Abstract
Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder and is associated with structural grey matter differences in the brain. We investigated the genetic background of some of these brain differences in a sample of 899 adults and adolescents consisting of individuals with ADHD and healthy controls. Previous work in an overlapping sample identified three ADHD-related grey matter brain networks located in areas of the superior, middle, and inferior frontal gyrus as well as the cerebellar tonsil and culmen. We associated these brain networks with protein coding genes using a statistical stability selection approach. We identified ten genes, the most promising of which were NR3C2, TRHDE, SCFD1, GNAO1, and UNC5D. These genes are expressed in brain and linked to neuropsychiatric disorders including ADHD. With our results we aid in the growing understanding of the aetiology of ADHD from genes to brain to behaviour.
Introduction
Attention-deficit/hyperactivity disorder (ADHD) is highly heritable, with twin studies suggesting genetic factors account for 70-80% of the variance. However, identifying specific causal genes remains a challenge. Traditional Genome-Wide Association Studies (GWAS) often require massive samples to detect the small effects of individual variants and frequently fail to explain the biological pathways leading to behavioral symptoms.
This study addresses this gap by examining intermediate phenotypes: the brain's physical structure. ADHD is associated with reduced grey matter volume in specific networks, such as the frontal gyrus and cerebellum. By investigating which specific genes drive structural differences in these ADHD-related brain networks, researchers aim to bridge the gap between broad genetic risk and physical brain architecture.
Methods
The researchers employed a multi-stage pipeline designed to maximize statistical power within a sample of 899 participants:
- Brain Network Identification: Independent Component Analysis (ICA) was used to identify three specific grey matter networks previously linked to ADHD symptoms and cognitive tasks, including the inferior frontal gyrus and the cerebellar tonsil.
- Gene-Level Scoring: Quantitative Genetic Scoring (QGS) aggregated genetic data into scores for 12,407 protein-coding genes, significantly reducing the number of variables compared to individual SNP analysis.
- Stability Selection (Randomized Lasso): To ensure robustness, the team ran 5,000 iterations on random subsamples. Genes were prioritized if they were selected consistently (at least 50% of the time), a method that is more resilient to the noise and false positives typical of smaller genetic studies.
Datasets
The primary data were drawn from the NeuroIMAGE and IMpACT Dutch cohorts. The final sample included 899 participants, comprising both individuals with ADHD and healthy controls across adult and adolescent age groups. The 1000 Genomes Project served as a reference population for genetic scoring, while the Human Protein Atlas was used to verify brain expression of the identified genes.
Key Insights and Ideas
- Stability Over Significance: Shifting the focus from traditional p-values to "selection probability" allowed for the detection of consistent biological signals that might be missed by standard methods due to small effect sizes.
- The Mediation Hypothesis: Post-hoc analysis suggests that the effect of these genes on ADHD symptoms is fully mediated by the identified brain networks. This implies these genes influence brain structure, which then manifests as behavioral symptoms.
- Evolutionary Pressure: Most identified associations (8 out of 10) were negative, meaning rarer genetic variants were associated with lower network loadings and higher ADHD symptoms. This supports the hypothesis that evolutionary pressure may select against variants that significantly alter these critical brain networks.
Key Results
The study identified 10 genes with a selection probability $\geq 0.5$. Collectively, these genes explained approximately 1% of the variance in ADHD symptoms—a modest but statistically significant contribution for a complex polygenic trait.
- Top Candidates:
- RELL1: The most consistent gene, linked to the cerebellar network and ADHD symptoms.
- NR3C2: Encodes the mineralocorticoid receptor, involved in stress response.
- TRHDE: Overlaps with TPH2, a gene known for its role in serotonin synthesis and ADHD.
- Brain Expression: 9 out of the 10 genes are expressed in the brain, with GNAO1, TRHDE, and UNC5D showing particularly high expression levels.
Limitations and Future Work
While 899 participants is substantial for neuroimaging, it remains small for genetic studies, necessitating replication in larger samples. Furthermore, the study focused exclusively on protein-coding regions, omitting potentially relevant non-coding regions. Future research should incorporate functional validation, such as cell-based models, to uncover the exact biological mechanisms by which these genes alter grey matter structure.
Summary updated