Genetic influences on food liking and food preference patterns in young adults: a genome-wide association study

This genome-wide association study of 2,784 young adults identifies 32 significant genetic variants influencing specific food likings and broader dietary patterns, revealing that while genetic influences operate at both individual and pattern levels, findings show limited replication in older populations.

The Gist

Imagine your taste buds are like a unique musical instrument. Some people are born with a violin that naturally sounds sweet, while others have a drum that beats with a preference for salty or spicy rhythms. For a long time, scientists have tried to figure out how much of this "musical style" is written in our DNA versus how much is just the result of what we've eaten or where we've lived.

Most previous studies looked at older adults—people who have been playing their instruments for decades. But by then, their music might have changed because they decided to eat healthier, or because their taste buds got tired. This study decided to tune in to a different group: young adults (around age 25). Think of this as catching the band right when they are forming their sound, before they've been influenced by too many outside factors.

Here is the story of what they found, broken down simply:

1. The Big Experiment: A Taste Test for 2,784 People

The researchers gathered a group of 2,784 young adults from a long-running study in the UK (called ALSPAC). They asked these people to rate how much they liked 97 different foods, from pizza and chocolate to Brussels sprouts and lentils. They used a scale from "I hate this" to "I love this."

Then, they looked at the DNA of every single participant to see if specific genetic "switches" (called variants) were linked to specific food likes.

2. The Two Main "Flavor Profiles"

Before looking at the DNA, the researchers noticed something interesting. When they mapped out the data, the foods didn't just sit randomly; they clumped together into two big "flavor tribes":

• The "Garden & Ocean" Tribe (PC1): This group includes people who love a mix of vegetables, seafood, avocados, and olives. It's like a fresh, diverse salad bowl.
• The "Meat & Fire" Tribe (PC2): This group loves bacon, burgers, steak, and sausages. It's like a hearty barbecue.

The study found that your DNA doesn't just decide if you like one specific thing (like "I love strawberries"); it often decides which entire tribe you belong to.

3. The Genetic "Remote Controls"

The study found 32 specific genetic switches that act like remote controls for our taste preferences.

• The "Broadcaster" Switches: Some switches control whole groups of foods.

  • Example: One switch (linked to lentils) seemed to turn up the volume on a whole playlist of foods: honey, yogurt, chili peppers, and even the "Garden & Ocean" tribe. If you have this switch, you likely enjoy a wide variety of plant-based and seafood foods.
  • Example: Another set of switches (linked to bacon and burgers) seemed to control the "Meat & Fire" tribe. If you have these, you probably crave a whole range of meaty, savory foods.

• The "Specific" Switches: Other switches were very picky.

  • Example: There was a switch specifically for grapefruit juice. It turned out that this gene is related to how we taste bitterness. If your DNA makes you super sensitive to bitter tastes, you probably hate grapefruit juice. This makes perfect sense because grapefruit is very bitter!

4. The "Bitter" Detective Story

The researchers also looked at a famous gene called TAS2R38, which is known to make some people taste bitter things (like Brussels sprouts) as incredibly strong and unpleasant.

• The Finding: In this young group, people with the "bitter-sensitive" version of the gene did indeed dislike Brussels sprouts more.
• The Twist: However, this gene didn't seem to affect their liking for broccoli or other veggies as much as expected. It's like having a super-sensitive nose for one specific smell, but not others.

5. The "Time Travel" Problem: Why didn't it match older studies?

The researchers compared their results with a massive study of older adults (the UK Biobank). They expected the genetic switches to match perfectly.

• The Reality: They barely matched! Only the grapefruit switch was the same in both groups.
• The Analogy: Imagine you ask a 25-year-old and a 60-year-old, "What music do you like?" The 25-year-old might say, "I love this new indie band." The 60-year-old might say, "I love classic rock." Even if they have the same musical DNA, their current preferences are shaped by their age, their life experiences, and what's popular right now.
• The Lesson: This suggests that our genetic taste preferences might change or get "masked" as we get older. The genes that drive a young person's love for a specific food might get drowned out by health concerns or habit in an older person.

The Bottom Line

This study is like taking a snapshot of our "taste DNA" before life gets too complicated. It tells us that:

1. Genes matter: Your DNA helps decide if you're a "veggie/seafood" person or a "meat" person.
2. It's not just one food: Some genes control your love for a whole category of foods, not just a single item.
3. Age changes the tune: What your genes make you like at 25 might look very different from what you like at 50.

So, the next time you find yourself loving a plate of spicy lentils while your friend can't stand them, remember: it's not just your childhood; it's your unique genetic playlist playing in the background.

Technical Summary

1. Problem Statement

While genetic factors are known to influence food preferences, existing Genome-Wide Association Studies (GWAS) have predominantly focused on middle-aged and older populations (e.g., UK Biobank, ages 37–73). In these older cohorts, dietary behaviors are often confounded by health-motivated changes, chronic disease management, and cumulative environmental exposures, potentially obscuring the "pure" hedonic and sensory drivers of food liking.

Young adulthood (specifically around age 25) represents a critical developmental window where long-term dietary habits are established, independence in food choice increases, and the prevalence of chronic disease is lower. However, the genetic architecture of food liking during this specific life stage remains largely unexplored. This study aims to fill that gap by investigating the genetic basis of food liking in young adults and determining whether genetic influences operate at the level of individual foods or broader dietary patterns.

2. Methodology

Study Cohorts and Data Sources:

• Primary Cohort (Discovery): The Avon Longitudinal Study of Parents and Children (ALSPAC). The analysis focused on 2,784 unrelated individuals of European ancestry (mean age 25.33 years, 65.25% female).
• Replication Cohort: The UK Biobank (European ancestry, ages 37–73). Summary statistics from a previously published GWAS on food liking (May-Wilson et al., 2022) were used for cross-cohort comparison.

Phenotyping:

• Food Liking: Assessed via the "Life at 25+" questionnaire in ALSPAC. Participants rated 97 distinct food items on a 9-point hedonic scale (1 = "Extremely dislike" to 9 = "Extremely like"). "Never tasted" responses were excluded from specific trait analyses.
• Principal Component Analysis (PCA): PCA was performed on the 97 food liking traits (after adjusting for age and sex) to derive two major dietary preference patterns:
  • PC1: Represented a preference for diverse plant-based and seafood foods (e.g., olives, avocado, spinach, salmon).
  • PC2: Represented a preference for meat-based foods (e.g., bacon, burgers, steak, sausages).

Genotyping and Statistical Analysis:

• Genotyping: ALSPAC participants were genotyped using Illumina platforms and imputed against the 1000 Genomes Phase 3 European reference panel.
• GWAS: Linear regression under an additive genetic model was performed for each of the 97 food traits and the two PCs using PLINK2.
  • Covariates: Age, sex, and the first 10 genetic principal components (to control for population stratification).
  • Significance Threshold: Genome-wide significance set at $p < 5 \times 10^{-8}$.
• Post-GWAS Analysis:
  • Clumping: Independent loci identified using a 500 kb window and $r^2 < 0.05$.
  • Cross-Trait Analysis: Examined whether lead Single Nucleotide Polymorphisms (SNPs) associated with one food also influenced other related foods or the derived PCs.
  • Exploratory Analysis: Investigated specific functional variants in the bitter taste receptor gene TAS2R38.
  • Replication: Significant ALSPAC hits were tested against UK Biobank summary statistics for corresponding food liking and dietary intake traits.

3. Key Contributions

• First Large-Scale GWAS in Young Adults: This is the first study to map the genetic architecture of food liking specifically in a young adult cohort (age ~25), providing a baseline less confounded by health-driven dietary modifications.
• Dual-Level Genetic Architecture: The study demonstrates that genetic influences on food liking operate at two distinct levels:
  1. Trait-Specific: Variants influencing the liking of specific individual foods (e.g., white rice, grapefruit).
  2. Pattern-Level: Variants influencing broader dietary preference patterns (e.g., plant/seafood vs. meat preferences).
• Identification of Novel Loci: The study identified 32 genome-wide significant associations across 24 traits, many of which had not been previously reported in food preference GWAS.

4. Key Results

Genome-Wide Significant Associations:

• Total Findings: 32 significant associations were identified across 24 food traits, including one association each for PC1 and PC2.
• Strongest Signal: The most significant association was for grapefruit juice liking at locus PRR4 (rs2264192, $p = 4.18 \times 10^{-11}$). This variant is in linkage disequilibrium ($r^2 = 0.81$) with a missense variant in the bitter taste receptor gene TAS2R30 (rs2599404).
• Other Notable Hits:
  • FAM20C variant associated with wholegrain cereal liking.
  • CACNA1G variant (encoding a T-type calcium channel) associated with milk chocolate liking.
  • Variants near STUM (cider), SLCO2A1 (PC1), and HSPA14 (PC2).

Cross-Trait and Pattern Analysis:

• Broad Patterns: Several variants showed pleiotropic effects across related food groups.
  • The lentil-associated SNP (rs76659918) was associated with PC1 and multiple plant/seafood items (honey, yogurt, chilli, aubergine, avocado, black olives).
  • Variants associated with bacon, burgers, and steak were linked to PC2 and other meat-based foods (roast chicken, ham, lamb, pork).
• Specificity: Some loci, such as those for white rice (rs28407649) and grapefruit, showed highly specific effects with minimal cross-trait associations.

Exploratory TAS2R38 Analysis:

• Bitter-sensitive alleles of TAS2R38 were significantly associated with lower liking for Brussels sprouts ($p < 2 \times 10^{-5}$), consistent with the known role of this gene in bitter taste perception. However, associations with other vegetables (e.g., broccoli) were not significant.

Replication in UK Biobank:

• Limited Replication: Only the grapefruit juice association (rs2264192) replicated robustly in the UK Biobank ($p = 7.73 \times 10^{-60}$).
• Interpretation: The lack of replication for other loci is attributed to age-related differences in taste perception, cumulative dietary exposure, and the fact that older adults' preferences may be influenced by health constraints rather than pure hedonic drive.

5. Significance and Implications

• Developmental Genetics: The study highlights that the genetic architecture of food preference is dynamic across the lifespan. Findings in young adults capture the biological and sensory determinants of liking before they are heavily modified by health behaviors.
• Mechanistic Insights: The identification of variants in genes related to sensory perception (TAS2R30, TAS2R38) and neuronal signaling (CACNA1G) reinforces the biological plausibility of these associations.
• Future Directions: The study suggests that future research should focus on harmonized measures across diverse age groups to disentangle age-specific genetic effects from general genetic predispositions. It also underscores the need for functional studies to elucidate the mechanisms of non-sensory genes (e.g., SLCO2A1, FAM20C) in dietary preferences.
• Public Health: Understanding the genetic basis of early dietary preferences could inform targeted interventions to promote healthier eating habits during the critical transition to adulthood.

Limitations:

• Modest sample size (N=2,784) compared to large-scale biobanks, potentially limiting power to detect small-effect variants.
• Self-reported data may be subject to recall bias or cultural interpretation of food names.
• Limited replication in the UK Biobank due to cohort age differences.
• Population homogeneity (European ancestry) limits generalizability to other ethnic groups.