The Evidence Layer for Scientific AI

Evidence supports feasibility, not safety

The strongest direct curing evidence is an in-vitro, single-enterococcal-strain experiment: resistance loss occurred after intercalating-agent exposure, but the study did not measure human toxicity, tissue distribution, microbiome disruption, pathogen fitness, relapse, or selection of compensatory mutants. Its 18 transconjugants also show that resistance-bearing plasmids can transfer into a plasmid-free recipient, confirming a manipulable but potentially hazardous system. Transferable resistance plasmid evidence

Why conjugation inhibition is plausible—but unpredictable

In a simplified surface-biomass model, transfer probability was the dominant determinant of transconjugant numbers, whereas plasmid loss exerted a smaller dampening effect; experimentally, high boundary intermixing produced 138 versus 91 transconjugants, and skin transfer in an older staphylococcal model exceeded broth transfer. These results support target engagement, but not clinical benefit: effects depend on species, plasmid, spatial structure, capsule, and host environment. Transfer-probability model and experiment

Safety boundary: the supplied studies justify laboratory and possibly tightly contained translational evaluation, but provide no evidence for administering a curing compound or conjugation inhibitor to infected people. A defensible progression would require demonstrated pathogen-selective activity, preservation of beneficial microbial communities, no increase in virulence or horizontal transfer, compatibility with antibiotics, durable plasmid suppression, and clinically meaningful endpoints in validated models before patient exposure. The recent Klebsiella study further warns that plasmid acquisition occurs alongside within-host adaptation and biofilm-associated phenotypes, so removing one plasmid may not remove persistence or invasive potential. Within-host plasmid and biofilm evolution

What would change the conclusion: reproducible multi-species data showing selective plasmid suppression without increased transfer, virulence, treatment failure, or microbiome injury would raise confidence; evidence of collateral transfer, rapid plasmid replacement, or host toxicity would reject the intervention concept.