Download Left Ventricular Mass

Left Ventricular Mass
Reliability of M-Mode and 2-Dimensional Echocardiographic Formulas
Saul G. Myerson, Hugh E. Montgomery, Michael J. World, Dudley J. Pennell
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
Abstract—The study of left ventricular (LV) hypertrophy is hindered by problems with LV mass measurement by
echocardiography. Both the M-mode and 2D area-length formulas for calculating LV mass assume a fixed geometric
shape, which may be a source of error. We examined this hypothesis by using cardiovascular magnetic resonance images
to eliminate the confounding effects of acoustic access and image quality. LV mass was measured directly in 212 healthy
subjects by means of a standard 3D cardiovascular magnetic resonance technique. LV mass was also calculated by using
the cube-function and area-length formulas with measurements from the magnetic resonance images. A comparison of
serial measurements was made by examining the changes in LV mass by all 3 techniques in those completing an exercise
program (n⫽140). The cube-function technique showed a consistent underestimation of LV mass of 14.3 g, and there
were wide 95% limits of agreement (⫾57.6 g and ⫾46.3 g for cube-function and area-length techniques, respectively)
when compared with 3D measurement. There were similarly wide limits of agreement for the change in mass (⫾55.2
g and ⫾44.8 g for cube-function and area-length, respectively). The assumption of geometric shape in the cube-function
and area-length formulas resulted in significant variation in LV mass estimates from direct measurement by using a 3D
technique. The technique cannot be recommended either at a single time point or for serial studies in small populations;
3D imaging techniques, such as cardiovascular magnetic resonance, are preferable. (Hypertension. 2002;40:●●●●●●.)
Key Words: myocardium 䡲 hypertrophy 䡲 magnetic resonance imaging 䡲 echocardiography
L
mass to autopsy findings for M-mode9,10 and 2D echocardiography.12,19 –21 The cubing involved in the formulas means
that small (⬍1 mm) differences in measurement can have
large effects on the calculated mass. Additionally, the assumption of shape may not be appropriate, even for normal
hearts, in which a substantial natural variation in shape may
exist, though the specific influence of this assumed geometry
on accuracy has not previously been examined. We examined
the M-mode and 2D formulas under the best possible circumstances of good image quality and normal hearts by applying
them to cardiovascular magnetic resonance (CMR) images
and comparing the results with direct measurement of the LV
mass by using the standard 3D CMR technique.
eft ventricular (LV) hypertrophy is an independent cardiovascular risk factor associated with significant excess
and morbidity and mortality rates.1–3 There is now evidence
for the effectiveness of antihypertensive agents, particularly
ACE inhibitors, in reducing LV mass,4 – 6 and this reduction in
LV mass appears to carry a favorable prognosis.7,8 However,
studying the prognostic implications of LV mass reduction is
hindered by the poor accuracy and reproducibility of LV
mass measurement by M-mode and 2D echocardiography.
The high 95% confidence limits for accuracy (⫾57 to 190 g
for M-mode,9 –12 ⫾61 to 80 g for 2D techniques11–13) and
interstudy reproducibility (⫾45 to 78 g for M-mode,10,14 –17
poorly assessed for 2D) result in the need for large numbers
of subjects in research studies.
The principal sources of error are considered to be image
quality, beam positioning, and the assumption of a uniform
geometric shape of the left ventricle, with LV mass calculated
from measurements made at 1 or 2 positions. Both M-mode
and 2D echocardiography assume a prolate ellipsoid shape,
with a ratio of long- to short-axis lengths of 2:1, which
provided the best simplified geometric model for LV mass
estimation.18 Formulas were developed for the calculation of
LV mass based on the regression equations of the calculated
Methods
Subjects
The study group comprised 212 male British Army recruits, free
from cardiovascular disease, who had CMR scans performed at the
start of military training, as part of a separate study.22 After a
10-week physical training program as part of standard training, 140
completed the course and had follow-up scans. In this way, both the
measurement of LV mass at a single time point and the detection of
changes over time could be compared.
Received April 22, 2002; first decision May 20, 2002; revision accepted August 23, 2002.
From the Cardiovascular Magnetic Resonance Unit, Royal Brompton Hospital (S.G.M., D.J.P.), London; Centre for Cardiovascular Genetics,
University College London (H.E.M.), London; and Royal Defense Medical College (M.J.W.), Gosport, Hampshire, UK.
Correspondence to Prof Dudley Pennell, Cardiovascular Magnetic Resonance Unit, Royal Brompton Hospital, Sydney Street, London SW3 6NP, UK.
E-mail [email protected]
© 2002 American Heart Association, Inc.
Hypertension is available at http://www.hypertensionaha.org
DOI: 10.1161/01.HYP.0000036401.99908.DB
1
2
Hypertension
November 2002
Cardiovascular Magnetic Resonance (3D Method)
Studies were performed with a custom-built mobile CMR scanner
situated at the military base for the study (0.5 Tesla; imaging
software: Surrey Medical Imaging Systems). All imaging was
ECG-gated and acquired in a standard manner as previously described.23 Briefly, a stack of contiguous short-axis cine images
covering the whole LV were obtained, and the myocardial volume
from each slice was summed to obtain the total volume, by using
Simpson’s rule. Multiplying this volume by the specific density of
myocardial tissue (1.05 g/cm3) gives the mass. This technique is well
validated, with high accuracy and reproducibility.16,17,24
M-Mode Measurements
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
From the CMR image set, the short-axis image slice just basal to the
papillary muscles was used, reproducing measurements obtained at
the level of the chordae tendineae, according to the American
Society of Echocardiography (ASE) guidelines for M-mode measurement.25 Measurements were made of the septal wall thickness
(SWT) and posterior wall thickness (PWT) and LV internal diameter
(LVID) from this slice. These were used in the cube-function
formula for ASE guidelines10 to calculate LV mass (the ASE-cube
formula):
LVmass( g)⫽0.80[1.04{(PWT⫹LVID⫹SWT)3⫺LVID3}]⫹0.6
Two-Dimensional Measurements
The area-length method12 was selected for comparison because it is
a simpler formula with easily obtainable measurements; it is commonly used and has been shown to be closer to the true LV mass than
other 2D methods.20 Measurements were made of the epicardial
(AEpi) and endocardial (AEndo) areas from the same short-axis slice as
the M-mode calculations, just basal to the papillary muscles. This is
close as possible to the high papillary muscle level suggested, given
the limitations of the CMR slice thickness (10 mm). The ventricular
long axes (epicardial, LEpi; and endocardial, LEndo) were measured
from the horizontal long-axis slice (equivalent to the echocardiographic 4-chamber view), maximizing accuracy of the apex position.
LV mass was calculated by using the following area-length
formula:12
LVmass( g)⫽1.05⫻{5/6( AEpi⫻LEpi)⫺( AEndo⫻LEndo)}
Statistical Analysis
The difference in LV mass measurement between methods was
calculated for each subject at baseline and for the change in LV
mass. Bland-Altman plots were performed to determine the 95%
limits of agreement between methods. Group means were compared
with an unpaired Student t test and the within-group change in means
by paired t test. Values are shown as mean⫾1 SD. Percentage
changes are shown using the standard 3D LV mass measured by
CMR as the reference. Although scatterplots are shown, correlation
coefficients were not calculated because these are a measure of the
linearity of the relation between 2 variables and do not indicate the
strength of agreement between techniques, for which Bland-Altman
plots are more appropriate.26
Results
Mean LV mass measurements at baseline were 168.8⫾35.4 g,
180.5⫾32.8 g, and 183.1⫾26.4 g for the M-mode, 2D, and
3D methods, respectively, though there was considerable
scatter when the methods were compared (Figure 1). The
mean difference between M-mode and 3D techniques was
⫺14.3 g (7.8%, P⬍0.0001), with 95% limits of agreement of
⫾57.6g (31.5%). The mean difference between the 2D
area-length formula and 3D techniques was ⫺2.6 g (1.4%,
P⬍0.0001), with 95% limits of agreement of ⫾46.3 g
(25.3%). Bland-Altman plots for these differences are shown
in Figure 2.
Figure 1. Scatterplots of LV mass by 3D-CMR technique versus
ASE-cube LV mass (a) and 2D area-length LV mass (b). Lines of
identity are shown.
LV mass increased with training by a mean of ⫹8.3⫾14.0
g (4.5%; P⬍0.0001) by 3D compared with ⫹7.8⫾28.2 g
(P⬍0.001) and ⫹8.9⫾22.5 g (P⬍0.0001), with the use of the
M-mode and 2D echocardiographic methods, respectively.
Although the mean change was similar for all groups, there
was considerable scatter in the results when comparing the
methods (Figure 3) and the Bland-Altman plots showed wide
95% limits of agreement (⫾55.2 g and ⫾44.8 g for M-mode
and 2D echocardiography, respectively; Figure 4).
Discussion
Despite the excellent image quality used here, calculations of
LV mass with the use of the ASE-cube (M-mode) and 2D
formulas showed considerable variation from the reference
standard of direct 3D imaging of the whole ventricle. The
ASE-cube method showed a consistent underestimation of
LV mass by ⬇14 g, though the mean values of the 2D method
were little different from 3D measurement. In addition, the
scatter of values was high for both formulas, indicating high
absolute differences between these techniques and 3D measurement (95% confidence intervals ⫾46.3 g for 2D; ⫾57.6
g for ASE-cube), which represent significant clinical differences. Group studies with M-mode and 2D methods are still
feasible, and our understanding of the dangers of LV hypertrophy come from several of these,1,2,7 but the increased
variability means that much larger sample sizes are needed.
Myerson et al
Reliability of Echocardiographic LV Mass
3
Figure 2. Bland-Altman plots showing 95% limits of agreement between 3D-CMR and ASEcube LV mass calculation (a) and 2D arealength LV mass calculation (b).
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
The mean values for the change in LV mass with training
were not significantly different between techniques, suggesting no systematic bias in serial measurements with the use of
the formulas. It also demonstrates that with large enough
group sizes, similar mean changes can be observed with all 3
techniques. However, the wide confidence limits for both
formula methods result in a reduced ability to detect changes
in smaller groups and the loss of the ability to detect
individual changes in LV mass below this level (45 to 55 g).
With direct measurements using 3D techniques, the same
change in LV mass can be detected with much smaller sample
sizes, or, alternatively, using the same sample size, smaller
degrees of change can be identified.23 This has a substantial
impact on the costs of performing studies.
Validity of Reference Standard
CMR LV mass may deviate slightly from the true LV mass,
and it lacks the gold-standard comparison of human autopsy
LV mass with in vivo data, but it has been shown to be highly
accurate and reproducible (95% confidence intervals ⫾15 to
19 g)16,17,24 and no better in vivo method exists. The purpose
of this study was to allow a direct comparison of the
differences between directly measured 3D quantification and
calculated values with an assumed geometric shape, using the
same data set to minimize image quality problems, and
the CMR data perform this function well. Furthermore, the
confidence intervals obtained here for the 2D and M-mode
methods are similar to previously published values for accuracy9 –13 and reproducibility14 of LV mass with the use of the
same formulas, suggesting that these differences are mostly
due to errors in the calculations rather than an incorrect
reference standard of 3D CMR.
Possible Reasons for Errors in LV
Mass Calculation
The use of a single geometric model to estimate LV mass
appears inappropriate. The original studies determining the
equations9,10,12 involved relatively few subjects (34, 52, and
21 subjects), with a wide age range (23 to 82 years) and
marked variation in LV mass (77 to 625 g). Although the
wide range studied facilitated a “best-fit” model, it is difficult
to expect equations drawn from small numbers to be applicable to all sections of the population. The LV mass derived
from these equations may be useful in large population
studies that have demonstrated prognostic differences1; however, it does not appear to predict LV mass accurately for
individuals nor to be good at detecting changes in individuals
and small groups. Previous studies have quoted a good
correlation of echocardiography-derived LV mass with postmortem values,9,10,12 but this is a poor measure of the strength
4
Hypertension
November 2002
that results in patients with abnormal ventricles would increase the variances we have shown between directly measured and calculated techniques.
The CMR short-axis image slice used for M-mode and 2D
analysis was as close as possible to the positions recommended for echocardiographic analysis, but the greater slice
thickness of CMR may encompass a wider area than the
echocardiographic equivalent, with averaging of the myocardial border positions. In practice, the degree of error is likely
to be small and is offset by the true short-axis alignment
(some misalignment is possible using echocardiography) and
good image quality.
CMR has the same limitations as any MRI, with a small
minority of people unable to undergo scanning for a variety
of reasons, and limited availability. These are covered in
more detail in a previous review.23
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
Perspectives
Figure 3. Scatterplots of change in LV mass over 10-week
physical training program by standard 3D-CMR technique vs
ASE-cube LV mass (a) and 2D area-length LV mass (b) (n⫽140).
Lines of identity are shown. Substantial scatter for individuals is
clear.
of agreement between techniques and indicates only the
linearity of the relation—the trend for one variable to increase
proportionately to another. If the variables are actually 2
different techniques measuring the same parameter, it is
expected that the 2 variables will be proportional to each
other, and little inference can be made about the accuracy of
either technique or their strength of agreement. Considerable
differences may exist in the absolute values from 2 techniques despite a very high correlation.26 Given that the prolate
ellipsoid correlated better than other simple models in the
original comparison studies,18 it is unlikely that any fixed
geometric shape will give an adequate representation of left
ventricular mass. These were also normal hearts, and altered
cardiac shape, slice positioning problems, and poor image
quality will all add to the error from fixed geometric
formulas.
Limitations
The study population involved only young men free from
cardiovascular disease. They were physically fit but not
athletes, and the LV mass index was toward the upper end of
the normal range for a population (96.8⫾16.9 g/m2 before
training). It is unlikely that the LV shape or variability in
shape in this group differs substantially from the rest of the
normal population. In addition, the absence of any cardiac
disease makes this a good group for study because it is likely
The use of improved techniques for the measurement of LV
mass and in particular, greater reproducibility for the changes
in mass, is likely to aid the study of LV hypertrophy and the
effect of pharmaceutical agents and other interventions on its
regression. It may also be possible to determine whether it is
the LV mass per se or some other factor for which LV mass
is a marker that is responsible for the serious adverse health
consequences of LV hypertrophy.
All 3 dimensional techniques—CMR, 3D echocardiography, and electron-beam computed tomography—share the
advantages of direct LV mass measurement, avoiding the use
of formulas, and may be used for the above studies. Each has
limitations: for 3D echocardiography, problems with limited
acoustic access and operator skill; for electron-beam computed tomography, ionizing radiation and limited slice positioning; and for CMR, the unsuitability of a minority of
patients, for example, with pacemakers. All share the problem
of limited availability. It is suggested, however, that for the
majority of subjects, CMR is the preferred method, with a
free choice of imaging planes, uniformly excellent image
quality, and lack of ionizing radiation.
Conclusions
With normal hearts and good image quality, calculated values
for LV mass with the use of echocardiographic formulas and
an assumed geometric shape of the left ventricle resulted in
significant variation from direct measurement with the use of
a 3D technique. The confidence limits of ⫾45 to 55 g
represent significant clinical differences and are sufficiently
large to prevent routine use of these techniques either at a
single time point or for serial studies in small populations,
where 3D imaging techniques are preferable. Irregularities in
shape and reduced image quality will increase these errors.
Acknowledgments
This study was supported by the British Heart Foundation (FS
97030), the Coronary Artery Disease Research Association
(CORDA), and the British Army. We especially thank the staff and
recruits at the Army Training Regiment, Bassingbourn, and the staff
at the CMR Unit for invaluable technical support.
Myerson et al
Reliability of Echocardiographic LV Mass
5
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
Figure 4. Bland-Altman plot showing 95% limits of agreement between 3D-CMR and ASEcube LV mass (a) and 2D area-length LV mass
(b) for change in LV mass over 10-week physical training program. Note substantial limits of
agreement.
References
1. Levy D, Garrison R, Savage D, Kannel W, Castelli W. Prognostic implications of echocardiographically determined left ventricular mass in the
Framingham Heart Study. N Engl J Med. 1990;322:1561–1566.
2. Sullivan J, Vander Zwaag R, el-Zeky F, Ramanathan K, Mirvis D. Left
ventricular hypertrophy: effect on survival. J Am Coll Cardiol. 1993;22:
508 –513.
3. Ghali J, Liao Y, Simmons B, Castaner A, Cao G, Cooper R. The
prognostic role of left ventricular hypertrophy in patients with or without
coronary artery disease. Ann Intern Med. 1992;117:831– 836.
4. Schlaich M, Schmieder R. Left ventricular hypertrophy and its regression:
pathophysiology and therapeutic approach: focus on treatment by antihypertensive agents. Am J Hypertens. 1998;11:1394 –1404.
5. Dahlof B, Pennert K, Hansson L. Reversal of left ventricular hypertrophy
in hypertensive patients: a metaanalysis of 109 treatment studies. Am J
Hypertens. 1992;5:95–110.
6. Lievre M, Gueret P, Gayet C, Roudaut R, Haugh MC, Delair S, Boissel
J-P. Ramipril-induced regression of left ventricular hypertrophy in treated
hypertensive individuals. Hypertension. 1995;25:92–97.
7. Verdecchia P, Schillaci G, Borgioni C, Ciucci A, Gattobigio R, Zampi I,
Reboldi G, Porcellati C. Prognostic significance of changes in left ventricular mass in essential hypertension. Circulation. 1998;97:48 –54.
8. Levy D, Salomon M, D’Agostino R, Belanger A, Kannel W. Prognostic
implications of baseline electrocardiographic features and their serial
changes in subjects with left ventricular hypertrophy. Circulation. 1994;
90:1786 –1793.
9. Devereux RB, Reichek N. Echocardiographic determination of left ventricular mass in man. Anatomic validation of the method. Circulation.
1977;55:613– 619.
10. Devereux R, Alonso D, Lutas E, Gottlieb G, Campo E, Sachs I, Reichek
N. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol. 1986;57:450 – 458.
11. Gopal A, Schnellbaecher M, Shen Z, Akinboboye O, Sapin P, King D.
Freehand three-dimensional echocardiography for measurement of left
ventricular mass: in vivo anatomic validation using explanted human
hearts. J Am Coll Cardiol. 1997;30:802– 810.
12. Reichek N, Helak J, Plappert T, Sutton MS, Weber KT. Anatomic
validation of left ventricular mass estimates from clinical twodimensional echocardiography: initial results. Circulation. 1983;67:
348 –352.
13. Byrd BF III, Finkbeiner W, Bouchard A, Silverman NH, Schiller NB.
Accuracy and reproducibility of clinically acquired two-dimensional
echocardiographic mass measurements. Am Heart J. 1989;118:133–137.
14. Gottdiener J, Livengood S, Meyer P, Chase G. Should echocardiography
be performed to assess effects of antihypertensive therapy? Test-retest
6
15.
16.
17.
18.
19.
20.
Hypertension
November 2002
reliability of echocardiography for measurement of left ventricular mass
and function. J Am Coll Cardiol. 1995;25:424 – 430.
Gosse P, Roudaut R, Dallocchio M. Is echocardiography an adequate
method to evaluate left ventricular hypertrophy regression? Eur Heart J.
1990;11:107–112.
Bottini P, Carr A, Prisant L, Flickinger F, Allison J, Gottdiener J.
Magnetic resonance imaging compared to echocardiography to assess left
ventricular mass in the hypertensive patient. Am J Hypertens. 1995;8:
221–228.
Germain P, Roul G, Kastler B, Mossard J, Bareiss P, Sacrez A.
Inter-study variability in left ventricular mass measurement: comparison
between M-mode echocardiography and MRI. Eur Heart J. 1992;13:
1011–1019.
Geiser E, Bove K. Calculation of left ventricular mass and relative wall
thickness. Arch Pathol. 1974;97:13–21.
Schiller NB, Skioldebrand CG, Schiller EJ, Mavroudis CC, Silverman
NH, Rahimtoola SH, Lipton MJ. Canine left ventricular mass estimation
by two-dimensional echocardiography. Circulation. 1983;68:210 –216.
Wyatt HL, Heng MK, Meerbaum S, Hestenes JD, Cobo JM, Davidson
RM, Corday E. Cross-sectional echocardiography. I. Analysis of math-
21.
22.
23.
24.
25.
26.
ematic models for quantifying mass of the left ventricle in dogs. Circulation. 1979;60:1104 –1113.
Stack RS, Ramage JE, Bauman RP, Rembert JC, Phillips HR, Kisslo JA.
Validation of in vivo two-dimensional echocardiographic dimension measurements using myocardial mass estimates in dogs. Am Heart J. 1987;
113:725–731.
Myerson S, Montgomery H, Whittingham M, Jubb M, World M,
Humphries S, Pennell D. Left ventricular hypertrophy with exercise and
the angiotensin converting enzyme gene I/D polymorphism: a randomised
controlled trial with losartan. Circulation. 2001;103:226 –230.
Myerson SG, Bellenger NG, Pennell DJ. The assessment of left ventricular mass by cardiovascular magnetic resonance. Hypertension. 2002;39:
750 –755.
Katz J, Milliken M, Stray-Gundersen J, Buja L, Parkey R, Mitchell J,
Peshock R. Estimation of human myocardial mass with MR imaging.
Radiology. 1988;169:495– 498.
Sahn DJ, DeMaria A, Kisslo J, Weyman A. Recommendations regarding
quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation. 1978;58:1072–1083.
Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet. 1986;1:307–310.
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
Left Ventricular Mass
Saul G. Myerson, Hugh E. Montgomery, Michael J. World and Dudley J. Pennell
Hypertension. published online October 7, 2002;
Downloaded from http://hyper.ahajournals.org/ by guest on June 15, 2017
Hypertension is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2002 American Heart Association, Inc. All rights reserved.
Print ISSN: 0194-911X. Online ISSN: 1524-4563
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://hyper.ahajournals.org/content/early/2002/10/07/01.HYP.0000036401.99908.DB.citation
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Hypertension can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located,
click Request Permissions in the middle column of the Web page under Services. Further information about
this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Hypertension is online at:
http://hyper.ahajournals.org//subscriptions/