This study provides three relevant and novel observations. First, it characterizes the systemic inflammatory network pattern (inflammome) in patients with COPD and distinguishes it from that of smokers with normal lung function and non-smokers. Secondly, it shows that systemic inflammation is not a constant feature in all COPD patients, since about a third of those studied here did not have any abnormal biomarker at baseline and about the same proportion remained ‘non-inflamed’ after one year of follow up. Finally, it identifies a subgroup of COPD patients with persistently elevated inflammatory biomarker levels that, despite relatively similar lung function impairment, had significantly increased all-cause mortality and exacerbation frequency. These inflamed patients may therefore constitute a novel distinct phenotype within the larger group of patients with COPD and could be the target of novel therapeutic strategies.
Several studies have previously reported elevated levels of circulating WBC, CRP, IL-6, IL-8, fibrinogen and TNFα in patients with clinically stable COPD 
. Yet, they were limited because of the relatively small numbers of patients studied, the large variability of values observed, the fact that measurements were mostly made on a single occasion, potential confounders such as smoking status and treatment with anti-inflammatory drugs were not considered and, importantly, the longitudinal relationship with relevant clinical outcomes of the disease could not be established because of their cross-sectional design. Our study overcomes these limitations and provides, therefore, novel information on the true prevalence of systemic inflammation in COPD and its importance in the progression of disease.
The inflammatory response is a complex network of multiple cell types and mediators 
which the emerging field of network medicine is only beginning to decipher 
. We used this approach 
to identify relationships between systemic inflammatory biomarkers (the inflammome) 
among smokers with and without COPD. We recognize that our results are incomplete but they showed that, at variance with current understanding 
, systemic inflammation is not a constant feature of COPD and that, when present for at least 1 year, it is associated with worse COPD outcomes at 3 years follow-up. Age, gender and smoking exposure were similar between non-inflamed and inflamed patients but the latter were more obese, dyspneic, had lower health related quality of life, more frequent exacerbations, worse exercise tolerance, a higher BODE index and reported more cardiovascular disease, despite similar use of statins (). Interestingly, although airflow limitation was slightly worse in patients with persistent inflammation, most pulmonary characteristics of COPD, such as the prevalence of chronic bronchitis, the degree of emphysema, the bronchodilator response and the rate of FEV1
decline during follow-up, were similar in both groups (). Logistic regression analysis identified age, BMI, current smoking, health status and airflow limitation as risk factors for persistent inflammation whereas gender, cumulative smoking exposure, presence of chronic bronchitis, prior exacerbation rate, use of ICS, history of cardiovascular disease, statin use, exercise tolerance and the presence of emphysema were excluded (). Taken together, these observations suggest that systemic inflammation in COPD need not parallel the severity of the lung disease and raises questions about its pulmonary origin (the “spill-over” hypothesis) 
. In contrast, the fact that persistently inflamed patients were more obese supports a potential systemic origin of inflammation 
, although other potential mechanisms, such as the presence of airway bacterial colonization 
and/or sleep apnea syndrome overlap 
cannot be excluded because they were not investigated in ECLIPSE. The origin of systemic inflammation in COPD remains to be determined. However, our findings are consistent with those of Garcia-Aymerich et al
, who using a different methodological approach (cluster analysis) also identified a “systemic” COPD subtype characterized by more systemic inflammation and a higher proportion of obesity in 342 COPD patients followed during 4 years 
An important observation of our study is that all-cause mortality (13% vs. 2%) and the annual rate of moderate/severe COPD exacerbations (1.5 vs. 0.9 per year) during the 3 year follow-up were higher (p<0.001) in the persistently inflamed patients, compared with non-inflamed patients. These observations are clinically relevant because the severity of airflow limitation has been used so far as the most important criteria to guide therapy in COPD 
, whereas our study shows that patients with similar levels of airflow limitation may have different outcomes depending on the presence or absence of persistent systemic inflammation. Indeed, a persistent elevation of systemic inflammatory biomarkers can occur even in patients with moderate airflow limitation (Figure S3
). In this context, it is worth noting that among the 220 patients identified in this study with persistent systemic inflammation (), 89 (40%) were frequent exacerbators according to the definition of Hurst et al
, an additional 61 (28%) had a single exacerbation, and the remaining 70 (32%) reported no exacerbations during the first year of follow up, suggesting that the frequent exacerbator phenotype 
and the persistently inflamed phenotype described here are not necessarily identifying the same individuals. Finally, given the limited efficacy of inhaled corticosteroids in reducing systemic inflammation in COPD 
, patients with persistent systemic inflammation may require a different therapeutic approach for the optimal management of their disease that will have to be explored in future studies.
Our study has several strengths and limitations. To date, it provides the largest longitudinal investigation of systemic inflammatory biomarkers in a group of stable, well characterized COPD patients and compares their results to those of smoking and non-smoking controls 
. This latter aspect proved important for the proper interpretation of the findings reported here, since the large biomarker variability observed required the establishment of upper normal values. Likewise, given the significant effect of smoking identified, any accurate interpretation of abnormal levels of inflammatory markers in COPD must take it into account. The fact that patients were followed prospectively for 3 years is another strength of our study because it not only allowed the assessment of the temporal stability of the biomarker levels but, importantly, the investigation of their relationship with clinically relevant outcomes, and thus the identification of a distinct subgroup of COPD patients with worse clinical outcomes associated with the persistence of systemic inflammation. Our study also has some potential limitations. First, this is a descriptive study, so our results only show associations and do not prove causality. Besides, since this is an exploratory analysis, we opted to identify as many possible differences for further investigation by not adjusting for multiple comparisons. Hence, our analyses and conclusions will need to be replicated either prospectively in a study powered for these hypotheses or in other cohorts that contain similar data. Second, the biology of the inflammatory response is complex and we studied only a limited panel of biomarkers. However, the biomarkers we chose correspond to those investigated by the majority of previous studies 
and are often and easily measured in clinical practice. Yet, we did not study markers of tissue repair, and it is likely that the balance between inflammation and repair is important for the pathobiology of COPD 
. Third, patients were recruited into ECLIPSE mostly from hospital clinics and were treated according to their local physician. These considerations need to be taken into account when comparing results with untreated patients or patients managed in primary care since no patients with mild airflow limitation (GOLD grade 1) were included in the study. Finally, mortality data refers to all-cause mortality since cause-specific mortality was not recorded in the study.
In conclusion, this study begins to describe the systemic inflammatory network pattern (inflammome) associated with COPD and how it differs from that of smokers with normal lung function. It also identifies a sub-group of COPD patients with persistently increased biomarkers levels that is associated with a higher incidence of exacerbations and worse survival despite similar lung impairment, suggesting that this constitutes a novel COPD phenotype 
. Future clinical trials will have to determine the best therapeutic strategy for these patients. This may have important therapeutic implications also for other major non-communicable diseases, including cardiovascular and metabolic diseases, also characterized by chronic low-level systemic inflammation