Author + information
- Received October 4, 2006
- Revision received March 6, 2007
- Accepted April 3, 2007
- Published online September 11, 2007.
- Thomas N. Martin, MD⁎,⁎ (, )
- Bjoern A. Groenning, MD⁎,
- Heather M. Murray, MSc†,
- Tracey Steedman, BSc⁎,
- John E. Foster, PhD⁎,
- Alex T. Elliot, PhD⁎,
- Henry J. Dargie, MD⁎,
- Ronald H. Selvester, MD‡,1,
- Olle Pahlm, MD, PhD§ and
- Galen S. Wagner, MD∥,2
- ↵⁎Reprint requests and correspondence:
Dr. Thomas N. Martin, Western Infirmary, Dumbarton Road, Glasgow, Strathclyde G206DW, Scotland.
Objectives The purpose of this study was to validate existing 12-lead electrocardiographic (ECG) ST-segment elevation myocardial infarction (STEMI) criteria in the diagnosis of acute myocardial infarction (AMI) and the application of similar ST-segment depression (STEMI-equivalent) criteria with contrast-enhanced cardiac magnetic resonance imaging (ceMRI) as the diagnostic gold standard.
Background The admission ECG is the cornerstone in the diagnosis of AMI, and ceMRI is a new diagnostic gold standard that can be used to validate existing and novel 12-lead ECG criteria.
Methods One hundred fifty-one consecutive patients with their first hospital admission for chest pain underwent ceMRI. The 116 patients without ECG confounding factors were included in this study, and AMI was confirmed in 58 (50%). The admission ECG was evaluated on the basis of the lead distribution of ST-segment deviation according to current American College of Cardiology/European Society of Cardiology (ACC/ESC) guidelines.
Results A sensitivity of 50% and specificity of 97% for AMI were achieved with the currently applied ST-segment elevation criteria. Consideration of ST-segment depression in addition to elevation increased sensitivity for detection of AMI from 50% to 84% (p < 0.0001) but only decreased specificity from 97% to 93% (p = 0.50). There were no significant differences in AMI location or size between patients meeting the 12-lead ACC/ESC ST-segment elevation criteria and those only meeting the ST-segment depression criteria.
Conclusions In patients admitted to hospital with possible AMI, the consideration of both ST-segment elevation and depression in the standard 12 lead-ECG recording significantly increases the sensitivity for the detection of AMI with only a slight decrease in the specificity.
It is important to achieve a rapid and accurate diagnosis regarding acute myocardial infarction (AMI) in patients with symptoms suggestive of an acute coronary syndrome (ACS), and the initial electrocardiogram (ECG) is the cornerstone of this decision-making process.
Troponin T or I are very sensitive markers for AMI and are now part of routine clinical practice in the diagnosis of patients with symptoms suggesting ACS. However, the infarction is in progress by the time the current routine biomarkers are detectable in venous blood. In addition, the noncardiac causes for troponin release are well documented (1). Contrast-enhanced magnetic resonance imaging (ceMRI) with its high spatial resolution for clinical detection of AMI provides a unique gold standard for evaluation of more universally available diagnostic methods (2). Indeed, location, transmurality, and size of AMI can also be determined with precision and reproducibility (3).
Clinical decisions for initiating reperfusion therapy are typically based on ECG criteria developed in the GUSTO (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries) series of trials (4); slightly revised criteria have more recently been introduced by the American College of Cardiology and the European Society of Cardiology (ACC/ESC) (5). However, it is well recognized that the sensitivities of these sets of 12-lead ECG criteria are suboptimal (6,7) An example of this deficiency is the routine under-detection of acute posterolateral myocardial infarction (MI) that is the typical result of occlusion of the left circumflex coronary artery (LCx) (8). The adverse risk associated with non–ST-segment elevation myocardial infarction (NSTEMI) is well documented (9), but many trials have failed to demonstrate the benefits of thrombolysis on the basis of alternative non–ST-segment elevation criteria (10–13). With the emergence of more targeted treatments such as percutaneous coronary intervention, the potential role of the admission ECG as a triage tool is increased.
Acute transmural ischemia caused by occlusion of a major coronary artery produces an epicardial injury current that can be detected as a deviation of the ST-segment toward the involved myocardial region (14). This deviation is ST-segment elevation when a region is “viewed” by the positive pole of an ECG lead but ST-segment depression when “viewed” by the negative pole. Acute occlusion of the left anterior descending (LAD) or the right coronary artery (RCA) therefore typically causes ST-segment elevation in chest leads V1to V4or in limb leads II, aVF, and III, respectively, and therefore the resulting myocardial infarcts are termed STEMI. However, acute occlusion of the nondominant LCx typically produces only ST-segment depression in the 12 standard ECG leads, and therefore, the resulting myocardial infarcts could be termed “STEMI equivalent” (15). This ST-segment depression in the standard leads would appear as ST-segment elevation in the negative counterparts of these leads (7,16).
The primary aim of this study is to demonstrate that the currently accepted GUSTO (4) and ACC/ESC ST-segment elevation criteria (5) from the 12-lead ECG have high specificity but low sensitivity for the diagnosis of acute myocardial infarcts and to investigate the diagnostic benefits of considering STEMI-equivalent ST-segment depression criteria. The secondary aim is to determine the incidence and extent of infarcts meeting ACC/ESC biochemical marker criteria that are not detected by ceMRI.
All patients admitted to the Western Infirmary Glasgow with symptoms suggesting an ACS between August 2002 and May 2003 were considered for this study. This hospital is the primary medical center for a population of 300,000 located in a metropolitan area in a surrounding mixed suburban/rural region in Western Scotland. Patients were excluded if they: had a past medical history of chest pain; had significant comorbidity; were unable or refused to consent for the study; had contraindications to ceMRI (ferrous implants, severe claustrophobia, pregnancy, were unable to remain supine for >45 min); or were transferred from another hospital.
Thirty-five patients (23%) were excluded from this study because of confounding factors in the admission ECG. These ECG factors include diagnostic evidence of prior AMI (n = 17); left ventricular (LV) hypertrophy (n = 9); atrial fibrillation with high ventricular rate (n = 1); Wolff-Parkinson-White syndrome (n = 1); bundle branch block (n = 5); right ventricular hypertrophy (n = 1); and excessive artifacts (n = 1). The study of the remaining 116 patients complies with the Declaration of Helsinki. The Ethics Committee of the North Glasgow University Hospitals NHS Trust approved the protocol for study, and all participants gave their written informed consent.
The diagnosis of STEMI for final analysis was made according to the stated criteria on the basis of the findings of the ECG core lab. The diagnosis of NSTEMI included all those patients with evidence of myocardial infarction as indicated by the presence of delayed hyperenhancement rather than by the biomarkers. All except 4 patients had positive biomarkers and a rise and fall over 3 separate time points (admission, 12 h, and at the time of ceMRI). Of the 4 with negative biomarkers (false positive by cardiovascular magnetic resonance [CMR] for AMI) 3 had no ST-segment deviation on the admission ECG. The 4th patient met the STEMI-equivalent criteria (ST-segment depression had resolved on the repeat ECG the following day) and had a diffuse pattern of delayed hyperenhancement associated with an inferior and posterolateral hypokinetic regional wall motion abnormality; coronary angiography was not performed.
The admission ECG was quantitatively evaluated in a core laboratory by 2 of the investigators, who were blinded to all other study data. The ST-segment measurements were made at the J point to the nearest 0.05 mV, and all differences were adjudicated in conference. Regarding “contiguity,” the spatial consideration of the 6 chest leads and the spatially based orderly sequence of the 6 standard limb leads were considered (17). The ECGs were classified according to the following 3 criteria:
1. GUSTO STEMI: ≥0.1 mV ST-segment elevation in ≥2 anatomically contiguous standard limb leads (among I, II, III, aVL, and aVF) or ≥0.2 mV in ≥2 contiguous standard precordial leads (4).
2. ACC/ESC STEMI: ≥0.1 mV ST-segment elevation in ≥ 2 anatomically contiguous standard limb leads (in their orderly sequence from aVL to III, including −aVR) or chest leads V4to V6, or ≥0.2 mV in chest leads V1to V3(5).
3. STEMI equivalent: ≥0.1 mV ST-segment depression in ≥2 anatomically contiguous leads or in 1 lead that is anatomically contiguous to a lead with ST-segment elevation criteria (e.g., ST-segment elevation in lead aVL and ST-segment depression in lead III and vice versa).
All patients received a ceMRI study as soon as they were considered to be sufficiently stable; the protocol in place in the coronary care unit at the time indicated that biomarker-positive patients were not permitted to leave the unit for 48 h. The median (interquartile range [IQR]) time to scan was 50 (31 to 79) h after the onset of symptoms.
The ceMRI examination was carried out with a 1.5-T whole-body scanner (Siemens Sonata, Erlangen, Germany) with a phased-array chest coil as the receiver during breath hold and gated to the ECG. For the measurement of LV dimensions, a steady-state free-precession (SSFP) sequence was used to acquire a short-axis cinematographic (CINE) stack of the LV (field of view [FOV] = 340 mm, slice thickness = 8 mm, interslice gap = 2 mm, repetition time [TR]= 47.5 ms, echo time [TE]= 1.58 ms, flip angle = 60°).
Immediately following this, ceMRI was performed with the standard segmented gradient-echo inversion-recovery sequence as described elsewhere (18–20) for the determination of location and size of AMI. Briefly, gadolinium-DTPA-BMA (GE Healthcare, Waukesha, Wisconsin), 0.2 mmol/kg, was administered intravenously, and delayed enhancement short-axis images were recorded 10 to 20 min later (FOV = 340 mm, slice thickness = 8 mm, interslice gap = 2 mm, TE = 4.3 ms, flip angle = 30°, optimum inversion time (TI) adjusted to null normal myocardium [range 200 to 300 ms]).
Identical short-axis slice positions were used for the CINE and the ceMRI investigations. Measurement of left ventricular ejection fraction (LVEF) and LV mass were evaluated with manual planimetry on commercially available Argus software (Siemens) by an observer blinded to all other clinical data. The ceMRI data were analyzed with CMR Tools (Imperial College, London). Regions of AMI by ceMRI were defined as those with hyperenhancement involving at least the subendocardium. The signal intensities of gadolinium-enhanced myocardium and normal myocardium were measured with computer-assisted planimetry and delineated manually (21).
The AMI location was determined with the 12-segment method described by Pahlm et al. (22) and Ideker et al. (23,24). The location was defined as the region of the LV containing the highest percent infarction; patients were classified as having multiple infarcts if there were ≥2 separate areas of delayed enhancement.
Transmurality was considered present when at least 1 segment had hyperenhancement extending across the entire myocardial wall (19). Microvascular obstruction (MVO) was considered present when at least 1 segment had a central core of hypoenhancement at least 10 min after contrast injection (25).
Blood sampling was performed for biomarkers (creatine kinase [CK], CK-myocardial band, troponin I, and C-reactive protein) on admission and 8 to 12 h later.
Statistical analyses were performed with SAS for Windows (version 8.2, Cary, North Carolina). Baseline demographic data and clinical characteristics for patients with or without AMI detected by ceMRI were compared with the 2-sample ttest (or Wilcoxon rank sum test for skewed measurements) for continuous variables and the chi-square test for categorical variables (or Fisher Exact test if expected counts <5). Graphical procedures and the relation of the SD to the mean were used to check whether the data were normally distributed. In each case the value for the SD was either greater or close to the mean value. The sensitivities and specificities of the GUSTO STEMI, ACC/ESC STEMI, and STEMI-equivalent criteria were compared with the ceMRI gold standard using the McNemar test. The 95% confidence intervals (CI) were calculated for each statistic from the binomial distribution. The McNemar test was applied to examine the observed differences in sensitivity and specificity between the STEMI and STEMI-equivalent methods for both the GUSTO and ACC/ESC criteria. Exact p values are reported. Summary statistics are presented as mean (SD) or median (IQR) for continuous variables and number (%) for categorical variables. The level of significance was set at p < 0.05.
The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
Table 1presents the baseline demographic and clinical characteristics of the study population divided into AMI (i.e., ceMRI positive for delayed enhancement; n = 58) and non-AMI (n = 58) groups. Patients in the AMI group were older (mean 59 vs. 53 years, p = 0.0077), and there was a shorter duration of time from the onset of chest pain to the ECG (median 2.7 vs. 7.4 h, p = 0.02). There was a longer interval until the initial ceMRI study (median 64 vs. 33 h, p < 0.0001) in the AMI group. Slightly more than one-half (54%) of the AMI group and few (7%) of the non-AMI group received reperfusion therapy. Any reference to reperfusion therapy is based on the GUSTO ECG criteria as dictated by the local protocol, and all patients received thrombolysis, except 2 who underwent primary percutaneous intervention. There were 8 patients with multiple infarcts, defined by having ≥2 distinct regions of hyperenhancement.
Table 2describes the clinical characteristics of the 10 patients falsely positive by ECG and/or troponin I for AMI considering ceMRI as the gold standard. In the 8 patients falsely positive by troponin I, the peak value was 4.4 ng/ml and the mean (SD) was 1.44 (1.38) ng/ml. In the 2 falsely positive patients by ECG alone, 1 had a clinical diagnosis of pericarditis and the other had neither clinical nor angiographic evidence of cardiac disease. Eight of these 10 patients could be considered to have had a “necrosette” (i.e., an extremely small MI with a low peak troponin I) (5).
Table 3compares the performances of the GUSTO and ACC/ESC STEMI criteria. Their performances are almost identical, with only a single patient in both the AMI and non-AMI groups being classified differently. Therefore, for the remainder of the data analysis, only the more recently published STEMI criteria (ACC/ESC) are considered. Only 50% (29 of 58) of the AMI group are identified by the ACC/ESC STEMI criteria, but an additional 34% (49 of 58) are identified by the STEMI-equivalent ST-segment depression criteria (p < 0.0001) as illustrated in Figure 1.Consideration of the STEMI-equivalent criteria reduced the specificity from 97% and 93% (p = 0.50). Positive and negative predictive values are also shown and have been calculated with an infarct prevalence of 50%.
Table 4presents the characteristics of the ceMRI-positive patients stratified according to detection by the ACC/ESC STEMI criteria (n = 29) and by the STEMI-equivalent ST-segment depression criteria only (n = 20). There is markedly longer delay from time of symptom onset until the initial ECG in the STEMI-equivalent group (median 5.5 h) than in the STEMI group (median 2.2 h [p = 0.0075]). The comparative ceMRI-determined AMI sizes in the 2 ECG diagnostic groups are similar (median 31 vs. 28 g or 24% vs. 26% of total LV mass). This is illustrated in Figure 1, in contrast to the trend (p = 0.29) to smaller AMI sizes in the group of 9 patients in whom even the STEMI-equivalent criteria were not present.
The AMI sizes estimated by the various biochemical markers are also similar between the 2 ECG detection groups. There was a tendency toward lower LVEF in those detected by the STEMI criteria (54%) than in those detected by the STEMI-equivalent criteria only (59%) (p = 0.14), but it should be noted that both these values are within normal limits. However, as indicated, most (86%) in the STEMI group but few (26%) in the STEMI-equivalent group received reperfusion therapy. There was no significant difference in the presence or absence of MVO or transmurality or presence of multiple infarcts between the 2 groups. There is also no significant difference in the distribution of infarct locations between the 2 groups.
The biochemical methods for detection of AMI have become so sensitive that it is necessary to develop clinically relevant subgroups. The 1999 ESC/ACC consensus conference produced such a broad definition, on the basis of historical information and biochemical marker data, that any ECG criteria were rendered superfluous for making the final diagnosis (5). The current availability of aggressive emergency therapies for establishing myocardial reperfusion requires a method for establishing a rapid working diagnosis of thrombotic AMI at the time of initial pre-hospital or emergency department presentation. The performance of such a method can be determined by observing outcomes that range from aborted infarction to hemorrhagic complications due to inappropriate therapy (19,20,25). An immediately available clinical test with high sensitivity and specificity for thrombotic AMI is required, and neither the history nor the biochemical markers achieve this goal. The ST-segment elevation has such high specificity for AMI that the currently accepted name of this patient subgroup is “ST-segment elevation MI,” conveniently termed “STEMI.” Indeed, the multiple randomized clinical trials in patients with ACS have either STEMI or NSTEMI inclusion criteria. This study aims to highlight that, in patients with acute chest pain, ST-segment depression is often indicative of AMI, but does not aim to make any recommendations about treatment strategies.
Contrast-enhanced MRI provides a gold standard for identification, localization, and quantification of AMI in both its acutely necrotic and chronically fibrotic states (3,20,21,24,26). Use of this method to confirm the high specificity and to determine the sensitivity of the existing ECG criteria requires the study of the broad population of consecutive patients admitted to a single medical center with symptoms compatible with ACS. The present study was designed and implemented for this purpose.
Thirty-five (23%) patients were excluded on the basis of ECG confounding factors. This removal of confounders so often characterized by the early repolarization that mimics the ST-segment deviation of acute transmural ischemia led to, for example, a surprising infarction rate by ceMRI of 29 of the 31 patients (94%) meeting the ACC/ESC ST-segment elevation threshold (23). When considering the total population, infarct by ceMRI was present in 32 of the 40 (80%) patients meeting the ST-segment elevation threshold. In addition, the sensitivity rose from 44% in the ACC/ESC STEMI group to 71% in the STEMI-equivalent group with a reduction in specificity from 91% to 86%. As expected, these values demonstrate a reduced diagnostic accuracy both overall and between groups, but interestingly there is very little difference in the positive predictive values (83% vs. 82%).
The prevalence of infarction in this study population was high at 50% and provided positive predictive values for ACC/ECC STEMI and STEMI-equivalent criteria of 94% and 92%. Considering the lower AMI prevalence of 20% that is more typical of the population presenting to the emergency room, the PPV falls to 85% and 75%, respectively. This extrapolation requires further investigation, if the results are to be applied more widely.
Eight patients were considered to have had multiple infarcts, 3 of whom had 3-vessel disease (1 with critical left main stem stenosis); 1 had proximal and distal lesions in a dominant RCA; 1 had distal left anterior descending disease; 1 had normal coronaries; and for the final 2 patients coronary angiography was not clinically indicated. The distribution of STEMI/NSTEMI was equal, and all had a rise and fall in cardiac enzymes in keeping with AMI. It is possible that the patients with 3-vessel disease had multiple “hot plaques,” but it is more likely that there were chronic infarcts. Because there were no well-validated ceMRI sequences for differentiating age of infarct at the time of recruitment, absence of either historical or ECG suggestion of infarct was required for study inclusion together with a rise and fall of troponin and/or CK-myocardial band.
Consideration of presenting STEMI or STEMI-equivalent criteria as “falsely positive” would be in error when the prompt administration of reperfusion therapy sufficiently aborts the AMI so that it is not detected by ceMRI. Fourteen patients presented to hospital within 1 h; 8 had an AMI by both ECG and ceMRI. None of the remaining 6 had significant ST-segment deviation, and the largest initial CK value was 134. Aborted AMI is therefore unlikely to be a significant confounder in this study.
Contrast-enhanced MR has been documented to achieve diagnosis via delayed enhancement only when more than approximately 1 cm3is infarcted (27), illustrating the current limitations due to spatial resolution. Despite this, it is still vastly superior to any other available noninvasive imaging technique. There were no patients with a troponin I >4.4 ng/ml who did not have delayed hyperenhancement, and the mean troponin I for these 8 patients was 1.44 (Table 4). This might indicate the upper limit of troponin that suggests a “necrosette,” and further studies on the significance of this change range using ceMRI in this lower-risk group might help in future risk stratification (28).
The similarity in infarct size by ceMRI between the STEMI and STEMI-equivalent groups highlights that significant infarcts are being missed by the ECG criteria used in typical clinical practice. It should be noted, however, that the naturally occurring AMI sizes of many of the 31 of 58 in the ceMRI-positive group who received reperfusion therapy would have been higher. This is further illustrated by the significantly lower time to presentation in the STEMI group (difference in median 3.3 h), suggesting that this group had more severe symptoms. Myocardial infarcts that are missed altogether by the ECG are smaller, although the difference does not reach statistical significance (p = 0.29, Kruskal-Wallis test) (Fig. 1).
The additional diagnostic information provided by the STEMI-equivalent criteria resulted in increased detection in the anterior location from 38% (5 of 13) to 69% (9 of 13), in the inferior location from 70% (14 of 20) to 95% (19 of 20), in the posterolateral location from 36% (4 of 11) to 82% (9 of 11), and in the inferior and posterolateral location from 17% (1 of 6) to 67% (4 of 6) (data from Tables 2 and 4).
In the present study, when maximal ST-segment deviation appeared as depression in V1, V2, or V3, 7 of the 9 patients had ceMRI-determined infarction in the posterolateral region of the LV; the remaining 2 had 3-vessel disease. This maximal ST-segment deviation direction was falsely positive for infarction in none of the individuals. The basal region of the LV is superiorly oriented in the thorax, directly opposite the inferior or diaphragmatic region. When ST-segment depression in “inferior leads” accompanies ST-segment elevation maximal in V2to V4, extension of the anterior infarction into the basal region has been documented, due to occlusion of the proximal LAD (29). In the present study, when maximal ST-segment deviation appeared as depression in II, III, and aVF, 4 of the 5 patients had ceMRI-documented infarction in the basal region of the LV. This ST-segment deviation direction was only falsely positive in 1 individual with a body mass index of 33 who subsequently had a normal coronary angiogram.
The term “ST-segment deviation MI” could be adapted with designation of the direction of the lead with the maximum either ST-segment elevation or ST-segment depression. The negative leads are as “real” as the positive leads, and indeed lead “−aVR” replaced +aVR in Sweden 30 years ago (30). With either alternative, the identification of the spatial direction of the ST-segment as indicated by the lead with the maximal deviation would be required.
This study has documented that in patients admitted to hospital with chest pain and no historical or ECG evidence for previous AMI only 50% of the AMI group are identified by the ACC/ESC STEMI criteria, but an additional 34% are identified by the STEMI-equivalent ST-segment depression criteria with ceMRI as the gold standard.
Future studies of the contribution of an increase in the number of leads and/or electrodes, T-wave changes, and ST-segment changes appearing in serial ECG recordings will be required. Addition of selected leads such as V4R, V7, or V8(31) or multi-lead body surface maps (32) have been shown to increase sensitivity. However, the ability to increase sensitivity via additional electrodes must be confirmed when “ST-segment deviation” replaces “ST-segment elevation” in AMI criteria for the standard 12-lead ECG.
This study is limited to the population of patients with their first episode of chest pain, who have been deemed to be at enough risk to merit hospital admission to diagnose AMI and who do not have ECG confounding factors; so it is not clear how these results would apply to the broader, less pure population.
- Abbreviations and Acronyms
- American College of Cardiology
- acute coronary syndrome
- acute myocardial infarction
- contrast-enhanced magnetic resonance imaging
- European Society of Cardiology
- left anterior descending coronary artery
- left circumflex coronary artery
- left ventricle/ventricular
- left ventricular ejection fraction
- microvascular obstruction
- non–ST-segment elevation myocardial infarction
- right coronary artery
- ST-segment elevation myocardial infarction
- Received October 4, 2006.
- Revision received March 6, 2007.
- Accepted April 3, 2007.
- American College of Cardiology Foundation
- Mahrholdt H.,
- Wagner A.,
- Holly T.A.,
- et al.
- The Joint European Society of Cardiology/American College of Cardiology Committee
- Freeman M.R.,
- Langer A.,
- Wilson R.F.,
- Morgan C.D.,
- Armstrong P.W.
- TIMI IIIB Investigators
- Kim R.J.,
- Chen E.L.,
- Lima J.A.,
- Judd R.M.
- Kim R.J.,
- Fieno D.S.,
- Parrish T.B.,
- et al.
- Kim R.J.,
- Fieno D.S.,
- Parrish T.B.,
- et al.
- Gerber B.L.,
- Rochitte C.E.,
- Melin J.A.,
- et al.
- Kwong R.Y.,
- Schussheim A.E.,
- Rekhraj S.,
- et al.
- Ricciardi M.J.,
- Wu E.,
- Davidson C.J.,
- et al.
- Diderholm E.,
- Andren B.,
- Frostfeldt G.,
- et al.