Improved risk stratification of patients with acute coronary syndromes using a combination of hsTnT,NT-proBNP and hsCRP with the GRACE score

Abstract

Background:

Clinical scores and biomarkers improve risk stratification of patients with acute coronary syndromes. However, little is known about their value in patients referred for coronary angiography.

Methods:

Consecutive patients admitted at four Swiss university hospitals with a diagnosis of acute coronary syndrome were enrolled into the SPUM-ACS Biomarker Cohort between 2009 and 2012. Patients were followed at 30 days and 1 year with assessment of adjudicated events including all-cause mortality and the composite of all-cause mortality or non-fatal recurrent myocardial infarction.

Results:

Events and biomarkers were analysed in 1892 patients (52.4% with ST-segment elevation myocardial infarction, 43.3% with non-ST-segment elevation myocardial infarction and 4.3% with unstable angina). Death at 30 days occurred in 35 patients (1.9%) and at 1 year in 80 patients (4.3%). The choice of troponin assay (conventional versus high sensitivity) to calculate the Global Registry of Acute Coronary Events (GRACE) score did not affect risk prediction. The prognostic accuracy of the GRACE score was improved when combined with three individual biomarkers including high sensitivity troponin T (hsTnT), N-terminal-pro B-type natriuretic peptide (NT-proBNP) and high sensitivity C-reactive protein (hsCRP) to yield a 9% increment (C-statistic 0.73–>0.82) for the discrimination of short-term risk for all-cause mortality. In contrast, the novel biomarkers placental growth factor (PlGF), soluble fms-like tyrosine kinase-1 (sFlt-1) and the ratio sFlt-1/PlGF did not improve risk stratification.

Conclusions:

In patients with acute coronary syndrome referred for coronary angiography, combinations of biomarkers including hsTnT, NT-proBNP and hsCRP with the GRACE score enhanced risk discrimination.

Clinical Trials Registration:

NCT01000701

Keywords

Acute coronary syndromes biomarkers risk stratification

Introduction

Biomarkers constitute an integral part in risk stratification and management of patients with acute coronary syndromes (ACS).¹ Soluble fms-like tyrosine kinase-1 (sFlt-1) is a receptor for the angiogenic factors vascular endothelial growth factor-A, vascular growth factor-B and placental growth factor (PlGF).² sFlt-1 and PlGF were initially described as biomarkers in pre-eclampsia^2,3 and inflammatory mediators of atherosclerosis.⁴ More recently, sFlt-1 was identified as a novel early ischaemic biomarker.^5,6 Furthermore, elevated levels of sFlt-1 in patients with ACS were associated with an increased risk of acute heart failure⁷ and adverse outcome⁸ similar to PlGF.^8–10 Little is known whether the addition of the biomarkers sFlt-1, PlGF, high sensitivity troponin T (hsTnT), N-terminal-pro B-type natriuretic peptide (NT-proBNP) and high sensitivity C-reactive protein (hsCRP) to the Global Registry of Acute Coronary Events (GRACE) score (which includes conventional troponins (cTn)) improves risk stratification of contemporary patients with ACS referred for coronary angiography.

Thus, we first investigated whether risk prediction of the GRACE score in patients with ACS is improved using hsTnT instead of cTn as an elevated cardiac marker. Second, we studied if prognostic accuracy improves beyond that provided by the GRACE score alone when adding hsTnT, NT-proBNP, hsCRP, PlGF or sFlt-1 in a real-world multicentre prospective Swiss ACS cohort.

Materials and methods

Patient population

The Special Program University Medicine Acute Coronary Syndromes and Inflammation (SPUM-ACS) was established by four Swiss university hospitals (Bern, Geneva, Lausanne and Zurich) prospectively to recruit and analyse a real-world cohort of patients with ACS.^11–13 Consecutive patients with a diagnosis of ACS who underwent coronary angiography were enrolled between December 2009 and October 2012. Inclusion criteria comprised both genders aged 18 years or older presenting within 5 days (preferably within 72 hours) after pain onset with a main diagnosis of ST-segment elevation myocardial infarction (STEMI), non-ST-segment elevation myocardial infarction (NSTEMI) or unstable angina. Enrolled patients had symptoms compatible with angina pectoris (chest pain, dyspnoea) and fulfilled at least one of the following criteria: (a) ECG changes such as persistent ST-segment elevation or depression, T-inversion or dynamic ECG changes, new left bundle branch block; (b) evidence of positive (predominantly conventional) troponin by local laboratory reference values (with a rise and/or fall in serial troponin levels); (c) known coronary artery disease, specified as status after myocardial infarction (MI), coronary artery bypass graft (CABG), or percutaneous coronary intervention (PCI) or newly documented 50% or greater stenosis of an epicardial coronary artery during the initial catheterisation. Exclusion criteria comprised severe physical disability, inability to comprehend study or less than 1 year of life expectancy for non-cardiac reasons. The study was approved by the local ethical committees and all patients gave written informed consent in compliance with the Declaration of Helsinki, and it is listed at ClinicalTrials.gov (identifier NCT01000701).

Clinical endpoints

Follow-up was performed at 30 days (phone call) and 1 year (clinical visit) with events adjudicated by three independent experts using prespecified event adjudication forms. The primary endpoint was all-cause mortality (cardiac, vascular, non-cardiovascular) during 30 days and 1 year follow-up. Supplementary Table 1 shows individual causes of death. The secondary endpoint comprised the composite of all-cause mortality or non-fatal recurrent MI as defined according to the universal definition of MI comprising both spontaneous and peri-procedural MI.¹⁴ In patients with unstable angina, peri-procedural MI was defined as newly documented pathologic Q waves in two or more contiguous ECG leads and elevated Muscle-Brain type Creatinine Kinase (CK-MB) or troponin greater than 1*Upper Reference Limit (URL) or in the absence of CK-MB and troponin by creatine kinase (CK) greater than 1*URL or according to Clinical Endpoint Adjudication Committee (CEC) decision upon clinical scenario in the absence of all three cardiac biomarkers. Alternatively, peri-procedural MI was defined as CK of 2*URL or more confirmed by CK-MB or troponin greater than 1*URL or according to CEC decision upon clinical scenario in the absence of CK-MB and troponin. In the absence of CK, peri-procedural MI was defined by CK-MB greater than 3*URL or in the absence of CK and CK-MB by troponin greater than 3*URL. A tertiary endpoint was acute heart failure developing during 1 year follow-up defined as NT-proBNP concentrations of 900 ng/L or greater as a validated surrogate in the absence of adjudicated clinical acute heart failure events.¹⁵

Clinical risk score calculation

The GRACE score was used to calculate both in-hospital and long-term predictions of mortality, and to assess if the GRACE score based on hsTnT (incorporated as an elevated cardiac marker in the calculation) was a better predictor of outcomes than the GRACE score based on cTn. The GRACE risk criteria used to assess the score for in-hospital mortality comprised age, heart rate, systolic blood pressure, initial serum creatinine, Killip class, cardiac arrest on admission, elevated cardiac markers (cTn/hsTnT) and ST-segment deviation.¹⁶ The GRACE risk criteria used to calculate the score for long-term mortality comprised age, heart rate, systolic blood pressure, initial serum creatinine, history of congestive heart failure, history of MI, elevated cardiac markers (cTn/hsTnT), ST-segment depression and no in-hospital PCI.¹⁷ The GRACE scores (short and long-term) were calculated using a program written in Stata statistical software (version 13; Stata Corp, College Station, TX, USA) and we used the standard scoring of the GRACE score as mentioned in the reference publications.^16,17

Biomarker measurements

Blood was drawn from the arterial sheath at the time of diagnostic coronary angiography and centrifuged at 2700 g for 10 minutes at room temperature to obtain serum, frozen and stored in aliquots at −80°C until serial measurement (no prior freeze–thaw cycles) in the Zurich core laboratory. To assess new-onset acute heart failure events after the index ACS, NT-proBNP was measured in serum ascertained from venous puncture at the 1 year follow-up visit. Aliquots were measured blinded to the patients’ data by means of numbered ID codes. Tests were run for all five biomarkers (sFlt-1, PlGF, hsTnT, NT-proBNP, hsCRP) in one serum aliquot from each patient using electrochemiluminescence immunoassays all analysed on a cobase e 602 reader (all Roche Diagnostics, Mannheim, Germany) with assay characteristics as reported by the manufacturer.

Statistical analyses

Baseline continuous variables are presented as means with standard deviations, and categorical variables as counts with percentages. The time to first event or composite events were analysed throughout, censoring patients at 30 days or 365 days, at death, or at last valid contact date, whichever came first. We also present a subgroup analysis for the tertiary endpoint of acute heart failure developing during 1-year follow-up. We only analysed patients who had a NT-proBNP concentration of less than 900 ng/L at index blood draw (T₁ at time of coronary angiography). Patients with NT-proBNP concentrations of 900 ng/L or greater at the index ACS event were not analysed (a) to exclude patients with possible chronic heart failure; (b) to control for other potential confounders such as renal function, age and body mass index.

Cox’s regression models were used to evaluate possible associations between outcomes as all-cause mortality and all-cause mortality or recurrent MI (at 30 days and 1 year follow-up) and dichotomised biomarkers (sFlt-1, PlGF, NT-proBNP, hsCRP and hsTnT) and GRACE scores. To study the associations between dichotomised biomarkers and acute heart failure we used logistic regression modelling. We used clinically relevant cut points for the biomarkers and for GRACE scores the medians were used. The hazard ratios (HRs) are not adjusted because we used the potential confounders for the calculation of GRACE scores. To validate the results, we repeated the analyses using quintiles of biomarkers and GRACE scores. We compared the predictive accuracy of GRACE scores (using cTn as an elevated cardiac markers) with its recalculation with hsTnT, also GRACE plus hsTnT, GRACE plus NT-proBNP and GRACE plus hsCRP and combinations of biomarkers and GRACE scores using receiver operating characteristics curves. In addition, to investigate the predictive ability of the combination of biomarkers with an existing model, we used the HRs of the multivariate Cox model to find the best combination. The added predictive ability of a new predictor combined with an existing model was assessed by the integrated discrimination improvement (IDI) index and net reclassification index (NRI), which are based on logistic models.¹⁸

Results

Prospective biomarker cohort 1: outcome data

Among 2168 patients recruited, 1892 patients had available biomarker data (Figure 1). Of those, 52.4% had STEMI, 43.3% NSTEMI and 4.3% unstable angina. Revascularisation by PCI was performed in 91.8% of patients and by CABG in 3.9% of patients, respectively (Table 1).

Table 1.

Baseline characteristics.

Parameter	Mean ± SD
Age (years)	63.73 ± 12.32; n=1892
⩾75 years	415 (21.9%); n=1892
Gender (female)	404 (21.4%); n=1892
Weight (kg)	80.25 ± 15.03; n=1871
Body mass index (kg/m²)	27.13 ± 4.32; n=1869
Medical history
Diabetes mellitus	342 (18.1%); n=1892
Hypertension	1112 (58.8%); n=1892
Hypercholesterolaemia	1172 (61.9%); n=1892
Current smoker	738 (39.7%); n=1860
Family history of coronary artery disease	477 (25.5%); n=1871
Renal failure (MDRD formula)^a	242 (12.8%); n=1886
History of stroke or TIA	67 (3.5%); n=1892
Previous myocardial infarction	292 (15.4%); n=1890
Previous PCI	339 (17.9%); n=1891
Previous CABG	111 (5.9%); n=1892
Clinical presentation
Unstable angina	81 (4.3%); n=1892
NSTEMI	819 (43.3%); n=1892
STEMI	992 (52.4%); n=1892
Index procedure
Intervention or surgery attempted^b	1769 (93.5%); n=1892
PCI	1619 (91.8%); n=1763
Any drug-eluting stent	1311 (74.4%); n=1763
Any bare-metal stent	335 (19.0%); n=1763
Any balloon angioplasty	201 (11.4%); n=1763
CABG	68 (3.9%); n=1763
Antithrombotic regimen during PCI
Unfractionated heparin	1802 (95.4%); n=1888
LMWH	110 (5.8%); n=1891
Bivalirudin	79 (4.2%); n=1891
Glycoprotein IIb/IIIa antagonists	491 (26.0%); n=1891
Baseline medication
Proton pump inhibitor	285 (15.2%); n=1881

Based on creatinine-estimated glomerular filtration rate clearance of <60 ml/min/1.73 m², using the modification of diet in renal disease (MDRD) formula. ^bIncludes six patients where the PCI was aborted (e.g. bail-out, only thrombus aspiration, gives n=1769 − 6=1763). Medications shown are not necessarily mutually exclusive.

CABG: coronary artery bypass graft; LMWH: low molecular weight heparin; NSTEMI: non-ST-segment elevation myocardial infarction; STEMI: ST-segment elevation myocardial infarction; PCI: percutaneous coronary intervention; TIA: transient ischaemic attack.

Table 2 shows the rate of adjudicated major adverse cardiovascular and cerebrovascular events. At 30 days, follow-up data were available in 99.1%. All-cause mortality was 1.9% (n=35) among whom 1.7% (n=32) were of cardiac origin, 1.5% had a reinfarction (n=28), 0.8% stent thrombosis (n=15), 1.8% underwent ischaemia-driven revascularisation (n=34) and 0.7% experienced a stroke (n=14).

Table 2.

Clinical outcomes.

Parameter	Number (%)
At 30 days
Death	35 (1.9)
Cardiac death	32 (1.7)
Reinfarction (any)	28 (1.5)
Revascularisation (any)	37 (2.0)
Revascularisation (ischaemia-driven)	34 (1.8)
Cerebrovascular events (any)	16 (0.9)
Stroke (any)	14 (0.7)
Ischaemic stroke	13 (0.7)
Intracerebral haemorrhage	0 (0)
Unclear aetiology	1 (0.1)
TIA	2 (0.1)
Definite stent thrombosis	15 (0.8)
All-cause mortality and MI (at 30 days)	62 (3.3)
At 1 year
Death	80 (4.3)
Cardiac death	64 (3.5)
Reinfarction (any)	67 (3.7)
Revascularisation (any)	122 (6.8)
Revascularisation (ischaemia-driven)	115 (6.4)
Cerebrovascular events (any)	35 (1.9)
Stroke (any)	28 (1.5)
Ischaemic stroke	25 (1.4)
Intracerebral haemorrhage	2 (0.1)
Unclear aetiology	1 (0.1)
TIA	8 (0.4)
Definite stent thrombosis	19 (1.0)
All-cause mortality and MI (at 365 days)	140 (7.6)

Depicted are counts (% incidence at follow-up 30 days or 1 year from life tables with censoring at last contact date). First event per outcome for each patient only.

TIA: transient ischaemic attack; MI: myocardial infarction.

At 1 year, follow-up data were available in 95.8% of patients. All-cause mortality was 4.3% (n=80) among whom 3.5% (n=64) were of cardiac origin, 3.7% had a reinfarction (n=67), 1.0% stent thrombosis (n=19), 6.4% underwent ischaemia-driven revascularisation (n=115) and 1.5% experienced a stroke (n=28). Data on medication compliance are shown in Supplementary Table 2.

Risk stratification of patients: prognostic accuracy of the GRACE score by choice of troponin assay

The prognostic accuracy of the GRACE score (short and long-term) was robust using cut points (Supplementary Tables 3–5) to assess all-cause mortality or the composite of all-cause mortality or recurrent MI during follow-up. Of note, performance of the GRACE score (both short and long-term) to predict any of the outcomes was not influenced by the choice of troponin (cTn as per local laboratory or hsTnT as measured by the core laboratory) that went into the calculation of the scores (Supplementary Table 6). The long-term GRACE score yielded similar prognostic accuracy to predict new-onset acute heart failure during 1 year follow-up (Supplementary Table 7). Quintile analysis yielded similar results (Supplementary Table 8 and for acute heart failure Supplementary Tables 9 and 10).

Prognostic accuracy of individual biomarkers in risk stratification of patients

Among individual biomarkers evaluated by univariate analysis, hsTnT and NT-proBNP provided significant discrimination of risk for all-cause mortality during short and long-term follow-up (Table 3). In contrast, hsCRP predicted all-cause mortality only during long-term but not during short-term follow-up. Multivariate analysis identified hsTnT as a significant independent predictor of short-term all-cause mortality, whereas NT-proBNP provided significant discrimination of risk for long-term all-cause mortality (Table 4).

Table 3.

Prognostic accuracy for all-cause mortality during follow-up: univariate analysis.

Parameter	Hazard ratio (95% CI)	P value	C-statistic
30 Days all-cause mortality
sFlt-1 (>850 vs. ⩽850 ng/l)	1.47 (0.75, 2.88)	0.256	0.55
PlGF (>20 vs. ⩽20 ng/l)	1.32 (0.58, 3.03)	0.507	0.53
NT-proBNP (>174 vs. ⩽174 ng/l)	5.01 (1.53, 16.35)	0.008	0.62
hsCRP (>2 vs. ⩽2 mg/l)	1.72 (0.82, 3.58)	0.148	0.56
hsTnT (>0.07 vs. ⩽0.07 µg/l)	14.06 (1.92, 102.72)	0.009	0.63
Ratio (sFlt-1/PlGF)	1.34 (0.68, 2.61)	0.394	0.54
1 Year all-cause mortality
sFlt-1 (>850 vs. ⩽850 ng/l)	1.23 (0.79, 1.90)	0.360	0.53
PlGF (>20 vs. ⩽20 ng/l)	1.43 (0.81, 2.50)	0.213	0.53
NT-proBNP (>174 vs. ⩽174 ng/l)	5.89 (2.56, 13.53)	<0.001	0.62
hsCRP (>2 vs. ⩽2 mg/l)	2.11 (1.27, 3.49)	0.004	0.58
hsTnT (>0.07 vs. ⩽0.07 µg/l)	2.91 (1.50, 5.64)	0.002	0.59
Ratio (sFlt-1/PlGF)	1.17 (0.76, 1.82)	0.478	0.52

N=1892 patients with all five biomarker measured at baseline.

NT-proBNP: N-terminal-pro B-type natriuretic peptide; hsCRP: high sensitivity C-reactive protein; hsTnT: high sensitivity troponin T; sFlt-1: soluble fms-like tyrosine kinase-1; PlGF: placental growth factor.

Table 4.

Multivariate hazard ratios for all-cause mortality during follow-up.

Parameter	Hazard ratio (95% CI)	P value
30 Days all-cause mortality
In-hospital GRACE score using cTn (>140 vs. ⩽140)	14.48 (3.47, 60.44)	<0.001
NT-proBNP (>174 vs. ⩽174 ng/l)	2.46 (0.72, 8.36)	0.150
hsCRP (>2 vs. ⩽2 mg/l)	1.11 (0.52, 2.37)	0.783
hsTnT (>0.07 vs. ⩽0.07 µg/l)	8.96 (1.21, 66.49)	0.032
1 Year all-cause mortality
Long-term GRACE score using cTn (>122 vs. ⩽ 122)	4.75 (2.62, 8.62)	<0.001
NT-proBNP (>174 vs. ⩽174 ng/l)	3.66 (1.54, 8.67)	<0.001
hsCRP (>2 vs. ⩽2 mg/l)	1.45 (0.86, 2.44)	0.160
hsTnT (>0.07 vs. ⩽0.07 µg/l)	1.73 (0.87, 3.41)	0.116

N=1892 patients with all biomarkers measured at baseline.

cTN: conventional troponins; NT-proBNP: N-terminal-pro B-type natriuretic peptide; hsCRP: high sensitivity C-reactive protein; hsTnT: high sensitivity troponin T.

Kaplan–Meier curves demonstrated a marked difference in risk for all-cause mortality at 1 year when patients were categorised into high or low risk based on the GRACE score (Figure 2(a)) or GRACE score combined with hsTnT and NT-proBNP, respectively (Figure 2(b)).

NT-proBNP, hsCRP and hsTnT were all significant predictors of the composite endpoint (all-cause mortality or recurrent MI) during both short and long-term follow-ups, with the best prognostic accuracy for NT-proBNP (Supplementary Table 11). Multivariate analysis identified NT-proBNP as a single biomarker that could adequately discriminate patients at risk for the composite endpoint during both short and long-term follow-ups albeit at a lower HR than the GRACE score (Table 5). For the tertiary endpoint acute heart failure, the results of our univariate analysis showed that the long-term GRACE score and NT-proBNP predict outcome better than the other biomarkers (Supplementary Table 7).

Table 5.

Multivariate hazard ratios for all-cause mortality or MI during follow-up.

Parameter	Hazard ratio (95% CI)	P value
30 Days all-cause mortality or MI
In-hospital GRACE score using cTn (>140 vs. ⩽140)	3.28 (1.80, 5.97)	<0.001
NT-proBNP (>174 vs. ⩽174 ng/l)	2.11 (0.99, 4.47)	0.052
hsCRP (>2 vs. ⩽2 mg/l)	1.50 (0.84, 2.68)	0.166
hsTnT (>0.07 vs. ⩽0.07 µg/l)	0.93 (0.50, 1.74)	0.820
1 Year all-cause mortality or MI
Long term GRACE score using cTn (>122 vs. ⩽122)	2.37 (1.78, 3.81)	<0.001
NT-proBNP (>174 vs. ⩽174 ng/l)	2.16 (1.31, 3.58)	0.002
hsCRP (>2 vs. ⩽2 mg/l)	1.67 (1.13, 2.46)	0.011
hsTnT (>0.07 vs. ⩽0.07 µg/l)	1.05 (0.69, 1.62)	0.808

N=1892 patients with all biomarkers measured at baseline. Conventional troponin (cTn) was used to calculate the GRACE score.

MI: myocardial infarction; NT-proBNP: N-terminal-pro B-type natriuretic peptide; hsCRP: high sensitivity C-reactive protein; hsTnT: high sensitivity troponin T.

Quintile analysis was used to validate the findings using cut point-based discrimination of risk during short and long-term follow-up with data shown in Supplementary Tables 12–15. PlGF, sFlt-1 and the ratio sFlt-1/PlGF had poor prognostic accuracy for the primary, secondary and tertiary endpoints during short and long-term follow-up, respectively (Table 3, Supplementary Tables 9, 11, 12 and 14).

Discrimination of risk for combinations of biomarkers with GRACE score

Combining individual biomarkers identified as predictors of risk with the reference model (GRACE score calculation using cTn) provided incremental discrimination of risk by improvements in the C-statistics, IDI and NRI (Table 6). The best prognostic accuracy was achieved for the combination of hsTnT with NT-proBNP and hsCRP to the GRACE score to predict both all-cause mortality or the composite endpoint during short and long-term follow-up, respectively. Similarly, in the subgroup analysis for the tertiary endpoint acute heart failure, the long-term GRACE score combined with NT-proBNP provided the best prognostic accuracy (Table 7).

Table 6.

Accuracy of risk prediction for individual combinations.

	C-statistic		IDI		NRI
	C-statistic	P value	IDI value	P value	NRI value	P value
30 Days all-cause mortality
GRACE score using cTn	0.73	–	Reference	–	Reference	–
GRACE score + hsTnT	0.78	<0.001	0.010	<0.001	1.352	<0.001
GRACE score + NT-proBNP	0.77	0.010	0.006	0.002	1.182	<0.001
GRACE score + hsCRP	0.76	0.120	0.002	0.292	0.613	<0.001
GRACE score + hsTnT + NT-proBNP	0.81	<0.001	0.011	<0.001	1.183	<0.001
GRACE score + hsTnT + NT-proBNP + hsCRP	0.82	<0.001	0.011	<0.001	0.614	<0.001
1 Year all-cause mortality
GRACE score using cTn	0.67	–	Reference	–	Reference	–
GRACE score + hsTnT	0.71	<0.001	0.010	<0.001	1.153	<0.001
GRACE score + NT-proBNP	0.72	<0.001	0.009	<0.001	1.191	<0.001
GRACE score + hsCRP	0.71	0.006	0.006	0.003	0.676	<0.001
GRACE score + hsTnT + NT-proBNP	0.74	<0.001	0.013	<0.001	1.192	<0.001
GRACE score + hsTnT + NT-proBNP + hsCRP	0.75	<0.001	0.015	<0.001	0.677	<0.001
30 Days all-cause mortality or MI
GRACE score using cTn	0.64	–	Reference	–	Reference	–
GRACE score + hsTnT	0.65	0.478	0.002	0.019	0.963	<0.001
GRACE score + NT-proBNP	0.68	0.011	0.004	<0.001	1.061	<0.001
GRACE score + hsCRP	0.68	0.042	0.003	0.025	0.632	<0.001
GRACE score + hsCRP + NT-proBNP	0.69	0.007	0.006	<0.001	1.062	<0.001
GRACE score + hsTnT + NT-proBNP + hsCRP	0.70	0.019	0.006	<0.001	0.631	<0.001
1 Year all-cause mortality or MI
GRACE score using cTn	0.62	–	Reference	–	Reference	–
GRACE score + hsTnT	0.63	0.050	0.005	0.001	0.99	0.020
GRACE score + NT-proBNP	0.65	<0.001	0.008	<0.001	1.049	<0.001
GRACE score + hsCRP	0.66	<0.001	0.008	<0.001	0.651	<0.001
GRACE score + hsCRP+ NT-proBNP	0.68	<0.001	0.012	<0.001	1.050	<0.001
GRACE score + hsTnT + NT-proBNP + hsCRP	0.68	<0.001	0.013	<0.001	0.650	<0.001

N=1892 patients with all biomarkers measured at baseline. The reference model is the GRACE score using cTn for the model tested.

IDI: integrated discrimination improvement index; NRI: net reclassification index; MI: myocardial infarction; cTn: conventional troponins; NT-proBNP: N-terminal-pro B-type natriuretic peptide; hsCRP: high sensitivity C-reactive protein; hsTnT: high sensitivity troponin T.

Table 7.

Accuracy of risk prediction for individual combination.

Parameters	C-statistic		IDI		NRI
Parameters	C-statistic	P value	IDI value	P value	NRI value	P value
Acute heart failure (1 year follow-up)
GRACE score using cTn	0.651	–	Reference	–	Reference	–
GRACE score using cTn + NT-proBNP	0.695	0.077	0.014	<0.001	0.562	<0.001

N=874 patients with available determination of acute heart failure. The reference model is the GRACE score using cTn for the model tested.

IDI: integrated discrimination improvement index; NRI: net reclassification index; cTn: conventional troponins; NT-proBNP: N-terminal-pro B-type natriuretic peptide.

Similar results were obtained when biomarkers were analysed as continuous variables rather than based on cut points (Supplementary Table 16). Moreover, separate analysis of patients with STEMI and NSTEMI showed that adding biomarkers improved risk stratification beyond the GRACE score in both subgroups. The best discrimination of risk was achieved for the combination of all three biomarkers with the GRACE score (Supplementary Tables 17 and 18).

Discussion

This prospective study conducted in a real-world multicentre cohort of patients admitted with a diagnosis of ACS confirmed by angiography confers the following major findings: (a) the addition of hsTnT, NT-proBNP and hsCRP improves risk discrimination for all-cause mortality and the composite endpoint of all-cause mortality or recurrent MI beyond the GRACE score during short and long-term follow-up; (b) the choice of assay for troponin (conventional versus high sensitivity) has no impact on prognostic accuracy of the GRACE score; (c) the novel biomarkers PlGF, sFlt-1 and the ratio sFlt-1/PlGF do not provide incremental prognostic value; (d) the combination of the GRACE score and NT-proBNP was a better predictor of new-onset acute heart failure during 1 year follow-up than the GRACE score alone.

Good prognostic performance of the GRACE score in the SPUM-ACS Biomarker Cohort

In our study the commonly used GRACE score proved to be a powerful tool for risk stratification of patients with ACS, albeit with a slightly lower prognostic performance (C-statistic 0.62–0.73) compared with the original GRACE registries (C-statistic ⩾0.75).^16,17 This discrepancy is likely to be attributable to the use of dichotomised rather than continuous/quintile-based parameters. In the current SPUM-ACS Biomarker Cohort all patients received coronary angiography in line with a high rate of coronary angiography performed in recent studies evaluating the GRACE score in patients with ACS;^19–21 however, our study was performed in a larger cohort at a very high rate of PCI (91.8%).

In light of the easy use of the GRACE score covering the entire spectrum of patients with ACS,^16,17 the time and expense of performing additional biomarker testing needs to be weighed with respect to their incremental value for patient management. Of note, the choice of troponin T assay (high sensitivity vs. conventional) that went into the calculation of the GRACE score had no impact on the prognostic accuracy of the score in our study. This may indicate that the gain in sensitivity for the detection of troponin T with the high sensitivity assay does not translate into improved risk stratification in low-risk patients when combined with clinical parameters comprised in the GRACE score.

Improved risk stratification in all patients with ACS when combining biomarkers with GRACE score

Among individual biomarkers tested in our study, hsTnT provides independent information beyond the GRACE score in risk discrimination of patients for short and long-term all-cause mortality. Studies addressing the incremental value of high sensitivity troponin assays compared with clinical risk scores are scarce and revealed conflicting results for different scores in selected ACS cohorts.^21–24 Our findings were obtained from the entire spectrum of patients with ACS (i.e. unstable angina/NSTEMI and STEMI) and extend prior data of patients with NSTEMI demonstrating improved risk stratification using highly sensitive troponins upon addition to the Thrombolysis in Myocardial Infarction (TIMI) score²⁴ or GRACE score,²¹ respectively. Two recent studies found an improved risk discrimination and reclassification of patients with NSTEMI on addition of NT-proBNP to the TIMI risk score incorporating ultra cTn²⁵ or the GRACE score and hsTnT,²¹ respectively. In our study, NT-proBNP provides incremental information beyond the GRACE score in risk stratification for all-cause mortality and the composite of all-cause mortality and recurrent MI, during short and long-term follow-up, respectively. Furthermore, our subgroup analysis also identifies NT-proBNP combined with the GRACE score as a suitable predictor for new-onset acute heart failure events during 1 year follow-up. Moreover, this is the first study of a large prospective cohort of patients across the entire spectrum of ACS that demonstrates a maximum benefit in risk stratification of patients beyond the GRACE score on addition of hsTnT, NT-proBNP and hsCRP combined. We found an improvement in discrimination of risk as per increment in C-statistic of up to 9% (11% in the STEMI subgroup) compared with the GRACE score alone (Table 6 and Supplementary Table 17). Furthermore, we show that adding these three common biomarkers to the GRACE score improved the prognostic accuracy of the GRACE score for all-cause mortality at 1 year in both the NSTEMI and STEMI subgroup of patients.

Poor prognostic value of angiogenic biomarkers in patients with ACS who received heparin

In the SPUM-ACS Biomarker Cohort, the prognostic performance of the angiogenic biomarkers sFlt-1, PlGF and the ratio of sFlt-1/PlGF were poor. This is in contrast to previous data which proposed sFlt-1 and PlGF as independent predictors for long-term mortality and recurrent MI.^8–10 This discrepancy may be attributable to different cut points in distinct patient groups, different assay technology and in particular to the use of heparin prior to blood draw in the SPUM-ACS cohort, as the latter increases sFlt-1 plasma levels.^26,27 In the study by Hochholzer et al.,⁸ patients with chest pain presenting to the emergency department were analysed in whom blood was drawn before the administration of heparin, using the same assay technology. sFlt-1 is characterised by rapid release kinetics in patients with ACS,^5,6 which may enable a more rapid ‘rule-in’ of MI. The optimal setting for the use of the angiogenic biomarkers remains to be identified. Patients presenting to the emergency department with chest pain prior to receiving heparin may constitute appropriate candidates in light of the heparin-induced release of the angiogenic markers.^26,27

Limitations

Recruitment was based on written informed consent impeding complete assessment of all patients with ACS despite major efforts to include as many patients as possible including those with cardiogenic shock. Although we recruited a total number of 1892 patients with complete biomarker measurements, only 80 all-cause mortalities were recorded and only 140 all-cause mortality or MI composite events were recorded, limiting the power of our study. In data analysis we primarily used dichotomised biomarkers and the GRACE score, which is useful in terms of clinical practice but we lose information compared to the continuous form of parameters. To overcome this limitation, we repeated the analysis using quintiles of biomarkers and the GRACE score (data presented in the supplement). As we did not have in-hospital outcome we used 30 days outcome to assess the prognostic accuracy of the in-hospital GRACE score. Biomarker measurements in this study were performed at the time of coronary angiography after the decision for invasive management had already been made and after heparin administration which is a known confounder of sFlt-1 concentration.^26,27 As biomarker concentrations were measured at different time points after the onset of symptoms in individual patients, temporal changes in concentrations were not accounted for. However, the aim of this study was to assess the clinical usefulness of biomarkers in a straightforward manner in an all-comer ACS patient population.

Footnotes

Acknowledgements

The authors appreciate the work of the clinical event adjudication committee for SPUM-ACS: Matthias Pfisterer, University of Basel (chair); Tiziano Moccetti, CardioCentro Lugano; Lukas Kappenberger, University of Lausanne, all Switzerland. They also thank the local study nurses (Anika Adam, Maja Müller, Christa Schönenberger, Therese Fahrni, Saskia Bühlmann, Geneviève Legault, Véronique Berset, Nicole Bonvin, Anne Bevand, Armelle Delort), the core lab technicians (Isabelle Peereboom, Monika Seiler, Anuschka Beccato), the central data monitors (Katja Heinimann, Daria Bochenek, Timon Spörri), the electronic data capturing system (2mt GmbH Ulm, Jürgen Nagler-Ihlein, Torsten Illmann), the research coordinator Lambertus J van Tits, Sven Trelle, Marcel Zwahlen, ISPM Bern and CTU Bern and the members of the local catheter teams for their invaluable work.

Collaborators:

Reto Auer, Daniela Buhl, David Carballo, Martin Fiedler, Thorsten Hornemann, Milosz Jaguszewski, Philipp Jakob, Ulf Landmesser, Willibald Maier, Lanja Saleh, Barbara Staehli, Giulio Stefanini, Christian Templin and Nicolas Vuilleumier

Conflict of interest

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RK received speaker fees from Eli Lilly, Servier and Bayer Healthcare. FM has received research grants to the institution from Amgen, AstraZeneca, Boston Scientific, Biotronik, Medtronic, MSD, Eli Lilly and St Jude Medical including speaker or consultant fees. SW has received research grants to the institution from Abbott, Boston Scientific, Biosensors, Biotronik, The Medicines Company, Medtronic and St Jude Medical and honoraria from Abbott, Astra Zeneca, Eli Lilly, Boston Scientific, Biosensors, Biotronik, Medtronic and Edwards. PJ has received research grants to the institution from Astra Zeneca, Biotronik, Biosensors International, Eli Lilly and The Medicines Company, and serves as an unpaid member of the steering group of trials funded by Astra Zeneca, Biotronik, Biosensors, St Jude Medical and The Medicines Company. AvE received speaker or consultant fees from Amgen, Astra-Zeneca, MSD and Sanofi-Aventis. TFL received research grants to the institution from AstraZeneca, Bayer Healthcare, Biosensors, Biotronik, Boston Scientific, Eli Lilly, Medtronic, MSD, Merck, Roche and Servier, including speaker fees by some of them. CMM received research grants to the institution from Eli Lilly, AstraZeneca, Roche and MSD including speaker or consultant fees. LR received speaker fees and research grants to the institution from St Jude Medical. All other authors have no conflict of interest to declare.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors received support from the Swiss National Science Foundation (SPUM 33CM30-124112); the Swiss Heart Foundation; the Foundation Leducq and the Foundation for Cardiovascular Research – Zurich Heart House, Zurich. The SPUM consortium was also supported by: Roche Diagnostics, Rotkreuz, Switzerland (providing the kits for the biomarkers); Eli Lilly, Indianapolis (USA); AstraZeneca, Zug; Medtronic, Münchenbuchsee; Merck Sharpe and Dome (MSD), Lucerne; Sanofi-Aventis, Vernier; St Jude Medical, Zurich (all Switzerland). SA is supported by the European Community’s Seventh Framework Program FP7/2011: Marie Curie Initial Training Network MEDIASRES (‘Novel Statistical Methodology for Diagnostic/Prognostic and Therapeutic Studies and Systematic Reviews’; ) with the grant agreement number 290025.

References

Scirica

. Acute coronary syndrome: emerging tools for diagnosis and risk assessment. J Am Coll Cardiol 2010; 55: 1403–1415.

Maynard

Min

Merchan

. Excess placental soluble fms-like tyrosine kinase 1 (sFlt1) may contribute to endothelial dysfunction, hypertension, and proteinuria in preeclampsia. J Clin Invest 2003; 111: 649–658.

Levine

Maynard

Qian

. Circulating angiogenic factors and the risk of preeclampsia. N Engl J Med 2004; 350: 672–683.

Luttun

Tjwa

Moons

. Revascularization of ischemic tissues by PlGF treatment, and inhibition of tumor angiogenesis, arthritis and atherosclerosis by anti-Flt1. Nat Med 2002; 8: 831–840.

Kapur

Heffernan

Yunis

. Elevated soluble fms-like tyrosine kinase-1 levels in acute coronary occlusion. Arterioscler Thromb Vasc Biol 2011; 31: 443–450.

Liebetrau

Nef

Dorr

. Release kinetics of early ischaemic biomarkers in a clinical model of acute myocardial infarction. Heart 2014; 100: 652–657.

Onoue

Uemura

Takeda

. Usefulness of soluble Fms-like tyrosine kinase-1 as a biomarker of acute severe heart failure in patients with acute myocardial infarction. Am J Cardiol 2009; 104: 1478–1483.

Hochholzer

Reichlin

Stelzig

. Impact of soluble fms-like tyrosine kinase-1 and placental growth factor serum levels for risk stratification and early diagnosis in patients with suspected acute myocardial infarction. Eur Heart J 2011; 32: 326–335.

Heeschen

Dimmeler

Fichtlscherer

. Prognostic value of placental growth factor in patients with acute chest pain. JAMA: the journal of the American Medical Association 2004; 291: 435–441.

10.

Oemrawsingh

Lenderink

Akkerhuis

. Multimarker risk model containing troponin-T, interleukin 10, myeloperoxidase and placental growth factor predicts long-term cardiovascular risk after non-ST-segment elevation acute coronary syndrome. Heart 2011; 97: 1061–1066.

11.

Klingenberg

Heg

Raber

. Safety profile of prasugrel and clopidogrel in patients with acute coronary syndromes in Switzerland. Heart 2015; 101: 854–863.

12.

Reiser

Klingenberg

Hof

. Circulating FABP4 is a prognostic biomarker in patients with acute coronary syndrome but not in asymptomatic individuals. Arterioscler Thromb Vasc Biol 2015; 35: 1872–1879.

13.

Gencer

Montecucco

Nanchen

. Prognostic value of PCSK9 levels in patients with acute coronary syndromes. Eur Heart J 2016; 37: 546–553.

14.

Thygesen

Alpert

Jaffe

. Third universal definition of myocardial infarction. Eur Heart J 2012; 33: 2551–2567.

15.

Januzzi

Jr. Chen-Tournoux

Moe

. Amino-terminal pro-B-type natriuretic peptide testing for the diagnosis or exclusion of heart failure in patients with acute symptoms. Am J Cardiol 2008; 101: 29–38.

16.

Granger

Goldberg

Dabbous

. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med 2003; 163: 2345–2353.

17.

Eagle

Lim

Dabbous

. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA: the journal of the American Medical Association 2004; 291: 2727–2733.

18.

Pencina

D’Agostino

Sr. D’Agostino

Jr. . Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 2008; 27: 157–172; discussion 207–212.

19.

Beygui

Silvain

Pena

. Usefulness of biomarker strategy to improve GRACE score’s prediction performance in patients with non-ST-segment elevation acute coronary syndrome and low event rates. Am J Cardiol 2010; 106: 650–658.

20.

Widera

Pencina

Meisner

. Adjustment of the GRACE score by growth differentiation factor 15 enables a more accurate appreciation of risk in non-ST-elevation acute coronary syndrome. Eur Heart J 2012; 33: 1095–1104.

21.

Widera

Pencina

Bobadilla

. Incremental prognostic value of biomarkers beyond the GRACE (Global Registry of Acute Coronary Events) score and high-sensitivity cardiac troponin T in non-ST-elevation acute coronary syndrome. Clin Chem 2013; 59: 1497–1505.

22.

Hochholzer

Reichlin

Twerenbold

. Incremental value of high-sensitivity cardiac troponin T for risk prediction in patients with suspected acute myocardial infarction. Clin Chem 2011; 57: 1318–1326.

23.

Meune

Drexler

Haaf

. The GRACE score’s performance in predicting in-hospital and 1-year outcome in the era of high-sensitivity cardiac troponin assays and B-type natriuretic peptide. Heart 2011; 97: 1479–1483.

24.

Bonaca

Scirica

Sabatine

. Prospective evaluation of the prognostic implications of improved assay performance with a sensitive assay for cardiac troponin I. J Am Coll Cardiol 2010; 55: 2118–2124.

25.

Scirica

Sabatine

Jarolim

. Assessment of multiple cardiac biomarkers in non-ST-segment elevation acute coronary syndromes: observations from the MERLIN-TIMI 36 trial. Eur Heart J 2011; 32: 697–705.

26.

Rosenberg

Buhimschi

Lockwood

. Heparin elevates circulating soluble fms-like tyrosine kinase-1 immunoreactivity in pregnant women receiving anticoagulation therapy. Circulation 2011; 124: 2543–2553.

27.

Searle

Mockel

Gwosc

. Heparin strongly induces soluble fms-like tyrosine kinase 1 release in vivo and in vitro – brief report. Arterioscler, Thromb, Vasc Biol 2011; 31: 2972–2974.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.38 MB