Structural Brain Network Modeling Identifies Three ADHD Biotypes

The Challenge of Heterogeneity in Neurodevelopmental Disorders

Attention-deficit/hyperactivity disorder (ADHD) presents with immense clinical variability, complicating efforts to establish consistent neurobiological markers. While structural neuroimaging has attempted to map brain maturation in neurodevelopmental conditions, large-scale analyses often yield heterogeneous results or fail to find uniform differences between patients and healthy controls [1]. This inconsistency largely stems from profound inter-individual variability in brain anatomy and functional organization, which traditional group-level comparisons tend to obscure [2]. To overcome these limitations, the field of precision psychiatry has increasingly focused on individualized prediction models that account for unique structural deviations [3]. A recent study offers fresh insights by mapping how individual brain networks deviate from typical developmental baselines, providing a framework to understand the biological diversity underlying clinical symptoms.

Mapping Brain Networks Against a Developmental Baseline

To identify robust stratification markers for children with ADHD, researchers conducted a case-control study leveraging multisite cross-sectional neurodevelopmental datasets, which included a longitudinal follow-up cognitive assessment for a subset of participants. Analyzed from November 2023 to January 2025, the data allowed investigators to determine whether normative modeling (a statistical technique establishing a baseline of typical brain development to flag individual deviations) of topological properties derived from brain morphometry similarity networks (maps detailing how different brain regions structurally relate to one another) could reliably stratify patients. The initial discovery cohort comprised 446 children with ADHD (mean age 11.5 years, standard deviation 2.6; 339 male, 76.0 percent) and 708 controls (mean age 11.0 years, standard deviation 2.3; 429 male, 60.6 percent).

Using structural imaging data, the researchers constructed these networks and developed normative models for three specific topological metrics: degree centrality, nodal efficiency, and participation coefficient. For the practicing physician, these metrics collectively measure how efficiently different brain regions connect, coordinate, and share information. By quantifying how each patient's brain network deviated from the healthy baseline across these three metrics, the team applied semisupervised clustering, an algorithm that groups patients based on both their biological imaging data and their clinical labels. This approach allowed the researchers to delineate putative biotypes and examine their distinct clinical profiles. To ensure these findings were not isolated to the initial group, model generalizability was assessed with external validation performed in an independent transdiagnostic cohort. This validation cohort included an additional 554 children with ADHD (mean age 10.1 years, standard deviation 2.8; 372 male, 67.1 percent) and 123 controls (mean age 10.1 years, standard deviation 3.0; 70 male, 56.9 percent). By confirming the network deviations in this separate group, the study provides clinicians with a biologically grounded framework for understanding the structural variations that drive different ADHD presentations.

Orbitofrontal Cortex Deviations and Three Distinct Biotypes

When comparing the structural imaging data, the researchers found that children with ADHD exhibited atypical hub organization across all 3 topological metrics compared to healthy controls. This indicates that the central nodes responsible for coordinating communication across brain networks were structurally altered. Specifically, the most significant case-control differences were primarily localized to a covarying multimetric component in the orbitofrontal cortex, a region critical for decision-making and impulse control.

Rather than treating the disorder as a single structural entity, the analysis revealed that three ADHD biotypes emerged, directly linking specific clinical presentations to distinct neural circuit deviations. The first biotype was severe-combined with emotional dysregulation, featuring widespread medial prefrontal cortex-pallidum alterations (n = 142). For clinicians, this maps onto patients who present with both severe inattention and hyperactivity complicated by pronounced mood lability, driven by structural deviations in pathways connecting the frontal lobe to the basal ganglia. The second biotype was predominantly hyperactive/impulsive, featuring anterior cingulate cortex-pallidum circuit alterations (n = 177). This structural variation in the anterior cingulate, an area involved in error detection and motor control, aligns with the physical restlessness and impulsivity seen in these patients. The third biotype was predominantly inattentive, featuring superior frontal gyrus alterations (n = 127), pointing to disruptions in a region essential for sustained attention and working memory. Ultimately, the researchers noted that each biotype was characterized by distinct clinical profiles and longitudinal trajectories. This suggests that these structural markers could eventually help physicians predict disease course and tailor interventions based on a patient's specific neurobiological subtype.

Neurochemical Correlates and Clinical Implications

To understand the underlying biology of these structural variations, the researchers investigated how the brain profiles of the biotypes were contextualized in terms of their neurochemical and functional correlates using large-scale databases. By cross-referencing the structural imaging data with established maps of brain chemistry and activity, the investigators found that the neural profiles of each biotype showed distinct neurochemical and functional correlates. For clinicians, this suggests that the three structural subtypes may be driven by different neurotransmitter systems or functional network disruptions, potentially explaining why patients respond differently to standard pharmacological treatments. To ensure these patterns were reliable, the investigators tested their model on an independent group of patients. The core findings were replicated in the validation cohort, demonstrating robust generalizability. This confirmation in a separate group indicates that these neurobiological signatures are stable across different patient populations rather than being artifacts of the initial sample.

In daily practice, ADHD is characterized by considerable clinical heterogeneity, and existing classification frameworks constrain the development of neurobiologically informed subtyping approaches. Current diagnostic manuals rely heavily on behavioral symptom checklists, which group biologically diverse patients into broad categories and complicate efforts to match specific neural deficits with targeted therapies. The researchers concluded that the integration of normative modeling with semisupervised clustering provided both dimensional and categorical insights into ADHD heterogeneity. By combining normative modeling (a statistical approach that quantifies how a single patient's brain deviates from a healthy developmental baseline) with semisupervised clustering (an algorithm that groups these biological deviations alongside clinical symptom labels), the study captures both the continuous spectrum of symptom severity and distinct biological categories. Ultimately, the identification of 3 distinct ADHD biotypes with unique clinical-neural profiles lays the groundwork for personalized management. For the practicing physician, these findings offer a biological foundation for moving beyond trial-and-error prescribing toward tailored interventions based on a child's specific neural circuitry.

References

1. Gharehgazlou A, Freitas C, Ameis S, et al. Cortical Gyrification Morphology in Individuals with ASD and ADHD across the Lifespan: A Systematic Review and Meta-Analysis. Cerebral Cortex. 2020. doi:10.1093/cercor/bhaa381

2. Forkel SJ, Friedrich P, Schotten MTD, Howells H. White matter variability, cognition, and disorders: a systematic review. Brain Structure and Function. 2021. doi:10.1007/s00429-021-02382-w

3. Pablo GSD, Studerus E, Vaquerizo‐Serrano J, et al. Implementing Precision Psychiatry: A Systematic Review of Individualized Prediction Models for Clinical Practice. Schizophrenia Bulletin. 2020. doi:10.1093/schbul/sbaa120