Attacking Heterogeneity in Schizophrenia by Deriving Clinical Subgroups From Widely Available Symptom Data

Abstract

Previous research has identified (1) a "deficit" subtype of schizophrenia characterized by enduring negative symptoms and diminished emotionality and (2) a "distress" subtype associated with high emotionality-including anxiety, depression, and stress sensitivity. Individuals in deficit and distress categories differ sharply in development, clinical course and behavior, and show distinct biological markers, perhaps signaling different etiologies. We tested whether deficit and distress subtypes would emerge from a simple but novel data-driven subgrouping analysis, based on Positive and Negative Syndrome Scale (PANSS) negative and distress symptom dimensions, and whether subgrouping was informative regarding other facets of behavior and brain function. PANSS data, and other assessments, were available for 549 people with schizophrenia diagnoses. Negative and distress symptom composite scores were used as indicators in 2-step cluster analyses, which divided the sample into low symptom (n = 301), distress (n = 121), and deficit (n = 127) subgroups. Relative to the low-symptom group, the deficit and distress subgroups had comparably higher total PANSS symptoms (Ps < .001) and were similarly functionally impaired (eg, global functioning [GAF] Ps < .001), but showed markedly different patterns on symptom, cognitive and personality variables, among others. Initial analyses of functional magnetic resonance imaging (fMRI) data from a 182-participant subset of the full sample also suggested distinct patterns of neural recruitment during working memory. The field seeks more neuroscience-based systems for classifying psychiatric conditions, but these are inescapably behavioral disorders. More effective parsing of clinical and behavioral traits could identify homogeneous target groups for further neural system and molecular studies, helping to integrate clinical and neuroscience approaches.

Keywords: Positive and Negative Syndrome Scale; cluster analysis; data-driven subgrouping; deficit syndrome; distress; schizophrenia.

Fig. 1.

Characteristics of low-symptom, high-distress, and deficit Positive and Negative Syndrome Scale (PANSS) subgroups. (a) Scatterplot of the sample across PANSS negative and distress symptom dimensions. (b) Percentage of individuals in each subgroup with different schizophrenia spectrum diagnostic subtypes. (c) Percentage of individuals in each subgroup with no history of suicidal gestures. (d) Within subgroup percentage taking antidepressants, mood stabilizers and sedatives, and taking 3 or more different medications. For detailed statistics, see tables 1 and 2.

Fig. 2.

Positive and Negative Syndrome Scale (PANSS) subgroups and prefrontal functional magnetic resonance imaging (fMRI) activation. The N-back task reliably activates a working memory network centered on the dlPFC, but also involving the anterior cingulate and parietal cortex, including in patients with schizophrenia. We display regional activation across these 3 brain areas per each patient symptoms subgroup—low symptoms, high distress, and deficit. Each bar represent in order: right middle frontal gyrus (Brodmann area [BA] 9) in blue, left anterior cingulate (BA 32) in tan, and left superior parietal lobule, (BA 39/40) in orange. fMRI activation represent beta weights extracted from the SPM F map (see Methods) and presented as mean ± 1 SE in arbitrary units (a.u.).