Neurostructural Heterogeneity in Youths With Internalizing Symptoms

Area of Science:

• Neuroscience
• Psychiatry
• Machine Learning

Background:

• Internalizing disorders like anxiety and depression are common in youth and show varied treatment responses.
• Symptom-based classifications do not fully capture the neurobiology of these disorders.
• Neurobiologically informed subtypes derived from brain imaging offer a promising alternative classification approach.

Purpose of the Study:

• To identify neurobiologically distinct subtypes of internalizing youth.
• To utilize advanced machine learning for analyzing brain imaging data.
• To understand the heterogeneity in internalizing disorders based on neurobiological markers.

Main Methods:

• Employed a semisupervised machine learning method (HYDRA) on structural brain imaging data from 1141 youths.
• Utilized cross-validation to confirm the stability of identified subtypes.
• Incorporated resting-state functional MRI and diffusion imaging to further characterize subtypes.

Main Results:

• Identified two stable subtypes of internalizing youth based on brain volume and cortical thickness.
• Subtype 1 exhibited smaller brain volumes, reduced cortical thickness, impaired cognition, and higher psychopathology.
• Subtype 1 also showed altered frontolimbic function and white matter integrity compared to subtype 2 and controls.

Conclusions:

• Two subtypes of internalizing youth were differentiated by brain structure, function, and white matter integrity.
• One subtype demonstrated significantly poorer functioning across multiple domains.
• Biologically grounded subtypes can aid in early intervention and prognosis assessment for internalizing disorders.