ST-segment depression is an electrocardiographicproxy for myocardial ischemia. When ST depression isproduced by exercise or other forms of stress, it caneffectively diagnose the presence of coronary artery dis-ease. In situations where the diagnosis of coronaryartery disease is established, the appearance of STdepression (ischemia) on the exercise electrocardiogram(ECG) may assist in the treatment of the patient.

Previous reports1-10 have confirmed that ST depressionduring exercise is not a reliable indicator of ischemia inthe presence of bundle branch block, left ventricularhypertrophy, and digitalis administration. Even when

these factors are excluded, ST-segment depressionappears often (10%) on the resting ECG in patients withdocumented coronary artery disease.11 It is not certain, inthis situation, whether additional ST depression during anexercise treadmill test (ETT) indicates ischemia.

Interpretation of stress-induced ST-segment depres-sion in patients with known coronary disease can bean important factor in risk stratification and overallmanagement if we have confidence that it representsischemia. Using the thallium perfusion scan as the“gold standard” for ischemia, we undertook this studyto assess the reliability of exercise ST-segment depres-sion as an indicator of ischemia in patients with restingST depression who have been diagnosed with coro-nary artery disease.

MethodsPatient recruitment

We analyzed prospectively collected data from theMulticenter Study of Myocardial Ischemia (MSMI). The MSMI

Ability of the exercise electrocardiogram test todetect ischemia in stable coronary artery diseasepatients with ST-segment depression on the restingelectrocardiogram Vijay G. Kalaria, MD,a and Edward M. Dwyer, MDb Rochester, N.Y., and Newark, N.J.

Background The appearance of ST-segment depression on the exercise electrocardiogram (ETT) generally indicatesmyocardial ischemia. Disorders that produce ST-segment depression on the resting electrocardiogram can confound theresults of the ETT. The purpose of this study was to assess the reliability of exercise ST-segment depression as an indicator ofischemia in patients with a resting electrocardiographic ST depression, presumably caused by coronary disease.

Methods We studied 882 patients 1 to 6 months after a hospitalization for myocardial infarction or unstable anginaas part of the Multicenter Study of Myocardial Ischemia. An ETT and thallium-201 perfusion scan were performed in allpatients. For the purpose of this study, the perfusion scan was considered the “gold standard” for ischemia. We correlatedthe ETT ST-segment responses with the perfusion scan in a group with a normal electrocardiogram baseline ST segment(NO ST DEP group) and in a group with resting electrocardiographic ST depression ≥0.5 mm (ST DEP group).

Results Although the frequency of ischemia by perfusion scan was similar in the ST DEP group (48%) to that in the NOST DEP group (40%), the ST DEP group had a significantly higher (p = 0.03) sensitivity (54%) in detecting ischemia than theNO ST DEP group (35%). The ST DEP group, however, had a lower (58% vs 74%) specificity than the NO ST DEP group(p = 0.056). Baseline characteristics of the two groups probably account for these differences. The overall diagnostic accuracyis similar (58% vs 56%) in the two groups.

Conclusion We conclude that in a stable population with known coronary disease, the presence of ST depression onthe resting electrocardiogram does not impair the detection of ischemia by the ETT, but may be associated with a higherfalse-positive rate.(Am Heart J 1998;135:901-6.)

From the aDivision of Cardiology, University of Rochester,and the bDivision ofCardiovascular Diseases, Department of Medicine, New Jersey Medical School andUniversity Hospital, University of Medicine and Dentistry of New Jersey.Submitted Aug. 1, 1997; accepted Nov. 6, 1997.Reprint requests: Edward M. Dwyer, MD, Director of Division of CardiovascularDiseases, MSB 1-538, UMDNJ-New Jersey Medical School, 185 South OrangeAve., Newark, NJ 07103.Copyright © 1998 by Mosby, Inc.0002-87-3/98/$5.00 + 0 4/1/87841

block; atrial fibrillation; implanted pacemakers; participationin other studies; psychologic factors; or physician refusal. Atotal of 2096 patients met the eligibility criteria from which936 (45%) gave informed consent. For each patient on theday of enrollment, a complete medical history was recordedand an exercise tolerance test with planar thallium-201scintigraphy was performed.

Data acquisitionAll the patients with a resting ST-segment depression of 0.5

mm or more in one of the three monitored leads (AVF, V2,and V5) were included in the resting ST-segment depressiongroup (ST DEP). The remainder of the patients comprised thegroup without resting ST-segment depression (NO ST DEP).

Exercise testing protocolExercise testing was performed by use of the Sheffield modi-

fication of the Bruce protocol with a 3-minute “stage 1/2”before the Bruce stage I. An exercise duration of 6 minutes wasthus equivalent to completion of Bruce stage I with an estimat-ed energy expenditure of 5 metabolic equivalents. Exercise wasterminated when the target heart rate (220 beats/min minusage in years) was reached or for severe fatigue or dyspnea,severe angina, hypertension, or ventricular tachycardia.

ECG leads aVF, V2, and V5 were monitored in all patients,but 12 leads were used in most. The monitored leads, bloodpressure, and heart rate were recorded with the patient inthe supine position, while standing, after hyperventilation,every 3 minutes during exercise, and immediately after and 2minutes after stopping. At peak exercise, 2.5 mCi of thallium-201 was injected intravenously, and the patient exercised foran additional minute. Planar myocardial imaging was startedwithin 5 minutes after the injection. Antiischemic medica-tions, including β-adrenergic blocking agents, were withheldthe day of the test. Interpretation of all exercise tests wascoordinated through the core laboratory at Jewish Hospital,St. Louis, Mo.

The ST segment was analyzed visually, assisted by magnifi-cation. The ST-segment location was expressed to the nearest0.5 mm with a precision of 0.2 mm. The reference point wasthe PQ segment. For both groups, a positive exercise test wasdefined as horizontal or down sloping ST depression of 1 mm(0.1 mV) or more additional ST depression in any of the threeleads (aVF, V2, or V5) during or after exercise measured at 0.08seconds after the J point. Because there were few patients ineither group who had more than 2 mm additional ST-segmentdepression (16% to 18%), we elected not to carry out a similaranalysis, using the criteria of 2 mm additional ST-segmentdepression for a positive stress test. There was excellent agree-ment on repeat analysis (n = 62, kappa 0.81).

Planar thallium-201 perfusion scanning protocolAcquisition and analysis were performed as previously

described.12 Images were evaluated in the core laboratory at

trial was designed to study the prognostic significance ofischemia detected by noninvasive testing in ambulatorypatients with stable coronary artery disease. Details of patientrecruitment, data acquisition, and core laboratories havebeen previously published.12 The study was approved in allinstitutions by each institutional committee on humanresearch, and informed consent was obtained in all patients.

Briefly, 936 patients were enrolled from 12 centers in theUnited States, Canada, and Israel (see Appendix) betweenJuly 1988 and May 1991, 1 to 6 (mean ± SD = 2.7 ± 1.4)months after their index coronary event. Patients of eithersex who were 21 years of age or older and had been admit-ted to a coronary care unit of the participating hospitalswith a documented myocardial infarction or unstable angi-na, with transient ST-T wave ECG changes, were eligible forenrollment. Exclusion criteria included the following: coro-nary bypass graft surgery after the index coronary event;angioplasty performed less than 1 month before enrollment;major comorbidity such as a malignancy or severe renal,hepatic, or cerebral disease; a musculoskeletal disorder thatwould prevent the patient from performing the exercisetest; use of cardiac glycosides or other drugs likely toinduce ST-segment changes on the ECG; bundle branch

American Heart JournalMay 1998Kalaria and Dwyer902

NO ST DEP ST DEPGroup Group

Clinical (n = 809) (n = 73) pcharacteristic (%) (%) Value

Clinical featuresMean age (yr) 57 ± 10 63 ± 11 0.0001Age > 60 yr 41 65 0.0001Male 77 73 NSPrior MI 18 29 0.02Hypertension 40 62 0.0001Diabetes (insulin dependent) 6 5 NSTobacco use 13 25 0.002Angina pectoris* 41 56 0.009CHF history 2 4 0.002

CCU data (at index event)Event typeUnstable angina 30 38 NSQ-wave MI 45 35 NS

Non-Q wave MI 25 27 NSThrombolytic treatment 33 15 0.001Peak CPK (>1000) 36 29 NSRales more than bibasilar 28 40 0.03

Treatment at enrollmentβ-Blockers 51 54 NSNitrates 42 52 NSDiuretics 10 20 0.005

NO ST DEP, Patient group with normal ST segment on resting ECG; ST DEP, patientgroup with ST-segment depression on resting ECG; CCU, coronary care unit; MI,

myocardial infarction; CPK, creatine phosphokinase; CHF, congestive heart failure.*In month before enrollment.

Table I. Baseline clinical characteristics of patients with andwithout resting ST-segment depression

Yale University. Final interpretation was based on quantita-tive circumferential profiles with visual overreads.13 Studieswere characterized as (1) normal, (2) with reversible defects(ischemia), (3) fixed defects (scar), or (4) both (scar plusischemia). Ischemia was evaluated as present or absent.Reread reliability as measured by the kappa statistic of 0.84in 41 patients was excellent.

Routine clinical information was missing in 0.2% of enrolledpatients, thallium-201 perfusion data in 4.3%, and exercisedata in 2.1%. Overall, 882 patients had complete exercise andperfusion data.

Statistical methodsBaseline clinical characteristics and ischemic variables

were compared between those patients with resting ST-segment depression and those without resting ST-segmentdepression by use of the chi-square and Fisher exact test(for dichotomous variables) or t tests (for continuous vari-ables). All the results are expressed as mean ± SD unlessotherwise specified. All the results are analyzed with use oftwo-tailed p values. Differences in sensitivity and specifici-ty of the exercise test were also expressed with 95% confi-dence limits (CL).

ResultsThere were 73 (8.3%) patients with 0.5 mm or more

ST-segment depression on the resting 12-lead ECGdone at the beginning of the exercise test. The mag-nitude of resting ST-segment depression was as fol-lows: 17 patients had 0.5 mm, 27 had 1 mm, 11 had1.5 mm, 13 had 2 mm, five had 2.5 mm, five had 3mm, and one patient had 4 mm ST-segment depres-sion. There were 809 patients (91.7%) with no ST-segment depression on the resting ECG.

Baseline characteristicsClinical data. Several significant differences among

these two groups are noteworthy. Patients with restingST-segment depression were older and had a signifi-cantly higher incidence of a prior myocardial infarction,history of hypertension, angina, congestive heart failure,and smoking (Table I).

CCU observations. Thrombolytic treatment was lesscommon in the ST DEP group (p < 0.001), whereaspresence of rales was more common (p = 0.03). In theST DEP group 62% had a myocardial infarction and 38%had unstable angina as their index event. In the NO STDEP group, 70% of the patients were admitted with amyocardial infarction as their index event and 30% withunstable angina (NS). The type and size of the infarctwere similar in the two groups.

Exercise tolerance test and thallium-201 perfusionscintigraphy

Exercise performance. The maximum heart rateachieved at peak exercise was 128 ± 20 beats/min in theST DEP group and 137 ± 23 beats/min in the NO STDEP group (p = NS). Similarly, the maximum systolicblood pressure (165 ± 27 vs 166 ± 30 mm Hg) andmaximum diastolic blood pressure (83 ± 12 vs 83 ± 13mm Hg) were not different (Table II).

Reason for termination. Fatigue was the most com-mon reason (51%) for the termination of the ETT inboth ST DEP and NO ST DEP groups. Severe angina(10% in both), shortness of breath (16% vs 23%), andreaching the target heart rate (16% vs 10%) were theother causes for the termination of the ETT. None ofthese were statistically different.

Thallium perfusion scan findings. We found evidenceof ischemia with a reversible defect on the thallium-201scan in 362 (41%) patients. Thirty-five (48%) of thepatients in the ST DEP group and 327 (40%) of patients inthe NO ST DEP group had a positive scan (p = 0.2).

Exercise ECG changes. With the criteria of 1 mm ormore additional ST-segment depression during exerciseor recovery, a total of 275 (31%) patients had a positiveexercise test. Abnormal ST-segment depression duringexercise or recovery occurred in 35 (48%) patients in

American Heart JournalVolume 135, Number 5, Part 1 Kalaria and Dwyer 903

Exercise No ST DEP ST DEP test Group Group

variable (n = 809) (n = 73) p Value

Exercise duration (min) 10.4 ± 3.1 8.5 ± 3.2 NSMaximum heart rate 137 ± 23 128 ± 20 NS(beats/min)

Maximum SBP (mm Hg) 165 ± 27 166 ± 30 NSMaximum DBP (mm Hg) 137 ± 23 128 ± 20 NSTermination cause (%)

Fatigue 51 51 NSAngina 10 10 NSDyspnea 16 23 NSTarget heart rate 16 10 NSOthers* 7 6 NS

Positive ETT (%)† 30 48 0.002Positive thallium-201 (%)‡ 40 48 NS

ETT, Exercise treadmill test; SBP, systolic blood pressure; DBP, diastolic blood pressure;NO ST DEP, patient group with normal ST segment on resting ECG; ST DEP, patientgroup with ST-segment depression on resting ECG.*Other reasons for termination of ETT were SBP >250 or <90 mm of Hg, dizziness, STdepression >5 mm.†Positive ETT: additional 1 mm or more horizontal or downsloping ST-segment depres-sion (measured at 80 msec from the J point) during exercise or recovery.‡A positive thallium-201 scan requires a reversible defect.

Table II. Exercise treadmill test data

the ST DEP group and 240 (30%) patients in the NO STDEP group (p = 0.002).

The magnitude of the additional ST-segment depres-sion in patients who had a positive exercise test by theabove criteria was distributed in both the groups as fol-lows: in the ST DEP group, 31 (82%) patients had 1 to1.9 mm, five (13%) had 2 to 2.9 mm, and two (5%)patients had ≥3 mm additional ST-segment depression.In a very similar distribution, the NO ST DEP group,206 (84%) had 1 to 1.9 mm, 31 (13%) had 2 to 2.9 mm,and nine (3%) patients had ≥3 mm additional ST-seg-ment depression. The distribution of the magnitude ofadditional ST-segment depression for a positive exercisetest was not significantly different between the twogroups.

ETT as an indicator of ischemiaIn the patients with baseline ECG ST depression, sen-

sitivity to detect ischemia was 54% (CL = 38 to 71) andthe specificity was 57% (CL = 42 to 74). In the patientswith normal baseline ECG, the sensitivity was 35% (CL= 29 to 40) and the specificity was 74% (CL = 70 to 78)(Table III).

The differences in sensitivity (54% vs 35%) betweenthe two groups was significantly different (p = 0.03).The differences in specificity (74% vs 57%) were mar-ginally different (p = 0.056).

The overall diagnostic accuracy (number of correctlyclassified patients/total number of patients) was similar:58% in the NO ST DEP group and 56% in the ST DEPgroup.

DiscussionIn earlier studies where the primary purpose of the

ETT was to diagnose the presence of coronary arterydisease, ST-segment depression on the resting ECGcaused by bundle branch block,5 digitalis administra-

tion,4 or left ventricular hypertrophy3 have been shownto confound the interpretation of ischemia. In patientswith known coronary artery disease, who have neithera bundle branch block, left ventricular hypertrophy, norare receiving digitalis, ST-segment depression on theresting ECG still occurs frequently. The basis for theresting ECG ST changes are not clear. A previous studyusing this same patient cohort11 demonstrated a relationof the resting ECG ST depression with hypertension,older age, elevated left ventricular end-diastolic pres-sure, and multivessel disease. It is possible that the rest-ing ECG ST changes could represent regions of chronicischemia, residual non-Q wave infarction, or left ven-tricular hypertrophy.

The purpose of this study was to evaluate the abilityof the ETT to detect the presence or absence ofischemia in patients with known stable coronary arterydisease when the ST segment of the resting ECG wasabnormally depressed. Several aspects of the design ofthe study are worth comment here. The patient cohortwas taken from the database of the Multicenter Study ofMyocardial Ischemia. The likelihood of the diagnosis ofcoronary artery disease is very high inasmuch as partici-pants were required to have a documented myocardialinfarction or unstable angina with transient ST or Twave changes. Two thirds of the patients had coronaryangiography before entry into the trial. All patients wereambulatory and stable. A resting ECG showed STdepression in 8% (n = 75) of the patients who qualifiedfor this analysis.

Patients with most of the usual factors that produceresting ECG ST-segment depression were excludedfrom the trial. There was no provision in the originalselection of patients to diagnose and exclude left ven-tricular hypertrophy. Many of the ECG criteria, otherthan voltage, that are commonly used to support thediagnosis of left ventricular hypertrophy (i.e., delayed

American Heart JournalMay 1998Kalaria and Dwyer904

NO ST DEP group ST DEP group*(n = 809)* (n = 73)

Variable n % CL n % CL p Value

Sensitivity 113/327 35 29 to 40 19/35 54 38 to 71 0.03Specificity 355/482 74 70 to 78 22/38 57 42 to 74 0.056Diagnostic accuracy† 468/809 58 55 to 61 91/73 56 45 to 68 NS

NO ST DEP, Patient group with normal ST segment on resting ECG; ST DEP, patient group with ST-segment depression on resting ECG; CL, 95% confidence limits.*NO ST DEP group had 327 (40%) positive thallium scans and ST DEP group had 35 (48%) positive thallium scans.†Diagnostic accuracy = number of correctly classified patients/total number of patients studied.

Table III. Detection of ischemia by ECG exercise test

ventricular conduction, ST depression, T wave inver-sion, left atrial abnormality, and poor R wave progres-sion), are common sequelae of ischemic heart diseaseas well. This would have resulted in frequent misclassi-fication. Therefore this subgroup is included in thisanalysis. Studies addressing this issue are consider-able14,15 and have concluded that inclusion of left ven-tricular hypertrophy moderately reduces both sensitivityand specificity.

A reversible perfusion defect on thallium-201 planarscintigraphy was used in this study to identify the pres-ence of exercise-induced ischemia. The ETT ST-seg-ment changes of patients with a normal baseline ECGand the group with abnormal (depressed ST segment)ECG were both compared to their respective perfusionscan results. Our intent was to determine if the groupwith ST depression on the resting ECG performed aswell in the detection of ischemia as the normal baselineECG group.

The baseline characteristics were different betweenthe two groups. Patients with ST depression on the rest-ing ECG were older, more hypertensive, had more angi-na before trial entry, had less frequent thrombolysis,and had greater incidence of congestive heart failure.The performance on the treadmill and symptomaticresponse to the exercise were similar between the twogroups. However, a positive (ST-segment depression ≥1mm greater than baseline) ETT was more common(48% vs 30%) in the group with the baseline ST-seg-ment depression (ST DEP). This finding coincides withthe higher incidence of an abnormal thallium perfusionscan (48% vs 40%) in the group with the abnormalbaseline ECG.

When the two groups were compared, the sensitivityfor ischemia (54%) in the ST DEP group was significant-ly higher (p = 0.03) than the sensitivity (35%) of the NOST DEP group. The difference in sensitivities probablyreflects the differences in baseline characteristics ofthese groups. The ST DEP group has a higher frequen-cy of hypertension, diabetes, congestive heart failure,and multivessel coronary disease,11 each of which couldcontribute to a more extensive ischemic response onthe ECG. Of some interest, the sensitivities in bothgroups, particularly the NO ST DEP group, were lowerthan values reported15 for exercise tests done for diag-nostic reasons. The reason for this difference is notapparent because our study was not designed toaddress this issue.

Specificity, on the other hand, was marginally (p = 0.056)reduced in the ST DEP group (58%) compared with the

NO ST DEP group (74%). One possible contributor toa lower specificity is the potential greater inclusion ofpatients with left ventricular hypertrophy into the STDEP group. Other confounding factors such as digitalisadministration and conduction abnormalities hadalready been excluded from the study.

We conclude that in a stable population with knowncoronary artery disease, the presence of ST depressionon the resting ECG does not impair the detection ofischemia by an ETT but does carry with it a higherpotential for false-positive studies, possibly related to ahigher incidence of left ventricular hypertrophy in thesepatients.

References1. Ellestad MH. Stress testing: principles and practice. New York: Davis;

1975.2. Wong HO, Kasser IS, Bruce RA. Impaired maximal exercise perfor-

mance with hypertensive cardiovascular disease. Circulation 1969;39:633-6.

3. Harris CN. Treadmill stress test in left ventricular hypertrophy. Chest1973;63:353-7.

4. Kawai C, Hultgren HN. The effects of digitalis upon the exercise elec-trocardiogram. Am Heart J 1964;68:409-12.

5. Surawicz B, Saito S. Exercise testing for detection of myocardialischemia in patients with abnormal electrocardiograms at rest. Am JCardiol 1978;41:943-51.

6. Nasrallah A, Garcia E, Benrey J, Hall RJ. Treadmill exercise testing inthe presence of nonspecific ST-T changes or digitalis effect: correlationwith coronary angiography. Am J Cardiol 1975;35(suppl II):II-160.

7. Meyers DG, Bendon KA, Hankins JH, Stratbucker RA. The effect ofbaseline electrocardiographic abnormalities on the diagnostic accuracyof exercise-induced ST segment changes. Am Heart J 1990;119:272-6.

8. Rotiman D, Jones WB, Sheffield LT. Comparison of submaximal exer-cise ECG test with coronary cineangiogram. Ann Intern Med 1970;72:641-7.

9. Miranda CP, Lehman KG, Forelicher VF. Correlation between restingST segment, depression, exercise testing, coronary angiography andlong-term prognosis. Am Heart J 1991;122:1617-28.

10. Kansal S, Rotiman D, Sheffield LT. Stress testing with ST-segmentdepression at rest. Circulation 1976;54:636-9.

11. Dwyer EM Jr, Case RB, Gillespie JA, Greenberg HM, Krone RJ,Lichstein E,et al. Adverse prognosis of ST depression of the restingelectrocardiogram in patients with stable coronary artery disease: ananalysis of the basis for this association. Ann Noninvas Electrocardiogr1996;1:54-62.

12. Moss AJ, Goldstein RE, Hall WJ, Bigger JT, Fleiss J, Greenberg H, etal. JAMA 1993;269:2379-85.

13. Wackers FJTh, Bodenheimer M, Fleiss JL, Brown M, and the MSSMIThallium-201 Investigators. Factors affecting uniformity in interpretationof planar thallium-201 imaging in a multicenter trial. J Am Coll Cardiol1993;21:1064-74.

14. Detrano R, Gianrossi R, Froelicher V. The diagnostic accuracy of theexercise electrocardiogram. A metaanalysis of 22 years of research.Prog Cardiovasc Dis 1989;32:173-206.

15. ACC/AHA Guidelines for Exercise Testing. J Am Coll Cardiol 1997;30:260-315.

American Heart JournalVolume 135, Number 5, Part 1 Kalaria and Dwyer 905

AppendixExecutive Committee for Substudy of the Multicenter

Study of Myocardial Ischemia: A.J. Moss, MD, W.J. Hall,MD, R. Raubertas, MD, M.W. Brown, MD, University ofRochester, Rochester, N.Y.; R.E. Goldstein, MD,Uniformed Services University of Health Sciences,Bethesda, Md.; J.T. Bigger, Jr., MD, J.L. Fleiss, MD, J.Coromilas, MD, College of Physicians and Surgeons,Columbia University, New York, N.Y.; H. Greenberg,MD, R. Case, MD, E.M. Dwyer, MD, Roosevelt-St. LukesHospital, New York, N.Y.; M. Bodenheimer, MD, LongIsland Jewish Hospital, New Hyde Park, N.Y.; R.J.

Krone, MD, R. Kleiger, MD, Jewish Hospital, Saint Louis,Mo.; F.I. Marcus, MD, University of Arizona HealthScience Center, Tucson, Ariz.; F.J. Th. Wackers, MD,Yale University School of Medicine, New Haven, Conn.;J. Benhorin, MD, S. Stern, MD, D. Tzivoni, MD, BikurCholim Hospital, Jerusalem, Israel; J.A. Gillespie,Highland Hospital, Rochester, N.Y.; J.J. Gregory, MD,Overlook Hospital, Summit, N.J.; E. Lichstein, MD,Maimonides Medical Center, Brooklyn, N.Y.; J. O.Parker, Kingston General Hospital, Kingston, Ontario,Canada; L. Van Voorhees, MD, Washington HospitalCenter, Washington, D.C.

American Heart JournalMay 1998Kalaria and Dwyer906