Download Unreliability of M-Mode Left Ventricular Dimensions for Calculating

Unreliability of M-Mode Left Ventricular
Dimensions for Calculating Stroke Volume
and Cardiac Output in Patients Without
Heart Disease*
Susan Rasmussen, R.N., M.S.N.;? Betty C . Corya, M.D.;$
JohnF . Phillips, M.D., F.C.C.P.;$ and Mary10 Black, B.A.$
Single-dimension left ventricular echocardiographic
measurements are currently being used in investigational studies as the basis for evaluating cardiac output
parameters in normal subjects, even though vaIidity of
the metbod for normal subjects has not been established. We prospectively compared stroke volume derived from M mode left ventricular dimensions (LVID)
to Fick stroke volume in 20 patients with no objective evidence of cardiac disease. Based on simultaneous
studies, stroke volume by Fick ranged from 39 to 121
c
ontroversy exists as to the accuracy and reliability of using single-dimension echocardiographic measurements for quantifying or estimating
ventricular
Despite multiple potential
sources of error in using a single dimension to estimate v o l ~ m e early
, ~ studies showed good correlation between stroke volume derived from M mode
echocardiographic left ventricular dimensions
(LVID) and stroke volume derived from angiographic610 or Fick methods.l17l2 The majority of
these early study groups were comprised of patients
with a variety of cardiac abnormalities, and statistical analyses were not applied to subjects grouped by
disease entities. Significant error in the LVID method for individual patient groups has since been
reported and errors partially accounted for by the
presence of left ventricular (LV) dyssynergy and
LV dilatation.
Single-dimension echocardiographic measure@Fromthe Krannert Institute of Cardiology- the Department of Medicine, Indiana University ~ c h o o of
i Medicine;
and the Veterans Administration Medical Center, Indianapolis.
Supported in part b the Herman C. Kramert Fund; by
grants HL 06308 an$ HL 07812 from the National Heart,
Lun and Blood Institute, National Institutes of Health;
the American Heart Association, Indiana Affiliate.
and
Inc.
?Assistant Professor of Cardiovascular Research.
SAssociate Professor of Medicine.
§Echocardiographic Research Technologist.
Reprint requests: Dr. Co a, I&
University School of
Medicine, 1100 West M i L n , Indianapolis 46202
ty
ml and cardiac output ranged from 3.9 to 10.4 L / h .
Comparing the LViD cubed method with Fick, the
correlation coefficient was r = .47 for stroke v o h e
and r = -36 for cardiac obtput LVID abohrte error
in cardiac output ranged -2.11 to +3.21 L/min. Use
of other published formulas for calculating stroke volume from LVID did not improve accuracy. These data
indicate tbat stroke volume and cardiac output cannot
be accurately me-asurtd or reliably estimated from M
mode left ventricular internal dimensions.
ments are currently being used to quantitate the
effects of pharmaceutical agents,l3 isometric-isotonic
(athletic) exercise,14 respiratory changes,'v5 aortic
root motion,15 and the aging process16 on cardiac
output parameters in normal subjects, even though
validity of the method has not been reported in
adults with no objective evidence of cardiac disease. In addition, single-dimension volume measurements have not been validated for other noncardiac
patient groups, yet data based on the LVID method
continue to be reported for patients with hypertension l7 and muscular dystrophy.18
The purpose of this study was to prospectively
compare LVID stroke volume to Fick stroke volume in patients with no objective evidence of
cardiac disease.
Simultaneous Fick cardiac output studies and hi mode
left ventricular echocardiographic studies were performed
on 20 patients who had normal angiographic and hemodynamic findings at cardiac catheterization. The study
group was composed of six women and 14 men, with ages
ranging from 26 to 58 years (mean age 42 years, median
age 39 years).
Echocardiograms
During simultaneous four-minute Fick cardiac output
studies, all patients were examined with a commercially
available echocardiograph having a repetition rate of 1,000
614 RASMUSSEN El A 1
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21288/ on 05/05/2017
CHEST, 81: 5, MAY, 1982
FIGURE1. Normal M mode echogram. Measurements for end-diastolic ( d ) and end-systolic
( s ) left ventricular internal dimensions (LVID).IVS = interventricular septum; PW =
posterior wall.
pulses per second and using a 2.25 MHz transducer with
a 12-mm crystal collimated to 7.5 cm. All echocardiographic
recordings were taken from the 4th or 5th intercostal space,
and echograms were recorded on either a portable or multichannel strip chart recorder, along with a simultaneous
ECG monitoring lead 2.
The left ventricular internal dimension was measured as
the vertical distance between the left side of the interventricular septum and the endocardial surface of the
posterior wall ( Fig 1 ). Enddiastole was taken at the peak
of the R wave on the ECC at or just below the level of
the posterior leaflet of the mitral valve. This echocardiographic location has been found to best reflect left septal
and posterior left ventricular endocardial echoes. The peak
of the R wave, which has been shown to correspond to the
mechanical onset of left ventricular contraction in normal
patients,ls was used to detennim the end of diastole.
Measurements were also taken at the onset of the QRS
and at the point in time when the maximum left ventricular
internal dimension was recorded. End-systole was defined as
the point in time when the shortest distance between the left
septal and posterior endocardial echoes was reached. Echograms were recorded at paper speeds of 25 and 50 mm/sec,
and 0.2-second time lines were recorded on all tracings. Diastolic and systolic measurements of the LV internal dimension were made throughout a respiratory cycle, and measurements were averaged to derive an average dimension
for calculatinn
- LVID stroke volume. The formulas and
regression equations used for calculating stroke volume
from the LVID are given in Table 1.
All echocardiographic measurements and calculations
were made by at least two observers (usually three ) without prior know!edge of Fick cardiac output results. Interobserver variability in measuring echocardiographic LV internal dimensions ranged from 0 to 2 1 mm.
the pulmonary artery and arterial samples were obtained
from the thoracic aorta via the right femoral artery. Venous
and arterial blood were simultaneously sampled over one
minute. beginning at 2): minutes and ending at 3%minutes
into the four-minute period. Expired air was collected in a
Douglas bag, and tbe quantity of air inspired was measured
using a Tissot spirometer. Simultaneous recordings of lead
2 ECGs and left ventricular echograms were taken over
four minutes while expired air was collected.
Stattstical Analysis
Echocardiographic end-diastolic measurements were compared using nonparametric analysis of variance. Stroke MIume and cardiac output data were compared using the
Mann-Whitney U test, Kolmogorov-Smirnov test, and linear
regression analyses.
There were no statistically significant differences
between end-diastolic dimension measurements
taken at the peak of the R wave compared with
measurements taken either at the onset of QRS or
a t the point in time of diastolic maximal size.
Measurements taken at the QRS onset were the
same or 1 to 2 mm less than those measurements
Table l-Formulu and Regression Equations Used for
Deriving Echocardiographic Stroke V d u m e From Left
Ventricular Internal Dimrnrioru
16.B.11
I ~ V I D-~I,VII),=
=
Cardiac Outputs
Cardiac output determinations were obtained by the
Fick technique prior to angiographic studies. Patients were
allowed to "practice" hreathing with the mouthpiece in
place and nose clamped for 30 to 60 seconds before collection of data began. Venous samples were obtained from
CHEST, 81: 5, MAY, 1982
LVID = E(*hoc*cirdiograptiicleft \-entric-ular internnl tlimension; d =md-dia*tolic; s ==rntl-systolic. *From Teichholz LE,
et al. Am J Cnrdiol 1976;37:7-11.
UNREUABILITY OF ECHOCARDIOGRAPHY 615
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21288/ on 05/05/2017
Table 2--Raw EcAocetiiogmpirk m d
Hemdynamic
Dana for 20 Normal Subject.
Echocardiographic
-
L I I D d , cm
Patient No.
1
LVTD& cm
(KO. of Complexes Measured)
Fick Cardiac Output,
L/min
Heart Rate. *
beats/min
Body Surface Area.
cm/m'
*Average measurement throughout respiratory cycles.
taken at the R wave. Measurements taken at the
point in time of maximal size were the same or 1
to 3 mm greater than those measurements taken
at the R wave. Measurements of left ventricular
internal dimensions taken at the peak of the R
wave were used for all other statistical analyses
involving end-diastolic measurements.
Table 2 lists raw data pertaining to the 20 cardiac output and stroke volume determinations both
by Fick and singledimension echocardiographic
left ventricular (LV) techniques. Figures 2 and 3
are LV echograms recorded during Fick cardiac
output determinations in two of the 20 subjects in
this study. All 20 patients were in sinus rhythm,
and heart rates ranged from 55 to 114 beats per
minute (mean, 74 beats per minute, median 72,
SD = 14) . Patient body surface area ( BSA ) ranged
1.42 to 2.24 cm/m2. LV dilatation is defined in our
laboratory as diastolic LVID (LVIDd) index > 3.2
cm/ m2 BSA. M-mode end-diastolic LVID ranged
from 3.6 to 5.8 cm; LVIDd index ranged from 2.41
to 3.09 cm/m2 BSA.
Stroke volume by Fick ranged from 39 to El
ml (mean, 82 ml, median 79 ml) and cardiac output ranged from 3.9 to 10.4 L/min (mean, 5.92;
median 5.74 ). Corrected for BSA, cardiac indices
ranged from 2.5 to 4.64 L/min/m2. Comparing the
LVID cubed method with Fick, the correlation
coefficient ( r ) for stroke volume was r = .47 (Fig
4) with an SEE of 23 ml and SD of 25. Correlation
for cardiac output was r = .36 (Fig 5; SEE -+ 1.59,
SD 1.65). LVID absolute error in stroke volume
ranged from -30 to +43 ml with a percent error
ranging from -45 to +34 percent. LVID error
in cardiac output ranged from -2.11 to +3.21
L/rnin. There were no statistically significant differences between the LVID cubed method and four
other published regression equations ( Table 1) or
between regression equations. Use of those other
published formulas for calculating SV from LVID
did not improve accuracy.
Results of this study should not detract from
the important use of LVIDd in assessing left venCHEST, 81: 5, MAY, 1982
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21288/ on 05/05/2017
FIGURE2. Left ventricular (LV) echogram recorded during simultaneous Fick cardiac output (CO) study. m = stroke volume.
tricular size and function in the clinical situation.
Normal values, corrected for body surface area,
are firmly e~tablished,~O-*l
and LVIDd is a reproducible mea~urement.~~
An increased LVIDd index
( > 3.2 cm/m2 BSA) is a very specific indicator
of left ventricular dilatation, and serial measurements are useful in patient management Pietro and
associatesP have reported that a change in LVIDd
of 5 percent or greater is not due to problems of
reproducibility and provides a basis for determin-
ing statistically significant serial changes in clinical
situations. Our study does contraindicate the use
of left ventricular internal dimensions for calculating stroke volumes.
There are numerous reasons for the poor correlation we found between LVID stroke volumes and
Fick stroke volumes. Validity of LVID stroke volume formulas depends on the following assump
tions: (1)the left ventricle approximates an ellipse;
(2) the left ventricle contracts svmmetrically along
FIGURE3. Left ventricular (LV) echogram recorded during simultaneous Fkk cardiac output (CO) study. Complexes 3. 4, 7, and 8 demonstrate obliteration of endocardia] surfaces
due to effect of inspiration. SV = stroke volume.
CHEST, 81: 5, MAY, 1982
UNREUABlLlTY OF ECHOCARDIOGRAPHY 617
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21288/ on 05/05/2017
140
-
120
-
clear from 2D that left ventricular wall excursion
or thickening is unequal in different regions of the
normal ventricle.2sThere is also a greater percentage
of error for the smaller cavity in systole compared
with diastole. For these reasons both systolic volumes and stroke volumes derived from LVID are
less reliable than are diastolic v o l ~ r n e s . ~
Error in stroke volume may be partially due to
movement of the ventricle relative to the echo
beam from end-diastole to end-systole.24 Spatial
orientation also varies with respiration. Brenner and
Waugh4 recently studied 30 normal subjects with
single-dimension echocardiography and reported
statistically significant variability in derived volumes associated with "quiet" phasic respiration
alone.
We are not aware of other prospective studies
comparing LVID to Fick stroke volumes in normal
subjects. Wyatt et ala have published interesting
experimental data comparing mathematic models
for quantifying left ventricular volume from the
two-dimensional echograms of formalin-fixed dog
ventricles. Under relatively controlled conditions,
with a nonbeating heart surrounded by mineral oil
and with the equivalent of two-dimensional guidance in obtaining left ventricular dimensions, the
best r value correlation for cubed left ventricular
dimensions and measured volumc was 0.837 with
a 40.4 SEE and a mean percent error of 49.9. These
data show that diastolic internal dimensions do
have a strong positive correlation with diastolic
volume. However, the large SEE and large mean
percent error under these controlled conditions indicate volume quantification, based on internal dimensions, would not be reliable. ii7yattet alZ5found
excellent correlations with fluid volume using 2D
short axis area and long axis length measurements,
and it is reasonable to expect that the left ventricular diastolic volumes from clinical 2D echograms
will prove clinically reliable.
Correlation of the LVID cardiac output to Fick
was lower than for stroke volume in our study
because of statistical methods. Fick stroke volume
was derived by dividing cardiac output by a heart
rate factor. Any time a number is divided, variance
from a line of identity is reduced, which results in
a higher r value.
This study indicates the necessity of developing
better echocardiographic methods for quantifying
stroke volume and cardiac output in normal subjects.
loo FICK SV
=-
lccl
a
a
60-
40
-
20
-
a
>
1
1
m
1
20
1
1
40
1
1
1
1
1
m
80
60
1
1
1
120
140
r d s v lccl
FIGURE
4. Fick stroke volume (SV) compared to SV using
cubed left ventricular internal dimensions (LVID) for 20
normal subjects.
the major axis with little shortening in the major
dimension; (3) the LVID represents the lateral
(mid) minor axis LV dimension at its maximal
circumference; and ( 4 ) the major axis is twice the
minor axis dimension. Reliability of LVID as a
true minor axis dimension also depends on correct
identification of left septa1 and posterior endocardial echoes and on proper positioning of the transducer on the chest wall.
The left ventricle does resemble an ellipse during diastole, but the systolic left ventricular shape
is more variable and does not look ellipsoid.
This has become more evident with the opportunity
of viewing large numbers of normal ventricles with
two-dimensional echocardiography ( 2D ) . It is also
-
.
1
0 5-
9
ra.38
SEE:169
mi 1.85
P = .121*9)
+
8 5
FEar m
IL/mml
.
,s
. . .:... ..
. .
6 9
%
5 5-
4.5-
3 5-
•
.
REFEREE~CES
I
15
I
I
43
I
I
LI
I
5
.
0
~
~
75
I
B
I
I
I
BI
I
ms
I
I
L V I ~co /L/mm)
F'XCURE5. Fick cardiac output (CO) compared to CO using
left ventricular internal dimensions ( LVID ) for 20 normal
subjects.
I
I
~
1 Linhart JW, Mintz GS, Segal BL, Kawai N, Kotler X I S .
Left ventricular volume measurement by echocardiography: fact or fiction? Am J Cardiol 1975; 36: 114-18
2 Schiller NB, Acquatella H, Ports TA, et al. Left ven-
618 RASMUSSEN E l A 1
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21288/ on 05/05/2017
CHEST, 81: 5, MAY, 1982
tricular volume from paired biplane two-dimensional
echocardiography. Circulation 1979; 60:547-55
3 Popp RL, Filly K, Brown OR, Harrison DC. Effect of
transducer placement on echocardiographic measurement of left ventricular dimensions. Am J Cardiol 1975;
35:537-40
4 Brenner JI, Waugh RA. Effect of phasic respiration on
left ventricular dimension and performance in a normal
population. Circulation 1978; 57: 122-27
5 Belenkie I. Beat-to-beat variability of echocardiographic
measurements of left ventricular end-diastolic diameter
and performance. JCU 1979; 7 :263-68
6 Pombo JF, Troy BL, Russell RO Jr. Left ventricular volumes and ejection fraction by echocardiography. Circulation 1971; 43 :480-90
7 Fortuin NJ, Hood \VP Jr, Sherman ME, Craige E. Determination of left ventricular volumes by ultrasound. Circulation 1971; 44 :575-84
8 Feigenbaum H, Popp RL, Wolfe SB, et al. Ultrasound
measurements of the left ventricle. Arch Intern Med
1972; 129:461-67
9 Murray JA, Johnston W, Reid JM. Echocardiographic
determination of left ventricular dimensions, volumes
and performance. Am J Cardiol 1972; 30:25257
10 ten Cate FJ, Kloster FE, vanDorp WG, Meester GT,
Roelandt J. Dimensions and volumes of left atrium and
ventricle determined by single beam echocardiography.
Br Heart J 1974; 36:737-46
11 Feigenbaum H, Nasser WK, Ross E, Elliott WC. Left
ventricular stroke volume measured by ultrasound. Circulation 1969; 40 ( suppl3) :79
12 Popp RL, Harrison DC. Ultrasonic cardiac echography
for determining stroke volume and valvular regurgitation. Circulation 1970; 41:493502
13 Martin MA, Fieller KRJ. Echocardiography in cardiovascular drug assessment. Br Heart J 1979; 41:536-43
14 Cohen JL, Gupta PK, Lichstein E, Chadda KD. The
heart of a dancer: noninvasive cardiac evaluation of
professional ballet dancers. Am J Cardiol 1980; 45:95965
CHEST, 81: 5, MAY, 1982
15 Lewis BS, Hasin Y, Pasternak R, Gotsman MS. Echocardiographic aortic root motion in ventricular volume
overload and the effect of mitral incompetence. Eur J
Cardiol 1979; 10:375-84
16 Gardin JM Henry WL, Savage DD, Ware JH, Bum C,
Borer JS. Echocardiographic measurements in normal
subjects: evaluation of an adult population without
clinically apparent heart disease. JCU 1979; 7:439-47
17 Schlant RC, Felner JM, Heyms6eld SB, et al. Echocardiographic studies of left ventricular anatomy and function in essential hypertension. Cardiovasc Med 1977;
hIay:477-91
18 Farah MG, Evans EB, Vignos PJ Jr. Echocardiographic
evaluation of left ventricular function in Duchenne's
muscular dystrophy. Am J Med 1980; 69:24854
19 Braunwald E, Fishman AP, Cournand A. Time relationship events in the cardiac chambers, pulmonary artery and aorta in man. Circ Res 1956; 4:100-07
20 Feigenbaum H. Echocardiography. 3rd ed. Philadelphia: Lea & Febiger, 1981
21 Corya BC, Rasmussen S. Echocardiography in evaluating patient with coronary artery disease. In: Clinical
handbook of ultrasound. New York: John Wiley and
Sons, 1978; 503-13
22 Pietro DA, Voelkel AG, Ray BJ, Parisi AV. Reproducibility of echocardiography: a study evaluating the
variability of serial echocardiographic measurements.
Chest 1981; 79: 29-32
23 Pandian N, Skorton D. Collins S, Burke E, Falsetti H,
Kerber R. Heterogeneity of segmental left ventricular
wall thickening and excursion in two-dimensional echocardiograms of normal humans. Am J Cardiol 1981;
47 :452
24 Bhatt DR, Isabel-Jones JB, Villoria GJ, et al. Accuracy
of echocardiography in assessing left ventricular dimensions and volume. Circulation 1978; 57:699-707
25 Wyatt HL,Heng MK, Meerbaum S, et al. Cross-sectional echocardiography: 11. Analysis of mathematic models
for quantifying volume of the formalin-fixed left ventricle. Circulation 1980; 61: 1119-25
UNREUABILITY OF ECHOCARDIOGRAPHY 619
Downloaded From: http://publications.chestnet.org/pdfaccess.ashx?url=/data/journals/chest/21288/ on 05/05/2017