- Email: [email protected]
Measurement of left ventricular mass in hypertrophic cardiomyopathy using MRI: Comparison with echocardiography
Mapwtic Resonance Imaging, Vol. 1 I, pp. 329-334, Printed in the USA. All rights reserved.
l
1993 Copyright 0
0730-725X/93 $6.00 + .oO 1993 Pergamon Press Ltd.
Original Contribution MEASUREMENT OF LEFT VENTRICULAR MASS IN HYPERTROPHIC CARDIOMYOPATHY USING MRI: COMPARISON WITH ECHOCARDIOGRAPHY JERRY D. ALLISON, FRED W. FLICKINGER, JOHN C. WRIGHT, DORTH G. FALLS III, L. MICHAEL PRISANT, THOMAS W. VONDOHLEN, AND MARTIN J. FRANK Medical College of Georgia, Augusta, GA 30912, USA Left ventricular mass (LVM) is an important consideration in the management of cardiac hypertrophy associated with hypertrophic cardiomyopathy (HCM), systemic hypertension, and other diseases. A brief MRI cardiac imaging procedure used to monitor regression of LVM during treatment would be beneficial in management of these patients, since echocardiograms cannot be obtained in all patients and since the volume of a hypertrophic heart can straightforwardly be assessed from a series of tomographic slices. The present study was designed to evaluate a brief cardiac MRI procedure for measurement of LVM in HCM and compare it to echocardiography. MRI images acquired in a simulated transverse body plane were used to evaluate the mass of the left ventricle in 6 ex vivo human hearts obtained at autopsy. The estimates of LVM by MRI in the ex-vivo hearts were within 8% of the actual LVM. MRI images were acquired to evaluate LVM in 5 normal subjects and 12 patients diagnosed with HCM. Echocardiography was accomplished on 4 of the normal subjects and 10 of the patients having HCM. There were no significant differences in LVM by MRI and echocardiographic techniques in normal subjects. Transverse MRI images acquired on normal subjects demonstrated that estimates of LVM are reproducible when repeated over 3-w to 3-mo intervals. Images selected for analysis represented the heart in an early diastolic phase. MRI and echocardiographic techniques demonstrated significant differences in LVM in HCM patients. Estimates of LVM in normal subjects and patients diagnosed with HCM were normalized for body weight. The LVM estimates for HCM patients were very significantly different than normal subjects. A short (20 min) in vivo cardiac MRI exam acquired in the transverse body plane can be used to accurately measure LVM. Normalized estimates of LVM may be a useful index for monitoring the progression or regression of LVM in HCM and hypertension and following aortic valve replacement for aortic stenosis. Keywords: Magnetic resonance imaging; Echocardiography;
INTRODUCTION
Left ventricular mass; Hypertrophic cardiomyopathy.
Left ventricular mass (LVM) is an important consideration in the management of cardiac hypertrophy associated with hypertrophic cardiomyopathy (HCM), systemic hypertension, and other diseases. It has been demonstrated that echocardiographic left ventricular hypertrophy is an independent cardiovascular risk factor.’ The purpose of this research is to evaluate a brief cardiac imaging procedure for measurement of LVM in HCM and compare it to echocardiography. Although echocardiography is not considered to be a gold standard in evaluation of LVM in HCM, it is routinely used.
Angiocardiography and echocardiography have been validated as methods for the determination of LVM in normal hearts. Calculation of LVM from 2D echocardiographic measurements use, among others, the arealength method or a truncated ellipsoidal mode1.2-4 Calculation of LVM from M-Mode echocardiographic measurements use, among others, a prolate ellipsoidal model.5 Although not all echocardiographic techniques for measurement of LVM utilize an ellipsoidal model, those that do may introduce inaccuracy if dilatation due to valvular or myocardial disease, asymmetric hypertrophy (especially HCM), or wall motion abnormalities are present. Approximately 15-20’70of patients referred for echocardiography have an inadequate
RECEIVEDS/27/92; ACCEPTED 12/2/92. Address correspondence to Jerry D. Allison, PhD, Medi-
cal College of Georgia, 112015th Street (AE-2018),Augusta, GA 30912. 329
330
Magnetic Resonance Imaging 0 Volume 11, Number 3, 1993
acoustic window needed to measure cardiac dimensions reproducibly. MRI estimates of LVM require no assumptions concerning left ventricular shape. The MRI assessment of LVM in men, using modified short axis images, normalized for body weight, agreed with normal limits derived from autopsy data.6 MRI estimation of LVM using short axis images has been shown to be accurate and reproducible in ex vivo cadaver hearts and normal subjects.’ Manual tracing and computerized edge detection of cardiac boundaries were accurate and had close correspondence when used to assess endocardial area, epicardial area, and left ventricular mass from short axis MRI images in excised animal hearts and normal human subjects.8 A study of LVM from short axis MRI images in dogs before and after surgically induced infarction concluded that “MRI accurately determines LV mass in both distorted and normal hearts.“’ Studies in canines using transverse images corrected for partial volume effects have demonstrated that in vivo MRI assessment of normal and hypertrophic canine hearts accurately predicts the actual postmortem weight of the left ventricle.‘O Serial evaluation of LVM from short axis MRI images demonstrated significant increases in LVM following surgically induced hypertrophy in canine puppies. ” Transverse MRI techniques correlated (r = 0.96) with echocardiography in assessment of right and left ventricular volumes in healthy individuals.6 Estimates of interventricular septal thickness using multiplanar MR images correlated (r = 0.91) with echocardiography in documenting regression of cardiac hypertrophy during ramipril treatment.” Evaluation of MRI for assessment of LVM in HCM has not been reported in humans. The lack of timely postmortem data makes validation of LVM data difficult because there is not an accepted reference standard. We evaluated MRI for measurement of LVM in HCM using the following strategy. A. Validation of methodology using ex vivo hearts: We validated that our methodology for measurement of LVM without correcting for partial volume effects accurately predicts the weight of the isolated left ventricle for ex vivo hearts imaged in simulated transverse body planes. B. Comparison of MRI and echocardiography in normal subjects: We compared transverse MRI techniques to echocardiographic techniques for measurement of LVM in normal human subjects. Examination of reproducibility and sensitivity to cardiac phase were included. C. Comparison of MRI and echocardiography in patients having HCM:
We compared transverse MRI techniques to echocardiographic techniques for measurement of LVM in patients having HCM. MATERIALS AND METHODS Six ex vivo hearts were imaged in a 1.5 T Signa MRI system (GE Medical Systems, Milwaukee, WI). The hearts were obtained at autopsy of young adults having no evidence of cardiac disease. The hearts were not preserved and were imaged within 24 hr of death. A pathologist used opaque paper to orient each heart within a Plexiglas box so that the geometric relationship of the heart to the orthogonal axes of the box were similar to those of the heart in conventional body imaging planes. The hearts were placed in the bore of the magnet in an orientation simulating in vivo cardiac imaging. Images were acquired in a simulated transverse body plane using a conventional spin-echo imaging technique (TR:600, TE:20, Axial Plane, 32 cm FOV, 1 excitation, 10 mm slice thickness, skip 0 mm, 128 matrix, frequency L/R, frequency wrap correction). Images were acquired in the transverse plane since it reduces the examination time. The voxel dimensions were 1Omm x 1.25mm x 1.25mm. Each of approximately 10 slices of the left ventricle were analyzed on an image processing workstation. The epicardial and endocardial walls were identified using a mouse-guided cursor (Fig. 1). The interface of chamber void and the myocardium is identified as the endocardial boundary. The interface of the epicardial fat and the myocardium is identified as the epicardial boundary. The interventricular septum was included as part of the left ventricular mass because (1) the echocardiographic techniques (Woythaler, ASE) used at our institution for measurement of LVM specifically include the interventricular septal thickness as well as the posterior wall thickness; and (2) the septum was included when the ex vivo left ventricles in this project were weighed. Identification of wall boundaries required approximately 45 s per slice (7.5 min per heart). Pixels (voxels) included in the left ventricular wall were totalled for all slices to determine the number of voxels comprising the left ventricle and hence the left ventricular volume. A specific gravity of 1.06 g/cc was used to calculate the LVM. The ex vivo left ventricles were weighed following separation from the other cardiac chambers, blood vessels and epicardial fat. Estimates of LVM in 5 normal subjects and 12 patients diagnosed with HCM were derived using the same methodology as the ex vivo hearts except that cardiac gating, respiratory compensation, and phase wrap correction were used during data acquisition. Subjects enrolled in the study were over the age of 20 in an attempt
Measurement of LV mass 0 J.D. ALLKON
331
ET AL.
Conventional echocardiographic techniques .were used to estimate linear cardiac dimensions and LVM in the normal subjects and patients having HCM. For the normal subjects, M-Mode and 2D estimates of LVM were calculated using Woythaleri3 and American Society of Echocardiography (ASE)14 methodologies. For patients having HCM, M-Mode estimates of LVM were calculated using the Woythaler and ASE methodologies. The estimates of LVM in normal subjects were divided by body weight to develop a normalized LVM. Body weight has been shown to be a better predictor of normal heart weight than height or body surface area.15 Data were not normalized for age or sex, since it has been reported that the heart weight in normal men does not change significantly with age, and since age related changes in heart weight for women are relatively small.” Normalizing LVM for body weight reduces the significance of differences between LVM in women and men.” RESULTS
Fig. 1. Transverse MR image from HCM patient at the cardiac base. This gated image was acquired 252 msec after the R wave (R-R = 923 msec). The area outlined represents myocardium consisting of 1459 voxels (1.72 mm x 1.72 mm x 10 mm thick): and contributes 45.6g to the total mass (302.9 g) of the left ventricle. Note that an elliptical model would not adequately describe this heart due to asymmetry of muscular hypertrophy.
to minimize any influence on cardiac measurements due to body growth. The inflow of fresh blood was satu-
rated using five slabs (anterior, inferior, superior, left, right). A multislice-multiphase cardiac gated spin-echo imaging technique was used. Eight to twelve contiguous slices (10 mm thick) that encompassed the full extent of the left ventricle were imaged. Images were acquired at each of 10 different times using the cardiac cycle for each slice location (64-144 images per acquisition). The effective TR was one R-R interval which resulted in an imaging time of approximately 20 min. Images selected for analysis represented a cardiac phase of approximately 252 msec after the R wave, since these images best defined the endocardial and epicardial walls of the left ventricle with a minimum of signal void and motion artifacts. Studies in canines have demonstrated that end systolic and late diastolic images are both accurate for estimating LVM.” In order to assess variability of the MRI technique for estimating LVM, four of the normal subjects were repeated with 3 wk to 3 mo between subsequent MRI examinations.
MRI evaluations of LVM in the ex-vivo hearts were within 8% of the mass determined by actually weighing the hearts as indicated in Fig. 2. There was not a significant difference between the two methods (p > .8).* The agreement between MRI and echocardiographic techniques for estimation of LVM in normal subjects was examined. There were no significant differences between: MRI and 2D Woythaler (p > .2); MRI and 2D ASE (p > . 1); MRI and M-Mode Woythaler (p > .4) and MRI and M-Mode ASE (p > .2). Of the four normal subjects that had echocardiography examinations, one could not be evaluated with M-Mode and another could not be evaluated with 2D techniques due to limited acoustic windows. The variability of the MRI technique for LVM data acquired on different days and at two slightly different cardiac phases was demonstrated in four normal subjects. Data acquired in normal subjects indicated that evaluation of LVM was not significantly different for images acquired at 190 ms and 252 ms following the R wave (p > . 1). The range of variation for images acquired at two cardiac phases on two separate days was 22.5 g as demonstrated in Table 1. Studies were repeated on four of the normal subjects. Data acquired in the same patient on separate days, having intervals *Using a 95% confidence interval, values ofp 2 .05 confirm the null hypothesis that there is not a significant difference between the sample means of the two groups. Values of p < .05 reject the null hypothesis and indicate that there is a significant difference between the sample means of the two groups.
Magnetic Resonance Imaging 0 Volume 11, Number 3, 1993
332
280 260 E240 z220 ; 200 ?I z 180 ‘$ 160 in 140
Ex-vivo LVM by weighing(g)
Fig. 2. Comparison (r = 0.99).
of MRI technique
for estimation
of the left ventricular
of 3 wk to 3 mo, showed no significant differences in estimation of LVM from transverse images (p > .05). LVM by MRI in patients having HCM was significantly larger than normal subjects (p < .OOl) as shown in Tables 1 and 2. LVM by M-Mode in patients having HCM was significantly larger than normal subjects
Table 1. Normal Time post R-wave (ms)
Subject
mass in ex vivo hearts to weighing
the hearts
(Woythaler p < .005; ASE p < .005). LVM by MRI in patients having HCM was significantly smaller than by M-Mode Woythaler (p c .005) and by M-Mode ASE (p < .Ol). The correlation between MRI and echocardiographic techniques was poor in patients having HCM (Woythaler r = 0.17; ASE r = 0.17).
subjects
MRI LVM (g)
LVM data
M-mode LVM (Woyth.) (g)
M-mode LVM (A=) (g)
2D LVM (Woyth.) (g)
2D LVM (A=) (g)
#l (02/13/90) #l (02/13/N) #l (05/04/90) # 1 (05/04/90)
134 134 140 140
183 243 190 252
92.7 108.5 114.2 113.2
*
*
148.7
163.8
#2 (02/16/90) #2 (02/16/90)
190 252 190 252
71.8 79.5 87.4 93.1
53.4
67.7
70.6
85.0
#2 (05/02/90)
106 106 105 105
#3 (02/27/90)
163
259
147.6
t
t
#4 (02/28/90) #4 (02/28/90) #4 (05/07/90)
190 252 190 252
91.2 90.1 112.6 100.4
130.3
145.0
100.5
#4(05/07/90)
130 130 133 133
#5(04/16/90) #5 (04/16/90) #5 (05/09/90) #5 (05/09/90)
95 95 95 95
190 252 190 252
81.2 84.6 82.5 79.6
109.9
124.6
*
#2 (05/02/!XI)
*Incomplete due to limited acoustic window. tEcho data not acquired.
t
t 115.1
*
Measurement of LV mass 0 J.D. Table 2. Hypertrophic cardiomyopathy LVM data
Patient
zi
Time post R-wave (ms)
MRI LVM (g)
#l
190
#1 #2 #3
190 180 120
190 252 252 190
#4 #5 #6 #7 #8 #9 #lO #ll #12
202 143 224 234 255 227 190 101 150
252 252 252 252 252 252 252 190 376
ALLISON ET AL.
333
Table 4. Hypertrophic cardiomyopathy normalized LVM data
patients
M-mode LVM (Woyth.) (g)
M-mode LVM (A=) (g)
394.4 392.7 400.7 302.9
396.1
378.7
* *
* *
302.4 220.2 416.4 347.4 391.0 363.3 212.8 164.6 239.3
336.0 367.0 * 616.7 540.9 631.2 707.8 383.1 t
319.2 349.9 * 597.2 522.1 611.6 687.4 365.9 t
MRI evaluation of LVM (g) 393 401 303 302 220 416 347 391 363 213 165 239
patients
Normalized MRI LVM (g/k) 4.56 4.91 5.57 3.30 3.39 4.09 3.27 3.38 3.53 2.47 3.28 3.51
*Incomplete due to limited acoustic window. tEcho data not acquired.
Normalized LVM by MRI in normal subjects had a mean value of 1.80 g/kg (+O.ZO; 1 SD) as indicated in Table 3. Normalized LVM by MRI in patients diagnosed as having cardiac hypertrophy had a mean value of 3.77 g/kg (kO.86; 1 SD) as indicated in Table 4. Normalized LVM by MRI was significantly larger in patients having HCM than normal subjects (p < .OOl) as shown in Tables 3 and 4. DISCUSSION Data acquired in ex vivo hearts demonstrate that transverse MRI images can be used to accurately as-
sess the mass of the left ventricle without correction for partial volume effects. Correction for partial volume effects was not necessary since the most caudal slice represented ‘a relatively small percentage of the total mass and since slice locations were selected to place the apex on a slice boundary. Accuracy of the
Table 3. Normal subjects normalized LVM data MRI evaluation. of LVM (g)
Normalized MRI LVM (g/kg)
113 79.5 148 90.1 84.6
1.79 1.67 2.01 1.54 1.98
technique should not be compromised by use on hypertrophic hearts since no assumptions of left ventricular shape are involved and since accuracy is expected to improve in proportion to the square root of the number of voxels involved. Data in normal subjects showed no significant differences in LVM between MRI and echocardiographic techniques. Data acquired in normal volunteers indicate LVM derived from transverse MRI images is reproducible and insensitive to the cardiac phase of the images selected for analysis. MRI and echocardiography demonstrated significant increases in LVM in HCM. LVM by MRI was significantly smaller than M-Mode measurement of LVM in HCM. We believe that MRI is an accurate technique for assessment of LVM in HCM since: 1. MRI techniques require no geometric assumptions. 2. Echocardiographic techniques in use at our hospital are among those that utilize an ellipsoidal model. The left ventricular myocardial wall shows irregular patterns of thickening (and thinning) in HCM and hence does not conform to the ellipsoidal model (as demonstrated in Fig. 1). Linear cardiac dimensions (left ventricular end-diastolic dimension, interventricular septal dimension, posterior wall dimension) are inadequate for describing the mass of a hypertrophied left ventricle since the left ventricle does not enlarge uniformly as a function of increases of selected linear dimensions. The Woythaler and ASE conventions involve the third power of linear cardiac dimensions which can produce significant errors since a few linear dimensions cannot adequately describe an irregular geometric shape.
334
Magnetic Resonance Imaging 0 Volume 11, Number 3, 1993
3. In vivo MRI techniques in hypertrophic canines have been shown to accurately predict the post mortem weight of the left ventricle.” The very significant difference between normalized LVM in normal subjects and HCM patients indicates that it may be a sensitive indicator of progression or regression of LVM during hyptertrophy of all etiologies - HCM, systemic hypertension, aortic stenosis, and so on. It has been demonstrated that interventricular septal thickness decreased from 19.6 mm to 15.2 mm in MRI studies of 32 hypertensive patients undergoing treatment with ramipril.‘* Echocardiography demonstrated a decrease from 18.8 mm to 14.6 mm in the same patient population. The decrease in interventricular septal thickness was on the order of 22%. Data in Tables 3 and 4 indicate that the normalized LVM can increase on the order of 100’70,producing a potentially more sensitive indicator of regression during treatment. Successful therapy of hypertension results in regression of LVM within 3 days to 6 wk.‘* A brief MRI cardiac imaging procedure used to monitor progression or regression of LVM would be beneficial in management of these patients, especially in patients with unattainable echocardiograms. The use of transverse imaging to reduce the examination time is an important consideration in routine monitoring of LVM. Subsequent research will explore the use of normalized LVM by MRI to .monitor regression during therapy of hypertensive patients.
4.
5.
6.
7.
8.
9.
10.
11.
12. Acknowledgments- Image processing was accomplished on a Sun Microsystems 4/24O(TAAC-1) computer (Sun Microsystems, Mountain View, CA) using MR Console 4.5 research software (General Electric Medical Systems, Milwaukee, WI). MR Console 4.5 facilitated selection of endocardial and epicardial walls using a mouseguided cursor in a user friendly environment.
13.
REFERENCES 14. Levy, D.; Anderson, K.M. ; Savage, D.D.; et al. Echocardiographically detected left ventricular hypertrophy: Prevalence and risk factors. Ann. Intern. Med. 108:713; 1988. Schiller, N.B.; Shah, P.M.; Crawford, M.; et al. Recommendations for quantitation of the left ventricle by twodimensional echocardiography. J. Am. Sot. Echo 2(S): 358-367; 1989. Reichek, N. Standardization in the measurement of left
15.
ventricular wall mass two-dimensional echocardiography. Hypertension Suppl 11: 11-30-11-32; 1987. Schiller, N.B. Considerations in the standardization of measurement of left ventricular myocardial mass by twodimensional echocardiography. Hypertension Supplll: 11-33-H-35; 1987. Reichek, N. Standardization in the measurement of left ventricular wall mass two-dimensional echocardiography. Hypertension Suppl 11: H-27-11-29; 1987. Sechtem, U.; Pflugfelder, P.W.; Gould, R.G.; Cassidy, M.M.; Higgins, C.B. Measurement of right and left ventricular volumes in healthy individuals with tine MR imaging. Radiology 163:697-702; 1987. Katz, J.; Milliken, M.C.; Stray-Gundersen, J.; Buja, L.M.; Parkey, R.W.; Mitchell, J.H.; Peshock, R.M. Estimation of human myocardial mass with MR imaging. Radiology 169:495-498; 1988. Fleagle, S.R.; Thedens, D.R.; Ehrhardt, J.C.; Scholz, T.D.; Skorton, D. J. Automated identification of left ventricular borders from spin-echo magnetic resonance images. Invest. Radiol. 26:295-303; 1991. Weiss, J.L.; Shapiro, E.P. ; Buchalter, M.B.; Beyar, R. Magnetic resonance imaging as a noninvasive standard for the quantitative evaluation of left ventricular mass, ischemia, and infarction. Ann. NY Acad. Sci. 601:95106; 1990. Caputo, G.R.; Tscholakoff, D.; Sechtem, U.; Higgins, C.B. Measurement of canine left ventricular mass by using MR imaging. AJR 148:33-38; 1987. Wang, J.Z.; Mezrich, R.S.; Scholz, P.; Als, A.; Douglas, F. MRI evaluation of left ventricular hypertrophy in a canine model of aortic stenosis. Invest. Radiol. 25:783788; 1990. Eichstaedt, H.; Danne, 0.; Langer, M.; Cordes, M.; Schubert, C.; Felix, R.; Schmutzler, H. Regression of left ventricular hypertrophy under ramipril treatment investigated by nuclear magnetic resonance imaging. J. Cardiovasc. Pharmacol. 13(3):S75-S80; 1989. Woythaler, J.N.; Singer, S.L.; Kwan, O.L.; et al. Accuracy of echocardiography versus electrocardiography in detecting left ventricular hypertrophy: Comparison with post-mortem mass measurements. J. Am. Coil. Cardiol. 2:305-311; 1983. Sahn, D.J.; DeMaria, A.; Kisslo, J.; Weyman, A. Recommendations regarding quantitation in M-mode echocardiography: Results of a survey of echocardiographic measurements. Circulation 58:1072-1083; 1978. Kitzman, D.W.; Scholz, D.G.; Hagen, P.T.; Ilstrup, D.M.; Edwards, W.D. Age-related changes in normal human hearts during the first 10 decades of life. Part 11 (Maturity): A quantitative anatomic study of 765 specimens from subjects 20 to 99 years old. Mayo Ctin. Proc. 63:137-146; 1988.
l
1993 Copyright 0
0730-725X/93 $6.00 + .oO 1993 Pergamon Press Ltd.
Original Contribution MEASUREMENT OF LEFT VENTRICULAR MASS IN HYPERTROPHIC CARDIOMYOPATHY USING MRI: COMPARISON WITH ECHOCARDIOGRAPHY JERRY D. ALLISON, FRED W. FLICKINGER, JOHN C. WRIGHT, DORTH G. FALLS III, L. MICHAEL PRISANT, THOMAS W. VONDOHLEN, AND MARTIN J. FRANK Medical College of Georgia, Augusta, GA 30912, USA Left ventricular mass (LVM) is an important consideration in the management of cardiac hypertrophy associated with hypertrophic cardiomyopathy (HCM), systemic hypertension, and other diseases. A brief MRI cardiac imaging procedure used to monitor regression of LVM during treatment would be beneficial in management of these patients, since echocardiograms cannot be obtained in all patients and since the volume of a hypertrophic heart can straightforwardly be assessed from a series of tomographic slices. The present study was designed to evaluate a brief cardiac MRI procedure for measurement of LVM in HCM and compare it to echocardiography. MRI images acquired in a simulated transverse body plane were used to evaluate the mass of the left ventricle in 6 ex vivo human hearts obtained at autopsy. The estimates of LVM by MRI in the ex-vivo hearts were within 8% of the actual LVM. MRI images were acquired to evaluate LVM in 5 normal subjects and 12 patients diagnosed with HCM. Echocardiography was accomplished on 4 of the normal subjects and 10 of the patients having HCM. There were no significant differences in LVM by MRI and echocardiographic techniques in normal subjects. Transverse MRI images acquired on normal subjects demonstrated that estimates of LVM are reproducible when repeated over 3-w to 3-mo intervals. Images selected for analysis represented the heart in an early diastolic phase. MRI and echocardiographic techniques demonstrated significant differences in LVM in HCM patients. Estimates of LVM in normal subjects and patients diagnosed with HCM were normalized for body weight. The LVM estimates for HCM patients were very significantly different than normal subjects. A short (20 min) in vivo cardiac MRI exam acquired in the transverse body plane can be used to accurately measure LVM. Normalized estimates of LVM may be a useful index for monitoring the progression or regression of LVM in HCM and hypertension and following aortic valve replacement for aortic stenosis. Keywords: Magnetic resonance imaging; Echocardiography;
INTRODUCTION
Left ventricular mass; Hypertrophic cardiomyopathy.
Left ventricular mass (LVM) is an important consideration in the management of cardiac hypertrophy associated with hypertrophic cardiomyopathy (HCM), systemic hypertension, and other diseases. It has been demonstrated that echocardiographic left ventricular hypertrophy is an independent cardiovascular risk factor.’ The purpose of this research is to evaluate a brief cardiac imaging procedure for measurement of LVM in HCM and compare it to echocardiography. Although echocardiography is not considered to be a gold standard in evaluation of LVM in HCM, it is routinely used.
Angiocardiography and echocardiography have been validated as methods for the determination of LVM in normal hearts. Calculation of LVM from 2D echocardiographic measurements use, among others, the arealength method or a truncated ellipsoidal mode1.2-4 Calculation of LVM from M-Mode echocardiographic measurements use, among others, a prolate ellipsoidal model.5 Although not all echocardiographic techniques for measurement of LVM utilize an ellipsoidal model, those that do may introduce inaccuracy if dilatation due to valvular or myocardial disease, asymmetric hypertrophy (especially HCM), or wall motion abnormalities are present. Approximately 15-20’70of patients referred for echocardiography have an inadequate
RECEIVEDS/27/92; ACCEPTED 12/2/92. Address correspondence to Jerry D. Allison, PhD, Medi-
cal College of Georgia, 112015th Street (AE-2018),Augusta, GA 30912. 329
330
Magnetic Resonance Imaging 0 Volume 11, Number 3, 1993
acoustic window needed to measure cardiac dimensions reproducibly. MRI estimates of LVM require no assumptions concerning left ventricular shape. The MRI assessment of LVM in men, using modified short axis images, normalized for body weight, agreed with normal limits derived from autopsy data.6 MRI estimation of LVM using short axis images has been shown to be accurate and reproducible in ex vivo cadaver hearts and normal subjects.’ Manual tracing and computerized edge detection of cardiac boundaries were accurate and had close correspondence when used to assess endocardial area, epicardial area, and left ventricular mass from short axis MRI images in excised animal hearts and normal human subjects.8 A study of LVM from short axis MRI images in dogs before and after surgically induced infarction concluded that “MRI accurately determines LV mass in both distorted and normal hearts.“’ Studies in canines using transverse images corrected for partial volume effects have demonstrated that in vivo MRI assessment of normal and hypertrophic canine hearts accurately predicts the actual postmortem weight of the left ventricle.‘O Serial evaluation of LVM from short axis MRI images demonstrated significant increases in LVM following surgically induced hypertrophy in canine puppies. ” Transverse MRI techniques correlated (r = 0.96) with echocardiography in assessment of right and left ventricular volumes in healthy individuals.6 Estimates of interventricular septal thickness using multiplanar MR images correlated (r = 0.91) with echocardiography in documenting regression of cardiac hypertrophy during ramipril treatment.” Evaluation of MRI for assessment of LVM in HCM has not been reported in humans. The lack of timely postmortem data makes validation of LVM data difficult because there is not an accepted reference standard. We evaluated MRI for measurement of LVM in HCM using the following strategy. A. Validation of methodology using ex vivo hearts: We validated that our methodology for measurement of LVM without correcting for partial volume effects accurately predicts the weight of the isolated left ventricle for ex vivo hearts imaged in simulated transverse body planes. B. Comparison of MRI and echocardiography in normal subjects: We compared transverse MRI techniques to echocardiographic techniques for measurement of LVM in normal human subjects. Examination of reproducibility and sensitivity to cardiac phase were included. C. Comparison of MRI and echocardiography in patients having HCM:
We compared transverse MRI techniques to echocardiographic techniques for measurement of LVM in patients having HCM. MATERIALS AND METHODS Six ex vivo hearts were imaged in a 1.5 T Signa MRI system (GE Medical Systems, Milwaukee, WI). The hearts were obtained at autopsy of young adults having no evidence of cardiac disease. The hearts were not preserved and were imaged within 24 hr of death. A pathologist used opaque paper to orient each heart within a Plexiglas box so that the geometric relationship of the heart to the orthogonal axes of the box were similar to those of the heart in conventional body imaging planes. The hearts were placed in the bore of the magnet in an orientation simulating in vivo cardiac imaging. Images were acquired in a simulated transverse body plane using a conventional spin-echo imaging technique (TR:600, TE:20, Axial Plane, 32 cm FOV, 1 excitation, 10 mm slice thickness, skip 0 mm, 128 matrix, frequency L/R, frequency wrap correction). Images were acquired in the transverse plane since it reduces the examination time. The voxel dimensions were 1Omm x 1.25mm x 1.25mm. Each of approximately 10 slices of the left ventricle were analyzed on an image processing workstation. The epicardial and endocardial walls were identified using a mouse-guided cursor (Fig. 1). The interface of chamber void and the myocardium is identified as the endocardial boundary. The interface of the epicardial fat and the myocardium is identified as the epicardial boundary. The interventricular septum was included as part of the left ventricular mass because (1) the echocardiographic techniques (Woythaler, ASE) used at our institution for measurement of LVM specifically include the interventricular septal thickness as well as the posterior wall thickness; and (2) the septum was included when the ex vivo left ventricles in this project were weighed. Identification of wall boundaries required approximately 45 s per slice (7.5 min per heart). Pixels (voxels) included in the left ventricular wall were totalled for all slices to determine the number of voxels comprising the left ventricle and hence the left ventricular volume. A specific gravity of 1.06 g/cc was used to calculate the LVM. The ex vivo left ventricles were weighed following separation from the other cardiac chambers, blood vessels and epicardial fat. Estimates of LVM in 5 normal subjects and 12 patients diagnosed with HCM were derived using the same methodology as the ex vivo hearts except that cardiac gating, respiratory compensation, and phase wrap correction were used during data acquisition. Subjects enrolled in the study were over the age of 20 in an attempt
Measurement of LV mass 0 J.D. ALLKON
331
ET AL.
Conventional echocardiographic techniques .were used to estimate linear cardiac dimensions and LVM in the normal subjects and patients having HCM. For the normal subjects, M-Mode and 2D estimates of LVM were calculated using Woythaleri3 and American Society of Echocardiography (ASE)14 methodologies. For patients having HCM, M-Mode estimates of LVM were calculated using the Woythaler and ASE methodologies. The estimates of LVM in normal subjects were divided by body weight to develop a normalized LVM. Body weight has been shown to be a better predictor of normal heart weight than height or body surface area.15 Data were not normalized for age or sex, since it has been reported that the heart weight in normal men does not change significantly with age, and since age related changes in heart weight for women are relatively small.” Normalizing LVM for body weight reduces the significance of differences between LVM in women and men.” RESULTS
Fig. 1. Transverse MR image from HCM patient at the cardiac base. This gated image was acquired 252 msec after the R wave (R-R = 923 msec). The area outlined represents myocardium consisting of 1459 voxels (1.72 mm x 1.72 mm x 10 mm thick): and contributes 45.6g to the total mass (302.9 g) of the left ventricle. Note that an elliptical model would not adequately describe this heart due to asymmetry of muscular hypertrophy.
to minimize any influence on cardiac measurements due to body growth. The inflow of fresh blood was satu-
rated using five slabs (anterior, inferior, superior, left, right). A multislice-multiphase cardiac gated spin-echo imaging technique was used. Eight to twelve contiguous slices (10 mm thick) that encompassed the full extent of the left ventricle were imaged. Images were acquired at each of 10 different times using the cardiac cycle for each slice location (64-144 images per acquisition). The effective TR was one R-R interval which resulted in an imaging time of approximately 20 min. Images selected for analysis represented a cardiac phase of approximately 252 msec after the R wave, since these images best defined the endocardial and epicardial walls of the left ventricle with a minimum of signal void and motion artifacts. Studies in canines have demonstrated that end systolic and late diastolic images are both accurate for estimating LVM.” In order to assess variability of the MRI technique for estimating LVM, four of the normal subjects were repeated with 3 wk to 3 mo between subsequent MRI examinations.
MRI evaluations of LVM in the ex-vivo hearts were within 8% of the mass determined by actually weighing the hearts as indicated in Fig. 2. There was not a significant difference between the two methods (p > .8).* The agreement between MRI and echocardiographic techniques for estimation of LVM in normal subjects was examined. There were no significant differences between: MRI and 2D Woythaler (p > .2); MRI and 2D ASE (p > . 1); MRI and M-Mode Woythaler (p > .4) and MRI and M-Mode ASE (p > .2). Of the four normal subjects that had echocardiography examinations, one could not be evaluated with M-Mode and another could not be evaluated with 2D techniques due to limited acoustic windows. The variability of the MRI technique for LVM data acquired on different days and at two slightly different cardiac phases was demonstrated in four normal subjects. Data acquired in normal subjects indicated that evaluation of LVM was not significantly different for images acquired at 190 ms and 252 ms following the R wave (p > . 1). The range of variation for images acquired at two cardiac phases on two separate days was 22.5 g as demonstrated in Table 1. Studies were repeated on four of the normal subjects. Data acquired in the same patient on separate days, having intervals *Using a 95% confidence interval, values ofp 2 .05 confirm the null hypothesis that there is not a significant difference between the sample means of the two groups. Values of p < .05 reject the null hypothesis and indicate that there is a significant difference between the sample means of the two groups.
Magnetic Resonance Imaging 0 Volume 11, Number 3, 1993
332
280 260 E240 z220 ; 200 ?I z 180 ‘$ 160 in 140
Ex-vivo LVM by weighing(g)
Fig. 2. Comparison (r = 0.99).
of MRI technique
for estimation
of the left ventricular
of 3 wk to 3 mo, showed no significant differences in estimation of LVM from transverse images (p > .05). LVM by MRI in patients having HCM was significantly larger than normal subjects (p < .OOl) as shown in Tables 1 and 2. LVM by M-Mode in patients having HCM was significantly larger than normal subjects
Table 1. Normal Time post R-wave (ms)
Subject
mass in ex vivo hearts to weighing
the hearts
(Woythaler p < .005; ASE p < .005). LVM by MRI in patients having HCM was significantly smaller than by M-Mode Woythaler (p c .005) and by M-Mode ASE (p < .Ol). The correlation between MRI and echocardiographic techniques was poor in patients having HCM (Woythaler r = 0.17; ASE r = 0.17).
subjects
MRI LVM (g)
LVM data
M-mode LVM (Woyth.) (g)
M-mode LVM (A=) (g)
2D LVM (Woyth.) (g)
2D LVM (A=) (g)
#l (02/13/90) #l (02/13/N) #l (05/04/90) # 1 (05/04/90)
134 134 140 140
183 243 190 252
92.7 108.5 114.2 113.2
*
*
148.7
163.8
#2 (02/16/90) #2 (02/16/90)
190 252 190 252
71.8 79.5 87.4 93.1
53.4
67.7
70.6
85.0
#2 (05/02/90)
106 106 105 105
#3 (02/27/90)
163
259
147.6
t
t
#4 (02/28/90) #4 (02/28/90) #4 (05/07/90)
190 252 190 252
91.2 90.1 112.6 100.4
130.3
145.0
100.5
#4(05/07/90)
130 130 133 133
#5(04/16/90) #5 (04/16/90) #5 (05/09/90) #5 (05/09/90)
95 95 95 95
190 252 190 252
81.2 84.6 82.5 79.6
109.9
124.6
*
#2 (05/02/!XI)
*Incomplete due to limited acoustic window. tEcho data not acquired.
t
t 115.1
*
Measurement of LV mass 0 J.D. Table 2. Hypertrophic cardiomyopathy LVM data
Patient
zi
Time post R-wave (ms)
MRI LVM (g)
#l
190
#1 #2 #3
190 180 120
190 252 252 190
#4 #5 #6 #7 #8 #9 #lO #ll #12
202 143 224 234 255 227 190 101 150
252 252 252 252 252 252 252 190 376
ALLISON ET AL.
333
Table 4. Hypertrophic cardiomyopathy normalized LVM data
patients
M-mode LVM (Woyth.) (g)
M-mode LVM (A=) (g)
394.4 392.7 400.7 302.9
396.1
378.7
* *
* *
302.4 220.2 416.4 347.4 391.0 363.3 212.8 164.6 239.3
336.0 367.0 * 616.7 540.9 631.2 707.8 383.1 t
319.2 349.9 * 597.2 522.1 611.6 687.4 365.9 t
MRI evaluation of LVM (g) 393 401 303 302 220 416 347 391 363 213 165 239
patients
Normalized MRI LVM (g/k) 4.56 4.91 5.57 3.30 3.39 4.09 3.27 3.38 3.53 2.47 3.28 3.51
*Incomplete due to limited acoustic window. tEcho data not acquired.
Normalized LVM by MRI in normal subjects had a mean value of 1.80 g/kg (+O.ZO; 1 SD) as indicated in Table 3. Normalized LVM by MRI in patients diagnosed as having cardiac hypertrophy had a mean value of 3.77 g/kg (kO.86; 1 SD) as indicated in Table 4. Normalized LVM by MRI was significantly larger in patients having HCM than normal subjects (p < .OOl) as shown in Tables 3 and 4. DISCUSSION Data acquired in ex vivo hearts demonstrate that transverse MRI images can be used to accurately as-
sess the mass of the left ventricle without correction for partial volume effects. Correction for partial volume effects was not necessary since the most caudal slice represented ‘a relatively small percentage of the total mass and since slice locations were selected to place the apex on a slice boundary. Accuracy of the
Table 3. Normal subjects normalized LVM data MRI evaluation. of LVM (g)
Normalized MRI LVM (g/kg)
113 79.5 148 90.1 84.6
1.79 1.67 2.01 1.54 1.98
technique should not be compromised by use on hypertrophic hearts since no assumptions of left ventricular shape are involved and since accuracy is expected to improve in proportion to the square root of the number of voxels involved. Data in normal subjects showed no significant differences in LVM between MRI and echocardiographic techniques. Data acquired in normal volunteers indicate LVM derived from transverse MRI images is reproducible and insensitive to the cardiac phase of the images selected for analysis. MRI and echocardiography demonstrated significant increases in LVM in HCM. LVM by MRI was significantly smaller than M-Mode measurement of LVM in HCM. We believe that MRI is an accurate technique for assessment of LVM in HCM since: 1. MRI techniques require no geometric assumptions. 2. Echocardiographic techniques in use at our hospital are among those that utilize an ellipsoidal model. The left ventricular myocardial wall shows irregular patterns of thickening (and thinning) in HCM and hence does not conform to the ellipsoidal model (as demonstrated in Fig. 1). Linear cardiac dimensions (left ventricular end-diastolic dimension, interventricular septal dimension, posterior wall dimension) are inadequate for describing the mass of a hypertrophied left ventricle since the left ventricle does not enlarge uniformly as a function of increases of selected linear dimensions. The Woythaler and ASE conventions involve the third power of linear cardiac dimensions which can produce significant errors since a few linear dimensions cannot adequately describe an irregular geometric shape.
334
Magnetic Resonance Imaging 0 Volume 11, Number 3, 1993
3. In vivo MRI techniques in hypertrophic canines have been shown to accurately predict the post mortem weight of the left ventricle.” The very significant difference between normalized LVM in normal subjects and HCM patients indicates that it may be a sensitive indicator of progression or regression of LVM during hyptertrophy of all etiologies - HCM, systemic hypertension, aortic stenosis, and so on. It has been demonstrated that interventricular septal thickness decreased from 19.6 mm to 15.2 mm in MRI studies of 32 hypertensive patients undergoing treatment with ramipril.‘* Echocardiography demonstrated a decrease from 18.8 mm to 14.6 mm in the same patient population. The decrease in interventricular septal thickness was on the order of 22%. Data in Tables 3 and 4 indicate that the normalized LVM can increase on the order of 100’70,producing a potentially more sensitive indicator of regression during treatment. Successful therapy of hypertension results in regression of LVM within 3 days to 6 wk.‘* A brief MRI cardiac imaging procedure used to monitor progression or regression of LVM would be beneficial in management of these patients, especially in patients with unattainable echocardiograms. The use of transverse imaging to reduce the examination time is an important consideration in routine monitoring of LVM. Subsequent research will explore the use of normalized LVM by MRI to .monitor regression during therapy of hypertensive patients.
4.
5.
6.
7.
8.
9.
10.
11.
12. Acknowledgments- Image processing was accomplished on a Sun Microsystems 4/24O(TAAC-1) computer (Sun Microsystems, Mountain View, CA) using MR Console 4.5 research software (General Electric Medical Systems, Milwaukee, WI). MR Console 4.5 facilitated selection of endocardial and epicardial walls using a mouseguided cursor in a user friendly environment.
13.
REFERENCES 14. Levy, D.; Anderson, K.M. ; Savage, D.D.; et al. Echocardiographically detected left ventricular hypertrophy: Prevalence and risk factors. Ann. Intern. Med. 108:713; 1988. Schiller, N.B.; Shah, P.M.; Crawford, M.; et al. Recommendations for quantitation of the left ventricle by twodimensional echocardiography. J. Am. Sot. Echo 2(S): 358-367; 1989. Reichek, N. Standardization in the measurement of left
15.
ventricular wall mass two-dimensional echocardiography. Hypertension Suppl 11: 11-30-11-32; 1987. Schiller, N.B. Considerations in the standardization of measurement of left ventricular myocardial mass by twodimensional echocardiography. Hypertension Supplll: 11-33-H-35; 1987. Reichek, N. Standardization in the measurement of left ventricular wall mass two-dimensional echocardiography. Hypertension Suppl 11: H-27-11-29; 1987. Sechtem, U.; Pflugfelder, P.W.; Gould, R.G.; Cassidy, M.M.; Higgins, C.B. Measurement of right and left ventricular volumes in healthy individuals with tine MR imaging. Radiology 163:697-702; 1987. Katz, J.; Milliken, M.C.; Stray-Gundersen, J.; Buja, L.M.; Parkey, R.W.; Mitchell, J.H.; Peshock, R.M. Estimation of human myocardial mass with MR imaging. Radiology 169:495-498; 1988. Fleagle, S.R.; Thedens, D.R.; Ehrhardt, J.C.; Scholz, T.D.; Skorton, D. J. Automated identification of left ventricular borders from spin-echo magnetic resonance images. Invest. Radiol. 26:295-303; 1991. Weiss, J.L.; Shapiro, E.P. ; Buchalter, M.B.; Beyar, R. Magnetic resonance imaging as a noninvasive standard for the quantitative evaluation of left ventricular mass, ischemia, and infarction. Ann. NY Acad. Sci. 601:95106; 1990. Caputo, G.R.; Tscholakoff, D.; Sechtem, U.; Higgins, C.B. Measurement of canine left ventricular mass by using MR imaging. AJR 148:33-38; 1987. Wang, J.Z.; Mezrich, R.S.; Scholz, P.; Als, A.; Douglas, F. MRI evaluation of left ventricular hypertrophy in a canine model of aortic stenosis. Invest. Radiol. 25:783788; 1990. Eichstaedt, H.; Danne, 0.; Langer, M.; Cordes, M.; Schubert, C.; Felix, R.; Schmutzler, H. Regression of left ventricular hypertrophy under ramipril treatment investigated by nuclear magnetic resonance imaging. J. Cardiovasc. Pharmacol. 13(3):S75-S80; 1989. Woythaler, J.N.; Singer, S.L.; Kwan, O.L.; et al. Accuracy of echocardiography versus electrocardiography in detecting left ventricular hypertrophy: Comparison with post-mortem mass measurements. J. Am. Coil. Cardiol. 2:305-311; 1983. Sahn, D.J.; DeMaria, A.; Kisslo, J.; Weyman, A. Recommendations regarding quantitation in M-mode echocardiography: Results of a survey of echocardiographic measurements. Circulation 58:1072-1083; 1978. Kitzman, D.W.; Scholz, D.G.; Hagen, P.T.; Ilstrup, D.M.; Edwards, W.D. Age-related changes in normal human hearts during the first 10 decades of life. Part 11 (Maturity): A quantitative anatomic study of 765 specimens from subjects 20 to 99 years old. Mayo Ctin. Proc. 63:137-146; 1988.