Abstract
Objective. Several genome scans have been performed for adult obesity. Because single formal genetic studies suggest a higher heritability of body weight in adolescence and because genes that influence body weight in adulthood might not be the same as those that are relevant in childhood and adolescence, we performed a whole genome scan.
Methods. The genome scan was based on 89 families with 2 or more obese children (sample 1). The mean age of the index patients was 13.63 ± 2.75 years. A total of 369 individuals were initially genotyped for 437 microsatellite markers. A second sample of 76 families was genotyped using microsatellite markers that localize to regions for which maximum likelihood binomial logarithm of the odd (MLB LOD) scores on use of the concordant sibling pair approach exceeded 0.7 in sample 1.
Results. The regions with MLB LOD scores >0.7 were on chromosomes 1p32.3-p33, 2q37.1-q37.3, 4q21, 8p22, 9p21.3, 10p11.23, 11q11-q13.1, 14q24-ter, and 19p13-q12 in sample 1; MLB LOD scores on chromosomes 8p and 19q exceeded 1.5. In sample 2, MLB LOD scores of 0.68 and 0.71 were observed for chromosomes 10p11.23 and 11q13, respectively.
Conclusion. We consider that several of the peaks identified in other scans also gave a signal in this scan as promising for ongoing pursuits to identify relevant genes. The genetic basis of childhood and adolescent obesity might not differ that much from adult obesity.
The number of whole genome scans for obesity and obesity-related phenotypes has rapidly increased after publication of the initial scan pertaining to a search for genes that influence percentage body fat in Pima Indians in 1997.1 The ethnically diverse populations include Pima Indians1–4; Mexican Americans5–7; European and African Americans8–11; French Canadians12,13; Old Order Amish14; and Europeans from France,15 Finland,16–18 the Netherlands,19 and Sweden.17 Several different chromosomal regions have been identified in these whole genome scans, some of which have been confirmed in independent whole genome or regional studies, including linkage to chromosomes 2p,7,15 7q,11,14,20–25 10p and q,9,15,26,27 and 20q.9,28
One of the highest heritability estimates for body mass index (BMI; kg/m2) has been determined in a twin study based on adolescents.29 These findings suggest that heritability might even be higher at this age than in adulthood, for which estimates derived from twin studies typically range in the magnitude of 0.6 to 0.8.30,31
Human obesity as a result of rare single gene mutations such as in the leptin32,33 and leptin receptor genes34 typically manifests early in life. Nonsense, frameshift, and functionally relevant missense mutations in the melanocortin-4 receptor gene, which occur with a frequency of 2% to 4% among extremely obese adolescents and adults, are often associated with (extreme) obesity during childhood.35–41 It has been estimated that only 40% of the genes that influence BMI at age 20 continue to do so at ages 40 and 60.42 Nevertheless, obesity, in particular extreme obesity in adolescence, commonly persists in adulthood,43 the risk being even higher when at least 1 parent is also obese.44
Currently, no published whole genome scan for obesity has been based on children or adolescents as index patients. In light of the potentially stronger genetic determination of childhood and adolescent BMI and the possibility of age-dependent genetic influences on body weight, genome-wide scans based on children and adolescents are of obvious interest. Furthermore, scans based on young probands entail the advantage that the parents can be readily ascertained, thus enabling more accurate determination of the identity by descent status. Because the typical complications of obesity, including non-insulin-dependent diabetes and hypertension, have not fully become manifest at adolescence, these complications cannot be co-assessed in a scan based on young siblings. Particularly, a scan based on adolescents is complicated by the fact that several obesity-related traits are influenced by pubertal status; serum leptin levels represent a good and well-characterized example45 of this phenomenon.
The standard definitions for obesity based on absolute BMI46 cannot be applied to children and adolescents. In our molecular genetic studies, we have used BMI centiles based on the representative German National Nutrition Survey47 to define the degree of obesity of our index patients in the age range 5 to 22 years and their siblings (eg,26,38,39,48). Different centiles have been used to define overweight and obesity in children and adolescents, including the 85th and 95th49 and the 90th and 97th50 centiles, respectively. On the basis of the extremely concordant sibling pair approach (ECSP51) we have ascertained obese index patients and their siblings via a BMI ≥95th for 1 sibling and ≥90th centile for the other(s). Under consideration of the parameters estimated in previous segregation analyses for obesity, we have shown that the ECSP is better suited than the extreme discordant sibling pair approach to detect linkage.52 On use of the ECSP approach, we have confirmed linkage of obesity to chromosome 10p based on a regional scan encompassing 93 families with 2 or more obese offspring.26
In this study, we present for the first time results of a whole genome scan based on 89 young obese affected sibling pairs. The 452 microsatellite markers were spaced at an average distance of 8.4 cM and included markers for fine mapping; maximum likelihood binomial logarithm of the odd (MLB LOD) scores53 were calculated to determine linkage on the basis of the ECSP approach. We were subsequently able to genotype an additional 76 families with 2 young obese offspring in chromosomal regions of interest identified in the first group.
METHODS
Ascertainment of obese index patients was performed at 3 German hospitals (Klinik Hochried, Murnau; Adipositas Rehabilitationszentrum Insula, Berchtesgaden; and Spessart Klinik, Bad Orb) that specialize in the inpatient treatment of extremely obese children and adolescents. Families that were willing to participate were included when 1) at least 1 offspring had an age- and gender-specific BMI centile ≥95, 2) at least 1 sibling had an age- and gender-specific BMI centile ≥90, and 3) the DNA of both biological parents was available. There were 77, 11, and 1 families with 2, 3, and 4 obese children, respectively. Accordingly, the total number of individuals genotyped for the whole genome scan was n = 369 (sample 1). During genotyping of sample 1, an additional 76 families (sample 2) were recruited as part of an ongoing ascertainment to enable a future genome scan based on 300 families. Sample 2 also fulfilled the aforementioned 3 criteria. Again, the majority of these families (n = 68) had 2 obese offspring; 6, 1, and 1 had 3, 4, or 5 obese siblings, respectively. These families (sample 2) were genotyped for those markers that contributed to peak regions identified in sample 1 as defined by a MLB LOD >0.70. Finally, families of samples 1 and 2 were genotyped for 15 additional markers that localize within the identified peak regions. These markers were chosen by applying informativity criteria. Descriptive statistics for both samples are presented in Tables 1 and 2. In single families in samples 1 and 2, an obese offspring was aged ≥18 years: the oldest index patient (22 years) had an 18-year-old sibling; 18 siblings of index patients who were younger than 17 years were older than 22 years. Age- and gender-adjusted BMI centiles were calculated from the large and representative German National Nutrition Survey.47 Written informed consent was given by all participants; in the case of minors, consent was given by their parents. This study was approved by the Ethics Committee of the University of Marburg.
Descriptive Statistics Based on 89 Families With at Least 1 ECSP Used to Perform a Whole Genome Scan (Sample 1)
Descriptive Statistics Based on 76 Families With at Least 1 ECSP (Sample 2) Used in an Attempt to Confirm Linkage to Regions With an MLB LOD score >0.70 as Detected in Sample 1
Genotyping
DNA was isolated from peripheral white blood cells using standard protocols.48 The Gene Mapping Centre panel of 372 highly polymorphic microsatellite markers with an average distance of 9.9 cM and an average heterozygosity of 0.78 was selected from the final Généthon linkage map as previously described.54 In brief, markers were amplified on microtiter plates in single reactions on Tetrad polymerase chain reaction (PCR) machines (MJ Research Biozym, Hessisch Oldendorf, Germany). All pre- and post-PCR pipetting steps were performed using robotic devices. PCR product pools were separated on ABI377XL (Applied Biosystems [ABI], Darmstadt, Germany) sequencers and on MegaBace sequencers (Pharmacia Amersham, Freiburg, Germany), respectively. Semiautomated genotyping was performed using the Genescan and Genotyper (ABI) software and the genetic profiler in the case of MegaBace data. Instrument allele calling was checked manually. All genotypes were subject to an automatic Mendelian check using the Linkrun routine (T. F. Wienker, unpublished), which in turn calls the program Unknown v5.20 from the Linkage Package.55 All allele sizes were standardized to known Centre d’Etude Polymorphisme Humain control individuals. Sixty-five additional markers were typed where the total information content56 was below 0.6 in our samples. Statistical analyses included Crimap to minimize false double recombinants. Peak regions from sample 1 with MLB LOD scores >0.70 were then typed with 2 additional flanking markers on each side of the peak, encompassing 10 cM on either side of the peak loci on chromosomes 1, 2, 4, 8, 9, 10, 11, 14, and 19.
Statistical Analysis
We conducted model-free linkage analysis because the mode of inheritance is unknown for obesity. Multipoint LOD score analysis was performed using the MLB statistics as implemented in MLBGH, Version 1.0.53 This test statistic is based on the binomial distribution of parental alleles among extremely concordant offspring and accounts for multiple sibships in a natural way.
Gender-averaged map distances (cM) from the Généthon map were transformed to recombination fractions and vice versa using Haldane’s map function. X chromosomal calculations were conducted with Genehunter, version 1.3.56
RESULTS
The whole genome scan performed with sample 1 based on all 452 markers did not reveal a MLB LOD score ≥2. Two peaks on chromosomes 8p and 19q surpassed a MLB LOD score of 1.5. MLB LOD scores of 2 additional peaks on chromosomes 2q and 11q were ≥1.0. Figure 1 demonstrates that 2 peaks on chromosomes 10p and 11q out of the total of 9 peak regions, for which sample 2 was also genotyped, revealed a MLB LOD score ≥0.5. In addition, Table 3 gives an overview of all 21 markers with 2-point LOD scores ≥0.70 in the first sample. These markers primarily contribute to the reported multipoint LOD scores.
MLB LOD scores for chromosomes 1 to 12 (A) and 13 to 22 and X (B) based on a genome scan of 89 families with 2 or more obese children (black lines) and MLB LOD scores obtained in an additional sample of 76 families in regions of interest (gray lines).
Two-Point MLB LOD Scores >0.70 in Sample 1 and Corresponding LOD Scores in Sample 2 (if Available)
DISCUSSION
To our knowledge, this study represents the first genome scan for adolescent obesity; the mean ages of the index patients and their siblings range between 13 and 15 years (Tables 1 and 2). Only single offspring were older than 18 years; in all sibships, 1 sibling was aged ≤18 years. Because the majority of the index patients had BMIs above the maximal BMI observed in the age- and gender-matched population-based reference group,47 it seems reasonable to assume that the onset of obesity dated before age 10 in most of the index patients and their siblings. We had hypothesized that a genome scan based on childhood- and adolescent-onset obesity has a greater potential to detect relevant chromosomal regions than a scan based on obese adult sibling pairs. This hypothesis stemmed from findings indicating a potentially higher genetic load in childhood and adolescent obesity.29 Furthermore, such young sibling pairs are more homogeneous with respect to age at onset, thus potentially limiting a major source of heterogeneity. Finally, and in contrast to most genome scans based on adult probands, we ascertained both parents of all of our young sibling pairs so that the parental phase was primarily used as source of information instead of allele frequencies.
Recently, Altmüller et al57 reviewed 101 published genome scans for complex disorders. They pointed out that most of the analyzed studies were not able to detect “significant” linkage according to the Lander and Kruglyak criteria.58 The results of our genome scan based on only 89 families fall within this category. Despite our failure to detect suggestive evidence for linkage according to the strict Lander and Kruglyak criteria,58 the following aspects need to be considered:
First, for 77 of our 89 families, only 2 obese offspring were ascertained. The advantage of a genome scan based on mostly single and independent sibling pairs is that the respective results can be considered more representative of families with obese offspring in a given population than a scan that includes a mixture of both small and large or only large sibships. However, because heterogeneity of obesity is evident, the reliance on single sibling pairs entails the disadvantage that a hypothetical major gene operative in a limited number of families cannot lead to a high LOD score in a small sample. For identifying such major genes, large pedigrees with several affected family members should be sampled.
Second, despite the low MLB LOD scores, it is of interest to observe that some of the peaks localize to the same or close to chromosomal regions that have previously been detected in other scans based on adult populations of European origin (Table 4). Thus, previously identified peaks on chromosomes 1p,19 2q,15 8q,9 10p,15 11q,9 and 14q2 also gave a peak signal in the current scan. The moderate size of our peaks is in line with considerations that substantially larger sample sizes are needed to replicate previous findings with LOD scores that fulfill the Lander and Kruglyak criteria.58 In addition, it is worthwhile to point out that with the exception of our peaks on chromosome 14q and 19p, all of the other peaks with a MLB ≥0.70 have previously been identified in 2 of the scans based on extremely obese probands of European origin.9,15 Because in both of these studies extremely obese adult index patients were ascertained, it is possible that several of these also had a childhood onset of their obesity.
Chromosomal Regions Identified Both in the Current and in Previous Genome Scans
Finally, despite the finding that study groups several times the size of the original sample need to be analyzed to confirm reliably the linkage regions,59 2 of our peaks (chromosome 10p and 11q) also showed up—albeit weakly—in our second sample. The chromosome 10p linkage currently can be considered one of the most consistent findings in obesity scans.15,26,27 Promising candidate genes localized within the chromosome 10 and 11 regions of interest include glutamic acid decarboxylase 2 (chromosome 10) and angiotensin receptor-like 1, ciliary neurotrophic factor, galanin, and uncoupling proteins 2 and 3 (www.ensembl.org/). Previously, we had not found evidence for association of a null allele of the ciliary neurotrophic factor gene with obesity.60 Furthermore, despite a positive association study pertaining to an uncoupling protein 2 promoter polymorphism,61 we were not able to replicate this finding; we also did not find evidence for linkage based on the families investigated in the current study.62
CONCLUSION
Whereas our scan did not reveal MLB LOD scores ≥2, we nevertheless consider that several of the previously identified peaks also gave a signal in this scan as promising for ongoing pursuits to identify genes within the respective chromosomal regions.
Acknowledgments
This work was supported by the Deutsche Forschungsgemeinschaft and by the German Bundesministerium für Forschung und Technologie (FKZ 01KW0006, 01GS0118, and 01KW9967).
We thank all probands for participation.
REFERENCES
- Copyright © 2003 by the American Academy of Pediatrics
In this issue
Jump to section
Related Articles
Cited By...
- Loss of mitochondrial protease OMA1 alters processing of the GTPase OPA1 and causes obesity and defective thermogenesis in mice
- Thematic Review Series: Glycerolipids. DGAT enzymes and triacylglycerol biosynthesis
- Quantitative Trait Loci for Fasting Glucose in Young Europeans Replicate Previous Findings for Type 2 Diabetes in 2q23-24 and Other Locations
- Genetics of obesity in Hispanic children.
- Human Galanin (GAL) and Galanin 1 Receptor (GALR1) Variations Are Not Involved in Fat Intake and Early Onset Obesity
- Identification of Two Novel Human Acyl-CoA Wax Alcohol Acyltransferases: MEMBERS OF THE DIACYLGLYCEROL ACYLTRANSFERASE 2 (DGAT2) GENE SUPERFAMILY
- A Genome-Wide Scan for Childhood Obesity-Associated Traits in French Families Shows Significant Linkage on Chromosome 6q22.31-q23.2
- APOE and TGF-{beta}1 genes are associated with obesity phenotypes