Statistical analysis

Descriptive statistics were calculated for all variables. Continuous variables are presented as the mean ± standard deviation (SD), and categorical variables are presented as numbers with frequencies. Laboratory data were presented as the median, interquartile range (IQR) and logarithmic scale values. The baseline variables were compared between the two groups with a one-sample t-test, and Pearson’s χ² test was performed for the categorical variables. Inflammatory parameters and pulmonary biomarkers were compared between the two groups consecutively (smokers with COPD and without AO) with the non-parametric Mann-Whitney U test. The percentage of smokers with COPD and without AO that exceeded the 90th percentile of the values of the healthy non-smokers was presented in radar plots.

To determine the prevalence of elevated cytokines, pulmonary pneumoproteins and other biomarkers, values exceeding the 90th percentile in the healthy non-smokers were considered abnormal. A model for the cigarette smoking-induced immune response for selected biomarkers was presented as a network layout, with each node corresponding to one of the six biomarkers, and its size was proportional to the prevalence of abnormal values (the 90th percentile of the healthy nonsmokers) for that particular biomarker in that particular group of subjects. Two nodes are linked if more than 1% of subjects in the network share abnormal values of these two biomarkers, its width being proportional to that proportion.

Subsequently all participants (cases and controls) were divided into the following groups: smokers with systemic inflammation (i.e., with 2 or more elevated biomarker levels in the upper quartile, including participants with COPD and without AO n=19) and smokers without systemic inflammation (i.e., with no elevated biomarkers, also including participants with COPD and without AO n=36). To verify the pattern of systemic inflammation, smokers with 3 or more elevated biomarker levels in the upper quartile (n=11) were additionally compared to smokers with no elevated biomarkers (n=36).

Smokers with a high cytokines profile (defined as in upper quartile for at least 2 (or 3) biomarkers) were considered as “inflamed” smokers. Smokers with 2+ (or 3+) elevated biomarker levels (“inflamed”) and smokers with no elevated biomarkers (non-inflamed) were compared between the two groups with the one-sample t-test for baseline variables and Pearson’s χ² test for categorical variables. Logistic regression was used to investigate factors contributing to systemic inflammation and high pneumoprotein levels in the whole group of smokers.

A group of heavy smokers (those who smoked more than 40 pack/years including participants with COPD and without AO) was identified.

Statistical significance was set at P<0.05 (a two-tailed probability value). SPSS version 20.0 (SPSS Inc., Chicago, IL, USA) was used for the statistical analyses.