Gut Microbiota Mediates the Protective Effects of SGLT-2 Inhibitors on Bipolar Disorder: an intermediary Mendelian randomization study.

Background: Bipolar disorder (BD), with a global prevalence of approximately 1%, is characterized by mood swings and social impairment, and involves factors such as genetics, brain structure abnormalities, and gut-brain axis dysregulation OBJECTIVES: The influence of SGLT-2 inhibitors and gut microbiota on BD remains unclear. We Investigate causal links between SGLT-2 inhibitors, gut microbiota, and BD risk using Mendelian randomization (MR). Identify mediating microbiota and elucidate connecting molecular pathways.

Methods: Two-sample MR utilized European GWAS data. Instrumental variables (p<5×10⁻⁵ for microbiota/BD) were analyzed primarily via inverse-variance weighted regression, with sensitivity analyses. Mediation analysis assessed SGLT-2 effects via microbiota. Functional enrichment (GO/KEGG), gene expression (GSE5388), and PPI network analyses identified pathways and hub genes.

Results: SGLT-2 inhibitors showed a significant negative causal effect on BD risk (OR = 0.104, 95% CI: 0.048-0.228, p<0.001), partially mediated by Clostridium sporosphaeroides abundance (proportion mediated = 4.1%; statistically significant at a nominal p-value threshold, though not after multiple testing correction. 368 of the 473 gut microbiota taxa showed evidence of a causal association with BD. Enrichment implicated calcium signaling (p<0.001) and neuroactive ligand-receptor pathways. Differential expression identified dysregulated GANC (p=0.048) in BD frontal cortices. PPI networks revealed hub genes (COQ2, PKM) involved in mitochondrial function and GPCR activity.

Conclusions: This study provides genetic evidence that SGLT-2 inhibitors may reduce BD risk, partly modulated by Clostridium sporosphaeroides. Identified pathways (e.g., calcium signaling) and dysregulated genes highlight mecolic-immune interplay in BD. Findings support evaluating SGLT-2 inhibitors and microbiota therapies in BD, though European-centric data requires broader validation.