A peer-reviewed study in Translational Psychiatry has identified three biologically distinct subtypes of ADHD in adolescents, based not on behavioral checklists but on measurable differences in brain structure. The findings challenge a long-standing assumption that ADHD is a single condition with varying symptom intensity, and they raise pointed questions about why standard treatments work well for some patients and fail others. If the results hold up in larger trials, they could eventually reshape how clinicians diagnose and treat one of the most common neurodevelopmental disorders.
Brain Scans Reveal Three Structural Patterns
The research team used semi-supervised deep learning to analyze cortical thickness patterns in adolescents drawn from the NIH’s large ABCD cohort, a longitudinal project following nearly twelve thousand children over time. Rather than sorting participants by the familiar behavioral categories of inattentive, hyperactive-impulsive, or combined presentation, the algorithm grouped them by how their cerebral cortex physically differed from age-matched peers without ADHD. This data-driven approach allowed the model to discover patterns that are not obvious from symptom checklists alone, focusing instead on measurable neuroanatomical differences.
The analysis produced three distinct structural clusters with abnormal cortical thickness: an under-developed subtype with thinner cortices across multiple regions, an over-developed subtype with thicker cortices, and a mixed subtype showing both patterns in different brain areas. What makes these groupings clinically significant is their link to treatment response. Adolescents in the under-developed group showed the poorest outcomes on stimulants, which remain the first-line pharmacological option in most practice guidelines. That pattern suggests a sizable fraction of young people diagnosed with ADHD may be receiving medication that is structurally mismatched to their brain profile, offering a concrete biological explanation for the uneven benefits that families and clinicians have reported for decades.
Why Symptom-Based Labels Fall Short
The current diagnostic framework, as summarized by the CDC, recognizes three presentation types based on which behaviors are most prominent: mostly inattentive, mostly hyperactive-impulsive, or a combined pattern of both. These symptom-based categories are useful for screening and for communicating with schools and families, but they describe outward behavior rather than underlying biology. Two children might both be labeled “combined presentation” while having very different cortical thickness profiles, different cognitive strengths and weaknesses, and different responses to medication, all of which are invisible in the current manual-based system.
This is not the first time data-driven methods have fractured ADHD into biologically or cognitively distinct groupings. An earlier analysis using latent class methods on neuropsychological testing identified three separable cognitive subgroups, characterized by differences in executive control, reward processing, and timing functions. Separately, quantitative EEG work has pointed to multiple brainwave-based profiles, and now the cortical thickness findings add a structural dimension to this picture. The convergence of independent approaches (cognitive testing, electrophysiology, and MRI-based imaging), all resolving ADHD into multiple clusters, strengthens the case that heterogeneity is genuine and measurable, not just a statistical artifact of noisy symptom reports.
Genetic Evidence Points to Multiple Biological Pathways
The subtyping work gains additional weight when viewed alongside genetics. A large genome-wide association study in Nature Genetics identified dozens of common risk loci for ADHD and confirmed that the condition is highly polygenic: hundreds or even thousands of variants each contribute a small amount of risk. The same analysis found substantial genetic overlap between ADHD and other psychiatric and cognitive traits, including mood disorders and educational attainment, which helps explain why ADHD so often co-occurs with anxiety, depression, and learning difficulties. If genetic liability is dispersed across many pathways, it is unlikely that all individuals with ADHD would show the same downstream brain changes.
More recent exome sequencing has begun to identify rare, higher-impact variants that tie ADHD risk even more directly to neuronal development, synaptic function, and comorbidity patterns. In this context, the cortical thickness subtypes look less like an anomaly and more like an expected outcome of diverse genetic routes converging on a shared behavioral label. Different constellations of common and rare variants could plausibly drive the under-developed, over-developed, and mixed cortical profiles observed in the Translational Psychiatry study. That genetic backdrop also raises the possibility that each structural subtype may have partially distinct trajectories for academic performance, emotional regulation, and vulnerability to additional psychiatric conditions, even when outward ADHD symptoms appear similar.
What This Means for Diagnosis and Treatment
For families and clinicians, the immediate implications are cautious but important. The new findings do not mean that every child with ADHD should suddenly receive an MRI, nor do they invalidate existing treatments that help many patients. Instead, they highlight that ADHD, as currently defined, is likely an umbrella term covering several related but distinct neurodevelopmental profiles. If future work replicates these subtypes in larger and more diverse samples, neuroimaging or less expensive proxy measures, such as refined cognitive testing or EEG markers, could eventually guide treatment selection before a long cycle of trial and error. For example, a patient whose profile matches the under-developed cortical subtype might be counseled that stimulants are less likely to be effective and that non-stimulant medications or intensive behavioral interventions should be considered earlier.
On the research side, the structural subtypes point toward a more stratified model of clinical trials. Rather than testing one medication across a broad, heterogeneous ADHD sample, future studies could enroll participants according to cortical or cognitive subtype and then examine whether specific treatments work better for particular groups. A logical next step, suggested by the existing evidence but not yet tested in randomized trials, would be to combine neuroimaging-based clustering with polygenic risk scores and detailed cognitive assessments. Such multimodal profiling could clarify whether the same biological subtype appears consistently across brain structure, genetics, and behavior, and whether that subtype predicts not only medication response but also long-term outcomes like school completion and mental health in adulthood.
Toward a More Precise Model of ADHD
Moving from a single, broad ADHD category to a set of biologically grounded subtypes would align this condition with trends already underway in other areas of medicine. Oncology, for instance, now routinely distinguishes cancers by molecular markers that guide targeted therapies, even when tumors arise in the same organ. A similar shift in ADHD, from surface-level symptoms to measurable brain and genetic markers, could eventually support more personalized care, reduce unnecessary exposure to ineffective medications, and improve the odds that each child receives an intervention matched to their specific neurodevelopmental profile. That transition will require not only replication of current findings but also practical tools that clinicians can use outside of research settings.
For now, the key message is that ADHD is more biologically diverse than its diagnostic label suggests. The identification of three cortical thickness subtypes, their differing responses to stimulant treatment, and their consistency with earlier cognitive and genetic work all point to a future in which “ADHD” is less a single disorder than a family of related conditions. As larger longitudinal datasets mature and analytic methods improve, researchers will be able to track how these subtypes evolve from childhood into adulthood and how they intersect with environmental factors such as stress, sleep, and education. That knowledge, in turn, could help families and clinicians move beyond one-size-fits-all strategies toward interventions that are both more effective and more humane.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
