Screening for asymptomatic coronary artery disease

946

J AM cou,CARDlOl 1983:1(3) 946--55

Screening for Asymptomatic CoronaryArteryDisease GREGORY S. UHL, MD, FACC*, VICTOR FROELICHER, MD, FACCt Albuquerque, New Mexico and San Diego, California

Because it will be some time beforeprimaryprethe lated to its use in populations with different prevalences vention ofcardiovascular disease is a reality, it is ad-of disease. Various techniques have been recommended visable to evaluate screening methdds for detecting latent to improve the sensitivity and specificityof exercise testcardiovascular disease. Because risk factor screeninging, andincluding other exercise measurements, computerized probability estimates, nuclear cardiology, cardiotechniques with the patient at rest have limited sensitivcardiac fluoroscopy and risk factor ity, exercise testing that brings out abnormalitieskymography, not present at rest deserves consideration. Numerous studies analysis. There is promise that these techniques will imhave shown the exercise electrocardiogram to have a prove attempts to screen asymptomatic subjects for corosensitivity of approximately 50% and aspecificityof nary disease. reo 90%. The different reportedpredictive values are

gram (3). Inaddition,theyconsideredthe sensitivityand One of the mostimportantscientificaccomplishments of the past two decadeshas been the identification of the risk specificityof theexerciseelectrocardiogram, cardiokymogfactors related to the epidemicof cardiovascular disease, raphy,cardiacfluoroscopy for coronaryarterycalcification However,given thevicissitudesof human nature, it seems and radionuclideimaging to arrive predictive at values in unlikelythat the steps prevent to coronaryheart disease will populationsubsets.Essentially,they derived a system of be taken, People only becomeconcernedwith their health decisionanalysis that prescreensindividualpatientsbefore " sick," Therefore,theearliestdetectionof when they are theyundergomoreexpensivetests. Thisenhancesthe preselecting a subgroup tests coronaryatherosclerosis is importantso that persons are dictive value of these noninvasiveby identified while they can still take some action againstwith theagreaterpre-testlikelihoodof disease(perhapswith so that the post-test probability of disease, Risk factor screeningand restdiagnostictechniques a 15 to 40%prevalence) an abnormal test will be raised to 60 to 80%. have limitedsensitivity, but mayidentifypopulations with The greatestweaknessof thisapproachis that the sena greaterprevalenceof coronarydiseasein whom more expensivescreeningtechniqueswould be morecost-effec- sitivity andspecificityof thesecondarytests are not yet tive (I), established(exceptfor thetreadmilland thalliumtests), and it is uncertain how they interact becauseof similar inadequacies. In addition, a stepapproach that uses risk Probability Analysis markers to identify a high risk group excludesthemajority Diamondand Forrester(2) reviewedthe literature and of persons who will eventuallyhavecoronarydisease.This derived tables to estimateprevalenceof coronarydisease approachconcentrates the preventiveimpact on the small, by age, sex,symptomsand theFraminghamriskequation, high risk group while ignoring the majorityof personsin which is basedon blood pressure,cholesterol,glucose the moderate risk range who contribute will largernumbers, intolerance, smoking history and the rest electrocardiobut at a lesser rate, to disease end points,

Risk FactorAnalysis for Pre-testStratification

From theDepartmentof Cardiology,LovelaceMedical Center, Albuquerque, New Mexico" and the Division of Cardiology,School of Medt This study was icine, Universityof California,San Diego. California. supported in part by the SpecializedCenterof Research on Ischemic Heart testingalone. Recent data Disease,National Institutes of Health Research Grant HL-17682 awardedMass screeningwith exercise from the U.S. Air Force School AerospaceMedicine of by the National Heart, Lung, and Blood Insntute.Bethesda,Maryland,to John Ross, Jr. MD. (USAFSAM) (4)demonstrate theproblemof notperforming Address for reprints: Victor Froehcher,MD, Cardiac Rehabilitation some type of screeningprocess before exercisetesting. Exand Exercise Testing, University Hospital, 225 DIckinson Street, San Diego, ercise tests were performedon 771 totallyasymptomatic California92103. iD1983 bythe AmericanCollegeof Cardiology

0735-1097/83/030946-10$03 00

SCREENING FOR CORONARY ARTERY DISEASE

J AM COLL CARDIOL 1983:1(3) 946--55

aircrewmenwith anormalrest electrocardiogram, whose only indicationfor such testingwas anattemptto detect latentcoronarydisease.Of these,618 underwenta double Masterstep testwhich revealed 42 personswith abnormal ST segmentresponses.In the 153subjectswho underwent treadmill testingonly and in the 42 with abnormal an double Mastertest whounderwentfollow-upmaximaltreadmill testingthere were only 20 abnormalresponders(10.2%). This prevalence of abnormalST segmentresponsesin men with anormalrestelectrocardiogram is similarto that reportedby Bruce et al. (5) and Cumming et al. (6).Cardiac catheterization in 17of the 20 menyieldedonly two cases of moderately severeangiographic c oronarydisease.T hus, mass screeningwith exercisetestingalone ishamperedby a low yield of true positiveresultsand anunacceptably high numberof falsepositiveresponses. Bruce et al. (5) recentlyreporteda 6 yearfollow-upof 2,365 clinicallyhealthymen (mean age 45 years) who underwentmaximalexercisetestingas part of the Seattle HeartWatchstudy.The Bruceprotocolwas used and lead CBs was monitored. Conventional risk factors were assessed at the timeof the initialexaminationin a subsetof the population. Follow-upwas obtainedby questionnaire, with morbiditydefinedashospitala dmissionfor acardiacevent. Forty-sevenmen (2%) experienced coronaryheartdisease morbidityor mortality. Univariate analysisof theindividual risk factors did not show a significantincreasein the 5 year o f coronarydiseaseevents.Only when the sum probability of risk factors in an individualman wasassesseddid conventionalrisk factorsbecomestatistically significantin relation to the e ventrate. MacIntyreet al.(7) performedmaximaltreadmilltests on 548 healthy,middle-agedformeraviatorsat the Naval AerospaceMedicalLaboratory.Of these,326 had one or more of threerisk factors(smoking,hypertension and hypercholesterolemia). Criteriafor coronarydiseaseafter an 8 yearfollow-upweresuddendeath,myocardialinfarction, anginaor coronarybypasssurgery.The predictive valueof an abnormalST segmentresponseto exercisewas 27.7% (I of 5) in the (5 of 18) inthosewith risk factors and 20% 222 menwithoutrisk factors. Serum lipid screening.In a study byWilliamset al. (8) including2,568asymptomatic men, a total cholesterol to highdensitylipoprotein cholesterol ratioof4.0identified o f coronaryarterydisa groupwith a very low prevalence ease;a ratioof 8.0 identifieda very high riskgroup.Uhl et al. (9)m easuredfastingtotalcholesterol and highdensity lipoproteins in 572asymptomatic aircrewmen. Of these,132 had anabnormalt readmilltest andunderwentcoronaryangiography.C oronarydisease(50% or greaterluminalnarrowing) was found in 16, with the rest having minimal (n= 14) or no(n= 102)coronarydisease.The 14 men with minimalcoronarydiseasehad totalcholesterol to high densitylipoprotein cholesterol ratios that were different no from

947

thoseof the normalsubjects(mean4.4versus4.75)while the mean ratio (± standard deviation)in thosewith significantdiseasewas 7.5 ± 1.6. Only 2 of the 16 with angiographiccoronarydiseasehad a totalcholesterolto high densitylipoprotein cholesterol ratioofless than6.0,whereas 4 of the 102 subjectswithnormalangiographic findingshad a ratiogreaterthan 6.0. Only 42 (9.5%) of 440 with a normaltreadmilltest had a ratio greaterthan6.0; 87% of thosewith coronarydiseasehad a ratiogreaterthan 6.0. A ratiogreaterthan6.0generated an odds ratio of 172. A sensitivitycannotbe limitationof this study is that true determined b ecauseonly thosewith anabnormaltreadmill testunderwentc oronaryangiography. Predictive v alueofST segmentdepression v ersusrisk factors.At theUSAFSAM, 255 totallyasymptomatic men underwentcardiaccatheterization becauseof at least 0.1 J point(10). SixtymY of ST depression80 ms past the five men had at least 50% coronaryarterynarrowing.T hus, thepredictive value of STsegmentchangeswas only 24%. Five risk factors were studied(smoking,hypertension, hyand percholesterolemia, family historyof arteriosclerosis glucoseintolerance), andunivariate analysisdid notincrease the predictivevalue.However,41 men had no risk factors and only 4of these hadcoronarydisease.The predictive valueof the STsegmentresponseincreasedas thenumber of risk factors increasedand the odds ratio was greaterthan 3: I withhypercholesterolemia alone or the presenceofthree factorand two risk factors.The presenceofat least one risk or moreexercisevariables identified aspredictive ( including 0.3 mY of ST depressionearly,persistent ST depression afterexerciseor exercisedurationunder10 minutes)idendiseasewith tifiedover halfthe casesoftwo or three vessel a predictivevalue of 84%.

Follow-upScreeningStudiesUsing ExerciseTesting Table1 summarizesfivefollow-upstudies(5-7,11-13) thatutilizedmaximalor near-maximal exercisetestingto screenasymptomatic personsfor latentcoronaryheartdisease and one (13) that evaluatedmen and women with atypicalchest pain. The patientsin these studieswere testedand followedup for thecoronaryheartdiseaseend pointsof angina,acutemyocardialinfarctionand sudden death.Angiographicfindings were not used as end points in thesestudies. Several aspects of these epidemiologic studies are similar. Threeused asinglebipolarchestlead formonitoring two used 3bipolarleads and one used all standard 12 leads. Two studiesused bicycleexerciseand, therefore, the subjectsmay haveperformed only submaximalstress;theothers used atreadmilland encouragedpatientsto exerciseto exhaustion.All studieswith theexceptionof one used 0.1 mY

948

J AM

cou.

CARDIOl 1983.1(3)9 46-55

UHl AND FROELICHER

Table 1. Resultsof ExerciseElectrocardiographic Testingin FiveProspective StudiesScreeningAsymptomaticMen for Latent CoronaryDiseaseand One Study EvaluatingMen andWomen With AtypicalChest Pain

Pnncipal Investigator Cummmg (6) Froelicher(13) Allen (II) Bruce (5)

no.

Age Range (yr)

40-65 20-54 >40 >25 (44) MacIntyre (7) 47-57 548 (52) Manca* (12) 947 >30 (men) (48) 508 >30 (women) (49) Average (men only) 510 1,390 356 2,365

Incidenceof CoronaryHeart Years of Follow-up Disease

% With Abnormal Exercise Test Sensinvuy(%)

Specificuy(%)

Relative Risk (times Predictive that for Value normal (%) subjects)

4.7% 3.3% 9.6% 20%

3.0 6.3 5.0 6.0

13 10 23 II

58 61 41 30

90 92 79 89

25 20 17 5

10 14 2 3

6.9%

8.0

4

16

97

26

4

5.0%

5.2

18

67

84

18

10

1.6%

5.2

28

88

73

5

15

13

46

89

19

7

*CalculatIonsbased onUSIng only ST segmentdepressiona,thecntenonfor abnormal Ages in parentheses are mean values (Adaptedfrom FroehcherV ExerciseTesting and Trammg New York: Le Jacq. 1983'96.)

of horizontal or downslopingST depressionas thecriterion the prevalenceof coronaryheart disease in the population of the exercise test is for abnormal; Cumming et al. (6) required 0.2 mY of STunder study, and that the specificity depression. Two ofthese studies deserve further comments, lower in women than in men. both because they includedwomen and one because the Asymptomatic men versuswomen. Allen et al. (1 I) recently reported a 5 year follow-up study of 888 asymppatients had atypical chest pain. Subjects with atypical chest pain. In Italy, Manca et tomatic men and women without known coronaryheart al. (13)studied947 men and 508 women who were referred disease who had undergonemaximal treadmill testing. No for exercisetesting because of atypical chest pain. Thosesubject was taking medicationand none had pathologicQ with typical symptomsof anginapectoris,valvulardisease, waves,otherknown cardiovascular abnormalities or exerhypertension,bundle branch block,arrhythmias, Wolffcise-induced chest pain. Maximal treadmill testing was perParkinson-White syndrome, left ventricular hypertrophyformed with using the Ellestadprotocol,and leads CM s, Y I and strain,significantrestrepolarization abnormalities and pre- a bipolarvertical lead were recorded.Subjects were exervious myocardialinfarctionwereexcluded.No patient re- cised until they reached 100%predicted of maximal heart ceivedcardiovascular drugs in the 2 weeks preceding ex-rate, fatigue or marked dyspnea.Subjects with major ST ercise testing. Supinebicycleexercisewas carried out after segment changes at rest were excluded,but in those with routinehyperventilation until at least 85% of thepredicted minor changes at rest, additional an 0.15 mY on exercise maximal heart rate was reached.The conventional12 elecwas required for a test to judgedabnormal. be trocardiographic leads wererecordedduring and after the The original population included 1,077 subjects,888 exercise test. The criterionfor an abnormal response was(82.5%) of whom were contacted forfollow-up.Of the 113 0.1 mY or moreof horizontalor downslopingST segment subjects who initially had an abnormal exercisetest, 105 depressionand waspresentin 18% of the men and 28% of (92.9%) were located. There was a I. incidence I% rate of the women. The overall incidencerate of coronary diseasecoronary heart disease per year. Only of221 2 men 40 years was 5% in the men and 1.6% in the women. The sensitivity of age or less developedheart disease end points and neither was 67% in the men and 88% in the women; specificity man had ST segment abnormalities. Hence, in this study, was 84% in the men and 73% in the women. The predictive the treadmill test did not predict coronaryheart disease in value of a positive test was 18% in men, but only 5%asymptomatic in men 40 years of age younger. or These results women. Men with a positive test had a relative risk of 10 contrast with those of a similar study of 563 men aged 30 fordevelopingclinicalmanifestations of coronary heart dis- to 39 years in whom there was a 1.4% incidence rate of easecomparedwith 15 for women with a positive test. This coronarydisease (13). The exercise electrocardiogram was study clearly shows how predictivevalue is influenced by found to have a 50% sensitivity, 95% specificity,13% pre-

949

J AM COLL CARDIOL 1983,1(3)'946-55

SCREENING FOR CORONARY ARTERY DISEASE

interview;2) typical angina during a neardictive value and a risk ratio of 17, and thus it stilltoris had positive on maximalbicycletest; and 3) an abnormal exercise electrovalue in this age range. Theexercisetest had a predictive value of 84% Of the 311 women followed up,developed 10 coronary cardiogram. if a slowlyascendingST segment was used ascriterion a heart disease end points, but ST segment depressionand R The higherpredictive value in this study may wave response did not predict these events. Exerciseforabnormal. duexclusion the of ration of 3 minutes or less, however,provedto be predictive; be relatedto the older age of the patients, men with rest ST-T changes and inclusion of men with 4 (30%) of 13 women with this response developed coronary the heart disease. When used as criterion a for an abnormal angina. These studies using "instant"epidemiologic end point ofangiography confirm the low predictive value test,exerciseduration of 3 minutes or less asymptomatic in populations,as women had asensitivityof 40%, specificity of 97%, pre- of theexercisetest in apparently healthy summarizedin Table 2. dictive value of % 31 and risk ratio of 15. However,we believe that exercisetesting may prove to be of value in screeningasymptomatic women. The limited follow-up of Other Electrocardiographic and Exercise 80% of the original populationand the low incidence of coronarydisease end points in women and in men younger Test Variables than 40 years of age are limitations of this study. New electrocardiographic abnormality. Other variables of the baseline electrocardiogram have beenevaluated Maximal or Near-Maximal Exercise Testing as markers of coronary disease. In the U.S. AirForce,the With Coronary Angiography annualevaluationof pilots involved serial rest electrocardiograms. The finding of a new abnormality resultedin Froelicheret al. (14)performedcardiaccatheterization referral to USAFSAM. In order to be returned to flying on 138asymptomatic men with an abnormal treadmill test. status, cardiac catheterization was required for any major Less thanone-thirdof the subjects had at least one lesion electrocardiographic abnormality. Because all men referred equal to orgreaterthan 50% luminal narrowing of a major performeda treadmill test, this could also coronaryartery(Table2). Barnard et al. (15) used near-for any reason catheterization. We (14)previouslyreported maximal treadmill testing to screen randomly selectedlead Los to cardiac prevalencerate ofangiographiccoronary artery disease Angelesfirefighters. Ten percent showed an abnormal aexofrisk only 22 and 18% in asymptomatic persons with new left ercise-induced electrocardiographic response despite few block, respectively.Pilots with factors for coronaryheartdisease,and six of these elected and right bundle branch serial repolarization changesunderwent exercisetesting and to undergocardiaccatheterization. One had severe three to whetheror not they vessel disease and anotherhad a 50%obstructionof the their response was useddetermine needed cardiac catheterization. Angiographic findings in those leftcircumflexcoronaryartery; the other four men had norrepolarization phenomenaand abnormal treadmal studies. Borer et al. (16) reported angiographic findings with specific in 30 asymptomatic persons withhyperlipidemia and an mill tests are listed in Table 3. One wonders how many had we abnormal exercise test. Only 37% were found to have ST an-segmentelevationnormalizingrestabnormalities; giographically documented coronaryheart disease. Erikssen were not aware of the meaning ofphenomenon this when et al. (17) reported angiographic findings in 105 men aged our studies were done and thus it could have been over40 to 59 years from a working population with one or more looked. In the study of Uhl et al. (10) moretwo-thirds than of the following criteria: questionnaire I) a for angina pee- of all patients had ST and T wave changes on electrorest

Table 2. Prevalence of AngiographicCoronaryArteryDiseasein ApparentlyHealthyMen WithExerciseTest-induced

ST SegmentDepression

Patients (no.) Principal lnvesngator Froehcher(14) Borer (16) Erikssen (17) Uhl (10)

Total 138 30 75 255

'''Working'' population rather than asymptomatic. many had angma. CAD = coronary artery diease.

Luminal Obstruction 30 II 48 65

With Any CAD 56 20 98

PredictiveValue for Significant AngiographicCAD (%)

26 37 84* 26

950

J AM cou,CARDlOl 1983:1(3).946--55

Table 3.AngiographicFindings inI) I Aircrewmen With an

UHL AND FROELI CHER

only by upsloping criteriaand thepredictivevalue was no worse than horizontaldownsloping or ST changes.In general, upsloping ST criteria increasesensitivity the but deSignificant crease the specificity of the exercise electrocardiogram. Mean Angiographic Exercise responsesin additionto ST depression. Annual Age CoronaryDisease Three recent studies have evaluatedseveral exercise test (%) (% ) Rest ECG n variablesotherthan ST segmentdepression . Allen et al. Normal 34 44 23.5 (II) evaluatedmany exercisetest variables including the PreviousST-T change amountofST segmentdepression , total duration ofexercise but current and R waveamplitudechange withexercise.The presence 21 43 23.8 ECG normal Low amplitude T of coronaryheart disease within 5 years was predicted 24 waves 42 25.0 by an abnormalST segmentresponse,an increase or no ST segment abnormal 32 44 46.9 change in R wave and an exercise duration of5 minutes or less: these variables had sensitivity a of 41, 47 and 27%, respectively(Table 4). With the test results interpreted cardiograms, yet thepredictivevalue of an additional 0.1 asabnormalwhen either ST or R wave criteria were present, mY of ST segmentdepressionwas arelativelyhigh 28%. sensitivitywas 65%. When all three criteria were present, Labile ST -T changeson standingor duringhyper- a sensitivity of 29% with a specificity of 100% was achieved. ventilation.OrthostaticST segmentdepression or that Exercise duration of less than 5 minutes was the most preassociated withpreexercise hyperventilation have been iden- dictive single variable . tified as markers of a falsely positive ST depressionwith In the study of Bruce et al.(5) , severalconventional risk exercise insymptomatic patients . Erikssen et al. (17) did factors and exercise test variablesstudied were and correnotconsiderfurther STdepressionwith exercise in patientslated withcoronaryevents at 6 year follow-up. The ST manifestinglabile repolarization changes as an abnormal segment criteria had a sensitivity of 30%, specificity of 89%. response. Morris and McHenry(18) performed serial stresspredictive value of 5% and a risk ratio of 3.5. Individually. tests on 900 presumably healthy men and identified 14 men the other exercise test variables were nopredictive more with labile ST-T changeswithstandingor hyperventilation than ST segment criteria(Table4). Only when at least one risk and abnormal ST depressionatexercise.At the 7 year fol- conventional risk factor and at least oftheexercise two low-up study , none had manifested a coronary event while predictors were present was a subgroup (about % of the I in 24 men withexercise-induced ST changes but no labile total population) identified with substantial a increase in ST-T wave before exercise,10 (42%) had a coronary event.relative risk (18 x ) developing of coronaryheart disease. In the studies of Froelicherand hisco-workers(13,14),the In the study from the USAFSAM (10) of 255 asymptomatic men with angiographic coronary disease end points. presence of orthostatic or hyperventilation-induced ST changes did not rule out the development ofcoronaryheart disease several exercise test variables were studied singly and in in men withabnormaltreadmilltest results. In the angio-combinationwith conventionalrisk factors.Increasing o f Uhl et al.(l0), 18% of patients with cor-amounts of maximal ST segment graphic study depression did notenhance onary disease and 15% of normal subjects had posturalthe STpredictivevalue over ST segment criteria alone until changes withstandingand thepredictivevalue of an ad- at least 0.3 mY of ST depressionoccurredin stage I or II ditional 0.1 mY of ST depressionwith exercise of 25%, of the USAFSAM protocol. Early onset of at least 0.1 mY which was no worse than the overall predictive value. Simof ST depressionwas not a goodpredictoro f coronary ilar percents ofabnormaland normal subjects (22 and 16%,lesions. An increase or no change in R wave amplitude after respectively) demonstrated ST and T changes with preex- exercisecomparedwith baseline R wave amplitudewas ercisehyperventilation with apredictivevalue of 26%. insensitive and not highly predictiveand detectedonly 4 Upslopingdepression.In the study of Allen et al. (II), (12%) of 33asymptomatic men withmultivesselcoronary the relative risk for coronaryevents in men demonstrating disease . The two most accurate predictors of angiographic an upsloping ST segmentduringexercisethat was depressed disease were total treadmill time less than 10 minutes and 0.1 mY at 80 ms after the J point was quite low. Only 3 persistence of at least 0.1 mY horizontal depression ST for of 49 men who met this criterionforabnormality developed at least 6 minutes after exercise . One finding very similar cardiac end points. In the study of Erikssen et al. (17), an to data from the Seattle Heart Watch study (5) was that a upsloping STsegmentthat was depressed at least an adcombination of one of the conventional risk factors plus two ditional 0 .15 mY at 80 ms was justas predictive as hori- or more of the exertional risk predictors(at least 0.3 mV zontal ordownslopingST depression . In the study of Uhl ST depression in the first 6 minutes, test duration under 10 et al. (10) the exercisetests of 9 of 65 men with coronaryminutes or persistent depression ST into 6 minutes of redisease,including 3 with multivessel disease, were positive covery) was very predictive.This combinationhad a senAbnormal Exercise Electrocardiogram Grouped According to Annual RestElectrocardiographic (ECG) Findings

cou,CA RDIOl 6-55 1983:1(3):94

J AM

SCR EENING FOR CO RONARY ARTE RY DISEASE

951

Table 4.Performance of ExerciseTestVariable s andRisk Factorsin DetectingCoronaryArteryDisease

Predictive Value

Risk Ratio

79 78 96 86 99 99 100

17 19 43 27 71 82 100

2 3 6 5 12 17

30 6 6 19 19

91 99 99 93

5 15 19 7 46

3.5 8 10 4 18

36 33 46 28 28 55

79 64

38

2

92

87 87 86

67 43 42 84

4 6 2 4

37

98

89

45

First Author

Abnormal Response

Sensitivity ('7c)

('k)

Allen (1 \)

ST! RW A TM time < 5 min ST! + RWA ST! + TM lime < 5 min RWA + TM time < 5 nun ST! + RWA + TM time < 5 min

41 47

Bruce(5)

ST! Angina on TM TM time < 6 min HRI 2':1 RF + 2':2 ExRP

Uhl (10)

2': 0.3mV ST Onset ST! ins tage1 TM time < 10 min PersistentST ! (6min) RW A 2': 1RF + 2':2ExRP 2':1 RF + 2':3 ExRP to detect multivessel disease

27

40 24 33 29

Specificity

23

II

I

ExRP = exercise nsk predictor: HRI = heart rate impanrnent: =RFnsk factor: RWA= R wave amplitude abnormalit y.1ST= ST segme nt depresvion. T=M treadmill test.

sitivity of55% , specificity of 86% and a predictive value exercise, but that the determination of wall motion abnor malities is much less reliable. This is disappointing because of 84% . Exercise-inducedarrhythmias. Very few studies in segmental dysfunction precedes global dysfunction during exercise and should be more specific for coronary disease. asymptomatic subjects have evaluated exercise-induced which is usually a regional disease. ventricular premature beats for detecting coronary disease. The only reported application of exercise radionuclide In the USAFSAM follow-up study ( 19) of 1.390men, only 39 (2. 1%) developed"ominous" arrhythmias. The risk ra-angiography in detecting coronary artery disease in asymp tio of developing coronary disease 6over years of follow- tomatic men compared global ejection fraction changes with up with these arrhythmias 3. was However. the predictive visual evaluation of segmental wall motion and the visual value was only 10% and sensitivity 6.7 only% .In the study interpretation of thallium perfusion scintigraphy in 32 men (4), 14asymptomatic men (7 .2% of the (21). The thallium perfusion study had a better sensitivity of Piepgrass et al. 6 con(92%) than did global (85%) or regional(62%) abnormalpopulation) developed ominous arrhythmias; only sented to further studies and 2 of these had coronary disease. ities by radionuclide angiography without loss of specificity In the study of U.S. Army personnel by Zoltick (20). et al. (95 versus85 and95%. respectively). However. most stud6 (2. 1%) of 287 subjects exhibited ominous ventricular aries have shown exercise ventriculography to have a much rhythmias. All six underwent thallium scintigraphy. poorer and specificity, making it unacceptable for screening ( I) . only the two with abnormal scans were found to have coroBattler et al. (22) reported on imaging gated blood pool nary artery disease at catheterization . Thus, arrhythmias ventriculograms after treadmill exercise. Tagged technetium induced by exercise testing have not been helpful inwas de- administered intravenously in order to perform setecting latent coronary disease in apparently healthy quential men. left ventriculography 10 in patients with coronary heart disease and 8 normal subjects. Ejection fraction was measured at 2 4,4 to to 6 and 8 to 10 minutes of recovery Radionuclide Stress Testing after a maximal treadmill test. At108 minutes to of reRadionuclide Angiography covery. all normal subjects but none of the patients with coronary heart disease had a higher ejection fraction tha Studies from many centers have shown that left ventricular ejection fraction measured by gated blood pool they angi-did at rest. Post-treadmill analysis could be more sen ography is accurate and reproducible both at rest and during sitive than the supine bicycle technique because of the highe

952

J AM COLL CARDIOL 1983,1(3)946-55

UHL AND FROELICHER

work loadachieved,and an increase in specificity couldMotion within its electromagnetic field causes a change in possibly beobtainedoverstandard ST segment analysis bythe frequencyof an oscillator,which isconvertedinto a proportional to the motion. The advantage the more accurate detectionof regional wall motion abnor-change in voltage of thecardiokymogram over theapexcardiogram and the malitiespossiblewithout motion artifact. kinetocardiogram is that it records absolute cardiac motion Thallium-201 Exercise Testing without chest motion, thus eliminatingthat distortion. Silverberget al. (27) reported their use cardiokyrnogof Caralis et al. (23) performedthallium-20lexercise testvolunteersand ing and coronaryangiographyto evaluateasymptomatic raphy after treadmill exercise in 27 healthy coronaryheart disease who persons withabnormalST segmentresponses to exercise 130 patients with suspected underwent coronary angiography. Recordingswere made testing. In3,496consecutivetreadmill tests, 22 persons within 2 minutes after exercise and every minute thereafter developed0.2 mY or moreasymptomatic horizontal ST for 10 minutes. Two sets of empiric criteria for an abnormal segmentdepression. Fifteenof these agreed to further evaluation with thalliumscintigraphy and coronary angiography, cardiokymographic pattern were defined in relation to known effects of ischemia on regional wall motion. The first abOf these 15personswith an abnormal exercise test, 5 had as systolicoutwardmonormal thallium scans with exercise,whereas 10 developed normality was definedparadoxical abnormality as developmentof total new defects. The thalliumexercisescans classified 13 of tion, and the second 15 patients properlywith one false negative and one false absence of inward motion, a resultant holosystolicoutward positive result. motion or systolic outward motion occurringfor less than Nolewajkaet al. (24)performed a thallium treadmill testthe entire period of ejection butpreceded not by inward on 58asymptomatic men as part of a screening study. The motion. For detecting coronary heart disease in patients wit risk forcoronaryheart disease was determinedusing the atypical chest pain, cardiokymogram the had a higher senFraminghamrisk equation(3). The riskcalculationwas sitivity, specificity and predictivevalue than did the elecgreater in those withabnormal an exercisestudycompared trocardiogram. However, there was difference no between with those who had a normal study. Six had an abnormal theelectrocardiogram and thecardiokyrnogram in asympthallium scan (five consistent withischemia,one with scar). tomatic patients. These results may explained be by the Three of the subjects with an abnormal thallium study underunusualdistribution of disease in this study; patients with wentcoronaryangiography,and all had normal coronaryatypical chest pain had a 30% prevalence of disease, whereas arteries.Surprisingly,two of these subjects had left bundle asymptomaticpatients had a 64% prevalenceof angiobranch block (one with exerciseonly, the other at rest),graphiccoronaryartery disease. which mayaccountfor the false positive scans (25). The Zoltick et al. (20) utilized cardiokymography as part of disappointment of these results was compoundedby the a serial testing evaluationof 287asymptomatic subjects. psychologicstresses to the personswho were told they had Type II cardiokymographic abnormalities occurredin 10 "abnormal"results. subjects,5 of whom hadcoronarydisease. This finding, Uhl et al. (26)performedsingle dose exercise and de-with asensitivity of 63%, specificity of 74% and predictive layed thallium-201 scintigramson 191aircrewmenbefore value of 50% was more effective than horizontal ST segment the menunderwentcoronaryangiographybecause of ab- depression(25, 89 and 50%,respectively), making it the normalexerciseelectrocardiograms. Of these men, 61 had second best single screeningtest(thalliumscintigraphy was significantangiographicdisease,for apredictivevalue of best).However,rememberthat 26% of normal subjects the electrocardiogram of 21 versus 74% forscintigraphy. would be false positive responders. The specificity of the computer-processed scans was 90%. There were mixed results in the 15 men with minimal disease « 50% occlusion):I0 had an abnormal scan and 5 had Coronary Artery Calcification on normal scan. The finding of high sensitivity and specificity Fluoroscopic Examination of thesecomputer-enhanced thalliumscintigramsin these Kelley, Langou and their co-workers(28,29)reported apparently healthy men is strong support for their use as a the use of cardiac fluoroscopy to detect coronaryartery second linescreeningprocedure.If both an abnormal exmen erciseelectrocardiogram and abnormal perfusionscintigram calcificationas a prescreeningtool inasymptomatic underwent exercise tests. In one study (28), had been required before angiographywas performed,136 before the men 129 healthy men (average age 49 years)evaluated were with of those free of coronarydisease would not have needed to cardiac fluoroscopy and then submaximal a exercise test. undergoangiography. Of the 108 subjects who completedthe exercise test, 37 (34%) had at least one calcified coronary artery. Of this Cardiokymography group of subjects with positive fluoroscopic findings, 13 (35%) had an abnormal ST segment response to the exercise The cardiokymograph is an electronic device that produces a representation of regional left ventricularwall motion. test. Of the 68 subjects with normal fluoroscopic results,

SCREENING FOR CORONARY ARTERY DISEASE

J AM COLL CARDIOL 1983.1(3):946-55

953

only 3 (4%) had an abnormal exercise response. Of theand 16 an "abnormalnondiagnostic"response, defined as % had subjects with an abnormal exercise test, 81calcifi- upsloping STchanges,occurred in 15 men. Six men had frequent exercise-induced ventricular premature beats. Thes cation of at least one coronary artery. In a second study (29), 13 men who had both coronary 26 men underwentcardiac fluoroscopy and thallium perartery calcification and an abnormal exercise test underwent fusionscintigraphy. Seven men had abnormal thallium scincoronary angiography. They had a mean age of 44 years; tigraphic findings, six underwent cardiaccatheterization and none had any symptoms or signs of coronary disease, one anddied of m a yocardial infarction. One man with a normal all had a normal rest electrocardiogram. Coronary arteri- thallium scan,ventriculararrhythmia on exercise testing ography revealed 12 men with clinically significant coronary with a normal ST segment response and a very high risk (>50% luminal narrowing in any major cor-index wascatheterized artery disease and was found to be free of disease. onary branch): single vessel disease in 4, double vessel in man with a low risk index and normal treadmill test, One 5 and triple vessel in 3 men. One man had only a minor cardiokymogram and fluoroscopic findings had a myocardial lesion. On a 3 year follow-up in these 13 patients, 3 infarction had after 6 months of follow-up.performance The of developedtypical angina and 1 had developed transmural these various tests is presented in Table 5. No patient had myocardial infarction. The combination of coronary artery coronarycalcification. An abnormal ST segment response calcification and an abnormal exercise test appears highly was insensitive and not highly predictive of coronary dispredictive of coronary heart disease. However, the sensiease.Cardiokymography had a 63% sensitivity, a 74% spectivity and specificity ofcombination the of these procedures ificity, a predictive value of 50% and was the most accurate remains to be clarified. Questions remaining are: I) didindividual the test. Risk factor analysis was predictive, not and three subjects with an abnormal exercise test and noonly cal-when there were two or more risk factors and an abcification have angiographic disease? and 2) what of the 24normalcardiokymogram was screening accuracy improved. subjects with calcification and a nonnal exercise electrocardiogram? In the study of Zoltick et al. (20), fluoroscopic evaluation Conclusions of 26 asymptomatic subjects who had abnormal exercise tests or ominous exercise-induced arrhythmias failed to re- Risk factoranalysisversusroutineexerciseelectroveal calcification in any of them, including the 8 with proven cardiogram.In screening for asymptomatic coronary heart coronary disease. Perhaps coronary calcification is detectdisease, risk factor analysis alone has a sensitivity of only able only in patients with long-standing coronary disease, approximately 20%. In the studies reviewed here, sensitivity that is, in older subjects. of the maximal exercise electrocardiogram ranged from 30 to 67%. Two recent studies (5,11) show markedly lower sensitivity and predictivevalues for the exercise test than previouslydocumented.In the second study (11), roughly Serial Testing Proceduresto DetectCoronary 20% of the subjects were lost to follow-up; hence the inArteryDisease cidence of coronary heart disease may well have been underRecently,the preliminarystage of a program of serialestimated. The Seattle Heart Watch study (5) is difficult to testing to detect latent coronary heart disease was completed interpret in light of preceding studies by Bruce and others. by the U.S.Army(20). Screening was considered necessary There was no predictive value for an abnormal exercise in the subgroup without risk factors. Unbefore initiating a mandatory exercise program for allelectrocardiogram perfortunately, not all of the population had a cholesterol serum sonnel older than 40 years. The screening tests were applied measurement, so the risk factor data were complete. not in a sequential manner in an attempt to eliminate low risk patients from further testing and enhance the pre-testThese like- latter two studies considered exercise responses test lihood of disease in the remaining subset. An initial history, other than ST segment depression. Future studies are neede physicalexamination and restelectrocardiogram were per- to substantiate these findings. formed on 285 men and 2 women over 40 years (mean agePredictivevalueof abnormalexerciseelectrocardio44). A fastingbiochemicalprofile was obtained and a riskgram. If serial exercise tests are performed the response factor index based on Framingham the data base was cal- will change from normal to abnormal in some of the persons who are developing coronary disease. This could increase culated.All subjects underwent maximal symptom-limited exercise testing. All were encouragedto exercise to ex- sensitivity and specificity. However, Morris and McHenry (18) reported that a change from a negative to a positiv haustionand the average oxygen consumptionwas 38.5mil kg per min (range 25 to 64). Pre- and postexercise carditest is no more predictive than is an initially abnormal tes and one subject has been observed whose changed test from okymogramswere performed. A risk factor index over 5.0 of was consideredabnonnal.Unfortunately, the ratio of cho- normal toabnonnalalthough he was freeangiographically disease (30). lesterol to high density lipoprotein cholesterol was not significant studexercise ied. An abnormal ST segment response occurred in 4 menThe predictive value of the abnormal maximal

954

cou,CARDIOl 1983,1(3)9 46-55

J AM

UHl AND FROELICHER

Table 5.Performance of Stratified RiskAnalysisin PredictingCoronaryArteryDiseasein ApparentlyHealthyPersons(20) Patients (8 withdisease)

Variable Risk mdex ~ 5 2 or more RF Abnormal TM TM PVCs Abnormal CKG 2 or more RF+ abnormal CKG

No, of Positive Results

(%)

(%)

Predictive Value

Odd, RatIo

II 15 4 21 10

125 625 25 75 62,5

53 47 98 21 74

18 33 50 29 50

016 L5 28 08 1.75

96

50

89

67

8,5

Sensttivity

Specificity

CKG = cardiokymogram, PVCs = significantpremature ventncular complexes:RF = conventional nsk factors,TM = treadmilltest

electrocardiogram rangedfrom 5 to 46% in the studies step should be a maximal exercisetest.Thalliumscintigor shouldbe reveiwed(Table I). An averageof 21% of the abnormal raphy, cardiac fluoroscopycardiokymography diagnostic the value of responders developed coronaryheartdiseaseo verthe follow- the third step, The lack of data on up period.Thus,more than 75% of the abnormalresponders these tools inasymptomaticpersonspreventsstrict recat this time, Good clinicaljudgmentmust be had a falsepositivetest. Thepossibilityexists that some of ommendations complication of producing thesesubjectshavecoronarydiseasethat has yet to manifestexercisedto avoid the iatrogenic by healthypeople.Theclinician itself, but angiographic studies inasymptomatic populations cardiac cripples mislabeling " needtoknow" againstthe highprobability havesupported this high false positiverate (Table 2). This must weigh the embarkingon the next level result isexplainedby the directrelationbetweenthe pre- of a false positive test before dictive value of an abnormalexercisetest response and the of the decision analysis tree. prevalence of coronary heart disease in the population tested, Thus, thepredictive value of the testenhanced is in subjects References with increasedrisk factors or markersfor coronaryheart I. FroelicherVF, Exercise Testing and Traimng.New York, Le Jacq, disease, 1983, Motivational effects of exercise testing. Exercisetest2, DiamondGA, Forrester JS, Analysis ofprobabihty as an aid in the ing mayproveto have a value asymptomatic in persons in clinical diagnosisof coronary-artery disease, N Engl J Med 1979; 300:1350-8, additionto itsscreeningcapabilities, Bruce et al. (31) ex3, Gordon T, Sorlie P, Kannel WB, Coronary heart disease,atheroamined themotivational effectsof exercisetesting for modthrombolitIc braininfarction, intermittent claudication-a multIvanate ifying risk factors and health habits. questionnaire A was analysis of some factors related to their mcidence: FrammghamStudy, sent tonearly3,000men aged 35 to 64 years who had 16-yearfollow-up,WashmgtonDC: U,S, GovernmentPrinting Office, 1971, section 27, undergonesymptom-limited treadmilltesting at least I year 4. Piepgrass SR, Uhl GS, HickmanJR Jr, Hopklrk JAC, Plowman K. earlier.The men were asked if the treadmilltestmotivated The hmitations of the exercise stress test indetection the of coronary them to stop s moking,increasedailyexercise,alter diet or artery disease apparently in healthy men. Aviat Space Environ Med takemedication forhypertension, There was a 69% response 1982;53:379-82. to thisquestionnaire, and 63% of therespondersindicated 5 Bruce RA, DeRouen TA. Hossack KF. Value of maximal exercise dl~ea,e 10 healthy that they had modifiedone or more risk factors and health tests in riskassessmentof pnmarycoronary heart men Five year',' expenenceof the Seattle Heart Watch study. Am J habits and that they attributed this changeto theexercise Cardrol 1980.46:371-8 test. In fact,greater a percento fsubjectswith a poor exercise 6. Cumming GR, Samm J, Borysyk L. Kich L Electrocardiographic changesdunngexercise inasymptomatic men 3-yearfollow-up.Can capacityreported a favorablemodification of risk factors or Med Assoc J 1975:112.578-81 health habits.

Recommendedthreestep approachto screening If a methodof secondaryprevention asymptomatic men.

7 MacintyreNR. Kunkler JR, Mitchell RE,ObermanA. GraybielA. Eight-yearfollow-up exercise electrocardiograms in healthy.middleaged aviators Aviat Space Environ Mcd :52:256-9 1981

were proved and availabletoday,we would offer the fol- 8. Wilharns P. Robmson D. Bailey A. High density lipoprotemand coronary risk f actor,m normal men Lancet 1979; 1:72-5. lowing three step approachto screeningfor asymptomatic coronaryheartdiseasein men over35 years old. First to 9. Uhl GS, TroxlerRG, Hickman JR Jr. Clark D.Angiographiccorrelation ofcoronaryartery disease With high densuyIipoprotem cholbe obtainedare ahistoryof angina,risk factor analysis esterol inasymptomatic men. Am J Cardiol 1981 :48.903-11. (includinglevel of highdensitylipoprotein cholesterol) and 10. Uhl GS, Hopkrrk JAC, Hickman JR Jr. Fischer1, Medina A. PreIt must berememberedthat the a restelectrocardiogram. dictiveimplications of multiple exercise variables detectmg m Signifsensitivity andpredictive value of these rest techniquesare icantcoronaryartery disease asymptomatIc m men. (10 press). low.If theyindicatethat the subject is at risk, the second II. Allen WH, Aronow WS,GoodmanP. Stmson P.Five-yearJ Cardrol

SCREENING FOR CORONARY ARTERY DISEASE

J AM COLL CARDIOL 19~3.1(3).946-55

Rehabfollow-upof maximaltreadmill stress test 10 asymptomatic men and women.Circulation1980;62:522-7. 12. MancaC, Cas LD, AlbertiniD, Baldi G, Visioh O. Differentprognostic value of exerciseelectrocardiogram in men andwomen.Cardiology1978;63:312-9.

955

22. BattlerA, Slutsky R, PfistererM, Ashburn W, FroelicherV. Left ventricular e jectionfracttonchangesduring recoveryfrom treadmill exercise:a preliminaryreportof a newmethodof detectmgcoronary arterydisease.Clin Cardiol1980;3:1 4-8.

13. FroelicherV, ThomasM, Pillow C, LancasterM An epidemiologic study ofasymptomatic men screenedwith exercisetesting for latent coronaryheartdisease.Am J Cardiol1974;34:770-81.

23. CarahsDG, Bailey I, KennedyHL, Pitt B.Thallium-201myocardial imagingin evaluation of asymptomatic individualswith ischaernicST segment depressionon exercise electrocardiogram. Br Heart J 1979;42:562-7.

14. FroelicherVF, ThompsonAJ, WolthuisR, et 'II.Angiographicfindings in asymptomaticaircrewmen with electrocardiographic abnormalities.Am J Cardiol1977;39:32-8.

24. NolewajkaAJ, Kostuk Wj, HowardJ, et 'II Thalliumstress myocardral rmaging: evaluation an of fifty-eight a symptomatic males. Clin Cardiol1981;4:134-8.

15. BarnardRJ, GardnerGE, Diaco NV. "Ischemic"heartdiseasein fire fighters withnormalcoronaryartenes.J Occup Med 1976;1 8:81824.

25. JohnsonRE, WilliamsBR, LibermanHA, Morris DC, Logue RB. Stressthalliumscintigraphy in patientsWith leftbundlebranchblock (abstr) Circulation1981;(suppl IV):IV -105

16. BorerJS, BrensikeJF, RedwoodDR, et 'II.Lurutationsof the electrocardiographic responseto exercisein predicting coronaryartery disease.N Engl J Med 1975;293:367-71.

17. ErikssenK, Enge I, ForfangK, Storstein0 False positivediagnostic tests andcoronaryangiographicfindings in 105presumablyhealthy males.Circulation1976;54:371-6.

26. UhlGS, Kay TN, HickmanJR Jr.Computer-enhanced thalliumscintigrams10 asymptomatic men with abnormalexercisetests. Am J 1037-47. Cardiol1981,48: 27. SilverbergRA, DiamondGA, Vas R, Tzrvoni D, Swan HJC, Forrester JS Noninvasivediagnosisof cornaryarterydisease:the cardiokymographicstresstest.Circulation1980;61:579-89.

18. Morns SN, McHenry PL. Role ofexercisetest109 10healthysubjects and pattents withcoronaryheartdisease.Am J Cardiol1978;42:65966.

28. Kelley MJ, Huang EK, Langou RA. Correlationof fluoroscopically detectedcoronaryarterycalcification wrthexercisestress testing in 1-6. asymptomatic men. Radiology1978; 129:

J, Lancaster 19. FroelicherVF, ThompsonAJ, Longo MR, Tnebwasser men for MC. Value ofexercisetest109for screenmgasymptomatic latentcoronaryarterydisease.Prog CardiovascDis 1976:1 8'265-76.

29. Langou RA, Huang EK, Kelley MJ,Cohen LS. Predictiveaccuracy of coronaryarterycalcification andabnormalexercisetest forcoronary arterydiseasein asymptomatic men. Circulation1980;62:1196-203.

20. ZoltickJM, Patton J, Vogel J, DanielsW, BedynekJL, Davia JE. Cardiovascular screeningevaluation to test for coronaryarterydisease in asymptomatic malesoverthe age of 40 (abstr).J Am Coli Cardiol 1982; (I 2):638. 21. UhlGS, Kay TN, HickmanJR Jr.Comparisonof exerciseradionuclide angiographyand thalliumperfusionimagmg in detectingcoronary diseasein asymptomatic men. JCardiacRehab 1982;2:118-24.

30. ThompsonAJ, FroelicherVF. Normal coronaryangiographyin an aircrewmanwith seriale xercisetestchanges.Aviat SpaceEnviron Med 1975;46:69-76. 31. Bruce RA, DeRouen TA, Hossack KF Pilot studyexammmgthe motivational effectsof maximale xercisetesting to modify risk factors and healthhabits.Cardiology1980;66:1 11-22.