Article Text
Abstract
Background The objective of this study was to determine if the extent of quantitative troponin elevation predicted mortality as well as in-hospital complications of cardiac arrest, new heart failure and cardiogenic shock.
Design 16 318 patients with non-ST-segment elevation acute coronary syndromes (NSTE ACS) from the Global Registry of Acute Coronary Events (GRACE) were included. The maximum 24 h troponin value as a multiple of the local laboratory upper limit of normal was used. The population was divided into five groups based on the degree of troponin elevation, and outcomes were compared. An adjusted analysis was performed using quantitative troponin as a continuous variable with adjustment for known prognostic variables.
Results For each approximate 10-fold increase in the troponin ratio, there was an associated increase in cardiac arrest, sustained ventricular tachycardia (VT) or ventricular fibrillation (VF) (1.0, 2.4, 3.4, 5.9 and 13.4%; p<0.001 for linear trend), cardiogenic shock (0.5, 1.4, 2.0, 4.4 and 12.7%; p<0.001), new heart failure (2.5, 5.1, 7.4, 11.6 and 15.8%; p<0.001) and mortality (0.8, 2.2, 3.0, 5.3 and 14.0%; p<0.001). These findings were replicated using the troponin ratio as a continuous variable and adjusting for covariates (cardiac arrest, sustained VT or VF, OR 1.56, 95% CI 1.39 to 1.74; cardiogenic shock, OR 1.87, 95% CI 1.61 to 2.18; and new heart failure, OR 1.57, 95% CI 1.45 to 1.71). The degree of troponin elevation was predictive of early mortality (HR 1.61, 95% CI 1.44 to 1.81; p<0.001 for days 0–14) and longer term mortality (HR 1.18, 95% CI 1.07 to 1.30, p=0.001 for days 15–180).
Conclusion The extent of troponin elevation is an independent predictor of morbidity and mortality.
- Myocardial infarction
- mortality
- shock
- heart failure
- prognosis
- angina - unstable
- NSTEMI
- STEMI
Statistics from Altmetric.com
Studies have demonstrated that elevation of troponin (ie, above threshold value) in patients with non-ST-segment elevation acute coronary syndromes (NSTE ACS) is an important predictor of mortality and provides important prognostic information beyond that provided by creatine kinase-MB (CK-MB).1–3 Indeed, troponin has been incorporated into a revised definition of myocardial infarction by the European Society of Cardiology and American College of Cardiology4 5 based in part on its prognostic value.
However, studies examining the relationship between the degree of troponin elevation, particularly among those with abnormal values, and mortality in NSTE ACS have shown differing results. Three analyses2 6 7 from randomised trials have demonstrated that the degree of troponin elevation was predictive of mortality, but a fourth analysis8 showed no significant differences in mortality in troponin tertiles among those with a positive result. Finally, a recent analysis from a large registry showed no independent association of degree of troponin elevation in those with an abnormal troponin and mortality after adjustment for the GRACE risk score; however. this analysis was based on initial troponin values.9
Furthermore, in patients with NSTE ACS, the degree of CK-MB elevation10 has been shown to predict in-hospital complications such as cardiogenic shock, cardiac arrest and congestive heart failure. However, for troponin, there is a paucity of data examining the relationship of the degree of elevation and the risk of these serious clinical events.
The objective of this analysis was to determine if the extent of troponin concentration elevation in patients presenting with NSTE ACS is independently associated with (1) in-hospital complications such as cardiogenic shock, new onset heart failure and cardiac arrest and (2) mortality after accounting for other established risk variables.11
Methods
Full details of the GRACE registry methodology have been published.12–14 In brief, GRACE is an international registry that has centres in 14 countries (Argentina, Australia, Austria, Belgium, Brazil, Canada, France, Germany, Italy, New Zealand, Poland, Spain, the UK and the USA). For this analysis we included 16 318 patients enrolled in 95 hospitals from 26 July 2000 to 31 December 2007 who had a presumed admission diagnosis of NSTE ACS, based on a clinical history of ACS accompanied by at least one of the following: ECG changes consistent with ACS (ST-segment elevation or left bundle branch block were excluded), serial increases in biochemical markers of cardiac necrosis (troponin, CK-MB) or documented history of coronary artery disease. Follow-up was done at ∼6 months after hospital discharge by interviewing patients, significant others or healthcare providers, or utilising medical records. Where required, study investigators received approval from their local hospital ethics or institutional review board.
Initial troponin T or I values at hospitalisation and peak troponin values in the first 24 h after hospitalisation were recorded. The frequency and timing of measurement of troponin were determined as per local practice by the treating physician. The maximum of these values was used for analysis and, for patients dying in the first 24 h with only one troponin measurement, the initial value was utilised. Troponin level was measured in each hospital's local laboratory and was expressed as the ratio of the troponin value to the local laboratory's upper limit of normal (ULN). The in-hospital outcomes of death, new-onset congestive heart failure, cardiogenic shock and cardiac arrest/sustained ventricular tachycardia (VT)/ventricular fibrillation (VF) were examined; in addition, we evaluated all-cause mortality at 14 days and 6 months. For in-hospital new-onset heart failure, the analysis excluded patients who had signs or symptoms of heart failure at presentation. Similarly, for in-hospital cardiac arrest/VT/VF the analysis excluded patients who presented to hospital with a cardiac arrest. Finally, for in-hospital cardiogenic shock, patients who presented to hospital with cardiogenic shock were excluded.
Outcome definitions
Definitions for outcomes have been previously published for the GRACE registry.1 For patients presenting with unstable angina, recurrent myocardial infarction (MI) was defined as recurrent ischaemic symptoms with elevation in biomarkers (troponin, CK-MB) >24 h after initial presentation. Recurrent MI in those patients who had an acute MI as the qualifying event was defined as recurrent ischaemic symptoms with re-elevation of CK-MB to above the ULN and increased by at least 50% over the previous value and, if not available, then re-elevation of total CK by either (1) >2× ULN and increased 25% over the previous value or (2) >1.5× ULN and increased by at least 50% over the previous value. Within 24 h after percutaneous coronary intervention (PCI), recurrent MI was defined as CK-MB >3× ULN and increased by at least 50% over the previous value if elevated. Within 24 h after coronary bypass surgery, recurrent MI was defined as CK-MB >5 ULN; if CK-MB was previously elevated, ECG changes consistent with MI were also required.
Cardiogenic shock was defined as patients with pulmonary oedema and hypoperfusion characterised by systolic blood pressure <80 mm Hg. Heart failure was defined as having symptoms and signs of Killip class II or greater heart failure. Cardiac arrest/VF was defined as either VF or rapid VT with haemodynamic instability, asystole or electrical mechanical dissociation requiring cardiopulmonary resuscitation. Sustained VT was defined as ECG documentation and requiring intervention.
Events during initial hospitalisation
All tests of significance were at α=0.05. For crude rates of in-hospital death, cardiac arrest/sustained VT/VF, cardiogenic shock and new congestive heart failure, the population was stratified into five groups of ∼10-fold increases based on the troponin ratio to the local laboratory ULN: (1) ratios >0.01 to ≤1 (ie, normal); (2) ratios >1 through to ratios of 10; (3) ratios >10 through to 100; (4) ratios >100 through to 1000; and (5) ratios >1000. We excluded those patients (n=10) with significantly outlying ratios. The unadjusted relationships of troponin elevation and rates of hospital events were evaluated using two-sided Cochran–Armitage tests for linear trend across the groups at α=0.05. A sensitivity analysis was performed by excluding the group of troponin ratios >1000 because this represented a small group (n=107) of extreme values.
To assess the relationship of troponin ratio and in-hospital cardiogenic shock, new congestive heart failure and cardiac arrest/sustained VT/VF, stepwise logistic regression models were used with log base 10 of the troponin ratio (as a continuous variable), adjusted for hospital cluster, and applicable GRACE risk score variables other than cardiac marker status (age, cardiac arrest prior to admission, ST depression, Killip class, systolic BP, pulse and creatinine level).11 For the outcome of new-onset heart failure, Killip class at presentation was not used in the model (patients presenting with heart failure were excluded for this analysis), for the outcome of cardiac arrest/sustained VT/VF, cardiac arrest at presentation was not used in the model (patients presenting with cardiac arrest were excluded for this analysis), and for the outcome of cardiogenic shock, cardiogenic shock at presentation was not used in the model (patients presenting with cardiogenic shock were excluded for this analysis).
Mortality
The unadjusted association of troponin and all-cause death was examined using Kaplan–Meier plots. To assess the adjusted relationship between log base 10 troponin ratio and death up to 180 days after admission, we used a Cox regression model, adjusted for hospital cluster and the GRACE risk score variables other than cardiac marker status (age, cardiac arrest prior to admission, ST depression, Killip class, systolic blood pressure, pulse and creatinine level).11 Initial Cox proportional hazard models showed that the effect of the log of the troponin ratio on death was not uniform over time. Therefore, the final Cox model separated the effect of log troponin ratio into early (0–14 days) and late mortality (15–180 days), each time period having a distinct HR (non-overlapping 95% CIs). The HR reported for late mortality (15–180 days) included only survivors to 15 days.
Results
The majority of patients included in this study had troponin I recorded (n=12 454; 76%) and a smaller proportion had troponin T (n=3832; 24%). Baseline characteristics in the five groups of troponin elevation are presented in table 1. Patients in the groups with the highest elevations of troponin had higher rates of ST depression and/or Q waves at presentation. Patients in the group with highest elevations of troponin were less likely to have had a prior myocardial infarction, prior PCI or coronary bypass surgery.
In-hospital events
As seen in table 2, unadjusted rates of in-hospital death, cardiogenic shock, new-onset congestive heart failure and cardiac arrest/sustained VT/VF increased progressively across the five troponin groups (p<0.0001 for trend). During the sensitivity analysis, the results were unchanged when the highest troponin group (n=107) was excluded from the trend test.
For unadjusted in-hospital myocardial (re)infarction there was also a statistically significant increasing risk with increasing troponin ratios (p=0.035). However, in the sensitivity analysis where the highest troponin group was excluded, the trend for increasing risk of re-infarction with increasing troponin ratios was no longer significant (p=0.13, as shown in table 2).
After adjustment for GRACE risk score variables, among patients without heart failure at presentation (2197 patients were excluded who presented with heart failure), the troponin ratio (as a continuous variable) was independently associated with the development of new congestive heart failure in-hospital (OR 1.57, 95% CI 1.45 to 1.71, as shown in figure 1). Similarly after adjustment, among patients not presenting in cardiogenic shock (55 patients were excluded who presented in cardiogenic shock), the troponin ratio was independently associated with the development of cardiogenic shock (OR 1.87, 95% CI 1.61 to 2.18, as shown in figure 1). Finally, after adjustment for GRACE risk score variables, among patients not presenting with cardiac arrest (164 patients were excluded who presented with cardiac arrest), the troponin ratio was independently associated with the development of cardiac arrest/sustained VT/VF (OR 1.56, 95% CI 1.39 to 1.74, as shown in figure 1). After adjustment, the troponin ratio was not independently associated with recurrent MI after 24 h (OR 1.08, 95% CI 0.97 to 1.21). The results were consistent with both troponin I and T for the in-hospital outcomes.
As patients can experience multiple events, it is important to note that 75% of patients with an outcome event during initial hospitalisation (1060/1414) had only one event. The proportion of patients with only one outcome event during initial hospitalisation was 77% for new congestive heart failure, 71% for recurrent MI, 37% for cardiac arrest/sustained VT/VF and 21% for cardiogenic shock. Of patients with cardiogenic shock, 46% had a cardiac arrest/sustained VT/VF during initial hospitalisation. The in-hospital mortality for patients with (1) new-onset congestive heart failure was 11.3%, (2) cardiogenic shock was 61.3%, (3) cardiac arrest/sustained VT/VF was 50.1% and (4) recurrent myocardial infarction was 14.7%.
Mortality
Using Kaplan–Meier estimation, unadjusted mortality for all patients presenting with NSTE ACS was 2.6% at 14 days and 7.3% at 180 days from presentation to hospital. Figure 2 demonstrates Kaplan–Meier curves for the different troponin groups for mortality and shows increasing mortality rates with increasing levels of troponin elevation. The highest troponin group had a marked increase in mortality rate compared with the other troponin groups.
In the Cox proportional hazards model, the troponin ratio was independently associated with increased mortality for days 0–14 from hospitalisation (HR 1.61, 95% CI 1.44 to 1.81; p<0.001) and for days 15–180 (HR 1.18, 95% CI 1.07 to 1.30; p=0.001). Other variables that significantly predicted mortality were age, Killip class, serum creatinine, cardiac arrest on presentation, ST depression, heart rate and systolic blood pressure at presentation (figure 3). The results for mortality were consistent for both troponin I and T.
Discussion
This study demonstrates that among patients presenting with NSTE ACS, the extent of elevation of troponin is an independent predictor of mortality. While previous studies have found a relationship with the degree of elevation of troponin and mortality, this study is the first to our knowledge to establish this relationship with cardiac arrest, cardiogenic shock and new-onset heart failure.
In-hospital complications
In two previous studies, the rate of ‘life-threatening’ ventricular arrhythmias after NSTE ACS ranged between 2.1% and 2.5%, with 80% of these events occurring within the first 12 h.15 16 In one of these studies (n=588),15 validated risk scores (GRACE, PURSUIT and TIMI) were not shown to be predictive of malignant ventricular arrhythmias; however, this study had limited power due to a small number of events. In another analysis from four randomised trials of glycoprotein IIb/IIIa inhibitors in acute coronary syndromes (n=26 416),16 independent predictors of VF were chronic obstructive lung disease, ST deviation, prior MI and prior hypertension. Neither of these studies15 16 specifically examined the relationship of quantitative troponin to ‘life-threatening’ ventricular arrhythmias.
Cardiogenic shock complicating NSTE ACS has been estimated to occur in 2.5% of patients, with an estimated mortality of 66% in a clinical trial setting.17 18 Age, multivessel disease and diabetes have been shown to be predictors of cardiogenic shock but, to the best of our knowledge, this study is the first to demonstrate that the degree of troponin elevation is associated with an increased risk of cardiogenic shock.
In patients presenting with NSTE ACS without heart failure in a large multicentre registry, the development of heart failure was associated with an adjusted OR of 4.93 for mortality, suggesting that this an important event.19 Our findings suggest that troponin is also a useful predictor of this unfavourable event.
Mortality
There is a significant amount of evidence supporting that an abnormal troponin value in patients with NSTE ACS predicts increased mortality.1–3 However, what is less clear is whether, among those with an abnormal value, the degree of troponin elevation independently predicts mortality. In a registry-based analysis of patients with NSTE ACS (n=2359), the degree of troponin elevation in those with an abnormal value at admission did not provide incremental information beyond that provided by the GRACE risk score; however, this analysis was based on initial values of troponin.9 In the FRISC (Fast Revascularization during Instability in CAD) II substudy (n=2329), there appeared to be a gradient of risk, with greater elevations of troponin concentrations having poorer prognosis with regards to mortality at 1 year.8 Finally, in the GUSTO IV (Global Use of Strategies To Open occluded arteries) trial (n=7168),7 the troponin quartile elevation independently predicted mortality (OR 1.63, 95% CI 1.43 to 1.87) after adjustment for covariates, and these findings have also been seen in smaller studies.2 6 The divergent findings of the previous studies could be related to chance or inadequate statistical power of smaller studies. The results of our study (n=16 318), one of the largest studies available, confirms that the degree of elevation independently predicts mortality as well as serious in-hospital complications.
These findings suggest that quantitative troponin values obtained within the first 24 h of admission in the spectrum of patients with acute coronary syndromes may be useful in identifying patients at increased risk for subsequent adverse in-hospital outcomes, including cardiac arrest, new-onset heart failure or cardiogenic shock, and death.
Limitations
One limitation of this study is that the measurements of troponin were not performed in a core laboratory with a single assay, and values were taken as multiples of local laboratory ULNs. Another limitation is that the timing of in-hospital complications (heart failure, cardiac arrest and cardiogenic shock) was not captured on case report forms and so the relationship of quantative troponin and these events may only be an association rather than causal. Another limitation is that the frequency of measurement of troponin was not mandated as per a protocol and was at the discretion of the treating physician. However, the external validity of these registry data may be greater as they come from a broad-based population, and the subjects are likely to be less selected than those populations included in randomised controlled trials. Another limitation is that the type of troponin assay, beyond I or T, was not captured on the case report form. A further limitation relates to the fact that non-fatal events were not centrally adjudicated by an independent committee; however, trained site abstractors used standard event definitions in the GRACE registry.
The extent of troponin concentration elevation appears to predict mortality as well as in-hospital complications such as cardiogenic shock, new heart failure and life-threatening ventricular arrhythmias, and can help assist clinicians in risk stratifying patients with NSTE ACS.
Acknowledgments
The authors thank the physicians and nurses participating in GRACE. Further information about the project, along with the complete list of participants, can be found at http://www.outcomes.org/grace.
References
Footnotes
See Editorial, p 169
Linked articles 212175.
Funding GRACE is supported by an unrestricted educational grant from Sanofi-Aventis to the Center for Outcomes Research, University of Massachusetts Medical School. Sanofi-Aventis had no involvement in the collection, analysis and interpretation of data, in the writing of the report and in the decision to submit the paper for publication.
Competing interests SSJ, DB, KAF, KAE, PGS, SGG were speakers for, and received reconsulting honoraria and a research grant from Sanofi-Aventis.
Ethics approval This study was conducted with the approval of the review boards of all participating hospitals.
Provenance and peer review Not commissioned; externally peer reviewed.