Major depressive disorder has long been treated as a single diagnosis, but growing evidence suggests it may be a collection of biologically distinct conditions masquerading under one label. A new case-control study published in Molecular Psychiatry homes in on one such subgroup: patients whose depression is accompanied by REM sleep behavior disorder (RBD), a condition characterized by the loss of normal muscle paralysis during dreaming, leading to dream enactment. The findings suggest that these patients carry a gut microbiome signature that closely resembles that of idiopathic RBD (iRBD), a known prodromal marker of Parkinson’s disease and related synucleinopathies.
What they found
Researchers led by Yuhua Yang, Yun Kwok Wing, and colleagues across multiple Hong Kong institutions recruited 420 participants in a four-group design: 124 healthy controls, 80 patients with major depressive disorder without RBD (MDD-only), 82 patients with both depression and RBD (MDD+RBD), and 134 patients with idiopathic RBD. All underwent clinical assessments including olfactory testing, motor examination, and stool sample collection for 16S rRNA gene sequencing and metagenomic analysis.
The clinical picture was striking. Patients with MDD+RBD showed elevated scores on the prodromal Parkinson’s disease likelihood ratio, documented olfactory deficits, and subtle motor signs compared with the MDD-only group. In other words, their clinical profile more closely resembled iRBD patients than patients with depression alone, suggesting that RBD in the context of depression is not simply a comorbidity but a marker of underlying neurodegenerative risk.
The microbiome data reinforced this distinction. Beta diversity analysis, which measures how distinct microbial communities are from one another, showed that the gut microbial composition of the MDD+RBD group was significantly different from both the healthy controls and the MDD-only group. Crucially, the MDD+RBD microbial profile closely paralleled that of iRBD.
Several specific taxonomic shifts drove this pattern. The MDD+RBD group showed enrichment of Akkermansia muciniphila, a mucin-degrading bacterium previously linked to Parkinson’s disease, and Ruthenibacterium lactatiformans. At the same time, they showed depletion of Faecalibacterium prausnitzii, a key short-chain fatty acid producer known for its anti-inflammatory properties. Depression-associated taxa such as Streptococcus parasanguinis and Actinomyces oris were also shifted in the MDD+RBD group.
At the functional level, metagenomic prediction revealed attenuated B-vitamin biosynthesis pathways and reduced polysaccharide degradation capacity in the MDD+RBD group, mirroring the functional profile seen in iRBD. These metabolic changes could have downstream effects on neurotransmitter synthesis, energy metabolism, and immune signaling.
A Random Forest classifier trained on microbial features distinguished MDD+RBD from MDD-only with an area under the curve of 0.73 in cross-validation and 0.79 in an independent validation dataset of 65 participants. While not yet at clinical grade, this performance suggests that microbiome-based biomarkers could eventually help identify depression patients who warrant closer monitoring for neurodegenerative disease.
Why it matters
The study proposes that depression with RBD represents a biologically distinct subtype of major depressive disorder, one that carries an elevated risk of progression to Parkinson’s disease or dementia with Lewy bodies. Identifying these patients early matters because standard antidepressant treatments have not been shown to modify neurodegenerative risk, and clinical trials of disease-modifying therapies for synucleinopathies increasingly target prodromal populations.
If the gut microbiome signature can be validated as a scalable screening tool, it could allow clinicians to stratify depression patients by neurodegenerative risk without requiring expensive imaging or specialized neurological assessment. A simple stool test, combined with clinical screening for RBD (which is often underdiagnosed), could flag high-risk individuals for further evaluation.
The study also adds to the growing recognition that the gut microbiome is sensitive to both brain state and neurodegenerative pathology. The convergence of microbial profiles between MDD+RBD and iRBD suggests that the same gut-brain axis disruptions seen in prodromal Parkinson’s may be operating in this depression subgroup, independent of depression severity or treatment history.
Limits
The study is cross-sectional, which means it cannot determine whether the observed microbiome differences predate the onset of depression or RBD, or whether they emerge as a consequence of these conditions. Longitudinal follow-up of the cohort will be essential to establish temporal ordering. The sample is drawn exclusively from a Hong Kong population, and generalizability to other ethnic and dietary backgrounds remains unknown. Confounders such as medication use, particularly antidepressants and benzodiazepines, which are known to affect both sleep and the microbiome, were addressed through statistical adjustment but cannot be fully eliminated in an observational design. The machine learning classifier, while promising, requires external validation in independent cohorts from diverse settings before it can be considered for clinical use.
Bottom line
Depression accompanied by REM sleep behavior disorder carries a gut microbiome signature indistinguishable from idiopathic RBD and distinct from depression alone, supporting the view that MDD+RBD is a biologically distinct depression subtype tied to neurodegenerative risk. Microbiome-based biomarkers may one day help identify these high-risk patients early.
Source: Yang Y, Li N, Zhou L, Gong S, He Z, Tang S, Ni J, Liu Y, Chan JWY, Or BPN, Lam SP, Zhang J, Chan PKS, Chen Z, Wong SH, Mok VCT, Chan NY, Chau SWH, Lai CKC, Scheperjans F, Wang J, Huang B, Wing YK. Gut microbiome signature distinguishes depression with REM sleep behavior disorder from depression alone: implications for neurodegeneration. Mol Psychiatry. 2026 Jul 23. doi: 10.1038/s41380-026-03773-3. PMID: 42493609.