comparisons.

The ability of all the studied parameters and their combinations to discriminate between ACO and the other categories was assessed by estimating sensitivity, specificity, positive and negative predicted values, Youden index (Youden index=sensitivity+specificity−1), the corresponding ROC curves and the 95% CI's.

In this study, we have found that SEPAR-ACO algorithm classifies in the ACO category all patients diagnosed with asthma who smoked and developed a fixed bronchial obstruction and almost a half of “eosinophilic COPD”. The definition of “eosinophilic COPD” depends on a somewhat arbitrary cut-off point of blood eosinophil count: 200cells/μl in this cohort of COAD patients and 300cells/μl in the proposed algorithm (it is for this reason why not every patient with “eosinophilic COPD” was classified as ACO). There is controversy about the exact blood eosinophils’ cut-off point that better reflects airway eosinophilia: Negewo et al. observed that patients are unlikely to have sputum eosinophilia if their blood eosinophil count is below 200cells/μl13 whereas Schleich et al. set this point at 162cells/μl to identify a sputum eosinophil count >3% and 215cells/μl in those treated with high dose of ICS.14 On the other hand, Kolsum et al. found that >250eosinophils/μl in blood reliably reflect lung eosinophilia.15 Whatever the exact limits are, it seems clear that eosinophilic COPD appears to be a distinct patient subgroup with a different inflammatory signature15 and a better response to corticosteroid treatment.16,17 It is therefore reasonable to separate this subgroup from “non-eosinophilic COPD” and to include it in the ACO's category, because it will ensure a therapeutic approach based on the use of ICs combined with bronchodilators and even biological drugs in a near future.18 However, we should take into account that SEPAR-ACO algorithm also encompasses patients with COPD and a current diagnosis of asthma, and it has been shown that these patients exhibit more atopic features, have better pulmonary function and they form a more heterogeneous group -from the inflammatory point of view- than e-COPD patients, including a significant proportion of subjects with a “Th2-low” pattern.19 This could be explained by the fact that the concept “COPD and a current diagnosis of asthma” relies on clinical characteristics (asthma diagnosis and smoking history) whereas e-COPD definition includes a biomarker of Th-2 inflammation.

Woodruff et al. have shown that lung function improvements expected with ICs were restricted to Th2-high asthma so, it is not unreasonable to speculate that not every asthma patient will need treatment with ICs.20 However, in our opinion, it seems far too early to deprive asthma patients from ICs at the present time and the most sensible option is to maintain this subgroup under the ACO definition.

Besides current diagnosis of asthma and blood eosinophilia, SEPAR-ACO's algorithm also considers a “very positive bronchodilator response” as a diagnostic criterion, since it has been published that a Th2 gene signature correlates with bronchodilator reversibility.21 However, no patient in our study was labelled as having ACO on the basis of a “very positive bronchodilator response” as a single diagnostic feature, because all patients with this characteristic showed either current asthma diagnosis or high blood eosinophilia.

Recently, Sinn and colleages made an alternative proposal to identify ACO.5 If we apply their proposed criteria in our cohort, we might classify in this category 11.7% of NSA group patients, 5.6% of COPD group patients, 77.3% of the SA group patients and 12.3% of COPD-e patients. Therefore, these criteria mainly identify as ACO asthmatics who have smoked and developed a “fixed bronchial obstruction”, but they set aside most patients with eosinophilic COPD. Since definitions of ACO vary across the different studies and expert recommendations,22 the most appropriate would be the one that includes more common “treatable traits”, enabling the clinician to accurately prescribe ICs in a particular patient and, thus, providing a more personalized therapeutic approach.

We have been unable to find clinical, functional or inflammatory variables – in solo or combined – capable of distinguishing ACO patients from other categories of COAD, beyond those included in the definition. This probably underscores the complexity and the great patient-to-patient variability of all forms of COAD, a fact that was already highlighted by other authors.23–25Apart from the arbitrariness of the blood eosinophil's cut-off point to define e-COPD – pointed out above – some other limitations deserve a comment. A large number of patients were receiving medications able to modify the expression of parameters like blood eosinophils and FENO (e.g. omalizumab and oral corticosteroids). In addition, we have chosen a limit of 20 pack-years of smoking exposure to define SA category and we recognize that it could be misleading, but we wanted to make sure that smoking-related inflammation is present in individuals with previous asthmatic inflammation.

In conclusion, the SEPAR-ACO consensus algorithm classifies as ACO all smoking patients with previous history of asthma who developed fixed bronchial obstruction and a considerable proportion of COPD patients with >300cells/μl, but apparently the criterion of high reversibility does not identify ACO patients by itself. SEPAR-ACO algorithm has the advantage that identifies patients who can benefit from treatment with ICS.

The project was endorsed by the COPD and Asthma Research Board (PII de EPOC y asma) of the Spanish Society of Pneumology and Thoracic Surgery (SEPAR). The project was funded by an unrestricted grant from Chiesi España S.A.U.

LPLL and BC made substantial contributions to conception and design of the study, and analysis and interpretation of data. They have both been involved in drafting the manuscript and revising it critically for important intellectual content. MM and VP have both been involved in drafting the manuscript and revising it critically for important intellectual content. All authors have read and approved the final manuscript.

The authors declare that they have no conflicts of interest related to this study.