ABSTRACT
Objective
BMI is widely used to define obesity and may not capture heterogeneity in body fat distribution. Genetic studies of obesity have largely focused on BMI but not visceral fat and fat‐free mass, which is particularly limited in East Asian populations. This study aimed to explore the genetic associations among body composition traits.
Methods
We conducted genome‐wide association studies (GWAS) of fat composition traits, including visceral fat, fat mass, and fat‐free mass, in 14,121 adults from a population‐based biobank. Bioelectrical impedance analysis was used for standardized body composition. GWAS were performed, followed by multinomial logistic regressions for associations across phenotype strata, adjusting for BMI and confounders.
Results
GWAS identified multiple SNPs significantly associated with visceral fat, fat mass, and fat‐free mass. A variant within the TTN‐rs794727539 was inversely associated with visceral fat and positively associated with fat‐free mass, independent of BMI. Additional loci showed trait‐specific associations, while many fat‐mass‐associated variants were attenuated after BMI adjustment, highlighting different genetic regulation of fat distribution.
Conclusions
We found that visceral fat, fat mass, and fat‐free mass were governed by partially distinct genetic mechanisms not fully captured by BMI. These results underscore the importance of incorporating fat composition phenotypes into genetic studies in East Asian populations.
Keywords: fat composition, FTO gene, genome‐wide association study (GWAS), SUSD4 gene, TTN gene
1. Introduction
Obesity is a major global public health challenge and a leading risk factor for cardiometabolic disease, disability, and premature mortality [1, 2, 3]. Although body mass index (BMI) remains the most widely used metric to define obesity, it does not adequately capture individual variation in body fat distribution or lean mass composition, both of which substantially influence metabolic and cardiovascular risk [4, 5, 6]. Individuals with similar BMI values may exhibit markedly different proportions of visceral adipose tissue, subcutaneous fat, and fat‐free mass, resulting in heterogeneous health outcomes.
Visceral adiposity has been shown to confer greater cardiometabolic risk than overall fat mass, whereas higher fat‐free mass, largely reflecting skeletal muscle, has been associated with improved metabolic health and reduced mortality [6, 7, 8]. These observations have prompted growing interest in body composition‐based phenotypes as more informative markers of obesity‐related risk than BMI alone [9, 10].
Genetic factors play a substantial role in obesity susceptibility, with genome‐wide association studies (GWAS) identifying more than 1000 loci associated with BMI and related traits [11]. However, many of these studies have focused on BMI, waist circumference, or waist to hip ratio as proxy measures of adiposity [11, 12]. Genetic investigations directly examining visceral fat, fat mass, and fat‐free mass remain limited, particularly in non‐European populations.
This gap is especially relevant for East Asian populations, who tend to exhibit higher body fat percentages and greater visceral adiposity at lower BMI levels compared with European populations [13, 14]. Although several GWAS conducted in East Asian cohorts have identified both shared and population‐specific obesity‐associated loci [15, 16], few studies have systematically examined the genetic determinants of body composition phenotypes measured using standardized instruments.
Using data from a large population‐based biobank, this study aims to investigate the genetic architecture underlying visceral fat, fat mass, and fat‐free mass using GWAS. By integrating detailed body composition phenotyping with comprehensive genotyping and adjusting for BMI and lifestyle factors, we seek to clarify the extent to which genetic determinants of obesity operate through fat distribution and lean mass rather than overall body size.
2. Methods
2.1. Study Design and Population
This study used data from the Taiwan Biobank (TWB) [17], a community‐based program predominantly comprising individuals of Han Chinese descent. After informed consent, participants underwent comprehensive physical examinations, provided urine and blood samples, and shared lifestyle information through face‐to‐face interviews with trained healthcare professionals. The analyses were based on genotyping data derived from peripheral blood samples and phenotypic measurements of body composition. Obesity‐related measurements were performed by professional nurses or certified laboratory scientists. Genotyping was performed using genomic DNA extracted from peripheral whole blood samples, following standardized TWB protocols. Whole blood was used as the primary source of DNA to ensure high‐quality genotyping and consistency across participants. The TWB utilizes a specialized chip (TWBv2.0) containing approximately 600,000–700,000 SNPs, facilitating GWAS to identify genetic markers associated with disease traits. The study received approval from the Institutional Review Board (IRB) of the TWB and Kaohsiung Medical University Hospital (IRB Number: KMUHIRB‐G(II)‐2,020,007). Because we used deidentified data, the consent form was waived by the IRB.
For this study, data were initially retrieved from 45,541 entries in the TWB database during a follow‐up period from January 2021 to June 2023, which included individuals with complete anthropometric measurements. After excluding 31,420 individuals who lacked TWBv2.0 chip data, the final study group comprised 14,121 participants with both comprehensive anthropometric and SNP data. The baseline characteristics in Table S1 showed the total TWB population with complete anthropometric measurements (n = 45,541) and the genetic testing subset (n = 14,121) with available TWBv2.0 chip data. The baseline characteristics were similar.
2.2. Anthropometric Measurements
Body composition parameters, including BMI, fat mass, fat‐free mass, and visceral fat level, were measured using TANITA bioelectrical impedance analysis (BIA) devices (DC‐430MA or BC‐420MA, Tanita Corporation, Tokyo, Japan). These validated tools are widely used in clinical and research settings [18, 19, 20, 21]. BIA technology estimates body composition by passing a low‐level electrical current through the body and analyzing the conductive and resistive properties of tissues.
BMI was calculated as weight in kilograms divided by height in meters squared. The Taiwan Ministry of Health and Welfare defined BMI of 24–26.9 kg/m2 for overweight and BMI ≥ 27 kg/m2 for obesity. These cutoffs are widely recognized as the most clinically appropriate thresholds for the Taiwanese population [22]. Unlike the standards established for Caucasian populations, these lower BMI thresholds reflect the higher prevalence of comorbidities and increased fat mass observed in Asians at comparable BMI levels [23, 24].
Fat mass, defined as the total fat content in the body—including both subcutaneous and visceral adipose tissue—was estimated using the BIA device's proprietary algorithms and expressed in kilograms, with measurements recorded to the nearest 0.1 kg. While fat‐free mass, representing the combined weight of all non‐fat components in the body—including skeletal muscle, bone, and total body water—was calculated using the analyzer's impedance‐based estimation models, with values also recorded to the nearest 0.1 kg. These estimation models have been validated against dual‐energy X‐ray absorptiometry (DXA), which serves as the reference method for body composition analysis [25, 26].
Visceral fat level was quantified as an index using the BIA device's algorithm, which integrates impedance measurements specifically from the abdominal region. This index, reported as a unitless scale ranging from 1 to 59, reflects the relative volume of visceral adipose tissue, with higher values indicating greater accumulation. The algorithm has been cross‐validated with imaging modalities, including magnetic resonance imaging (MRI), to ensure accuracy [27, 28, 29].
2.3. Genotyping and Quality Control
Genotyping was conducted using the Affymetrix TWB 2.0 SNP chip (Thermo Fisher Scientific, Santa Clara, CA, USA), a custom‐designed array optimized for Han Chinese genetic variation. The TWB 2.0 array contains approximately 690,000 markers aligned to the GRCh38 reference genome, including coding sequence variants, protein‐altering variants, and GWAS markers. The array was specifically enriched for rare and functional alleles, leveraging whole‐genome sequencing (WGS) data from 946 TWB participants [30]. This design aimed to enhance the detection of population‐specific risk alleles relevant to Han Chinese individuals.
Quality control (QC) on genotyping data was performed in PLINK 1.9 (https://www.cog‐genomics.org/plink/), filtering samples and SNPs based on sex ambiguity, call rates below 95%, Hardy–Weinberg equilibrium deviation (p < 1 × 10−5), and minor allele frequencies (MAF) < 0.01 [31]. Associations between genetic variants and body composition traits were analyzed.
2.4. Statistical Analysis
GWAS were conducted using PLINK (version 1.9) to evaluate the associations between SNPs and body composition traits. A linear regression model was employed, treating body composition traits (e.g., fat mass, fat‐free mass, visceral fat level) as continuous variables, allowing for the full variability of the data to be utilized in identifying potentially significant SNPs [32]. Manhattan plots were generated in R (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria).
Post‐GWAS analyses, focusing on significant SNPs and their associations within stratified body composition categories, were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). To evaluate the associations between SNPs and body composition traits, multinomial logistic regression models were applied. Each body composition trait, including visceral fat level, fat‐free mass, and fat mass, was stratified into three categories (low, intermediate, and high) based on percentile thresholds. For visceral fat level and fat mass, the low category was defined as below the 50th percentile, the intermediate category as the 50th–75th percentiles, and the high category as above the 75th percentile. In contrast, for fat‐free mass, the high category was defined as above the 50th percentile, the intermediate category as the 25th–50th percentiles, and the low category as below the 25th percentile. Participants in the low category for visceral fat level and fat mass were considered relatively healthy, characterized by lower visceral fat and fat mass levels. Conversely, for fat‐free mass, participants in the high category were regarded as relatively healthy, as this group reflects higher levels of muscle and lean tissue. Descriptive statistics across these three categories were performed using one‐way ANOVA for continuous variables and the chi‐square test for categorical variables.
SNPs were analyzed under the additive genetic inheritance model. While dominant and recessive models were also explored, only the additive model yielded statistically significant results. Therefore, the findings are presented using the additive model to best reflect the observed associations. Adjusted odds ratios (AORs) with 95% confidence intervals (CIs) were estimated to assess the associations for intermediate versus low and high versus low groups. The models included covariates such as age, sex, educational level, marital status, smoking habits, alcohol consumption, betel‐nut use, physical activity, and BMI.
All statistical tests were two‐sided, with p < 0.05 considered statistically significant unless otherwise specified.
3. Results
GWAS identified several SNPs linked to visceral fat level, fat‐free mass, and fat mass, as illustrated by Manhattan plots (Figure 1). In these plots, each SNP is represented by its chromosomal position (x‐axis) and the negative logarithm of its p value (y‐axis), with genome‐wide significance defined as p < 5 × 10−8 (marked by a horizontal line) and suggestive associations for p values between 5 × 10−8 and 1 × 10−5. Figure 1A shows 10 SNPs significantly associated with visceral fat, with rs794727539 on chromosome 2 exhibiting the strongest signal. Similarly, Figure 1B highlights 8 SNPs associated with fat‐free mass, including rs794727539, suggesting a shared genetic role in regulating both traits. In contrast, Figure 1C reveals 66 SNPs linked to fat mass, with a notable cluster on chromosome 16 indicating a potential genetic hot spot for adiposity traits.
FIGURE 1.
Manhattan plots for visceral fat, level, fat‐free mass, and fat mass. (A) This Manhattan plot depicts the GWAS results for visceral fat level. Among the significant findings, SNP rs794727539, located on chromosome 2, demonstrates a strong association with visceral fat level. Other significant SNPs are scattered across different chromosomes. (B) This Manhattan plot illustrates the GWAS results for fat‐free mass. Significant SNPs are distributed across multiple chromosomes, with rs794727539 on chromosome 2 showing associations with both fat‐free mass and visceral fat level, indicating a shared genetic influence. (C) This Manhattan plot displays the GWAS results for fat mass. Unlike the patterns observed for visceral fat level and fat‐free mass, a distinct cluster of SNPs on chromosome 16 exhibits genome‐wide significance. This suggests that chromosome 16 may serve as a critical genetic hot spot for fat mass regulation. However, many associations were attenuated after adjusting for BMI, emphasizing the complexity of fat mass regulation. For details, refer to Tables S2–S4. [Color figure can be viewed at wileyonlinelibrary.com]
A summary of the genotype distributions of SNPs associated with visceral fat level, fat‐free mass, and fat mass, respectively, of the study population is shown in the Supporting Information tables. In brief, across all traits, the genotype frequencies demonstrated consistency with Hardy–Weinberg equilibrium, validating the genetic quality control process. The detailed data are shown in Tables S2–S4.
In Table 1, participants were categorized into low (< 50th percentile, n = 6841), intermediate (50th–75th, n = 3670), and high (> 75th percentile, n = 3610) groups based on visceral fat level, which increased significantly from 5.0 (SD = 1.7) to 14.8 (SD = 2.2) (p < 0.001). The high visceral fat group was older (55.1 vs. 43.1 years, p < 0.001), had a higher proportion of males and married individuals, and exhibited increased rates of smoking (4.0% to 17.1%), alcohol consumption (3.4%–15.3%), and betel‐nut use (0.6%–14.7%), along with a higher BMI, 21.6 (SD = 2.2) to 27.4 (SD = 3.9) kg/m2, p < 0.001.
TABLE 1.
Characteristics of participants by visceral fat level categories.
| Total participants (n = 14,121) | Low (n = 6841) | Intermediate (n = 3670) | High (n = 3610) | p |
|---|---|---|---|---|
| Visceral fat level | 5.0 (1.7) | 9.2 (1.1) | 14.8 (2.2) | < 0.001 |
| Age, years | 43.1 (13.1) | 52.2 (13.3) | 55.1 (12.3) | < 0.001 |
| Sex (females), n (%) | 6342 (92.7%) | 2463 (67.1%) | 203 (5.6%) | < 0.001 |
| Educational level, n (%) | < 0.001 | |||
| Elementary school or below | 152 (2.2%) | 207 (5.6%) | 99 (2.7%) | |
| Junior high school | 321 (4.7%) | 241 (6.6%) | 179 (5.0%) | |
| Senior high school | 1676 (24.5%) | 989 (27.0%) | 892 (24.7%) | |
| College or above | 4692 (68.6%) | 2233 (60.8%) | 2440 (67.6%) | |
| Marital status, n (%) | < 0.001 | |||
| Single/divorced/widowed | 2638 (38.6%) | 1268 (34.6%) | 751 (20.8%) | |
| Married | 4203 (61.4%) | 2402 (65.4%) | 2859 (79.2%) | |
| Smoking, n (%) | < 0.001 | |||
| No smoking | 5935 (86.8%) | 3101 (84.5%) | 2728 (75.6%) | |
| Passive smoking | 631 (9.2%) | 311 (8.5%) | 266 (7.4%) | |
| Current smoking | 275 (4.0%) | 258 (7.0%) | 616 (17.1%) | |
| Alcohol, n (%) | 233 (3.4%) | 219 (6.0%) | 551 (15.3%) | < 0.001 |
| Betel‐nut, n (%) | 44 (0.6%) | 103 (2.8%) | 529 (14.7%) | < 0.001 |
| Physical activity, n (%) | 2462 (36.0%) | 1415 (38.6%) | 1580 (43.8%) | < 0.001 |
| BMI, kg/m2 | 21.6 (2.2) | 26.0 (3.1) | 27.4 (3.9) | < 0.001 |
Note: Values are given as mean (SD) and number (%).
In Table 2, participants were divided into low (< 25th percentile, n = 3448), intermediate (25th–50th, n = 3596), and high (> 50th percentile, n = 7077) groups based on fat‐free mass, which increased significantly from 35.5 (SD = 1.8) kg to 53.1 (SD = 7.8) kg (p < 0.001). Older individuals were more common in the low fat‐free mass group (53.9 vs. 49.5 years, p < 0.001), and nearly all participants in this group were female (99.8%), compared to only 28.2% in the high group. Higher educational attainment, as well as a greater prevalence of smoking and alcohol consumption, was also observed in the high fat‐free mass group.
TABLE 2.
Characteristics of participants by fat‐free mass categories.
| Total participants (n = 14,121) | Low (n = 3448) | Intermediate (n = 3596) | High (n = 7077) | p |
|---|---|---|---|---|
| Fat‐free mass, kg | 35.5 (1.8) | 39.8 (1.1) | 53.1 (7.8) | < 0.001 |
| Age, years | 53.9 (13.3) | 51.0 (12.4) | 49.5 (13.4) | < 0.001 |
| Sex (females), n (%) | 3442 (99.8%) | 3569 (99.3%) | 1997 (28.2%) | < 0.001 |
| Educational level, n (%) | < 0.001 | |||
| Elementary school or below | 199 (5.8%) | 139 (3.9%) | 120 (1.7%) | |
| Junior high school | 280 (8.1%) | 195 (5.4%) | 266 (3.8%) | |
| Senior high school | 1035 (30.0%) | 992 (27.6%) | 1530 (21.6%) | |
| College or above | 1934 (56.1%) | 2270 (63.1%) | 5161 (72.9%) | |
| Marital status, n (%) | < 0.001 | |||
| Single/divorced/widowed | 1237 (35.9%) | 1259 (35.0%) | 2161 (30.5%) | |
| Married | 2211 (64.1%) | 2337 (65.0%) | 4916 (69.5%) | |
| Smoking, n (%) | < 0.001 | |||
| No smoking | 3066 (88.9%) | 3186 (88.6%) | 5510 (77.9%) | |
| Passive smoking | 291 (8.5%) | 301 (8.4%) | 618 (8.7%) | |
| Current smoking | 91 (2.6%) | 109 (3.0%) | 949 (13.4%) | |
| Alcohol, n (%) | 91 (2.6%) | 108 (3.0%) | 804 (11.4%) | < 0.001 |
| Betel‐nut, n (%) | 6 (0.2%) | 8 (0.2%) | 662 (9.4%) | < 0.001 |
| Physical activity, n (%) | 1454 (42.2%) | 1314 (36.5%) | 2689 (38.0%) | < 0.001 |
| BMI, kg/m2 | 21.5 (2.7) | 23.6 (3.2) | 25.9 (4.0) | < 0.001 |
Note: Values are given as mean (SD) and number (%).
In Table 3, participants were classified into low (< 50th percentile, n = 6981), intermediate (50th–75th, n = 3561), and high (> 75th percentile, n = 3579) groups by fat mass, which increased from 13.5 (SD = 2.9) kg in the low group to 28.4 (SD = 6.7) kg in the high group (p < 0.001), while BMI rose from 21.6 (SD = 2.2) kg/m2 to 28.9 (SD = 3.5) kg/m2 (p < 0.001). Although age varied only slightly (50.8–52.2 years), the high fat mass group had a higher proportion of females (73.1% vs. 57.7%) and lower levels of physical activity (42.5% vs. 30.3%, p < 0.001), with minimal differences in smoking and alcohol consumption.
TABLE 3.
Characteristics of participants by fat mass categories.
| Total participants (n = 14,121) | Low (n = 6981) | Intermediate (n = 3561) | High (n = 3579) | p |
|---|---|---|---|---|
| Fat mass, kg | 13.5 (2.9) | 19.7 (1.3) | 28.4 (6.7) | < 0.001 |
| Age, years | 50.8 (13.8) | 52.2 (12.6) | 50.1 (12.7) | < 0.001 |
| Sex (females), n (%) | 4028 (57.7%) | 2364 (66.4%) | 2616 (73.1%) | < 0.001 |
| Educational level, n (%) | < 0.001 | |||
| Elementary school or below | 159 (2.3%) | 146 (4.1%) | 153 (4.3%) | |
| Junior high school | 319 (4.5%) | 209 (5.9%) | 213 (6.0%) | |
| Senior high school | 1597 (22.9%) | 978 (27.4%) | 982 (27.4%) | |
| College or above | 4906 (70.3%) | 2228 (62.6%) | 2231 (62.3%) | |
| Marital status, n (%) | < 0.001 | |||
| Single/divorced/widowed | 2359 (33.8%) | 1013 (28.4%) | 1285 (35.9%) | |
| Married | 4622 (66.2%) | 2548 (71.6%) | 2294 (64.1%) | |
| Smoking, n (%) | 0.188 | |||
| No smoking | 5863 (84.0%) | 2957 (83.0%) | 2944 (82.3%) | |
| Passive smoking | 565 (8.1%) | 319 (9.0%) | 324 (9.0%) | |
| Current smoking | 553 (7.9%) | 285 (8.0%) | 311 (8.7%) | |
| Alcohol, n (%) | 500 (7.2%) | 281 (7.9%) | 222 (6.2%) | 0.02 |
| Betel‐nut, n (%) | 303 (4.3%) | 194 (5.4%) | 179 (5.0%) | 0.033 |
| Physical activity, n (%) | 2964 (42.5%) | 1409 (39.6%) | 1084 (30.3%) | < 0.001 |
| BMI, kg/m2 | 21.6 (2.2) | 24.7 (1.9) | 28.9 (3.5) | < 0.001 |
Note: Values are given as mean (SD) and number (%).
Multinomial logistic regression analyses (presented in Tables 4, 5, 6) adjusted for confounders including age, sex, education, marital status, smoking, alcohol consumption, betel‐nut use, physical activity, and BMI. The crude odds ratio (OR) without adjustment of these confounders was shown in Tables S5–S7. For visceral fat (Table 4), rs794727539 on chromosome 2 showed a dose‐dependent protective effect: each additional minor allele (C) was associated with a reduced likelihood of high visceral fat (AOR = 0.32, 95% CI: 0.11–0.93, p = 0.036), while no significant association was observed in the intermediate group. In the case of fat‐free mass (Table 5), rs794727539 also demonstrated a positive association (AOR = 1.83, 95% CI: 1.10–3.04, p = 0.020), suggesting a shared genetic influence on visceral fat and lean tissue. Additionally, rs1418150 (AOR = 1.20, 95% CI: 1.02–1.41, p = 0.024) was significantly linked to increased fat‐free mass, while rs7145690 (AOR = 0.63, 95% CI: 0.42–0.96, p = 0.030) and rs11621139 (AOR = 0.91, 95% CI: 0.82–1.00, p = 0.043) showed negative associations, all significant only in the high versus low comparison, highlighting a dose‐dependent genetic effect. For fat mass (Table 6), rs11942302 was significantly associated with the intermediate group (AOR = 1.13, 95% CI: 1.01–1.26, p = 0.034), but not with the high group, suggesting that its effect may be modulated by additional factors. Moreover, a cluster of SNPs on chromosome 16, evident in the Manhattan plot (Figure 1C), indicates a potential genetic hot spot for fat mass regulation. However, many associations were attenuated after adjusting for BMI, emphasizing the complexity of fat mass regulation. To further characterize the genomic context and cross‐trait relationships of the identified variants, we summarized all significant SNPs with their genomic annotations in Table S8. Notably, only one locus, rs794727539 located in the TTN gene, was shared between visceral fat and fat‐free mass, suggesting a potential pleiotropic effect. In contrast, fat mass‐associated SNPs were predominantly clustered within the FTO region on chromosome 16, where 66 variants were located within a single linkage disequilibrium (LD) block, indicating a BMI‐driven genetic architecture. These findings highlight the distinct genetic architectures underlying different body composition traits.
TABLE 4.
Additive genetic model for SNP associations with visceral fat level.
| SNP | Intermediate versus Low** | High versus Low** | ||
|---|---|---|---|---|
| AOR* (95% CI) | p | AOR* (95% CI) | p | |
| rs794727539 | 0.46 (0.21, 1.02) | 0.056 | 0.32 (0.11, 0.93) | 0.036 |
| rs28551468 | 1.06 (0.90, 1.25) | 0.470 | 1.21 (0.95, 1.54) | 0.132 |
| rs1705492 | 1.05 (0.87, 1.27) | 0.624 | 0.99 (0.75, 1.32) | 0.957 |
| rs78367007 | 1.03 (0.74, 1.44) | 0.861 | 1.00 (0.60, 1.67) | 0.991 |
| rs145950663 | 1.80 (0.95, 3.41) | 0.069 | 1.32 (0.55, 3.21) | 0.536 |
| rs144492569 | 1.29 (0.74, 2.25) | 0.373 | 1.12 (0.49, 2.56) | 0.791 |
| rs1035209 | 0.98 (0.79, 1.20) | 0.823 | 0.89 (0.65, 1.22) | 0.478 |
| rs12002598 | 0.97 (0.81, 1.17) | 0.771 | 0.97 (0.74, 1.28) | 0.836 |
| rs151114847 | 1.08 (0.61, 1.90) | 0.799 | 1.32 (0.54, 3.21) | 0.541 |
| rs9348935 | 1.20 (0.84, 1.71) | 0.320 | 0.97 (0.58, 1.62) | 0.901 |
Adjusted odds ratios (AORs) with 95% confidence intervals (CIs) and p values were estimated using a multinomial logistic regression model based on single‐nucleotide polymorphisms (SNPs) under additive inheritance models, adjusting for age, sex, educational level, marital status, smoking habits, alcohol consumption, betel‐nut use, physical activity, and BMI.
The “Low” group served as the reference category for all comparisons.
TABLE 5.
Additive genetic model for SNP associations with fat‐free mass.
| SNP | Intermediate versus Low** | High versus Low** | ||
|---|---|---|---|---|
| AOR* (95% CI) | p | AOR* (95% CI) | p | |
| rs794727539 | 1.12 (0.75, 1.68) | 0.585 | 1.83 (1.10, 3.04) | 0.020 |
| rs112721073 | 0.93 (0.73, 1.19) | 0.570 | 0.73 (0.50, 1.04) | 0.083 |
| rs3865563 | 1.01 (0.91, 1.12) | 0.875 | 0.96 (0.83, 1.10) | 0.515 |
| rs6054036 | 1.01 (0.91, 1.12) | 0.826 | 0.96 (0.84, 1.10) | 0.575 |
| rs7145690 | 0.85 (0.64, 1.13) | 0.269 | 0.63 (0.42, 0.96) | 0.030 |
| rs11621139 | 0.94 (0.88, 1.01) | 0.116 | 0.91 (0.82, 1.00) | 0.043 |
| rs12666524 | 1.03 (0.96, 1.11) | 0.414 | 1.07 (0.97, 1.18) | 0.193 |
| rs1418150 | 1.02 (0.91, 1.15) | 0.750 | 1.20 (1.02, 1.41) | 0.024 |
Adjusted odds ratios (AORs) with 95% confidence intervals (CIs) and p values were estimated using a multinomial logistic regression model based on single‐nucleotide polymorphisms (SNPs) under additive inheritance models, adjusting for age, sex, educational level, marital status, smoking habits, alcohol consumption, betel‐nut use, physical activity, and BMI.
The “Low” group served as the reference category for all comparisons.
TABLE 6.
Additive genetic model for SNP associations with fat mass.
| SNP | Intermediate versus Low** | High versus Low** | ||
|---|---|---|---|---|
| AOR* (95% CI) | p | AOR* (95% CI) | p | |
| rs10468280 | 0.98 (0.85, 1.14) | 0.832 | 1.13 (0.90, 1.41) | 0.292 |
| rs11075989 | 0.98 (0.85, 1.14) | 0.816 | 1.13 (0.90, 1.41) | 0.296 |
| rs11075990 | 0.97 (0.84, 1.13) | 0.721 | 1.12 (0.89, 1.40) | 0.328 |
| rs11075991 | 0.98 (0.84, 1.14) | 0.766 | 1.12 (0.89, 1.40) | 0.329 |
| rs11075992 | 0.98 (0.84, 1.13) | 0.746 | 1.12 (0.90, 1.40) | 0.321 |
| rs11075993 | 0.93 (0.81, 1.08) | 0.339 | 1.03 (0.83, 1.28) | 0.795 |
| rs11642015 | 0.95 (0.82, 1.10) | 0.461 | 1.08 (0.87, 1.35) | 0.504 |
| rs12149574 | 0.94 (0.81, 1.09) | 0.402 | 1.05 (0.85, 1.31) | 0.645 |
| rs12149832 | 0.95 (0.82, 1.10) | 0.513 | 1.04 (0.84, 1.30) | 0.700 |
| rs1421085 | 0.95 (0.82, 1.10) | 0.496 | 1.09 (0.87, 1.36) | 0.451 |
| rs17817449 | 0.97 (0.84, 1.13) | 0.722 | 1.12 (0.90, 1.40) | 0.319 |
| rs17817497 | 0.97 (0.83, 1.12) | 0.668 | 1.13 (0.90, 1.41) | 0.301 |
| rs17817712 | 0.98 (0.84, 1.13) | 0.758 | 1.12 (0.89, 1.40) | 0.328 |
| rs17817964 | 0.99 (0.86, 1.14) | 0.839 | 1.16 (0.94, 1.43) | 0.179 |
| rs3751812 | 0.98 (0.85, 1.14) | 0.801 | 1.12 (0.90, 1.40) | 0.311 |
| rs3751814 | 0.98 (0.84, 1.14) | 0.780 | 1.12 (0.90, 1.40) | 0.308 |
| rs55872725 | 0.97 (0.84, 1.13) | 0.683 | 1.11 (0.89, 1.39) | 0.370 |
| rs56094641 | 0.93 (0.80, 1.08) | 0.343 | 1.05 (0.84, 1.31) | 0.693 |
| rs56313538 | 0.98 (0.84, 1.13) | 0.737 | 1.12 (0.89, 1.40) | 0.344 |
| rs62033400 | 0.97 (0.84, 1.13) | 0.703 | 1.12 (0.89, 1.40) | 0.328 |
| rs62033403 | 0.97 (0.84, 1.13) | 0.721 | 1.12 (0.89, 1.40) | 0.332 |
| rs62048402 | 0.95 (0.82, 1.11) | 0.534 | 1.08 (0.87, 1.35) | 0.480 |
| rs66908032 | 0.96 (0.83, 1.11) | 0.588 | 1.11 (0.89, 1.39) | 0.367 |
| rs7185735 | 0.98 (0.84, 1.13) | 0.747 | 1.12 (0.90, 1.40) | 0.324 |
| rs7202116 | 0.98 (0.85, 1.14) | 0.831 | 1.13 (0.90, 1.41) | 0.292 |
| rs7202296 | 0.98 (0.84, 1.14) | 0.777 | 1.12 (0.90, 1.41) | 0.302 |
| rs72803697 | 0.98 (0.84, 1.13) | 0.756 | 1.12 (0.89, 1.40) | 0.327 |
| rs72805611 | 0.95 (0.82, 1.10) | 0.487 | 1.06 (0.85, 1.31) | 0.623 |
| rs72805613 | 0.94 (0.81, 1.09) | 0.408 | 1.04 (0.84, 1.29) | 0.716 |
| rs8043757 | 0.97 (0.84, 1.13) | 0.716 | 1.11 (0.89, 1.39) | 0.366 |
| rs8050136 | 0.97 (0.84, 1.13) | 0.690 | 1.11 (0.89, 1.39) | 0.341 |
| rs8051591 | 0.98 (0.84, 1.13) | 0.750 | 1.12 (0.90, 1.40) | 0.310 |
| rs8063057 | 0.97 (0.84, 1.13) | 0.731 | 1.13 (0.91, 1.42) | 0.270 |
| rs9923233 | 0.98 (0.84, 1.13) | 0.735 | 1.12 (0.90, 1.40) | 0.326 |
| rs9923312 | 0.97 (0.84, 1.13) | 0.734 | 1.12 (0.90, 1.40) | 0.320 |
| rs9926289 | 0.98 (0.84, 1.13) | 0.740 | 1.12 (0.89, 1.40) | 0.333 |
| rs9935401 | 0.97 (0.84, 1.13) | 0.732 | 1.11 (0.89, 1.38) | 0.374 |
| rs9936385 | 0.97 (0.84, 1.13) | 0.723 | 1.12 (0.89, 1.39) | 0.337 |
| rs9939609 | 0.98 (0.84, 1.13) | 0.748 | 1.12 (0.90, 1.40) | 0.320 |
| Affx‐12,870,033 | 1.00 (0.88, 1.14) | 0.971 | 1.10 (0.91, 1.34) | 0.331 |
| Affx‐23,390,696 | 1.01 (0.79, 1.30) | 0.929 | 1.03 (0.69, 1.52) | 0.904 |
| Affx‐26,070,446 | 0.97 (0.72, 1.30) | 0.827 | 1.22 (0.79, 1.88) | 0.376 |
| rs10494163 | 0.99 (0.89, 1.11) | 0.899 | 0.95 (0.80, 1.12) | 0.514 |
| rs10820391 | 0.98 (0.88, 1.09) | 0.725 | 0.98 (0.83, 1.15) | 0.759 |
| rs11075985 | 1.00 (0.88, 1.14) | 0.965 | 1.11 (0.91, 1.35) | 0.304 |
| rs1121980 | 1.01 (0.89, 1.16) | 0.852 | 1.11 (0.91, 1.35) | 0.300 |
| rs11934511 | 1.02 (0.79, 1.31) | 0.885 | 1.08 (0.73, 1.61) | 0.688 |
| rs11942302 | 1.13 (1.01, 1.26) | 0.034 | 1.15 (0.97, 1.37) | 0.105 |
| rs149688127 | 0.78 (0.49, 1.24) | 0.291 | 0.83 (0.40, 1.72) | 0.624 |
| rs1558901 | 1.01 (0.88, 1.15) | 0.928 | 1.12 (0.93, 1.37) | 0.238 |
| rs16952522 | 1.00 (0.84, 1.19) | 0.988 | 1.12 (0.87, 1.45) | 0.376 |
| rs266585 | 1.07 (0.85, 1.33) | 0.579 | 1.18 (0.84, 1.64) | 0.347 |
| rs28425208 | 1.11 (0.99, 1.25) | 0.080 | 1.14 (0.96, 1.37) | 0.136 |
| rs7201850 | 1.02 (0.89, 1.16) | 0.816 | 1.11 (0.91, 1.35) | 0.318 |
| rs7206629 | 1.01 (0.89, 1.16) | 0.846 | 1.11 (0.91, 1.35) | 0.319 |
| rs883354 | 1.04 (0.79, 1.37) | 0.767 | 0.84 (0.54, 1.29) | 0.417 |
| rs9923147 | 1.00 (0.88, 1.14) | 0.998 | 1.11 (0.91, 1.35) | 0.295 |
| rs9923544 | 1.00 (0.88, 1.14) | 0.986 | 1.10 (0.91, 1.34) | 0.327 |
| rs9928094 | 1.00 (0.88, 1.14) | 0.963 | 1.12 (0.92, 1.36) | 0.264 |
| rs9933509 | 1.01 (0.89, 1.15) | 0.855 | 1.11 (0.91, 1.35) | 0.304 |
| rs9937053 | 1.00 (0.88, 1.14) | 0.997 | 1.11 (0.91, 1.35) | 0.291 |
| rs9937354 | 1.00 (0.88, 1.14) | 0.985 | 1.11 (0.92, 1.35) | 0.280 |
| rs9939973 | 1.00 (0.88, 1.14) | 0.968 | 1.10 (0.91, 1.34) | 0.320 |
| rs9940128 | 1.00 (0.87, 1.13) | 0.938 | 1.11 (0.91, 1.34) | 0.312 |
| rs9940646 | 1.00 (0.88, 1.13) | 0.951 | 1.10 (0.91, 1.34) | 0.317 |
| rs999730 | 1.01 (0.91, 1.13) | 0.847 | 0.92 (0.77, 1.09) | 0.320 |
Adjusted odds ratios (AORs) with 95% confidence intervals (CIs) and p values were estimated using a multinomial logistic regression model based on single‐nucleotide polymorphisms (SNPs) under additive inheritance models, adjusting for age, sex, educational level, marital status, smoking habits, alcohol consumption, betel‐nut use, physical activity, and BMI.
The “Low” group served as the reference category for all comparisons.
In sex‐stratified analyses, rs794727539 was not significantly associated with visceral fat in either females (β = −0.017, 95% CI: −0.091 to 0.057, p = 0.652) or males (β = 0.052, 95% CI: −0.150 to 0.253, p = 0.615). For fat‐free mass, a borderline association was observed in females (β = −0.457, 95% CI: −0.913 to −0.001, p = 0.050), whereas no significant association was detected in males (β = 0.294, 95% CI: −0.387 to 0.975, p = 0.398) in Table S9. In the interaction model, no significant association between rs794727539 and visceral fat was observed in females (β = 0.107, p = 0.245), whereas a significant SNP‐by‐sex interaction was detected (β = −0.397, p = 0.002) in Table S10. The estimated effect in males, derived from the interaction term, was negative (β = −0.290). For fat‐free mass, rs794727539 showed a borderline association in females (β = −0.531, p = 0.070), and no significant SNP‐by‐sex interaction was observed (p = 0.140).
Overall, these findings underscore the intricate genetic architecture underlying adiposity traits and the necessity of considering multiple confounding factors to accurately assess genetic contributions to body composition.
4. Discussion
In this large population‐based study, we identified distinct and partially overlapping genetic determinants of visceral fat, fat mass, and fat‐free mass, extending current obesity genetics research beyond BMI‐centric paradigms. Our findings demonstrate that body composition traits reflect heterogeneous biological processes that are not fully captured by conventional anthropometric indices.
A key finding was the identification of a variant within the TTN locus that exhibited a dual association with reduced visceral fat accumulation and increased fat‐free mass, independent of BMI. While TTN is best known for its role in sarcomere structure and muscle integrity, emerging evidence from animal models suggests that TTN variants may influence the balance between lean tissue accretion and fat deposition [33, 34]. Although human data linking TTN to adiposity traits remain sparse, our results support a potential role for muscle‐related genetic pathways in regulating visceral fat accumulation, consistent with growing recognition of skeletal muscle as a central regulator of metabolic health.
This finding highlights a potential decoupling of adiposity and lean mass genetics. Among the identified loci, rs794727539 in the TTN gene emerged as a unique shared signal for both visceral fat and fat‐free mass, suggesting a potential pleiotropic effect. The identification of rs794727539 in the TTN gene as a shared locus for visceral fat and fat‐free mass provides novel insights into the genetic regulation of body composition. TTN has been extensively linked to cardiomyopathies and cardiovascular function, with truncating variants representing the most common genetic cause of dilated cardiomyopathy [35]. TTN encodes titin, a key structural protein in cardiac and skeletal muscle, and has been extensively linked to cardiomyopathies and cardiac function [36]. The observed association of this variant with reduced visceral fat and increased fat‐free mass suggests a potential biological pathway linking muscle integrity, adiposity distribution, and cardiometabolic health. Given that visceral adiposity is a well‐established risk factor for cardiovascular disease, this finding raises the possibility that rs794727539 may represent a shared genetic determinant influencing both body composition and cardiovascular risk. This observation is consistent with emerging evidence that genetic variants affecting muscle biology may also influence metabolic and cardiovascular traits through systemic pathways [37]. However, direct evaluation of cardiovascular outcomes was beyond the scope of the present study. Future investigations integrating genetic data with longitudinal cardiovascular phenotypes are warranted to further elucidate this relationship. This observation is consistent with emerging evidence that genetic variants affecting muscle biology may also influence metabolic and cardiovascular traits through systemic pathways.
To further explore whether the observed genetic effects differed by sex across body composition traits, a significant SNP‐by‐sex interaction was observed for rs794727539 in relation to visceral fat, indicating sex‐dependent genetic effects. While no significant association was detected in females, the effect estimate in males was negative and greater in magnitude. This pattern suggests that the influence of this variant on visceral fat may differ between sexes, potentially reflecting underlying biological differences in fat distribution and metabolic regulation. Sex differences in adipose tissue distribution and metabolism are well established, with males and females exhibiting distinct patterns of visceral fat accumulation and associated metabolic risk [38]. In contrast, although rs794727539 showed a suggestive negative association with fat‐free mass in females, no significant SNP‐by‐sex interaction was observed. This indicates that, unlike visceral fat, the genetic effect of this variant on fat‐free mass does not significantly differ between sexes. Together, these findings suggest that the genetic mechanisms underlying adiposity distribution may be more sensitive to sex‐specific factors, whereas the influence on lean mass appears to be more consistent across sexes. Skeletal muscle biology and metabolic regulation exhibit sex‐related differences; however, certain aspects of lean mass regulation appear to be relatively conserved across sexes [39]. Notably, the significant interaction observed despite nonsignificant stratified results suggests that effect heterogeneity may be better captured in the combined interaction model. The lack of significance in sex‐stratified analyses may be attributable to reduced statistical power after stratification, while the combined analysis provides more stable effect estimates.
In contrast, many loci associated with fat mass, particularly those clustered within the FTO region, showed attenuated associations after adjustment for BMI. This observation aligns with prior studies demonstrating that FTO primarily influences overall adiposity and energy intake rather than fat distribution per se [40, 41, 42]. These findings underscore the importance of distinguishing genetic effects on total body size from those on regional fat deposition when interpreting obesity GWAS results.
We observed limited overlap between loci associated with visceral fat and those associated with fat mass, supporting the hypothesis that central adiposity and overall adiposity are governed by partially distinct genetic mechanisms [14, 43]. This distinction may help explain why individuals with comparable BMI values can exhibit markedly different cardiometabolic risk profiles.
In our study, although the identified SNPs did not overlap with genes previously reported in GWAS focusing on fat distribution in Asian populations, we employed BIA measurements to quantify fat mass, visceral fat level, and even fat‐free mass. This approach, combined with adjustments for BMI and lifestyle factors, allowed us to refine the identification of SNPs more specifically associated with fat distribution. By incorporating these additional adjustments, our findings provide a more precise perspective on genetic contributors to fat distribution, potentially addressing gaps in previous research that lacked comparable adjustments.
Our study has several strengths, including a relatively large sample size, standardized body composition assessment, and comprehensive adjustment for lifestyle factors. However, limitations should be acknowledged. First, the cross‐sectional design precludes causal inference. Second, bioelectrical impedance analysis (BIA), while practical for large‐scale epidemiological studies, is less precise than imaging‐based methods such as MRI or CT for quantifying visceral fat. Third, all findings were derived from a single East Asian cohort (TWB), and external validation in an independent East Asian population was not performed. Therefore, the generalizability of our findings should be interpreted with caution. In addition, differences in body composition measurements and phenotype definitions across available biobanks currently limit direct replication. Future studies using independent East Asian cohorts with harmonized body composition phenotypes are warranted to validate these findings and confirm the robustness of the identified genetic associations. Finally, residual confounding by unmeasured factors such as diet or sleep patterns cannot be excluded.
In conclusion, this study using data from 14,121 participants with BIA phenotyping provides robust evidence that genetic regulation of obesity extends beyond BMI to include distinct pathways influencing visceral fat and lean mass. The findings enhance our understanding of genetic determinants for body composition traits, including visceral fat, fat‐free mass, and fat mass. We identified several SNPs linked to these traits. Notably, rs794727539 in the TTN gene was associated with reduced visceral fat and increased fat‐free mass, suggesting a role in balancing adipose and lean tissues. Additional SNPs in the SUSD4 gene and others showed varied effects, emphasizing the polygenic nature of these traits. Adjustments for BMI and lifestyle factors significantly influenced SNP associations, highlighting the complex interplay between overall adiposity, behavior, and specific fat distribution. Future research should focus on elucidating gene–environment interactions through functional studies and comprehensive genetic analyses to better inform precision medicine strategies aimed at reducing obesity‐related health risks, particularly in East Asian populations.
Author Contributions
Conceptualization, H.‐M.L., F.‐W.L., and H.‐Y.C.; methodology, K.‐H.L., S.‐J.C., F.‐W.L., and H.‐Y.C.; software, K.‐H.L. and H.‐M.L.; validation, S.‐J.C. and Y.‐S.C.; formal analysis, K.‐H.L. and H.‐M.L.; investigation, H.‐M.L., K.‐H.L., S.‐J.C., F.‐W.L., Y.‐S.C., and H.‐Y.C.; resources, H.‐Y.C. and F.‐W.L.; data curation, K.‐H.L., F.‐W.L., and H.‐Y.C.; writing – original draft preparation, H.‐M.L. and S.‐J.C.; writing – review and editing, H.‐Y.C. and F.‐W.L.; visualization, H.‐M.L. and Y.‐S.C.; supervision, F.‐W.L. and H.‐Y.C.; project administration, F.‐W.L. and H.‐Y.C.; funding acquisition, H.‐Y.C. All authors have read and agreed to the published version of the manuscript.
Funding
The study was supported by funds from the Taiwan National Science and Technology Council, NSTC 114–2314‐B‐037‐060, and Kaohsiung Medical University Hospital (KMUH114‐4R85). This work was also supported partially by the Research Center for Precision Environmental Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan, from the Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan and by a Kaohsiung Medical University Research Center Grant (KMU‐TC114A01). The funder did not influence the results/outcomes of the study despite author affiliations with the funder.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: Baseline characteristics of total and genetic testing group.
Table S2: Genotype distribution of SNPs related to visceral fat level.
Table S3: Genotype distribution of SNPs related to fat free mass.
Table S4: Genotype distribution of SNPs related to fat mass.
Table S5: Crude odds ratios for SNP associations with visceral fat level.
Table S6: Crude odds ratios for SNP associations with fat‐free mass.
Table S7: Crude odds ratios for SNP associations with fat mass.
Table S8: Cross‐trait and trait‐specific SNP annotation from GWAS of body composition traits.
Table S9: Sex‐stratified associations of rs794727539 with visceral fat and fat‐free mass.
Table S10: SNP‐by‐sex interaction effects of rs794727539 on visceral fat and fat‐free mass.
Acknowledgments
We would like to thank the National Core Facility for Biopharmaceuticals (NCFB) and the National Center for High‐Performance Computing (NCHC) of the National Institutes of Applied Research (NIAR) of Taiwan for providing computational and storage resources. This short text also acknowledges the Taiwan Biobank (TWB), specific colleagues, institutions, and agencies that aided the efforts of the authors.
Data Availability Statement
The data could be applied to use via an application proceeding at Taiwan Biobank (https://www.biobank.org.tw/english.php).
References
- 1. World Health Organization , “Obesity and Overweight,” updated December 8, (2025), https://www.who.int/news‐room/fact‐sheets/detail/obesity‐and‐overweight.
- 2. Masood B. and Moorthy M., “Causes of Obesity: A Review,” Clinical Medicine (Lond) 23 (2023): 284–291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. World Health Organization , “Obesity: Health Consequences of Being Overweight,” published March 1, (2024), https://www.who.int/news‐room/questions‐and‐answers/item/obesity‐health‐consequences‐of‐being‐overweight.
- 4. Shi T. H., Wang B., and Natarajan S., “The Influence of Metabolic Syndrome in Predicting Mortality Risk Among US Adults: Importance of Metabolic Syndrome Even in Adults With Normal Weight,” Preventing Chronic Disease 17 (2020): E36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Escobedo‐de la Peña J., Ramírez‐Hernández J. A., Fernández‐Ramos M. T., González‐Figueroa E., and Champagne B., “Body Fat Percentage Rather Than Body Mass Index Related to the High Occurrence of Type 2 Diabetes,” Archives of Medical Research 51 (2020): 564–571. [DOI] [PubMed] [Google Scholar]
- 6. Kang P. S. and Neeland I. J., “Body Fat Distribution, Diabetes Mellitus, and Cardiovascular Disease: An Update,” Current Cardiology Reports 25 (2023): 1555–1564. [DOI] [PubMed] [Google Scholar]
- 7. Antonio‐Villa N. E., Bello‐Chavolla O. Y., Vargas‐Vázquez A., et al., “Increased Visceral Fat Accumulation Modifies the Effect of Insulin Resistance on Arterial Stiffness and Hypertension Risk,” Nutrition, Metabolism, and Cardiovascular Diseases 31 (2021): 506–517. [DOI] [PubMed] [Google Scholar]
- 8. Bosy‐Westphal A., Braun W., Geisler C., Norman K., and Müller M. J., “Body Composition and Cardiometabolic Health: The Need for Novel Concepts,” European Journal of Clinical Nutrition 72 (2018): 638–644. [DOI] [PubMed] [Google Scholar]
- 9. Piche M. E., Poirier P., Lemieux I., and Despres J. P., “Overview of Epidemiology and Contribution of Obesity and Body Fat Distribution to Cardiovascular Disease: An Update,” Progress in Cardiovascular Diseases 61 (2018): 103–113. [DOI] [PubMed] [Google Scholar]
- 10. Jastreboff A. M., Kotz C. M., Kahan S., Kelly A. S., and Heymsfield S. B., “Obesity as a Disease: The Obesity Society 2018 Position Statement,” Obesity (Silver Spring) 27 (2019): 7–9. [DOI] [PubMed] [Google Scholar]
- 11. Loos R. J. F. and Yeo G. S. H., “The Genetics of Obesity: From Discovery to Biology,” Nature Reviews. Genetics 23 (2022): 120–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Sun C., Kovacs P., and Guiu‐Jurado E., “Genetics of Body Fat Distribution: Comparative Analyses in Populations With European, Asian and African Ancestries,” Genes 12 (2021): 841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Akiyama M., Okada Y., Kanai M., et al., “Genome‐Wide Association Study Identifies 112 New Loci for Body Mass Index in the Japanese Population,” Nature Genetics 49 (2017): 1458–1467. [DOI] [PubMed] [Google Scholar]
- 14. Wong H. S., Tsai S. Y., Chu H. W., et al., “Genome‐Wide Association Study Identifies Genetic Risk Loci for Adiposity in a Taiwanese Population,” PLoS Genetics 18 (2022): e1009952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Nakayama K. and Inaba Y., “Genetic Variants Influencing Obesity‐Related Traits in Japanese Population,” Annals of Human Biology 46 (2019): 298–304. [DOI] [PubMed] [Google Scholar]
- 16. Sun C., Kovacs P., and Guiu‐Jurado E., “Genetics of Obesity in East Asians,” Frontiers in Genetics 11 (2020): 575049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Feng Y. A., Chen C. Y., Chen T. T., et al., “Taiwan Biobank: A Rich Biomedical Research Database of the Taiwanese Population,” Cell Genomics 2 (2022): 100197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Vasold K. L., Parks A. C., Phelan D. M. L., Pontifex M. B., and Pivarnik J. M., “Reliability and Validity of Commercially Available Low‐Cost Bioelectrical Impedance Analysis,” International Journal of Sport Nutrition and Exercise Metabolism 29 (2019): 406–410. [DOI] [PubMed] [Google Scholar]
- 19. Verney J., Schwartz C., Amiche S., Pereira B., and Thivel D., “Comparisons of a Multi‐Frequency Bioelectrical Impedance Analysis to the Dual‐Energy X‐Ray Absorptiometry Scan in Healthy Young Adults Depending on Their Physical Activity Level,” Journal of Human Kinetics 47 (2015): 73–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Ritchie J. D., Miller C. K., and Smiciklas‐Wright H., “Tanita Foot‐To‐Foot Bioelectrical Impedance Analysis System Validated in Older Adults,” Journal of the American Dietetic Association 105 (2005): 1617–1619. [DOI] [PubMed] [Google Scholar]
- 21. Kyle U. G., Bosaeus I., De Lorenzo A. D., et al., “Bioelectrical Impedance Analysis‐Part II: Utilization in Clinical Practice,” Clinical Nutrition 23 (2004): 1430–1453. [DOI] [PubMed] [Google Scholar]
- 22. Health Promotion Administration, Taiwan Ministry of Health and Welfare (MOHW) , Taiwan's Obesity Prevention and Management Strategy, (MOHW, 2018), https://www.hpa.gov.tw/Pages/EBook.aspx?nodeid=3813. [Google Scholar]
- 23. World Health Organization , “Appropriate Body‐Mass Index for Asian Populations and Its Implications for Policy and Intervention Strategies,” Lancet 363 (2004): 157–163. [DOI] [PubMed] [Google Scholar]
- 24. Lui D. T. W., Ako J., Dalal J., et al., “Obesity in the Asia‐Pacific Region: Current Perspectives,” Journal of Asian Pacific Society of Cardiology 3 (2024): e21. [Google Scholar]
- 25. Beeson W. L., Batech M., Schultz E., et al., “Comparison of Body Composition by Bioelectrical Impedance Analysis and Dual‐Energy X‐Ray Absorptiometry in Hispanic Diabetics,” International Journal of Body Composition Research 8 (2010): 45–50. [PMC free article] [PubMed] [Google Scholar]
- 26. Feng Q., Bešević J., Conroy M., Omiyale W., Lacey B., and Allen N., “Comparison of Body Composition Measures Assessed by Bioelectrical Impedance Analysis Versus Dual‐Energy X‐Ray Absorptiometry in the United Kingdom Biobank,” Clinical Nutrition ESPEN 63 (2024): 214–225. [DOI] [PubMed] [Google Scholar]
- 27. Heymsfield S. B., “Algorithm Development for Estimating Visceral Fat Rating, Columbia University College of Physicians and Surgeons, Tanita Institute Contract Study” (2004).
- 28. Wang Z., “Japanese‐American Differences in Visceral Adiposity and a Simplified Estimation Method for Visceral Adipose Tissue,” abstract 518‐P, North American Association for the Study of Obesity Annual Meeting, (2004).
- 29. Tanita Corporation , Tanita Technical Bulletin: Visceral Fat Measurement (Tanita Corporation, 2013), https://support.tanita.eu/support/solutions/. [Google Scholar]
- 30. Wei C.‐Y., Yang J.‐H., Yeh E.‐C., et al., “Genetic Profiles of 103,106 Individuals in the Taiwan Biobank Provide Insights Into the Health and History of Han Chinese,” NPJ Genomic Medicine 6 (2021): 10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Turner S., Armstrong L. L., Bradford Y., et al., “Quality Control Procedures for Genome‐Wide Association Studies,” Current Protocols in Human Genetics Chapter 1 (2011): Unit1.19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Jo J., Ha N., Ji Y., et al., “Genetic Determinants of Obesity in Korean Populations: Exploring Genome‐Wide Associations and Polygenic Risk Scores,” Briefings in Bioinformatics 25 (2024): bbae389. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Braglia S., Davoli R., Zappavigna A., et al., “SNPs of MYPN and TTN Genes Are Associated to Meat and Carcass Traits in Italian Large White and Italian Duroc Pigs,” Molecular Biology Reports 40 (2013): 6927–6933. [DOI] [PubMed] [Google Scholar]
- 34. Li Y., Cheng G., Yamada T., Liu J., Zan L., and Tong B., “Effect of Expressions and SNPs of Candidate Genes on Intramuscular Fat Content in Qinchuan Cattle,” Animals (Basel) 10 (2020): 10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Johnson R., Fletcher R. A., Peters S., et al., “Titin‐Related Familial Dilated Cardiomyopathy: Factors Associated With Disease Onset,” European Heart Journal 46 (2025): 5240–5257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Gohlke J., Lindqvist J., Hourani Z., et al., “Pathomechanisms of Monoallelic Variants in TTN Causing Skeletal Muscle Disease,” Human Molecular Genetics 33 (2024): 2003–2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Skriver S. V., Krett B., Poulsen N. S., et al., “Skeletal Muscle Involvement in Patients With Truncations of Titin and Familial Dilated Cardiomyopathy,” JACC. Heart Failure 12 (2024): 740–753. [DOI] [PubMed] [Google Scholar]
- 38. Karastergiou K., Smith S. R., Greenberg A. S., and Fried S. K., “Sex Differences in Human Adipose Tissues—The Biology of Pear Shape,” Biology of Sex Differences 3 (2012): 13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Wymer D. T., Patel K. P., W. F. Burke, 3rd , and Bhatia V. K., “Phase‐Contrast MRI: Physics, Techniques, and Clinical Applications,” Radiographics 40 (2020): 122–140. [DOI] [PubMed] [Google Scholar]
- 40. Loos R. J. and Yeo G. S., “The Bigger Picture of FTO: The First GWAS‐Identified Obesity Gene,” Nature Reviews Endocrinology 10 (2014): 51–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Huang C., Chen W., and Wang X., “Studies on the Fat Mass and Obesity‐Associated (FTO) Gene and Its Impact on Obesity‐Associated Diseases,” Genes & Diseases 10 (2023): 2351–2365. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Poosri S., Boonyuen U., Chupeerach C., Soonthornworasiri N., Kwanbunjan K., and Prangthip P., “Association of FTO Variants rs9939609 and rs1421085 With Elevated Sugar and Fat Consumption in Adult Obesity,” Scientific Reports 14 (2024): 25618. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Rask‐Andersen M., Karlsson T., Ek W. E., and Johansson Å., “Genome‐Wide Association Study of Body Fat Distribution Identifies Adiposity Loci and Sex‐Specific Genetic Effects,” Nature Communications 10 (2019): 339. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: Baseline characteristics of total and genetic testing group.
Table S2: Genotype distribution of SNPs related to visceral fat level.
Table S3: Genotype distribution of SNPs related to fat free mass.
Table S4: Genotype distribution of SNPs related to fat mass.
Table S5: Crude odds ratios for SNP associations with visceral fat level.
Table S6: Crude odds ratios for SNP associations with fat‐free mass.
Table S7: Crude odds ratios for SNP associations with fat mass.
Table S8: Cross‐trait and trait‐specific SNP annotation from GWAS of body composition traits.
Table S9: Sex‐stratified associations of rs794727539 with visceral fat and fat‐free mass.
Table S10: SNP‐by‐sex interaction effects of rs794727539 on visceral fat and fat‐free mass.
Data Availability Statement
The data could be applied to use via an application proceeding at Taiwan Biobank (https://www.biobank.org.tw/english.php).