BrHeart3r 1984; 52: 266-71

Significance of pulmonary valve prolapseA cross sectional echocardiographic study

PETER J ROBINSON, RICHARD K H WYSE, FERGUS J MACARTNEY

From the Department ofPaediatric Cardiology, The Hospitalfor Sick Children, London

SuMMARY Many patients with congenital heart disease now undergo cardiac surgery based solely onclinical and echocardiographic findings, but those with intracardiac shunts still frequently requirecardiac catheterisation because there is no reliable non-invasive method of measuring the pulmonaryartery pressure. Blinded to the haemodynamic results two independent observers retrospectivelystudied the cross sectional echocardiograms of 59 patients with uncomplicated ventricular septaldefect to assess whether diastolic backward bowing of the pulmonary valve leaflets towards the rightventricular outflow tract (pulmonary valve prolapse) was associated with pulmonary hypertension.There was considerable interobserver variation in the diagnosis of pulmonary valve prolapse, butconcordance was achieved in 27 cases. Mean pulmonary artery systolic and mean and diastolicpressures and the ratios of aortic to pulmonary artery mean pressures were all significantly higher forthe group with pulmonary valve prolapse diagnosed by both observers than for the group without,thus showing an association between pulmonary valve prolapse and pulmonary hypertension.Further studies are warranted to determine the usefulness of this cross sectional echocardiographicsign in routine clinical practice.

In many patients with congenital heart disease therelevant intracardiac anatomy can be clearly demon-strated by modem high resolution cross sectionalechocardiography, but cardiac catheterisation remainsnecessary in many cases to assess the degree of pul-monary hypertension. In an attempt to obviate theneed for these often repeated invasive proceduresmany studies have been performed predominantlyusing M mode echocardiographic observations ofpulmonary valve motion. A few non-invasive markersof increased pulmonary artery pressure have beenidentified.'-5 None of these has, however, found aroutine place in clinical practice.

It is well known that severe pulmonary hyperten-sion can cause clinically evident pulmonary regurgita-tion. We had noted during routine cross sectionalechocardiography that some patients with pulmonaryhypertension showed diastolic backward bowing orprolapse of the pulmonary valve leaflets and presumedthat this might ultimately result in pulmonary incom-petence. This observation prompted us to test thehypothesis that the cross sectional echocardiographic

Requests for reprints to: Professor F J Macartney, The Hospital forSick Children, Great Ormond Street, London WC1N 3JH.

Accepted for publication 24 April 1984

appearance of pulmonary valve prolapse would beassociated with pulmonary hypertension in a group ofpatients with a single anatomical lesion (that is, ven-tricular septal defect) and, if this was confirmed, toassess the reliability of detecting pulmonary hyperten-sion by this approach.

Patients and methods

To define pulmonary valve prolapse one of us retro-spectively studied the cross sectional echocardiogramsof 38 patients with atrioventricular septal defects andknown haemodynamics as well as 30 clinically normalpatients who had not undergone cardiac catheterisa-tion. These constituted the learning set, and theresults are not presented. Pulmonary valve prolapsewas defined as diastolic backward bowing of the pul-monary valve towards the right ventricular outflowtract beyond a horizontal line joining the points ofattachment of the pulmonary valve leaflets to theirannulus (Figs. 1 and 2).Our initial study population consisted of all patients

with simple ventricular septal defects (excluding thosewith atrioventricular septal defects) who had under-gone both cardiac catheterisation and cross sectionalechocardiography at our hospital between September

266

on 21 April 2019 by guest. P

rotected by copyright.http://heart.bm

j.com/

Br H

eart J: first published as 10.1136/hrt.52.3.266 on 1 Septem

ber 1984. Dow

nloaded from

Sign#icance of echocardiographic pulmonary valve prolapse

RVOT

MPA

Fig. 1 Diagrammatic representation of the definition ofpulmonary valve prolapse as seen in the parastermal short axiscross sectional echocardiographic cut. Dotted line joins the pointsofatachments of the pulmonary valve leets to the annulus.RVOT, right ventricular ou#¶ow tract; MPA, main pulmonaryartery; PVP, diastolic position ofpulmonary valve leklets whenthere was pulmonary valve prolapse; N, diastolic position ofpulmonary valve leflets when there was no pulmonary valveprolapse.

1979 and June 1983. Patients with coexistent persis-tent ductus arteriosus were, however, excluded sincediastolic blood flow from aorta to main pulmonaryartery may affect pulmonary valve motion.6 We alsoexcluded patients who were studied after repair of theventricular septal defect or who had pulmonarystenosis.

Fifty nine patients met these criteria, and theseyielded 61 echocardiograms for review. Theseechocardiograms were examined by the same observerwho had formulated the definition of pulmonary valveprolapse and by a second echocardiographer who hadnot previously studied pulmonary valve motion indetail. Although both observers used the samedefinition of pulmonary valve prolapse, observer 1was prepared to identify the points on the oppositesides of the pulmonary trunk from which the pulmo-nary valve leaflets originated on different cross sec-tional echocardiographic frames, whereas observer 2required that both these points be identified on thesame stop frame. A classification of unknown wasused if a decision could not be made because the qual-ity of the echocardiogram prevented adequate visual-isation of the pulmonary valve. When the echocar-diograms were reviewed neither observer was awareeither of the other's opinion or of the patient'scatheterisation data.

All cross sectional echocardiograms were recordedusing an Advanced Technology Laboratory (ATL)500 Mark V mechanical sector scanner with a 3-0MHz transducer and recorded on Sony U-matic 1inch tape. These were reviewed on a Sony U-maticvideo cassette recorder (model VO-5800 ps) which hasthe facility for isolated stop frame analysis and vati-

Fig. 2 Subcostal short axis cross sectional echocardiograhic cutsof(a) normalpulmonary valve end diastolic position and (b) enddiastolic position ofpulmonary valve klqlets when there ispulmonary valve prolapse. RVOT, right ventricular outlet tract;MPA, main pulmonary artery; PV, pulmonary valve lWlets;RV, right ventricle; LV, kft ventrick; VSD, ventricular septaldefect; s, superior; a, anterior; r, right.

able play back speed regulation. The pulmonary valvewas usually best visualised in the parasternal shortaxis cut. Occasionally, the subcostal right ventricularoutflow short axis cut or suprasternal examination ofthe long axis of the pulmonary trunk gave the bestview of pulmonary valve motion and was thereforeused. Cardiac catheterisation and angiocardiography

267

on 21 April 2019 by guest. P

rotected by copyright.http://heart.bm

j.com/

Br H

eart J: first published as 10.1136/hrt.52.3.266 on 1 Septem

ber 1984. Dow

nloaded from

Table 1 Comparison ofmean haemodynamic values between categories ofechocardiograms according to whether pulmonary valveprolapse was present, absent, or unknown for two different observers

Category Pulmonary artery pressures (mm Hg) Ratio ofPA mean Ratio ofPA systolic to Aoto Ao mean pressure systolic pressure

Systolic Diastolic Mean

Observer IPresent (Y) 70 29 45 0-76 0-82Absent (N) 35 16 22 0-29 0-34p value <0.001 <0 001 <0-001 <0-001 <0-001

Present (Y) 70 29 45 0-76 0-80Unknown (U) 50 17 31 0-50 0 60p value <0-05 <0-01 <0 05 NS NS

Absent (N) 35 16 22 0-29 0-34Unknown (U) 50 17 31 0-50 0-60p value NS NS NS NS NS

Present (Y) 70 29 45 0-76 082Absent (N) andunknown (U) 41 16 26 0-37 0 43

p value <0.001 <0-001 <0-001 <0001 <0-001

Observer 2Present (Y) 69 27 44 0-74 0 79Absent (N' 57 24 36 0-60 0-69Difference NS NS NS NS NS

Present (Y) 69 27 44 0-74 079Unknown (U) 55 23 36 056 062Difference/p value <0.05 NS NS <0-05 <0-05

Absent (N) 57 24 36 069 079Unknown (U) 55 23 36 0-62 0-65Difference NS NS NS NS NS

Present (Y) 69 27 44 079 074Absent (N) andunknown (U) 56 23 35 0-65 0-58

Difference/p value <0 05 NS <0 05 <0 05 <0 05

PA, pulmonary artery; Ao, aortic.

were performed using standard techniques. Oxygenconsumption was not routinely measured.

In analysing the results, the haemodynamic vari-ables were categorised according to whether pulmo-nary valve prolapse was present (Y), absent (N), orunknown (U) for observer 1, observer 2, and forthe combination of both the observers. Thehaemodynamic variables examined were pulmonaryartery systolic pressure, pulmonary artery mean pres-sure, pulmonary artery diastolic pressure, the ratio ofpulmonary to aortic systolic pressures, and the ratio of

Table 2 Observers' opinions of whether pulmonary valveprolapse was present, absent, or unknown. Figures are numbersofechocardiograms

Observer I

Present Absent Unknown Total(1') (N) (U)

Observer 2:Present (Y) 17 1 1 19Absent (N) 11 5 2 18Unknown (U) 12 7 5 24Total 40 13 8

pulmonary to aortic mean pressures. Haemodynamicresults were analysed on an Amdahl mainframe com-puter using a statistical package (SPSS version 7-0,Northwestern University) and the standard unpaired ttest.

Results

The time between performance of the cross sectionalechocardiograms and cardiac catheterisation was0-912 (mean 23*5) days. This interval was -1 day in39(64%) patients, 2 days in nine (15%), and >2 daysin 13 (21%). When those whose echocardiogramswere recorded 6 1 day before the haemodynamicstudy were considered, the overall results were thesame as for the total group. For this reason all 61echocardiograms (59 patients) are included in theanalysis.

Table 1 shows the haemodynamic values for theechocardiograms according to whether pulmonaryvalve prolapse was present (Y), absent (N), orunknown (U).

268 Robinson, Wyse, Macannq

on 21 April 2019 by guest. P

rotected by copyright.http://heart.bm

j.com/

Br H

eart J: first published as 10.1136/hrt.52.3.266 on 1 Septem

ber 1984. Dow

nloaded from

Signjifcance of echocardiographic pulmonary valve prolapse

Table 3 Comparison ofhaemodynamic values where there was observer concordance on whether pulmonary valve prolapse waspresent or absent

Pulmonary valve Pulmonary artery pressures (mm Hg) Ratio ofPA mean Ratio ofPA systolic toAoprolapse to Ao mean pressure systolic pressure

Systolic Diastolic Mean

Present (Y):Mean 71 28 47 0-78 082Range 35-100 15-50 25-70 0-5-1-0 044-1.0No of patients 17 17 17 14 15

Absent (N):Mean 38 14 24 0-30 0-36Range 25-50 10-20 15-30 0-19-0-39 0-21-0-53No of patients 5 5 5 5 5

Significance* <0-001 <0-02 <0 005 <0-001 <0-001

*Student's unpaired t test.

OBSERVER 1Pulmonary valve prolapse was diagnosed as present(Y) on 40 echocardiograms, absent (N) on 13 andunknown (U) on eight. The arithmetic means of allthe haemodynamic values were significantly different(p<0001) when the category (Y) was compared withcategory (N). When category (Y) was compared withcategory (U) all the arithmetic means were alsosignificantly different (p<005) except for the pulmo-nary to systemic systolic and mean pressure ratios.

1 00-

0--.,

a 075

o

U)

ci

E

0

< 025-

0

0

0

0@

0

0@

0

0

0

0

YY NN UU

Fig. 3 Comparison between pulmonary valve prolapse groupswhere there was observer concordance for the ratio ofpulmonaryarteiy mean pressure to aortic mean pressure (PA.Ao meanpressure ratio). YY, both observers agree pulmonary valveprolapse present; NN, both observers agree pulmonary valveprolapse absent; UU, both observers agree neither pulmonaryvalve prolapse nor its absence could be diagnosed. In three ofthe17 patients in whom there was observer agreement that there wasprolapse (YY) no aortic mean pressure was availabk so no pointsare plotted.

There was no significant difference betweenhaemodynamic values within categories (N) and (U).When category (Y) was compared with the combinedcategories (N) and (U) the mean values weresignificantly different (p<0.00l)

OBSERVER 2Pulmonary valve prolapse was diagnosed as present(Y) on 19 echocardiograms, absent (N) on 18, andunknown (U) on 24. No statistically significant differ-ence was present in the haemodynamic values for thecategories (Y) and (N). When category (Y) was com-pared with category (U) only pulmonary artery systolicpressure, pulmonary to aortic systolic ratio, and pul-monary to aortic mean ratio showed significant differ-ences (p<0.05). The mean values for categories (N)and (U) were not significantly different for any of thehaemodynamic variables. When categories (N) and(U) were combined and compared with category (Y)significant differences (p<0-05) were found for pul-monary artery mean pressure, pulmonary artery sys-tolic pressure, pulmonary artery to aortic systolicpressure ratio, and pulmonary artery to aortic meanpressure ratio.

COMBINED RESULTSThere was agreement between observers in 27 of the61 (41%) reviewed echocardiograms (Table 2); in 17pulmonary valve prolapse was identified as present(Y), in five as absent (N), and in five as unknown (U).Observer 2 was three times as likely to be undecidedas observer 1. For these 27 cases; the haemodynamicvalues for category (Y) were compared with those forcategory (N) (Table 3). There were significantly dif-ferent values for all haemodynamic variables but therewas considerable overlap in all the ranges except thatfor pulmonary artery to aortic mean pressure ratio; noechocardiogram in category (N) had a ratio >0-40 andnone in category (Y) had a ratio <0-5 (Fig. 3). Thisfinding was highly significant (p<0-001).

269

on 21 April 2019 by guest. P

rotected by copyright.http://heart.bm

j.com/

Br H

eart J: first published as 10.1136/hrt.52.3.266 on 1 Septem

ber 1984. Dow

nloaded from

270

Discussion

Despite major advances in the non-invasive diagnosisof congenital heart disease due to the introduction ofcross sectional echocardiography, cardiac catheterisa-tion is still necessary in many cases for the reliablediagnosis of pulmonary hypertension. In our unitmany patients now undergo surgery based on clinicaland cross sectional echocardiographic findings alone,7but we, like others,' still find it necessary for many

patients with left to right shunts to have invasivemeasurement of pulmonary artery pressure.

Previous attempts to predict pulmonary hyperten-sion non-invasively have been unreliable or beyondthe routine resource and time constraints of most car-

diac centres. Most of this work has involved the use ofsystolic time intervals measured from surface elec-trocardiography combined with M mode echocardi-ography of the tricuspid and pulmonary valves.Unfortunately, many of the results have been conffict-ing, as Silverman et al have previously discussed.'They tested the predictive value of the ratio of rightventricular pre-ejection period to right ventricularejection time and found no statistically significant cor-

relation between this ratio and either pulmonary vas-

cular resistance or pulmonary artery mean pressure.'

They also discussed the conffict in published studiesregarding the reliability of pulmonary valve "a" wave

amplitude in the prediction of pulmonary hyperten-sion.Boyd et al have highlighted other limitations of sys-

tolic time intervals.2 They noted that a raised rightatrial pressure, overt right heart failure, or right bun-dle branch block changed the measurements so thatthere was no longer a correlation with pulmonaryartery pressure.

Other M mode echocardiographic features of pul-monary valve motion have been related to pulmonaryartery pressure but again with confficting results. Lewand Karliner found that mid-systolic notching of thepulmonary valve was 1000/o specific for pulmonaryhypertension,3 but Heger and Weyman noted that isolated dilatation of the main pulmonary artery maycause the same finding in the absence of pulmonaryhypertension.8 Nanda et al found a correlation be-tween the pulmonary valve e-f slope and pulmonaryhypertension,4 but Weyman et a19 and Lew and Kar-liner3 found it an unreliable sign owing to overlapwith normal values. The only M mode echocardio-graphic predictor of pulmonary hypertension yet to bechallenged is the finding of Serwer et a15 that inpatients with ventricular septal defect the absence ofany right to left shunt on contrast echocardiographywas strong evidence against pulmonary hypertension.

Other factors may also affect some of these vari-ables, further reducing their reliability in predicting

Robinson, Wyse, Macartney

pulmonary artery pressure. Kerber et al showed thatcardiac output, heart rate, and concurrent drugtreatment each had an effect on the ratio of right ven-tricular pre-ejection period to ejection time.'0 Fur-thermore, Green and Popp studied pulmonary valvemotion in relation to other anatomical factors with Mmode and cross sectional echocardiography and foundthat there was a relation between the motion of theposterior aortic wall and that of the pulmonary valvein late diastole. They also showed that the size of thepulmonary valve "a" wave was related to motion ofthe entire cardiac base as well as to left atrial volumechanges.

Stevenson et al used Doppler ultrasound to predictpulmonary hypertension in patients with persistentductus arteriosus.'2 They found that when ductalblood flow occurred throughout the whole of diastolethere was a normal pulmonary pressure. Neverthe-less, there was a non-linear correlation betweendegrees of raised pulmonary pressure and the degreeof abbreviation of ductal flow. More recently, using anomogram described by Burstin,'3 Hatle et al showeda linear correlation between pulmonary artery systolicpressure and the interval between pulmonary valveclosure and tricuspid valve opening identified byDoppler ultrasound.'4 Limitations of this techniquewere similar to those in previous studies with systolictime intervals. It proves unreliable when there is anincreased right atrial pressure, increased right ven-tricular end diastolic pressure, tricuspid regurgita-tion, very low cardiac output, or pulmonary regurgi-tation, all of which can occur with chronic pulmonaryhypertension. Furthermore, recording the time ofpulmonary valve closure was not always easy in theirstudy. Great accuracy in this is vital since a 10 mserror in timing will create an error of 10 mm Hg inpressure estimation. This technique is thereforeunlikely to become useful clinically because of thedifficulty in making the measurements and the largescope for error.

Studies assessing a non-invasive measurementagainst a gold standard are best tested with the inde-pendent opinions of more than one observer, eachblinded to the true results. Few of the studiesreviewed above fulfilled these criteria so it is perhapsnot surprising that so few of the results reported havebeen reproducible by other workers. In only two ofthe studies2 12 was it stated that the echocardiographicmeasurements or observations were made before inva-sive testing-that is, blinded to the true results. Somemissed the opportunity to test interobserver error,instead using the consensus of all investigators toarrive at their opinions.-' 12

In our study two independent observers comparedthe occurrence of pulmonary valve prolapse with vari-ous haemodynamic pressures and ratios which might

on 21 April 2019 by guest. P

rotected by copyright.http://heart.bm

j.com/

Br H

eart J: first published as 10.1136/hrt.52.3.266 on 1 Septem

ber 1984. Dow

nloaded from

Significance of echocardiographic pulmonary valve prolapse

be expected to affect diastolic pulmonary valve one indicant of pulmonary hypertension along withmotion. Pulmonary arteriolar resistance was not several other preferably independent non-invasiveincluded among these because oxygen uptake is not measures a multivariate model might have a higherroutinely measured during cardiac catheterisation in predictive value.our laboratories and assumed values in small patients In conclusion, we have shown for the first time thatwith varying degrees of sedation and cardiac output the cross sectional echocardiographic demonstrationare unreliable.15 of pulmonary valve prolapse is associated with pul-Our results clearly show that pulmonary valve pro- monary hypertension. Nevertheless, further prospec-

lapse is associated with pulmonary hypertension. tive studies are required to evaluate its use as a reliableAlthough it is not clear whether systolic, diastolic, or echocardiographic sign in patients with left to rightmean pressure is the most important determinant shunts.there was no overlap for the pulmonary artery to aor- PJR is supported by a Weilcome Research Fellowshiptic mean pressure ratio between the groups with and and FJM and RKHW by the Vandervell and Britishwithout prolapse. In theory, diastolic (closing) pres- Heart Foundations. The study was also supported insure is the most relevant. part by the Child Health Research Appeal Trust.The results also show that without further

improvement in the resolution of cross sectional Referencesechocardiography it is unlikely that the observation of 1 Silverman NH, Snider AR, Rudolph AM. Evaluation of pulmo-the presence or absence of pulmonary valve prolapse nary hypertension by M-mode echocardiography in children withwill prove to be either a specific or sensitive marker ventricular septal defect. Circulation 1980; 61: 1125-32.for pulmonary hypertension. The thickness of the 2 Boyd MJ, Williams IP, Turton CWG, Brooks N, Leach G, Mil-pulmonary valve in most of these patients was such as lard FJC. Echocardiographic method for the estimation of pul-tolie between the *Lmits of the a*dal and lateral resolu- monary artery pressure in chronic lung disease. Thorax 1980; 35:to lie between the limits Of the axial and lateral resolu- 914-9.

tion of the equipment. As a result, in suprasternal 3 Lew W, Karliner JS. Assessment of pulmonary valve echogram incuts the coapted edges of the leaflets often appeared in normal subjects and in patients with pulmonary arterial hyperten-diastole as a free floating line in the centre of and sion. Br HeartJ 1979; 42: 147-61.

parallel with the pulmonary t.k, t.e .remainder 4 Nanda NC, Gramiak R, Robinson TI, Shah PM. Echocardiog-parallel with the pulmonary trunk, the remainder of raphic evaluation of pulmonary hypertension. Circulation 1974;the leaflets being invisible. By contrast, in short axis 50: 575-81.cuts the most easily identified portion of the leaflets 5 Serwer GA, Armstrong BE, Anderson PAW, Sherman D, Bensonwas neither the coapted segments nor the origin of the DW Jr, Edwards SB. Use of contrast echocardiography for evalu-ation of right ventricular hemodynamics in the presence of yen-leaflets but the region in between, lying in the short tricular septal defects. Circulation 1978; 58: 327-36.axis of the pulmonary truklk. 'Such views as were 6 Fisher EA, Sepehri B, Barron S, Hastreiter AR. Echocardiog-obtained of the origin of the leaflets from the pulmo- raphic diastolic flutter of the pulmonary valve in isolated patentnary trunk were often only fleeting. It is clear that the ductus arteriosus. Chest 1982; 81: 74-7.requirement for the leaflet origins on both sides of the 7 Stark J, Smallhorn J, Huhta J, et al. Surgery for congenital heartrequirmentorth leaflt oriins o both ides f the defects diagnosed with cross-sectional echocardiography. Circula-

pulmonary trunk to be seen simultaneously (which is tion 1983; 68 (suppl II): 129-38.essential if the recognition of prolapse is to be truly 8 Heger JJ, Weyman AE. A review of M-mode and cross-objective) resulted in the classification of many cases, sectiona echocardiographic findings of the pulmonary valve. Jbohobje tive) pstv n reneaiepoas a

Clin Ultrasound 1979% 7: 98-107.both of true positive and true negative prolapse, as 9 Weyman AE, Dillon JC, Feigenbaum H, Chang S. Echocardiog-unknown. For a technique to be generally adopted it raphic patterns of pulmonic valve motion with pulmonary hyper-is essential that it should be as objective as possible. tension. Circulation 1974; 50: 905-10.

There are, however, four reasons why we believe 10 Kerber RE, Martins JB, Barnes R, Manuel WJ, Maximov M.Effects of acute hemodynamic alterations on pulmonic valvepulmonary valve prolapse may merit further study as motion. Experimental and clinical echocardiographic studies.a marker for pulmonary hypertension. Firstly, the Circulation 1979; 60: 1074-81.resolution of cross sectional echocardiographic 11 Green SE, Popp RL. The relationship of pulmonary valve motionequipment will doubtless improve and may make vis- to the motion of surrounding cardiac structures: a two-

dimensional and dual M-mode echocardiographic study. Circula-ualisation of the entire pulmonary valve a feasible tion 1981; 64: 107-12.proposition in all patients. Secondly, this was a 12 Stevenson JG, Kawabori I, Guntheroth WG. Noninvasive detec-retrospective study. Had the primary intention of the tion of pulmonary hypertension in patent ductus arteriosus byechocardiographic examination been to demonstrate pulsed Doppler echocardiography. Circulation 1979; 60: 355-9.13 Burstin L. Determination of pressure in the pulmonary artery bythe pulmonary valve, the results might well have been external graphic recordings. Br HeartJ 1967; 29: 396-404.better. Thirdly, the association between pulmonary 14 Hatle L, Angelsen BAJ, Tromsdal A. Non-invasive estimation ofvalve prolapse and pulmoiary hypertension might pulmonary artery systolic pressure with Doppler ultrasound. Brhave been closer had echocardiograms been recorded Hearj 1981; 45: 157-65.15 Kappagoda CT, Greenwood P, Macartney FJ, Linden RJ.at the time of cardiac catheterisation in every case. Oxygen consumption of children with congenital diseases of theFinally, if pulmonary valve prolapse were taken as heart. Clin Sci Mol Med 1973; 4S: 107-14.

271

on 21 April 2019 by guest. P

rotected by copyright.http://heart.bm

j.com/

Br H

eart J: first published as 10.1136/hrt.52.3.266 on 1 Septem

ber 1984. Dow

nloaded from