- Email: [email protected]
The new high-sensitivity cardiac troponin T assay improves risk assessment in acute coronary syndromes
Clinical Investigations
Acute Ischemic Heart Disease
The new high-sensitivity cardiac troponin T assay improves risk assessment in acute coronary syndromes Bertil Lindahl, MD, PhD, a Per Venge, MD, PhD, b and Stefan James, MD, PhD a Uppsala, Sweden
Background Cardiac troponins are currently the markers of choice for diagnosis of acute myocardial infarction and risk assessment in acute coronary syndrome (ACS). With the introduction of the new high-sensitivity cardiac troponin T (hs-cTnT) assay, it has become possible to measure cTnT even in healthy subjects. However, how the hs-cTnT assay compares with the old cTnT assay for risk assessment in ACS is still unknown. Methods Cardiac troponin T levels were measured with the new hs-cTnT assay and the old third-generation cTnT assay in serum samples collected 48 hours after randomization in 1,452 randomly selected ACS patients enrolled in the GUSTO-IV trial. During 30 days of follow-up, deaths and myocardial infarctions were recorded. At 12 months, only all-cause mortality was collected. Results The 16% of the patients that had levels higher than the 99th percentile cutoff for hs-cTnT but less than for cTnT had a similar 1-year mortality as the 60% that were positive for both assays (9.2% vs 10.7%, P = .52) and a higher 1-year mortality compared with the 24% that were negative for both assays (9.2% vs 2.6%, P = .001). For death or acute myocardial infarction at 30 days, the group that was positive only for hs-cTnT had an intermediate risk compared with the groups negative or positive for both assays (2.4%, 5.2%, and 8.7%; P b .001). Conclusion The new hs-cTnT assay, compared with the old cTnT assay, identified more patients with myocardial damage and who were at an increased risk for new cardiac events. (Am Heart J 2010;160:224-9.)
Cardiac troponins (cTns) are ideal biochemical markers for detection of myocardial cell injury because of their high sensitivity and specificity. In patients with symptoms suggestive of acute coronary syndrome (ACS), measurement of cTns in blood is currently the procedure of choice for diagnosis of acute myocardial infarction (AMI)1 and for risk assessment.2 Numerous studies have been published showing, almost without exception, that troponin elevation is predictive of new cardiac events, both short and long term.3 We have previously shown that more sensitive cTn assays improve the risk assessment.4 With the introduction of the high-sensitivity cardiac troponin T (hs-cTnT) assay, it has become possible to measure cTnT even in healthy subjects5 and substantially shorten the time to diagnosis AMI compared with the previous-generation
a
From the Department of Medical Sciences, Cardiology, and Uppsala Clinical Research Center, University of Uppsala, Sweden, and bDepartment of Medical Sciences, Clinical Chemistry, University of Uppsala, Uppsala, Sweden. Submitted March 14, 2010; accepted May 6, 2010. Reprint requests: Bertil Lindahl, MD, PhD, Uppsala Clinical Research center, University Hospital, SE-751 85 Uppsala, Sweden. E-mail: [email protected] 0002-8703/$ - see front matter © 2010, Mosby, Inc. All rights reserved. doi:10.1016/j.ahj.2010.05.023
cTnT assay.5,6 However, the impact of the hs-cTnT assay on risk assessment is still unknown. Furthermore, although the 99th percentile level of a healthy reference population is chosen for diagnosis of AMI, the optimal cutoff level for risk stratification is also unknown and might be substantially lower.7 Therefore, the aim of this study was to compare the new hs-cTnT assay with the third-generation cTnT assay for risk assessment in ACS. For that purpose, we have analyzed the levels in a subsample of the Global Use of Strategies to Open coronary arteries IV trial (GUSTO IV) cohort in which outcome data of 1-year mortality are available, as well as the composite end point death and/or MI at 30 days.
Patients and methods Patient selection and funding The GUSTO-IV trial included 7,800 patients with non–STelevation ACS from 458 centers in 24 countries during 1999 and 2000. The design and main results of the trial have been published8; the main result showed that the abciximab infusion compared with placebo had a completely neutral effect on primary as well as on secondary end points. Eligible patients were ≥21 years of age with one or more episodes of angina lasting ≥5 minutes, within 24 hours of admission, and either a positive cTnT or cardiac troponin I (cTnI) test result (higher than the upper limit of normal for the local assay) or ≥0.5 mm
American Heart Journal Volume 160, Number 2
of transient or persistent ST-segment depression. The study was conducted in a double-blind fashion with patients randomly assigned to 3 treatment groups: abciximab infusion for 24 hours or 48 hours or corresponding placebo infusion. All patients received 150 to 325 mg aspirin orally once a day for long-term treatment as well as either unfractionated heparin infusion for 48 hours (n = 6,826) or subcutaneous dalteparin every 12th hour for 5 to 7 days (n = 974). Coronary angiography was not to be performed during or within 12 hours after the completion of study agent infusion. During 30 days of follow-up, mortality and rate of adjudicated myocardial infarctions were recorded. At 12 months, only all-cause mortality was collected. For this study, we selected randomly serum samples from a cohort of 1,452 patients. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper, and its final contents. The GUSTO-IV trial was funded by Centocor Inc. The hs-cTnT substudy was supported by funding from Uppsala University Hospital and Roche Diagnostics.
Laboratory analyses Venous blood samples were collected in evacuated tubes via a direct venous puncture at baseline and 48 hours after randomization. Aliquots of the serum samples were stored at −70°C and were analyzed at the Department of Clinical Chemistry, Uppsala University Hospital, Sweden. In 48 hours, serum samples' hs-cTnT levels were measured by the Elecsys_hs_TnT assay on a Cobas instrument (Roche Diagnostics, Basel, Switzerland) with, according to the manufacturer, a detection limit of 3 ng/L, 99th percentile level of 14 ng/L, and a 10% coefficient of variation (CV) level of 13 ng/L; and cTnT levels were determined by a third-generation assay on an Elecsys (Roche Diagnostics) with a detection limit of 10 ng/L (0.01 μg/L), 99th percentile level b0.01 μg/L (b10 ng/L), and a 10% CV level of 30 ng/L (0.03 μg/L). In the randomization serum samples, NT-pro–brain natriuretic peptide (NT-proBNP) (Roche Diagnostics) and Creactive protein (CRP) (Immulite CRP, Diagnostic Products Corporation, Los Angeles, CA) levels were determined. Serum creatinine was analyzed at the local laboratories at randomization. The creatinine clearance rate was calculated with the Cockcroft and Gault equation, as follows: [(140 − age) × weight (in kilograms)]/serum creatinine (in micromoles per liter).9
Statistics Medians were shown with 25th to 75th percentiles for continuous variables. Differences in categorical baseline variables between groups were evaluated with χ2 test. Differences in sensitivity were evaluated with the McNemar test. Differences between median values for continuous variables were evaluated with Mann-Whitney U or Kruskal-Wallis tests as appropriate. Odds ratios regarding 1-year mortality, death/AMI at 30 days, and percutaneous coronary intervention (PCI) rate at 30 days for different groups based on the results of the hs-cTnT and cTnT assays and adjusted for estimated creatinine clearance were calculated by logistic regression analyses. All calculations were performed by the statistical software SPSS 15 (SPSS Inc, Chicago, IL).
Lindahl et al 225
Results Samples at 48 hours from randomization The median hs-cTnT level was 128 ng/L (25th-75th percentile 15-605 ng/L). Seventy-six percent of the patients had an hs-cTnT ≥14 ng/L (99th percentile), and 77% had an hs-cTnT ≥13 ng/L (10% CV). For cTnT (n = 1,439), the median level was 108 ng/L (25th-75th percentile b10-722 ng/L); and 60% of the patients had a cTnT ≥10 ng/L (99th percentile), and 58% had a cTnT ≥30 ng/L (10% CV). The correlation between hs-cTnT and cTnT levels was high (Spearman ρ = 0.91). In the 1,439 patients with both hs-cTnT and cTnT determined at 48 hours, 340 (23.6%), 231 (16.1%), and 868 (60.3%) had levels less than the 99th percentile for both hscTnT and cTnT (neg/neg group), higher than the 99th percentile for hs-cTnT and less than for cTnT (pos/neg group), and higher than the 99th percentile for both hs-cTnT and cTnT (pos/pos group), respectively. No patient had a level higher than the 99th percentile for cTnT but less than for hs-cTnT. The baseline characteristics and findings of the patients in the 3 groups are shown in Table I. Patients in the pos/neg group compared with the neg/neg group were older; were more often men; and more often had diabetes mellitus, poorer renal function, and higher baseline levels of CRP and NT-proBNP. Outcome in relation to the 99th percentile levels for hs-cTnT and cTnT The 99th percentile level of the hs-cTnT assay separated low- from high-risk patients better than the 99th percentile level of the cTnT assay (Table II), with an odds ratio for death at 1 year of 4.29 (95% CI 2.15-8.56) and 2.16 (95% CI 1.41-3.31), respectively. The corresponding odds ratio for death/AMI at 30 days was 3.57 (95% CI 1.71-7.44) and 2.62 (95% CI 1.58-4.34), respectively. By using the 99th percentile cutoff and the hs-cTnT assay, 114 (93%) of the total of 123 deaths during the first year were identified, compared with 93 deaths (76%) using the cTnT assay (P b .001). For the total of 94 deaths or AMIs at 30 days, the corresponding figures were 86 (91%) and 74 (79%), respectively (P b .001). The outcome in the 3 groups that were neg/neg, pos/ neg, and pos/pos for the hs-cTnT and cTnT assays, respectively, are shown in Figure 1. The 1-year mortality rate and PCI rate at 30 days were significantly higher in the pos/neg group compared with the neg/neg group, whereas there were no significant differences between the pos/neg and the pos/pos groups. For the rate of death or AMI at 30 days, there was a stepwise increase in the rates between the 3 groups; however, the difference between the neg/neg and the pos/neg groups did not reach statistical significance (P = .07). Because there were significant differences in the baseline renal function (Table I) between the 3 groups, we adjusted for the estimated creatine clearance in a multivariable
American Heart Journal August 2010
226 Lindahl et al
Table I. Baseline characteristics and findings according to levels higher than the 99th percentile for hs-cTnT and cTnT in the 48-hour samples (n = 1439) hs-cTnT/cTnT Variable (%)
neg/neg
n 340 Age, sex, delay Age (y)‡ 63 (54-70) Men 41 Coronary risk factors Hypertension 55 Current smoker 18 Diabetes mellitus 15 Previous and current cardiac disease Previous MI 25 Angina 57 Previous PCI 10 Previous CABG 7.4 CHF 5.9 Medication at admission at randomization Aspirin 82 β-Blocker 59 Calcium antagonist 26 Long-acting nitrate 53 ACE inhibitor 36 Biochemical markers at randomization Hs-CRP (mg/L) 2.7 (1.3-5.2) NT-proBNP (ng/L) 165 (74-362) § Creatinine clearance 71 (57-88)
P value⁎
P value†
66 (58-74) 68
b.001 b.001
b.001 b.001
60 18 24
50 25 23
.018 .004 .012
.27 .87 .014
29 53 10 9.6 7.5
33 42 9.6 9.0 6.0
.022 b.001 .93 .58 .68
.22 .35 .91 .35 .45
85 52 26 53 42
81 54 28 41 29
.45 .25 .80 b.001 b.001
.29 .12 .98 .94 .10
4.1 (1.8-9.2) 551 (277-1278) 61 (49-79)
4.6 (2.3-10.3) 981 (359-2451) 64 (49-83)
b.001 b.001 b.001
b.001 b.001 b.001
pos/neg
pos/pos
231
868
69 (60-76) 58
CABG, Coronary artery bypass grafting; CHF, congestive heart failure; ACE, angiotensin-converting enzyme. ⁎ Overall P value. † P value for comparison between neg/neg and pos/neg groups. ‡ Median and interquartile range. § Estimated creatinine clearance (Cockcroft and Gault equation) in milliliters per minute.
Table II. Rates (percentage) of death at 1 year and of death/AMI at 30 days in patients with values less than or higher than the 99th percentile for the cTnT and hs-cTnT assay, respectively
cTnT ≤10 N10 hs-cTnT b14 ≥14
Death
P value
OR (95% CI)
Death/AMI
P value
OR (95% CI)
5.3 10.7
b.001
2.16 (1.41-3.31)
3.5 8.7
b.001
2.62 (1.58-4.34)
2.6 10.4
b.001
4.29 (2.15-8.56)
2.3 7.9
b.001
3.57 (1.71-7.44)
OR, Odds ratio.
model. The resulting adjusted odds ratios are shown in Table III.
Outcome in relation to different levels of hs-cTnT The relation between the level of hs-cTnT and outcome is illustrated in Figure 2. There was an almost linear increase in 1-year mortality by increasing quintiles of hscTnT, whereas the relationship with AMI at 30 days showed an inverted U-shaped curve, with the highest risk in the second and third quintiles. We also evaluated the predictive value regarding death or AMI at 30 days of even lower cutoff levels than the
99th percentile level (Table IV). In a multivariable stepwise logistic regression analysis, 10 ng/L was chosen as the best cutoff level of all the cutoff values in Table IV for short-term risk prediction, with an odds ratio of 8.34 (95% CI 2.62-26.5). An hs-cTnT value ≥10 ng/L identified 91 (97%) of the 94 short-term events, compared with 86 (91%) using the 99th percentile cutoff value of 14 ng/L (P = .063).
Discussion The main finding of the present study in an ACS population is that the new hs-cTnT assay, compared with
American Heart Journal Volume 160, Number 2
Lindahl et al 227
Figure 1
Figure 2
Mortality at 1 year, death or AMI at 30 days, and PCI at 30 days in patients with both cTnT and hs-cTnT less than the 99th percentile level; with cTnT less than but hs-cTnT higher than the 99th percentile level; and with both cTnT and hs-cTnT higher than the 99th percentile level, respectively.
Proportion of patients (percentage) with AMI at 30 days (P = .001) and death at 1 year (P b .001), respectively, in relation to quintiles of hs-cTnT.
Table IV. Death or MI at 30 days (from 48 hours to 30 days) in relation to different cutoff levels of hs-cTnT hs-cTnT, ng/L
Table III. Odds ratios (95% CI) adjusted for creatinine clearance for 1-year mortality, death/AMI at 30 days, and PCI at 30 days in relation to results of the hs-cTnT and cTnT assays
hs-cTnT neg/ cTnT neg hs-cTnT pos/ cTnT neg hs-cTnT pos/ cTnT pos
Death
Death/AMI
PCI
1y
30 d
30 d
1
1
1
2.99 (1.33-6.72)
2.03 (0.81-5.07)
2.46 (1.49-4.06)
3.66 (1.81-7.40)
3.63 (1.72-7.64)
2.32 (1.54-3.51)
the old, less sensitive, and less precise cTnT assay, identifies more patients with myocardial damage and who are at an increased risk for new cardiac events both in the short and long term. Thus, these very minor myocardial damages, not previously possible to detect, are also important to identify. The comparison of the 2 cTnT assays showed, as expected, that they correlated well overall. However, using the 99th percentile level as cutoff, the hs-cTnT assay identified in absolute terms 16% (in relative terms 27%) more patients with evidence of myocardial damage compared with the old cTnT assay, which is in accordance with a recent study evaluating a new highsensitivity cTnI assay.10 The additional patients diagnosed with the hs-cTnT showed a slightly higher clinical risk profile because they were older, were more often men, and had more often diabetes mellitus than those negative with both assays. They also had a poorer renal function
b14 (99th percentile) ≥14 b13 (10% CV) ≥13 b10 ≥10 b8 ≥8 b6 (≈20% CV) ≥6
Death/AMI, % 2.3 7.9 2.1 7.9 1.0 8.0 1.2 7.6 1.3 7.2
(8/342) (86/1092) (7/331) (87/1103) (3/292) (91/1142) (3/243) (91/1191) (2/157) (92/1277)
P value
OR (95% CI)
b.001
3.57 (1.71-7.44)
b.001
3.96 (1.82-8.65)
b.001
8.34 (2.62-26.5)
b.001
6.62 (2.08-21.1)
.005
6.02 (1.47-24.7)
and higher baseline levels of CRP and NT-proBNP. Importantly, the 1-year mortality among these patients was not different from those with cTnT elevations higher than the 99th percentile with both assays, but was higher than those without cTnT elevation higher than the 99th percentile with either assay. For short-term cardiac events (death/AMI), the additionally identified patients had an intermediate risk. These finding are in line with what has previously been shown in comparisons between less and more sensitive cTn assays.4 An interesting observation in the present study is the N2-fold higher use of PCI in the additionally identified patients compared with hs-cTnT– negative patients (Figure 1). Because the treating physicians had access to neither hs-cTnT result nor cTnT result, the additionally identified patients must have given some other signal prompting the physicians to perform a PCI. Taken together, these results strongly support the clinical relevance of identifying even very minor myocardial damage in patients with a clinical picture of ACS.
American Heart Journal August 2010
228 Lindahl et al
In any fair comparison of methods, the choice of cutoffs is crucial. We chose the 99th percentile of the upper reference level for the comparison between the cTnT and hs-cTnT assays, as the 99th percentile level is mandated as the troponin cutoff level for the diagnosis of AMI.1 However, there is no biologically plausible reason why the 99th percentile level should constitute a specific prognostic threshold. On the contrary, the results of the present study indicate that the risk starts to rise at substantially lower levels of hs-cTnT (Table IV, Figure 2), in accordance with what has been seen with other sensitive cTn assays.7 Furthermore, the optimal cutoff level may differ considerably between different clinical end points. As we also previously have shown,11 there seems to be a more linear relation between the long-term mortality and troponin level, whereas the risk for a new AMI seems to be highest among those with only minor elevations. What is then the optimal cutoff level for risk stratification? There is no unequivocal answer to that question, simply because a cutoff is always a trade off between sensitivity and specificity and there is no universal agreement on how one should weigh the risk of false positives against false negatives. Furthermore, the “optimal” cutoff level might differ between the end point of interest as discussed above. Nevertheless, using the hs-cTnT assay for risk stratification, the 99th percentile level seems to be a fair compromise bearing in mind that a lower cutoff level such as 10 ng/L might be used to identify a group with very low risk of new cardiac events and in whom even the diagnosis of ACS might be challenged.10,12
Limitations There are some limitations of the present study. Firstly, it was undertaken in a clinical trial population in which all patients should have an “objective” sign of ischemia, that is, ST-T changes and/or elevation of troponin measured with any type of assay available at the time including the Spectral qualitative cTnI test (Spectral Diagnostics Inc, Toronto, Canada) in addition to a typical clinical presentation to be enrolled in the study. Thus, the patients in the present study probably had a higher pretest risk than a typical chest pain population seen in the emergency department. However, that implies that the negative predictive value of a negative hs-cTnT result is probably even better in routine use. Secondly, the results are based on analyses of cTnT and hs-cTnT in samples taken on the second day after onset of symptoms because we had no more frozen inclusion samples for analysis of hs-cTnT available. That implies that some patients may have had a cTnT higher than the 99th percentile level at inclusion but had become cTnT negative at the second day. However, most probably, there would also have
been some patients among those negative for hs-cTnT at day 2 that would have had an hs-cTnT value higher than the 99th percentile at inclusion not lasting until day 2, if we had had the chance to measure it. Therefore, the comparison between the 2 assays in the present study, based on analyses on the same day 2 samples, probably reflects the difference in prognostic capacity between the 2 assays adequately. Thirdly, the analysis of hs-cTnT was performed in frozen, never previously thawed samples, stored for up to 10 years. We cannot completely exclude some deviation in hs-cTnT concentration due to the long storage. However, data on the stability of frozen samples of 29 patients (hs-cTnT range 5-14 pg/mL) stored for 3 years showed excellent stability with a mean deviation of only 3% (personal communication, Dr Jochen Jarausch, Roche Diagnostics GmbH). Fourthly, the use of early revascularization in the GUSTO-IV trial was lower compared with what is the current standard in most countries. However, other studies have verified that elevated cTn is associated with an adverse prognosis also in an era of frequent early revascularizations.13 Furthermore, although differences in rate of early revascularization might influence the absolute levels of death at 1 year and death/AMI at 30 days in the cTn-positive groups, it is unlikely that it will influence the outcome in the hs-cTnT–positive/cTnTpositive group and hs-cTnT–positive/cTnT-negative group differently.
Disclosures Bertil Lindahl has received research grants from Roche Diagnostics and Radiometer A/S and lecture fees from Roche Diagnostics and Siemens Healthcare Diagnostics, and is member of the scientific advisory boards of Siemens Healthcare Diagnostics and Beckman Coulter Inc. Per Venge has received research grants from Siemens Healthcare Diagnostics, Abbott Diagnostics, Beckman Coulter Inc, Radiometer A/S, and Roche Diagnostics. Stefan James has no disclosures relevant to this paper. Funding sources: The GUSTO-IV trial was funded by Centocor. The hs-cTnT substudy was supported by funding from Uppsala University Hospital (ALF funding) and Roche Diagnostics.
References 1. Thygesen K, Alpert JS, White HD, on behalf of the Joint ESC/ACCF/ AHA/WHF Task Force for the Redefinition of Myocardial Infarction. Universal definition of myocardial infarction. Circulation 2007;116: 2634-53. 2. Morrow DA, Cannon CP, Jesse RL, et al. National Academy of Clinical Biochemistry Laboratory Medicine practice guidelines: clinical characteristics and utilization of biochemical markers in acute coronary syndromes. Circulation 2007;115:e356-75.
American Heart Journal Volume 160, Number 2
3. Ottani F, Galvani M, Nicolini FA, et al. Elevated cardiac troponin levels predict the risk of adverse outcome in patients with acute coronary syndromes. Am Heart J 2000;140:917-27. 4. Venge P, Lagerqvist B, Diderholm E, et al. Clinical performance of three cardiac troponin assays in patients with unstable coronary artery disease (a FRISC II substudy). Am J Cardiol 2002;89:1035-41. 5. Reichlin T, Hochholzer W, Bassetti S, et al. Early diagnosis of myocardial infarction with sensitive cardiac troponin assays. N Engl J Med 2009;361:858-67. 6. Giannitsis E, Kurz K, Hallermayer K, et al. Analytical validation of a high-sensitivity cardiac troponin T assay. Clin Chem 2009 [Epub ahead of print]. 7. Eggers KM, Jaffe AS, Lind L, et al. Value of cardiac troponin I cutoff concentrations below the 99th percentile for clinical decision-making. Clin Chem 2009;55:85-92. 8. GUSTO-IV-ACS-investigators. Effect of glycoprotein IIb/IIIa receptor blocker abciximab on outcome in patients with acute coronary syndromes without early coronary revascularisation:
Lindahl et al 229
9. 10.
11.
12.
13.
the GUSTO IV-ACS randomised trial. Lancet 2001;357: 1915-24. Cockcroft D, Gault M. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41. Venge P, Johnston N, Lindahl B, et al. Normal plasma levels of cardiac troponin I measured by the high-sensitivity cardiac troponin I access prototype assay and the impact on the diagnosis of myocardial ischemia. J Am Coll Cardiol 2009;54:1165-72. Lindahl B, Diderholm E, Lagerqvist B, et al. Mechanisms behind the prognostic value of troponin T in unstable coronary artery disease: a FRISC II substudy. J Am Coll Cardiol 2001;38:979-86. Wilson S, Sabatine M, Braunwald E, et al. Detection of myocardial injury in patients with unstable angina using a novel nanoparticle cardiac troponin I assay: observations from the PROTECT-TIMI 30 Trial. Am Heart J 2009;158:386-91. Eggers KM, Dellborg M, Oldgren J, et al. Risk prediction in chest pain patients by biochemical markers including estimates of renal function. Int J Cardiol 2008;128:207-13.
Acute Ischemic Heart Disease
The new high-sensitivity cardiac troponin T assay improves risk assessment in acute coronary syndromes Bertil Lindahl, MD, PhD, a Per Venge, MD, PhD, b and Stefan James, MD, PhD a Uppsala, Sweden
Background Cardiac troponins are currently the markers of choice for diagnosis of acute myocardial infarction and risk assessment in acute coronary syndrome (ACS). With the introduction of the new high-sensitivity cardiac troponin T (hs-cTnT) assay, it has become possible to measure cTnT even in healthy subjects. However, how the hs-cTnT assay compares with the old cTnT assay for risk assessment in ACS is still unknown. Methods Cardiac troponin T levels were measured with the new hs-cTnT assay and the old third-generation cTnT assay in serum samples collected 48 hours after randomization in 1,452 randomly selected ACS patients enrolled in the GUSTO-IV trial. During 30 days of follow-up, deaths and myocardial infarctions were recorded. At 12 months, only all-cause mortality was collected. Results The 16% of the patients that had levels higher than the 99th percentile cutoff for hs-cTnT but less than for cTnT had a similar 1-year mortality as the 60% that were positive for both assays (9.2% vs 10.7%, P = .52) and a higher 1-year mortality compared with the 24% that were negative for both assays (9.2% vs 2.6%, P = .001). For death or acute myocardial infarction at 30 days, the group that was positive only for hs-cTnT had an intermediate risk compared with the groups negative or positive for both assays (2.4%, 5.2%, and 8.7%; P b .001). Conclusion The new hs-cTnT assay, compared with the old cTnT assay, identified more patients with myocardial damage and who were at an increased risk for new cardiac events. (Am Heart J 2010;160:224-9.)
Cardiac troponins (cTns) are ideal biochemical markers for detection of myocardial cell injury because of their high sensitivity and specificity. In patients with symptoms suggestive of acute coronary syndrome (ACS), measurement of cTns in blood is currently the procedure of choice for diagnosis of acute myocardial infarction (AMI)1 and for risk assessment.2 Numerous studies have been published showing, almost without exception, that troponin elevation is predictive of new cardiac events, both short and long term.3 We have previously shown that more sensitive cTn assays improve the risk assessment.4 With the introduction of the high-sensitivity cardiac troponin T (hs-cTnT) assay, it has become possible to measure cTnT even in healthy subjects5 and substantially shorten the time to diagnosis AMI compared with the previous-generation
a
From the Department of Medical Sciences, Cardiology, and Uppsala Clinical Research Center, University of Uppsala, Sweden, and bDepartment of Medical Sciences, Clinical Chemistry, University of Uppsala, Uppsala, Sweden. Submitted March 14, 2010; accepted May 6, 2010. Reprint requests: Bertil Lindahl, MD, PhD, Uppsala Clinical Research center, University Hospital, SE-751 85 Uppsala, Sweden. E-mail: [email protected] 0002-8703/$ - see front matter © 2010, Mosby, Inc. All rights reserved. doi:10.1016/j.ahj.2010.05.023
cTnT assay.5,6 However, the impact of the hs-cTnT assay on risk assessment is still unknown. Furthermore, although the 99th percentile level of a healthy reference population is chosen for diagnosis of AMI, the optimal cutoff level for risk stratification is also unknown and might be substantially lower.7 Therefore, the aim of this study was to compare the new hs-cTnT assay with the third-generation cTnT assay for risk assessment in ACS. For that purpose, we have analyzed the levels in a subsample of the Global Use of Strategies to Open coronary arteries IV trial (GUSTO IV) cohort in which outcome data of 1-year mortality are available, as well as the composite end point death and/or MI at 30 days.
Patients and methods Patient selection and funding The GUSTO-IV trial included 7,800 patients with non–STelevation ACS from 458 centers in 24 countries during 1999 and 2000. The design and main results of the trial have been published8; the main result showed that the abciximab infusion compared with placebo had a completely neutral effect on primary as well as on secondary end points. Eligible patients were ≥21 years of age with one or more episodes of angina lasting ≥5 minutes, within 24 hours of admission, and either a positive cTnT or cardiac troponin I (cTnI) test result (higher than the upper limit of normal for the local assay) or ≥0.5 mm
American Heart Journal Volume 160, Number 2
of transient or persistent ST-segment depression. The study was conducted in a double-blind fashion with patients randomly assigned to 3 treatment groups: abciximab infusion for 24 hours or 48 hours or corresponding placebo infusion. All patients received 150 to 325 mg aspirin orally once a day for long-term treatment as well as either unfractionated heparin infusion for 48 hours (n = 6,826) or subcutaneous dalteparin every 12th hour for 5 to 7 days (n = 974). Coronary angiography was not to be performed during or within 12 hours after the completion of study agent infusion. During 30 days of follow-up, mortality and rate of adjudicated myocardial infarctions were recorded. At 12 months, only all-cause mortality was collected. For this study, we selected randomly serum samples from a cohort of 1,452 patients. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper, and its final contents. The GUSTO-IV trial was funded by Centocor Inc. The hs-cTnT substudy was supported by funding from Uppsala University Hospital and Roche Diagnostics.
Laboratory analyses Venous blood samples were collected in evacuated tubes via a direct venous puncture at baseline and 48 hours after randomization. Aliquots of the serum samples were stored at −70°C and were analyzed at the Department of Clinical Chemistry, Uppsala University Hospital, Sweden. In 48 hours, serum samples' hs-cTnT levels were measured by the Elecsys_hs_TnT assay on a Cobas instrument (Roche Diagnostics, Basel, Switzerland) with, according to the manufacturer, a detection limit of 3 ng/L, 99th percentile level of 14 ng/L, and a 10% coefficient of variation (CV) level of 13 ng/L; and cTnT levels were determined by a third-generation assay on an Elecsys (Roche Diagnostics) with a detection limit of 10 ng/L (0.01 μg/L), 99th percentile level b0.01 μg/L (b10 ng/L), and a 10% CV level of 30 ng/L (0.03 μg/L). In the randomization serum samples, NT-pro–brain natriuretic peptide (NT-proBNP) (Roche Diagnostics) and Creactive protein (CRP) (Immulite CRP, Diagnostic Products Corporation, Los Angeles, CA) levels were determined. Serum creatinine was analyzed at the local laboratories at randomization. The creatinine clearance rate was calculated with the Cockcroft and Gault equation, as follows: [(140 − age) × weight (in kilograms)]/serum creatinine (in micromoles per liter).9
Statistics Medians were shown with 25th to 75th percentiles for continuous variables. Differences in categorical baseline variables between groups were evaluated with χ2 test. Differences in sensitivity were evaluated with the McNemar test. Differences between median values for continuous variables were evaluated with Mann-Whitney U or Kruskal-Wallis tests as appropriate. Odds ratios regarding 1-year mortality, death/AMI at 30 days, and percutaneous coronary intervention (PCI) rate at 30 days for different groups based on the results of the hs-cTnT and cTnT assays and adjusted for estimated creatinine clearance were calculated by logistic regression analyses. All calculations were performed by the statistical software SPSS 15 (SPSS Inc, Chicago, IL).
Lindahl et al 225
Results Samples at 48 hours from randomization The median hs-cTnT level was 128 ng/L (25th-75th percentile 15-605 ng/L). Seventy-six percent of the patients had an hs-cTnT ≥14 ng/L (99th percentile), and 77% had an hs-cTnT ≥13 ng/L (10% CV). For cTnT (n = 1,439), the median level was 108 ng/L (25th-75th percentile b10-722 ng/L); and 60% of the patients had a cTnT ≥10 ng/L (99th percentile), and 58% had a cTnT ≥30 ng/L (10% CV). The correlation between hs-cTnT and cTnT levels was high (Spearman ρ = 0.91). In the 1,439 patients with both hs-cTnT and cTnT determined at 48 hours, 340 (23.6%), 231 (16.1%), and 868 (60.3%) had levels less than the 99th percentile for both hscTnT and cTnT (neg/neg group), higher than the 99th percentile for hs-cTnT and less than for cTnT (pos/neg group), and higher than the 99th percentile for both hs-cTnT and cTnT (pos/pos group), respectively. No patient had a level higher than the 99th percentile for cTnT but less than for hs-cTnT. The baseline characteristics and findings of the patients in the 3 groups are shown in Table I. Patients in the pos/neg group compared with the neg/neg group were older; were more often men; and more often had diabetes mellitus, poorer renal function, and higher baseline levels of CRP and NT-proBNP. Outcome in relation to the 99th percentile levels for hs-cTnT and cTnT The 99th percentile level of the hs-cTnT assay separated low- from high-risk patients better than the 99th percentile level of the cTnT assay (Table II), with an odds ratio for death at 1 year of 4.29 (95% CI 2.15-8.56) and 2.16 (95% CI 1.41-3.31), respectively. The corresponding odds ratio for death/AMI at 30 days was 3.57 (95% CI 1.71-7.44) and 2.62 (95% CI 1.58-4.34), respectively. By using the 99th percentile cutoff and the hs-cTnT assay, 114 (93%) of the total of 123 deaths during the first year were identified, compared with 93 deaths (76%) using the cTnT assay (P b .001). For the total of 94 deaths or AMIs at 30 days, the corresponding figures were 86 (91%) and 74 (79%), respectively (P b .001). The outcome in the 3 groups that were neg/neg, pos/ neg, and pos/pos for the hs-cTnT and cTnT assays, respectively, are shown in Figure 1. The 1-year mortality rate and PCI rate at 30 days were significantly higher in the pos/neg group compared with the neg/neg group, whereas there were no significant differences between the pos/neg and the pos/pos groups. For the rate of death or AMI at 30 days, there was a stepwise increase in the rates between the 3 groups; however, the difference between the neg/neg and the pos/neg groups did not reach statistical significance (P = .07). Because there were significant differences in the baseline renal function (Table I) between the 3 groups, we adjusted for the estimated creatine clearance in a multivariable
American Heart Journal August 2010
226 Lindahl et al
Table I. Baseline characteristics and findings according to levels higher than the 99th percentile for hs-cTnT and cTnT in the 48-hour samples (n = 1439) hs-cTnT/cTnT Variable (%)
neg/neg
n 340 Age, sex, delay Age (y)‡ 63 (54-70) Men 41 Coronary risk factors Hypertension 55 Current smoker 18 Diabetes mellitus 15 Previous and current cardiac disease Previous MI 25 Angina 57 Previous PCI 10 Previous CABG 7.4 CHF 5.9 Medication at admission at randomization Aspirin 82 β-Blocker 59 Calcium antagonist 26 Long-acting nitrate 53 ACE inhibitor 36 Biochemical markers at randomization Hs-CRP (mg/L) 2.7 (1.3-5.2) NT-proBNP (ng/L) 165 (74-362) § Creatinine clearance 71 (57-88)
P value⁎
P value†
66 (58-74) 68
b.001 b.001
b.001 b.001
60 18 24
50 25 23
.018 .004 .012
.27 .87 .014
29 53 10 9.6 7.5
33 42 9.6 9.0 6.0
.022 b.001 .93 .58 .68
.22 .35 .91 .35 .45
85 52 26 53 42
81 54 28 41 29
.45 .25 .80 b.001 b.001
.29 .12 .98 .94 .10
4.1 (1.8-9.2) 551 (277-1278) 61 (49-79)
4.6 (2.3-10.3) 981 (359-2451) 64 (49-83)
b.001 b.001 b.001
b.001 b.001 b.001
pos/neg
pos/pos
231
868
69 (60-76) 58
CABG, Coronary artery bypass grafting; CHF, congestive heart failure; ACE, angiotensin-converting enzyme. ⁎ Overall P value. † P value for comparison between neg/neg and pos/neg groups. ‡ Median and interquartile range. § Estimated creatinine clearance (Cockcroft and Gault equation) in milliliters per minute.
Table II. Rates (percentage) of death at 1 year and of death/AMI at 30 days in patients with values less than or higher than the 99th percentile for the cTnT and hs-cTnT assay, respectively
cTnT ≤10 N10 hs-cTnT b14 ≥14
Death
P value
OR (95% CI)
Death/AMI
P value
OR (95% CI)
5.3 10.7
b.001
2.16 (1.41-3.31)
3.5 8.7
b.001
2.62 (1.58-4.34)
2.6 10.4
b.001
4.29 (2.15-8.56)
2.3 7.9
b.001
3.57 (1.71-7.44)
OR, Odds ratio.
model. The resulting adjusted odds ratios are shown in Table III.
Outcome in relation to different levels of hs-cTnT The relation between the level of hs-cTnT and outcome is illustrated in Figure 2. There was an almost linear increase in 1-year mortality by increasing quintiles of hscTnT, whereas the relationship with AMI at 30 days showed an inverted U-shaped curve, with the highest risk in the second and third quintiles. We also evaluated the predictive value regarding death or AMI at 30 days of even lower cutoff levels than the
99th percentile level (Table IV). In a multivariable stepwise logistic regression analysis, 10 ng/L was chosen as the best cutoff level of all the cutoff values in Table IV for short-term risk prediction, with an odds ratio of 8.34 (95% CI 2.62-26.5). An hs-cTnT value ≥10 ng/L identified 91 (97%) of the 94 short-term events, compared with 86 (91%) using the 99th percentile cutoff value of 14 ng/L (P = .063).
Discussion The main finding of the present study in an ACS population is that the new hs-cTnT assay, compared with
American Heart Journal Volume 160, Number 2
Lindahl et al 227
Figure 1
Figure 2
Mortality at 1 year, death or AMI at 30 days, and PCI at 30 days in patients with both cTnT and hs-cTnT less than the 99th percentile level; with cTnT less than but hs-cTnT higher than the 99th percentile level; and with both cTnT and hs-cTnT higher than the 99th percentile level, respectively.
Proportion of patients (percentage) with AMI at 30 days (P = .001) and death at 1 year (P b .001), respectively, in relation to quintiles of hs-cTnT.
Table IV. Death or MI at 30 days (from 48 hours to 30 days) in relation to different cutoff levels of hs-cTnT hs-cTnT, ng/L
Table III. Odds ratios (95% CI) adjusted for creatinine clearance for 1-year mortality, death/AMI at 30 days, and PCI at 30 days in relation to results of the hs-cTnT and cTnT assays
hs-cTnT neg/ cTnT neg hs-cTnT pos/ cTnT neg hs-cTnT pos/ cTnT pos
Death
Death/AMI
PCI
1y
30 d
30 d
1
1
1
2.99 (1.33-6.72)
2.03 (0.81-5.07)
2.46 (1.49-4.06)
3.66 (1.81-7.40)
3.63 (1.72-7.64)
2.32 (1.54-3.51)
the old, less sensitive, and less precise cTnT assay, identifies more patients with myocardial damage and who are at an increased risk for new cardiac events both in the short and long term. Thus, these very minor myocardial damages, not previously possible to detect, are also important to identify. The comparison of the 2 cTnT assays showed, as expected, that they correlated well overall. However, using the 99th percentile level as cutoff, the hs-cTnT assay identified in absolute terms 16% (in relative terms 27%) more patients with evidence of myocardial damage compared with the old cTnT assay, which is in accordance with a recent study evaluating a new highsensitivity cTnI assay.10 The additional patients diagnosed with the hs-cTnT showed a slightly higher clinical risk profile because they were older, were more often men, and had more often diabetes mellitus than those negative with both assays. They also had a poorer renal function
b14 (99th percentile) ≥14 b13 (10% CV) ≥13 b10 ≥10 b8 ≥8 b6 (≈20% CV) ≥6
Death/AMI, % 2.3 7.9 2.1 7.9 1.0 8.0 1.2 7.6 1.3 7.2
(8/342) (86/1092) (7/331) (87/1103) (3/292) (91/1142) (3/243) (91/1191) (2/157) (92/1277)
P value
OR (95% CI)
b.001
3.57 (1.71-7.44)
b.001
3.96 (1.82-8.65)
b.001
8.34 (2.62-26.5)
b.001
6.62 (2.08-21.1)
.005
6.02 (1.47-24.7)
and higher baseline levels of CRP and NT-proBNP. Importantly, the 1-year mortality among these patients was not different from those with cTnT elevations higher than the 99th percentile with both assays, but was higher than those without cTnT elevation higher than the 99th percentile with either assay. For short-term cardiac events (death/AMI), the additionally identified patients had an intermediate risk. These finding are in line with what has previously been shown in comparisons between less and more sensitive cTn assays.4 An interesting observation in the present study is the N2-fold higher use of PCI in the additionally identified patients compared with hs-cTnT– negative patients (Figure 1). Because the treating physicians had access to neither hs-cTnT result nor cTnT result, the additionally identified patients must have given some other signal prompting the physicians to perform a PCI. Taken together, these results strongly support the clinical relevance of identifying even very minor myocardial damage in patients with a clinical picture of ACS.
American Heart Journal August 2010
228 Lindahl et al
In any fair comparison of methods, the choice of cutoffs is crucial. We chose the 99th percentile of the upper reference level for the comparison between the cTnT and hs-cTnT assays, as the 99th percentile level is mandated as the troponin cutoff level for the diagnosis of AMI.1 However, there is no biologically plausible reason why the 99th percentile level should constitute a specific prognostic threshold. On the contrary, the results of the present study indicate that the risk starts to rise at substantially lower levels of hs-cTnT (Table IV, Figure 2), in accordance with what has been seen with other sensitive cTn assays.7 Furthermore, the optimal cutoff level may differ considerably between different clinical end points. As we also previously have shown,11 there seems to be a more linear relation between the long-term mortality and troponin level, whereas the risk for a new AMI seems to be highest among those with only minor elevations. What is then the optimal cutoff level for risk stratification? There is no unequivocal answer to that question, simply because a cutoff is always a trade off between sensitivity and specificity and there is no universal agreement on how one should weigh the risk of false positives against false negatives. Furthermore, the “optimal” cutoff level might differ between the end point of interest as discussed above. Nevertheless, using the hs-cTnT assay for risk stratification, the 99th percentile level seems to be a fair compromise bearing in mind that a lower cutoff level such as 10 ng/L might be used to identify a group with very low risk of new cardiac events and in whom even the diagnosis of ACS might be challenged.10,12
Limitations There are some limitations of the present study. Firstly, it was undertaken in a clinical trial population in which all patients should have an “objective” sign of ischemia, that is, ST-T changes and/or elevation of troponin measured with any type of assay available at the time including the Spectral qualitative cTnI test (Spectral Diagnostics Inc, Toronto, Canada) in addition to a typical clinical presentation to be enrolled in the study. Thus, the patients in the present study probably had a higher pretest risk than a typical chest pain population seen in the emergency department. However, that implies that the negative predictive value of a negative hs-cTnT result is probably even better in routine use. Secondly, the results are based on analyses of cTnT and hs-cTnT in samples taken on the second day after onset of symptoms because we had no more frozen inclusion samples for analysis of hs-cTnT available. That implies that some patients may have had a cTnT higher than the 99th percentile level at inclusion but had become cTnT negative at the second day. However, most probably, there would also have
been some patients among those negative for hs-cTnT at day 2 that would have had an hs-cTnT value higher than the 99th percentile at inclusion not lasting until day 2, if we had had the chance to measure it. Therefore, the comparison between the 2 assays in the present study, based on analyses on the same day 2 samples, probably reflects the difference in prognostic capacity between the 2 assays adequately. Thirdly, the analysis of hs-cTnT was performed in frozen, never previously thawed samples, stored for up to 10 years. We cannot completely exclude some deviation in hs-cTnT concentration due to the long storage. However, data on the stability of frozen samples of 29 patients (hs-cTnT range 5-14 pg/mL) stored for 3 years showed excellent stability with a mean deviation of only 3% (personal communication, Dr Jochen Jarausch, Roche Diagnostics GmbH). Fourthly, the use of early revascularization in the GUSTO-IV trial was lower compared with what is the current standard in most countries. However, other studies have verified that elevated cTn is associated with an adverse prognosis also in an era of frequent early revascularizations.13 Furthermore, although differences in rate of early revascularization might influence the absolute levels of death at 1 year and death/AMI at 30 days in the cTn-positive groups, it is unlikely that it will influence the outcome in the hs-cTnT–positive/cTnTpositive group and hs-cTnT–positive/cTnT-negative group differently.
Disclosures Bertil Lindahl has received research grants from Roche Diagnostics and Radiometer A/S and lecture fees from Roche Diagnostics and Siemens Healthcare Diagnostics, and is member of the scientific advisory boards of Siemens Healthcare Diagnostics and Beckman Coulter Inc. Per Venge has received research grants from Siemens Healthcare Diagnostics, Abbott Diagnostics, Beckman Coulter Inc, Radiometer A/S, and Roche Diagnostics. Stefan James has no disclosures relevant to this paper. Funding sources: The GUSTO-IV trial was funded by Centocor. The hs-cTnT substudy was supported by funding from Uppsala University Hospital (ALF funding) and Roche Diagnostics.
References 1. Thygesen K, Alpert JS, White HD, on behalf of the Joint ESC/ACCF/ AHA/WHF Task Force for the Redefinition of Myocardial Infarction. Universal definition of myocardial infarction. Circulation 2007;116: 2634-53. 2. Morrow DA, Cannon CP, Jesse RL, et al. National Academy of Clinical Biochemistry Laboratory Medicine practice guidelines: clinical characteristics and utilization of biochemical markers in acute coronary syndromes. Circulation 2007;115:e356-75.
American Heart Journal Volume 160, Number 2
3. Ottani F, Galvani M, Nicolini FA, et al. Elevated cardiac troponin levels predict the risk of adverse outcome in patients with acute coronary syndromes. Am Heart J 2000;140:917-27. 4. Venge P, Lagerqvist B, Diderholm E, et al. Clinical performance of three cardiac troponin assays in patients with unstable coronary artery disease (a FRISC II substudy). Am J Cardiol 2002;89:1035-41. 5. Reichlin T, Hochholzer W, Bassetti S, et al. Early diagnosis of myocardial infarction with sensitive cardiac troponin assays. N Engl J Med 2009;361:858-67. 6. Giannitsis E, Kurz K, Hallermayer K, et al. Analytical validation of a high-sensitivity cardiac troponin T assay. Clin Chem 2009 [Epub ahead of print]. 7. Eggers KM, Jaffe AS, Lind L, et al. Value of cardiac troponin I cutoff concentrations below the 99th percentile for clinical decision-making. Clin Chem 2009;55:85-92. 8. GUSTO-IV-ACS-investigators. Effect of glycoprotein IIb/IIIa receptor blocker abciximab on outcome in patients with acute coronary syndromes without early coronary revascularisation:
Lindahl et al 229
9. 10.
11.
12.
13.
the GUSTO IV-ACS randomised trial. Lancet 2001;357: 1915-24. Cockcroft D, Gault M. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41. Venge P, Johnston N, Lindahl B, et al. Normal plasma levels of cardiac troponin I measured by the high-sensitivity cardiac troponin I access prototype assay and the impact on the diagnosis of myocardial ischemia. J Am Coll Cardiol 2009;54:1165-72. Lindahl B, Diderholm E, Lagerqvist B, et al. Mechanisms behind the prognostic value of troponin T in unstable coronary artery disease: a FRISC II substudy. J Am Coll Cardiol 2001;38:979-86. Wilson S, Sabatine M, Braunwald E, et al. Detection of myocardial injury in patients with unstable angina using a novel nanoparticle cardiac troponin I assay: observations from the PROTECT-TIMI 30 Trial. Am Heart J 2009;158:386-91. Eggers KM, Dellborg M, Oldgren J, et al. Risk prediction in chest pain patients by biochemical markers including estimates of renal function. Int J Cardiol 2008;128:207-13.