🧬 biology

Genetic architecture of eating disorder risk and specific symptoms in the general adult population

This study utilizes genome-wide association analyses of 212,000 individuals to demonstrate that eating disorder symptoms in the general population are driven by distinct genetic variants linked to appetite regulation and cognitive-behavioral traits, suggesting that a symptom-based approach offers valuable insights for improving intervention and prevention strategies beyond traditional diagnosis-based methods.

Original authors: Katri Pärna, Tuuli Sedman, Hanna Kariis, Uku Vainik, René Mõttus, Triin Laisk, Reedik Mägi, Kadri Kõiv, Kelli Lehto

Published 2026-07-09

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Katri Pärna, Tuuli Sedman, Hanna Kariis, Uku Vainik, René Mõttus, Triin Laisk, Reedik Mägi, Kadri Kõiv, Kelli Lehto

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine your body's appetite and eating habits are like a complex orchestra. Sometimes, the music is perfect; other times, certain instruments get out of tune, leading to a chaotic symphony known as an eating disorder. While doctors usually only study the "loud crashes" of the orchestra (clinical diagnoses like Anorexia or Bulimia), this study decided to listen to the entire band, including the quiet, off-key notes that happen in everyday people who haven't been diagnosed with a disorder.

Here is what the researchers found, explained simply:

1. The "SCOFF" Detective Tool

The researchers used a five-question checklist called SCOFF (Sick, Control, One stone, Fat, Food) to ask 212,000 adults in Estonia about their eating habits. Think of this like a smoke detector. It doesn't tell you why the fire started, but it tells you if there's smoke.

The Big Finding: They found that about 25% of the people in the study had enough "smoke" (symptoms) to be considered at higher risk for an eating disorder, even though most of them had never been diagnosed by a doctor.

2. The Genetic Blueprint: Two Different Types of "Smoke"

The study looked at the DNA of these people to see if there was a genetic "instruction manual" for these eating problems. They discovered that the genetic signals were very different depending on which symptom you asked about.

The "Cognitive-Behavioral" Signals (The Loud Noises):
Two specific symptoms stood out in the DNA:
1. "Food dominates my life" (obsessing over food).
2. "I've lost control over how much I eat" (binge eating).
  These symptoms had a strong genetic blueprint. The researchers found specific genetic "switches" (genes like FTO and MC4R) that are famous for regulating appetite. It's as if the genetic instruction manual for these people has a specific chapter on "appetite regulation" that is written in a way that makes them feel like food is taking over their brain.
The "Weight-Compensatory" Signals (The Quiet Whispers):
The other three symptoms were:
1. "I make myself sick" (vomiting).
2. "I feel fat even when others say I'm thin."
3. "I lost a lot of weight quickly."
  Crucially, the study found almost no genetic signals for these. It's as if the genetic blueprint for these specific behaviors is either missing, very faint, or so complex that the current tools couldn't read it. The researchers suggest these behaviors might be more about immediate reactions or environment rather than a strong, inherited genetic code.

3. The "Double-Edged Sword" of Genes

The study found that the genes linked to "losing control" and "food dominating" didn't just affect eating; they were connected to a whole other set of traits. Imagine a master key that opens many doors:

Physical Doors: It opens the door to obesity, high blood pressure, and type 2 diabetes.
Mental Doors: It opens the door to anxiety, depression, ADHD, and trouble sleeping.
Lifestyle Doors: It links to smoking and spending too much time on screens.

The researchers found that people with these specific genetic risks were much more likely to have these other health issues, even if they didn't have a formal eating disorder diagnosis.

4. Predicting the Future (The "Crystal Ball")

The team built a "genetic score" (a Polygenic Risk Score) based on these findings. They tested this score on a different group of people to see if it could predict who would struggle with uncontrolled eating (the feeling of "I can't stop once I start").

The Result: The score worked. People with a higher genetic score were significantly more likely to report that they have trouble stopping once they start eating. This proves that the genetic signals they found are real and meaningful.

5. The Main Takeaway

The study concludes that for the general population, the genetic risk for eating problems is mostly driven by the mental struggle with food (feeling out of control, food taking over your life) rather than the physical attempts to fix it (vomiting, extreme dieting).

In simple terms:
If you think of eating disorders as a house on fire, current medicine often waits until the roof is burning (clinical diagnosis) to call the fire department. This study suggests that the "smoke detectors" (the genetic signals for losing control and food obsession) are already going off in many people's homes long before the roof catches fire. These "smoke signals" are strongly linked to other health issues like heart disease and diabetes, suggesting we need to pay attention to these early warning signs in everyday people, not just those with a formal diagnosis.

Technical Summary: Genetic Architecture of Eating Disorder Risk and Specific Symptoms in the General Adult Population

Problem Statement
Eating disorders (EDs) are complex psychiatric conditions with significant somatic comorbidity and elevated mortality. While previous genome-wide association studies (GWAS) have identified risk variants for clinical diagnoses (primarily anorexia nervosa), the genetic architecture of ED symptoms as they manifest in the general population—outside clinical thresholds—remains unclear. Subclinical symptoms, such as disordered eating behaviors and attitudes, are more prevalent than clinical diagnoses and exist along a severity spectrum. Current diagnosis-based methods may miss individuals at risk who have not reached clinical thresholds, limiting early identification and intervention strategies. There is a need to understand the genetic underpinnings of specific ED symptoms in the general population to improve detection and prevention.

Methodology
The study utilized data from the Estonian Biobank (EstBB), a population-based resource covering approximately 20% of the adult population (N ≈ 212,000).

Phenotyping: ED symptoms were assessed using the modified Estonian version of the SCOFF screening tool (Sick, Control, One stone, Fat, Food), a five-item questionnaire. A sum-score ≥ 2 indicated "higher ED risk." The five items were analyzed individually: "food dominates," "lost control," "lost > 6 kg," "feel fat," and "make yourself sick." These were categorized into "cognitive-behavioral" symptoms (higher ED risk, lost control, food dominates) and "weight-compensatory" symptoms (lost > 6 kg, feel fat, make yourself sick).
GWAS Design: Six genome-wide association studies were conducted: one for higher ED risk and one for each of the five individual symptoms. The analysis included approximately 8.5 million SNPs after quality control. GWASs were performed using REGENIE software, adjusting for age, sex, and 10 genetic principal components. Notably, BMI was not adjusted for in the main analysis to avoid collider bias, as BMI may lie on the causal pathway between SNPs and ED risk.
Sensitivity Analyses: Sex-stratified and BMI-stratified (underweight/normal vs. overweight/obese) GWASs were conducted.
Genetic Correlation & Heritability: Linkage Disequilibrium Score Regression (LDSC) was used to estimate SNP-based heritability ( $h^2_{SNP}$ ) and genetic correlations ( $r_g$ ) between ED traits and 59 external phenotypes (anthropometric, psychiatric, lifestyle, personality, and somatic traits).
Gene Mapping: Significant loci were annotated and mapped to genes using FUMA GWAS, Open Targets Genetics, and Mouse Genome Informatics.
Polygenic Risk Scores (PRS): PRSs were calculated for all six phenotypes using PRS-CS-auto. These were tested in an independent EstBB subsample (n=11,218) for association with "uncontrolled eating" (difficulty stopping once eating started).
Phenome-Wide Association Studies (PheWAS): PRSs were linked to 1,278 ICD-10 diagnoses from electronic health records (2004–2023) in an independent subsample (n=71,594) to map medical diagnostic profiles associated with genetic liability.

Key Results

Genome-Wide Significant Signals: The study identified 14 independent genome-wide significant SNPs across three GWASs: higher ED risk, "lost control," and "food dominates." No significant signals were detected for the weight-compensatory symptoms ("lost > 6 kg," "feel fat," "make yourself sick").
Gene Prioritization: Lead SNPs mapped to intronic regions of genes including FTO and MC4R (linked to weight-related traits), ROBO2 (linked to sensory processing), and others such as GPR61, TMEM18, GSTM1, NRN1, and FAIM2. Mouse model data suggested disruptions in these genes relate to distorted eating patterns, polyphagia, and sensory capabilities.
Heritability: SNP-based heritability estimates ranged from 2% to 8%. The highest heritability was observed for "higher ED risk" and "lost control" (8% each), followed by "food dominates" (7%). Weight-compensatory symptoms showed lower heritability (2–5%).
Genetic Correlations:
- Strong genetic correlations existed between higher ED risk, "lost control," and "food dominates" ( $r_g$ up to 1.00).
- Cognitive-behavioral symptoms showed moderate-to-strong positive genetic correlations with anthropometric traits (especially waist circumference, BMI), psychiatric traits (ADHD, alcohol use disorder, depression, PTSD), lifestyle factors (smoking, screen time), and somatic traits (cardiometabolic conditions, asthma).
- Weight-compensatory symptoms exhibited fewer and less consistent genetic correlations.
PRS Validation: In the independent subsample, all PRSs predicted uncontrolled eating. The strongest association was with "lost control" (PRS_ControlLoss), where a one standard deviation increase raised the odds of uncontrolled eating by nearly 20% (OR=1.18).
PheWAS Findings: PRSs for cognitive-behavioral symptoms were associated with a broad range of medical conditions (24–28 significant diagnoses), predominantly cardiometabolic. The strongest association across all PRSs was with obesity (ICD-10: E66). Other significant associations included Type 2 diabetes, hypertensive heart disease, gout, and knee osteoarthritis. Interestingly, even PRSs for weight-compensatory symptoms showed their strongest association with obesity.

Significance and Claims
The authors claim this study provides the first genome-wide association analysis of self-reported ED risk and specific symptoms in a general adult population. The primary contribution is the demonstration that the genetic liability and health burden related to ED symptoms in the general population are predominantly driven by cognitive-behavioral symptoms (loss of control, food dominating life) rather than weight-compensatory symptoms.

The paper argues that a symptom-based approach complements current diagnosis-based methods. By identifying genetic variants associated with subclinical symptoms, the study highlights a substantial gap between individuals with higher ED risk (25.7% in the sample) and those with a clinical diagnosis (~1%). The findings suggest that focusing solely on diagnosed cases misses a considerable proportion of individuals with disordered eating behaviors. The identification of shared genetic architecture with cardiometabolic and psychiatric conditions underscores the broad health impact of these symptoms. Ultimately, the authors posit that a symptom-informed genetic framework can better map phenotypic heterogeneity, facilitating early intervention and prevention strategies for a diverse set of disordered eating behaviors before they deteriorate into severe clinical conditions.