Author + information
- Received August 7, 2002
- Revision received October 31, 2002
- Accepted November 11, 2002
- Published online March 19, 2003.
- Lene Holmvang, MD*,* (, )
- Peter Clemmensen, MD, PhD*,
- Bertil Lindahl, MD, PhD†,
- Bo Lagerqvist, MD†,
- Per Venge, MD†,
- Galen Wagner, MD‡,
- Lars Wallentin, MD, PhD, FACC† and
- Peer Grande, MD, PhD*
- ↵*Reprint requests and correspondence:
Dr. Lene Holmvang, The Heart Center 2141, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen Ø Denmark.
Objectives The aim of the present study was to evaluate whether the effect of an early invasive treatment strategy differed between patients sub-grouped according to their severity of myocardial ischemia, as evaluated by quantitative electrocardiographic (ECG) analysis at the time of presentation. The present study is a sub-study of the previously published Fast Revascularization during InStability in Coronary artery disease trial (FRISC-II).
Background An early invasive treatment strategy has been shown to be the preferable treatment for non–ST-segment elevation acute coronary syndromes (ACS). The population of patients with unstable coronary artery disease is heterogeneous regarding both the underlying pathology and prognosis. Early risk stratification is important to select patient subgroups that will benefit the most from a given treatment.
Methods In 2,201 patients with non–ST-segment elevation ACS, the ischemic burden at hospital admission was determined by quantitative measurements of ST-T-segment deviations on the ECG. The patients were subsequently sub-grouped in tertiles based on the amount of ST-segment deviation. The primary end point for this analysis was death or myocardial infarction (MI) within one year after study inclusion.
Results The invasive treatment strategy produced a reduction of ∼50% in death or MI among the patients with intermediate or major ST-segment deviation. The findings were independent of age, gender, or troponin T status. The patients with confounding factors precluding ST analysis had a poor outcome regardless of the treatment strategy.
Conclusions Ischemic burden on the admission ECG identifies patients with ACS who benefit the most from an invasive treatment strategy. When the standard ECG is scrutinized with complete quantitative measurements, it provides independent information on prognosis and benefit of treatment.
Patients with unstable coronary artery disease (CAD) are heterogeneous regarding both their underlying pathology and prognosis. Thus, early risk stratification is important in selecting patient subgroups that will benefit the most from a given treatment. Biochemical markers of myocardial damage, such as troponin T (TnT) and I, have been found to identify high-risk patients and even to indicate who will or will not benefit from aggressive anti-thrombotic treatment (1–3).
We and other investigators have previously found that admission electrocardiographic (ECG) variables carry similar or even better prognostic information than troponin measurements in patients with acute coronary syndromes (ACS) (4,5). In the Fast Revascularization during InStability in Coronary artery disease trial (FRISC-II) population, qualitative analysis of ECG data predicted the treatment outcome, either alone or in combination with TnT data (6,7). Although efforts have been made into finding the appropriate cut-off levels for troponin to predict prognosis and treatment success, the quantitative findings on the admission ECG have not yet been systematically investigated regarding their ability to identify treatment responders.
Accordingly, the aim of the present study was to test the hypothesis that quantitative analysis of the admission ECG can help identify patients with non–ST-segment elevation ACS who might benefit the most from an invasive treatment strategy. The hypothesis is that the patients with the highest ischemic burden, as determined by ST-segment deviation on the standard ECG, have the greatest benefit, defined as improvement in mortality and morbidity, from an aggressive treatment strategy.
The present work is a sub-study of the FRISC-II trial, a randomized study of an early invasive versus early non-invasive strategy in unstable CAD. Details on the study design have been published elsewhere (8). Briefly, patients were included in the study if they had chest pain within the last 48 h, in combination with either elevation of cardiac biomarkers or ST depression or T-wave inversion on the ECG. The patients were randomized to one of four treatment groups: invasive treatment and long-term dalteparin; invasive treatment and placebo; non-invasive treatment and long-term dalteparin; and non-invasive treatment and placebo. The design and results from the long-term dalteparin versus placebo arm of the study have been described in detail (9).
In the invasive arm, the target was to perform the revascularization procedures (where indicated) within seven days after admission to the hospital. Non-invasive treatment included coronary angiography only when a patient had refractory or recurrent symptoms or a positive exercise test before discharge.
The primary outcome of the FRISC-II trial was death or myocardial infarction (MI) at six months, but the patients were followed up for at least one year. Secondary end points were angina, re-admission to the hospital, myocardial revascularization, and bleeding. Myocardial infarction was defined by the occurrence of two of three of the classic criteria: typical chest pain, diagnostic ECG, or elevation of biochemical markers of infarction (8). When the present sub-study was carried out, one-year follow-up data were available. Because these data confirmed the findings from the six-month analysis documenting a benefit of an invasive treatment strategy, one-year follow-up data were used in this sub-study.
A 12-lead ECG was obtained at admission, at randomization, after the intervention, or at hospital discharge, and then at three-month and six-month visits, as well as when there was any suspicion of angina or MI. The admission ECGs were quantitatively analyzed in the ECG core laboratory at the Department of Medicine B, Rigshospitalet, Copenhagen University Hospital, before any other data, including end point data, were revealed. The Copenhagen ECG core laboratory was established post hoc, and ECG data originating in this core laboratory were not included in the FRISC-II main data base. The presence and amount of ST-segment deviation and T-wave amplitudes were measured in all leads except aVR, and the patients were subsequently sub-grouped in tertiles based on the amount of cumulative ST-segment deviation and number of leads with ST-segment deviation exceeding 0.5 mm on the admission ECG. A fourth subgroup included patients with confounding factors excluding reliable ST-segment analysis. Criteria used to define the confounding factors are shown in Appendix A.
At randomization, blood samples were obtained from all patients and stored for central analysis of TnT. Quantitative TnT analysis was performed centrally at the biomarker core laboratory at Uppsala University Hospital, using the Elecsys TnT third-generation assay (Roche Diagnostics, Basel, Switzerland).
The invasive strategy required coronary angiography within seven days after study inclusion. All angiograms were evaluated locally. An obstruction of at least 50% was considered significant. Revascularization was encouraged in all patients with an obstruction of 70% or more in any significant coronary artery. The study protocol recommended percutaneous coronary intervention (PCI) if there were one or two target lesions, and coronary artery bypass graft surgery (CABG) in the presence of three-vessel or left main CAD.
Data analysis was performed based on the “intention-to-treat” principle. Relative risk ratios were calculated by univariate analysis, whereas odds ratios (ORs) were determined by multivariate logistic regression analysis. For risk and ORs, 95% confidence intervals for the combined end point of death and MI after one year of follow-up, in relation to treatment strategy, were calculated. The ORs were adjusted for the clinical variables of age, gender, smoking, previous MI, diabetes, hypertension, and admission TnT level, all pre-defined as important baseline characteristics in the main study (8). The timing of end points is illustrated by Kaplan-Meier plots.
To assess a possible interaction between ST-segment deviation and treatment, multiple logistic regression analysis was used. The following terms were included: ST-segment deviation × randomized treatment, age, previous MI, diabetes, smoking, hypertension, TnT, and randomized treatment in itself. The ST-segment variable was transformed to a dichotomous variable (summed ST-segment deviation <3 mm and ≥3 mm at the 50th percentile) for a more simple interpretation.
A comparison of baseline characteristics, troponin values, and angiographic data between the three ST-segment deviation groups was done by using the Kruskall-Wallis analysis of variance median test (continuous variables) or the Pearson chi-square test (categorical variables), as appropriate. Comparisons between the ECG confounder group and any of the three ST-segment deviation groups were done by using the Mann-Whitney Utest or chi-squared test.
The STATISTICA software program was used for data processing and statistical analysis, except for the logistic regression analysis and the test for interaction, which were performed using the SPSS program (version 9.0).
The ECG data were available to the Copenhagen core laboratory in 2,238 patients participating in the invasive versus non-invasive treatment study. In 37 patients (1.6%), parts of the admission ECG were missing or the ECG was of very poor technical quality. Thus, the population consisted of 2,201 patients with evaluable ECGs.
The patients were classified into subgroups according to the total amount of ST-segment deviation (the summated deviation in 11 leads and the total number of leads with ST-segment deviation ≥0.5 mm) at admission. Tertile values were used as cut-off limits to obtain three groups of almost equal size. Finally, 504 patients (23%) had confounding factors on the ECG excluding ST-segment analysis. Table 1lists the presence of these confounding factors and the number of events occurring in this sub-population. Tables 2 and 3present the benefit of an invasive treatment strategy when the patients are sub-grouped ⇓by cumulative ST-segment deviation and by the number of leads with >0.5-mm ST-segment deviation. There was a definite benefit of the invasive treatment strategy in reducing the risk of the combined end point death or MI among the patients with intermediate or major ST-segment deviation, defined either by the total amount of deviation or the number of leads with ST-segment deviation. Even when corrected for other baseline variables (gender, age, hypertension, smoking, diabetes, previous MI, and TnT >0.03 μmol/l) in multivariate logistic regression analysis, the risk reduction obtained by the invasive strategy was not affected. The invasive strategy reduced mortality among all the subgroups except for the patients with minor ST-segment changes defined by summed deviation. The numbers were too small, however, to obtain statistical significance. The patients with confounding ECG factors that precluded ST-segment analysis had a relatively poor outcome, regardless of the treatment strategy. The timing of the combined end point of death/MI is presented as Kaplan-Meier plots (Figs. 1A to 1Dand Figs. 2A to 2C).
Interaction analysis of the variable summed ST-segment deviation × randomized treatment came out as significant for predicting death or MI at 12 months (p = 0.006) in a multivariate logistic regression analysis including randomized treatment in itself, TnT status, and the aforementioned baseline variables.
Figure 3illustrates the benefit of invasive treatment when the patients are classified into smaller subgroups according to admission ECG findings.
Patient characteristics in relation to the ECG at presentation
Table 4shows the differences in baseline characteristics when the patients are classified according to the amount of ST-segment deviation at admission. The patients with ECG confounders and with the most ST-segment deviation at admission had a higher risk profile than the remainder of the patients. Similar results were obtained when the patients were grouped according to the number of leads with ST-segment deviation instead of the total amount (data not shown).
ST-segment deviation and TnT
Randomization TnT levels rose with increasing ST-segment deviation, and both the median TnT level and the number of TnT-positive patients were higher among the patients with the most ST-segment and ECG confounders (Table 5).
Angiographic findings and invasive procedure
Coronary angiography was performed within seven days in 1,077 (98.6%) of the 1,092 patients in the invasive group. Fifteen patients (1.4%) did not undergo angiography. Table 6depicts the angiographic findings among the patients in the four subgroups. More patients in the minor ST-segment deviation group had either no significant vessel disease or one-vessel disease, whereas two- or three-vessel disease and left main stenosis were 50% more common among the patients with major ST-segment depression. The mode of invasive procedure differed significantly between the subgroups. As a consequence of the angiographic findings, the patients with minor ST-segment changes randomized to the invasive strategy were more likely to undergo PCI, whereas CABG was the favored procedure among patients with major ST-segment deviation (Table 7).
Invasive procedures and procedure-related acute MI
Because of the different definitions of procedure-related acute MI during CABG (in which only new Q waves were referred to as MI) and PTCA (in which elevation of biomarkers was sufficient for diagnosing acute MI) (8), the occurrence of procedure-related acute MI was separately registered. Between 47% and 66% of all end point acute MIs were procedure-related; however, despite more frequent use of PCI for revascularization among patients with minor ST-segment depression, the procedure-related acute MI frequency was not significantly increased among these patients.
Clinical course of patients randomized to a non-invasive strategy
Of the 1,109 patients randomized to a non-invasive strategy, 314 (28%) underwent coronary angiography within the first 30 days, because of either a positive exercise test or clinical symptoms. The patients who crossed over to an invasive strategy had significantly more ischemia on the admission ECG than the patients who continued to be managed conservatively (median ST-segment deviation 5 mm [range 2.5 to 7.5] vs. 4 mm [range 2.5 to 6.5], p = 0.001).
Non-invasive risk stratification
In patients with ACS, the amount of ischemia, as detected by ECG changes at admission, is an indicator of poor prognosis (10–12).
These previous findings were confirmed in the present study. The amount of ST-segment depression at admission was related to poor outcome among patients treated according to conventional principles. The rate of death or MI at one-year follow-up was doubled among the patients with major ST-segment deviation, compared with the patients with only minor or no ST-segment changes. ST-segment deviation was associated with a higher risk profile, higher TnT levels, and more extensive CAD determined by angiography.
Admittedly, risk and treatment stratification based solely on troponin measurement and the ECG might exclude some high-risk patients from beneficial therapies. Hence, in the Thrombolysis in Myocardial Infarction (TIMI) risk score system, age (>65 years), risk factors for CAD, known coronary artery stenosis >50%, ST-segment deviation at admission, at least two previous attacks of angina, use of aspirin within the last seven days, and elevated biomarkers have all been found to carry independent prognostic information (13). The Platelet IIb/IIIa Underpinning the Receptor for Suppression of Unstable Ischemia Trial (PURSUIT) found the most important baseline indicators of poor prognosis to be age, increased heart rate and blood pressure, ST-segment depression, signs of heart failure, and elevated cardiac markers (14). The present study confirms the findings from previous studies that ST-segment deviation on the admission ECG is an important risk indicator in patients admitted with an ACS. However, several of the other variables from the TIMI risk score seem to be associated with ST-segment deviation.
Benefit of invasive treatment
It has been shown that the mere presence of ST-segment depression at a threshold of 0.5 mm identifies patients at high risk of death on long-term follow-up after unstable angina/non–Q-wave MI (11). The present study suggests that invasive treatment is efficacious across wide ranges of extent and magnitude of ST-segment changes, but the magnitude of this effect does seem to increase with the quantity of the ST-segment deviation. Also, the extent of CAD increases substantially with an increasing amount of ST-segment deviation (i.e., patients with major ST-segment deviation had a threefold higher incidence of left main stenosis than the patients with only minor ST-segment changes).
We have previously shown that the inability to perform analysis of the ECG at discharge is associated with a poor outcome after an episode of unstable CAD (15). Because disqualifying ECG changes, such as bundle branch block and hypertrophy, will usually be present both at admission and discharge, the results of the present study are not surprising. The patients with ECG confounders resemble the patients with major ST-segment depression in terms of their TnT values, angiographic findings, and choice of revascularization procedure. Despite the fact that these “high-risk” patients seemed to benefit somewhat less from an invasive treatment strategy than did the patients with major ST-segment deviation, judged by the combined end point, mortality alone was reduced by 42% by the invasive strategy. The ECG confounders are probably indicators of several severe underlying conditions like advanced age, previous acute MI, and heart failure, leading to the same high risk as in patients with major ST-segment depression with similar TnT values.
The explanation for the apparent smaller benefit from invasive treatment among the patients with minor ST-segment deviation could hypothetically be sought in the pre-defined primary end point criteria. Peri-procedural MI was diagnosed based primarily on biomarker elevation after PCI, whereas the diagnosis of peri-operative MI during CABG required new Q-waves on the ECG. Because CABG was more common among the patients with major ST-segment depression (due to more severe CAD), the “risk” of a peri-procedural MI could paradoxically be lower. However, by counting the rate of procedure-related acute MI versus “spontaneous” acute MI, this was not the case. Thus, the higher rate of end points among the invasively treated patients in the minor ST-segment depression group could not be explained by a higher rate of PCI-related infarcts. Controversies still exist about the clinical relevance of PCI-related infarcts (16–18). A recent analysis of data from PURSUIT revealed that peri-procedural infarcts were associated with the same prognosis as spontaneous infarcts, as long as the infarcts were of similar size, as determined by biomarker levels (19). These findings were confirmed in the Treat Angina with Aggrastat and determine Cost of Therapy with an Invasive or Conservative Strategy–Thrombolysis In Myocardial Infarction-18 (TACTICS–TIMI-18) study, where larger infarcts (creatine kinase, MB isoenzyme >3× the upper limit of normal [ULN]) had the same prognosis, regardless of origin, whereas the smaller spontaneous infarcts (1 to 3× ULN) had a 10-fold increase in mortality within six months, compared with PCI-related infarcts of similar size (20). Whether small PTCA-related infarcts have a similar or better prognosis than conventional Q-wave infarcts associated with CABG remains to be shown.
Compared with the recently published TACTICS–TIMI-18 (21), in FRISC-II, only very few patients received treatment with a glycoprotein IIb/IIIa inhibitor outside the catheterization laboratory. The very early invasive treatment strategy used in TACTICS produced the lowest event rate ever in a trial of non–ST-segment elevation ACS. The results of previous trials of glycoprotein IIb/IIIa inhibitors (22), in combination with the low event rates in TACTICS, should lead to a recommendation of glycoprotein inhibitors for all patients with non–ST-segment elevation ACS. Whether this treatment would have further increased the benefit of the invasive treatment strategy in FRISC-II is speculative, however.
In the TACTICS–TIMI-18 trial, ST-segment changes on the standard ECG predicted a benefit from the early aggressive treatment strategy, and so did TnT. When stratified only according to the presence or absence of ST-segment changes at baseline, the patients without ST-segment changes had no benefit of an early invasive treatment strategy. Similarly, subgroup analysis showed that the 52% of patients in FRISC-II without ST-segment changes at baseline also had similar outcomes with the use of either strategy (8).
Within the last years of the past century, “the troponin hypothesis” gained support based on many trials on the ACS, although the potential information from standard ECG analysis has not been explored as extensively. Numerous studies have tried to determine the best cut-off levels for the various biochemical markers to identify high-risk patients with ACS. To do justice to the ECG, similar efforts must be put forth into identifying the ECG variables containing the most information.
Previous studies on the prognostic value of ECG analysis have mostly concentrated on ST-segment deviation analysis, and the results on prognostic information of isolated T-wave inversion are conflicting. A future sub-study on the FRISC-II population will hopefully reveal whether additional prognostic information can be extracted from careful analysis of the T wave.
The present study is based on a post hoc analysis of data from the FRISC-II trial. Subgroup analyses may be underpowered; hence, the results of post hoc analysis should be handled with care. The troponin cut-off level at 0.03 to 0.1 μg/l and the ST-segment thresholds were established based on a review of the data not tested prospectively. Also, the different rates of CABG and PCI in the subgroups are potential confounding factors that could have influenced the occurrence of end points. Even though the results of the present study indicate that an invasive treatment strategy is beneficial only in patients with a certain extent of ST-segment deviation on their admission ECG, such a conclusion cannot be made without a randomized trial aimed at this specific question. Based on the FRISC-II results, an invasive treatment strategy should be recommended for all patients with ACS who fulfill the original inclusion criteria of the trial.
The present study supports the fact that the standard ECG remains an easy available, inexpensive method of risk stratifying patients with acute chest pain. When ECG data are carefully handled, the ECG provides independent information on prognosis and benefit of treatment. Thus, the present study suggests that quantitative ECG data and troponin measurements should be considered as complementary; they constitute the two most useful non-invasive tests in patients with non–ST-segment elevation ACS. ST-segment depression should be included in the risk models for assessment of prognosis in ACS. Furthermore, when resources for early revascularization therapy are limited, careful analysis of the ECG could help identify patients who might benefit the most from an invasive treatment strategy and, just as important, which patients can initially be allocated to a less aggressive approach starting with antithrombic treatment and referred to elective cardiac catheterization based on stress testing and symptoms.
Definition of confounders
LBBB = left bundle branch block
RBBB = right bundle branch block
LAHB = left anterior hemi block
LPHB = left posterior hemi block
LVH = left ventricular hypertrophy
RVH = Right ventricular hypertrophy
WPW = Wolf-Parkinson-White syndrome
LBBB = left bundle branch block: QRS duration >120 ms and typical configuration
RBBB = right bundle branch block: QRS duration >120 ms and typical configuration
LAHB = left anterior hemi-block: QRS axis between −45° and −90° and rS configuration in leads II, III, and aVF
LPHB = left posterior hemi-block: QRS axis >120° and rS configuration in leads I and aVL
LVH = left ventricular hypertrophy: at least two of the following criteria (1,2):
1. Sokolow-Lyon: S amplitude in lead V1+ R amplitude in lead V5or V6≥35 mm
2. Cornell voltage: S amplitude in lead V3+ R amplitude in lead aVL (men: ≥28 mm, women ≥20 mm)
3. Romhilt-Estes: scoring system—possible LVH = 4 points; definite LVH = 5 points; ST-segment deviation opposite QRS complex = 3 points: S amplitude in leads V1to V3≥25 mm = 3 points; R amplitude in leads V4to V6≥25 mm = 3 points; R or S amplitude in leads I, II, III, aVL, or aVF ≥20 mm = 3 points; terminal (downsloping) P-wave duration ≥40 ms = 3 points; left axis >−15° = 2 points; and QRS duration ≥90 ms = 1 point
RVH = right ventricular hypertrophy: R (lead V1or V2) + S (lead V6or I) − S (lead V1) ≥0.7 mV
WPW = Wolf-Parkinson-White syndrome: PR ≤120 ms, wide QRS complex, delta wave
Low voltage: R + S amplitude ≤0.5 mV in leads I, II, III, aVL, and aVF and R + S amplitude ≤1.0 mV in the precordial leads
☆ The FRISC-II trial was economically supported by Pharmacia. The ECG Core Laboratory at the Heart Center, Rigshospitalet, was supported by a grant Danish Heart Foundation.
- acute coronary syndrome(s)
- coronary artery bypass graft surgery
- coronary artery disease
- electrocardiogram, electrocardiographic
- Fast Revascularization during InStability in Coronary artery disease trial
- myocardial infarction
- percutaneous coronary intervention
- percutaneous transluminal coronary angioplasty
- troponin T
- upper limit of normal
- Received August 7, 2002.
- Revision received October 31, 2002.
- Accepted November 11, 2002.
- American College of Cardiology Foundation
- Fragmin in Unstable Coronary Artery Disease (FRISC) Study Group,
- Lindahl B.,
- Venge P.,
- Wallentin L.
- Platelet Receptor inhibition in Ischemic Syndrome Management (PRISM) Study Investigators,
- Heeschen C.,
- Hamm C.W.,
- Goldmann B.,
- Deu A.,
- Langenbrink L.,
- White H.D.
- TRIM Study Group,
- Holmvang L.,
- Luscher M.S.,
- Clemmensen P.,
- Thygesen K.,
- Grande P.
- GUSTO-IIa Investigators,
- Newby L.K.,
- Christenson R.H.,
- Ohman E.M.,
- et al.
- Diderholm E.,
- Andren B.,
- Frostfeldt G.,
- et al.
- Holmvang L.,
- Clemmensen P.,
- Wagner G.,
- Grande P.
- Platelet IIb/IIIa Antagonism for the Reduction of Acute Global Organization Network (PARAGON-A) and GUSTO-IIb Investigators,
- Kaul P.,
- Fu Y.,
- Chang W.C.,
- et al.
- PURSUIT Investigators,
- Boersma E.,
- Pieper K.S.,
- Steyerberg E.W.,
- et al.
- Sätre H.,
- Holmvang L.,
- Wagner G.S.,
- Lindahl B.,
- Wallentin L.
- Integrilin (eptifibatide) to Minimize Platelet Aggregation and Coronary Thrombosis-II (IMPACT-II) Investigators,
- Tardiff B.E.,
- Califf R.M.,
- Tcheng J.E.,
- et al.
- Califf R.M.,
- Abdelmeguid A.E.,
- Kuntz R.E.,
- et al.
- Akkerhuis K.M.,
- Alexander J.H.,
- Tardiff B.E.,
- et al.
- ↵Sadanandan S, Cannon CP, Dibattiste PM, et al. Outcomes of patients with spontaneous versus post PCI CK-MB elevation: data from the TACTICS–TIMI-18 trial (abstr). Presented at: AHA Scientific Session 2001. ID #3290. Available at: www.aha.agora.com/abstractviewer/results.asp
- Boersma E.,
- Akkerhuis K.M.,
- Theroux P.,
- Califf R.M.,
- Topol E.J.,
- Simoons M.L.