Soluble urokinase plasminogen activator receptor predicts mortality in exacerbated COPD

Chronic obstructive pulmonary disease (COPD) is worldwide a leading cause of morbidity and mortality, posing a considerable socioeconomic burden to society [1]. COPD is a highly complex disease with symptoms not only related to the respiratory system. In many patients with COPD, the course of the disease will be progressive with a gradual decline in health status and increasing frequency and severity of acute exacerbations (AECOPD). This progressive course of the disease is associated with a substantial increase in all-cause mortality. In the clinical assessment of the severity of the disease, the GOLD classification is often applied [1]. However, improved risk prediction in patients with COPD would be of great value, in particular if it additionally would facilitate identification of terminal ill patients in need of enhanced preventive efforts, palliative care or other interventions.

Several biomarkers have, in studies of stable COPD, been associated with disease severity and mortality risk [2]. A history of frequent exacerbations of COPD is per se a marker of poor prognosis [3]. However, diagnostic and/or prognostic markers in unstable disease have so far provided variable results. In the ECLIPSE cohort, previous exacerbations, older age, high white blood cell count, emphysema, severe airflow limitation and poor health status were associated with increased risk of future hospitalizations [4]. A meta-analysis of routine blood-tests in AECOPD found that anaemia, hypoalbuminemia and elevated levels of the cardiac biomarkers pro-BNP and troponin-T, but not high-sensitivity C-reactive protein (hs-CRP) predicted mortality from COPD [5]. By contrast, a recent systematic review found CRP to be the only robust biomarker showing consistently elevated levels in AECOPD compared with control groups [6]. Recently, a readmission and mortality risk prediction score, comprising previous admissions, extended dyspnoea grade, age and heart failure but no biomarkers, has been proposed [7].

Previous studies have shown that the urokinase plasminogen activator receptor (uPAR) is involved in the pathogenesis (both small airway fibrosis and emphysema) of COPD through complex molecular pathways and gene expression patterns [8‐10]. Soluble urokinase plasminogen activator receptor (suPAR) is the soluble form of the uPAR receptor and can be measured in blood/plasma. suPAR is considered an inflammatory biomarkerand is elevated in both acute and chronic illness. It is a non-specific risk marker with a diagnostic cut-off value of approximately 3 ng/ml [11]. However, little is known about suPAR in relation to disease status and mortality in COPD. suPAR has previously in a small study been found to be superior to CRP and fibrinogen in the prediction of AECOPD and in monitoring treatment of exacerbation [12]. Furthermore, suPAR levels are increased in stable COPD compared to healthy controls [13]. Finally, suPAR has been shown to predict a range of chronic conditions and mortality in the general population as well as being associated with risk of readmission and mortality in a large group of acutely admitted medical patients, with AECOPD constituting a subgroup of these [14, 15].

In this study, we aimed to investigate the value of suPAR levels in plasma upon hospital admission as a prognostic biomarker for short-term mortality in COPD patients based on a large, unselected population of acutely admitted medical patients.

SuPAR was measured in > 95% of all acutely admitted patients. During the inclusion period, a total of 717, 2573 and 265 patients with AECOPD, any COPD and Respiratory failure due to COPD, respectively, were admitted to the AMU and had suPAR measured. The median age was 72.8, 73.4 and 73.4 years, respectively, and median length of stay was 3.2, 2.2 and 5.9 days, respectively. In all three groups there were slightly more women than men who were admitted with COPD or respiratory failure (Table 1). In the group with AECOPD, 6.4% died within 30 days from the index admission, and 13.4% died within 90 days. For patients with any COPD or Respiratory failure the numbers were 9.1, 14.4 and 20.8% and 26.8%, respectively.

This retrospective cohort study in a population of patients admitted to a University hospital with AECOPD or any acute medical condition with underlying COPD showed that the inflammatory biomarker suPAR was independently associated with all-cause mortality. Furthermore, we showed that median suPAR levels were higher among those who died within 90 days of admission compared with those who survived in all three groups of respiratory patients. Interaction analyses showed that these effects were not mediated through concomitantly elevated CRP, although this biomarker was also independently associated with mortality, but less consistently compared to suPAR. This is consistent with previous findings and the value of CRP in combination with other biomarkers as indicators of poor prognosis in COPD has also recently been confirmed in a study comprising patients from the COPDGene- and the ECLIPSE cohorts [6, 20].

To our knowledge, this is the first large-scale study investigating the relationship between this recently introduced prognostic biomarker and COPD. The mortality rate in the present study did not differ from other studies of mortality in relation to severe AECOPD [21]. The magnitude of the association corresponds well with other studies that specifically have addressed the role of suPAR as prognostic indicator of mortality in acutely ill patients [15, 22, 23]. However, the vast majority of previous studies have investigated suPAR in cohorts of patients with systemic inflammatory response syndrome (SIRS) or sepsis in an intensive care unit setting, which makes comparisons of the present study with previous findings challenging. The sensitivity and specificity of suPAR regarding mortality in the present study is comparable with pooled values from a recent systematic review of suPAR and bacterial infections [24]. A very recent study of chronic heart failure (CHF), a condition which has the chronic, systemic inflammation in common with COPD, found that elevated suPAR levels approximately doubled the risk of subsequent mortality during 3 years of follow-up [25]. This was, however, stable patients in an outpatient setting. A Danish study of 449 patients consecutively admitted to hospital with chest pain and followed for up to 6.6 years found that suPAR was associated with increased all-cause mortality independent of other variables reflecting myocardial ischaemia (HR 1.93) [26]. Only one small cross-sectional study comprising 43 patients has measured suPAR upon hospitalisation with AECOPD and found it to be higher than in healthy controls [12] and also that suPAR correlated negatively with level of FEV₁. Furthermore, suPAR levels decreased after 7 days, but whether this was a treatment effect or occurred spontaneously remains unknown. Two other small studies of stable COPD patients matched with healthy control subjects found opposite effects regarding suPAR levels [13, 27]. Thus, the value of adding suPAR to routine blood samples in hospitalised AECOPD for subsequent risk prediction has so far been largely unexplored. It has previously been demonstrated that the cell-bound form of suPAR (uPAR) is highly expressed in the small airway epithelia of patients with COPD compared with normal controls and that increased uPAR expression in the small airway epithelium of patients with COPD participates in an active epithelial-mesenchymal transition process [8, 9]. Assuming that increased uPAR expression results in increased uPAR cleavage and suPAR generation, the observations by Wang and co-workers are supported by our observation of increased suPAR and worse prognosis in COPD.

Given the pathophysiological mechanisms of most cases of AECOPD, namely a flare-up in acute local and systemic inflammatory mediators and airway bacterial or viral infection, [28] a high level of an unspecific inflammatory biomarker such as suPAR is expected. As suPAR is a very stable biomarker, it may be less affected by the exacerbation and more reflective of the underlying chronic condition, which may explain its high prognostic power in AECOPD.

In AECOPD no single biomarker or clinical test, apart from sustained respiratory acidosis measured in an arterial puncture, can accurately predict the immediate or short-term (less than 3 months) outcome [29]. Previous studies have proposed either a panel of biomarkers, a composite measure of clinical features and blood tests e.g. as a “prediction score tool” or simply an assessment of demographic variables for determining the risk for hospital readmission or death from COPD [2‐7]. In terms of novel biomarkers of prognosis in COPD, the study by Stolz et al. [2] found that simultaneously elevated plasma levels of the three biomarkers adrenomedullin, arginine vasopressin and atrial natriuretic peptide were associated with increased mortality risk during a 5-year follow-up in patients with stable COPD. An assessment of adrenomedullin alone or in combination with composite measures such as age, grade of dyspnea and degree of airflow obstruction conveyed an increased mortality risk comparable to the observations in the present study [30]. However, the patients in this study were derived from several cohorts of patients who also had stable disease. Furthermore, a number of the scoring tools that have been developed for predicting future risk of readmission or death from AECOPD are quite complex or include measures of cardiovascular comorbidity, which may not be readily accessible in the acute setting [7, 31]. None of the scoring systems seem to agree upon, which few key parameters are considered superior in the prediction of severe and/or fatal incidents of AECOPD. The latest publication proposes an 8-item scoring tool for predicting 1-year mortality after a first hospitalization for AECOPD [32]. The variables associated with increased risk included older age, male sex, having a severe exacerbation, longer duration of the disease and hospital stay, prior hospital admissions, a diagnosis of cancer and a higher Charlson score. A number of these were already predefined as important confounders and included as covariates in the present study [15]. Interestingly, a study aimed at predicting 12-month mortality in elderly patients presenting to the emergency ward, by asking the attending physicians the so-called “surprise question” (“Would you be surprised whether this patient died in the next 12 months?”) suggested a fair sensitivity but poor specificity [33]. When combined with other determinants of unfavourable outcome, although not including any blood-based biomarkers, results remained unchanged.

For admission with respiratory failure and COPD as secondary diagnosis we found no statistical significant relationship between suPAR and mortality, although there was a tendency when entering suPAR as a continuous variable for death by 90 days (p = 0.06, Table 3). This may in part be due to the small numbers, especially when analysis is performed by suPAR quartiles. Nevertheless, this finding is somewhat unexpected and needs further investigation. Patients with acute, hypercapnic respiratory failure have an even higher risk of in-hospital or short-term mortality than patients belonging to either of the other groups, [29] which is why a stronger relationship would be anticipated in this group. The 30- and 90 days mortality rates were approximately doubled in the Respiratory failure group compared with the other groups, indicating that the negative results from the survival analyses represent a statistical lack of power rather than a true association.

Limitations of the present study include lack of data on smoking status as suPAR is confounded by smoking [34]. The Danish patient registries are validated and diagnoses are registered upon hospital discharge or in-hospital death but some inaccuracy may occur, although misclassification of the diagnoses in question is normally not a major source of bias [35]. Furthermore, this study did not discriminate the patients by other indicators of increased mortality risk such as number of previous exacerbations, need for non-invasive ventilation or referral to intensive care unit, level of FEV₁, dyspnoea grade, BMI or socioeconomic status. However, regarding FEV₁, the new GOLD guidelines recognize that a hospitalisation for COPD is more predictive of worsened prognosis than lung function [1]. Nevertheless we cannot rule out that patients with high levels of suPAR also had lower FEV₁. Also, measurement of blood-eosinophils was only available in a minority of the patients. Another limitation is that suPAR values were only measured at hospital admission, and thus it was not possible to assess whether this biomarker changed according to therapy or disease course in general. In future studies of AECOPD it would also be useful to analyse how suPAR can contribute to already existing prediction scores for readmission and/or death.

We have shown that suPAR levels are elevated in patients admitted to a large university hospital with COPD as primary or secondary diagnosis, who died within 30- or 90 days of follow-up compared with patients, who survived. The risk of death within 30- or 90 days after admission was doubled with increasing levels of suPAR for patients with AECOPD or any medical condition and underlying COPD, which is consistent with other studies. suPAR seems to be independent of- and better than CRP in discriminating an adverse outcome. It is therefore suggested that measurement of suPAR can be used as an indicator of poor prognosis in patients with more severe acute exacerbation of COPD.