Clinical Science (1992) 82, 493-504 (Printed in Great Britain) 493

Platelet cytosolic free calcium concentration, total plasma calcium concentration and blood pressure in human twins: a genetic analysis

P. D. WILLIAMS, I. B. PUDDEY, N. G. MARTIN* and L J. BEILIN Department of Medicine. University of Western Australia. Perth. Western Australia. Australia. and *Queensland Institute of Medical Research. Brisbane. Queensland. Australia

(Received 17 October 1991 n January 1992; accepted 22 January 1992)

l. We used path analysis and maximum-likelihood model fitting to evaluate the relative contributions of genetic and environmental factors to the relationships observed between level of blood pressure and both total plasma calcium concentration and platelet cytosolic free calcium concentration in 109 twin pairs. 2. Total plasma calcium concentration was positively associated with systolic (r= 0.26, P< 0.00 1) but not diastolic blood pressure, a relationship which remained significant after adjustment for albumin, age and body mass index. A relationship between platelet cytosolic free calcium concentration and both systolic and diastolic blood pressure (r=0.17 and r=O.13, respectively, P~ 0.05) was no longer significant after adjustment for age and body mass index. 3. Additive genetic influences, unique environmental effects and age contributed to 60%, 30% and 10% of the variance in systolic blood pressure, respectively. Additive genetic effects explained 78% of the variance in plasma total calcium concentration and at least 48% of the variance in platelet cytosolic free calcium concentration in females and 37% in males. 4. Bivariate factor models provided evidence of genetic, but not environmental, co-variation of total plasma calcium concentration and systolic blood pressure, suggesting that a common genetic factor (or factors) contributes to their univariate relationship. In contrast, there was evidence of environmental, but not genetic, co­variation of platelet cytosolic free calcium concentration and systolic blood pressure, suggesting that some of the individual experiences specific to each twin may be caus­ing these two traits to vary together. S. The possible confounding effects of adiposity and environmental factors should be considered in future studies investigating the role of intracellular calcium levels in the pathogenesis of hypertension.

Key words: blood pressure. calcium. hypertension. platelets.

INTRODUCTION

The essential role of calcium as a second messenger in a number of the processes that regulate blood pressure, including regulation of muscle contraction, secretion of hormones and neurotransmitters, platelet aggregation and regulation of metabolic pathways, has made it an attrac­tive nexus for the myriad of biochemical alterations that are associated with essential hypertension. In a number of epidemiological studies, several derangements of calcium biochemistry have been described in association with increased levels of blood pressure. A positive association between blood pressure and total plasma calcium concen­tration in cross-sectional population studies [1-3 I has been one of the most consistent observations. Paradoxi­cally, serum ionized calcium levels have been reported to be lower [4, 51 in case-control studies of patients with essential hypertension. However, this has not been a consistent finding, with other investigators reporting either higher [2, 61 or normal [7,81 serum ionized calcium levels in hypertensive patients. These inconsistencies have been explained by one author in terms of differences in plasma renin activity [9). Other likely explanations include patient selection, inadequate matching of subjects and differences in methodology. More consistent observations have been those of elevated levels of circulating para­thyroid hormone and increased urinary calcium excretion [5, 10, 11 I, giving rise to a theory of possible renal calcium wasting in hypertensive patients [ 10, II).

As well as these abnormalities of extracellular calcium homoeostatic mechanisms, recent studies also suggest increased intracellular calcium levels in hypertensive patients, consistent with the hypothesis that a generalized membrane defect (possibly inherited) results in an increase in cytosolic free calcium concentration ([ Ca ~ + );)

in vascular smooth muscle, manifesting as increased peripheral vascular resistance 112). Given the inaccessi-

Abbreviations: [Cal' t. cytosolic free calcium concentration; DZ, dizygous; MZ, monozygous. Correspondence: Dr P. D. Williams. University Department of Medicine. Royal Perth Hospital. 3S Victoria Square. Perth. Western Australia. 6000. Australia.

494 P. D. Williams et al.

bility of human vascular smooth muscle, attention has focused on platelets as a potential model for events at the vascular smooth muscle level. Erne el al. (13) were the first to report that platelet ICa~+)i was significantly higher in hypertensive patients than in normotensive control subjects. A number of investigators have reproduced this finding [14-201, although the degree of overlap between hypertensive patients and normotensive subjects appears considerably greater than that first reported by Erne el al. 113).

These studies have left open the question of whether the increase in [Ca2 +]; in platelets is an epi-phenomenon secondary to an increase in blood pressure or a cellular marker of an underlying genetic abnormality in hyper­tensive patients. It is unlikely that the precise nature of the relationship between [Ca l +]; and blood pressure can be further clarified until the relative importance of genetic and environmental influences on ICal +)i are established. In the present study, the contribution of genetic and environmental factors to the variance of platelet [Cal + t. total plasma calcium concentration and blood pressure in healthy, normotensive twins has been examined. In addition, we have investigated the relative importance of these factors in determining the inter-relationships between blood pressure and intra- or extra-cellular calcium levels.

METHODS

Subjects

One hundred and nine pairs of monozygous (MZ) and dizygous (OZ) twins aged between 17 and 65 years were recruited through local community newspaper advertising and the Australian National Health and Medical Research Council Twin Registry. Subjects were screened to exclude any pairs in which one or both twins were receiving drug treatment for chronic diseases including hypertension. cardiac failure, angina pectoris or diabetes mellitus. Subjects with renal or liver impairment or a history of myocardial infarction were also excluded. The protocol for this study was approved by the University of Western Australia's Committee for Human Rights. and informed consent was obtained from all participants.

Study design

Both members of a twin pair attended on the same day on two occasions, 7 days apart. Pairs were seen between 07.00 and 10.00 hours. On the first visit. a questionnaire designed to assess zygosity was administered to each twin independently [21). In a subset of 42 like-sexed twin pairs examined by DNA fingerprinting [22), one pair was misclassified by the questionnaire, indicating a likely error rate of 2A'X •. The misclassified twin pair were reclassified to the correct zygosity group before genetic analysis. Our standard health and lifestyle questionnaire was also administered to each twin [23). Finally. a fasted blood sample was obtained from each subject, without stasis and after 5 min recumbency. On the second visit, suhjects

were weighed using a calibrated beam balance and measured for height with a stadiometer. Blood pressure was recorded at least 10 min after arrival at a clinic by semi-automatic, non-invasive sphygmomanometry (Dinamap Vital Signs Monitor 845XT; Critikon Inc, Tampa, FL, U.S.A.) with readings taken at 2 min intervals for 20 min with subjects resting supine. After discarding the first reading for each subject, systolic and diastolic blood pressures were calculated as the means of 10 supine measurements.

Measurement of [CaH ];

Platelet and lymphocyte [CaH 1 was measured by a modification of previously published methods [24, 25). Samples from twin pairs were assayed in parallel. Whole blood was collected into acid citrate dextrose (75 mmol/I trisodium citrate, 38 mmol/I citric acid and 139 mmol/I D­

glucose, 1-9 voL). Cells were separated by a two-step zonal centrifugation process (Ficoll-Paque followed by a sucrose gradient) (26). The leucocyte suspension con­tained at least 90% mononuclear leucocytes. Assuming a cell volume of 230 fl for lymphocytes and 7 fl for plate­lets, platelets contributed approximately 4.9% of the total cell volume to the lymphocyte suspension. The platelet suspension contained one lymphocyte for every 167 platelets on average. This is equivalent to approximately 20% of the total cell volume of the platelet suspension. These results are in accordance with other studies (27).

Platelet and lymphocyte suspensions (I x lOx cells/ml and 2 x 106 cells/ml, respectively) were incubated with 1 !lmol/I Fura-2 acetoxymethyl ester in dimethyl sulph­oxide at 3rC for 30 min in duplicate. Cells were washed to remove extraneous dye and were resuspended and equilibrated for 30 min in a simplified saline (145 mmol/I NaCl) buffer with 1 mmol/I calcium. Fluorescence was measured in a 3000 u.v.-visible spectrofluorimeter (Hitachi. Tokyo, Japan) using a thermostatically con­trolled (3rC) quartz cuvette with a magnetic stirrer at 340 and 380 nm excitation wavelengths and a fixed emission wavelength of 500 nm. [Ca2 +); was calculated as described by Grynkiewicz [24] based on the equation:

[Ca2 +)i = Kd x [(R - RminJ/(Rmax. - R)] x b

where R is the ratio of the fluorescence of the intact cell suspension at 340 and 380 nm, Rmin. and Rmax. are the ratios for the fluorescence intensities of the Fura-2 free acid at the same wavelengths in nominally zero Ca2 + and at saturating Ca2+, respectively, and b is the ratio of fluorescence of Fura-2 free acid at 380 nm in zero and saturating Cal +. The dissociation constant (Kd ) of the Cal + -Fura-2 complex was assumed to be 224 nm at 3rC in the cytosolic environs.

Correction for autofluorescence was made by subtract­ing the fluorescence of an equal concentration of unloaded cells plus dimethyl sulphoxide from the cells loaded with Fura-2 under the same conditions. [Ca~ + /; is expressed in nmol/I and the values given arc the means from duplicates. For platelets and lymphocytes, intra­assay variability was approximately 5°/.. and 6% and inter-

Platelet calcium and blood pressure in twins 495

assay variability was approximately 9% and 7%, respectively.

Other biochemical analyses

Total plasma calcium and plasma albumin concentra­tions were analysed on a SMAClI autoanalyser, using Technicon reagents and methods (Bayer Diagnostics, Melbourne, Australia), by the staff of the Department of Biochemistry at the Royal Perth Hospital.

Statistical analysis

Much of the preliminary statistical work was carried out using the Statistical Package for the Social Sciences (SPSS Inc., U.S.A.). Before analysis, the frequency distri­bution of each variable was examined to determine if any deviations from Gaussian distributions occurred. Platelet [Ca2 + Ii was not normally distributed and was reciprocated to normality.

Associations between phenotypes were analysed by Pearson product-moment correlations and multiple linear regression. In order to explore the possible relationships between platelet {Ca~+li and blood pressure, we have regarded the observations in this study as independent. However, this is not strictly true, since the data are from twin pairs. This means that the correlations calculated are usually correct but the estimates of these correlations are deflated. Therefore, the P values are less than conserva­tive.

Using the health and lifestyle questionnaire, a number of lifestyle variables encompassing such traits as alcohol consumption (ealculated in ml of ethanol per week from a retrospective diary of the number and volume of all drinks consumed in the preceding week), level of physical activity (assessed from self-reported number of days per week on which a variety of vigorous physical activities were undertaken), marital status, education, smoking status and caffeine consumption (cups of tea and/or coffee consumed per week) were considered as poten­tially relevant to blood pressure. A screening step which searched for associations between these categorical life­style variables and the continuous dependent variables was performed using Spearman rank correlations. Each variable was considered for further evaluation if the correlation was significant at the 10% level. The next step involved regressing a given phenotype on the lifestyle variables selected from the univariate correlations. The variables were entered in a stepwise manner retaining only those terms significant at the 5% level.

Genetic analysis

When the 109 twin pairs were grouped by sex and zygosity, there were 44 MZ females, 31 DZ females, 14 MZ males, nine DZ males and II DZ pairs of opposite sex. One-way analysis of variance and group contrasts were used to assess differences in means across the five twin groups. Male-female and MZ-DZ differences in variance for each variable were assessed by using r·:'tests. The presence of genotype-environment interaction was

assessed by examining the correlation of the absolute intrapair differences, a measure of environmental influences within families, with the absolute intra pair sums, a measure of genetic and shared environmental effects, in MZ twins for each variable [28).

Path analysis and maximum-likelihood model fitting were used to partition and evaluate the contribution of genetic and environmental factors to the total phenotypic variance for each trait. The methodology was based upon the standard biometric model described by links & Fulker (28) and Eaves et al. (29). The raw data from the five twin groups were summarized into co-variance-variance matrices using the pre-processor package PRELlS 1.12 (30). A series of models of genetic and environmental contributions was specified and fitted to the co-variance matrices using L1SREL 7.16, a program for linear struc­tural equation modelling (31, 32).

Path analysis was used to specify the expected contri­butions of genes and environment to the observed co­variances and variances of the five groups of twins. Fig. 1 represents a simple univariate path diagram including three possible sources of variation. These are: A, additive genetic variance, i.e. the sum of the average effects of all the genes that influence a phenotype; C, variance due to the environmental effects common to both members of a twin pair; E, variance due to environmental effects unique to each twin. Basically MZ twins share all the genetic effects and therefore the correlation between genotypes of MZ twins for 1.0. Assuming that mating is random and that genetic non-additivity, i.e. dominance or epistasis, is not present, DZ twins share on average half the additive gene effects and have a genetic correlation of 0.5. The correlation between common environmental effects is 1.0 for MZ and DZ twins, since the common environment should act equally on each twin and on both twin types. The unique environmental effects, which include measurement error, are uncorrelated as indicated by the absence of a connecting arc.

Maximum-likelihood estimation was used to fit the data to each model. It provides a X2 test of the overall goodness of fit of the data to the model and estimates of the model parameters. The overall X2 tests the agreement between the observed and the predicted statistics. A large X2 indicates a poor degree of fit. If the model fits well then the data are consistent with the model. However, if the model fits badly then it is rejected in light of the data. A full model accounting for additive genetic effects and both common and unique environmental effects was fitted and a test of the overall goodness of fit was examined. If the model fitted. then the relative importance of A and C were assessed by dropping each parameter in turn from the model and determining if there was a significant reduction in the goodness of fit. Age, a potential source of variation contributing to the difference between, but not within pairs, was included where appropriate [331.

This method led to a preferred, most parsimonious model, estimates of the variance attributable to each parameter (A, C. £ and age) and a test of the significance of each. The parameters are expressed as the estimated proportion of the total phenotypic variance which was

496 P. D. WiUiams et al.

1.0 (MZ) 0.5 (OZ)

~ AI .....----...... A2

EI CI " E2

\/ \lu

PI 1'2

Genes

Environment

Phenotype

Fig. I. Simple univariate path diagram induding three possible sources of variation for same-sex twin pairs. The phenotypes of the [wins (PI. 1'2) are modelled as being determined by additive effects (AI. A2). environmental effects common to both twins (CI. C2) and environ· mental effects unique to each twin (E I. E2). The correlation of genetic effeas on twin I and twin 2 is 1.0 for MZ twins and 0.5 for OZ twins. The correlation of common environmental effeas is 1.0 for both MZ and OZ twins. The unique environmental effects are uncorrelated.

calculated by dividing the square of the estimate of each parameter by the sum of the square of all the estimates.

Differences in genetic and environmental influences between the sexes were tested by deriving estimates for each sex separately and then constraining the estimates to be equal across the two groups. If the residual Xl of the model obtained from the males and females jointly was significantly greater than the sum of the residual Xl of the males and females separately, this indicated sex hetero­geneity, i.e. that the estimates of genetic and environ­mental influences were different for males and females. When sex heterogeneity was indicated, the analyses were performed on each sex separately. When there was no significant difference in the estimates between males and females, the DZ pairs of opposite sex were included in the analysis.

RESULTS Means, 95% confidence intervals and tests for sex

differences are presented in Table 1. The males were sig­nificantly younger, heavier and taller than the females. The blood pressures ranged from 96/53 mmHg to 144/ 92 mmHg. Although systolic blood pressure was signifi­cantly higher in males than in females, there was no sex difference in the means for diastolic blood pressure (Table I).

Systelic blood pressure was positively associated with age and weight in females (r = 0.34, I' < 0.00 1 and r=0.20, 1'=0.014, respectively) and showed similar, although non-significant, trends in males (r= 0.22, 1'=0.101 and r=0.16, 1'=0.246, respectively). Diastolic blood pressure was positively associated, in both males and females, with age (r=0.62, 1'<0.001 and r=0.49, 1'<0.001, respectively) and weight (r=0.16, 1'=0.046 and r=0.24, 1'=0.070, respectively).

The methodology for determination of [Cal + I. employed in this study did not correct for possible leakage of Fura-2 from the cell into the extracellular

medium. As a consequence, the absolute levels of platelet [Ca2 + Ii may be artificially elevated. However, the absolute levels of platelet [Ca2+L seen in this study are similar to those reported in other studies investigating normotensive singletons [13, 171 and the associations observed in this study are not appreciably different from those reported by groups which have [19. 34) or have not [20, 35J corrected for the possible leakage.

There was no discernible sex difference in means for platelet [Cal + L between males and females (Table 1). Platelet (CaH Ii wa.. negatively associated with total plasma calcium concentration in females and tended towards a similar association in males (Table 2). There was no association between platelet (Ca2+1 and plasma albumin concentration in either sex (Table 2).

In the whole population, platelet [CaH l was positively associated with both systolic blood pressure (r=0.166, 1'=0.015) and dia .. tolic blood pressure (r=0.134, £>=0.05). The relationship with systolic blood pressure largely resulted from the positive association seen in females (Table 2}.ln males, platelet [Cal + 1 was unrelated to systolic blood pressure but showed a positive. although non-significant, correlation with diastolic blood pressure (Table 2).

Platelet [Ca2 +1 in females was also positively related to weight and body mass index (Table 2). After adjustment for age and body mass index by multiple regression, the relationship between platelet [Ca2+Ji and systolic blood pressure in females was attenuated (£>=0.075). When this analysis was repeated for males and females combined, after adjustment for sex, age and body mass index, platelet [Ca2+li was not a significant predictor of the variance in either systolic or diastolic blood pressure.

Lymphocyte (Ca2+); was strongly correlated with platelet [Ca2 +li (r=0.354, £><0.001) but the correlations with both systolic (r=0.124, £>=0.07) and diastolic (r= 0.112, P= 0.087) blood pressure did not achieve con­ventionallevels of significance.

Total plasma calcium, plasma albumin and albumin­adjusted total plasma calcium concentrations were signifi­cantly greater in males than in females (Table 1). Total plasma calcium concentration was positively related to systolic blood pressure in the total population (r=0.261, P<O.OOl) and in females (r=0.220, P=0.005) and followed the same direction for males (r=0.138, not sig­nificant). Diastolic blood pressure was not significantly correlated with total plasma calcium concentration in the total popUlation (r=0.03, not significant), males or females (r= -0.086, not significant, and r= - 0.094, not significant, respectively). After adjusting for the effects of albumin, age and body ma<;s index in a multiple regression analysis of the data for females, total plasma calcium con­centration remained an independent predictor of systolic blood pressure (1'=0.055). When this analysis was repeated on the combined data set with sex included as a dummy variable, total plasma calcium concentration was a significant predictor of systolic blood pressure (1'= 0.020).

In stepwise multiple regression analysis including the lifestyle variables, age, sex and heavy alcohol consumption

Platelet calcium and blood pressure in twins 497

Table I. Desaiptive variables for males and females. Values are means and 95% confidence intervals with n equal to the number of observations. Independent two-tailed t-testS were used to test for signifICant differences in means between males and females taking group variances into account. Statistical significance: • P '0.05 ••• P '0.0 I •••• P '0.00 I.

Mean

Age (years) 33.1 Weight (kg) 76.0

Body mass index (kg/ml) 24.4 Height (m) 176.6

Systolic blood pressure (mmHg) 1219

Diastolic blood pressure (mmHg) 68.7

Platelet [CaH ); (nmolll) 90.2 Lymphocyte [Cal +]. (nmolll) 84.5 Total plasma calcium concn. (mmolll) 2.39

Plasma albumin concn. (gil) 45.6

Albumin·adjusted plasma calcium conen. (mmolll)

Table 2. Pearson product-moment correlations (r) between platelet [Cal+]1 and other variables in males and females. n is the number of observations. Statistical significance: ·P'0.05. ··P'O.OI.

Males females

n n

Age (years) 0.013 57 0.108 156 Weight (kg) -0.048 57 0.248" 154 Body mass index (kg/ml) -0.086 57 0.187' 154 Systolic blood pressure (mmHg) 0.013 57 0.219** 156 Diastolic blood pressure (mmHg) 0.181 57 0.117 156 Total plasma calcium concn. (mmolll) -0.130 57 -0.158* 156 Plasma albumin concn. (gil) -0.107 57 -0.047 156

(i.e. > 280 ml ethanol per week) were the only significant predictors of systolic blood pressure (F3•203 = 18.34, P<O.OOI, R2=0.21). Age, heavy alcohol consumption and regular vigorous exercise (inversely related) were significant predictors of diastolic blood pressure (fJ.203=30.38, P<O.OOI, R2=0.31). Despite the signifi­cant positive univariate relationships of body mass index with both systolic and diastolic blood pressure, body mass index was not an independent predictor of blood pressure in multivariate regression. However, given that body mass index may be a confounder of both exercise and alcohol consumption, i.e. people who exercise are leaner and people who drink are heavier, the regression analysis was repeated with body mass index and age preforced into the equation. Correcting for the effects of obesity, removed the significant effect of heavy drinking on systolic blood pressure. It did not affect the predictors of diastolic blood pressure.

Lifestyle factors with a univariate correlation with platelet [Ca2 + Ii significant at less than the 10% level included tertiary versus non-tertiary education (r=0.14,

2.27

Males females

95% confidence n Mean 95% confidence n interval interval

30.4.35.8 57 37.6' 35.4. 39.8 161

73.5.78.5 57 60.5'" 59.1.61.9 159

23.6.25.2 57 23.0" 22.4.23.6 159

174.8. 178.4 57 162.3'" 161.3. 163.2 159 120.7. 125.1 57 116.2**' 114.6. 117.8 161

66.2.71.2 57 69.0 67.8.70.2 161

85.5.95.4 57 90.2 86.9.93.7 156

81.6.87.4 57 84.2 82.6.85.8 159

2.37.2.41 57 2.34**' 2.32.136 160

45.0.46.2 57 44.4'" 44.0,14.8 160

2.265. 2.275 57 2.25* 2.245.2.255 160

P=0.044) and tea consumption (r= -0.12, P=0.076). None was found to be significant independent predictors of platelet [Ca2+1i when included with age in a stepwise multiple regression analysis.

Total plasma calcium concentration was not associated with any of the lifestyle variables.

Genetic analysis

Using one-way analysis of variance and group contrasts there \Vas no significant difference in means or variances between MZ and DZ pairs of the same sex for any variable. There was no indication of genotype-environ­ment interaction for any variable. There was a significant correlation between the absolute MZ intrapair difference and the MZ intrapair sums for total plasma calcium concentration. However, this interaction could be attributed to the choice of scale since it was removed by logarithm 10 transformation of the data. The genetic analysis for total plasma calcium concentration was performed on the transformed data. The Pearson correla­tion coefficients and the variance-co-variance matrices (twin 1 versus twin 2) for all variables by zygosity and sex are listed in Table 3. The finding of larger correlations for the MZ twin pairs in comparison with DZ twins of the same sex is consistent with a significant contribution of genetic effects to their variance. This was analysed further by path analysis and maximum-likelihood model fitting, the results of which are listed below for each phenotype.

Systolic blood pressure. A sex difference in means is not important with regard to the genetic analysis as the PRELlS program generates co-variance matrices in which deviations are measured from their own sex mean. There was no indication of sex heterogeneity (X 2 = 4.39, 4 degrees of freedom, P= 0.356) and so all five twin groups were included in the univariate genetic analysis for systolic blood pressure. Age contributed significantly to the variance in systolic blood pressure for females

498 P. D. Williams et al.

Table 1. Co-variance matrices and correlations. Abbreviations: TI, twin I; n, twin 2; DZF, dizygous female pairs; MZF, monozygous female pairs; DZM. dizygous male pairs; MZM. monozygous male pairs; DZOS, dizygous opposite sex pairs. The blood pressure variables have age included in the matrices (i.e. 1 xl). Variances are given on the leading diagonal of each of the 3 x 3 or 2 x 2 matrices, co-variances in the lower triangle and correlations (bold italics) in the upper off-diagonal entries.

Systolic blood pressure Diastolic blood pressure Platelet (Cal +]. T olal plasma calcium concn.

Age TI T2 Age TI

DZF Age 201.899 0.28 0.22 201.899 0.58 TI 31.627 61.272 0.36 65.601 64.456 T2 29.489 27.074 90.167 43.549 29.909

MZF Age 172.133 0.43 0.52 172.133 0.49 TI 57.623 103.894 0.74 50.401 61.855 T2 68.795 75.754 101.351 39.460 36.923

DZM Age 95.694 0.40 0.13 95.694 0.77 TI 35.153 79.639 0.01 68.865 83.184

T2 7.718 0.174 39.177 28.872 40.509

MZM Age 107.670 0.34 0.57 107.670 0.63 TI 28.619 67.164 0.77 63.575 94.994 T2 46.080 49.352 60.457 77.781 89.385

DZOS Age 177.055 0.09 0.26 177.055 0.64 TI 10.495 82.013 0.35 75.780 79.596 T2 33.575 30.552 90.801 73.142 11.030

(Xl = 14.12, 1 degree of freedom, P<O.OOl) with a similar trend in males (X2 = 3.1, 1 degree of freedom, P= 0.078) and was accordingly included as a parameter in each modeL The model which contained only unique environ­mental effects could not adequately explain the data (X2=60.87, 27 degrees of freedom, P<O.OOl). The full model, which included age, additive genetic influences and common and unique environment effects, provided a satisfactory fit to the data (X 2 = 26.74. 25 degrees of freedom, 1'= 0.369). The reduced model from which genetic effects were excluded had a significant loss of fit compared with the full model (X2 = 12.22, 1 degree of freedom, 1'= 0.00 1), indicating a significant additive genetic contribution to the variance in systolic blood pressure. When the common environmental effects were excluded from the full model, there was no loss of fit (X2 = 0.00), suggesting that common environmental influences were not significantly different from zero and therefore, need not be included in the most parsimonious model. Since the DZ correlation was less than half the MZ correlation for systolic blood pressure (Table 3), dominance should be included in the model. The effects of genetic dominance and common environment are con­founded in data from twin pairs reared together [36.371 and therefore these effects cannot be estimated simul­taneously. A model which included additive genetic effects, dominance effects and unique but not common environmental influences was fitted to the data

T2 TI T2 TI T2

0.44 0.54 4.881 0.64 0.197x 10'1 O.SO

47.555 3.620 6.594 0.110 x 10'1 0.245 X 10-1

0.42 0.66 6.965 0.90 0.244 x 10-1 0.75

50.963 6.284 6.934 0.152xlO-1 0.168xI0- 1

0.30 0.46 4.690 0.82 0.112 x 10-1 0.63

94.244 4.003 5.023 O.ll2xIO- 1 0.284 x 10-1

0.74 0.90 3.333 0.96 0.841 x 10". 0.69

102.954 3.077 3.073 0.73IxI0-· 0.135 x 10-1

0.5/ 0./2 10.449 0.70 0.209 x 10'1 0.07

114.791 4.831 4.604 0.209 x 10-· 0.389 X 10-1

(X2 = 24.59,25 degrees offreedom, 1'=0.486). However, there was no significant evidence for dominance (X 2 =2.15, 1 degree of freedom, 1'=0.143) and so the model of best fit included age, additive genetic effects and unique environmental influences (X2 = 26.74, 26 degrees of freedom, 1'= 0.423).

Table 4 reports the model of best fit for each variable, the percentages of the variance attributable to each para­meter in that model and the likelihood ratio tests of sig­nificance for each parameter. Additive genetic effects accounted for 60% of the variance, 10% was due to age and the remaining 30% was attributed to unique environ­mental factors.

Diastolic blood pressure. For both diastolic blood pressure and platelet [Ca2 + t there were significant dif­ferences in variance between the sexes (Fi2.15~ = 1.66, /'= 0.037, and F.<!>.I!>II = 1.70, /'= 0.0 11. respectively). A significant difference in variance between the sexes may indicate that genetic or environmental factors are acting unequally upon these variables. Therefore, data for males and females for both these variables were analysed separately.

With the data for diastolic blood pressure in males. the pattern followed that seen for systolic blood pressure. The model which best described the data contained age, additive genetic influences and unique environmental factors (X 2 = 6.10, g degrees of freedom, /'= 0.636 J. Additive genetic effects explained 52(1'0 of the total

Platelet calcium and blood pressure in twins 499

Table 4. Most appropriate model for each variable and per. centage of the variance attributable to each parameter in that model. Abbreviations: SBP. systolic blood pressure; DBP. diastolic blood pressure; A. additive genetic variance; C. common environmental variance; f. unique environmental variance. likelihood ratio tests of significance for each parameter are included. Statistical significance: ··P~O.OI. ···P~O.OOI.

Model of Variance (% of the total) best fit

Age A E

SSP Age+A+E 10'" 60'" 30'"

DBP Females Age+A+E 23'" 43 14'" Males Age+A+E 38'" 52**' 10'"

Platelet [Cal + ),

Females A+E+C 48'** 10'*'

Males A+E+C 37*' 3***

Total plasma calcium concn. A+E 78'" n'"

C

42

60

variance, 38% was explained by age and 10% was attributed to environmental factors specific to the individ­ual (Table 4 ).

With respect to the data for females, the contribution from age was again significant (X2 = 24.96, 1 degree of freedom, P<O.OO1). The full model containing all para­meters explained the data well (X 2 =2.87, 7 degrees of freedom, P= 0.897); however, excluding A or C did not result in a significant loss of fit (X 2 = 1.07, 1 degree of freedom, P=0.301 and X2 =0.38, 1 degree of freedom, P= 0.538, respectively). Excluding both parameters led to a significant loss of fit (X2 = 20.83,2 degrees of freedom, p< 0.001). This indicates that there is a significant family contribution but there is insufficient power to ascribe the variance to either genetic or common environmental effects. Since dropping A from the model gave a slightly better X2 and considering the models chosen for the blood pressures thus far, the preferred model was that which included age, unique environmental and additive genetic variance (X 2=3.25, 8 degrees of freedom, P=0.918). Age, genetic influence and unique environmental effects accounted for 23%, 43% and 34% of the variance in diastolic blood pressure in this female population, respectively (Table 4 ).

Platelet reaH /;. The complete model with A, Cand E was the model of best fit for both females and males (Xl = 1.49, 3 degrees of freedom, P= 0.685 and X2 = 0.73, 3 degrees of freedom, P= 0.867, respectively). Excluding A from the data for females and males incurred a signifi­cant loss of fit (X2 = 12.09, 1 degree offreedom, P< 0.00 1 and X2 = 8.04, 1 degree of freedom, /'=0.005, respec­tively), indicating a significant genetic component of the variance. Excluding Cdid not result in a significant loss of fit in either females or males (X2 = 2.76, 1 degree of frcedom, /'= 0.100 and X2 = 2.0 I, I degree of freedom, /'= 0.156, respectively). We chose to include the common environmental effects in the model of best fit for both sexes, since its contribution to the variance was relatively

large and approached statistical significance. In females, additive genetic effects accounted for 48% of the vari­ability of platelet [Cal + t. while unique environmental factors accounted for 10% and a common environmental influence for 42% (Table 4). In males the estimates were 37%, 3% and 60%, respectively (Table 4).

Total plasma calcium concentration. Total plasma calcium concentration was not associated with age in females (r=0.04, P= 0.582). Although there was a signifi­cant negative association between total plasma calcium concentration and age in males (r= -0.39, /'=0.003), age did not contribute significantly to the variance of total plasma calcium in males (Xl = 0.72, 1 degree of freedom, /'=0.396). All twin groups were included in the analysis. since there was no indication of sex heterogeneity (X 2 = 2.04, 2 degrees of freedom, /'= 0.361). The full model explained the data (X 2 = 18.34, 12 degrees of free­dom, /'= 0.106). Excluding the additive genetic effects resulted in a significant loss of fit (X2 = 13.71, 1 degree of freedom, P< 0.00 1), whereas excluding the common environmental influence led to no loss of fit (X 2 = 0.00). The DZ correlation was less than half the MZ correlation for total plasma calcium concentration (Table 3); how­ever, a formal test for dominance was not significant (X2 = 0.20, 1 degree of freedom, P=0.655). Therefore, the model which described the data best contained A and E (X2= 18.34, 13 degrees of freedom, P=0.145). Additive genetic effects accounted for 78% and unique environmental effects accounted for 22% of the vari­ability in total plasma calcium concentration (Table 4).

Bivariate analysis

The methods used to estimate the genetic effects in a trait can also be used to determine whether some or all of the factors that contribute to the variance in 'one trait could also contribute to the variance in the other trait. In particular, we wish to test for the presence of genetical co­variance between two traits and to determine whether this arises because one trait influences the other, vice versa, both or neither. Fig. ,2 shows a path diagram for a general bivariate factor model. Only the model for one of the twins is depicted and the whole structure is replicated for the second twin.

Bivariate models were specified for platelet [Ca2 +1i and systolic blood pressure, and for total plasma calcium con­centration and systolic blood pressure. Only the data from females were used, since the small number of males was subject to large sampling errors and their inclusion may have led to spurious results. In addition, the inclusion of such small numbers of pairs would not appreciably increase the resolving power of the study.

Table 5 shows the bivariate models fitted to the data for platelet [Ca2 +1i and systolic blood pressure in females. The bivariate models included age, genetic effects and unique environmental influences for both variables and common environmental effects for platelet [Ca2 + Ii' When the genetic correlation was excluded from the model (hI! = 0), there was no significant loss of fit (X 2 = O.{)4. I degree of freedom, /'= 0.842). indicating no genetic co-

500 P. D. Williams et al.

G ~8~'1"" ..

0°0·-V V I MZIO.SOZ

j'

' .... 1 ___ _

Fig. 2. A general bivariate factor model for the resemblance of twin pairs for two traits. This model includes a general additive genetic factor. A,. which influences trait I (platelet [ea2+];) and trait 2 (systolic blood pressure) to the extent hc and he and a general common environmental factor. C" which has the corresponding paths 'c and 'e. In addition. this model allows for a general unique environmental factor. f,. which influences the two traits to the extent ec and ee. and a general age effect with paths Oc and 0B' Each of these influences are common to both phenotypes and would dictate that the two traits are perfectly correlated. Therefore, we also allow for the possibility that there are influences specifIC to one or other trait. Unfortunately. in the bivariate case. it is not possible to estimate loadings on a common factor and on two specific influences. so we only estimate one set of specifics. Since this choice is arbitrary. the specific unique environment. common environment and additive genetic influences (E,. C. and A •• respectively) were placed on blood pressure. Because of this arbitrariness. less interest is placed on the values of the parameter estimates than on their statistical significance. In addition. paths ; and ;' represent the direct effects of the first trait on the second trait and vice versa. In this Figure only the model for one of the twins is depicted with the whole structure replicated for the second twin. The full path diagram would have a double-headed arrow indicating a correlation between the additive genetic sources of variance, both AI and A,. for both twin I and twin 2. with a value of 1.0 for HZ twins and 0.5 for DZ twins. Similarly. the general and specific common environmental sources of variance are correlated 1.0 for HZ and DZ twins. By definition. the unique environmental effects are uncorrelated. Various hypotheses may be tested by setting different combinations of these paths to zero and observing the effect on the fit. This is formally tested by Xl likelihood ratio tests or the difference between Xl goodness of fit of the full and reduced models.

va nation between platelet [CaH)i and systolic blood pressure. When ell = 0, i.e. excluding the unique environ­mental correlation, there was a significant loss of fit (Xl = 3.63, 1 degree of freedom, P= 0.056), indicating the presence of unique environmental co-variation between platelet [Ca~ +); and systolic blood pressure.

nate between the factor and direction of causation models.

The bivariate models for total plasma calcium concen­tration and systolic blood pressure contained age, genetic effects and unique environmental influences (Table 6). The simple bivariate factor model explained the data well in females. When the genetic effects and unique environ­mental correlations were excluded from the model simul­taneously, there was a tendency towards a significant loss of fit (Xl = 4.63, 2 degrees of freedom, /'= 0.099), implying co-variation of our two phenotypes. Excluding the unique environmental correlation did not lead to a loss of fit (X1=0.83, I degree of freedom, /'={).362). Excluding the genetic correlation led to an increase in X1, which was not significant (Xl = 2.39, 1 degree of freedom, /'=0.122). This observation of a greater increase in X1 when the genetic correlation was excluded is evidence suggestive, although not unequivocally, of genetic co-

The second type of bivariate model tests a more restrictive hypothesis, that platelet [Cal + li has a direct effect on systolic blood pressure, and the alternative hypothesis, that systolic blood pressure causes variation in platelet [CaH li' This is done by allowing j and j' to be free while setting hll = ell = ell = O. This reciprocal causation hypothesis also described the data well (Table 5). Further to this, we tested the individual path sub­models to determine which is more likely to be the true direction of causation. Table 6 shows that both direction of causation submodels fit the data well. We do not have the statistical power in this small population to discrimi-

Platelet calcium and blood pressure in twins 501

Table 5. Bivariate models fitted to the platelet [Cal+ 1. and systolic blood pressure data from the female twin pairs. This Table includes the Xl 'goodness of fit' index and the specific submodels to test for co-variation, genetic co-variation and unique environmental co-variation. Hypothesis I: 'co·variation' (model I versus model 4), likelihood ratio Xl = 6.92, 3 degrees of freedom, P = 0.074. Hypothesis 2: 'genetic co-variation' (model I versus model 2), fikelihood ratio Xl=O.04. I degree of freedom, P=0.842. Hypothesis 3: 'unique environmental co-variation' (model I versus model 3). likelihood ratio Xl = 3.63. I degree of freedom, P =0.056. Abbreviation: SSP, systolic blood pressure.

Model i 'goodness Degrees of P value of fit' index freedom

I. Simple bivariate (i=i' =0) 12.18 19 0.878

2. Set hs =0 12.22 20 0.908

3. Set es =0 15.81 20 0.728 4. Seths=(s=es=O 19.10 22 0.639

5. Reciprocal causation 12.34 20 0.904 (set hs =(s = es =0)

6. SSP> platelet [Cal< 1( i = 0) 12:46 21 0.926 7. Platelet [Cal + Ii> SSP (i' = 0) 13.62 21 0.885

variation between total plasma calcium concentration and systolic blood pressure. Similar to the platelet rCa2 +J; results, each of the direction of causation models fits the data well, but with insufficient resolving power to discriminate between the factor and direction of causation bivariate models.

DISCUSSION

We have found a positive association between systolic blood pressure and both platelet [CaH]i and total plasma calcium concentration in 109 pairs of normotensive twins. This observation has extended previous studies of this relationship by allowing precise definition and quantifica­tion of the contribution of additive genetic effects to each of these phenotypes.

The magnitude of the cross-sectional relationships between platelet [CaZ +]i and blood pressure reported in this study were quite small. We have already mentioned that caution is indicated in the interpretation of these associations because our twin subjects could not be con­sidered a random sample and the observations within a twin pair cannot, in the strictest sense, be considered independent. However, similar caution is necessary in any assessment of previous reports, where cross-sectional relationships have generally been determined from case­control studies. Such an approach combines arbitrarily selected hypertensive patients and normotensive subjects who have already been clearly demarcated by a deciding variable, making analysis of linear relationships difficult. In fact. no true population-based studies of the relationship between platelet [Ca! +]; and blood pressure have been conducted and in case-control studies, the entire spectrum from very weak to extraordinarily high correlations have been reported. No study has been able to

reproduce the strong correlations first reported by Erne

Table 6. Bivariate models fitted to the plasma c:aIdum concentra­tion and systolic blood pressure data from the female twin pairs. This Table includes the Xl 'goodness of fit' index and the specifIC submodels to test for co-variation, genetic co-variation and unique environmental co-variation. Hypothesis I: 'co-variation' (model I versus model 'I), likelihood ratio Xl =4.63, 2 degrees of freedom, P=0.099. Hypothesis 2: 'genetic co·variation' (model I versus model 2), likefihood ratio Xl=139, I degree of freedom, P=0.122. Hypothesis 3: 'unique environmental co·variation' (model I versus model 3), likelihood ratio Xl = 0.83, I degree of freedom, P = 0.362. Abbreviation: SSP, systolic blood pressure.

Model X I 'goodness Degrees of Pvalue of fit' index freedom

I. Simple bivariate (i = i' =0) 13.93 21 0.872 2. Set hs =0 16.32 22 0.800 3. Setes=O 14.76 22 0.872

4. Set hs=es=O 18.56 23 0.727 5. Reciprocal causation 13.93 21 0.872

(set hs =(B = eB =0)

6. SSP> total plasma calcium 13.97 22 0.902 concn. (i=O)

7. Total plasma calcium concn. > SSP 1·4.21 22 0.894 (i'=O)

et al. [13]. These approached unity and it has been com­mented [20] that such correlations were stronger than those seen for blood pressure between two visits in anyone. individual. In contrast, Hvarfner et al. [17], after correct­ing for the influences of parathyroid hormone, ionized CaH and body mass index, estimated that [CaH]i levels could only explain 5% of the variability in mean blood pressure. In general weaker relationships, as observed in the present study, have been shown.

We also report a positive correlation between platelet [CaH ]; and both body mass index and weight. Prior to this report there has been the suggestion of an association between platelet [Caz+]; and weight [38], and an observa­tion of a decrease in platelet [Ca2 +]i in obese subjects on a weight-reducing diet [39]. Recently, a positive association between obesity and erythrocyte [Ca2 +]i' as measured by '~F-n.m.r. has been reported [40]. Contrary to this, a negative association between erythrocyte total calcium content and both age and weight has also been observed [41]. However, these latter results may be specific for the total calcium assay, which is physiologically less meaningful than [Caz+];.

After correcting for any effects of age, sex and body mass index, we were no longer able to demonstrate a relationship between platelet [Ca2 +1 and blood pressure. However, with no a priori hypothesis regarding the relationships betwecn platelet [Ca~ + ]i' obesity and blood pressure, the inclusion of these variables in multiple regression models may be obscuring a true association, as the effect of adiposity on blood pressure may ultimately be mediated by changes in [Cal + Ii levels. Our study and others [411 have raised the possibility that weight and/or obesity may have contributed to the differences often observcd in case-control studies where subjects have not

502 P. D. Williams et at.

been matched for weight or body mass index. In support of this, the one study in which hypertensive patients and normotensive subjects were well matched for weight showed no significant difference in platelet [Caz+)i between these groups (38).

Using path analysis and maximum-likelihood estima­tion procedures, this study has demonstrated significant genetic influences for systolic and diastolic blood pressure, platelet [Ca2 +)i and total plasma calcium con­centration. Twin studies have now been utilized for over half a century [42) to examine the interplay of genetic and environmental factors influencing blood pressure vari­ability. Overall, studies have suggested that genetic effects playa fairly major role, contributing approximately 50% to the individual variability of blood pressure in humans [43-47). We have confirmed these findings in our study, where additive genetic effects were found to account for 60% of the variability in systolic blood pressure in both males and females and 53% and 43% of the variability in diastolic blood pressure in males and females, respec­tively. In support of some [45, 48) but not all [43) studies, we have shown that additive genetic effects were less influential to diastolic than systolic blood pressure vari­ability. Indeed, while age was shown to contribute approximately 10% to the variability in systolic blood pressure, this increased to approximately 20-30% for diastolic blood pressure. It has been suggested that there is an increasing impact of individual environmental influences on blood pressure with age [49). We have also confirmed the finding of other authors [48, 49] that there were no significant contributions of common, i.e. shared, family environmental factors to blood pressure levels in this twin popUlation.

We report that a significant proportion of the variance in platelet [Caz+)i can also be attributed to additive genetic effects. The model of best fit featured genetic influences and both unique and common environ­mental effects. According to this model, genetic effects explained approximately 48% of the variance in females and 37% in males. This is the first study that has investi­gated relative genetic and environmental influences on platelet [Ca2 +)i' Since in both females and males the common environmental effects were non-significant, the more parsimonious model containing only genetic and unique environmental factors could also adequately explain our data and would indicate that the contribution to platelet [Ca2 + I; variability from genetic influences could be as high as 89% for females and 96% for males. We have chosen" to include the common environment parameter in our model of choice, since its contribution to the variance almost approaches statistical significance and represents a relatively large component. Samples for determination of platelet [Caz+)i for each twin pair were collected and analysed together and such a large non­significant contribution from common environment may have resulted from shared analytical variation rather than being a true retlection of the contribution of shared twin environment.

This twin method was employed to test for the presence of genetic co-variation between platelet ICa2 '1.

and systolic blood pressure and to determine whether this arises because one trait influences the other, vice versa, both or neither. It was hoped that this design would provide a novel opportunity to address the question of whether the rise in platelet [Ca2 +1 is primary or secondary to the rise in blood pressure. If a genetic factor was influencing [Caz+l, as assessed in platelets, and thereby influencing blood pressure, we would expect to observe genetic co-variation of these two traits. However, we were unable to demonstrate genetic co-variation between platelet [Ca2 + L and systolic blood pressure in this normotensive twin population. Insufficient statistical power may provide one explanation for this result, but this seems less likely since unique environmental co­variation of platelet [Ca2 +)i could be demonstrated with the same sample size. Our data suggest that a generalized inherited abnormality is not the explanation for the association between platelet [Ca2 + 1 and blood pressure.

The tlnding of unique environmental co-variation between platelet [Ca2 +)i and systolic blood pressure suggests that some of the experiences unique to each twin may be causing the two traits to vary together. Since body mass index was related to both platelet [Ca2 +1 and blood pressure and its variability has been shown to be partially due to unique environmental influences [50), we can speculate that it may have contributed to the unique environmental co-variation observed between platelet [CaH)i and blood pressure.

Shear stress forces [51), increased presence of circulat­ing agonists or contact with endothelial lesions could result in increases in platelet [Ca2+1 secondary to the pressure rise. In the present study, the majority of subjects have blood pressures which would not normally be regarded as 'hypertensive'. Nevertheless, it is conceivable that pressure-related shear forces might influence platelet function increasingly throughout the entire range of blood pressures. In addition, subjects with 'low normal' pressures have been shown to have lower circulating plasma adrenaline levels than those with 'high normal' pressures [52) and circulating adrenaline may lead to platelet activation. Unfortunately, it is difficult to isolate these effects since [Ca2+)i levels regulate a number of interrelated events in the platelet. In this twin study our bivariate factor and direction of causation models all fitted the data equally well, preventing us from further resolving this issue of whether the increase in platelet [Ca2 + Ii is a primary or secondary event.

In the present study, total plasma calcium concentra­tion was positively related to systolic blood pressure even after correction for sex, albumin, age and body mass index. In support of other studies [53), genetic intluences accountcd for a large percentage of the variability of total plasma calcium concentration. In addition, the data suggest that there may be genetic co-variation between total plasma calcium concentration and systolic blood pressure. This would imply that a genetic factor influences the relationship betwcen these two variables. Duc to a lack of resolving power, we were unable to

distinguish between the factor models and the unique direction of causation models. However, a direct effect of

Platelet calcium and blood pressure in twins 503

Ca2 + on the peripheral vasculature has been suggested by calcium infusion studies [54).

ft is important for the twins to be representative of the general population with respect to the traits of interest if we wish to extrapolate the results found in them to the general population. Since the mean levels and ranges for males and females for our variables are similar to those reported for normotensive singletons in the general popu­lation [13.17.381. it would seem reasonable to assume that. with respect to these variables. our twins are under the same genetic and environmental influences as the rest of the popUlation.

If there is a large genetic influence for a particular trait it is generally considered that the trait cannot be easily modified by alterations to the environment. This is a misconception. This genetic influence relates only to the variance about the mean and thus a substantial change in the mean may be effected by an environmental change while the variation about the mean and the various sources affecting that variation remain as before. Thus a finding of significant additive genetic variance for blood pressure does not preclude the importance of the modifi­cation of environment in the management of hyper­tension. Given the positive association we have reported between weight and platelet [Ca1 + t. these comments may also be relevant to the mean levels of platelet [Ca2 +)i reported in previous studies of hypertensive patients.

In summary. the present study has confirmed the previously reported positive associations between blood pressure and both platelet [Ca1 +)i and total plasma calcium concentration in 109 pairs of normotensive twins. and has highlighted the important role of weight and body mass index as potential confounders of these relation­ships. We have demonstrated that genetic influences sig­nificantly contribute to not only the variance in systolic and diastolic blood pressure but also to the variance in total plasma calcium concentration and platelet [Ca1+t. and have found evidence suggestive of genetic co-varia­tion of total plasma calcium concentration and systolic blood pressure. The finding of significant unique environ­mental co-variation between platelet [Ca2 + t and systolic blood pressure indicates that environmental factors. including adiposity. need consideration in any future studies investigating the role of intracellular calcium levels in the pathogenesis of hypertension.

ACKNOWLEDGMENTS

We are grateful to Dr John Wetherall and Mr David Groth at Curtin University of Technology. Perth. Western Australia. for their assistance with the DNA fingerprint­ing. and the National Heart Foundation of Australia and the National Health and Medical Research Council of Australia Twin Registry for support.

REFERENCES

I. Kesteloot. H. & Geboers. J. Calcium and blood pressure. Lancet 1982; i, 813-15.

2. Tillman. D.M. & Semple. P.F. Calcium and magnesium in essential hypertension. Clin. Sci. 1988; 75, 395 -402.

1 lind, L., Jakobssen, S., Lithell, H .. Wengle. B. & ljunghall. S. Relation of serum calcium concentration to metabolic risk factors for cardiovascular disease. Br. Med. J. 1988; 297, 960-1

4. McCarron. DA low serum concentrations of ionized calcium in patients with hypertension. N. Engl. J. Med. 1982; 307, 226-8.

S. Hvarfner. A. Bergstrom. R .• Morlin. c.. Wide. L. & ljunghall. S. Relationship between calcium metabolic indices and blood pressure in patients with essential hypertension as compared with a healthy population. J. Hypertens. 1987; 5, 451-6.

6. Fogh·Andersen. N., Hedegaard. L.. Thode. J. & Siggard-Andersen. O. Sex· dependent relation between ionized calcium in serum and blood pressure. Clin. Chem. 1984; 30, 116-18.

7. Shore. A.C.. Booker. J., Sagnella. GA. Markandu, N.D. & MacGregor. GA Serum ionized calcium and pH; effects of blood storage. some physiological influences and a comparison between normotensive and hypertensive subjects. J. Hypertens. 1987; 5, 499-505.

8. Buckley, B.M .• Smith. S.c.. Beevers. M .• Beevers. D.G. & McKiernan. M.J. Lack of evidence of low ionized calcium levels in systemic hypertension. Am. J. Cardiol. 1987; 59,878-80.

9. Resnick. L.M .• Laragh. J.H .. Sealey. J.E. & Alderman. M.H. Divalent cations in essential hypertension; relations between serum ionized calcium. magnesium and plasma renin activity. N. Engl.]. Med. 1983; 309, 888-91.

10. McCarron. DA, Pingree. P .A .• Rubin. R.J .• Gaucher. S.M .• Molitch, M. & Kritzik. S. Enhanced parathyroid function in essential hypertension: a homeostatic response to a urinary calcium leak. Hypertension 1980; 2, 162-8.

II. Strazzullo. P .• Nunziata. V .• Cirillo. M. et al. Abnormalities of calcium metabolism in essential hypertension. Clin. Sci. 1983; 65, 137-41.

12. Bing, R.F .• Heagerty, A.M .• Thumon. H. & Swales. J.D. Ion transport in hypertension; are changes in the cell membrane responsible! Clin. Sci. 1986; 71,225-30.

13. Erne. P .. Bolli. P., Burgisser. E. & Buhler. F.R. Correlation of platelet calcium with blood pressure: effect of antihypertensive therapy. N. Engl. J. Med. 1984; 310, 1084-8.

1·4. Bruschi. G .• Bruschi. M.E .• Caroppo. M .• Orlandini. G .• Spaggiari, M. & Cavatona. A Cytoplasmic free [Cal +] is increased in the platelets of spontaneously hypertensive rats and essential hypertensive patients. Clin. Sci. 1985; 68,179-84.

15. Le Quan Sang. K.H .• Benlian. P .• Kanawati. c.. Montenary·Garestier. T .• Meyer. P. & Devynck. M. Platelet cytosolic free calcium concentration in primary hypertension. J. Hypertens. 1985; 3 (Suppl. 3). S33-6.

16. Lechi. A .• Lechi. c.. Bonadonna. G. et al. Increased basal and thrombin induced free calcium in platelets of essential hypertensive patients. Hyper. tension 1987; 9, 230-5.

17. Hvarfner. A .• Larsson. R .• Morlin. C. et a!. Cytosolic free calcium in platelets; relationships to blood pressure and indices of systemic calcium ,metabolism. J. Hypertens. 1988; 6, 71-2.

18. Barr. S.M. & Reid. J1. Basal platelet intracellular free calcium concentrations in untreated hypertensive patients and normotensive controls. J. Hum. Hypertens. 1989; 3, 263-6.

19. Pritchard. K .• Raine. A.E.G .. Ashley. c.c. et al. Correlation of blood pressure in normotensive and hypertensive individuals with platelet but not lymphocyte intracellular free calcium concentrations. Clin. Sci. 1989; 76, 631-5.

20. Cooper. R.S .• Shamsi. N. & Katz. S. Intracellular calcium and sodium in hypertensive patients. Hypertension 1987; 9, 224-9.

21. Magnus. P .. Berg. K. & Nance. W.E. Predicting zygosity in Norwegian twin pairs born 1915-1960. Clin. Genet. 1983; 24,103-12.

22. Jeffreys. A.J .• Wilson. V. & Thein. S.L. Hypervariable 'minisatellite' regions in human DNA. Nature (London) 1985; 314, 67-73.

23. Arkwright. P.D .. Beilin. L.J .. Rouse. I.. Armstrong. B.K. & Vandongen. R. Effects of alcohol use and other aspects of lifestyle on blood pressure levels and prevalence of hypertension in a working population. Circulation 1982; 66,60-6.

24. Grynkiewicz. G .. Poenie. M. & Tsien. R.Y. A new generation of Cal' indicators with greatly improved fluorescence properties. J. BioI. Chem. 1985; 260, 3440-50.

2S. Pollock. W.K. & Rink. T.J. Thrombin and ionomycin can raise platelet cytosolic Cal' to micro molar levels by discharge of internal Cal' stores; studies using Fura·2. Biochem. Biophys. Res. Commun. 1986; 139,308-14.

26. Perper. R.J.. Zee. TW. & Mickelson. M.M. Purification of lymphocytes and platelets by gradient centrifugation. J. Lab. Clin. Med. 1968; 72, 842-8.

27. Shore. A.C.. Beynon. je.. Jones. J.c.. Markandu. N.D .. Sagnella. GA &

504 P. D. Williams et aI.

MacGregor. GA Mononuclear leucocyte intracellular free calcium - does it correlate to blood pressure! J. Hypertens. 1985; 3, 18]-7.

28. Jinks. J.L. & Fulker. D.W. Comparison of the biometrical genetical. mava and classical approaches to the analysis of human behaviour. Psychol. Bull. 1970; 73, ]11-49.

29. Eaves. L.J .. last, KA. Young. P A & Martin. N.G. Model fitting approaches to the analysis of human behavior. Heredity 1978; 41, 249-320.

30. Joreskog. K.G. & Sorbom. D. Prefis - a program for multivariate data screening and data summarization. 2nd ed. Mooresville. Il: Scientific Software. 1988.

31. loreskog. K.G. & Sorbom. D. lisrel 7 - User's reference guide. 1st ed. Mooresville. Il: ScientifIC Software. 1989.

32. Neale. M.C .• Heath. A.e.. Hewitt. J.K .• Eaves. L.I. & Fulker. D.W. Fitting genetic models with USREl: hypothesis testing. Behav. Genet. 1989; 19, 37-49.

33. Neale. M.e. & Martin. N.G. The effects of age. sex and genotype on self­report drunkenness following a challenge dose of alcohol. 8ehav. Genet. 1989; 19,63-78.

34. Astarie. e.. levenson. I .• Simon. A .• Meyer. P. & Devynck. M. Platelet cytosolic proton and free calcium concentrations in essential hypertension. I. Hypertens. 1989; 7, 458-91.

35. Brickman. A.S .• Nyby. M.D .• von Hungen. K .. Eggena. P. & Tuck. M.L. Calcitropic hormones. platelet calcium and blood pressure in essential hypertension. Hypertension 1990; 16, 5 I 5-22.

36. Grayson. D.A. Twins reared together: minimizing shared environmental effects. Behav. Genet. 1989; 19,59]-604.

37. Hewitt. J.K. Of biases and more in the study of twins reared together: a reply to Grayson. 8ehav. Genet. 1989; 19,605-8.

38. Dominiczak. A.K .. Morton. J.J .• Murray. G. & Semple. P.F. Platelet cytosolic free calcium in essential hypertension: responses to vasopressin. Clin. Sci. 1989; 77, 183-8.

39. Scherrer. U .• Torriani. S .. Nussberger. I. et al. Weight reduction of obese patients reduces free cytosolic calcium of platelets together with ambulatory blood pressure. I. Hypertens. 1987; 5 (Suppl. 5). S629.

40. Resnick. L.M .• Gupta. R.K .• Bhargava. K.K .• Gruenspan. H .• Alderman. M.H. & laragh. J.H. Cellular ions in hypertension. diabetes. and obesity. A nuclear magnetic resonance spectroscopic study. Hypertension 1991; 17,951-7.

41. Bramley. P.M .• Paulin. I.M. & Millar. JA Intracellular cations and trans­membrane cation transport in essential hypertension: the importance of controlled clinical observations. J. Hypertens. 1986; 4, 589-96.

42. Stocks. P. A biometric investigation of twins and their brothers and sisters. Ann. Eugenics 1930; 4, 49-108.

43. Feinleib. M .• Garrison. R.. Somani. N .. Rosenman. R. & Christian, J. Studies of hypertension in twins. In: Paul. 0 .• ed. Epidemiology and control of hypertension. New York: Grune and Stratton. 1975: 3-17.

44. Mcllhaney. M.L.. Shaffer.I.W. & Hines. EA The heritability of blood pressure: an investigation of 200 pairs of twins using the cold pressOf" test. Johns Hopkins Med. J. 1975; 116,57-64.

45. Sorhani. N.O .• Feinleib. M .• Garrison. R.J .• Christian. J.C. & Rosenman. R.H. Genetic variance in blood pressure. Acta Genet. Med. GemeIIol 1976; 25, 137-44.

46. Rose. R.J .• Fulker. D.W •• Miller. J.z.. Grim. C.E. & Christian, J.e. Heritability of systolic blood pressure. Analysis of variance in MZ twin parentS and their children. Acta Genet. Med. Gemellol. 1980; 29, 143-9.

47. Tishler. P.V .• lewitter. F.I .• Rosner. B. & Speizer. F.E. Genetic and environ­mental control of blood pressure in twins and their family members. Acta Genet. Med. Gemellol. 1987; 36, 455-66.

48. Schieken. R.M .• Eaves. L.J .• Hewitt. J.K. et al. Univariate genetic analysis of blood pressure in children (The Medical College of Virginia Twin Study). Am. I. Cardiol. 1989; 64, 1333-7.

49. Sims. I .• Hewitt. I.K .• Kelly. KA. Carroll. D. & Turner. J.R. Familial and individual influences on blood pressure. Acta Genet. Med. Gemellol. 1986; ]5,7-21.

SO. Stunkard. A.I .• Harris. J.R .• Pedersen. N.L. & McCleam. G.E. The body-mass index of twins who have been reared apart. N. Engl.J. Med. 1990; 322, 1483-7.

5 I. Levenson. I .• Devynck. M .• Pithois-Merli. I.. le Quan Sang. K.H .• Filitti. V. & Simon. A. Dynamic association between artery shear flow condition and platelet cytosolic free Ca2' concentration in human hypertension. Clin. Sci. 1990; 79, 613-18.

52. Beilin.l.J .• Vandongen. R.. Arkwright, P.D. & Davidson. L. Adrenal and sympathetic nervous activity in subjectS with 'Iow' and 'high' normal blood pressure. J. Hypertens. 198]; I, 13-18.

53. Whitfield. J.B. & Martin. N.G. The effectS of inheritance on constituents of plasma: a twin study on some biochemical variables. Ann. Clin. Biochem. 1984; 21, 176-83.

54. Overbeck. H.W .• Pamnani. M.B. & Derifield. R.S. Similar vasoconstrictor responses to calcium in normotensive and essential hypertensive man. Proc. Soc. Exp. Bioi. Med. 1975; 149, 519-2S.