--- module: 095-03 language: en chapter: 95 title: "Microbiome, Microbial Ecology, Evolution, and Host–Pathogen Co-adaptation" module_title: "Dysbiosis, microbial metabolites, microbiome therapeutics, and causal inference" source_sha256: a6a5dbdc92da42ad99ba251e1848622501066c14977506c2cbbbbdd21da8433c --- # Dysbiosis, metabolites, and causal inference ## What dysbiosis does and does not mean ### Altered community associated with disease ### No universal taxonomic definition ### Loss of function or expansion of harmful activity ### Illness, diet, drugs, transit, inflammation all change it #### Association must not be read as cause ## Sampling and measurement limits ### Stool is accessible but sheds luminal material ### Time, storage, oxygen, extraction, transit shift results ### Relative abundance can rise while counts are unchanged ## Sequencing and complementary methods ### Sixteen-S amplifies marker regions #### Usually genus or approximate species only ### Shotgun metagenomics resolves genes and strains #### Detects dead organisms, cannot prove expression ### Culture proves viability and allows phenotype testing ### Quantitative PCR is sensitive for sought targets only ### Absolute quantification prevents compositional misreading ## Short-chain fatty acids and bile acids ### Anaerobic fermentation produces short-chain fatty acids #### Butyrate fuels colonocytes, shapes barrier and immunity #### Stool level is production minus uptake ### Bile acids conjugated by liver, modified by microbes #### Antibiotics reduce conversion, altering C. difficile #### Protective or injurious by species and site ## Other metabolites and drug interactions ### Tryptophan metabolism yields indoles ### Choline and carnitine become trimethylamine, then the oxide #### Diet and kidney clearance confound the association ### Protein fermentation gives ammonia, phenols, sulfide ### Cross-feeding: one species' waste is another's substrate ### Microbes activate prodrugs and inactivate active drugs #### Response association may reflect severity or diet #### Mechanism needs pathway isolation and exposure change ## Inflammation and Clostridioides difficile ### Inflammation raises oxygen and electron acceptors #### Facultative organisms exploit the new conditions #### They are markers and amplifiers, not original cause ### C. difficile shows ecological causation #### Antibiotics disrupt colonisation resistance #### Spores germinate, vegetative cells expand, toxins injure #### Toxigenic detection without diarrhoea may be colonisation ## Microbiome therapeutics ### Faecal transplantation restores community function #### Highly effective in selected recurrent infection #### Screening reduces but cannot remove unknown risk #### Success here does not justify unregulated broad use ### Probiotics are strain, dose, and host specific #### Evidence does not transfer between products ### Prebiotics are selectively used substrates #### Fibres may raise fatty acids but worsen bloating ### Diet acts on microbes and host directly #### Short-term taxonomic change is not an outcome ### Precision tools: narrow agents, phage, engineered bacteria #### Phage therapy faces range, immunity, manufacturing #### Ecological replacement can open unintended niches ## Study designs and causal inference ### Association studies confounded by drugs, diet, geography #### Regression cannot control unmeasured factors #### Behaviour change before diagnosis gives reverse causation ### Longitudinal studies give order, not causality #### Preclinical physiology may drive early change ### Randomised interventions are stronger if they hit the mediator #### Diet and antibiotics act on the host too ### Gnotobiotic animals show biological possibility #### Donor communities change in a new host #### Transfer is support, not proof for human disease ### Mendelian randomisation limited by noise and pleiotropy ### Classifiers may predict from treatment or geography ### Koch's postulates fit polymicrobial disease poorly #### Use temporality, dose, manipulation, reversibility #### A community function can be causal across taxa ## Biomarkers, claims, and the hierarchy ### Biomarkers need locked protocols and external validation #### They must improve decisions beyond clinical data #### Report performance, comparator, population, indeterminates ### Personalised claims often exceed evidence #### No single optimal community exists #### Relative abundance advice medicalises normal variation #### Strongest use: pathogen detection and selected transplant ### Therapeutics must define the intended mechanism #### Judge by patient outcome and safety #### Benefit possible without permanent engraftment ### Causal hierarchy: measure, associate, order, reproduce #### Show mechanism, perturb it, improve an outcome #### Dysbiosis is a hypothesis, not a diagnosis