Reading notes
Most microbiome case–control papers report genus-level associations as if they were specific to one disease. Duvallet and colleagues re-analysed 28 published 16S studies with a consistent pipeline and asked which signals replicate across studies and which recur across diseases, a different question from any single cohort.
Two dysbiosis patterns
Some conditions show enrichment of disease-associated bacteria relative to controls; others show depletion of health-associated bacteria. The distinction matters for interpretation and for what kind of intervention might be plausible in research (narrow antimicrobials vs replacement probiotics), neither follows directly from a consumer report line item.
| Pattern | Example diseases | Consistent genera (cross-study) |
|---|---|---|
| Enrichment | Colorectal cancer (CRC) | Fusobacterium, Porphyromonas, Peptostreptococcus, Parvimonas, Enterobacter, higher in CRC patients in two or more of four CRC studies |
| Depletion | IBD | Butyrate-producing Clostridiales, especially five genera in Ruminococcaceae and Lachnospiraceae depleted in IBD patients in at least two of four IBD studies |
For IBD, several individual studies found overall microbiota differences, but no single microbe was consistently associated with IBD across all studies, community shift without a universal IBD biomarker taxon.
Diarrhoea and IBD
Diarrhoeal illness (including CDI and non-CDI diarrhoea) showed the strongest and most consistent alpha-diversity reduction across studies. Microbiome shifts were dominated by:
- ↑ Proteobacteria and upper-gut–adapted taxa (Lactobacillaceae, Enterobacteriaceae)
- ↓ Bacteroidetes and some Firmicutes, including butyrate-producing Clostridia (Ruminococcaceae, Lachnospiraceae)
IBD patients, who often have diarrhoeal symptoms, showed overlapping enrichment of upper-gut taxa (Lactobacillus, Enterobacteriaceae) and lower alpha diversity than controls (Crohn’s in three studies, ulcerative colitis in two), but the diversity drop was less drastic than in active diarrhoea studies.
Diseases with little or no consistent signal
| Condition | Finding |
|---|---|
| Obesity | Little to no difference between obese and lean microbiomes across studies; only weak genus-level agreement (e.g. Roseburia and Mogibacterium enriched in obesity in two of three studies; Anaerovorax, Oscillibacter, Pseudoflavonifractor, Clostridium IV depleted in two studies) |
| Autism, rheumatoid arthritis | Reported Prevotella associations from individual papers did not hold in this conservative re-analysis |
| HIV | No consistent case–control difference across HIV studies once confounders were considered; Prevotella enrichment in one landmark study tracked sexual behaviour (MSM) rather than HIV status alone |
Non-specific responders (~51% of associations)
On average, 51% of genus-level associations in a dataset were genera linked to more than one disease. A general sickness/health signal exists even across different diseases (Supplementary Note 3).
Health-associated across diseases
- Clostridiales (especially Lachnospiraceae and Ruminococcaceae), depleted in multiple sick cohorts; 17 of 24 non-specific health-associated genera fell in this order
Disease-associated across diseases
- Lactobacillales, all five non-specific disease-associated genera in this order were enriched in cases across multiple diseases; adapted to lower pH / upper GI and likely enriched with faster transit and redox/pH disruption in the lower gut rather than as specific pathogens
Ubiquitous low-abundance taxa
- Escherichia/Shigella, Streptococcus, and other Enterobacteriaceae / Streptococcaceae are common at low frequency in healthy stool; presence/absence of non-specific disease genera may be more informative than small relative-abundance swings
Implication for research and reports: Future case–control work should ask whether an association is specific to the disease of interest or responds to a shared symptom (diarrhoea), antibiotic exposure, or a general sickness signature. Putative causal or diagnostic biomarkers should prioritize taxa outside the shared-response set.
Alpha diversity is not a universal disease marker
Prior work proposed reduced alpha diversity as a reliable dysbiosis indicator. In this re-analysis, individual studies often showed diversity–disease links that lost significance when pooled across studies, except for diarrhoea and perhaps IBD (Supplementary Figs. 6–8). Obesity and most other conditions showed no consistent diversity reduction.
Confounders that break simple narratives
- Antibiotic use, diseases often treated with antibiotics (diarrhoea, IBD) have microbiome shifts confounded with medication; disease-associated genera may reflect treatment, not disease alone
- Symptom overlap, diarrhoea-driven taxa (Proteobacteria, Lactobacillales) appear in IBD and other conditions with loose stools
- Demographics and behaviour, HIV/Prevotella is the clearest example; weak single-study associations may be false positives that meta-reanalysis deflates
What this means on consumer reports
- A low diversity score is most interpretable in the context of active diarrhoea or possible IBD, not as a universal “sick vs healthy” flag
- High Lactobacillus or Enterobacteriaceae on a stool panel may reflect transit/redox more than a probiotic win or acute infection
- Low Ruminococcaceae / Lachnospiraceae overlaps IBD and general sickness patterns, not proof of IBD without calprotectin and clinical workup
- CRC-style enrichment (Fusobacterium, etc.) is one of the few patterns with cross-study agreement in this paper, still not a screening test
- Obesity, IBS, and many functional labels cannot be read reliably from microbiome shifts alone if this meta-analysis is representative
For pages that apply these ideas: Dysbiosis, Alpha diversity, IBD vs functional gut, Opportunistic bacteria lists.