Background

Food choice is a complex process that can impact various aspects of health, including our body composition. Numerous factors may influence our food choices, including the taste of food, intrapersonal determinants such as perceptions, beliefs, attitudes, and motivations, interpersonal determinants (such as significant others), social, cultural, and environmental determinants (such as food availability, the market, etc.), and economic determinants [1, 2]. Taste is one of the most crucial determinants of food choices; however, perceptions and preferences for different tastes vary widely among individuals [3]. Genetic polymorphisms in genes involved in taste perception, at least in part, can explain these interindividual variations [3, 4].

Food preferences are influenced by a multitude of environmental, cultural, nutritional, and genetic factors [5, 6]. The initial indications of the genetic impact on food preferences were observed through investigations involving families and twins [7, 8]. In recent decades, significant progress in molecular genetics has transformed the understanding of individual variations across various aspects of human behavior. These breakthroughs empower researchers with the means to conduct extensive genetic association studies, enabling a deeper exploration of the involvement of particular gene loci in sensory perceptions, food preferences, liking or disliking, as well as habits related to food intake on a larger scale [9, 10]. here is a relatively large number of studies that have investigated the association between single nucleotide polymorphisms (SNPs) in different genes [11] especially taste receptors for sweet and umami (T1R) genes [12]. However, the association between food preference and genes seems to be much more complicated, and probably much more genes are involved in this regard [13,14,15].

Food hedonic questionnaires are often used to assess food preferences. These questionnaires gauge how much a person "likes" or "wants" a particular product [16, 17]. Previous studies have reported that some of the significant food preferences include sweet and savory snacks, high-protein foods, and fatty foods. Additionally, it has been noted that food preferences differ across gender and age groups [18,19,20].

Recent advances in genetics and the development of genome-wide association studies (GWAS) have brought a unique opportunity to gain a more holistic view of the impact of genes on food preferences [21]. Therefore, this study was designed as a systematic review to evaluate the genetic aspects of food preference in human studies among adults.

Methods

Study design and search strategy

The study was conducted following the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [22]. The PRISMA checklist can be found in Supplementary Table S1. The study protocol has been registered with PROSPERO (CRD42022352920). In August 2022, searches were conducted on seven electronic databases: PubMed, Scopus, Cochrane Library, Web of Science, ClinicalTrials.gov, Embase, and OpenGrey. The searches involved a combination of key terms related to genetics and food preferences (see Supplementary Methods 1). There were no language restrictions in our search. Additionally, for the grey literature search, we assessed conference papers. If a study met the necessary criteria, we contacted the corresponding author to obtain the full text or required information. Our search also included review publications, editorials, letters to editors, conference papers, and the references of all the included studies. Ethical approval from the local institutional ethics committee was not required for this study, as we used previously published data.

Eligibility criteria

Studies were included in this analysis if they were conducted among human subjects with adult participants (> 18 years). To be eligible for inclusion, studies needed to incorporate both food preferences and genotypes. Studies that solely assessed preferences for basic tastes (using glucose or salt solutions and not food) or alcoholic drinks (without food) were excluded. Studies that assessed people's food intake or eating habits without referencing food preferences were also not included. Additionally, studies that only explored the relationship between heredity and food preferences without examining specific genes were excluded from the analysis.

Study selection

Initially, researchers conducted the search process in electronic databases. In the second stage, two researchers (MR and JH) independently performed the initial screening of the studies entered into the Endnote software. Finally, in the subsequent stage, a secondary screening was conducted by examining the full text of the articles based on the inclusion and exclusion criteria, and the final articles were selected for inclusion in this systematic review. During this stage, researchers primarily used article titles and abstracts as selection criteria. Any inconsistencies between the two researchers in the selection of studies were resolved through re-examination by each of them and consultation with a third person (RA).

Data extraction

Two researchers, MR and JH, independently extracted the necessary information, including (a) study-related variables (first author's name, publication year, sample size, study design, the presence of a control group, and its general description), evaluated genes, the method for assessing food preferences, and the main results.

Quality assessment

Using the Newcastle–Ottawa Scale (NOS) for observational studies, we assessed the risk of bias and rated the quality of the included research [23]. The scale employs a "star system" with a maximum of ten points, assigning points for factors such as study group selection, group comparability, exposure measurement, and result measurement. A study with five or more points was considered to be of high grade [24].

Statistical analysis

Due to substantial variations in the outcomes under investigation, a meta-analysis was not feasible. Instead, this study was conducted as a systematic review.

Results

Characteristics of included studies

Figure 1 displays a PRISMA flow diagram summarizing the inclusion procedure. In total, 8,510 items were identified using our systematic searching method. Of these, 2,634 were eliminated as duplicate records. The titles and abstracts of the remaining 5,876 records were screened, resulting in 203 articles being included in the next step. In the second phase of screening and after reviewing the full text of the studies, 50 studies met the necessary conditions for inclusion in this systematic review [9, 14, 21, 25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71]. Table 1 provides a summary of the features of the studies that were part of this systematic review. Participants were from different geographic regions, including: Japan [14, 25, 27, 39, 55, 64, 65, 68, 71],USA [32, 33, 42, 44, 60], Brazil [26], Finland [9, 38, 41], Czech Republic [28, 29, 50], Netherlands [30], Malaysia [31, 52], Caucasus [34], Israel [36], Turkey [37], Italy [21, 40, 43, 48, 53, 65], Australia [45, 51], UK [46, 49, 56, 57, 62, 67], India [47], Sweden [35, 54], Korea [58, 59], Hungary [61], Pakistan [63], and Spain[66]. The age range of the people examined in the studies was between 18 and 70 years. Most of the studies were conducted on both men and women. Also, most of the studies were conducted on healthy people. However, in four studies, patients with migraine [9], gestational diabetes mellitus (GDM) [50], obese [28, 29] and metabolic syndrome [66] were evaluated. In most studies, standard food questionnaires such as food frequency questionnaire (FFQ) or 24-h food recall were used to evaluate food preferences. However, in some studies, specific questionnaires of food preferences or interviews were applied. Also in term of evaluated genes, a wide variety of genes and SNPs have been evaluated in these studies.

Fig. 1
figure 1

Preferred reporting items for systematic reviews and meta-analyses flow diagram of search process

Full size image
Table 1 Characteristics of the studies
Full size table

Quality of included studies

The results of the quality assessment of studies using the NOS tool are shown in Supplementary Table 2. Among all the evaluated studies, 9 studies were of poor quality (score less than 5) [26, 29, 31, 37, 42, 51, 68,69,70] and the rest of the studies had good quality (overall score ≥ 5 points).

Association between genes and food preference to macronutrients (carbohydrate, fat, and protein)

As shown in Table 2, in some studies, researchers have evaluated the relationship between genes and the preference to consume macronutrients. Prado-Lima et al., in a cross-sectional study, evaluated the association between the serotonin receptor 5-HT2A gene and preference for micro and macronutrient intake. They found that participants with the TT genotype of the T102C polymorphism of the 5-HT2A receptor gene had higher protein intake and a higher tendency toward beef compared with CC or TC subjects [26]. In a different study, Bauer et al. looked at the relationship between a few SNPs and macronutrient intake and discovered statistically significant relationships between five of the twelve SNPs that were situated in or close to the genes SH2B1, KCTD15, MTCH2, NEGR1, and BDNF and consumption of macronutrients. The risk allele at rs7498665 (SH2B1) was linked to higher intakes of total fat (1.08 g/d energy-adjusted; 95% CI: 0.36, 1.81), saturated fat (0.60 g/d; 95% CI: 0.22, 0.97), and monounsaturated fat (0.37 g/d; 95% CI: 0.04, 0.69). For the risk alleles of the SNPs in or close to KCTD15 and NEGR1, a reduction in monounsaturated fat intake was seen, but carriers of the risk allele for NEGR1 also had reduced intakes of saturated fat. Furthermore, individuals who carry this SNP in or near KCTD15 have been found to consume less fat and more mono- and disaccharides and total carbohydrates [30]. Also, It has been reported in a study by Han et al., that G protein-coupled receptor TAS1R1 and TAS1R3 polymorphisms were associated with macronutrient intake. They found that participants with CC alleles of the TAS1R3 rs307355 and rs35744813 consumed a higher amount of protein than T carriers. Additionally, people who had the TAS1R1 SNP rs34160967's GG genotype ingested more fat and calories than those who had the A genotype [51]. Researchers looked at the relationship between angiotensinogen (AGT) gene polymorphisms and food preferences in a study of the Japanese population and discovered that individuals with the MM/MT genotype of AGT Met235Thr in comparison to those with the TT genotype consumed more total lipids, cholesterol, and unsaturated free fatty acids. Also, they didn’t find a significant correlation between AGT polymorphism (rs7079) and the ACE I/D with food preference [71].

Table 2 Overview of genetic association studies related to food preference
Full size table

However, in some studies, there was no meaningful connection between SNPs and macronutrient intake. Bienertova-Vasku et al. reported that there wasn’t any significant association between examined polymorphisms (LEP –2548 G/A, LEPR Gln223Arg, POMC RsaI and AvaI, Arg51Gln and Leu72Met in ghrelin gene, APM1 T94G) with abnormal eating patterns [28]. Also, rs1761667 G > A in the CD36 protein's genetic variant and the consumption of fat or other kinds of macronutrients or on the choice of food among non-obese males and females were not shown to be significantly correlated by Choi et al. [59]. In line with this study, Keller et al. demonstrated a relationship between the intake of extra fats and oils and the CD36 gene variant rs1761667 [33].

Association between genes and food preference for sweet, salt and fatty foods

Mennella et al. examined the relationship between genetic variation in the TAS2R38 gene and food preference in a cross-sectional study, and they discovered that genotypes at the TAS2R38 locus substantially linked with a higher liking for sucrose and sweet-tasting meals and beverages, including cereals [69]. In line with this finding, Keskitalo et al. reported a significant association between a locus on chromosome 16 and a preference for 5 sweet foods (chocolate, candy, ice cream, sweet desserts, and sweet pastry) [9]. Another study looked into the relationship between the leptin gene (LEP) and leptin receptor gene (LEPR) polymorphisms and food preferences. It found that the LEP A19G and LEPR R109K polymorphisms are connected to a desire for sweet foods [28]. Also, Kawafune et al. examined the association between the 12q24 locus and sweet taste preference in the Japanese population and found a significant correlation [14]. In a population-based study among Korean adults, the researchers evaluated the association between genetic risk scores(GRS) which contained 8 SNPs (TAS1R2_rs61761364, SLC2A5_rs11121306, SLC2A7_ rs769902, SLC2A5_rs765618,

TRPM5_rs1965606, TRPV1_rs224495, TRPV1_ rs8065080, and TRPV1_rs8078502) and sweet taste preference and they found a 1.30-folds increase in GRS was associated with higher sweet taste preference [58]. Moreover, in another study on Hungarian general and Roma populations, researchers looked at the relationships between taste and dietary preferences and the polymorphisms TAS1R3, CD36, SCNN1B, TRPV1, TAS2R38, TAS2R19, and CA6. The findings revealed a significant association between CA6 rs2274333 and a preference for raw kohlrabi and salt, CD36 rs1527483 and a preference for fat, TAS2R19 rs10772420 and a preference for grapefruit, and TAS2R38 rs713598 and a preference for the amount of sugar added [47].

In some studies, it has been reported that the adiponectin encoding gene (ADIPOQ gene) especially the rs822396 SNP, is related to confectionery intake [39]. Another study didn’t find a significant correlation between TAS1R2 and GLUT2-related SNPs with sweet liking scores [43].

Association between genes and food groups' preferences

In some studies, researchers evaluated the effects of genetic variation on food groups’ preferences. Ozawa et al. in 2002 evaluated the association between genetic variation in human leukocyte antigen (HLA) genes (DRB1, DQA1, and DQB1) and cows' milk preference and discovered a negative correlation between the prevalence of the HLA-DQA1*0102 allele and liking for cows milk [25]. In another population-based study among subjects from the Caucasus and Central Asia, in 2012, the preference for certain food items was examined by scientists who assessed the relationship between genetic variations with the TAS1R2, TAS1R2, PCLB2, TPRV1, and ITPR3. It has been shown that there are significant correlations between TAS1R2 and TAS1R3 variants and liking Vodka, white wine, and lamb meat, PCLB2 gene and preference for Hot Tea, TPRV1 gene and liking of beet, and ITPR3 gene and liking of both lamb meat and sheep cheese [34]. Another study conducted by Brunkwall et al. investigated how variations in the fat mass and obesity-associated gene (FTO) are linked to dietary preferences in individuals without any health issues. They found that A-allele carriers reported a higher intake of some energy-dense foods such as biscuits and pastries but lower consumption of soft drinks in comparison with TT allele carriers [35].

In a cross-sectional study, the investigators didn’t find any significant correlation between the bitter taste receptor gene hTAS2R38 and food choices [37]. Also, there wasn’t any significant correlation between CD36 protein SNPs (rs1527479 and rs1984112) with fats and oil, as well as dairy consumption frequencies [45].

Furthermore, there have been investigations into the association between genetic variations and the level of craving for unhealthy foods. Wallace et al. conducted a cross-sectional study to explore the link between the dopamine-related catechol-O-methyltransferase (COMT) gene and the appeal of "unhealthy" foods. The study discovered that individuals with Val/Val and Val/Met genotypes of the COMT gene showed a higher desire for objectively identified "unhealthy" food items compared to those with the Met/Met genotype [44].

The correlation between genetic variations and the inclination to consume bitter-tasting foods

Multiple studies have examined the connection between genetic variations and individuals' inclination towards consuming vegetables. Shen et al. conducted a cross-sectional study to investigate the connection between TAS2R38 and gustin (CA6) gene variations and their correlation with a preference for brassica vegetables. The study revealed that individuals possessing the TAS2R38 AVI/AVI genotype exhibited a greater preference for brassica vegetables. Also, they found that both PAV/PAV and AVI/AVI subjects consumed more total vegetables and brassica vegetables than PAV/AVI cariers [46]. Pirastu et al. in a population-based study among an Italian population evaluated the Genome-Wide Association (GWAS) with common food likings and reported that seven loci were associated with vegetables. They found that some SNPs such as rs10050951, rs8034691 and rs28849980 were associated with artichokes liking. Also, they found two loci for broccoli liking (rs2530184 located in a gene desert region on chromosome 17 and rs9832668 located on chromosome 3 close to the RYBP gene), and finally one locus on chromosome 8, very close to the CSMD1 gene (rs138369603) which was associated with chicory liking. Moreover, in terms of bitter foods, they found 3 loci, one for dark chocolate (rs73082019), one for coffee (rs145671205), and one for liver liking (rs34088951) [21]. Risso et al. evaluated 183 volunteers from four geo-linguistic groups and found a significant correlation between rs860170 (TAS2R16) and the desire to consume bitter vegetables including broccoli, mustard, and beer [48]. In another study, Perna et al. found that RS713598 SNP of the TAS2R38 gene was associated with a higher preference for beer [53]. Similar findings were reported between caveolin 1-related SNPs and liking alcoholic beverages [65]. In another study, Hayes et al. showed that SNPs in TAS2R3, TAS2R4, and TAS2R5 are significantly correlated with the desire to consume bitter coffee and alcohol intake [32]. Moreover, a significant association was shown between TAS2R gene-related SNPs and coffee liking [40].

However, the results of some studies were contradictory. Deshaware et al. in a study among Indian subjects didn’t find any significant correlation between bitter taste receptor gene TAS2R38 polymorphisms and food preference for vegetable or bitter foods [47].

Discussion

This study is the initial systematic review conducted among adults, with the aim of analyzing how genetic variation influences food preferences. Dietary behavior in people is influenced by various factors, and one of the most significant factors is genetics [72]. Food preference take form in the course of fetal development, and eating habits undergo changes as time progresses. This intricate characteristic is influenced by a combination of genetic and environmental elements. The sensory attributes of ingested food play a crucial role in shaping dietary habits, with taste recognized as a primary influencer of food choices and dietary patterns [3]. Chemical compounds present in food trigger specialized taste receptors, and these receptors can be influenced by genetic variations, resulting in individual variations in taste and preferences. The perception of bitter, sweet, and umami is associated with G-protein-coupled receptors [73, 74], while salt and sour tastes are governed by ion channels [75]. In this regard, SNPs in taste receptor genes are among the most studied polymorphisms [76]. For instance, sugar consumption in humans has been linked to sweet taste receptor (TAS1R2) alleles [77, 78]. Additionally, the consumption of vegetables, oil, and sweets has been associated with the genetic variation of the bitter taste receptor TASR38 [79].

The effect of TAS1R and TAS2R gene families on dietary behavior, specifically the preference for sweet and bitter tastes, has been investigated in the majority of studies examining the relationship between genetic variants and dietary behavior. Monosodium glutamate (MSG) is commonly used in humans to stimulate the heterodimeric G protein-coupled receptors TAS1R1 and TAS1R3 to perceive umami tastes [74, 80]. There are three proteins in the TAS1R family receptors, TAS1R1, TAS1R2, and TAS1R3, encoded by their respective genes, TAS1R1, TAS1R2, and TAS1R3. It has been reported that TAS1R2 and TAS1R3 create a heterodimer, and this heterodimer through the connections and effects on specific receptors, lead to responds to sweet tastes, including sugars, artificial sugars, d-amino acids, and some proteins, such as miraculin [81, 82]. Previous studies have shown that individuals with the GG genotype of the TAS1R1 SNP rs34160967 and the CC genotype of the TAS1R3 SNP rs307377 had lower levels of MSG threshold [83]. Furthermore, certain studies have indicated a notable association between the TAS1R2 gene and an increased preference for vodka and white wine [34, 84]. Indeed, research has proven that alcohol and sucrose activate the identical gustatory neural pathway [85]. Brasser et al. in an animal study found that in knockout mouse models for the TAS1R3 gene, the preference for alcohol and the amount of alcohol consumption significantly reduced [86]. In line with these findings, Hinrichs et al. found that TAS2R16 has an impact on alcohol liking [87].

Additional research has examined how variations in genes related to obesity and adipose tissue impact individuals' food preferences. Among these genes, the fat mass and obesity-associated gene (FTO) holds particular significance. Brunkwall et al. demonstrated that individuals with the FTO A-allele may not only display a greater appetite overall but also exhibit a preference for certain food categories. Specifically, they showed a tendency for higher consumption of biscuits, cereals, high-fat meat, and pastry, while having a lesser preference for soft drinks compared to those with the TT allele [35]. Also, it has been reported in one study among children that the A-allele carriers consume higher amounts of energy-dense foods. In a study by Hwang et al. on the UK Biobank sample, a significant association between the rs11642841 variant of FTO and total sugar intake was reported [56]. These results suggest that some variants of FTO can increase the risk of obesity by increasing the tendency to consume sweet and high-calorie foods.

The findings from studies examining the relationship between genes and the preference for fat taste and fat intake are inconsistent. It is believed that the CD36 gene, responsible for producing the fatty acid translocase, plays a role in detecting fatty acids in the mouth. This protein acts as a scavenger receptor and facilitates the transfer of long-chain fatty acids into cell membranes, which is crucial in the breakdown of fats. Considerable research focus has been directed toward the analysis of fat taste perception in relation to two SNPs, namely rs1761667 and rs1527483, situated within the CD36 gene [88]. The CD36 gene encodes the fatty acid translocase, a crucial element in the initial step of fat metabolism, responsible for transporting long-chain fatty acids (LCFA) across cell membranes. Some studies indicated that individuals with the AA genotype of rs1761667 in the CD36 gene had higher thresholds for perceiving lipid taste compared to those with the GG genotype [33, 89]. In a particular study conducted on the UK population, no notable association was observed between TAS2R38 diplotypes or CD36 rs1761667 and the consumption of dietary fat. However, researchers did discover a meaningful relationship between the sensitivity to bitter taste and the intake of saturated fatty acids [62]. Another genetic variant that has been associated with the tendency to consume oil and fatty foods is rs6661761 located within the BPNT1. Although precise data about BPNT1 gene function is not available, It is highly expressed throughout the brain and is severely suppressed by lithium, a medication that is frequently used to treat bipolar disorder. In rats conditioned to anhedonic responses, lithium has been demonstrated to recover hedonic responses to appetizing stimuli through the nucleus accumbens [90]. Some researchers suggest that liking oil and fatty foods might be linked to the reward of palatable foods through the nucleus accumbens [21]. There is a linear negative association between the function of this nucleus and obesity risk. So, BPNT1 is known as a good indicator for understanding the physiology underlying the liking of palatable foods and the activation of the reward system [91].

Based on our knowledge, this study is the first systematic review that evaluates the association between genetic variants and food preference. Our study had several strengths, including conducting a grey literature search, no language restrictions in the screening of studies, and also examining a wide range of genetic variants. However, this study had limitations that should be considered in the interpretation of the results. Although the aim of this study was to perform a meta-analysis, due to the high heterogeneity between the examined variants and the impossibility of pooling the data, this was not possible, and the study was written in a systematic review form. On the other hand, the tools used in the primary studies to evaluate dietary tendencies were diverse (some studies used self-filled questionnaires, some online questionnaires, interviews, and other methods), which affects the accuracy of the results.

Conclusion

In conclusion, this systematic review study showed that there was a considerable association between some of the genetic variants with food preferences among adults. Some of these genetic variants increase or decrease the desire for sweet and fatty foods, and some affect the choice of food groups. However, due to the high heterogeneity in the investigated variants, more studies are needed to investigate these genetic variants more closely and to identify the mechanisms involved in the observed effects.