The left atrial volume curve can be assessed from pulmonary vein and mitral valve velocity tracings

The left atrial volume curve can be assessed from pulmonary vein and mitral valve velocity tracings After instantaneous left atrial volume was defined as the net difference between the forward-flowing blood from the lungs and the blood flowing through the mitral valve, we constructed the left atrial volume curve by sampling the Doppler mitral valve and the right upper pulmonary vein velocity from an apical four-chamber view in eight normal subjects and 11 patients with heart disease. The instantaneous mitral valve flow was estimated as mitral valve velocity X annular area (derived from the same view), whereas the pulmonary venous flow was obtained as right upper pulmonary vein velocity X pulmonary vein area, where pulmonary vein area = mitral valve velocity integral X mitral valve area) + pulmonary vein velocity integral. The left atrial volume curve can then be derived as: [(instantaneous pulmonary venous flow - mitral valve flow) + left atrial volume assessed at end diastole by two-dimensional echocardiography]. Biplane angiographic left atrial volume curves, available in four of 11 patients, compared morphologically very closely with the noninvasive curves, whereas the correlation coefficient for maximum (end-systolic) and filling (maximum minus minimum) left atrial volumes obtained from the Doppler-derived curve and the corresponding two-dimensional echocardiographic estimates was 0.95 (p < 0.001, standard error of the estimate = 11.9 ml), the dispersion of the data increased with decreasing volumes. These data demonstrate that combined Doppler mitral valve and pulmonary vein velocities can be used to construct the left atrial volume curve in human beings. The approach described, besides providing a tool for further noninvasive evaluation of the left atrial function, offers the opportunity for relating the continuous pulmonary venous flow to the intermittent filling of the ventricle through the mitral orifice in diastole, underlining the complex role that the left atrial cavity plays in this process. (AM HEART J 1994;127:888-98.)

Paolo Marino, MD, Antonia M. Prioli, MD, Gianni Destro, MD, Isabella LoSchiavo, MD, Giorgio Golia, MD, and Piero Zardini, MD Verona, Italy

The function of the left cardiac atrium has been investigated extensively,le4 with most of the interest focused on indexes derived from changes in its pressure, whereas information on the left atria1 volume was limited to an indirect assessment of the atria1 contribution to left ventricular filling. An easy way to assessthe left atria1 volume continuously in fact is not available at present and can be obtained by invasive investigations only,5-7 yet it would be highly desirable because it may allow better understanding of the atria1 function in normal and diseased states.

From Received

the Cardiology

Division,

for publication

Reprint requests: tedra e Divisione Stefani 1, 37126 Copyright :z’ 1994 0002.8703/94/$3.00

University

Feb. 8, 1993;

of Verona, accepted

Italy.

Aug. 6, 1993.

Paolo Marina, MD, Laboratorio di Ecocardiografia, Clinicizzata di Cardiologia, Universita’ di Verona, Verona, Italy. by Mosby-Year + 0 4/l/52140

Book,

Inc.

CatP. le A.

It has previously been shown that the left ventricular volume curve can be derived from the Dopplerdetermined mitral flow velocity integral X the mitral cross-sectional area8 and that the results thus obtained correlate significantly with the estimates of the left ventricular filling volume obtained by thermodilution.g A similar approach could be applied to pulmonary vein velocities, defining the pulmonary venous flow as the product of the pulmonary veins’ integral x the pulmonary veins’ area. If this reasoning is correct, then a unifying mechanism of left atria1 dynamics could be presented where instantaneous left atria1 volume could be defined as the net difference between the forward-flowing blood from the lungs and the blood flowing through the mitral valve. In the present study we explore this concept, and we anticipate that a reliable noninvasive left atria1 volume curve can be obtained by this approach, which links the continuous flow of the pulmonary veins to

Volume 127. Number 4, Part 1 American Heart Journal

Marina

et al.

887

Table I. Patient data Patient (initials)

Age (Yr)

Sex

AP vs PA FG cs GD LG PG MM RS MC VM RA ZG MG GL MMA VIS TF

69 67 64 51 44 57 44 42 48 56 62 28 29 29 33 5 7 4 15

M M M M M M M M M M M M M M M M M M F

BPG, Bypass grafting; opathy; MI, myocardial

Previous MI

Angina

No No No No No No No No No Yes No No No No No No No No No

Anterior Anterior Anterior Anterior No No No Inferior Inferior No No No No No No No No No No

Idio tong CM’, idiopathic congestive cardiomyopathy; IHD, ischemic infarction; PZ’CA, percutaneous transluminal coronary angioplasty.

the intermittent diastolic filling through the mitral orifice.

of the ventricle

METHODS Normal subjects and patients (Table I). Eight normal subjects(meanage,18.8 f 11.5 years) and 11patients with heart disease(meanage,54.9 + 9.3 years) who had clinical evidence of ischemic heart disease(n = 6) or idiopathic (n = 1) and ischemic (n = 4) congestive cardiomyopathy formed the study population. The patient selection criteria adopted were only directed toward collecting a wide range of left ventricular and left atria1 volumes in good quality examinations. Four patients (two of whom had experienced a previous inferior myocardial infarction) had undergonea recent (within 6 months) percutaneoustransluminal coronary angioplasty, and another had undergone surgical revascularization. Another patient had effort angina. All four patients with ischemiccardiomyopathy had experienceda previous myocardial infarction. All patients were in sinusrhythm; there wasno clinical or echocardiographic evidence of aortic or mitral regurgitation except in the patients with cardiomyopathy, who exhibited mild mitral regurgitation during cardiac auscultation and pulsed Doppler flow imaging. Echocardiographic data. Echocardiographic data were recorded on a Toshiba SSH-160 or SSH-140 echocardiograph (Toshiba Medical Co. Ltd., Tokyo, Japan) equipped with a 3.75 or 2.5 MHz phased-array transducer. The pulse repetition rate usedwas4 KHz, and the width of the sample volume was 3 mm. A medium filter setting was used. Two-dimensional echocardiographic imagesand Doppler

heart

disease:

Clinical diagnosis Isch Cong CMP Isch Cong CMP Isch Cong CMP Isch Cong CMP Idio Cong CMP IHD (previous IHD (previous IHD (previous IHD (previous IHD IHD (previous Normal Normal Normal Normal Normal Normal Normal Normal Isch Gong CMP,

ischemic

PTCA) PTCA) PTCA) PTCA) BPG)

congestive

cardiomy-

signalswere recorded on a 3/4inch (Sony VO-5800P) videotape. A standard four-chamber view of the heart from the apical window was recorded first, with minor adjustments madeto maximize the mediolateral diameter of the mitral anulus.Pulsed-waveDoppler measurementswere then obtained with the Doppler beamalignedperpendicular to and bisecting the plane of the mitral anulus, with the sample volume placed just on the ventricular side of the plane of the anu1us.sSmall adjustments were made to the transducer angulation to obtain the highest peaksof filling velocities. Subsequently, the imagewasexpanded to include the roof of the left atrium in the view and thus clearly delineate the orifices of the pulmonary veins. The Doppler samplevolume was then moved from the mitral plane toward the roof of the atrium, at the junction between the interatrial septum and the wall of the atria1 cavity. To detect the pulmonary venous flow, the sample volume was placed at the orifices of the pulmonary veins,‘Owith the best vein for registering flow usually being the right upper paraseptal vein. In four normal subjects, flow from more than one pulmonary vein was alsoobtainable by rotating the transducer slightly and angulating the Doppler sampling direction to minimize the angle(<20 degrees)formed with the vessel.llAfter Doppler recordingswerecompleted, the left ventricle and the left atria1 cavity werealsoimaged in an apical long-axis view by rotating the transducer 90 degreesclockwiseto obtain a view asperpendicular aspossible to the four-chamber view. Recordingswere madeduring quiet respiration, keeping the extent of the lateral rotation of the patient fixed and

888

Marino

et al.

American

April 1994 Heart Journal

.---.--_Pulmonary vein --- -- Mitral valve

Time (ms) Fig. 1. Original cleft) and digitized fright) Doppler mitral and pulmonary venous flow velocity profiles of patient LG. The two digitized tracings have been superimposed according to the end of the QRS complex. Difference in the R-R interval between the two tracings is equal to 2.1 a,’ ,c of their mean. An ECG tracing is also shown at the bottom of the figure.

holding the position of the transducer on the thoracic wall as constant as possible during the entire procedure. For each patient the diameter of the mitral anulus was measured off-line from the video frame containing the maximal separation of the leaflets (first or second frame after mitral valve opening) on the orthogonal views on 3 consecutive beats8 with a commercially available computer system (Cardio 80, Kontron Instruments, Inc., Everett, Mass.). Mitral anulus area was calculated according to the formula: 3.14 x (Dl X D2) + 4, where Dl = diameter of the mitral anulus in the apical long-axis view and D2 = diameter of the mitral anulus in the four-chamber view. Left ventricular apex-to-base (L) and anterior-to-posterior wall distance (D), perpendicular to L at its midpoint, were also measured in the long-axis view for the same beats at end systole, defined as the preceeding frame with the smallest left ventricular cavity area. The orthogonal minor left ventricular dimension was derived as the septum-to-lateral wall distance measured at the midpoint of the ventricular long axis obtained from the four-chamber view. Left ventricular systolic biplane volume was then computed according to the formula U x 3.14[(Dl X D2) f 41 x L,12 where L and Dl are the major and minor axes of the ventricle from the apical long-axis view and D2 is the minor axis derived from the four-chamber view. Biplane left atria1 volume cavity was also calculated in the same expanded views according to the formula: 8 X Al X A2 + 3 x 3.14 x L, where Al and A2 = area of the left atrium in fourchamber and long-axis view, and L = distance from the mitral plane to the roof of the atria1 cavity in four-cham-

ber view.13 Minor irregularities and dropouts of the atria1 surface were handled according to the general curvature of the existing atrial outline. Analysis of data. For each patient, mitral and right upper pulmonary vein velocities of three sequential sinus beats were digitized and averaged (Fig. 1) according to the end of the QRS complex on the ECG. The end instead of the beginning of the QRS complex was chosen in order to digitize more easily the negative component of the pulmonary vein flow velocity, the ascending limb of which is synchronous with QRS. Furthermore, because the mitral and the pulmonary vein flow velocities could not be recorded simultaneously, they were analyzed by matching records at approximately the same R-R interval,” which never exceeded 5 % of their mean (range, 0.2% to 4.1% ). Mitral tracings were digitized along the darkest line on the Doppler spectrum by means of a careful linear extrapolation to baseline (when necessary) of the ascending curve of the E wave and the descending curve of the atria1 contraction (A) wave. The velocity signals that occurred during periods of diastasis were also digitized. Doppler tracings of the pulmonary veins were similarly digitized, taking into account also the retrograde velocities at the time of atrial contraction.14 For each point of the Doppler profile, mitral inflow velocity was multiplied by maximal annular area (as measured in early diastole) to yield filling rate, which was then integrated to obtain cumulative filling vo1ume.s Similarly, the pulmonary venous flow was obtained as pulmonary vein velocity x pulmonary vein area, where pulmonary vein area = (mitral valve velocity

Volume 127, Number 4, Part 1 American Heart Journal

A ew

Marino et al.

-

T- . -

160__ 1

Transmitral Pulmonary Left atrial

889

flow venous flow volume

..........

curve

loo-/ / /

60-0

’ 0

/

----

---_

/ / 1I 200

I 400

600

I 600

\

\

\

I

1 1200

1000

soo300 B B T260 -260 --

z

200

--

150

--

Left ventricular volume curve - - Left atrial volume curve

100 .

0 60--

--------\ \ \

/’

o!

I I

0

200

1 I

1 I

1 I

1 I

400

600

600

1000

Time from

I

1200

end of QRS (ms)

Fig. 2. Transmitral flow, derived left ventricular filling volume curve, pulmonary veins’ flow curve, and

left atria1 volume curve in patient LG.

integral X mitral valve area) + pulmonary vein velocity integral. The left atria1 volume curve was then derived as instantaneouspulmonary venous flow minus mitral valve flow (Fig. 2, A). The two-dimensional echocardiographic estimatesof ventricular volume at end systole and atria1 volume at end diastole werethen usedto quantify the ventricular and left atria1 volumes at the beginning of the respective volume curves (Fig. 2, B). Then, to determine how correctly the volume curves predicted filling volumes and maximum ventricular and atria1 volumes, the largest valuesof the ventricular and atria1 curves and the differences between maximum and minimum values were regressed against the corresponding echocardiographic estimates, assessed at previous end diastole and subsequentsystole. Angiographic data. On the samemorning asthe echocardiographic examination, those four patients who had undergonea previous percutaneoustransluminal coronary angioplasty (Table I) underwent an invasive investigation as part of a protocol for the assessmentof the long-term

efficacy of the angioplastic procedure. Informed consent was obtained from all patients before the procedure. Briefly, after left ventriculography and coronary angiography were performed, an 8F Bergman catheter (Arrow International, Inc., Reading, Pa.) waspositioned in the pulmonary artery trunk through the right femoral vein, and 40 ml of nonionic contrast media (iopamidole)wasinjected at a rate of 20 ml/set in mid expiration, avoiding the Valsalva maneuver.7Simultaneousbiplane cineangiograms(anteroposterior and left lateral projections) of the left atrium were obtained at a speedof 50frames/set with a 35 mm cinecameramounted on a 27 cm imageintensifier in the anteroposterior plane and a 17cm imageintensifier in the left lateral plane. During the injection and afterward, a lead II ECG wasalsorecordedon a direct-writing recorder (Micor; SiemensMedical Corp., Iselin, N.J.). Calibration wasperformed by measuringthe catheter diameter on the film. The angiographicsilhouettes of the left atrium weretraced at 20 msec intervals on a film projector (Cipro 35N,

890

Marina et al.

American

\/

-701 0

C “i

2

80

al

200

400

Time

ID 100

RUPV

I

LA -.- LA - - LA LV

from from i from (from

(ms)

400

Time

600

(ms)

RUPV flow-MV LUPV flow-MV LLPV flowMV flow)

200

Time

April 1994 Heart Journal

flow) flow)

600

(ms)

Fig. 3. Zoomed four-chamber view of the left atrium in a single patient (A). The orifices of three pulmonary veins are shown, together with the corresponding pulmonary venous and mitral flow velocities (6). Pulmonary flow velocities are morphologically very comparable among veins, apart from the direction of blood flow in the lower left vein, where it moves away from the transducer. Minor differences in velocity do not translate into major differences in computed flow (C) or in derived left atria1 volume curves (D). LA, Left atrium; LLPV, left lower pulmonary vein; LUPV, left upper pulmonary vein; LV, left ventricle; MV, mitral valve; RUPV, right upper pulmonary vein.

Siemens Medical Corp.) for at least one cardiac cycle, starting with the first well-opacified left atria1 contour after the pulmonary injection. The simultaneous ECG recording excluded the presence of ectopic beats during the levophase of the angiogram. The mitral valvular plane in diastole was identified in each frame according to its systolic position. The biplane left atrial volume was computed as described by Sauter et a1.2 and plotted sequentially to construct a left atria1 volume curve that encompassed little more than one cardiac cycle in each patient and was subsequently smoothed with a 3-point filter. A complete cardiac cycle was then defined on the left atria1 volume curve as composed of those sequential frames exhibiting a progressively increasing and decreasing volume between two minimum volumes of comparable size. A number of time points, ranging from a minimum of 38 to a maximum of 61, identified the atria1 volume curve in the four patients. Then, in order to more easily to compare the angiographic left atria1 volume curve with that derived from the Doppler tracings, the number of time points of each noninvasive

curve (originally ZOO) was equalized to that of the corresponding angiographic curve by means of a linear interpolation. Statistics. All of the results are expressed as means + 1 SD. A linear regression analysis (least squares method) was used to regress the maximum value of the ventricular (end-diastolic) and atrial (end-systolic) volume curves and their respective filling volumes versus the corresponding values assessed by means of two-dimensional echocardiography. Differences between two sequential measurements of the ventricular and atria1 volumes and of the mitral anulus area performed 4 months apart were assessed by paired t tests. Two-sided p values are reported. Statistical analyses were performed on a 16-bit personal computer with NWA STATPAK (version 3.1; Northwest Analytical, Portland, Ore.) statistical software. RESULTS Echocardiographic formed in less than

data. The 20 minutes

recordings were perIn for each patient.

volume 127, Number 4, Part 1 American Heart Journal

Marino et al.

891

Fig. 4. Integral (A) and corresponding calculated area (C) of the right upper pulmonary vein velocity curve

(x axis) plotted against the same parameters obtained in the same subjects from the other Although differences among the three veins for the velocity integral and the calculated the fmsl results in terms of left atrial maximum volume (8) and of atria1 filling volume parable. LA, Left atrium; LLPV, left lower pulmonary vein; LUPV, left upper pulmonary upper pulmonary vein. those four patients in whom more than one vein was sampled, recording of flow velocity was obtainable from all the pulmonary veins except the one in the lower right position, which, being explorable from the parasternal approach only, was not considered in the present study. An example of the three pulmonary veins’ velocities obtained from the apical approach in a single patient is shown in Fig. 3. Flow velocity was morphologically very comparable among veins (Fig. 3, B), apart from the direction of flow in the lower left vein where blood moved away from the transducer. Small differences in the angle between the beam and the flows were probably responsible for minor differences in the absolute mean value of the various velocity curves (Fig. 3, B), although possible, real minor changes in flow velocity, mainly in the upper left vein compared with the other two, could not be completely ruled out. These differences in velocity, however, do not translate into major differences in flow.

two veins (y axis). area are evident, (D) are very comvein; RUPV, right

If we look at the different pulmonary flow curves L tained from each vein (Fig. 3, C) or at the derivea atria1 volume curves (Fig. 3, D), differences are minor, This is confirmed in Fig. 4 where on the left, the integral of the right upper pulmonary vein velocity curve and the corresponding calculated pulmonary vein area is plotted against the same parameters obtained from the other two veins (Fig. 4, A and C), whereas on the right, the derived maximum left atria1 volume and the atria1 filling volume are shown (Fig. 4, B and D). Although differences among the three veins for the velocity integral and the calculated area are evident, the final results in terms of maximum volume and left atria1 filling volume are very similar. This suggests the possibility of an interchangeable use of one of the three veins for the computation of the left atria1 volume curve. For the entire group of subjects participating in the study, the end-diastolic and filling left ventricular

892

Mark

et al,

American

April 1994 Heart Journal

maximum volume r=0.98 p(.OOl filling volume r=0.70 p(.OOl

Overall r=0.98 p( .OOl y=o.97x+23.7 SEE= 16.6 ml

,-A 50

loo

150

260

200

300

350

LV Volume by 2Decho (ml) o

maximum volume r=0.96 p(.OOl filling volume r=0.67 p( .Ol

_A

O/

,

Overall

-

r=0.95 p( .OOl y=O.94x+ 11.8 SEE= 11.9 ml

F=-23

B

“oy

i

l&l

LA Volume

9k

lA0

lie

lie

by 2Decho (ml)

Fig. 5. Plots of estimates of maximum and filling left ventricular (A) and left atria1 volumes (B) for the entire population assessed at two-dimensional echocardiography (x axis) and with the Doppler-derived technique (y axis). There is a highly significant correlation for both maximum and filling volumes between the two techniques at both ventricular and atria1 levels. The identity line is shown in both panels. LA, Left atrium; LV, left ventricle; 2Decho, two-dimensional echocardiography.

volumes (taken as the largest value of the ventricular volume curve and as the difference between maximum and minimum value, respectively) correlated significantly with the corresponding two-dimensional echocardiographic estimates (Fig. 5, A). The overall correlation was 0.98 (p < 0.001, y = 0.97x + 23.7, standard error of the estimate [SEE] = 16.6 ml). The correlation coefficient for the filling volume only was0.70 (p < 0.001, y = 0.68x + 41.2, SEE = 15.7 ml) and for the maximum volume, 0.98 (p < 0.001, y = 0.96x + 27.4, SEE = 16.9 ml). Also, the atria1 maximum (end-systolic) and filling volumes (taken as the difference between maximum and minimum volumes) correlated significantly with the corresponding echocardiographic estimates (Fig. 5, B), with an overall correlation of 0.95 (p < 0.001, y = 0.94x + 11.6, SEE = 11.9 ml). The correlation coefficient for the atria1 filling volumes only was 0.67 (p < 0.01, y = 0.67x + 19.5, SEE = 11.4 ml) and for

the maximum volume, 0.96 (p < 0.001, y = 0.94x + 11.5, SEE = 11 ml). To better determine the accuracy of one method (Doppler echocardiography) compared with the other (two-dimensional echocardiography) for both maximum and filling volumes, plots of the average between the two methods (V volume) on the x axis were compared with their difference on the y axis, expressed proportionally to their mean [(A volume t mean volume) X 1001for both the ventricle (Fig. 6, A) and the atrium (Fig. 6, B) according to the method of Bland and Altman.r5 Doppler echocardiography systematically overestimated two-dimensional echocardiography (mean error: 20.6 % f 18.9% for the ventricle [Fig. 6, A] and 21.0% t 28.8% for the atrium [Fig. 6, II]); the difference between the two methods increased with decreasing volumes. As in the comparison between the two methods, maximum volumes simply stretch the data between

Volume 127, Number 4, Part 1 American Heart Journal

Marino

et al.

89

0 maximum volume 0 filling volume 20 -L

fin V-

mean+OSD

0 0

0"

0

0

mean-&SD -100

I 0

1, 50

I1 150

,1 100

m t:

s

,1 250

Mean LV Volume

B 100 v i? 5

II 200

0 maximum volume 0 filling volume

.o

o-80 --100 0

1 350

(ml)

--

60 80 -40-20 --

-60 --

1 300

mean+BSD

0

0

* 0 l. a

I 30

mean-2SD II 120

,I 90

1I 60

Mean

LA Volume

1 I 150

II 180

(ml)

Fig. 6. Plots of the average between Doppler and two-dimensional echocardiographicestimatesof maximum and filling volumes: on the x axis, against their difference, on the y axis, expressedproportionally to their meanfor both the ventricle (A) and the atrium (B). Doppler systematically overestimatestwo-dimensionalechocardiography(meanerror: 20.6% k 18.9% for the ventricle [A] and 21.0% +- 28.8% for the atrium [B]), the difference between the two methods increasingwith decreasingvolumes. The mean difference 2 2 standard deviations are shown in both panels.LA, Left atrium; LV, left ventricle.

more extreme values but do not change the information obtained from the filling volumes only. We analyzed the relationship between the ratio of the Doppler-derived or the echocardiographic-derived filling volume divided by the echocardiographic volume of the cavity at the beginning of the filling curve (filling volume + systolic volume for the ventricle and filling volume + diastolic volume for the atrium) on the y axis and the echocardiographic volume of the cavity at the beginning of the filling curve on the x axis (Fig. 7). This was done to eliminate the potential bias derived from measuring volumes of cavities with unequal sizes. As expected, there was an inverse exponentional relationship between the initial volume and the ratio of filling volume to initial volume for both the ventricle (Fig. 7, A) and the atrium (Fig. 7, B). The correlation coefficients were higher for the ventricle (Doppler technique: r = -0.98, p < 0.001, y = 2.53Exp[-O.O09x],SEE = 0.14;echographic technique: r = -0.95, p < 0.001, y = 1.83Exp[-0.008x],

SEE = 0.23) (Fig. 7, A) than for the atrium (Doppler: r = -0.92, p < 0.001, y = 3.09Exp[-0.021x], SEE = 0.36; two-dimensional echocardiography: r = -0.92, p < 0.001, y = 2.29Exp[-0.021x], SEE = 0.36 (Fig. 7, B). The dispersion of the data, however, as reflected from the SEES, was comparable between the two techniques at both levels, although the spreading of the data for small atria1 volumes was larger for Doppler than for two-dimensional echocardiography (Fig. 7, B). Comparison between volume measurements.

echo-Doppler

and angiographic

In those four patients who underwent pulmonary cineangiography, there was such a similarity in shape between Doppler-derived and angiographic estimates of the left atria1 volume curve (Fig. 8) that we considered it unethical to enroll more patients in that extended protocol. Angiography, however, systematically overestimated the atria1 volume relative to the noninvasive method. This systematic offset is probably secondary to the volume of

994

Marino

et al.

American

April 1994 Heart Journal

5-

02Decho r=-0.95 p(.OOl y= 1.83Exp(-0.008x)

4--

SEE=0.23 3 --

l Doppler r=-0.98

p(.OOl

2--

o+ 0

50

100

150

200

LV systolic volume

02Decho

I 300

250

(ml)

r=-0.92

p(.OOl

y=2.29Exp(-0.021x) SEE=0.36 l Doppler r--O.92 p(.OOl

@..W G 3

-4

O-

o

30

60

LA diastolic

120

00

volume

I 1so

(ml)

Fig. 7. Regressionof the ratio of the Doppler-derived and of the echocardiographic-derivedfilling volume divided by the echocardiographicvolume of the cavity at the beginning of the filling curve (systolic volume for the ventricle and diastolic volume for the atrium) on the y axis and the initial echocardiographicvolume on the x axis. As expected, there is an inverse exponentional relationship between the initial volume and the ratio of filling volume to initial volume for both the ventricle (A) and the atrium (B). The correlation coefficients are higher for the ventricle (Doppler technique: r = -0.98, p < 0.001, y = 2.5Exp [-0.009x], SEE = 0.14; echocardiographic technique: r = -0.95, p < 0.001, y = 1.8Exp[-0.008x], SEE = 0.23, A) than for the atrium (Doppler: r = -0.92, p < 0.001, y = 3.1Exp[-0.21x], SEE = 0.36; twodimensionalechocardiography:r = -0.92, p < 0.001,y = 2.3Exp[-0.021~1, SEE = 0.36, B). The dispersion of the data, however, as reflected by the SEES,is comparable between the two techniques at both levels, although the spreadingof the data for smallatria1 volumeswaslarger for Doppler than for two-dimensional echocardiography(B). Arrows indicate patients GL, MMA, VIS, TF (Table I). For details seetext. LA, Left atrium; LV, left ventricle; 2Decho, two-dimensional echocardiography.

the atria1 appendage, which is included in the angiographic silhouette, but it is excluded from the initial echocardiographic atria1 volume. Furthermore, the angiographic outer margin of the opacified cavity does not necessarily represent the endocardial surface of the wall because angiography minimizes wall thickness and consequently increases cavity dimensions, as previously reported.16 Reproducibility of echocardiographic measurements. The mean percentage of difference between two measurements (performed 4 months apart by

the same operator in 10 subjects) averaged 5.9 % +- 4.5% for the end-systolic left ventricular volume (p = 0.46), 4.8% + 2.3% for the end-diastolic left atria1 volume (p = 0.38), and 3.1% + 3.3% for the mitral anulus area (p = 0.11). Alternatively, the correlation coefficient between the two measurements was 0.98 for the end-systolic ventricular volume (p < 0.001, SEE = 16.8 ml), 0.99 for the end-diastolic left atria1 volume 0, < 0.001, SEE = 5.1 ml), and 0.71 for the mitral anulus area (p < 0.05, SEE = 0.8 cm2).

Volume 127, Number 4, ParI 1 American Heart Journal

Marino

o-0 A-A

et al.

895

Doppler Angiography

0 ‘0

O0

Q-3

30-

1

11

21

31

41

90

4

60

30

TIME

(# of angiographic

1

11

21

31

41

51

61

frames)

Fig. 8. Individual Doppler-derived and angiographicleft atria1 volume curves in those four patients who

underwent the angiographicstudy. There is a strong similarity between the two estimatesof the atria1 volume curve for eachpatient. Angiography systematically overestimatesthe Doppler-derived curve in all patients. DISCUSSION

Characterization of the atrium as a contracting chamber is difficult because its volume is difficult to measure. Although the atria1 diameter can easily be measured continuously during M-mode echocardiography,17 whether this single measurement adequately expresses the corresponding variation in the atria1 volume is open to question. We have previously shown that the left ventricular diastolic volume curve can be derived from the Doppler-determined mitral flow velocity integral x the mitral cross-sectional area (mitral diastolic flow) and that ventricular volume measurements thus obtained correlate signiflcantly with the angiographic estimates.8 In the present study we have applied the same concept to the pulmonary veins’ velocity, defining the pulmonary venous flow as the product of the pulmonary veins’ integral X the pulmonary veins’ area, where the veins’ area is obtained by dividing the mitral diastolic flow by the pulmonary veins’ velocity integral. The left atria1 volume curve, derived as the net instantaneous difference between the two flows (with the two-dimensional echocardiographic estimate of

the atria1 volume at end-diastole used to quantify the volume of the chamber at the beginning of the curve), compares morphologically with the angiographic curve in those four patients from whom this information was available; furthermore, for the entire population, the maximum values of the Doppler-derived curves and the calculated filling volumes correlate significantly with the corresponding echocardiographic estimates of the atria1 volume. Some degree of inaccuracy, however, in the estimation of the atria1 stroke volume by the Doppler technique at small atria1 cavity sizes must be acknowledged. Such discrepancy might reflect inaccuracies in the estimation of the pulmonary veins’ flow. Alternatively, inadequacy of the echocardiographic technique could also be postulated because the contribution of the atria1 appendage to filling is not accounted for by two-dimensional echocardiography. Such a contribution, which is included in the flow computation by the Doppler technique, might not be so trivial at small atria1 volumes when distensibility in the appendage, reported to be greater than in the body of the cavity,18 could counterbalance the small

April

896

Marino et al.

size of the cavity. Such a consideration would be supported by the fact that most of the difference in the estimation of the atria1 filling volume by the two techniques comes from the youngest patients in our study group (GL, MMA, VIS, TF), those marked with an arrow in Fig. 7, B (A ratio 1.51 ? 0.90 vs 0.16 f 0.37, p < 0.001). Pulmonary venous flow. Detection of pulmonary venous flow is relatively easy. Sampling of the right upper pulmonary vein from an expanded apical four-chamber view including the left atrium allows the recording of the pulmonary venous flow in the majority of patients. g,l”,r4 In sinus rhythm three different phases in the pulmonary venous flow velocity tracings can be identified: (1) systolic, (2) early diastolic, and (3) phase of flow reversal at atria1 contraction.14 Although the degree of systolic and diastolic forward and backward flow in the extraparenchymal pulmonary veins is controlled by the complex relations existing between the veins and the atria1 and ventricular cavities, it seemsreasonable to think that the net difference between pulmonary venous and mitral flow must describe the dynamic behavior of the left atria1 volume. Although it is true that Doppler echocardiography measures velocity and not flow, it has been shown that the integral of flow velocity x the cross-sectional area through which flow is sampled correlates significantly with corresponding flow. lg To make this assumption ture, some requirements must be fullfilled: (1) flow must be laminar in the area interrogated, (2) velocity profile must be flat, and (3) Doppler angle must be less than 20 degrees so that the cosine of the angle can be assumed to be approximately 1.20Although these conditions are probably met in the normal mitral valve and at the pulmonary veins’ orifices,14, 2othe estimation of the area through which blood is flowing may be a major difficulty. 2oIn fact, although a good correlation between thermodilution and Doppler stroke volume estimates with the mitral velocity integral and the mitral annular diameter has been reported, a 15 % interobserver variability in measuring mitral cross-sectional area from the apical approach has been noted.21 The reproducibility of the mitral anulus area, however, averaged 7.7 % +- 6.2 % in our previous studys and 3.1% f 3.3% in the present study, suggesting a reasonable intraobserver reproducibility for the mitral valve area when assessedfrom the anulus size measured in the four-chamber view. Left atrial volume curve. The instantaneous changes of left atria1 volume during one cardiac cycle are shown in Fig. 2, B. As the ventricular systole begins,

American

Heart

1994

Journal

the left atria1 volume progressively increases, reaching a maximum in the vicinity of end systole, after which it decreases rapidly at first and then more slowly during diastasis. At the end of the diastasis period the left atrium begins to contract actively, expelling blood into the left ventricle and thus allowing refilling of the cavity with the next beat. The left atria1 volume curve, however, does not provide a measure of the amount of blood entering the left ventricle from the atrium during diastole. In the phase of passive atria1 emptying and atria1 diastasis, blood also flows from the pulmonary veins to the ventricle (Fig. 2, A). Furthermore, during active atria1 emptying, some blood may flow back to the pulmonary veins. The simultaneous availability of the left atria1 and left ventricular volume curves allows a precise definition of the contribution of the left atrium to the ventricular filling process, highlighting the triple function of the atria1 cavity: reservoir, conduit, and pump.22>23Quantification of these different functions in all the patients in whom the mitral and pulmonary venous flow velocities can be recorded would thus be possible, whereas the transesophageal approach would overcome difficulties in patients with poor echogenicity,24 extending the potential analysis of the left atria1 function to intraoperative studies in which estimates of left atria1 pressure would also be easily available.24, 25 Limitations. Several limitations of this study must be pointed out. The use of the mitral valve area in the computation of flow may be a source of errors. Ormiston et a1,26using multiple calibrated apical views in normal subjects have shown that the mitral valve area changes in shape and size during the cardiac cycle, with a 12% gradual increase in cross-sectional area from early to late diastole. The assumption of a constant diastolic shape for the mitral anulus, which we made in the present study would result in a systematic underestimation of flow and consequently of left ventricular volume at end diastole and of left atria1 volume at end systole. Regression of volume data derived from the echo-Doppler method versus the corresponding two-dimensional echocardiographic data did not show any systematic underestimation, but rather a slight overestimation (Fig. 5, A and B). This finding, together with the consideration that such a change in area represents a relatively small change in diameter because it is related to the square root of area, suggests that the assumption of a constant mitral area through diastole is acceptable for clinical purposes. Changes in shape and section during each cardiac

Volume 127, Number 4, Part 1 American Heart Journal

cycle have also been reported for the pulmonary veins as a result of the filling demands of the left atrium.27 These changes in shape (from an elliptical to a circular configuration in relation to a progressive increase in the left atria1 pressure), providing the major component of the overall compliance of the pulmonary venous system, enable the left ventricular stroke volume to remain relatively unaffected by beat changes in right ventricular stroke output.27 Obviously, cyclic changes in the area of the pulmonary veins or nonuniform cyclic variations among the various veins may affect the calculation of the instantaneous pulmonary venous flow. In practice, however, the indirect way we estimated the pulmonary veins’ area, based on the assumption of a net final balance of flow to and from the left atrium through constant orifice sizes, gives the data a consistency derived from the strong similarity between the angiographic and noninvasive estimates of the left atria1 volume curve in those four patients who underwent hemodynamic investigations (Fig. 8) and also the similarity among the atria1 volume curves obtained in the same patient by sampling different veins (Fig. 3). The presence of relevant mitral insufficiency or any other form of abnormal atria1 or ventricular filling is another source of error for the method we have proposed, whereas the presence of irregular rhythm may prevent the collection of useful data because of the impossibility of synchronizing mitral and pulmonary vein velocities. This becomes a limiting factor for the assessment of left atria1 function in those patients who have atria1 fibrillation or other types of atria1 or ventricular arrhythmias. Finally, limitations inherent in the reference standard used to measure the left atria1 volume as related to angiography have to be acknowledged, including the relative planar nature of the angiographic images and the difficulties in defining boundaries of the opacified cavities. Conclusions. In conclusion, we have shown that the left atria1 volume curve can be obtained as the net instantaneous difference between the pulmonary veins’ flow and the mitral valve flow during the entire cardiac cycle. The result is a noninvasive, reproducible curve very comparable to the angiographic one with maximum value and computed filling volume significantly correlated with the corresponding twodimensional echocardiographic estimates, although a larger spreading of data at the atria1 than at the ventricular level must be acknowledged. The approach described, besides providing a tool for further noninvasive evaluation of the left atria1 function, offers the opportunity for relating the continuous pulmonary

Marino

et al.

997

venous flow to the intermittent filling of the ventricle through the mitral orifice in diastole, underlining the complex role that the left atria1 cavity plays in this process. REFERENCES 1. Braunwald

2. 3. 4.

5.

6.

7.

8.

9. 10.

11.

12.

13.

14. 15. 16.

17.

E, Frahm CJ. Studies on Starling’s law of the heart. IV. Observations on the hemodynamic functions of the left atrium in man. Circulation 1961;24:633-42. Sauter HJ, Dodge HT, Johnston RR, Graham TP. The relationship of left atria1 pressure and volume in patients with heart disease. AM HEART J 1964;67:635-42. Nolan SP, Dixon SH, Fisher RD, Morrow AG. The influence of atria1 contraction and mitral valve mechanics on ventricular filling. AM HEART J 1968:77:784-91. Rahimtlola SH, Ehsani A, Iliad Sinno M, Loeb HS, Rosen KM, Gunnar RM. Left atria1 transport function in myocardial infarction: importance of its booster pump function. Am J Med 1975;59:686-94. Natsume T, Nakajima K, Shibata H, Yokouchi M, Deguchi F, Shimomura K, Ikeda M. Left ventricular and left atria1 function in patients with borderline and established hypertension. Jpn Circ J 1983;47:230-9. Toma Y, Matsuda Y, Moritani K, Ogawa H, Matsuzaki M, Kusukawa R. Left atria1 filling in normal human subjects: relation between left atria1 contraction and left atria1 early filling. Cardiovasc Res 1987;21:255-9. Yamaguchi M, Arakawa M, Tanaka T, Takaya T, Nagano T, Hirakawa S. Study on left atria1 contractile performance: participation of Frank-Starling mechanism. Jpn Circ J 1987;51:1001-9. Marino P, Destro G, Barbieri E, Zardini P. Early left ventricular filling: an approach to its multifactorial nature using a combined hemodynamic-Doppler technique. AM HEART J 1991;122:132-41. Choong CY, Herrman HC, Weyman AE, Fifer MA. Preload dependence of Doppler-derived indexes of left ventricular diastolic function in humans. J Am Co11 Cardiol 1987;10:800-8. Keren G, Sherez J, Megidish R, Levitt B, Laniado S. Pulmonary venous flow pattern-its relationship to cardiac dynamics: a pulsed Doppler echocardiographic study. Circulation 1985;71:1105-12. Keren G, Meisner JS, Sherez J, Yellin EL, Laniado S. Interrelationship of mid-diastolic mitral valve motion, pulmonary venous flow, and transmitral flow. Circulation 1986;74:36-44. Wyatt HL, Heng MK, Meerbaum S, Hestenes JD, Cobo JM, Davidson RM, Corday E. Cross-sectional echocardiography: analysis of mathematic models for quantifying mass of left ventricle in dogs. Circulation 1979;60:1104-13. Ren JF, Kotler M, DePace N, Mintz G, Kimbiris D, Kalman P, Ross J. Two-dimensional echocardiographic determination of left atria1 emptying volume: a noninvasive index in quantifying the degree of nonrheumatic mitral regurgitation. J Am - Co11 eardiol ?983;4:729-36. Smallhorn JF. Freedom RM. Ollev PM. Pulsed Doonler echocardiographic assessment of extraparenchymal pulmonary vein flow. 3 Am Co11 Cardiol 1987;9:573-9. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10. Rackley CE, Hood WP. Measurements of ventricular volume, mass and ejection fraction. In: Grossman W, ed. Cardiac catheterization and angiography. Philadelphia: Lea and Febiger, 1975:176-86. Matsuzaki M, Tamitani M, Toma Y, Ogawa H, Katayama K, Matsuda Y, Kusukawa R. Mechanism of augmented left atria1 I

”

II

April

Miki et al.

18.

19.

20. 21.

22.

American

pump function in myocardial infarction and essential hypertension evaluated by left atria1 pressure-dimension relation. Am J Cardiol 1991;67:1121-6. Davis CA III, Rembert JC, Greenfield JC Jr. Compliance of left atrium with and without left atrium appendage. Am J Physiol 1990;259(suppl):H1006-10. Huntsman LL, Stewart DK, Barnes SR, Franklin SB, Colocousis JS, Hessel EA. Noninvasive Doppler determination of cardiac output in man: clinical validation. Circulation 1983; 67593-601. Gardin JM. Estimation of volume flow by Doppler echocardiography. Echocardiography 1987;4:17-28. Lewis JF, Kuo LC, Nelson JG, Limacher MC, Quinones MA. Pulsed Doppler echocardiographic determination of stroke volume and cardiac output: clinical validation of two new methods using the apical window. Circulation 1984;70:425-31. Matsuda Y, Toma Y, Ogawa H, Matsuzaki M, Katayama K, Fujii T, Yoshino F, Moritani K, Kumada T, Kusukawa R. Importance of left atrial function in patients with myocardial infarction. Circulation 1983;67:566-71.

Heart

1994

Journal

23. Nikolic SD, Biasucci LM, Tanoue T, Liao K, Solomon S, Jondeau G, Nanna M, Frater RWM, Yellin EL. Separation of the left atria1 reservoir, conduit and pump functions by controlled atrial filling [Abstract]. Circulation 1990;82(suppl 111):606. 24. Nishimura RA, Abel MD, Hatle LK, Tajik AJ. Relation of pulmonary vein to mitral flow velocities by transesophageal Doppler echocardiography: effect of different loading conditions. Circulation 1990;81:1488-97. 25. Kuecherer HF, Muhiudeen IA, Kusumoto FM, Lee E, Mouliner LE, Cahalan MK, Schiller NB. Estimation of mean left atria1 pressure from transesophageal pulsed Doppler echocardiography of pulmonary venous flow. Circulation 1990;82:1127-39. 26. Ormiston JA, Shah PM, Tei C, Wong M. Size and motion of the mitral valve anulus in man: a two-dimensional echocardiographic method and findings in normal subjects. Circulation 1981;64:113-20. 27. Rajagopalan B, Bertram CD, Stallard T, Lee GJ. Blood flow in pulmonary veins. III. Simultaneous measurements of their dimensions, intravascular pressure and flow. Cardiovasc Res 1979;13:684-92.

Echocardiographic findings in 104 professional cyclists with follow-up

study

To assess the effect of long-term athletic training on the heart, 104 professional cyclists and 40 sedentary controls (69 younger cyclists and 26 controls aged 20 to 39 and 35 older cyclists and 14 controls aged 40 to 60) were examined by using M-mode and pulsed Doppler echocardiography. Cyclists had larger and more hypertrophied left ventricle than did controls (p < 0.001) and had normal percentages of fractional shortening (%FS). The ratio of left ventricular late-to-early diastolic peak filling velocity (A/R) of younger cyclists was normal, but the AIR of older cyclists was larger than that of controls (p < 0.001). Of the 104 cyclists, 95 continued cycling and were reexamined 2 years later; 9 of 40 older cyclists retired and were reexamined 20 * 6 months after retirement. During the follow-up period for the active cyclists, left ventricular dilatation, hypertrophy, and %FS of both younger and older cyclists and the AIR of younger cyclists did not change. However, the AIR of older cyclists increased (p < 0.01). For the nine retired cyclists, left ventricular dimension decreased @ < O.OOl), left ventricular wall thickness and %FS did not change, and AIR increased (p < 0.05) after retirement. We concluded that (1) cyclists had large and hypertrophied left ventricles with normal systolic function, and (2) some cyclists with long-term athletic training may have partly irreversible left ventricular hypertrophy with impaired left ventricular diastolic filling. (AM HEART J 1994;127:696-905.)

Takahiko Miki, MD, Yoshiyuki Yokota, MD, Toshihiko Seo, MD, and Mitsuhiro Yokoyama, MD Kobe, Japan

From The First Department of Internal Medicine, Kobe University School of Medicine. Received for publication June 11, 1992; accepted Aug. 20, 1993. Reprint requests: Yoshiyuki Yokota, MD, The First Department of Internal Medicine, Kobe University School of Medicine, 7-5-l Kusunoki-cho, Chuo-ku, Kobe 656, Japan. Copyright 3 1994 by Mosby-Year Book, Inc. 000%6703/94/$3.00

898

+ 0

4/l/52244

In the last few decades several echocardiographic studies have been performed to assessthe cardiac functions and anatomy of a variety of athletes.l-ii According to these reports, left ventricular hypertrophy induced by intensive training was not accompanied by impaired left ventricular systolic and diastolic function. However, these were studies of the