Circulating soluble urokinase plasminogen activator

Original Article

| doi: 10.1111/j.1365-2796.2010.02252.x

Circulating soluble urokinase plasminogen activator receptor predicts cancer, cardiovascular disease, diabetes and mortality in the general population J. Eugen-Olsen1,*, O. Andersen1,2,*, A. Linneberg3, S. Ladelund1, T. W. Hansen4, A. Langkilde1, J. Petersen1, T. Pielak1, L. N. Møller3, J. Jeppesen5, S. Lyngbæk5, M. Fenger6, M. H. Olsen3, P. R. Hildebrandt5, K. Borch-Johnsen7,8, T. Jørgensen3,9 & S. B. Haugaard1,10 From the 1Clinical Research Centre, Copenhagen University, Hvidovre Hospital, Hvidovre; 2Department of Infectious Diseases, Copenhagen University, Hvidovre Hospital, Hvidovre; 3Research Centre for Prevention and Health, Copenhagen University Hospital, Glostrup; 4Department of Clinical Physiology, Nuclear Medicine, and PET, Faculty of Health Sciences, Copenhagen University Hospital, Rigshospitalet, Copenhagen; 5 Department of Medicine, Faculty of Health Sciences, Copenhagen University Hospital, Glostrup; 6Department of Biochemistry, Copenhagen University, Hvidovre Hospital, Hvidovre; 7Steno Diabetes Center, Gentofte; 8Faculty of Health Science, University of Aarhus, Aarhus; 9Faculty of Health Science, University of Copenhagen, Copenhagen; and 10Department of Endocrinology, Copenhagen University, Hvidovre Hospital, Hvidovre, Denmark

Abstract. Eugen-Olsen J, Andersen O, Linneberg A, Ladelund S, Hansen TW, Langkilde A, Petersen J, Pielak T, Møller LN, Jeppesen J, Lyngbæk S, Fenger M, Olsen MH, Hildebrandt PR, Borch-Johnsen K, Jørgensen T, Haugaard SB (Copenhagen University, Hvidovre Hospital, Hvidovre; Copenhagen University Hospital, Glostrup; Copenhagen University Hospital, Copenhagen; Copenhagen University Hospital, Glostrup; Copenhagen University, Hvidovre Hospital, Hvidovre; Steno Diabetes Center, Gentofte; University of Aarhus, Aarhus; University of Copenhagen, Copenhagen; Copenhagen University, Hvidovre Hospital, Hvidovre, Denmark). Circulating soluble urokinase plasminogen activator receptor predicts cancer, cardiovascular disease, diabetes and mortality in the general population. J Intern Med 2010; doi: 10.1111/j.1365-2796.2010.02252.x. Background. Low-grade inflammation is thought to contribute to the development of cardiovascular disease (CVD), type-2 diabetes mellitus (T2D), cancer and mortality. Biomarkers of inflammation may aid in risk prediction and enable early intervention and prevention of disease. Objective. The aim of this study was to investigate whether plasma levels of the inflammatory biomarker soluble urokinase plasminogen activator receptor (suPAR) are predictive of disease and mortality in the general population. Design. This was an observational prospective cohort study. Cohort participants were included from June 1993 to December 1994 and followed until the end of 2006.

Setting. General adult Caucasian population. Participants. The MONICA10 study, a population-based cohort recruited from Copenhagen, Denmark, included 2602 individuals aged 41, 51, 61 or 71 years. Measurements. Blood samples were analysed for suPAR levels using a commercially available enzyme-linked immunosorbent assay. Risk of cancer (n = 308), CVD (n = 301), T2D (n = 59) and mortality (n = 411) was assessed with a multivariate proportional hazards model using Cox regression. Results. Elevated baseline suPAR level was associated with an increased risk of cancer, CVD, T2D and mortality during follow-up. suPAR was more strongly associated with cancer, CVD and mortality in men than in women, and in younger compared with older individuals. suPAR remained significantly associated with the risk of negative outcome after adjustment for a number of relevant risk factors including C-reactive protein levels. Limitation. Further validation in ethnic populations other than Caucasians is needed. Conclusion. The stable plasma protein suPAR may be a promising biomarker because of its independent association with incident cancer, CVD, T2D and mortality in the general population. Keywords: biomarker, low-grade inflammation, prognosis, risk, suPAR.

ª 2010 Blackwell Publishing Ltd

1

J. Eugen-Olsen et al.

|

suPAR: a novel risk marker for major disease and mortality

Introduction ‘Low-grade inflammation’ (LGI) is an undefined subclinical chronic inflammatory state, which is thought to contribute to the development of cardiovascular disease (CVD) [1], type 2 diabetes mellitus (T2D) [2], cancer [3] and Alzheimer’s disease [4]. The most commonly used biomarker of LGI is C-reactive protein (CRP) measured using a high-sensitivity (hsCRP) assay; the plasma CRP concentration is associated with an increased risk of CVD [5] and, in some studies, with cancer mortality and total mortality [6]. The urokinase plasminogen activator receptor (uPAR) is expressed on a number of different cells; in particular, on vascular endothelial cells, monocytes, neutrophils and activated T-cells. It is involved in several immune functions including migration, adhesion, angiogenesis, fibrinolysis and cell proliferation [7–9]. The uPAR is released from cells during inflammatory stimulation to generate soluble uPAR (suPAR) that is a highly flexible molecule [10] with intrinsic chemotactic properties [11, 12]. Soluble urokinase plasminogen activator receptor levels are positively correlated with pro-inflammatory biomarkers such as tumour necrosis factor-a and leucocyte counts [13] and with C-reactive protein levels [14]. An elevated suPAR level is thought to reflect activation of the inflammatory and immune systems and has been associated with poor clinical outcomes in patients suffering from various infectious diseases [15–18] as well as in those with certain types of cancers [19–21]. As suPAR is likely to be a central player in the mechanisms of LGI and shows stable kinetics both in vivo [13] and in vitro [22], we investigated the potential of suPAR as a risk marker for common diseases and death in the general population. This study included 2602 men and women aged 41, 51, 61 or 71 years who were living in the community in the vicinity of Copenhagen and who were enrolled into the Danish MONICA10 cohort study in 1993–1994 to be followed for a median of 13 years. Cohort description The Danish contribution to the international MONICA project (MONItoring trends and determinants of CArdiovascular disease), a study conducted under the auspices of the World Health Organisation (WHO), was undertaken between 1982 and 1991. The Danish MONICA1 population survey took place

2

ª 2010 Blackwell Publishing Ltd Journal of Internal Medicine

at the Research Center for Prevention and Health in Glostrup from 1982 to 1984 and included 4807 individuals born in 1923, 1933, 1943 or 1953 who were randomly selected from 11 municipalities within Copenhagen County [23, 24]. The participation rate was 78.7%. In 1993–1994, 2656 formerly invited individuals (55%) agreed to participate in MONICA10 (see flow chart in Fig. 1). For this, blood pressure measurements and plasma samples obtained between June 1993 and December 1994 were available from 2605 participants who completed a self-administered questionnaire on CVD risk factors, medical history and lifestyle habits including smoking and physical activity. All participants gave written consent and the study was conducted in accordance with the second Declaration of Helsinki and approved by the Ethics Committee for Copenhagen County. Blood pressure was measured using a random-zero mercury sphygmomanometer: two measurements were taken whilst the participant was sitting down and at rest for 5 min; the mean value was used for analysis. Laboratory measurements Fasting concentrations of HDL cholesterol and total cholesterol were measured using enzymatic colorimetric methods (Roche, Mannheim, Germany), as previously described [25]. Fasting concentrations of blood glucose were analysed by standard methods [26]. CRP levels were measured using a particle-enhanced immunoturbidimetric hsCRP assay (Roche ⁄ Hitachi, Basel, Switzerland) with a range of 0.1–20 mg L)1 and a detection limit of 0.03 mg L)1 as previously described [27]. Plasma levels of suPAR were measured using the commercially available suPARnostic kit, according to the manufacturer’s instructions (ViroGates, Copenhagen, Denmark). The intra-assay variation was 2.75% and inter-assay variation was 9.17%. Sixteen samples showed more than 10% variation, and were therefore re-measured. The kit standard curve was validated to measure suPAR levels between 0.6 and 22.0 ng mL)1. The technician who measured the samples (TP) and the head of the laboratory (JEO) were blinded to the identity of the patient samples. The clinical database was released by the Research Centre for Prevention and Health (AL and LNM) after having received the suPAR data. The duration of sample freezing did not appear to have a major influence on the plasma level of suPAR as indicated by the lack of correlation between suPAR levels and the date of


|


MONICA (including from 1982–1984) 4807 randomly chosen from CPR register 226 of non-Danish origin excluded 4581 invited 3785 included (79%) 428 died 23 immigrated MONICA10 (including from 1993–1994) 4130 invited 2656 included (64%) 51 with no available samples 3 with suPAR below or above validated assay range

2602 with suPAR measurement

Cancer: 60 excluded due to previous cancer diagnosis and 144 excluded for missing one or more covariates CVD: 151 excluded due to previous CVD diagnosis and 147 excluded for missing one or more covariates Diabetes: 81 excluded due to baseline diabetes and 161 excluded for missing one or more covariates Mortality: 240 excluded for missing one or more covariates

Data available for multivariate endpoint analysis •Cancer: n = 2398 •CVD: n = 2304 •Diabetes: n = 2360 •Mortality: n = 2362

Fig. 1 Flow-chart of participation in MONICA and MONICA 10.

plasma sampling from 14 June 1993 to 2 December 1994 (q = 0.001, P = 0.96). However, long-term freezer storage may lead to water evaporation and increased protein levels [28]. Evaporation is likely to induce nondifferential misclassification that would tend to draw the risk estimates to the null hypothesis. The median sample freezer storage duration was 13.6 years (range, 12.8–14.3). Outcomes Information regarding morbidity at baseline and during follow-up was obtained from the Danish National Patient Register (NPR) [29]. Mortality data were obtained from the Danish National Death Register. Participants were followed from the time of blood sampling (1993–1994) to 31 December 2006. At the time of blood sampling, the participants were 41, 51, 61 or 71 years old. We investigated four different outcomes in the study population. 1 Cancer of any kind. Both nonfatal and fatal cancers were included in the cancer end-point analysis (WHO international classification of diseases (ICD)-10 codes C00–C97) that included lung cancers (C32, C33, C34 and D38.1), gastrointestinal cancers (C00–

C26), prostate cancer (C61) and breast cancer (C50). Patients with a registered diagnosis of cancer prior to the time of entry into the study were excluded from the analysis of this end-point. 2 CVD. The CVD end-point was the combination of cardiovascular death (ICD-8 codes 390–448 or ICD10 codes I00–I79 and R95–R99), ischaemic heart disease (ICD-8 codes 410–414 or ICD-10 codes I20–I25) and stroke (ICD-8 codes 431, 433 and 434 or ICD-10 codes I61 and I63), as previously described in this cohort [30]. Patients with a prior diagnosis of myocardial infarction or stroke, or those taking digoxin or nitrates, were excluded from the analysis of this end-point. 3 T2D. Individuals with self-reported diabetes at baseline, hospitalizations with discharge diagnosis including diabetes prior to the first examination (ICD-8 code 250 and ICD-10 codes E10–E14) or a fasting plasma glucose level above 6.9 mmol L)1, or use of antidiabetic drugs at the baseline examination were all classified as prevalent cases of diabetes and consequently excluded from the analysis of this endpoint. Incident cases of diabetes were defined as individuals who during the follow-up were registered for


3

|


the first time in the NPR with ICD-8 code 250 or ICD10 codes E10–E14. 4 Death from any cause. Analysis was restricted to 2362 individuals with complete information for all studied explanatory variables. The number of individuals diagnosed with either cancer, CVD or T2D at baseline is shown in Fig. 1. Statistical analysis End-points are presented within birth cohort- and gender-specific suPAR-quartiles, and the association was tested with the chi-square test. A Cox proportional hazards model with age as time and delayed entry was fitted to all four end-points. The proportional hazards assumption was checked by examination and test of the Schoenfeld residuals. suPAR and CRP were estimated in birth cohorts as the Cox regression model control showed an interaction between suPAR ⁄ CRP and age at examination (nonproportionality). Controls revealed that linear scoring of suPAR on the log hazard scale was indeed appropriate. Results are presented as hazard ratios with 95% confidence intervals and P-values. Furthermore tests for trend are presented for suPAR and CRP. Tests for departure from trend were preformed (data not shown) and were nonsignificant in all cases. Age- and gender-adjusted Spearman partial correlation was used to determine the relation between plasma levels of suPAR ⁄ CRP and selected cardiometabolic risk factors. Cumulative incidence plots were made for all disease endpoints. Death was considered a competing end-point rather than censoring. Our cumulative incidence plots thus show the lifetime risk of disease, given disease-free status at the age of 41. Gender- and age (cohort)-specific suPAR quartiles were used in the cumulative incidence plots. An analysis of death and suPAR quartiles is presented in the form of an age-specific Kaplan–Meier plot. The area between the curves for the first and fourth quartiles was calculated as an estimate of the difference in expected lifetime. Role of the funding source. The suPARnostic kits were kindly donated by ViroGates A ⁄ S. ViroGates had no role in the design of the study, in the collection, analysis or interpretation of data or in the decision to submit the manuscript for publication. Copenhagen University Hospital Hvidovre holds patents on the

4


use of suPAR in prognostics with Drs Eugen-Olsen, Andersen and Haugaard cited as inventors. Results Plasma suPAR levels in the MONICA10 cohort Plasma samples from 2605 individuals were measured and suPAR was quantified in 2602 of these samples (Fig. 1); three samples were excluded because they fell below (n = 2) or above (n = 1) the validated range of the suPAR assay. The median plasma suPAR level was 4.03 ng mL)1 (range, 1.3–19.9). suPAR levels increased with age and were higher amongst women {n = 1292; median suPAR, 4.26 ng mL)1 [interquartile range (IQR), 3.60–5.13]} than men (n = 1310; median suPAR, 3.84 ng mL)1 (IQR, 3.14–4.71), P < 0.001). Smoking significantly increased suPAR in all age groups as shown in Fig. 2. Baseline cohort characteristics by quartiles of suPAR are shown in Table 1. Follow-up and end-points We studied four different end-points in the population: cancer, CVD, T2D and death. Participants were followed for end-points from the time of blood sam-

Men

Women

9 8 7

suPAR (ng/ml)


6 5 4 3 2 1 41

51

61 Age

71

41

51

61

71

Age

Fig. 2 Box-plot of suPAR levels amongst nonsmokers (white boxes) and current smokers (grey boxes) according to age and gender. Boxes represent 25–75% percentiles and whispers are 5–95% percentiles. P < 0.05 for all age groups.

Glucose (mmol L)1)

Metabolic variables

Diastolic blood pressure (mmHg)

Nonsmoker ⁄ ex-smoker ⁄

variables

Smoking status

5.4 (3.8–8.1)

75 ⁄ 5 ⁄ 20

82 (67–101)

123 (102–157)

5.8 (3.9–8.8) 82 (66–100)

128 (103–164) 61 ⁄ 5 ⁄ 34

Data are presented as median and 5–95% percentiles, or frequency (in percentage).

regular smoker (%)

Systolic blood pressure (mmHg)

Haemodynamic

Leucocytes (1 · 109 cells L)1)

variables

1.50 (0.30–7.97)

87 (69–111)

88 (70–108) 1.13 (0.23–6.79)

Waist circumference (cm)

C-reactive protein (mg L)1)

25.2 (20.7–33.8)

4.7 (4.2–5.6)

1.10 (0.56–2.90)

1.45 (0.94–2.33)

6.08 (4.48–8.16)

2.0

6.8

7.1

25.3 (20.5–32.0)

4.7 (4.1–5.6)

Inflammatory

Body mass index (kg m)2)

1.37 (0.93–2.09)

Triglycerides

(mmol L)1) 1.15 (0.61–3.06)

HDL cholesterol

variables

1.2

History of cancer (%) 5.95 (4.55–7.86)

2.9

History of CVD (%)

Total cholesterol

6.0

Treatment of hypertension (%)

2.6

51 (41–71)

50.3

‡3.4–£4.0 ng mL

50 ⁄ 4 ⁄ 46

81 (65–99)

129 (101–163)

6.4 (4.5–10.3)

1.98 (0.40–13.92)

87 (69–108)

25.7 (20.3–33.9)

4.8 (4.1–6.4)

1.22 (0.66–3.13)

1.40 (0.91–2.23)

6.19 (4.50–8.07)

3.1

6.0

10.3

4.9

61 (41–71)

56.2

>4.0–£4.9 ng mL

)1

30 ⁄ 2 ⁄ 68

81 (65–102)

131 (102–171)

7.3 (4.6–11.3)

3.20 (0.44–17.44)

87 (68–111)

25.2 (19.6–34.7)

4.8 (4.0–6.7)

1.31 (0.68–3.28)

1.33 (0.85–2.24)

6.14 (4.58–8.12)

3.5

9.1

14.3

5.9

61 (41–71)

58.2

>4.9–19.9 ng mL)1

Range:

4th (n = 650) P-value