📄 psychiatry and clinical psychology

Lifespan brain structural variation reveals shared organization across mental health conditions

By analyzing over 22,000 individuals across six continents, this study demonstrates that structural brain deviations in diverse mental health conditions follow a systematic organization constrained by the brain's intrinsic architecture, revealing shared transdiagnostic vulnerabilities that reconcile individual variability with common neurobiological mechanisms.

Original authors: Hettwer, M. D., Saberi, A., Shafiei, G., Manoli, A., De Boer, A. A., Alnaes, D., Alonso, P., Arango, C., Assaf, M., Avram, M., Balachander, S., Banaj, N., Basgöze, Z., Batistuzzo, M. C., Bauduin, S.

Published 2026-08-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hettwer, M. D., Saberi, A., Shafiei, G., Manoli, A., De Boer, A. A., Alnaes, D., Alonso, P., Arango, C., Assaf, M., Avram, M., Balachander, S., Banaj, N., Basgöze, Z., Batistuzzo, M. C., Bauduin, S. E. E. C., Benedetti, F., Bertolin, S., Besteher, B., Biagi, L., Blair, R. J., Blair, K., Bölte, S., Borgwardt, S., Bosco, P., Brambilla, P., Bravi, B., Brennan, B. P., Bruin, W. B., Busatto, G. F., Cairns, M. J., Calderoni, S., Calhoun, V., Calvo, R., Cano, M., Carr, V. J., Carruthers, S. P., Caruana, G. F., Caseras, X., Catts, S. V., Chi, I.-J., Cobia, D., Colombo, F., Couto, M. B., Crespo-Facor

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The human brain is not a static organ; it changes continuously from childhood through old age, growing and refining its structure in ways that are unique to each person. When scientists look at mental health conditions like schizophrenia, autism, or depression, they often find a confusing picture: people with the same diagnosis can have very different brain structures, while people with different diagnoses can share similar brain features. This makes it hard to understand what is truly happening inside the mind. For years, researchers have tried to find a common thread, asking whether these diverse conditions might actually follow a shared, underlying pattern that is hidden beneath the surface of individual differences. To answer this, scientists have developed a way to map the brain's normal growth, creating a kind of standard chart that shows how brain features typically change with age and sex. By comparing an individual's brain to this standard, they can measure exactly how much that person's brain has drifted from the expected path.

A massive international team of researchers, working through a global collaboration known as the ENIGMA consortium, decided to use this approach to look at the brains of more than 22,000 people. They gathered data from 172 different scanning sites across six continents, including over 10,000 individuals diagnosed with one of six major mental health conditions and nearly 12,000 people without any such diagnosis. The conditions studied were schizophrenia, autism, bipolar disorder, obsessive-compulsive disorder, generalized anxiety, and major depression. The researchers focused on three specific measurements of brain structure: the thickness of the brain's outer layer, the surface area of that layer, and the volume of deeper brain structures. Instead of just comparing averages between groups, they calculated how much each person's brain differed from the typical lifespan trajectory for their age and sex. This allowed them to see the unique "fingerprint" of deviation for every single person in the study.

What the team discovered was that these deviations were not random. Even though the specific brain regions affected varied from person to person, the differences tended to cluster along two major, predictable pathways that run through the brain's architecture. One of these pathways stretched from the emotional and memory centers of the brain toward the higher-level thinking areas at the top and back. The other pathway ran from the areas that handle basic senses and movement toward the complex association areas that integrate information. These pathways mirror the brain's natural organization, reflecting how different regions are connected to one another and how they mature over time. The researchers found that the brain's intrinsic wiring acts like a scaffold, guiding where and how structural changes occur, regardless of the specific diagnosis a person carries.

The study also revealed that while the brain changes were subtle on average, some individuals showed extreme deviations in specific regions. These extreme changes were rare and highly individualized, meaning two people with the same diagnosis might have extreme changes in completely different parts of the brain. However, even these rare, extreme changes were not scattered randomly. They tended to occur in regions that are densely connected to many other parts of the brain, forming a network of co-occurring alterations. This suggests that when the brain is pushed to its limits, the changes still respect the brain's natural connectivity, concentrating in the hubs that link different systems together.

Perhaps the most striking finding was how much overlap there was between the different diagnoses. When the researchers compared the brain structures of people with different conditions, they found that the distributions of their brain deviations were remarkably similar, often overlapping by more than 90 percent in many brain regions. The differences between the diagnoses were not found in entirely separate parts of the brain, but rather in the degree and direction of the changes along the same shared pathways. The conditions were most distinct from one another in the brain's association areas, which are involved in complex thinking and social processing, but they were nearly indistinguishable in the regions handling basic senses and movement. This indicates that mental health conditions are not distinct, separate categories with unique biological signatures, but rather represent different points along a shared spectrum of brain variation.

The researchers also looked at whether these brain patterns related to how sick a person felt. They found that for several conditions, the more severe a person's symptoms were, the stronger their brain deviations were along these shared pathways. This connection held true across different diagnoses, suggesting that the severity of the illness is tied to the magnitude of the structural change along these common axes. The study did not find that the brain changes were caused by the specific label of a diagnosis, but rather that the diagnosis reflects a particular expression of a broader, systematic variation in brain structure.

This work challenges the traditional view that mental health conditions are distinct diseases with unique biological causes. Instead, it suggests that the brain's own architecture constrains how mental health conditions manifest. The brain's natural organization, including how its regions are connected and how they develop over a lifetime, creates a limited set of ways in which it can be altered. When a person develops a mental health condition, their brain changes along these pre-existing lines of vulnerability. This does not mean that all conditions are the same, but rather that they share a common neurobiological foundation. The findings provide a new framework for understanding mental illness, moving away from rigid categories and toward a more integrated view that recognizes the shared biological roots of diverse symptoms. By understanding these shared patterns, scientists hope to eventually develop better ways to predict risk and tailor treatments based on an individual's specific position within this shared landscape of brain variation.

Technical Summary: Lifespan Brain Structural Variation Reveals Shared Organization Across Mental Health Conditions

Problem Statement

Understanding the neurobiological basis of neurodevelopmental and psychiatric conditions (NDPCs) is hindered by two central paradoxes: individuals with the same diagnosis often exhibit divergent brain alterations, while different diagnoses frequently share overlapping neurobiological patterns. Furthermore, psychiatric diagnoses often co-occur, suggesting shared risk factors and mechanisms that transcend traditional categorical boundaries. While dimensional frameworks (e.g., HiTOP, RDoC) attempt to address this heterogeneity, it remains unknown whether the vast variability in brain structural alterations across individuals and diagnoses follows a systematic, shared organizational principle rooted in the brain's intrinsic architecture.

Methodology

The study leveraged the largest transdiagnostic psychiatric neuroimaging sample to date, aggregated through the ENIGMA consortium.

Sample: The dataset included 10,132 individuals diagnosed with schizophrenia (SCZ), autism spectrum disorder (ASD), bipolar disorder (BP), obsessive-compulsive disorder (OCD), generalized anxiety disorder (ANXG), or major depressive disorder (MDD), alongside 11,998 reference participants (RC) without psychiatric or neurological diagnoses. Participants ranged from 5 to 80 years old and were scanned across 172 sites on six continents.
Imaging Metrics: Cortical thickness (CT), surface area (SA), and subcortical volumes (SV) were derived using FreeSurfer and summarized across 68 cortical and 14 subcortical regions (Desikan atlas).
Normative Modeling: The authors employed hierarchical Bayesian regression (HBR) using the PCNtoolkit to model lifespan trajectories of brain structure based on age, sex, and site. A reference cohort (RC_train, n=9,532) was used to establish expected trajectories, while a held-out test set (RC_test, n=2,466) served as an unbiased benchmark. Individual deviations from these normative trajectories were quantified as z-scores.
Analytic Approaches:
- Principal Component Analysis (PCA): Applied to absolute deviation scores to identify shared cortical axes of co-variation.
- Connectome Alignment: Deviation axes were correlated with normative connectome hubs (derived from Human Connectome Project data) and the sensorimotor-to-association (S-A) axis of cortical hierarchy.
- Extreme Deviation Analysis: Identified regions where individuals exhibited deviations exceeding the 2.5th/97.5th percentiles ( $|z| > 1.96$ ) and mapped "co-alteration hubs" where extreme deviations co-occur within connected networks.
- Distribution Overlap: Quantified the shared area under density functions of deviation scores across diagnostic groups to measure transdiagnostic similarity.
- Symptom Correlation: Linked individual deviation scores along identified axes to symptom severity scores (e.g., PANSS, Y-BOCS, ADOS-CSS).

Key Contributions

Systematic Organization of Deviations: The study demonstrates that structural deviations across diverse NDPCs are not random but converge along specific cortical axes that recapitulate the brain's intrinsic organization (connectome hubs, maturational timing, and cytoarchitectonic diversity).
Transdiagnostic Convergence: It provides evidence that brain structural alterations across major psychiatric diagnoses overlap substantially, suggesting that diagnostic categories capture different expressions of a shared underlying neurobiological architecture rather than distinct, categorical signatures.
Network Embedding of Extreme Variability: Even rare, highly individualized extreme deviations are not scattered randomly; they are systematically embedded within the brain's typical connectome organization, specifically clustering in densely connected network hubs.
Symptom Scaling: The magnitude of deviations along these shared axes correlates with symptom severity across multiple diagnoses, independent of the specific diagnostic label.

Results

Cortical Axes of Deviation: PCA revealed two primary axes of structural deviation:
- PC1: Spanned from (para)limbic to dorsal heteromodal cortex. This axis aligned with normative connectome hubs and captured the magnitude of transdiagnostic group shifts.
- PC2: Contrasted unimodal (idiotypic) with ventral heteromodal/(para)limbic cortex. This axis aligned with the sensorimotor-to-association axis and captured within-diagnosis variability.
Diagnostic Specificity: While all NDPCs followed these axes, their expression varied. SCZ, ASD, and OCD showed stronger deviations along PC1 (towards heteromodal poles), whereas ANXG showed an opposite trend. ASD and OCD exhibited greater variability along PC2.
Extreme Deviations: Extreme deviations ( $|z| > 1.96$ ) were rare, with at most 9.13% of individuals in a specific diagnosis sharing an extreme deviation in the same region. However, these extreme deviations preferentially occurred in regions that are densely connected (hubs), particularly in the dorsal prefrontal and parietal cortex.
Transdiagnostic Overlap: Distribution overlap analysis showed high similarity between diagnostic groups (79.2% to 98.5% overlap across the cortex). Overlap was highest in paralimbic, visual, and motor areas, while differentiation between diagnoses was more pronounced in the heteromodal cortex.
Symptom Severity: In ASD, SCZ, and MDD, higher symptom severity was positively associated with stronger deviations along the PC1 axis.

Significance and Claims

The paper claims to provide large-scale evidence that structural deviations across NDPCs are systematically constrained by the brain's intrinsic architecture. The authors argue that the neurobiology of mental health diagnoses is better understood as graded variation within a shared organizational architecture rather than a set of categorically distinct, diagnosis-specific signatures.

Key implications highlighted by the authors include:

Reconciliation of Variability: This framework reconciles the paradox of high within-diagnosis heterogeneity with transdiagnostic similarities by positing that individual deviations unfold along pre-existing degrees of freedom defined by the brain's connectome and maturational gradients.
Mechanistic Insight: The findings suggest that susceptibility to psychopathology may be anchored in neurobiological properties (e.g., metabolic demand of hubs, plasticity of late-maturing cortex) that are systematically arranged within the cortical hierarchy.
Shift in Perspective: The results motivate a shift from region- and diagnosis-centric perspectives toward a systems-level understanding of mental health, supporting dimensional models of psychopathology.
Limitations: The authors modestly note that while the study identifies shared patterns, it is cross-sectional and limited by the phenotypic data available from primary studies (e.g., lack of harmonized medication dosages or detailed comorbidity data). They also acknowledge that extreme deviations, while network-embedded, are too heterogeneous to serve as standalone biomarkers for specific diagnoses.

The study concludes that understanding mental health variation as rooted in the brain's intrinsic architecture offers a principled framework for investigating shared mechanisms of risk and resilience that transcend traditional diagnostic boundaries.