Full research paper

European Journal of Preventive

Cardiology

Circulating biomarkers for long-term 2020, Vol. 27(6) 570–578

! The European Society of
Cardiology 2019
cardiovascular risk stratification in Article reuse guidelines:
sagepub.com/journals-permissions
apparently healthy individuals from DOI: 10.1177/2047487319885457
journals.sagepub.com/home/cpr
the MONICA 10 cohort

Charles Edward Frary1,2,*, Marie Kofoed Blicher3,*,

Thomas Bastholm Olesen3, Jacob Volmer Stidsen1,
Sara Vikström Greve4, Julie KK Vishram-Nielsen5,

Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021

Susanne Lone Rasmussen6,7, Michael Hecht Olsen2,8
and Manan Pareek9,10

Abstract
Aims: The aim of this study was to examine whether high-sensitivity C-reactive protein (hs-CRP), N-terminal pro-brain
natriuretic peptide (NT-proBNP), and soluble urokinase plasminogen activator receptor (suPAR) carried incremental
prognostic value in predicting cardiovascular morbidity and mortality beyond traditional risk factors in apparently healthy
individuals.
Methods and results: This was a prospective population-based cohort study comprising 1951 subjects included in the
10-year follow-up of the MONItoring of trends and determinants in CArdiovascular disease (MONICA) study, between
1993 and 1994. The principal endpoint was death from cardiovascular causes. Secondary endpoints were death from any
cause, coronary artery disease, heart failure, and cerebrovascular disease. Predictive capabilities of each of the three
biomarkers were tested using Cox proportional-hazards regression, Harrell’s concordance index (C-index), and net
reclassification improvement (NRI). Study participants were aged 41, 51, 61, or 71 years, and equally distributed between
the two sexes. During a median follow-up of 18.5 years (interquartile range: 18.1–19.0), 177 (9.1%) subjects died from a
cardiovascular cause. Hs-CRP (adjusted standardized hazard ratio (HR): 1.37, 95% confidence interval (CI): 1.17–1.60),
NT-proBNP (HR: 1.90, 95% CI: 1.58–2.29), and suPAR (HR: 1.35, 95% CI: 1.17–1.57) were all significantly associated with
cardiovascular deaths after adjustment for age, sex, smoking status, systolic blood pressure, and total cholesterol
(p < 0.001 for all). Furthermore, all three biomarkers were significantly associated with significant NRI. However, only
NT-proBNP significantly raised the C-index in predicting death from cardiovascular causes when added to the risk factors
(C-index 0.860 versus 0.847; p ¼ 0.02).
Conclusions: Hs-CRP, suPAR, and particularly NT-proBNP predicted cardiovascular death and may enhance prognos-
tication beyond traditional risk factors in apparently healthy individuals.

1
Department of Endocrinology, Odense University Hospital, Odense, 8
Denmark Centre for Individualized Medicine in Arterial Diseases, Department of
2
Cardiology Section, Department of Internal Medicine, Holbaek Hospital, Regional Health Research, University of Southern Denmark, Odense,
Holbaek, Denmark Denmark
9
3
Department of Internal Medicine, Hospital of Little Belt, Kolding, Department of Cardiology, North Zealand Hospital, Hilleroed,
Denmark Denmark
10
4
Section of Endocrinology, Department of Internal Medicine, Odense Department of Internal Medicine, Yale New Haven Hospital, Yale
University Hospital, Svendborg, Denmark University School of Medicine, New Haven, CT, USA
5
Centre for Cardiac, Vascular, Pulmonary, and Infectious Diseases,
Rigshospitalet, University of Copenhagen, Copenhagen, Denmark *These authors contributed equally as co-first authors.
6
Department of Clinical Physiology and Nuclear Medicine, Herlev Corresponding author:
Hospital, University of Copenhagen, Copenhagen, Denmark Manan Pareek, Department of Internal Medicine, Yale New Haven
7
Center for Clinical Research and Prevention, Bispebjerg and Hospital, Yale University School of Medicine, 20 York Street, New Haven,
Frederiksberg Hospital, The Capital Region of Denmark, Copenhagen, CT 06510, USA.
Denmark Email: mananpareek@dadlnet.dk
Frary et al.
Keywords
Assessment, risk, biomarkers, cardiovascular diseases, C-reactive protein, natriuretic peptide, brain, receptor, urokinase
plasminogen activator
Received 11 August 2019; accepted 9 October 2019
Introduction
Traditional risk assessment tools, such as the
Systematic COronary Risk Evaluation (SCORE) and
the Framingham Risk Score (FRS), calculate an
individual’s 10-year risk of cardiovascular mortality
and/or myocardial infarction based on clinical charac-
teristics and laboratory tests.1,2 Given the limited dis-
criminative ability of these approaches,3 several
additional non-circulating and circulating biomarkers
have been examined as means of refining risk stratifica-
tion, with potentially promising results.4,5 These bio-
markers include high-sensitivity C-reactive protein
(hs-CRP), N-terminal pro-brain natriuretic peptide
(NT-proBNP), and soluble urokinase plasminogen acti-
vator receptor (suPAR).6–11 Accordingly, the aim of
this study was to examine whether these three circulat-
ing biomarkers carried incremental prognostic value in
predicting cardiovascular events beyond traditional
cardiovascular risk factors in apparently healthy
individuals.
Methods
Study design
Between 1982 and 1984, a random sample of 4807 men
and women aged 30, 40, 50, or 60 years and residing in
the Greater Copenhagen Area were invited to take part
in the MONItoring of trends and determinants in
CArdiovascular disease (MONICA) project.12 A total
of 3971 individuals (82.6%) chose to participate.
Approximately 10 years later, between 1993 and 1994,
3785 of the original subjects (now aged 41, 51, 61, or 71
years), still living in the surrounding area, were asked to
participate in a follow-up examination, MONICA 10.4
A total of 2656 (70.2%) subjects participated. The study
complied with the Declaration of Helsinki and was
approved by the regional ethics committee. All partici-
pants provided written and verbal informed consent. For
the present analysis, all patients with a history of cardio-
vascular disease or diabetes (n ¼ 142), all those who
received anti-diabetic, anti-hypertensive, or lipid-lower-
ing therapy (n ¼ 312), or anyone who had missing data
for any of the three biomarkers were excluded (n ¼ 251).
Accordingly, the final sample consisted of 1951 subjects.
571
Baseline evaluation
A complete medical history (i.e. conditions, medi-
cations, and lifestyle, including smoking habits) was
collected using a self-administered questionnaire.
Height and weight were measured, and body mass
Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021
index was calculated. Blood pressure was measured
twice using a random zero mercury sphygmoman-
ometer in the sitting position after 5 minutes of rest
(arm at heart level), and mean systolic and diastolic
blood pressures were recorded. Heart rate was counted
over 15 seconds. Blood samples obtained after an over-
night fast were used for measurement of fasting plasma
glucose, serum total cholesterol, high-density lipopro-
tein (HDL) cholesterol, and triglycerides, followed
by the calculation of low-density lipoprotein (LDL)
cholesterol.
Cardiovascular biomarkers
Immediately after sample collection, the serum was
frozen at 20 C and then thawed for analysis in July
2003. Serum CRP was analyzed using a highly sensitive
particle-enhanced immunoturbidimetry assay (Roche/
Hitachi, Roche Diagnostics, Basel, Switzerland). The
limit of detection was 0.03 mg/l, and the normal
range was 0.1–20 mg/l. The concentration of serum
NT-proBNP was determined using the Elecsys
2010 sandwich immunoassay (Roche Diagnostics,
Mannheim, Germany). The measured range was
5.1–34,927 pg/ml. Plasma suPAR was measured using
the suPARnostic enzyme-linked immunosorbent assay
(ViroGates, Copenhagen, Denmark), obtaining levels
between 0.6 and 22.0 ng/ml.
Endpoints
The principal endpoint was death from cardiovascular
causes, which required any cardiovascular diagnosis to
have been recorded as the primary cause of death
(Supplemental Table S1). Additional analyses were per-
formed for the secondary endpoints of fatal ischemic
stroke or coronary artery disease, death from any
cause, coronary artery disease, heart failure, and cere-
brovascular disease. Data on cardiovascular diagnoses
at hospital discharge were collected from the Danish
572
National Health Registry, a method that has
been validated previously.13–16 The diagnostic
codes employed to cover the secondary, non-fatal
endpoints are shown in Supplemental Table S2. Only
primary diagnoses were considered. Information on
mortality was obtained from the Danish Civil
Registration System. Follow-up continued until
December 31, 2012.
Statistical analyses
Distributions of continuous variables were visually
inspected using histograms. Normally distributed vari-
ables were presented as means and standard deviations,
non-normally distributed variables as medians and
interquartile ranges. Categorical variables were pre-
sented as frequencies and percentages. Groupwise
comparisons were performed using independent
samples t-test, Mann–Whitney U-test, or Pearson’s
2-test, as appropriate.
For visual purposes, receiver operating characteristic
(ROC) plots for hs-CRP, NT-proBNP, and suPAR, in
predicting death from cardiovascular causes, were ren-
dered from univariable binary logistic regression
models – that is, without the inclusion of traditional
cardiovascular risk factors. Given their positive
skewness, biomarker concentrations were logarithmic-
ally transformed for survival analyses. Cox propor-
tional-hazards regression models and Harrell’s
concordance index (C-index) were employed to
estimate the individual discrimination ability for each
biomarker.17 An uncensored approach for handling
ties was assumed. Models were adjusted for age,
sex, smoking status, systolic blood pressure, and
total cholesterol – that is, the individual variables that
comprise SCORE.2 We also performed sensitivity
analyses in which we further adjusted for fasting
plasma glucose and urine albumin-to-creatinine
ratio. Hazard ratios (HRs) and corresponding 95%
confidence intervals (CIs) were reported for one stand-
ard deviation increase in the logarithmically transformed
biomarker concentrations. Model calibration – whether
observed rates differed significantly from expected event
rates – was assessed using the Gronnesby and Borgan
likelihood-ratio test.18 The ability of each biomarker to
improve prognostication beyond individual SCORE
variables was also examined using continuous net reclas-
sification improvement (NRI).19
Two-sided p-values < 0.05 were considered statistic-
ally significant. We did not adjust for multiple compari-
sons. The statistical package Stata/IC 15 (StataCorp
LP, College Station, TX, USA) was used for all
computations.
European Journal of Preventive Cardiology 27(6)
Results
Baseline variables
The baseline characteristics, stratified for the occur-
rence of cardiovascular death, are summarized in
Table 1. Male sex, active smoking status, higher age,
total and LDL cholesterols, triglycerides, fasting
plasma glucose, and systolic and diastolic blood pres-
sures were associated with incident events.
Concentrations of hs-CRP, NT-proBNP, and suPAR
were also higher in the group of individuals that died
from a cardiovascular cause.
Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021
Biomarkers and event prediction
During a median of 18.5 years (interquartile range:
18.1–19.0 years), 478 (24.5%) individuals died, 177
(37.0%) from a cardiovascular cause, corresponding
to an event rate of the principal endpoint of 9.1%.
Figure 1 illustrates the ROC curves for the three bio-
markers, in predicting death from cardiovascular cause.
Hs-CRP (adjusted HR: 1.37, 95% CI: 1.17–1.60), NT-
proBNP (adjusted HR: 1.90, 95% CI: 1.58–2.29), and
suPAR (adjusted HR: 1.35, 95% CI: 1.17–1.57) were all
significantly associated with this endpoint (p < 0.001 for
all). Sex significantly interacted with the association
between hs-CRP and the principal endpoint
(p ¼ 0.005). However, no other interactions between
biomarkers and age or sex were identified (p > 0.05).
Whether the biomarkers retained their significance
depended on the specific endpoint. For example, all
three were associated with fatal ischemic stroke or cor-
onary artery disease and with death from any cause,
but only NT-proBNP was significantly associated
with heart failure, while hs-CRP appeared to have a
stronger association with coronary artery disease than
both NT-proBNP and suPAR (Table 2).
Discrimination, calibration, and reclassification
Models were generally well calibrated. Only NT-
proBNP significantly raised the C-index in predicting
death from cardiovascular causes when added to
SCORE variables (C-index 0.860 versus 0.847;
p ¼ 0.02). However, all biomarkers improved prognos-
tication when assessed using NRI. Additionally,
suPAR increased the C-index in predicting death
from any cause (C-index 0.856 versus 0.845;
p < 0.001), while both suPAR and hs-CRP aided prog-
nostication based on NRI. Finally, NT-proBNP carried
a significant NRI for predicting heart failure and hs-
CRP did so for fatal ischemic stroke or coronary artery
disease, but the C-index differences compared with
Frary et al. 573

Table 1. Baseline clinical and biochemical characteristics.

Subjects not experiencing Subjects experiencing

death from a cardiovascular death from a cardiovascular
cause (n ¼ 1774) cause (n ¼ 177) p-value

Age, years 51.9  10.1 63.9  8.5 <0.001

Age categories, years <0.001
41 628 (35.4%) 7 (4.0%)
51 545 (30.7%) 24 (13.6%)
61 408 (23.0%) 57 (32.2%)
71 193 (10.9%) 89 (50.3%)
Female sex 916 (51.6%) 58 (32.8%) <0.001

Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021

Active smokers 814 (45.9%) 104 (58.8%) <0.001
Body mass index, kg/m2 25.6  4.0 26.0  4.1 0.20
Systolic blood pressure, mmHg 126  17 139  19 <0.001
Diastolic blood pressure, mmHg 81  10 84  10 <0.001
Heart rate, bpm 65  10 69  11 <0.001
Total cholesterol, mmol/l 6.1  1.1 6.4  1.1 <0.001
HDL cholesterol, mmol/l 1.5  0.4 1.4  0.5 0.12
LDL cholesterol, mmol/l 4.0  1.0 4.3  1.0 <0.001
Triglycerides, mmol/l 1.3  0.7 1.5  0.7 <0.001
Fasting plasma glucose 4.7  0.5 4.9  0.5 <0.001
Urine albumin-to-creatinine ratio, mg/mmol 0.65  5.89 3.11  16.80 <0.001
Calculated SCORE 4.0  5.8 14.0  11.7 <0.001
hs-CRP, mg/l 1.50 (0.72–3.21) 2.97 (1.37–5.01) <0.001
NT-proBNP, pmol/l 43.14 (19.93–80.04) 87.31 (39.67–180.92) <0.001
suPAR, ng/ml 3.89 (3.27–4.73) 4.55 (3.78–5.49) <0.001
Categorical variables are presented as n (%), and continuous variables are given as mean  standard deviation (normally distributed variables) or median
(interquartile range) (non-normally distributed variables).
HDL: high-density lipoprotein; hs-CRP: high-sensitivity C-reactive protein; LDL: low-density lipoprotein; NT-proBNP: N-terminal pro-brain natriuretic
peptide; SCORE: Systematic COronary Risk Evaluation; suPAR: soluble urokinase plasminogen activator receptor.

SCORE variables only were not statistically significant
(heart failure: C-index 0.857 versus 0.829; p ¼ 0.15; fatal
ischemic stroke or coronary artery disease: 0.855 versus
0.848; p ¼ 0.55) (Table 3).
Sensitivity analyses
Further adjustments for fasting plasma glucose and the
urine albumin-to-creatinine ratio did not appreciably
alter the results (Supplemental Tables S3 and S4).
Discussion
In this population-based cohort study of apparently
healthy men and women, higher baseline values of hs-
CRP, NT-proBNP, and suPAR were each associated
with an increased long-term risk of cardiovascular
death. NT-proBNP significantly improved discrimin-
ation ability when added to traditional cardiovascular
risk factors, while all three biomarkers improved
reclassification.
As shown in prior analyses of MONICA 10,
hs-CRP, NT-proBNP, and suPAR display different
associations with incident cardiovascular disease and
may, individually or in combination, improve the
predictive capabilities of conventional risk prediction
algorithms.9,20–22 In addition to benefitting from a con-
siderably extended follow-up time, our study also relied
on continuous biomarker concentrations and individ-
ual SCORE variables to maximize the prognostic value
of each risk factor and avoid the adverse consequences
of grouping patients into threshold-based categories,
which may inflate the apparent benefit of added bio-
markers. Considering prior biomarker studies, in the
West of Scotland Coronary Preventive Study, both
hs-CRP and NT-proBNP provided improved discrim-
ination and reclassification for cardiovascular disease
and death among middle-aged men with hypercholes-
terolemia at moderate risk.23 NT-proBNP was more
strongly associated with fatal than non-fatal cardiovas-
cular disease, but less so with non-cardiovascular death.
Other studies have found the association between
574 European Journal of Preventive Cardiology 27(6)

1.00 High-sensitivity C-reactive protein

both hs-CRP and suPAR are markers of non-specific
inflammation, which is associated with atheroscler-
osis.21,22,26,27 CRP is an acute-phase protein released
0.75

by hepatocytes in response to interleukin-6 and has

multiple proinflammatory and anti-inflammatory
Sensitivity

roles, including aiding in foreign pathogen recognition

0.50

and activation of the complement system. suPAR is

mainly expressed on cells of the immune system and
0.25

plays a role in the plasminogen-activating pathway,

modulation of cell adhesion, migration, and prolifer-
0.00

0.00 0.25 0.50 0.75 1.00 ation. Our findings support this, as NT-proBNP aided
1- specificity in the correct reclassification of patients who experi-
Area under ROC curve = 0.6461 enced cardiovascular death and heart failure, respect-

Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021

N-terminal pro-brain natriuretic peptide
ively, but the numerically large discrimination
1.00

improvement for heart failure was not statistically sig-

nificant, presumably due to the limited number of
0.75

events. Conversely, hs-CRP and suPAR aided in the

Sensitivity

reclassification of death from any cause, and suPAR

0.50

also significantly improved discrimination ability for

this endpoint. This complies with the fact that
0.25

both inflammatory biomarkers can be increased in

various (e.g. infectious and malignant) settings.28,29
0.00

Interestingly, the 2016 European Society of

0.00 0.25 0.50 0.75 1.00
1- specificity
Cardiology/European Association for Cardiovascular
Area under ROC curve = 0.6992 Prevention and Rehabilitation (ESC/EACPR) guide-
Soluble urokinase plasminogen activator receptor lines on cardiovascular disease prevention inferred
1.00

that such mechanistic biomarker classifications were

not relevant.30
0.75

The exact role of circulating biomarkers among indi-

viduals without known cardiovascular disease remains
Sensitivity
0.50

unclear.31 There appears to be evidence of selective

reporting bias,32 which is particularly problematic for
observational studies with negative findings.33 External
0.25

validation of new risk prediction models is infrequently

performed,34 presumably for practical reasons as the
0.00

0.00 0.25 0.50
1-specificity
Area under ROC curve = 0.6536
Figure 1. ROC plots for biomarkers in predicting death from
cardiovascular causes.
NT-proBNP and cardiovascular events to be independ-
ent of echocardiographic measures of left ventricular
structure and function.24,25 Finally, in the Malmö
Diet and Cancer Study, middle-aged individuals in
the upper suPAR concentration quartile had a signifi-
cantly greater risk of coronary events or ischemic
stroke than those in the lower quartile.26
Hs-CRP, NT-proBNP, and suPAR represent at least
two distinct pathophysiological pathways. BNP is
released predominantly from ventricular myocytes in
response to increased wall stress – that is, volume
expansion with resultant pressure overload, and has
vasodilatory and diuretic properties.9 Conversely,
0.75 1.00 same biomarker combinations and assays may not be
available. Even when statistically significant discrimin-
ations are shown, they are often of modest magni-
tude.35–37 These shortcomings are reflected in the 2016
ESC/EACPR guidelines, which do not recommend
routine assessment of circulating biomarkers for refine-
ment of cardiovascular disease risk stratification
(class III recommendation, level of evidence B).30
In theory, such biomarkers, if able to reliably predict
incident cardiovascular events, could be implemented
into automated risk calculators that also include
traditional risk factors, supplemental information
from electronic health records, and genetic dispositions,
to deliver a more precise baseline risk estimate
and allow for appropriately individualized preventive
therapy.38 In particular, NT-proBNP seems to be
consistently associated with adverse outcomes and
provide improved discrimination, even when assessed
by the somewhat insensitive measure of the
C-index.9,23–25,37,39 However, besides showing
Frary et al. 575

Table 2. Unadjusted and adjusted hazard ratios for standardized, logarithmically transformed biomarker concentrations, for each
endpoint.

Unadjusted hazard ratio Adjusted hazard ratioa

(95% confidence interval) p-value (95% confidence interval) p-value

Death from cardiovascular causes

hs-CRP 1.61 (1.39–1.85) <0.001 1.37 (1.17–1.60) <0.001
NT-proBNP 2.49 (2.09–2.96) <0.001 1.90 (1.58–2.29) <0.001
suPAR 1.69 (1.48–1.92) <0.001 1.35 (1.17–1.57) <0.001
Death from any cause
hs-CRP 1.47 (1.35–1.61) <0.001 1.27 (1.16–1.40) <0.001
NT-proBNP 1.67 (1.51–1.85) <0.001 1.25 (1.12–1.40) <0.001

Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021

suPAR 1.67 (1.55–1.81) <0.001 1.39 (1.27–1.52) <0.001
Fatal ischemic stroke or coronary artery disease
hs-CRP 1.69 (1.36–2.10) <0.001 1.47 (1.15–1.88) 0.002
NT-proBNP 2.49 (1.90–3.26) <0.001 1.79 (1.35–2.39) <0.001
suPAR 1.74 (1.43–2.11) <0.001 1.42 (1.14–1.77) 0.002
Coronary artery disease
hs-CRP 1.38 (1.18–1.61) <0.001 1.26 (1.07–1.50) 0.007
NT-proBNP 1.17 (0.99–1.38) 0.06 1.22 (1.01–1.47) 0.04
suPAR 1.30 (1.11–1.51) <0.001 1.19 (1.003–1.42) 0.046
Heart failure
hs-CRP 1.51 (1.11–2.04) 0.008 1.30 (0.92–1.83) 0.13
NT-proBNP 2.63 (1.82–3.80) <0.001 2.33 (1.56–3.46) <0.001
suPAR 1.28 (0.95–1.74) 0.11 0.96 (0.66–1.40) 0.83
Cerebrovascular disease
hs-CRP 1.26 (1.09–1.45) 0.002 1.08 (0.93–1.26) 0.33
NT-proBNP 1.84 (1.56–2.17) <0.001 1.40 (1.17–1.67) <0.001
suPAR 1.49 (1.31–1.70) <0.001 1.24 (1.07–1.44) 0.004
Hazard ratios and corresponding 95% confidence intervals are reported for one standard deviation increase in the logarithmically transformed
biomarker concentrations.
a
Adjusted for sex, age, smoking status, systolic blood pressure, and total cholesterol.
hs-CRP: high-sensitivity C-reactive protein; NT-proBNP: N-terminal pro-brain natriuretic peptide; suPAR: soluble urokinase plasminogen activator
receptor.

associations with future risk, this also requires evidence
for biomarker-guided management algorithms in pri-
mary preventive settings, including cost-effectiveness.
In other words, biomarker assessment should lead
to specific interventions that ultimately improve
prognosis.
Strengths and limitations
Our study is notable for its assessment of three bio-
markers, the very long follow-up duration, and the
assessment of both discrimination and reclassification
measures. Interestingly, the three biomarkers differed
in their reclassification abilities for the different end-
points, but it should be mentioned that the direct clin-
ical interpretation and implications hereof remain
unclear.40,41 It is also unclear to what extent the
self-exclusion of subjects that chose not to participate
affected the study population, although comparable
participation rates have been noted in similarly con-
ducted population-based studies.25,42 Non-attendees
may have carried a greater overall risk burden as
they had similar body mass indexes, but were more
often smokers than attendees.43,44 Furthermore, our
study population may have been subject to the
healthy cohort effect. The selection mode based on
specific ages also prevented the use of age as a con-
tinuous variable. Our principal endpoint differed
slightly from that originally used to derive the
SCORE model since we also included death from
non-atherosclerotic cardiovascular causes, particu-
larly heart failure.2 However, risk factors for these
different conditions appear to be similar.45 The gen-
eralizability of the results may be limited, given the
576 European Journal of Preventive Cardiology 27(6)

Table 3. Concordance indices, calibration statistics, and net reclassification for standardized, logarithmically transformed biomarker
concentrations, for each endpoint.

Net reclassification
C-indexa p-value Calibrationb p-value improvement

Death from cardiovascular causes

hs-CRP 0.851 vs 0.847 0.22 0.01 >0.99 0.349 (0.154 to 0.527)
NT-proBNP 0.860 vs 0.847 0.02 1.63 0.65 0.314 (0.312 to 0.460)
suPAR 0.850 vs 0.847 0.23 0.46 0.93 0.222 (0.074 to 0.440)
Death from any cause
hs-CRP 0.849 vs 0.845 0.07 8.00 0.53 0.289 (0.166 to 0.406)
NT-proBNP 0.846 vs 0.845 0.48 5.49 0.79 0.060 (–0.061 to 0.186)

Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021

suPAR 0.856 vs 0.845 <0.001 8.68 0.47 0.377 (0.250 to 0.507)
Fatal ischemic stroke or coronary artery disease
hs-CRP 0.855 vs 0.848 0.55 0.84 0.36 0.419 (0.075 to 0.643)
NT-proBNP 0.859 vs 0.848 0.11 0.79 0.37 0.199 (–0.082 to 0.395)
suPAR 0.857 vs 0.848 0.14 0.13 0.72 0.266 (–0.006 to 0.572)
Coronary artery disease
hs-CRP 0.723 vs 0.719 0.35 0.66 0.72 0.239 (–0.095 to 0.385)
NT-proBNP 0.720 vs 0.719 0.91 4.26 0.12 –0.038 (–0.220 to 0.176)
suPAR 0.725 vs 0.719 0.11 1.95 0.38 0.172 (–0.231 to 0.332)
Heart failure
hs-CRP 0.839 vs 0.829 0.14 0.69 0.41 0.315 (–0.428 to 0.624)
NT-proBNP 0.857 vs 0.829 0.15 0.08 0.77 0.432 (0.113 to 0.795)
suPAR 0.829 vs 0.829 0.99 1.70 0.19 0.041 (–0.408 to 0.446)
Cerebrovascular disease
hs-CRP 0.740 vs 0.740 0.83 2.78 0.43 –0.181 (–0.199 to 0.174)
NT-proBNP 0.741 vs 0.740 0.82 3.45 0.33 0.138 (–0.052 to 0.303)
suPAR 0.744 vs 0.740 0.24 3.58 0.31 0.148 (–0.085 to 0.299)
a
For models with and without the biomarker.
b
Gronnesby and Borgan likelihood-ratio statistic.
C-index: concordance index; hs-CRP: high-sensitivity C-reactive protein; NT-proBNP: N-terminal pro-brain natriuretic peptide; suPAR: soluble uro-
kinase plasminogen activator receptor.

relatively homogenous study population. Biomarker
concentrations were measured in blood samples that
had been frozen for 10 years. Although hs-CRP,
NT-proBNP, and suPAR appear to remain stable in
frozen samples,21,46,47 a resulting underestimation of
their relative importance cannot be ruled out.48
Moreover, although multi-marker approaches have
gained attention in later years,8,10,25 our main results
did not warrant exploration thereof. Finally, we did
not have access to emerging biomarkers like growth
differentiation factor 15 and soluble ST2 or genomic
biomarkers, the evaluation of which would have been
desirable.49,50
Conclusion
Hs-CRP, suPAR, and particularly NT-proBNP pre-
dicted cardiovascular death and may enhance
prognostication beyond traditional risk factors in
apparently healthy individuals.
Author contribution
CEF and MKB conceptualized the hypothesis, designed the
work, interpreted the data, drafted the manuscript, and
revised the manuscript critically for important intellectual
content. TBO, JVS, SVG, JKV, and SLR interpreted the
data and revised the manuscript critically for important intel-
lectual content. MHO conceptualized the hypothesis,
designed the work, interpreted the data, and revised the
manuscript critically for important intellectual content.
MP conceptualized the hypothesis, designed the work,
analyzed and interpreted the data, and revised the manuscript
critically for important intellectual content. All authors have
read and approved the final version of the manuscript and
agree to be accountable for all aspects of the work in ensuring
that questions related to the accuracy or integrity of any part
of the article are appropriately investigated and resolved.
Frary et al.
Declaration of conflicting interests
The author(s) declared the following potential conflicts of
interest with respect to the research, authorship, and/or pub-
lication of this article: Dr Manan Pareek discloses the follow-
ing relationships – Advisory Board: AstraZeneca; Speaker’s
Fee: AstraZeneca, Bayer, and Boehringer Ingelheim. Dr
Michael Hecht Olsen discloses that he has received a part-
time clinical research grant from the Novo Nordisk
Foundation. The remaining authors have no disclosures to
report.
Funding
The author(s) received no financial support for the research,
authorship, and/or publication of this article.
References
1. Wilson PW, D’Agostino RB, Levy D, et al. Prediction of
coronary heart disease using risk factor categories.
Circulation 1998; 97: 1837–1847.
2. Conroy RM, Pyorala K, Fitzgerald AP, et al. SCORE
Project Group. Estimation of ten-year risk of fatal cardio-
vascular disease in Europe: the SCORE project. Eur Heart
J 2003; 24: 987–1003.
3. Mortensen MB, Sivesgaard K, Jensen HK, et al.
Traditional SCORE-based health check fails to identify
individuals who develop acute myocardial infarction.
Dan Med J 2013; 60: A4629.
4. Sehestedt T, Jeppesen J, Hansen TW, et al. Risk prediction
is improved by adding markers of subclinical organ
damage to SCORE. Eur Heart J 2010; 31: 883–891.
5. Wang A, Arver S, Boman K, et al. Testosterone, sex hor-
mone-binding globulin and risk of cardiovascular events: a
report from the Outcome Reduction with an Initial
Glargine Intervention trial. Eur J Prev Cardiol 2019; 26:
847–854.
6. Ridker PM, Rifai N, Rose L, et al. Comparison of
C-reactive protein and low-density lipoprotein cholesterol
levels in the prediction of first cardiovascular events.
N Engl J Med 2002; 347: 1557–1565.
7. Wang TJ, Larson MG, Levy D, et al. Plasma natriuretic
peptide levels and the risk of cardiovascular events and
death. N Engl J Med 2004; 350: 655–663.
8. Wang TJ, Gona P, Larson MG, et al. Multiple biomarkers
for the prediction of first major cardiovascular events and
death. N Engl J Med 2006; 355: 2631–2639.
9. Olsen MH, Hansen TW, Christensen MK, et al.
N-terminal pro-brain natriuretic peptide, but not high sen-
sitivity C-reactive protein, improves cardiovascular risk
prediction in the general population. Eur Heart J 2007;
28: 1374–1381.
10. Zethelius B, Berglund L, Sundstrom J, et al. Use of mul-
tiple biomarkers to improve the prediction of death from
cardiovascular causes. N Engl J Med 2008; 358:
2107–2116.
11. Diederichsen MZ, Diederichsen SZ, Mickley H, et al.
Prognostic value of suPAR and hs-CRP on cardiovascu-
lar disease. Atherosclerosis 2018; 271: 245–251.
577
12. Tunstall-Pedoe H, Kuulasmaa K, Mahonen M, et al.
Contribution of trends in survival and coronary-event
rates to changes in coronary heart disease mortality:
10-year results from 37 WHO MONICA project popula-
tions. Monitoring trends and determinants in cardiovas-
cular disease. Lancet 1999; 353: 1547–1557.
13. Johnsen SP, Overvad K, Sorensen HT, et al. Predictive
value of stroke and transient ischemic attack discharge
diagnoses in The Danish National Registry of Patients.
J Clin Epidemiol 2002; 55: 602–607.
14. Madsen M, Davidsen M, Rasmussen S, et al. The validity
of the diagnosis of acute myocardial infarction in routine
statistics: a comparison of mortality and hospital dis-
charge data with the Danish MONICA registry. J Clin
Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021
Epidemiol 2003; 56: 124–130.
15. Schmidt M, Schmidt SA, Sandegaard JL, et al. The
Danish National Patient Registry: a review of content,
data quality, and research potential. Clin Epidemiol
2015; 7: 449–490.
16. Sundboll J, Adelborg K, Munch T, et al. Positive predict-
ive value of cardiovascular diagnoses in the Danish
National Patient Registry: a validation study. BMJ
Open 2016; 6: e012832.
17. Harrell FE Jr, Califf RM, Pryor DB, et al. Evaluating the
yield of medical tests. JAMA 1982; 247: 2543–2546.
18. Gronnesby JK and Borgan O. A method for checking
regression models in survival analysis based on the risk
score. Lifetime Data Anal 1996; 2: 315–328.
19. Pencina MJ, D’Agostino RBS and Steyerberg EW.
Extensions of net reclassification improvement calcula-
tions to measure usefulness of new biomarkers. Stat
Med 2011; 30: 11–21.
20. Olsen MH, Hansen TW, Christensen MK, et al. New risk
markers may change the HeartScore risk classification
significantly in one-fifth of the population. J Hum
Hypertens 2009; 23: 105–112.
21. Sehestedt T, Lyngbaek S, Eugen-Olsen J, et al. Soluble
urokinase plasminogen activator receptor is associated
with subclinical organ damage and cardiovascular
events. Atherosclerosis 2011; 216: 237–243.
22. Lyngbaek S, Marott JL, Sehestedt T, et al.
Cardiovascular risk prediction in the general population
with use of suPAR, CRP, and Framingham Risk Score.
Int J Cardiol 2013; 167: 2904–2911.
23. Welsh P, Doolin O, Willeit P, et al. N-terminal pro-
B-type natriuretic peptide and the prediction of primary
cardiovascular events: results from 15-year follow-up of
WOSCOPS. Eur Heart J 2013; 34: 443–450.
24. Dietl A, Stark K, Zimmermann ME, et al. NT-proBNP
predicts cardiovascular death in the general population
independent of left ventricular mass and function:
insights from a large population-based study with long-
term follow-up. PLoS One 2016; 11: e0164060.
25. Pareek M, Bhatt DL, Vaduganathan M, et al. Single and
multiple cardiovascular biomarkers in subjects without a
previous cardiovascular event. Eur J Prev Cardiol 2017;
24: 1648–1659.
26. Persson M, Engstrom G, Bjorkbacka H, et al. Soluble
urokinase plasminogen activator receptor in plasma is
associated with incidence of CVD. Results from the
578
Malmo Diet and Cancer Study. Atherosclerosis 2012; 220:
502–505.
27. Munkhaugen J, Otterstad JE, Dammen T, et al. The
prevalence and predictors of elevated C-reactive protein
after a coronary heart disease event. Eur J Prev Cardiol
2018; 25: 923–931.
28. Pepys MB and Hirschfield GM. C-reactive protein: a crit-
ical update. J Clin Invest 2003; 111: 1805–1812.
29. Thuno M, Macho B and Eugen-Olsen J. suPAR: the
molecular crystal ball. Dis Markers 2009; 27: 157–172.
30. Piepoli MF, Hoes AW, Agewall S, et al. 2016 European
Guidelines on cardiovascular disease prevention in clin-
ical practice: The Sixth Joint Task Force of the European
Society of Cardiology and Other Societies on
Cardiovascular Disease Prevention in Clinical Practice
(constituted by representatives of 10 societies and by
invited experts): Developed with the special contribution
of the European Association for Cardiovascular
Prevention & Rehabilitation (EACPR). Eur J Prev
Cardiol 2016; 23: NP1–NP96.
31. Ioannidis JP and Tzoulaki I. Minimal and null predictive
effects for the most popular blood biomarkers of cardio-
vascular disease. Circ Res 2012; 110: 658–662.
32. Tzoulaki I, Siontis KC, Evangelou E, et al. Bias in asso-
ciations of emerging biomarkers with cardiovascular dis-
ease. JAMA Intern Med 2013; 173: 664–671.
33. Song F, Parekh-Bhurke S, Hooper L, et al. Extent of
publication bias in different categories of research
cohorts: a meta-analysis of empirical studies. BMC Med
Res Methodol 2009; 9: 79.
34. Siontis GC, Tzoulaki I, Castaldi PJ, et al. External val-
idation of new risk prediction models is infrequent and
reveals worse prognostic discrimination. J Clin Epidemiol
2015; 68: 25–34.
35. Melander O, Newton-Cheh C, Almgren P, et al. Novel
and conventional biomarkers for prediction of incident
cardiovascular events in the community. JAMA 2009;
302: 49–57.
36. Blankenberg S, Salomaa V, Makarova N, et al.;
BiomarCaRE Investigators. Troponin I and cardiovascu-
lar risk prediction in the general population: the
BiomarCaRE consortium. Eur Heart J 2016; 37:
2428–2437.
37. Natriuretic Peptides Studies Collaboration. Natriuretic
peptides and integrated risk assessment for cardiovascu-
lar disease: an individual-participant-data meta-analysis.
Lancet Diabetes Endocrinol 2016; 4: 840–849.
European Journal of Preventive Cardiology 27(6)
38. Pletcher MJ and Moran AE. Cardiovascular risk assess-
ment. Med Clin North Am 2017; 101: 673–688.
39. Wang TJ, Larson MG, Keyes MJ, et al. Association of
plasma natriuretic peptide levels with metabolic risk fac-
tors in ambulatory individuals. Circulation 2007; 115:
1345–1353.
40. Leening MJ, Vedder MM, Witteman JC, et al. Net reclas-
sification improvement: computation, interpretation, and
controversies: a literature review and clinician’s guide.
Ann Intern Med 2014; 160: 122–131.
41. Kerr KF, Wang Z, Janes H, et al. Net reclassification
indices for evaluating risk prediction instruments: a crit-
ical review. Epidemiology 2014; 25: 114–121.
42. Diederichsen AC, Sand NP, Norgaard B, et al.
Downloaded from https://academic.oup.com/eurjpc/article/27/6/570/5924758 by guest on 05 July 2021
Discrepancy between coronary artery calcium score and
HeartScore in middle-aged Danes: the DanRisk study.
Eur J Prev Cardiol 2012; 19: 558–564.
43. Gerdes LU, Bronnum-Hansen H, Madsen M, et al.
Trends in selected biological risk factors for cardiovascu-
lar diseases in the Danish MONICA population, 1982-
1992. J Clin Epidemiol 2000; 53: 427–434.
44. Gerdes LU, Bronnum-Hansen H, Osler M, et al. Trends
in lifestyle coronary risk factors in the Danish MONICA
population 1982–1992. Public Health 2002; 116: 81–88.
45. Khan SS, Ning H, Shah SJ, et al. 10-year risk equations
for incident heart failure in the general population. J Am
Coll Cardiol 2019; 73: 2388–2397.
46. Olsen MH, Hansen TW, Christensen MK, et al.
N-terminal pro brain natriuretic peptide is inversely
related to metabolic cardiovascular risk factors and the
metabolic syndrome. Hypertension 2005; 46: 660–666.
47. Olsen MH, Christensen MK, Hansen TW, et al. High-
sensitivity C-reactive protein is only weakly related to
cardiovascular damage after adjustment for traditional
cardiovascular risk factors. J Hypertens 2006; 24:
655–661.
48. Kugler KG, Hackl WO, Mueller LA, et al. The impact of
sample storage time on estimates of association in bio-
marker discovery studies. J Clin Bioinforma 2011; 1: 9.
49. Ibrahim NE and Januzzi JL Jr. Established and emerging
roles of biomarkers in heart failure. Circ Res 2018; 123:
614–629.
50. Li X, Lin Y and Zhang R. Associations between endo-
thelial nitric oxide synthase gene polymorphisms and the
risk of coronary artery disease: a systematic review and
meta-analysis of 132 case-control studies. Eur J Prev
Cardiol 2019; 26: 160–170.