| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Hypertension. 2005;45:80.)
© 2005 American Heart Association, Inc.
Scientific Contributions |
From the Department of Biological Psychology (N.K., W.G., D.P., D.I.B., E.J.C.G.), Vrije Universiteit Amsterdam; and the Department of Psychiatry (H.R.), University of Groningen, the Netherlands.
Correspondence to Nina Kupper, Vrije Universiteit, Department of Biological Psychology, van der Boechorststraat 1, 1081 BT, Amsterdam, the Netherlands. E-mail hm.kupper{at}psy.vu.nl
| Abstract |
|---|
|
|
|---|
Key Words: blood pressure monitoring hypertension, genetic antihypertensive agents genetics twins
| Introduction |
|---|
|
|
|---|
To date, only 4 twin studies1013 and 1 family study14 reported heritability estimates for daytime or 24-hour ABP. Estimates ranged from 22% to 62% for systolic blood pressure (SBP) and from 38% to 63% for diastolic blood pressure (DBP). With the exception of studies by Vinck et al12 and Fagard et al,13 sample sizes for the twin analyses have been rather small, ie, at most, 66 pairs in total. Thus, there is a relative paucity of adequately powered twin studies on ambulatory measures. One way of increasing statistical power is to include singleton siblings. Such an extended twin design15 further provides an optimal design to address the question whether results from twin studies on the genetics of ABP may be generalized to the singleton population, because it matches twins and singletons for familial factors like socioeconomic status (SES), diet habits, and maternal behaviors during pregnancy.
Existing twin and family studies of ABP have excluded subjects using antihypertensive medication,12,13 or have been performed with analyses of normotensive subjects only,10,11 thereby removing an important part of the population variance of interest.16 The present study estimated the genetic influences on hypertensive status and ambulatory SBP and DBP in a large sample of twins and their singleton siblings. To examine the effects of exclusion, the genetic analyses on ABP were first performed on normotensive subjects only, secondly after exclusion of medicated hypertensive subjects, and finally without any exclusion.
| Methods |
|---|
|
|
|---|
Procedures
Subjects were visited in the morning before going to work and were requested to refrain from intense physical activity on the preceding and the ambulatory monitoring days. A Spacelabs 90207 ABP monitor (Redmont, Wash) and an ambulatory ECG/ICG recorder,17 which includes a vertical accelerometer, were attached to the subject and the operation was explained. Arm circumference was measured to choose the appropriate arm-cuff size. BP measurements were initiated automatically every 30 minutes. Before inflating, the device gave an auditory 2-tone beep to warn participants to keep their arm as still and relaxed as possible. Subjects were unable to observe their own BP readings. The monitor was programmed to retake a measurement 2 minutes after a misreading. Every 30 (±10) minutes, subjects were prompted by an auditory beep to write down a chronological account of activity (eg, deskwork, housekeeping, watching TV), posture (lying, sitting, standing, walking, and bicycling), and location (eg, at home, at work, at a public place). When they went to bed, participants removed the BP monitor. The signal from the vertical accelerometer was combined with the diary information to check the diary entries on posture and physical activity for accuracy.
Data Reduction
Previous recommendations for excluding artifacts and outliers from ambulatory recordings were followed.18 The reported times of diner and lunch, awakening, and bedtime were used to compute mean SBP and DBP across all readings in the morning, afternoon, and evening. To assess the confounding of different physical activity patterns on ABP levels, we also computed the average ABP on the 3 periods of the day using only BP values obtained during sitting activities. Applying European Society of Hypertension criteria, hypertension was considered present when subjects were currently using prescribed antihypertensive medication or when mean daytime ABP was higher than 135/85 mm Hg.19
Statistical Analysis
Heritability estimates of hypertension and daytime ABP were obtained from structural equation modeling of the MZ and DZ/sib variances and covariances using the structural equation program Mx.20 Hypertension heritability was assessed using a liabilitythreshold model, which assumes a latent, normally distributed liability to disease that is manifest as a categorical phenotype.21 For ABP, the best-fitting trivariate model was used to estimate the relative contribution of genetic and environmental influences to the variance of the 3 daytime period means of SBP and DBP. A detailed description of the model-fitting procedures is found in the online supplement available at http://www.hypertensionaha.org.
These analyses on ABP were performed 3 times, in 3 different sets of subjects. In the normotensive set, we excluded all subjects with hypertension diagnosed (ABP >135/85) and subjects using antihypertensive medication. In the second and unmedicated set, we only excluded the 29 subjects using antihypertensive medication. The third set included all available subjects. To obtain ABP values in these medicated subjects, drug class-specific treatment effect averages, obtained from a recent large systematic review of the effect of antihypertensive treatment on ambulatory BP,6 were added to the observed pressures. Unlike substitution methods, this adjustment makes no assumptions regarding the underlying reasons for treatment and keeps the relative ranking of the treated subjects intact.
| Results |
|---|
|
|
|---|
13.5 hours) took place during the recording period, of which, on average, 13 (±5) were during sitting posture (
6.5 hours, which is 48% of the total monitoring period). Although the sample was previously selected based on the presence of at least 2 family members with extreme scores on personality questionnaires, their scores did not correlate significantly with hypertension diagnosis, or with SBP and DBP. Throughout the day, men had significantly higher SBP and DBP than women. At all 3 periods, age was significantly correlated with SBP (0.24 to 0.33) and DBP (0.31 to 0.35). Both sex and age were kept as covariates in all further model-fitting analyses. Body mass index correlated significantly with both SBP (0.24 to 0.31) and DBP (0.20 to 0.25). Because several studies reported a genetic covariation between BP and body mass index,4,22,23 and because genetic variance is removed when shared genes influence both variable and covariate, we intentionally did not include body mass index as covariate.
Hypertension
In our sample, 115 (14.5%) of the 792 subjects received a hypertension diagnosis based on their ambulatory recording, with 29 of them receiving antihypertensive medication. Among the hypertensive were 31 MZ twins. The homogeneity of covariances over the sexes and between DZ twins and siblings was confirmed, and no evidence for a difference in twin pair versus singleton sibling pair correlation in hypertensive risk was found. The liability threshold was higher for women compared with men, and higher for younger versus older subjects. The sex- and age-corrected tetrachorical correlations for the liability dimension were 0.62 for MZ twins and 0.29 for DZ twins, indicating genetic influences. This was confirmed by model fitting, the results of which are shown in Table 1. Leaving out both shared environmental and genetic influences from the model (E-model) caused a large increase in
2, indicating a significant worsening of the fit. This shows a clear influence of familial factors on hypertension. Statistical power was insufficient to discriminate between genetic influences and shared environmental influences, but given the pattern of twin correlations and the values of Akaikes Information Criterion of the AE and CE models, it is most likely that the AE model is the preferred model. In the AE model, variance in hypertension diagnosis is, for 61%, explained by genetic influences.
|
Ambulatory BP
Table 2 presents the mean ABP for the 3 periods of the day in the normotensive, unmedicated, and full set of subjects. It illustrates the impact of excluding subjects with hypertension and/or using antihypertensive medication. Although the means do not dramatically change (2% to 4.5%), strict exclusion brings about a 30% to 39% reduction in the standard deviation in comparison to the "true" population. Exclusion of medicated subjects led to much smaller changes in means (0.5% to 0.5%) and moderate reductions in standard deviations (1% to 10.5%). Next, the resemblance between MZ twins and between DZ twins or sibling pairs was examined by calculating age-adjusted Pearson correlations, stratified by sex, as shown in Table 3. Throughout, a larger MZ than DZ correlation is evident, suggesting the presence of additive genetic and unique environmental influences.
|
|
Multivariate Genetic Analyses
The means and variances of both SBP and DBP were equal for MZ and DZ twins and for singleton siblings. Importantly, we found no twinsingleton differences in ABP in all 3 sets of subjects, suggesting that results obtained in twins can be generalized to singletons.
Multivariate model-fitting resulted in the preference for a model without shared environmental factors (AE model) over the full model (ACE model) for all 3 sets of subjects. Statistical power was sufficient to discern the AE and CE model, because quantitative analyses have higher statistical power than the ordinal analyses performed for hypertension status. Although there was sufficient power (at ß=0.80,
=0.05) to detect effects of 23% or higher, no significant common environmental effect was found. We further tested the hypothesis that a common latent trait was underlying BP at all 3 periods of the day (common pathway model). This model was preferred over the initial independent pathway model in all 3 sets of subjects. We found no specific genetic influences for each of the daily periods, and the largest part of unique environmental influences was also common to all 3 periods. Table 4 shows the common pathway estimates for A, which corresponds to the heritability, and for E, which corresponds to the influence of the common environmental factor. The specific E estimates represent unique environmental influences that are specific to each of the periods of the day.
|
The highest heritability estimates were found in the full set of subjects. In comparison to the results on the normotensive subjects, heritability estimates were substantially higher (SBP, +7% to 12%; DBP, +8% to 15%). Furthermore, confidence intervals were wider for the normotensive subjects compared with the full set of subjects. An increase in heritability was already seen when unmedicated subjects were included, but only when medicated subjects were included did we find nonoverlapping confidence intervals with the normotensives.
Individual differences in daily physical activity on the measurement day did not confound the genetic analyses of ABP. When these analyses were repeated using BP measurements obtained during sitting activities, the results were essentially unchanged.
| Discussion |
|---|
|
|
|---|
Exclusion of hypertensive subjects clearly distorted MZ and DZ twin correlations, as well as the variances. Excluding medicated subjects further increased the distortion, although the effect was only very minor in this population. Our results showed that restricting the sample to normotensives only not only caused a decrease in total variance but also specifically reduced heritability estimates. With this result, we extend the earlier findings of Cui et al26 on conventional office BP to prolonged BP measurements in naturalistic settings. The effect of excluding groups of subjects, on grounds of hypertension and/or medication, is undesirable because the reduced heritability estimates directly lead to a loss of power in linkage studies. This effect was convincingly illustrated by Hunt et al,27 who showed that removing medicated subjects from the sample led to the disappearance of a quantitative trait locus for conventional office SBP on chromosome 6.
The substantial heritability of office BP has motivated many large-scale efforts to identify hypertension-predisposing genes through linkage28 approaches. Because ABP is a better predictor of target organ damage,7 cardiovascular morbidity, and mortality8,9 than office BP, these gene finding attempts may be well-served by using ABP. Here, we show that genetic variance of ABP is sufficiently high to justify its use in genome searches. The repeated-measures structure inherent in ambulatory monitoring brings further advantages. We separated the entire ambulatory recording into 3 daily periods, allowing for the possibility that different genetic factors would affect BP regulation during leisure (evening) and work (morning, afternoon) periods. For both SBP and DBP, a common factor influenced all 3 periods. There were no separate genetic factors influencing BP at different periods of the day. From a gene finding point of view, the common genetic factor structure is advantageous on 2 accounts. The repeated-measures structure increases statistical power to find genes in linkage analysis.29 Additionally, these genes, by virtue of having a pervasive influence on SBP or DBP across all situations, will also have the largest clinical relevance.
Limitations
An important limitation to our study is the lack of night-time recordings. We opted not to burden our subjects by asking them to continue wearing the BP monitor at night. In our experience, this causes large attrition in nonpatient populations. In family-based studies, the loss of a single subject is more hard-felt than in population samples, in which an additional randomly drawn subject can be easily recruited without loss to the overall study design. It is possible, however, that different genetic factors come into play during the day than at night. Some indication for this possibility, although the confidence intervals of the estimates for the 3 daily periods were overlapping, is seen in the systematically lower heritability in the evening compared with morning and, particularly, afternoon recordings.
Our sample consisted of young white adults aged mainly between 20 and 40, with an age range between 15 and 81 years. Therefore, generalization of our results beyond the main age range or to a population of different ethnicity should be performed only hesitantly. A recent family study performed an age-stratified longitudinal genetic analysis of office BP and found little variation in the genetic architecture over time.30 This suggests that a common set of genes may be contributing to the observed variation in BP across a wide age range. In our sample, separating the analyses of ABP over multiple age cohorts would have compromised statistical power to detect, for instance, twinsingleton differences.
Perspectives
Genetic contribution to the variance in BP is very likely to be polygenic, with only very small contributions of the individual QTLs to the final hypertensive risk. Future gene-finding studies, particularly linkage studies, should therefore aim to maximize statistical power. In this regard, ABP monitoring has a number of advantages over conventional office measurements. Heritability of ABP is comparable to office BP, but the genes underlying ABP are likely to have better predictive validity for target organ damage,7 cardiovascular morbidity, and mortality.8,9 Ambulatory recording has an intrinsic multivariate nature, and the repeated, highly correlated BP measures can be exploited to reduce measurement error and improve the estimation of the latent genetic factor. Our findings further show that a part of the genetic variance in ABP is lost when hypertensive (and/or medicated) subjects are excluded. In future linkage and association studies, such exclusion should be avoided.
| Acknowledgments |
|---|
Received September 9, 2004; first decision September 30, 2004; accepted October 27, 2004.
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
D. T. O'Connor, G. Zhu, F. Rao, L. Taupenot, M. M. Fung, M. Das, S. K. Mahata, M. Mahata, L. Wang, K. Zhang, et al. Heritability and Genome-Wide Linkage in US and Australian Twins Identify Novel Genomic Regions Controlling Chromogranin A: Implications for Secretion and Blood Pressure Circulation, July 15, 2008; 118(3): 247 - 257. [Abstract] [Full Text] [PDF] |
||||
![]() |
N.-Y. Wang, J. H. Young, L. A. Meoni, D. E. Ford, T. P. Erlinger, and M. J. Klag Blood Pressure Change and Risk of Hypertension Associated With Parental Hypertension: The Johns Hopkins Precursors Study Arch Intern Med, March 24, 2008; 168(6): 643 - 648. [Abstract] [Full Text] [PDF] |
||||
![]() |
H. Zhu, W. Yan, D. Ge, F. A. Treiber, G. A. Harshfield, G. Kapuku, H. Snieder, and Y. Dong Relationships of Cardiovascular Phenotypes With Healthy Weight, at Risk of Overweight, and Overweight in US Youths Pediatrics, January 1, 2008; 121(1): 115 - 122. [Abstract] [Full Text] [PDF] |
||||
![]() |
N. Kupper, J. Denollet, E. J. C. de Geus, D. I. Boomsma, and G. Willemsen Heritability of Type-D Personality Psychosom Med, September 1, 2007; 69(7): 675 - 681. [Abstract] [Full Text] [PDF] |
||||
![]() |
E. J. C. De Geus, N. Kupper, D. I. Boomsma, and H. Snieder Bivariate Genetic Modeling of Cardiovascular Stress Reactivity: Does Stress Uncover Genetic Variance? Psychosom Med, May 1, 2007; 69(4): 356 - 364. [Abstract] [Full Text] [PDF] |
||||
![]() |
J.-J. Hottenga, J. B. Whitfield, D. Posthuma, G. Willemsen, E. J.C. de Geus, N. G. Martin, and D. I. Boomsma Genome-Wide Scan for Blood Pressure in Australian and Dutch Subjects Suggests Linkage at 5P, 14Q, and 17P Hypertension, April 1, 2007; 49(4): 832 - 838. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. M. McCaffery, H. Snieder, Y. Dong, and E. de Geus Genetics in Psychosomatic Medicine: Research Designs and Statistical Approaches Psychosom Med, February 1, 2007; 69(2): 206 - 216. [Abstract] [Full Text] [PDF] |
||||
![]() |
B. K. Rana, P. A. Insel, S. H. Payne, K. Abel, E. Beutler, M. G. Ziegler, N. J. Schork, and D. T. O'Connor Population-Based Sample Reveals Gene-Gender Interactions in Blood Pressure in White Americans Hypertension, January 1, 2007; 49(1): 96 - 106. [Abstract] [Full Text] [PDF] |
||||
![]() |
N. C. B. B. Taarnhoj, M. Larsen, B. Sander, K. O. Kyvik, L. Kessel, J. L. Hougaard, and T. I. A. Sorensen Heritability of retinal vessel diameters and blood pressure: a twin study. Invest. Ophthalmol. Vis. Sci., August 1, 2006; 47(8): 3539 - 3544. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. M. Seasholtz, J. Wessel, F. Rao, B. K. Rana, S. Khandrika, B. P. Kennedy, E. O. Lillie, M. G. Ziegler, D. W. Smith, N. J. Schork, et al. Rho Kinase Polymorphism Influences Blood Pressure and Systemic Vascular Resistance in Human Twins: Role of Heredity Hypertension, May 1, 2006; 47(5): 937 - 947. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hypertension Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2005 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |