Skoltech scientists and their colleagues from other Russian medical research centers and mental health institutions have confirmed they can reliably distinguish patients with psychiatric disorders from healthy individuals based on nothing more than a blood sample. Moreover, the team’s latest findings indicate that the blood plasma levels of certain molecules called lipids can help differentiate patients with a major depressive disorder from those with schizophrenia.

Published in the journal Biomolecules, the research findings make it possible to enhance NeurOmix, an innovative system for early diagnosis of mental health risks based on a blood sample, developed earlier by Skoltech scientists and their collaborators. Studies spanning nearly 5,000 test subjects indicated that the test is 93% accurate.

“We know that serious psychiatric disorders leave molecular traces in the body, yet to this day all mental health diagnoses are made based on symptoms and the evaluation of a patient’s condition by a qualified physician,” says study co-author Research Scientist Anna Tkachev from Skoltech Neuro. “As soon as objective biological markers of diseases are introduced, mental health professionals will have an entirely new set of tools to bolster their subjective assessments. Besides that, biomarkers can make a difference for patient screening or differential diagnosis early on, at a point when symptoms are either absent or ambiguous.”

Backed by a Russian Science Foundation grant and the Moscow Center for Innovative Technologies in Healthcare, the team investigated blood plasma samples from 416 patients with common psychotic and affective disorders and 272 healthy individuals from two different cohorts, which came from a psychiatric hospital in Ufa, Russia, and from a Moscow-based mental health clinic.

Using a sophisticated analytic tool called a mass spectrometer, the researchers profiled the levels of a large number of lipids in the samples. They found 107 lipids that exhibited significant changes common to patients with schizophrenia and major depressive disorder. The variations of a further 37 lipids proved specific to either of the two diseases.

These findings enabled the team to train a machine learning algorithm that could distinguish schizophrenia from MDD with a success rate of 83%. Technically, the relevant performance metric is known as ROC AUC.

The lead author of the study, Anastasia Golubova — a Skoltech PhD student and a research intern at Skoltech Neuro — commented: “Previous studies have shown that the blood lipid composition of psychiatric patients differs significantly from that of healthy people. However, the specificity of these lipidomic alterations for particular disorders has remained uncertain. Our research demonstrates that lipid profiles can not only indicate the presence of a disease but also distinguish between two common psychiatric groups: psychotic (or schizophrenic) and mood (or depressive) disorders. This finding can facilitate the use of blood lipid tests as a helpful tool for accurate diagnostics of mental disorders, especially for disputable cases, and contribute to the development of precision medicine.”