Neuroanatomical Subtypes of Schizophrenia Reveal Divergent Atrophy Trajectories and Distinct Associations with Long-Term Clozapine Use
Xiao Chang, Yuchao Jiang, Bochao Danae Lin, Shitong Xiang, Jie Zhang, Guusje Collin, Paul Wirsching, Florian Schlagenhauf, Yijie Zhao, Hengfen Gong, Yanqun Zheng, Long-Biao Cui, Jijun Wang, Tianye Jia, Andreas Heinz, Jurjen J Luykx, Jianfeng FengAbstract
Background and Hypothesis
Schizophrenia is a heterogeneous disorder, and biology-informed stratification may provide insights into individual differences in disease course and treatment response. We hypothesized that neuroimaging-based subtyping could reveal biologically distinct subgroups characterized by divergent brain atrophy trajectories and unique clinical profiles.
Study Design
We applied the Subtype and Stage Inference framework to baseline structural MRI data from the 6-year longitudinal Genetic Risk and Outcome of Psychosis cohort (n = 467), including patients, unaffected siblings and healthy controls. We characterized subtype-specific longitudinal brain atrophy and medication profiles. Subtype-specific atrophy patterns were correlated with external molecular maps of neurotransmitter density and schizophrenia-related gene expression.
Study Results
Two distinct neuroanatomical subtypes were identified: subtype 1 (65.5%), characterized by widespread cortical volume loss at baseline, and stable volume decrease over 6-year follow-up. Conversely, subtype 2 (34.5%) showed baseline volume decrease confined in the hippocampus and amygdala, but accelerated volume loss over follow-up. Clinically, subtype 1 patients had a significantly higher prevalence of clozapine use than subtype 2 (36% vs. 17%; chi2 = 5.62, P = 0.02), and brain atrophy of subtype 1 mirrored patterns observed in clozapine users. The two subtypes showed divergent spatial associations with neurotransmitter systems and risk-gene expression profiles.
Conclusions
We identified two neuroanatomical subtypes of schizophrenia showing divergent brain atrophy over 6-years. The subtypes may represent biologically distinct forms of schizophrenia, as they show different medication profiles and associations with molecular signatures. Our results provide a candidate framework for parsing disease heterogeneity and may offer insights for future stratified treatment strategies.