A Multivariate Approachfor Interpreting Treadmill Exercise Tests

in Coronary Artery Disease

JULIAN L. BERMAN, M.D., JOSHUA WYNNE, M.D, AND PETER F. COHN, M.D.

SUMMARY To determine the value of a multivariate approach for the analysis of the treadmill exercise tolerance test(ETT), 237 patients referred for evaluation of chest pain who underwent a standard Bruce protocol ETT and coronary

arteriography were studied. Predictive value of a positive ETT was 0.78 (43/55) using 1.0-1.9 mm ST segment depressioncriterion, 0.97 (59/61) using > 2.0 mm ST segment depression. When the 1.0-1.9 mm ST criterion was combined with peaksystolic blood pressure-heart rate product (double product) < 23,000, exercise duration less than 6 minutes, and ST depres-sion for greater than 3 minutes into recovery, predictive value improved to 0.89 in 18 patients with any two of the above.Predictive value for multivessel disease was also improved using non-ST criteria. Predictive value of a negative ETT forabsence of coronary artery disease was 0.60 (29/48), and was 0.86 (12/14) if double product was > 30,000. Presence ofchest pain during ETT did not improve predictive value of any type of test. Digitalis ingestion in 33 patients was not associ-ated with decreased predictive value of a positive test. These data suggest that the predictive value of both positive andnegative ETT in a symptomatic population can be improved with a multivariate approach.

THE DEFINITIVE DIAGNOSTIC TEST forevaluation of chest pain syndromes is the coronaryarteriogram. Because this is an in-hospital procedurewith a small but definite morbidity and mortality,many investigators have attempted to identify factorsin the history and noninvasive evaluation of patientswith chest pain that would identify those patients witha high likelihood of having coronary artery disease.'The exercise test is the most widely used of theavailable noninvasive procedures for identifyingpatients with probable coronary artery disease. It issafe, not difficult to perform, and relatively inexpen-sive. Evaluation of ECG changes, specifically ST seg-ment depression during or after exercise, has been thetraditional criterion for evaluating the exercisetolerance test. However, problems with false positiveand negative tests, especially in patients with atypicalor no angina pectoris,2 6 have stimulated efforts to ex-

amine other aspects of the exercise response besidesthe presence or absence of exercise-induced ST seg-ment depression in order to increase the diagnosticutility of the test.7-" In this study, we have attemptedto supplement the traditional ST segment criterion byanalyzing the importance of the following additionalvariables in the interpretation of the exercise toler-ance test: 1) anginal chest pain induced by exercise; 2)heart rate-blood pressure product (double product) atpeak exercise; 3) persistence of ST segment depressionduring recovery from exercise; and 4) depth of STdepression.

From the Cardiovascular Division, Department of Medicine,Peter Bent Brigham Hospital and Harvard Medical School, Boston,Massachusetts.

Supported by USPHS Grant HL-7049.Presented in part at the 27th Annual Scientific Sessions,

American College of Cardiology, Anaheim, California, March,1978.Address for reprints: -Peter F. Cohn, M.D., Cardiovascular Divi-

sion, Peter Bent Brigham Hospital, 721 Huntington Avenue,Boston, Massachusetts 02115.

Received April 4, 1978; revision accepted May 24, 1978.

Nlaterials and Methods

The records of all patients undergoing cardiaccatheterization at the Peter Bent Brigham Hospital,Boston, Massachusetts, from January 1, 1975 to July1, 1977 were retrospectively reviewed. Patients un-dergoing catheterization for chest pain syndromeswho had maximal, symptom-limited Bruce protocol1'exercise tolerance tests performed at the Peter BentBrigham Hospital or its outpatient facility wereselected for a further evaluation. Electrocardiogramswith bundle branch block in the resting tracing werenot included in this series. The exercise protocol in-volved recording the supine, upright and post-hyperventilation (60 sec) 12-lead ECG at rest. AllECGs at rest and during and after exercise were takenusing the Mason-Likar modification of the standardECG lead placements.'2 No patients had orthostaticor post-hyperventilation ECG changes. ECGs werethen taken, along with cuff blood pressures, every 3minutes during exercise. Supine ECGs and cuff bloodpressures were taken at 0, 1, 3, 5, and 8 minutes afterexercise, or until ECG returned to baseline. A positivetest was defined as one showing 1 mm of ST segmentelevation or depression (of any morphology and in anylead except aVR) 80 msec after the end of the QRScomplex any time after exercise was begun. If therewas resting ST segment depression or elevation, a 1mm increase was required for a positive test. Anegative test was one without this degree of ST seg-ment change and in which the patient achieved 85% ofthe maximum predicted heart rate for his age. Allother tests were considered nondiagnostic.The peak blood pressure was usually measured just

before termination of exercise, before the patientreturned to the supine position. Presence or absence ofexercise-induced angina (defined as either typicalanginal distress or the patient's usual anginalequivalent) was recorded, as were peak blood pres-sure, heart rate and depth and duration of ST segmentchange during recovery. No systematic attempt wasmade to measure ST segment changes during exercise,

505

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

VOL 58, No 3, SEPTEMBER 1978

TABLE: 1. Predictive Value of Positive Exercise Tolerance Tests for Coronary Artery Disease and Multivessel DiseasePredictive Value for CAD Predictive Value for MVD

All ST; ST; < 2mm ST; >2mm All ST; ST; < 2mm ST i > 2mm

All tests 0.88 (102 0.78 (43 0.97 (59 0.71 ( 83 0.56 (31 0.85 (52(n =116) 116) 55) 61) 116) 505) 61)

Peak double product* 0.94 (70 0.86 (24 1.0 (46 0.75 (63 0.71 (20 0.93 (43(n =74) 74) 28) 46) 84) 28) 46)

Duration of exercise < 6 min. 0.92 (48 0.80 (16 1.0 (32 0.85 (44 0.60 (12 1.0 (32(Stage I-II) (n = 52) 52) 20) 32) 52) 2) 32)

STI persisting > 3 minutes into 0.94 (73 0.89 (24 0.96 (49 0.71 (56 0.48 (13 0.84 (43recovery (n = 78) 78) 27) 51) 78) 27) 51)

Peak double product* and duration of 0.93 (38 0.79 (11 1.0 (27 0.88 (36 0.64 ( 9 1.0 (27exercise < 6 min (Stage I-IH) (n = 41) 41) 14) 27) 41) 14) 27)

Peak double product* and ST[ persisting 0.96 (49 0.87 (13 1.0 (36 0.90 (46 0.67 (10 1.0 (36> 3 minutes into recovery (n = 51) 51) 15) 36) 51) 15) 36)

Duration of exercise < 6 min (Stage I-II) 0.97 (30 0.88 (7 1.0 (23 0.90 (28 0.62 (5 1.0 (23and STI persisting > 3 minutes into 31) 8) 23) 31) 8) 23)recovery (n = 31)

Two of above criteria 0.97 (59 0.89 (16 1.0 (43 0.90 (55 0.78 (14 0.95 (41(n =61) 61) 18) 43) 61) 18) 43)

All three of above criteria 0.96 (27 0.83 (5 1.0 (22 0.88 (24 0.33 (2 1.0 (22(n= 28) 28) 6) 22) 28) 6) 22)

One or more of above criteria 0.90 ( 92 0.82 (36 0.97 (56 0.82 ( 84 0.66 (29 0.95 (55(n = 102) 102) 44) 58) 102) 44) 58)

Two or three of above criteria 0.93 (68 0.82 (23 1.0 (45 0.74 (54 0.50 (14 0.89 (40(n = 23) 73) 28) 45) 73) 28) )* = > 23,000.Abbreviations: ETT = exercise tolerance test; CAD = coronary artery disease; MVD = multivessel disease.

due to the frequency of baseline wandering and muscleartifact.Coronary arteriograms were interpreted by three

staff cardiologists and cardiovascular radiologists, andthe consensus of their findings as recorded on the finalcatheterization report was used to determine thenumber of obstructed coronary arteries. Seventy per-cent luminal diameter narrowing was consideredobstructive. Involvement of four possible vessels wasconsidered: 1) the main left coronary artery, con-sidered equivalent to two vessels if obstructed; 2) theleft anterior descending and/or any of its diagonals orramus medianus branch; 3) the left circumflex and/orany of its obtuse marginal branches; and 4) the rightcoronary, posterior descending, and/or any of itsacute marginal branches. Coronary arteriograms wereusually performed within 48 hours, but never morethan several weeks, of the exercise test.The presence and number of obstructed coronary

arteries was compared with the exercise data. Thepredictive value of a positive exercise tolerance test isdefined as the percentage of the positive testsassociated with at least one significantly obstructedcoronary artery (true positive/all positive), and that ofa negative exercise tolerance test is defined as thepercentage without at least one significantlyobstructed artery (true negative/all negative). Thepredictive value for multivessel coronary artery dis-ease is defined in a similar manner, with the require-ment of at least two significantly obstructed coronary

arteries for positive tests and no or one significantlyobstructed artery for negative tests.

Results

Sixty-eight patients (31 men, 37 women) had noevidence of obstructive coronary artery disease, 45 (32men, 13 women) had one vessel obstruction, 59 (51men, eight women) had two vessel obstruction, 65 (57men, eight women) had three vessel obstruction, andseven (five men, two women) had left main coronaryartery obstruction. Of a total of 204 exercise toler-ance tests in patients not on digitalis glycosides, 116(89 men, 27 women) were positive, 48 (31 men, 17women) were negative, and 40 (25 men, 15 women)were nondiagnostic. Unless otherwise stated, none ofthe analyses included patients taking digitalis.

Predictive Value of Positive Exercise Tolerance Tests forCoronary Artery Disease and Multivessel Coronary Disease(table 1)

The predictive value of a positive exercise tolerancetest for coronary disease was 0.88 (102/116) overall,0.78 (43/55) if ST depression was less than 2 mm (lesspositive exercise tolerance test), 0.97 (59/61) if it wasgreater than or equal to 2 mm (markedly positive ex-ercise tolerance test*). For multivessel disease, the

*The eight patients with ST segment elevation were also con-sidered to have markedly positive tests, and are included in this sub-group from this point on.

CIRCULATION506

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

TREADMILL EXERCISE TESTS AND CAD/Berman et al.

TABLE 2. Influence of Exercise Induced Anginal Chest Pain on Predictive Value of Positive Exercise ToleranceTest

Predictive value for CADST1 < 2mm ST5 > 2mm

Pain

No pain

Peak double product> 23,000

Pain

No pain

Duration of exercise< 6 min. (Stage 1-11)

Pain

No pain

STI persisting > 3minutes into recovery

Pain

No pain

One or more of above Pain

No pain

0.8 (2227)

0.64 (1828)

0.75 (1216)

0.75 ( 912)

0.67 (69)

0.70 ( 710)

0.77 (1013)

0.92 (1112)

0.85 (2226)

0.78 (1418)

1.0 (3030)

0.94 (2931)

1.0 (2424)

1.0 (88)

1.0 (1717)

1.0 (1515)

1.0 (2929)

0.95 (2021)

1.0 (2927)

0.97 (27.28)

Predictive value for MVDST;7 < 2mm ST;7 . 2mm

0.52 (14 0.87 (2627) 30)

0.32 ( 928)

0.5 ( 816)

0.58 ( 71-2)

0.44 (49)

0.40 ( 410)

0.81 (2531)

1.0 (2424)

0.75 (68)

1.0 (1717)

1.0 (1515)

0.23 ( 3 0.93 (2713) 29)

0.83 (1012)

0.76 (1621)

0.42 (11 0.97 (2826) 29)

0.56 (1018)

0.82 (2328)

Two or three of above Pain

No pain

0.71 (10 1.0 (2814) 28)

0.93 (13 1.0 (1714) 17)

0.50( 7 1.0 (2814) 28)

0.79 (1114)

1.0 (2828)

All three of above Pain

No pain

0.50 (24)

1.0 (22)

1.0 (1212)

1.0 (1010)

0.25 (1 1.0 (124) 12)

Abbreviations: ETT = exercise tolerance test; CAD = coronary artery disease; MVD = multivessel disease.

figures are 0.71 (83/116), 0.56 (31/55), and 0.85(52/61), respectively. By empirical observation, peaksystolic blood pressure-heart rate product (doubleproduct) < 23,000, exercise duration < 6 minutes,and persistence of ST segment change for > 3 minutesinto recovery provided the best separation of truepositive from false positive tests. Using theseparameters, predictive value of the less positive exer-cise tolerance tests could be raised to 0.89 (16/18).The predictive value of positive tests for multivesseldisease improved from 0.56 to 0.78 (14/18) for lesspositive, and from 0.85 to 1.0 (36/36) for markedlypositive exercise tolerance tests when the non-ST seg-ment parameters were added. This last change fromthe overall predictive value achieved statisticalsignificance at the P < 0.05 level using the chi squaretest. However, the increases in predictive value wereachieved in progressively smaller subsets of the entirepopulation.

Influence of Chest Pain on Positive Exercise Tolerance Tests(table 2)

Exercise-induced angina did not appear to be usefulin increasing the overall predictive value of less

positive exercise tolerance tests, nor of markedlypositive exercise tolerance tests. The combination ofchest pain with the exercise parameters (doubleproduct, exercise duration, or the persistence of STsegment change) did not enhance the improved predic-tive value seen with combinations of these parametersalone, either for coronary artery disease or for mul-tivessel coronary artery disease.

Hypotension with Exercise

Eleven patients had a decrease of blood pressure atpeak exercise (mean 10 mm Hg, range 5-22 mm Hg).None of these patients exercised for more than 6minutes. Ten had coronary artery disease, seven hadelectrocardiographically positive tests, and nine hadtwo vessel, three vessel or left main coronary arteryobstruction.

Predictive Value of Negative Exercise Tolerance Test forAbsence of Coronary and Multivessel Disease

Tables 3 and 4 summarize the predictive value ofnegative exercise tolerance tests for the absence ofsignificant coronary artery disease and for the absenceof multivessel coronary artery disease, respectively.

All tests

1.0 (22)

1.0 (10)10)

507

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

VOL 58, No 3, SEPTEMBER 1978

TABLE 3. Predictive Values of Negative Exercise Tolerance Tests for the Absence of Significant Coronary Ob-structive Lesions

Double productExercise duration All > 24,000 > 26,000 > 28,000 > 30,000

All (48) 0.60 (29 0.71 (27 0.74 (26 0.70 (19 0.86 (1248) 38) 35) 27) 14)

> 6 min (42) 0.57 (24 0.58 (21 0.61 (20 0.63 (17 0.86 (1242) 36) 33) 27) 14)

> 8 min (27) 0.56 (15 0.59 (14 0.59 (14 0.63 (13 0.73 ( 827) 24) 24) 23) 11)

> 10 min (16) 0.69 (11 0.67 (10 0.67 (10 0.73 ( 8 1.00 (616) 15) 15) 11) 6)

Predictive value for the absence of overall coronaryartery disease increased from its overall value of 0.60(29/48) to 0.86 (12/14) if a double product of greaterthan 30,000 was achieved, and for multivessel coro-nary artery disease from 0.81 (39/48) to 1.0 (14/14).No notable increase in predictive value was associatedwith increased exercise duration, and increased exer-cise duration did not appear to enhance the improvedpredictive power of exercise tolerance tests with higherdouble product (tables 5 and 6).

Anginal chest pain occurred in only seven negativeexercise tolerance tests too few to assess its value indiagnosis. The increased predictive value with greaterdouble product was essentially unchanged in the groupwithout angina during exercise, as was the lack of ad-ditional predictive influence of the exercise duration.

Nondiagnostic Exercise Tolerance Tests

In the patients with inadequate heart rate response,predictive value for coronary artery disease was 0.52(21/40). No combination of exercise parameters andchest pain improved this lack of separation of normalpatients from those with significant coronary arterydisease.

Correlation of Exercise Parameters with the Extent ofCoronary Disease

Table 7 shows the mean ± SEM for double productand exercise duration for all patients whose exercisewas not limited by extracardiac problems. Patientswith coronary disease and multivessel disease who hadpositive exercise tolerance tests had significantly lowerdouble products at peak exercise than those with no

coronary disease (P < 0.02, unpaired t test). Patientswith multivessel disease and positive exercisetolerance tests also had significantly lower peak dou-ble product than those with one vessel disease (P< 0.01, unpaired t test), who did not differ signifi-cantly in this parameter from normals.

For exercise duration, the identical trends werenoted, but failed to reach statistical significance (0.1 >P > 0.05, unpaired t test for all similar comparisons).

Male vs Female Patients

For the positive exercise tolerance test, predictivevalue of the markedly positive exercise tolerance testfor coronary disease was high for both sexes: 0.95(53/55) for males, 1.0 (6/6) for females. For mul-tivessel disease, the equivalent figures are 0.84 (46/55)and 1.0 (6/6), respectively. However, the less positiveexercise tolerance test was significantly more predic-tive of coronary disease and multivessel disease inmales. The values are 0.91 (30/33) vs 0.46 (9/19) and0.55 (18/33) vs 0.21 (4/19) (both P < 0.01, chi squaretest). When the values of exercise duration, doubleproduct, and duration of ST segment change (andvarious combinations thereof) (table 1) for the smallgroup of females were tested, no tendency toward in-creased predictive value was noted, but the numbers inany subgroup were quite small.

For the negative exercise tolerance test, there was atendency toward increased predictive value in females,both for absence of coronary artery disease and for theabsence of multivessel coronary artery disease. Thevalues for no coronary disease are 0.71 (12/17) vs 0.55(17/31), and for no multivessel disease 0.88 (15/17)

TABLE 4. Predictive Value of Negative Exercise Tolerance Tests of Multivessel DiseaseDouble product

Exercise duration All > 24,000 > 26,000 > 28,000 > 30,000

All 0.81 (39 0.87 (33 0.91 (32 0.93 (25 1.0 (1448) 38) 35) 27) 14)

> 6 min 0.86 (36 0.89 (32 0.90 (30 0.93 (25 1.0 (1442) 36) 33) 27) 14)

> 8 min 0.85 (23 0.93 (27 0.97 (28 0.94 (26 1.0 (1127) 2 2 9 ) 7) 11)

> 10 min 0.88 (14 0.87 (13 0.93 (14 0.91 (10 1.0 (616) 15) 15) 11) 6)

508 CIRCULATION

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

TREADMILL EXERCISE TESTS AND CAD/Berman et al.

TABLE 5. Influence of Exercise Induced Chest Pain on Predictive Value of Negative Exercise Tolerance Tests(ETT) of Significant Coronary Artery Disease.

Double productAll > 24,000 > 26,000 > 28,000 > 30,000

Negative ETT with 0.43 (3 0.50 (3 0.50 (3 0.50 (2 0.67 (2angina 7) 6) 6) 4) 3)

Negative ETT without 0.63 (26 0.75 (24 0.69 (20 0.74 (17 0.91 (10angina 41) 32) 29) 23) 11)

TABLE 6. Predictive Value of Angina in Negative Exercise Tolerance Tests for Absence of Multivessel DiseaseDouble product

All > 24,000 > 26,000 > 28,000 > 30,000

Angina 0.71 (5 0.67 (4 0.67 (4 0.75 (3 1.0 (37) 6) 6) 4) 3)

No angina 0.83 (34 0.91 (29 0.97 (28 0.96 (22 1.0 (1141 ) 32) 29) 23) 11 )

versus 0.77 (25/31). In those patients with inadequateheart rate response, males were much more likely tohave coronary disease. The predictive value for cor-

onary artery disease in males is 0.64 (16/25), and infemales 0.2 (3/15) (P < 0.01, chi square test).

Digitalis Group

Of the 33 patients taking digitalis, 26 men and sevenwomen aged 42-66, 22 had at least mm increase inresting ST depression (with one false positive test),giving a predictive value for coronary artery disease of0.95. (The false positive result occurred in a patientwith an increase of 2.5 mm over the resting ST seg-ment depression.)

There were four negative exercise tolerance tests(three were false negative) and seven nondiagnostictests, four in patients with coronary artery disease.Two patients with multivessel disease had hypoten-sion. This percentage (6%) is similar to the percentageof hypotension in the non-digitalis patient group(l1/204, 5.4%). The virtual absence of false positiveexercise tolerance tests and the small number ofnegative and nondiagnostic exercise tolerance testsprecluded further analysis.

DiscussionPositive Exercise Tests

The positive treadmill exercise test is highlypredictive of coronary artery disease in a symptomaticpopulation where the disease is prevalent. In a numberof studies using both maximal and submaximal typesof exercise, the predictive value of a positive test was0.88 to 0.96,1' 12-23 while in another24 it was 0.73. Ouroverall predictive value is 0.88. As the frequency of thedisease in the population falls, the false positive testbecomes more of a problem.25 In most studies, a

greater likelihood of a false positive result is seen inthe population with a less positive (< 2 mm STdepression) exercise test.7 However, the parameters ofexercise we evaluated (double product, exercise dura-

tion and duration of ST segment change in recovery),appeared individually and in combination to enhancethe predictive value of this subset of positive tests,both for overall coronary artery disease and for mul-tivessel disease. Although overall predictive values arehigh, some combinations of parameters did achievestatistically significant improvement in predictivevalue for multivessel disease.

Previous studies have also shown a tendency forthe proportion of false positive tests to rise with in-creasing maximum heart rate and double productachieved, 9 14, 22, 26 at least in male patients.'0 Lowheart rates achieved for a given duration of exercisehave been related to increased prevalence of coronarydisease and to subsequent coronary events indepen-dent of ST segment changes.27 28 Hypotensiveresponses to low level of exercise clearly point toischemic left ventricular dysfunction.29 30 In theanalysis of the postoperative coronary bypass graftpatient, increased heart rate and double product withmaximal exercise have been correlated with the ade-quacy of revascularization,31 32 but have also beenseen in patients showing clinical improvement despiteclosure of all grafts.33 Maximum heart rate and double

TABLE 7. Mean Values for Exercise Duration and DoubleProduct for Exercise Tolerance Test Subgroups

Double Exerciseproduct duration

(mm Hg-min-10-) (min)

CAD +Test 20.6 = 0.7* Mean SEM) 6.4 0.3-Test 21.8 t 1.1 7.6 0.4

No CAD +Test 23.8 t 2.2* 8.1 1.3-Test 23.1 - 1.1 7.6 0.4

One vessel +Test 24.0 - 1.6t 7.5 0.6CAD -Test 23.2 1.6 8.7 0.6

MVD + Test 20.0 0.7t- 6.2 0.3-Test 20.7 1.3 7.0 0.6

*P <0.02.tP <0.01.+ = positive test; - = negative test.Abbreviations: CAD = coronary artery disease; MVD = multivessel

disease.

509

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

VOL 58, No 3, SEPTEMBER 1978

product of normal individuals, as well as in those withcoronary artery disease, correlate with maximummyocardial oxygen uptake.34' 3 When the blood flowto the left ventricle is limited, as it is in ischemic heartdisease, myocardial ischemia is likely to occur atlower levels of myocardial oxygen consumption.

For the Bruce protocol, exercise duration alsorelates to ventricular function,28 and in most studies,the prevalence of coronary disease in an ECG positivetest group decreases as exercise duration in-creases.9' 13 14 In one study, which examined falsepositive exercise tests by multivariate analysis, exer-cise duration was not found to be a useful predictor offalse positivity.10

Persistence of ST changes into recovery has alsobeen correlated with increasing prevalence andseverity of coronary lesions.8' 13 In our study, the datasuggest that part of this association may be due to thetendency for the less positive exercise ECGs to nor-malize more quickly and to be associated more oftenwith normal vessels or less severe coronary artery dis-ease.

Negative Exercise Tests

There is a wider variation of predictive valuesfor the absence of significant coronary disease. Inthe studies alluded to previously, the range is0.36-0.92.1' 12-24 However, most of the valuesclustered around 0.60, the predictive value found inthis study. A higher achieved double product corre-lated with increasing predictive value for a negativetest, both for absence of coronary disease and multi-vessel disease. Predictive value was not further im-proved with increased exercise duration, in contrast toimprovement found by other workers.'4 We found adecreased frequency of negative tests in patients withmultivessel disease (also noted by others3' 8 13. 20. 26),but not any particular association of old myocardialinfarction and false negative tests. This associationhas led to the suggestion that the false negative test isoften true negative, infarction having "cured" thepatient's ischemia,32 36. so that coronary disease ex-ists without active ischemia.

In both positive and negative exercise tests, thelower peak double product achieved with multivesselcoronary disease suggests that diffuse cardiacischemia limits exercise capacity (in the absence oflimiting leg ischemia or other extracardiac causes ofexercise limitation). This presumably occurs either bydevelopment of ischemic ST changes with or withoutchest pain, by limitation of cardiac output leading tofatigue, or by increase in left ventricular fillingpressure leading to pulmonary congestion anddyspnea.28 It may be hypothesized that suchlimitations in cardiac function may prevent sufficientlocal ischemia to cause ST changes and/or chest painin some patients.

Contrary to several other studies,20 36, 38 we wereunable to show enhancement of the predictive valuefor coronary disease of positive tests, nor a decrease ofthat for absence of coronary disease of negative tests if

there was exercise-induced chest pain. However, thenumbers were small in most of the subgroups in thisanalysis.

Nondiagnostic Exercise Tolerance Tests

Little has been written about the exercise test whichis nondiagnostic because of low peak heart rate andabsence of ST segment change. It has been suggestedthat, because left ventricular dysfunction may preventsufficient exercise stress to bring out ischemia on theECG, most nondiagnostic tests would occur inpatients with coronary disease.20 27 We found that thenondiagnostic exercise tolerance test predicted neitherpresence nor absence of coronary disease with orwithout the use of non-ST segment exerciseparameters.

Male vs Female Patients

Most of the studies which have dealt with the valueof exercise testing in females as opposed to males,whether of symptomatic6' 22 39 or asymptomatic,presumably normal females,40 have shown decreasedpredictive value of the positive exercise test in femalesand some increased predictive value of the negativetest. One study came to different conclusions,41 butour data agrees with the other studies cited.Interestingly, in the analysis of the nondiagnostic ex-ercise test, most of the women had normal coronaryarteries and most of the men had coronary disease.This statistically significant finding suggests that themechanism for poor exercise performance in the twosexes is often different - usually cardiac in men andusually extracardiac in women.

Digitalis Therapy

We hypothesized that the digitalized patient wouldoffer a clear opportunity to demonstrate the usefulnessof exercise parameters as the predictive value of thepositive test in this group is said to be diminished.42'However, we found that the predictive value of apositive test in this group was 0.95, using a 1 mm in-crease in ST depression as a criterion for positivity. Aprevious preliminary report also showed no increase infalse positivity with digoxin therapy.44 The use of agreater increase in ST depression than 1 mm has beensuggested as a way to improve predictive value.20' 45The lack of uniformity in findings suggests a place forfurther analysis of ECG changes and exerciseparameters in the digitalized patient.

ConclusionsIt appears that the use of the simple, conveniently

measured exercise cuff blood pressure, in addition tothe duration of exercise and ST segment change afterexercise, can enhance the predictive value of positiveand negative exercise tests in a population with chestpain. An exception is the . 2 mm ST segmentcriterion. This criterion alone has been highly predic-tive of coronary artery disease in the present study aswell as others7' 20 using the treadmill, confirming work

510 CIRCULATION

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

TREADMILL EXERCISE TESTS AND CAD/Berman et al.

reported earlier from our laboratory using the two-step test.46 As the non-ST segment parameters of exer-cise relate to limitations of overall cardiac function (inthe absence of extracardiac symptoms) they shouldalso be useful in asymptomatic and atypically symp-tomatic populations where the lesser frequency of cor-

onary disease leads to decreased predictive value forECG changes alone.25 In addition, the findings thatpredictive values for multivessel disease are con-

sistently higher using non-ST segment exerciseparameters should help identify patients with such dis-ease in these populations. In order to enhance nonin-vasive diagnosis of coronary artery disease, studies in-volving larger patient groups are needed to determinethe most useful values for double-product, exerciseduration and other exercise parameters.

References

1. Cohn PF, Vokonas PS, Williams RA, Herman MV, Gorlin R:A quantitative clinical index for the diagnosis of symptomaticcoronary artery disease. N Engl J Med 286: 501, 1972

2. Fortuin NJ, Weiss JL: Exercise stress testing. Circulation 56:699, 1977

3. Borer JS, Brensike JF, Redwood DR, Itscoitz SB, PassmaniER, Stone NJ, Levy RI, Epstein SE: Limitations of the elec-trocardiographic response to exercise in predicting coronaryartery disease. N Engl J Med 293: 367, 1975

4. Redwood DR, Borer JS, Epstein SE: Whither the ST segmentduring exercise (editorial). Circulation 54: 703, 1976

5. Sheffield LT, Reeves TJ, Blackburn H, Ellestad MH,Froelicher VF, Roitman D, Kansal S: The exercise test inperspective (editorial). Circulation 55: 681, 1977

6. Goodin RR, Graham JM, Gwinn JS Jr, Marden RR, McMar-tin DE, Flowers NC: Exercise stress testing in patients withchest pain and normal coronary arteriography, with review ofthe literature. Cath Cardiovasc Diagn 1: 251, 1975

7. Rifkin RD, Hood WB Jr: Bayesian analysis of electrocar-diographic exercise stress testing. N Engl J Med 297: 681, 1977

8. Goldschlager N, Selzer A, Cohn KE: Treadmill stress tests as

indicators of presence and severity of coronary artery disease.Ann Intern Med 85: 277, 1976

9. Carvalho A, Amsterdam EA, DeMaria AN, Mason DT: Re-evaluation of exercise electrocardiography in diagnosis of cor-

onary artery disease: significance of the heart rate at which STdepression occurs. (abstr) Circulation 53 (suppl II): 11-206,1976

10. Ellestad MH, Savitz S, Bergdall D, Teske J: The false positivestress test: multivariate analysis of 215 subjects withhemodynamic, angiographic and clinical data. Am J Cardiol40: 681, 1977

11. Bruce RA, Hornstein TR: Exercise stress testing in evaluationof patients with ischemic heart disease. Progr Cardiovasc Dis11: 371, 1969

12. Mason RE, Likar I, Biern RO, Ross RS: Multiple lead exerciseelectrocardiogram. Circulation 36: 517, 1967

13. Chaitman BR, Bourassa MG, Wagniart P, Corbara F,Ferguson RJ: Improved efficiency of treadmill exercise testingusing a multiple lead ECG system and basic hemodynamic ex-

ercise response. Circulation 57: 71, 197814. McNeer JF, Margolis JR, Leo KL, Peter RH, Kong Y, Behar

VS, Wallace AG, McCants BS, Rosati RA: The role of the ex-

ercise test in the evaluation of patients for ischemic heart dis-ease. Circulation 57: 64, 1978

15. Kassebaum DG, Sutherland KI, Judkins MP: A comparison ofhypoxemia and exercise electrocardiography in coronary arterydisease: diagnostic precision of the methods correlated with cor-

onary cineangiography. Am Heart J 75: 759, 196816. Roitman D, Jones WB, Sheffield LT: Comparison of sub-

maximal exercise ECG tests with coronary cineangiography.Ann Intern Med 72: 641, 1970

17. Ascoop CA, Simoons ML, Egmond WG, Bruschke AVG: Ex-ercise test, history and serum lipid levels in patients with chestpain and normal electrocardiogram at rest: comparison of find-ings and coronary arteriography. Am Heart J 82: 609, 1971

18. Keleman MH, Gililan RE, Bouchard RJ, Heppner RL, War-basse JR: Diagnosis of obstructive coronary disease by max-imal exercise and atrial pacing. Circulation 48: 1227, 1973

19. Martin CM, McConahay DR: Maximal treadmill exercise elec-trocardiogram. Circulation 46: 956, 1972

20. Bartel AG, Behar VS, Peter RH, Orgain ES, Kong Y: Gradedexercise stress tests in documented coronary artery disease. Cir-culation 49: 348, 1974

21. Rios JC, Hurwitz LE: Electrocardiographic responses to atrialpacing and multistage treadmill exercise testing. Am J Cardiol34: 661, 1974

22. Detry J-MR, Kapita BM, Cosyns J, Sottiaux B, Brasseur LA,Rousseau MF: Diagnostic value of history and maximal exer-cise electrocardiography in men and women suspected of cor-

onary heart disease. Circulation 56: 756, 197723. Vieweg WVR, Alpert JS, Johnson AD, Hagan AD: Electrocar-

diographic exercise stress testing and coronary arteriography:correlation in 114 men with chest pain. Western J Med 127:199, 1977

24. Froelicher VF, Thompson AJ, Longo MR Jr, Triebwasser JH,Lancaster MC: Value of exercise testing for screening asymp-tomatic men for latent coronary artery disease. Progr Car-diovasc Dis 18: 265, 1976

25. Zohman LR, Kattus AA: Exercise testing in the diagnosis ofcoronary heart disease: a perspective. Am J Cardiol 40: 243,1977

26. Tonkon MJ, Miller RR, DeMaria AN, Vismara LA, Amster-dam EA, Mason DT: Multifactor evaluation of the deter-minants of the ischemic electrocardiographic response to max-imal treadmill testing in coronary disease. Am J Med 62: 339,1977

27. Ellestad MH, Wan MKC: Predictive implications of stresstesting: follow-up of 2700 subjects after maximum treadmill ex-

ercise testing. Circulation 51: 363, 197528. Bruce RA: Exercise testing for evaluation of ventricular func-

tion. N EngI J Med 296: 671, 197729. Thomson PD, Keleman MH: Hypotension accompanying the

onset of exertional angina: a sign of severe compromise of leftventricular blood supply. Circulation 52: 28, 1975

30. Morris SN, Phillips JF, Jordan JW, McHenry PL: Incidenceand significance of decreases in systolic blood pressure duringgraded treadmill exercise testing. Am J Cardiol 41: 221, 1978

31. Lawrie GM, Morris GC Jr, Howell JF, Ogura JW, SpencerWH III, Cashion WR, Winters WL, Beazley HL, ChapmanDW, Peterson PK, Lie JT: Results of coronary bypass more

than five years after operation in 434 patients: clinical, treadmilland angiographic correlations. Am J Cardiol 40: 665, 1977

32. Bartel AG, Behar VS, Peter RH, Orgain ES, Kong Y: Exercisestress testing in evaluation of aortocoronary bypass surgery:report of 123 patients. Circulation 48: 141, 1973

33. Block TA, Murray JA, English MT: Improvement of exercisecapacity after unsuccessful myocardial revascularization. Am JCardiol 40: 673, 1977

34. Amsterdam EA, Hughes JL III, DeMaria AN, Zelis R, MasonDT: Indirect assessment of myocardial oxygen consumption inthe evaluation of mechanisms and therapy of angina pectoris.Am J Cardiol 33: 737, 1974

35. Gobel FL, Nordstron LA, Nelson RR, Jorgenson CR, WangY: The rate pressure product as an index of myocardial oxygenconsumption during exercise in patients with angina pectoris.Circulation 57: 549, 1978

36. Kattus AA: Exercise electrocardiography: recognition of theischemic response, false positive and negative patterns. Am JCardiol 33: 721, 1974

37. Kramer N, Susmano A, Shekelle RB: The "false negative"treadmill exercise test and left ventricular dysfunction. Circula-tion 57: 763, 1978

38. Cole JP, Ellestad MH: Significance of chest pain during tread-mill exercise: correlation with coronary events. Am J Cardiol41: 227, 1978

39. Sketch MH, Mohiuddin SM, Lynch JD, Zencka AE, Runco V:

511

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

VOL 58, No 3, SEPTEMBER 1978

Significant sex differences in the correlation of electrocar-diographic exercise testing and coronary arteriograms. Am JCardiol 36: 169, 1975

40. Cumming GR, Dufresne C, Kich L, Samm J: Exercise elec-trocardiographic patterns in normal women. Br Heart J 35:1055, 1973

41. Linhart JW, Laws JG, Satinsky JD: Maximal treadmill exer-cise electrocardiography in female patients. Circulation 50:1173, 1974

42. Kawai C, Hlultgren HN: The effect of digitalis upon the exerciseelectrocardiogram. Am Heart J 68: 408, 1964

43. LeWinter MM, Crawford MH, O'Rourke RA, Karliner JS:The effect of oral propranolol, digoxin, and combination

therapy on the resting and exercise electrocardiogram. AmHeart J 93: 202, 1977

44. Senat A, Johnson S, Klodnycky M, Loeb H, Gunnar R: Effectsof digoxin on the ST segment during treadmill exercise testing.(abstr) Circulation 50 (suppl III): 111-245, 1974

45. Nasrallah AT, Garcia E, Benrey J, Hall RJ: Treadmill exercisetesting in the presence of digitalis therapy or nonspecific ST-Tchanges: correlation with coronary angiography. Cath Car-diovasc Diagn 1: 375, 1975

46. Cohn PF, Vokonas PS, Most AS, Herman MV, Gorlin R:Diagnostic accuracy of two-step postexercise ECG: results in305 subjects studied by coronary arteriography. JAMA 220:501, 1972

Alterations in Calcium Levels of Coronary Sinus BloodDuring Coronary Arteriography in the Dog

CHARLES B. HIGGINS, M.D. AND WALTER SCHMIDT

SUMMARY Intracoronary administration of contrast materials causes myocardial depression which is related to severalphysiochemical properties of the contrast solution. The role of variations in ambient calcium ions (Ca ) in mediating thiseffect was evaluated in 19 anesthetized dogs. Sodium meglumine diatrizoate caused decreases in left ventricular peak systolicpressure (LVPSP), -12.6 ± 3.2%, and dp/dt at a left ventricular pressure (LVP) of 40 mm Hg, -14.3 ± 4.1%. The totalcalcium (Ca,) decreased from 10.2 ± 0.2 to 6.5 ± 0.2 mg%, while Ca++ decreased from 4.6 ± 0.1 mg% to 2.3 ± 0.7 mg%.In the presence of systemic hypocalcemia the myocardial depressant actions of this contrast material were accentuated.

Intracoronary administration of contrast material with added Ca+-, calcium sodium meglumine metrizoate, caused no

myocardial depression. Total calcium decreased only slightly (10.2 ± 0.2 to 9.1 ± 0.2 mg%), while Ca-t increased(4.8 ± 0.1 to 5.1 ± 0.2 mg%). During systemic hypocalcemia, the calcium metrizoate compound induced increases inLVPSP and dp/dt/LVP40.

Thus, contrast materials caused myocardial depression which, at least in part, was related to reductions of ambientcalcium through a dilutional and binding action. The addition of Ca- to monomeric contrast materials reversed themyocardial depressant action and produced a transient rise in ambient Ca-.

ACUTE LEFT VENTRICULAR power failure andelectromechanical dissociation are rare but disastrouscomplications of coronary arteriography. Currently-utilized ionic contrast materials have a direct depress-ing action on the ventricular myocardium which isrelated at least in part to the hyperosmolarity1 andhigh sodium ion content4 I of the solution. This solu-tion, which displaces blood during coronary arteriog-raphy, is also deficient in calcium ions (Ca++), whichhave a critical role in contraction of the sarcomere.6-10Moreover, it has been suggested that the materialmost frequently used for coronary arteriography con-tains chelating properties which might further reduceambient Ca++ in the myocardium.1' It is well-established from early investigations6 9 that excita-tion-contraction coupling depends on the presence offree Ca++, and electromechanical dissociation ensues

From the Department of Radiology, University of California,San Diego, California.

Dr. Higgins is a recipient of U.S. Public Health Service ResearchCareer Development Award Grant #1 K04 HI 00201 from theNational Heart and Lung Institute.Address for reprints: Charles B. Higgins, M.D., Department of

Radiology, University of California Medical Center, 225 W.Dickinson Street, San Diego, California.

Received March 17, 1978; revision accepted May 19, 1978.

upon lowering ambient Ca+- in perfusate of isolatedheart preparations. Recently, contrast material hasbeen implicated as a causative factor in electro-mechanical dissociation occurring during coronaryarteriography." The purpose of this study was: 1) todetermine the extent of and the mechanism by whichcalcium is lowered in the coronary circulation duringcoronary arteriography; 2) to compare the severityand time sequence of depression in calcium levels in a"control state" and an induced acute heart failurestate; 3) to assess the influence of preexisting systemichypocalcemia on the myocardial depression inducedby contrast materials; and 4) to compare the effects oncalcium levels and myocardial contractility of a con-trast material containing calcium ions with currentlyused non-calcium-containing contrast materials.

MethodsNineteen mongrel dogs weighing 23-32 kg were

anesthetized with pentobarbital sodium, 25 mg/kg,mechanically ventilated at 12 cycles/min, and rightthoracotomies were performed. Teflon catheters wereinserted into the left ventricle through an apical inci-sion and positioned with the single end hole just belowthe aortic valve. The position was confirmed byfluoroscopic observation during the hand injection of a

512 CIRCULATION

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from

J L Berman, J Wynne and P F Cohndisease.

A multivariate approach for interpreting treadmill exercise tests in coronary artery

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 1978 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation doi: 10.1161/01.CIR.58.3.505

1978;58:505-512Circulation. 

http://circ.ahajournals.org/content/58/3/505the World Wide Web at:

The online version of this article, along with updated information and services, is located on

  http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer information about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. FurtherEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculationpublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:

by guest on May 18, 2018

http://circ.ahajournals.org/D

ownloaded from