Neurostructural subgroup in 4291 individuals with schizophrenia identified using the subtype and stage inference...

Area of Science:

• Neuroimaging
• Psychiatric Genetics
• Computational Psychiatry

Background:

• Psychiatric conditions lack precise biological definitions.
• Subtyping mental disorders can improve treatment efficacy.
• Neuroimaging offers insights into brain structure alterations in schizophrenia.

Purpose of the Study:

• To identify distinct neurostructural subtypes in schizophrenia using machine learning.
• To map the trajectory of gray matter changes in schizophrenia subtypes.
• To explore a biologically-based taxonomy for psychiatric disorders.

Main Methods:

• Analysis of cross-sectional brain images from 4,222 individuals with schizophrenia and 7,038 healthy controls.
• Utilized the Subtype and Stage Inference (SuStaIn) algorithm for subgroup identification.
• Pooled data from 41 international cohorts (ENIGMA, non-ENIGMA, public datasets).

Main Results:

• Identified two reproducible neurostructural subtypes of schizophrenia.
• Subtype 1: Early cortical gray matter loss with striatal enlargement.
• Subtype 2: Early subcortical gray matter loss (hippocampus, striatum).

Conclusions:

• Machine learning can define biologically distinct subtypes of schizophrenia.
• These subtypes exhibit unique gray matter change trajectories.
• An imaging-based taxonomy may refine psychiatric disorder constructs.