| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Southwest Foundation for Biomedical Research (W.-C.H., B.D.M., J.L.S.), San Antonio, Texas 78245; GlaxoWellcome, Inc. (P.L.S.J., M.G.E., M.J.W., D.K.B.), Research Triangle Park, North Carolina 27709; University of Maryland School of Medicine (T.I.P., A.R.S.), Baltimore, Maryland 21201; and Axys Pharmaceuticals (H.S., C.J.B.), La Jolla, California 92037
Address all correspondence and requests for reprints to: Dr. Braxton D. Mitchell, Division of Endocrinology, Diabetes, and Nutrition, University of Maryland School of Medicine, 725 West Lombard Street, Room S-420, Baltimore, Maryland 21201. E-mail: bmitchel{at}medicine.umaryland.edu
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
gene), 14q (waist: lod = 1.80), and 16p (leptin:
lod = 1.72; BMI: lod = 1.68). We also tested for linkage to
BMI-adjusted leptin concentrations and observed suggestive evidence for
linkage on chromosome 10p (lod = 2.73), approximately 10–20 cM
telomeric from obesity loci previously reported in French and German
Caucasians. Two additional linkage signals for this trait were observed
on chromosomes 7q (lod = 1.77, 
20 cM from the leptin gene) and
14q (lod = 2.47). Follow-up studies may be warranted to pursue
some of these linkage signals, especially those detected near known
obesity candidate genes, and those in regions coinciding with linkage
signals reported previously. |
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Human obesity is a multifactorial syndrome influenced by genetic, physiological, behavioral, and socio-cultural factors (5). Twin and family studies have shown that genes contribute substantially to variation in body fat accumulation. Research into the genetic determinants of obesity has progressed rapidly in the past decade. Genes responsible for several rare Mendelian forms of obesity have been identified (6), although specific genes contributing to the common form of obesity are largely unknown.
Segregation analyses suggest that single genes with relatively large effects as well as polygenes with smaller effects contribute to the development of obesity. Studies in several populations suggest that major genes inherited in a recessive manner may account for 35–45% of the variation in obesity-related traits [e.g. BMI, fat mass, and percentage of body fat (percent fat)] after adjusting for the effects of sex and age (7, 8, 9). However, other studies have failed to implicate major gene effects, inferring instead a large contribution from multiple genes, each having small effects individually (10, 11, 12). Genome-wide linkage studies have been carried out in at least four populations in an attempt to localize obesity susceptibility genes, and chromosomal regions with suggestive linkage have been reported in several populations, including Mexican Americans (13, 14), Pima Indians (15, 16, 17), and Caucasians (18, 19). In this paper we report the results of a genome scan of obesity and obesity-related traits conducted in a genetically homogeneous founder population, the Old Order Amish (OOA) in Lancaster County, PA. We also present the results of a simulation study conducted to estimate the power to detect linkage in this unique sample.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The OOA is a founder population of European origin. Nearly all of the 30,000 OOA currently alive can trace their lineages back 12–14 generations to some 200 founding couples (20, 21). The OOA are a rural-living population and are characterized by their eschewal of technological innovation and their strong interest in their ancestry and genealogical relationships. Moreover, there is considerable homogeneity in the Amish lifestyle (22).
Recruitment for the Amish Family Diabetes Study began in early 1995 with the goal of identifying genes influencing the risk of type 2 diabetes and related traits. The study protocol was approved by the institutional review board at the University of Maryland School of Medicine, and informed consent was obtained from each study participant. With the help of liaisons from the OOA community, we identified individuals with type 2 diabetes. These probands and their family members, aged 18 yr and older, were invited to participate. Between February 1995 and February 1997, 691 subjects received examinations at the Amish Research Clinic in Strasburg, PA, or in their homes. Because the recruitment is ongoing, analyses in this paper were restricted to 672 individuals in whom diabetes status was determined and on whom a genome scan was performed. The recruitment strategy and study design have been described in detail previously (22).
Phenotypes
Anthropometric measurements (including BMI, waist circumference, and percent fat) were obtained at the time of clinic or home visit. BMI was calculated as weight (kilograms) divided by height (meters) squared. A measuring tape was used to obtain minimum waist girth at the natural indentation or at the level midway between the iliac crests (and the lower edge of the rib cage if no natural indentation was present) to the nearest 0.5 cm. The percent fat was estimated by bioelectroimpedance using the resistance method (23). A weak radio frequency signal was applied to four electrodes attached to the subject’s extremities, thus allowing measurement of the resistances. Sex-specific formulas were then applied to estimate the fat-free mass based on the magnitude of the resistance and body weight. The percent body fat was estimated as total body weight minus fat-free mass, expressed as a proportion of body weight. Serum leptin concentrations were measured from fasting blood samples by RIA (Linco Research, Inc., St. Louis, MO). Its interassay coefficient of variation was 4.25%. Subjects with a BMI of 30 kg/m2 or more were defined as obese.
Genotypes
DNA was extracted from leukocytes, and a screening set of 357 highly polymorphic microsatellite short tandem repeat markers was genotyped from the ABI Prism Linkage Mapping Set (PE Applied Biosystems/Perkin-Elmer Corp., Foster City, CA). The mean marker heterozygosity was 0.75 (range, 0.33–0.91). The average intermarker interval was 10.2 cM. The largest gap between markers was 25.4 cM, occurring on chromosome 7. The genotyping error rate based on blind replicates was 0.16%, on the average.
Statistical analysis
Although all subjects can be related by tracing their ancestors back multiple generations, we divided the sample into 28 discrete families to reduce computational burden. These 28 families ranged in size from 3–69 subjects and provided a large number of relative pairs, including 606 parent-offspring pairs, 1460 sibling pairs, 2035 avuncular (aunt/uncle-niece/nephew) pairs, and 1781 first cousin pairs.
Quantitative trait multipoint linkage analysis was carried out using a
variance components methodology. We partitioned variation in the
obesity phenotype into components attributable to environmental
covariates, the additive effects of genes (i.e. residual
heritability), and a specific quantitative trait locus (QTL;
i.e. the linkage component). These analyses were conducted
using maximum likelihood procedures as implemented in the SOLAR program
(24). The residual heritability was modeled as a function
of the expected genetic covariances between relatives, and the QTL
effect was modeled as a function of the identity by descent
relationships at the marker locus. The hypothesis of linkage was
evaluated by the likelihood ratio test, which tests whether the
locus-specific effect is significantly greater than zero
(i.e. Ho:
2QTL = 0 vs.
HA: 2QTL) > 0). In
each model, we simultaneously adjusted for the effects of sex,
sex-specific age, and age (2). The total number of
individuals included for each trait-specific analysis ranged from
613–662. To reduce the skewness of the phenotype distributions,
extreme outliers were excluded from analysis. Leptin levels were
transformed by their natural logarithm. We also analyzed obesity as a
dichotomous trait, using a cut-point of BMI of 30
kg/m2 or greater to classify subjects as obese.
For this analysis, we employed an extension of the variance components
methodology, which is also implemented in the SOLAR program. The method
is based on a threshold model that assigns an individual a specific
affection status if the underlying genetically determined risk or
liability exceeds a specific threshold on a normally distributed
liability curve (25). The latent liability is assumed to
have an underlying multivariate normal distribution with equal unit
variances of liability in both the general population and relatives of
affected individuals.
After initially obtaining log odds (lod) scores for linkage by the
likelihood ratio test, we then evaluated the probability of obtaining
false positive results by generating a large number of unlinked markers
and evaluating evidence for linkage to these simulated markers. We
simulated each unlinked marker locus to have five equally frequent
alleles, and then assigned genotypes to all pedigree members based on
the genotypes simulated for the founding members of the pedigree. We
then conducted linkage analysis of each obesity-related trait with each
simulated unlinked marker. The unlinked marker loci were simulated
using PAP4 software (26). We simulated a total of 5000
unlinked markers and defined the probability of obtaining a false
positive result as the proportion of the 5000 replicates for which we
obtained a lod score higher than that observed for the original linked
locus. All lod scores reported in this paper were obtained by
converting the empirical P value obtained by simulation to
its corresponding lod score (lod =
2/[2 x
ln10]).
Power estimation
We performed a simulation study to evaluate the power of our sample to detect linkage. For this study we simulated genotypes for every pedigree member and phenotypes for those actually examined. The simulated genotypes consisted of a five-allele marker locus that was tightly linked to a two-allele obesity-causing QTL with a recombination fraction equal to zero. Allelic effects at the QTL were modeled to be additive. Eight different generating configurations with varying levels of heritability attributable to the QTL were simulated. For each generating configuration, we assumed the total heritability of the phenotype to be 40%, which approximates that estimated for BMI in our sample. We then simulated the heritability at the QTL for values ranging from 5–40%, in increments of 5%. For simulated QTL effects of less than 40%, a residual heritability was included so that the total heritability, (h2QTL + h2residual), was kept constant at 40%. The size of these gene effects is a function of the separation between the means of the two homozygous genotypes (in SD units) and the allelic frequencies at the QTL.
Two hundred replicates were simulated for each of the eight generating configurations. We analyzed each replicate dataset using variance component linkage analysis. The power to detect linkage was defined as the proportion of replicates for which we obtained a lod score of 2.0 or greater.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
). Approximately 9% of men and
13% of women had diabetes (age-adjusted P = 0.03).
Mean levels of BMI, waist circumference, percent fat, and leptin
concentrations were significantly higher in women than in men (age and
diabetes status-adjusted P < 0.001). The correlations
(r) between these traits ranged from 0.30 (waist circumference and
percent fat) to 0.83 (leptin and percent fat). Leptin concentrations
were strongly correlated with BMI (r = 0.66); however, as shown in
Fig. 1, there was considerable variation
in leptin concentrations even among individuals with the same BMI.
BMI-adjusted leptin levels were highly correlated with both percent fat
(r = 0.60) and the original (i.e. unadjusted) leptin
concentrations (r = 0.76), suggesting a potential link between
BMI-adjusted leptin levels and obesity.
|
|
and Fig. 2 show the maximum lod score for each
trait and the chromosomal locations corresponding to these peaks. Peak
lod scores were 1.68 (P = 0.0027) for BMI (chromosome
16p), 1.80 (P = 0.0020) for waist circumference
(chromosome 14q), 1.61 (P = 0.0032) for percent fat
(chromosome 3p), and 1.72 (P = 0.0024) for leptin
concentrations (chromosome 16p). There were no prominent linkage
signals detected for obesity when defined as a dichotomous trait
(having a BMI 
30 kg/m2; maximum lod = 1.05
on chromosome 8q). When we also adjusted for the effect of diabetes
status in our linkage analysis, these results were virtually unchanged
in terms of both magnitude of the lod score and location of the linkage
peak.
|
|
, the strongest evidence for linkage was observed on
chromosome 10 (71 cM) with a lod score of 2.73 (P =
0.0002). In addition, lod scores greater than 1.5 were observed in two
other chromosomal regions: one on chromosome 7 (lod = 1.77 or
P = 0.0022, at 164 cM) and the other on chromosome 14
(lod = 2.47 or P = 0.0004, at 86 cM). As before,
the results were virtually unchanged after adjusting for the effect of
diabetes status. Results from the multipoint linkage analyses for 22
autosomes are shown in Fig. 2.
The results of our power simulation are summarized in Fig. 3. These results indicate that our sample
provides good power (78%) to detect linkage (at lod scores >3.0) for
a QTL accounting for 30% or more of the total trait variance. For a
QTL accounting for 25% of the total trait variance, we would have 52%
power to detect a lod score greater than 3.0 and 76% power to detect a
lod score of 2.0 or greater. For a QTL accounting for 20% of the
phenotypic variance, the power decreases dramatically (32% power to
detect a lod score >3.0 and 52% to detect a lod score >2.0).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
40 yr) have mean BMI levels 1–2
kg/m2 higher than those of U.S. white females in
the same age group (1). Despite the well recognized
genetic contribution to human obesity, the specific genes involved are
unknown. We did not detect any lod scores at the significant level
(P < 0.05 at a genome-wide context) for any of the
traits we analyzed, perhaps reflecting the complexity of genetic
mechanisms and/or the small effects of individual genes even in this
relatively homogeneous population. However, the power of our family
collection to detect QTL effects accounting for as high as 20–25% of
total phenotypic variance was still quite low.
Peak lod scores for the primary obesity phenotypes that we analyzed
(i.e. BMI, percent fat, waist circumference, and leptin)
ranged between 1.6 and 1.8, occurring on chromosomes 3, 14, and 16. The
inclusion of the diabetes status in our linkage model did not change
the linkage results appreciably, suggesting the linkage signals we
observed in this study may reflect genetic influences on obesity alone,
rather than on both obesity and diabetes. Peak evidence for linkage to
percent fat (lod = 1.61) occurred near marker D3S3608
on chromosome 3p, less than 2 cM away from a linkage reported by Lee
et al. (19) to this same trait
(P = 0.011 estimated by single point analysis for
marker D3S1286). This region is of particular interest
because it corresponds to the location of the peroxisome
proliferator-activated receptor-
(PPAR) gene, which maps to
chromosome 3p25–24.2, approximately 2 cM telomeric from the peak
signal observed. The product of this gene is a nuclear
hormone receptor that stimulates adipogenesis (27) and
increases insulin sensitivity (28). To date, four variants
causing amino acid changes and one silent mutation have been
identified, and these variants have been associated in some populations
with obesity and altered insulin sensitivity (29, 30, 31, 32, 33).
The highest lod scores we observed in our genome scan were for the trait, BMI-adjusted leptin levels. The peak lod score we observed was 2.73 (P = 0.0002), occurring on chromosome 10p. In addition, we observed lod scores of 2.47 and 1.77, occurring on chromosomes 14q and 7q, respectively. An obvious obesity candidate gene in the region of linkage on chromosome 7 is the leptin gene (LEP). LEP has been mapped to 7q31.3, which is within the broad range of the chromosome 7 linkage signal. Rare mutations in LEP have been shown to cause leptin deficiency and extreme obesity in humans (34, 35). A linkage study conducted in Mexican Americans also reported linkage of several obesity-related traits to the gene LEP region (36).
The relevance of BMI-adjusted leptin levels to obesity is controversial. Leptin is secreted by adipose tissue and appears to signal the central nervous system regarding available energy stores. The correlation between serum leptin concentration and BMI is high in the OOA (r = 0.66), as it is in other populations. This may reflect resistance to the action of leptin (37). Although an absolute deficiency of leptin is associated with extreme obesity in both mice (38) and humans (34, 35), there is also a subset of individuals who have a relative deficiency of leptin given their level of fatness (39, 40). There is some indication that a relative deficiency of leptin may be a risk factor for weight gain. Prospectively, low leptin levels predict weight gain in Pima Indians (41). Moreover, weight-reduced postobese subjects have lower leptin levels and lower rates of fat oxidation than weight-matched (but never obese) subjects, possibly explaining the propensity of weight-losers to regain weight they lost (42). The issue remains controversial, however, as at least two other prospective studies have reported that low leptin levels do not predict weight gain (43, 44).
The causes of relative deficiency (or relative excess) of leptin levels
for a given BMI are largely unknown. To our knowledge, this is the
first demonstration that BMI-adjusted leptin concentrations are
significantly heritable, suggesting that there may be genetic
influences on this trait. Presumably, a defect in leptin regulation
could lead to secretory insufficiency, analogous to pancreatic ß-cell
defects leading to deficiencies in insulin secretion. Leptin regulation
is complex. Factors shown to influence the expression of leptin include
intracellular glucosamine concentrations, tumor necrosis factor,
insulin, glucocorticoids, interleukin-1m, c/EBP, and PPAR
(45).
To date, genome-wide linkage studies of obesity-related traits have
been reported in at least four other populations (13, 14, 15, 16, 17, 18, 19).
A QTL influencing serum leptin concentrations on chromosome 2p21 was
identified in Mexican American and French subjects (13, 18), a QTL on chromosome 8p for BMI in Mexican Americans
(14), one locus on chromosome 10p for obesity in French
subjects (18, 46), one QTL on chromosome 11q for BMI and
percent body fat in Pima Indians (15, 16), and a QTL on
chromosome 20q for BMI in white American (19).
Interestingly, our peak signal for BMI-adjusted leptin concentrations
is approximately 21 cM from a QTL influencing obesity (lod = 4.85)
in a French population (18) and within 10–20 cM of an
obesity locus reported in a German study (46). The
linkages reported in the French and German studies were both telomeric
to that observed for BMI-adjusted leptin levels in the Amish. In the
German study the maximum multipoint lod score was 2.32, approximately
19 cM from the Amish peak, and the peak two-point lod score was 2.45
(to marker TCF8), approximately 9 cM away (46).
It is difficult to evaluate whether a putative gene underlying the
linkage signal in one study (e.g. the OOA) represents the
same gene(s) as those detected by linkage in other studies
(e.g. in the French and German studies). Simulation studies
using an affected sibling pair design indicate that even when the
genetic effect is strong (e.g. lod >3 with a 
s of 2) and
the sample size is large (400 affected sibling pairs, 10 cM map
density), location estimates can be as far as 20 cM away from the true
locus location (47, 48).
In summary, we obtained suggestive evidence for linkage to BMI-adjusted
leptin concentrations on three chromosomal regions, yet little evidence
for linkage was detected for other obesity-related traits. Of
particular interest are regions on chromosomes 3, 7, and 10. These
regions have already been shown to be linked to extreme obesity and/or
obesity-related traits by one or more studies in other populations. The
linkage signals on chromosomes 3 and 7 reside in regions where known
candidate genes for obesity are located (PPAR and
LEP, respectively). The linkage signals for the trait
BMI-adjusted leptin concentrations may warrant further investigations
for their potential relevance to the development of human obesity.
|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Present address: Alameda Laboratories Pfizer, Inc.,
Alameda, California 94502.
3 Present address: National Center for Genome Resources, Santa Fe,
New Mexico 87505.
Received September 29, 2000.
Revised December 6, 2000.
Accepted December 6, 2000.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
2, a lipid-activated
transcription factor. Cell. 79:1147–1156.[CrossRef][Medline]
: a key regulator of adipogenesis
and systemic insulin sensitivity. Eur J Med Res. 2:457–464.[Medline]
2 gene with obesity in two Caucasian
populations. Diabetes. 47:1806–1808.[Medline]
gene influences plasma leptin levels in obese humans. Hum Mol
Genet. 7:435–440. associated with severe
insulin resistance, diabetes mellitus and hypertension. Nature. 402:880–883.[Medline]
2
(PPAR-2) is associated with measures of obesity in Mexican
Americans. Int J Obes. 24:522–524.
This article has been cited by other articles:
 |
|
 |
|
  W. Yang, T. Kelly, and J. He Genetic Epidemiology of Obesity Epidemiol. Rev., June 12, 2007; (2007) mxm004v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. Groves, E. Zeggini, M. Walker, G. A. Hitman, J. C. Levy, S. O'Rahilly, A. T. Hattersley, M. I. McCarthy, and S. Wiltshire Significant Linkage of BMI to Chromosome 10p in the U.K. Population and Evaluation of GAD2 as a Positional Candidate Diabetes, June 1, 2006; 55(6): 1884 - 1889. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W.-D. Li, C. Dong, D. Li, C. Garrigan, and R. A. Price A genome scan for serum triglyceride in obese nuclear families J. Lipid Res., March 1, 2005; 46(3): 432 - 438. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. I. Pollin, K. Tanner, J. R. O'Connell, S. H. Ott, C. M. Damcott, A. R. Shuldiner, J. C. McLenithan, and B. D. Mitchell Linkage of Plasma Adiponectin Levels to 3q27 Explained by Association With Variation in the APM1 Gene Diabetes, January 1, 2005; 54(1): 268 - 274. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. A. Lakka, T. Rankinen, S. J. Weisnagel, Y. C. Chagnon, T. Rice, A. S. Leon, J. S. Skinner, J. H. Wilmore, D.C. Rao, and C. Bouchard A Quantitative Trait Locus on 7q31 for the Changes in Plasma Insulin in Response to Exercise Training: The HERITAGE Family Study Diabetes, June 1, 2003; 52(6): 1583 - 1587. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Saar, F. Geller, F. Ruschendorf, A. Reis, S. Friedel, N. Schauble, P. Nurnberg, W. Siegfried, H.-P. Goldschmidt, H. Schafer, et al. Genome Scan for Childhood and Adolescent Obesity in German Families Pediatrics, February 1, 2003; 111(2): 321 - 327. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. I Steinle, W.-C. Hsueh, S. Snitker, T. I Pollin, H. Sakul, P. L St Jean, C. J Bell, B. D Mitchell, and A. R Shuldiner Eating behavior in the Old Order Amish: heritability analysis and a genome-wide linkage analysis Am. J. Clinical Nutrition, June 1, 2002; 75(6): 1098 - 1106. [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 |
, The locations of linkage
signals reported in the literature. PPARG and
LEP are the GenBank designations for the PPAR