Cortical gene patterns diverge in adolescent-onset ADHD and conduct problems
Adolescent-onset ADHD and conduct problems were the only mental health groups in this longitudinal analysis that showed significant neurodevelopmental change across five neuroimaging modalities. The finding matters because it separates two externalizing conditions from anxiety, depressive, oppositional defiant, and somatic problems, which did not show the same detectable cortical signal over the two-year follow-up window.
The study drew on longitudinal data from the Adolescent Brain Cognitive Development study and examined structural and functional cortical changes in youth who developed specific mental health problems over two years, compared with healthy controls. Imaging data came from multiple modalities, including structural MRI, resting-state fMRI, and task-based fMRI. The analytic design then linked those spatial brain-change patterns to cortical gene-expression profiles from the Allen Human Brain Atlas using partial least squares regression.
Imaging change concentrated in two externalizing phenotypes
The cortical neurodevelopmental signal was not diffuse across all mental health outcomes. It appeared in the ADHD group and the conduct-problem group, while the other diagnostic categories did not show significant neurodevelopmental alterations in this framework. In practical terms, the study points to a selective developmental architecture rather than a uniform brain-response pattern across adolescent psychiatric symptoms.
A direct brain-to-transcriptome link strengthened the interpretation. PLS analysis showed significant spatial associations between the neurodevelopmental changes and regional gene expression distributions, with Spearman correlations greater than 0.64. In both conditions, the brain-imaging phenotype aligned with molecular maps in a way that suggests these cortical changes sit within a measurable biological context rather than reflecting random developmental noise.
The separation between phenotypes was striking. ADHD and conduct problems shared the broad category of externalizing behavior, yet their transcriptional signatures were not the same. Each condition carried its own cortical molecular context, and the pathways implicated in one did not simply mirror the other.
Mitochondrial biology dominated the ADHD signal
In ADHD, pathway enrichment pointed toward energy metabolism inside the mitochondria. The strongest terms included mitochondrial respiration, oxidative phosphorylation, aerobic electron transport, ATP synthesis, and mitochondrial organization. A negative enrichment signal also involved synaptic protein interaction pathways.
This pattern is biologically coherent. Cortical development in adolescence is energetically demanding, and the brain relies heavily on mitochondrial function to sustain synaptic remodeling, neuronal signaling, and circuit refinement. A transcriptomic signature centered on mitochondrial respiration and ATP production suggests that ADHD-related cortical development may be shaped by altered cellular energy handling rather than by a single neurotransmitter pathway alone.
The negative enrichment of synaptic protein interaction pathways adds another layer. It suggests that the ADHD imaging phenotype may not only track with energy-production machinery, but also with coordinated shifts in synaptic organization and protein-network regulation. In a developmental disorder where attention, impulse control, and executive function are all under pressure, that combination is clinically plausible without overstating causality.
Conduct problems showed an inverse transcriptional alignment
Conduct problems produced a different molecular pattern. All significant pathways were enriched among genes whose cortical expression patterns were inversely aligned with the conduct-problem imaging phenotype. The enriched pathways included translational machinery, ribosomal function, mitochondrial respiration, mitochondrial translation, protein targeting to mitochondria, and broader RNA and protein regulatory processes.
This is not a simple replication of the ADHD result. The conduct-problem signature still touched mitochondrial biology, but it did so alongside a broader set of protein-synthesis and cellular-maintenance pathways. The inverse alignment means the cortical imaging phenotype corresponded to genes expressed in the opposite spatial direction, a detail that points to a qualitatively distinct developmental biology compared with ADHD.
The translational and ribosomal signals matter because they speak to cellular construction and maintenance. During adolescence, cortical circuitry continues to mature through synaptic pruning, myelination-related processes, and ongoing refinement of network architecture. A profile involving translation, ribosomal function, and mitochondrial protein targeting suggests a developmental state in which the cell’s energy supply and protein-production systems are tied closely to the observed cortical phenotype.
What the cross-modal association adds to adolescent psychiatry
This analysis used a rare bridge between brain imaging and transcriptomics. By pairing longitudinal neuroimaging from the ABCD cohort with gene-expression maps from the human cortex, the investigators moved beyond describing where the brain changes were located and toward asking which molecular programs might be spatially linked to those changes.
The result is not a diagnostic test and not a treatment guide. It is a biological map. For ADHD and conduct problems, that map suggests that adolescent-onset externalizing symptoms may arise in the setting of distinct cortical molecular environments, with mitochondrial energy metabolism standing out for ADHD and a broader inverse translational profile defining conduct problems.
The broader implication is conceptual but important. Adolescence is a period of rapid brain reorganization, and psychiatric symptoms that emerge during this window may reflect different layers of developmental disruption depending on the syndrome. Here, the molecular context was detectable only for two externalizing groups, which argues against treating all adolescent mental health problems as biologically interchangeable.
Limits built into the dataset and analytic design
The study’s strength lies in its multimodal, longitudinal design, but the biology was inferred indirectly. Gene-expression data came from the Allen Human Brain Atlas rather than from living participants in the same cohort, so the transcriptomic links are spatially matched associations rather than direct within-person tissue measurements.
The analysis also focused on cortical patterns. Subcortical systems, developmental timing outside the two-year window, and symptom severity gradients were not part of the core abstracted findings. The work therefore identifies robust cortical associations in two conditions, but it does not close the case on mechanism.
Even so, the practical value is clear at the level of hypothesis generation. The study narrows the search for molecular correlates of adolescent-onset externalizing psychopathology and places mitochondria at the center of the ADHD signal while pointing to a distinct, inverse transcriptional architecture in conduct problems. Future work will need to test whether these cortical gene-expression signatures persist across longer follow-up, map onto symptom trajectories, and remain reproducible in independent cohorts.