Shared genetic architecture between anorexia nervosa and metabolomic biomarkers suggest underlying causal pathways

This study reveals that anorexia nervosa shares a distinct genetic architecture with 142 metabolomic biomarkers, particularly lipid-related ones, which are driven by developmental and inflammatory processes and linked to cardiometabolic traits through a causal pathway mediated by body mass index (BMI).

The Gist

Imagine your body is a massive, bustling city. In this city, Anorexia Nervosa (AN) isn't just a problem with the "Mayor's office" (the brain); it's a city-wide power outage that affects the power grid, the water supply, and the traffic systems all at once.

For a long time, scientists thought Anorexia was purely a psychological issue—a glitch in the brain's software. But this new research suggests that the "hardware" (your body's metabolism) is deeply tangled up with the "software" (your mind).

Here is the story of what this paper discovered, explained simply:

1. The "Fingerprint" Match

Scientists took a giant magnifying glass to the DNA of people with Anorexia and compared it to the DNA of people with various body chemicals (called metabolites). Think of these metabolites as the "exhaust fumes" or "fuel efficiency ratings" of your body's engine.

They found a massive match. Out of 249 different body chemicals, 142 of them shared a genetic fingerprint with Anorexia. It's like finding that the same specific tool is used to build both a house (the body) and a painting (the mind).

2. The Great "Opposite" Dance

Here is the twist: The relationship was backwards.

• The "Healthy" Dance: Usually, when people have high Body Mass Index (BMI), Type 2 Diabetes, or heart issues, their body chemistry looks a certain way (let's call it the "Heavy" pattern).
• The Anorexia Dance: People with Anorexia had the exact opposite chemical pattern.

If you imagine the "Healthy/Heavy" pattern as a red light, Anorexia is flashing a green light. The study found that the genetic code for Anorexia is almost the mirror image of the code for obesity and diabetes. It's as if the body is running a "reverse engine" compared to what we see in common weight-related diseases.

3. The "Double-Edged Sword" of Fat

The researchers zoomed in on fats and lipids (the fuel in your body). They found that Anorexia is linked to a mix of "good" and "bad" fat signals.

• Some signals said, "Make more fat!"
• Others said, "Stop making fat!"

It's like a car with a gas pedal and a brake pedal both being pressed at the same time. This confusion in the body's fuel system seems to be a core part of why the disorder happens. The genes involved are like the mechanics who are supposed to manage the car's oil and fuel, but they are sending mixed signals.

4. The "Middleman" (BMI)

The study asked a big question: Does Anorexia change the body's fuel, or does the body's fuel change Anorexia?

They found that the answer is a bit of both, but Body Mass Index (BMI) is the middleman.

• The Chain Reaction: Your genes influence your BMI. Your BMI then changes your body's fuel (metabolites). And because Anorexia is so tightly linked to low BMI, the fuel changes happen because of the weight loss, which then feeds back into the disorder.
• Think of it like a domino effect: The genetic dominoes fall, knocking over the "Low Weight" domino, which then knocks over the "Body Chemistry" domino.

5. Why This Matters: A New Map for Treatment

For years, treating Anorexia has been like trying to fix a broken car by only looking at the driver's seat. This study suggests we need to look under the hood.

• The "Fuel" Clue: Because the body's fuel system is so broken, maybe we can fix the disorder by fixing the fuel. The study suggests that instead of just trying to force someone to eat more, we might need to look at precision nutrition—giving the body the specific "fuel" (like certain amino acids or fats) it is genetically starving for.
• The "Reverse" Medicine: Since Anorexia is the opposite of diabetes and heart disease, maybe the drugs used to treat those conditions could be tweaked to help Anorexia patients. For example, if a drug lowers cholesterol for a diabetic, maybe a drug that raises specific healthy fats could help an Anorexia patient restore their weight and health.

The Bottom Line

This paper tells us that Anorexia is a metabo-psychiatric disorder. It's not just in the head; it's in the blood, the liver, the muscles, and the brain all at once.

By understanding that the body's "fuel system" is genetically wired to be the opposite of common weight-gain diseases, scientists now have a new map. They can stop guessing and start targeting the specific biological switches that are stuck, offering hope for treatments that actually work to restore both body and mind.

Technical Summary

1. Problem Statement

Anorexia Nervosa (AN) is a severe, often chronic psychiatric disorder with high mortality rates and limited effective treatments. While AN is known to be highly heritable and shares genetic architecture with cardiometabolic traits (e.g., BMI, Type 2 Diabetes), the specific relationship between AN and circulating metabolomic biomarkers remains largely unexplored.

• The Gap: Previous studies have linked AN to broad anthropometric measures, but the specific biological pathways involving individual metabolites (lipids, amino acids, etc.) are unknown.
• The Hypothesis: There is a shared genetic etiology between AN and systemic metabolism that operates through distinct biological pathways, potentially mediated by Body Mass Index (BMI), offering new targets for treatment and weight restoration.

2. Methodology

The study employed a multi-step genomic and bioinformatic approach using large-scale Genome-Wide Association Study (GWAS) summary statistics.

• Data Sources:
  • AN: GWAS of 16,992 cases and 55,525 controls (Psychiatric Genomics Consortium).
  • Metabolomics: 249 circulating metabolomic biomarkers (lipoproteins, fatty acids, amino acids, etc.) derived from NMR profiling in the UK Biobank and Estonian Biobank (total $N \approx 300,000$).
  • Comparators: GWAS data for Anxiety (ANX), BMI, Type 2 Diabetes (T2D), and Coronary Artery Disease (CAD).
• Statistical Analyses:
  1. Global Genetic Correlation: Used LD Score Regression (LDSC) to estimate genome-wide genetic correlations ($r_g$) between AN and all 249 metabolites, comparing these profiles against ANX, BMI, T2D, and CAD.
  2. Variant Discovery: Applied Conjunctional False Discovery Rate (condFDR/conjFDR) to identify specific shared genetic variants between AN and metabolites, capturing both concordant (same direction) and discordant (opposite direction) effects.
  3. Genetic Overlap Quantification: Used Bivariate Gaussian Mixture Modeling (MiXeR) to estimate the proportion of shared causal variants and their directionality for prioritized metabolites.
  4. Functional Annotation:
     • Variant-to-Gene Mapping: Mapped lead variants to genes using Open Targets Genetics.
     • Gene Set Enrichment: Performed Gene Ontology (GO) analysis on mapped genes.
     • Tissue Enrichment: Analyzed gene expression patterns across 30 tissues (GTEx v8) and specific brain regions (Allen Human Brain Atlas).
  5. Causal Inference: Conducted Bidirectional Mendelian Randomization (MR) to test causal relationships between AN, BMI, and metabolomic markers.

3. Key Results

A. Genetic Correlations and Profiles

• Significant Overlap: Significant genetic correlations were found between AN and 142 of the 249 metabolomic biomarkers.
• Opposite Directionality: The genetic correlation profile of AN with metabolites was opposite in direction to that of cardiometabolic traits (BMI, T2D, CAD). For instance, while high BMI correlates positively with certain lipids, AN correlates negatively with them (and vice versa).
• Distinct from Anxiety: The AN-metabolomic profile showed strong negative correlations with cardiometabolic traits ($r_s \approx -0.94$), whereas the Anxiety-metabolomic profile showed positive correlations. This suggests AN and Anxiety have distinct biological underpinnings regarding systemic metabolism.
• Specific Markers: Branched-chain amino acids (leucine, valine) showed significant negative genetic correlations with AN but not with Anxiety.

B. Shared Genetic Variants

• Discovery Boost: The condFDR approach identified up to 16 shared variants for specific metabolites (e.g., Cholesteryl esters in HDL and VLDL particle size), effectively doubling the number of genome-wide significant loci discovered for AN compared to the original GWAS (which found 8 loci).
• Directionality Mix: A majority (57.9%) of shared variants had discordant effects (opposite directions) between AN and metabolites. MiXeR analysis confirmed a mixture of positive and negative effects for prioritized markers like HDL-cholesteryl esters (HDL-CE) and VLDL particle size.

C. Biological Pathways and Tissue Enrichment

• Pathways: Gene Ontology analysis of shared genes revealed enrichment in:
  • Lipid-related cell signaling.
  • Developmental growth (increase in organism/cell size).
  • Inflammation (cytokine production).
  • Canonical cellular processes (e.g., tyrosine kinase signaling).
• Tissue Expression: Shared genes were significantly enriched in 10 of 20 tissues, including the brain, pancreas, liver, heart, kidneys, skeletal muscle, and female reproductive organs.
• Key Genes: Notable shared genes included CTNNB1 (Beta-catenin, involved in Wnt signaling and neurodevelopment), ARIH2, and CTC1.

D. Causal Relationships (Mendelian Randomization)

• AN $\leftrightarrow$ Metabolites: No direct causal relationship was found between AN and metabolomic markers in either direction.
• The Role of BMI:
  • Strong bidirectional causality exists between AN and BMI (AN causes lower BMI; lower BMI increases AN risk).
  • BMI mediates the relationship: The study suggests that the association between AN and metabolomic markers is likely indirect, acting through their respective influences on BMI. AN influences BMI, which in turn alters metabolite levels.

4. Key Contributions

1. Reconceptualization of AN: Provides strong evidence supporting the view of AN as a "metabo-psychiatric" disorder, where systemic metabolic dysregulation is genetically intertwined with psychiatric pathology.
2. Novel Genetic Loci: Successfully identified new genetic variants associated with AN by leveraging cross-trait analysis with metabolomics, expanding the known genetic architecture beyond the original GWAS.
3. Differentiation from Anxiety: Demonstrated that while AN and Anxiety are comorbid, their genetic relationships with systemic metabolism are fundamentally different (opposite directions), suggesting distinct therapeutic targets.
4. Mechanistic Insight: Identified specific biological processes (lipid signaling, developmental growth, inflammation) and tissues (reproductive organs, liver, brain) involved in AN etiology.
5. Mediation Model: Clarified that BMI acts as a central mediator in the causal pathway between AN and metabolic markers, rather than a direct causal link between AN and specific metabolites.

5. Significance and Implications

• Treatment Innovation: The findings suggest that metabolism-informed treatments could be viable for AN. Since AN shows an opposite metabolic profile to cardiometabolic diseases, therapies that increase specific lipids or amino acids (rather than lowering them, as done for heart disease) might aid in weight restoration.
• Precision Nutrition: The specific link to branched-chain amino acids (valine, leucine) and lipid markers offers a basis for developing precision nutrition strategies to address appetite regulation and satiety in AN patients.
• Drug Repurposing: The study hints that medications used for cardiometabolic conditions (e.g., lipid-altering drugs) might have utility in AN if their mechanism of action is reversed (e.g., lipid-increasing agents).
• Developmental Perspective: The enrichment of developmental growth pathways suggests that sensitive periods (e.g., puberty) where these biological processes are active may be critical windows for AN onset and intervention.

Limitations: The study relies on a metabolomic panel biased toward lipids (NMR), limiting the scope to other metabolite classes (e.g., proteins). Additionally, while MR suggests mediation, it does not prove direct causality, and findings may not generalize to atypical AN (where BMI is not low).