Heritability of plasma renin activity and plasma concentration ... - Nature

Journal of Human Hypertension (2002) 16, 417–422  2002 Nature Publishing Group All rights reserved 0950-9240/02 $25.00 www.nature.com/jhh

ORIGINAL ARTICLE

Heritability of plasma renin activity and plasma concentration of angiotensinogen and angiotensin-converting enzyme WJ Vinck1, RH Fagard1, R Vlietinck2 and P Lijnen1 1

Hypertension and Cardiovascular Rehabilitation Unit, Department of Molecular and Cardiovascular Research, University of Leuven, Belgium; 2Department of Human Genetics, Faculty of Medicine, University of Leuven (KU Leuven), Belgium

The purpose of the present investigation was to describe the relative impact of genes and environment on the variance of the plasma constituents of the renin angiotensin system. We ascertained 56 male and 80 female adult same-sex twin pairs from the Flemish population. Plasma renin activity (PRA), the concentration of angiotensinogen (AGT) and angiotensinconverting enzyme (ACE) were measured, and path analysis was applied, after transformation toward normality. For PRA and AGT significant heritability was only detected in the male subgroup, with heritability estimates of 66% and 90%, respectively. Angiotensin-

converting enzyme concentration was determined by additive genes for 43% of its variance, by shared environmental influences for 42%, and by specific environmental influences for 15%. The high heritability found for AGT is compatible with the results of earlier studies linking the M235T polymorphism of the angiotensinogen gene to plasma AGT levels. For PRA, we are the first to show significant heritability. Our results regarding ACE confirm the findings in other populations. Journal of Human Hypertension (2002) 16, 417–422. DOI: 10.1038/sj/jhh/1001410

Keywords: blood pressure; renin-angiotensin system; heritability; twin studies

Introduction The renin-angiotensin-aldosterone system (RAAS) plays a pivotal role in the day-to-day regulation of renal salt and water balance and is involved in the pathophysiology of many forms of secondary hypertension. Mendelian forms of secondary hypertension, such as glucocorticoid-remediable aldosteronism,1 have been attributed to polymorphisms of genes in the RAAS, which have also been considered as candidate genes for essential hypertension. Jeunemaitre et al2 have shown linkage of the angiotensinogen (AGT) M235T polymorphism with AGT levels and with essential hypertension, a finding which has not been consistently confirmed in other studies.3 Despite the wealth of information regarding pathophysiology and genetics, data on the relative importance of genetic and environmental influences in the variation of the main components of the RAAS are scarce. Williams et al4 and Rossi et Correspondence: W Vinck, MD, Hypertension and Cardiovascular Rehabilitation Unit, U.Z. Gasthuisberg, Campus Onderwijs en Navorsing, Herestraat 49, B-3000 Leuven, Belgium. E-mail: Wouter.Vinck얀med.kuleuven.ac.be Received 7 August 2001; revised 14 February 2002; accepted 18 February 2002

al5 have reported insignificant heritability estimates for plasma renin activity (PRA), using pedigree analysis and twin analysis, respectively. To the best of our knowledge, the heritability of AGT has never been assessed using the twin method. Jeunemaitre et al2 have shown that the level of AGT is determined by the M235T polymorphism of the AGT gene, but have not determined the relative importance of genetic and environmental sources of variation in the variance of AGT levels. Both Busjahn et al6 and Rossi et al5 have demonstrated significant heritability for angiotensin-converting enzyme (ACE) levels, both using twin analysis. Furthermore, it could be shown that the ACE I/D polymorphism is responsible for part of this genetic variance.6 Using twin studies to define which phenotypes of the RAAS have significant genetic variance may help to elucidate the genetic background of essential hypertension. Phenotypes for which no significant genetic variance can be demonstrated are unlikely to qualify as intermediate phenotypes for a common disorder with a high genetic variance such as essential hypertension. The purpose of the present study was to investigate the heritability of PRA and plasma AGT and ACE concentration, in a sample of middle-aged and elderly same-sex twin pairs. This

Heritability of RAAS components WJ Vinck et al

418

age range was chosen to match the typical age of onset of essential hypertension.

Methods Study population A total of 1087 adult twin pairs was identified from three city registries in Flanders. Twins were randomly selected from this database according to certain in- and exclusion criteria. Twin pairs were excluded when one or both members had died, had moved out of the area, could not be reached (eg, address not available), or refused to participate. Medical exclusion criteria were the use of antihypertensives, diabetes mellitus, heart failure (NYHA classes III and IV), haemodialysis therapy and chronic obstructive pulmonary disease. Twins in which accurate zygosity diagnosis was not possible were excluded from the analysis. The blood pressure data of this and a younger cohort, were reported previously.7 Protocol The study protocol was approved by the Ethics Committee of the Faculty of Medicine of the Catholic University of Leuven. The twins were examined in pairs in an air-conditioned laboratory. When the examination could not be arranged for both members on the same day, they were scheduled on different days, but always at the same time of the day and within 4 weeks. First, the subjects underwent a clinical examination including measurement of height and weight, skinfolds, waist–hip ratio, and blood pressure (BP), first in the supine and afterwards in the sitting position. After 15 min of rest, a 30-ml blood sample was withdrawn from an antecubital vein in the sitting position, and immediately processed. The heparinised tubes were centrifuged at 2000 rpm for 3 min at a temperature of 4°C. The plasma was frozen using liquid nitrogen and stored at −80°C. The biochemical analyses were performed for all samples simultaneously. Also, ambulatory blood pressure was recorded and 24-h urine was collected as described in Vinck et al.7 Biochemical measurements: PRA was measured by a commercially available kit (Angiotensin I RIA coated tube, RADIM, Pomazia, Italy), and expressed as ng/ml per hour. Plasma AGT was measured as the maximal quantity of angiotensin-I generated after 1 h of incubation at 37°C, in the presence of an excess of human renin (0.0062 IRU, Calbiochem, San Diego, CA, USA), and was expressed as ng/ml. Plasma ACE was measured using a spectrophotometric method utilising the synthetic tripeptide substrate N-[3-(2-furyl)acryloyl]-L-phenylalanylglycylglycine (PFAPGG), and expressed as units per litre. One unit of ACE activity was defined as the amount of enzyme that catalyzes the formation of 1 Journal of Human Hypertension

micromole of PFAPGG per min, at 37°C. When abnormal values were obtained, duplicate and triplicate measurements were performed. Zygosity: Zygosity was determined using blood groups (ABO, Rhesus C,D and E, Du if D negative and MNS blood group) and a validated questionnaire.8 Hypervariable genetic markers (DXYS17, 33.6, YNZ-22, HUMvWF, HUMTHO1, D21S11, HPRT, LDLR, GYPA, HGBB, D7S8 and GC) were determined in some subjects. A different result for at least two of the marker loci was taken as sufficient proof of dizygosity. Since birth order was not always known, twin members of each pair were assigned number 1 or 2 using a random number algorithm. Statistical analysis Database management and statistical analysis were performed with the SAS software (SAS Institute Inc, Cary, NC, USA). Genetic analysis was performed by use of the MX path analysis software.9 Depending on whether the distribution was normal or not, characteristics of the subjects were presented as mean ± s.d. or median and range, and differences among the means of male and female subjects were calculated using an unpaired t-test or a Wilcoxon rank-sum test, respectively. In path analysis various assumptions are made. The normality assumption is necessary for the ␹2 distribution of the variance components, which is obtained by the method of maximum likelihood, to be valid.10 Unbiased estimation of the variance components requires the assumptions of equal means and equal variances across zygosity groups.10,11 The normality of the data was assessed using normal probability plots and Shapiro-Wilk’s statistics. A nested hierarchical ANOVA model was implemented using SAS-GLM software12 to test for differences between the means of MZ and DZ twins. Tests for equality of total within-group variance between the MZ and DZ twins were performed according to procedures described by Christian et al.11 We tested for a greater common environment in the MZ than in the DZ twins by assessing whether the MZ lived more frequently together than the DZ twins. Two-sided Pvalues of 0.05 or smaller were considered significant. Prior to genetic modelling, the raw PRA, AGT and ACE values were transformed toward normality, using the method described by Box and Cox.13 Thus we applied a maximum likelihood algorithm to the data to choose the optimal transformation from a very general subset of transformations, each identified by a parameter ␭. It should be noted that most classically known data transformations are defined by particular values for ␭. We analysed the data using covariance structure modelling. When a particular model holds in reality, this will affect the covariance structure of MZ and that of DZ twins in a particular way, so that by examining the observed


covariance matrices of both MZ and DZ twins, the real underlying model can be found. We used a model for the covariance structure involving additive (A) and dominant (D) genetic variance components and shared (C) as well as non-shared (E) environmental variance components, and we modelled the covariance structure of both twin members using the possible combinations of these model parameters (E, AE, CE, ACE and ADE). The maximum likelihood method was used to find the model parameters and expected covariance matrices for both MZ and DZ twins. Then ␹2 values were computed representing the agreement between the observed and expected covariance matrices. These were first used as goodness of fit indices. A P value of 0.05 or smaller indicated a lack fit to the data, and led to rejection of the model. The Akaike Information Criterion (AIC),14 which equals the ␹2 goodness of fit index minus twice the number of degrees of freedom, was used to compare the relative fit and parsimony of the remaining models. Likelihood ratio ␹2 statistics among nested models were used to assess the significance of the individual paths. To assess the significance of the additive genetic variance component, we checked both whether the AE model fitted significantly better than the E model, and whether the ACE model fitted significantly better than the CE model. The latter test checks whether additive genetic influences are still necessary in the presence of a common environmental variance component, with which it is most easily confounded.15 Only when both tests were significant, we concluded that an additive genetic influence was present. Two-sided P-values were significant at the 0.05 level. Since oestrogens affect plasma levels of PRA and AGT,16 for these hormones the analysis was performed in males and females separately. The analysis was performed both on the uncorrected phenotype and after linear correction for significant confounders. The confounders that were considered for each of the phenotypes in a stepwise regression were gender, the use of oral contraceptives or oestrogen substitution, age, body mass index (BMI), 24-h urinary sodium content, creatinine clearance, use of tobacco, use of alcohol, physical activity on a scale from 1 to 10 and the alternative components of the RAAS. All genetic analyses were replicated in the full dataset and in a subset of twins characterised by a very reliable zygosity diagnosis (dizygosity proven by a difference in blood groups, and monozygosity proven by the more strict B-criterion).

Results Sample The data from 136 same-sex twin pairs with reliable zygosity were complete and were used in the present analysis. Of the 1087 twin pairs identified from the town registers, 877 pairs were of the same

sex, and 693 pairs were contacted. Of these 497 were excluded because of death (n = 356), interfering disease (n = 7), antihypertensive drug treatment (n = 54), uncertain zygosity (n = 4) or because they could not be reached (n = 76). In 60 pairs one or both members refused to participate, leaving 136 pairs for analysis. The monozygous male, monozygous female, dizygous male and dizygous female groups comprised 22, 32, 34 and 48 twin pairs respectively. In this sample, 62 twin pairs were classified dizygous by a difference in blood groups or DNA markers (61/1 pairs). One twin pair was classified as monozygous because of a medical history of fetal twin transfusion syndrome.17 The remaining 73 twin pairs were classified according to the algorithm proposed by Sarna and Kaprio.8 Of these 64 pairs could be classified by the very strict B-criteria (46 MZ, 18 DZ), and nine pairs were classified by the A-criteria (seven MZ and two DZ pairs). In 16 twin pairs, the blood sample of one or both members could not be rapidly processed, due to technical problems. In few cases (three for PRA and six for AGT), the laboratory reported insufficient sample quality or quantity. Finally, PRA was available in 117 twin pairs, AGT in 114 and ACE in 120 twin pairs.

419

Characteristics Data on age, sex, anthropometric characteristics, blood pressure and biochemical measurements of MZ and DZ twins, in whom all data were available are given in Table 1. Mean levels of PRA, AGT and ACE did not differ significantly between MZ and DZ twins (0.39 ⬍ P ⬍ 0.70). For the other anthropometric characteristics, we did not find any significant differences between MZ and DZ twins. Haseman and Elston’s F test did not reveal significant differences in the variance of PRA, AGT or ACE between MZ and DZ twins (0.08 ⬍ P ⬍ 0.32). Monozygous twins did not live together significantly more frequently than the DZ twins, so that this source of environmental covariance cannot account for greater MZ than DZ twin concordance. Both PRA, AGT and ACE levels were significantly different between males and females (Table 2). Stepwise regression of the phenotypes on the available covariates revealed a weak but significant positive association between PRA and the use of tobacco, an association between AGT and sex (Table 2), and an increase of ACE with advancing age. Normal probability plots showed a tendency for skewness to the right for the three phenotypes. Before the genetic analyses, for PRA the square-root transformation was applied, and for ACE the logarithmic transformation was able to normalise the data. The AGT values were normalised using a Box-Cox transformation with parameter −1.56. After transformation, the distribution of the phenotypic data was approximately normal. Journal of Human Hypertension


420

Table 1 Characteristics of the subjects Variable Number of subjects Sex ratio (males/females) Living apart/together (pairs) Age (yr)a Weight (kg)b Height (cm)a BMI (kg/m2)a SBP (mm Hg)b DBP (mm Hg)b 24-h syst ABP (mm Hg)b 24-h diast ABP (mm Hg)b PRA (ng/ml/h)a ACE (IU/1)a AGT (ng/ml)a Nau (mmol/24 h)a

Monozygous

Dizygous

P

94 40/54 45/2 44 (32–76) 65.9 ± 9.7 165 (150–188) 23.1 (18.3–35.4) 124.6 ± 12.2 74.6 ± 9.3 117.5 ± 8.3 72.9 ± 6.8 1.1 (0.07– 4.45) 34.5 (20–63) 295.5 (146–1626) 145.5 (42–319)

132 62/70 63/3 43 (31–68) 68.6 ± 13.7 166 (147–199) 24.4 (15.8– 48) 126.8 ± 16.3 77.4 ± 10.8 118.6 ± 10.8 74.0 ± 7.9 1.03 (0.01– 4.71) 35 (17–85) 259 (168–2457) 142.5 (38–342)

– 0.703c 0.657c 0.566d 0.198d 0.930d 0.173d 0.342d 0.138d 0.456d 0.315d 0.699d 0.730d 0.395d 0.444d

a Median and range;b mean ± s.d.; cFisher exact test; dhierarchical ANOVA. BMI = body mass index; SBP = systolic blood pressure; DBP = diastolic blood pressure, ABP = ambulatory blood pressure; PRA = plasma renin activity; ACE = serum angiotensin-converting enzyme concentration; AGT = serum angiotensinogen concentration; Nau = 24-h urinary sodium excretion.

Table 2 Characteristics of male and female subgroups Variable Number of subjects Age (yr)a BMI (kg/m2)a PRA (ng/ml/h)a ACE (IU/1)a AGT (ng/ml)a Nau (mmol/24 h)a

Male

Female

P

94 47.3 (32–75) 25.1 ± 4.4 1.44 (0.07– 4.45) 41.3 (19–85) 320.2 (152–1862) 156.3 ± 58.3

132 43.9 (31–73) 23.2 ± 3.1 1.12 (0.01– 4.71) 34.3 (17–63) 474.0 (146–2457) 140.5 ± 48.6

– 0.009a ⬍0.001b 0.003a ⬍0.001a ⬍0.001a 0.118a

a Median and range; P-value based on Wilcoxon rank-sum test; b mean ± s.d; P-value based on unpaired t-test. BMI: = body mass index; PRA: = plasma renin acivity; ACE = serum angiotensin-converting enzyme concentration; AGT = serum angiotensinogen concentration; Nau = 24-h urinary sodium excretion.

Results of the genetic analyses All results were obtained using the transformed levels of PRA, AGT and ACE. The point estimates of the variance components and the inferential statistics are shown in Table 3. For the log-transformed levels of ACE, the ACE model was chosen, and both additive genetic and shared environmental influences significantly contributed to the variance, for 43% and 42% of the variance, respectively. For the

square-root-transformed levels of PRA and for the Box-Cox-transformed levels of AGT, there was no significant contribution of additive genes to the variance in the overall sample. According to the AIC, the CE-model performed best. The application of linear corrections for significant confounders (including sex) did not change these results (all point estimates within 8%). For PRA and AGT, additional analyses were performed in the male and female subgroups, and in premenopausal twins

Table 3 Variance components of the constituents of the renin angiotensin system Dataset

Model

␹2

Overall dataset Men Women Overall dataset Overall dataset Men Women

CE AE CE ACE CE AE CE

13.28 2.96 4.01 1.124 12.89 1.79 0.23

Variable Sqrt (PRA) Log (ACE) BC (AGT)

AIC −6.72 −17.04 −15.99 −16.88 −7.11 −18.21 −19.77

A (%)

C (%)

E (%)

PAE versus E

0 66 0 43 0 90 0

34 0 36 42 63 0 41

66 34 64 15 37 10 59

⬍0.001 ⬍0.001 0.03 ⬍0.0019 ⬍0.001 ⬍0.001 0.001

PACE versus CE 0.64 0.01 1 0.003 0.73 0.003 0.65

sqrt(PRA) = square root transformation of plasma renin activity; Log(ACE) = Log transformation of angiotensin converting enzyme concentration; BC(AGT) = Box-Cox transformation of angiotensinogen concentration; A = additive genes; C = common environmental influences; E = specific environmental influences; AIC = Akaike Information criterion. Journal of Human Hypertension


421

Table 4 Variance components of the constituents of the renin angiotensin system, using the most strict zygosity criteria Variable Sqrt(PRA)* Log(ACE) BC(AGT)*

Model

␹2

AIC

A (%)

C (%)

E (%)

PAE versus E

PACE versus CE

AE AE AE

1.61 7.69 1.64

−18.39 −12.31 −18.36

69 81 87

0 0 0

31 19 13

⬍0.001 ⬍0.001 ⬍0.001

0.022 0.003 0.027

*Only male estimates are reported.

concordant for menopausal status and use of estrogen preparations. In men, a highly significant additive genetic effect could be demonstrated for both PRA and AGT, and the AE model was chosen (Table 2). For these phenotypes, the ACE model fitted only marginally better, at the cost of using one more degree of freedom. For AGT, in the ACE model the contribution of additive genes decreased from 90 to 56% of the variance, whereas for PRA, in the ACE model, the additive genetic effect was the same as in the AE model. In the women, and the premenopausal female twin pairs, on the contrary, no significant contribution of additive genes to the variance of PRA and AGT could be demonstrated. For the female subsample, the CE model was chosen, and for the premenopausal female twin pairs, the E model could not be rejected. In the postmenopausal twin pairs additional analyses were impossible due to limited numbers of twin pairs concordant for oestrogen use. The replication of all analyses in the subset of twins characterised by the most strict criteria of zygosity diagnosis, yielded the same results, except for ACE (Table 4). There we notice that when the most strict zygosity criteria were applied, the additive genetic variance component and the common environmental variance component were confounded, and that the preferred model only included a genetic and a specific environmental influence.

Discussion Our study showed that additive genes significantly contribute to the variance of PRA, AGT and ACE. For ACE, an additive genetic contribution was present in the entire sample, on top of a significant contribution of a common environment. For PRA and AGT, in the overall sample neither the contribution of additive genes, nor the contribution of a common environment were significant, and the CE model was preferred. When for PRA and AGT the sexes were considered separately, we noted a significant contribution of additive genes in males, and the absence of any significant effects in females. For PRA, this is the first report of a significant heritability. Former reports failed to demonstrate significant heritability for PRA,18 possibly because male and female subjects were not analysed separately. The influence of additive genes on the variance of PRA may be due to a genetic control on

either of the mechanisms involved in the secretion of renin. Renin is secreted by the juxtaglomerular apparatus, in reponse to normal or abnormal phenomena that reduce arterial BP, renal perfusion, or sodium chloride load to the macula densa. These include changes in posture or effective circulating fluid volume (as in sodium depletion, haemorrhage, heart failure, nephrotic syndrome, and cirrhosis). Obviously, in our study involving healthy twins, examined in the same posture, salt intake is an important potential confounder. If it were genetically determined, this could explain our findings. However, we did not find a significant relationship between salt intake and PRA, nor did we find a significant contribution of additive genes to the variance of salt intake (data not shown). In former studies, in which salt intake was experimentally controlled, heritability of PRA was not significant. Moreover, in our study, the heritability of PRA remains significant after correction for salt intake (data not shown). Further studies, both standardising salt intake and taking into account gender may clarify this issue. The high genetic variance found for PRA implies that it cannot be ruled out as a factor responsible for the genetic variance of blood pressure. Also for AGT we were able to show significant heritability in male subjects, with a high heritability estimate, even when the less parsimonious ACE model was chosen. This agrees with earlier reports on the effect of the M235T polymorphism of the AGT gene on AGT levels.2 In combination with the biochemical evidence that AGT is the rate limiting step in the RAAS, and the physiological evidence that the RAAS is an infinite gain negative feedback system for blood pressure, AGT qualifies as a very good intermediate phenotype for essential hypertension, despite recent negative reports.3 The fact that for PRA and for AGT no significant heritability could be demonstrated in female subjects requires further investigation. Possibly the known (cyclic) effect of oestrogen endowment on plasma renin and AGT16,19 explains why we did not find a significant heritability in females. To detect a genetic influence on PRA and AGT in women, it may be necessary to investigate postmenopausal women who are not taking oral oestrogen substitution, to examine women on the same day of the menstrual cycle, or to correct for oestradiol levels. If, after appropriate correction, any genetic influence were to be found in female subjects as well, it would be of interest to compare the amount of the variance Journal of Human Hypertension


422

attributable to additive genes in males and females, which would require studying opposite-sex twins. If no significant heritability can be demonstrated for these hormones (eg, after correction for oestrogen levels), they are unlikely to explain the genetic basis of essential hypertension in women. The heritability of plasma ACE has been repeatedly estimated at about 40–50%.5,6 This estimate was confirmed in our twins. Also the impact of common environmental has been reported previously. We have not been able to identify the nature of the common environmental effects on ACE. It is known that plasma ACE levels are very stable within an individual,20 yet different among individuals. This suggests a complete genetic determination, or early environmental effects. To identify the common environmental influence we have demonstrated, early environmental effects may require further investigation. In conclusion, the present study demonstrated significant genetic variance for both PRA, and the concentration of AGT and ACE.

Acknowledgements The study was partly supported by the National Fund for Medical Research NFWO, Brussels (Belgium). W Vinck was the beneficiary of a fellowship grant from Pfizer Inc (Belgium). R Fagard is holder of the Prof. Amery Chair in Hypertension Research, founded by Merck Sharp & Dohme (Belgium). The kind hospitality of the Limburgs Universitair Centrum and the expert technical assistance of Mrs AM Martin are gratefully acknowledged.

References 1 Lifton RP et al. A chimaeric 11 betahydroxylase/aldosterone synthase gene causes glucocorticoid-remediable aldosteronism and human hypertension. Nature 1992; 355: 262–265. 2 Jeunemaitre X, et al. Molecular basis of human hypertension: role of angiotensinogen. Cell 1992; 71: 169– 180. 3 Brand E, et al. Evaluation of the angiotensinogen locus in human essential hypertension. A European study. Hypertension 1998; 31: 725–729.

Journal of Human Hypertension

4 Williams RR et al. Genetic traits related to hypertension and electrolyte metabolism. Hypertension 1991; 17(Suppl 1): I69–I73. 5 Rossi GP, et al. Genetic determinants of plasma ACE and renin activity in young normotensive twins. J Hypertens 1999; 17: 647–655. 6 Busjahn A et al. Angiotensin-convering enzyme and angiotensinogen gene polymorphisms, plasma levels, cardiac dimensions. A twin study. Hypertension 1997; 29: 165–170. 7 Vinck WJ, Fagard RH, Loos R, Vlietinck R. The impact of genetic and environmental influences on blood pressure variance across age-groups. J Hypertens 2001; 19: 1007–1013. 8 Sarna S, Kaprio J, Sistonen P, Koskenvuo M. Diagnosis of twin zygosity by mailed questionnaire. Hum Mol Genet 1978; 28: 241–254. 9 Neale MC. Mx: statistical Modeling. 2nd edn. 1994, pp 1–142. 10 Haseman JK, Elston RC. The estimation of genetic variance from twin data. Behav Genet 1970; 1: 11–19. 11 Christian JC. Testing twin means and estimating genetic variance. Basic methodology for the analysis of quantitative twin data. Acta Genet Med Gemellol 1979; 28: 35– 40. 12 The SAS Institute Inc. The GLM Procedure 4th edn. 1989, pp 891–996. 13 Box GEP, Cox DR. An analysis of transformations. J R Statist Soc B 1964; 26: 211–252. 14 Akaike H. Factor analysis and AIC. Psychometrika 1987; 52: 317–332. 15 Christian JC, Norton JA, Sorbel J, Williams CJ. Comparison of analysis of variance and maximum likelihood based path analysis of twin data: partitioning genetic and environmental sources of covariance. Genet Epidemiol 1995; 12: 27–35. 16 Skinner SL, Lumbers ER, Symonds EM. Alterations by oral contraceptives of normal menstrual changes in plasma renin activity, concentration and substrate. Clin Sci 1969; 36: 67–76. 17 Vlietinck R. The Determination of Zygosity of Twins. PhD thesis, University of Leuven (KULeuven), 1986. 18 Jeunemaitre X et al. Familial correlations of plasma renin in hypertensive sibling pairs. J Hypertens 1997; 15: S86–S86. 19 Danser JAH et al. Determinants of interindividual variation of renin and prorenin concentrations: evidence for a sexual dimorphism of (pro)renin levels in humans. J Hypertens 1998; 16: 853–862. 20 Alhenc-Gelas F, Weare JA, Johnson Jr RL, Erdo¨ s EG. Measurement of human converting enzyme level by direct radioimmunoassay. J Lab Clin Med 1983; 101: 83–96.