Genetic Regulation of Hemodynamic Variables During Dynamic Exercise
The MCV Twin Study
Background Both resting and exercise levels of blood pressure in individuals have been used as predictors of adult hypertension. One possible mechanism underlying the relation between childhood resting and exercise blood pressure and future blood pressure is a set of genes expressed in childhood that persists to regulate adult blood pressure.
Methods and Results To investigate the genetic relation of blood pressure and heart rate during both rest and exercise, we asked: (1) Are the genes that regulate resting hemodynamic variables the same genes that regulate these variables during exercise? (2) How much of the variance in exercise hemodynamic variables is genetic and how much is environmental? (3) Do the genetic and environmental influences on hemodynamic responses change with increasing levels of exercise? To determine how genetic and environmental effects expressed at rest influenced responses during dynamic exercise, a genetic analysis was conducted by fitting a series of models to the covariance matrices with the use of the LISREL VII program.
Conclusions We found that all the genetic effects expressed at the later stages of exercise can be explained by genetic effects expressed at rest and at the first stage of exercise. The environmental effects appear to be workload specific and include errors of measurement.
Both resting and exercise levels of blood pressure in children have been used as predictors of adult hypertension. The use of these variables as predictors relies on an assumption that there are shared mechanisms both within an individual for the resting and exercise childhood variables and between these juvenile variables and the individual's subsequent adult levels of blood pressure. One possible mechanism underlying the relation between childhood resting and exercise blood pressure and future blood pressure is a set of genes expressed in childhood that persists to regulate adult blood pressure. Whether such blood pressure–regulating genes exist and, if so, the magnitude of their genetic influence, is not known.
Studies of twins show that genes play a prominent role in determining blood pressure level.1 2 3 These studies reveal a significant familial aggregation of blood pressure measurements, with heritability estimates of 40% to 70%. Clinical studies confirm that the offspring of hypertensive parents are more likely to develop hypertension in adulthood than are the offspring of normotensive parents.4 Thus, a family history of hypertension, either through genetic or environmental mechanisms, predicts a subsequent increased risk of hypertension in the children of those families.
In an 11-year-old twin population we have previously shown that genes influence resting systolic blood pressure, diastolic blood pressure, and heart rate.5 Studies of adult twins confirm the clinical importance of genetic factors in hypertension both during childhood and adult years.6
Many studies of heart rate and blood pressure during exercise assume that the regulatory mechanisms for these variables are the same at rest as well as during exercise.7 Additionally, they assume that the same genetic effects regulate both child and adult levels of blood pressure and heart rate. Therefore, studies of children's blood pressure and heart rate responses to the stress of exercise have been used to predict adult blood pressure levels and the likelihood of hypertension.8 However, fundamental questions about these assumptions remain. To investigate the genetic relation of blood pressure and heart rate during both rest and exercise, we asked: (1) Are the genes that regulate resting hemodynamic variables the same genes that regulate these variables during exercise? (2) How much of the variance in exercise hemodynamic variables is genetic and how much is environmental? (3) Do the genetic and environmental influences on hemodynamic responses change with increasing levels of exercise?
Using school rosters, we identified 259 white twin pairs living in the Commonwealth of Virginia. Descriptions of our study were sent by the school, and affirmatively replying families were invited to participate. All twins were examined as close as possible to their eleventh birthdays. For the purposes of analysis, the twins were divided into five groups by zygosity and sex within a pair (monozygotic males and females, dizygotic males and females, and dizygotic opposite sex twin pairs). The designation twin 1 or twin 2 was by birth order except in the case of opposite sex twins, in which females were designated twin 1. All twins were healthy by history and physical examination and were not taking medications that could affect heart rate or blood pressure levels. This study was approved by the Institutional Review Board for human subject studies.
Dynamic exercise testing was performed on a bicycle ergometer modified to provide a special low saddle with adjustable handle bars and peddle height for children. The bicycle ergometer was calibrated by the manufacturer to indicate the workload (actual power output in kg/m per minute). The children were asked to cycle on the electronically braked ergometer, which maintained a given power output within the range of 60 to 80 cycle revolutions per minute. Heart rate was measured with a cardiotachometer and the blood pressure measured by oscillometry. Resting blood pressure was measured twice and the results averaged. Measurements of blood pressure could only be determined once during each stage of exercise because of the time required for oscillometric measurement. The measurements were begun at rest to accustom the children to the apparatus. The child began to cycle at a workload of 150 kg/m per minute and the workload was increased every 2 minutes by 100 kg/m per minute. The child was encouraged to continue exercising either until physical or mental fatigue supervened or the heart rate reached 170 beats per minute. All children achieved a workload of 550 kg/m per minute.
Reliability of Measurements
Data from a pilot study of 20 subjects who performed the bicycle exercise protocol showed excellent reliability. The test-retest reliability coefficients during exercise were systolic blood pressure, r=.91; diastolic blood pressure, r=.84; and heart rate, 0.97.
Pooled t tests were computed to compare male and female means for systolic and diastolic blood pressures and heart rate. The rest versus the stage 4 (450 kg/m per minute) differences for the three variables were compared with a paired t test. The effects of sex, zygosity, and their interaction on the slopes of the three cardiovascular variables were tested with repeated-measures analyses with the use of the polynomial transformation (GLM procedure in SAS).9 Rest and the four stages of exercise were the five factor levels in the repeated analysis, and orthogonal polynomial contrasts were specified between the factor levels. To avoid nonindependent observations, the means of twin 1 and twin 2 were used. Pearson correlation analysis was used to generate the total variances of the variables within each of the five conditions of rest, 150, 250, 350, and 450 kg/m per minute. An effect of the date of visit on the variables under study was noticed. This effect was regressed out of the data before genetic analyses were carried out.
The analysis of twin data provides a means of separating possible genetic and environmental contributions to traits. The method is based on the biological differences between monozygotic (MZ), or identical twins (who have all of their genes in common), and dizygotic (DZ), or nonidentical twins (who on average share 50% of their genes). Zygosity was assessed initially by a questionnaire and confirmed by dermatoglyphic analysis and testing of the twins and their parents for the ABO, MNS, Rh, Kell, Fy, Hp, Tf, Hb, PGM, AP, G-6-PD, Ct, and LDH systems. HLA typing was also performed. With this battery of polymorphisms, the probability of dizygosity for concordant pairs typically is <.001.10
Genetic influences on a given trait are implicated if the within-pair similarities for monozygotic twins are approximately twice those for dizygotic twins. Several factors can lead to deviations from the expectations of this simple additive genetic model. An example is the influence of the shared family environment, which can act to increase the similarity of dizygotic twins, relative to the monozygotic twin pair resemblance. Shared environmental factors that could influence blood pressure might be the salt content of the meals eaten by the family or the habit of family members to engage in leisure time physical activity together. The effects of individual experiences, on the other hand, will make family members less similar. These nonshared environmental effects reduce correlations between monozygotic twins.
A twin model allowing for a trait to be determined by additive genetic and nonshared environmental influences will necessarily yield different expectations for MZ versus DZ twin similarities than a model assuming that genetic factors as well as shared and nonshared environmental effects play a role (the full model). These models of different expectations can be compared against observed patterns of similarity (variance/covariance matrices) for the two (MZ and DZ) twin groups. The significance of genetic factors can be tested by comparing the approximate χ2 test of goodness of fit of the full model with that of the model from which the genetic contributions have been deleted. Along the same lines of reasoning, the shared environmental influences can be left out of the model to determine the significance of its effects. By eliminating both the effects of additive genes and the shared environment, a model accounting for nonshared environmental effects only could be tested as sufficient to explain the data. All possible models necessarily include the influences of the nonshared environment. This parameter also reflects unreliability caused by measurement error.
Data from male and female twin pairs offer the possibility to test for sex differences in genetic and environmental influences on a trait. To be able to conclude that the magnitudes of genetic and environmental effects differ for the sexes, models including separate parameters for males and females ideally would need to yield a significantly better fit than models without sex-specific estimates. When the differences in fit between models are nonsignificant, the criterion of parsimony is used; that is, the preferred model will have a relatively good fit while needing relatively few parameters to yield a satisfactory solution.
The multivariate analysis of twin data seeks to decompose the relationships between multiple variables into their genetic and environmental components.11 Often, genetic correlations between moderately heritable variables can be quite large, even though environmental correlations are small.12 Thus, a modest correlation such as those described for blood pressure in children and youth13 may reflect the fact that uncorrelated environmental changes over time may partially obscure the fact that the same genes are actually expressed throughout the whole period of measurement and result in only moderate phenotypic tracking correlations.
Starting with the variance/covariance matrices for the five twin groups under the conditions of rest and increasing physical load, we fitted a model that represented the covariance structure of the genetic and environmental covariances separately. The five measures of reactivity, at rest and at four increasing loads, are considered as an ordered set. These are V1 through V5 in Fig 1⇓. The model assumes that there are genetic factors (G1) that affect reactivity at rest (V1) and, potentially, all subsequent measures. A second set of genetic influences, G2, is assumed to be expressed at the start of exercise (V2) and possibly have continuing effects on all subsequent measures (V3 through V5). Analogous genetic effects are included to represent any additional genetic effects expressed de novo on later measures. A similar structure is assumed to underlie the influences of the environmental factors, E1 through E5, on the observed measures, V1 through V5. This way of decomposing a series of correlated measures is sometimes referred to as the Cholesky decomposition.12
The contribution of the latent genetic and environmental factors (G1 through G5, E1 through E5) to the covariances in the outcome measures is assessed by the path coefficients relating the latent variables to the five measures of reactivity. In so far as one latent variable (eg, G1) affects several outcomes, the measures will correlate genetically or environmentally. In practice, the Cholesky decomposition is a saturated model for the genetic and environmental covariance structures; that is, all necessary parameters are included but not all may contribute significantly. Fig 2⇓ includes only additive genetic effects (G1) and nonshared environmental effects (E). Other effects, such as those of genetic dominance or the shared environment, may be included if needed. Allowance may also be made for sex differences in the parameters if necessary.
The LISREL VII program14 was used to obtain optimal estimates of the genetic and environmental parameters. The program yields approximate χ2 tests of goodness of fit. Akaike's information criterion (AIC), which is the χ2 value minus twice the degrees of freedom, offers a quick first approach to judging the fit of the models. Models with lower (ie, larger negative) values are said to fit better by AIC than do models with higher values.
Of the 259 twin pairs studied, 148 were monozygotic and 111 were dizygotic. Data analysis was limited to submaximal exercise to minimize the potentially confounding effects of anaerobic exercise. All subjects comfortably reached stage 4 of the protocol. As Table 1⇓ indicates, all variables were equal in magnitude in boys and girls, with the exception that girls had higher diastolic blood pressures than boys (P<.01).
Table 2⇓ describes the group mean differences for heart rate and blood pressures between rest and stage 4 of bicycle exercise. Both systolic blood pressure and heart rate increased significantly (P≤.05) from rest to stage 4 for all five zygosity groups. Diastolic blood pressure increased significantly for monozygotic females and opposite sex twins. It was similar at rest and during exercise for monozygotic males, dizygotic males, and dizygotic females. Repeated-measures ANOVA revealed significant differences between boys and girls (P≤.05) for the slopes of heart rate and for the resting stage and first stage of physical load (150 kg/m per minute of systolic blood pressure). Heart rate and blood pressure means were found not to differ significantly between monozygotic and dizygotic twins. For heart rate, the interaction between sex and zygosity was found to be significant for all stages except rest.
Table 3⇓ summarizes the results of the model-fitting analyses for the three variables systolic blood pressure, diastolic blood pressure, and heart rate. Models with genetic and/or shared environmental effects explain the data significantly better than models including only unique environmental effects. The shared environmental effects can be omitted in all three response variables without adversely affecting the fit. The most parsimonious model showed similar parameter estimates for both boys and girls. The term “reduced HE” refers to the inclusion of only two genetic common factors, those affecting rest and exercise and those affecting exercise alone. The paths for the subsequent exercise stages included in the Cholesky decomposition did not contribute to the significance of the model, hence the designation, “reduced.”
In all three cases, based on the joint criteria of parsimony and goodness of fit, the best-fitting model required genetic effects and within-family environmental effects but no effects of the shared family environment. The best fit of the models was for diastolic blood pressure (χ2251=282.7), while those for heart rate and systolic blood pressure fit less well (χ2251=379.5, 360.9, respectively).
Fig 2⇑ plots the change in the total variance and its genetic and environmental components for each of the three measures as the workload increases. Since within a measure, the units remain constant over workloads, the total variance increases with increasing workload for all three hemodynamic variables. For heart rate, measured in beats per minute, we note that the increase in variance (bpm2) is almost entirely due to an increase in genetic variance, while the environmental variance remains virtually constant. For the two blood pressure measures, the increase in variance (mm Hg2) is a reflection of increasing genetic variance and increasing sensitivity to the environment, including errors of measurement.
Figs 3 and 4⇓⇓ show that the genetic and environmental variances, respectively, may be partitioned into the components of the model in Fig 1⇑. The pattern of genetic effects is shown in Fig 3⇓. The genetic variance for each variable is composed of two components: a “resting” and a “working” component. The resting component is expressed mainly at rest and decays during exercise.
For heart rate, the impact of genes expressed during rest decreases with increasing workload. The exercise genetic component is not only expressed at low loads (150 kg/m per minute) but persists throughout exercise. For both heart rate and systolic blood pressure, the genetic component increases during exercise. The resting genetic effects on diastolic blood pressure persist but do not decrease during exercise. There appears to be a genetic component of diastolic blood pressure separate from rest, which operates during exercise, but these genetic effects do not seem to increase significantly during exercise.
Fig 4⇑ shows the magnitude of the environmental contributions to the total variance at each exercise stage and the relative persistence of these effects at subsequent stages. The environmental effects expressed at rest continue to contribute during subsequent exercise stages, but their impact diminishes. Similarly, “new” environmental effects detected at 150 kg/m per minute persist, but with decreasing effect. This pattern most likely represents load-specific errors of measurement. The environmental effects, though small, are significant and are specific for each exercise stage.
In summary, all the genetic effects expressed at the later stages of exercise can be explained by genetic effects expressed at rest and at the first stage of exercise. Thus, two distinct and independent genetic processes explain the cardiovascular response to exercise. The first set of genes is expressed at rest and persists with decreasing impact as the load increases. A second set of genetic effects is activated at the start of exercise (150 kg/m per minute) and generally exerts an increasing effect as exercise load increases. The environmental effects appear to be workload specific and include errors of measurement.
The model that we used for the changing roles of genes and environment in the cardiovascular response to increasing workload uses established approaches to the genetic modeling of multiple measures and developmental change.11 15
In a previous study, we have shown that in normotensive 11-year-old children there is a significant genetic effect on resting systolic blood pressure, diastolic blood pressure, and heart rate.5 The present study confirms these earlier observations on resting blood pressure and heart rate and reveals that genetic effects also play an important role in the regulation of these variables during exercise. However, the resting and exercise genetic effects appear to be distinct one from another.
To understand the changes that occur in the genetic and environmental effects during exercise, we used repeated-measurement genetic models that use the exercise intervals as the time periods. The resting genetic contributions diminish with exercise. A set of “work” genes becomes manifest during early exercise and its effect increases with increasing exercise load. The environmental contribution to heart rate and blood pressure suggests workload-specific environmental regulation.
The cardiovascular response to exercise in children is affected by age, sex, ethnicity, exercise protocol, and training.16 Dynamic exercise increases cardiac output by as much as 400% to 500%. The increase in cardiac output, which results mainly from increases in heart rate and to a lesser extent from increases in stroke volume and cardiac contractility, is accompanied by a decrease in peripheral vascular resistance.17 Our study does not allow the partition of the increase in blood pressure into components of cardiac output and peripheral vascular resistance. Blood pressure measured during exercise reflects the interaction of these hemodynamic events. Systolic blood pressure increases while diastolic pressure changes little. Heart rate increases progressively during exercise and correlates linearly with oxygen consumption.
In children, both resting and exercise blood pressures have been used as predictors of adult hypertension. Resting systolic blood pressure readings in school age children track well, while diastolic readings track somewhat less well.18 Lauer et al19 described the importance of relative growth in the tracking of blood pressure and predicted that ≈5% of the children appeared to be tracking toward established hypertension. Van Lenthe et al20 have reviewed recent work using model fitting to interpret the tracking of blood pressure in children. They confirmed tracking of blood pressure in both children and adolescents but found much lower tracking correlations in children.
In addition to tracking, high cardiovascular reactivity to stressors such as mental stress21 may predict future hypertension. In a study of the effects of exercise on blood pressure, Schieken et al22 found that during exercise, the body size–adjusted systolic blood pressure in children in the highest blood pressure quintile remained the highest during bicycle ergometry. The pressure product (systolic blood pressure times heart rate), moreover, increased the most in children whose resting blood pressures were in the highest quintile. Falkner et al23 found exaggerated physiological responses to various stressors in children from hypertensive families. They hypothesized that these resulted from changes in the autonomic nervous system. The significance of changes in blood pressure during exercise is unknown, but an exaggerated systolic blood pressure response has been associated with left ventricular hypertrophy and may be a marker for future hypertension.24
We find that the increase in variance in cardiovascular response during exercise is a function of both genetic and environmental influences. However, the effects of the environment are largely specific to individual loads, and their persistence from one load to the next is relatively slight. By contrast, the increase in genetic variance can be explained principally by the fact that two separate genetic systems are needed to account for the variability in response. One is expressed at rest and persists throughout exercise. The second set of genes is expressed at the start of exercise with effects that persist and increase as the load increases. Over the range of submaximal exercise, there is little evidence that increasing the load leads to the expression of totally new genetic effects. Mahoney et al25 studied the forearm vascular resistance in children selected with low, average, and high blood pressures. They found that the vasodilator function was less in children with higher levels of blood pressure compared with those who had lower levels of blood pressure. Studies of the genetics of variables controlling vascular resistance or those controlling maximal exercise oxygen consumption might provide important clues to the link between resting and exercise blood pressures. Longitudinal genetic studies of the interrelationship of rest and exercise responses performed in later adolescence and young adulthood would provide additional information about the predictive value of hemodynamic variables during exercise to predict adult resting blood pressure.
This study was sponsored in part by grant R01-HL-31010 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md.
- Received February 27, 1996.
- Revision received April 29, 1996.
- Accepted April 30, 1996.
- Copyright © 1996 by American Heart Association
Feinleib M, Garrison R, Borhani N, Rosenman R, Christian J. Studies of hypertension in twins. In: Paul O, ed. Epidemiology and Control of Hypertension. Chicago, Ill: Yearbook Medical; 1975.
Havlik RH, Garrison RH, Katz SH, Ellison RC, Feinleib M, Myrianthopoulos NC. Detection of genetic variance blood pressure of seven-year-old twins. Am J Epidemiol. 1979;109:512-516.
Heinberg A, Magnus P, Berg K, Nance WE. Blood pressure in Norwegian twins. In: Twin Research 3, Part C. New York, NY: Alan R. Liss; 1981:163-168.
Ewell LW, Corey LA, Winter PM, Boughman JA, Nance WE. Blood pressure studies on monozygotic twins and their families. In: Nance WE, Allen G, Parisi P, eds. Twin Research: Clinical Studies. New York, NY: Alan R. Liss; 1981:29-38.
Fixler DE, Laird WP, Dana K. Usefulness of exercise stress testing for prediction of blood pressure trends. Pediatrics. 1985;75:1071-1075.
SAS Institute Inc. SAS/STAT User's Guide, Version 6. 4th ed, vol 1. Cary, NC: SAS Institute Inc; 1989:2.
Magnus P, Berg K, Nance WE. Predicting zygosity in Norwegian twin pairs born 1915 to 1916. Clin Genet. 1984;24:102-112.
Neale MC, Cardon LR, eds. Methodology for Genetic Studies of Twins and Families. Dordrecht, Netherlands: Kluwer Academic Publishers; 1992.
Martin NG, Eaves LJ. The genetic analysis of covariance structure. Heredity (Edinburgh). 1977;38:79-95.
Lauer RM, Clarke WR. Childhood risk factors for high adult blood pressure: the Muscatine Study. Pediatrics. 1989;84:633-641.
Jo¨reskog KG, So¨rbom D. LISREL V11: A Guide to the Program and Applications. 2nd ed. Chicago, Ill: SPSS, Inc; 1989.
Blomquist CG. Exercise physiology: clinical aspects. In: Wenger NK, ed. Exercise and the Heart. Philadelphia, Pa: FA Davis Co; 1978:1-12.
Voors AW, Foster TA, Frerichs RR, Webber LS, Berenson GS. Studies of blood pressures in children, ages 5-14 years, in a total biracial community: the Bogalusa Heart Study. Circulation. 1976;54:319-326.
Lauer RM, Clarke WR, Beaglehole R. Level, trend, and variability of blood pressure during childhood: the Muscatine Study. Circulation. 1984;69:242-249.
Falkner B, Onesti G, Angelakos ET, Fernandes M, Langman C. Cardiovascular response to mental stress in normal adolescents with hypertensive parents. Hypertension. 1979;1:23-30.
Schieken RM, Clarke WR, Lauer RM. The cardiovascular responses to exercise in children across the blood pressure distribution: the Muscatine Study. Hypertension. 1983;5:71-78.
Falkner B, Kushner H, Khalsa DK, Canessa M, Katz S. Sodium sensitivity, growth and family history of hypertension in young blacks. J Hypertens. 1986;4:S38-S41.