关键词: 肠道菌群, 精神分裂症, 随机森林, 宏基因组, 非靶向代谢组

Abstract:

Objective To analyze the differences in gut microbiota（GM） metabolomics and their correlations with schizophrenia（SZ），and to provide a reference for early diagnosis and targeted intervention. Methods Fecal and blood samples were collected from 6 SZ patients（SZ group） and 6 healthy subjects（healthy control group） at Hulunbuir Third People's Hospital in 2023. Fecal samples were used for metagenomic sequencing（mNGS），and blood samples were used for metabolomics analysis by liquid chromatography-mass spectrometry. Random forest machine learning method was used to analyze the characteristics of the microbiota. The different species in the 2 groups were analyzed，and based on the significantly defferent species an early diagnosis model for SZ was consructed. Receiver operating characteristic（ROC） curve was used to evaluate the efficiency for diagnosing SZ. Results The mNGS results showed that there was statistical significance in 18 biomarkers such as Ruminococcus，Phascolarctobacterium，Lachnospira，Megasphaera and Megamonas between the 2 groups（P<0.05）. The metabolism pathways of glycine，serine and threonine were enriched in SZ group（P<0.05），and the 5-hydroxytryptamine receptor agonist/antagonist pathway was enriched（P<0.05）. The fatty acid biosynthesis pathway and steroid biosynthesis pathway were enriched in healthy control group（P<0.05）. The areas under curves（AUC） of Lachnospira，Phascolarctobacterium and Megamonas for diagnosing SZ were 1.000，0.090 and 0.909，respectively. Conclusions For SZ patients，GM，the pathways of 5-hydroxytryptamine receptor agonist/antagonist，fatty acid biosynthesis and steroid biosynthesis pathway have changes，which are related to SZ to some extent.

Key words: Gut microbiota, Schizophrenia, Random forest, Metagenomics, Non-targeted metabolomics