Biomarker case-detection and prediction with potential for functional psychosis screening: development and validation of a model related to biochemistry, sensory neural timing and end organ performance.

• Main Authors: Stephanie eFryar-Williams, Jorg E. Strobel
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2016-04-01
• Series: Frontiers in Psychiatry
• Subjects: Biomarkers1; schizoaffective8; psychosis2; mental illness3; model4; case detection5
• Online Access: http://journal.frontiersin.org/Journal/10.3389/fpsyt.2016.00048/full

The Mental Health Biomarker Project aimed to discover case-predictive biomarkers for functional psychosis. In a retrospective, cross-sectional study, candidate marker results from 67, highly-characterized symptomatic participants were compared with results from 67 gender and age matched controls. Urine samples were analysed for catecholamines, their metabolites and hydroxylpyrolline-2-one, an oxidative stress marker. Blood samples were analyzed for vitamin and trace element cofactors of enzymes in the catecholamine synthesis and metabolism pathways. Cognitive, auditory and visual processing measures were assessed using a simple 45 minute, office-based procedure. Receiver Operating Curve (ROC) and Odds Ratio analysis discovered biomarkers for deficits in folate, vitamin D and B6 and elevations in free copper to zinc ratio, catecholamines and the oxidative stress marker. Deficits were discovered in peripheral visual and auditory end-organ function, intra-cerebral auditory and visual processing speed and dichotic-listening performance. 15 ROC biomarker variables were divided into 5 functional domains. Through a repeated ROC process, individual ROC variables, followed by domains and finally the overall 15 set model, were dichotomously scored and tallied for abnormal results upon which it was found that ≥ 3 out of 5 abnormal domains achieved an AUC of 0.952 with a sensitivity of 84 per cent and a specificity of 90 percent. Six additional middle ear biomarkers in a 21 biomarker set increased sensitivity to 94% percent. Fivefold cross-validation yielded a mean sensitivity of 85% for the 15 biomarker set. Non-parametric regression analysis confirmed that ≥ 3 out of 5 abnormally scored domains predicted > 50% risk of case-ness whilst 4 abnormally-scored domains predicted 88% risk of case-ness and 100% diagnostic certainty was reached when all 5 domains were abnormally scored. These findings require validation in prospective cohorts and other mental illness states. They have potential for case-detection,-screening, -monitoring and targeted personalized management. They unmask unmet needs within the functional psychosis condition and suggest new biological understandings of psychosis phenomenology.