Background
Recent investigations showed an increased mortality in COPD patients with elevated uric acid (UA) levels [
1,
2] and described UA as an independent predictor of 30-day mortality of acute exacerbations [
2]. UA is known to be associated with markers of systemic inflammation [
3], bronchoconstriction by stimulation of endothelin-1 [
4,
5], as well as oxygen desaturation [
6].Both lower and higher levels of UA have been described as risk factors for airway obstruction [
7,
8], also in addition to restrictive pattern linked to UA [
9]. Therefor the worse outcome of COPD patients with hyperuricemia seems to involve a number of factors including systemic inflammation, oxygen desaturation and lung function alterations. Hyperuricemia is also associated with an increasing risk of coronary heart disease [
10], a comorbidity that is relevant for mortality in COPD patients [
11]. As an overall marker of functional capacity in COPD the 6-min walk distance (6-MWD) is well established and known to be a stronger predictor of mortality than other markers of severe COPD [
12]. In view of the complexity of the disease and interdependence of parameters, the functional changes related to uric acid [
2,
13] may well include changes in 6-MWD, in addition to associations with comorbidities and exacerbations, even if common risk factors such as age, gender, smoking and body-mass index (BMI) have been taken into account. However, the multiple associations between parameters may render it difficult to quantify the causal role of UA and to separate direct and indirect effects from each other. This can be done using path analysis as a tool to integrate and cross-check the results of conventional regression analyses that have been performed for single outcome measures but never put into a comprehensive picture.
Based on this the aim of this study was to identify the role of the biomarker UA for spirometric parameters, 6-min walk distance, exacerbation rate and cardiovascular comorbidities in COPD while taking into account the fact that these outcome measures are related to each other. The hypothesis was that UA has direct effects on these measures that cannot be explained by their mutual relationships and common risk factors. Such information could be helpful to understand pathophysiological mechanisms and the responses to therapy. For this purpose we used data from the German COPD cohort study COSYCONET (COPD and Systemic Consequences-Comorbidities Network).
Discussion
Previous studies have indicated a role of uric acid (UA) for mortality, exacerbations and lung function in COPD [
1,
2]. The results of our cross-sectional analysis are in line with the different findings and more closely identify its role within the network of functional parameters, exacerbations, cardiovascular comorbidities and risk factors. Beyond conventional regression analyses we used the approach of path analysis to account for both direct and indirect effects of UA as closely as possible. Higher levels UA were linked to higher airway obstruction in terms of FEV
1, lower physical capacity in terms of 6-MWD and a greater number of cardiovascular comorbidities. Both FEV
1 and 6-MWD mediated an indirect effect of UA on exacerbations as defined by GOLD 2017. In the absence of hyperuricemia-specific medication there was even a direct link from UA to exacerbations, probably due to the fact that untreated patients had higher uric acid levels (see Fig.
1). Overall, our results were not critically dependent on the inclusion of patients with the diagnosis of hyperuricemia or the presence of specific mediation. The major role was played by UA, strongly suggesting a causative role of UA itself for clinically relevant outcomes in COPD.
The observed association of lung function with UA is in line with previous results regarding FVC and FEV
1 in lung healthy subjects [
9], or FEV
1 in COPD patients [
2]. In our study population we found both FVC and FEV
1 to be linked to UA. In parallel to findings in patients with pulmonary hypertension [
21], we observed that 6-MWD was also linked to UA in patients with COPD. Moreover, UA levels have been shown to be related to exacerbations in patients with COPD [
2], as well as cardiovascular comorbidities [
22]. Both findings were confirmed by our data which therefore are fully compatible with the known link between UA and mortality [
1].
UA levels are known to be influenced by a variety of factors, among them overweight, age and male gender. These risk factors were also significant predictors in our data. In the path analysis gender was implicit and taken into account by the use of appropriately adjusted values. As excretion rate has an effect on UA levels [
9], we tentatively also included creatinine as a biomarker in additional analyses. Importantly, this did not lead to a change of the role played by UA in our network. Higher values of BMI were associated with higher values of FEV
1, which is a common finding in COPD [
19]. At the same time, there were associated with higher levels of UA, however UA itself had a negative effect on FEV
1, pointing towards greater airflow limitation. Taken together, this observation again underlines that UA itself has effects that are explained by common risk factors like BMI.
Thus, patients with hyperuricemia showed different ranges of UA depending on the presence of specific medication as illustrated in Fig.
1. The associations identified by path analysis became even stronger when excluding patients with specific medication. Remarkably, most of the previous studies investigated the role of UA in COPD irrespective of a diagnosis of hyperuricemia. As there were no major differences when including or excluding these patients, our data support this approach. The most important factor seemed to be the level of UA not the diagnosis. On the other hand, our observations regarding the additional link between UA and exacerbations indicate that the presence of hyperuricemia-specific medication can modulate the results, suggesting that future analyses should take account of diagnosis, medication and biomarker. As a secondary finding, Fig.
1 demonstrates the effectiveness of hyperuricemia-specific therapy in our study population but also shows that UA levels were still higher than in patients without the diagnosis of hyperuricemia.
Beyond lung function and physical capacity, the rate and severity of exacerbations are determinants of the prognosis in COPD [
12,
23]. We coded exacerbations through a binary variable equivalent to the difference between the AB and CD groups in the recent GOLD recommendations [
15]. This was motivated by the aim to use a definition close to that established for therapeutical decisions. Exacerbations defined in this way were dependent on lung function and by themselves had a negative effect on 6-MWD, in accordance with previous data [
24,
25]. The link to 6-MWD appears plausible since exacerbations often lead to an irreversible deterioration of clinical state. Despite the indirect effects of UA mediated via both exacerbations and FEV
1, it was also directly associated with 6-MWD. We do not know whether this could be partially due persistent motoric impairment from gout arthritis even in the absence of acute episodes.
In the path analysis model exacerbations were dependent on FEV
1. The independent effect of UA on exacerbations was statistically significant only when excluding patients with hyperuricemia-specific medication. This seems understandable due to the reduction in the range of variation of UA compared to patients with hyperuricemia without specific UA-lowering medication. These observations support the hypothesis that UA influences the risk/severity of COPD exacerbations. Taking into account the different paths linking UA to exacerbations, our findings suggested that an increase in UA level by 2 mg/dl was linked to a shift by about 5% from the low to the high exacerbation category. A potential pathophysiological link could be endothelin-1 which is known to be elevated in hyperuricemia, but also in asthma and COPD exacerbations [
5]. Furthermore, pro-inflammatory effects of UA in terms of TNF-alpha activation have been described [
26]. The associations of the cardiovascular comorbidity count with UA but also functional measures and exacerbations suggest that UA triggers pro-inflammatory processes that are relevant for cardiovascular diseases [
22] which in turn have an impact on COPD prognosis [
27]. Taken together, it might well be that UA exerts part of its effects on functional state and mortality of COPD via a set of inflammatory compounds.
Limitations and strength
Obvious limitations of the study are its cross-sectional and non-interventional design. Furthermore, we did not have data on the history of clinical manifestations of gout, in terms of acute episodes. Patients with the physician-based diagnosis of hyperuricemia are potentially more health-conscious and may therefore have a better physical capacity and prognosis, which could influence our results. Moreover, there was no information on the specific medical reasons why patients received hyperuricemia-specific treatment. Also, there was no information on the duration of the hyperuricemia-specific treatment. This, however, appeared to be of minor importance, since the major results were independent of excluding or including these patients. Long-term follow-up data could reveal whether the relationships observed in a cross-sectional analysis are also reflected in the relationships between the changes of parameters within the follow-up period. The strength of our study was its large sample size which allowed the use of sophisticated statistical methods with the aim to disentangle the role of UA from other influencing factors. Moreover we could rely on comprehensive, high quality data of lung function and patients’ clinical characteristics.
Acknowledgements
We thank all patients of COSYCONET for their kind cooperation and all study centers for their excellent work. Moreover, we are grateful to the Scientific Advisory Board of COSYCONET for continuing support und helpful recommendations. The members of the board are: Edwin J.R. van Beek (Clinical Research Imaging Centre (CRIC), The Queen’s Medical Research Institute, University of Edinburgh, UK), Klaus Friedrich Rabe (LungenClinic Grosshansdorf, Zentrum für Pneumologie und Thoraxchirurgie, Grosshansdorf, Germany), Joseph M. Antó (Universitat Pompeu Fabra, Barcelona, Spain), Philippe Grenier (French Society of Radiology (SFR), Paris, France), Norbert Krug (Frauenhofer Institut für Toxikologie und experimentelle Medizin, Hannover, Germany), Michael Kiehntopf, Universitätsklinikum Jena, Institut für klinische Chemie und Laboratoriumsmedizin, Jena, Germany), Jørgen Vestbo (University of Manchester and South Manchester University Hospital NHS Foundation Trust, Manchester, UK), Emiel F. Wouters (Maastricht University Medical Center, Maastricht, The Netherlands).