ABSTRACT

Background: Advances in neuroimaging techniques such as multi-voxel pattern analysis (MVPA) for analysis and arterial spin labelling (ASL) for data acquisition have made it possible to observe the effects of intranasal oxytocin (IN-OT) administration in Bulimia Nervosa/Binge Eating Disorder (BN/BED) and Clinical High Risk for Psychosis (CHR-P) in a promising non-invasive way via rCBF changes. A pattern recognition algorithm employing MVPA, was used to distinguish between BN/BED and CHR-P patients to see the ability of the classifier when labelling the disorders.

Methods: Datasets from three different double blind placebo controlled studies, two cross over (within subjects) and one mixed study were used involving 17 healthy men, 30 CHR-P men and 25 healthy women &26 BN/BED women. rCBF was used as a measure of IN-OT’s effects within the brain and ASL was used to measure this change in perfusion. Difference images for oxytocin and placebo were obtained for all participants before using pairwise cross-validation to distinguish between the different datasets.

Results: The classifier only distinguished between BN/BED and CHR-P patients with an 55.5 % accuracy. This low insignificant score was explained by the classifier’s inability to distinguish between healthy women and BN/BED patients (23.5%) and the classifier’s only by chance ability to distinguish between CHR-P and healthy men (56.4%).

Conclusions: The classifier was unable to distinguish between CHR-P and BN/BEED patients. Further studies specifically investigating oxytocin’s effects in either BN/BED and CHR-P patients with a gender specific research is required at different doses and time-points, whilst ensuring circular analysis doesn’t occur.