To our knowledge, the present study is the first one to show increase over time in systemic inflammation, as measured by plasma IL-6 concentration, in COPD patients. Similar to previous studies [8, 10], our results showed that CRP mean values did not change. Furthermore, increased IL-6 levels were associated with mortality and also with reduced exercise tolerance at baseline and after three years. Therefore, our three-year follow-up study showed that IL-6 could be part of the assessment of COPD progression, and increases in serum values are markers of poorer outcomes.
COPD is characterized by a specific pattern of inflammation involving increased numbers of CD8+ T lymphocytes, neutrophils and macrophages in small and large airways and in lung parenchyma and pulmonary vasculature . Alveolar macrophages have a crucial part in orchestrating this inflammation through the release of inflammatory cytokines, such as IL-6, that attract neutrophils into the airways. IL-6 regulates many pathways that could contribute to its effect on inflammatory disease progression. During CD4 T cell differentiation, IL-6 promotes IL-17 and IL-21 production, and suppresses regulatory T cell function. The downstream effect of IL-6 is deposition of matrix, antibody complexes and proteases in the targeted tissue and, consequently, tissue destruction . However, given the cross-sectional nature of most studies, the role of IL-6 over time is not clear. Kolsum et al.  showed that the IL-6 mean values did not change significantly during the one-year period, and there was moderate repeatability of IL-6 between the two visits. By contrast, our results reported that IL-6 mean values increased significantly after three years compared to baseline measurements. We also found that patients who died presented higher plasma IL-6 concentration when compared to survivors. Furthermore, increased IL-6 at baseline was associated with mortality after adjustments for age, gender, BODE index, SpO2 and comorbidity index. Association between serum IL-6 and prognosis is not conclusive considering the results of recent studies [24, 25]. Mehrotra et al.  showed that IL-6 was a significant predictor of mortality in 268 elderly subjects with obstructive airway disease. Another study  showed association between mortality and levels of WBC counts, IL-6, fibrinogen, CCL-18, CRP, IL-8, and SP-D in 1843 COPD patients studied over three years. Using C statistics, only IL-6 independently added predictive power to the basic clinical model; however, the addition of all the biomarkers in the panel significantly increased the ability of clinical variables to predict mortality in patients with COPD. By contrast, Waschki et al.  did not find an association between mortality and levels of IL-6 in 170 outpatients with stable COPD.
CRP mean values did not change over time, although 21% of the patients presented with >3 mg/L changes in CRP after 3 years. We also found no association between CRP levels and mortality or differences in outcome variables between patients who survived at least 3 years with and without increases in CRP >3 mg/L (data not shown). In line with our findings, de Torres et al.  showed that baseline serum CRP did not correlate with mortality in patients with moderate to very severe COPD after a three year follow up study. In addition, Pinto-Plata et al.  also reported that CRP mean level did not change over a 17-month interval. In contrast, epidemiologic studies showed an association between baseline levels of systemic inflammatory markers and COPD progression [27, 28]. In a study with mild to moderate COPD patients, baseline serum levels of CRP were divided into quintiles. After five years of follow-up, it was observed that the highest quintile of CRP was a predictor of mortality compared with the lowest quintile . In the study by Dahl et al. , a baseline serum CRP greater than 3 mg/L was associated with increased risk of hospitalization and death after 8 years of follow-up in COPD patients.
As reported in previous studies, we found that CRP levels were positively correlated with plasma IL-6 concentration [4, 5, 8]; however, while IL-6 levels increased, mean values of CRP did not change over time. Biologically, IL-6 is a primary cytokine regulator of CRP in the liver  and may play a salient role in the systemic inflammatory response in COPD . IL-6 has been studied as a COPD marker, since it is increased in low weight COPD patients , and its higher levels were associated with lower levels of lung function independent of confounders such as age and smoking . It also plays a critical role in hematopoiesis, causing thrombocytosis and leukocytosis with IL-6 overexpression . In addition, studies have showed that serum IL-6 is a powerful independent predictor of future cardiovascular events in Japanese patients with multiple cardiovascular risk factors  and in older adults from a 9-year cohort , and the authors suggest that its prognostic value is superior to that of CRP.
We also found that IL-6 was negatively correlated with 6MWD at baseline and after 3 years. Although we cannot prove a causal relation, the association would suggest a persistence deleterious effect of IL-6 on physical performance of COPD patients and further investigation is warranted. In agreement with our results, Brinkley et al.  found that IL-6 levels were associated with poorer physical function, independent of age, gender, race, and body composition in older adults across multiple comorbidities, including COPD patients. Yende at al.  also showed that IL-6 was an independent predictor of reduced exercise capacity in elderly individuals with an obstructive pattern and with normal spirometry results. In a cross-sectional analysis, Garrod et al.  evaluated the values of TNF-α, IL6 and CRP in 41 patients with COPD and found a negative association between CRP and 6MWD, similar to other studies [4, 10]. These results provide evidence that chronic inflammation and impaired physical function are related in various age-related diseases. Our study, in line with this hypothesis, shows that inflammation is persistent and associated with exercise tolerance over time in COPD patients.
We also explored the relationship between number of exacerbations and plasma IL-6 and CRP concentration; however, no association was found. Agusti et al.  showed that the annual rate of exacerbations during the 3 year follow-up were higher in the persistently inflamed patients, compared with non-inflamed patients. However, the logistic regression did not show association between that annual rate of exacerbations and the presence of persistent systemic inflammation (defined as in upper quartile at baseline and after one-year for at least 2 biomarkers).
The strengths of this study are the wide range of disease severities designed to evaluate changes in systemic inflammation and its relationship to disease prognosis outcomes over a considerable period of time. Despite of the small number of patients, this study shows the association of IL-6 with prognosis in COPD patients and reinforces the role of this mediator in the evaluation of patients’ outcomes. In addition, taking in consideration the difference in values of serum IL-6 between healthy subjects and smokers in previous study , a sample of 40 COPD patients would be necessary (α = 0.05; power = 80%). However, a large sample size could allow the identification of a reliable cutoff value of IL-6 to predict poorer prognosis. We measured the systemic inflammatory mediators at baseline and after three years; therefore our study has limitations regarding the predictor value of serum IL-6 for shorter periods of time.