Background
Chronic obstructive pulmonary disease, COPD, is today considered a heterogeneous syndrome, including several clinical phenotypes [
1]. Under-diagnosis is common; around 20–30% of all cases are identified by healthcare. Most of the current knowledge regarding the pathophysiological mechanisms in COPD is based on highly selected study populations [
2,
3]. Thus, it is unclear to what extent the results are generalisable to COPD in the population.
The discovery of alpha-1-antitrypsin deficiency (AATD) and its association with emphysema in smokers provided the concept of imbalance between proteases and anti-proteases and its contribution to COPD development [
4]. Cross-sectional studies on selected COPD-populations have observed increased levels of matrix metalloproteinase-9 (MMP-9) in COPD, and an association between the level of MMP-9 and FEV
1 [
5‐
7]. We have recently shown an inverse relationship between FEV
1 and serum MMP-9 in a cross-sectional study of a population-based COPD-cohort [
8]. This indicates that increased proteolytic activity is associated with disease severity not only in selected COPD populations. However, MMP-9 should be considered in relation to the tissue inhibitor of metalloproteinases-1 (TIMP-1), because of its inhibiting effect on MMP-9 [
6,
9].
There are a few longitudinal studies within this topic, of small and selected study populations. These studies indicate that MMP-9 is related to lung function decline in COPD [
10]. Also, in a selected population of patients with AATD-associated emphysema, higher plasma MMP-9 levels were associated with disease progression [
11]. Still, the link between proteolytic imbalance and long-term COPD disease progress has not yet been evaluated in population-based studies. We believe that our population-based COPD cohort provides an excellent basis for a longitudinal follow-up, when it comes to addressing the prognostic value of MMP-9 in COPD [
8].
The first aim of this population-based study was to evaluate the biomarkers MMP-9, TIMP-1 and MMP-9/TIMP-1 ratio in serum. The second aim was to put this in relation to disease progress and prognosis, assessed as lung function decline and mortality respectively. The third aim was to do this among subjects with and without COPD.
Discussion
In this population-based study, subjects with and without COPD showed a different risk factor pattern when evaluating biomarkers in relation to prognosis, assessed as mortality. In non-COPD, male sex, burden of smoking, heart disease and MMP-9/TIMP-1 ratio increased the risk for death, and serum concentration of TIMP-1 had a protective effect. Meanwhile in COPD, age, current smoking and increased serum concentration of MMP-9 and MMP-9/TIMP-1 ratio increased the risk for death. These findings bring additional verification of MMP-9’s impact on COPD, previously explored in epidemiological, [
11,
20‐
23] as well as recent genetic studies [
24] This supports earlier findings of MMP-9’s relation to mortality in other groups of subjects [
25,
26]. Though the present results imply that protease-anti-protease imbalance has a prognostic impact both among subjects with and without COPD, their risk factor patterns did differ. In COPD, the biomarker pattern indicates that an increased proteolytic activity is associated with an increased risk for death and the significance of these biomarkers’ importance for the prognosis in COPD merits further evaluation.
Based on the results in a selected population of COPD-patients [
10], we expected to find a correlation between the selected biomarkers and lung function decline in COPD, but this was not the case. Still, when comparing the 2005-biomarkers among those who did and those who did not attend in 2010, a greater protease-anti-protease imbalance was indicated among non-participants. Likewise, those who died during follow-up had higher MMP-9 and MMP-9/TIMP-1 ratio in 2005 than those who lived and participated in follow-up, and the levels were higher in COPD than non-COPD. Notwithstanding the association between biomarkers and lung function decline in non-COPD, this could not be proven in COPD. However, we assume that there is an influence also among those with COPD that cannot be demonstrated here, due to the loss of follow up of those with the highest serum biomarker levels. The findings imply a healthy survivor effect, which could offer an explanation to the somewhat unexpected absence of a correlation between biomarkers and lung function decline in COPD, when assessed over a five-year period. The lower mortality and lower biomarker levels associated with mortality in non-COPD than COPD provides a different basis for evaluating decline in lung function in relation to the biomarkers and may contribute to the observed significant relationship in non-COPD.
To the best of our knowledge, this is the first time these biomarkers have been evaluated in relation to prognosis and assessed as risk factors for death in a population-based COPD study. In both groups, MMP-9 levels and MMP-9/TIMP-1 ratio in 2005 were higher among those who died compared to those who survived. Furthermore, among those with COPD, these biomarkers were associated with an increased risk for death, also after adjustment for confounders, and independent of disease severity, assessed as FEV1% of predicted. This suggests that markers of increased proteolytic imbalance may be an important disease mechanism in COPD and an issue that deserves further evaluation. In non-COPD subjects, other risk factors appear more consequential as risk factors for death, and TIMP-1 had a protective effect.
Lung function decline in COPD is proposed for evaluation across several years, as FEV
1 values may naturally fluctuate between different examinations carried out at shorter time intervals [
27]. Similar to the last point, the progression of COPD has been found to be heterogeneous and thus a long interval of follow up may be more indicative of lung function decline in a population than a shorter interval [
28]. In the current study, decline in lung function was assessed over a period of approximately five years, which can be considered satisfactory.
Phenotyping studies have explained the heterogeneity in COPD and might represent an opportunity to enhance diagnosis, predict outcomes and personalize treatments in patients. Suggested COPD phenotypes are based on clinical parameters such as; emphysema severity measured by CT, spirometry, nutritional status, exercise capacity and exacerbation frequency [
29,
30]. Presumably, one parameter could be associated to many phenotypes, and there is probably an overlap between phenotypes in the population. Based on the knowledge of AATD and COPD [
4], it seems appropriate to associate markers of increased proteolytic imbalance to greater parenchymal degradation and emphysema development. It is unclear to what extent various parameters and phenotypes define the complex syndrome of COPD, as presented in the population. However, proteolytic imbalance appears to be a factor that is significant on a population level [
22,
31], and the results generate hypotheses for future studies of COPD phenotypes.
The distribution of COPD severity in our cohort is comparable to what has been reported from other population-based studies, including individuals with predominantly mild- to moderate COPD [
20,
32,
33]. COPD was based on post-bronchodilator spirometry, and both internal and external validity are considered good. In a population-based study, the results will not be affected by the under-diagnosis of COPD, allowing a discussion of generalizability to COPD in the general population.
Limitations of this study that merit further discussion are firstly that the analysis of MMP-9 and TIMP-1 includes both total and pro-enzyme levels. Secondly, serum MMP-9 levels may not reflect overall MMP-9 airway activity, since enzyme levels do not directly reflect enzyme activity. Further, in the present study, all samples have been collected during field studies and stored at − 20 °C. Thus, a decrease in MMP-9 levels across time, cannot be excluded [
4]. The possible effects of storage are expected to affect all samples in an equivalent manner, as they were all collected within the same year, but the geometric decrease in measurable enzymes should be the same for all samples. The measured absolute values may be influenced by storage, but not likely the results, regarding the observed correlations. Furthermore, prognosis was assessed as all-cause mortality and the association between these biomarkers and specific cause of death would be of interest. However, unfortunately, data on cause-specific mortality were not available when this manuscript was completed.
Conclusions
In this population-based study, we could not demonstrate the expected association between the measured biomarkers and lung function decline among subjects with COPD. This result may, after analyses of non-participation, be explained by a healthy survivor effect. Increased MMP-9 serum levels and MMP-9/TIMP-1 ratio, indicating increased proteolytic imbalance, were associated with mortality in both subjects with and without COPD. Among subjects with COPD, both these biomarkers were associated with an increased risk for death also when adjusted for common confounders. We propose that the prognostic value of systemic markers of increased proteolytic imbalance can be translated to a population level by this study, which includes predominantly mild and moderate COPD. Future longitudinal studies are important for the further understanding of MMP-9 and TIMP-1 in relation to the pathogenesis and prognosis of different COPD phenotypes.
Acknowledgements
The authors acknowledge Bo Lundbäck, the founder of the OLIN-studies and initiator of the OLIN COPD-study. The authors thank Helena Backman and Hans Stenberg for statistical support, and Ann-Christin Jonsson, Sigrid Sundberg and Linnea Hedman for data collection.