The M-Factor: Cross-disorder analysis of shared gut microbial features across psychiatric and non-psychiatric conditions

Background

Psychiatric and neurodevelopmental disorders, including anxiety disorders (Anx), major depressive disorder (MDD), bipolar disorder (BD), schizophrenia (SZ), attention-deficit/hyperactivity disorder (ADHD), and autism spectrum disorder (ASD), share substantial clinical and genetic overlap and frequently co-occur with gastrointestinal (GI), metabolic, autoimmune, and neurodegenerative diseases. This cross-disorder overlap points to shared underlying biology, with chronic low-grade inflammation as one of the most consistently proposed mechanisms. The gut microbiome, given its central role in immune regulation, represents a plausible contributor to these cross-disorder links. Although numerous studies report microbiome alterations within individual disorders, a systematic cross-diagnostic comparison of gut microbiome alterations across these disorders is currently lacking.

Aims

The M-Factor project addresses this gap by aiming to(i) identify robust, cross-disorder gut microbial features spanning psychiatric and non-psychiatric conditions.

Methods

We integrate gut microbiome data from twelve in-house datasets (N = 3,420) alongside publicly available datasets from the European Nucleotide Archive and Sequence Read Archive (N=3,634). The combined dataset includes both 16S rRNA gene sequencing and shotgun metagenomics. All data are processed using a harmonized bioinformatic pipeline to minimize technical heterogeneity. Included psychiatric conditions comprise Anx, MDD, BD, SZ, ADHD, and ASD, while non-psychiatric conditions include inflammatory bowel disease, type 2 diabetes, obesity, multiple sclerosis, Alzheimer’s disease, and Parkinson’s disease. Studies are required to include at least 50 cases and 50 controls. Within each dataset, differential abundance is assessed using logistic regression, modeling disease status as the outcome and genus-level abundance (centered log-ratio transformed; prevalence >10%) as the predictor. Summary statistics are used for (i) cross-disorder Pearson correlations of effect sizes to quantify concordance in direction and magnitude, and (ii) random-effects meta-analyses to identify consistently altered taxa. Analyses are performed across different levels of diagnostic grouping (e.g., individual disorders to broader categories) to determine at which level shared microbial patterns emerge.

Results

Preliminary analyses include five 16S datasets from the European adult population: ADHD (n = 155), MDD (n = 1,476), comorbid psychiatric disorders (n = 405), Parkinson’s disease (n = 464), and multiple sclerosis (n = 414). Across these datasets, 230 genera exceeded 10% prevalence, with 183 genera present in at least three datasets retained for downstream analysis. Cross-disorder correlations of effect sizes were consistently positive and statistically significant (r = 0.30–0.70, all p<.0002, Table 1), indicating a shared direction of microbial associations across conditions. Random-effects meta-analyses identified 64 genera as significantly associated across disorders (FDR q < 0.05), supporting the presence of a robust transdiagnostic microbial signal. Hungatella, Clostridium innocuum group, and Anaerotruncus showed the strongest positive associations (OR ≥ 1.28 q ≤ 4.17e-08), while RF39, Lachnospiraceae NK4A136 group, and Clostridium sensu stricto 1 showed the strongest negative associations (OR ≤ 0.85, q ≤ 8.72e-4).

Conclusions

These findings provide preliminary evidence for a shared gut microbial signature spanning psychiatric, (neuro)immune, and neurodegenerative disorders. Ongoing work will expand dataset inclusion, evaluate the impact of key confounders (e.g., medication use, diet), and incorporate functional pathway analyses to determine whether shared taxonomic shifts translate into convergent biological functions.

References

No references