Plasma Angiotensin II, Predisposition to Hypertension, and Left Ventricular Size in Healthy Young Adults
Background We studied the correlates of left ventricular mass (LVM) in 84 healthy young adults aged 16 to 24 years from the general population. Subjects were selected according to predisposition to hypertension into four groups with either high or low personal blood pressures and either high or low parental blood pressures.
Methods and Results LVM was measured by echocardiography, and measurements of blood pressure, heart rate, body dimensions, and plasma concentrations of components of the renin-angiotensin system were made under resting conditions. LVM was similar in individuals predisposed to hypertension (high personal and parental blood pressures) and those with contrasting predisposition (low personal and parental pressures). Regression analysis of the combined groups showed that LVM correlated closely with body size, particularly lean body mass (r=.69, P<.0001) and systolic (r=.35, P<.0001) but not diastolic blood pressure. Plasma angiotensin II (r=.39, P<.0001), renin (r=.302, P<.01), and angiotensin-converting enzyme (r=.22, P<.05) showed significant correlation with LVM. Multiple regression analysis revealed that plasma angiotensin II was the most important component of the renin-angiotensin system and that its effect was independent of systolic blood pressure and body size.
Conclusions These findings provide evidence in humans that angiotensin II exerts a direct effect on myocardial size. This association may have important implications for the complications and treatment of left ventricular hypertrophy.
Echocardiographic studies have revealed LVH to be a common and important cardiovascular risk factor.1 This is partly because LVH is more frequent in older people and in individuals with hypertension, obesity, myocardial infarction, and valve disease.2 However, LVH also appears to have an independent effect on cardiovascular outcomes.1 Therefore, an improved understanding of the origins and pathophysiology of LVH may have important implications for the prevention and treatment of heart disease.
Values for LVM aggregate in families in a pattern that suggests a high degree of genetic determination.3 4 Biometric studies suggest that the genetic component of variation in LVM is partly the result of genes that influence body size or blood pressure and thereby LVM. This hypothesis is supported by observations that normotensive young adults with a family history of hypertension have increased LVM.5 6 7 8 This relationship also may involve a shared genetic effect on body size.9 However, after correction for known causes of LVH, there is a component of the genetic variation of heart size that remains unexplained. It has been suggested that certain genes have a specific effect on LVM.3 4 The underlying physiology in humans is not well understood but may involve the renin-angiotensin system.
Both angiotensin10 and aldosterone11 appear to affect cardiac size independent of blood pressure. The greater reversal of LVH for a given fall in blood pressure with ACE inhibitors compared with other antihypertensive agents12 is also consistent with a pressure-independent effect mediated by the renin-angiotensin system. Furthermore, the DD genotype for the ACE gene, which is associated with increased ACE levels,13 has been reported to be more prevalent in subjects with ECG evidence of LVH.14 However, the same is not true when LVM is measured by echocardiography.15
In this study we examined the possibility that the activity of the renin-angiotensin system or familial patterns of blood pressure could explain variation in left ventricular size in healthy young adults.
Ascertainment and Sampling Design
The four corners approach was used to select young adults from the general population with contrasting genetic predisposition to high blood pressure.16 All participants were white and drawn from an area served by two group general practices based at the Ladywell Medical Centre in the west of Edinburgh, Scotland.17 Parental blood pressures were recorded during the screening phase of the Medical Research Council Trial of Treatment of Mild Hypertension in 1978 to 1979, during which 76% of adults aged 35 to 64 years registered at the medical center took part. The screened adults included 1809 pairs of husbands and wives. The study population was formed by 1473 couples still registered with the practices in 1986. Blood pressure was measured by research nurses in 864 young adults aged 16 to 24 years from 603 families. Offspring comprised 74% of all children in this age group related to these parents by blood.
Blood pressure scores were calculated for both parents and offspring (parents on treatment were given a score equivalent to the upper 5% of the distribution) with the arithmetic average of systolic and diastolic pressures and adjusting for age and sex. A scatter diagram was then constructed with combined parental blood pressure scores on one axis and offspring scores on the other.16 Four groups of approximately 50 offspring were identified from each corner of the scatter diagram. The cutoffs for each of the corners corresponded to the lower or upper 30% of the distributions of combined parental or offspring scores.
Among the four groups, the greatest contrast in genetic predisposition existed between offspring with high blood pressure who came from families in which both parents had high blood pressure and offspring with low blood pressure who came from families in which both parents had low blood pressure. Comparison of these groups was considered the most powerful test within this study design of the association between candidate phenotypes and predisposition to high blood pressure.
Clinical and Hormonal Measurements
Informed consent was obtained from all participants, and all procedures were carried out according to the guidelines of the Western Infirmary Ethics Committee. Two hundred one young adults attended the medical center for measurements of blood pressure, weight, and height. Blood pressure was measured twice after 10 minutes of recumbent rest with the Hawksley random zero sphygmomanometer by research nurses who were unaware as to which offspring group individuals belonged. Subjects were allowed to rest supine for 25 minutes with a butterfly needle in situ before blood was taken for estimations of angiotensinogen, PRA, ACE, angiotensin II, and aldosterone.17
Left Ventricular Mass and Lean Body Mass
On a subsequent admission to the Western Infirmary, Glasgow, 100 of the original 201 subjects underwent echocardiographic examinations, of which 84 were deemed technically satisfactory. Echocardiography was performed as described previously with the use of a Hewlett-Packard Ultrasound Unit with a 2.5-MHz transducer.18 M-mode recordings were made according to the convention of the American Society of Echocardiography, with two-dimensional images to direct the M-mode sweep. These data were used to assess left ventricular posterior wall thickness, interventricular septal thickness, and left ventricular diastolic diameter, measured distal to the tips of the mitral valve leaflets at the peak of the R wave on the ECG. Measurements were made over three consecutive cardiac cycles, and mean values were calculated. LVM was calculated by the cube formula with the Penn convention and corrected in some analyses for body surface area to yield the LVMI. The reproducibility of echocardiographic measurements in a single subject was calculated by measuring LVM on 10 occasions in the same healthy volunteer. The coefficient of variation was 2.9%.
Participants had measurements of skinfold thickness in the triceps, biceps, and subscapular and suprailiac regions. LBM (or fat-free mass) was calculated from the skinfold measurements according to validated algorithms.19
Because the distribution of most variables was skewed to the right, data are summarized as medians and IQR. Differences between the groups were analyzed by nonparametric Kruskal-Wallis one-way ANOVA. Before parametric tests, variables were normalized by logarithmic transformation. In the case of angiotensinogen, values were standardized both for sex and for use of the oral contraceptive pill. Tests of linear association between variables were made with multiple regression techniques.20 The effects of sex were assessed by including a dummy variable (0, female; 1, male) in the linear regression analyses. Standardized residuals derived from linear regression were used to determine independent relationships between particular variables.
The data relate to 84 offspring in whom echocardiography was technically successful. The basic characteristics of these subjects are shown in Table 1⇓. As a result of selection, there were significant differences in offspring and parental blood pressures. These differences are reproducible and were observed each time we measured blood pressure in this population.21 The combined parental z scores indicate that on average, parental pressures in the low and high families were 1 SD below and above the mean, respectively. There were no significant differences in the male:female ratio between the groups. Offspring with high blood pressures were heavier than offspring with low blood pressures irrespective of whether parental blood pressures were high or low (Table 1⇓). These values are consistent with the previously reported characteristics of the entire 201 offspring.17
There were no significant differences in either the raw LVM or the LVMI between the four offspring groups (Table 1⇑). The 95% confidence intervals for differences in means between the two groups with the greatest contrast in predisposition encompassed zero for both LVM (−33.9 to 20.0 g) and LVMI (−14.1 to 10.9 g/m2). The left ventricular diastolic diameter and thickness of the posterior wall and septum also were similar in all four groups (data not shown). No significant difference in LVM was observed between the 43 offspring of parents with high blood pressure scores (LVM, 145 g; IQR, 52) and the 41 offspring of parents with low blood pressure scores (LVM, 156 g; IQR, 57), nor was there a significant difference in LVM in 43 offspring with high (LVM, 158 g; IQR, 64) versus 41 offspring with low blood pressure scores (LVM, 145 g; IQR, 52).
Because selection according to predisposition to hypertension did not appear to influence LVM, we combined the four groups to increase statistical power in the regression studies to identify other relevant phenotypic correlates of LVM. These analyses were made with and without inclusion of parental and personal blood pressure score grouping in the regression analysis (see below).
Left Ventricular Mass Correlates Closely With Body Size
The distribution of LVM (Fig 1⇓) for the combined groups exhibited a skew toward upper values, consistent with previous studies of healthy population samples.22 We found a positive association between LVM and body size. The simple correlation coefficients relating anthropometric indices and LVM were in descending order: LBM, 0.691 (P<.0001); body surface area, 0.610 (P<.0001); height, 0.535 (P<.0001); weight, 0.519 (P<.0001); and body mass index, 0.198 (P<.05). There was a high degree of intercorrelation between these variables (data not shown). Stepwise multiple regression analysis of LVM in relation to the different ways of describing body habitus revealed that after inclusion of LBM, no other index of body size contributed significantly to the variation in LVM.
Men had greater LVM than women (LVM men: 170 g, IQR, 52.7; LVM women: 123 g, IQR, 30; P<.0001), and LVMI was also greater (P<.0001) in men (92.5 g/m2; IQR, 26.7) than women (73.2 g/m2; IQR, 23.1). A regression analysis including sex indicated that sex differences in LVM could be explained by differences in LBM between men and women. When individual LVM was expressed per kilogram of LBM, the differences between men (2.97 g/kg; IQR, 0.76) and women (2.82 g/kg; IQR, 0.93) were no longer significant (P=.30).
LBM correlated with left ventricular diastolic internal diameter (r=.64, P<.0001) and the thickness of the left ventricular posterior wall (r=.35, P<.003) and the interventricular septum (r=.32, P<.002).
Relationship of Left Ventricular Mass and Blood Pressure
LVM was correlated with pulse pressure and systolic pressure (Table 2⇓). The simple correlation coefficient was slightly greater between LVM and pulse pressure. However, pulse pressure is derived mathematically from systolic pressure and therefore the two are intercorrelated, making it difficult to determine their relative biological importance. LVM did not correlate with either diastolic pressure or heart rate (Table 2⇓). There also were positive intercorrelations between both diastolic and systolic pressures and between heart rate and systolic pressure. In multiple regression analysis, pulse pressure or systolic pressure was the only significant hemodynamic variable associated with LVM. Pulse and systolic pressures were higher in men and heart rates lower in men than women, but a regression analysis taking sex into account indicated that sex differences did not explain entirely the correlation between pulse pressure and LVM.
The left ventricular diastolic internal diameter correlated both with pulse pressure (r=.26, P=.01) and with systolic pressure (r=.29, P<.005). The thickness of the left ventricular posterior wall correlated significantly with pulse pressure (r=.29, P=.007) but not systolic pressure (r=.15). Neither pulse pressure (r=.16) nor systolic pressure (r=.15) correlated with the interventricular septal dimension.
Left Ventricular Mass and the Renin-Angiotensin System
LVM showed significant positive correlations with plasma ACE, PRA, and plasma angiotensin II in increasing order of strength of association (Table 3⇓). No significant relationship was seen between LVM and either plasma angiotensinogen or aldosterone. PRA and to a lesser extent plasma angiotensinogen were significantly correlated with plasma angiotensin II. Plasma ACE correlated with PRA but not with plasma angiotensin. Plasma renin, ACE, and angiotensin II were higher in men than women. Multiple regression analysis showed that plasma angiotensin II was the only significant independent variable when all measured components of the renin-angiotensin system were considered in either simultaneous or stepwise analysis. This was also true when sex was included in the analysis, indicating that the relationship with angiotensin II was independent of sex.
Plasma angiotensin II was significantly correlated with the left ventricular diastolic internal diameter (r=.36, P<.001) and the thickness of the left ventricular posterior wall (r=.31, P<.003) but not the interventricular septum (r=.13).
Multiple Regression Analysis
Although LVM correlated significantly with body size, blood pressure, and angiotensin II, it is possible that these effects may not be independent. For example, angiotensin may influence LVM through blood pressure. Therefore, we undertook a multiple regression analysis that included the five independent variables: LBM, pulse pressure, systolic blood pressure, plasma angiotensin II, and sex.
In an initial comparison we found a positive univariate correlation (r=.22, P<.03) between LBM and plasma angiotensin II. In addition, pulse pressure correlated with angiotensin II (r=.20, P<.04) and LBM (r=.19, P<.05). However, systolic pressure did not correlate with either angiotensin II (r=.08) or LBM (r=.15).
When all five variables were entered simultaneously, only LBM (P<.0001) and plasma angiotensin II (P<.007) showed a significant independent association with LVM. In a stepwise multiple regression analysis we found significant independent correlations of LVM with LBM (partial regression coefficient: B=2.55 [SE=0.41], P<.0001), plasma angiotensin II (B=1.87 [SE=0.55], P<.001), and systolic blood pressure (B=0.83 [SE=0.35], P<.02). The standardized partial regression coefficients (β) indicate that the relative importance of these variables in descending order is LBM (β=0.53), plasma angiotensin II (β=0.29), and systolic blood pressure (β=0.19). Together these three independent variables accounted for 57% of the variation in LVM.
To depict the independent nature of the association between plasma angiotensin II and LVM, we generated a partial regression plot shown in Fig 2⇓. In effect, this plots LVM against plasma angiotensin II after both have been adjusted (expressed as standardized residuals) for LBM and systolic pressure. This revealed a significant positive correlation (partial correlation coefficient=.35, P=.002) between plasma angiotensin II and LVM independent of LBM and systolic pressure.
Finally, to determine any possible influence of predisposition to hypertension, parental and personal blood pressure score groupings were entered as dummy variables (low, 0; high, 1) into the multiple regression analysis. Neither made a significant contribution and did not alter substantially the independent effects of LBM, systolic blood pressure, and plasma angiotensin II.
This study demonstrates for the first time a significant correlation between plasma angiotensin II and LVM in healthy young adults. This relationship is independent of both body size and blood pressure and therefore consistent with a direct effect of angiotensin on cardiac tissue growth in normal humans. The independent correlation between plasma angiotensin II and the thickness of the posterior left ventricular wall also supports this hypothesis. Posterior wall thickness also has been shown to correlate with plasma angiotensin II in patients with essential hypertension, and this relationship was also independent of blood pressure and body size.23
A direct trophic action of angiotensin II has been suggested by the augmentation of cardiac muscle cell growth in vitro by angiotensin II.24 The in vivo effects of angiotensin II are more difficult to describe because of changes in cardiac afterload that influence heart size. However, angiotensin II infusion has been shown to increase myocardial protein synthesis and the weight of heterotopically transplanted hearts in which the left ventricle is largely isolated from changes in systemic hemodynamics.10 We confirm an association between systolic blood pressure and LVM.2 7 25 26 However, the relationship between plasma angiotensin II and LVM in our study was not confounded by systolic blood pressure because there was no direct correlation between these variables and they were independent in the multiple regression analysis.
Plasma ACE and PRA also correlated with LVM. Interestingly, the strength of association, as reflected by the magnitude of the correlation coefficient, increased step by step down the renin-angiotensin cascade to angiotensin II but was not observed for aldosterone. This implies that for LVM, angiotensin II represents the biologically significant component of the renin-angiotensin cascade. Although angiotensinogen was not associated directly with LVM, it correlated with plasma angiotensin II. PRA correlated with both LVM and plasma angiotensin II. Both findings are consistent with a cascade effect in which angiotensinogen and PRA determine angiotensin II and thereby LVM. Previous studies in older hypertensive subjects with LVH found no relationship between PRA and LVM,27 and it may be that in pathological situations other factors such as high blood pressure or the effects of previous antihypertensive treatment might alter any underlying physiological relationship.
In contrast, plasma ACE correlated with LVM but not with plasma angiotensin II, suggesting that ACE affects LVM through a separate mechanism. For example, increased ACE might lower myocardial bradykinin levels, thereby contributing to myocardial growth.28 An involvement of ACE is also suggested by the association between ECG evidence of LVH in normotensive men and the ACE DD genotype,14 which is known to increase plasma ACE levels.13 However, other echocardiographic studies have not linked the ACE DD genotype with increased LVM.15 The possibility that ACE inhibitors may cause an additional reduction in heart size beyond that expected from antihypertensive effects is also relevant.12
The renin-angiotensin system is intimately involved in sodium metabolism, and it has been suggested that high sodium intake may contribute to cardiac hypertrophy.29 In fact, we observed a positive correlation (data not shown) between urinary sodium excretion and LVM, but this effect was accounted for by differences in body size, with larger subjects excreting more sodium per day on average. A primary role for sodium intake would not explain the association between high angiotensin II and increased LVM since high sodium intake would be expected to suppress plasma angiotensin II. Nevertheless, it is possible that angiotensin II, through a sodium-retaining action, may increase volume load and therefore left ventricular chamber size.
LBM showed the strongest relationship with LVM, correlating with both the internal diastolic dimension and the wall thickness of the left ventricle and also accounting for the difference in LVM between the sexes. This was also found in previous comparisons of anthropometric variables,30 but the explanation of the predominance of LBM is not clear. LBM is composed largely of skeletal muscle and bone and related to overall body size. The skeletal muscle component may explain the closer correlation of LBM and LVM. Blood flow to skeletal muscle is considerably greater than that to fat so that variation in muscle bulk, which depends on physical activity, has a potentially greater effect on cardiac output. Exercise training and fitness are known to exert important effects on heart size.31 Therefore, LBM is likely to reflect the influence of both physical activity and body size on LVM.
Of the hemodynamic variables measured, only systolic blood pressure and pulse pressure correlated with LVM. The magnitude of these correlations was similar; however, pulse pressure (unlike systolic pressure) also correlated significantly with plasma angiotensin II and LBM. This probably explains why systolic pressure but not pulse pressure emerged as a significant independent correlate of LVM in the multiple regression analysis. Interestingly, systolic pressure correlated with left ventricular capacity, as indicated by the internal diastolic dimension but not with posterior wall or interventricular septal thickness. In the physiological range, it is possible that increased stroke volume may cause higher systolic pressure rather than the higher pressures eliciting a hypertrophic response characteristic of pressure overload. Both stroke volume and arterial stiffness, which affect pulse and systolic pressures, have been shown to be independent predictors of LVM in previous human studies.32 33
We did not observe a significant relationship between LVM and the genetic predisposition to hypertension. This may seem at odds with the finding that systolic pressure correlates with LVM. However, our categorization of offspring depends on blood pressure scores derived from both systolic and diastolic pressures of both parents and children. These variables may strengthen the assessment of genetic predisposition but dilute the specific influence of personal systolic pressure. Our findings contrast with previous studies in which an increase in LVM was seen in normotensive offspring with a family history of hypertension. The reason for this discrepancy is not obvious. We believe that we had adequate statistical power to detect reported increases in LVM because the number of our subjects with contrasting high or low parental pressure were greater5 7 8 or similar6 to previous reports. In these studies, the magnitude of difference between individuals with a positive parental history and those with a negative parental history of hypertension has been equivalent to approximately 1 SD of LVM. Based on grouping according to parental pressures, we have power of 80% to detect a difference of 1 SD, with 1% level of statistical significance. Furthermore, our sampling from opposite ends of the scale of familial resemblance based on measured blood pressures of both generations offers greater contrast in predisposition than selection based on the presence or absence of a reported family history of hypertension.21
We have confirmed the importance of body size and systolic blood pressure in the determination of LVM in healthy young adults and have discovered a significant relationship between plasma angiotensin II and LVM that is independent of blood pressure and body size. This relationship holds both in men and in women. The apparent involvement of several steps in the renin-angiotensin cascade provides impetus for the further study of a number of possible genetic candidates including the renin and ACE genes that might explain the familial aggregation of LVM.
Selected Abbreviations and Acronyms
|LBM||=||lean body mass|
|LVH||=||left ventricular hypertrophy|
|LVM||=||left ventricular mass|
|LVMI||=||left ventricular mass index|
|PRA||=||plasma renin activity|
These studies were funded by the Scottish Home and Health Department and the British Heart Foundation. Prof Harrap was supported during this work as a Neil Hamilton Travelling Fellow of the National Health and Medical Research Council of Australia. The authors wish to acknowledge the important cooperation of Dr David W. Holton and Dr Hugh E. Edwards from the Ladywell Medical Centre, Dr B. Leckie for estimation of plasma renin, and Prof Pierre Corvol for arranging analysis of angiotensinogen and the angiotensin-converting enzyme.
- Received August 29, 1995.
- Accepted October 23, 1995.
- Copyright © 1996 by American Heart Association
Levy D, Anderson KM, Savage DD, Kannel WB, Christiansen JC, Castelli WP. Echocardiographically detected left ventricular hypertrophy: prevalence and risk factors: the Framingham Heart Study. Ann Intern Med. 1988;108:7-13.
Post WS, Larson MG, Myers RH, Galderisi M, Levy D. Heritability of left ventricular mass. Circulation. 1994;90(suppl I):I-283. Abstract.
Nielsen JR, Oxhöj H. Echocardiographic variables in progeny of hypertensive and normotensive parents. Acta Med Scand. 1985;693:61-64.
Galderisi M, Celentano A, Tammaro P, Mureddu GF, Garafalo M, de Simone G, de Divitiis O. Ambulatory blood pressure monitoring in offspring of hypertensive patients: relation to left ventricular structure and function. Am J Hypertens. 1993;6:114-120.
Ravogli A, Trazzi S, Villani A, Mutti E, Cuspidi C, Sampieri L, De Ambroggi L, Parati G, Zanchetti A, Mancia G. Early 24-hour blood pressure elevation in normotensive subjects with parental hypertension. Hypertension. 1990;16:491-497.
Suurküla MB, Wikstrand J, Berglund G, Sivertsson R. Body weight is more important than family history of hypertension for left ventricular function. Hypertension. 1991;17:661-668.
Geenen DL, Malhotra A, Scheuer J. Angiotensin II increases cardiac protein synthesis in adult rat heart. Am J Physiol. 1993;265(Heart Circ Physiol 34):H238-H243.
Brilla CG, Pick R, Tan LB, Janicki JS, Weber KT. Remodeling of the rat right and left ventricles in experimental hypertension. Circ Res. 1990;67:1355-1364.
Rigat B, Hubert C, Alhenc-Gelas F, Cambien F, Corvol P, Soubrier F. An insertion/deletion polymorphism in the angiotensin I-converting enzyme gene accounting for half the variance of serum enzyme levels. J Clin Invest. 1990;86:1343-1346.
Kupari M, Perola M, Koskinen P, Virolainen J, Karhunen PJ. Left ventricular size, mass, and function in relation to angiotensin-converting enzyme gene polymorphism. Am J Physiol. 1994;267(Heart Circ Physiol 36):H1107-H1111.
Watt GCM. Methods of identifying individuals with contrasting predisposition to high blood pressure. In: Hofman A, Grobbee DE, Schalekamp MADH, eds. Early Pathogenesis of Primary Hypertension. New York, NY: Elsevier Science Publishers BV; 1987:69-79.
Watt GCM, Harrap SB, Foy CJW, Holton DW, Edwards HE, Davidson HR, Connor JM, Lever AF, Fraser R. Abnormalities of glucocorticoid metabolism and the renin-angiotensin system: a four corners approach to the identification of genetic determinants of blood pressure. J Hypertens. 1992;10:473-482.
SPSS Base System User’s Guide. Chicago, Ill: MJ Norusis/SPSS Inc; 1990:244-301.
Schmieder RE, Messerli FH, Garavaglia GE, Nunez B, MacPhee AA, Re RN. Does the renin-angiotensin-aldosterone system modify cardiac structure and function in essential hypertension? Am J Med. 1988;84(suppl 3A):136-139.
Baker KM, Aceto JF. Angiotensin II stimulation of protein synthesis and cell growth in chick cells. Am J Physiol. 1990;259(Heart Circ Physiol 28):H610-H618.
Schieken RM, Clarke WR, Lauer RM. Left ventricular hypertrophy in children with blood pressures in the upper quintile of the distribution. Hypertension. 1981;3:669-675.
Rostrup M, Smith G, Bjørnstad H, Westheim A, Stokland O, Eide I. Left ventricular mass and cardiovascular reactivity in young men. Hypertension. 1994;23(suppl I):I-168-I-171.
Devereux RB, Savage DD, Drayer JIM, Laragh JH. Left ventricular hypertrophy and function in high, normal, and low-renin forms of essential hypertension. Hypertension. 1982;4:524-531.
Linz W, Weimer G, Schölkens BA. Bradykinin prevents left ventricular hypertrophy in rats. J Hypertens. 1993;11(suppl 5):S96-S97.
Schmeider RE, Messerli FH, Garavaglia GE, Nunez BD. Dietary salt intake: a determinant of cardiac involvement in essential hypertension. Circulation. 1988;78:951-956.
Ganau A, Devereux RB, Pickering TG, Roman MJ, Schnall PL, Santucci S, Spitzer MC, Laragh JH. Relation of left ventricular hemodynamic load and contractile performance to left ventricular mass in hypertension. Circulation. 1990;81:25-36.