Plant-based and high protein diets are associated with a lower risk of obesity among Koreans at a high genetic risk of obesity.

doi:10.21203/rs.3.rs-2076727/v1

Download PDF

Article

Plant-based and high protein diets are associated with a lower risk of obesity among Koreans at a high genetic risk of obesity.

https://doi.org/10.21203/rs.3.rs-2076727/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background/Objectives: Obesity has become a severe public health challenge globally. The present study aimed to identify separate and interactive dietary, genetic, and other factors that increase the risk of obesity, as measured by body fat mass.

Subjects/Methods: We utilized a genome-wide association study to identify genetic variants associated with high-fat mass(obesity; n=10,502) and combined them to generate polygenic risk scores(PRS) of genetic variants interacting with each other in adults aged over 40(n=58,701). We then evaluated dietary and lifestyle factors in subjects to assess what factors might help overcome a genetic propensity for higher body fat.

Results: The three-SNP model included brain-derived neurotrophic factor(BDNF)_rs6265, fat mass and obesity-associated protein(FTO)_rs1421085, and SEC16B_rs509325. ADCY3_rs6545790 and BAIAP2_rs35867081 minor alleles increased their gene expression in the visceral and subcutaneous adipocytes, but ADCY3_rs6545790 and BAIAP2_rs35867081 minor alleles decreased their gene expression in the hypothalamus. In the three-SNP model, the PRS was associated with body fat mass by 1.408 and 1.396 times after adjusting covariates 1 and 2, respectively. However, when separating subjects by PRS of the three-SNP model, a plant-based diet was the most significant factor associated with low body fat, followed by high protein diets and lower energy intakes. They could offset the effects of high genetic risk for high body fat.

Conclusions: Modulating nutrient intakes might overcome a high genetic risk for obesity. Dietary choices favoring more plant-based and higher protein foods might help prevent increased body fat in Asians and potentially people of other ethnicities with high polygenetic risk scores.

Body fat

polygenic risk scores

plant-based diet

vitamin C

vitamin D

fiber

Obesity is a multi-factored pathological condition involving genetic, environmental, and behavioral factors. Despite intense research efforts over many years, no practical solutions for the prevention and treatment of obesity have been realized. It has been estimated that 40–75% of the variability in BMI is genetic [1]. However, known combinations of single-nucleotide polymorphisms (SNPs) can only predict about 6% of the BMI variability in people of European descent [2]. Although genetic susceptibility to obesity is incompletely described, genome-wide association studies (GWAS) have identified patterns of genetic variations that increase the risk of obesity [3]. Applicability of these genetic factors among all populations is impossible since there are ethnic differences in the risk alleles [4], and many study results may not apply to people of different races. It is also unknown whether genetic predispositions to obesity plays a causative or permissive role in most obesity. There have always been examples of people with severe obesity due to inherited traits, but that is uncommon. The recent rapid global increases in obesity rates suggest a more permissive role since the genomes of populations are slow to evolve [5]. It is well known that many people who carry obesity-related genes do not become obese, and for most people, genetics do not equate to destiny [5].

Many studies use BMI as a measure of obesity, but BMI may not represent obesity equally in people, especially in those with a high muscle mass or different fat distributions. It has been demonstrated that people of different ethnicities with similar BMI may exhibit rather significant differences in fat mass [6]. BMI reflects lean body mass and fat mass, but obesity is best defined by fat mass, which is difficult to measure accurately. Furthermore, it is difficult to separate subcutaneous fat from the more harmful visceral fat. However, body fat mass can be determined by several methods, including dual-energy X-ray absorptiometry (DEXA), bioelectrical impedance analysis (BIA), and hydrodensitometry. DEXA and hydrodensitometry can estimate body fat more accurately than BIA [7]. Furthermore, DEXA can predict visceral and subcutaneous fat mass separately, but DEXA and hydrodensitometry are not applicable for epidemiological studies [8]. The BIA method is substantially accurate for measuring body fat mass. Therefore, we conducted a GWAS study in a large Korean cohort and evaluated dietary and lifestyle factors that interact with genetic predispositions to obesity defined by body fat mass measured by the BIA method.

The association of fat mass with genetics needs to be conducted in large cohorts since numerous dietary and lifestyle factors make small but statistically significant contributions to fat mass. In the Korean Genome and Epidemiology Study (KoGES), fat mass was measured in the Ansan/Ansung cohort (n = 8842) by BIA, and estimated in a city hospital-based cohort study (n = 58,701) by a prediction model generated by a machine learning approach. In an attempt to isolate more significant effects that may be useful as effective real-life interventions, we took the analysis a step further and developed polygenic risk scores (PRS) for obesity and separated the population according to those with high and low PRSs, and then divided each into those with high and low body fat composition. We hypothesized that investigating people with a high PRS that do not develop obesity would provide information on the most important dietary factors required to maintain a lean body type, even in those predisposed to obesity. This study aimed to determine the fundamental approaches to avoiding obesity in the Korean population.

Participants and setting

The institutional review board (IRB) of the Korean National Institute of Health and the IRB of Hoseo University approved the KoGES and the present study (KBP-2015-055 and 1041231-150811-HR-034-01, respectively). All participants signed a written informed consent form. In this study, 58,701 participants aged 40–74 years were recruited from a large city hospital-based cohort that formed part of the Korean Genome and Epidemiology Study (KoGES) conducted during 2010–2014 [9]. Adults aged over 40 (n = 8842) that participated in the Ansan/Ansung plus rural cohorts during 2001–2002 were used as the replicate study for exploring fat mass-related genetic variants. People with a disease history that might influence energy metabolisms, such as cancers, thyroid diseases, chronic kidney disease, and brain-related diseases, were excluded from the study. The 4,802 and 591 participants were excluded from the urban hospital-based cohort and Ansan/Ansung plus rural cohorts, respectively.

Demographic, Anthropometric, And Biochemical Parameters Of The Participants

The information on age, gender, education, income, alcohol and coffee consumption, physical activity, and smoking history was received during a health interview. Alcohol intake was calculated as the daily amount (g/day) by multiplying the frequency of drinking and the amount at an event [10]. The smoking status was categorized into current smoker (smoked at least 20 cigarettes in the past six months), past smoker (no smoking for at least six months), and never-smoker [10]. The coffee intake was assessed as the weekly drinking frequency and was categorized into three groups by the tertiles of daily coffee intake. Regular physical activity was defined as more than 30 min of moderate physical activity for three or more days per week.

The anthropometric characteristics (height and weight) were measured at the initial visit, as described previously [10]. Body fat and skeletal muscle masses were not determined using the Inbody 3.0 (Cheonan, Korea) based on the BIA method in Ansan/Ansung cohort. However, they were estimated using a method described in a previous study by using a machine learning prediction model generated from the Ansan/Ansung cohort [11] because the appendicular skeletal muscle and fat masses were not measured in the city hospital-based cohort. A doctor measured blood pressure three times via a sphygmomanometer under resting conditions, and the average systolic blood pressure (SBP) and diastolic blood pressure (DBP) were reported. After fasting for more than 12 h, blood was collected in heparin-treated and non-treated tubes, and the plasma glucose concentrations were measured using a Hitachi 7600 Automatic Analyzer (Hitachi, Tokyo, Japan). HbA1c in heparin-treated blood was measured using an automatic analyzer (ZEUS 9.9; Takeda, Tokyo, Japan). Serum total cholesterol, high-density lipoprotein cholesterol (HDL-C), triglyceride, alanine aminotransferase (ALT), aspartate aminotransferase (AST), and creatinine levels were assessed using a Hitachi 7600 Automatic Analyzer. The serum high-sensitive C-reactive protein (hs-CRP) concentrations were measured using a high-sensitivity ELISA kit (Thermofisher, Waltham, MA, USA). The white blood cells (WBCs) were counted from the EDTA-treated blood.

Definition Of Obesity And Metabolic Syndrome

Obesity was defined according to total body-fat mass: men with ≥ 25% (n = 3298) and women with ≥ 30% (n = 7204) considered to be obese and assigned to the high-BF group (a case) in accordance with Korean definitions of obesity. Metabolic syndrome is a cluster of energy, glucose, and lipid disorders categorized according to the 2005 revised National Cholesterol Education Program-Adult Treatment Panel III criteria for Asia [12, 13]. The criteria for MetS were as follows: 1) abdominal obesity (waist circumference ≥ 90 cm for men and ≥ 85 cm for women), 2) elevated fasting blood glucose level (≥ 100 mmol/L) or current use of anti-diabetic medication, 3) elevated blood pressure (average systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg) or current blood pressure medication use, 4) low HDL-C level (< 40 mg/dL for men and < 50 mg/dL for women), or 5) elevated serum triglyceride level (≥ 150 mmol/L) or current use of anti-dyslipidemic medication. Participants meeting three or more criteria were considered to have MetS.

Usual Food Intake Using A Semi-quantitative Food Frequency Questionnaire (Sqffq)

During the last 12 months of the interview, the food intake of each participant was determined using an SQFFQ designed for the Korean diet, and their accuracy and reproducibility were validated with the three-day food records in four seasons for Koreans [14]. The SQFFQ includes 106 food items that Koreans commonly consume, and the food frequencies were categorized into never or seldom, once per month, two to three times monthly, once or twice weekly, three or four times weekly, five or six times weekly, daily, twice daily, and ≥ 3 times daily. The food amount at a meal was scored as more than, equal to, or less than the regular portion size visualized by photographs of 103 foods. The participants checked the frequencies and the portion size of 106 food items in the SQFFQ. The daily food intake was calculated by multiplying the median of the weekly-consumed frequencies by portion sizes. The food intake is given in grams/day. The daily energy, carbohydrates, fat, protein, vitamin, and mineral intakes were calculated from the SQFFQ results using Can-Pro 2.0 nutrient assessment software designed by the Korean Nutrition Society.

Dietary Patterns By Principal Components Analysis

For dietary pattern analysis, 103 food items in the SQFFQ were classified into 30 predefined food groups, as reported previously [15]. The dietary patterns were designed using principal component analysis of the 30 food groups based on eigenvalues > 1.5, and four dietary patterns explained the criteria [15]. The orthogonal rotation procedure (varimax) yielded four dietary patterns uncorrelated with each other, and foods with ≥ 0.40 factor-loading values were considered the predominant contributors to the assigned dietary pattern [15]. Supplemental Table 1 lists the foods in each dietary pattern. These patterns indicate the participant’s diet types, and they were divided into the Korean-balanced diet (KBD), plant-based diet (PBD), Western-style diet (WSD), or rice-based diet (RBD) according to including foods with ≥ 0.40 factor-loading values.

Dietary inflammatory index (DII)

DII, an index of the pro-inflammatory potential of diets, was calculated from the equation with assigned food and nutrient intakes using their dietary inflammatory weights, including the energy, 32 nutrients, four food products, four spices, and caffeine, as described previously [16]. Because the SQFFQ did not include garlic, ginger, saffron, and turmeric, their intakes were excluded from the DII calculation. DII was calculated by multiplying the dietary inflammatory scores of the 38 food and nutrient components by the daily intakes, and the sums of the scores of 38 items were divided by 100.

Genotyping Using A Korean Chip And Quality Control

The Center for Genome Science at the Korea National Institute of Health determined the participants' genotypes in the Ansan/Ansung and city hospital-based cohorts. The genomic DNA was isolated from whole blood, and genotypes were measured using a Korean Chip (Affymetrix, Santa Clara, CA) designed to examine the disease-related single nucleotide polymorphisms (SNPs) in Koreans [9]. The genotyping accuracy was estimated using Bayesian Robust Linear Modeling in the Mahalanobis Distance Genotyping Algorithm. The inclusion criteria of the genotyping accuracy, missing genotype call rate, and heterozygosity were ≥ 98%, < 4%, and < 30%, respectively, and the data showed no gender bias. The genetic variants that were included satisfied the Hardy-Weinberg equilibrium (HWE) at P > 0.05 and minor allele frequency (MAF) at > 5% [17]. Manhattan and quantile-quantile (Q-Q) plots indicated the accuracy of GWAS data using the Fastman library in the R program [17]. A Manhattan plot of genetic variants was displayed with the negative logarithms of the association p-values for type 2 diabetes. A Q-Q plot is a probability plot to show the goodness of fit of the actual data distribution to the theoretical data distribution. The Q-Q plot of genotype data displayed the quantile distribution of observed p-values (on the y-axis) versus the quantile distribution of expected p-values (on the x-axis). The Q-Q plot was constructed to assure that the lambda value of Q-Q plot was close to 1 and confirmed that the GWAS genotypes were ideal. The pathways linked to the genetic variants associated with type 2 diabetes having P-values < 0.05 for Bonferroni correction were selected using the MAGMA gene-set analysis in SNP2GENE of FUMA web application, available through the git repository (https://github.com/Kyoko-wtnb/FUMA-webapp/).

Selection of the genetic variants that influence obesity defined by fat mass and the best model with SNP-SNP interactions

Figure 1 presents the procedure of selecting genetic variants for high body fat risk and investigating the best model for SNP-SNP interactions. The GWAS was conducted to explore genetic variants associated with obesity risk in the urban hospital-based cohort (P < 5X10^− 5). From the GWAS associated with obesity risk, 1992 genetic variants were selected at P < 5×10⁻⁵. We eliminated 387 genetic variants that did not meet MAF (< 1%) and HWE (P < 0.05). In the gene name search using g:Profiler (https://biit.cs.ut.ee/gprofiler/snpense), 178 SNPs were not identified with gene names, and 126 gene names of the 1427 genetic variants were identified. The linkage disequilibrium (LD) analyses were performed on the SNPs of the 1427 genetic variants using Haploview 4.2 in PLINK. The potential genetic variants in the same chromosome were not strongly correlated (D’ <0.2). The SNPs with high D’ values were not included in the generalized multifactor dimensionality reduction (GMDR) because they provided the same information on the genetic impact. There were 19 SNPs in 18 genes selected, and genes associated with fat mass were selected using HuGE Navigator (https://phgkb.cdc.gov/PHGKB/ hNHome.action).

Among the 14 genetic variants selected for obesity risk, ten SNPs with an SNP-SNP interaction were selected automatically by GMDR. The best SNP-SNP interaction model was selected in a sign rank test of trained balanced accuracy (TRBA) and testing balanced accuracy (TEBA) while adjusting for the covariates using a GMDR program and a P-value threshold of 0.05 [10]. The covariates used were age, gender, residence area, education, and income for models 1 and 2, plus energy intake, alcohol intake, regular exercise, and smoking status. Ten-fold cross-validation was also checked for cross-validation consistency (CVC) because the sample size was larger than 1000 [18], and 10 out of 10 in the CVC met the perfect cross-validation criteria.

The risk allele number of each SNP was counted to generate the PRS of the best model. For example, the genetic score for the SNP was 2, 1, and 0 when the participants had AA, AG, and GG of one SNP, and the A allele was the risk allele, respectively. The polygenic risk score (PRS) of the best model was assessed by summing the number of the risk alleles from each selected SNP in the best gene-gene interaction model [3]. The PRS in the three and six SNP models was divided into three categories according to the number of risk alleles; they were classified as Low-PRS, Middle-PRS, and High-PRS when the number of risk alleles in the PRS was 0–2 (n = 19,686), 3–4 (n = 30,513), and ≥ 5 (n = 3,629) in the three-SNP model and 0–5 (n = 27,212), 6–7 (n = 20375), and ≥ 8 (n = 1822) in the six-SNP model, respectively. Among the best models to meet the p-value of the sign test and CVC, the model with the lowest SNP number (three-SNP model) was used to interact with the lifestyle parameters.

Expression Quantitative Trait Locus (Eqtl) Analysis

The eQTL analysis is a direct approach to estimating the candidate gene expression of the genetic variants at risk loci. The allele variants are involved in the corresponding gene expression, and the expression of candidate susceptible genes with risk alleles is estimated to influence various diseases. Gene expressions of the genetic variants related to the abdominal obesity risk was identified by eQTL analysis in the Genotype-Tissue Expression (GTE) × eQTL calculator (https://gtexportal.org/home/testyourown, accessed on July 19, 2021). The gene expressions in the subcutaneous and visceral adipose tissues, skeletal muscles, liver, and brain were calculated using the GTE × eQTL calculator.

Statistical analysis

The statistical analysis was performed using SAS (version 9.3; SAS Institute, Cary, NC, USA). A sample size of 53,828 was sufficient to achieve significance at α = 0.05 and β = 0.99 at an odds ratio of 1.05 in the logistic analysis using a G-power calculator. The descriptive statistics for categorical variables, such as gender and dietary habits, were obtained by determining the frequency distributions, which were analyzed statistically according to the immunity groups of the classification variables using a chi-square test. Descriptive statistics of the continuous variables were analyzed as the adjusted means with standard deviations after adjusting for the covariates. The statistical differences among the gender and insulin resistance groups were compared using a two-way analysis of covariance (ANCOVA) [19]. Multiple comparisons of the groups were performed using a Tukey’s test.

The association of insulin resistance on metabolic parameters was examined by logistic regression analysis with the Low-BF group as the reference after adjustment for covariates. The results are presented as the odds ratios (ORs) and 95% confidence intervals (CI) of each biochemical parameter for the High-BF and Low-BF groups. The first model was generated after adjusting for age, residence area, survey year, lean body mass, education, and income. The second model was produced with the adjustments for covariates in model 1 and the energy intake, physical activity, smoking status, and alcohol consumption.

The lifestyle-related parameters were categorized into the high or low groups using the criteria defined by the predesignated cutoffs, such as the dietary reference intake or 30th percentiles of each variable, to determine the interactions between the fat mass and lifestyle parameters. Two-way ANCOVA was used to analyze the interactions between the fat mass groups and the lifestyle parameters, including dietary intake, smoking, and physical activity. The main effects were insulin resistance and lifestyle-related parameters with their interaction terms after adjusting for covariates. The ORs and 95% CI of fat mass with lifestyle-related parameters were also calculated by logistic regression analysis in the High-BF and Low-BF groups of the lifestyle-related parameters. The significant difference in the High-BF percentage was analyzed using the PRS groups in the χ² test in the low- and high groups of lifestyle-related parameters.

Demographic characteristics and lifestyles according to genders and fat mass

Age was higher in the high-BF group than the low-BF group for both genders. Education and income also affected body fat mass for both genders: The participants with less than a high school education and lower income had higher incidences of high body fat mass (Table 1). Energy intake minimally affected fat mass, and carbohydrate and fat proportions showed similar trends as energy intake. In dietary fat composition, MUFA intake was higher, and PUFA intake was lower in the high-BF group than in the low-BF group. Vitamin C, fiber, and Ca intakes were lower in the high-BF group than in the low-BF group only in women, but vitamin D intake showed the same trend in both genders (Table 1). DII was the same between fat mass groups in both genders, although DII was higher in men than women. Among dietary patterns, the number of men consuming a KBD was higher in the high-fat mass group, but it was the opposite in women. The number of participants with High PBD was lower in the high-BF group only in women, but those with a high WSD were higher in the high-BF group only in men (Table 1). The participants who were smokers were lower in the high-BF group than the low-BF group, only in men. Alcohol intake did not differ between low- and high-BF groups for either gender. However, the number of participants who participated in regular exercise was lower in the high-BF group than in the low-BF group (Table 1).

Table 1

Demographic parameters and nutrient intake according to genders and body-fat mass
	Men (n = 19444)		Women (n = 34384)
	Low-fat mass (n = 16146)	High-fat mass (n = 3298)	Low-fat mass (n = 27180)	High-fat mass (n = 7204)
Age (year)	55.7 ± 0.06^b	56.9 ± 0.13^a	51.9 ± 0.05^d	53.8 ± 0.09^c***+++###
Education ≤ Middle school High school ≥Collage	1318 (13.2) 7540 (76.2) 1040 (10.5)	343 (16.8)^‡‡‡ 1491 (72.8) 214 (10.5)	4290 (20.3) 15479 (73.4) 1322 (6.27)	1727 (28.4)^‡‡‡ 4074 (67.0) 276 (4.54)
Income ≤$2000 $2000–4000 >$4000	1258 (8.24) 6565 (42.8) 7479 (49.0)	251 (7.76)^‡ 1311 (40.5) 1672 (51.7)	2848 (11.2) 10903 (43.0) 11579 (45.7)	868 (12.4)^‡‡‡ 3369 (48.0) 2777 (39.6)
Energy (EER %)	86.0 ± 0.06^b	85.4 ± 0.13^c	104 ± 0.05^a	104 ± 0.09^a***+++
Carbohydrates (En %)	71.6 ± 0.08	71.3 ± 0.17	71.7 ± 0.06	71.6 ± 0.12⁺⁺
Proteins (En%)	13.3 ± 0.03^b	13.4 ± 0.05^b	13.6 ± 0.02^a	13.6 ± 0.04^a***#
Fat (En%)	13.9 ± 0.06	14.2 ± 0.12	13.9 ± 0.04	14.1 ± 0.09
SFA (En%)	4.46 ± 0.02	4.55 ± 0.06	4.45 ± 0.02	4.43 + 0.04
MUFA (En%)	5.62 ± 0.03^a	5.80 ± 0.06^a	5.45 ± 0.02^b	5.48 ± 0.05^b***+
PUFA (En%)	3.26 ± 0.03^a	3.21 ± 0.05^a	3.12 ± 0.02^b	3.07 ± 0.04^b**
Cholesterol (mg/d)	169 ± 1.07	173 ± 2.25	170 ± 0.81	171 ± 1.52
Vitamin C (mg/d)	93.8 ± 0.68^c	91.6 ± 1.50^c	121 ± 0.53^a	115 ± 1.05^b***++
Vitamin D (ug/d)	5.68 ± 0.05^c	5.39 ± 0.10^d	6.94 ± 0.04^a	6.57 ± 0.07^b***+++
Fiber (g/d)	14.6 ± 0.09^b	14.1 ± 0.20^b	15.3 ± 0.07^a	14.4 ± 0.14^b***+++
DII (scores)	-19.96 ± 0.02^b	-21.07 ± 0.04^b	-21.26 ± 0.01^a	-21.34 ± 0.03^a**+#
Ca (mg/d)	417 ± 3.42^b	414 ± 5.56^b	491 ± 2.31^a	484 ± 3.81^a***
KBD (Yes, %)	6671 (39.6)	1430(41.7)^‡	9165 (30.1)	2303(28.8)^‡
PBD (Yes, %)	3488 (20.7)	710 (20.7)	12346 (40.6)	3032(37.9)^‡‡‡
WSD (Yes, %)	8489 (50.4)	1933 (56.3)^‡‡‡	10333 (34.0)	2788(34.8)
RMD (Yes, %)	5376 (31.9)	1089(31.7)	10370 (34.1)	2736(34.2)
Non-Smoking (Yes, %) Former smoking (Yes, %) Smoking(Yes, %)	4712 (29.3) 6770 (42.1) 4612 (28.7)	834 (25.3)^‡‡‡ 1546 (46.9) 917 (27.8)	26201 (96.8) 326 (1.2) 532 (1.97)	6959 (96.7) 90 (1.25) 149 (2.07)
Alcohol drinking (g/day)	35.1 ± 0.40^b	38.9 ± 0.84^a	5.33 ± 0.29^c	5.51 ± 0.56^c***+++###
Regular exercise (Yes, %)	10155 (60.5)	1797 (52.4) ^‡‡‡	16350 (54.0)	3674 (45.9) ^‡‡‡
Values represented adjusted means + standard error for continuous variables and the number and percentages of participants for categorical variables. Adjusted for age, BMI, education, income, energy intake, alcohol intake, smoking, and total activity. High body-fat (case) was defined as 25% and 30% of body weight for men and women respectively. En%, the percentage based in energy intake.
^* Significantly different by genders at P < 0.05, ^ at P < 0.01, ^* at P < 0.001.
⁺ significantly different by fat mass at P < 0.05, ⁺⁺ at P < 0.01, ⁺⁺⁺ at P < 0.001.
^# significant interactions at P < 0.05, ^## at P < 0.01, ^### at P < 0.001.
^‡Significantly different from Low fat mass group at P < 0.05.
^a,b,c Different letters on the bar indicated significant differences among the groups in Tukey’s test at p < 0.05.

Anthropometric And Biochemical Parameters

The participants with high body fat mass were shorter and had higher BMIs and waist circumferences but lower SMI than those with low body fat mass (Table 2). Height and SMI were inversely associated with body fat mass, and BMI and waist circumferences were positively linked to body-fat mass. MetS was higher in the high-BF group than in the low-BF group and positively associated with body fat mass by 6.29 times (Table 2). Serum glucose and blood HbA1c concentrations were higher in the high-BF group than in the low-BF group, while they were positively related to body fat mass only for women (Table 2). Furthermore, dyslipidemia, including hypercholesterolemia, hyper-LDL-cholesterolemia, hypo-HDL-cholesterolemia, and hypertriglyceridemia, was also positively linked to fat mass (Table 2). However, hypertension was less associated with fat mass. eGFR was lower in high-fat mass than low-fat mass in men, not women. Serum AST and ALT concentrations, known as liver damage indexes, were higher in the high-BF group than in the low-BF group, whereas they were positively associated with fat mass (Table 2). Serum hs-CRP concentrations were higher in the High-fat mas s group than in the Low-fat mass group, and they were positively related to fat mass (Table 2).

Table 2

Adjusted means and association of anthropometric and biochemical variables in the urban hospital-based cohort according to genders and body-fat mass
	Men (n = 19444)				Adjusted ORs and 95% CI
	Low-fat mass (n = 16146)	High-fat mass (n = 3298)	Low-fat mass (n = 27180)	High-fat mass (n = 7204)	Adjusted ORs and 95% CI
Height (cm)¹	169.0 ± 0.04^b	167.1 ± 0.09^a	157.0 ± 0.03^d	154.9 ± 0.07^c***+++#	3.093 (2.881–3.320)
BMI (mg/kg²)²	23.9 ± 0.02^c	26.9 ± 0.05^a	22.9 ± 0.02^d	26.1 ± 0.03^b***###+++	18.05 (17.03–19.14)
Waist (cm)³	84.3 ± 0.06^c	91.7 ± 0.13^a	76.7 ± 0.05^d	83.4 ± 0.09^b***###+++	5.038 (4.794 5.293)
SMI (%) ⁴	7.84 ± 0.01^a	7.50 ± 0.01^b	7.05 ± 0.003c	6.68 ± 0.006^d***###+++	0.858 (0.745–0.954)
Fat mass (%)⁵	22.4 ± 0.05^d	26.3 ± 0.09^c	29.8 ± 0.04^b	33.9 ± 0.06^a***+++
MetS (%)⁶	2393(14.2)	1205 (35.1)^‡‡‡	2933 (9.65)	1769 (22.1)^‡‡‡	6.289 (5.833–6.780)
Serum glucose (mg/dL)⁷	98.2 ± 0.17^b	99.6 ± 0.37^a	93.2 ± 0.13^d	94.1 ± 0.25^c***+++	1.178 (1.103–1.257)
HbA1c (%)⁸	5.72 ± 0.01^b	5.77 ± 0.01^a	5.70 ± 0.01^b	5.69 ± 0.01^b***##	1.454 (1.321–1.601)
Serum total cholesterol⁹	190.4 ± 0.31^d	194.9 ± 0.68^c	199.4 ± 0.24^b	207.3 ± 0.46^a***++	1.073 (1.001–1.150)
Serum HDL¹⁰	49.2 ± 0.11^c	49.8 ± 0.24^c	55.8 ± 0.08^a	57.6 ± 0.16^b***+++###	1.210 (1.135–1.290)
Serum LDL¹¹	113 ± 0.28^b	114 ± 0.62^c	121 ± 0.22^a	126 ± 0.42^a***+++###	1.089 (1.007–1.177)
Serum TG¹²	142 ± 0.72^b	158 ± 1.59^a	115 ± 0.57^c	117 ± 1.07^c***+++###	1.229 (1.156–1.308)
Serum hsCRP¹³	0.153 ± 0.004^b	0.187 ± 0.007^a	0.118 ± 0.003^c	0.154 ± 0.005^b***+++	1.266 (1.008–1.589)
SBP (mmHg)¹⁴	125.0 ± 0.14^a	124.9 ± 0.24^a	121.0 ± 0.10^b	121.4 ± 0.20^b***	1.105 (1.040–1.175)
DBP (mmHg)¹⁵	78.1 ± 0.10^a	77.9 ± 0.16^a	74.4 ± 0.07^b	74.6 ± 0.13^b***#	1.040 (0.948–1.140)
eGFR (ml/min)¹⁶	84.5 ± 0.16^b	83.0 ± 0.26^c	87.2 ± 0.11^a	87.7 ± 0.23^a***+###	0.774 (0.687–0.873)
Serum AST (U/L)¹⁷	24.6 ± 0.24^b	25.7 ± 0.40^a	23.1 ± 0.16^c	23.2 ± 0.34^c**+++	1.425 (1.250–1.624)
Serum ALT(U/L)¹⁷	25.1 ± 0.24^b	27.9 ± 0.39^a	20.5 ± 0.16^c	20.9 ± 0.33^b***+++###	1.333 (1.227–1.448)
Values represented adjusted means + standard error for continuous variables and the number and percentages of participants for categorical variables. Adjusted for age, BMI, education, income, energy intake, alcohol intake, smoking, and total activity. High body-fat (case) was defined as 25% and 30% of body weight for men and women respectively. MetS, metabolic syndrome
The cutoff points of the reference for logistic regression with adjusted for age, BMI, education, income, energy intake, alcohol intake, smoking, and total activity.
were as following: ¹<172.5 cm for men and < 160 cm for women; ²<25 kg/m2 for BMI; ³< 90 cm for men and 85 cm for women waist circumferences; ⁴ <29.0% for men and 22.8% for women in skeletal muscle index (SMI; defined as appendicular skeletal muscle mass/weight);⁵< 25% for men and 30% for women for fat mass; ⁶Metabolic syndrome (MetS); ⁷<126 ml/dL fasting serum glucose plus diabetic drug intake; ⁸<6.5% hbA1c plus diabetic drug intake; ⁹<230 mg/dL plasma total cholesterol concentrations; ¹⁰>40 mg/dL for men and 50 mg/dL for women plasma HDL cholesterol; ¹¹<160 mg/dL plasma LDL cholesterol concentrations; ¹²<150 mg/dL plasma triglyceride concentrations; ¹³<0.5 mg/dL serum high sensitive-C-reactive protein (hs-CRP) concentrations; ¹⁴<140 mmHg SBP, ¹⁵< 90 mmHg DBP plus hypertension medication; ¹⁶ Estimated glomerular filtration rate (eGFR) < 70; ¹⁷Aspartate aminotransferase < 40 U/L; ¹⁸Alanine aminotransferase < 35 U/L.
^* Significantly different by genders at P < 0.05, ^ at P < 0.01, ^* at P < 0.001.
⁺ significantly different by fat mass at P < 0.05, ⁺⁺ at P < 0.01, ⁺⁺⁺ at P < 0.001.
^# significant interactions at P < 0.05, ^## at P < 0.01, ^### at P < 0.001.
^‡Significantly different from Low fat mass group at P < 0.05.
^a,b,c Different letters on the bar indicated significant differences among the groups in Tukey’s test at p < 0.05.

Polygenetic Variants And Their Interactions Related To The Body Fat Mass

The statistical association of genetic variants with fat mass is shown in a Manhattan plot in Fig. 2A. The Q-Q plot (Fig. 2B) showed the quantile distribution of the log of observed p values versus the quantile distribution of the log of expected p values and the lambda value was 1.116. These results suggested that genetic variants for fat mass were in an acceptable range, although they might be somewhat inflated.

Genes of the genetic variants associated with fat mass risk were highly involved in the sodium ion transmembrane transport pathway in the biological process of gene ontology (GO) (Table 3). Interestingly, the genes were associated with myoblast proliferation and the apoptotic process of neurons in the biological process of GO. In the curated gene sets, the biocarta flumazenil pathway and Reactome signaling to p38 via RIT and RIN were associated with genes related to fat mass (Table 3). These pathways were related to toxic compound metabolism by MAPK and RAS.

Table 3

Pathways related to genetic variants for body fat mass
Pathways	No. of genes	Beta	std	P value	P value bon
GO BP: GO sodium ion transmembrane transport	133	0.323	0.027	0.081	3.18e-05
GO BP: GO myoblast proliferation	17	0.693	0.021	0.179	5.41e-05
Curated gene sets: biocarta flumazenil pathway	8	1.208	0.025	0.326	0.000104
GO BP: GO neuron apoptotic process	224	0.211	0.023	0.058	0.000129
Curated gene sets: reactome rmts methylate histone arginine	60	0.387	0.022	0.108	0.000172
GO MF: GO peptide hormone binding	47	0.488	0.024	0.137	0.000183
GO BP: GO positive regulation of vascular endothelial cell proliferation	13	0.778	0.020	0.219	0.000193
GO BP: GO somatic stem cell population maintenance	70	0.359	0.022	0.103	0.000258
Curated gene sets: reactome signaling to p38 via rit and rin	5	1.318	0.022	0.382	0.000272
GO BP: GO chromatin remodeling	157	0.230	0.021	0.067	0.000274
GO, gene ontology; BP, biological process; MF, Molecular function

Ten genetic variants with the genetic variant-genetic variant interaction were presented in Table 4. The ten genetic variants were rs509325_SEC16B, rs6545790_ADCY3, rs7560575_PSME4, rs2196476_SLIT2, rs6265_BDNF, rs587056_FARP1, rs1421085_FTO, rs35867081 _BAIAP2, rs60259426 _SYMPK, and rs6089240_CDH4. The genes were known to be involved in obesity. They met the MAF > 0.05 and p-value for HWE > 0.05, whereas the adjusted ORs were 0.9278–1.35, and their p values were < 5X10E-06 and < 0.0069 in the city hospital-based cohort and Ansan/Ansung cohort plus rural cohort.

Table 4

The characteristics of the ten genetic variants of genes related to fat mass in adults using the generalized multifactor dimensionality reduction analysis
CHR¹	SNP²	Location	Mi³	Ma4	OR⁵	SE⁶	P value for OR⁷	P value for OR⁸	Genes	Feature	⁸MAF	⁹HWE
1	rs509325	177894591	G	T	1.123	0.0145	1.37E-15	8.28E-04	SEC16B	Intron	0.2845	0.5035
2	rs6545790	25109302	A	G	1.063	0.0133	4.87E-06	4.75E-04	ADCY3	Intron	0.4374	0.4064
2	rs7560575	54142030	C	T	0.9889	0.0670	2.33E-07	1.94E-02	PSME4	Transcript	0.0112	0.2631
4	rs2196476	20270600	G	A	1.097	0.0191	1.26E-06	0.0069	SLIT2	Intron	0.135	0.4579
11	rs6265	27679916	C	T	0.923	0.0133	2.45E-10	0.0048	BDNF	Missense	0.4588	0.1481
13	rs587056	98976374	T	C	1.35	0.0636	2.33E-06	0.0067	FARP1	Intron	0.0106	0.5496
16	rs1421085	53800954	T	C	1.173	0.0197	6.24E-16	2.82E-06	FTO	Transcript	0.1245	0.4604
17	rs35867081	79047278	G	A	1.065	0.0137	3.99E-06	0.0052	BAIAP2	Transcript	0.3651	0.2094
19	rs60259426	46340832	G	A	0.938	0.0134	1.54E-07	0.00072	SYMPK	Transcript	0.4202	0.3135
20	rs6089240	60152260	A	G	0.9278	0.0126	1.60E-08	5.81E-04	CDH4	Intron	0.4641	1.0
¹Chromosome; ²Single nucleotide polymorphism; ³Minor allele; ⁴Major allele; ⁵Odds ratio; ⁶Standard error; ⁷P-value for OR after adjusting for age, gender, residence area, survey year, daily energy intake, levels of education and income; ⁸Minor allele frequency; ⁹Hardy-Weinberg equilibrium.

The gene expression according to the alleles of the selected SNPs in different tissues from GTEx v8.

The expression of ADCY3 with the minor allele of rs6545790 was elevated with 0.21 of slope in visceral fat (P = 4.1E-14) and 0.23 in subcutaneous fat (P = 2.2E-19) (Fig. 3). By contrast, it significantly decreased with − 0.18 and − 0.28 of slope in the amygdala (P = 0.0064) and brain cortex (P = 5.4E-10), respectively. The SEC16B expression with the minor allele of rs509325 increased with the slope of 0.24 in the adrenal gland. The BAIAP2_expression with the minor allele of rs35867081 was elevated with 0.066 (P = 0.025) and − 0.14 (P = 0.000035) of slope in the subcutaneous adipose tissue and the hippocampus, respectively (Fig. 3). However, the minor allele of BDNF_rs6265was not significantly associated with gene expression (P = 0.55), although it was a missense mutation.

The Best Model Of Genetic Variants With Snp-snp Interaction For Body Fat Mass

The best models with genetic variant-genetic variant interactions influencing fat mass were selected when satisfying < 0.05 p-value for the sign test of TEBA and CVC 10/10. The models to meet the criteria included 3-, 6-, 7-, 8-, and 10-SNPs (Supplementary Table S1). The three-SNP model included BDNF_rs6265, FTO_rs1421085, and SEC16B_rs509325. In the three-SNP model, the PRS was associated with body fat mass by 1.408 and 1.396 times after adjusting for covariates 1 and 2, respectively (Fig. 4). The six-SNP model contained ADCY3_rs6545790, BDNF_rs6265, SEC16B_rs509325, BAIAP2_rs35867081, SYMPK_rs60259426, and CDH4_rs6089240. The PRS of the six-SNP model was related to body fat mass by 1.36 and 1.37 times, respectively. However, the adjusted ORs in the six-SNP model were decreased compared to the three-SNP model (Fig. 4). Therefore, the three-SNP model was used as the best model for body fat mass risk.

Interaction Of Prs With Lifestyles To Influence Body Fat Mass

Energy intake interacted with PRS of the three-SNP model to affect body fat mass, and PRS was positively associated with body fat mass risk only in low- and high-energy intakes by 1.228 and 1.157 times, respectively (Table 5). The proportion of the participants was higher in the High-PRS group than in the Low-PRS group in both low- and high-energy intake (Fig. 5A). Therefore, the participants with high-PRS increased fat mass in low energy intake, and the genetic impact was negated with high energy intake. Protein intake also interacted with PRS to influence fat mass (Table 5). A high protein intake could prevent the increase of fat mass in the participants in the high-PRS group compared to those in the low-PRS group (Fig. 5B). Other nutrient intakes did not interact with PRS affecting fat mass. Among four dietary patterns, only PBD intake interacted with PRS for fat mass (Table 5). The high body fat mass proportion was higher in the participants with high PRS than those with low PRS in both low- and high-PBD intake. However, the increase was much more significant in the low-PBD intake than in the high-PRB intake (Fig. 5C). Exercise and smoking status did not interact with PRS to affect fat mass. Therefore, these results suggest that Koreans with high PRS can decrease their risk of high body fat by consuming a PBD diet with less energy and high protein intakes.

Table 5

Adjusted odds ratios for fat mass risk by polygenetic risk scores of the three-SNP model (PRS) for gene-gene interaction after covariate adjustments according to the patterns of lifestyles.
	Low- PRSBM (n = 19,686)	Medium-PRSBM (n = 30,513)	High-PRSBM (n = 3,629)	Gene-nutrient interaction P value
Low energy¹ High energy	1	1.150 (1.081 1.223) 1.102 1.021 1.189 )	1.228 1.088 1.387 )^* 1.157 0.997 1.348 )	0.0231
Low CHO² 70 High CHO	1	1.045 (0.893 1.223) 0.991 (0.927 1.060	1.006 (0.866 1.169) 1.079 (1.014 1.149)	0.3493
Low protein³ 13 High protein	1	1.123 (1.052–1.198) 0.961 (0.881 1.049)	1.255 (1.103–1.427)^** 1.019 (0.938–1.107)	0.0272
Low fat⁴ 15 Moderate fat	1	1.000 (0.926 1.080) 0.994 (0.896 1.102)	1.088 (1.013 1.169)^* 1.028 (0.932 1.133)	0.0664
Low alcohol⁵ 20 High alcohol	1	1.139 (1.075 1.208) 1.129 (1.050 1.213)	1.218 (1.086 1.366) 1.204 (1.046 1.386)	0.8083
Low KBD⁶ High KBD	1	0.997 (0.938 1.061) 1.002 (0.929 1.080)	1.067 (1.007 1.130) 1.083 (1.009 1.162)	0.5368
Low PBD⁶ High PBD	1	1.132 (1.081 1.185) 1.023 (0.961 1.088)	1.258 (1.126 1.406) 1.110 (1.006 1.224)	0.0026
Low WSD⁶ High WSD	1	0.997 (0.938 1.061) 0.998 (0.927 1.075)	1.067 (1.007 1.130) 1.074 (1.002 1.151)	0.1356
Low RMD⁶ High RMD	1	1.112 (1.057 1.169) 1.159 (1.096 1.225)	1.197 (1.134 1.263 1.254 (1.128 1.395)	0.8419
Low exercise⁷ High exercise	1	1.034 (0.949 1.128) 0.997 (0.938 1.061)	1.103 (1.018 1.196) 1.067 (1.007 1.130)	0.1778
Non-smoke⁸ Smoke	1	0.997 (0.938 1.061) 1.071 (0.948 1.210)	1.067 (1.007 1.130) 1.041 (0.926 1.169)	0.1328
Values were expressed as odds ratio and 95% confidence intervals. PRS of the 3-SNP model (BDNF_rs6265, FTO_rs1421085, and SEC16B_rs509325) was divided into three categories (0–2, 3–4, and ≥ 5) by three groups as the low, medium, and high groups of the best model of GMDR. The cutoff points of the parameters were defined as: ¹ <estimated energy intake, ² <70% carbohydrate (CHO), ³<13% protein, ⁴<15% fat, ⁵< 20 g/day alcohol, ⁶ <75th percentile, ⁷ 30 min moderate exercise for 3 times a week, ⁸ smoking. Multivariate regression models include the main effects, interaction terms of gene and main effects (energy and nutrient intake), and potential confounders including sex, age, BMI, the status of smoking and drinking, levels of income and education, physical activity, energy intake, percent intake for carbohydrate and fat. Reference was the low-PRS. KBD, Korean-style balanced diet; PBD, plant-based diet (ovovegetarian); WSD, Western-style diet; RMD, rice-main diet.

The present study investigated genetic, lifestyle, and dietary factors that differentiate people with healthy fat mass levels from those with excessive body fat. The study utilized a genome-wide association study (GWAS) followed by an analysis of PRS for high body fat mass (obesity) and comparisons of body fat mass status in subjects according to high- and low-risk scores. This approach was used to highlight what dietary and lifestyle patterns might lead to an individual being lean despite a high genetic risk of obesity and vice versa.

The three genetic variants that constituted the best model for genetic risk for high body fat were variants associated with BDNF, FTO, and SEC16B in the present study. Genetic variants that result in decreased production of BDNF have been shown to increase the obesity risk significantly in humans [20]. The BDNF_rs6265 variant causes a methionine substitution for valine which results in multiple physiological and psychological changes, including impaired fat oxidation in muscle, type 2 diabetes, depression, and disruption of hunger and satiation signals [21, 22]. Interestingly, the consequences of the BDNF_rs6265 are associated with obesity risk, and it interacts with lifestyles, including nutrient intake [23]. Furthermore, it interacts with other obesity-related genetic variants to exacerbate obesity risk, and PRS, including BDNF_rs6265, interact with lifestyles to modulate the obesity risk [3]. These results suggest that neuronal circuit integrity, survival, and synaptic plasticity may be associated with obesity risk.

The FTO gene, located at chromosome 16q12.2, encodes for the fat mass and obesity-associated protein. The FTO protein functions as a nucleic acid demethylase, and thus it is involved in controlling the methylation of DNA and RNA. However, the FTO_rs1421085 non-coding variant does not affect the expression or the functionality of the FTO protein. Mutated FTO (rs1421085) interacts with Iroquois-class homeodomain proteins, IRX-3 and IRX-5, which are also linked to obesity in humans and animals [24, 25]. Laber et al. [26] have demonstrated that the FTO_rs 1421085 interaction with IRS-3 and IRX-5 modulates numerous organismal phenotypes related to diet-induced weight gain and observes that the FTO_rs1421085 genetic variant mediates cross-species conserved changes in steroid patterns following nutritional challenges.

The third member of the best model group, SEC16B_rs509325, is a variant of the SEC16B gene, also known as RGPR-p117, a gene that encodes a protein known as regucalcin that is characterized by its leucine zipper motif. Regucalcin is known to be a transporter of proteins from the endoplasmic reticulum, but evidence to date has not demonstrated the role of the transporter in lipid metabolism [27]. Nevertheless, variants of the SEC16B gene have consistently been shown to be linked to human obesity, especially in Asian populations [3, 28]. Although there has not been any link to fat metabolism established for SEC16B, it is linked to transport peptides involved in regulating satiety, including neuropeptide Y and proopiomelanocortin [3, 29]. All three genes composing the best model affect different aspects of metabolism.

The most important finding of the current study was that the primary dietary factors linked to low body fat content, other than lower energy intake, were high dietary protein consumption and, even more potent, a more plant-based diet. Among the four dietary patterns studied, most had no significant effect on the risk for high body fat content; however, diets considered low in plant-based foods and low in protein resulted in a significantly higher percentage of obese subjects among those with medium- and high- PRS. This result seems counterintuitive since plant-based diets are typically lower in protein than diets high in animal foods [30]. However, these results concur with studies in other ethnic groups that have found that vegetarian and/or plant-based diets support a leaner body type [31, 32].

There are various hypotheses about how a plant-based diet may help maintain a healthy body weight. First, a plant-based diet has a lower energy density, resulting in greater satiety with a lower energy consumption [18, 32]. Furthermore, plant-based diets contain lower fat and higher fiber which may impact the microbiome, which is known to have substantive impacts on energy utilization in humans [33–35]. However, the microbiome research has not progressed to the point that changes in the gut microbiome can be considered a cause or cure for excess body weight. There is also evidence that various phytochemicals from plant-based foods and dietary supplements may have anti-obesity effects by various mechanisms [19]. It has been shown that the anti-inflammatory properties of plant-based diets can prevent obesity and even result in a healthier obese phenotype in people who become obese [36]. However, this study had no clear association between high body fat mass and dietary inflammation index. Even though it has been difficult to demonstrate specific phytochemicals that profoundly affect body fat mass, it has been previously demonstrated that a phytochemical index score was associated with a lower prevalence of obesity in the Korean population [37]. Therefore, the accumulation of current evidence supports the finding in this study that consuming a plant-based diet provides significant benefits for managing body fat.

Previous studies have also found the association of higher protein intake with lower body fat mass. It has also been observed in previous studies, and a recent meta-analysis and systematic review found that diets high in protein (18–59 energy %) were associated with an average of 1.6 kg decrease in body weight [38]. The study did not determine a more precise dietary protein level needed for weight reduction. However, the efficacy of increased protein to support weight loss does have support from a human trial that used biscuits enriched with plant-based protein and found that the group with the protein-added biscuits lost significantly more body weight and fat mass than the control group [39].

In this study, having a plant-based diet had the most significant association with lower body fat, followed by a high protein diet. It suggests that a plant-based diet high in protein might be optimal for the Korean population and possibly people of other ethnic groups. The findings support the possibility that substituting plant protein for animal protein can decrease the incidence of type 2 diabetes, probably by decreasing obesity-induced inflammation [40]. It was also demonstrated that elderly Korean men with concurrent sarcopenia, osteoporosis, and obesity had the same total and animal protein intake but a significantly lower intake of plant protein than controls [41].

This research revealed that modest differences in dietary choices could have a significant impact on fat accumulation in the Korean population. Other studies have shown that people who adhere to vegetarian diets are less obese than their omnivorous counterparts [42] and that vegetarians are healthier even at the same BMI [43]. However, the subjects in the present study with a plant-based diet were not necessarily vegetarians. A plant-based diet in this study indicated that plant foods contributed relatively more to the total energy intake than animal foods, but the subjects did not adhere to any particular dietary restrictions. The lack of a clearly defined diet could be both a strength and a limitation of the study. There was more variability in the dietary type and potentially less of an effect than if all subjects with a plant-based diet were vegetarians. On the other hand, this represents a more “real world” scenario for many people and shows that modest changes can have profound effects.

This study suggested that BDNF_rs6265, FTO_rs1421085, and SEC16B_rs509325 interact with high body fat mass, that and plant-based and high protein diets are beneficial for maintaining a lower body fat in adults with a high PRS of the three genetic variants. This study demonstrated that the discovery of PRS-environment interactions in body fat was applicable in the clinical setting. Therapies targeting the PRS could help prevent obesity and manage weight by applying personalized nutrition programs that emphasize a plant-based diet high in protein.

Ethics statement

CONFLICT OF INTEREST

The authors have no potential conflict of interest.

DATA AVAILABILITY

The data can be retrieved from the corresponding author on reasonable request.

AUTHOR CONTRIBUTIONS

SP conceived, designed and analyzed the data. JWD wrote the draft of the article. All authors contributed to the article and approved the submitted version

Funding

This study was funded by the National Research Foundation of Korea (2019R1A2C1007203).

Jacob R, Bertrand C, Llewellyn C et al. Dietary Mediators of the Genetic Susceptibility to Obesity-Results from the Quebec Family Study. J Nutr. 2022;152:49–58.
Yengo L, Sidorenko J, Kemper KE et al. Meta-analysis of genome-wide association studies for height and body mass index in ∼700000 individuals of European ancestry. Hum Mol Genet. 2018;27:3641–9.
Park S, Yang HJ, Kim MJ et al. Interactions between Polygenic Risk Scores, Dietary Pattern, and Menarche Age with the Obesity Risk in a Large Hospital-Based Cohort. Nutrients. 2021;13.
Konuma T, Okada Y. Statistical genetics and polygenic risk score for precision medicine. Inflamm Regen. 2021;41:18.
Hu FB. 437Genetic Predictors of Obesity. In: Hu FB (ed) Obesity Epidemiology. Oxford University Press, 2008, p 15.
Heymsfield SB, Peterson CM, Thomas DM et al. Why are there race/ethnic differences in adult body mass index-adiposity relationships? A quantitative critical review. Obes Rev. 2016;17:262–75.
Hamilton-James K, Collet T-H, Pichard C et al. Precision and accuracy of bioelectrical impedance analysis devices in supine versus standing position with or without retractable handle in Caucasian subjects. Clin Nutr ESPEN. 2021;45:267–74.
Fang W-H, Yang J-R, Lin C-Y et al. Accuracy augmentation of body composition measurement by bioelectrical impedance analyzer in elderly population. Medicine 2020;99:e19103.
Kim Y, Han BG. Cohort Profile: The Korean Genome and Epidemiology Study (KoGES) Consortium. Int J Epidemiol. 2017;46:e20.
Park S. Association between polygenetic risk scores related to sarcopenia risk and their interactions with regular exercise in a large cohort of Korean adults. Clin Nutr. 2021;40:5355–64.
Wu X, Park S. An Inverse Relation between Hyperglycemia and Skeletal Muscle Mass Predicted by Using a Machine Learning Approach in Middle-Aged and Older Adults in Large Cohorts. J Clin Med. 2021;10.
Park S, Kim K, Lee BK et al. A Healthy Diet Rich in Calcium and Vitamin C Is Inversely Associated with Metabolic Syndrome Risk in Korean Adults from the KNHANES 2013–2017. Nutrients. 2021;13.
Yang HJ, Song MY, Kim MJ et al. Associations between metabolic syndrome and urinary Na-to-K ratio and glomerular filtration rate in middle-aged adults regardless of Na and K intakes. Clin Exp Nephrol. 2020;24:1015–24.
Park S, Daily JW, Song MY et al. Gene-gene and gene-lifestyle interactions of AKAP11, KCNMA1, PUM1, SPTBN1, and EPDR1 on osteoporosis risk in middle-aged adults. Nutrition. 2020;79–80:110859.
Park S, Kang S. A Western-style diet interacts with genetic variants of the LDL receptor to hyper-LDL cholesterolemia in Korean adults. Public Health Nutr. 2021;24:2964–74.
Hébert JR, Shivappa N, Wirth MD et al. Perspective: The Dietary Inflammatory Index (DII)-Lessons Learned, Improvements Made, and Future Directions. Adv Nutr. 2019;10:185–95.
Liu M, Park S. A Causal Relationship between Vitamin C Intake with Hyperglycemia and Metabolic Syndrome Risk: A Two-Sample Mendelian Randomization Study. Antioxidants. 2022;11.
Burns KF, Robson SM, Raynor HA. A low-energy-dense dietary pattern: a narrative review. Nutr Rev. 2022 doi: 10.1093/nutrit/nuac037..
Li X, Zheng L, Zhang B et al. The Structure Basis of Phytochemicals as Metabolic Signals for Combating Obesity. Front Nutr. 2022;9:913883.
Mou Z, Hyde TM, Lipska BK et al. Human Obesity Associated with an Intronic SNP in the Brain-Derived Neurotrophic Factor Locus. Cell Rep. 2015;13:1073–80.
Daily JW, Park S. Interaction of BDNF rs6265 variants and energy and protein intake in the risk for glucose intolerance and type 2 diabetes in middle-aged adults. Nutrition. 2017;33:187–94.
Nguyen VT, Hill B, Sims N et al. Brain-derived neurotrophic factor rs6265 (Val66Met) single nucleotide polymorphism as a master modifier of human pathophysiology. Neural Regen Res. 2023;18:102–6.
Ma X-Y, Qiu WQ, Smith CE et al. Association between BDNF rs6265 and Obesity in the Boston Puerto Rican Health Study. J Obes. 2012;2012:102942.
Smemo S, Tena JJ, Kim KH et al. Obesity-associated variants within FTO form long-range functional connections with IRX3. Nature. 2014;507:371–5.
Claussnitzer M, Dankel SN, Kim K-H et al. FTO Obesity Variant Circuitry and Adipocyte Browning in Humans. N Engl J Med 2015;373:895–907.
Laber S, Forcisi S, Bentley L et al. Linking the FTO obesity rs1421085 variant circuitry to cellular, metabolic, and organismal phenotypes in vivo. Sci Adv. 2021;7.
Yamaguchi M, Murata T. Involvement of regucalcin gene promoter region-related protein-p117, a transcription factor, in human obesity. Biomed Rep. 2017;6:374–8.
Ng MC, Tam CH, So WY et al. Implication of genetic variants near NEGR1, SEC16B, TMEM18, ETV5/DGKG, GNPDA2, LIN7C/BDNF, MTCH2, BCDIN3D/FAIM2, SH2B1, FTO, MC4R, and KCTD15 with obesity and type 2 diabetes in 7705 Chinese. J Clin Endocrinol Metab. 2010;95:2418–25.
Bhattacharyya D, Glick BS. Two mammalian Sec16 homologues have nonredundant functions in endoplasmic reticulum (ER) export and transitional ER organization. Mol Biol Cell. 2007;18:839–49.
Domić J, Grootswagers P, van Loon LJC et al. Perspective: Vegan Diets for Older Adults? A Perspective On the Potential Impact On Muscle Mass and Strength. Adv Nutr. 2022;13:712–25.
Vang A, Singh PN, Lee JW et al. Meats, processed meats, obesity, weight gain and occurrence of diabetes among adults: findings from Adventist Health Studies. Ann Nutr Metab. 2008;52:96–104.
Fuente González CE, Chávez-Servín JL, de la Torre-Carbot K et al. Relationship between Emotional Eating, Consumption of Hyperpalatable Energy-Dense Foods, and Indicators of Nutritional Status: A Systematic Review. J Obes. 2022;2022:4243868.
Riedl RA, Burnett CML, Pearson NA et al. Gut Microbiota Represent a Major Thermogenic Biomass. Function. 2021;2:zqab019.
Armstrong LE, Casa DJ, Belval LN. Metabolism, bioenergetics and thermal physiology: influences of the human intestinal microbiota. Nutr Res Rev. 2019;32:205–17.
Li D, Li Y, Yang S et al. Diet-gut microbiota-epigenetics in metabolic diseases: From mechanisms to therapeutics. Biomed Pharmacother. 2022;153:113290.
Mohamadi A, Shiraseb F, Mirzababaei A et al. Circulating Inflammatory Markers May Mediate the Relationship between Healthy Plant-Based Diet and Metabolic Phenotype Obesity in Women: A Cross-Sectional Study. Int J Clin Pract. 2022;2022:8099382.
Im J, Kim M, Park K. Association between the Phytochemical Index and Lower Prevalence of Obesity/Abdominal Obesity in Korean Adults. Nutrients. 2020;12:2312.
Hansen TT, Astrup A, Sjödin A. Are Dietary Proteins the Key to Successful Body Weight Management? A Systematic Review and Meta-Analysis of Studies Assessing Body Weight Outcomes after Interventions with Increased Dietary Protein. Nutrients. 2021;13:3193.
Binou P, Yanni AE, Kartsioti K et al. Wheat Biscuits Enriched with Plant-Based Protein Contribute to Weight Loss and Beneficial Metabolic Effects in Subjects with Overweight/Obesity. Nutrients. 2022;14:2516.
Li J, Glenn AJ, Yang Q et al. Dietary Protein Sources, Mediating Biomarkers, and Incidence of Type 2 Diabetes: Findings From the Women's Health Initiative and the UK Biobank. Diabetes Care. 2022;45:1742-53.
Choi MK, Bae YJ. Protein intake and osteosarcopenic adiposity in Korean adults aged 50 years and older. Osteoporos Int. 2020;31:2363–72.
Orlich MJ, Fraser GE. Vegetarian diets in the Adventist Health Study 2: a review of initial published findings. Am J Clin Nutr. 2014;100 Suppl 1:353s-8s.
Slywitch E, Savalli C, Duarte AC et al. Obese Vegetarians and Omnivores Show Different Metabolic Changes: Analysis of 1340 Individuals. Nutrients. 2022;14:2204.

There is NO conflict of interest to disclose.

supplementaryTableS1.docx
Supplementary Table S1. The characteristics of the ten genetic variants of genes related to fat mass in adults

Download PDF

Version 1

posted

You are reading this latest preprint version

Plant-based and high protein diets are associated with a lower risk of obesity among Koreans at a high genetic risk of obesity.

Status:

Version 1

Abstract

Figures

Introduction

Methods

Participants and setting

Demographic, Anthropometric, And Biochemical Parameters Of The Participants

Definition Of Obesity And Metabolic Syndrome

Usual Food Intake Using A Semi-quantitative Food Frequency Questionnaire (Sqffq)

Dietary Patterns By Principal Components Analysis

Dietary inflammatory index (DII)

Genotyping Using A Korean Chip And Quality Control

Selection of the genetic variants that influence obesity defined by fat mass and the best model with SNP-SNP interactions

Expression Quantitative Trait Locus (Eqtl) Analysis

Statistical analysis

Results

Demographic characteristics and lifestyles according to genders and fat mass

Anthropometric And Biochemical Parameters

Polygenetic Variants And Their Interactions Related To The Body Fat Mass

The gene expression according to the alleles of the selected SNPs in different tissues from GTEx v8.

The Best Model Of Genetic Variants With Snp-snp Interaction For Body Fat Mass

Interaction Of Prs With Lifestyles To Influence Body Fat Mass

Discussion

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1