henhaobio
Sign in

Nature GeneticsGeneticsPeer-reviewed journal article

Discovery Of The First Genome-Wide Significant Risk Loci For ADHD

Ditte DemontisiD, Raymond Kenney WaltersiD, Joanna MartiniD, Manuel MattheiseniD, Thomas D. AlsiD, Esben AgerboiD, Gísli Baldursson, Rich Belliveau, Jonas Bybjerg‐GrauholmiD, Marie Bækvad‐HanseniD, Felecia E. CerratoiD, Kimberly Chambert, Claire Churchhouse, Ashley Dumont, Nicholas ErikssoniD, Michael J. GandaliD, Jacqueline I. GoldsteiniD, Katrina L. GrasbyiD, Jakob GroveiD, Ólafur Ó. GuðmundssoniD, Christine Søholm HanseniD, Mads E. HaubergiD, Mads V. HollegaardiD, Daniel P. HowriganiD, Hailiang HuangiD, , Benjamin M. NealeiD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

Attention-Deficit/Hyperactivity Disorder (ADHD) is a highly heritable childhood behavioral disorder affecting 5% of school-age children and 2.5% of adults. Common genetic variants contribute substantially to ADHD susceptibility, but no individual variants have been robustly associated with ADHD. We report a genome-wide association meta-analysis of 20,183 ADHD cases and 35,191 controls that identifies variants surpassing genome-wide significance in 12 independent loci, revealing new and important information on the underlying biology of ADHD. Associations are enriched in evolutionarily constrained genomic regions and loss-of-function intolerant genes, as well as around brain-expressed regulatory marks. These findings, based on clinical interviews and/or medical records are supported by additional analyses of a self-reported ADHD sample and a study of quantitative measures of ADHD symptoms in the population. Meta-analyzing these data with our primary scan yielded a total of 16 genome-wide significant loci. The results support the hypothesis that clinical diagnosis of ADHD is an extreme expression of one or more continuous heritable traits.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (3 Jun 2017) · CC BY 4.0

Genome-wide association meta-analysis of 20,183 ADHD cases and 35,191 controls finds 12 significant loci; signal concentrates in conserved regions, constrained genes, and CNS regulatory elements.

Twin studies place ADHD heritability at 70-80% and SNP heritability estimates span 0.10 to 0.28, yet earlier GWASs had found no robustly associated common variants. Pooling iPSYCH and PGC cohorts (55,374 individuals) yields the first genome-wide significant ADHD loci and ties clinical diagnosis genetically to population-level ADHD symptom traits.

Key results

  1. ShownTwelve independent loci reach genome-wide significance for ADHD, the first robust common-variant associations. Effects are homogeneous across cohorts, no locus shows a genome-wide significant secondary signal, and credible sets at three loci, including FOXP2 on chromosome 7, map to single genes.

    304 variants in 12 loci passed P<5×10^−8 among 8,047,421 tested variants; no marker showed significant between-study heterogeneity; conditional tests of two putative secondary signals found none independent; index-SNP odds ratios 1.077 to 1.198; chr12 credible set includes a non-lead DUSP6 missense variant, and most chr15 credible-set variants are SEMA6D fibroblast eQTLs. · Genome-wide significantly associated ADHD risk…; Biological annotation of significant loci; Discussion; Conditional analysis

  2. ShownCommon variants carry substantial liability-scale heritability for ADHD. It is enriched in evolutionarily conserved regions and central nervous system (CNS) regulatory elements, and loss-of-function intolerant genes are associated as a set.

    Liability-scale h^2snp=0.216 (SE=0.014; P=8.18×10^−54) assuming 5% prevalence; polygenic effects account for 88% of test-statistic inflation; conserved-region enrichment P = 8.49 x 10^−10; CNS regulatory enrichment 2.44 (SE=0.35; P = 5.81 x 10^−5); 2,932 constrained genes (pLI ≥0.9) beta=0.062, P=2.6x10^−4. · Polygenic Architecture of ADHD; Analysis of gene sets; Gene-set analyses

  3. ShownGenetic effects agree between the Danish population cohort and the PGC case/control and trio cohorts. Polygenic risk scores track ADHD risk in a dose-dependent way across score deciles.

    LD score rg between European PGC and iPSYCH 1.17 (SE = 0.2); bivariate GREML rg for PGC case/control vs trio 1.02 (SE = 0.32); five-fold iPSYCH cross-validation gave mean maximal Nagelkerke R^2 5.5% (SE = 0.0012) and OR per standardized score 1.56 (95% CI 1.53 to 1.60); odds ratios rose across PRS deciles in iPSYCH and PGC. · Homogeneity of effects between cohorts; Genetic correlations between ADHD samples; Polygenic Risk Scores for ADHD

  4. ShownADHD shares common-variant genetic risk with depression, lower educational attainment and childhood IQ, smoking, obesity-related traits, reproductive traits, and parental mortality.

    38 of 220 tested phenotypes significant (P<2.27×10^−4): major depressive disorder rg=0.42 (P= 7.38×10^−38), years of schooling rg=−0.53 (P= 6.02x10^−80), childhood IQ −0.41, ever smoking rg=0.48, BMI rg=0.26, waist-to-hip ratio rg=0.30, HDL cholesterol rg=−0.22. · Genetic correlation with other traits; Discussion; Genetic correlations of ADHD with other traits

  5. ShownThe clinical ADHD signal shows moderate concordance with self-reported diagnosis and stronger agreement with population ADHD symptom scores, which supports diagnosis as the extreme of continuous heritable traits. Joint meta-analysis yields 16 loci in total.

    23andMe (5,857 cases, 70,393 controls): rg=0.653 (SE=0.114), below 1 (P=2.35x10^−3); 71 of 94 clumped loci sign-concordant (P=1.25x10^−6) but only 7 of 11 genome-wide loci (P=0.386); heterogeneity at rs12410155 (I^2=97.2). EAGLE (17,666 children): rg=0.943 (SE = 0.204); 10 of 11 genome-wide loci concordant (P =0.0159). Joint meta-analysis: 16 loci. · Replication of GWAS loci; Replication of GWAS loci

Methods, data and reproducibility

Logistic-regression GWAS run per cohort (Ricopili QC; SHAPEIT/IMPUTE2 imputation to 1000 Genomes phase 3) in 12 cohorts and pooled by fixed-effects inverse-variance meta-analysis. Follow-up used LD score regression for heritability, partitioning and genetic correlation, polygenic risk score cross-validation, credible-set fine-mapping, MAGMA gene-set tests, and replication in 23andMe and EAGLE.

  • iPSYCH Danish population-based case/control cohort; register-based ICD10 F90.0 diagnoses; Illumina PsychChip genotypes processed in 23 batches · human · n = 14,584 cases and 22,492 controls
  • 11 PGC ADHD cohorts from Europe, North America and China, with case/control and parent-offspring trio designs · human
  • Combined post-QC primary GWAS meta-analysis across 12 cohorts (8,047,421 variants after filtering) · human · n = 20,183 cases and 35,191 controls (55,374 individuals)
  • European-ancestry subset meta-analysis used for LD score, credible-set and MAGMA analyses · human · n = 19,099 cases and 34,194 controls
  • 23andMe GWAS of self-reported ADHD or attention deficit disorder diagnosis (replication) · human · n = 5,857 cases and 70,393 controls
  • EAGLE consortium meta-analysis of childhood ADHD symptom rating scales (replication) · human · n = 17,666 children < 13 years of age
  • LD Hub public GWAS summary statistics plus an updated PGC major depressive disorder GWAS for genetic correlation · human · n = 219 phenotypes; MDD 130,664 cases and 330,470 controls
  • 1000 Genomes Project Phase 3 imputation reference and precomputed European LD scores for HapMap3 SNPs · human (1000 Genomes Project) · n = 378 European-ancestry individuals (LD scores)
Methods & tools
Ricopili (PGC QC, imputation and association pipeline), SHAPEIT and IMPUTE2 (phasing and imputation), PLINK v1.9, EIGENSOFT (principal component outlier removal), METAL (version 2011-03-25), LD score regression (ldsc) with LD Hub, GCTA bivariate GREML, MAGMA 1.05, FM-summary R script (Maller et al. credible sets), Ensembl Variant Effect Predictor (GRCh37), FUMA (GTEx/BIOS eQTL and Roadmap chromHMM annotation), MSigDB 6.0 Gene Ontology sets and ExAC pLI constraint scores
Software & environment
PLINK v1.9; METAL version 2011-03-25; MAGMA 1.05; MsigDB 6.0 Gene Ontology sets; Ensembl VEP on GRCh37 with Gencode v19; NCBI 37.3 gene definitions; 1000 Genomes Phase 3 reference panel.
Reproducibility gaps
The text gives no download location for the ADHD meta-analysis summary statistics or for individual-level iPSYCH, PGC or 23andMe genotypes; materials are requested through the corresponding authors; the MDD comparison GWAS was still unpublished (submitted).
Limitations (6)
  • The 12 loci account for only a small share of common-variant liability: index-SNP odds ratios are modest (1.077 to 1.198), while all common SNPs together explain much more (h^2snp = 0.22). · stated by the authors
  • The lead chromosome 1 locus lies in a gene-rich 250kb block of strong LD, Hi-C and eQTL data point to several alternative genes, and it shows significant heterogeneity with 23andMe (rs12410155 I^2=97.2), so its causal source is unresolved. · stated by the authors
  • Credible set annotations at several loci remain cryptic, and three FOXP2 downstream target gene sets showed no association with ADHD, which the authors suggest may reflect limited power or unknown FOXP2 functions. · stated by the authors
  • Inflation beyond polygenic signal is modest but significant (LD score intercept 1.0362, SE=0.0099; P=2.27×10^−4) and may reflect cryptic relatedness or population stratification. · stated by the authors
  • Genetic correlations may reflect any mix of colocalization, pleiotropy, shared mechanisms and causal links between traits, and the direction of selective pressure implied by the reproductive and parental mortality correlations is unclear. · stated by the authors
  • Partitioned heritability, genetic correlations, credible sets and gene-set tests use European-ancestry results only, and Danish iPSYCH supplies 14,584 of 20,183 cases, so transfer to other ancestries and ascertainment settings is untested. · noted in this summary

Read this if you work on psychiatric or neurodevelopmental genetics and want the first ADHD risk loci, their heritability and genetic-correlation profile, or a way to meta-analyze case/control with quantitative-trait GWAS.

Citations
2,202
FWCI
35.82
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISRovira '20Zhong '25Demontis '19Liao '19Olfson '23Powell '21
77% semantic matchGenetics
Paula Rovira, Ditte Demontis, Cristina Sánchez‐Mora+51 more·Neuropsychopharmacology·Apr 12, 2020· 111 citations