Echocardiographic Measurement of Left Ventricular Mass and Volume in Normotensive and Hypertensive Rats Giovanni de Simone, Donald G Wallerson, Massimo Volpe, and Richard B. Devereux
Although rats are commonly used to study left ventricular (LV) hypertrophy, measurement of LV mass and dimensions has required killing the rat. To determine the accuracy of echocardiography in rats, blinded crosssectional area (CSA) and LV mass measurements using either the cube function (LVM) or an elliptical model (LVMel) from high resolution M-mode echocardiograms were com pared to necropsy LV weight (0.28 to 1.5 g), in 41 normotensive (body weight 116 to 762 g) and 17 hy pertensive rats (350 to 560 g). Postmortem chamber volumes in 28 normal rats (0.02 to 0.19 mL) were also compared to echocardiographic volumes derived from the elliptical model. Correlation with LV weight was r = 0.87 for LVM, 0.90 for CSA and 0.93 for LVMel (all Ρ < .00001). Comparison of hypertensive and body-weight-matched normoten sive rats revealed the upper normal limit for both LVMel and CSA to have 89% sensitivity and
S
ince 1966, rats have been employed in more than 11,000 studies of arterial hypertension. Studies on rats have provided basic knowledge about the pathophysiology of arterial hyperten1
From the Department of Medicine and the Hypertension Center, The New York Hospital—Cornell Medical Center, New York City. This article is based on data presented at the American Society of Hypertension annual meeting, May 10 to 12, 1989. Supported in part by grant HL18323 from the National Heart, Lung and Blood Institute, Bethesda, Maryland. Dr. de Simone is the recipi ent of the 1st Lederle International Cardiovascular Award. Address correspondence and reprint requests to Dr. Richard B. Devereux, Division of Cardiology, Box 222, The New York Hospital —Cornell Medical Center, 525 East 68th Street, New York, NY, 10021.
100% specificity for detection of post mortem LV hypertrophy. Necropsy LV volumes were more closely related to systolic echocardiographic vol umes than to diastolic volumes (r = 0.78 ν 0.71, both Ρ < .00001), compatible with the effects of post mortem contracture. Stroke volume determined invasively in 5 Wistar rats by thermodilution was similar to that obtained using elliptical model echo volumes in 5 rats of the same body size (0.35 ± 0.05 ν 0.30 ± 0.06 mL/beat). Echocardiography can be used to evaluate LV structure and function in rats and to detect in vivo LV anatomic differences induced by hypertension. Am J Hypertens 1990;3:688-696
Left ventricular hypertrophy, arterial hypertension, cardiac function, cardiac output, Wistar rat. KEY WORDS:
sion, and have contributed particularly to knowledge of left ventricular (LV) hypertrophy in hypertension and other types of cardiac overload. " Nevertheless, until now, assessment of LV hypertrophy in rats has required killing the animal, making it impossible to study the in vivo geometry or function of the left ventricle under different conditions of overload, hypertrophy or neuro humoral stimulation. A method that has proved useful for determination of LV mass and other geometric and functional variables in humans, and also in large and small animals, is echocardiography. " Since the most commonly used animal in studies on arterial hyperten sion is the rat, the ability to study rat heart adaptation by echocardiography would facilitate a range of studies 2
4
5
12
Downloaded from http://ajh.oxfordjournals.org/ at Michigan State University on March 1, 2015
Necropsy Validation
concerning different aspects of the heart in hyperten sion, as well as in other types of overload. Accordingly, we studied the accuracy of ante mortem echocardio graphic determination of LV mass and chamber vol umes in rats in comparison to necropsy findings after the death of the animals. We also compared cardiac output measurements obtained by echocardiography and by invasive procedures. METHODS
]ivs
EDD
A
]PWT
Echocardiographic Method Echocardiograms were performed between 3 and 5 PM with the animals in left
....J..............
w.= 0 . 3 0 mm
1 cm
]ivs
EDD
Β
]PWT
FIGURE 1. Time motion strip-chart record of a 0.3 mm wire suspended in a standard tissue mimicking phantom; it was mea sured as 0.31 ± 0.01 mm using a computer-assisted digitizing system.
FIGURE 2. Μ mode echocardiographic tracings from two nor motensive rats. A: body weight = 425 g; B: body weight = 250 g. IVS = interventricular septum, EDD = end-diastolic dimension, PWT = posterior wall, ESD = end-systolic dimension.
Downloaded from http://ajh.oxfordjournals.org/ at Michigan State University on March 1, 2015
Animals Fifty-eight one to eight month old rats were studied, including 41 normotensive Wistar, three stroke-prone spontaneously hypertensive rats (SHRsp) and 14 renovascular hypertensive rats (nine 1-kidney 1-clip [lklc] and five 2-kidney 1-clip [2klc]). Blood pressure recording in hypertensive rats was carried out 1 day to 2 weeks before the echocardiographic study. Dif ferent degrees of hypertension were present, according to the caliber of the renal artery clip: 0.2 mm for three lklc, which formed together with SHRsp a group with severe hypertension (systolic blood pressure 170 to 215 mm Hg, mean 192 ± 9); 0.5 mm for six l k l c and all of the 2k lc, which formed a group with moderate hy pertension (systolic blood pressure was 147 to 172 mm Hg, mean 159 ± 5). Body weight was 116 to 763 g in normotensive and 350 to 560 g in hypertensive rats.
decubitus position, after light anesthesia with intramus cular ketamine (50 mg/kg). A Hewlett-Packard (Andover, MA) 770720 echocardiographic system was used, equipped with a 5 mHz, shallow focus, 21211A phased-array transducer, which was placed on the left hemithorax, which had been previously shaved. Tests performed using a standard tissue mimicking phantom had shown that this transducer can resolve 0.3 mm wires suspended in the phantom at 3, 2, 1 and 0.5 cm intervals; M-mode tracings performed at low gain showed a clear definition of the wire interfaces (Figure 1): the mean and standard deviation of 9 measurements performed on the image of a 0.3 mm wire using the same digitizer pen used for study tracings was extremely accurate (0.31 ± 0.01 mm). Short axis 2D views of the left ventricle at or just below the tip of the mitral valve leaflets were used to obtain targeted M-mode recordings on strip-chart paper at 100 mm/sec (Figure 2). Two-dimensional recordings
13
5,7
12
Wistar rats, by filling the chambers with isotonic solu tion at atmospheric pressure. Left Ventricular Weight and Internal Chamber Vol umes To determine echocardiographic LV mass, we used either the standard cube function formula or an elliptical model. Cube function LV mass was obtained by the formula: LVM (mg) = 1.04 X [(EDD + PWT + IVS) - EDD ] 3
3
where LVM is LV mass and 1.04 is the specific gravity of myocardium. LVM by the elliptical model (LVMel) was based on the assumption that, especially with a heart rate as high as in these animals (214 to 529 beats/min), LV geometry is best represented as a three-dimensional, prolate, ellip soidal shell where the external diameters of the shell were considered as minor and major axis. Based on measurements by Streeter and Hanna, the shell thick ness at the base and apex was assumed to be 5 5 % of the equatorial values. Echocardiographic EDD and wall thicknesses were used to represent the external minor axis. To derive the long axis measurement, an empiric echocardiographic method was used, based on ne cropsy measurements in 27 normotensive rats. A multi ple regression model allowed estimation of post mortem long axis measurement, using the square root of body weight in kg (BW), systolic posterior wall thickness (PWTs) and fractional shortening (FS = (EDD - ESD)/ (EDD) X 100) as independent variables (multiple r = 0.94). To estimate in vivo end-diastolic long axis length, a constant (2.04), representing the mean ratio between EDD and post mortem necropsy short axis dimension, was incorporated in the regression equation. Accord ingly, for each rat long axis (LA) was: 15
16
LA = 2.04 X (5.46 X VBW + 1.43 PWTs - 0.07 X FS + 8.31)
14
Necropsy Method After the echocardiographic study, rats were killed with an intraperitoneal pentobarbital overdose and the hearts were removed and trimmed of pericardium, fat and blood vessels. The atria and the right ventricular free wall were carefully removed and the left ventricle, including the interventricular septum, was dried and weighed on an analytical balance. After the described procedure, the post mortem LV internal long axis was measured in 27 Wistar rats using manual calipers, and left ventricular volume was assessed in 28
R-square = 0.88; SEE = 0.67 mm; Ρ < .00001 Thus, end-diastolic external volume (Ve) was: Ve = 3 . 1 4 / 6 X (EDD + IVS + PWT)
2
X (LA + PWT + IVS) and end-diastolic internal volume (Vi) was: Vi = 3 . 1 4 / 6 X EDD X [LA + 0.45 X (PWT + IVS)]. 2
From those formulae we derived LVMel, as: LVMel (mg) = 1.04 X (Ve - Vi) Furthermore, to test an index of myocardial mass in dependent of volumetric assumptions, we also calcu lated myocardial crosssectional area (CSA) : 17
CSA ( m m ) = 3.14 X [(EDD + PWT + IVS)/2] -3.14X(EDD/2) 2
2
2
Downloaded from http://ajh.oxfordjournals.org/ at Michigan State University on March 1, 2015
were not used for measurement because at the rapid heart rates each frame would encompass 2 0 % to 5 0 % of a cardiac cycle, effectively precluding "real time" imag ing. In accord with standards proposed for human stud ies, only M-mode echocardiograms with well defined continuous interfaces for septum and posterior wall were accepted for measurements (Figure 2), a criterion that resulted in exclusion of tracings from 4 of 62 ani mals (6.5%). As seen in two rats, extreme bradycardia to cardiac arrest can occur if the transducer pressure on the thorax is not extremely gentle; thus, during perform ance of the echocardiograms, particular attention was paid to avoid excessive pressure. One rat died after anesthesia. End-diastolic LV dimension (EDD) and interventricu lar septal and posterior wall thicknesses (IVS and PWT) were measured by the leading edge method at the time of maximum diastolic dimension, independently of the ECG, because the very high heart rate (over 500 beats/ minute in several cases) caused the R wave onset to occur during LV filling. Penn convention measure ments were not made because a previous study from our laboratory suggested no advantage to them over leading edge measurements in echocardiographic stud ies in small animals. End-systolic dimension (ESD) was measured at the time of the maximum anterior mo tion of the posterior wall (Figure 2). Interfaces were marked by two observers blinded to the knowledge of necropsy findings, weight and type of rat. Measure ments were done by a digitizer-pen connected with a computer, calculating the mean of measurements on three or more cardiac cycles. The interobserver variabil ity was assessed in 10 randomly chosen tracings by se lecting three cardiac cycles and having two investigators performing blinded measurements without marking the tracings, by using the digitizer pen. The coefficient of correlation and the standard error of estimate (SEE) was calculated for measurements of EDD, IVS and PWT; as indicated by Bland and Altman. Because the correla tion between two measurements does not assess the agreement between them, the "limits of agreement" were also calculated, as the 9 5 % confidence interval of the difference between the two observers' measure ments.
the fat/lean body mass ratio increasing with age and body weight.
EDVel = 3 . 1 4 / 6 X EDD X [ 2 X (EDD + PWT X 2) - 1.1 X PWT)]
Echocardiographic Left Ventricular Findings Echo cardiographic IVS ranged from 0.9 to 2.7 mm, PWT from 0.8 to 2.3 mm, EDD from 3.7 to 8.4 mm and ESD from 0.9 to 3.8 mm. Correlation between measurements by two observers was very close (r = 0.98, SEE = 0.07 mm, for IVS; r = 0.98, SEE = 0.06 mm, for PWT; and r = 0.997, SEE = 0.04 mm, for EDD; all P < .00001) and the limits of agreement very small (— 0.14 to 0.12 mm for I V S , - 0 . 1 3 to 0.14 mm for PWT a n d - 0 . 0 5 to 0.23 mm for EDD). Post mortem LV weight was closely correlated to echocardiographically measured IVS (r = 0.72; Ρ < .00001) and PWT (r = 0.84; Ρ < .00001), and less closely to EDD (r = 0.45; Ρ < .0004). Close correlations were also found between necropsy LV weight and de rived echocardiographic indices including LVM, LVMel and CSA. Necropsy LV weight was most closely related to LVMel (0.23 to 1.39 g; r = .93, SEE = 0.11 g; Figure 3), CSA (15 to 59 mm ; r = 0.90, SEE = 0.12 g, Figure 3) and cube function LVM (0.16 to 1.1 g; r = 0.87, SEE = 0.14 g) (all Ρ < .00001). As with the necropsy findings, close correlations ex isted between body weight and echocardiographic LV measurements in normotensive rats (r = 0.93, SEE = 0.11 g for LVMel; r = 0.90, SEE = 4.4 mm for CSA; r = 0.89, SEE = 0.1 g for LVM; all Ρ < .00001), whereas no correlation was found in hypertensive ani mals.
2
ESVel = 3 . 1 4 / 6 X E S D X [ 2 X (ESD + PWTs X 2) - 1.1 X PWTs] 2
where EDVel is end-diastolic volume, ESVel is end-sys tolic volume and PWTs is systolic posterior wall thick ness. Hemodynamic Method For comparison with echocar diographic estimates of stroke volume and cardiac out put, these variables were measured invasively by a pre viously published thermodilution technique in a separate group of 5 Wistar rats anesthetized with ketamine (60 mg/kg) and acepromazine (0.3 m g / k g ) . Si multaneous readings of cardiac output (mL/min) and stroke volume (mL/beat) were provided by the cardiac output computer (Cardiomax II, Columbus Instru ments, Columbus, Ohio). Each measurement was done in triplicate. 18
Statistical Analysis Data are expressed as mean ± one standard deviation. Echocardiographic and ana tomic findings were compared using linear regression analysis, also employed to study the relation of body weight to either the echocardiographic or anatomic vari ables. Student's t test was used to compare invasively determined cardiac output and stroke volume in 5 rats and echocardiographic determined cardiac output and stroke volume in a subgroup of normotensive rats matched for body weight, as well as to compare ana tomic and echocardiographic findings between hyper tensive rats and a subgroup of normotensive rats matched for age and body weight. Further comparison between moderately and severely hypertensive rats was done using one-way analysis of variance and Tukey's post hoc test. The null hypothesis was rejected at P < .05. RESULTS Necropsy Findings Necropsy LV weight was 0.28 to 1.26 g in normotensive rats and 0.84 to 1.5 g in hyper tensive rats. Post mortem LV volume in 28 normoten sive rats was 0.02 to 0.19 mL. Necropsy long axis in 27 normotensive rats was 9 to 14 mm. As expected, LV weight was closely correlated with body weight in nor motensive rats (r = 0.95; SEE = 0.08 g; Ρ < .00001), whereas no correlation was found in hypertensive rats (r = 0.15, P > .5) because of the superimposed effect of LV overload. Normalization of LV weight for body weight resulted in a strong decrease of the LV weight/ body weight ratio with increasing body weight (r = - 0 . 7 4 ; SEE = 0.19 g/kg; Ρ < .00001), possibly due to
2
2
Left Ventricular Volumes and Determination of Car diac Output Post mortem LV volume measured in 28 normotensive rats was related to echocardiographic EDD (3.7 to 8.4 mm; r = 0.63; SEE = 0.04 mL; P < .0003) and end-systolic dimension (0.9 to 3.8 mm; r = 0.70; SEE = 0.04 mL; Ρ < .00001). Correlations im proved when LV volumes were calculated from echocardiographic data. Echocardiographic volumes were significantly correlated to post mortem volume (0.02 to 0.19 mL). End-systolic echocardiographic vol umes (0.005 to 0.16 mL) were more closely related to post mortem LV volume (r = 0.78, SEE = 0.003 mL, Ρ < .0001) than were echocardiographic end-diastolic volumes (0.08 to 0.77 mL; r = 0.71, SEE = 0.01 mL, P
Although rats are commonly used to study left ventricular (LV) hypertrophy, measurement of LV mass and dimensions has required killing the rat. To determine the accuracy of echocardiography in rats, blinded crosssectional area (CSA) and LV mass measurements using either the cube function (LVM) or an elliptical model (LVMel) from high resolution M-mode echocardiograms were com pared to necropsy LV weight (0.28 to 1.5 g), in 41 normotensive (body weight 116 to 762 g) and 17 hy pertensive rats (350 to 560 g). Postmortem chamber volumes in 28 normal rats (0.02 to 0.19 mL) were also compared to echocardiographic volumes derived from the elliptical model. Correlation with LV weight was r = 0.87 for LVM, 0.90 for CSA and 0.93 for LVMel (all Ρ < .00001). Comparison of hypertensive and body-weight-matched normoten sive rats revealed the upper normal limit for both LVMel and CSA to have 89% sensitivity and
S
ince 1966, rats have been employed in more than 11,000 studies of arterial hypertension. Studies on rats have provided basic knowledge about the pathophysiology of arterial hyperten1
From the Department of Medicine and the Hypertension Center, The New York Hospital—Cornell Medical Center, New York City. This article is based on data presented at the American Society of Hypertension annual meeting, May 10 to 12, 1989. Supported in part by grant HL18323 from the National Heart, Lung and Blood Institute, Bethesda, Maryland. Dr. de Simone is the recipi ent of the 1st Lederle International Cardiovascular Award. Address correspondence and reprint requests to Dr. Richard B. Devereux, Division of Cardiology, Box 222, The New York Hospital —Cornell Medical Center, 525 East 68th Street, New York, NY, 10021.
100% specificity for detection of post mortem LV hypertrophy. Necropsy LV volumes were more closely related to systolic echocardiographic vol umes than to diastolic volumes (r = 0.78 ν 0.71, both Ρ < .00001), compatible with the effects of post mortem contracture. Stroke volume determined invasively in 5 Wistar rats by thermodilution was similar to that obtained using elliptical model echo volumes in 5 rats of the same body size (0.35 ± 0.05 ν 0.30 ± 0.06 mL/beat). Echocardiography can be used to evaluate LV structure and function in rats and to detect in vivo LV anatomic differences induced by hypertension. Am J Hypertens 1990;3:688-696
Left ventricular hypertrophy, arterial hypertension, cardiac function, cardiac output, Wistar rat. KEY WORDS:
sion, and have contributed particularly to knowledge of left ventricular (LV) hypertrophy in hypertension and other types of cardiac overload. " Nevertheless, until now, assessment of LV hypertrophy in rats has required killing the animal, making it impossible to study the in vivo geometry or function of the left ventricle under different conditions of overload, hypertrophy or neuro humoral stimulation. A method that has proved useful for determination of LV mass and other geometric and functional variables in humans, and also in large and small animals, is echocardiography. " Since the most commonly used animal in studies on arterial hyperten sion is the rat, the ability to study rat heart adaptation by echocardiography would facilitate a range of studies 2
4
5
12
Downloaded from http://ajh.oxfordjournals.org/ at Michigan State University on March 1, 2015
Necropsy Validation
concerning different aspects of the heart in hyperten sion, as well as in other types of overload. Accordingly, we studied the accuracy of ante mortem echocardio graphic determination of LV mass and chamber vol umes in rats in comparison to necropsy findings after the death of the animals. We also compared cardiac output measurements obtained by echocardiography and by invasive procedures. METHODS
]ivs
EDD
A
]PWT
Echocardiographic Method Echocardiograms were performed between 3 and 5 PM with the animals in left
....J..............
w.= 0 . 3 0 mm
1 cm
]ivs
EDD
Β
]PWT
FIGURE 1. Time motion strip-chart record of a 0.3 mm wire suspended in a standard tissue mimicking phantom; it was mea sured as 0.31 ± 0.01 mm using a computer-assisted digitizing system.
FIGURE 2. Μ mode echocardiographic tracings from two nor motensive rats. A: body weight = 425 g; B: body weight = 250 g. IVS = interventricular septum, EDD = end-diastolic dimension, PWT = posterior wall, ESD = end-systolic dimension.
Downloaded from http://ajh.oxfordjournals.org/ at Michigan State University on March 1, 2015
Animals Fifty-eight one to eight month old rats were studied, including 41 normotensive Wistar, three stroke-prone spontaneously hypertensive rats (SHRsp) and 14 renovascular hypertensive rats (nine 1-kidney 1-clip [lklc] and five 2-kidney 1-clip [2klc]). Blood pressure recording in hypertensive rats was carried out 1 day to 2 weeks before the echocardiographic study. Dif ferent degrees of hypertension were present, according to the caliber of the renal artery clip: 0.2 mm for three lklc, which formed together with SHRsp a group with severe hypertension (systolic blood pressure 170 to 215 mm Hg, mean 192 ± 9); 0.5 mm for six l k l c and all of the 2k lc, which formed a group with moderate hy pertension (systolic blood pressure was 147 to 172 mm Hg, mean 159 ± 5). Body weight was 116 to 763 g in normotensive and 350 to 560 g in hypertensive rats.
decubitus position, after light anesthesia with intramus cular ketamine (50 mg/kg). A Hewlett-Packard (Andover, MA) 770720 echocardiographic system was used, equipped with a 5 mHz, shallow focus, 21211A phased-array transducer, which was placed on the left hemithorax, which had been previously shaved. Tests performed using a standard tissue mimicking phantom had shown that this transducer can resolve 0.3 mm wires suspended in the phantom at 3, 2, 1 and 0.5 cm intervals; M-mode tracings performed at low gain showed a clear definition of the wire interfaces (Figure 1): the mean and standard deviation of 9 measurements performed on the image of a 0.3 mm wire using the same digitizer pen used for study tracings was extremely accurate (0.31 ± 0.01 mm). Short axis 2D views of the left ventricle at or just below the tip of the mitral valve leaflets were used to obtain targeted M-mode recordings on strip-chart paper at 100 mm/sec (Figure 2). Two-dimensional recordings
13
5,7
12
Wistar rats, by filling the chambers with isotonic solu tion at atmospheric pressure. Left Ventricular Weight and Internal Chamber Vol umes To determine echocardiographic LV mass, we used either the standard cube function formula or an elliptical model. Cube function LV mass was obtained by the formula: LVM (mg) = 1.04 X [(EDD + PWT + IVS) - EDD ] 3
3
where LVM is LV mass and 1.04 is the specific gravity of myocardium. LVM by the elliptical model (LVMel) was based on the assumption that, especially with a heart rate as high as in these animals (214 to 529 beats/min), LV geometry is best represented as a three-dimensional, prolate, ellip soidal shell where the external diameters of the shell were considered as minor and major axis. Based on measurements by Streeter and Hanna, the shell thick ness at the base and apex was assumed to be 5 5 % of the equatorial values. Echocardiographic EDD and wall thicknesses were used to represent the external minor axis. To derive the long axis measurement, an empiric echocardiographic method was used, based on ne cropsy measurements in 27 normotensive rats. A multi ple regression model allowed estimation of post mortem long axis measurement, using the square root of body weight in kg (BW), systolic posterior wall thickness (PWTs) and fractional shortening (FS = (EDD - ESD)/ (EDD) X 100) as independent variables (multiple r = 0.94). To estimate in vivo end-diastolic long axis length, a constant (2.04), representing the mean ratio between EDD and post mortem necropsy short axis dimension, was incorporated in the regression equation. Accord ingly, for each rat long axis (LA) was: 15
16
LA = 2.04 X (5.46 X VBW + 1.43 PWTs - 0.07 X FS + 8.31)
14
Necropsy Method After the echocardiographic study, rats were killed with an intraperitoneal pentobarbital overdose and the hearts were removed and trimmed of pericardium, fat and blood vessels. The atria and the right ventricular free wall were carefully removed and the left ventricle, including the interventricular septum, was dried and weighed on an analytical balance. After the described procedure, the post mortem LV internal long axis was measured in 27 Wistar rats using manual calipers, and left ventricular volume was assessed in 28
R-square = 0.88; SEE = 0.67 mm; Ρ < .00001 Thus, end-diastolic external volume (Ve) was: Ve = 3 . 1 4 / 6 X (EDD + IVS + PWT)
2
X (LA + PWT + IVS) and end-diastolic internal volume (Vi) was: Vi = 3 . 1 4 / 6 X EDD X [LA + 0.45 X (PWT + IVS)]. 2
From those formulae we derived LVMel, as: LVMel (mg) = 1.04 X (Ve - Vi) Furthermore, to test an index of myocardial mass in dependent of volumetric assumptions, we also calcu lated myocardial crosssectional area (CSA) : 17
CSA ( m m ) = 3.14 X [(EDD + PWT + IVS)/2] -3.14X(EDD/2) 2
2
2
Downloaded from http://ajh.oxfordjournals.org/ at Michigan State University on March 1, 2015
were not used for measurement because at the rapid heart rates each frame would encompass 2 0 % to 5 0 % of a cardiac cycle, effectively precluding "real time" imag ing. In accord with standards proposed for human stud ies, only M-mode echocardiograms with well defined continuous interfaces for septum and posterior wall were accepted for measurements (Figure 2), a criterion that resulted in exclusion of tracings from 4 of 62 ani mals (6.5%). As seen in two rats, extreme bradycardia to cardiac arrest can occur if the transducer pressure on the thorax is not extremely gentle; thus, during perform ance of the echocardiograms, particular attention was paid to avoid excessive pressure. One rat died after anesthesia. End-diastolic LV dimension (EDD) and interventricu lar septal and posterior wall thicknesses (IVS and PWT) were measured by the leading edge method at the time of maximum diastolic dimension, independently of the ECG, because the very high heart rate (over 500 beats/ minute in several cases) caused the R wave onset to occur during LV filling. Penn convention measure ments were not made because a previous study from our laboratory suggested no advantage to them over leading edge measurements in echocardiographic stud ies in small animals. End-systolic dimension (ESD) was measured at the time of the maximum anterior mo tion of the posterior wall (Figure 2). Interfaces were marked by two observers blinded to the knowledge of necropsy findings, weight and type of rat. Measure ments were done by a digitizer-pen connected with a computer, calculating the mean of measurements on three or more cardiac cycles. The interobserver variabil ity was assessed in 10 randomly chosen tracings by se lecting three cardiac cycles and having two investigators performing blinded measurements without marking the tracings, by using the digitizer pen. The coefficient of correlation and the standard error of estimate (SEE) was calculated for measurements of EDD, IVS and PWT; as indicated by Bland and Altman. Because the correla tion between two measurements does not assess the agreement between them, the "limits of agreement" were also calculated, as the 9 5 % confidence interval of the difference between the two observers' measure ments.
the fat/lean body mass ratio increasing with age and body weight.
EDVel = 3 . 1 4 / 6 X EDD X [ 2 X (EDD + PWT X 2) - 1.1 X PWT)]
Echocardiographic Left Ventricular Findings Echo cardiographic IVS ranged from 0.9 to 2.7 mm, PWT from 0.8 to 2.3 mm, EDD from 3.7 to 8.4 mm and ESD from 0.9 to 3.8 mm. Correlation between measurements by two observers was very close (r = 0.98, SEE = 0.07 mm, for IVS; r = 0.98, SEE = 0.06 mm, for PWT; and r = 0.997, SEE = 0.04 mm, for EDD; all P < .00001) and the limits of agreement very small (— 0.14 to 0.12 mm for I V S , - 0 . 1 3 to 0.14 mm for PWT a n d - 0 . 0 5 to 0.23 mm for EDD). Post mortem LV weight was closely correlated to echocardiographically measured IVS (r = 0.72; Ρ < .00001) and PWT (r = 0.84; Ρ < .00001), and less closely to EDD (r = 0.45; Ρ < .0004). Close correlations were also found between necropsy LV weight and de rived echocardiographic indices including LVM, LVMel and CSA. Necropsy LV weight was most closely related to LVMel (0.23 to 1.39 g; r = .93, SEE = 0.11 g; Figure 3), CSA (15 to 59 mm ; r = 0.90, SEE = 0.12 g, Figure 3) and cube function LVM (0.16 to 1.1 g; r = 0.87, SEE = 0.14 g) (all Ρ < .00001). As with the necropsy findings, close correlations ex isted between body weight and echocardiographic LV measurements in normotensive rats (r = 0.93, SEE = 0.11 g for LVMel; r = 0.90, SEE = 4.4 mm for CSA; r = 0.89, SEE = 0.1 g for LVM; all Ρ < .00001), whereas no correlation was found in hypertensive ani mals.
2
ESVel = 3 . 1 4 / 6 X E S D X [ 2 X (ESD + PWTs X 2) - 1.1 X PWTs] 2
where EDVel is end-diastolic volume, ESVel is end-sys tolic volume and PWTs is systolic posterior wall thick ness. Hemodynamic Method For comparison with echocar diographic estimates of stroke volume and cardiac out put, these variables were measured invasively by a pre viously published thermodilution technique in a separate group of 5 Wistar rats anesthetized with ketamine (60 mg/kg) and acepromazine (0.3 m g / k g ) . Si multaneous readings of cardiac output (mL/min) and stroke volume (mL/beat) were provided by the cardiac output computer (Cardiomax II, Columbus Instru ments, Columbus, Ohio). Each measurement was done in triplicate. 18
Statistical Analysis Data are expressed as mean ± one standard deviation. Echocardiographic and ana tomic findings were compared using linear regression analysis, also employed to study the relation of body weight to either the echocardiographic or anatomic vari ables. Student's t test was used to compare invasively determined cardiac output and stroke volume in 5 rats and echocardiographic determined cardiac output and stroke volume in a subgroup of normotensive rats matched for body weight, as well as to compare ana tomic and echocardiographic findings between hyper tensive rats and a subgroup of normotensive rats matched for age and body weight. Further comparison between moderately and severely hypertensive rats was done using one-way analysis of variance and Tukey's post hoc test. The null hypothesis was rejected at P < .05. RESULTS Necropsy Findings Necropsy LV weight was 0.28 to 1.26 g in normotensive rats and 0.84 to 1.5 g in hyper tensive rats. Post mortem LV volume in 28 normoten sive rats was 0.02 to 0.19 mL. Necropsy long axis in 27 normotensive rats was 9 to 14 mm. As expected, LV weight was closely correlated with body weight in nor motensive rats (r = 0.95; SEE = 0.08 g; Ρ < .00001), whereas no correlation was found in hypertensive rats (r = 0.15, P > .5) because of the superimposed effect of LV overload. Normalization of LV weight for body weight resulted in a strong decrease of the LV weight/ body weight ratio with increasing body weight (r = - 0 . 7 4 ; SEE = 0.19 g/kg; Ρ < .00001), possibly due to
2
2
Left Ventricular Volumes and Determination of Car diac Output Post mortem LV volume measured in 28 normotensive rats was related to echocardiographic EDD (3.7 to 8.4 mm; r = 0.63; SEE = 0.04 mL; P < .0003) and end-systolic dimension (0.9 to 3.8 mm; r = 0.70; SEE = 0.04 mL; Ρ < .00001). Correlations im proved when LV volumes were calculated from echocardiographic data. Echocardiographic volumes were significantly correlated to post mortem volume (0.02 to 0.19 mL). End-systolic echocardiographic vol umes (0.005 to 0.16 mL) were more closely related to post mortem LV volume (r = 0.78, SEE = 0.003 mL, Ρ < .0001) than were echocardiographic end-diastolic volumes (0.08 to 0.77 mL; r = 0.71, SEE = 0.01 mL, P
