Abstract
Asthma is characterised by chronic airway inflammation and remodelling, which can be (partially) suppressed by inhaled corticosteroids (ICSs). Plasminogen activator inhibitor-1, encoded by the SERPINE1 gene, is the key inhibitor of the plasminogen activator system, which affects tissue repair and remodelling.
We studied associations between a functional SERPINE1 -675 4G/5G promoter polymorphism and asthma development, severity and response to ICSs.
Longitudinal cohorts of 281 asthmatics and their nonasthmatic spouses, and the general population (n=1,390) were studied. No significant associations were found with asthma development and immunoglobulin (Ig)E levels, or with forced expiratory volume in 1 s (FEV1) in nonasthmatic controls. Asthmatic subjects carrying the SERPINE1 5G allele had higher IgE and lower lung function levels at follow-up, lower maximally attained lung function levels, and faster lung function decline compared with individuals with the 4G/4G genotype. ICS treatment showed an immediate improvement in FEV1 in asthmatics carrying the 5G allele. However, these asthmatics still had the fastest rate of FEV1 decline after initiating ICS treatment. Finally, the 5G allele was associated with a lower prevalence of complete asthma remission at follow-up.
These findings suggest that SERPINE1 is not an asthma susceptibility gene, but rather affects the severity, progression and long-term ICS response in asthma.
- Asthma remission
- asthma severity
- genetic
- inhaled corticosteroids
- lung function decline
- plasminogen activator inhibitor-1
Asthma is characterised by chronic airway inflammation, which contributes to characteristic structural changes, referred to as airway remodelling. A proportion of patients with asthma develops persistent airflow limitation [1–4] or shows an accelerated lung function decline [2, 5]. Conversely, some individuals outgrow their asthma [6, 7]. However, in patients in clinical asthma remission, ongoing airway inflammation is present, with a risk of relapse of symptoms, but a small subset of asthmatics may show complete remission later in life [8].
Airway remodelling in asthma is the process in which injured epithelial layers are replaced by extracellular matrix (ECM) instead of parenchymal cells of the same cell type. This pathological remodelling eventually leads to changes in the airway structure, including ECM deposition, subepithelial fibrosis, smooth muscle hypertrophy and goblet cell hyperplasia [9]. An imbalance between proteases and their inhibitors in response to inflammation contributes to remodelling [10, 11]. The plasminogen activator system (PAS) is a regulator of ECM proteolysis, both directly through plasmin formation and indirectly through plasmin-mediated activation of matrix metalloproteinases (MMPs). Plasminogen activator inhibitor (PAI)-1 is a key regulator of PAS, inhibiting both fibrinolysis and MMPs in the lungs. Serum and sputum PAI-1 levels are higher in asthmatic than in healthy subjects [12, 13]. Increased activity of PAI-1 has been associated with lung fibrosis in murine models [14, 15] and PAI-1 is thought to play an essential role in tissue repair and remodelling [14, 16, 17]. PAI-1 is synthesised by many cells relevant to asthma and human mast cells release functionally active PAI-1 when stimulated by immunoglobulin (Ig)E receptor cross-linking [18].
We have previously shown in a genome-wide screen that elevated IgE is linked to chromosome 7q21 in a Dutch families asthma study [19]. A number of candidate genes are located in this region, one of these being SERPINE1, which encodes PAI-1. Elevated serum PAI-1 levels and the SERPINE1 -675 4G/5G promoter polymorphism [20, 21] have previously been associated with elevated serum IgE levels in allergic diseases and allergic asthma [12, 22]. The 4G/5G polymorphism has also been associated with both the development of asthma [12, 23, 24] and asthma severity (lower lung function and increased airway hyperresponsiveness (AHR)) [12]. Importantly, the SERPINE1 4G/5G polymorphism influences PAI-1 expression, with the 4G allele being associated with higher PAI-1 levels [20, 21].
In summary, there is an association between IgE levels and a region on chromosome 7 that harbours the SERPINE1 (PAI-1) gene [19]. There is evidence for elevated PAI-1 levels in asthma [12], and a role of PAI-1 in cell migration and tissue repair [14, 16, 25–27], and airway remodelling [17], and SERPINE1 polymorphisms are associated with asthma development and severity [12, 23, 24]. Therefore, we investigated associations of SERPINE1 with asthma, IgE and airway remodelling, as reflected by asthma remission and progression, including effects on forced expiratory volume in 1 s (FEV1) and FEV1 decline in asthma patients. Since inhaled corticosteroids (ICSs) can reduce airway inflammation and lung function decline [5], we also assessed the interaction between SERPINE1 and ICS use. To this end, we analysed a unique longitudinal population of patients with moderate-to-severe asthma and their spouses, and a large, independent, longitudinal population-based cohort as a control group.
METHODS
Study populations
Asthma population
A cohort of 281 patients diagnosed with symptomatic asthma was initially studied in 1962–1975 at Beatrixoord Hospital (Haren, the Netherlands), a regional referral centre for patients with obstructive airways disease. At that time, all were aged <45 yrs and had AHR to histamine (30-s method; provocative concentration causing a 20% fall in FEV1 (PC20) ≤32 mg·mL−1) [28]. Between 1991 and 1999, the patients were extensively re-examined, as well as 200 spouses of these asthmatic probands [7, 29]. At the time of testing, all participants had no exacerbation. Maintenance asthma and allergy medication was stopped during the 2 weeks prior to the study, except for oral corticosteroid use [7].
General population
A sample of 1,390 people was selected from 2,467 subjects participating in the last survey of the Vlagtwedde–Vlaardingen cohort study in 1989–1990 [30]. This general population-based cohort study of exclusively Caucasian individuals of Dutch descent started in 1965. Participants have been followed up for 25 yrs, with surveys being performed every 3 yrs (maximum of seven surveys per participant) [31]. We genotyped DNA samples of 1,390 subjects with >1,500 ng isolated DNA available. There were no differences in characteristics at the last survey between the selected and nonselected groups [30].
The Medical Ethics Committee of the University Medical Center Groningen (Groningen, the Netherlands) approved all studies and all participants gave their written informed consent.
Clinical evaluation
Asthma population
Probands underwent a standardised, comprehensive evaluation for asthma at initial testing, including lung function, AHR to histamine and a symptom questionnaire [7, 29]. After initial testing, the asthmatic probands had routine check-ups for their asthma at least once a year. Data on lung function and corticosteroid use during check-ups were extracted from the medical records. Lung function data during hospital stays because of asthma exacerbations or during pregnancies were not used [5]. In the re-examination between 1991 and 1999, the same standardised methodology was used and spouses were also tested using this methodology. In addition, serum total IgE levels were measured and blood samples for DNA isolation were taken.
General population
During all surveys, information on respiratory symptoms and smoking status was collected using a questionnaire [32, 33], and lung function testing was performed [34]. In addition, at the last survey in 1989–1990, serum total IgE was measured and blood samples for DNA isolation were taken.
DNA extraction and genotyping
DNA was isolated from peripheral blood leukocytes using standard methods. All subjects were genotyped for the SERPINE1 -675 4G/5G polymorphism (rs1799889). For a detailed description of the methods used, refer to the online supplementary material.
Statistical analyses
The Chi-squared test was used to test for Hardy–Weinberg equilibrium. Case–control analyses (Chi-squared test) were performed on the association between the SERPINE1 4G/5G polymorphism and the presence of asthma, according to the algorithm previously published by Panhuysen et al. [29] (see online supplementary material). The genotype distributions of the asthmatic probands and asthmatic spouses of the asthma population were compared with the genotype distribution of: 1) the nonasthmatic spouses of the asthma population; and 2) the nonasthmatic subjects from the general population. The nonasthmatic spouses were a suitable control group as they shared environmental factors with the asthma cases, therefore excluding this confounder. In the general population cohort, the asthma phenotype according to the algorithm could not be established and, therefore, asthma was defined as ever having experienced an asthma attack, as reported on the questionnaire.
Associations between the SERPINE1 4G/5G polymorphism and serum total IgE were determined within the patient and control groups using one-way ANOVA. Linear regression analysis was used to evaluate the effect of the SERPINE1 polymorphism on FEV1, adjusted for age, height, sex, pack-years of smoking exposure and steroid use.
The maximally attained level of FEV1 during young adulthood (i.e. the plateau phase [35]) was determined as the highest level of FEV1 reached between the ages of 20 and 35 yrs, and could only be determined in the probands of the asthma population (see online supplementary material). A linear regression analysis was used to determine the effect of the SERPINE1 polymorphism on the maximally attained level of FEV1 adjusted for age at maximum, sex, height and pack-years of smoking.
Linear mixed-effect models were used to investigate the effect of the SERPINE1 4G/5G polymorphism on FEV1 decline in the probands of the asthma and general populations [5, 36]. The age of 30 yrs was the starting point of the analyses because, at that age, the maximum lung function level is generally achieved and lung function starts to decline after this point [35] (online supplementary material).
Finally, the association between complete asthma remission at the follow-up visit and the SERPINE1 genotype was determined in the asthma population. Complete asthma remission was defined as the absence of wheezing and asthma attacks, presence of normal lung function (post-bronchodilator FEV1 >90% predicted), absence of AHR (PC20 ≥32 mg·mL−1), and no use of inhaled or oral corticosteroids.
As our analyses were hypothesis driven and the outcome variables were not independent from each other (e.g. the different lung function parameters are correlated), we did not apply a sequential, classical, Bonferroni multiple-testing correction. Rather, we used an α of 0.025, and results with a p-value between 0.025 and 0.05 were regarded as borderline significant.
Linear mixed-effect models of FEV1 decline were conducted with S-plus 7.0 (Insightful Corp, Seattle, WA, USA). All other analyses were conducted with SPSS (version 16; SPSS Inc., Chicago, IL, USA).
RESULTS
The clinical characteristics and the genotype distribution of the study populations are shown in table 1. The SERPINE1 4G/5G polymorphism was in Hardy–Weinberg equilibrium. 20 out of the 200 spouses of the probands had asthma themselves, and were included in the analyses of asthma and asthma phenotypes. In the general population, 143 (10.3%) out of the 1,390 included subjects had ever had an asthma attack.
Association with asthma and IgE
The case–control analyses did not show statistically significant differences in genotype distribution between asthma cases and nonasthmatic controls (online supplementary table E1).
In probands and spouses with asthma, patients with the 5G allele (4G/5G or 5G/5G) had a higher serum total IgE level (geometric mean for 4G/4G 58.9 IU·L−1 (sd 4.1%), and for 4G/5G and 5G/5G 97.1 IU·L−1 (sd 4.6%); p=0.019). There were no significant associations between the genotypes and IgE levels in nonasthmatic spouses, or in asthmatic or nonasthmatic subjects from the general population.
Association with lung function
Regression analyses showed that the 5G allele was significantly associated with lower FEV1 in asthma probands and affected spouses (n=301) after correction for confounding variables (240 subjects had data on SERPINE1; table 2). A trend for the same association was observed in asthma subjects from the general population (online supplementary table E2). No associations between SERPINE1 genotype and lung function level were found in the control populations (online supplementary table E2).
The 5G allele was associated with a lower maximally attained FEV1, i.e. asthmatic probands with the 4G/5G or 5G/5G genotype had a 30.4 cL (95% CI 1.8–59.0 cL) lower FEV1 level compared with those with the 4G/4G genotype (borderline significant; p=0.04).
In the asthma probands, the mean FEV1 decline was substantially faster in subjects with the 4G/5G or 5G/5G genotype (47.7 mL·yr−1, 95% CI 31.1–64.2 mL·yr−1) than the 4G/4G genotype (32.4 mL·yr−1, 95% CI 13.4–51.3 mL·yr−1) (fig. 1), a difference of 15.3 mL·yr−1 (95% CI 2.1–28.6 mL·yr−1; p=0.024) (online supplementary table E3).
Forced expiratory volume in 1 s (FEV1) and SERPINE1 -675 4G/5G polymorphism (dominant genetic model). Results of linear mixed-effect model analyses. Estimates and p-values from the analyses of the dominant genetic model (4G/4G group compared with the combined 4G/5G and 5G/5G groups). Corrected for use of inhaled corticosteroids. Data are presented as mean (95% CI). #: p=0.024.
As expected, there was an immediate improvement in FEV1 after the start of ICS treatment [37]. However, this was only present in individuals having the 4G/5G or 5G/5G genotypes: FEV1 increased by a mean of 189.9 mL (95% CI 30.5–349.2 mL). This contrasts with asthmatics with the 4G/4G genotype, i.e. a fall of -63.1 mL (95% CI -298.3–172.1 mL) (fig. 2) and a difference of 253.0 mL (95% CI 7.9–498.0 mL) (borderline significant; p=0.044).
Change in forced expiratory volume in 1 s (FEV1) after starting treatment with inhaled corticosteroids (ICSs) stratified by the SERPINE1 -675 4G/5G polymorphism (dominant genetic model). Results of linear mixed-effect model analyses. FEV1 levels before the start of ICSs were set to zero per genotype group. Estimates and p-values from the analyses of the dominant genetic model (4G/4G group compared with the combined 4G/5G and 5G/5G groups) of the level of FEV1 before the start of ICSs compared with the level of FEV1 after the start of ICSs. Data are presented as mean (95% CI). #: p=0.044.
ICS use also slowed the rate of FEV1 decline over time by 31.2 mL·yr−1 (95% 12.8–49.5 mL·yr−1). This effect was comparable between the genotypes (4G/4G versus 4G/5G or 5G/5G; p=0.28).
In the general population, SERPINE1 genotypes were not significantly associated with FEV1 decline over time (4G/4G versus 4G/5G or 5G/5G: -0.8 mL·yr−1 (95% CI -3.7–2.0 mL·yr−1) faster decline in nonasthmatics and 0.2 mL·yr−1 (95% CI -9.1–9.5 mL·yr−1) less rapid decline in asthmatics).
Association with complete asthma remission
The prevalence of complete asthma remission at follow-up in the 281 asthmatic probands was 11.7% (n=33). Complete asthma remission was more prevalent in subjects with the 4G/4G genotype (20.3%) compared with subjects with the 4G/5G genotype (10.9%) or the 5G/5G genotype (3.8%) (p=0.025).
The prevalence of complete asthma remission at follow-up in the 281 asthmatic probands was 11.7% (n=33). Complete asthma remission was more prevalent in subjects with the 4G/4G genotype (20.3%) compared with subjects with the 4G/5G genotype (10.9%) or the 5G/5G genotype (3.8%) (p=0.025).
DISCUSSION
This study suggests that the functional SERPINE1 4G/5G polymorphism is associated with the severity, but not the presence, of asthma. Asthmatics with the 5G allele had a significantly higher serum total IgE level, a lower FEV1, a lower maximally attained FEV1 during young adulthood and a faster annual FEV1 decline. Data in the general population confirmed the association between the 5G allele and lower FEV1 in asthmatics only. Interestingly, treatment with ICSs was associated with an immediate FEV1 improvement in asthmatics with the SERPINE1 5G genotype, a finding that was absent in those with the 4G/4G genotype. Notably, we suggest the 5G allele was also associated with a lower prevalence of complete asthma remission at follow-up.
Our genetic findings could be a direct result of differences in PAI-1 activity in airway remodelling. PAI-1 has been described to regulate tissue response and repair, by PAS inhibition [17]. Higher levels of PAI-1 at the inflammatory site in association with pronounced airway remodelling may, therefore, lead to a lower lung function. The 4G/5G polymorphism directly influences the level and activity of PAI-1 in plasma, and individuals with the 4G allele have higher plasma PAI-1 levels and activity [20, 21, 38, 39]. Differences in PAI-1 levels between the genotypes can therefore be involved in the development and progression of asthma. We show that asthmatics with the 4G/5G or 5G/5G genotype had a lower lung function level than those carrying the 4G/4G genotype. Unfortunately, we have no data on serum PAI-1 levels and are thus unable to confirm higher PAI-1 levels in individuals with the 4G allele. Although some studies have described higher PAI-1 plasma levels to be associated with the 4G allele [20, 21, 38, 39], this has not been invariably shown [12, 40]. Moreover, Stevens et al. [41] showed that PAI-1 expression and activity was increased in epithelial brushings of children with asthma, which was not reflected in plasma. Thus, the end result of SERPINE1 genotypes on the in situ activity, which may affect airway wall and lung tissue remodelling, remains to be determined. Furthermore, active smoking, alcohol, obesity, high serum triglycerides, male sex and age seem to increase PAI-1 levels, whilst regular exercise has been associated with lower PAI-1 levels [42–46]. This makes interpreting plasma PAI-1 level differences between the genotype groups difficult.
In our analysis, we found no significant difference in genotype distribution between asthma cases and nonasthmatic controls. Although this analysis is based on relatively few subjects, the remarkable similarity of the genotype distributions of the asthma cases and nonasthmatic controls implies that low study power is not driving this nonsignificance.
Our longitudinal study design allowed us to investigate the effect of SERPINE1 4G/5G polymorphism on the rate of FEV1 decline, especially in relation to the start of ICS treatment. As expected, we found a significant initial improvement in FEV1 after the start of ICS treatment. Interestingly, this improvement was largest in asthmatics having the 4G/5G or 5G/5G genotype, and absent in the 4G/4G genotype group. This effect could not be driven by the fact that asthmatics carrying the 5G allele had a lower level of lung function than the 4G/4G genotype group, since this was accounted for in the analysis.
From a pathophysiological perspective, the beneficial effects of the 5G genotype in ICS response might be due to differences in PAI-1 levels in the lungs. However, asthmatic subjects carrying the 5G allele had a more rapid annual decline in FEV1. ICS treatment did not reduce the difference between the genotype groups and subjects carrying the 5G allele still had a significantly more rapid annual FEV1 decline during ICS treatment compared to those carrying the 4G/4G genotype. This implies that the short-term effects of ICS on FEV1 may be differentially regulated in a pathological respect than their long-term effects.
There exists an imbalance between MMPs and their inhibitors in asthma that contributes to airway remodelling [10, 11, 47], an important mechanism of accelerated lung function decline. Human mast cells are an important source of PAI-1 [18], a major inhibitor of MMPs, and pronounced activity of PAI-1 was reported to associate with pulmonary fibrosis [14–17]. Thus, PAI-1 may affect airway remodelling in asthma. Cho et al. [18] showed that sensitised mast cells release a considerable amount of PAI-1 and this was associated with blocking of fibrinolysis, thereby promoting fibrin and collagen deposition, which is a feature of airway remodelling [18]. It is tempting to speculate that ICS use will lead to a reduction of PAI-1 activity, since steroids suppress degranulation and cytokine production in mast cells potently in vitro in a time-dependent manner [48–51]. So far, conflicting results have been reported on the effects of steroids on PAI-1 production by cells in vitro, which may largely be due to the type of cell being studied and timing of effect measurements. Incubation with glucocorticoids for 2–8 h resulted in increased PAI-1 levels in keratinocytes and human lung fibroblasts [52–54], and partial suppression of fibrinolytic activity in pulmonary alveolar epithelial cells [55]. In contrast, after an initially increased PAI-1 activity by glucocorticosteroids in human adipose tissue fragments, a significant reduction in PAI-1 activity and mRNA expression occurred after a further 48-h incubation [56]. Finally, there is most probably no simple answer to the question of in what way ICS use influences PAI-1 levels and activity at the bronchial epithelial level, especially since other anti-inflammatory effects of ICSs, such as suppression of cytokine release, are also influencing PAI-1 expression and activity. Thus, short-term beneficial effects on lung function may result from regulation at different cellular levels rather than the lack of difference in FEV1 decline with glucocorticosteroids in relation to PAI-1 genotypes.
Many studies have investigated which factors determine the progression or severity of asthma, especially since severe asthma largely drives the economic costs of asthma management. However, it is also of importance to get insight in the driving factors of asthma remission. Here, we found that asthma remission, defined as the absence of asthma symptoms, normal lung function, absence of hyperresponsiveness and no use of asthma treatment, occurred in 11% of our asthma population. For the first time, we suggest that remission of asthma may be genetically determined. Asthmatics with the 5G allele of SERPINE1 had the lowest prevalence of remission and those with the 4G/4G genotype the highest prevalence. Again, this may reflect differences in chronic airway inflammation and remodelling between genotypes.
Higher serum IgE levels have been associated with the 4G allele of the SERPINE1 4G/5G polymorphism [12, 57]. We found higher IgE levels in those carrying the 5G allele. However, we found no association of the SERPINE1 4G/5G polymorphism with serum total IgE in the control groups or with asthma in the case–control analyses. Therefore, we believe SERPINE1 is not an asthma susceptibility gene and does not solely carry the risk of higher IgE levels in asthma. Thus, SERPINE1 is not the gene by which we can explain our previously found linkage on chromosome 7q21 [19].
In accordance with previous studies, the 4G allele was the most prevalent allele in our populations [12, 23, 24, 58]. In our study, we found the 5G allele carried the risk of development of more severe asthma, reflected by lung function impairment, response to ICS use and lower prevalence of asthma remission. Others found that the 4G allele was associated with asthma and lower lung function in asthmatic patients [12]. These discrepant findings may be due to population stratification. Asthma is a complex disease in which genetic and environmental factors contribute to the asthma phenotype. Gene–environment interactions are extremely complex and not always linear, such that the same genetic variants might be associated with opposite phenotypes in different environments [59]. This might be the case in our population; although we have no direct signs of which contributing factors may account for the conflicting findings. Another explanation may be that the real causal variant in the SERPINE1 gene is not the polymorphism we genotyped but another polymorphism that is in linkage disequilibrium (LD) with the 4G/5G polymorphism. This LD structure may differ between populations and, therefore, we found different risk allele to previous studies.
A recent linkage and association study identified the plasma urokinase plasminogen activator receptor (PLAUR) gene in chromosome 19q13.1-3 as a potential asthma susceptibility gene [60]. Polymorphisms in the PLAUR gene were associated with asthma, AHR, FEV1 and plasma levels of PLAUR (also known as uPAR) [60]. PLAUR interacts with urokinase plasminogen activator, resulting in enhanced activation of cell-bound plasminogen [61] and, therefore, plays a role in PAS. PLAUR has been implicated in many physiological processes, including cell migration, proliferation, differentiation and tissue fibrosis [62]. These findings further support those of the present study on SERPINE1 polymorphism and asthma, implying that PAS is a candidate pathway for enhanced airway remodelling in asthma.
We conclude that our findings may suggest a role of the SERPINE1 gene in the progression, remission and severity of asthma, probably via the effects of PAI-1 on airway inflammation and remodelling. Clearly, functional studies need to address the exact contribution of PAI-1 in this process and the interaction with ICS treatment.
Footnotes
This article has supplementary material available from www.erj.ersjournals.com
Support Statement
This study was supported by the Netherlands Asthma Foundation (grants AF 93.66 and AF 3.2.00.38).
Statement of Interest
Statements of interest for D.S. Postma and G.H. Koppelman can be found at www.erj.ersjournals.com/site/misc/statements.xhtml
- Received November 27, 2010.
- Accepted March 15, 2011.
- ©ERS 2011
REFERENCES
