Differences in the genetic architecture of common and rare variants in childhood, persistent and late-diagnosed attention-deficit hyperactivity disorder

Abstract

Attention-deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder with onset in childhood (childhood ADHD); two-thirds of affected individuals continue to have ADHD in adulthood (persistent ADHD), and sometimes ADHD is diagnosed in adulthood (late-diagnosed ADHD). We evaluated genetic differences among childhood (n = 14,878), persistent (n = 1,473) and late-diagnosed (n = 6,961) ADHD cases alongside 38,303 controls, and rare variant differences in 7,650 ADHD cases and 8,649 controls. We identified four genome-wide significant loci for childhood ADHD and one for late-diagnosed ADHD. We found increased polygenic scores for ADHD in persistent ADHD compared with the other two groups. Childhood ADHD had higher genetic overlap with hyperactivity and autism compared with late-diagnosed ADHD and the highest burden of rare protein-truncating variants in evolutionarily constrained genes. Late-diagnosed ADHD had a larger genetic overlap with depression than childhood ADHD and no increased burden in rare protein-truncating variants. Overall, these results suggest a genetic influence on age at first ADHD diagnosis, persistence of ADHD and the different comorbidity patterns among the groups.

Figure 1.

Genetic correlations of ADHD subgroups with major psychiatric disorders and phenotypes representing domains with high genetic correlation with ADHD, estimated by linkage disequilibrium (LD) score regression. Results from genome-wide association analyses of ADHD subgroups was used including childhood (N=14,878 individuals), late-diagnosed (N=6,961 individuals) and persistent (N=1,473 individuals) attention-deficit hyperactivity disorder (ADHD) against 38,303 control individuals. Error bars (horizontal lines) represent standard errors. *Indicates a significant difference (after Bonferroni correction) with a two-sided P-value lower than P=0.0013 in the genetic correlation observed for childhood ADHD compared with late-diagnosed ADHD.

Figure 2.

Results from polygenic score (PGS) analysis demonstrating the association of PGS with childhood (N=14,878 individuals), persistent (N=1,473 individuals) and late-diagnosed attention-deficit hyperactivity disorder (ADHD) (N=6,961 individuals). PGSs for psychiatric disorders: autism spectrum disorder (ASD), depression (MDD), schizophrenia (SZ), bipolar disorder (BD), anorexia, obsessive compulsive disorder (OCD), alcohol use disorder (AUD) and cannabis use disorder (CUD). PGSs for five phenotypes representing domains highly correlated with ADHD: educational attainment (EA), insomnia, mother’s age at death and age of first birth (AOB). On the y-axis is the beta from multi-nominal regression against controls (N=38,303 individuals); error bars (vertical lines) represent standard errors (see also Supplementary Table 10). Significant pair-wise comparisons (after Bonferroni correction) with a two-sided P-value lower than P=0.0013 are given in the right corner of the header: ‘C’ indicates childhood ADHD, ‘L’ indicates late-diagnosed ADHD, ‘P’ indicates persistent ADHD and the direction of the difference in the betas is given by ‘>‘ (see also Supplementary Table 9).

Figure 3.

The load of rare protein truncating variants (rPTVs) and rare synonymous (rSyn) in three gene-sets (1) ‘Highly constrained’, genes intolerant to loss-of-function mutations (pLI > 0.9); (2) ‘Denovo’, highly constrained genes which in another study has been found to be enriched with de novo mutations in individuals with neurodevelopmental disorders; and (3) ‘Low constrained’, genes tolerant to loss-of-function mutations (pLI score <0.1). The y-axis represents the beta from multiple logistic regression of childhood (N=4,987 individuals), persistent (N=748 individuals) and late-diagnosed (N=1,915 individuals) ADHD with comparison to controls (N=8,649 individuals), error bars (vertical lines) represent the standard error. For pair-wise comparisons of attention-deficit hyperactivity disorder (ADHD) subgroups, see Supplementary Table 12.