Author + information
- Received October 20, 2008
- Revision received January 30, 2009
- Accepted March 10, 2009
- Published online July 21, 2009.
- Kai M. Eggers, MD, PhD⁎,⁎ (, )
- Bo Lagerqvist, MD, PhD⁎,
- Per Venge, MD, PhD†,
- Lars Wallentin, MD, PhD⁎ and
- Bertil Lindahl, MD, PhD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Kai M. Eggers, Department of Medical Sciences, Cardiology, University Hospital Uppsala, S-751 85 Uppsala, Sweden
Objectives The aim of this study was to assess risk prediction by different biomarkers in patients with an ongoing non–ST-segment elevation acute coronary syndrome (NSTE-ACS) and after clinical stabilization.
Background Different biomarkers reflect different aspects of the pathobiology in NSTE-ACS. However, there is little information regarding their relative prognostic value during the time course of disease.
Methods The N-terminal pro-brain natriuretic peptide (NT-proBNP), C-reactive protein (CRP), cardiac troponin I (cTnI), and the estimated glomerular filtration rate (eGFR) were measured at randomization and after 6 weeks and 6 months in 877 NSTE-ACS patients included in the FRISC (FRagmin and fast revascularization during InStability in Coronary artery disease) II trial. The biomarkers' prognostic value during 5-year follow-up was evaluated by Cox regression models, calculation of the c-statistics, and estimation of the net reclassification improvement (NRI).
Results Among the biomarkers measured at randomization, NT-proBNP was the strongest predictor for mortality (adjusted hazard ratio [HR]: 1.7; 95% confidence interval [CI]: 1.3 to 2.1; p < 0.001). Even during follow-up, NT-proBNP demonstrated the strongest association to the composite end point of death/myocardial infarction (adjusted HR at 6 weeks: 1.5; 95% CI: 1.3 to 1.7; p < 0.001; adjusted HR at 6 months: 1.4; 95% CI: 1.2 to 1.7; p = 0.001). Even CRP was independently predictive at 6 months for the composite end point (adjusted HR: 1.3; 95% CI: 1.1 to 1.5; p = 0.003). Only 6-week results of NT-proBNP provided significant incremental prognostic value to established risk indicators regarding the composite end point (c-statistics 0.69 [p = 0.03]; NRI 0.11 [p = 0.03]).
Conclusions The NT-proBNP is an independent risk predictor in patients with ongoing NSTE-ACS and after clinical stabilization. The CRP exhibits increasing predictive value at later measurements. However, only NT-proBNP provided incremental prognostic value and might therefore be considered as a complement for early follow-up controls after NSTE-ACS.
Biochemical markers play a major role for risk assessment in patients with an ongoing non–ST-segment elevation acute coronary syndrome (NSTE-ACS). In particular the cardiac troponins are generally recognized as important risk indicators but also markers of left ventricular performance (i.e., N-terminal pro-brain natriuretic peptide [NT-proBNP]), inflammation (i.e., C-reactive protein [CRP]), and renal function (i.e., estimated glomerular filtration rate [eGFR]) that provide strong prognostic information (1,2). Emerging evidence furthermore suggests that these biomarkers are prognostically useful even in the late phase after a NSTE-ACS (3–8), with the natriuretic peptides having been most extensively evaluated (3,4). However, the evolution of the prognostic implications of different biomarkers during and after an NSTE-ACS has not been assessed so far.
The purpose of this study was thus to investigate the prognostic value of cardiac troponin I (cTnI), NT-proBNP, CRP, and the eGFR relative to each other and in the context of other established risk indicators in patients with an ongoing NSTE-ACS and at 6 weeks and 6 months after clinical stabilization.
Patients and study design
The protocol and main results of the FRISC (FRagmin and fast revascularization during InStability in Coronary artery disease) II trial have been published elsewhere (9,10). This trial was a prospective, multicenter study randomizing 3,489 patients with NSTE-ACS between 1996 and 1998 in a factorial design to an early invasive or noninvasive strategy and to 3-month treatment with dalteparin or placebo. Patients were included in case of symptoms of unstable coronary artery disease with objective signs of myocardial ischemia, such as electrocardiographic changes (ST-segment depression ≥0.1 mV or T-wave inversion ≥0.1 mV) or elevated biochemical markers of myocardial necrosis. Major exclusion criteria were increased risk of bleeding, serum creatinine >150 μmol/l, percutaneous coronary intervention during the last 6 months, and a decision to perform coronary angiography or percutaneous coronary intervention before randomization. Patients with a history of previous open heart surgery, advanced age, or poor general health and those included after completion of recruitment to the invasive versus noninvasive arm were treated primarily noninvasively and randomized only regarding prolonged treatment with dalteparin or placebo. In the invasive strategy, the aim was to perform coronary angiography and, if appropriate, revascularization within 7 days from admission. Patients randomized to the noninvasive strategy underwent coronary angiography only in case of refractory or recurrent angina or if they showed signs of severe ischemia on a pre-discharge exercise test. Informed consent was obtained from all patients, and the protocol was approved by all local ethics committees.
Patients were followed after randomization by outpatient visits after 6 (range 4 to 7) weeks and 3 and 6 months. At these instances, a blood sampling program was performed at selected study centers including 1,380 patients. Further follow-up was performed by telephone contacts after 12 and 24 months. Thereafter and up to 5 years after randomization, all information on events was based on mandatory National Registries run by the Swedish Health Authority.
The cTnI, NT-proBNP, and CRP were determined in frozen (−70°C), not previously thawed, samples of ethylenediaminetetraacetic acid-plasma obtained at randomization and follow-up at 6 weeks and 6 months. For follow-up measurements, only results from patients without an acute myocardial infarction (AMI) or a coronary revascularization procedure during the previous 14 days were considered. The cTnI was measured with the Access AccuTnI assay (Beckman Coulter, Inc., Fullerton, California). For randomization samples, the original version of this assay was used, and results were dichotomized at its 99th percentile of 0.04 μg/l (11). Samples obtained during follow-up were analyzed with the refined assay version, which is characterized by an improved analytical performance at low cTnI concentrations with 0.014 μg/l as the lowest concentration measurable with a coefficient of variation <10% and 0.006 μg/l as the lower level of detection (5). Following previous experiences, follow-up cTnI results were dichotomized at 0.01 μg/l (5).
The NT-proBNP was measured at all instances with the Elecsys proBNP sandwich immunoassay on an Elecsys 2010 instrument (Roche Diagnostics, Mannheim, Germany). The CRP was analyzed with the Immulite CRP assay (Diagnostic Products Corp., Los Angeles, California). Serum creatinine was determined locally, and the eGFR was calculated according to the 4-variable version of the Modification of Diet in Renal Disease formula (12).
The study end points for the present analysis were total mortality and AMI, alone or as composite, after each respective measurement instance. For patients with an AMI during the first 6 months of follow-up, the time to the next AMI or end of follow-up was calculated.
The prognostic value of the tested biomarkers was investigated by different Cox proportional hazard models. Adjustment was made for age, sex, diabetes, a history of heart failure and a history of AMI (model 1). Covariates were defined as follows:
Diabetes: history of diabetes (randomization results) and/or antidiabetic treatment and/or a pre-test fasting glucose >6.1 mmol/l (follow-up results).
Heart failure: history of heart failure (randomization results) and/or left-ventricular ejection fraction <0.45 during the index hospital stay (follow-up results).
Previous AMI: history of AMI (randomization results) and/or AMI as index event (defined as cardiac troponin T >0.035 μg/l at randomization ) and/or recurrent AMI during follow-up and before the respective measurement instance (follow-up results).
Additional adjustment was made for the 4 tested biomarkers (model 2) and for coronary revascularization before the respective measurement instance during follow-up (model 3). We decided to focus on coronary revascularization instead for the randomized treatment strategy with regard to some crossover between the randomization arms during follow-up. The proportional hazard assumptions were checked by computing log[−log(event)] plots. The functional form of the association between outcome and each respective biomarker was assessed by the inspection of generalized additive model plots, which revealed nonlinear associations between the eGFR (all measurements) and CRP (at randomization). Therefore we dichotomized the eGFR at 75 ml/min/1.73 m2(13) and excluded randomization CRP results from all multivariable analyses, because the nonlinear association could not be compensated by mathematical transformations. The cTnI was dichotomized on the basis of the respective cutoffs, whereas NT-proBNP and CRP (at 6 weeks and 6 months) were entered as ln transformed variables.
The incremental prognostic value of the tested biomarkers during follow-up to the covariates applied in model 1 and regarding the composite end point was tested first, by calculating the respective c-statistics. The differences in c-statistics were estimated with the method described by Antolini et al. (14). The Hosmer-Lemeshow statistic was used to assess the goodness of fit of the applied models. Second, the increased discriminative value of the biomarkers regarding the composite end point was studied by assessing the net reclassification improvement (NRI) as described by Pencina et al. (15). This method determines the difference in the probabilities of a subject to belong to predefined risk categories before and after the addition of a specific marker. Two models were used for prognostic classification based on estimated risk tertiles and estimated risks of <10%, 10% to 19.9%, and ≥20%, respectively.
Continuous variables are described as medians with 25th and 75th percentiles and were compared with the Mann-Whitney Utest. Categoric variables are expressed as frequencies and percentages. Differences between categoric variables were analyzed with the chi-square test. In all tests, a p value <0.05 was considered significant. The statistical software packages SPSS version 14.0 (SPSS Inc., Chicago, Illinois) and SAS version 9.1 (SAS Institute Inc., Cary, North Carolina) were used.
Randomization results for all 4 biomarkers were available in 877 subjects. The sample sizes at 6 weeks and 6 months are given in Figure 1.
The clinical characteristics of the sample population and the results of the tested biomarkers at the different measurement instances are demonstrated in Table 1.In total, 442 patients (50%) had been randomized to 3-month treatment with dalteparin. Three hundred sixty-five patients (42%) had been randomized to the invasive strategy, 365 to the noninvasive strategy (42%), and 147 patients (16%) had not been randomized regarding these strategies. Coronary revascularization had been performed at 6 weeks in 455 patients (52%) of the entire sample population and in 514 patients (60%) at 6 months.
During 5-year follow-up, 79 patients (9.0%) died, 152 patients (17%) had at least 1 recurrent AMI, and 198 patients (23%) suffered the composite end point. The occurrence of events is illustrated in Figure 1.
At randomization, NT-proBNP was the strongest predictor for mortality and cTnI >0.04 μg/l for both AMI and the composite end point (Table 2).The eGFR at randomization did not provide independent prognostic information. When added as a ln transformed continuous variable, CRP neither altered the hazard ratios for the other tested markers considerably nor emerged as an independent predictor of events (data not shown).
Even during follow-up, NT-proBNP was the strongest predictor for mortality and AMI. The CRP was also predictive for death and AMI, in particular at 6 months. The cTnI and the eGFR provided only limited prognostic value and before final adjustment. These associations remained unchanged when follow-up cTnI results were entered into the multivariable analysis as ln transformed variables (data not shown).
When the tested biomarkers were added to prognostic models based on conventional risk indicators, only NT-proBNP at 6 weeks provided incremental value associated with a significant increase of the c-statistics from 0.66 to 0.69 (p = 0.03). Even at 6 months, addition of NT-proBNP resulted in the greatest increase of the c-statistics but without reaching statistical significance (Table 3).The NT-proBNP at 6 weeks was also the only marker to provide a significant NRI (Figs. 2Aand 2B). This was due to a substantial proportion of patients being downgraded regarding their risk category after addition of NT-proBNP to the risk models (19% of all patients in both models).
The prognostic importance of biomarkers reflecting different pathophysiologic entities related to risk is well established in the acute phase of NSTE-ACS. However, the evidence regarding the evolution of the prognostic information provided by biomarkers at later stages of the disease is still somewhat limited. Our results extend previous data from the FRISC II study (3,16), demonstrating that NT-proBNP is a strong predictor of adverse events in NSTE-ACS, both throughout the entire 6-month sampling period and even in the context of other clinical risk indicators, including biomarkers of cardiomyocyte necrosis, inflammation, and renal dysfunction. The NT-proBNP levels were particularly predictive for mortality, probably because they reflect larger infarct size, progressive remodeling, and thus a more pronounced degree of myocardial dysfunction (17,18). However, even myocardial ischemia augments the synthesis of the natriuretic peptides (19), which might explain the association between NT-proBNP upon early follow-up and recurrent ischemic events.
Among the biomarkers measured at randomization, cTnI was the strongest predictor for recurrent AMI but exhibited no independent association with mortality. This finding probably depends on the choice of the 99th percentile as a prognostic cutoff. This cutoff has been shown to reflect an increased likelihood of an unstable plaque with downstream microembolization of thrombotic material, which translates into a higher risk of recurrent ischemic events (20,21). The risk for mortality, in contrast, is known to increase at higher troponin peak levels, given the relation of troponin leakage to the amount of infarcted myocardium (18,22).
Contrasting with the acute situation, cTnI elevation at a threshold below the 99th percentile was not related to recurrent AMI but predicted mortality. This indicates that this lower cutoff might be more sensitive to detect left ventricular dysfunction after NSTE-ACS (23). However, cTnI was no longer prognostic after adjustment for other biomarkers, including NT-proBNP, probably because the natriuretic peptides more appropriately reflect impaired cardiac performance (18). This does not exclude the possibility that cTnI might exhibit stronger associations with risk at even later time points, given its independent relation to mortality in a recently assessed elderly community population (24).
Even CRP emerged as an independent predictor when measured during follow-up and became, despite decreasing levels, more prognostic over time. Late after NSTE-ACS, CRP levels might reflect an increased myocardial cytokine activation leading to remodeling, paving the way for heart failure and thereby to poor outcome (25). However, given the prognostic value of CRP independent of NT-proBNP, it is more probable that the prognostic implications of CRP late after NSTE-ACS are mediated by extramyocardial mechanisms such as a chronic low-grade inflammation in the coronary vasculature, as described in other stable populations (26).
We did not find, somewhat unexpectedly, associations between renal function and prognosis, which contrasts with previous studies in patients with ongoing or recent NSTE-ACS (8,13). We assume that these findings are biased, due to the exclusion of patients with renal failure in the FRISC II study.
Our findings raise the question of whether and which biochemical markers should be assessed in the stable phase after NSTE-ACS to improve prognostication. Current follow-up routines focus on symptomatic assessment and modification of cardiovascular risk factors to halt the progression of atherosclerosis. However, these measures are in part operator dependent, and conventional risk factors might not completely account for the individual risk profile, which also is known to change over time (27).
We found that measurement of NT-proBNP at 6 weeks might be particularly useful with regard to its incremental value to conventional risk indicators. Noteworthy, the assessment of NRI indicated that NT-proBNP levels allowed for the reclassification of a relative large subset of patients without events to lower categories of risk. Early measurement of NT-proBNP might also be useful for the identification of some high-risk subjects with persistent left ventricular dysfunction after NSTE-ACS in whom optimization pharmacological treatment is mandatory. C-reactive protein, in contrast, provided no incremental prognostic information in the tested models. However, with regard to its independent association with events at 6 months and the considerably improved calibration of our risk model at this time point after addition of CRP, testing for this biomarker could be useful for later follow-up controls when results would be integrated into more sophisticated risk prediction tools than the exploratory model applied in our analysis.
This study is based on an NSTE-ACS population randomized more than 10 years ago (i.e., before the current era of intense secondary prevention with adenosine diphosphate inhibition, angiotensin-converting enzyme inhibitors, and high-dose statins) and therefore needs to be validated in a contemporary sample. Furthermore, we assessed a pre-selected patient population, and the sample size was somewhat limited. Therefore we cannot exclude the possibility that CRP and the eGFR might exhibit stronger associations with risk when assessed in a larger and more unselected population. Due to a lack of remaining samples, we could not perform re-analyses of cTnI at randomization with the refined version of the AccuTnI assay. Thus, cTnI results were obtained with 2 iterations of this assay with different prognostic cutoffs, which limits the transferability of results obtained at randomization to follow-up results. The focus on coronary revascularization versus no revascularization might have weakened the relationships between the biomarkers and risk, because the decision to revascularize patients from other than the invasive arm was based on clinical decisions with a potential higher rate of sicker patients with higher biomarker levels being treated with this prognostically beneficial therapy.
Cardiac troponin I, NT-proBNP, and CRP exhibit different and changing relations to adverse events in patients with NSTE-ACS, both in the acute phase and after clinical stabilization. The NT-proBNP was independently predictive for adverse outcome throughout the entire 6-month observation period and should be considered for improvement of risk stratification during early follow-up. Even measurement of CRP might be a useful tool for risk stratification at a later time point (e.g., after 6 months). The eventual incremental value of slight elevations of troponin determined with highly sensitive assays will need further evaluation in larger patient cohorts with stable coronary artery disease.
The authors wish to thank Hans Garmo, Karin Jensevik, and Johan Lindbäck for their statistical support.
Continuing Medical Education (CME) is available for this article.
This study was supported by grants from the Swedish Heart and Lung Foundation (Stockholm, Sweden) and the Erik, Karin and Gösta Selander Foundation (Uppsala, Sweden). The reagents for the troponin I assay were provided by Beckman Coulter, Inc. (Fullerton, California). Dr. Venge has received research honoraria from Abbott, Beckman Coulter, Roche, and Siemens Diagnostics. Dr. Lindahl has served as a consultant for Beckman Coulter and Siemens and has received honoraria for educational lectures for Beckman Coulter, Siemens, and Roche.
- Abbreviations and Acronyms
- acute myocardial infarction
- C-reactive protein
- cardiac troponin I
- estimated glomerular filtration rate
- net reclassification improvement
- non–ST-segment elevation acute coronary syndrome
- N-terminal pro-brain natriuretic peptide
- Received October 20, 2008.
- Revision received January 30, 2009.
- Accepted March 10, 2009.
- American College of Cardiology Foundation
- Heeschen C.,
- Hamm C.W.,
- Bruemmer J.,
- Simoons M.L.
- James S.K.,
- Lindahl B.,
- Siegbahn A.,
- et al.
- Lindahl B.,
- Lindbäck J.,
- Jernberg T.,
- et al.
- Eggers K.M.,
- Lagerqvist B.,
- Venge P.,
- Wallentin L.,
- Lindahl B.
- Jose P.,
- Skali H.,
- Anavekar N.,
- et al.
- Levey A.S.,
- Greene T.,
- Kusek J.W.,
- Beck G.J.,
- MDRD Study Group
- Jernberg T.,
- Lindahl B.,
- Siegbahn A.,
- et al.
- Sabatine M.S.,
- Morrow D.A.,
- de Lemos J.A.,
- et al.
- Heeschen C.,
- van den Brand M.J.,
- Hamm C.W.,
- Simoons M.L.
- Lindahl B.,
- Diderholm E.,
- Lagerqvist B.,
- Venge P.,
- Wallentin L.,
- FRISC II Investigators
- Ono K.,
- Matsumori A.,
- Shioi T.,
- Furukawa Y.,
- Sasayama S.
- Westerhout C.M.,
- Fu Y.,
- Lauer M.S.,
- et al.