Clinical subtypes that predict conversion to psychosis: A canonical correlation analysis study from the ShangHai At Risk for Psychosis program

• Published in: Australian and New Zealand journal of psychiatry 2020-05, Vol.54 (5), p.482-495
• Main Authors: Zhang, TianHong, Tang, XiaoChen, Li, HuiJun, Woodberry, Kristen A, Kline, Emily R, Xu, LiHua, Cui, HuiRu, Tang, YingYing, Wei, YanYan, Li, ChunBo, Hui, Li, Niznikiewicz, Margaret A, Shenton, Martha E, Keshavan, Matcheri S, Stone, William S, Wang, JiJun
• Format: Article
• Language: English
• ISSN: 0004-8674; 1440-1614
• DOI: 10.1177/0004867419872248

Objective: Since only 30% or fewer of individuals at clinical high risk convert to psychosis within 2 years, efforts are underway to refine risk identification strategies to increase their predictive power. The clinical high risk is a heterogeneous syndrome presenting with highly variable clinical symptoms and cognitive dysfunctions. This study investigated whether subtypes defined by baseline clinical and cognitive features improve the prediction of psychosis. Method: Four hundred clinical high-risk subjects from the ongoing ShangHai At Risk for Psychosis program were enrolled in a prospective cohort study. Canonical correlation analysis was applied to 289 clinical high-risk subjects with completed Structured Interview for Prodromal Syndromes and cognitive battery tests at baseline, and at least 1-year follow-up. Canonical variates were generated by canonical correlation analysis and then used for hierarchical cluster analysis to produce subtypes. Kaplan–Meier survival curves were constructed from the three subtypes to test their utility further in predicting psychosis. Results: Canonical correlation analysis determined two linear combinations: (1) negative symptom and functional deterioration-related cognitive features, and (2) Positive symptoms and emotional disorganization-related cognitive features. Cluster analysis revealed three subtypes defined by distinct and relatively homogeneous patterns along two dimensions, comprising 14.2% (subtype 1, n = 41), 37.4% (subtype 2, n = 108) and 48.4% (subtype 3, n = 140) of the sample, and each with distinctive features of clinical and cognitive performance. Those with subtype 1, which is characterized by extensive negative symptoms and cognitive deficits, appear to have the highest risk for psychosis. The conversion risk for subtypes 1–3 are 39.0%, 11.1% and 18.6%, respectively. Conclusion: Our results define important subtypes within clinical high-risk syndromes that highlight clinical symptoms and cognitive features that transcend current diagnostic boundaries. The three different subtypes reflect significant differences in clinical and cognitive characteristics as well as in the risk of conversion to psychosis.