Impact of Genes and Environment on Obesity and Cardiovascular Disease

Abstract

Obesity and abdominal obesity have been closely related to cardiovascular outcomes, and recent evidence has indicated that environmental and genetic factors act in concert in determining the risks of these conditions. Improving adherence to healthy lifestyle habits and healthy dietary patterns can at least partly counteract genetic variations related to risks of obesity and cardiovascular disease (CVD). Other factors, such as epigenetic alterations, may also modulate a relationship between genetic susceptibility and these disorders. In this review, we highlight data from recent studies on gene and environmental risk factors for obesity and CVD, and describe how these findings might inform understanding of the complex roles of interactions between genes and environmental factors in the development of obesity and CVD.

The prevalence of obesity has been rising dramatically worldwide, resulting in a variety of hazardous health problems such as cardiovascular disease (CVD) and premature death (1). Abdominal obesity, which is usually assessed by waist circumference or waist-to-hip ratio (WHR), is associated with increased risks of cardiovascular events and mortality, independent of general obesity measured by body mass index (BMI) (2–5). Taking advantage of the rapid advances in genomic techniques, genomewide association studies (GWASs) have contributed to revealing the genetic architecture of “common” types of general obesity and abdominal obesity, as well as body fatness. Common genetic variants at more than 500 loci for adiposity have been identified to date (6), and a large-scale analysis by the Genetic Investigation of Anthropometric Traits (GIANT) consortium has identified low-frequency and rare variants to determine adiposity and energy metabolism (7). Similar collaborative efforts have led to the identification of genetic variants associated with risks of coronary artery disease (CAD) and stroke (8–15). These findings enable assessment of overall genetic susceptibility by polygenetic risk scores (GRSs) that are sums of all the risk alleles related to specific obesity phenotypes, in magnitude comparable to the monogenic mutations (16).

On the other hand, genetic variations may explain a small proportion of variations in certain phenotypes, and it has been suggested that the effect of gene-environment interaction, in which the genetic variants trigger the occurrence of diseases in high-risk environments (17), may explain part of the missing variance. The epidemic of obesity has coincided with a profound shift in our living environment, such as unhealthy dietary patterns, a sedentary lifestyle, physical inactivity, and poor sleeping habits (18). In addition to these modifiable risk factors, environmental agents, such as endocrine-disrupting chemicals (EDCs), can also alter body weight, adipose tissue expansion, circulating lipid profiles, and adipogenesis (19). The obesogenic EDCs include nonsteroidal estrogens, parabens, phthalates, polychlorinated biphenyls, organotin, and bisphenols (19). Notably, exposure to EDCs in early life may be related to an increased risk of obesity-related disorders later in life (20, 21).

A study has shown that the magnitude of a relationship between obesity and the genetic risk may be stronger in more recent birth cohorts than in earlier years of birth cohorts, suggesting that the genetic predisposition to obesity may have a more influential effect in more recent obesogenic environments (22). Results of a recent GWAS also suggest that environmental risk factors such as smoking and physical activity may alter the genetic susceptibility to overall adiposity and body fat distribution (23, 24). We discuss findings on genetic and environmental risk factors for obesity and CVD that would contribute to understanding the complex architecture of interactions between genes and environmental factors in the development of obesity and CVD.

Genetics of Obesity and CVD: Overview

The number of susceptibility loci of obesity has grown dramatically; ≈200 loci for BMI (25, 26), 50 loci for WHR (27), and >10 loci for body fat percentage (28) have been identified and replicated in European populations. In the GIANT consortium, a total of 97 loci for BMI have been identified, and 77 loci reached genomewide significance among individuals of European descent (25). Cross-sectional data of the GWAS suggest that combining loci for BMI explains only a small proportion (∼3%) of the phenotypic differences (25, 26). In a longitudinal analysis of women in the Nurses’ Health Study, each 10-allele increment in GRS of BMI (based on the 97 loci) was associated with an average BMI gain of 0.54 kg/m² during early adulthood (29). In a recent GWAS of BMI in Japanese persons (n = 173,430), 51 loci for BMI were identified, including five loci on the X chromosome and two loci with sex-dependent effects (26). Combining the results with GWASs of Europeans resulted in the identification of 61 new loci, bringing the total number of known susceptibility loci to >200 (26). The investigation of diverse populations contributed to identifying novel loci and narrowing the candidate genomic region in their study (26).

For abdominal obesity, which is defined by WHR after adjustment for BMI, a total of 49 loci have been identified in a GWAS by the GIANT consortium (27). European ancestry (n = 210,088) sex-combined analysis identified 39 of the 49 loci, and European ancestry sex-specific analysis identified nine additional loci, eight of which were new and important only in women (27). A higher GRS for abdominal obesity, which was generated on the basis of these findings, has been related to increased risks of various cardiometabolic diseases, such as type 2 diabetes and coronary heart disease (CHD) (30, 31). Another meta-analysis of 114 studies (n = >320,000 individuals) investigated genetic loci with age- or sex-dependent effects on body size and shape (32). For BMI, 15 loci showed important age-specific effects, of which 11 had more substantial effects in younger (<50 years) than in older adults (≥50 years) (32). For WHR, 44 loci showed sex-specific effects, of which 28 had larger effects in women than in men, five had more substantial effects in men than in women, and 11 had opposite effects between women and men. On the other hand, no age-dependent effects were identified for WHR (32).

In a GWAS of 100,716 individuals for body fat percentage (28), a total of 12 loci have been identified; eight of these loci were previously associated with overall adiposity, and four loci (in or near COBLL1/GRB14, IGF2BP1, PLA2G6, and CRTC1) were novel (28). When they compared effects of the 12 loci on body fat percentage and BMI, seven loci (TOMM40/APOE, IRS1, SPRY2, COBLL1/GRB14, IGF2BP1, PLA2G6, and CRCT1) showed a more substantial effect on body fat percentage than on BMI, suggesting that these variants may be primarily associated with body fat. The remaining five loci (FTO, TMEM18, MC4R, SEC16B, and TUFM/SH2B1) showed stronger associations with BMI than with body fat percentage, suggesting potential effects on both fat and lean mass (28).

A recent meta-analysis of exome-targeted genotyping data (n = 718,734) showed associations of rare and low-frequency coding variants with BMI (7). In the study, researchers identified 14 coding variants in 13 genes, of which eight variants have not previously been implicated in human obesity. Pathway analyses showed a critical role for neuronal processes and implicated adipocyte and energy expenditure biology. The study also reported associations for two variants in GIPR; two variants in MC4R and KSR2 loci previously were observed to be mutated in extreme obesity. Although each variant contributes little to the BMI variation in general populations, the effect sizes of these rare variants may be ∼10 times larger than those of common variants at an individual level (7). The most considerable effect was observed in carriers of an MC4R mutation introducing a stop codon; the carriers of the mutation weighed on average 7 kg more than noncarriers. (7).

Large collaborative studies, including the Myocardial Infarction Genetics Consortium (33), Coronary Artery Disease (C4D) Genetics Consortium (34), the Coronary Artery Disease Genomewide Replication and Meta-Analysis (CARDIoGRAM) consortium (15), the CARDIoGRAMplusC4D Consortium (14), and other studies, have successfully identified genomewide significant loci associated with CAD. A GWAS of 60,801 CAD cases and 123,504 control subjects showed that among 48 loci previously reported at genomewide levels of significance, a total of 47 loci showed nominally significant associations with CAD, and 36 loci were at genomewide significance level. In this GWAS, 10 CAD loci were identified; three lower-frequency and moderately well-imputed single nucleotide polymorphisms (SNPs) in LPA and APOE showed strong associations (13). Also, key SNPs in the APOE and PCSK9 genes have been strongly associated with CAD, and evidence suggests these variants may mediate their effects on CAD via low-density lipoprotein cholesterol (LDL-C)–linked pathways (13). In the Myocardial Infarction Genetics and CARDIoGRAM Exome Consortia, investigators identified six new loci associated with CAD at genomewide significance level: on 2q37 (KCNJ13-GIGYF2), 6p21 (C2), 11p15 (MRVI1-CTR9), 12q13 (LRP1), 12q24 (SCARB1), and 16q13 (CETP) (9). In a study investigating the relationship between protein truncating variants in CETP and CHD, carriers of the protein truncating variants displayed higher levels of high-density lipoprotein cholesterol (HDL-C), and lower levels of LDL-C and triglycerides, as well as lower risk for CHD, as compared with noncarriers (35). In another GWAS, a total of 15 new genomic regions associated with CAD were identified, and nine genomewide significant loci were found only in Europeans (10). In that study (10), of the 58 previously reported CAD loci, 34 showed directionally concordant associations with at least nominal significance (P < 0.05) in either European- or all-ancestry cohorts. In a GWAS among participants of the UK Biobank, a total of 15 novel loci for CAD were identified, including CCDC92, which likely affects CAD through insulin resistance pathways (11). To date, a total of 95 CAD-associated regions have been identified. A more recent GWAS included 34,541 CAD cases and 261,984 control subjects from the UK Biobank as the discovery set, a total of 88,192 cases and 162,544 controls from the CARDIoGRAMplusC4D as the replication set. In this study, 64 new loci were identified (8). Although common variants that have a strong effect for CAD have been reported [e.g., those in the CDKN2A-CDKN2B (9p21): odds ratio, 1.3 per risk allele; (15) and LPA loci: odds ratio, 1.5 per risk allele (36)], a larger sample size of subjects and a meta-analytic approach to data derived from large studies have led to the identification of additional loci with a moderate to low effect for the outcome. Some studies (35, 37–39), considering potential physiological pathways behind diseases, further investigated candidate loci. The identified CAD-susceptible loci also have shared genetic associations with other cardiovascular traits (13), but the underlying biological mechanisms remain mostly unknown.

Aortic valve stenosis is common valvular heart disease. In a GWAS of 2,457 Icelandic aortic valve stenosis cases and 349,342 control subjects with a follow-up analysis of 4,850 cases and 451,731 control subjects of European ancestry, two new loci on chromosome 1p21 near PALMD (rs7543130) and chromosome 2q22 in TEX41 (rs1830321) were identified (40). For stroke, in a recent multiancestry GWAS of 521,612 individuals, 32 genomewide significant loci for stroke were identified, 22 of which were new. Genomewide significance level was observed for 18 loci (12 novel loci) for any stroke, 20 (12 novel loci) for any ischemic stroke, 6 loci (3 novel loci) for large-artery atherosclerotic stroke, 4 loci (2 novel loci) for cardioembolic stroke, and 2 for small-vessel stroke (41).

On the basis of these findings, GRSs have been created by combining multiple SNPs associated with CVD (42). Accumulating evidence suggests the GRSs for common diseases, such as CAD, atrial fibrillation, type 2 diabetes, inflammatory bowel disease, and breast cancer, may be useful to identify individuals with a risk equivalent to monogenic mutations (16). In a study of men with hyperlipidemia enrolled in a randomized controlled trial for the primary prevention of CHD (43), statin therapy conferred greater relative benefit among those at high genetic risk (assessed by a 57-SNP GRS for CHD) compared with other participants, suggesting the use of the GRS may be useful for identifying individuals at higher risk for coronary events (43). Nonetheless, more research would be needed to validate these findings and to assess the cost-effectiveness. Also, needed would be consideration of the potential interplay between genetic susceptibility and lifestyle, socioeconomic status, and other factors.

Interplay Between Genetics and the Environment

Epidemiological studies have identified a variety of modifiable risk factors for obesity and CVD, such as sugar-sweetened beverages, fried foods, poor diet quality, sedentary lifestyle, and physical inactivity (18). Accumulating evidence has suggested that such environmental risk factors may modify the genetic risk of obesity and related diseases (Table 1). Recent GWASs have also underscored the importance of taking into account gene-environment interactions in the discovery of new genetic variants (23, 24). Several reviews have summarized major findings of studies on gene-environment interactions and discussed the clinical implications (62–65). Here, we highlight data from more recent studies examining dietary, lifestyle, and other risk factors that may modify the genetic risks of adiposity and CVD.

Table 1.

Diet and Lifestyle Factors That May Modify the Genetic Risk of Obesity and Cardiovascular Disease

Factor	Genetic Risk	Main Outcome	Study
Sugar-sweetened beverages
	GRS of 32 SNPs for BMI	BMI	Qi et al. (44)
	GRS of 50 SNPs for BMI, WC, WHR	Change in body weight, WC, and BMI-adjusted WC	Olsen et al. (45)
	GRS of 30 SNPs for BMI	BMI	Brunkwell et al. (46)
	FTO genotype (rs9939609)	BMI	Livingstone et al. (47)
	SNPs at the chromosome 9p21 locus: rs10757274, rs4977574, rs2383206, and rs1333049	Nonfatal myocardial infarction	Zheng et al. (48)
Fried foods	GRS of 32 SNPs for BMI	BMI	Qi et al. (49)
Dietary saturated fatty acids	GRS of 63 obesity-associated SNPs	BMI	Casas-Agustench et al. (50)
No/low consumption of wine	SNPs rs4977574 at the 9p21 locus	CVD	Hindy et al. (51)
Vegetable	SNPs rs4977574 at the 9p21 locus	CVD	Hindy et al. (51)
Prudent diet (high in raw vegetables and fruits)	SNPs rs2383206 at 9p21 locus	Myocardial infarction/CVD	Do et al. (52)
Diet score	GRS of 32 SNP for BMI; GRS of 14 SNPs for WHR	BMI and BMI-adjusted WHR	Nettleton et al. (53)
Diet-quality changes	GRS of 97 SNPs or 77 SNPs for BMI	Changes in BMI and changes in body weight	Wang et al. (54)
Physical activity
	GRS of 32 SNP for BMI	BMI	Qi et al. (55)
	GRS of 12 SNPs for obesity; FTO rs1121980	BMI	Ahmed et al. (56)
	GRS of 69 SNPs for BMI	BMI	Tyrrell et al. (57)
	FTO rs9941349	BMI	Graff et al. (24)
Physical activity (e.g., stair climbing, frequencies of light, moderate, or vigorous exercise)	GRS of 94 SNPs for BMI	BMI	Rask-Andersen et al. (58)
Waking pace	GRS of 94 SNPs for BM.	BMI	Rask-Andersen et al. (58)
Grip strength, cardiorespiratory fitness	GRS of ≈60 SNPs for CHD	CHD	Tikkanen et al. (59)
Sedentary lifestyle, such as television watching
	GRS of 32 SNPs for BMI	BMI	Qi et al. (55)
	GRS of 94 SNPs for BMI	BMI	Rask-Andersen et al. (58)
	GRS of 69 SNPs for BMI	BMI	Tyrrell et al. (57)
Physical activity changes	GRS of 77 SNPs for BMI; GRS of 12 SNPs for body fat percentage	Changes in BMI and changes in body weight	Wang et al. (60)
Healthy lifestyle	GRS of 50 SNPs for CAD	CHD	Khera et al. (61)

Abbreviation: WC, waist circumference.

Diet quality

The obesity epidemic has coincided with overnutrition and unhealthy dietary patterns during the past decades. Although continuing efforts have been made to improve diet quality, the overall dietary quality remains poor in the US population (66). In addition to reducing intake of unhealthy foods such as sugar-sweetened beverages and fried foods that are a major source of calorie intake, the adherence to healthier dietary patterns and better diet quality have been recommended to prevent obesity and CVD (67). For example, studies have shown that the Alternate Healthy Eating Index (AHEI), which is based on foods and nutrients (68), is predictive of risks of major chronic diseases such as obesity, CHD, and type 2 diabetes (69–71), and is also predictive of healthy aging (72). The AHEI includes assessments of vegetables, fruits, nuts and legumes, whole grains, red or processed meat, sugar-sweetened beverages, alcohol, sodium, trans fat, long-chain ω-3 fatty acids, and other polyunsaturated fats. As compared with a single assessment of individual foods, diet scores consider combinations and quantities of different foods, which may well capture synergistic and cumulative effects of foods. Among participants of the Nurses' Health Study and the Health Professionals Follow-Up Study, changes in diet quality as assessed by the AHEI, Alternate Mediterranean Diet, and the Dietary Approaches to Stop Hypertension diet were consistently associated with risks of mortality and cardiovascular events (69, 70). Plant-based diets emphasize the consumption of healthy plant foods and discourage most or all animal products; vegetarian diets typically have beneficial nutrients for cardiovascular health, such as a high amount of dietary fiber, plant-based bioactives, and less saturated fatty acids (73). Vegetarian diets and plant-based diets are also related to lower risks of CHD (74, 75) and metabolic risk factors for CVD (76–78). In a nested case-control study of men and women, better adherence to the Mediterranean dietary pattern was associated with lower risk of obesity among individuals with risk alleles of the FTO gene, as compared with subjects with lower adherence to the dietary pattern and lower genetic susceptibility to obesity (79). In a study of data from 18 cohorts of European ancestry (53), associations of GRSs based on 32 BMI- and 14 WHR-associated SNPs with a composite diet score representing healthy diet (which was calculated on the basis of self-reported intakes of whole grains, fish, fruits, vegetables, nuts/seeds and red/processed meats, sweets, sugar-sweetened beverages, and fried potatoes) have been investigated. The diet score did not modify the association of the BMI-GRS with BMI, whereas there was nominal evidence that a higher diet score (representing a healthier diet) strengthened the association of WHR-GRS with BMI-adjusted WHR (53). In a recent study of participants in the Nurses' Health Study and the Health Professionals Follow-Up Study, there were consistent interactions between changes in diet quality scores and GRS-BMI for long-term changes in BMI and body weight (54). The study findings suggested that improving adherence to healthy dietary patterns assessed by the AHEI-2010 and Dietary Approaches to Stop Hypertension score attenuated the genetic association with long-term weight gain (54). Intriguingly, the beneficial effect of improved diet quality on weight regulation was more pronounced among people at high genetic risk for obesity (54). Such a combined effect of beneficial bioactivities of healthy dietary patterns may partly explain the modifying effect on genetic predisposition to weight gain. In addition to a large sample size and reliable measurements of dietary intake and outcomes, research on interactions between genetic variations and dietary intake may require a long-term follow-up to capture potential effects of exposures for the outcomes. In earlier studies, it was noted that the risk of CVD conferred by chromosome 9p21 SNPs were attenuated by intakes of healthful plant-based foods, such as vegetables and fruits (51, 52) whereas higher intake of sugar-sweetened beverages exacerbated the effects of chromosome 9p21 variants on CAD (48). However, these studies only analyzed a single food or did not consider a number of loci that have been identified in the recent GWASs.

Stimulators of food intake

Previous results in GWASs also highlight the potential of genetic variation for determining preference of macronutrient intake. For example, MC4R deficiency represents common monogenic obesity disorder (80), and in a study, a common variant (rs17782313) near the MC4R gene was significantly associated with higher intakes of total energy and dietary fat (81). Also, the MC4R genotype was related to greater long-term weight changes and increased risk of diabetes in women (81). A GWAS has identified that the genetic variants near the FGF21 and FTO loci were among the top associations with intakes of dietary protein and carbohydrate (82, 83). Among overweight and obese adults who participated in a 2-year weight-loss dietary intervention trial [the Preventing Obesity Using Novel Dietary Strategies (POUNDS) Lost Trial], the FTO genotype was found to interact with dietary protein in relation to changes in appetite control (84). Another study in the POUNDS Lost trial also found that dietary protein intake modified the relation between the MC4R genotype and measures of appetite during the diet intervention (85). The GWAS consistently showed that the genetic variation in the FGF21 region was associated with carbohydrate or fat intake (82, 83). Blood levels of FGF21 increase in response to carbohydrate intake (86, 87), and several recent studies provided possible mechanisms for the role of FGF21 in regulating sweet preference (88–90). FGF21 stimulates insulin sensitivity, energy expenditure, weight loss, and associated with cardiometabolic abnormalities (91, 92). A genetic variant in the FGF21 region was significantly associated with improvements in central adiposity and body fat composition among overweight and obese people in the POUNDS Lost trial, and the associations were modified by carbohydrate and fat intakes (93). A study by Solon-Biet et al. (94) introduced the Geometric Framework, a nutritional modeling platform, and 25 different types of diets, and showed that maximal FGF21 elevation occurred with low-protein, high-carbohydrate intake, suggesting that metabolic effects of FGF21 depend on macronutrient context. The Geometric Framework can integrate key aspects of nutritional systems (e.g., nutrients, foods, diets, appetites, and nutritional homeostatic physiology) and map the relationship between nutrient intakes and health outcomes (95). This method has been used in observational and experimental studies in humans to understand dietary determinants of chronic diseases associated with obesity (96, 97).

Physical activity and sedentary lifestyle

One of the most reproducible findings in gene-environment interactions for obesity is that physical activity and sedentary lifestyle may modify the genetic risk of obesity (55–57, 98, 99). In particular, a relationship between the FTO genotype and obesity is attenuated among people with high levels of physical activity (98). Results of a large-scale study among >200,000 adults confirmed the interactions between genetic variants and physical activity on adiposity; the BMI-increasing effect was attenuated by up to 30% in physically active individuals compared with inactive individuals (24). A sedentary lifestyle, as assessed by prolonged hours of watching television, also strengthens the genetic effect of obesity (55). The findings have been replicated in different studies (57, 58); in the Hispanic Community Health Study/Study of Latinos, finding suggested individuals with lower levels of physical activity and more sedentary behavior may be more susceptible to genetic effects on adiposity, supporting the roles of gene–physical activity and gene–sedentary behavior interactions in the development of obesity (100). In a study of >362,000 participants from the UK Biobank, there were also significant interactions between the GRS based on 94 SNPs for BMI and several factors related to physical activity, such as usual walking pace, stair climbing, and television watching, as well as frequencies of light, moderate, and vigorous exercise (58). In particular, the association of the GRS with BMI was 2.5 times higher among participants who reported having a slow walking pace compared with participants who reported having a brisk walking pace (58). Wang et al. (60) examined whether changes in physical activity over time modified the genetic susceptibility to long-term weight gain. The study examined associations of GRS for BMI (n = 77 SNPs) and GRS for body fat percentage (n = 12 SNPs) with long-term changes in body weight and BMI over 20 years of follow-up among US men and women (60). Their study showed that changes in physical activity significantly interacted with the body-fat GRS for changes in BMI, showing that higher physical activity levels attenuated the genetic associations of body fat percentage (60). In a study of the UK Biobank cohort, higher levels of grip strength, physical activity, and cardiorespiratory fitness were associated with lower risks of cardiovascular events (59). Also, it was found that the association of grip strength with CHD risk was strongest among those with the lowest GRS of CHD (P for interaction = 0.0002). A similar interaction pattern was observed between GRS and cardiorespiratory fitness for CHD risk (P for interaction = 0.03), although physical activity as assessed by the International Physical Activity Questionnaire did not show significant differences for CHD across GRS (P for interaction = 0.52) (59). Taken together, evidence has consistently demonstrated that physical activity and sedentary behavior may modify the genetic risk of obesity, weight gain, and cardiovascular outcomes.

Healthy lifestyle patterns

Although individual lifestyle factors are related to the incidence of CVD, adherence to a healthy lifestyle pattern, which is a combination of several behaviors, has been related to the burden of obesity and CVD (101, 102). It has been estimated that ∼73% of CHD incident cases were attributable to poor adherence to healthy lifestyle behaviors (a combination of not smoking, normal BMI, physical activity ≥2.5 h/wk, television viewing ≤7 h/week, higher adherence to healthy diet, and moderate alcohol intake) (101).

A study examined whether the genetic risk (based on 50 SNPs for CAD) was modified by the adherence of healthy lifestyle behaviors in relation to CHD incidence (61). Among participants at high genetic risk, a favorable lifestyle was associated with a 46% lower risk of coronary events than was an unfavorable lifestyle (61). However, there was no evidence of a significant interaction between the genetic risk and lifestyle factors in the study (61). In the UK Biobank cohort (103), genetic predisposition and combined health behaviors had an additive effect for the incidence of CVD, and there was no significant interaction between the genetic risk and health behaviors for the CVD risk. In another study of the UK Biobank cohort, healthy lifestyle score (which was calculated by BMI, healthy diet, sedentary lifestyle, alcohol consumption, smoking, and urinary sodium excretion levels) was associated with CVD events regardless of blood pressure–associated GRS (104). In comparison with an unfavorable lifestyle, there was a ∼30% lower risk of CVD among participants in low, middle, and high genetic risk groups, without significant interactions between the blood pressure–associated GRS and the healthy lifestyles (104). This study did not identify statistically significant gene-healthy lifestyle interactions in the development of CVD. In results of the Look AHEAD (Action for Health in Diabetes) Study among overweight/obese individuals with type 2 diabetes, a GRS comprising 153 SNPs significantly predicted cardiovascular outcomes, and the intensive lifestyle intervention, which included weight loss and physical activity, did not modify the genetic association (105). This evidence is limited to assessing whether people at higher genetic risk can benefit more from reducing CVD risk by adherence to healthier lifestyles as compared with those at lower genetic risk. Also, whether adherence to healthier lifestyles may modify the genetic risk of complications in patients with type 2 diabetes would be an important clinical issue.

Vitamin D

Epidemiological and mechanistic studies have provided strong support for a potentially protective effect of vitamin D on metabolic abnormalities (106). An inverse relationship between serum 25-hydroxyvitamin D levels and BMI has been reported in adults, except for women living in developing countries (107). In the Nurses’ Health Study, higher 25-hydroxyvitamin D levels were associated with a lower CHD risk (108). Previous GWASs of serum 25-hydroxyvitamin D concentrations identified four genomewide significant loci (GC, NADSYN1/DHCR7, CYP2R1, and CYP24A1) (109, 110). A recent GWAS of European populations expanded the sample size and identified two additional loci harboring genomewide significant variants at rs8018720 in SEC23A, and rs10745742 in AMDHD1 (111). Data from a Mendelian randomization analysis indicated that the relation between 25-hydroxyvitamin D and risk of type 2 diabetes was unlikely to be causal (112). However, findings of another, more recent Mendelian randomization study involving approximately 82,500 adults living in China support the causality between vitamin D levels and risk of type 2 diabetes (113). In the combined analysis of >58,000 cases and 370,000 control subjects in European and Chinese adults, genetically instrumented increase in vitamin D status was related to a 14% lower risk of diabetes (113). Mendelian randomization studies and randomized clinical trials have not shown significant effects of vitamin D on cardiovascular events, but these trials were not designed to investigate cardiovascular outcomes in vitamin D–deficient individuals (114).

Lipids and lipoprotein

In a GWAS of 188,577 individuals of European ancestry in the Global Lipids Genetics Consortium (115), a total of 157 loci (including 62 novel loci) associated with lipid levels have been identified. These loci for lipids were also associated with other cardiometabolic traits such as CAD, type 2 diabetes, and adiposity measures (115). Activation of LPL was associated with lower levels of triglycerides and reduced risks for CAD and type 2 diabetes without increasing liver fat (116). In a population-based prospective cohort from Northern Sweden [the Gene-Lifestyle interactions and Complex Traits Involved in Elevated Disease Risk (GLACIER) Study], seven novel genomic regions were associated with long-term (10-year) changes in blood lipids, of which three also increased the CAD risk (117). In the study, chr19:50121999 at APOE was associated with changes in triglycerides and multiple SNPs in the APOA1/A4/C3/A5 region at genomewide significant level (117). SNP rs7412 at APOE was associated with changes in total cholesterol levels in the GLACIER study (117). In an updated study from the Global Lipids Genetics Consortium and Stroke Genetics Network, a causal role of LDL-C, HDL-C, and triglycerides in the development of ischemic stroke were examined through the Mendelian randomization approach (118). The study used 185 genomewide lipids-associated SNPs as instrumental variables (76 SNPs for LDL-C, 86 SNPs for HDL-C, and 51 SNPs for triglycerides) and showed a causal relationship between LDL-C and ischemic stroke. A 1-SD genetically elevated LDL-C level was associated with a 12% increased risk of ischemic stroke and a 28% increased risk of large-artery atherosclerosis stroke. The study also showed an inverse relationship between HDL-C and small-artery occlusion stroke (118). In a genomewide association meta-analysis of survival of 606,059 parents, HDL-C levels were one of the most positively genetically correlated factors with lifespan (119). These studies have shown that genetic variants determining lipid metabolism may contribute the cardiovascular outcomes by moderating lipid traits, but potential interplays among diet and lifestyles and genetic risk in relation to CVD need to be investigated further.

Epigenetics

Epigenetics can be defined as heritable changes that affect gene expression through mechanisms not associated with alterations in the DNA sequence. The most-studied epigenetic modification is methylation of cytosine in CpG dinucleotides in DNA. The alterations in CpG sites may be predominantly the consequence, rather than the cause, of obesity (120). Epigenome-wide association studies have identified methylation at numerous CpG sites associated with adiposity (120–124). In the first epigenome-wide analysis of methylation at CpG sites in relation to BMI among European populations (121), elevated BMI was associated with increased methylation at the HIF3A locus in blood cells and adipose tissue (121). A study of African Americans confirmed previously identified methylation loci for obesity and related traits (CPT1A, ABCG1, and HIF3A), and also identified numerous additional novel loci related to both DNA methylation in blood and adipose tissue and adiposity traits (123). According to results of the Nurses' Health Study and the Health Professionals Follow-Up Study, the DNA methylation variant in the HIF3A was significantly related to BMI changes, showing interactions between dietary intake of total or supplemental vitamin B2, vitamin B12, and folate (125). In two dietary intervention trials, dietary fat significantly modified the genetic effect of HNF1A on weight loss and reduction in waist circumference (126). In an epigenome-wide association study, CpG site, cg11024682 (intronic to sterol regulatory element binding transcription factor 1, SREBF1), was associated with BMI (124). Interestingly, DNA methylation–associated genetic variant rs752579 in the SREBF1 gene significantly interacted with dietary intake of food-source B vitamins on 6-year change in BMI among participants of the Women's Health Initiative Memory Study (127). These results suggest that habitual intake of food-source B vitamins may modify the effect of DNA methylation–related genetic variant on long-term changes in adiposity (127). The latest epigenome-wide association study identified changes in DNA methylation in 187 genetic loci, and the disturbances in DNA methylation (assessed by a methylation risk score) were also predictive of future type 2 diabetes (120). In the epigenome-wide association study, the NFATC2IP genetic polymorphism (rs11150675), cis-DNA methylation at cg26663590 CpG sites was identified as causally related to BMI (120). Among overweight and obese adults who participated in a 2-year weight-loss dietary intervention trial, there were significant interactions between intake of dietary fat and the genetic and transcriptional (ILMN_1725441) variations at the NFATC2IP locus for 2-year weight change (128). The genetic variants related to DNA methylation may contribute to regulation of obesity, and the associations may be modified by different dietary intake. However, the details of this process and physiological mechanisms remain to be clarified. Another epigenome-wide analysis (129) examined the association of DNA methylation with metabolic traits in humans, using adipose tissue samples from the Metabolic Syndrome in Men cohort. The study examined a total of 32 clinical traits related to diabetes and obesity, and identified 18 candidate genes, including known and novel genes associated with diabetes and obesity traits. This study suggests that profiling DNA methylation in adipose tissue may become a powerful tool for understanding the molecular effects of metabolic syndrome on adipose tissue (129).

In conclusion, unhealthy dietary habits, physical inactivity, and sedentary lifestyle are major modifiable risk factors related to obesity and CVD, and more studies are investigating gene-environment interactions associated with these diseases. Improving diet quality, increasing physical activity levels, and reducing sedentary time may reduce risks of weight gain and adiposity, even among participants with a high genetic risk of the disease. On the other hand, most of the previous studies did not identify gene-environmental interactions relative to the incidence of CHD and stroke at statistically significant levels. More sufficient follow-up, larger sample size, accurate assessments of environmental risk factors, and valid statistical methods for testing interaction are critical in studying the gene-environmental interactions in the development of obesity and CVD. Nonetheless, many challenges exist to clarify the underlying mechanisms of and the interplay between genetic and environmental factors on these disorders, and to translate the scientific findings into prevention and treatment practices.

Glossary

Abbreviations:

AHEI

Alternate Healthy Eating Index

BMI

body mass index

CAD

coronary artery disease

CHD

coronary heart disease

CVD

cardiovascular disease

EDC

endocrine-disrupting chemical

GRS

genetic susceptibility by polygenetic risk score

GWAS

genomewide association study

HDL-C

high-density lipoprotein cholesterol

LDL-C

low-density lipoprotein cholesterol

POUNDS

Preventing Obesity Using Novel Dietary Strategies

SNP

single nucleotide polymorphism

WHR

waist-to-hip ratio