A new study based on MRI data from a large cohort of children diagnosed with ADHD and controls has identified three distinct ADHD “biotypes.” These biotypes indicate various ways in which specific and contrasting patterns of variation in brain structure and function manifest across a large set of ADHD patients. Such information can potentially inform future efforts to create more effective patient-specific therapies.

The results, validated in the same study by testing them against data from a large independent cohort of young people with ADHD, provide insights into the complexity of ADHD, in terms of how it manifests in the brain, affecting neurodevelopment, and how it produces distinct clusters of symptoms in different individuals.

Like many other neuropsychiatric disorders, ADHD is “characterized by considerable clinical heterogeneity,” note authors of the new study, adding that this heterogeneity—the diversity of symptoms experienced across the total population of patients—“extends beyond” the diagnostic definitions offered in the DSM-5 manual used by mental health professionals. Typically, ADHD patients are diagnosed in a binary way, their symptoms understood to be predominantly either “inattentive” or “hyperactive/impulsive.” Although this schema has been useful, at the analytical level it is widely thought to oversimplify the presumably diverse neurobiological mechanisms underlying the disorder. Efforts to define robust ADHD subtypes, meanwhile, are growing in sophistication.

Some past efforts to define ADHD subtypes have sought to identify characteristic symptom combinations. While this is done for the sake of clarity and simplicity in order to guide therapy, such efforts haven’t tried to account for neurobiological variations among individuals—between those with and without ADHD, but also variations between patients. Such “inter-individual” variations are not evident on the surface, and their impact on how the disorder manifests is unknown.

The new study, which was led by Nanfang Pan, M.D., Ph.D., and Qiyong Gong, M.D., Ph.D., both of West China Hospital of Sichuan University, People’s Republic of China, took a different approach, based mainly on advanced modeling algorithms that utilize data from MRI scans and consider them in the light of a variety of neurobiological and clinical data from various databases.

Three BBRF grantees were members of the team: Melissa P. DelBello, M.D., 2006 BBRF Independent Investigator and 2004 and 2001 Young Investigator; Robert K. McNamara, M.D., 2010 BBRF Independent Investigator and 2000 and 1998 Young Investigator; and Manpreet K. Singh, M.D., 2016 BBRF Independent Investigator and 2008 Young Investigator.

Data from 1,154 young people made up the “discovery” cohort—those whose data was used to arrive at new insights, prior to the portion of the study devoted to validating the results in an independent group. The discovery cohort consisted of 446 children with ADHD, about three-fourths male and 11.5 years old, on average; and 708 controls, 60% male and 11 years old, on average. The validation cohort was made up of 554 children with ADHD, two-thirds male and about 11 years old, and 123 controls, about 57% male and 10 years old.

The main method of analysis used is called normative modeling, which uses mathematical analysis of (in this case) imaging data revealing brain structure to directly address the question of how different individuals with a given illness manifest it in different ways. Some of the variations between patients, as well as between patients and controls, are ultimately considered within a normal range, while others are considered atypical. The task then is to try to discern patterns and correlate these with different clusters of symptoms experienced in those with ADHD whose scans are being analyzed.

What the team calls their “brain-first approach”—seeking patterns within the brain rather than in outward symptoms experienced by patients—yielded comprehensive neuroimaging-derived characterization of “subtle network-level deviations” in ADHD brains. The analysis identified the brain’s orbitofrontal cortex as an area that “may serve as a fundamental network anchor point in ADHD pathophysiology that transcends biotype boundaries.” This, they said, aligns with current understanding of the orbitofrontal cortex’s role in orchestrating the transition between impulsive and reflective behaviors in the context of goal-directed actions.

As for the three ADHD biotypes that the team identified: each was linked with specific abnormalities in neural circuits. Biotype 1 (“severe, combined with emotional regulation”) was associated with dysregulation in fronto-striatal circuitry. In Biotype 2 (“predominantly hyperactive-impulsive”), alterations in a circuit connecting the anterior cingulate cortex and the palladium “may drive hyperactive/impulsive behaviors through a dysregulated action-mode network.” In Biotype 3 (“predominantly inattentive”), alterations in the superior frontal gyrus “may selectively impair sustained attention” via interaction with the default mode network.

The common underlying pathology across all three biotypes, the team stressed, was dysfunction of the orbitofrontal cortex. Deviations in specific regions appeared to shape symptoms specific to individuals. This, they said, may explain why certain neural signatures have consistently appeared in past neuroimaging studies comparing people with ADHD vs. controls while other signatures may “emerge only through biotype stratification.”

In the full analysis provided in the paper, the three distinct MRI-derived ADHD biotypes were validated in the independent cohort. In the team’s view, circuit-specific mechanisms identified in analyses like theirs “may provide potential targets for stratified interventions tailored to the unique network dysfunction profile of each biotype.”

They noted that future research may reveal that the biotypes they identified, assuming they are further validated, may prove to be distinct points along a broad, underlying continuum of alterations in the brain that attend ADHD rather than “distinct diagnostic entities.”