关键词:
Neurocognition
Schizophrenia
Cluster Analysis
Cortical Thinning
Neuroanatomy
摘要:
Background: Cortical thinning may index the neurobiological pathogenesis of schizophrenia. However, case-control studies ignore the neurobiological, symptomatic, and cognitive heterogeneity of the illness. To organize heterogeneity, unsupervised machine learning approaches can derive neuroanatomical subgroups with cortical thickness data and identify clinical and cognitive correlates. Methods: Demographic, clinical (PANSS), cognitive (MCCB) and functional competence (COALS) measurements were collected from 63 healthy controls and 73 patients diagnosed with a schizophrenia spectrum disorder. They underwent 3T MRI scans, which provided 148 regional thickness values with FreeSurfer. Agglomerative hierarchical cluster analysis was carried out using Ward’s method with Euclidean distance on age-corrected thickness values. The optimal number of clusters was determined by 16 validity indices. Results: 12/16 validity indices suggested 2 clusters. The first subgroup (n1=53) was primarily composed of patients (75%), displayed significantly thinning cortex across 110/148 regions (p<0.001;84/148 regions with Bonferroni correction), and was more cognitive impaired than the other group (MCCB 30.23 vs. 37.84). Approximately two-thirds of subgroup 2 (n2=83) were healthy controls. Despite displaying similar cortical thickness, patients were more cognitively impaired than controls in subgroup 2 (MCCB 29.36 vs. 43.44) Conclusions: Cortical thinning is associated with cognitive impairment, but only in a subgroup of schizophrenia patients. Cognitive impairment is present in patients with normal-range cortical thickness, which implies a partial dissociation between cognition and this aspect of brain structure. The findings illustrate the potential value of generating neuroanatomical subtypes to pursue novel hypotheses about the pathogenesis of schizophrenia. Supported By: CIHR Keywords: Neurocognition, Schizophrenia, Cluster Analysis, Cortical Thinning, Neuroanatomy