Multianalyte markers of schizophrenia and bipolar disorder: A preliminary study

Previous studies have identified altered molecular profiles in blood samples from individuals with schizophrenia and with bipolar disorder using multianalyte immunoassay platforms but there has been little comparison of the two groups in the same investigation. A total of 337 participants including 146 with schizophrenia, 79 with bipolar disorder, and 112 non-psychiatric controls had a blood sample drawn from which 166 analytes were measured. The initial dataset was split; classification models were developed in a training dataset and their performance evaluated in a test dataset. Principal component analysis was used to generate factor scores that were then compared between the groups. In a training set, a total of 7 independent factors were generated using 29 markers that were both normally distributed and significantly associated with diagnosis. Many of these analytes are components of the immune system and involved in the inflammatory response to infectious agents and foreign antigens. Two of the seven principal component scores discriminated between individuals with schizophrenia and with bipolar disorder; additional factors distinguished individuals with either schizophrenia or bipolar disorder from control individuals, while two factors were not significantly different between any of the diagnostic groups. In a test dataset, the schizophrenia vs. control Receiver Operating Curve (ROC) analysis shows an overall accuracy of 77% for schizophrenia vs. bipolar disorder, 84% for schizophrenia vs. controls, and 72% for bipolar disorder vs. controls. An increased understanding of the role of altered pathways in serious psychiatric disorders may lead to novel methods for disease diagnosis and therapy.