Risk Factors Associated with Irreversible Airway Obstruction in Asthma: A Systematic Review and Meta-Analysis

Irreversible airway obstruction (IAO) is a subtype of asthma and relates to poorer prognosis in some asthma patients. However, the prevalence and risk factors for IAO are unknown. A systematic review regarding controlled clinical studies (cohort, case-control studies) on IAO asthma in adult and/or children affected by asthma/early wheeze was performed. Eighteen papers were identified in this study. It was reported that the incidence of IAO at random effects or fixed effects in severe asthma and nonsevere asthma was 0.54 (95% CI: 0.45–0.62) and 0.16 (95% CI: 0.12–0.20), respectively. In IAO asthma, the pooled odds ratio (OR) related to smoking exposure was 2.22 (95% CI: 1.82–2.73), the OR for male, smoking, and fractional exhaled nitric oxide (FENO) was 2.22 (95% CI: 1.82–2.7), 1.79 (95% CI: 1.46–2.19), and 2.16 (95% CI: 1.05–4.43), respectively, suggesting these factors increase the risk of IAO. However, a decreased OR in IAO asthma was observed due to rhinitis (OR = 0.31, 95% CI: 0.24–0.40), atopy (OR = 0.584, 95% CI: 0.466–0.732), and atopic dermatitis (OR = 0.60, 95% CI: 0.42–0.85), indicating these factors are associated with reduced risk of IAO. IAO in asthma is associated with gender, smoking, FENO, rhinitis, atopy, and atopic dermatitis.


Introduction
Asthma is a multifactorial, heterogeneous, and chronic inflammatory disease, which is characterized with the symptoms of cough, shortness of breath, wheeze, and chest tightness. The majority of patients with reversible airflow limitation asthma can be easily controlled by regular medication. However, in a small subset of asthmatic patients, even aggressive treatment fails to control the disease. The prevalence of these patients with so-called irreversible airway obstruction (IAO) asthma is not known.
Backman et al. first reported the patients with IAO asthma [1]. Lange et al. [2] found that asthmatic patients had greater declines in forced expiratory volume in 1 second (FEV1) over time compared to those who did not have asthma. Ulrik and Backer [3] found that IAO adults patients usually are associated with moderate to severe asthma. Some risk factors for IAO development have been defined; IAO asthma is associated with more severe disease [4][5][6] and is a predictor of overall mortality in asthmatic patients [7]. The reduction of risk factors for IAO may be the key strategy to decrease the morbidity in severe asthmatic patients with IAO. Vonk et al. [8] found that 41% of asthmatic patients did not have airway obstruction and 16% of asthmatic patients had IAO. However, studies about the prevalence and risk factors for IAO in asthmatic patients are limited and often contradictory. Martin et al. demonstrated that males have higher possibility than females to develop into asthma patients [9]. Ten Brinke et al. [6] found that the onset of asthma in adults increased AHR and sputum eosinophilia, which increased the risk factor for IAO asthma. Apostol et al. [10] found that smoking is a risk factor for IAO asthma. Aspirin sensitivity has been identified as a risk factor for IAO asthma [11].
Our meta-analysis was designed to assess the prevalence and risk factors of IAO in a well-defined group of patients with asthma/nonsevere asthma and to examine the clinical characteristics. In this study, we analyzed age, gender, disease duration, rhinitis, atopic dermatitis, aspirin sensitivity, smoking history and atopic status, and some clinical characteristics including markers of airway inflammation, fractional exhaled nitric oxide (FENO), and percentage of eosinophils and neutrophils in induced sputum.

Methods
According to the rules of the meta-analysis of Observational Studies in Epidemiology group, this meta-analysis was conducted [12]. The literature related to the terms "airways obstruction" OR "irreversible airway obstruction" OR "fixed airflow obstruction" OR "airway remodeling" AND "asthma" was collected from Ovid Medline, Pubmed, and ISI database by two independent reviewers (Z. LL. and G. J., last update in March, 2013). English was included in this meta-analysis. References were also manually searched to identify additional published or unpublished data and were evaluated by experts.
Studies of our meta-analysis included the following inclusion criteria: (1) primary outcome of incidence or risk factors of IAO asthma; (2) follow-up of at least 70% of patients; (3) all studies that used self-reported, physician-diagnosed asthma or new symptoms and/or medication use compatible with asthma as the criteria for incident asthma diagnosis. We excluded the studies that used single asthma symptom as an outcome and did not establish asthma diagnosis. Moreover, we did not consider those studies that did not report the odds risk. Study data sources were examined to ensure that every included dataset was unique. Stata 12.0 (Stata Corporation, College Station, TX) [13] was used to calculate ORs using inverse-variance weighted, random-effects meta-analysis [14,15]. Randomeffects methodology was chosen to analyze within-study and between-study variations. Heterogeneity of data was evaluated using the -statistic [15]. We weighted the study-specific adjusted log odd risks for cohort studies and the studyspecific adjusted log odds ratios for case-control studies by inversing their variance to compute a pooled odds risk and 95% confidence intervals (CIs) on a forest plot [16], and publication bias was evaluated [16,17]. Using data from studies that provided mean repeat length of specific case groups, we calculated standardized mean difference (SMD) to compare IAO cases with reversible airway obstruction (RAO) controls. The SMD and 95% CI were calculated in each group of cases. A plan was established prior to performing sensitivity analyses for identified issues relative to study quality, if necessary, rather than applying weights to studies in the meta-analysis based simply on quality scoring criteria [12].
2 -based statistic evaluated heterogeneity and considered the statistical significance at 2 -value less than 50%. When the 2 -value is less than 50%, fixed-effect model was used; Otherwise, a random-effects model was applied. The significance of the pooled OR was determined by the -test, and less than 0.05 was considered statistically significant. Publication bias was analyzed by several methods: (i) funnel plots; (ii) Egger's test that was also used to statistically assess publication bias. All statistical tests were performed by using the Revman 5.0 software and Stata 12.0.

Incidence of IAO Asthma
Twelve articles with the data of risk factors for IAO were retrieved [3-6, 8, 11, 33, 34, 36, 38, 39, 41]. Of the total 13,498 patients, 16% of patients were diagnosed with IAO. Four articles were involved in severe asthma [4][5][6]34], in which 54% of patients were analyzed by asthma subtypes and IAO asthma using the random-effects model. The incidence of IAO asthma in non-severe asthma patients was 16% with the random-effects model ( Figure 1) or 8% with the fixed-effect model.

Smoking.
Eight articles were retrieved based on smoking for IAO asthma patients in a cohort study [4-6, 33-35, 37, 39]. The amount of heterogeneity was decreased ( 2 = 38.2%) when we set our analysis to case-control studies; the patients exposed to smoking showed the pooled OR of 1.79 (95% CI: 1. 46-2.19). Moreover, we did not find any evidence of publication bias ( value of Egger's symmetry test: 0.606) ( Figure 3).
Although the result showed the heterogeneity, both of them showed that aspirin sensitivity was an important risk factor for IAO.

FENO.
Two studies related to FENO were included [6,34]. There was no heterogeneity; the result from fixed-effect    model was similar to that from random-effects model. These two studies yielded a protective pooled odds risk of 2.16 with 95% CI: 1.05 to 4.43.

Atopic Dermatitis.
Two studies focused on atopic dermatitis [5,34]. There was no heterogeneity, so fixed-effect pooled model was used. These two studies yielded a protective pooled odds risk of 0.60 with 95% CI: 0.42 to 0.85.

Others.
The results showed that age at onset, asthma duration, sputum eosinophils, blood eosinophils, and atopy were not important discriminators for airway remodeling in cohort studies.
Five studies found that OR age at onset is another risk factor in IAO asthma patients with OR of 1.04 (95% CI: 1.01-1.06) by the fixed-effect model [4,5,8,36,40], but significant heterogeneity ( 2 = 92.4%) was observed among studies and the random-effects model showed different result (OR = 1.02, 95% CI: 0.88-1. 19). The amount of heterogeneity did not decrease when the sensitivity analysis was performed (data not shown). We also did not find any evidence of publication bias ( value of Egger's symmetry test: 1.00).

Discussion
To our knowledge, this is the first meta-analysis study for the prevalence and the risk factors of IAO. Our findings support the high prevalence of IAO in severe asthma. IAO is highly associated with smoking and male asthma patients have twofold greater risk of IAO. Furthermore, the asthma patients with higher FENO and aspirin sensitivity also have two-fold greater risk of IAO. However, atopic dermatitis and rhinitis and atopy were significantly related to the decreasing IAO in severe asthma.
The concept of IAO implies that some changes in the structure and function of the airway occur and reduce expiratory airflow, and these changes cannot be restored to the prior state by endogenous mechanism or any treatment. Some indirect evidence demonstrated that progressive airflow obstruction may cause IAO because severe or  progressive airflow obstruction occurs in some IAO patients in spite of ongoing anti-inflammatory (glucocorticoid) or bronchodilator treatment. The studies on the pathogenesis of airway remodeling in asthma [43], such as airway wall thickness, allergic airway inflammation, epithelial-driven models of airway remodeling, bronchial neovascularization, and physiological consequences of airway remodeling, support explaining the reasons of IAO.
Some studies have explored the risk factors for the development of IAO in asthma. The reduction in lung function in children and adults may contribute to the development of IAO. Cigarette smoking is related to the development of IAO asthma and has been associated with a rapid decline in lung function in asthma [2,44]. It might be related to smoking-induced changes in airway inflammation, with a predominance of neutrophils instead of eosinophils [45].   Jiang et al. [46] suggested that atopy in itself is not related to asthma. However, asthma and atopy usually occur in parallel. Since atopic asthma is usually mild and occurs early, it is more commonly accompanied by other allergic diseases such as allergic rhinitis [47]. A reduced prevalence of rhinosinusitis in patients with more severe asthma may be due to the spontaneous regression of rhinosinusitis during the course of asthma [48]. Mascia et al. [11] showed that adults with aspirin-sensitive asthma have more severe disease and airway obstruction than aspirin-insensitive asthma patients. It may be possible that aspirin-sensitive asthma patients have longer disease duration. Chronic inflammation of the airways may also contribute to IAO development, so it may explain why FENO is the major factor of IAO.
Two major mechanisms for airway remodeling including reticular basement membrane (RBM) and subepithelial basement membrane thickening have been reported [49,50]. Another common finding in remodeled airways is the presence of increased airway smooth muscle mass [51]. Findings of increased RBM thickness and airway smooth muscle area on biopsies of severe asthma patients have intrigued the interest in the use of noninvasive imaging techniques, such as computed tomography (CT) [52], to assess the changes in the bronchial wall thickness and structure. The third dogma is that the structure changes that cause the functional changes in the airway wall are primarily responsible for IAO. In the future, we still use the risk factors to explore and identify the mechanisms of remodeling.
The limitations of this meta-analysis should also be addressed. First, some studies are excluded due to the absence of original dada, which may lead to selection bias. Thus, these studies are usually excluded. Second, all eligible studies are published in English from selected databases. It is possible that some relevant studies published in other languages are missed. Third, most studies are from Asians and Caucasians; thus our study may be applicable to non-Africans only. Fourth, asthma is a heterogeneous disease, and the therapy may affect risk factors. Fifth, we should also analyze the possibility of publication bias. Publication bias can result in the disappearance of some studies with negative results.
In conclusion, this study is the first to report risk factors that may be related to IAO in asthma patients. We found that smoking, male gender, FENO, and the absence of rhinitis, atopy, and atopic dermatitis were more likely to lead to persistent IAO. In addition, our findings further demonstrated that smoking is an important cause of irreversible airway damage. Further we need more studies to clarify the underlying risk factors of the development of IAO in subjects with asthma. The search for known and novel therapies that can directly target individual components of the remodeling process and hopefully lead to an improvement in the treatment of airflow limitation and the prognosis in severe asthma will be made.