The Genome-Wide Association Study (GWAS) meta-analysis

The Genome-Wide Association Study (GWAS) meta-analysis by Watson ym. «Watson HJ, Yilmaz Z, Thornton LM, ym. Genome-wide ...»1 included 33 datasets comprising 16 992 cases and 55 525 controls of European ancestry from 17 countries. Participants included individuals from the PGC-ED (The eating disorders workgroup of Psychiatric Genomics Concortium), the ANGI (Anorexia Nervosa Genetic Initiative), archived samples from the GCAN/WTCCC (Genetic Consortium for Anorexia Nervosa / The Wellcome Trust Case Control Consortium), samples from cases of anorexia nervosa from the UK Biobank, and additional controls from Poland.

Case definitions established a lifetime diagnosis of anorexia nervosa via hospital or register records, structured clinical interviews, or online questionnaires based on standardized criteria (DSM-III-R / DSM-IV / ICD-8 / ICD-9 / ICD-10), whereas in the UK Biobank, cases self-reported a diagnosis of anorexia nervosa. Controls were carefully matched for ancestry, and some, not all, control cohorts were screened for lifetime eating and/or some or all psychiatric disorders. Given the relative rarity of anorexia nervosa, large unscreened control cohorts were deemed appropriate for inclusion.

GWAS for Anorexia Nervosa tested hundreds of thousands of genetic variants across large number of genomes to find those statistically associated with the disorder.

GWAS identified the following eight risk loci that exceeded genome-wide significance (P <5 x 10^-8):

• The lead SNP of rs9821797 (in chromosome 3 locus 1, multigenic with 111 genes, nearest gene NCKIPSD, P = 6.99 × 10⁻¹⁵, OR=1.17)
• The lead SNP of rs6589488 (in chromosome 11 locus 2, single gene, nearest gene CAM1, P = 6.31 × 10⁻¹¹, OR=1.14)
• The lead SNP of rs2287348 (in chromosome 2 locus 3, multigenic with 13 genes, nearest gene ASB3 and ERLEC1, P = 5.62 × 10⁻⁹, OR=1.11)
• The lead SNP of rs2008387 (in chromosome 10 locus 4, single gene, nearest gene MGMT, P = 1.73 × 10⁻⁸, OR=1.08)
• The lead SNP of rs9874207 (in chromosome 3 locus 5, single gene, nearest gene FOXP1, P = 2.05 × 10⁻⁸, OR=1.08)
• The lead SNP of rs10747478 (in chromosome 1 locus 6, single gene, nearest gene PTBP2, P = 3.13 × 10⁻⁸, OR=1.08)
• The lead SNP of rs370838138 (in chromosome 5 locus 7, intergenic, nearest gene CDH10, P = 3.17 × 10⁻⁸, OR=1.08)
• The lead SNP of rs13100344 (in chromosome 3 locus 8, intergenic, nearest gene NSUN3, P = 4.21 × 10⁻⁸, OR=1.08).

The authors evaluated three ways to connect anorexia nervosa-associated loci identified by GWAS to genes:

1. regulatory chromatin interactions,
2. relationship to brain expression quantitative trait loci (eQTLs), and
3. the standard approach of gene location within a GWAS locus. The significant anorexia nervosa-associated loci implicated 121 brain-expressed genes.

Four single-gene loci were confirmed by eQTL analyses, chromatin interaction studies or both. These were the locus-intersecting genes CADM1 (locus 2, in chromosome 11), MGMT (locus 4, in chromosome 10), FOXP1 (locus 5, in chromosome 3) and PTBP2 (locus 6, in chromosome 1). For locus 5, eQTL data implicated a distal gene, GPR27. One intergenic locus (locus 7, in chromosome 5) had no eQTL or chromatin interactions, whereas the other intergenic locus (locus 8, in chromosome 3) had eQTL connections to PROS1 and ARL13B. Two complex multigenic loci had many brain-expressed genes and dense chromatin and eQTL interactions that precluded identification of any single gene (locus 1, in chromosome 3; locus 3, in chromosome 2).

The clearest evidence and connections were for the single-gene loci that intersected with CADM1, MGMT, FOXP1 and PTBP2, and the authors concluded that these genes may have a role in the etiology of anorexia nervosa.

Liability-scale SNP heritability (SNP-h²) was estimated using linkage disequilibrium score regression. Assuming a lifetime prevalence of 0.9–4%, SNP-h² was 11–17%, supporting the polygenic nature of anorexia nervosa. Polygenic risk score analyses using a leave-one-out approach indicated that the polygenic risk score captures approximately 1.7% of the phenotypic variance on the liability scale for discovery. The authors did not observe differences in polygenic architecture between anorexia nervosa subtypes with binge eating (2381 cases, 10249 controls) or without (2262 cases, 10254 controls), or between males (447 cases, 20347 controls) and females (14898 cases, 27545 controls). Similar to females, males in the highest polygenic risk score decile had 4.1 (95% CI 2.6–6.6) times the odds of anorexia nervosa than those in the lowest decile.

SNP-based genetic correlations (SNP - r_g) were tested with external traits. Bonferroni-significant SNP - r_g assorted into five trait categories: psychiatric and personality, physical activity, anthropometric traits, metabolic traits and educational attainment.

Bonferroni-corrected positive SNP- r_g values were associated with the following:

• obsessive compulsive disorder (SNP - r_g ± s.e. = 0.45 ±0.08; P =4.97×10⁻⁹),
• major depressive disorder (SNP - r_g ± s.e. = 0.28 ±0.07; P =8.95×10⁻⁵),
• anxiety disorders (SNP - r_g ± s.e. = 0.25 ±0.05; P =8.90×10⁻⁸), and
• schizophrenia (SNP - r_g ± s.e. = 0.25 ±0.03; P =4.61×10⁻¹⁸).
• physical activity (SNP - r_g ± s.e. = 0.17 ±0.05; P =1.00×10⁻⁴)
• educational attainment (SNP - r_g ± s.e. = 0.25 ±0.03; P =1.69×10⁻¹⁵)
• fat mass (SNP - r_g ± s.e. = −0.33 ±0.03; P =7.23×10⁻²⁵),
• fat-free mass (SNP - r_g ± s.e. −0.12 ±0.03; P =4.65×10⁻⁵),
• BMI (SNP - r_g ± s.e. = −0.32 ±0.03; P =8.93×10⁻²⁵),
• obesity (SNP - r_g ± s.e. = −0.22 ±0.03; P =2.96×10⁻¹¹),
• type 2 diabetes (SNP - r_g ± s.e. = −0.22 ±0.05; P=3.82×10⁻⁵),
• fasting insulin (SNP - r_g ± s.e. −0.24 ±0.06; P =2.31×10⁻⁵),
• insulin resistance (SNP - r_g ± s.e. = −0.29 ±0.07; P =2.83×10⁻⁵)
• leptin (SNP - r_g ± s.e. = −0.26 ±0.06; P =4.98×10⁻⁵),
• HDL cholesterol (SNP - r_g ± s.e. = 0.21±0.04; P=3.08×10⁻⁷).

Of note, the metabolic and anthropometric r_g with anorexia nervosa were more pronounced than are in other psychiatric disorders; educational attainment and related constructs were not seen for IQ (Intelligence Quotient).

Systems biology analyses of the results revealed notable observations. Gene-wise analysis with MAGMA prioritized 79 Bonferroni-corrected significant genes, most within the multigenic locus on chromosome 3. MAGMA indicated an association with NCAM1, the expression of which increases in response to food restriction in a rodent activity-based anorexia nervosa model. Partitioned heritability analysis showed considerable enrichment of SNP-h² in conserved regions. Cell type group-specific annotations revealed that the overall SNP-h² is significantly enriched for tissues of the central nervous system. One biological pathway, Gene Ontology, was significant; it is related to regulation of embryonic development (32 genes, P=1.39×10⁻⁷), which contains two Bonferroni-corrected significant genes on chromosome 3, CTNNB1 and DAG1. CTNNB1 encodes catenin β-1, which is part of adherens junctions and a component of Wnt signaling, and DAG1 encodes dystroglycan, a receptor that binds extracellular matrix proteins. This pathway points to a potential role of developmental processes in the etiology of this complex phenotype (although currently speculative). Genes associated with anorexia nervosa were enriched for expression in most brain tissues, particularly the cerebellum, which has a notably high proportion of neurons. Among 24 brain cell types from mouse brain, significant enrichment was found for medium spiny neurons and pyramidal neurons from hippocampal CA1. Both medium spiny and pyramidal neurons are linked to feeding behaviors, including food motivation and reward. 36 genes were predicted to be differentially expressed in GTEx tissues or blood, with the expression of MGMT predicted to be downregulated in the caudate. The authors noted that these results represent the first indications of specific pathways, tissues and cell types that may mediate genetic risk for anorexia nervosa.

The study had 80% power to detect an odds ratio of 1.09–1.19 (additive model, 0.9% lifetime risk, α =5×10⁻⁸, minor allele frequency =0.05–0.5). Typical of complex-trait GWAS, the test statistic inflation (λ=1.22) was observed consistent with polygenicity, with no evidence of significant population stratification according to the linkage disequilibrium intercept and attenuation ratio. Meta-analysis results were completed for autosomes and the X chromosome.

The authors concluded that genetic architecture of anorexia nervosa mirrors its clinical presentation, showing significant genetic correlations with psychiatric disorders, physical activity, and metabolic (including glycemic), lipid and anthropometric traits, independent of the effects of common variants associated with body-mass index. The authors also suggest that these results encourage a reconceptualization of anorexia nervosa as a metabo-psychiatric disorder. Elucidating the metabolic component is suggested to represent a critical direction for future research, and paying attention to both psychiatric and metabolic components is suggested to be a key to improving outcomes.