TL;DR
This study uses structural covariance analysis to investigate brain organization in conduct disorder (CD), finding distinct patterns between childhood-onset (CO-CD) and adolescence-onset (AO-CD) subtypes. The findings suggest that CO-CD is associated with widespread increases in brain connectivity, while AO-CD shows a decrease, offering new insights into the neurodevelopmental origins of these disorders.
What Problem Does This Paper Address?
Understanding the neurobiological underpinnings of conduct disorder (CD) is crucial, especially given its significant impact on individuals and society. CD is not a monolithic condition; it's characterized by distinct subtypes, primarily childhood-onset (CO-CD) and adolescence-onset (AO-CD), which are hypothesized to have different developmental trajectories and etiologies. While previous research has identified structural brain differences in CD, it has often yielded inconsistent results, particularly when comparing these subtypes. This paper addresses the challenge of characterizing the coordinated development of brain regions in CD, moving beyond examining individual regions in isolation. Structural covariance, which measures how the size or thickness of different brain regions covary, is a powerful tool for this, as it reflects shared developmental processes. The authors aim to determine if structural covariance patterns can differentiate between CO-CD and AO-CD, and whether these patterns support the neurodevelopmental origins hypothesis for both subtypes.
Methods
The study employs a technique called Mapping Anatomical Correlations Across Cerebral Cortex (MACACC) to analyze structural covariance. This method quantifies interregional correlations in cortical thickness. Here's a breakdown of the key steps:
- Cortical Thickness Measurement: High-resolution structural MRI scans were acquired. The FreeSurfer software was used to segment the brain and parcellate the cortex into 34 regions of interest (ROIs) per hemisphere (totaling 68 ROIs). For each ROI, an average cortical thickness value was extracted.
- Structural Covariance Calculation: For each participant, cortical thickness data were mean-centered across regions to account for scaling effects. Then, interregional correlations (Pearson's R-values) were computed between all pairs of ROIs.
- Group Comparisons: To test for differences in structural covariance between groups (Healthy Controls (HCs), CO-CD, AO-CD), a novel statistical approach was used. This involved building a null distribution using a random sampling technique and correcting for multiple comparisons using a false-discovery-rate (FDR) procedure.
- Covariate Control: Analyses were conducted while controlling for potential confounds such as age, intelligence quotient (IQ), and attention-deficit/hyperactivity disorder (ADHD) symptoms. Estimated total intracranial volume (eTIV) was also regressed out in a separate analysis to rule out group differences in overall brain size.
The MACACC approach is highlighted as being particularly sensitive to detecting widespread changes in brain organization, potentially revealing abnormalities that univariate analyses might miss.
Datasets
The study utilized two independent datasets:
- Cambridge Sample: 58 male youths with CD and 25 age- and sex-matched healthy controls (HCs). Participants were aged 16-21 years.
- Southampton Sample: 37 male youths with CD and 32 age- and sex-matched healthy controls (HCs). Participants were aged 13-18 years.
These independent samples were crucial for replicating findings and ensuring the robustness of the results, despite variations in MRI scanners and acquisition parameters.
Key Insights and Ideas
- Subtype Specificity: The most striking finding is the clear differentiation in structural covariance patterns between CO-CD and AO-CD. This suggests that these subtypes, while both falling under the umbrella of CD, have distinct neurodevelopmental trajectories.
- Widespread Brain Alterations: Structural covariance abnormalities in CD appear to be more diffuse and widespread across the cortex than previously thought, affecting multiple networks rather than isolated regions. This aligns with the idea of global disruptions in brain maturation.
- Developmental Trajectories: The observed patterns of increased (CO-CD) and decreased (AO-CD) interregional correlations support the hypothesis that both subtypes have neurodevelopmental origins, but potentially through different mechanisms or at different developmental stages.
- MACACC's Sensitivity: The study underscores the utility of structural covariance analysis (MACACC) in revealing subtle yet significant differences in brain organization, particularly in complex neurodevelopmental disorders, that might be missed by other neuroimaging methods.
- Clinical Relevance: The distinct structural covariance profiles could potentially serve as biomarkers for distinguishing between CO-CD and AO-CD, aiding in more precise diagnosis and classification.
Key Results
- Number of Correlations:
- CO-CD: Displayed a significantly higher number of significant interregional correlations in cortical thickness compared to HCs and AO-CD participants in both samples.
- AO-CD: Displayed significantly fewer interregional correlations than HCs and CO-CD participants.
- Strength of Correlations: Significant group differences in the strength of interregional correlations were observed across numerous regions in both samples.
- Replication: The patterns of findings were highly consistent across the two independent Cambridge and Southampton samples, even with different MRI parameters.
- Covariate Independence: The observed group differences in structural covariance remained significant even after controlling for IQ and ADHD symptoms, suggesting these are not solely driven by these factors.
- Overlap: A substantial overlap in significant interregional correlations was found between the two samples for each group, further supporting the robustness of the findings.
Limitations and Future Work
- Retrospective Onset Classification: The classification of CO-CD and AO-CD relied on retrospective accounts of age-of-onset, which can be subject to recall bias. While efforts were made to mitigate this, it remains a limitation.
- Sex Differences: The study sample was exclusively male. Future research should investigate whether these findings generalize to females with CD, as sex differences in antisocial behavior and brain development are known.
- MRI Acquisition: The use of a single excitation in the MRI anatomical pulse sequences might have reduced signal-to-noise ratio and accuracy.
- Subcortical Structures: The MACACC approach, as applied here, focuses on cortical thickness and does not assess subcortical structures or covariance between cortical and subcortical regions, which are also implicated in CD.
- Future Directions:
- Integrate structural covariance with functional connectivity measures.
- Combine cortical thickness analysis with diffusion tensor imaging (DTI) to examine white matter microstructure.
- Investigate whether similar changes in structural covariance are observed in adulthood.
- Examine the impact of environmental adversity and substance use on structural covariance.
- Conduct longitudinal studies to track developmental trajectories of structural covariance and functional connectivity.
Summary updated