Pleiotropic Genetic Effects on LDL Size, Plasma Triglyceride, and HDL Cholesterol in Families Karen L. Edwards, Michael C. Mahaney, Arno G. Motulsky, Melissa A. Austin Abstract—The interrelationships among low density lipoprotein (LDL) particle size, plasma triglyceride (TG), and high density lipoprotein cholesterol (HDL-C) are well established and may involve underlying genetic influences. This study evaluated common genetic effects on LDL size, TG, and HDL-C by using data from 85 kindreds participating in the Genetic Epidemiology of Hypertriglyceridemia (GET) Study. A multivariate, maximum likelihood– based approach to quantitative genetic analysis was used to estimate the additive effects of shared genes and shared, unmeasured nongenetic factors on variation in LDL size and in plasma levels of TG and HDL-C. A significant (P,0.001) proportion of the variance in each trait was attributable to the additive effects of genes. Maximum-likelihood estimates of heritability were 0.34 for LDL size, 0.41 for TG, and 0.54 for HDL-C. Significant (P,0.001) additive genetic correlations (rG), indicative of the shared additive effects of genes on pairs of traits, were estimated between all 3 trait pairs: for LDL size and TG rG520.87, for LDL size and HDL-C rG50.65, and for HDL-C and TG rG520.54. A similar pattern of significant environmental correlations between the 3 trait pairs was also observed. These results suggest that a large proportion of the well-documented correlations in LDL size, TG, and HDL-C are likely attributable to the influence of the same gene(s) in these families. That is, the gene(s) that may contribute to decreases in LDL size also contribute significantly to higher plasma levels of TG and lower plasma levels of HDL-C. These relationships may be useful in identifying genes responsible for the associations between these phenotypes and susceptibility to cardiovascular disease in these families. (Arterioscler Thromb Vasc Biol. 1999;19:2456-2464.) Key Words: pleiotropy n LDL size n HDL cholesterol n triglycerides n families
N
umerous studies have shown that plasma lipids and lipoproteins are associated with increased risk of cardiovascular disease (CVD), including smaller-size LDL particles,1–3 elevated plasma triglyceride (TG),4,5 and decreased levels of HDL cholesterol (HDL-C).6 However, whether these associations with CVD are independent of one another remains controversial.1,5,7 Each of these cardiovascular risk factors also appears to be genetically influenced. For example, two of the most common forms of familial hyperlipidemia among families with documented coronary heart disease (CHD) have been classified as familial combined hyperlipidemia (FCHL) and familial “monogenic” hypertriglyceridemia (FHTG),8 –11 both of which are characterized by variable and obligatory hypertriglyceridemia, respectively. Although the specific genetic bases of both FCHL and FHTG are poorly understood, studies indicate major gene effects on TG and apo B in FCHL families.12,13 In addition, there is also considerable evidence for genetic influences on LDL subclass phenotypes12,14 –18 and HDL-C.19 –22 Statistical and physiological relationships between measures of LDL size, plasma TG, and HDL-C are well
established3,23–25 and may represent a high-CHD-risk lipid/ lipoprotein phenotype with common underlying genetic influences. Sprecher et al26,27 first proposed that the combination of high TG and low HDL-C constituted an inherited “conjoint trait” associated with increased risk of CHD in families of hypertriglyceridemic or hypercholesterolemic probands participating in the Lipid Research Clinics Family Study. An atherogenic lipoprotein phenotype characterized by small dense LDL, low plasma levels of HDL-C, high plasma TG levels, and elevated apo B levels was previously described by Austin et al.28 Furthermore, a multivariate lipid factor characterized primarily by LDL size, plasma TG, and HDL-C25 has been identified and was shown to prospectively predict CVD.29 By comparing female monozygotic and dizygotic twins, this lipid factor was shown to be heritable and possibly linked to both the lipoprotein lipase gene and the hormone-sensitive lipase gene.30,31 Although the evidence supports common genetic influences on these cardiovascular risk factors, no study to date has simultaneously evaluated the genetic relationships among the 3 risk factors; LDL size, plasma TG, and HDL-C in high-risk families.
Received December 8, 1998; revision accepted March 23, 1999. From the Department of Epidemiology (K.L.E., M.A.A.), School of Public Health and Community Medicine, and the Department of Medicine (A.G.M.), Division of Medical Genetics, University of Washington, Seattle; and the Department of Genetics (M.C.M.), Southwest Foundation for Biomedical Research, San Antonio, Tex. Correspondence to Karen L. Edwards, PhD, Department of Epidemiology, Box 357236, School of Public Health and Community Medicine, University of Washington, 1959 NE Pacific Ave, Seattle, WA 98195. E-mail:
[email protected] © 1999 American Heart Association, Inc. Arterioscler Thromb Vasc Biol. is available at http://www.atvbaha.org
Downloaded from http://atvb.ahajournals.org/ by guest on July 2, 2015 2456
Edwards et al
Quantitative Genetic Analysis of Lipids/Lipoproteins
Thus, the purpose of this investigation was to evaluate the evidence for shared genetic influences between phenotypic variation in LDL size, plasma levels of TG, and HDL-C in families at increased risk for CVD32 by using quantitative genetic analysis.
2457
Statistical Analysis
Because TG and HDL-C distributions were positively skewed (2.16 and 0.96, respectively), the natural log transformations (ln) of these risk factors were used for all analyses. Transformation reduced the skewness in plasma TG and HDL-C to 0.12 and 20.01, respectively. Untransformed values are presented in descriptive tables and figures for ease of interpretation. For the quantitative genetic analysis, pedigree and phenotypic data Methods were prepared using the computer package PEDSYS.37 Statistical Study Sample genetic analysis was conducted using the modified version of the The families included in this cohort study are primarily white and are Pedigree Analysis Package, version 3.0,38 which is based on estabpart of the Genetic Epidemiology of Hypertriglyceridemia (GET) lished genetic theory of partitioning the total phenotypic variance of Study, which is focused on understanding the risk of CVD in the a trait into genetic and environmental components. In this approach, common familial forms of elevated TGs, ie, FCHL and FHTG, and the univariate heritability (h2) of an individual trait represents the on elucidating the underlying genetic influences of these disorders.33 proportion of the total phenotypic variance due to additive genetic Family ascertainment for the GET Study was based on 2 family effects.39 The residual heritability is used here to reflect the proporstudies conducted at the University of Washington, Seattle, in the tion of variance attributable to additive genetic effects after accountearly 1970s.8,10 In brief, the first family study was published in ing for age and sex effects. 197310 and focused on families identified through a hyperlipidemic Extension of univariate quantitative genetic analysis to the multifamily member (proband) surviving a myocardial infarction.10 The variate state has been described in detail.38,40 – 43 In this approach, the second study was published in 1976 and identified families through total phenotypic correlation between pairs of traits is partitioned into a proband with hypertriglyceridemia but without coronary disease.8 the additive genetic (rG) and random environmental (rE) components In these baseline studies, FCHL was characterized by variable by using maximum-likelihood methods implemented in the modified expression of hypercholesterolemia or hypertriglyceridemia in famversion of the Pedigree Analysis Package.38 The 3 pairs of traits used ily members, while FHTG was characterized by families in which all in this trivariate quantitative genetic analysis were (1) LDL size and affected relatives had an elevated plasma TG level but not an TG, (2) LDL size and HDL-C, and (3) TG and HDL-C. The additive elevated cholesterol level. Thus, elevation in plasma TG levels in genetic and environmental correlations between these pairs of traits affected relatives was always seen in FHTG and commonly observed were obtained from the genetic and environmental variance-covariin FCHL at baseline. ance matrices, estimated by modeling the joint distributions of the As part of the GET Study, living individuals from a total of 85 traits as a function of their population means; their covariates and families were recontacted between 1994 and 1997. Of the 85 regression coefficients; the additive genetic values; random environfamilies, 58 (68.2%) had been classified as FCHL and 27 (31.8%) as mental deviations; and the degree of relationship among the individFHTG at baseline. In the 20 years since the baseline studies were uals in this sample. initiated, many family members had died, including those with the The genetic and environmental correlations represent the additive highest TG levels,32 and could not be resampled. New blood samples effects of shared genes (pleiotropy) and of shared, unmeasured were obtained from surviving family members, including probands environmental (nongenetic) factors on the phenotypic covariance for and their spouses; siblings and spouses of siblings; and grandchileach pair of traits, respectively. An estimate of the total phenotypic dren, nieces, and nephews of the proband. For the current analysis, correlation (rP) between 2 traits was obtained using the following eligible family members were those who were over age 18 at equation: rP5[=h12=h22rG]1[=(12h12)=(12h22)rE],1 where h12 follow-up, who were not pregnant, and who were not too ill to and h22 represent the residual heritability of each trait in the pair. participate. Each eligible family member was asked to complete and The significance of each of the estimated parameters (eg, h2, rG, and return a self-administered medical history questionnaire by mail and rE) was evaluated by likelihood-ratio tests by comparing the ln likelito provide a fasting blood sample for lipid and lipoprotein determihood of a model in which the parameter is estimated to the ln likelihood nations. Approximately 49% of participants lived in Washington of a more restricted model in which the same parameter is set to zero. state, with the remaining 51% residing in 28 other states. All The likelihood-ratio test yields a statistic that is asymptotically distribinformation was kept confidential and was not shared with other uted approximately as a x2 with degrees of freedom equal to the family members. Each participant provided written, informed condifference in the number of estimated parameters and is calculated as sent at the time they were enrolled in the follow-up study, and all 22(ln likelihoodrestricted model2ln likelihoodgeneral model). Polygenic pleiotmethods used to contact family members were approved by the ropy is indicated by a rG value that is significantly different from zero. University of Washington Internal Review Board. An additional likelihood-ratio test was performed for each significant genetic correlation to test for complete pleiotropy (rG51.0). In this test, Laboratory Measurements the unrestricted model, wherein rG is estimated, is compared with a All subjects were instructed to fast overnight. Blood samples were more restricted model wherein rG is fixed at either 1.0 or 21.0. When centrifuged at 2500 rpm for 15 minutes using a standardized rG is significantly different from ¦1.0¦, pleiotropy is interpreted as protocol. For participants living .100 miles from the Seattle area, incomplete.22 plasma was separated and shipped on ice for overnight delivery to Finally, to obtain an estimate of the shared additive effects of the University of Washington. genes on each of the residual phenotypes in a pair, the squared Lipid determinations including plasma TGs and total HDL-C were additive genetic correlation for that pair is multiplied by the performed at the Core Laboratory, Northwest Lipid Research Labheritability estimate for each trait. Similarly, the proportion of oratories in Seattle, a Centers for Disease Control and Prevention– residual phenotypic variance in an individual trait that is due to certified lipid laboratory. HDL-C and plasma TG were measured by shared effects of unmeasured environmental factors is obtained by enzymatic techniques (Abbott Laboratories) according to the stanmultiplying the squared environmental correlation by 1 minus the dardized procedures of the Lipid Research Clinics protocol.34 heritability estimate. Nondenaturing gradient gel electrophoresis was performed on To better meet assumptions of normality, individuals with plasma isolated LDL using 2% to 14% polyacrylamide gradient gels, as TG .4.5 SDs above the mean (TG .905 mg/dL) were excluded previously described.23,35,36 The estimated diameters of LDL subfrom all analyses (n55). Thus, this analysis focuses on 780 individclasses were calculated on the basis of a calibration curve conuals in 85 families who provided a fasting blood sample and structed from high-molecular-weight standards run on the same completed a medical history questionnaire, including 78 spouse gel.35 The size of the major LDL subclass, denoted LDL peak pairs, 501 parent-offspring pairs, and 518 sibling pairs. Quantitative particle diameter (LDL size), was identified from the isolated LDL genetic analysis was first performed in the whole data set and then gel and used as a continuous variable in the quantitative genetic separately in the FCHL and FHTG families. However, because there from http://atvb.ahajournals.org/ guest on differences July 2, 2015when the quantitative genetic analwere noby substantial analysis as a measure of LDL sizeDownloaded heterogeneity.
2458
Arterioscler Thromb Vasc Biol.
October 1999
TABLE 1. Descriptive Summary (Mean6SD) for AGE, LDL Size, Plasma TG, and HDL-C in Relatives and Spouses by Sex Relatives
Age, y
Spouses
Total
Men (n5255)
Women (n5298)
Men (n583)
Women (n5144)
Men (n5338)
Women (n5442)
46.3615.1
45.9614.9
54.1616.3
57.4613.6
48.2615.7
49.7615.5
LDL size, Å
262.469.2
267.568.3
264.569.2
268.468.9
262.969.2
267.868.5
TG, mg/dL
185.56136.0
144.3691.2
158.56130.8
145.6689.0
178.86135.1
144.7690.4
40.9611.3
52.5615.6
41.4611.4
52.3614.2
41.4611.3
52.8615.2
HDL-C, mg/dL
Not including probands.
ysis was run separately in FCHL and FHTG families, only results based on the combined sample are presented.
Results The average age of participants in this study was 49.0 years, and more than half of the participants were female (56.9%). As shown in Table 1, spouses tended to be older than relatives, although there were no differences in age between male and female relatives or between male and female spouses. Unadjusted measures of LDL size, plasma TG, and HDL-C for the individuals included in this analysis by relative type and sex are also shown in Table 1. Men tended to have smaller-size LDL particles and lower HDL-C levels than women, with similar patterns in relatives and spouses. The average level of plasma TG was higher in men than women and was higher in male relatives compared with male spouses. Plasma TG levels did not appear to be different in female relatives and spouses, and this may be due to increased mortality among relatives with high plasma TG levels relative to spouses.32 Figures 1 through 3 present the frequency distributions for all individual study subjects (n5780) for unadjusted measures of LDL size, plasma TG, and HDL-C, respectively. The mean LDL size was 265.7 Å (SD 9.13) in this sample (Figure
1). Figure 1 also shows a suggestive bimodal distribution of LDL size. The overall mean plasma TG level was 1.8061.27 mmol/L (159.5 mg/dL [SD 113.16]; median 1.46 mmol/L [129 mg/dL]) and was highly skewed (Figure 2). Mean HDL-C was 1.2460.38 mmol/L (47.8 mg/dL [SD 14.77]; median 1.16 mmol/L [45.0 mg/dL]) and was also positively skewed (Figure 3). The maximum-likelihood estimates of the mean effects and variance components from the quantitative genetic analysis are presented in Table 2. Likelihood-ratio tests indicate significant (P,0.001) residual heritabilities for each of the traits. Approximately one third of the residual variance in LDL size (h250.34), slightly greater than one third of the residual variance in plasma TG (h250.41), and more than half of the residual variance in HDL-C (h250.54) were attributable to additive genetic effects. This table also includes maximum-likelihood parameter estimates for significant (P,0.001) covariate effects. Based on the coefficients for sex ( b 2sex), mean values for females were 268.06 Å, 0.054 mmol/L (4.78 mg/dL), and 0.102 mmol/L (3.94 mg/dL) compared with 262.98 Å, 0.056 mmol/L (4.98 mg/dL), and 0.095 mmol/L (3.67 mg/dL) in males for LDL size, ln TG, and ln HDL-C, respectively. Significant age effects were also
Figure 1. Frequency distribution of LDL size in 780 family members. Mean LDL size was 265.7 Å (SD 9.13).
Downloaded from http://atvb.ahajournals.org/ by guest on July 2, 2015
Edwards et al
Quantitative Genetic Analysis of Lipids/Lipoproteins
2459
Figure 2. Frequency distribution of plasma TG in 780 family members. Mean plasma TG was 159.5 mg/dL (SD 113.16), median 129 mg/dL.
detected for each of the 3 traits. Overall, age and sex together accounted for 1.3%, 7.6%, and 17.6% of the total variance in LDL size, ln TG, and ln HDL-C, respectively. In the multivariate analysis, all genetic correlations were significantly different from zero (P,0.001) based on likelihood-ratio tests (Table 3). The negative genetic correlation (rG520.87) between LDL size and plasma TG was the strongest, suggesting that genes in common contribute to decreases in LDL size and increases in plasma TG. The positive genetic correlation (rG50.65) between LDL size and HDL-C was modest and suggests that shared genes that increase LDL size also increase HDL-C. The negative genetic
correlation between plasma TG and HDL-C (rG520.54) was also significant. Based on likelihood-ratio tests, the hypothesis of complete pleiotropy (rG561.0) was rejected for all genetic correlations (P,0.001). When squared, the additive genetic correlation between 2 traits in a pair provides an estimate of the additive genetic variance for each trait in the pair that is due to effects of the same gene(s). The squared additive genetic correlation between LDL size and plasma TG (rG2) was 0.757. Thus, almost 76% of the additive genetic variance in LDL size is shared with TG. The squared additive genetic correlation between LDL size and HDL-C (rG2) was 0.423, suggesting that nearly
Figure 3. Frequency distribution of HDL-C in 780 family members. Mean HDL-C was 47.8 mg/dL (SD 14.77), median 45.0 mg/dL.
Downloaded from http://atvb.ahajournals.org/ by guest on July 2, 2015
2460
Arterioscler Thromb Vasc Biol.
October 1999
TABLE 2. Maximum-Likelihood Parameter Estimates From the Multivariate Quantitative Genetic Analysis of LDL Size, Plasma TG, and HDL-C in 85 Kindreds Parameter
LDL Size (6SE)
In TG (6SE)
m
262.9860.52
4.9860.04
3.6760.02
s
8.7760.23
0.6160.02
0.2860.01 0.5460.07
2
h *
0.3460.07
0.4160.07
b2Sex*
5.0860.61
20.2060.04
In HDL (6SE)
0.2760.02
b2Agem*
20.02860.029
0.00460.002
1.13102560.001
b2Agef*
20.07360.026
0.01360.002
1.23102360.001
m Indicates the mean value in males; h , residual heritability. Sex was scaled so that the regression coefficient (b2sex) represents the effect of having the covariate present (female) compared with having it absent. Age was scaled by subtracting out the mean covariate value so that the regression coefficient (b2age) represents the effect associated with a 1-unit deviation from the mean level of the covariate in the overall sample. m indicates male; f, female. *All parameters are significant at P,0.001. 2
50% of the additive genetic variance in each of these traits is due to shared genes. The squared additive genetic correlation between plasma TG and HDL-C (rG2) was 0.292, suggesting that almost 30% of the additive genetic variance in each trait is due to genes shared between the pair. Unmeasured environmental correlations were also significant (P,0.001) between all pairs of traits. Consistent with the genetic correlations, the environmental correlations between LDL size and plasma TG and between plasma TG and HDL-C were negative, whereas the environmental correlation between LDL size and HDL-C was positive (Table 3). When squared, the environmental correlations provide an estimate of the proportion of the random environmental variance (ie, that proportion of the shared residual phenotypic variance not attributable to additive genetic effects or covariates) for each trait that is due to the effects of the same unmeasured environmental factors. Phenotypic correlations were estimated by using the equation described in Methods and are shown in Table 3. Estimates of the contributions of shared additive effects of genes and of shared unmeasured environmental factors on each of the residual phenotypes in a pair are shown in Figures 4A through 4C. Figure 4A presents the results for LDL size and TG, indicating that 26% of the residual variance in LDL size was due to additive genetic effects that are shared with plasma TG, while 19% of the residual variance was attributable to unmeasured environmental effects shared with TG. For plasma TG, 31% of the residual variance was attributable
to additive genetic effects shared with LDL size, and 17% was attributable to shared unmeasured environmental factors. However, the sum of the 2 proportions was less than 1 for each trait, suggesting that other nonshared factors contributed to the residual phenotypic variance in each trait of the pair: 55% for LDL size and 52% for TG (Figure 4A). Similar results were obtained for the other trait pairs; however, the proportion of nonshared residual variance was greater for traits in these pairs (Figures 4B and 4C). Figure 4B indicates that 14% and 23% of the residual variance in LDL size and HDL-C, respectively, were due to shared additive genetic effects between the pair. Shared unmeasured environmental effects accounted for 15% and 11% of the residual variance in LDL size and HDL-C, respectively. Approximately 30% of the residual variance of each trait in this pair was shared; however, the majority of the residual variance was due to other nonshared factors, 71% and 66% for LDL size and HDL-C, respectively. Similar results were observed for TG and HDL-C (Figure 4C). For both TG and HDL-C, 71% of the residual variance was attributable to other nonshared factors. Shared factors accounted for slightly less than 30% of the residual variance in each trait; additive genetic effects accounted for 12% and 16% of the residual variance, and shared unmeasured environmental factors accounted for 17% and 13% of the residual variance in TG and HDL-C, respectively.
Discussion
Although phenotypic associations between LDL size, TG, and HDL-C are well documented, this is the first demonstration that the observed phenotypic associations are largely genetic. That is, the results presented here demonstrate strong genetic correlations between each possible pair of these traits among families at increased risk for CVD.32 The negative genetic correlation between LDL size and plasma TG shows that genetic influences that decrease LDL size also result in increases in plasma TG levels. The positive genetic correlation between LDL size and HDL-C was also significant, as was the negative genetic correlation between HDL-C and plasma TG, indicating that genes in common explained a substantial proportion of the genetic covariance in each pair of traits. However, based on rejecting the hypotheses of complete pleiotropy, these results indicate that additional nonshared genes also contributed to the variation in each of the traits. Environmental correlations between each pair of traits were also significant, although generally not as large as the corresponding genetic correlations. However, they still provide evidence of shared random environmental effects among this set of phenotypically related lipid and lipoprotein risk TABLE 3. Maximum-Likelihood Estimates of the Additive factors, indicating the existence of other important covariates Genetic (rG) and Environmental (rE) Correlations From the that were not included in this analysis. At least 40% of the Multivariate Quantitative Genetic Analysis of LDL Size, 1n TG, residual phenotypic variance in each of the 3 traits is and 1n HDL-C in 85 Kindreds and the Phenotypic attributable to nonshared effects that almost certainly include Correlations (rP)* environmental factors and may even include some dominance Phenotype Pairs rG6SE rE6SE rP genetic effects that were not modeled in this statistical genetic LDL size–TG 20.8760.06 20.5360.05 20.66 approach. Importantly, identification and addition of signifiLDL size–HDL-C 0.6560.09 0.4860.06 0.54 cant environmental factors to the genetic model would TG–HDL-C 20.5460.09 20.5360.07 20.53 decrease the residual environmental contribution to the phenotypic variance and increase the relative importance of the *All genetic (rG) and environmental (rE) correlations are significant at Downloaded from http://atvb.ahajournals.org/ guest already on July 2, 2015 geneticby effects detected. P,0.001.
Edwards et al
Quantitative Genetic Analysis of Lipids/Lipoproteins
2461
Figure 4. A, Results of quantitative genetic analysis of LDL size and TG. Percentages of residual phenotypic variance in each trait due to shared additive genetic, shared unmeasured environmental, and nonshared effects are displayed. B, Results of quantitative genetic analysis of LDL size and HDL-C. Percentages of residual phenotypic variance in traits due to shared additive genetic, shared unmeasured environmental, and nonshared effects are displayed. C, Results of quantitative genetic analysis of TG and HDL-C. Percentages of residual phenotypic variance in traits due to shared additive genetic, shared unmeasured environmental, and nonshared effects are displayed.
Univariate residual heritability estimates for the lipids and possible pair of these variables was strong and highly lipoproteins in this sample of high-risk families are consistent statistically significant (P,0.01): 20.71 for TG and LDL with previous reports. For example, in a sample of Mexicansize, 20.57 for TG and HDL-C, and 0.61 for LDL size and American families participating in the San Antonio Family HDL-C3 compared with rP520.66, rP520.53, and rP50.54, 21 respectively, reported here. Heart Study, Mahaney et al reported heritability estimates of 0.53 and 0.55 for ln TG and ln HDL-C, respectively, The results presented in this study are also consistent with compared with estimates from the current study of 0.41 and a previous quantitative genetic analysis of TG and HDL-C. 0.54, respectively. Estimates of heritability for LDL size, Based on data from Mexican-American families participating plasma TG, and HDL-C based on female twins are also in the San Antonio Family Heart Study, Mahaney et al21 2 14 showed that shared genes contributed to the covariation of consistent with the estimates presented here, h 50.30, h250.50 to 0.65, and h250.73,44 respectively. Based on data TG and HDL. The magnitude of the genetic correlation from male twins, the heritability estimate for HDL-C was between TG and HDL-C reported in the current study of 0.36.45 high-risk families is remarkably similar (rG520.54 versus The phenotypic correlations estimated in this study are also 20.53, respectively).21 Heller et al46 also found significant consistent with the well-documented patterns between these genetic correlations between serum lipids and apolipoprolipids and lipoproteins.23–25 For example, in a recent report teins, including TG and HDL-C, in elderly Swedish twins. Downloaded from http://atvb.ahajournals.org/ by guest on current July 2, 2015 from the Physicians’ Health Study, the correlation for each The results of the study extend these previous reports
2462
Arterioscler Thromb Vasc Biol.
October 1999
by including a measure of LDL size and demonstrate com3 related traits, suggesting pleiotropic effects beyond those described in the current article. mon genetic influences between LDL size and both HDL-C In addition, when pleiotropy is due primarily to shared and plasma TG. additive effects of polygenes,57 such traits with high genetic The hypothesis of common genetic influences on pairs of and environmental correlations can also be used as covariates lipids and lipoproteins was first suggested by Goldstein et in univariate segregation and linkage analyses. Specifically, al.10 Based on data from the original family study published by removing common genetic effects of other traits, the in 1973, Goldstein et al suggested that the FCHL phenotype likelihood of detecting genes influencing the unique residual was due to variable expression of a single dominant gene and component of an individual trait may be increased. For not segregation of 2 separate genes.10 The hypothesis of example, on the basis of the results presented in this study, common underlying genetic influences is further supported both HDL-C and plasma TG could be included as covariates by recent results from linkage studies of candidate genes in analyses of LDL size to account for shared additive genetic involved in lipid and lipoprotein metabolism. Two recent and random environmental contributions to the variance in reports, 1 based on families ascertained in Finland and the LDL size. Evidence for linkage with the residual trait would other based on a mouse model of FCHL, have strongly represent a gene unique to that trait. Importantly, the pleiosuggested the existence of a novel gene for FCHL on tropic effects identified in this study can thus be used to chromosome 1.47,48 In addition, Dallinga et al49 showed that effectively increase the likelihood of detecting quantitative DNA variations in the apo AI-CIII-AIV gene cluster modify trait loci. plasma TGs, LDL-C, and apo CIII levels in families with Finally, although these results were interpreted as repreFCHL. Recent sib-pair linkage analysis also provides evisenting polygenic pleiotropy, individual loci exerting subdence for common genetic influences on LDL size, TG, and stantial effects on 1 of these 3 phenotypes may be responsible HDL-C. Preliminary evidence for linkage between the apo B for some or all of the pleiotropy detected in this analysis. gene and a lipid factor identified by factor analysis and However, it is important to emphasize that the analyses characterized by LDL size, plasma TG, and HDL-C is based presented here do not explicitly distinguish between the on data from the Kaiser Women Twins Study.31 Furthermore, 50 pleiotropic effects of single, major loci and the pleiotropic using data from the same women twins, Austin et al also effects of multiple loci. Thus, shared additive effects of provided evidence for linkage between the apo B gene and individual loci may well be included in the genetic correlaeach of the following individual risk factors: LDL size, TG, tions. Additionally, the effects of closely linked loci exhibitHDL-C, and apo B levels. Several studies have also reported ing strong linkage disequilibrium could also contribute to the linkage between the apo B gene and individual measures of genetic correlations and the well-documented pattern of circulating lipids and lipoproteins, including measures of phenotypic associations between these lipids and lipoproLDL heterogeneity51 and TG.52 Several other candidate teins. Linkage and other statistical genetic studies are curgenes, including the LDL receptor, lipoprotein lipase, and apo rently underway to test and distinguish between these hypothAI-CIII-AIV loci, have been shown to be associated with eses in this population. lipids and lipoproteins.53–55 Additionally, the lipoprotein Overall, these results suggest that correlations between lipase gene has recently been linked to LDL size, with LDL size, plasma TG, and HDL-C are strongly influenced by simultaneous effects on TG and HDL-C in lipoprotein lipase– 56 shared genes and thus may be jointly involved in genetic deficient families. susceptibility to CVD. Localizing and identifying these Together these findings suggest that each of the 3 lipid and gene(s) may lead to a better understanding of the role of LDL lipoprotein measures reflect a common underlying process size, plasma TG, and HDL-C in susceptibility to CVD in that is controlled, at least in part, by shared genes and provide these high-risk families. strong support for the existence of genes with pleiotropic effects influencing covariation in plasma lipids and lipoproAcknowledgments teins. Thus, it may be appropriate to consider the aggregate This research was supported by National Institutes of Health grants effects of these risk factors in predicting CHD, as has been HL49513 (to M.A.A.) and DK09645 (to K.L.E.) and was performed recently reported in type 2 diabetes.23 The use of multivariate during Dr Austin’s tenure as an Established Investigator of the traits is an important alternative to traditional approaches of American Heart Association. We would like to thank the families for estimating independent effects when the traits have a comtheir participation in this study. The authors also wish to thank Andy Louie for his technical assistance. mon etiologic pathway. This may be particularly true when considering genetically related risk factors. References Furthermore, from the standpoint of genetic mapping, 1. Gardner CD, Fortmann SP, Krauss RM. Association of small low-density large-magnitude genetic correlations observed in the current lipoprotein particles with the incidence of coronary artery disease in men study, particularly those between LDL size and plasma TG, and women. JAMA. 1996;276:875– 881. 2. Lamarche B, Tchernof A, Mauriege P, Cantin B, Dagenais GR, Lupien can be used to delimit major-locus hypotheses, including PJ, Despres JP. Fasting insulin and apolipoprotein B levels and lowmajor-locus pleiotropy and linkage, that can then be tested by density lipoprotein particle size as risk factors for ischemic heart disease. multivariate segregation analysis and combined segregation JAMA. 1998;279:1955–1961. and linkage analysis.21,57 For example, using a multivariate 3. Stampfer MJ, Krauss RM, Ma J, Blanche PJ, Holl LG, Sacks FM, Hennekens CH. A prospective study of triglyceride level, low-density lipid phenotype characterized by LDL size, plasma TG, and lipoprotein particle diameter, and risk of myocardial infarction. JAMA. HDL-C may be advantageous when pleiotropy is due primar1996;276:882– 888. 57 ily to major loci. Evidence for linkage with the multivariate 4. Criqui MH, Heiss G, Cohn R, Cowan LD, Chirayath MS, Bangdiwala S, http://atvb.ahajournals.org/ by guest JulyDR 2, Jr, 2015 Kritchevsky S, on Jacobs O’Grady HK, Davis CE. Plasma triglycerlipid phenotype could representDownloaded a gene thatfrom is common to all
Edwards et al
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
Quantitative Genetic Analysis of Lipids/Lipoproteins
2463
25. Edwards KL, Austin MA, Newman B, Mayer E, Krauss RM, Selby JV. ide level and mortality from coronary heart disease. N Engl J Med. 1993;328:1220 –1225. Multivariate analysis of the insulin resistance syndrome in women. ArteHokanson JE, Austin MA. Plasma triglyceride level is a risk factor for rioscler Thromb. 1994;14:1940 –1945. cardiovascular disease independent of high-density lipoprotein cholester26. Sprecher DL, Feigelson HS, Laskarzewski PM. The low HDL cholesterol level: a meta-analysis of population-based prospective studies. J Carol/high triglyceride trait. Arterioscler Thromb. 1993;13:495–504. diovasc Risk. 1996;3:213–219. 27. Sprecher DL, Hein MJ, Laskarzewski PM. Conjoint high triglycerides Gordon DJ, Probstfield JL, Garrison RJ, Neaton JD, Castelli WP, Knoke and low HDL cholesterol across generations: analysis of proband hyperJD, Jacobs DR Jr, Bangdiwala S, Tyroler HA. High-density lipoprotein triglyceridemia and lipid/lipoprotein disorders in first-degree family cholesterol and cardiovascular disease: four prospective American members. Circulation. 1994;90:1177–1184. studies. Circulation. 1989;79:8 –15. 28. Austin MA, King MC, Vranizan KM, Krauss RM. Atherogenic NIH Consensus Development Panel on Triglyceride, H-DL, and Corolipoprotein phenotype: a proposed genetic marker for coronary heart nary Heart Disease. Triglyceride, high-density lipoprotein, and coronary disease risk. Circulation. 1990;82:495–506. heart disease. JAMA. 1993;269:505–510. 29. Austin MA, Rodriguez BL, McKnight B, Curb JD, Sharp DS. LowBrunzell JD, Schrott HG, Motulsky AG, Bierman EL. Myocardial density lipoprotein (LDL) particle size and plasma triglyceride as preinfarction in the familial forms of hypertriglyceridemia. Metabolism. dictors of coronary heart disease in older Japanese-American men. Cir1976;25:313–320. culation. 1997;96(suppl I):I-337. Abstract. Genest J, Martin-Munley S, McNamara J, Ordovas J, Jenner J, Myers R, 30. Edwards KL, Newman B, Mayer E, Selby JV, Krauss RM, Austin MA. Silberman S, Wilson P, Salem D, Schaefer E. Familial lipoprotein disHeritability of factors of the insulin resistance syndrome in women twins. orders in patients with premature coronary artery disease. Circulation. Genet Epidemiol. 1997;14:241–253. 1992;85:2025–2033. 31. Edwards KL, Talmud PJ, Luong LA, Haddad L, Day INM, Newman B, Goldstein JL, Schrott HG, Hazzard WR, Bierman EL, Motulsky AG. Louie AW, Krauss RM, Humphries SE, Austin MA. Sib-pair linkage Hyperlipidemia in coronary heart disease, II: genetic analysis of lipid analysis of factors of the insulin resistance syndrome (IRS) in dizygotic levels in 176 families and delineation of a new inherited disorder, women twins. Can J Cardiol. 1997;13(suppl B):165B. Abstract. combined hyperlipidemia. J Clin Invest. 1973;52:1544 –1568. 32. Austin MA, McKnight B, Edwards KL, Wallace KJ, Bradley CM, Schaefer EJ, Genest JJ Jr, Ordovas JM, Salem DN, Wilson PW. Familial McNeely MJ, Psaty BM, Brunzell JD, Motulsky AG. Twenty-year carlipoprotein disorders and premature coronary artery disease. Atherosclediovascular disease (CVD) mortality in familial combined hyperlipidemia rosis. 1994;108(suppl):S41–S54. and familial hypertriglyceridemia: an historical cohort study. Circulation. Jarvik GP, Brunzell JD, Austin MA, Krauss RM, Motulsky AG, Wijsman 1997;96(suppl I):I-142. Abstract. E. Genetic predictors of FCHL in 4 large pedigrees: influence of apoB 33. Austin MA. Plasma triglyceride and coronary heart disease: epidemilevel major locus predicted genotype and LDL subclass phenotype. Arteologic, statistical, and genetic issues. In: Proceedings of the International rioscler Thromb. 1994;14:1687–1694. Congress of Triglycerides: The Role in Diabetes and Atherosclerosis. Cullen P, Farren B, Scott J, Farrall M. Complex segregation analysis Athero Rev. 1991:65–70. provides evidence for a major gene acting on serum triglyceride levels in 34. Lipid Research Clinics Program Manual of Laboratory Operations. Lipid 55 British families with familial combined hyperlipidemia. Arterioscler and Lipoprotein Analysis. Bethesda, Md: National Institutes of Health; Thromb. 1994;14:1233–1249. 1974. Austin MA, Newman B, Selby JV, Edwards K, Mayer EJ, Krauss RM. 35. Nichols AV, Krauss RM, Musliner TA, eds. Nondenaturing PolyacrylGenetics of LDL subclass phenotypes in women twins: concordance, amide Gradient Gel Electrophoresis. Orlando, Fla: Academic Press; heritability, and commingling analysis. Arterioscler Thromb. 1993;13: 1986. 687– 695. 36. Krauss RM, Burke DJ. Identification of multiple subclasses of plasma Austin MA, Jarvik GP, Hokanson JE, Edwards K. Complex segregation low density lipoproteins in normal humans. J Lipid Res. 1982;23:97–104. analysis of LDL peak particle diameter. Genet Epidemiol. 1993;10: 37. Dyke B. PEDSYS: A Pedigree Data Management System. San Antonio, 599 – 604. Tex: Population Genetics Laboratory, Southwest Foundation for BioAustin MA, Brunzell JD, Fitch WL, Krauss RM. Inheritance of low medical Research; 1992. density lipoprotein subclass patterns in familial combined hyperlipid38. Blangero J, Konigsberg L. Multivariate segregation analysis using the emia. Arteriosclerosis. 1990;10:520 –530. mixed model. Genet Epidemiol. 1991;8:299 –316. Bredie SJ, Kiemeney LA, de Haan AF, Demacker PN, Stalenhoef AF. 39. Falconer DS. Introduction to Quantitative Genetics. 3rd ed. Essex, UK: Inherited susceptibility determines the distribution of dense low-density Longman; 1989. lipoprotein subfraction profiles in familial combined hyperlipidemia. 40. Boehnke M, Moll PP, Kottke BA, Weidman WH. Partitioning the variAm J Hum Genet. 1996;58:812– 822. ability of fasting plasma glucose levels in pedigrees: genetic and envide Graaf J, Swinkels DW, de Haan AF, Demacker PN, Stalenhoef AF. ronmental factors. Am J Epidemiol. 1987;125:679 – 689. Both inherited susceptibility and environmental exposure determine the 41. Comuzzie AG, Blangero J, Mahaney MC, Mitchell BD, Stern MP, low-density lipoprotein-subfraction pattern distribution in healthy Dutch MacCluer JW. Genetic and environmental correlations among skinfold families. Am J Hum Genet. 1992;51:1295–1310. measures. Int J Obes Relat Metab Disord. 1994;18:413– 418. Coresh J, Beaty TH, Prenger VL, Xu J, Kwiterovich PO Jr. Segregation 42. Hopper JL, Mathews JD. Extensions to multivariate normal models for analysis of HDL3-C levels in families of patients undergoing coronary pedigree analysis. Ann Hum Genet. 1982;46:373–383. arteriography at an early age. Arterioscler Thromb Vasc Biol. 1995;15: 43. Lange K, Boehnke M. Extensions to pedigree analysis, IV: covariance 1307–1313. components models for multivariate traits. Am J Med Genet. 1983;14: Heller DA, de Faire U, Pedersen NL, Dahlen G, McClearn GE. Genetic 513–524. and environmental influences on serum lipid levels in twins. N Engl 44. Friedlander Y, Austin MA, Newman B, Edwards K, Mayer D-EI, King J Med. 1993;328:1150 –1156. MC. Heritability of longitudinal changes in coronary-heart-disease risk Mahaney MC, Blangero J, Comuzzie AG, VandeBerg JL, Stern MP, factors in women twins. Am J Hum Genet. 1997;60:1502–1512. MacCluer JW. Plasma HDL cholesterol, triglycerides, and adiposity: a 45. Christian JC, Carmelli D, Castelli WP, Fabsitz R, Grim CE, Meaney FJ, quantitative genetic test of the conjoint trait hypothesis in the San Antonio Norton JA Jr, Reed T, Williams CJ, Wood PD. High density lipoprotein Family Heart Study. Circulation. 1995;92:3240 –3248. cholesterol: a 16-year longitudinal study in aging male twins. ArterioMahaney MC, Blangero J, Rainwater DL, Comuzzie AG, VandeBerg JL, sclerosis. 1990;10:1020 –1025. Stern MP, MacCluer JW, Hixson JE. A major locus influencing plasma 46. Heller DA, Pedersen NL, de Faire U, McClearn GE. Genetic and envihigh-density lipoprotein cholesterol levels in the San Antonio Family ronmental correlations among serum lipids and apolipoproteins in elderly Heart Study: segregation and linkage analyses. Arterioscler Thromb Vasc twins reared together and apart. Am J Hum Genet. 1994;55:1255–1267. Biol. 1995;15:1730 –1739. 47. Castellani LW, Weinreb A, Bodnar J, Goto AM, Doolittle M, Mehrabian Austin MA, Mykkanen L, Kuusisto J, Edwards KL, Nelson C, Haffner M, Demant P, Lusis AJ. Mapping a gene for combined hyperlipidaemia SM, Pyorala K, Laakso M. Prospective study of small LDLs as a risk in a mutant mouse strain. Nat Genet. 1998;18:374 –377. factor for non-insulin dependent diabetes mellitus in elderly men and 48. Pajukanta P, Nuotio I, Terwilliger JD, Porkka KV, Ylitalo K, Pihlajamaki women. Circulation. 1995;92:1770 –1778. J, Suomalainen AJ, Syvanen AC, Lehtimaki T, Viikari JS, Laakso M, Burchfiel CM, Curb JD, Sharp DS, Rodriguez BL, Arakaki R, Chyou PH, Taskinen MR, Ehnholm C, Peltonen L. Linkage of familial combined Yano K. Distribution and correlates of insulin in elderly men: the hyperlipidaemia to chromosome 1q21– q23. Nat Genet. 1998;18: Honolulu Heart Program. Arterioscler Thromb Vasc Biol. 1995;15: Downloaded from http://atvb.ahajournals.org/ by guest on July 2, 2015 2213–2221. 369 –373.
2464
Arterioscler Thromb Vasc Biol.
October 1999
49. Dallinga-Thie GM, van Linde-Sibenius Trip M, Rotter JI, Cantor RM, Bu X, Lusis AJ, de Bruin TW. Complex genetic contribution of the Apo AI-CIII-AIV gene cluster to familial combined hyperlipidemia: identification of different susceptibility haplotypes. J Clin Invest. 1997;99: 953–961. 50. Austin MA, Talmud PJ, Luong LA, Haddad L, Day IN, Newman B, Edwards KL, Krauss RM, Humphries SE. Candidate-gene studies of the atherogenic lipoprotein phenotype: a sib-pair linkage analysis of DZ women twins. Am J Hum Genet. 1998;62:406 – 419. 51. Rotter JI, Bu X, Cantor RM, Warden CH, Brown J, Gray RJ, Blanche PJ, Krauss RM, Lusis AJ. Multilocus genetic determinants of LDL particle size in coronary artery disease families. Am J Hum Genet. 1996;58: 585–594. 52. Houlston RS, Snowden C, Laker MF, Alberti KG, Humphries SE. Variation in the apolipoprotein B gene and development of type 2 diabetes mellitus. Dis Markers. 1991;9:87–96. 53. Nishina PM, Johnson JP, Naggert JK, Krauss RM. Linkage of atherogenic lipoprotein phenotype to the low density lipoprotein receptor locus on the
54.
55.
56.
57.
short arm of chromosome 19. Proc Natl Acad Sci U S A. 1992;89: 708 –712. de Bruin TW, Mailly F, van Barlingen HH, Fisher R, Castro Cabezas M, Talmud P, Dallinga-Thie GM, Humphries SE. Lipoprotein lipase gene mutations D9N and N291S in four pedigrees with familial combined hyperlipidaemia. Eur J Clin Invest. 1996;26:631– 639. Dallinga-Thie GM, Bu XD, van Linde-Sibenius Trip M, Rotter JI, Lusis AJ, de Bruin TW. Apolipoprotein A-I/C-III/A-IV gene cluster in familial combined hyperlipidemia: effects on LDL-cholesterol and apolipoproteins B and C-III. J Lipid Res. 1996;37:136 –147. Hokanson JE, Brunzell JD, Jarvik GP, Wijsman EM, Austin MA. Linkage of low-density lipoprotein size to the lipoprotein lipase gene in heterozygous lipoprotein lipase deficiency. Am J Hum Genet. 1999;64: 608 – 618. Comuzzie AG, Mahaney MC, Almasy L, Dyer TD, Blangero J. Exploiting pleiotropy to map genes for oligogenic phenotypes using extended pedigree data. Genet Epidemiol. 1997;14:975–980.
Downloaded from http://atvb.ahajournals.org/ by guest on July 2, 2015
Pleiotropic Genetic Effects on LDL Size, Plasma Triglyceride, and HDL Cholesterol in Families Karen L. Edwards, Michael C. Mahaney, Arno G. Motulsky and Melissa A. Austin Arterioscler Thromb Vasc Biol. 1999;19:2456-2464 doi: 10.1161/01.ATV.19.10.2456 Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 1999 American Heart Association, Inc. All rights reserved. Print ISSN: 1079-5642. Online ISSN: 1524-4636
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://atvb.ahajournals.org/content/19/10/2456
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Arteriosclerosis, Thrombosis, and Vascular Biology can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular Biology is online at: http://atvb.ahajournals.org//subscriptions/
Downloaded from http://atvb.ahajournals.org/ by guest on July 2, 2015
