Identification, replication, and fine-mapping of Loci associated with adult height in individuals of african ancestry

Adult height is a classic polygenic trait of high heritability (h ² ∼0.8). More than 180 single nucleotide polymorphisms (SNPs), identified mostly in populations of European descent, are associated with height. These variants convey modest effects and explain ∼10% of the variance in height. Discovery efforts in other populations, while limited, have revealed loci for height not previously implicated in individuals of European ancestry. Here, we performed a meta-analysis of genome-wide association (GWA) results for adult height in 20,427 individuals of African ancestry with replication in up to 16,436 African Americans. We found two novel height loci (Xp22-rs12393627, P = 3.4×10⁻¹² and 2p14-rs4315565, P = 1.2×10⁻⁸). As a group, height associations discovered in European-ancestry samples replicate in individuals of African ancestry (P = 1.7×10⁻⁴ for overall replication). Fine-mapping of the European height loci in African-ancestry individuals showed an enrichment of SNPs that are associated with expression of nearby genes when compared to the index European height SNPs (P<0.01). Our results highlight the utility of genetic studies in non-European populations to understand the etiology of complex human diseases and traits.     

Assumptions and Properties of Limiting Pathway Models for Analysis of Epistasis in Complex Traits

For most complex traits, results from genome-wide association studies show that the proportion of the phenotypic variance attributable to the additive effects of individual SNPs, that is, the heritability explained by the SNPs, is substantially less than the estimate of heritability obtained by standard methods using correlations between relatives. This difference has been called the “missing heritability”. One explanation is that heritability estimates from family (including twin) studies are biased upwards. Zuk et al. revisited overestimation of narrow sense heritability from twin studies as a result of confounding with non-additive genetic variance. They propose a limiting pathway (LP) model that generates significant epistatic variation and its simple parametrization provides a convenient way to explore implications of epistasis. They conclude that over-estimation of narrow sense heritability from family data (‘phantom heritability’) may explain an important proportion of missing heritability. We show that for highly heritable quantitative traits large phantom heritability estimates from twin studies are possible only if a large contribution of common environment is assumed. The LP model is underpinned by strong assumptions that are unlikely to hold, including that all contributing pathways have the same mean and variance and are uncorrelated. Here, we relax the assumptions that underlie the LP model to be more biologically plausible. Together with theoretical, empirical, and pragmatic arguments we conclude that in outbred populations the contribution of additive genetic variance is likely to be much more important than the contribution of non-additive variance.     

Simultaneous Discovery,Estimation and Prediction Analysis of Complex Traits Using a Bayesian Mixture Model

Gene discovery, estimation of heritability captured by SNP arrays, inference on genetic architecture and prediction analyses of complex traits are usually performed using different statistical models and methods, leading to inefficiency and loss of power. Here we use a Bayesian mixture model that simultaneously allows variant discovery, estimation of genetic variance explained by all variants and prediction of unobserved phenotypes in new samples. We apply the method to simulated data of quantitative traits and Welcome Trust Case Control Consortium (WTCCC) data on disease and show that it provides accurate estimates of SNP-based heritability, produces unbiased estimators of risk in new samples, and that it can estimate genetic architecture by partitioning variation across hundreds to thousands of SNPs. We estimated that, depending on the trait, 2,633 to 9,411 SNPs explain all of the SNP-based heritability in the WTCCC diseases. The majority of those SNPs (>96%) had small effects, confirming a substantial polygenic component to common diseases. The proportion of the SNP-based variance explained by large effects (each SNP explaining 1% of the variance) varied markedly between diseases, ranging from almost zero for bipolar disorder to 72% for type 1 diabetes. Prediction analyses demonstrate that for diseases with major loci, such as type 1 diabetes and rheumatoid arthritis, Bayesian methods outperform profile scoring or mixed model approaches.     

Genome-Wide Contribution of Genotype by Environment Interaction to Variation of Diabetes-Related Traits

While genome-wide association studies (GWAS) and candidate gene approaches have identified many genetic variants that contribute to disease risk as main effects, the impact of genotype by environment (GxE) interactions remains rather under-surveyed. To explore the importance of GxE interactions for diabetes-related traits, a tool for Genome-wide Complex Trait Analysis (GCTA) was used to examine GxE variance contribution of 15 macronutrients and lifestyle to the total phenotypic variance of diabetes-related traits at the genome-wide level in a European American population. GCTA identified two key environmental factors making significant contributions to the GxE variance for diabetes-related traits: carbohydrate for fasting insulin (25.1% of total variance, P-nominal = 0.032) and homeostasis model assessment of insulin resistance (HOMA-IR) (24.2% of total variance, P-nominal = 0.035), n-6 polyunsaturated fatty acid (PUFA) for HOMA-β-cell-function (39.0% of total variance, P-nominal = 0.005). To demonstrate and support the results from GCTA, a GxE GWAS was conducted with each of the significant dietary factors and a control E factor (dietary protein), which contributed a non-significant GxE variance. We observed that GxE GWAS for the environmental factor contributing a significant GxE variance yielded more significant SNPs than the control factor. For each trait, we selected all significant SNPs produced from GxE GWAS, and conducted anew the GCTA to estimate the variance they contributed. We noted the variance contributed by these SNPs is higher than that of the control. In conclusion, we utilized a novel method that demonstrates the importance of genome-wide GxE interactions in explaining the variance of diabetes-related traits.     

Performance of genotype imputation for rare variants identified in exons and flanking regions of genes

Genotype imputation has the potential to assess human genetic variation at a lower cost than assaying the variants using laboratory techniques. The performance of imputation for rare variants has not been comprehensively studied. We utilized 8865 human samples with high depth resequencing data for the exons and flanking regions of 202 genes and Genome-Wide Association Study (GWAS) data to characterize the performance of genotype imputation for rare variants. We evaluated reference sets ranging from 100 to 3713 subjects for imputing into samples typed for the Affymetrix (500K and 6.0) and Illumina 550K GWAS panels. The proportion of variants that could be well imputed (true r²>0.7) with a reference panel of 3713 individuals was: 31% (Illumina 550K) or 25% (Affymetrix 500K) with MAF (Minor Allele Frequency) less than or equal 0.001, 48% or 35% with 0.0010.05. The performance for common SNPs (MAF>0.05) within exons and flanking regions is comparable to imputation of more uniformly distributed SNPs. The performance for rare SNPs (0.01相似文献   

Estimation of Epistatic Variance Components and Heritability in Founder Populations and Crosses

Genetic association studies have explained only a small proportion of the estimated heritability of complex traits, leaving the remaining heritability “missing.” Genetic interactions have been proposed as an explanation for this, because they lead to overestimates of the heritability and are hard to detect. Whether this explanation is true depends on the proportion of variance attributable to genetic interactions, which is difficult to measure in outbred populations. Founder populations exhibit a greater range of kinship than outbred populations, which helps in fitting the epistatic variance. We extend classic theory to founder populations, giving the covariance between individuals due to epistasis of any order. We recover the classic theory as a limit, and we derive a recently proposed estimator of the narrow sense heritability as a corollary. We extend the variance decomposition to include dominance. We show in simulations that it would be possible to estimate the variance from pairwise interactions with samples of a few thousand from strongly bottlenecked human founder populations, and we provide an analytical approximation of the standard error. Applying these methods to 46 traits measured in a yeast (Saccharomyces cerevisiae) cross, we estimate that pairwise interactions explain 10% of the phenotypic variance on average and that third- and higher-order interactions explain 14% of the phenotypic variance on average. We search for third-order interactions, discovering an interaction that is shared between two traits. Our methods will be relevant to future studies of epistatic variance in founder populations and crosses.     

Complex Variation in Measures of General Intelligence and Cognitive Change

Combining information from multiple SNPs may capture a greater amount of genetic variation than from the sum of individual SNP effects and help identifying missing heritability. Regions may capture variation from multiple common variants of small effect, multiple rare variants or a combination of both. We describe regional heritability mapping of human cognition. Measures of crystallised (g_c) and fluid intelligence (g_f) in late adulthood (64–79 years) were available for 1806 individuals genotyped for 549,692 autosomal single nucleotide polymorphisms (SNPs). The same individuals were tested at age 11, enabling us the rare opportunity to measure cognitive change across most of their lifespan. 547,750 SNPs ranked by position are divided into 10, 908 overlapping regions of 101 SNPs to estimate the genetic variance each region explains, an approach that resembles classical linkage methods. We also estimate the genetic variation explained by individual autosomes and by SNPs within genes. Empirical significance thresholds are estimated separately for each trait from whole genome scans of 500 permutated data sets. The 5% significance threshold for the likelihood ratio test of a single region ranged from 17–17.5 for the three traits. This is the equivalent to nominal significance under the expectation of a chi-squared distribution (between 1df and 0) of P<1.44×10⁻⁵. These thresholds indicate that the distribution of the likelihood ratio test from this type of variance component analysis should be estimated empirically. Furthermore, we show that estimates of variation explained by these regions can be grossly overestimated. After applying permutation thresholds, a region for g_f on chromosome 5 spanning the PRRC1 gene is significant at a genome-wide 10% empirical threshold. Analysis of gene methylation on the temporal cortex provides support for the association of PRRC1 and fluid intelligence (P = 0.004), and provides a prime candidate gene for high throughput sequencing of these uniquely informative cohorts.     

A latent variable partial least squares path modeling approach to regional association and polygenic effect with applications to a human obesity study

Genetic association studies are now routinely used to identify single nucleotide polymorphisms (SNPs) linked with human diseases or traits through single SNP-single trait tests. Here we introduced partial least squares path modeling (PLSPM) for association between single or multiple SNPs and a latent trait that can involve single or multiple correlated measurement(s). Furthermore, the framework naturally provides estimators of polygenic effect by appropriately weighting trait-attributing alleles. We conducted computer simulations to assess the performance via multiple SNPs and human obesity-related traits as measured by body mass index (BMI), waist and hip circumferences. Our results showed that the associate statistics had type I error rates close to nominal level and were powerful for a range of effect and sample sizes. When applied to 12 candidate regions in data (N = 2,417) from the European Prospective Investigation of Cancer (EPIC)-Norfolk study, a region in FTO was found to have stronger association (rs7204609∼rs9939881 at the first intron P = 4.29×10⁻⁷) than single SNP analysis (all with P>10⁻⁴) and a latent quantitative phenotype was obtained using a subset sample of EPIC-Norfolk (N = 12,559). We believe our method is appropriate for assessment of regional association and polygenic effect on a single or multiple traits.     

Predicting diabetic nephropathy using a multifactorial genetic model

Aims

The tendency to develop diabetic nephropathy is, in part, genetically determined, however this genetic risk is largely undefined. In this proof-of-concept study, we tested the hypothesis that combined analysis of multiple genetic variants can improve prediction.

Methods

Based on previous reports, we selected 27 SNPs in 15 genes from metabolic pathways involved in the pathogenesis of diabetic nephropathy and genotyped them in 1274 Ashkenazi or Sephardic Jewish patients with Type 1 or Type 2 diabetes of >10 years duration. A logistic regression model was built using a backward selection algorithm and SNPs nominally associated with nephropathy in our population. The model was validated by using random “training” (75%) and “test” (25%) subgroups of the original population and by applying the model to an independent dataset of 848 Ashkenazi patients.

Results

The logistic model based on 5 SNPs in 5 genes (HSPG2, NOS3, ADIPOR2, AGER, and CCL5) and 5 conventional variables (age, sex, ethnicity, diabetes type and duration), and allowing for all possible two-way interactions, predicted nephropathy in our initial population (C-statistic = 0.672) better than a model based on conventional variables only (C = 0.569). In the independent replication dataset, although the C-statistic of the genetic model decreased (0.576), it remained highly associated with diabetic nephropathy (χ² = 17.79, p<0.0001). In the replication dataset, the model based on conventional variables only was not associated with nephropathy (χ² = 3.2673, p = 0.07).

Conclusion

In this proof-of-concept study, we developed and validated a genetic model in the Ashkenazi/Sephardic population predicting nephropathy more effectively than a similarly constructed non-genetic model. Further testing is required to determine if this modeling approach, using an optimally selected panel of genetic markers, can provide clinically useful prediction and if generic models can be developed for use across multiple ethnic groups or if population-specific models are required.     

Contribution of Large Region Joint Associations to Complex Traits Genetics

A polygenic model of inheritance, whereby hundreds or thousands of weakly associated variants contribute to a trait’s heritability, has been proposed to underlie the genetic architecture of complex traits. However, relatively few genetic variants have been positively identified so far and they collectively explain only a small fraction of the predicted heritability. We hypothesized that joint association of multiple weakly associated variants over large chromosomal regions contributes to complex traits variance. Confirmation of such regional associations can help identify new loci and lead to a better understanding of known ones. To test this hypothesis, we first characterized the ability of commonly used genetic association models to identify large region joint associations. Through theoretical derivation and simulation, we showed that multivariate linear models where multiple SNPs are included as independent predictors have the most favorable association profile. Based on these results, we tested for large region association with height in 3,740 European participants from the Health and Retirement Study (HRS) study. Adjusting for SNPs with known association with height, we demonstrated clustering of weak associations (p = 2x10^-4) in regions extending up to 433.0 Kb from known height loci. The contribution of regional associations to phenotypic variance was estimated at 0.172 (95% CI 0.063-0.279; p < 0.001), which compared favorably to 0.129 explained by known height variants. Conversely, we showed that suggestively associated regions are enriched for known height loci. To extend our findings to other traits, we also tested BMI, HDLc and CRP for large region associations, with consistent results for CRP. Our results demonstrate the presence of large region joint associations and suggest these can be used to pinpoint weakly associated SNPs.     

Associations between APOE variants and metabolic traits and the impact of psychological stress

Objective

In a previous study, we observed that associations between APOE rs439401 and metabolic traits were moderated by chronic stress. Thus, in a population of stressed and non-stressed Danish men, we examined whether associations between APOE rs439401 and a panel of metabolic quantitative traits, all metabolic traits which may lead to T2D and CVD were moderated by psychological stress.

Methods

Obese young men (n = 475, BMI≥31.0 kg/m²) and a randomly selected control group (n = 709) identified from a population of 141,800 men were re-examined in two surveys (S-46: mean age 46, S-49: mean age 49 years) where anthropometric and biochemical measures were available. Psychological stress factors were assessed by a self-administered 7-item questionnaire. Each item had the possible response categories “yes” and “no” and assessed familial problems and conflicts. Summing positive responses constituted a stress item score, which was then dichotomized into stressed and non-stressed. Logistic regression analysis, applying a recessive genetic model, was used to assess odds ratios (OR) of the associations between APOE rs439401 genotypes and adverse levels of metabolic traits.

Results

The APOE rs439401 TT-genotype associated positively with BMI (OR = 1.09 [1.01; 1.17]), waist circumference (OR = 1.09 [1.02; 1.17]) in stressed men at S-46. Positive associations were observed for fasting plasma glucose (OR = 1.42 [1.07; 1.87]), serum triglycerides (OR = 1.41 [1.05; 1.91]) and with fasting plasma insulin (OR = 1.48 [1.05; 2.08]) in stressed men at S-49. Rs439401 TT-genotype also associated positively with surrogate measures of insulin resistance (HOMA-IR; OR = 1.21 [1.03; 1.41]) and inversely with insulin sensitivity (Stumvoll index; OR = 0.90 [0.82; 0.99], BIGTT-S_I; OR = 0.60 [0.43; 0.85]) in stressed men. No significant associations were observed in non-stressed men, albeit the estimates showed similar but weaker trends as in stressed men.

Conclusion

The present results suggest that the APOE rs439401 TT-genotype is associated with an adverse metabolic profile in a population of psychologically stressed Danish men.     

��-Catenin Is Genetically and Biologically Associated with Cortical Cataract and Future Alzheimer-Related Structural and Functional Brain Changes

Multiple lines of evidence suggest that specific subtypes of age-related cataract (ARC) and Alzheimer disease (AD) are related etiologically. To identify shared genetic factors for ARC and AD, we estimated co-heritability of quantitative measures of cataract subtypes with AD-related brain MRI traits among 1,249 members of the Framingham Eye Study who had a brain MRI scan approximately ten years after the eye exam. Cortical cataract (CC) was found to be co-heritable with future development of AD and with several MRI traits, especially temporal horn volume (THV, ρ = 0.24, P<10⁻⁴). A genome-wide association study using 187,657 single nucleotide polymorphisms (SNPs) for the bivariate outcome of CC and THV identified genome-wide significant association with CTNND2 SNPs rs17183619, rs13155993 and rs13170756 (P<2.6×10⁻⁷). These SNPs were also significantly associated with bivariate outcomes of CC and scores on several highly heritable neuropsychological tests (5.7×10⁻⁹≤P<3.7×10⁻⁶). Statistical interaction was demonstrated between rs17183619 and APP SNP rs2096488 on CC (P = 0.0015) and CC-THV (P = 0.038). A rare CTNND2 missense mutation (G810R) 249 base pairs from rs17183619 altered δ-catenin localization and increased secreted amyloid-β_1–42 in neuronal cell culture. Immunohistopathological analysis of lens tissue obtained from two autopsy-confirmed AD subjects and two non-AD controls revealed elevated expression of δ-catenin in epithelial and cortical regions of lenses from AD subjects compared to controls. Our findings suggest that genetic variation in delta catenin may underlie both cortical lens opacities in mid-life and subsequent MRI and cognitive changes that presage the development of AD.     

Milk intake and total dairy consumption: associations with early menarche in NHANES 1999-2004

Background

Several components of dairy products have been linked to earlier menarche.

Methods/Findings

This study assessed whether positive associations exist between childhood milk consumption and age at menarche or the likelihood of early menarche (<12 yrs) in a U.S sample. Data derive from the National Health and Nutrition Examination Survey (NHANES) 1999–2004. Two samples were utilized: 2657 women age 20–49 yrs and 1008 girls age 9–12 yrs. In regression analysis, a weak negative relationship was found between frequency of milk consumption at 5–12 yrs and age at menarche (daily milk intake β = −0.32, P<0.10; “sometimes/variable milk intake” β = −0.38, P<0.06, each compared to intake rarely/never). Cox regression yielded no greater risk of early menarche among those who drank milk “sometimes/varied” or daily vs. never/rarely (HR: 1.20, P<0.42, HR: 1.25, P<0.23, respectively). Among the 9–12 yr olds, Cox regression indicated that neither total dairy kcal, calcium and protein, nor daily milk intake in the past 30 days contributed to early menarche. Girls in the middle tertile of milk intake had a marginally lower risk of early menarche than those in the highest tertile (HR: 0.6, P<0.06). Those in the lowest tertiles of dairy fat intake had a greater risk of early menarche than those in the highest (HR: 1.5, P<0.05, HR: 1.6, P<0.07, lowest and middle tertile, respectively), while those with the lowest calcium intake had a lower risk of early menarche (HR: 0.6, P<0.05) than those in the highest tertile. These relationships remained after adjusting for overweight or overweight and height percentile; both increased the risk of earlier menarche. Blacks were more likely than Whites to reach menarche early (HR: 1.7, P<0.03), but not after controlling for overweight.

Conclusions

There is some evidence that greater milk intake is associated with an increased risk of early menarche, or a lower age at menarche.     

Statistical Power to Detect Genetic (Co)Variance of Complex Traits Using SNP Data in Unrelated Samples

We have recently developed analysis methods (GREML) to estimate the genetic variance of a complex trait/disease and the genetic correlation between two complex traits/diseases using genome-wide single nucleotide polymorphism (SNP) data in unrelated individuals. Here we use analytical derivations and simulations to quantify the sampling variance of the estimate of the proportion of phenotypic variance captured by all SNPs for quantitative traits and case-control studies. We also derive the approximate sampling variance of the estimate of a genetic correlation in a bivariate analysis, when two complex traits are either measured on the same or different individuals. We show that the sampling variance is inversely proportional to the number of pairwise contrasts in the analysis and to the variance in SNP-derived genetic relationships. For bivariate analysis, the sampling variance of the genetic correlation additionally depends on the harmonic mean of the proportion of variance explained by the SNPs for the two traits and the genetic correlation between the traits, and depends on the phenotypic correlation when the traits are measured on the same individuals. We provide an online tool for calculating the power of detecting genetic (co)variation using genome-wide SNP data. The new theory and online tool will be helpful to plan experimental designs to estimate the missing heritability that has not yet been fully revealed through genome-wide association studies, and to estimate the genetic overlap between complex traits (diseases) in particular when the traits (diseases) are not measured on the same samples.     

Genome-wide association analysis identifies variants associated with nonalcoholic fatty liver disease that have distinct effects on metabolic traits

Nonalcoholic fatty liver disease (NAFLD) clusters in families, but the only known common genetic variants influencing risk are near PNPLA3. We sought to identify additional genetic variants influencing NAFLD using genome-wide association (GWA) analysis of computed tomography (CT) measured hepatic steatosis, a non-invasive measure of NAFLD, in large population based samples. Using variance components methods, we show that CT hepatic steatosis is heritable (∼26%–27%) in family-based Amish, Family Heart, and Framingham Heart Studies (n = 880 to 3,070). By carrying out a fixed-effects meta-analysis of genome-wide association (GWA) results between CT hepatic steatosis and ∼2.4 million imputed or genotyped SNPs in 7,176 individuals from the Old Order Amish, Age, Gene/Environment Susceptibility-Reykjavik study (AGES), Family Heart, and Framingham Heart Studies, we identify variants associated at genome-wide significant levels (p<5×10⁻⁸) in or near PNPLA3, NCAN, and PPP1R3B. We genotype these and 42 other top CT hepatic steatosis-associated SNPs in 592 subjects with biopsy-proven NAFLD from the NASH Clinical Research Network (NASH CRN). In comparisons with 1,405 healthy controls from the Myocardial Genetics Consortium (MIGen), we observe significant associations with histologic NAFLD at variants in or near NCAN, GCKR, LYPLAL1, and PNPLA3, but not PPP1R3B. Variants at these five loci exhibit distinct patterns of association with serum lipids, as well as glycemic and anthropometric traits. We identify common genetic variants influencing CT–assessed steatosis and risk of NAFLD. Hepatic steatosis associated variants are not uniformly associated with NASH/fibrosis or result in abnormalities in serum lipids or glycemic and anthropometric traits, suggesting genetic heterogeneity in the pathways influencing these traits.     

Common SNPs in FTO Gene Are Associated with Obesity Related Anthropometric Traits in an Island Population from the Eastern Adriatic Coast of Croatia

Background

Multiple studies have provided compelling evidence that the FTO gene variants are associated with obesity measures. The objective of the study was to investigate whether FTO variants are associated with a broad range of obesity related anthropometric traits in an island population.

Methodology/Principal Findings

We examined genetic association between 29 FTO SNPs and a comprehensive set of anthropometric traits in 843 unrelated individuals from an island population in the eastern Adriatic coast of Croatia. The traits include 11 anthropometrics (height, weight, waist circumference, hip circumference, bicondilar upper arm width, upper arm circumference, and biceps, triceps, subscapular, suprailiac and abdominal skin-fold thicknesses) and two derived measures (BMI and WHR). Using single locus score tests, 15 common SNPs were found to be significantly associated with “body fatness” measures such as weight, BMI, hip and waist circumferences with P-values ranging from 0.0004 to 0.01. Similar but less significant associations were also observed between these markers and bicondilar upper arm width and upper arm circumference. Most of these significant findings could be explained by a mediating effect of “body fatness”. However, one unique association signal between upper arm width and rs16952517 (P-value = 0.00156) could not be explained by this mediating effect. In addition, using a principle component analysis and conditional association tests adjusted for “body fatness”, two novel association signals were identified between upper arm circumference and rs11075986 (P-value = 0.00211) and rs16945088 (P-value = 0.00203).

Conclusions/Significance

The current study confirmed the association of common variants of FTO gene with “body fatness” measures in an isolated island population. We also observed evidence of pleiotropic effects of FTO gene on fat-free mass, such as frame size and muscle mass assessed by bicondilar upper arm width and upper arm circumference respectively and these pleiotropic effects might be influenced by variants that are different from the ones associated with “body fatness”.     

Genome-wide population-based association study of extremely overweight young adults--the GOYA study

Background

Thirty-two common variants associated with body mass index (BMI) have been identified in genome-wide association studies, explaining ∼1.45% of BMI variation in general population cohorts. We performed a genome-wide association study in a sample of young adults enriched for extremely overweight individuals. We aimed to identify new loci associated with BMI and to ascertain whether using an extreme sampling design would identify the variants known to be associated with BMI in general populations.

Methodology/Principal Findings

From two large Danish cohorts we selected all extremely overweight young men and women (n = 2,633), and equal numbers of population-based controls (n = 2,740, drawn randomly from the same populations as the extremes, representing ∼212,000 individuals). We followed up novel (at the time of the study) association signals (p<0.001) from the discovery cohort in a genome-wide study of 5,846 Europeans, before attempting to replicate the most strongly associated 28 SNPs in an independent sample of Danish individuals (n = 20,917) and a population-based cohort of 15-year-old British adolescents (n = 2,418). Our discovery analysis identified SNPs at three loci known to be associated with BMI with genome-wide confidence (P<5×10⁻⁸; FTO, MC4R and FAIM2). We also found strong evidence of association at the known TMEM18, GNPDA2, SEC16B, TFAP2B, SH2B1 and KCTD15 loci (p<0.001), and nominal association (p<0.05) at a further 8 loci known to be associated with BMI. However, meta-analyses of our discovery and replication cohorts identified no novel associations.

Significance

Our results indicate that the detectable genetic variation associated with extreme overweight is very similar to that previously found for general BMI. This suggests that population-based study designs with enriched sampling of individuals with the extreme phenotype may be an efficient method for identifying common variants that influence quantitative traits and a valid alternative to genotyping all individuals in large population-based studies, which may require tens of thousands of subjects to achieve similar power.