Quantitative troponin elevation does not provideincremental prognostic value beyond comprehensiverisk stratification in patients with non–ST-segmentelevation acute coronary syndromesKi-Dong Lim, MD,a Andrew T. Yan, MD,a Amparo Casanova, MD, PhD,b Raymond T. Yan, MD,a

Aurora Mendelsohn, PhD,b Sanjit Jolly, MD,a David H. Fitchett, MD,a,b Anatoly Langer, MD, MSc,a,b andShaun G. Goodman, MD, MSca,b for the Canadian ACS Registry II Investigators Toronto, Ontario, Canada

Background The aim of this study was to evaluate whether quantitative cardiac troponin (cTn) assessment canimprove risk stratification in a spectrum of patients with non–ST-segment elevation (NSTE) acute coronary syndrome (ACS)using the validated Global Registry of Acute Cardiac Events (GRACE) risk model.

Methods The Canadian ACS Registry II is a prospective, multicenter study that enrolled patients admitted to hospital witha suspected NSTE ACS within 24 hours of symptom onset. Of the total 2297 patients, those with elevated cTn (n = 1013)were further stratified into tertiles of cTn ranges. Our primary end point was death and our secondary end point was acomposite of death or/and recurrent myocardial infarction at 1-year follow-up.

Results Multivariable analysis adjusting for validated predictors of death confirmed the independent prognostic valueof any abnormal cTn (vs normal) for death (adjusted odds ratio 2.28, 95% CI 1.49-3.49, P b .001) and for thecomposite outcome (adjusted odds ratio 2.18, 95% CI 1.61-2.95, P b .001) at 1 year. With quantitative assessment, thegradient of mortality risk with increasing cTn level was not evident after adjusting for other prognosticators. Quantitative(compared to qualitative) assessment of cTn level did not improve either the GRACE risk model discrimination for 1-year death.

Conclusions Any cTn elevation is associated with higher rate of death at 1 year, but its quantitative assessment did notprove as important as its mere presence as an independent long-term prognosticator in a nonclinical trial, “real-world” NSTEACS population. (Am Heart J 2008;155:718-24.)

Historically, the diagnosis of acutemyocardial infarction(MI) was based on a combination of patient symptoms,electrocardiogram (ECG) changes, and biochemicalmarkers characteristic for myocardial injury.1 For manyyears, the biochemical gold standard for diagnosing MIhad been creatine kinase (CK) and its MB fraction(CK-MB). Recently, assays for more sensitive and specificbiochemical markers such as cardiac troponins T and I(cTnT and cTnI) have been developed and are now the

gold standard for the diagnosis of acute MI. In 2000, a jointcommittee of the European Society of Cardiology and theAmerican College of Cardiology defined acute MI as thepresence of peak cTn in serum that exceeds the 99thpercentile of the values for a reference control group.2

Previous studies have clearly demonstrated the adverseprognostic significance of an abnormal troponin.3,4

Furthermore, abnormal cTn was shown to be better thanCK/CK-MB as an independent predictor of 1-year mortal-ity in patients with acute coronary syndrome (ACS).5

In contrast to the well-known direct relationshipbetween the degree of CK and CK-MB elevation, infarctsize, and prognosis,6,7 it is unclear whether a similarrelationship exists for quantitative cTn levels. Previousretrospective analyses of adverse outcomes in selected,clinical trial patients have suggested a direct, almost linearrelationship in the short-term, whereas long-term follow-up data have demonstrated an inverted U-shapedrelationship between the degree of cTn elevation andadverse outcome.8-11 The reason for this difference in therelationship between the degree of cTn elevation andshort- and long-term clinical outcomes remains unclear.

From the aTerrence Donnelly Heart Centre, Division of Cardiology, St Michael's Hospital,University of Toronto, Toronto, Ontario, Canada, and bCanadian Heart Research Centre,Toronto, Ontario, Canada.A list of participating Canadian ACS Registry Investigators and Coordinators may be foundin the Arch Intern Med 2007;167:1009-16.The Canadian ACS Registry II was sponsored by the Canadian Heart Research Centre,Pfizer Canada Inc, Sanofi-Synethelabo Canada Inc, and Bristol-Myers Squibb Canada Inc.Submitted August 17, 2007; accepted November 9, 2007.Reprint requests: Shaun G. Goodman, MD, MSc, Division of Cardiology, St Michael'sHospital, 30 Bond Street, Room 6-034, Toronto, Ontario, Canada M5B 1W8.E-mail: goodmans@smh.toronto.on.ca0002-8703/$ - see front matter© 2008, Published by Mosby, Inc.doi:10.1016/j.ahj.2007.11.012

Moreover, the utility of quantitative cTn elevation as aprognosticator in a less selected “real-world” registrypopulation is unclear.The purpose of our study was to examine the long-term

prognostic value of different degrees of cTn elevationamong patients enrolled in the Canadian ACS Registry II.Furthermore, we sought to determine whether quantita-tive value of cTn provides incremental prognosticinformation beyond the GRACE risk model. Wehypothesized that there is a direct and independentrelationship between the degree of cTn elevation andadverse outcome across the spectrum of non–ST-segmentelevation (NSTE) ACS.

MethodsStudy design and objectivesThe Canadian ACS Registry II is a multicenter, prospective,

observational study that enrolled patients with suspected NSTEACS from 36 centers across Canada between October 2002 andDecember 2003. Patients were eligible for the study if (1) theywere !18 years old on presentation; (2) they were admitted tohospital with a suspected NSTE ACS (defined by symptomsconsistent with acute cardiac ischemia within 24 hours ofonset); and (3) the qualifying ACS was not precipitated by asignificant concurrent event such as trauma, gastrointestinalbleeding, or surgery. There were no other specific exclusioncriteria, and consecutive patient enrolment was encouraged atall sites. At each site, the designated physician or studycoordinator recorded patient demographic and clinical data,relevant laboratory results, inhospital treatment, outcome, anddischarge diagnosis and medications on standardized case reportforms, which were then scanned directly into an electronicdatabase at the Canadian Heart Research Centre. Central datachecks were performed and queries were sent for correction.The study cohort comprised all patients with cTn levels

measured at admission to hospital. The primary end point wasdeath at 1 year, and the secondary end point was a compositeoutcome of death or/and (re-) MI at 1 year (including inhospitaldeath or [re-] MI). In assessing the prognostic value of cTn, wechose all-cause mortality as our primary end point because it isthe most robust clinical outcome measure, and ascertaining thecause of death is known to be inaccurate. Furthermore, all-causemortality is the outcome predicted by the validated GRACE riskscore.12 We did not assess (re-) MI alone as an outcome measurebecause the diagnosis of MI itself is dependent on abnormal cTn.Data were collected on hospital admission, at discharge, and at1-year follow-up. All biochemical markers were measured ateach participating hospital using its own assays and referenceranges. Inhospital events were recorded on standardized casereport forms, and follow-up data were obtained by telephoneinterview for hospital survivors. To determine the independentprognostic value of cTn elevation, we examined both unad-justed and adjusted outcomes based on previously validated ACSrisk models from GRACE.12-15

Variables and statistical analysisPatients were classified in 4 groups, based on initial cTn levels

at admission to hospital (normalized to the number of times the

upper limit of normal [!ULN] of the local laboratory): normalvalues of cTn (b1.0 ! ULN), and ascending tertiles of initial cTnvalues (!ULN) for those with abnormal cTn.Categorical data are described using percentages, and

continuous variables are described with quartiles (median [25th,75th percentiles]). All descriptive statistics are reportedseparately for each cTn group. Categorical variables werecompared using !2 test, whereas continuous variables werecompared among groups using Mann-Whitney U test or Kruskal-Wallis test, where appropriate. Linear trend was explored usingCochran-Armitage test and Spearman correlation. Correlationswere assessed via Spearman-" correlation coefficients.Bivariate logistic regression analyses were performed to study

the prognostic value of increasing quantitative cTn levels.Multivariable logistic regression (with backward elimination forP > .10) was performed to determine the independentprognostic value of cTn levels, after adjustment for previouslyvalidated prognostic factors. These prognosticators were firstforced into the regression model, and then 3 other variables(history of previous MI, previous congestive heart failure [CHF],and admission cTn) were added as covariates with backwardstepwise elimination, as these factors had proved to beindependent predictors of mortality at 6-month postdischarge.16

Possible interactions with variables related to cTn (percuta-neous coronary intervention [PCI] and coronary artery bypassgraft [CABG]) were explored. Adjusted OR and 95% CI wereestimated from the fitted logistic regression models. Hosmer-

Figure 1

Flow chart of the patients in the study.

Lim et al 719American Heart JournalVolume 155, Number 4

Lemeshow goodness-of-fit test and the c-statistic were used asmeasures of calibration and discrimination of the models,respectively. Comparison between c-statistics was performedwith SAS (SAS Institute, Cary, NC) using the method by DeLonget al.17 Significance level was # = .05. Data processing andstatistical analyses were performed using SPSS version 14.0(SPSS Inc, Chicago, IL).

ResultsOf the total 2359 patients enrolled in the registry, the

study cohort comprised all patients with cTn levelsmeasured at admission to hospital (N = 2297, 2.6%excluded) (Figure 1). Vital status and other outcomes at1 year were known for 2072 (90.2%) patients in thestudy cohort.

Quantitative cTnOverall, median initial cTn was 1 ! ULN (25th and 75th

percentiles 0.25 and 4.03 ! ULN, respectively). The study

population was divided into 4 groups: normal (n = 1284),cTn "1 ! ULN; low (n = 336), cTn >1 and "2.8;intermediate (n = 340), cTn >2.8 and "12.0; and high(n = 337), cTn >12.0. Descriptive statistics per group areshown in Tables I and II.

End points: death and composite (death or/and [re]MI) at 1-year follow-upOutcome measures are summarized in Table III. Both

end points were positively related with higher levels ofcTn, and significance remained after categorization ofthe continuous variable. Any abnormal cTn wasassociated with death (unadjusted OR 2.30, 95% CI 1.62-3.28, P b .001) and with the composite end point(unadjusted OR 2.35, 95% CI 1.81-3.04, P b .001) at1 year. Similar results were obtained with cTn levelscategorized in 4 groups. Unadjusted OR comparing eachcategory with the next lower, however, did not confirmthis trend.

Table I. Baseline characteristics related to normalized initial quantitative troponin levels (N = 2297 !)

cTn levels

Normal (n = 1284):cTn !1 ! ULN

Low (n = 336):cTn >1 and!2.8 ! ULN

Intermediate (n = 340):cTn >2.8 and!12.0 ! ULN

High (n = 337):cTn >12.0 ! ULN P value†

DemographicsSex (% females) 35.1 34.2 35.6 32.3 .48Age‡ 66 [56,75] 68 [57,75] 67 [57,75] 67 [56,76] .14

Medical history (%)Current smoker 19.2 21.4 24.1 30.0 b.001Previous aspirin use 60.8 51.8 42.6 40.4 b.001Previous angina 64.7 50.0 43.2 37.7 b.001Previous angiographic stenosis 62.4 63.3 53.1 51.6 .006Previous MI 37.5 33.0 28.8 25.5 b.001Previous congestive heart failure 9.4 10.7 10.0 7.7 .55Previous stroke 9.9 12.5 9.4 10.1 .97Diabetes mellitus 24.7 24.1 29.1 32.6 .002Hypertension 60.0 59.2 56.8 59.1 .48Hyperlipidemia 60.7 50.9 47.4 49.6 b.001Previous PCI 26.0 18.8 15.3 11.0 b.001Previous CABG 17.9 11.0 9.1 8.9 b.001

Signs and symptomsSystolic blood pressure (mm Hg)‡ 148 [130,168] 150 [129,172] 146 [123,167] 145 [125,165] .077Diastolic blood pressure (mm Hg)‡ 80 [70, 91] 83 [72, 92] 82 [73, 94] 81 [70, 93] .026Heart rate (beats/min)‡ 74 [64, 87] 78 [67, 92] 80 [66, 93] 82 [70, 96] b.001Killip class (%) b.001I 88.2 81.8 80.7 75.8II 9.0 13.1 12.2 16.6III-IV 2.8 5.1 7.0 7.5

Cardiac arrest (%) 0.4 1.2 0.3 0.3 .85ECG findings (%)ST elevation/LBBB 8.9 7.5 8.8 12.2 .15ST depression 15.4 20.7 25.9 22.8 b.001

LaboratoryCreatinine (!mol/L)‡ 90 [77,106] 91 [78,108] 91 [78,105] 91 [77,110] .009

LBBB, Left bundle branch block.!N varies due to missing values (8.7% Killip class, 1.5% creatinine, b1% the others).†Linear trend (Cochran-Armitage test for categorical variables, Spearman correlation for ordinal and numerical variables).‡Median [25th, 75th percentiles].

720 Lim et alAmerican Heart Journal

April 2008

Multivariable regression analysis, after adjustment forvalidated predictors of death at 1 year, showed indepen-dent prognostic value of qualitative troponin (normal vsabnormal) for death (adjusted OR 2.28, 95% CI 1.49-3.49,P b .001) and for the composite end point (adjusted OR2.18, 95% CI 1.61-2.95, P b .001). With quantitativeassessment of cTn, each group of abnormal (ie, all 3tertiles) cTn was also independently associated withdeath (Table IVA) and the composite end point (Table VAHowever, comparing each cTn category with the nextlower category did not show a significantly highergradient of risk with increasing cTn levels for death(Table IVB) or the composite end point (Table VB).Inhospital PCI and CABG were assessed for interactionwith cTn, but they were removed from the model in thebackward stepwise procedure.The multivariable model fitted with the validated

prognosticators including qualitative cTn (normal vs

abnormal cTn) showed excellent discrimination abilityand good calibration for death (c-statistic 0.842; Hosmer-Lemeshow, P = .74) and for the composite outcome(c-statistic 0.739; Hosmer-Lemeshow, P = .27) at 1 year. Incomparison, the model fitted using the 4 groups of cTn(including normal cTn and all 3 tertiles of abnormal cTn)showed no further improvement for predicting death(c-statistic 0.845; Hosmer-Lemeshow, P = .69; compar-ison of c-statistics P = .18) or for the composite outcome(c-statistic 0.742; Hosmer-Lemeshow, P = .17; compar-ison of the c-statistics P = .27).

DiscussionIn this multicenter, prospective observational study of a

broad spectrum of patients with NSTE ACS, we con-firmed that abnormal cTn is an independent predictor of1-year death and death/MI; however, quantitative

Table III. Outcomes at 1-year follow-up (including inhospital events) related to normalized initial quantitative troponin levels: death and thecomposite end point (re-) MI or/and death

CTn levels

Normal (ULN !1) Low Intermediate High Total P for trend

Death (n = 2072) 52 (4.5) 24 (7.6) 34 (10.9) 33 (11.1) 143 (6.9) b.001Death or/and (re-) MI (n = 2026) 107 (9.5) 48 (15.6) 68 (22.1) 62 (21.8) 285 (14.1) b.001

Values are shown as n (%).

Table II. Management patterns and inhospital events related to normalized initial quantitative troponin levels (N = 2297 !)

cTn levels

Normal (n = 1284):cTn !1 ! ULN

Low (n = 336):cTn >1 and!2.8 ! ULN

Intermediate (n = 340):cTn >2.8 and!12.0 ! ULN

High (n = 337):cTn >12.0 ! ULN P value †

Medications at discharge (%)Aspirin 86.6 83.3 91.2 90.5 .015Oral anticoagulants 9.1 9.8 7.4 8.6 .52Clopidogrel 50.3 54.2 57.9 57.9 .002" Blockers 81.7 76.8 84.1 84.9 .15ACE inhibitors 60.4 58.6 65.0 73.0 b.001Angiotensin receptor blockers 10.6 8.9 9.4 5.9 .016Diuretics 26.9 31.8 27.4 26.7 .91Statins 75.7 69.3 77.9 77.7 .38

Cardiac catheterization/interventions † (%)Catheterization 55.7 63.1 67.1 66.4 b.001PCI 23.8 26.3 34.1 29.9 b.001CABG 8.5 11.0 12.9 10.6 .043

Inhospital events (%)(re-) MI 2.4 4.2 10.0 8.9 b.001CHF 6.7 12.2 13.5 16.6 b.001(re-) angina with ECG changes 11.6 8.7 17.2 13.6 .064Stroke 0.2 0.0 0.6 1.2 .017Major bleeding 1.2 2.1 2.1 0.9 .78Death 0.5 1.5 4.7 4.2 b.001

! N varies due to missing values (all b1% except for [re-] angina with ECG changes, with 7% missing values).† Linear trend (Cochran-Armitage test).

Lim et al 721American Heart JournalVolume 155, Number 4

assessment of cTn elevation did not provide incrementalprognostic value after adjustment for other establishedrisk factors.The biochemical characteristic and utility of cTn as a

diagnostic marker of myocardial injury have been welldescribed.1,18,19 Because of its increased specificitycompared to other biochemical markers such as CK-MB,cTn has now become the preferred marker of MI.2,20 Inthe recent GRACE substudy by Goodman et al21

examining 17232 patients with ACS, adding cTn-positivestatus to the traditional CK evaluation resulted in up to

29% additional patients meeting the revised criteria foracute MI.In addition to its use in diagnosis, abnormal cTn has

been shown to have important prognostic value inpatients with ACS.3-5,22,23 Even after adjustment forclinical predictors of outcome, abnormal cTn level wasshown to be an adverse prognostic indicator, suggestingindependent predictive value of abnormal cTn over andabove other clinical predictors of outcome.5 In a largemeta-analysis by Ottani et al3 including a total of 18982patients, positive cTn predicted worse outcome across

Table IVA. Multivariable regression analysis for death at 1 year,after adjustment for validated prognosticators

Mortality at 1 year

OR 95% CI P value

Age 1.05 1.03-1.07 b.001Systolic blood pressure(per 10 mm Hg increase)

0.94 0.88-1.00 .063

Heart rate(per 10 beats/min increase)

1.11 1.02-1.20 .016

Killip class (reference=class I)II 0.87 0.50-1.54 .64III 3.61 1.93-6.77 b.001IV 3.54 0.97-12.98 .056

Creatinine(per 10!mol/L increase)

1.03 1.01-1.05 b.001

ST deviation 2.31 1.54-3.46 .001Previous MI 1.85 1.21-2.82 .004Previous CHF 2.92 1.76-4.83 b.001cTn levels ! (n = 2297 total, of which 1284 had normal cTn)Low (n = 336): cTn >1and !2.8 ! ULN

1.66 0.92-2.99 .090

Intermediate (n = 340):cTn >2.8 and!12.0 ! ULN

2.47 1.44-4.24 .001

High (n = 337):cTn >12.0 ! ULN

2.84 1.64-4.89 b.001

!Three increasing levels of abnormal troponin values compared with normal(ULN !1).

Table IVB. Multivariable regression analysis for mortality at 1year, after adjustment for validated prognosticators (each troponingroup is compared with the lower group)

Mortality at 1 year

OR 95% CI P value

cTn levels !Low (n = 336): cTn >1 and!2.8 ! ULN

1.67 0.93-2.94 .090

Intermediate (n = 340):cTn >2.8 and !12.0 ! ULN

1.49 0.79-2.78 .22

High (n = 337):cTn >12.0 ! ULN

1.15 0.64-2.04 .64

!Each increasing level of abnormal troponin values compared with the lower categoryof troponin values.

Table VA. Multivariable regression analysis for composite endpoint (death or/and [re-] MI) at 1 year, after adjustment forvalidated prognosticators

Death or/and (re) MI at 1year

OR 95% CI P value

Age 1.02 1.01-1.04 b.001Systolic blood pressure(per 10 mm Hg increase)

0.98 0.93-1.02 .30

Heart rate(per 10 beats/min increase)

1.06 0.99-1.13 .084

Killip Class (reference = class I)II 1.13 0.74-1.72 .58III 3.27 1.86-5.75 b.001IV 2.10 0.64-6.91 .22

Creatinine (per 10 !mol/L increase) 1.03 1.01-1.04 b.001ST deviation 2.10 1.56-2.82 b.001Previous MI 1.48 1.09-2.02 .013Previous CHF 1.82 1.19-2.79 .006cTn levels ! (n = 2297 total, of which 1284 had normal cTn)Low (n = 336): cTn >1 and!2.8 ! ULN

1.70 1.12-2.58 .013

Intermediate (n = 340):cTn >2.8 and !12.0 ! ULN

2.64 1.80-3.87 b.001

High (n = 337): cTn >12.0 ! ULN 2.25 1.50-3.37 b.001

!Three increasing levels of abnormal troponin values compared with normal(ULN !1).

Table VB. Multivariable regression analysis for composite endpoint (death or/and [re-] MI) at 1 year, after adjustment forvalidated prognosticators

Death or/and (re-) MIat 1 year

OR 95% CIP

value

cTn levels !Low (n = 336): cTn >1 and !2.8 ! ULN 1.69 1.12-2.56 .013Intermediate (n = 340): cTn >2.8 and!12.0 ! ULN

1.56 0.98-2.44 .060

High (n = 337): cTn >12.0 ! ULN 0.85 0.55-1.32 .48

!Each increasing level of abnormal troponin values compared with the lower categoryof troponin values.

722 Lim et alAmerican Heart Journal

April 2008

the whole spectrum of ACS. In this composite analysis,the unadjusted odds ratio of death or MI at 30 days was3.44 (P b .001) for patients with abnormal cTn, with arange of approximately 3-fold increase in risk for patientswith ST-segment elevation MI and 5-fold increase in riskfor patients with NSTE MI.3

In contrast to the direct relationship between CK andCK-MB elevation and risk of death and (re-) MI,6,7 thestudies examining quantitative cTn as a prognosticator ofadverse events in patients with ACS have shown mixedresults. Short-term follow-up studies show a linearrelationship between quantitative troponin and mortal-ity,8-10 whereas long-term follow up data suggest aninverse U-shaped relationship.11 Antman et al9 examined1404 patients from TIMI IIIB trial, which enrolledpatients with resting angina with documented evidenceof coronary artery disease, and found an increase in riskof mortality with increasing absolute levels of cTnI at 42days. After adjustment for age and ST-segment depres-sion on ECG, there was a significant increase in risk ratiowith each increase of 1 ng/mL in cTnI (P = .03).Similarly, Ohman et al10 studied 855 patients fromGUSTO-IIa trial which included patients with angina andischemic ECG changes and found 30-day mortality to besignificantly correlated with cTnT levels with highercTnT predicting higher risk of death in a near linear anddirect relationship.Lindahl et al11 examined 2329 patients from the FRISC-

II trial, which enrolled patients with angina associatedwith ischemic ECG changes or elevated cTnT. At 12months, the composite of death and MI was lowest in thelowest (6.8%), modest in the highest (11.5%), and highestin the 2 intermediate cTnT groups (20.3%, 17.6%, P b.001).11 The inverted U-shaped relationship describingtheir composite outcome was largely attributable tohigher rates of recurrent MI in the 2 intermediate cTngroups. This raises the hypothesis that an unstablecoronary lesion without complete infarct-related arteryocclusion may be present in patients with an intermedi-ate rise in their cTn as compared to the highest cTn groupwho are more likely to have had infarct-related arteryocclusion and are therefore at lower risk for reocclusionand reinfarction.11

Most recently, Westerhout et al24 published a substudyof a large clinical trial population from GUSTO-IVincluding 7800 patients with ischemic chest pain andeither elevated cTn or ST-segment depression on ECG toassess the prognostic value of quantitative ST-segmentdepression and selected biomarkers. In their multivari-able regression analyses, the degree of ST-segmentdepression and cTnT elevation was found to be anindependent predictor of 30-day death (P b .001). Forpredicting 1-year mortality, however, brain natriureticpeptide was found to be more important than cTnT. Apotential limitation of this substudy, as with many of theprevious studies described above, is that their multi-

variable regression model did not include validatedpowerful prognosticators of all-cause mortality in NSTEACS such as Killip class and systolic blood pressure.25

The present study extends the findings from previousclinical trials.8-10,24 Our results reflect the Canadian ACSRegistry II population of less selected patients across thebroad spectrum of ACS, in comparison to the highlyselected clinical trial patients, and, therefore, ourfindings may be more generalizable to the “real-world”population. Moreover, we examined longer-term follow-up of adverse events up to 1-year post hospitaldischarge/transfer, whereas the majority of previousstudies examined short-term outcomes limited to eitherinhospital or shortly postdischarge status. In contrast tothe clinical trial substudies examining the prognosticvalue of cTn, the selection of covariates in our multi-variable analyses was based on the validated GRACE riskmodel,12-15 which includes powerful predictors such asage, heart rate, systolic blood pressure, Killip class,serum creatinine, and ST deviation, which is a morerobust assessment of the independent predictive valueof quantitative cTn elevation.26 This is reflected by ourmultivariable model including qualitative cTn (abnormalvs normal) for 1-year mortality, which showed goodmodel discrimination (c-statistic 0.842).To more rigorously test for the quantitative cTn level as

an independent predictor of adverse events, we com-pared in our multivariable regression model each cTngroup to the next lower category of quantitative cTngroup to test for statistically significant incremental riskwith each higher cTn category. Although abnormal cTnremained an independent predictor of mortality andthere was a trend toward increased risk with higher cTngroups, the gradient of risk with increasing cTn elevationwas not observed after adjustment for other predictors ofadverse outcome. To our knowledge, this is the onlystudy reported to date that applies such statistical rigor inanalyzing the prognostic value of quantitative cTn. Itshould be noted that our division of normalized cTncategories is new and not previously standardized. Thelack of the previously reported8-10,24 risk gradient in ourstudy may reflect the rigor of our analysis and therobustness of our conclusion.A limitation of our study is that cTnI and cTnT levels

were measured by local laboratories using differentassays. Whereas the majority of the participating centersused the 99th percentile cutoff for their ULN, somecenters used an “MI cutoff” for their ULN which madenormalization a challenge in a few cases. Patientsexcluded from the study (n = 62) did not differ from thestudy cohort in most of the baseline variables. Patientslost to follow-up at 1 year (n = 225, out of 2297) weresimilar to those for whom follow-up data at 1 year wereavailable in their baseline characteristics, except for age.Finally, inadequate power is a possibility given the smallersize of our study. Therefore, our findings should be

Lim et al 723American Heart JournalVolume 155, Number 4

confirmed in other large, unselected patient populationswith NSTE ACS.In conclusion, the results of our study suggest that

across the broad spectrum of NSTE ACS, abnormal cTnelevation is associated with increased risk of death or thecomposite outcome of death or/and (re-) MI at 1 year. Incontrast to previously reported clinical trial studies,quantitative cTn does not significantly improve theprognostic value of qualitative cTn for adverse outcomebeyond that of a comprehensive validated risk modelincluding age, hemodynamic variables, creatinine, andECG changes. Therefore, the routine use of quantitativecTn to improve risk stratification of unselected patientswith ACS beyond validated risk score assessment may notbe warranted.

References1. Malasky BR, Alpert JS. Diagnosis of myocardial injury by biochemical

markers: problems and promises. Cardiol Rev 2002;10:306-17.2. Alpert JS, Thygesen K, Antman E, et al. Myocardial infarction

redefined—a consensus document of The Joint European Society ofCardiology/American College of Cardiology Committee for theredefinition of myocardial infarction. J Am Coll Cardiol 2000;36:959-69.

3. Ottani F, Galvani M, Nicolini FA, et al. Elevated cardiac troponinlevels predict the risk of adverse outcome in patients with acutecoronary syndromes. Am Heart J 2000;140:917-27.

4. Heidenreich PA, Alloggiamento T, Melsop K, et al. The prognosticvalue of troponin in patients with non–ST elevation acute coronarysyndromes: a meta-analysis. J Am Coll Cardiol 2001;38:478-85.

5. Yan AT, Yan RT, Tan M, et al. Troponin is more useful than creatininekinase in predicting one-year mortality among acute coronarysyndrome patients. Eur Heart J 2004;25:2006-12.

6. Alexander JH, Sparapani RA, Mahaffey KW, et al. Associationbetween minor elevations of creatine kinase-MB level and mortality inpatients with acute coronary syndromes without ST-segment eleva-tion. PURSUIT Steering Committee. Platelet Glycoprotein IIb/IIIa inUnstable Angina: Receptor Suppression Using Integrilin Therapy.JAMA 2000;283:347-53.

7. Savonitto S, Granger CB, Ardissino D, et al. The prognostic value ofcreatine kinase elevations extends across the whole spectrum of acutecoronary syndromes. J Am Coll Cardiol 2002;39:22-9.

8. Lindahl B, Venge P, Wallentin L, for the FRISC Study Group. Relationbetween troponin T and the risk of subsequent cardiac events inunstable coronary artery disease. Circulation 1996;93:1651-7.

9. Antman EM, Tanasijevic MJ, Thompson B, et al. Cardiac-specifictroponin I levels to predict the risk of mortality in patients with acutecoronary syndromes. N Engl J Med 1996;335:1342-9.

10. Ohman EM, Armstrong PW, Christenson RH, et al, for the GUSTO-IIaInvestigators. Cardiac troponin T levels for risk stratification in acutemyocardial ischemia. N Engl J Med 1996;335:1333-41.

11. Lindahl B, Diderholm E, Lagerqvist B, et al, for the FRISC-IIInvestigators. Mechanisms behind the prognostic value of troponin T

in unstable coronary artery disease: a FRISC-II substudy. J Am CollCardiol 2001;38:979-86.

12. Granger CB, Goldberg RJ, Dabbous O, et al. Predictors of hospitalmortality in the global registry of acute coronary events. Arch InternMed 2003;163:2345-53.

13. de Araujo Goncalves P, Ferreira J, Aguiar C, et al. TIMI, PURSUIT,and GRACE risk scores: sustained prognostic value and interactionwith revascularization in NSTE-ACS. Eur Heart J 2005;26:865-72.

14. Yan AT, Yan RT, Tan M, et al. Risk scores for risk stratification in acutecoronary syndromes: useful but simpler is not necessarily better. EurHeart J 2007;28:1072-8.

15. Tang EW, Wong CK, Herbison P. Global Registry of Acute CoronaryEvents (GRACE) hospital discharge risk score accurately predictslong-term mortality post acute coronary syndrome. Am Heart J 2007;153:29-35.

16. Eagle KA, Lim MJ, Dabbous OH, et al. A validated prediction modelfor all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA 2004;291:2727-33.

17. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areasunder two or more correlated receiver operating characteristic curves:a nonparametric approach. Biometrics 1988;44:837-45.

18. Babuin L, Jaffe AS. Troponin: the biomarker of choice for the detectionof cardiac injury. CMAJ 2005;173:1191-202.

19. Jaffe AS, Ravkilde J, Roberts R, et al. It's time for a change to atroponin standard. Circulation 2000;102:1216-20.

20. Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA guidelinesfor the management of patients with ST-elevation myocardialinfarction—executive summary: a report of the American College ofCardiology/American Heart Association Task Force on PracticeGuidelines (Writing Committee to Revise the 1999 Guidelines for theManagement of Patients With Acute Myocardial Infarction). Circu-lation 2004;110:588-636.

21. Goodman SG, Steg PG, Eagle KA, et al. The diagnostic andprognostic impact of the redefinition of acute myocardial infarction:lessons from the Global Registry of Acute Coronary Events (GRACE).Am Heart J 2006;151:654-60.

22. Hamm CW, Ravkilde J, Gerhardt W, et al. The prognostic value ofserum troponin T in unstable angina. N Engl J Med 1992;327:146-50.

23. Olatidoye AG, Wu AH, Feng YJ, et al. Prognostic role of troponin Tversus troponin I in unstable angina pectoris for cardiac events withmeta-analysis comparing published studies. Am J Cardiol 1998;81:1405-10.

24. Westerhout CM, Fu Y, Lauer MS, et al. Short- and long-term riskstratification in acute coronary syndromes: the added value ofquantitative ST-segment depression and multiple biomarkers. J AmColl Cardiol 2006;48:939-47.

25. Khot UN, Jia G, Moliterno DJ, et al. Prognostic importance of physicalexamination for heart failure in non–ST-elevation acute coronarysyndromes: the enduring value of Killip classification. JAMA 2003;290:2174-81.

26. Yan AT, Jong P, Yan RT, et al, for the Canadian Acute CoronarySyndromes registry investigators. Clinical trial–derived risk modelmay not generalize to real-world patients with acute coronarysyndrome. Am Heart J 2004;148:1020-7.

724 Lim et alAmerican Heart Journal

April 2008