Value of Myoglobin, Troponin T, and CK-MBmass in Ruling Out an Acute Myocardial Infarction in the Emergency Room
Background Ruling out acute myocardial infarction (AMI) on the basis of rapid assays for cardiac markers will allow early triage of patients and cost-effective use of available coronary care facilities.
Methods and Results We studied the value of myoglobin, creatine kinase (CK)–MBmass, and troponin T in ruling out an AMI in the emergency room in 309 consecutive patients presenting with chest pain. The gold standard for AMI was the combination of history, ECG, and a typical curve of the CK-MB activity (CK-MBact). Myoglobin was the earliest marker, and its negative predictive value (NPV) was significantly higher than for CK-MBmass and troponin T from 3 to 6 hours after the onset of symptoms (myoglobin versus CK-MBmass, P<.03; myoglobin versus troponin T, P<.01). The NPV of myoglobin reached 89% 4 hours after the onset of symptoms. The NPV of CK-MBmass reached 95% 7 hours after the onset of symptoms. Troponin T was not an early marker for ruling out AMI, and NPV changed over time, together with CK-MBact. The early NPV was higher in a subgroup of patients with a low probability of the presence of AMI for the three markers. Cardiac markers rise earlier in patients with large infarcts than in patients with small infarcts as indicated by the cumulative proportion of the marker above the upper reference limit at each time point (myoglobin, P=.04; CK-MBmass, P=.013; troponin T, P=.016).
Conclusions For ruling out AMI in the emergency room, myoglobin is a better marker than CK-MBmass or troponin T from 3 until 6 hours after the onset of symptoms, but the maximal NPV reaches only 89%. At 7 hours, the NPV of CK-MBmass is 95%. The test characteristics are influenced by the probability of the presence of AMI in the patients studied and by the size of their AMI. Infarct size of AMI patients should be reported in studies evaluating cardiac markers.
In many patients presenting with chest pain at the emergency room, cardiac markers in serum are measured in the early hours after the onset of symptoms not to detect AMI but rather to exclude MI. Early ruling out of AMI will allow better and more cost-effective triage and management of these patients. According to World Health Organization criteria, the diagnosis of AMI is established when there is typical chest pain suggestive of myocardial ischemia, ST-segment changes or the development of new Q waves on the ECG, and abnormal cardiac enzymes detectable in the peripheral blood. To rule out AMI, one frequently must rely on the measurement of cardiac enzymes or biochemical markers when patients suspected of AMI present with atypical chest pain or a nondiagnostic ECG.
New assays with excellent performance and short turnaround time will allow rapid detection of elevated cardiac markers. The diagnosis of AMI can be established on the basis of these assays as early as 1.5 to 3 hours after the onset of symptoms. Several biochemical markers for the early detection of myocardial damage have been proposed, of which troponin T,1 2 3 4 myoglobin,5 6 7 8 9 10 11 12 13 14 and CK-MBmass6 15 16 17 18 19 are the most promising candidates. However, the emphasis in many studies of the diagnostic properties of these markers has been on the detection rather than the ruling out of AMI. Serum levels of these markers change rapidly in the early hours after the onset of AMI; therefore, sensitivity and specificity of any particular marker change rapidly over time. Moreover, if blood samples are drawn from patients already admitted to a CCU because of suspected AMI, many patients will have large AMI. Infarct size may influence early sensitivity and specificity of the cardiac marker under study, and management decisions in these patients, such as the initiation of thrombolytic therapy, usually are not based on cardiac marker results.
We studied the value of rapid assays for myoglobin, troponin T, and CK-MBmass in ruling out AMI in a large group of consecutive patients in the first 24 hours after the onset of chest pain in the emergency room. The study was designed to focus on patients with a relatively low probability of AMI. We took hourly blood samples during the first 8 hours after the onset of symptoms and every 4 hours thereafter. We asked the attending physician to identify the patients in whom rapid marker results would influence decision making. Also, establishing the exact time of onset of symptoms received special attention.
All patients with chest pain were seen at the CER, a facility equipped with ECG monitoring where patients can remain under observation for a maximum of 24 hours. Patients with chest pain suggestive of myocardial ischemia within 12 hours after the onset of symptoms were eligible for the study, and all gave informed consent before inclusion. Exclusion criteria were severe skeletal muscle damage or trauma, cardiac resuscitation, and inability or refusal to give informed consent.
Patients with ECG abnormalities on admission that were strongly suggestive of AMI were immediately transferred to the CCU and received standard therapy, including thrombolysis, at the discretion of the physician. Patients with new episodes of chest pain accompanied by ECG changes during their stay in the CER were diagnosed as having severe unstable angina and were referred to the CCU for treatment with intravenous nitrates and heparin. Blood sampling from these patients continued at the CCU for 24 hours. All other patients remained in the CER for observation and blood sampling. The attending physician noted the time of onset of symptoms (t0), frequently after consulting the patients’ relatives, and gave an estimate of the accuracy of t0. After the initial histories, physical examinations, and ECGs were taken and before any laboratory results were reported, the physician recorded an estimate of the probability of the presence of myocardial infarction on a four-point scale.
Blood samples were drawn with an indwelling intravenous catheter at 3, 4, 5, 6, 7, 8, 12, 16, 20, and 24 hours after the onset of symptoms.
Routine chemistry and hematology were taken from the first sample. Only the CK-MBact results of samples at 6, 12, 20, and 24 hours after t0 were reported to the physician. Myoglobin, CK-MBmass, and troponin T results were not reported.
The protocol was approved by the Ethics Committee of our institution.
Diagnosis of AMI
The final diagnosis of AMI was established at hospital discharge on the basis of the patient’s clinical history and symptoms, ECG abnormalities, and typical rise and fall in the serum CK-MBact curve, with a peak exceeding 8 U/L using the results from all time points.
Blood was collected in 10-mL heparin-coated tubes and centrifuged without delay. Cells were discarded, and plasma was used to perform the myoglobin turbidimetric assay and the CK-MBact assay. The remaining plasma was stored at −20°C until further analysis.
CK-MBact was measured with an ion exchange column chromatography method as reported previously.20 In short, CK-MM and CK-MB fractions were separated on a Sephadex A-50 column. After separation, the activities were measured with a commercial kit (CPK, acetylcysteine-activated, product 124184, Boehringer Mannheim). The reaction was started by the addition of creatine phosphate, and the activities were measured at 340 nm in a spectrophotometer. The upper reference limit of CK-MBact in this assay is 8 U/L.
CK-MBmass assay was performed with the immunochemical method as implemented on the ACS-180 analyzer (CIBA Corning). The upper reference limit was 8.0 ng/mL; linearity was from 0 to 500 ng/mL. The turnaround time of the assay (the time from blood sampling until the availability of the result) was 45 minutes.
The myoglobin assay (Turbiquant myoglobin, Behringwerke) for use with the Behring Turbitimer analyzer was used for rapid immunoturbidimetric determination of myoglobin concentrations in plasma. This assay is based on polystyrene particles coated with rabbit anti-human myoglobin antibodies, which form agglutinates with myoglobin present in serum or plasma. The agglutination causes an increase in turbidity measured with a photometer. The measurement range was 50 to 650 ng/mL. The upper reference limit was 90 ng/mL. The turnaround time of the assay was 20 minutes.
Troponin T was measured with an ELISA method (Boehringer Mannheim, product 1289055) on an ES300 analyzer (Boehringer Mannheim). The upper reference limit was 0.1 ng/mL; the linearity range of this determination was 0 to 15 ng/mL. The turnaround time of the assay was 2.5 hours.
Physicians’ Estimate of AMI
The attending physicians were asked to estimate the probability of the presence of an AMI (>75%, 50% to 75%, 25% to 50%, or <25%) using the data available on admission without any marker results. They also estimated the accuracy of t0 on a three-point scale: accurate within 15 minutes, accurate within 60 minutes, or not accurate within 60 minutes.
To test the hypothesis that patients with large infarcts have an early rise of the three markers, the patients were divided into three groups according to infarct size as estimated by the CK-MBact peak. Infarct size categories were defined prospectively: small, CK-MB peak <60 U/L; medium, CK-MB peak between 61 and 120 U/L; and large, CK-MB peak >120 U/L.
Differences between the NPVs of the three markers at 3, 4, 5, and 6 hours after the onset of symptoms were calculated with the χ2 test for differences in proportions. For the cumulative proportion of patients with a marker above the upper reference limit, the marker was considered to be normal or abnormal for each patient. The time until the marker rose above the upper reference limit was calculated. Differences between the cumulative proportion of patients with a marker above the upper reference limit were then tested with the generalized Wilcoxon rank-sum test for time failure data. Differences between the cumulative proportions of abnormal markers comparing patients with small, medium, and large infarcts were calculated with the generalized Mann-Whitney U test for time failure data. A value of P<.05 was considered statistically significant.
In total, 317 patients were included in the study. Four patients were excluded because of an inability to draw blood from the indwelling catheter. Because of incorrect or incomplete data collection, 4 other patients were excluded. Therefore, 309 patients formed the study group, with 89% of the patients admitted within 6 hours after the onset of symptoms. The median time interval between onset of chest pain and admission was 135 minutes.
Table 1⇓ gives the patient characteristics. Three patients had impaired renal function with plasma creatinine levels >120 μmol/L. Of these, 2 had AMI, and the myoglobin curve showed a rise above the reference limit and subsequent fall to normal values. One patient with impaired renal function had no AMI, and the myoglobin levels in this patient were normal. These 3 patients were therefore not excluded from analysis.
Acute Myocardial Infarction
A final diagnosis of AMI was present in 163 patients. Table 2⇓ summarizes the sizes and locations of the MIs. CK-MBact peak value was 30±15 U/L (mean±SD) in the subgroup with small AMI, 82±21 U/L in the subgroup with medium AMI, and 211±78 U/L in the subgroup with large AMI.
Fifty-four patients were diagnosed as having severe unstable angina (Braunwald class IIIB) and transferred to the CCU. Twenty-four patients in this group (44%) had at least one troponin T sample >0.1 ng/mL at some time point during the first 24 hours.
Physician’s Estimate of AMI and Time of Onset of Symptoms
Table 3⇓ shows the estimate of the attending physician of the probability of the presence of AMI on admission compared with the final diagnosis of AMI. In 102 patients, the probability of AMI being present was estimated to be >75%. Ninety-five of these patients were finally diagnosed as having AMI. In the remaining 207 patients with intermediate or low probability of AMI, early marker results may have been helpful in decision making.
In reconstructing the exact time of t0, the attending physician sometimes interviewed the patients’ relatives. The t0 was estimated to be accurate within 15 minutes in 189 patients and between 15 and 60 minutes in 103 patients. In 17 patients, t0 could not be determined accurately within 1 hour.
For every time point, the sensitivity, specificity, PPV, and NPV of a single sample was calculated for each marker (Table 4⇓). Specificity and PPV of the CK-MBact do not reach 100%, although the gold standard for AMI was based on the CK-MBact results. In the gold standard for AMI, however, the abnormal CK-MBact result was defined as a typical rise and fall of the whole CK-MBact curve, and the present analysis is based on single time points. Some false-positive CK-MBact results may have resulted from assay error or the presence of macro-CK. Also, the CK-MBact level in some patients with small AMI was already falling below 9 U/L at 20 and 24 hours after the onset of chest pain. This caused the sensitivity to fall from 96 at 16 hours to 92 at 24 hours and the NPV to fall from 94 to 86, respectively.
Table 5⇓ shows the sensitivity, specificity, NPV, and PPV of the subgroup of 207 patients who were estimated by the attending physician to have a probability of AMI of <75% on admission.
The NPV of each marker is plotted against time after the onset of symptoms in Fig 1⇓. Each point represents the probability of AMI not being present in the presence of a single test result below the upper reference limit at that time point. The NPV of myoglobin was significantly different from the other markers at 3, 4, 5, and 6 hours after the onset of symptoms (myoglobin versus CK-MBact, P<.002; myoglobin versus CK-MBmass, P<.03; myoglobin versus troponin T, P<.01). The NPV of CK-MBmass was significantly different from CK-MBact at 4 and 6 hours after the onset of symptoms (P<.02) and from troponin T at 6 hours after the onset of symptoms (P<.01). There was no difference in the NPV of CK-MBact and troponin T at 3, 4, 5, and 6 hours after the onset of symptoms. Fig 2⇓ gives the NPV in the subgroup of patients with a low probability of the presence of AMI. As the prevalence of AMI decreases in this subgroup, the NPV in the early hours is increased for all markers. The NPV of myoglobin was significantly different from CK-MBact at 4 and 5 hours after the onset of symptoms (P<.008). The NPV of myoglobin was significantly different from troponin T at 6 hours after the onset of symptoms (P<.03). There was no significant difference in the NPVs at all other time points in this subgroup.
Cumulative Proportions of Markers Above the Upper Reference Limit
Because serial samples were taken of all patients, it was possible to calculate at what time point the markers rose above the upper reference limit. Fig 3⇓ shows the cumulative proportion of patients with AMI with a sample above the upper reference limit, comparing CK-MBmass, myoglobin, and troponin T. The cumulative proportion of patients with a myoglobin result above the upper reference limit was significantly higher in the early hours than for CK-MBmass and troponin T (P<.0001). At 6 hours, 97% of patients with AMI had a myoglobin result above the upper reference limit. The cumulative proportion of CK-MBmass was significantly earlier than troponin T (P<.0001).
Fig 4⇓ shows for each marker the cumulative proportion of patients with a sample above the upper reference limit, with the patients divided into subgroups with small, medium, and large infarcts. The cumulative proportion of patients with an abnormal myoglobin, CK-MBmass, and troponin T in the subgroup with small infarcts was significantly lower than in the subgroup with large infarcts (myoglobin, P=.04; CK-MBmass, P=.013; troponin T, P=.016).
It is estimated that in one third of patients AMI is not recognized by the physician or the patient because chest pain is atypical or absent.21 22 The ECG can be misleading in 8% of all patients with AMI and indeterminate in another 12%.23 As many as 50% of patients with AMI may initially arrive at emergency rooms with nondiagnostic ECGs.24 To rule out AMI on the basis of cardiac markers, some time must elapse after the onset of symptoms. This time is needed for the marker to rise above the upper reference limit of the assay used and the turnaround time of the particular assay. In hospitals with “normal” emergency room facilities, this means that these ruled-out AMI patients will be admitted to the CCU and occupy up to 30% to 50% of the available CCU beds.25 Ruling out AMI early on the basis of cardiac markers may prevent unnecessary CCU admissions.
We demonstrated with frequent serial sampling that the NPVs of myoglobin, CK-MBmass, and troponin T change rapidly in the early hours after the onset of symptoms. Several studies report myoglobin to be a useful marker for MI on admission.7 9 14 16 With a turnaround time of 20 minutes, AMI can be ruled out on a single myoglobin sample with 89% certainty at 5 hours after t0 (Fig 1⇑). However, the NPV of myoglobin declines rapidly after 7 hours in contrast to the other markers with slower appearance and clearance kinetics. This is explained by the return to normal values of myoglobin after 7 hours. The NPV of CK-MBmass increased earlier than that of CK-MBact and reached 95% at 7 hours. The turnaround time of this assay was 45 minutes; therefore, AMI can be ruled out with 95% certainty with CK-MBmass 8 hours after t0. The NPV of troponin T increased relatively late, comparable to that of CK-MBact, reaching 92% 12 hours after the onset of symptoms. An assay with a long turnaround time will delay this even further. Therefore, we conclude that troponin T appears not to be a suitable early marker for ruling out AMI. The specificity of troponin T was <90% after 6 hours owing to 24 patients with unstable angina who had elevated troponin T, as was described in other reports.3 17 26 Although these patients may be at increased risk of subsequent cardiac events, it currently is unclear whether a change in management of these patients from current clinical practice will improve their prognoses.
Table 3⇑ shows that of the 102 patients in whom the attending physician estimated the probability of the presence of AMI as >75%, 95 patients (93%) were confirmed as having AMI. We analyzed the subgroup of 207 patients with a relatively low probability of AMI in whom cardiac markers could have influenced patient management. The attending physician estimated the probability of the presence of AMI as <75%. The NPV in this subgroup was higher in the early hours than in the total patient group for all three markers (Fig 2⇑). This occurs because patients with obvious (larger) AMI and their early (false-negative) samples were excluded, and the overall prevalence of AMI decreased. Thus, the NPV of myoglobin reached 94% at 5 hours and the NPV of CK-MBmass reached 94% at 6 hours in this subgroup. Again, the NPVs of troponin T and CK-MBact are comparable in this subgroup.
The cumulative proportion of patients with AMI with a sample above the upper reference limit at each time point illustrates the early rise of myoglobin in AMI patients. The cumulative proportion in the analysis for the subgroups with small, medium, and large AMI clearly demonstrates that CK-MBmass and troponin T rise later in patients with small infarcts, the patients most likely to be retained in the emergency room for ruling out AMI (Fig 4⇑). This means that the size of the infarcts of the patients studied will influence the sensitivity and specificity in the first 8 hours after the onset of symptoms for CK-MBmass and troponin T, while this effect is less obvious for myoglobin.
As reported by Mair et al,15 CK-MBmass increases 1 hour earlier than the total CK or CK-MBact, depending on the upper reference limit of the assay. In that study, however, the size of the infarcts was not reported. Because 35 of 36 AMI patients were eligible for thrombolytic therapy and CK-MBmass values returned to normal a median of 74 hours after the onset of symptoms, these patients probably suffered large infarcts. Their data on the cumulative proportion of first abnormal values show higher proportions at 4 to 7 hours after the onset of symptoms than our data, but the more rapid increase of the CK-MBmass as a result of thrombolytic therapy may explain this difference. Bakker et al19 reported on the value of the CK-MBmass in the early diagnosis of AMI, but they divided the patients into only two groups: admission within 4 hours or between 4 and 12 hours after the onset of chest pain. Our data demonstrate that in the interval between 4 and 12 hours the sensitivity of CK-MBmass ranges from 63% to 99% and the NPV from 69% to 99%, depending on the exact time interval from the onset of symptoms. In the study of Bakker et al,19 infarct size was not reported, but 75 of 154 patients had Q-wave AMI, and 50 received thrombolytic therapy. Katus et al2 and Ravkilde et al27 reported 100% sensitivity of troponin T for the detection of AMI and 28% and 69% specificity. In those studies, plasma was sampled on admission and at 6-hour intervals. Katus et al2 did not report infarct size. In the study of Ravkilde et al,27 average peak CK-MBact was 118 U/L, ranging from 12 to 528 U/L. Therefore, their reported sensitivity for troponin T may have been overestimated by late admission or a predominance of patients with large infarcts.
It is important to take the onset of symptoms as the reference point in time for the evaluation of the rapid changes in diagnostic value of biochemical markers in the early hours. Puleo et al25 showed that patients with MI reach the hospital sooner than patients without MI. This can lead to potential bias if time of admission is taken as the reference time point. To the best of our knowledge, this is the first study that explicitly estimated the accuracy of the reported time of onset of symptoms. The majority of the reported time points in our study are estimated to be accurate within 1 hour.
To decide the clinical usefulness of these markers in ruling out AMI, the turnaround time of the assay should be taken into account.
Our data indicate that for the ruling out AMI in the emergency room in the early hours after the onset of symptoms, myoglobin was a better marker than CK-MBmass and troponin T. The NPV of myoglobin was significantly better that those of the other markers 3, 4, 5, and 6 hours after the onset of symptoms. Maximum NPV of myoglobin reached 89% to 94%, depending on the patient population studied. After 7 hours, however, the NPV of CK-MBmass already reached 95%.
For all markers the early NPV was higher in the subgroup of patients with a low probability of the presence of AMI, but the differences between markers were smaller.
All three markers increased earlier in patients with large infarcts, and differences in reported values of sensitivity and specificity in the literature may be explained by differences in infarct size of the patients studied and the time of early sampling relative to t0.
Selected Abbreviations and Acronyms
|AMI||=||acute myocardial infarction|
|CCU||=||coronary care unit|
|CER||=||cardiac emergency room|
|NPV||=||negative predictive value|
|PPV||=||positive predictive value|
This study was supported by a grant from the Dutch Heart Foundation (grant No. 90-290). Boehringer Mannheim, Behringwerke, and CIBA Corning kindly supplied the reagents used in this study. We thank the physicians and nursing staff of the CCU and CER of the Academic Medical Center for their enthusiasm and efforts to include patients in this study. Special appreciation is due Els van Dongen for secretarial work and Marriette Verstappen and Jan van Straalen for their contributions to data processing and performance of the marker assays.
- Received May 22, 1995.
- Revision received July 17, 1995.
- Accepted August 3, 1995.
- Copyright © 1995 by American Heart Association
Mair J, Artner-Dworzak E, Lechleitner P, Smidt J, Wagner I, Dienstl F, Puschendorf B. Cardiac troponin T in diagnosis of acute myocardial infarction. Clin Chem. 1991;37:845-852.
Katus HA, Remppis A, Neumann FJ, Scheffold T, Diederich KW, Vinar G, Noe A, Matern G, Keubler W. Diagnostic efficiency of troponin T measurements in acute myocardial infarction. Circulation. 1991;83:902-912.
Gerhardt W, Katus HA, Ravkilde J, Hamm CW. S-troponin-T as a marker of ischemic myocardial injury. Clin Chem. 1992;38:1194-1195.
Bakker AJ, Koelemay MJW, Gorgels JPMC, Van Vlies B, Smits R, Tijssen JGP, Haagen FDM. Troponin T and myoglobin at admission: value of early diagnosis of acute myocardial infarction. Eur Heart J. 1994;15:45-53.
Adams JE, Abendschein DR, Jaffe AS. Biochemical markers of myocardial injury. Circulation. 1993;88:750-763.
Ellis AK. Serum protein measurements and the diagnosis of acute myocardial infarction. Circulation. 1991;83:1107-1109.
Ohman EM, Casey C, Bengtson JR, Pryor D, Tormey W, Horgan JH. Early detection of acute myocardial infarction: additional diagnostic information from serum concentrations of myoglobin in patients without ST elevations. Br Heart J. 1990;63:335-338.
Stone MJ, Waterman MR, Harimoto D, Murray G, Willson N, Platt MR, Blomquist G, Willerson JT. Serum myoglobin level as diagnostic test in patients with acute myocardial infarction. Br Heart J. 1977;39:375-380.
Mair J, Artner-Dworzak E, Lechleitner P, Morass B, Smidt J, Wagner I, Dienstl F, Puschendorf B. Early diagnosis of acute myocardial infarction by a newly developed rapid immunoturbidimetric assay for myoglobin. Br Heart J. 1992;68:462-468.
Vrenna L, Castaldo AM, Castaldo P, Giardiello D, Di Giacomo C, Esposito LP, Romano F. Comparison between nephelometric and RIA methods for serum myoglobin, and efficiency of myoglobin assay for early diagnosis of myocardial infarction. Clin Chem. 1992;38:789-790.
Mair J, Smidt J, Artner-Dworzak E, Lechleitner P, Dienstl F, Puschendorf B. Rapid diagnosis of myocardial infarction by immunoturbidimetric myoglobin measurement. Lancet. 1991;337:1343-1344.
Ohman EM, Casey C, Bengston JR, Pryor D, Tormey W, Horgan JH. Early detection of acute myocardial infarction: additional diagnostic information from serum concentrations of myoglobin in patients without ST elevation. Br Heart J. 1990;63:335-338.
Gerhardt W, Katus H, Ravkilde J, Hamm C, Jorgensen PJ, Peheim E, Ljungdahl L, Lofdahl P. S-troponin T in suspected ischemic myocardial injury compared with mass and catalytic concentrations of S-creatine kinase isoenzyme MB. Clin Chem. 1991;37:1405-1411.
Van Blerk M, Maes V, Huygens L, Derde M-P, Meert R, Gorus FK. analytical and clinical evaluation of creatine kinase MB mass assay by IMx: comparison with MB isoenzyme activity and serum myoglobin for early diagnosis of myocardial infarction. Clin Chem. 1992;38:2380-2386.
Bakker AJ, Gorgels JPMC, Van Vlies B, Koelemay MJW, Smits R, Tijssen JGP, Haagen FDM. Contribution of creatine kinase MB mass concentration at admission to early diagnosis of acute myocardial infarction. Br Heart J. 1994;72:112-118.
Kannel WB. Prevalence and clinical aspects of unrecognized myocardial infarction and sudden unexpected death. Circulation. 1987;75(suppl II):II-4-II-5.
Grimm RH, Tillinghast S, Daniels K, Neaton JD, Mascioli S, Crow R, Prizker M, Prineas RJ. Unrecognized myocardial infarction: experience in the Multiple Risk Factor Intervention trial (MRFIT). Circulation. 1987;75(suppl II):II-6-II-7.
Turi ZG, Rutherford JD, Roberts R, Muller JE, Jaffe AS, Rude RE, Parker C, Raabe DS, Stone PH, Hartwell TO, Lewis SE, Parkey RW, Gold HK, Robertson RL, Sobel BE, Willerson JT, Braunwald E. Electrocardiographic, enzymatic and scintigraphic criteria of acute myocardial infarction as determined from a study of 726 patients (a MILIS study). Am J Cardiol. 1985;55:1463-1468.
Ravkilde J, Horder M, Gerhardt W, Ljungdahl L, Petterson T, Tryding N, Möller BH, Hamfelt A, Graven T, Asberg A, Helin M, Penttila I, Thygesen K. Diagnostic performance and prognostic value of serum troponin T in suspected acute myocardial infarction. Scand J Clin Lab Invest. 1993;53:677-685.