Therapeutics & platforms
Bacteriophage therapy
Receptor recognition, the self-amplifying dose, resistance by receptor loss, and the reasons a self-replicating biological agent does not fit the clinical trial machinery built for fixed doses.
A bacteriophage is an obligate parasite of bacteria, and its therapeutic properties follow almost entirely from how it finds a host. The tail fibres or receptor-binding proteins of a lytic phage recognise a specific surface structure — the O-antigen of lipopolysaccharide, wall teichoic acid, a pilus, an outer-membrane porin such as OmpF or LamB, sometimes a capsule polysaccharide it must first digest with its own depolymerase. Binding is a molecular recognition event, so the host range is often narrower than a species: strains of Pseudomonas aeruginosa isolated from two patients may share almost nothing a single phage can bind. That is the origin of the field’s central selling point — the gut and skin microbiota are untouched — and of its central logistical problem.
The dose that makes itself
After injection of its genome the phage hijacks host transcription, assembles progeny, and lyses the cell, releasing on the order of tens to a couple of hundred particles per burst. Pharmacokinetically this is unlike any small molecule. The agent amplifies where the target density is high and disappears where it is not, so exposure is not proportional to the administered dose; below a threshold inoculum replication may not sustain itself at all, and above it the effective concentration is set by the bacteria. Clearance works against this from the other side: phages are proteinaceous particles, taken up by the mononuclear phagocyte system within hours and neutralised by antibodies after repeated exposure, so the window is real but bounded.
Resistance is fast, and that is partly the point
Because entry depends on a single receptor, a bacterium escapes by altering or losing it — and mutants appear within hours in vitro. What makes this less fatal than it sounds is that the receptors are usually functional: an O-antigen, a pilus, an efflux-pump component. Losing them costs the bacterium something. The clearest demonstrated case is the phage OMKO1, which binds an outer-membrane protein of the MexAB–OprM efflux system, so escape mutants trade phage resistance for restored antibiotic sensitivity. Selecting phages to steer resistance into a corner, rather than merely to kill, is now a design principle. Cocktails targeting non-overlapping receptors serve the same end.
Why the evidence base is thinner than the mechanism
Several constraints are not biological but structural. Temperate phages must be excluded outright: a genome capable of lysogeny can transduce virulence or resistance genes between strains. Bulk lysis releases endotoxin, so Gram-negative preparations need purification and dose staging. Above all, a self-replicating, strain-matched, evolving product does not fit a fixed-dose randomised trial. PhagoBurn, the EU-funded burn-wound trial, remains the instructive failure: the administered cocktail lost titre during storage, and patients received doses orders of magnitude below intent, so the trial answered nothing about phage efficacy. Belgium’s magistral framework, which treats phages as pharmacy preparations compounded to an active-substance monograph, is currently the only route that matches the product’s actual nature — and it produces case series, not controlled evidence.