AUTHOR=Liu Linzi , Deng Zijing , Liu Wen , Liu Ruina , Ma Tao , Zhou Yifang , Wang Enhui , Tang Yanqing TITLE=The gut microbiota as a potential biomarker for methamphetamine use disorder: evidence from two independent datasets JOURNAL=Frontiers in Cellular and Infection Microbiology VOLUME=13 YEAR=2023 URL=https://www.frontiersin.org/journals/cellular-and-infection-microbiology/articles/10.3389/fcimb.2023.1257073 DOI=10.3389/fcimb.2023.1257073 ISSN=2235-2988 ABSTRACT=Background

Methamphetamine use disorder (MUD) poses a considerable public health threat, and its identification remains challenging due to the subjective nature of the current diagnostic system that relies on self-reported symptoms. Recent studies have suggested that MUD patients may have gut dysbiosis and that gut microbes may be involved in the pathological process of MUD. We aimed to examine gut dysbiosis among MUD patients and generate a machine-learning model utilizing gut microbiota features to facilitate the identification of MUD patients.

Method

Fecal samples from 78 MUD patients and 50 sex- and age-matched healthy controls (HCs) were analyzed by 16S rDNA sequencing to identify gut microbial characteristics that could help differentiate MUD patients from HCs. Based on these microbial features, we developed a machine learning model to help identify MUD patients. We also used public data to verify the model; these data were downloaded from a published study conducted in Wuhan, China (with 16 MUD patients and 14 HCs). Furthermore, we explored the gut microbial features of MUD patients within the first three months of withdrawal to identify the withdrawal period of MUD patients based on microbial features.

Results

MUD patients exhibited significant gut dysbiosis, including decreased richness and evenness and changes in the abundance of certain microbes, such as Proteobacteria and Firmicutes. Based on the gut microbiota features of MUD patients, we developed a machine learning model that demonstrated exceptional performance with an AUROC of 0.906 for identifying MUD patients. Additionally, when tested using an external and cross-regional dataset, the model achieved an AUROC of 0.830. Moreover, MUD patients within the first three months of withdrawal exhibited specific gut microbiota features, such as the significant enrichment of Actinobacteria. The machine learning model had an AUROC of 0.930 for identifying the withdrawal period of MUD patients.

Conclusion

In conclusion, the gut microbiota is a promising biomarker for identifying MUD and thus represents a potential approach to improving the identification of MUD patients. Future longitudinal studies are needed to validate these findings.