A o L V OH yC aU

7A na P A y o

i

T h rE cAlessandro Salustri, MD, PhD, Marcel J.M, Kofflard, MD, Jos R.T.C Roelandt, MD, PhD,

Youssef Nosir, MD, Giuseppe Trocino, MD, David Keane, MB, MRCPI, Wim B. Vletter, BSC,and Folkert J. Ten Cate MD, PhD

Thisstud analyzesthealterationsinsizeandgeometryJof the1 ventricular(LV)outflowtractthatoccurinhy-

pertrophiccardiomyopathy(HC)usingtransthoracic3-dimensionalechocardiography.Transthoracic3-dimen-sionalechocardiographywasperformedin patienkwithHC(4 aftermyectomy)andin 10normalsub&ts.imageswere acquiredwith the rotationalapproach,withelectrocardiographicandrespiratorygating.Fromthe3-dimensionaldatasets,short-axisparallelslicingoftheLVoutflow a 1 performedat theonsetof systole.Foreachslice,cross-sectionalareaand maximaland minimaldiameterwerecalcu-lated.Reconstructionof theLVoutflowtractcouldbedisplayedin 3 dimensionsin all patients,allowingori-entationandcieardefinitionof theirregulargeometry.InpatientswithHC,theminimalLVoutflowtractcross-sectionalareawassmallerthaninnormalsub@cts(2.3

Postmortem studies and intraoperative findingsindicate that 2-dimensional echocardiography

may fail to give the full picture of the left ven-tricular (LV) outflow tract in patients with hy-pertrophic cardiomyopathy (HC).l-C Three-di-mensional echocardiography is a new imagingmodality which provides unique information onthe spatial geometry of a given structure.7 Basedon our experience in an unselected population, 8’9we believe that the alterations in size and geom-etry of the LV outflow tract in patients with HCcould be more accurately analyzed with precor-dial 3-dimensional echocardiography. With theseconcepts in mind, we designed this study to gainfurther insight into the complex geometry of theLV outflow tract in patients with HC. With thisaim, analysis of the LV outflow tract was per-formed using 3-dimensional datasets obtained af-ter acquisition of precordial echocardiographicimages.

From theDepartmentof Cardiology, Thoraxcenter, University HospitalRotterdam-Dijkzigt and Erasmus University, Rotterdam,The Nether-lands. Dr. Nosir is supportedby the Nuffic, The Netherlands, and theDepartment of Cardiology, A1-Azhar University Hospital, Cairo,Egypt. Dr Keane is the recipientof a travel grant from Peel MedicalResearch Trust, London, United Kingdom. Manuscript received No-vember 20, 1995; revised manuscript receivedand acceptedFeb-ruary 20, 1996.

Address for correspondence:Folkert Ten Catel MD{ Thoraxcen-ter, Ba 350, Dr. Molewaterplein, 40, 3015 GD Rotterdam,TheNetherlands.

t 1.0vs5.0 t .0.9cm2,p <0.0001).Theratiobetweenmaximalandminimalcross-sectionalareaswashigher”in patientswithHCthanin normalsub@cts(2.6 t 0.9vs 1.4* 0.2, p <0.0001). ThemtiobetweenmaximalandminimaldiameterofthesmallestcrosssectionoftheLVoutflowtractwasalsosignificantlyhigherinpatienkwithHCthaninnormalsub@cts(1.6* 0.3vs1.2~ 0.1,p <0.001); a valueof 1.36separatednormalsubieckfromHCpatientswithoutpreviousmyectomy.Inconclu-sicm,precordial3-dimensionalechocardiographyal-lowsdetailedqualitativeand quantitativeinformationontheLVoufflowtract.PatientswithHCarechamcter-izedby a highlyeccentricandasymmetricshapeof theLVoutflowtract,and by a smallerminimalcross-sec-tionalareathanthatseenin normalsub@s.

(AmJCardiol1996;78:462-468)

METHODSstudypatients:We prospectively selected 17 pa-

tients (13 men and 4 women; mean t SD age 39 t15 years, range 19 to 65) with HC referred to theoutpatient clinic of our institution for routine trans-thoracic echocardiographic follow-up. High-qualityimages were a prerequisite for inclusion in this study.The diagnosis of HC was based on M-mode and 2-dimensional echocardiographic demonstration of anondilated hypertrophic left ventricle in the absenceof other cardiac or systemic disease that could pro-duce LV hypertrophy.10According to a previously

established classification, 1 the patterns of distribu-tion of LV hypertrophy were: type I, 1 patient; typeII, 3 patients; type III, 12 patients; and type IV, 1patient. Systolic anterior motion of the mitral valvewas present in 12 patients and its severity was eval-uated semiquantitatively from O= absence, to 3+ =contact with the interventricular septum during sys-tole.5 At the time of the echocardiographic study, apressure difference was calculated from Doppler LVoutflow tract velocity recordings, and obstruction(gradient >30 mm Hg under basal conditions) wasdetected in 4 patients. A septal myectomy had beenperformed in 4 patients. Ten asymptomatic subjectswithout evidence of LV hypertrophy were also stud-ied for comparison. These controls were aged 20 to49 years (mean 28 f 8) and 8 were men.

Examinationprocedure:Two-dimensional echocar-diographic studies were performed with a commer-

4 60 1 9b E x c e rM e d iI nA ir i g hr e s e r

0 0 0P S 0 0 0

FIGURE1.Analysisofbftverrtricular(LWoutflowtractina patientwithhy-pertrophiccardiomyopathy.Fromthe3-dimensionaldatoset,parallelslicingoftheLVoutflowtract

‘*alonQtiis(~m~K:~:;:::rdiagraphy)isPerfarmdat1mmin-tenfals,attheonsetofventricularsystale.IntheIefl 3 represrsn-tativecut-planesareindicated.tiecorrespondingcross-sectional2-di-mensionalimageswiththemanuallytracedendacardialcontoursareshowninthem p Thefinalwireframemaskdisplayrepresentedinthe & rotatedonthescreenalongthe3mainaxesforversatilequalitative3-dimensionalevaluation.AML= anteriormitralleaflet;AV = aorticvalve;IVS= in-twventricularseptum;.

cially available system (Vingmed CFM 750 IHorten,Norway] or Toshiba Sonolayer SSH-140A [Tokyo,Japan]) equipped with a 3.5 MHz transducer, whilethe patient was lying in the 45° left recumbent po-sition. After follow-up 2-dimensional echocardio-graphic study, the probe was positioned either at theleft parasternal or apical window for acquisition oftomographic images of the LV outflow tract for 3-dimensional reconstruction. Patient movement dur-ing the image acquisition can be prevented by thor-oughly explaining the procedure before the study.The operator has to find the central axis of rotationso that the conical datasets encompass the LV out-flow tract. During the acquisition, movements of thetransducer holder must be avoided diligently. Mirrorimages of the first and the final cut-plane indicate acorrect 180° rotation. All subjects gave informedconsent.

Three-dimensionalechocardiagraphy:IMAGE ACQUI-SITION:After the standard 2-dimensional examina-tion, the video output of the echocardiographic sys-tem is interfaced to the acquisition system(Echo-scan, TomTec GmbH, Munich, Germany)and the transducer is fixed into a cylindrical holder.A step-motor mounted on this holder can rotate theprobe around its longitudinal axis via a wheel-workinterface. The step-motor is connected to the acqui-sition system which commands the rotation of theprobe at 2° intervals over a span of 180°.Respiratoryand electrocardiographic gating are performed afterthe operator has selected the end-expiratory phase bythoracic impedance measurement and an adequateRR interval. Thus, only beats falling in the predeter-mined RR interval and at the end-expiratory phaseare selected by the steering logic of the system andacquired. Cycles that do not meet the preset rangesare rejected. Ninety sequential cross-sections are ac-quired, each during a complete heart cycle, encom-passing a 3-dimensional conical volume. After dis-tance calibration, images are stored in the computermemory for subsequent analysis.

IMAGE PROCESSING:The raw data are resampledoff-line according to their temporal and spatial lo-

cation. The coordinates of each point are convertedfrom a polar to a rectangular system, ,md the spacebetween contiguous points is electronically filledwith a trilinear cylindric interpolation. Several al-gorithms are used to reduce noise and artifacts whichcan be created by patient and probe movements.

IMAGE DISPLAYANALYSIS:Images were analyzedas follows:

1. After thorough examination of the 3-dimen-sional datasets with analysis of cardiac cross-sec-tions in any desired plane ( “anyplane echocardiog-raphy” ), a cut-plane is selected where the LVoutflowtract is visualized along its longitudinal axis.

2. The onset of ventricular systole is selected, asthe first frame during the cardiac cycle in which themitral valve appears closed.

3. Electronic parallel slicing ( “paraplane echo-cardiography’ ) of the LV outflow tract perpendic-ular to the vertical axis is performed at 1 mm inter-vals, from the hinge point of the anterior mitralleaflet to the point of coaptation of the mitral leaflets,and the corresponding 2-dimensional images (boththe stop frame at onset of systole and the dynamicsequence) are displayed.

4. On the stop frame image at the onset of systole,the endocardial contour of the cross-section is man-ually traced for automatic area measurement, andmaximal and minimal diameters are also measured.

5. Finally, reconstruction of the LV outflow tractat the onset of ventricular systole is displayed in wireframe mode and the representative image can be ro-tated on the screen for versatile 3-dimensional eval-uation (Figure 1).

From analysis of the 3-dimensional datasets, thefollowing measurements of the LV outflow tract atthe onset of systole were considered: ( 1) minimalcross-sectional area; (2) ratio between maximal andminimal cross-sectional area (max/min cross-sec-tional area), as an index of the “eccentricity” of theLV outflow tract (the higher the value, the greaterthe variations of the cross-sectional area throughoutthe length of the LV outflow tract; (3) ratio betweenmaximal (lateromedial) and minimal (anteroposter-

CARD10MYOPATHY/3-D ECHOCARDIOGRAPHY IN HYPERTROPHIC CARDIOMYOPATHY 463

TABLEI D e m o g r a pP a t iC h a r a c t ea M e a s uo t L V e n tO uT i P aW H yC a r d i o m y o

P a t i eN u m bA g( y& S eT yL T h eG r aS C ( cM aC M D

1 1 9I v 58 3+ 2.9 2.0 1.92 21M I — <lo 2+ 2.4 2.5 1.73 22M I — <lo 1+4

3.0 1.8 1.424M iv — <lo 2+

52.9 2,4

25M1.6

Ill v <lo 2+ 2.7 2,46 31F Ill

1.6s,v 50 3+

71,3 4.2

38F2.0

II v <lo 08

2.4 2.7 1.739F Ill v <lo 0 3.2 2.1 1.6

9 A II v 20 1+ 1.9 1.710 46F I

1A V <lo 2+ 0.7 4.1

111.4

5 3I <lo 1+ 1 .12

3.8 1.963M II M– 100 3+ 1.5 3,0

132.6

65M I v 35 3+ 2.0 3.0 1.6M e a? S 3 ~ 1 2 ? 0.7 2.7? 0.8 1.75t 0,314 31M* Ill — <lo 0 2.1 2.415

1.334M* Ill s <lo 1+ 0.9 3.9

161.2

44M* I v <lo 0 4.7 1.517

1.359M* HI v <lo 0 3.0 1.6 1.3

M e a~ S 4 ~ 1 2 f 1 2 * 1 1 + 0M e at S ( o v e r3 * 1 2 t 1 2 ? 0 1 f 0

* P e a m

A = a C = c ra = l v h M = m M = m S = s = s

a m V = v

ior) diameter (max/min diameter) at the level of thecross-section with the minimal area, as an index ofthe “asymmetry” of the LV outflow tract (a ratio of1 indicates a circular shape, with the highest valuescorresponding to the ,more elliptical shape of thecross sections).

Inter- and intraobserver reproducibility for themeasurements of LV outflow tract with 3-dimen-sional echocardiography was assessed in all 27 pa-tients. To assess interobserver variability, two ob-servers (A.S. and Y.N.) independently,measured theoutflow tract area from the 3-dimensional datasetswithout prior knowledge of clinical data and withoutpreelection of cut-planes. In addition, LV outflowtract cross-sectional area measurements were per-formed by 1 observer (A.S.) on 2 occasions (3months apart, without preelection of cut-planesfrom the 3-dimensional datasets) to assess intraob-server variability.

analysis:Values are given as mean ~SD. Student’s unpaired ttest was used to comparethe differences between HC and control subjects.Values of p <0.05 were considered to be significant.Reproducibility of the LV outflow tract measure-ments was expressed in terms of mean differencesand 95’-ZOconfidence intervals.12

RESULTSThree-dimensional acquisition could be per-

formed successfully in all patients. Echocardio-graphic acquisition of the image of the LV outflowtract for 3-dimensional reconstruction was per-formed either from the parasternal (n = 20) or apical(n = 7) windows, according to the image quality.The examination including the calibration proce-dure, selection of the optimal axis of rotation, a num-

464 THE AMERICAN JOURNAL OF CARDIOLOGY” VOL 78

ber of test runs, and the actual image acquisition re-quired approximately 10 minutes ~nadd_itionto thetime required for the standard 2-dimensional echo-cardiogram. Three-dimensional reconstruction of theimages was possible and of good quality in all pa-tients. The time required for post-processing the rawdata, and reconstruction and analysis of the imageswas approximately 20 minutes. Demographics,echocardiographic characteristics, and measure-ments of the LV outflow tract in each patient withHC are reported in Table I.

Leftventricularoutflowtractinpatientswittshypertro-phiccardiomyopathyversusnormalsubiects:CROSS-SEC-TIONAL AREA: The values of minimal cross-sec-tional area of the individual subjects are plottedin Figure 2. The minimal LV outflow tract cross-sectional area calculated with 3-dimensionalechocardiography was significantly smaller in pa-tients with HC than in the control group (2.3 ~1.0 vs 5.0 * 0.9 cm2, p <0.0001 ). Thirteen of the17 patients with HC had a value smaller than con-trols, and 2 of the 4 HC patients (nos. 6 and 17)with higher values were evaluated after myec-tomy. After correction for body surface area thevalues were 1.3 t 0.5 and 2.7 * 0.6 cm2, respec-tively (p <0.0001).

SHAPE:From the 3-dimensional datasets, the re-constructed LV outflow tract could be displayed asobserved from different viewpoints. This simplifiesvisualization of the geometry and shape as well asthe localization of the narrowing of the LV outflowtract in patients with HC (Figure 3). A similar dis-play in a normal patient is shown in Figure 4. Someexamples of 3-dimensional reconstruction of the LVoutflow tract in normal subjects and in patients withHC are shown in Figure 5.

AUGUST 15, 1996

2cm

p<().t)()ol

6-

11$

❑

0 4-:

zE.- +v ❑ ++

E Io-

2- Va+

?+❑+

0 I I I

HCM NA ,,..-,. , .. , ,-”. . , ., ,.,. . , “L

FIGUREZ.mm,malcross-secnonalarea ventricularouttlawtractinpatientswithhypertrophiccardiomyopathy(HCM)andinnormalsub@k (0). + = patientswithoutobstruction;❑ = patienkaftermyectomy;V = withobstruction.

FIGURE3. Inthispcstienkwithhyperirophiccardiomyopathy,thewireframedisplq ofthereconstructedleftventricular(LMaut-flowtractisrepresenkdasobservedfromdifferentviewpoink.Theviewot0° correspondsiothecut-lanerepresentedinthe

Lcentral Theimagesabiuineda r incremental60°clock-wiserotationaredisplayedinthecorrespondingpanels.Thereisaneccentricandas

rmetricshapeoftheLVoutflowtract.From

theseimagesitisa soapparentthatthenarrowinislocalizedat1.themiddle-caudadpartoftheLVoutflowtractan ISmainlyre-

latedtoa reductionintheanteraposteriordiameter.

MAX/MINCROSS-SECTIONALAREA(ECCENTRICITYINDEX):The max/rnin cross-sectional area of the LVoutflow tract derived from the 3-dimensional data-sets are displayed in Figure 6. From this figure, it is

A

FIGURE4. Three-dimensionalreconstructionoftheleftventricular(LWoutflowtractina normalsubiect,withthesamedisplayasinFigure3. Noteiheuniformi~(indicatedb t minimalvariationsofifrediametersthroughoutthelengthoftheLVoutflowtract)aswellasthesymmetry(indicaieclbythesimilardiametersof indi-vidualcross-sectionsfromdifferentviewpoink)oftheLVt

apparent that patients with HC had higher ratios (2.6k 0.9), with a broad range of values (from 1.5 to4.2) indicative of many irregular different shapes ofthe LV outflow tracts. In contrast, normal subjectshad smaller ratios ( 1.4 ~ 0.2), with a narrow rangeof values (from 1.1 to 1.6). In particular, each of thecontrols had a ratio of s1.6, whereas 15 of 17 pa-tients with HC had a ratio of >1.6. The 2 patients

CARD10MYOPATHY/3-D ECHOCARDIOGRAPHY IN HYPERTROPHIC CARDIOMYOPATHY 465

\

FIGURE5.Three-dimensionalreconstructionofthekftventricularoutflowtractwithwireframedisplayinnormalsubiects(AtoCJandinpatientswithhypertraphiccardiomyopathy(0toF).Thedifferentirregularconfigu-rationof iheleftventricularoutflowtractinpatientswithhypertrophiccar-diomyopathycanbeevaluated.

with max/min cross-sectional area of <1.6 (patients16 and 17) were evaluated after myectomy. Thus, anoutflow tract area ratio of 1.6 appeared to separatepatients with from subjects without HC.

MAXmnNDIAMETER(ASYMMETRYINDEX):The in-dividual values of maxhnin diameter of the LV out-flowtract cross-sectionare displayedin Figure 7. Thisratio was significantlyhigher inpatients with HC thanin normal subjects (1.6 * 0.3 vs 1.2 t 0.1, p =0.001). Of interest,patientsevaluatedafter myectomyhad the lowest values; conversely, the highest valueswere found in patients with HC and LV outflowtractobstruction. An index of 1.36 appeared to separate

normal subjects from patients with HC whodid not undergo operation.

Reproducibilityanalysisofihree-dimensionalechocardiography:INTEROBSERVER VARI-ABILITY:The difference between the 2 ob-servers for measurements of the LV out-flow tract was compared with the averageof the 2 measurements for each patient.The mean difference between the mea-surements of the 2 observers was 0.04 cmz(95% confidence interval – 0.08 to 0.12cm2) for cross-sectional area.

INTRAOBSERVERVARIABILITY;The differ-ence between the 2 measurements made bythe same observer was compared with theaverage of the 2 measurements for each pa-tient. The mean difference between the 2measurements was 0.13 cm2 (95% confi-dence interval – 0.01 to 0.25 cm2) for cross-sectional area.

DISCUSSIC)NHypertrophic cardiomyopathy is a disease with a

great individual variability and “no two hearts arealike.” 13The results of the present study indicatethat 3-dimensional echocardiography allows visual-ization of the varied complex geometry of the LVoutflow tract in patients with HC. With quantitativeanalysis of the 3-dimensional datasets, we demon-strated that in patients with HC the minimal cross-sectional area is smaller than that in normal subjects.In addition, most patients with HC have an irregularshape of the LV outflow tract as demonstrated by aneccentricity index of =1.5. In normal subjects, this

p<o.0001

v+

❑+

+❑

()

I+

1-

0 I I I

HCM N

FIGURE6. Ratiobetweenmaxima!andminimalcross-sectionalarea(max/minCSA)oftheleftventricularout80wtractinpotienkwithhypertraphiccardiomyopathy(HCM)andinnormalsub@cts.A highmtioindicatesaneccentricshapeoftheleftventricularOUFflowti”ct.S@sals asinFigure2.

466 THEAMERICANJOURNALOF Cardiology@ VOL.78 AUGUST15, 1996

p<o.oolv

1 -

Ii310

0’ 1 I I

HCM N

FIGURE7. Ratiabetweenmaximalandminimaldiameter(max/mindiameter)measuredatthecross-sectionoftheleftventricularoutilowtractwiththeminimalareainpatientswithhyper&ophiccardiomyo&hy(HCM)andi n s A highratioindicatesanasymmetricshapeoftheleftventricularoutflowtract.SymbolsasinFigure2.

index is always s 1.6, indicating a uniform shape ofthe LV outflow tract, without significantvariation ofthe cross-sectional area throughout its length. Wehave also demonstrated that the cross-sectional shapeof the LV outflow tract is more elliptical in patientswith HC than in normal subjects, as indicated by ahigher ratio of max/min diameter measured at theplane of the minimal cross-sectional area. This find-ing is in agreement with the concept that in HC thehypertrophic ventricular septum narrows the LV out-flow tract mainly along its anteroposterior diameter.Of interest, this asymmetry index was highest in pa-tients with HC and obstruction of the LV outflowtract at rest. In contrast, in HC neither the minimalcross-sectional area nor the eccentricity index of theLV outflow tract separated this subgroup. This find-ing indicates that for similar cross-sectional area theasymmetry of the LV outflow tract plays an impor-tant role in determining the presence of significantobstruction at rest.

Patients who had undergone myectomy had aminimal cross-sectional area similar to other patientswith HC, including those with obstruction (Figure2). However, from Figure 7 it is clear that aftermyectomy the asymmetry index was lowest, indi-cating that the surgical remodeling of the LV outflowtract was adequate for the relief of the obstructiondespite the finding that the cross-sectional area re-mained small compared with that in normal controls,and remained in the same range as that of the otherHC patients without previous myectomy. Thus, inpatients with HC, precordial 3-dimensional echocar-diography has the potential to play a major role intailoring the standard surgical resection of the inter-ventricular septum to the individual patient’s anat-

omy, which is crucial for safe and efficacious per-formance of myectomy, and also for evaluation ofthe results of surge~.

Evaluationof leftventricularoutflowtractusingthree-dimensionalechacardiography:Previous experiencewith 3-dimensional echocardiography for the evalu-ation of the LV outflow tract was based on mor-phologic analysis with volume-rendered display.14Incontrast, for quantitative analysis of the LV outflowtract performed in the present study, we selected amultitude of cut-planes (‘‘anyplane echocardiogra-phy” ) and performed a parallel scanning of the LVoutflow tract at 1 mm intervals ( “paraplane echo-cardiography’ ). This rate of sampling of the datasetis similar to the analysis done with magnetic reso-nance imaging or computed tomography, and allowsdetailed spatial information,15While some displaymodalities, such as volume rendering, are indicatedmore for representation of anatomic details, 16-18thewire frame display format appears particularly suitedto 3-dimensional reconstruction of the cardiac cavi-ties, where areas, volumes, size, and shape can beadequately evaluated.

Limitationsof three-dimensionalechocardiagraphyIn this study, patients were selected on the basis ofhigh-quality images at 2-dimensional echocardiog-raphy, which yielded a success rate of 3-dimensionalreconstruction of 100%. The same results cannot beexpected from an unselected population where poorquality precordial images may prevent adequatequality of the reconstruction.

Echocardiographic images were acquired by anexperienced technician and reconstruction was per-formed by a cardiologist after a learning period of>50 studies. This previous experience of 3-dimen-

CARD10MYOPATHY/3-D ECHOCARDIOGRAPHY IN HYPERTROPHIC CARDIOMYOPATHY 447

sional echocardiography prevented artifacts in the 3-dimensional datasets, limited the time required bothfor acquisition and reconstruction, and resulted in aminimal variability for the measurements.

Acknowledgment:The expert technical assistanceof Rena Frowijn, BSC, in the preparation of the il-lustrations is greatly acknowledged.

1 H e nW LC l aC ER o b eW CM o rA E p sS D i f fid i s t r i b uo m y u c a ra b n o r m ai p a t iw io b s t ra rn oo b s t r u ca s y m m es e ph y p e r t( A SE c h o c a r dag r oa n a t of i n d iC i r c u k1 9 7 4 ; 5 0 : 42 M a rB JB o nR OC a nR I IL eM E p sS H y p e rc a r d i o m y o pI n t e r r e l ao c l i nm a n i f e s tp a t h o p ha rt h e r a( F i ro t wp a r tN E nJ k f1 9 83 1 6 : 7 83 P a r rJ AM a rT JM a rB JD e v e l oa rd e t e r mo d y no b s t r u ct l ev e n t r i co u t fi y o up a t iw ih y p e rc ad i o m y o p aC i r c u l a1 9 9 2 ; 8 5 : 1 34 K h rH G‘ R o b eW CM a rB JM o r p h o ld e t e r mo e cc a r d i o g r ap a t t eo m i tv a ls y s ta a t em o ti o b s th y p e r t r oc a r d i o m y oC i r c u l1 9 98 7 : 1 5 75 S p i r iP M a rB JS i g n i f io l ev e n t ro u tt rc r os e c t i o ra ri h y p e r t rc a r d i o m ya t w o - d i me c h o cg m p ha s s e s s mC 1 9

6.M c I n tC LM a rB JC u r ro p e r at r e a to o b s t rh y pt r o p hc a r d i o m y o pC i r c u l1 9 8 8 ; 7 8 : 4

7 S a lA R o eJ RT h r er e co t hw ir o t aa c q um ea c la p6 H J1 9 9 5 ;2 ) :8 R o eJ RS aA B e kL B rN V W Pt h r e e -e c h n cw a r oi p ma ri n ie x p eE c h o c1 1 29 S a hA S p iS M cJ W eW R oJ T rt h r e e -e c h o ci a p aw c oh de aJ C C a1 92 6 :10.M zB E pS H a d io n oA J C1 9 7 9 ; 4 31IM aB G o tJ E pS P aa s io di b uo l v e n th y pi h y pc a rA wa nt w o - de c h n cs o 1 p aA J C1 9 8 1 ; 41 B lJ A lD S t am ef a sa gbt m e to c l im e aL a1 1 :13.R o bC R oW M n rf eP C 1 214.F yD L u dA S aS D P S iM S D Lv e n to ut ro b sd eb a t h rec a r du sr o tt r a ra cE c1 9l l : @15.S a fA R o eJ RL a tm v ib t ra gE c hC i r1 9 11 R o eJ Rt C F V lW T M U ldt h r e e -v i s uo t h w a m ut ri m at r a nJ S E c h1 7 :1 P a rN R o eJ N aN S L C Q A JS c hS V aM L uA M G V M D td i m ee c h o cm ea c lp oE cr a1 91 : 21 S c hS C Q A zJ P aN S io i nv i s u ao c as t ma s o d ys a wr e at b r e e -e c h o cA J C1 7 3

THE AMERICAN JOURNAL OF CARDIOLOGY” VOL 78 AUGUST 15, 1996