| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
From the Clinical Research Centers |
Division of Biostatistics (Y.H., T.R., D.C.R.), Washington University School of Medicine, St. Louis, Missouri 63110-1093; Physical Activity Sciences Laboratory (J.G., J.-P.D., L.P., C.B.), Laval University, Québec GIK 7P4, Canada; Lipid Research Center (J.-P.D.), Laval University, Quebéc GIV 492, Canada; Diabetes Research Unit (A.N.), Laval University, Quebéc GIV 492, Canada; School of Kinesiology and Leisure Studies (A.S.L.) University of Minnesota, Minneapolis, Minnesota 55455; Department of Kinesiology (J.S.S.), Indiana University, Bloomington, Indiana 46202; Department of Health and Kinesiology (J.H.W.), Texas A&M University, College Station, Texas 77843; and Departments of Genetics and Psychiatry (D.C.R.), Washington University School of Medicine, St. Louis, Missouri 63110
Address all correspondence and requests for reprints to: Dr. Yuling Hong, Division of Biostatistics, Campus Box 8067, Washington University School of Medicine, St. Louis, Missouri 63110. E-mail: yuling{at}wubios.wustl.edu
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
AVF was assessed by computed tomography scan, and fasting insulin was measured by RIA in 512 members of 98 sedentary Caucasian families participating in the HERITAGE Family Study. Baseline data, collected before exercise training, were used in the present investigation. A bivariate familial correlation model was applied to evaluate whether there are familial influences that are common to insulin and AVF before and after adjustment for total fat mass (FM), and to assess the overall heritability of insulin and AVF. The maximal heritability for AVF, before and after adjustment for total FM, was 42% and 50%, respectively; and for insulin, it was 21%. Interestingly, 29% of the familial influences on insulin were also common to AVF, whereas 14% of the familial influences on AVF were shared by insulin. Furthermore, after AVF was adjusted for total FM, these common familial influences were increased to 48% and 20%.
Genes and/or familial nongenetic factors with pleiotropic effects seem to influence both AVF and plasma insulin levels to a certain degree. Genes involved in the regulation of lipid storage and mobilization in the abdominal fat depot are potential candidates for these genetic pleiotropic effects.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
More interestingly, AVF, a useful abdominal obesity phenotype, is considered by some to be one of the most important predictors of NIDDM and atherosclerosis (3, 10, 11, 22). Furthermore, the correlation between insulin resistance and AVF has been reported to be slightly stronger than correlations between insulin resistance and other adiposity indices in some studies (23, 24, 25), although not in all (26). Genetic influences on AVF have been reported (27, 28, 29), and these effects are slightly stronger than those on sc adipose tissue and are thought to be different from those involved in determining total body FM (28, 29). In addition, there are indications of major gene effects on AVF, although the issue has not been fully resolved yet (30, 31). However, whether insulin resistance and AVF share common genetic or environmental etiologies has not yet been explored.
In the present investigation, baseline data from the HERITAGE Family Study (32) on fasting insulin and AVF, as measured by computed tomography scan, were used. One unique feature of this data set is the control of a potentially confounding source (namely, physical activity levels), because these families were selected, among other criteria, to be sedentary. Cross-trait familial correlations between fasting insulin and AVF, before and after adjustment for total FM (AVF-FM), were evaluated to determine whether there are genetic and/or familial environmental influences common to both insulin resistance and central obesity.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The HERITAGE Family Study is a multicenter exercise-training study. The main objective of the study is to assess the role of genetic factors in the cardiovascular, metabolic, and hormonal responses to aerobic exercise training in sedentary families. The HERITAGE sampling procedure, outlining the inclusion and exclusion criteria, has been described in detail elsewhere (32).
In brief, several criteria were used to screen subjects for participation. First, children were required to be 17–40 yr old, and parents were required to be 65 yr old or less, to reduce maturation (low end) and aging (high end) complications. Second, families were required to be sedentary, defined at baseline as having not engaged in regular vigorous physical activity over the previous 6 months (i.e. any activity lasting 30 min or more and involving a rate of energy expenditure of 7 METS or more (1 MET = 3.5 mL O2 uptake per kg body weight per min and represents the rate of energy expenditure at rest) in individuals 50 yr old or more, or 8 METS or more for younger individuals, and occurring more than once a week. Families with some nonsedentary members were included, provided that the nonsedentary individual(s) remained inactive for at least 6 months. Third, individuals with a BMI greater than 40 kg/m2 were usually excluded because of metabolic abnormalities and exercise difficulties associated with extreme obesity, unless shown to be capable of exercising on a cycle ergometer. Fourth, individuals with blood pressure levels greater than 159 mm Hg systolic and/or 99 mm Hg diastolic also were excluded. Fifth, individuals with any condition or disease that was life-threatening or that could be aggravated by cycle exercise were excluded (e.g. a malignancy; uncontrolled endocrine and metabolic disorders, including diabetes; definite or possible coronary heart disease; and chronic or recurrent respiratory problems). Finally, individuals requiring lipid-lowering or antihypertensive drugs were excluded.
In all, 98 nuclear families of Caucasian descent, each with both biological parents, and at least two biological children, were used for the present study. The sample size was 512 subjects. Families of African-American descent were also recruited, but their data are not reported here.
Measures
All participants underwent a battery of tests both before and after completing the 20-week standardized exercise training program. Results are limited to the baseline (preexercise training) evaluation in the present study.
Fasting plasma insulin. Blood samples were collected under EDTA, and the tubes were centrifuged at 1000 x g, at a temperature of 4 C, for 10 min. Plasma was kept frozen at -20 C until the time of assay. Plasma insulin levels were measured by RIA after polyethylene glycol separation, as described by Desbuquois and Aurbach (33). Polyclonal antibodies that cross-react more than 90% with proinsulin (and presumably, with its conversion intermediates) were used (34). Therefore, in this study [as in others (34, 35)], insulin refers to immunoreactive insulin (defined as the sum of insulin, proinsulin, and split-proinsulin). In the present cohort, with normal fasting glucose levels and no history of diabetes, it is estimated that about 10% of the immunoreactive insulin is in the form of proinsulin and its conversion intermediates (35). Insulin levels were treated as missing, for three individuals with insulin antibodies, four individuals with extremely low glucose disappearance rate, and one individual with both conditions. All the assays were performed at a central laboratory in Quebec. The intra- and interassay coefficients of variation were 7.7% and 10.3%, respectively.
AVF, abdominal sc fat (ASF), and total FM. AVF and ASF levels were measured by computed tomography scan (36). Subjects were examined in a supine position with their arms stretched above the head. The abdominal scan was obtained between the fourth and fifth lumbar vertebrae. The attenuation interval used in the quantification of the areas of adipose tissue ranged from -190 to -30 Hounsfield units. The AVF area was defined by drawing a line within the muscle wall surrounding the abdominal cavity. The ASF area was calculated by subtracting the amount of visceral fat from the total abdominal fat area. Underwater weighing was performed to determine FM, which was converted to percent fat by using the equation of Siri (37). A correction was made for residual lung volume by the oxygen dilution method (38). At the Laval University Clinical Center, residual lung volume was assessed by the helium-dilution technique (39).
Statistical analysis
Because fasting insulin tends to have a skewed distribution, the data were transformed using a natural logarithm before any data analysis.
Age adjustment. Adjustments for the effects of age on fasting insulin and AVF were carried out separately in the four sex-by-generation groups (fathers, mothers, sons, and daughters) using a stepwise multiple-regression procedure. Age, age2, and age3 were included in the regression model. The significance level for retaining the terms in the stepwise regression analysis was 5%. Age was a significant predictor for AVF in fathers (age term accounting for 6.3% of the variance), mothers (age3 term explaining 4.7% of the variance), sons (age term accounting for 30.1% of the variance), and daughters (age3 term explaining 17.1% of the variance). When FM was added into the regression model, FM and age were significant predictors for AVF in fathers (FM and age terms accounting for 29.9% and 3.2% of the variance, respectively), mothers (FM explaining 55.9% of the variance), sons (FM, age, and age3 terms accounting for 73.3%, 4.7%, and 0.9% of the variance, respectively), and daughters (FM and age3 explaining 57.5% and 4.9% of the variance). Age was also a significant predictor for fasting insulin in daughters (age and age3 terms accounting for 11.4% of the variance) but not in fathers, mothers, and sons. Standardized residuals from the best regression models were used in the following bivariate familial correlation analysis.
Bivariate familial correlation model. The bivariate correlation model is the multivariate extension of the univariate case. In the univariate case, there are four types of family members (f, fathers; m, mothers; s, sons; and d, daughters), leading to eight interindividual correlations (fm, fs, fd, ms, md, ss, sd, and dd). In addition to estimating these interindividual correlations for each trait, the bivariate analysis also evaluates the inter- and intraindividual cross-trait correlations. For example, the interindividual cross-trait correlation between father’s trait 1 and son’s trait 2 (f1s2) and the intraindividual cross-trait correlation in fathers (f12) are estimated. Here, subscript 1 refers to fasting insulin, and subscript 2 refers to AVF or AVF-FM. These cross-trait correlations are the focus of the current study.
The computer program SEGPATH (40) was used to estimate the familial
correlations based on maximum likelihood methods. SEGPATH is a
general-purpose program that can be used to generate any linear model
for analyzing pedigree data and is based on flexible,
model-specification syntax. Here, the statistical method of analysis
fits the model directly to the family data, under the assumption that
the phenotypes in a family follow jointly a multivariate normal
distribution. The application of SEGPATH for bivariate analysis has
been published elsewhere (16). In summary, 34 correlations are
estimated in the bivariate analysis (in contrast to 8 in the univariate
analysis). There are 18 cross-trait correlations: 14 interindividual,
for siblings (s1s2,
s1d2, s2d1, and
d1d2), parent-offspring
(f1s2, f2s1,
f1d2, f2d1,
m1s2, m2s1,
m1d2, and m2d1), and
spouse (f1m2 and f2m1);
and 4 intraindividual (s12, d12,
f12, and m12). The remaining 16 are
interindividual correlations within a trait, 8 for each of the 2 traits
(e.g. s1d1,
s1s1, d1d1,
f1s1, f1d1,
m1s1, m1d1, and
f1m1 for trait 1) (see Table 1
for details).
|
).
The significance of each set of familial correlations is tested by
comparing the log likelihood of a reduced model, where some of the
correlations are fixed to zero, against the log likelihood obtained
from the general model, where all familial correlations are estimated.
The likelihood ratio test, which is the difference between twice the
log likelihoods, is distributed as a 
2. The degrees of
freedoms are given by the difference in the number of parameters
estimated in the two nested models. A 2 with a
P value of less than 0.05 is taken to suggest that the set
of familial correlations (set to zero under a null hypothesis) is
significant. Moreover, a P value less than 0.10, but greater
than 0.05, will be considered here as indicating a correlation of
marginal significance. The most parsimonious model is derived from
combining all nonrejected models.
|
for details about model tests). The maximal cross-trait heritability
can be computed using the following equation: h2 =
(rsibling + rparent-offspring) (1 +
rspouse) / (1 + rspouse + 2
rspouserparent-offspring), where, r represents
the interindividual cross-trait correlations, and the heritability is
adjusted for the degree of spouse resemblance, if present. The
heritability for each trait also can be estimated, based on the above
mentioned equation, where r represents interindividual correlations
within a trait. It should be noted that this is a generalized
heritability: both genetic and familial environmental (if significant)
effects are included. |
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
presents the means and
SDs of the unadjusted fasting insulin and AVF levels by
four sex and generation groups. Based on a comparison of
SEs, there are sex and generation differences in fasting
insulin and AVF. In general, parents have higher levels of fasting
insulin and AVF than offspring, and males have higher levels of these
two variables than females, in both generations. Means and
SDs of BMI, FM, and ASF, across four sex and generation
groups, are also given in Table 3, for comparison purposes.
|
. There are marginal cross-trait
correlations in siblings and parent-offspring (models 5 and 6, with
P = 0.061 and 0.092, respectively). The cross-trait
spouse correlations are not significant (model 7, with
P = 0.325), suggesting that any cross-trait family
resemblance is likely to be caused by genetic influences in common to
fasting insulin and AVF. Model 8 shows that there are significant
intraindividual cross-trait correlations (P < 0.001),
confirming the significant phenotypic correlations between fasting
insulin and AVF. The hypothesis of no cross-trait correlations, at all
(model 10, with P < 0.001), was rejected, and that of
no interindividual cross-trait correlations (model 9, with
P = 0.069) was borderline, as expected (given the
results from the individual tests). Together, this pattern endorses
familial etiologies that are shared by fasting insulin and AVF. In
addition, no sex or generation differences in the familial correlations
were found (models 2, 3, and 4).
|
. The intraindividual cross-trait
correlation (0.484) was significant, on the basis of SE
comparisons, in accordance with the results from likelihood ratio
tests. Although not all cross-trait correlations in siblings and
parent-offspring are significant, on the basis of SE
comparisons, they were marginally significant as a group, based on the
likelihood ratio test. The maximum cross-trait heritability between
fasting insulin and AVF is 6% (calculated from the equation mentioned
previously) (Table 6). Because the
cross-trait spouse correlations were not significant, this heritability
may be predominantly genetic in etiology. Based on the interindividual
correlations within trait, heritabilities for fasting insulin and AVF
are calculated as 21% and 42%, respectively (Table 6). Thus, of the
heritability of 21% for fasting insulin, 6% is caused by genetic
influences in common with AVF, suggesting that 29% (6% of 21%) of
genetic influences on insulin are shared by AVF, whereas only 14% (6%
of 42%) of the genetic influences on AVF are shared by fasting
insulin. Therefore, although a modest part of the genetic etiology of
fasting insulin is related to that of AVF, only a small part of the
genetic etiology of AVF seems to be related to that of fasting
insulin.
|
|
). The cross-trait correlations in siblings,
between insulin and AVF-FM, were significant (model, 5 with
P = 0.01), whereas the cross-trait parent-offspring
correlations were not significant (model 6, with P =
0.29). The cross-trait spouse correlation remained nonsignificant
(model 7, with P = 0.33). Parameter estimates from the
most parsimonious model are given in Table 5. From these estimates, we
calculated the heritability estimate for AVF-FM as 50% and the
cross-trait heritability as 10%. Thus, 48% (10% of 21%) of the
genetic influences on insulin are in common to AVF-FM, whereas 20%
(10% of 50%) of the genetic influences on AVF-FM are shared by
insulin. |
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The present estimates of common genetic and/or familial environmental influences on AVF and insulin are based on cross-trait familial resemblance among relatives. Although it is difficult to distinguish between genetic and familial environmental factors in intact nuclear families, the pattern of significant sibling and parent-offspring cross-trait correlations (but no spouse cross-trait correlations) found in the present study suggests that common factors for AVF and insulin are likely to be predominantly genetic. Interestingly, the magnitudes of common genetic influences shared by insulin and AVF increased after AVF was adjusted for FM, suggesting that genetic influences on AVF are partly independent from those of FM.
There has been some discussion about which adiposity phenotype is the best correlate of insulin resistance (26, 41). AVF seems to be a better predictor of insulin resistance and NIDDM (3, 10, 11, 22), as compared to other adiposity indices, although several studies have not controlled the AVF data for the concomitant effects of FM (41), and there are contradictory results (26). The present finding, that the magnitudes of common genetic influences shared by insulin and AVF increased after AVF was adjusted for FM, also indirectly supports this notion that AVF is an important independent correlate of insulin resistance.
Several other studies have investigated the common genetic and environmental influences on diabetes or insulin resistance (fasting insulin, 2-h insulin, insulin-to-glucose ratio, and homeostasis (HOMA) insulin resistance index) and adiposity phenotypes (BMI, WHR, sum of skinfold thicknesses, or various ratios of skinfold thicknesses). Moderate-to-substantial shared-genetic influences were observed for BMI, with NIDDM or HOMA insulin resistance index, in studies of middle-aged and elderly Caucasian twins from Sweden and the USA (17, 19). The pleiotropic effects of genes influencing fasting insulin levels and adiposity measures (BMI, WHR, and subscapular/triceps skinfold thickness ratio) were also found to be moderate in Mexican Americans, in a three-generation family heart study (18). The magnitude of the common genetic influences on insulin levels and adiposity seems to be lower in the present study than in the previous reports (17, 18, 19); however, the present study is the first to consider AVF. Because the genetic influences on AVF are reported to be different from those on sc adipose tissue and, in part, independent of total body FM, the pattern of common genetic influences may also be different.
Differences among the study results may also relate to different study designs and sample sources. Age-related differences in insulin resistance and obesity have been reported (42). These differences could indicate that genetic effects are also age-dependent, so that genes related to both adiposity and insulin resistance may turn on later in life. For example, in a follow-up twin study, a major gene effect was reported to be responsible for the age-related change of diastolic blood pressure (43). Alternatively, genes associated with adipose tissue accretion, over time, may be detectable well before those specifically involved in tissue-specific insulin resistance become manifest, i.e. the expression of the obesity-related genes would precede the expression of those associated with the insulin resistance phenotypes. However, longitudinal data are needed to identify properly the age-related changes in the genetic influences that are in common between adiposity and insulin, and such data are lacking. Another reason that may explain differences across studies is exclusion criteria (health condition, activity levels, etc.). Strict exclusion criteria were applied in the present study. Subjects with NIDDM, extreme obesity, hypertension, and definite or possible coronary heart disease were not included in the HERITAGE study. Because of the fact that these diseases or disease conditions may entail high genetic susceptibility, low estimates of heritability may have been obtained in this study, in comparison with other studies.
The overall heritabilities of AVF and AVF-FM, from the univariate analysis (42% and 50%), are comparable with those found in previous studies (28, 29, 31), where the heritability ranged from 47–56%. The slight differences of heritability estimates for AVF and AVF-FM may reflect the fact that subjects with newly diagnosed diabetes or with insulin antibodies were excluded in the present study. However, the heritability estimate of 21% for insulin in the present study is lower than reported in several other studies, in which the heritability estimates ranged from 40–53% (18, 44, 45, 46, 47); but it is comparable with several others, including a Dutch study with a heritability of 21% in relatively younger subjects (48). Again, the age-related nature of genetic influences could be part of the reason for different heritability estimates across studies. Longitudinal studies could provide clearer answers. Another factor leading to differences across studies is gene-environment interaction. Heritability estimates could be biased if gene-environment interactions are age-dependent.
Genes or genetic markers with effects on both insulin and obesity have been reported (20, 21), an observation that provides some support for the findings from these multifactorial analyses. Neuroendocrine mechanisms are considered to be among the primary reasons for a high AVF level and insulin resistance or hyperinsulinemia (49). Several genes, related to the hypothalamic-pituitary-adrenal axis, could be perceived as useful candidates to be investigated. For instance, a polymorphism at the glucorticoid receptor gene has been reported to contribute to the accumulation of AVF, particularly in normal-weight people (50). Whether this gene also affects the insulin levels remains to be determined. Another reasonable candidate gene for the common genetic influences on AVF and fasting insulin could be the lipoprotein lipase (LPL) gene. Alterations in LPL activity levels have been found in insulin-resistant individuals (51, 52). In addition, LPL gene polymorphisms are known to be associated with both diabetes and severity of coronary artery disease (53). The LPL gene or other genes affecting LPL activity, particularly in visceral adipocytes, could be of interest as candidate genes for the common genetic influences on insulin and AVF. Moreover, as is well established, lipolytic rates vary among human fat depots (54), and these variations in lipid mobilization from the adipose tissue may play a role in the etiology of insulin resistance and the metabolic syndrome (55). Of particular interest in the search for candidate genes with pleiotropic effects on fasting insulin and AVF are the observations that lipolysis is higher in visceral adipocytes than in abdominal sc and gluteal-femoral sc tissues and that visceral adipocytes are more resistant to the antilipolytic effects of insulin (54, 55).
In conclusion, genes and/or familial nongenetic factors with pleiotropic effects seem to influence both AVF and plasma insulin levels. However, these genes and/or familial nongenetic factors have only low effects on the covariation between AVF and fasting insulin levels. Among potential candidate genes to consider, those involved in the regulation of lipid storage and mobilization in the abdominal fat depot should be investigated, with the aim of defining these genetic pleiotropic effects.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received June 4, 1998.
Revised August 19, 1998.
Accepted August 24, 1998.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. Ibanez, L. Suarez, A. Lopez-Bermejo, M. Diaz, C. Valls, and F. de Zegher Early Development of Visceral Fat Excess after Spontaneous Catch-Up Growth in Children with Low Birth Weight J. Clin. Endocrinol. Metab., March 1, 2008; 93(3): 925 - 928. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W.-C. Hsueh, K. D. Silver, T. I. Pollin, C. J. Bell, J. R. O'Connell, B. D. Mitchell, and A. R. Shuldiner A Genome-Wide Linkage Scan of Insulin Level Derived Traits: The Amish Family Diabetes Study Diabetes, October 1, 2007; 56(10): 2643 - 2648. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. S. Fox, J. M. Massaro, U. Hoffmann, K. M. Pou, P. Maurovich-Horvat, C.-Y. Liu, R. S. Vasan, J. M. Murabito, J. B. Meigs, L. A. Cupples, et al. Abdominal Visceral and Subcutaneous Adipose Tissue Compartments: Association With Metabolic Risk Factors in the Framingham Heart Study Circulation, July 3, 2007; 116(1): 39 - 48. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. D. Driscoll, G. E. Meininger, K. Ljungquist, C. Hadigan, M. Torriani, A. Klibanski, W. R. Frontera, and S. Grinspoon Differential Effects of Metformin and Exercise on Muscle Adiposity and Metabolic Indices in Human Immunodeficiency Virus-Infected Patients J. Clin. Endocrinol. Metab., May 1, 2004; 89(5): 2171 - 2178. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Parker, J. Meyer, S. Lewitzky, J. S. Rennich, G. Chan, J. D. Thomas, M. Orho-Melander, M. Lehtovirta, C. Forsblom, A. Hyrkkö, et al. A Gene Conferring Susceptibility to Type 2 Diabetes in Conjunction With Obesity Is Located on Chromosome 18p11 Diabetes, March 1, 2001; 50(3): 675 - 680. [Abstract] [Full Text]
|
|
|
|
 |
|
 A. H. Kissebah, G. E. Sonnenberg, J. Myklebust, M. Goldstein, K. Broman, R. G. James, J. A. Marks, G. R. Krakower, H. J. Jacob, J. Weber, et al. Quantitative trait loci on chromosomes 3 and 17 influence phenotypes of the metabolic syndrome PNAS, December 19, 2000; 97(26): 14478 - 14483. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
