Author + information
- Received March 23, 2007
- Revision received August 10, 2007
- Accepted September 7, 2007
- Published online January 22, 2008.
- Evangelos Giannitsis, MD⁎,1,4,
- Henning Steen, MD⁎,4,
- Kerstin Kurz, MD⁎,
- Boris Ivandic, MD⁎,
- Anke C. Simon, MD⁎,
- Simon Futterer, MD⁎,
- Christian Schild, MD⁎,
- Peter Isfort, MD⁎,
- Allan S. Jaffe, MD, FACC†,3,⁎ ( and )
- Hugo A. Katus, MD, FACC⁎,2
- ↵⁎Reprint requests and correspondence:
Dr. Allan S. Jaffe, Mayo Clinic and Mayo Medical School, CV Division, Gonda 5, 200 First Street SW, Rochester, Minnesota 55905.
Objectives We compared single-point cardiac troponin T (cTnT) measurements with parameters from serial sampling during 96 h after acute myocardial infarction with magnetic resonance imaging measured infarct mass.
Background Contrast-enhanced magnetic resonance imaging (CE-MRI) allows exact quantification of myocardial infarct size. Clinically, measurement of cardiac biomarkers is a more convenient alternative.
Methods The CE-MRI infarct mass was determined 4 days after primary percutaneous coronary intervention in 31 ST-segment elevation myocardial infarction (STEMI) and 30 non–ST-segment elevation myocardial infarction (NSTEMI) patients. All single-point, peak, and integrated area under the curve (AUC) cTnT values were plotted against CE-MRI infarct mass.
Results All single-point and serial cTnT values were significantly higher in STEMI than in NSTEMI (p < 0.01) patients. Except for the admission values, all single-point values on any of the first 4 days, peak cTnT and AUC cTnT were found to correlate comparably well with infarct mass. Among single-point measurements, cTnT on day 4 (cTnTD4) showed highest correlation and performed as well as peak cTnT or AUC cTnT (r = 0.66 vs. r = 0.65 vs. r = 0.69). Receiver-operator characteristic analysis demonstrated that cTnTD4 >0.84 μg/l predicted infarct mass above median as well as peak cTnT >1.57 μg/l or AUC cTnT (receiver-operator characteristic for AUC: 0.839 vs. 0.866 vs. 0.893). However, estimation of infarct mass with cTnTD4, peak cTnT, and AUC cTnT was worse in patients with NSTEMI (r = 0.36, r = 0.5, r = 0.36) than in STEMI (r = 0.75 vs. r = 0.65 vs. r = 0.76).
Conclusions All single-point cTnTs, except on admission, give a good estimation of infarct size and perform as well as peak cTnT or AUC cTnT. Infarct estimation by single-point measurements, particularly cTnTD4, may gain clinical acceptance because the measurement is easy and inexpensive.
Prognosis after acute myocardial infarction (AMI) is strongly determined by the extent of myocardial injury (1). Currently, several imaging techniques, including thallium sestamibi, positron emission tomography, and contrast-enhanced magnetic resonance imaging (CE-MRI), are used for quantification of infarct size (2). Among these techniques, some prefer cardiovascular magnetic resonance (CMR) imaging because it allows noninvasive assessment of myocardial function and viability with high spatial resolution and is superior to single-photon emission computed tomography (SPECT) for the identification of subendocardial AMI (3,4). Furthermore, the technique is highly sensitive and permits quantification of small areas of myocardial injury attributable to native coronary artery disease and/or percutaneous coronary interventions (5,6). However, the use of cardiac magnetic resonance (MR) imaging for quantification of infarct size is limited by availability and high cost. Therefore, estimating infarct size from concentrations or activities of cardiac proteins in peripheral blood, as has been done for years, is a convenient alternative (2).
Previous work depended on creatine kinase-MB fraction (CK-MB), but it is now clear that cardiac troponin T (cTnT) or cardiac troponin I (cTnI) possess better sensitivity and specificity and are, therefore, the preferred markers for diagnosis and risk stratification of patients with acute coronary syndrome and AMI (7,8). There is strong evidence to indicate that both serial and single-point measurements of cTnT correlate well with infarct size as measured at postmortem in animal experiments (9,10) and/or with various imaging techniques in clinical studies (11–13). However, MR imaging studies to determine the optimal timing for single-point measurements and compare such measurements to serial measurements are currently not available.
Accordingly, the present study sought to determine the best single-measurement time point and its performance as compared to peak and cumulative cTnT values (area under the curve [AUC]) values after AMI. Peak and AUC values were determined from daily serial sampling over 96 h, and all values were correlated to infarct size that was quantified using contrast-enhanced CE-MRI.
Only patients with first AMI were enrolled. Using 12-lead standard electrocardiogram, AMI patients were categorized into acute ST-segment elevation myocardial infarction (STEMI) and non–ST-segment elevation myocardial infarction (NSTEMI). We defined STEMI by the presence of ST segment elevation ≥0.1 mV in unipolar leads or elevation ≥0.2 mV in anterior wall leads and was retrospectively confirmed by elevated cTnT above 0.03 μg/l. Non–STEMI was defined using the diagnostic criteria of the European Society of Cardiology/American Heart Association Task Force on Myocardial Infarction and required an elevation of cTnT above 0.03 μg/l, which is the 10% coefficient of variation value for this assay used clinically at our institution on at least 1 occasion within 24 h after the ischemic index event, with a rise or fall during subsequent sampling (8). Patients with renal failure having a glomerular filtration rate below 60 ml/min/1.73 m2 were excluded because it might result in overestimation of infarct size and because of potential toxicity related to gadolinium use (14). Blood samples were taken on admissions and daily for 96 h. After collection, blood samples were centrifuged immediately and serum was stored at −20°C until analysis. We measured cTnT on the same Elecsys analyzer (Roche Diagnostics, Mannheim, Germany) using the third-generation assay (15). The detection limit of the assay is 0.01 μg/l. The decision limit used for diagnosis of AMI is 0.03 μg/l, with an imprecision of <10%. The laboratory staff responsible for measurements were blinded to the patient data.
We measured single-point cTnT at admission and after 24, 48, 72, and 96 h. Peak cTnT and cumulative cTnT over 96 h were determined retrospectively from serial samples. A value was defined as peak if it was the highest in the concentration time course and if there was at least 1 lower value before and after this maximum value.
All decisions regarding acute reperfusion therapy and clinical management of patients were left to the discretion of the attending cardiologist, who was not involved in the study. The study protocol was approved by the local ethical committee of the University of Heidelberg, and all patients gave informed consent before study enrollment.
Cardiac magnetic resonance imaging (MRI) was performed in all patients before discharge unless they had contraindications, withdrew their consent, experienced claustrophobia, or reported shortness of breath prohibiting adequate breath holding during the MRI examination.
Cardiac MRI was performed in a 1.5-T whole-body MRI scanner (Philips Achieva Intera, Philips, Best, the Netherlands). Assessment of resting left ventricular function was determined by cine images using a steady-state free precession sequence in continuous short-axes planes covering the whole left ventricle from base to apex, as well as 2- and 4-chamber views. Ten minutes after gadolinium contrast injection (Gd = 0.2 mmol/kg of gadopentetate dimeglumine [Schering, Berlin, Germany]), 3 volume stacks of delayed contrast-enhanced images covering the whole left ventricle were planned on previous short-, 2-, and 4-chamber axes. End-diastolic and -systolic volumes with resulting ejection fraction were generated manually using short-axes volumetry. Infarct size was visually defined as area of delayed hyperenhancement on short-axes views and determined manually by delineation of hyperenhanced versus normally saturated dark myocardium. Magnetic resonance imaging analysis included quantification of absolute infarct size (g) and left ventricular function in all patients.
Means with standard deviation or medians with 25th and 75th percentiles were calculated to describe continuous variables. Given that cTnT values and infarct size values were not normally distributed, correlations were calculated using Spearman rank correlation. Paired and unpaired correlations were analyzed for differences using z-statistics. Receiver-operator characteristic (ROC) curves were constructed to examine the relationship between biomarker estimates and CE-MRI infarct size. For ROC analysis, the patients were categorized into 2 data sets: those with CE-MRI infarct size above median and those below median. Finally, multivariate logistic regression analysis was performed to test for significant predictors of infarct size, including variables with a univariate p < 0.1. For all analyses, a p value <0.05 was regarded as statistically significant. All statistical analyses were performed using MedCalc Version 9.2 for Windows (MedCalc Software, Mariakerke, Belgium).
A total of 61 patients with AMI were included. Of these, 31 patients had a STEMI and 30 patients had a NSTEMI. Median time from onset of symptoms to reperfusion was 477 min (95% confidence interval [CI] 252.7 to 608.6 min) for the entire cohort, 224 min (95% CI 124.2 to 362.1 min) for STEMI, and 709.5 min (95% CI 516.9 to 1,332.75 min) for NSTEMI. Median door-to-balloon-times were 77.5 min (95% CI 49.95 to 124.6 min) for the entire cohort, 46.5 min (95% CI 35.4 to 70.4 min) for STEMI, and 221 min (95% CI 129.1 to 377.3 min) for NSTEMI.
On initial angiogram, patients with STEMI tended to more often have an occluded infarct-related artery (Thrombolysis In Myocardial Infarction flow grade 0 or 1) than patients with NSTEMI (61.3% vs. 43%, p = 0.08). Primary percutaneous coronary intervention (PCI) in STEMI or early PCI in NSTEMI patients was attempted in 60 of 61 patients and was successful in 58 of 61 patients (95.1%). Reperfusion success, defined as post-procedural Thrombolysis In Myocardial Infarction flow grade 2 or 3, was comparable in STEMI and NSTEMI (96.8% vs. 93.3%).
Baseline clinical characteristics of the entire study population are displayed in Table 1.
We performed CE-MRI after a median of 4 days (range 3 to 4). Patients with a STEMI had a significantly higher mean infarct mass (mean [95% CI]: 34.24 g (23.19 to 45.29 g) vs. 12.61 g (7.79 to 17.44 g), p = 0.0009], relative infarct size (20.32% [15.24 to 25.40] vs. 8.05% [5.53 to 10.56], p = 0.0001), and a trend to lower LV ejection fraction (54.45% [50.53 to 58.37] vs. 60.21% [56.60 to 63.83], p = 0.089).
A total of 280 samples were obtained from 61 patients, yielding a mean of 4.6 samples per patient. Median cTnT values with corresponding 95% CI for the median were 1.7 μg/l (1.09 to 2.5) on day 1, 1.64 μg/l (1.01 to 1.98) on day 2, 1.49 μ/l (0.93 to 1.89) on day 3, 1.27 μg/l (0.82 to 1.81) on day 4, and 23.0 μg/l (19 to 27.82) for peak cTnT value.
By infarct type, single-point cTnT values at 24 h (2.67 μg/l vs. 0.6 μg/l, p < 0.0001), 48 h (2.1 μg/l vs. 1.07 μg/l, p = 0.0019), 72 h (2.02 μg/l vs. 0.83 μg/l, p = 0.0017), and 96 h (1.86 μg/l vs. 0.76 μg/l, p = 0.027) were significantly higher in STEMI than in NSTEMI. Likewise, serial values, including AUC cTnT, (203.3 vs. 73.4, p = 0.0006), peak cTnT (3.08 μg/l vs. 1.08 μg/l, p = 0.0001), peakCKact (1,645 IU vs. 739 IU, p = 0.0008), and peak CK-MB mass (40.8 μg/l vs. 24.9 μg/l, p = 0.043), were significantly higher in STEMI than in NSTEMI. Consistently, time-to-peak cTnT values (20.5 h vs. 36.5 h, p = 0.0025) were shorter in STEMI than in NSTEMI.
Regression coefficients measured by Spearman rank correlation between biomarker estimates and absolute MRI infarct mass for the entire cohort are displayed in Figure 1.
Measurements of cTnT on any single point after admission correlated comparably well with infarct mass and were not different from peak cTnT or integration of the AUC from serial sampling.
The ROC analysis displays the performance of cTnT at 96 h, peak, and AUC values for prediction of infarct size above median, that is, 13.6 g (Fig. 2), demonstrating similar AUC values for cTnT at 96 h, peak cTnT, and AUC cTnT.
Impact of infarct size on performance of biomarker estimates
Three categories were formed to test the impact of infarct size on the performance of biomarker estimates. Spearman correlation coefficients between biomarker estimate and categories were compared (Table 2). Patients with NSTEMI or CE-MRI infarct size below the median (13.6 g) showed a substantially worse performance of cTnT than patients with STEMI, anterior MI, or non-anterior MI.
Multivariate analysis for prediction of infarct size
All significant univariate predictors were included in the multiple logistic regression model to examine their independent predictive value for estimation of infarct size above median (13.6 g). Keeping all other variables constant, all single-point and peak cTnT were tested to obtain the model with the highest chi-square value. In this final model, cTnT on Day 4 and presence of STEMI versus NSTEMI were the only independent significant predictors of larger infarct size (Table 3). A secondary analysis within STEMI or NSTEMI gave no meaningful results owing to the small number of patients.
Our data indicate that measurement of cTnT at a single point on any of the first 4 days or using the peak-value correlates well with infarct mass determined by CE-MRI. These data points perform as well as serial sampling estimates, including integration of AUC. We would suggest that the 96-h value would be convenient to use clinically, and it would be helpful for clinical trials to also use a similar value. However, correlations of all values are similar, although with a different slope, so clinicians can use any of the results if they so choose, and with time they should be able to develop a good estimate for infarct size.
Our results are not nearly as robust in patients with small infarcts, and that is likely the reason for the poorer correlations with NSTEMI, which were smaller by both cTnT measurements and by CE-MRI. This may be related to the variety of pathophysiologies that exist in patients with NSTEMI (16). In addition, however, microinfarcts associated with minor elevations of cTnT may escape visualization with CE-MRI or be difficult to quantify owing to insufficient visualization caused by inadequate spatial resolution or partial volume effects. Interestingly, Ricciardi et al. (5) found more promising results with visualization of a mean infarct size as low as 2.0 g. In that study, 5 of 9 patients with late hyperenhancement had suffered a side-branch occlusion during PCI, but none developed Q-wave infarction. Interestingly, new wall motion abnormalities were detected in only 1 patient with late hyperenhancement. The reason for this disparity is unclear but may involve the use of a single breath-hold 3-dimensional inversion recovery sequence with slightly lower spatial resolution in our study, and, conversely, a potential underestimation of infarct size because of incomplete coverage of the left ventricle coming from single-slice inversion recovery technique in the study of Ricciardi et al. (5).
Single versus serial cTnT for estimation of AMI
The most accurate methods for estimation of infarct size in the pre-reperfusion era were based on complete recovery of CK or its isoforms (17,18) or LD isoenzymes released from the infarcted myocardial tissue (19,20). Measurement of maximum enzyme values of CK or CK-MB proves convenient but has a propensity to overestimate infarct size, particularly when spontaneous reperfusion, PCI, or thrombolysis-induced reperfusion leads to an earlier and higher maximum CK level at identical amounts of left ventricular impairment (21–23). In addition, the peak value is influenced by release ratio, which is much higher after reperfusion (24); the plasma half-life of the cardiac constituent (25); and the magnitude of the AMI itself (26–28). In addition, CK and, to a lower extent, CK-MB mass lack absolute cardiospecificity, thus increasing the likelihood of overestimation of infarct size attributable to skeletal muscle damage. Other investigators have advocated for lactate dehydrogenase, which is cardio-unspecific, or the more cardiospecific derivative hydroxybutyrate dehydrogenase (HBDH) because these enzymes demonstrate a slower release from the myocyte and may manifest a more consistent release ratio despite successful reperfusion (19). By using HBDH at 72 h, others have found (20) minimal underestimation of infarct size compared to HBDH at 96 or 120 h. However, as noted by Sobel et al. (29), there is substantial HBDH in red cells, and hemorrhage can occur after reperfusion, confounding this measure of infarct size as well.
At present, the cardiac troponins are the preferred cardiac markers for the diagnosis of AMI owing to their superior sensitivity and specificity (8). In addition, unlike the small cytosolic cTnT fraction that may be prone to an accelerated release from the infarcted myocardium after successful reperfusion, the majority of the protein resides in the gradually released structurally bound fraction that manifests a more consistent release ratio (21,30). Experimental work has shown that the increases in serum troponin parallel the loss in tissue concentrations of troponin (31) and that blood values of cardiac troponin T correlate with histochemical infarct size in the non-reperfused experimental AMI (9,10). Interestingly, single-point measurements at 96 h after left anterior descending coronary artery ligation were as effective as measurement of cumulative release requiring tight serial sampling over 120 h or peak concentrations with retrospective location of the peaks, requiring at least 3 samples (9). Consistently, an increasing number of imaging studies have confirmed that the more convenient single-point cTnT or cTnI perform as well as cumulative release in humans (11–13). The SPECT or CE-MRI infarct size has been found to correlate well with single-point values of cTnT at approximately 72 h (12,13) and 96 h (11) and with single-point values of cTnI between 24 and 48 h (13,32). This is likely a reflection of the stability of the release ratio related to the structurally bound pool.
In the present study, the best single-point cTnT value was identified at 96 h and performed as effectively as peak cTnT and AUC cTnT over 96 h. For the latter, it is reasonable to speculate that performance of cumulative measurements could be improved by expansion of serial measurements beyond 96 h. However, prolongation of sampling is likely to reduce the attractiveness and acceptance of the method.
In the multivariate logistic regression analysis, high cTnT values and presence of STEMI emerged as significant independent predictors of larger infarct size. Interestingly, LV ejection fraction, which was preserved in most patients, failed to predict infarct size. This may be attributable to the fact that we could include only patients with first infarctions, and those who could not breath-hold because of respiratory difficulties could not be imaged.
Performance of single or serial cTnT in small AMI
In our study, prediction of infarct size with either single or serial cTnT was considerably better in patients with larger CE-MRI infarct size (i.e., mean infarct mass 34.24 g in STEMI) compared with smaller infarct size including NSTEMI, non-anterior infarct location, or CE-MRI infarct size below median (mean infarct mass 12.61 g, 18 g, and 13.6 g, respectively). A previous report had already alluded to the finding that correlation between infarct size and cTnT measured 72 to 96 h after infarct onset is inferior in NSTEMI than in STEMI (10). It has been speculated that the poorer correlation was due to suboptimal timing of sampling in NSTEMI and in part to inadequate visualization with current CE-MRI technology. Both likely play a role. Several additional reasons may also contribute. First, time-release curves of large STEMI differ markedly from small NSTEMI or microinfarcts and thus may have an impact on serial measurement algorithms, likely because of fluctuations in perfusion and thus variable release of marker. Thus, single-point measurements, for example, for peak, are less unequivocal and occur earlier for NSTEMI (48 or 72 h) than for STEMI (72 or 96 h). This latter finding is consistent with the prior report of earlier peaking of markers in patients with NSTEMI of Morrison et al (33). Second, NSTEMI comprises a heterogenous group of AMI, including strictly posterior AMI, owing to occlusion of the left circumflex artery as well as microinfarcts, which may hamper optimal timing even within the class of NSTEMI. Third, image resolution of MRI may be too low to allow visualization of very small infarcts or impair accurate quantification of small infarcts. Other possible reasons for incorrect quantification of small infarcts by CE-MRI include partial volume effects and higher intra- and interobserver variability at the low end of spatial resolution.
Although not addressed in the present study, as we had very few patients with prior AMI, it is likely that troponin and CMR will not correlate nearly as closely in patients with prior AMI than in those with first AMI because CE-MRI will measure both areas of damage. However, this would suggest synergism between the 2 methods, with 1 providing an estimate of the most recent insult and CMR integrating damage over time.
Single measurement of cTnT on any of the first 4 days, particularly on day 4 after onset of AMI, gives a good estimation of infarct size. Single-point measurements are convenient, easy, and inexpensive and may gain clinical acceptance because they are as effective as serial measurements. However, our findings may not be extrapolated to very small AMI unless imaging technologies have improved to allow exact visualization of small infarct areas. In addition, more efforts are required to refine sampling protocols in the heterogenous class of NSTEMI.
↵1 Dr. Giannitsis has received financial support for studies from Roche Diagnostics and is a consultant to Roche Diagnostics.
↵2 Dr. Katus developed the assay for cTnT and holds a patent jointly with Roche Diagnostics. He has received research support from Roche Diagnostics.
↵3 Dr. Jaffe is a consultant and has received research support from Dade-Behring and Beckman-Coulter. He is a consultant to Ortho Diagnostics and has consulted over time for most of the major diagnostic companies.
↵4 Drs. Giannitsis and Steen contributed equally to this article.
- Abbreviations and Acronyms
- acute myocardial infarction
- area under the curve
- contrast-enhanced magnetic resonance imaging
- confidence interval
- creatine kinase-MB fraction
- cardiovascular magnetic resonance
- cardiac troponin I or T
- magnetic resonance
- magnetic resonance imaging
- non–ST-segment elevation myocardial infarction
- percutaneous coronary intervention
- receiver-operator characteristic
- single-photon emission computed tomography
- ST-segment elevation myocardial infarction
- Received March 23, 2007.
- Revision received August 10, 2007.
- Accepted September 7, 2007.
- American College of Cardiology Foundation
- Gibbons R.J.,
- Valeti U.S.,
- Araoz P.A.,
- Jaffe A.S.
- Kim R.J.,
- Fieno D.S.,
- Parrish T.B.,
- et al.
- Ricciardi M.J.,
- Wu E.,
- Davidson C.J.,
- et al.
- Selvanayagam J.B.,
- Porto I.,
- Channon K.,
- et al.
- Katus H.A.,
- Remppis A.,
- Neumann F.J.,
- et al.
- Alpert J.S.,
- Thygesen K.,
- Antman E.,
- Bassand J.P.
- Steen H.,
- Giannitsis E.,
- Futterer S.,
- Merten C.,
- Juenger C.,
- Katus H.A.
- Licka M.,
- Zimmermann R.,
- Zehelein J.,
- Dengler T.J.,
- Katus H.A.,
- Kubler W.
- Panteghini M.,
- Cuccia C.,
- Bonetti G.,
- Giubbini R.,
- Pagani F.,
- Bonini E.
- Hallermayer K.,
- Klenner D.,
- Vogel R.
- Roberts R.,
- Henry P.D.,
- Sobel B.E.
- Witteveen S.A.G.,
- Hemker H.C.,
- Hollaar L.,
- Hermens W.T.
- Hackworthy R.A.,
- Sorenson S.G.,
- Fitzpatrick P.G.,
- et al.,
- APSAC Investigators
- Baardman T.,
- Hermens W.T.,
- Lenderink T.,
- et al.
- Willems G.M.,
- Muijtjens A.M.,
- Lambi F.H.,
- et al.
- Sobel B.E.,
- LeWinter M.M.
- Ricchiuti V.,
- Sharkey S.W.,
- Murakami M.M.,
- Voss E.M.,
- Apple F.S.
- Ingkarnisorn W.P.,
- Rhoads K.L.,
- Aletras A.H.,
- Kellman P.,
- Arai A.E.