Longitudinal inference of hippocampal volume and morphometric hippocampal-cortical covariance in psychosis.
TL;DR
These findings might contribute to elucidating distinct hippocampal patterns underlying heterogeneous manifestations of psychosis for a subgroup of patients, as not all patients were assigned a staging pattern for hippocampal volume or connectivity.
Abstract
Schizophrenia and related psychoses are characterized by reductions in hippocampal volume and morphometric hippocampal-cortical covariance. No study to date has inferred the temporal relationship between deviations in subfield and adjacent white matter volume and connectivity across psychosis. To explore this relationship, we sampled cross-sectional neuroimaging data from 174 patients and 120 non-clinical controls from two studies on first- and multi-episode psychosis. We applied Subtype and Stage Inference (SuStaIn), a machine learning algorithm combining clustering and disease progression modeling, to eight measures of hippocampal volume and morphometric hippocampal-cortical covariance (subfields and white matter regions per hemisphere) combined. We then performed in-depth SuStaIn analyses for 18 volumes (five subfields and four white matter regions per hemisphere) and 18 connectivity measures (graph-theory-based participation coefficient) separately. Some patients and controls showed no deviations (no increase or decrease) from control means in volume and covariance. For individuals who did show deviations, we inferred data-driven stages from deviations in volume toward covariance. In-depth SuStaIn analyses inferred three distinct volume patterns (intact volume, subfield-dominant, whiter matter-dominant) followed by one of four possible connectivity patterns (intact connectivity, whiter matter-dominant, subfield-dominant, subiculum-dominant). These subtypes showed significant differences in age, patient-control ratio and associations between data-driven stages and memory. In previous work, altered hippocampal volume and morphometric covariance distinctively precede cognitive and clinical symptoms of psychosis. While not all patients were assigned a staging pattern for hippocampal volume or connectivity, our findings might thus contribute to elucidating distinct hippocampal patterns underlying heterogeneous manifestations of psychosis for a subgroup of patients.