Electrocardiographic Localization of Coronary Artery

Transcription

1 Electrocardiographic Localization of Coronary Artery Narrowings: Studies During Myocardial Ischemia and Infarction in Patients with One-vessel Disease RICHARD M. FUCHS, M.D., STEPHEN C. ACHUFF, M.D., LouISE GRUNWALD, B.A., FRANK C.P. YIN, M.D., PH.D., AND LAWRENCE S.C. GRIFFITH, M.D. SUMMARY To investigate the accuracy of the 12-lead ECG in localizing the site of coronary artery narrowings, we reviewed abnormal ECGs obtained during myocardial infarction, spontaneous angina or exercise stress testing in 134 patients with angiographically documented one-vessel disease. The presence of Q waves, ST-segment elevation and T-wave inversion in leads I, avl and V1-V4 were all highly correlated with the presence of left anterior descending coronary artery disease (p < 0.001), and the same ECG findings in leads II, III and avf were associated with right (RCA) or circumflex coronary artery (LCx) narrowings (p < 0.001). In contrast, ST depression alone was not useful in predicting the site of coronary artery narrowing. Q waves correctly identified the location of the coronary disease in 98% of cases, ST elevation in 91%, T-wave inversion in 84%, and ST depression in 60%. No electrocardiographic criteria distinguished RCA from LCx disease, even in patients with a right-dominant circulation. These findings should lead to a better understanding of the value and limitations of the 12-lead ECG in localizing coronary artery disease. ABNORMALITIES in the 12-lead ECG are often used to localize the anatomic site of myocardial infarction and ischemia in patients with coronary artery disease. 1-3This practice is based largely on autopsy series correlating the site of myocardial infarction with the location of Q waves on antemortem ECGs: Q waves in the precordial leads VI-V4 appear to reflect anterior Wai, infarction; Q waves in leads II, III and avf inferior wall infarction; and Q waves in leads I, avl, V5 and V6 lateral wall infarction. 9 These same ECG findings are often assumed to correlate with coronary artery anatomy as well. Anterior wall infarction is usually attributed to disease of the left anterior descending coronary artery (LAD), inferior wall infarction to disease of the right coronary artery (RCA) except in patients with left-dominant systems, and lateral wall infarction to disease of the left circumflex (LCx) coronary artery.' Unfortunately, there is only limited documentation for these correlations between the location of coronary artery narrowings or occlusions and the findings of Q waves during myocardial infarction. '01 I There is even less documentation of the accuracy of electrocardiographic ST-segment or T-wave changes in identifying the site of injury or ischemia during infarction, rest angina or stress testing. Studies are often confounded by the inclusion of patients with multivessel coronary artery disease, which makes it difficult to determine in which vascular distribution From the Cardiovascular Division, Department of Medicine, The Johns Hopkins Medical Institutions, Baltimore, Maryland. Presented in part at the 51 st Scientific Sessions of the American Heart Association, Dallas, Texas, November Dr. Fuchs is the recipient of a fellowship grant from the American Heart Association. His present address: Department of Medicine, The New York Hospital-Cornell Medical Center, 525 East 68th Street, New York, New York Address for correspondence: Lawrence S. C. Griffith, M.D., Division of Cardiology Department of Medicine, The Johns Hopkins Hospital, 600 North Wolfe Street, Baltimore, Maryland Received January 19, 1982; revision accepted May 10, Circulation 66, No. 6, ischemia occurred. Nonetheless, patients who develop ST depression in leads II, III and avf during angina are commonly referred to as having "inferior wall ischemia" and are often presumed to have RCA disease; similar associations are often drawn between ST depression in anterior precordial leads and LAD disease, and between ST depression in so-called lateral leads I and avl and LCx disease. ' We have not found such associations to be uniformly valid. Therefore, we undertook the present study to evaluate the value and limitations of the 12-lead ECG in localizing the site of coronary artery disease. One hundred thirty-four patients with angiographically documented one-vessel coronary disease and abnormal ECGs recorded during myocardial infarction, spontaneous rest angina or treadmill exercise testing were evaluated, and electrocardiographic-angiographic correlations were obtained. Methods Patient Selection The records of all adult cardiac catheterization studies performed at the Johns Hopkins Hospital between January 1971 and April 1981 were reviewed; patients with a final diagnosis of one-vessel coronary artery disease were selected for further evaluation. For each case, the coronary angiographic films were reviewed independently by two observers without knowledge of the clinical or electrocardiographic findings. At least four views of the left coronary system and at least two views of the right coronary system were routinely obtained in each patient. Patients were considered to have one-vessel disease only if both observers described a 70% or greater diameter narrowing in one of the three major coronary arteries (LAD, RCA or LCx) without narrowing > 40% in the other two coronary arteries or their branches. No patient with significant narrowing in the left main coronary artery was included. Hospital records and stress test results were sought for all these patients. 1168

2 ECG LOCALIZATION OF CORONARY NARROWING/Fuchs et al The study population consisted of all 134 patients who met the following criteria: one-vessel coronary disease as defined above; no significant valvular or congenital heart disease; no bundle branch block or left ventricular hypertrophy on resting ECG; and an abnormal 12-lead ECG recorded during myocardial infarction documented by cardiac enzyme elevations, spontaneous angina at rest, with normal cardiac enzymes and reversible electrocardiographic changes or angina during exercise testing. Ten patients had positive stress tests after recovering from a myocardial infarction, and three patients had ECGs recorded during both rest angina and stress testing; thus, 147 instances of abnormal ECGs form the basis for this report. Exercise Tests All tests were standard graded treadmill exercise tests using a modified Bruce protocol. Electrodes were placed according to the method of Mason and Likar. 12 The exercise protocol was a submaximal test; the patient walked for 1 minute at 00 incline at 1 mph, then 2 minutes at 100 and 2 mph, then 3 minutes each at 150 and 3 mph, 200 and 3 mph and 22 and 4 mph. Twelvelead ECGs were recorded at rest, during each minute of exercise, upon termination of exercise and every minute thereafter until the ECG returned to baseline. Tests were terminated when at least one of the following end points was reached: moderate or severe angina, severe dyspnea, claudication or fatigue; 2 mm of horizontal or downsloping ST-segment depression or 2 mm of ST elevation; or 90% of the predicted maximal heart rate for age. Electrocardiogram All ECGs were reviewed independently by two of the investigators; only changes graded as abnormal by both were included. Serial ECGs recorded during the course of myocardial infarction were analyzed, and 11 leads (all except avr) were evaluated individually according to a modification of the Minnesota code.'3 Q waves were considered significant when their duration was at least 0.04 second in any lead or at least 0.03 second in association with a Q/R ratio greater than 1:3 in leads I, II and V2-V6. ST-segment elevation was considered significant if the J point was elevated more than 1 mm and the ST segment remained elevated more than 1 mm 0.08 second beyond the J point. Similarly, ST-segment depression was considered significant if the J point was depressed by more than 1 mm and the ST segment was horizontal or downsloping and remained 1 mm below the baseline 0.08 second beyond the J point. T-wave inversions were considered significant in leads I, II and V2-V6 when the net amplitude was negative; in leads III, av, and av, when the QRS was isoelectric or mainly upright and the net T- wave amplitude was negative; and in lead VL when the net amplitude was negative and represented a change from a previous tracing. The same criteria were used for ST-segment change and T-wave inversion during angina at rest and for ST-segment change during stress testing. Statistical Evaluation Data were tabulated comparing the prevalence of Q waves, ST elevation, ST depression and T-wave inversion in each electrocardiographic lead in patients with one-vessel disease involving the LAD, RCA and LCx. To examine the importance of "reciprocal" ST changes, additional tables were prepared showing the prevalence of ST depression in tracings with and without concomitant ST elevation in any lead. The prevalence of each ECG change was compared between patients with LAD and non-lad (RCA or LCx) disease, lead by lead, with the chi-square test and, where appropriate, by Fisher's exact test. To avoid type II errors when examining 11 leads in each of three clinical situations (i.e., myocardial infarction, unstable angina and exercise testing), differences were considered significant if p < Results Patient Population The study group consisted of 96 men and 38 women, ages years (mean + SD 48 ± 9 years). Of 71 patients with isolated LAD disease, 42 had had a myocardial infarction, seven had angina at rest and 30 had positive stress tests. Of 47 patients with one-vessel RCA disease, 29 had had a myocardial infarction, four had angina at rest and 19 had positive stress tests. Of 16 patients with disease limited to the LCx, eight had had a myocardial infarction, one had angina at rest and seven had positive stress tests. Seven patients with LCx disease had a right-dominant circulation, six had a balanced circulation, and three had a left-dominant circulation. Myocardial Infarction During the course of myocardial infarction, the location of Q waves was highly predictive of the location of the obstructed coronary artery (table 1). Twentynine patients with LAD disease developed significant Q waves in one or more of leads I, avl and V,-V4; with one exception, Q waves did not develop in these leads in patients with RCA or LCx disease (p < for each lead). In contrast, Q waves in leads II, III and avf reflected infarction in the RCA or LCx territory and occurred in only one patient with LAD disease (p < 0.001). Q waves were found in leads V5 and V6 in patients with both LAD and non-lad disease. Four of the six patients with LCx disease and Q waves during infarction had a right-dominant circulation; two had a balanced circulation. All six of these patients showed significant Q waves in at least two of the "inferior" leads (II, III and avf), and none developed Q waves in lead I or avl. ECG criteria did not distinguish between patients with infarction due to RCA disease and those with infarction due to LCx disease. The associations between the location of ST-segment elevation during myocardial infarction and the location of the obstructed coronary artery were also quite strong, but exceptions were found more frequently than with Q waves (table 2). ST-segment elevation

3 1170 CIRCULATION VOL 66, No 6, DECEMBER 1982 TABLE 1. Q Waves During Myocardial Infarction I L V1 V2 V3 V4 V5 V6 II III F LAD (n=29) 10* 15* 24* 27* 26* 23* * 1* 1* RCA (n=25) LCx (n=6) *Significant difference in prevalence of Q waves in that lead between LAD and non-lad patients, p < Abbreviations: LAD = left anterior descending coronary artery; RCA = right coronary artery; LCx = left circumflex coronary artery. in leads I, avl and V -V5 during myocardial infarction correlated with the presence of LAD disease (p < for each lead); ST-segment elevation in leads LI, III and avf was associated with RCA or LCx disease (p < 0.005). All patients with LCx disease and ST elevation during myocardial infarction had elevation in at least two of the inferior leads (II, III and avf). Three of the LCx patients had a right-dominant circulation; two had a balanced circulation. T-wave inversion in leads I, avl and V2-V6 during myocardial infarction was strongly associated with LAD disease (p < in each case) (table 3), and T-wave inversion in leads II, III and avf occurred predominantly in patients with RCA or LCx disease (p < 0.005). All five patients with LCx disease and T- wave inversions during infarction had inversion in at least two of the inferior leads. One of these patients had a right-dominant circulation, one left-dominant, and three balanced. The exceptions to the electrocardiographic-angiographic associations were more frequent with T-wave inversion than with Q waves or STsegment elevation, but the association was still strong. ST-segment depression during myocardial infarction followed a different pattern. Depression in leads I and avl was associated with RCA or LCx disease, and depression in leads III and avf was associated with LAD disease (all p < 0.001) (table 4). These ST depressions probably represent "reciprocal" changes. Of the 42 patients with ST depression during myocardial infarction, 37 also had simultaneous ST elevation in other leads. Patients with ST elevation in leads I, avl and V1-V5 tended to have ST depression in leads IL, III and avf, whereas patients with ST elevation in leads II, III and avf tended to have ST depression in leads I and avl. Rest Angina Patients in whom ECGs were recorded during rest angina tended to have ST elevation and T-wave inversion in the same leads as during myocardial infarction (tables 2 and 3). The number of ECGs recorded during rest angina was small, however, and significant associations were present only between T-wave inversion in lead V4and LAD disease (p < 0.005) and between T- wave inversion in lead avf and RCA or LCx disease (p < 0.005). Only five patients had ST depression during rest angina; no significant associations were present (table 4). TABLE 2. ST-segment Elevation I L VI V2 V3 V4 V5 V6 II III F Overall LAD (n= 54) 16t 24t 31t 45t 40t 34t 26t 16 6t 2t 3t RCA (n=23) LCx (n=6) During myocardial infarction LAD (n=36) 15t 17t 24t 34t 31t 29t 25t 15 6t 2t 3t RCA (n=16) LCx (n=5) During rest angina LAD (n=4) RCA (n=2) LCx (n = 0) During stress testing LAD(n=14) * 0 RCA (n=5) LCx (nl1) Symbols indicate a significant difference in prevalence of ST-segment elevation in that lead between LAD and non- LAD patients: *p < tp < p <

4 ECG LOCALIZATION OF CORONARY NARROWING/Fuchs et al TABLE 3. T-wave Inversion I L V1 V2 V3 V4 V5 V6 H III F Overall LAD (n=45) 33t 39t 12* 34t 35t 40t 39t 33* 7t 9t 2t RCA (n=30) LCx (n=6) During myocardial infarction LAD (n=38) 29* 34* 9 28* 29* 33* 32* 29* 7* 7* 2* RCA (n=27) LCx (n=5) During rest angina LAD (n=7) * * RCA (n=3) LCx (n= 1) Symbols indicate a significant difference in prevalence of T-wave inversion in that lead between LAD and non-lad patients: *p < tp < Exercise Testing All 56 patients with positive stress tests had typical angina pectoris during exercise; ST-segment elevation developed in 20 of these patients. Angiographic correlations with the leads in which ST elevation occurred during exercise testing revealed a pattern similar to the correlations with ST elevation during myocardial infarction (table 2), but a significant association was found only between ST elevation in lead III and RCA disease (p < 0.01). ST-segment depression developed in 54 of the 56 patients with positive stress tests. There was no correlation between the diseased coronary artery and the leads in which ST depression occurred (table 4). This lack of correlation was unaffected by excluding patients who also had ST-segment elevation during stress testing, in whom ST depression might be assumed to be reciprocal. ST-segment Elevation The data on ST-segment elevation can also be examined primarily with respect to electrocardiographic changes, regardless of the clinical event associated with those electrocardiographic changes. When ECGs recorded during myocardial infarction, rest angina and stress testing were combined, ST-segment elevation in leads I, avl and V-V5 appeared to be associated with LAD disease (p < 0.005) (table 2), and ST elevation TABLE 4. ST-segment Depression I L VI V2 V3 V4 V5 V6 II III F Overall LAD (n=49) 2* 0* * 33* RCA (n=37) LCx (n=ll) During myocardial infarction LAD (n=1i7) 0* 0* * 10* RCA (n=17) LCx (n=4) During rest angina LAD (n=3) RCA (n=2) LCx (n = 0) During stress testing LAD (n=29) RCA (n=18) LCx (n=7) *Significant difference in prevalence of ST-segment depression in that lead between LAD and non-lad patients, p <

5 1172 CIRCULATION VOL 66, No 6, DECEMBER 1982 in leads II, III or avf with RCA or LCx disease (p < 0.001). T-wave Inversion When ECGs recorded during infarction and rest angina were combined, T-wave inversion in leads I, avl, V1 V6 was associated with LAD disease (p < 0.005, table 3), and T-wave inversion in leads II, III or avf was associated with RCA or LCx disease (p < 0.001). ST Depression and Reciprocal Changes When ECGs recorded during infarction, rest angina and stress testing were combined, ST depression in leads I and avl was associated with RCA or LCx disease (p < 0.001) (table 4), and ST depression in leads III and avf with LAD disease (p < 0.001). These findings are opposite to those found with Q waves, ST elevation and T-wave inversion and are probably due to a high prevalence of "reciprocal" STsegment depression. If tracings with ST depression alone are examined (excluding tracings with ST elevation in some leads and depression in others), these significant associations disappear (table 5). Of the 57 patients in whom ST elevation was recorded during the course of myocardial infarction, 31 (54%) had ST depressions in other leads on the same tracing. These "reciprocal" changes were seen less often in patients with infarctions due to LAD occlusion (39%) than in patients with LCx (80%) or RCA (88%) disease. In patients with reciprocal changes, ST depression was noted in lead II, III or avf in every case of LAD disease, and in lead I or avl in every case of RCA disease. ST depression in the precordial leads was seen in eight of 15 patients with RCA disease. In the four cases of reciprocal change due to LCx infarction, ST depression was found in at least three precordial leads in all four patients and in lead I or avl in three of the four (table 6). Eighteen of the 20 patients with ST-segment elevation during stress testing had significant ST depression in other leads on the same record. The distribution of these ST-segment depressions was similar to that of the reciprocal changes found during myocardial infarction: ST depression in lead II, III or avf was generally associated with ST elevation in lead avl or V1-V4 and LAD disease. ST depression in lead I, avl or V2-V4 was generally associated with ST elevation in lead II, III or avf and RCA or LCx disease. In this group, ST depression in leads V and V6 was found frequently with one-vessel disease of the LAD, RCA and LCx. Variations in Coronary Anatomy, Collateral Vessels and Severity of Coronary Stenosis Inferior ST-segment depression during stress testing of patients with one-vessel LAD disease might have been due to the presence of an unusually large LAD that extended around the cardiac apex to supply the distal inferior wall. Our data showed that this was not the case: 15 of the 18 patients with one-vessel LAD disease in which the LAD wrapped around the cardiac apex had ST depression in at least one inferior lead during stress testing, while nine of 11 patients with LAD obstruction and short LADs (stopping at or before the apex) developed ST depression in lead II, III or avf during stress testing. Similarly, the presence or absence of angiographically visible collateral vessels arising from the RCA or LCx and supplying the diseased LAD coronary artery territory did not correlate with the presence of inferior ST depression during myocardial infarction or stress testing. Of 29 patients with LAD disease and ST depression during stess testing, eight had angiographically visible collaterals. Six of these eight (75%) had ST depression in leads II, III or avf, while 18 (86%) of the 21 patients without collaterals had ST depression in at least one inferior lead. Finally, some coronary nafrowings in "nondiseased" arteries that appeared to involve no more than 40% of the lumen diameter might have been hemodynamically significant. If so, some of our patients might have had two- or three-vessel disease. However, excluding all patients with any detectable narrowing in the nondiseased vessels did not alter the conclusions: The location of Q waves, ST elevation and T-wave TABLE 5. ST-segment Depression Excluding Tracings with Simultaneous ST-segment Elevation I L V1 V2 V3 V4 V5 V6 II III F Overall LAD (n=22) RCA (n= 15) LCx (n=6) Differences between LAD and non-lad patients were not significant. TABLE 6. in Other Leads During Myocardial Infarction I L VI V2 V3 V4 V5 V6 II HI F LAD (n=14) RCA(n=15) LCx (n=4) "Reciprocal" Changes: Prevalence of ST-segment Depression in Tracings with Simultaneous ST Elevation

6 ECG LOCALIZATION OF CORONARY NARROWING/Fuchs et al inversion remained predictive of the location of coronary disease, but ST depression alone was not predictive. Predictive Value Table 7 illustrates the usefulness of electrocardiographic location of Q waves and ST-T changes in localizing coronary artery disease in this selected patient population. The presence of a Q wave in two or more of leads I, avl and V,-V4 during myocardial infarction was 100% predictive of LAD disease; ST elevation in two or more of these leads during infarction, rest angina or stress testing was 94% predictive; and T-wave inversion in two or more of these leads during infarction or rest angina was 83% predictive. Similarly, the finding of Q waves in two or more of leads II, III and avf during infarction was 97% predictive of RCA or LCx disease; ST elevation in two or more of these leads during infarction, rest angina or stress testing was 90% predictive; and T-wave inversion in two or more of these leads during infarction or rest angina was 89% predictive. Excluding patients who had simultaneous ST elevation and depression, ST depression in two or more of leads I, avl and V,-V4 during infarction, rest angina or stress testing was 60% predictive of LAD disease; ST depression in two or more of leads II, III and avf was 56% predictive of RCA or LCx disease. By these criteria, the ECG location of Q waves, ST elevation and T-wave inversion were highly predictive of the anatomic location of coronary disease (p < in each case), but ST depression was not predictive (p > 0.25). Discussion The present study shows that (1) the development of Q waves, ST elevation or T-wave inversion in lead I, avl or V,-V4 is highly predictive of LAD disease; (2) the development of Q waves, ST elevation or T-wave inversion in lead II, III or avf is highly predictive of the presence of RCA or LCx disease, regardless of whether a right, left or balanced distribution is present; (3) RCA disease cannot be distinguished from LCx disease by ECG criteria; (4) "reciprocal" ST depression frequently develops during infarction and stress testing; and (5) the ECG location of ST depression is not very useful in predicting the location of coronary artery disease. These conclusions are based on a study of patients with one-vessel coronary disease in which many tracings were recorded during myocardial infarction and during stress testing and smaller numbers during rest angina. Although this study did not include a large number of patients with angina at rest, the patterns of ST-segment and T-wave changes in this group generally paralleled those recorded during infarction and stress testing; hence, the above conclusions probably apply to all three clinical circumstances. Investigation of patients with one-vessel coronary disease is important, because only in such patients can one be reasonably certain in which vascular distribution ischemia or infarction is occurring. TABLE 7. One-vessel Disease: Electrocardiographic-Angiographic Correlations Present Present in 2 or in 2 or more of more of leads I, leads avl, II, III, V1-V4 avf p Q waves (n = 57) LAD RCAILCx <.0 <0.001 ST elevation (n = 70) LAD 44 3 <.0 RCA/LCx 3 26 <0.001 T-wave inversion (n = 78) LAD RCA/LCx <.0 <0.001 ST depression (n = 82) LAD RCAfLCx <.0 <0.001 ST depression excluding tracings with ST elevation (n= 35) LAD RCA/LCx >02 >0.25 The correlation between the leads in which pathologic Q waves occur and the anatomic location of the infarction was noted as long ago as and has been repeatedly confirmed.5-9 With the advent of coronary angiography, these correlations were extended to include coronary artery anatomy. 10'" Early studies were limited, however, by the frequent occurrence of multivessel disease; furthermore, ECGs were not examined lead by lead, and leads I and avl were not evaluated in many cases. Our data from a more homogeneous patient population show that infarctions termed anterior, anteroseptal, anterolateral and lateral are all caused by LAD occlusion, whereas inferior or diaphragmatic infarctions are due to RCA or LCx disease. Contrary to some suggestions,' "lateral" infarctions with changes in leads I and avl did not occur in patients with LCx occlusions. Q waves in these leads may reflect infarction of the LAD diagonal territory. We are aware of no systematic study relating the ECG location of ST-segment elevation during myocardial infarction to the pathologic site of injury or the angiographic site of coronary disease in patients. Experimental studies have shown that ST-segment elevation on the surface ECG develops during epicardial or transmural myocardial injury.2' 3, 1-17 ST-segment mapping studies during myocardial infarction'8 and analyses of electrocardiographic-angiographic correlations during coronary spasm'9-2' or exercise-induced ST-segment elevation22-26 have generally supported the belief that ST elevation in precordial leads indicates transmural ischemia or injury in the LAD distribution, and ST elevation in the "inferior" leads (II, III and avf) indicates transmural ischemia or injury in the RCA or LCx distribution. In the present study, we

7 1174 CIRCULATION VOL 66, No 6, DECEMBER 1982 confirmed these findings and further analyzed the correlations lead by lead. Most studies have indicated that ST-segment elevation during stress testing is uncommon, although many of these studies did not record 12 leads during the test The finding that more than one-third of the patients who underwent stress testing in the current study had ST elevation may be related to the monitoring of all 12 leads and use of patients with one-vessel disease only. This finding was not related exclusively to the presence of prior infarction; only five of 20 patients with ST elevation during stress testing had evidence of prior infarction. Dunn et al.25' 26 examined a similar population of patients with one-vessel disease and found a similar incidence of ST-segment elevation during stress testing. Although T-wave changes have long been recognized as a marker for myocardial ischemia and infarction,' - few data are available as to their usefulness in locating the anatomic site of the perfusion deficit. The present study shows that the location of new T-wave inversions on the resting ECG during myocardial infarction is useful in predicting the site of coronary artery disease; in some patients, this appears to be true during episodes of rest angina as well. The present study has shown that ST-segment depression alone does not accurately reflect the site of ischemia, probably because ST-segment depression occurs frequently both as a primary change due to subendocardial ischemia or infarction and as a secondary "reciprocal" change. ST elevation during myocardial infarction is sometimes accompanied by reciprocal ST depression in "opposite" leads.30 The present study defined the frequency of this reciprocal change in patients with one-vessel disease and indicated that reciprocal changes are seen more often with RCA and LCx (88% and 80%) than with LAD disease (39%) during myocardial infarction. Our finding of the nonspecificity of the site of ST depression during stress testing for localizing the likely region of myocardial ischemia (i.e., the region perfused by the vessel with significant stenosis) is in accord with some studies25' 36 but at variance with others.37-3 Ours, however, is the largest series of positive stress tests in patients with one-vessel disease, where localization of ischemia is more certain. Much has been written about which ECG leads are most "sensitive" for detecting ischemia during stress testing.40'41 Our findings, like others, show that leads VP,V6 and II are the leads with the highest prevalence of positivity (table 4). However, approximately oneeighth of all positive stress tests were negative in these three leads. Monitoring all 12 leads significantly increases the yield of positive exercise stress testing and permits the localization of coronary disease in patients with ST-segment elevation during exercise. We believe that the nonspecificity of ST-segment depression in localizing the site of ischemia or infarction is due to the multiple ways in which ST depression can be produced. Animal experiments have shown that subendocardial ischemia or injury can produce ST depression in leads overlying the area of damage, and transmural ischemia or injury can produce ST depression in distant "reciprocal" leads.''6' 30, 42 Thus, ST depression in lead II, III or avf might be due to subendocardial ischemia in the distribution of the LCx or RCA or to transmural ischemia in the distribution of the LAD. There is no clear means of differentiating the two possibilities based on the ECG alone. Studies involving ST-segment mapping'8 43 and vectorcardiography33 34 in patients confirm these findings and emphasize the variability in translation of epicardial leads into body surface precordial and limb leads.44 The development of ST elevation in significant numbers of patients during exercise in our study suggest that transmural ischemia may develop in some patients undergoing stress testing, and ST depression in these cases may be reciprocal.22 23,30 In some instances, small areas of transmural ischemia not manifest in the 12- lead ECG by at least 1 mm of ST-segment elevation may produce reciprocal ST depression in "opposite" leads on the surface ECG. A more sensitive lead system or a more sensitive criterion for ST elevation (e.g., 0.5 mm in some leads) might reveal the "primary" ST elevation in some of these cases. An alternative explanation, while unlikely, is that ischemia may actually develop in a region supplied through a nonstenotic coronary artery (e.g., due to coronary spasm or abnormal wall stress). In patients with acute transmural myocardial infarction, several authors have attempted to predict the presence of a second jeopardized vascular territory by the presence of ST depression in "opposite" electrocardiographic leads, often with disparate results. For example, anterior precordial ST-segment depression in the setting of acute inferior myocardial infarction has been variously attributed to concomitant anterior (LAD) disease45' I or to extensive posterolateral (RCA or LCx) infarction However, the high prevalence of "reciprocal" ST depression in patients with onevessel disease suggests that localization of additional regions at risk for ischemia or infarction may not be possible by examining ST-segment depression alone. This is particularly important in populations with a high prevalence of multivessel disease, such as patients with myocardial infarction. Our data suggest that the presence of ischemia or infarction in more than one vascular territory may only be reliably identified when Q waves, ST elevation or T-wave inversion are present in both anterior and inferior leads. We examined the correlation between ECG changes and coronary anatomy, but did not directly examine which myocardial region was ischemic or injured. Given the moderate degree of interpatient variability in the myocardial area supplied by a given coronary artery,49 it is not surprising that abnormalities in certain leads - notably V5 and V6 -were not very useful in localizing the site of myocardial infarction, even when Q waves occurred in these leads. Changes in the other leads correlated strongly with narrowings in specific coronary arteries. In conclusion, we emphasize the specificity of elec-

8 ECG LOCALIZATION OF CORONARY NARROWING/Fuchs et al trocardiographic-anatomic correlations in patients with Q waves, ST-segment elevation and T-wave inversion during myocardial infarction and ischemia. We suggest closer scrutiny of ST segments for elevation during stress testing (particularly in little-used leads, such as avl. V1 and V2) and greater caution in the use of such terms as "anterior" and "inferior" ischemia based solely on ST-segment depression. Acknowledgment The authors gratefully acknowledge the statistical advice of Clayton Kallman and the assistance of Janet Lewis in preparation of the manuscript. References 1. Hutter AM, De Sanctis RW: The evaluation and management of patients with angina pectoris. In The Practice of Cardiology, edited by Johnson RA, Haber E, Austin WG. Boston, Little, Brown, 1980, pp Lipman BS, Massie E: Clinical Scalar Electrocardiography. Chicago, Year Book, 1956, pp Chung EK: Electrocardiography: Practical Applications with Vectorial Principles. New York, Harper & Row, 1980, pp Wolferth CC, Wood FC: Acute cardiac infarction involving anterior and posterior surfaces of the left ventricle. Arch Intern Med 56: 77, Myers GB, Klein HA, Stofer BE: Correlation of electrocardiographic and pathologic findings in large anterolateral infarcts. Am Heart J 36: 535, Myers GB, Klein HA, Hiratzka T: Correlation of electrocardiographic and pathologic findings in large anterolateral infarcts. Am Heart J 36: 838, Myers GB, Klein HA, Stofer BE: VII. Correlation of electrocardiographic and pathologic findings in lateral infarction. Am Heart J 37: 374, Myers GB, Klein HA, Hiratzka T: V. Correlation of electrocardiographic and pathologic findings in posterior infarction. Am Heart J 38: 547, Savage RM, Wagner GS, Ideker RE, Podolsky SA, Hackel DB: Correlation of postmortem anatomic findings with electrocardiographic changes in patients with myocardial infarction. Circulation 55: 279, Williams RA, Cohn PF, Vokonas PS, Young E, Herman MV, Gorlin R: Electrocardiographic, arteriographic and ventriculographic correlations in transmural myocardial infarction. Am J Cardiol 31: 595, Hamby RI, Hoffman I, Hilsenrath J, Amtablian A, Shanies S, Padmanabhorn VS: Clinical, hemodynamic and angiographic aspects of inferior and anterior myocardial infarctions in patients with angina pectoris. Am J Cardiol 34: 513, Mason RE, Likar I: A new system of multiple lead exercise electrocardiography. Am Heart J 71: 196, Rose GA, Blackburn H: Electrocardiographic reading codes (Appendix D). In The Coronary Drug Project: Design, Methods and Baseline Results. Circulation 47 (suppl 1): 1-39, Hellerstein HK, Katz LN: The electrical effects of injury at various myocardial locations. Am Heart J 36: 184, Ekmekci A, Toyoshima H, Kwoczynski JK, Nagaya T, Prinzmetal M: Angina pectoris. IV. Clinical and experimental difference between ischemia with ST elevation and ischemia with ST depression. Am J Cardiol 7: 412, Guyton RA, McClenathan JH, Newman GE, Michaelis LL: Significance of subendocardial ST segment elevation caused by coronary stenosis in the dog. Epicardial ST segment depression, local ischemia and subsequent necrosis. Am J Cardiol 40: 373, Vincent GM, Abildskov JA, Burgess MJ: Mechanisms of ischemic ST segment displacement. Circulation 56: 559, Holland RP, Brooks H: TQ-ST segment mapping: critical review and analysis of current concept. Am J Cardiol 40: 110, MacAlpin R: Relation of coronary arterial spasm to sites or organic stenosis. Am J Cardiol 46: 143, DeServi S, Specchia G, Angoli L, Bramucci E, Mussini A, Mariani GP, Salerno J, Bobba P: Coronary arterial spasm in angina at rest associated with transient ST segment changes. Clin Cardiol 3: 54, Yasue H, Omote S, Takizawa A, Masao N, Hyon H, Nishida S, Horie M: Comparison of coronary arteriographic findings during angina pectoris associated with ST elevation or depression. Am J Cardiol 47: 539, Fortuin NJ, Friesinger GC: Exercise-induced ST segment elevation. Am J Med 49: 459, Longhurst JC, Kraus WL: Exercise-induced ST elevation in patients without myocardial infarction. Circulation 60: 616, Fox KM, Selwyn AP, Shillingford JP: Projection of electrocardiographic signs in precordial maps after exercise in patients with ischemic heart disease. Br Heart J 42: 416, Dunn RF, Bailey IK, Uren R, Kelly DT: Exercise-induced ST segment elevation: correlation of thallium-201 myocardial perfusion scanning and coronary arteriography. Circulation 61: 989, Dunn RF, Freedman B, Kelly DT, Bailey IK, McLaughlin A: Exercise-induced ST segment elevation in leads VI or a VL: a predictor of anterior myocardial ischemia and left anterior descending coronary artery disease. Circulation 63: 1357, Waters DD, Chaitman BR, Bourassa MG, Tubau JF: Clinical and angiographic correlates of exercise-induced ST segment elevation. Circulation 61: 286, Sriwattanokomen S, Ticzon AR, Zubritzky SA, Blobner CG, Rice M, Duffy FC, Lanna EF: ST segment elevation during exercise: electrocardiographic and arteriographic correlation in 38 patients. Am J Cardiol 45: 762, Chaitman BR, Waters DD, Theroux P, Hanson JS: ST segment elevation and coronary spasm in response to exercise. Am J Cardiol 47: 1350, Wolferth CC, Bellet S, Livezey MM, Murphy FD: Negative displacement of the RS-T segment in the electrocardiogram and its relationship to positive displacement: an experimental study. Am Heart J 29: 220, Ariskog NH, Bjork L, Bjdrk VO, Hallen A, Strom G: Physical work capacity, ECG reaction to work test and coronary angiogram in coronary artery disease. Acta Med Scand (suppl 472): 9, Tubau JF, Chaitman BR, Bourassa MG, Lesperance J, Dupras G: Importance of coronary collateral circulation in interpreting exercise test results. Am J Cardiol 47: 27, Ascoop CA, Distelbrink CA, DeLang P, Van Bemmel JH: Quantitative comparison of exercise vectorcardiograms and findings at selective coronary arteriography. J Electrocardiol 7: 9, Blomqvist CG, Gaffney FA, Atkins JM, Nixon JV, Mullins CB, Willerson JT: The exercise ECG and related physiological data as markers of critical coronary artery lesion. Acta Med Scand (suppl 615): 51, Chaitman BR, Bourassa MG, Wagniart P, Corbara F, Ferguson RJ: Improved efficiency of treadmill exercise testing using a multiple lead ECG system and basic hemodynamic exercise response. Circulation 57: 7, Dunn RF, Freedman B, Bailey IK, Uren RF, Kelly DT: The value of exercise electrocardiography and exercise thallium-20 1 myocardial perfusion scanning in determining the presence of single vessel coronary artery obstruction. Am J Cardiol 48: 837, Herman MV, Elliott WC, Gorlin R: An electrocardiographic, anatomic and metabolic study of zonal myocardial ischemia in coronary heart disease. Circulation 35: 834, Robertson D, Kostuk WJ, Ahuja SP: The localization of coronary artery stenosis by 12 lead ECG response to graded exercise test: support for intercoronary steal. Am Heart J 91: 437, Fox KM, Selwyn A, Oakley D, Shillingford JP: Relation between the precordial projection of ST segment changes after exercise and coronary angiographic findings. Am J Cardiol 44: 1068, Mason RE, Likar 1, Biern RO, Ross RS: Multiple-lead exercise electrocardiography. Circulation 36: 517, Phibbs BP, Buckels LT: Comparative yield of ECG leads in multistage stress testing. Am Heart J 90: 275, Kato K, Fukuda H, Koyama S: Depression of the ST segment in epicardial electrocardiogram associated with experimental major coronary artery constriction. J Electrocardiol 1: 167, Madias JE, Hood WB: Precordial ST segment mapping. 4. Experience with mapping of ST segment depression in anterior transmural

9 1176 CIRCULATION VOL 66, No 6, DECEMBER 1982 myocardial infarction. J Electrocardiol 9: 315, Santinga JT, Brymer JF, Smith F, Flora J: The influence of lead strength on the ST changes with exercise electrocardiography (correlative study with coronary arteriography). J Electrocardiol 10: 387, Shah PK, Pichler M, Berman DS, Maddahi J, Peter P, Singh BN, Swan HJC: Noninvasive identification of a high risk subset of patients with acute inferior myocardial infarction. Am J Cardiol 46: 915, Salcedo JR, Baird MG, Chambers RJ, Beanland DS: The significance of reciprocal S-T changes in anterior precordial leads in acute inferior myocardial infarction: concomitant left anterior descending coronary artery disease? Am J Cardiot 48: 1003, Myers GB, Klein HA, Hiratzka T: Correlation of electrocardiographic and pathologic findings in posterolateral infarction. Am Heart J 18: 837, Goldberg H, Borer JS, Jocobstein JC, Klayer J, Scheido S, Alonso D: Anterior S-T segment depression in acute inferior myocardial infarction: an indicator of posterolateral infarction. Am J Cardiol 48: 1009, Kalbfleisch H, Hort W: Quantitative study on the size of coronary artery supplying areas postmortem. Am Heart J 94: 183, 1977 High-density Lipoprotein Cholesterol and Prognosis After Myocardial Infarction Prepared for the Coronary Drug Project Research Group by KENNETH G. BERGE, M.D., PAUL L. CANNER, PH.D., AND ADRIAN HAINLINE, JR., PH.D. SUMMARY The Coronary Drug Project was a randomized, placebo-controlled trial of lipid-influencing drugs in men who had recovered from one or more documented myocardial infarctions. Determinations of high-density lipoprotein (HDL) cholesterol were made at baseline in a group of 354 men randomized to the placebo group. Five-year mortality was highest (33.0%) in men with baseline serum HDL cholesterol levels of less than 35 mg/dl; it was 15.9%, 17.7%, and 21.8% in men with levels of 35-39, 40-44, and 45 mg/dl, respectively (for the linear inverse relationship between HDL cholesterol and 5-year mortality, p = 0.029). Adjustment for 40 baseline variables had a minimal effect on this relationship (p = 0.042). IN RECENT YEARS, several epidemiologic studies have indicated that low levels of high-density lipoprotein (HDL) cholesterol in the blood are associated with an increase in incidence and severity of coronary heart disease and with ischemic cerebrovascular disease.1-9 To our knowledge, no prospective data have been reported regarding possible prognostic significance of HDL cholesterol levels determined after recovery from acute myocardial infarction (MI). Limited data available from the baseline period of the Coronary Drug Project allow an assessment of HDL cholesterol in this circumstance. Methods The Coronary Drug Project10 was a randomized, double-blind, placebo-controlled trial of lipid-influencing drugs in the secondary prevention of coronary heart disease. Men ages years (mean 52.4 years) who had recovered from one or more documented MIs were recruited by 53 project clinical centers. Men with The Coronary Drug Project was carried out as a collaborative study supported by research grants and other funds from the National Heart, Lung, and Blood Institute. From the Department of Internal Medicine, Mayo Clinic, Rochester, Minnesota; the Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore, Maryland; and the Clinical Chemistry Division, Center for Environmental Health, Center for Disease Control, Atlanta, Georgia. Address for correspondence: Paul L. Canner, Ph.D., Division of Clinical Investigation, Department of Epidemiology and Preventive Medicine, University of Maryland, 600 Wyndhurst Avenue, Baltimore, Maryland Received February 2, 1982; revision accepted May 5, Circulation 66, No. 6, chronic conditions other than coronary heart disease were excluded from the study; thus, 88% of the deaths were from cardiovascular causes. As part of their baseline clinical assessment, three sets of lipid determinations were carried out under fasting conditions in the morning, after a low-fat meal the previous evening. During the early phases of the enrollment period, baseline lipid analyses included determination of the HDL cholesterol to be used in later assessment of adherence to the estrogenic treatment groups. This procedure was discontinued after such sets of data were obtained on 1038 men, including 354 men randomized to the placebo group. The latter group forms the basis of this report. Laboratory Methods Total cholesterol was determined by the AutoAnalyzer N-24a method modified to give results comparable to the method of Abell et al Serum HDL cholesterol was determined by the method of Walton and Scott,'3 which was modified to permit measurement of cholesterol by the AutoAnalyzer on the supernatant obtained by precipitation of low-density lipoproteins by a reagent composed of dextran sulfate, barbital and calcium chloride. Serum triglyceride was determined on the AutoAnalyzer using the chromotropic acid reaction with a silicic acid chloroform extract of serum Univariate and multivariate linear regression analyses were used to determine the relationship of baseline HDL cholesterol levels to 5-year mortality in the placebo group, both unadjusted and adjusted for 40 base-