Multivariate Genetic Analysis of Blood Pressure

Transcription

1 1780 Multivariate Genetic Analysis of Blood Pressure and Body Size The Medical College of Virginia Twin Study Richard M. Schieken, MD; Michael Mosteller, PhD; Monica M. Goble, MD; William B. Moskowitz, MD; John K. Hewitt, PhD; Lindon J. Eaves, PhD, DSc; and Walter E. Nance, MD, PhD Background. In subjects of all ages, those who weigh the most often have the highest blood pressure. Thus, in epidemiological studies, weight is the most important correlate of blood pressure. Using the data from the Medical College of Virginia Twin Study, we asked these questions: 1) Do the same genetic paths that regulate body size also regulate systolic and diastolic blood pressure? 2) Are there distinct genetic pathways that regulate each of these variables? 3) Does environment play a major regulatory role? 4) Are the correlations among these variables mainly due to genetic or environmental effects? 5) Do genetic paths that regulate body size mediate the correlation between systolic blood pressure and diastolic blood pressure? Methods and Results. We ascertained 253 Caucasian twin pairs living in the Commonwealth of Virginia. The average age was 11.2±0.2 years. We used multivariate path analyses to investigate the genetic relations among systolic blood pressure, diastolic blood pressure, and body size. We found that there was a highly significant genetic relation between systolic blood pressure and body size and between systolic and diastolic blood pressure. There are genetic paths that are shared within these two sets of variables, but in each case, the paths for each pair appear to be separate from one another. Conclusions. These analyses provide a method to partition correlation coefficients found in epidemiological studies into genetic and environmental components. The correlations found among these three variables are in large part due to these genetic relations. We found no genetic relation between diastolic blood pressure and body size. (Circulation 1992;86: ) KEY WORDS * cardiology, preventive * hypertension * screening * children Systolic and diastolic blood pressure are important predictors of atherosclerotic disease in adults.1"2 In children, blood pressure may predict an increased risk of both hypertension and atherosclerosis with onset in adulthood.3 In subjects of all ages, those who weigh the most often have the highest blood pressure. Thus, in epidemiological studies, weight is the most important correlate of blood pressure.45 Furthermore, weight loss often lowers blood pressure in obese adults and children.67 Physicians recognize the relation between weight and blood pressure and often recommend weight loss to patients at risk for sustained hypertension. Before recommending weight loss as an intervention to lower blood pressure in children who are not frankly obese, the relation between weight and blood pressure needs to be more carefully defined. Essential diastolic hypertension can occur in lean children, making a direct mechanism linking weight with systolic and diastolic From the Children's Medical Center, Division of Pediatric Cardiology, and the Department of Human Genetics of the Medical College of Virginia. Supported by the National Institutes of Health, National Heart, Lung, and Blood Institute (RO1-HL and R29-HL-38878). Address for reprints: Richard M. Schieken, MD, Box 26, MCV Station, Medical College of Virginia, Richmond, VA Received October 7, 1991; revision accepted August 19, blood pressure unlikely.8 In addition, the variability of both blood pressure and weight in children is highly influenced by genes.9"10 The intent of this report is to quantitate the genetic and environmental contributions 1) to the relation of body size with blood pressure; 2) to the relation of systolic with diastolic blood pressure, and 3) to search for a common relation among these three variables. Using the data from the Medical College of Virginia Twin Study, we performed both regression and path analyses. We asked these questions: 1) Do the same genetic paths that regulate body size also regulate systolic and diastolic blood pressure? 2) Are there distinct genetic pathways that regulate each of these variables? 3) Does environment play the major regulatory role? 4) Are the correlations among these variables mainly due to genetic or environmental effects? 5) Do genetic paths that regulate body size mediate the correlation between systolic blood pressure and diastolic blood pressure? Methods Population We ascertained 341 twin pairs living in the Commonwealth of Virginia using school rosters for identification. Descriptions of our study were sent by the school and

2 Schieken et al Genetics of Blood Pressure and Body Size 1781 affirmatively replying families were invited to participate. All twins were examined as close as possible to their 11th birthdays. Within the total population of 341 twin pairs, there were 253 Caucasian twin pairs that made up the sample for this study. For the purposes of analysis, the twins were divided into five groups by zygosity and sex within a pair. The designation "twin 1" or "twin 2" was by birth order except in the case of opposite sex twins, in whom the male twin was designated as twin 1. The average age was 11.2±0.2 years. Zygosity Determination Zygosity was assessed initially by a questionnaire and confirmed by 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 > Blood Pressure A casual blood pressure was measured using a mercury sphygmomanometer with the appropriate compression cuff with the subject in the sitting position. The chosen cuff was big enough to circle at least half of the upper arm without overlapping. The rubber bladder rested over the artery being compressed and had sufficient width to cover at least two thirds of the upper arm. The pressure within the compression cuff indicated by the level of the mercury column at the murmur of the first- and fourth-phase Korotkoff sounds was recorded. The blood pressure was recorded three times and averaged. The interobserver intraclass correlation coefficients for blood pressure were r=0.89 and r=0.84 for systolic and diastolic blood pressures, respectively. Anthropometric Measurements Weight was recorded in kilograms on an electronic scale that was calibrated using standardized weights. Height was measured using a stadiometer. Body mass index (BMI) was calculated as (weight in kilograms)/ (height in meters)2. Tanner Stage Pubertal staging of all children was performed using a five-scale score based on Tanner's criteria.12 Statistical Analyses Descriptive statistics. Mean differences within a twin pair were tested with a paired t test and overall malefemale differences with a pooled t test. To summarize the correlation of variables across twins, correlation coefficients were pooled using a z-score procedure developed by Fisher.13 Natural log transformations were used to normalize the distributions of weight and BMI, both of which were positively skewed. To investigate influences of weight on the correlation of systolic and diastolic blood pressure, we performed a linear regression of both systolic blood pressure and diastolic blood pressure on weight and analyzed the resulting residuals. Similar regression analyses were performed on blood pressure and BMI. Multivariate genetic analysis. We used multivariate path analysis to investigate the nature of the interrelations between the genetic and environmental determi- TABLE 1. Test Results Comparing Boys and Girls From All Twin Pairs Mean SE SD p SBP (mm Hg) Girls Boys DBP (mm Hg) Girls Boys <0.01 Weight (kg) Girls Boys BMI (kg/m2) Girls Boys SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, body mass index. n=260 female twins, 246 male nants of systolic blood pressure, diastolic blood pressure, and weight. In addition, these same analyses were performed using BMI instead of weight (see "Appendix"). The effects included in the models were labeled as either specific or common. We used methods similar to those published in a previous study.'4 Briefly, specific effects are those that influence only one of the variables, i.e., systolic blood pressure, diastolic blood pressure, weight, or BMI. Common effects are those that influence two or more of these variables. Both common and specific effects were also categorized as familial effects or nonshared environmental effects. The familial effects, those that increase the twin correlation, included genetic and shared environmental effects. Shared environmental effects are those affecting both members of a twin pair. Nonshared environmental effects influence one twin but not the other. Measurement error is included in the nonshared environmental effects. A model was formulated to partition the genetic and environmental variances of systolic blood pressure into components specific to systolic blood pressure and those common to systolic blood pressure and the other variables; a comparable model was formulated to partition the variances of diastolic blood pressure. In addition, these models were used to partition the correlation of systolic and diastolic blood pressure into genetic and environmental components (see "Appendix").15 The LIS- REL VII program'6 was used to obtain maximum likelihood estimates of model parameters and a x2 goodness of fit statistic. Pairwise comparisons of alternative models were performed. The simplest model that adequately explained the data was selected as the best one.'7 Results Means Table 1 compares the mean values of systolic and diastolic blood pressure, untransformed weight, and BMI by sex. Diastolic blood pressure was higher in girls than in boys. Table 2 presents a comparison of systolic blood pressure, diastolic blood pressure, untransformed weight, and BMI by sex. A significant difference existed for BMI in the male-female dizygotic twins. The group

3 1782 Circulation Vol 86, No 6 December 1992 TABLE 2. Means and Standard Errors of Systolic Blood Pressure, Diastolic Blood Pressure, and Untransformed Weight and Body Mass Index by Zygosity and Sex of Twins, Probability Level for Paired t Test Comparing Twin 1 and Twin 2, and Number of Twin Pairs Per Group SBP DBP WT BMI Twin 1 Twin 2 Twin 1 Twin 2 Twin 1 Twin 2 Twin 1 Twin 2 Monozygotic twins Male/male Mean SE p n Female/female Mean SE p n Dizygotic twins Male/male Mean SE p n Female/female Mean SE p n Male/female Mean SE p n SBP, systolic blood pressure; DBP, diastolic blood pressure; WT, untransformed weight; BMI, body mass index. mean Tanner stage for girls (2.15) was higher than for the Tanner stage for boys (1.69, p<o.o1). Correlations The Pearson product-moment correlation coefficients appear in Table 3 by zygosity and sex group for all of the variable twin combinations. The across-twins-withinvariable monozygotic coefficients were larger than the across-twins-within-variable dizygotic coefficients for TABLE 3. all three variables, suggesting that genetic effects make a significant contribution to the observed variation. The across-twins-across-variables correlations provide insight into sources of covariation between systolic and diastolic blood pressure and weight. These acrosstwins-across-variables correlations for the monozygotic pairs were higher than those for the dizygotic pairs, for systolic blood pressure-diastolic blood pressure, and systolic blood pressure-weight, wherein the dizygotic Pearson Product-Moment Correlations and Pooled Correlations Across twins-within variable Across twins-across variables* n SBP1-SBP2 DBP1-DBP2 WT1-WT2 SBP-DBP SBP-WT DBP-WT Monozygotic twins Male/male t Female/female t Dizygotic twins Male/male t t 0.lot 0.29t -0.14t Female/female t t 0.02t -0.llt Male/female t 0.04t t -0.04t 0.07t SBP, systolic blood pressure; DBP, diastolic blood pressure; WT, weight. *Pooled average of two correlations: corr (variable 1 in twin 1, variable 2 in twin 2) and corr (variable 2 in twin 1, variable 1 in twin 2). tcorrelation not significantly different from zero at p=0.05.

4 Schieken et al Genetics of Blood Pressure and Body Size 1783 TABLE 4. Phenotypic Correlations of Systolic Blood Pressure and Diastolic Blood Pressure With Weight and With Body Mass Index Weight Body mass index r p (Ho:p=0) r p (Ho:p=0) Systolic blood pressure 0.40 < <0.01 Diastolic blood pressure Based on a random sample of one twin per pair, n=253. correlations were not different from zero in either boys or girls. This may indicate that genetic effects are important in bringing about covariation between the sets of variables. No significant correlation was found either in monozygotic or dizygotic pairs for diastolic blood pressure-weight. One twin from each pair was chosen randomly. The unadjusted correlation within an individual for systolic blood pressure and diastolic blood pressure was significant (r=0.33, p<0.01). The correlations of the blood pressure variables and weight variables within an individual are given in Table 4. For systolic blood pressure and weight, the unadjusted correlation coefficient (r=0.40) suggested that 16% of the variation in systolic blood pressure was accounted for by variation in weight. The correlations for diastolic blood pressure with weight and diastolic pressure with BMI were not significantly different from zero. After sequentially regressing both systolic blood pressure and diastolic blood pressure on weight and systolic blood pressure and diastolic blood pressure on BMI, the correlations of the residuals showed no significant change from their unadjusted value. This suggests that neither weight nor BMI were important factors in the covariation of systolic blood pressure with diastolic blood pressure. Multivariate Genetic Analyses The best-fitting multivariate model for systolic blood pressure is depicted in Figure 1. The magnitude of the effects on systolic blood pressure and weight were equal in boys and girls. The genetic effect on diastolic blood pressure was larger in boys than in girls, which is indicated by the relative width of the solid arrows in the upper and lower portions of Figure 1. Table 5 lists the parameter estimates for the model depicted in Figure 1. Table 6 is a tabulation of the variance of systolic blood pressure partitioned according to source, based on the best-fitting model. Overall, genetic effects accounted for 63.8% of the variance of systolic blood pressure. The percentage of the total variance caused by genetic effects common to systolic blood pressure and weight was 11.2%. Likewise, the percentage of the total variance caused by genetic effects common to systolic blood pressure and diastolic blood pressure was 8.7%. Genetic SBP - Best Model X2= , d.f.. 92, P=.141 Males Females FIGURE 1. The best-fittingmodelfora trivariategenetic analysis ofsystolic bloodpressure (SBP), diastolic bloodpressure (DBP), and weight (WT) that allows the variance of SBP to be partitioned. The relative magnitude ofa given effect on an observed variable is indicated by the width of the arrow connecting them: _, ; -, ;, ; -, ; Nonsolid arrows represent causal paths that were fixed at zero (dropped from the model) during the model fitting process. Depicted here is the model for the unlike-sex dizygotic twin pairs; however, the results for like-sex male pairs are the same as thosefor the male subjects shown here. Likewise, the results for like-sex female pairs are the same as those for the female subjects shown here. Table 5 lists the parameter estimates for each path. In Table 6, the variance of SBP is partitioned into its genetic and environmental sources.

5 1784 Circulation Vol 86, No 6 December 1992 TABLE 5. Parameter Estimates and Standard Errors From the Best-Fitting Trivariate Model Male twins Female twins WT DBP SBP WT DBP SBP WT, DBP, SBP 0.941± ± ±0.049 DBP, SBP ± ± ± ±0.066 SBP ± ±0.046 WT, DBP, SBP 0.324± ± DBP, SBP ± ± ± ±0.047 SBP ± ±0.033 Correlation of genetic effects between boys and girls in unlike-sex twin pairs WT, DBP, SBP 0.337±0.135 DBP, SBP 0.045±0.249 SBP 0.227±0.247 WT, log-transformed weight; DBP, diastolic blood pressure; partition the variance of SBP (Figure 1). SBP, systolic blood pressure. Estimates are for the model designed to effects specific to systolic blood pressure accounted for 43.9% of the total variance. Nonshared environmental effects that were specific to systolic blood pressure accounted for 34.3% of the total variance. Those nonshared environmental effects common to systolic blood pressure and diastolic blood pressure were 1.9%. There were no nonshared environmental effects common to systolic blood pressure and weight. We found no evidence for shared environmental effects for any of the variables. A separate model was designed to partition the sources of variation of diastolic blood pressure (Figure 2). The parameter estimates for this model appear in Table 7. Because we observed sex differences in the magnitude of the genetic effects on diastolic blood pressure, the variance of diastolic blood pressure was partitioned separately for boys and girls (Table 8). The percentage of the total variance caused by genetic effects common to systolic and diastolic blood pressure was different in boys (7.4%) and girls (10.9%). We found no evidence for a genetic effect common to TABLE 6. Variance Partition of Systolic Blood Pressure by Source Variance of total of source Boys and girls WT, SBP DBP, SBP SBP Total DBP, SBP SBP Total Total WT, weight (log-transformed); SBP, DBP, diastolic blood pressure. Based analysis of SBP, DBP, and WT. systolic blood pressure; on a trivariate genetic diastolic blood pressure and weight. Based on the best-fitting model, 74% of the correlation between systolic blood pressure and diastolic blood pressure in boys was due to common genes (see "Appendix"). In girls, the genetic component of the systolic blood pressure-diastolic blood pressure correlation was 67%. In both sexes, the systolic blood pressure-weight correlation was attributed entirely to genetic effects. Quantitative parameter estimates for the model and sources of variation for the analyses of the blood pressures and BMI were similar to weight and are provided in Tables 9 and 10. Parameter estimates for the diastolic blood pressure model and sources of variation for BMI are provided in Tables 11 and 12. Discussion We found that there was a strong genetic relation between systolic blood pressure and weight and between systolic and diastolic blood pressure. There are genetic paths that are shared between these two sets of variables but in each case, the genetic paths for each pair appear to be separate from one another. The correlations found among these variables in large part are due to these genetic relations. We found no evidence for a genetic relation between diastolic blood pressure and weight or diastolic blood pressure and BMI. A strength of this study is its population, a large sample of same-aged, preadolescent twins of either sex. The power of the sample population allows us to resolve the genetic relations of these variables. One important limitation of the study is that we do not have a large enough sample of black children to fully characterize the genetic and environmental contributions to the variance of the blood pressure-weight relation by race. In both adults and children, there is a consistent relation between weight and blood pressure. The Framingham study demonstrated that middle-aged adults who weighed the most were the most likely to have elevated blood pressure.'8 These people were at a higher risk for the development of stroke and coronary heart disease. Initial analyses from the Framingham investigators concentrated on the predictive effects of blood pressure. Weight was considered in the analysis as

6 Schieken et al Genetics of Blood Pressure and Body Size 1785 DBP - Best Model X 2 = 10958, d.f.= 92, P=.102 Males Females LWTL LSBPi DBPi FIGURE 2. The best-fitting modelfor a trivariate genetic analysis of systolic blood pressure (SBP), diastolic bloodpressure (DBP), and weight (WT) that allows the variance ofdbp to be partitioned. The relative magnitude ofa given effect on an observed variable is indicated by the width ofthe arrow connecting them:, ;, ; _, ; -, ;..., Nonsolid arrows represent causal paths that were fixed at zero (dropped from the model) during the model fitting process. Depicted here is the modelfor the unlike-sex dizygotic twin pairs; however, the results for like-sex malepairs are the same as those for the male subjects shown here. Likewise, the results for like-sex female pairs are the same as those for the female subjects shown here. Table 7 lists the parameter estimates for each path. In Table 8, the variance ofdbp is partitioned into its genetic and environmental sources. a confounding variable to be controlled in order to explore the full impact of blood pressure. Only later, when weight was analyzed, holding blood pressure constant as a confounding variable, was weight found to have an independent predictive value for coronary heart disease.19 We investigated a model that included a path from a set of genes that influenced both weight and systolic blood pressure. This path was found to account for a significant proportion of the variance of systolic blood pressure. The model also accounted for the strong correlation of weight and systolic blood pressure. Hanis et ap20 showed in the Muscatine Study that a large proportion of the observed familial aggregation of systolic blood pressure could be explained by the familial TABLE 7. Parameter Estimates and Standard Errors From the Best-Fitting Trivariate Model Male twins Female twins WT SBP DBP WT SBP DBP WT, SBP, DBP ± ± ± SBP, DBP ± ± ±0.065 DBP ± ±0.065 WT, SBP, DBP ± SBP, DBP ± ± ± ±0.051 DBP ± ±0.035 Correlation of genetic effects between boys and girls in unlike-sex twin pairs WT, DBP, SBP 0.338±0.135 DBP, SBP 0.172±0.229 DBP 0.213±0.272 WT, weight (log-transformed); SBP, systolic blood pressure; DBP, diastolic blood pressure. These estimates are for the model designed to partition the variance of DBP (Figure 2).

7 1786 Circulation Vol 86, No 6 December 1992 TABLE 8. Variance Partition of Diastolic Blood Pressure by Source Variance of total of source Boys SBP, DBP DBP Total SBP, DBP DBP Total Total Girls SBP, DBP DBP Total SBP, DBP DBP Total Total WT, log-transformed weight; SBP, systolic blood pressure; DBP, diastolic blood pressure. Based on a trivariate genetic analysis of SBP, DBP, and WT. aggregation of weight. They found that 30% of the full sibling correlation was due to effects of weight. Their analyses included both shared genetic and shared environmental factors that contribute to the covariance of weight and systolic blood pressure. In contrast, removing the effects of weight on diastolic blood pressure did little to reduce the genetic variance. This suggests that genetic paths that regulate diastolic TABLE 9. Parameter Estimates and Standard Errors From the Best-Fitting Trivariate Model blood pressure are not common to weight. In the Bogalusa Heart Study, obese and very obese children had significantly higher systolic blood pressures than nonobese children; however, only the very obese children had significantly higher diastolic blood pressures. In a longitudinal study of children from Bogalusa, La., after dividing groups by ponderosity, there was a negative correlation between year 1 ponderosity and year 6 diastolic blood pressure, suggesting that ponderosity does not predict elevated diastolic blood pressure. This observation was true for black children and white children of either sex. Overweight adults frequently have elevated diastolic blood pressure, and this may suggest that the genes that regulate diastolic blood pressure in children may be different from those found in adults. Removing the effects of weight or BMI did little to reduce the correlation of systolic and diastolic blood pressure. We believe that this reflects mechanisms other than weight that influence common regulatory pathways. Perhaps this relation depends on genes shared in regulation of the connective tissue content of the vessel walls. Correlations found in epidemiological studies identify important relations. The relation in adults of weight to both systolic and diastolic blood pressure has been the foundation of the recommendation for weight loss in hypertensive patients. In children, a similar correlation exists between weight and systolic blood pressure. In this current study of children, we found that this relation depends almost entirely on the presence of genetic paths common to systolic blood pressure and weight. Our findings may explain in part, the lower correlations of diastolic blood pressure compared with systolic blood pressure with either weight or weight loss that have been seen in other studies.22 Weight or ponderosity in children and in adults may affect blood pressure via separate genetic mechanisms. Studies of obese subjects frequently demonstrate that weight loss reduces blood pressure.23 One possible explanation is that genes for obesity are different than genes that regulate normal body weight. Moll et a124 have demonstrated that obesity may result from genes other than those that regulate normal body weight. Male twins Female twins BMI DBP SBP BMI DBP SBP BMI, DBP, SBP DBP, SBP ± ± ±0.067 SBP ± ±0.047 BMI, DBP, SBP 0.366± ± DBP, SBP ± ± ± ±0.047 SBP ± ±0.033 Correlation of genetic effects between boys and girls in unlike-sex twin pairs BMI, DBP, SBP 0.407±0.132 DBP, SBP 0.042±0.248 SBP 0.208±0.242 BMI, log-transformed body mass index; DBP, diastolic blood pressure; SBP, systolic blood pressure. These estimates are for the model designed to partition the variance of SBP.

8 Schieken et al Genetics of Blood Pressure and Body Size 1787 TABLE 10. Variance Partition of Systolic Blood Pressure by Source Variance of total of source Boys and girls BMI, SBP DBP, SBP SBP Total BMI DBP, SBP SBP Total Total BMI, log-transformed body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure. Based on a trivariate genetic analysis of SBP, DBP, and BMI. "Obesity genes" may affect blood pressure via different mechanisms than "normal weight genes." Therefore, our findings do not preclude a recommendation for weight loss in obese children to lower their blood pressure. This multivariate genetic analysis allows us to begin to understand the epidemiological association of weight and blood pressure. The advantage of multivariate genetic path analysis is that it allows us to consider the effects of systolic blood pressure, diastolic blood pressure, and weight simultaneously and to analyze genetic and environmental effects separately. Total variance can be partitioned into those effects, whether genetic or environmental, that affected one variable specifically or two or more variables concurrently. We were thus able to characterize the relation between weight and systolic blood pressure and the relation between systolic and diastolic blood pressure. Our analysis clearly documented that, in nonobese children, neither the genetic nor the environmental effects of weight influenced diastolic blood pressure. Moreover, because no shared genetic or environmental effects accounting for the variance of diastolic blood pressure were common to systolic blood pressure and weight, the relation between diastolic and systolic blood pressure must depend on regulatory mechanisms that do not directly affect weight. These analyses provide a method to partition the correlation coefficients found in epidemiological studies into genetic and environmental components. In this study, virtually all of the correlation between weight and systolic blood pressure and a large part of the correlation between systolic blood pressure and diastolic blood pressure in both boys and girls was explained by genetic paths. Further studies are required to understand the genetic mechanisms responsible for the covariance of weight and blood pressure. Preventive recommendations should be based on the findings of such studies. Acknowledgments We acknowledge the technical expertise of A. Cook, L. Stevenson, B. Toms, L. Davidson, W. Smith, and L. Vermes. Appendix Results of Multivariate Analysis of Systolic and Diastolic Blood Pressure and BMI Tables 9-12 are equivalent to Tables 5-8 except that they are based on analyses of systolic blood pressure, diastolic blood pressure, and BMI instead of systolic blood pressure, diastolic blood pressure, and weight. Partitioning the Correlation Between Systolic Blood Pressure and Diastolic Blood Pressure Path analysis allows the estimation of the correlation between two observed variables by multiplying the coefficients along any path(s) connecting them, and, if there is more than one such path, summing the products. For example, if systolic blood pressure (SBP) and diastolic blood pressure (DBP) in male twins are considered in Figure 1, there are two paths connecting them: one through the effect labeled "Genetic Effects on DBP and SBP in Males," the other through the effect labeled "Environmental Effects on DBP and SBP in Males." To estimate the portion of the correlation of SBP and DBP caused by genetic effects within an individual male, the two relevant coefficients (see Table 5) are multiplied: TABLE 11. Parameter Estimates and Standard Errors From the Best-Fitting Trivariate Model Male twins Female twins BMI SBP DBP BMI SBP DBP BMI, SBP, DBP 0.926± ± ± ± SBP, DBP ± ± ± ±0.064 DBP ± ±0.066 BMI, SBP, DBP 0.366± ± SBP, DBP ± ± ± ±0.051 DBP ± ±0.035 Correlation of genetic effects between boys and girls in unlike-sex twin pairs BMI, SBP, DBP 0.407±0.132 SBP, DBP 0.170±0.222 DBP BMI, log-transformed body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure. These estimates are for the model designed to partition the variance of DBP.

9 1788 Circulation Vol 86, No 6 December 1992 TABLE 12. Variance Partition of Diastolic Blood Pressure by Source Variance of total of source Boys BMI, SBP SBP, DBP DBP Total BMI SBP, DBP DBP Total Total Girls BMI, SBP SBP, DBP DBP Total BMI SBP, DBP DBP Total Total BMI, log-transformed body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure. Based on a trivariate genetic analysis of SBP, DBP, and BMI x291=0.252 Similarly, to estimate the portion of the correlation caused by environmental effects that influence both SBP and DBP: 0.646x0.134=0.087 The total correlation is the sum of these two: =0.339 The percent of the correlation caused by genetic effects common to SBP and DBP is (0.252/0.339)x 100=74.3% The percent of the correlation caused by environmental effects common to SBP and DBP is (0.087/0.339)x 100=25.7% References 1. The Pooling Project Research Group: Relationship of blood pressure, serum cholesterol, smoking habit, relative weight and ECG abnormalities to incidence of major coronary events: Final report of the pooling project. J Chronic Dis 1978;31: Stamler J, Neaton JD, Wentworth D: Blood pressure (systolic and diastolic) and risk of fatal coronary heart disease. Hypertension 1989;13: Lauer RM, Clarke WR: Childhood risk factors for high adult blood pressure: The Muscatine Study. Pediatrics 1989;84: Burns TL, Moll PP, Lauer RM: The relation between ponderosity and coronary risk factors in children and their relatives: The Muscatine Ponderosity Family Study. Am J Epidemiol 1989;129: Webber LS, Voors AW, Srinivasan SR, Frerichs RR, Berenson GS: Occurrence in children of multiple risk factors for coronary artery disease: The Bogalusa Heart Study. Prev Med 1979;8: Reisin E, Abel R, Modan M, Silverberg DS: Effect of weight loss without salt restriction on the reduction of blood pressure in overweight hypertensive patients. N Engl JMed 1978;298: Rocchini AP, Katch V, Anderson J, Hinderliter J, Becque D, Martin M, Marks C: Blood pressure in obese adolescents: Effect of weight loss. Pediatrics 1988;82: Burke GL, Freedman DS, Webber LS, Berenson GS: Persistence of high diastolic blood pressure in thin children: The Bogalusa Heart Study. Hypertension 1986;8: Schieken RM, Eaves LJ, Hewitt JK, Mosteller M, Bodurtha JN, Moskowitz WB, Nance WE: Univariate genetic analysis of blood pressure in children: The Medical College of Virginia Twin Study. Am J Cardiol 1989;64: Bodurtha JN, Mosteller M, Hewitt JK, Nance WE, Eaves LJ, Katz S, Schieken RM: Genetic analysis of anthropometric measures in 11-year-old twins. Pediatr Res 1990;28: Miller JZ, Norton JA, Nance WE: Vitamin C and growth. JAMA 1977;238: Tanner JM: Growth at Adolescence, 2nd ed. Oxford, Blackwell Scientific Publications, 1962, p Fisher RA: On the probable error of a coefficient of correlations deduced from a small sample. Metron 1921;1: Verhaaren HA, Schieken RM, Mosteller M, Hewitt JK, Eaves LJ, Nance WE: Bivariate genetic analysis of left ventricular mass and weight in pubertal twins: The Medical College of Virginia Twin Study. Am J Cardiol 1991;68: Falconer DS: Introduction to Quantitative Genetics. New York, Longman Inc, Joreskog KG, Sorbom D: LISREL VII: A Guide to the Program and Applications, 2nd ed. Chicago, SPSS Inc, Neale MC, Heath AC, Hewitt JK, Eaves LJ, Fulker DW: Fitting genetic models with LISREL: Hypothesis testing. Behav Genet 1989; 19: Kannel WB, LeBauer EJ, Dawber TR, McNamara PM: Relation of body weight to development of coronary heart disease: The Framingham Study. Circulation 1967;35: Hubert HB, Feinleib M, McNamara PM, Castelli WP: Obesity as an independent risk factor for cardiovascular disease: A 26-year follow-up of participants in the Framingham Heart Study. Circulation 1983;67: Hanis CL, Sing CF, Clarke WR, Schrott HG: Multivariate models for human genetic analysis: Aggregation, coaggregation, and tracking of systolic blood pressure and weight. Am J Hum Genet 1983; 35: Aristimuno GG, Foster TA, Voors AW, Srinivasan SR, Berenson GS: Influence of persistent obesity on cardiovascular risk factors: The Bogalusa Heart- Study. Circulation 1984;69: Clarke WR, Woolson RF, Lauer RM: Changes in ponderosity and blood pressure in childhood: The Muscatine Study. Am J Epidemiol 1986;124: Reisin E, Frohlich ED, Messerli FH: Cardiovascular changes after weight reduction in obesity hypertension. Ann Intern Med 1983;98: Moll PP, Burns TL, Lauer RM: The genetic and environmental sources of body mass index variability: The Muscatine Ponderosity Family Study. Am J Hum Genet 1991;49: