Abstract
Most genome-wide association studies are confined to middle-aged populations. It is unclear whether associations between single nucleotide polymorphisms (SNPs) and obesity persist in old age. We aimed to relate 10 body mass index (BMI)–associated SNPs to weight, BMI, % fat, visceral and subcutaneous adipose tissue in Health ABC and AGES-Reykjavik comprising 4,846 individuals of European Ancestry, and 1,139 African Americans over age 65. SNPs were scaled using effect estimates from candidate SNPs. In Health ABC, a SNP near GNPDA2 was modestly associated with weight and SAT area (p = .008, p = .001). Risk score (sum of scaled SNPs) was associated with weight, BMI, and SAT area (p < .0001 for all), but neither GNPDA2 nor risk score was associated with weight, BMI, visceral adippose tissue, subcutaneous adipose tissue, or % fat in AGES-Reykjavik. In African Americans, a SNP near SEC16B was weakly associated with weight (p = .04). In this sample of older adults, no BMI-associated SNPs were associated with weight or adiposity.
Key Words: Obesity, Aging, Genetics, SNPs.
OVER the past 20 years, the prevalence of obesity, defined as body mass index (BMI) greater than 30kg/m2, has risen dramatically. This is a worldwide trend, affecting men and women of all ages and racial backgrounds. In the United States, nearly one third of adults are obese (1). In European countries like Iceland, obesity rates are over 20% (2), comparable to rates observed in the United States in the 1990s.
Twin and adoption studies have shown that genetic factors play a role in the development of obesity (3). Genome-wide association studies have identified common variants at numerous loci that associate with fat mass, weight, and obesity (4–6). However, these studies have largely been confined to middle-aged European ancestry (EA) populations, and it is unclear whether these relationships persist in old age or in African Americans (AA). With the increasing prevalence of obesity and population aging, it is important to understand factors associated with obesity in old age. The purpose of this study is to conduct a candidate gene analysis of associations between BMI-related loci, weight, and detailed adiposity measures in older adults.
Methods
Study Design and Participants
The Health, Aging and Body Composition Study (Health ABC) is a prospective study of 3,075 community-dwelling, initially well-functioning, AA and EA men and women aged 70–79 years at baseline. Participants were recruited from a random sample of white Medicare beneficiaries and all black Medicare-eligible residents in the Memphis, Tennessee, and Pittsburgh, Pennsylvania areas. A subsample of participants were genotyped, 1,139 AAs and 1,663 EAs. Replication was performed in 3,219 Icelandic men and women who participated in the Age Gene/Environment Susceptibility-Reykjavik study (AGES-Reykjavik). AGES-Reykjavik is a single center population study of 5,764 individuals aged 76–95 nested within the Reykjavik Study. All participants signed informed consent forms approved by institutional review boards. Detailed recruitment and study design of both studies have been published (7,8).
Anthropometric Measurements
Baseline weight was measured using a digital scale (AGES-Reykjavik) or balance beam scale (Health ABC), and height was measured using a stadiometer. BMI was calculated in kilogram per meter square. Participants in Health ABC were asked to recall their usual body weight at age 50. Midlife weight was measured in AGES-Reykjavik as part of the Reykjavik Study.
Computed tomography imaging of the abdomen at the level of the L4/L5 vertebrae was performed in both studies. Visceral adipose tissue and subcutaneous adipose tissue were estimated from a single 10-mm thick transaxial section. Adipose areas (in centimeter square) were calculated by multiplying the number of pixels of a given tissue by the pixel area. Visceral adipose tissue was manually distinguished from subcutaneous adipose tissue by tracing along the facial plane defining the internal abdominal wall. Percent body fat was measured with dual-energy x-ray absorptiometry (Hologic QDR4500A, Waltham, NY) in Health ABC and was calculated from bioelectric impedance (Xitron Hydra ECF/ICF Bio-Impedance Spectrum Analyzer) in AGES-Reykjavik.
Genetic Analysis
In Health ABC, participants were genotyped by the Center for Inherited Disease Research (CIDR) using a Illumina Human1M-Duo BeadChip array. Genomic data were extracted from buffy coat using PUREGENE DNA Purification Kit. Genotyping was successful on 1,151,215 single nucleotide polymorphisms (SNPs). Both cohorts used MACH software (version 1.0.16) to impute autosomes and criteria of minor allele frequency ≥1%, call rate ≥97%, and Hardy weinberg equilibrium (HWE) p ≤ 10–6 for SNPs used in imputation. AGES-Reykjavik genotyping was conducted on a subsample of 3,219 participants with the Illumina 370CNV BeadChip array. In Health ABC, six SNPs were imputed in EAs, rs15558902, rs571312, rs10938397, rs10767664, rs12444979, and rs543874, and six SNPs were imputed in AAs, rs1558902, rs2867125, rs571312, rs10938397, rs10767664, and rs543874. The quality of imputation was greater than 0.9750 for all SNPs in EAs and AAs. In AGES-Reykjavik, seven SNPs were imputed, rs1558902, rs571312, rs10938397, rs10767664, rs7359397, rs12444979, and rs543874. The quality of imputation was similar to Health ABC and ranged from 0.925 to 0.9998. Additional genotyping and imputation details have been published for Health ABC (9,10) and AGES-Reykjavik (11,12).
Candidate SNPs associated with obesity were identified from literature on BMI-related loci that met the following criteria: (a) sample size of both discovery and replication greater than 5,000; (b) no missing effect estimates (odds ratios, β, or confidence intervals); (c) minor allele frequency of at least 1% and 4%) p < 1 × 10−8 for both discovery and replication phases. If SNPs were in linkage disequilibrium (r 2 > .30), the SNP most strongly related to BMI based on effect estimates was included (13). Reference frequencies of candidate SNPs were compared with dbSNP build 36 of the public human genome reference to determine alleles corresponding to the forward strand. In the case of reversed allele frequency, the direction of the effect estimate was flipped. Allele counts were scaled by the effect estimates (per allele change in BMI) of candidate SNPs (effect estimate multiplied by allele frequency), which were applied to genotype data to create scaled doses for each SNP. This allows for incorporation of effects of different magnitude as the effect size for each SNP varies (14). A multilocus genetic risk score for each individual was constructed by summing the number of risk alleles for each scaled SNP.
Data Analysis
Associations were determined using linear regression, assuming additive effects of allele dosage. Midlife weight, weight in old age, BMI, percent body fat, and visceral and subcutaneous adipose tissue were analyzed as continuous dependent variables, sex, age, and study site within Health ABC as covariates, and each SNP and risk score as the predictor variable. The number of participants varies between anthropometric measurements due to missing data. All analyses were performed separately according to race. To account for the genetic substructure in the AAs, principal component analysis with 10 components was used. Use of principal component analysis to correct for population stratification is detailed by Price and coworkers (15) and has been utilized in previous genome-wide association studies, which included AAs from Health ABC (16,17). Significance was set at p < .0005 using the Bonferroni correction method based on the number of SNPs tested per phenotype per cohort to account for multiple testing burden. Analyses were conducted with R version 2.14.0 and SPSS version 19.0.
Results
From our literature search, we identified 10 candidate SNPs from study by Speliotes and coworkers (4) that met our inclusion criteria. The effect size of each SNP from Speliotes and coworkers (4) is shown in Table 1 along with the position, nearest gene, and minor allele frequencies in Health ABC and AGES-Reykjavik.
Table 1.
Candidate SNPs Associated With BMI
| SNP | Position (bp) | Nearest Gene | Health ABC | β* | Health ABC | AGES-Reykjavik | |||
|---|---|---|---|---|---|---|---|---|---|
| Al 1 | Al 2 | MAF | R 2 | MAF | R 2 | ||||
| European ancestry | |||||||||
| rs1558902 | 52,361,075 | FTO | A | T | 0.39 | 0.4268 | 0.9990 | 0.4069 | 0.9103 |
| rs2867125 | 604,210 | TMEM18 | C | T | 0.31 | 0.1795 | 0.9999 | 0.1613 | 0.9997 |
| rs571312 | 55,980,115 | MC4R | A | C | 0.23 | 0.2282 | 0.9997 | 0.2578 | 0.9644 |
| rs10938397 | 44,877,284 | GNPDA2 | A | G | −0.18 | 0.4387 | 0.9959 | 0.4008 | 0.871 |
| rs10767664 | 27,682,562 | BDNF | A | T | 0.19 | 0.2147 | 0.9983 | 0.1842 | 0.9993 |
| rs2815752 | 72,585,028 | NEGR1 | A | G | 0.13 | 0.3575 | 1.0000 | 0.4185 | 1.0000 |
| rs7359397 | 28,790,742 | SH2B1 | C | T | −0.15 | 0.3722 | 1.0000 | 0.4299 | 0.9996 |
| rs12444979 | 19,841,101 | GPRC5B | C | T | 0.17 | 0.1504 | 0.9947 | 0.104 | 0.9975 |
| rs2241423 | 65,873,892 | MAP2K5 | A | G | −0.13 | 0.2477 | 1.0000 | 0.204 | 0.9988 |
| rs543874 | 176,156,103 | SEC16B | A | G | −0.22 | 0.1643 | 0.9995 | 0.2083 | 0.9897 |
| African American | |||||||||
| rs1558902 | 52,361,075 | FTO | T | A | −0.39 | 0.102 | 0.8767 | ||
| rs2867125 | 604,210 | TMEM18 | T | C | −0.31 | 0.1165 | 0.9972 | ||
| rs571312 | 55,980,115 | MC4R | A | C | 0.23 | 0.3546 | 0.9980 | ||
| rs10938397 | 44,877,284 | GNPDA2 | G | A | 0.18 | 0.2318 | 0.9940 | ||
| rs10767664 | 27,682,562 | BDNF | A | T | 0.19 | 0.0716 | 0.9996 | ||
| rs2815752 | 72,585,028 | NEGR1 | A | G | 0.13 | 0.4381 | 1.0000 | ||
| rs543874 | 176,156,103 | SEC16B | A | G | −0.22 | 0.249 | 0.9995 | ||
Note. MAF, minor allele frequency; R2, imputation quality from MACH.
*Per allele change in kilogram per meter squared from Speliotes and coworkers (4).
A total of 6,018 participants had weight and genotype data. Baseline characteristics of participants in Health ABC and AGES-Reykjavik with measured weight are shown in Table 2. On average, participants had heavy body weights. In Health ABC, 68.2% of participants were overweight or obese, and 67.5% of participants in AGES-Reykjavik were overweight or obese. Participants in both studies averaged weight gain from midlife (6.08% and 4.12%, respectively).
Table 2.
Characteristics of Participants With Baseline Weight in Health ABC and AGES-Reykjavik
| Health ABC | AGES-Reykjavik | |
|---|---|---|
| No. of participants | 2,802 | 3,216 |
| Women, n (%) | 1,435 (51.2) | 1,865 (58.0) |
| African American, n (%) | 1,139 (40.7) | — |
| Age, mean (SD) | 73.6 (2.87) | 76.4 (5.45) |
| Weight in kg, mean (SD) | 75.9 (14.9) | 75.8 (14.7) |
| Body mass index in kg/m2, mean (SD) | 27.4 (4.77) | 27.1 (4.44) |
| <25, n (%) | 891 (31.8) | 1,042 (32.4) |
| 25.0–29.9, n(5) | 1,196 (42.7) | 1,465 (45.5) |
| ≥30.0, n (%) | 715 (25.5) | 709 (22.0) |
| % Weight change from midlife, mean (SD) | +6.08 (14.0) | +4.12 (13.0) |
| Smoking status, n (%)* | ||
| Never | 1,206 (43.0) | 1,351 (42.0) |
| Former | 1,299 (46.4) | 1,456 (45.3) |
| Current | 293 (10.5) | 409 (12.7) |
*Data missing for four participants. Cohort characteristics are shown for reference and were not statistically compared.
Table 3 shows associations between SNPs, weight in midlife and old age, and adiposity measures in EAs. In Health ABC, a SNP at rs10938397, near GNPDA2, associated with midlife weight, weight in old age (p = .008), and subcutaneous adipose tissue (p = .001), but it did not meet the statistical threshold and was not associated with any measure of weight or adiposity in AGES-Reykjavik. The overall risk score from the 10 SNPs was associated with midlife weight, weight in old age, BMI, and subcutaneous adipose tissue (p < .0001 for all) in Health ABC but not AGES-Reykjavik. There were weak trends (p < .05) between four SNPs and midlife weight in Health ABC and between one SNP and midlife weight in AGES-Reykjavik. Of these five SNPs, none of the associations carried over to old age. In AAs, a SNP at rs543874, near SEC16B, was associated with weight in midlife (p = .004) and old age (p = .05, Table 4), but the signal did not meet the statistical threshold. Risk score was also modestly associated with midlife weight (p = .02). No other associations with risk score or SNPs were observed.
Table 3.
Association of BMI-Related SNPs and Anthropometric Traits in European Ancestry Participants
| SNP | Nearest Gene | Health ABC | AGES-Reykjavik | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Midlife weight, kg | Weight, kg | BMI, kg/m2 | Midlife weight, kg | Weight, kg | BMI, kg/m2 | ||||||||||||||
| N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | ||
| rs1558902 | FTO | 1,630 | 0.03 | .44 | 1,663 | 0.03 | .30 | 1,662 | 0.06 | .03 | 3,207 | 0.02 | .26 | 3,216 | 0.03 | .22 | 3,216 | 0.03 | .09 |
| rs2867125 | TMEM18 | 1,630 | 0.05 | .11 | 1,663 | 0.03 | .25 | 1,662 | 0.04 | .09 | 3,207 | −0.02 | .47 | 3,216 | −0.01 | .68 | 3,216 | −0.01 | .69 |
| rs571312 | MC4R | 1,630 | 0.06 | .08 | 1,663 | 0.02 | .45 | 1,662 | 0.03 | .22 | 3,207 | 0.05 | .01 | 3,216 | 0.03 | .13 | 3,216 | 0.03 | .12 |
| rs10938397 | GNPDA2 | 1,630 | 0.11 | .001 | 1,663 | 0.08 | .008 | 1,662 | 0.05 | .03 | 3,207 | −0.01 | .64 | 3,216 | −0.01 | .74 | 3,216 | −0.01 | .49 |
| rs10767664 | BDNF | 1,630 | 0.06 | .06 | 1,663 | 0.03 | .26 | 1,662 | 0.02 | .51 | 3,207 | 0.04 | .09 | 3,216 | 0.05 | .01 | 3,216 | 0.04 | .01 |
| rs2815752 | NEGR1 | 1,630 | 0.07 | .03 | 1,663 | 0.03 | .35 | 1,662 | 0.02 | .38 | 3,207 | 0.03 | .10 | 3,216 | 0.02 | .40 | 3,216 | 0.02 | .34 |
| rs7359397 | SH2B1 | 1,630 | 0.09 | .005 | 1,663 | 0.04 | .23 | 1,662 | 0.01 | .82 | 3,207 | 0.04 | .07 | 3,216 | 0.02 | .23 | 3,216 | 0.02 | .31 |
| rs12444979 | GPRC5B | 1,630 | 0.02 | .46 | 1,663 | 0.06 | .03 | 1,662 | 0.06 | .01 | 3,207 | 0.02 | .32 | 3,216 | 0.00 | .90 | 3,216 | 0.01 | .51 |
| rs2241423 | MAP2K5 | 1,630 | 0.08 | .01 | 1,663 | 0.04 | .21 | 1,662 | 0.00 | .99 | 3,207 | −0.01 | .87 | 3,216 | −0.02 | .23 | 3,216 | −0.02 | .21 |
| rs543874 | SEC16B | 1,630 | 0.08 | .02 | 1,663 | 0.05 | .08 | 1,662 | 0.04 | .09 | 3,207 | −0.01 | .74 | 3,216 | −0.01 | .55 | 3,216 | −0.01 | .46 |
| Score | 1,630 | 0.18 | <.0001 | 1,663 | 0.12 | <.0001 | 1,662 | 0.11 | <.0001 | 3,207 | −0.02 | .40 | 3,216 | −0.01 | .77 | 3,216 | 0.00 | .80 | |
| % Body Fat | Visceral Adipose Tissue, cm2 | Subcutaneous Adipose Tissue, cm2 | % Body Fat | Visceral Adipose Tissue, cm2 | Subcutaneous Adipose Tissue, cm2 | ||||||||||||||
| rs1558902 | FTO | 1,600 | 0.03 | .44 | 1,604 | 0.01 | .63 | 1,574 | 0.05 | .06 | 2,418 | 0.02 | .45 | 3,172 | 0.01 | .79 | 3,172 | 0.01 | .15 |
| rs2867125 | TMEM18 | 1,600 | 0.03 | .35 | 1,604 | 0.01 | .80 | 1,574 | 0.04 | .15 | 2,418 | −0.04 | .20 | 3,172 | −0.01 | .75 | 3,172 | −0.01 | .76 |
| rs571312 | MC4R | 1,600 | 0.04 | .27 | 1,604 | −0.03 | .34 | 1,574 | 0.04 | .19 | 2,418 | 0.03 | .35 | 3,172 | 0.01 | .67 | 3,172 | 0.01 | .23 |
| rs10938397 | GNPDA2 | 1,600 | 0.09 | .01 | 1,604 | 0.01 | .61 | 1,574 | 0.09 | .001 | 2,418 | −0.01 | .65 | 3,172 | 0.00 | .97 | 3,172 | 0.00 | .05 |
| rs10767664 | BDNF | 1,600 | −0.02 | .61 | 1,604 | −0.02 | .41 | 1,574 | 0.03 | .22 | 2,418 | 0.01 | .65 | 3,172 | 0.02 | .32 | 3,172 | 0.02 | .01 |
| rs2815752 | NEGR1 | 1,600 | 0.00 | .93 | 1,604 | −0.01 | .65 | 1,574 | 0.01 | .70 | 2,418 | 0.04 | .21 | 3,172 | 0.01 | .61 | 3,172 | 0.01 | .05 |
| rs7359397 | SH2B1 | 1,600 | −0.02 | .63 | 1,604 | −0.01 | .77 | 1,574 | 0.01 | .79 | 2,418 | 0.00 | .91 | 3,172 | 0.00 | .96 | 3,172 | 0.00 | .007 |
| rs12444979 | GPRC5B | 1,600 | 0.07 | .05 | 1,604 | 0.03 | .25 | 1,574 | 0.03 | .30 | 2,418 | 0.01 | .84 | 3,172 | −0.01 | .59 | 3,172 | −0.01 | .58 |
| rs2241423 | MAP2K5 | 1,600 | 0.00 | .92 | 1,604 | 0.01 | .63 | 1,574 | 0.01 | .63 | 2,418 | 0.02 | .59 | 3,172 | 0.02 | .35 | 3,172 | 0.02 | .36 |
| rs543874 | SEC16B | 1,600 | 0.03 | .40 | 1,604 | 0.00 | .92 | 1,574 | 0.05 | .08 | 2,418 | −0.04 | .22 | 3,172 | −0.02 | .38 | 3,172 | −0.02 | .74 |
| Score | 1,600 | 0.08 | .02 | 1,604 | 0.01 | .82 | 1,574 | 0.12 | <.0001 | 2,418 | 0.00 | .94 | 3,172 | 0.00 | .96 | 3,172 | 0.00 | .59 | |
Note. Linear regression, values adjusted for age, sex, and study site (site adjustment only in Health ABC).
Table 4.
Association of BMI-Related SNPs and Anthropometric Traits in African Americans
| SNP | Nearest Gene | Health ABC | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Midlife Weight, kg | Weight, kg | BMI, kg/m2 | % Body Fat | Visceral Adipose Tissue, cm2 | Subcutaneous Adipose Tissue, cm2 | ||||||||||||||
| N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | N | Allele Effect Size (β) | p | ||
| rs1558902 | FTO | 1,080 | 0.02 | .54 | 1,139 | −0.02 | .48 | 1,139 | −0.05 | .11 | 1,099 | −0.11 | .84 | 1,096 | 0.03 | .32 | 1,055 | −0.02 | .53 |
| rs2867125 | TMEM18 | 1,080 | 0.02 | .58 | 1,139 | −0.02 | .49 | 1,139 | −0.02 | .61 | 1,099 | −0.02 | .71 | 1,096 | −0.07 | .02 | 1,055 | −0.02 | .59 |
| rs571312 | MC4R | 1,080 | 0.01 | .75 | 1,139 | 0.02 | .45 | 1,139 | 0.03 | .29 | 1,099 | 0.08 | .08 | 1,096 | 0.03 | .32 | 1,055 | 0.04 | .32 |
| rs10938397 | GNPDA2 | 1,080 | 0.05 | .16 | 1,139 | 0.02 | .61 | 1,139 | 0.02 | .52 | 1,099 | 0.00 | .95 | 1,096 | −0.01 | .67 | 1,055 | 0.02 | .62 |
| rs10767664 | BDNF | 1,080 | 0.04 | .18 | 1,139 | 0.00 | .99 | 1,139 | 0.00 | .95 | 1,099 | −0.04 | .41 | 1,096 | −0.05 | .10 | 1,055 | 0.00 | .98 |
| rs2815752 | NEGR1 | 1,080 | −0.01 | .69 | 1,139 | −0.03 | .35 | 1,139 | −0.02 | .54 | 1,099 | −0.03 | .57 | 1,096 | −0.03 | .35 | 1,055 | −0.01 | .71 |
| rs543874 | SEC16B | 1,080 | 0.09 | .004 | 1,139 | 0.06 | .05 | 1,139 | 0.04 | .15 | 1,099 | 0.03 | .53 | 1,096 | 0.02 | .55 | 1,055 | 0.03 | .36 |
| Score | 1,080 | 0.08 | .02 | 1,139 | 0.01 | .67 | 1,139 | 0.01 | .88 | 1,099 | 0.02 | .65 | 1,096 | −0.02 | .59 | 1,055 | 0.01 | 0.72 | |
Note. Linear regression, values adjusted for age, sex, and study site (site adjustment only in Health ABC).
Discussion
In this analysis, we investigated the association between BMI-related SNPs, weight, and fat mass in older EA and AAs. Our findings suggest that genes related to BMI in midlife do not track into old age. Although we found modest associations between loci and anthropometric measurements within Health ABC, there were no SNPs that replicated in AGES-Reykjavik. Within Health ABC participants, the SNPs associated with weight appear to differ between EA and AAs, which is in accordance with other studies that show inconsistent replication of obesity variants between EAs and AAs (18,19).
SNPs that were modestly associated with midlife weight did not show the same association with weight or adiposity in old age, suggesting the absence of a carryover effect of BMI-related SNPs into old age. Early lifetime environmental effects may also help to explain the discordance between BMI-related SNPs in midlife and old age. The participants in Health ABC and AGES-Reykjavik were born earlier than some of the participants in the Speliotes study (4) from which the candidate SNPs were identified, and thus environmental factors may have differed. Similarly, discordance between Health ABC and AGEs-Reykjavik may be related to differences in environmental factors as industrialization in Iceland occurred later than in the United States.
Previous studies have also reported discordance between SNPs in midlife and old age. Qi and coworkers (20) found decreased associations between an FTO variant and BMI in individuals older than 65 years compared with younger age groups. Kilpeläinen and coworkers (21) reported age-dependent effects of an obesity-related variant in PCSK1 such that the contribution to obesity in individuals older than 59 years was insignificant. A weakening association between candidate genes and obesity with age may also explain why associations in Health ABC were not replicated in AGES-Reykjavik participants who were on average older (66–95 years vs. 70–79 years) but otherwise had similar demographic characteristics as Health ABC participants.
Age-related changes in weight that result in individuals transitioning away from their midlife phenotype may underlie the diminishing associations between SNPs, weight, and BMI. In Health ABC and AGES-Reykjavik, weight fluctuation from midlife to old age was common and only one third of the participants maintained their weight within 5%. Likewise, age-related body composition changes may account for the null associations between BMI-related SNPs, visceral adipose tissue area, subcutaneous adipose tissue area, and % body fat we observed. Even in the absence of BMI changes, fat accumulates with age (22,23). Fat distribution also changes with age; in a comparison of older and middle-age men, older men had significantly higher amounts of abdominal adipose tissue despite having lower BMI (24).
A limitation of this study was the application of candidate SNPs from a genome-wide association study that was predominately EA compared with our study population that included AAs as there were no genome-wide association studies of obesity-related traits in AAs that met our selection criteria for SNPs. We also lacked access to a second cohort of AAs for replication. Additionally, the AGES-Reykjavik cohort was included in the replication stage of Speliotes and coworkers (4) from which we identified candidate genes although they comprised only 1.3% of the population (3,219 of 249,796 participants). An important limitation in this analysis is our sample size. Although we had a group of thousands of persons, it is a relatively small sample size for testing associations with candidate genes. It is possible that we had inadequate power to detect associations given the small effect size of some of the SNPs. Our analysis may have benefited from additional participants; however, cohorts of older adults with genetic data and detailed adipose measures are limited, and previous studies with smaller sample sizes than ours have identified significant associations between loci and BMI (25,26). Strengths of this study include the use of two large well-characterized population cohorts with data in midlife and old age and the use of adiposity measures from computed tomography imaging, dual-energy x-ray absorptiometry, and bioelectric impedance.
In conclusion, SNPs previously associated with BMI in middle-aged populations were not associated with weight or adiposity in older EAs and AAs. This suggests that genetic factors related to obesity may vary with age.
Funding
This work was supported by National Institute on Aging (NIA) Contracts N01-AG-6-2101; N01-AG-6-2103, N01-AG-6-2106, NIH contract N01-AG-1-2100, by the Intramural Research Program of the NIA, Hjartavernd (the Icelandic Heart Association), and by the Althingi (Icelandic Parliament). The genome-wide association study was funded by NIA grant R01AG032098-01A1 to Wake Forest University Health Sciences, and genotyping services were provided by the Center for Inherited Disease Research (CIDR). CIDR is fully funded through a federal contract from the National Institutes of Health to The Johns Hopkins University, contract number HHSN268200782096C. This research was supported in part by the Intramural Research Program of the NIH, National Institute on Aging. The participation of MAN and MFK was supported in part by the Intramural Research Programs of the National Institute on Aging: project number Z01 AG000932-04 (2011). Portions of the work of the IPDGC utilized the high-performance computational capabilities of the Biowulf Linux cluster at the National Institutes of Health, Bethesda, MD. (http://biowulf.nih.gov). RAM is supported by a Banting Postdoctoral Fellowship.
References
- 1. Ogden CL, Carrol MD, Kit BK, Flegal KM. Prevalence of Obesity in the United States, 2009–2010. NCHS data brief, no 82. Hyattsville, MD: National Center for Health Statistics; 2012. [PubMed] [Google Scholar]
- 2. OECD (2011). Health at a Glance 2011: OECD Indicators Paris, France: OECD Publishing; 2011. http://dx.doi.org/10.1787/health_ glance-2011-en [Google Scholar]
- 3. Maes HH, Neale MC, Eaves LJ. Genetic and environmental factors in relative body weight and human adiposity. Behav Genet. 1997; 27: 325–351 [DOI] [PubMed] [Google Scholar]
- 4. Speliotes EK, Willer CJ, Berndt SI, et al. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet. 2010; 42: 937–948 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Heid IM, Jackson AU, Randall JC, et al. Meta-analysis identifies 13 new loci associated with waist-hip ratio and reveals sexual dimorphism in the genetic basis of fat distribution. Nat Genet. 2010; 42: 949–960 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Thorleifsson G, Walters GB, Gudbjartsson DF, et al. Genome-wide association yields new sequence variants at seven loci that associate with measures of obesity. Nat Genet. 2009; 41: 18–24 [DOI] [PubMed] [Google Scholar]
- 7. Visser M, Goodpaster BH, Kritchevsky SB, et al. Muscle mass, muscle strength, and muscle fat infiltration as predictors of incident mobility limitations in well-functioning older persons. J Gerontol A Biol Sci Med Sci. 2005; 60: 324–333 [DOI] [PubMed] [Google Scholar]
- 8. Harris TB, Launer LJ, Eiriksdottir G, et al. Age, Gene/Environment Susceptibility-Reykjavik Study: multidisciplinary applied phenomics. Am J Epidemiol. 2007; 165: 1076–1087 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Lettre G, Palmer CD, Young T, et al. Genome-wide association study of coronary heart disease and its risk factors in 8,090 African Americans: the NHLBI CARe Project. PLoS Genet. 2011; 7: e1001300 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Fox CS, Liu Y, White CC, et al . Genome-wide association for abdominal subcutaneous and visceral adipose reveals a novel locus for visceral fat in women. PLoS Genet. 2012; 8: e1002695 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Johnson AD, Kavousi M, Smith AV, et al. Genome-wide association meta-analysis for total serum bilirubin levels. Hum Mol Genet. 2009; 18: 2700–2710 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Psaty BM, O’Donnell CJ, Gudnason V, et al. Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium: Design of prospective meta-analyses of genome-wide association studies from 5 cohorts. Circ Cardiovasc Genet. 2009; 2: 73–80 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hindorff L, MacArthur J. (European Bioinformatics Institute), Wise A, et al. A catalog of published genome-wide association studies www.genome.gov/gwastudies Accessed March 2012
- 14. Ripatti S, Tikkanen E, Orho-Melander M, et al. A multilocus genetic risk score for coronary heart disease: case-control and prospective cohort analyses. Lancet. 2010; 376: 1393–1400 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006; 38: 904–909 [DOI] [PubMed] [Google Scholar]
- 16. Qayyum R, Snively BM, Ziv E, et al. A meta-analysis and genome-wide association study of platelet count and mean platelet volume in african americans. PLoS Genet. 2012; 8: e1002491 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. David SP, Hamidovic A, Chen GK, et al. Genome-wide meta-analyses of smoking behaviors in African Americans. Transl Psychiatry. 2012; 2: e119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Scuteri A, Sanna S, Chen WM, et al. Genome-wide association scan shows genetic variants in the FTO gene are associated with obesity-related traits. PLoS Genet. 2007; 3: e115 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Grant SF, Li M, Bradfield JP, et al. Association analysis of the FTO gene with obesity in children of Caucasian and African ancestry reveals a common tagging SNP. PLoS ONE. 2008; 3: e1746 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Qi L, Kang K, Zhang C, et al. Fat mass-and obesity-associated (FTO) gene variant is associated with obesity: longitudinal analyses in two cohort studies and functional test. Diabetes. 2008; 57: 3145–3151 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Kilpeläinen TO, Bingham SA, Khaw KT, Wareham NJ, Loos RJ. Association of variants in the PCSK1 gene with obesity in the EPIC-Norfolk study. Hum Mol Genet. 2009; 18: 3496–3501 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Zamboni M, Zoico E, Scartezzini T, et al. Body composition changes in stable-weight elderly subjects: the effect of sex. Aging Clin Exp Res. 2003; 15: 321–327 [DOI] [PubMed] [Google Scholar]
- 23. Beaufrère B, Morio B. Fat and protein redistribution with aging: metabolic considerations. Eur J Clin Nutr. 2000; 54(suppl 3):S48–S53 [DOI] [PubMed] [Google Scholar]
- 24. Zamboni M, Armellini F, Harris T, et al. Effects of age on body fat distribution and cardiovascular risk factors in women. Am J Clin Nutr. 1997; 66: 111–115 [DOI] [PubMed] [Google Scholar]
- 25. Hunt SC, Hasstedt SJ, Xin Y, et al. Polymorphisms in the NPY2R gene show significant associations with BMI that are additive to FTO, MC4R, and NPFFR2 gene effects. Obesity (Silver Spring). 2011; 19: 2241–2247 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Shiri-Sverdlov R, Custers A, van Vliet-Ostaptchouk JV, et al. Identification of TUB as a novel candidate gene influencing body weight in humans. Diabetes. 2006; 55: 385–389 [DOI] [PubMed] [Google Scholar]