The present study is a part of the EPI-SCAN study, which was a multicentre, cross-sectional, population-based, observational study conducted at 11 sites in 10 Spanish cities (Barcelona, Burgos, Córdoba, Huesca, Madrid (two areas), Requena, Sevilla, Oviedo, Vic and Vigo) representing different geographic, climatic and socio-economic regions. The main objective of the study was to investigate the prevalence of COPD in Spain in individuals between 40 and 80 years of age. The protocol of the EPI-SCAN study has been published elsewhere.4 Fieldwork and all methods have been described previously.5 The corresponding ethics committees approved the study and all participants gave written informed consent.

The final recruited population consisted of 3802 non-institutionalized participants from 40 to 80 years of age. The definition of COPD was a post-bronchodilator FEV1/FVC ratio <0.70. Subjects considered not to have COPD had a post-bronchodilator FEV1/FVC ratio ≥0.70.6 The blood sample we obtained from all participants classified as having COPD. To avoid excessive testing of the non-COPD population, each study center collected equal number of blood samples from non-COPD subjects recruited consecutively in order to avoid selection bias. Exclusion criteria included a previous diagnosis of acute myocardial infarction, angina, congestive heart failure, cancer, hepatic cirrhosis, chronic renal failure, rheumatoid arthritis or any other systemic inflammatory disease. In addition, we identified a control group of non-COPD subjects after applying the same comorbidity exclusion criteria and specific exclusion criteria of any respiratory symptoms as per the European Coal and Steel Community (ECSC) questionnaire and regularly prescribed medications.7Fig. 1 illustrates our cohort.

Fig. 1.

Flow diagram of total number of patients included in and excluded from the EPI-SCAN study and patients with available data on plasma AAT levels.

(0.12MB).

At each study center, lung function tests were performed using the same equipment according to the current recommendations.8 Spirometry was performed before and 15–30min after salbutamol inhalation. Baseline dyspnea was assessed by the Modified Medical Research Council (MMRC) scale,9 and subjects completed the ECSC questionnaire of respiratory symptoms.10

Blood samples were collected using standardized procedures and stored at −80°C before sending for the centralized analysis to a single laboratory (Hospital Clinic, Barcelona, Spain). Plasma levels of TNFα, IL-6 and IL-8 were determined in duplicate with a high sensitivity enzyme-linked immunosorbent assays (Biosource, Nivelles, Belgium). We assessed plasma levels of CRP by the latex-enhanced immunonephelometry (Siemens, Dublin, Ireland), AAT by a particle-enhanced immunonephelometry (Siemens, Marburg, Germany), albumin by the bromocresol green method (Siemens, Dublin, Ireland) and fibrinogen by using a coagulation analyzer (Roche, Manheim, Germany). Specific details about the analytic procedures described elsewhere.7

Data analysis performed using R, a free software environment for statistical computing and graphics. Statistical significance between subject categories with regard to demographic characteristics and laboratory values was analyzed with ANOVA. Specific pairwise comparisons of subject categories were assessed with the t-test. Least squares linear regression was applied to assess the associations of the dependent variables, such as AAT levels and lung function (post-bronchodilator spirometry), with various other clinical parameters as independent variables. Modeling COPD status as a function of AAT levels and other clinical parameters was performed using logistic regression.

Out of 3802 participants, 837 (22%) had available plasma levels of AAT. Of them, 303 (36.2%) had COPD diagnosis, 222 (26.5%) had respiratory symptoms but not COPD, and 312 (37.3%) were healthy, with normal spirometry and without respiratory symptoms, considered as a reference group. The mean (SD) age of cohort was 58.0 (11.3) years. The main characteristics of the participants described in Table 1.

There were more men within the COPD group than in those with respiratory symptoms or healthy (73.6% vs 53.6% and 44.2%, respectively). Moreover, COPD group was older and had more smokers and ex-smokers relative to other groups. BMI was not significantly different among the groups. As expected, lung function was significantly impaired in COPD individuals compared to the healthy reference group, and there was a wide range of COPD severity in our cohort. Although to a lesser degree, lung function was also decreased in individuals with respiratory symptoms compared to the reference group.

In the entire cohort, levels of AAT were 1.51 (0.47)g/L, with no significant differences between men and women. Notably, a subset of cases (57 or 6.8%) exhibited particularly low levels of plasma AAT (below 1g/L): 20 within COPD, 9 within those having respiratory symptoms and 28 within the reference group. Unfortunately, in this cross-sectional study, genotypes of AAT were not determined and therefore the correlation between genotype and low AAT protein levels could not be analyzed. Removing low-AAT cases as outliers did not affect the observations, and these 57 cases were included in the final analysis. When compared to reference group, subjects with respiratory symptoms or COPD showed higher plasma levels of AAT (p<0.001). As illustrated in Table 1, analysis of other plasma proteins revealed that COPD subjects had significantly higher plasma levels of CRP, IL-6 and fibrinogen but lower albumin concentrations if compared to the reference group. Plasma levels of TNFα were higher in the COPD and respiratory symptoms groups compared to reference, whereas IL-8 did not significantly differ among the groups.

We next examined the relationship between plasma AAT levels and clinical/laboratory measures. For this, we applied three types of linear regression models: univariate, correcting for CRP, and multivariate. As shown in Table 2 and Fig. 2, plasma AAT levels were significantly affected by cohort (healthy, respiratory symptoms or COPD), smoking status (smokers but not ex-smokers), lung function tests, and levels of CRP, IL-6, IL-8, fibrinogen and albumin, but not by TNFα.

Fig. 2.

The multivariate relationship between AAT levels and clinical measures (graphical presentation of the data analysis presented in Table 2).

(0.4MB).

Levels of AAT were higher in individuals with COPD or with respiratory symptoms compared to the reference group, an effect that persisted after age and CRP correction, as well as in the multivariate model. Age did not show an impact on AAT levels. Plasma levels of CRP, IL-6, IL-8 and albumin were positively associated, whereas levels of fibrinogen were negatively associated with AAT. After correcting for CRP levels, most parameters remained significantly associated with AAT levels with the exception of FEV1 (pre and post-bronchodilators). The multivariate analysis included smoking status; post FEV/FVC, CRP, IL-6, IL-8, fibrinogen and albumin. Even when modeled jointly, each of these parameters contributed significantly to the multivariate model. This implies that there is an independent relationship of each of these factors to plasma AAT levels.

We also examined whether our data above support the relationship between plasma AAT levels and lung function. To do this, we modeled lung function measures, FEV1 and FEV1/FVC, as a function of AAT and other relevant analytics in both univariate and multivariate models (Table 3). In the univariate model, FEV1 was significantly and inversely associated with AAT levels, but the effect was weaker relative to factors like sex, BMI and smoking status. When combining these terms in a multivariate model, the above independent effect of AAT was diminished to non-significant level (p<0.32). However, the association between plasma AAT levels and lung function, as measured by FEV1/FVC ratio, was more robust. Here the inverse association was stronger and remained independently significant even when correcting for other associated factors such as age, sex, smoking status, CRP, TNFα, fibrinogen and albumin. Incidental to the associations of AAT with lung function, we also observe significant decreases in lung function with increased age, smoking, BMI, and CRP. Lung function (FEV1) was positively associated with being female and albumin.

A number of outliers with a high AAT level and high FEV1/FVC (Fig. 2), allows suggesting that responders with high AAT may actually be protected from severe COPD when adjusting for smoking and age. Therefore, only for individuals with high AAT levels, we investigated the associations between AAT and FEV1/FVC. We were unable to observe any indications of protection although the low numbers of cases available leaves us underpowered for any conclusive statements.

We have explored whether there is a dose response relationship between smoke and AAT level. By using AAT levels and smoking status as an interaction term in the model, we observed no significant interactions in this cohort. However, we observed for the cohort as a whole, that pack-years were associated with decreased lung function in our multivariate model (p<3.3 E10−7). When used in an interaction term with AAT, we also observed that this interaction was significant (p<0.06 for FEV1 and 0.009 for FEV/FVC).

Based on the findings above, we were interested in the association of AAT levels with COPD in the context of the other measured clinical variables. To this end, we modeled COPD diagnosis as a function of plasma AAT levels and other measured parameters using binomial logistic regression (Table 4). The risk for COPD significantly increased with higher AAT levels in both the univariate and multivariate models, with odds ratios of 1.8 and 1.5, respectively. The univariate model demonstrated that sex, smoking status and CRP levels were associated with COPD. In the multivariate model, these parameters continued to be associated with COPD probability, demonstrating that they act independently. Similarly, we looked at the AAT levels as a risk for respiratory symptoms. Here, we did not observe a significant association with AAT after accounting for other factors, such as smoking status, TNFα, IL-8, and IL-6. Herein smoking status appears to be a strongest risk factor for respiratory symptoms in a non-COPD population.