Receptor Structure Shapes Host Range but Incompletely Predicts High Activity in Klebsiella pneumoniae Phage Cocktails

Rational design of phage cocktails typically relies on the lytic spectrum. However, susceptibility criteria vary widely among studies, ranging from qualitative lysis in spot tests to quantitative endpoints based on serial-dilution titration or efficiency of plating.

We investigated this discrepancy using four commercial phage cocktails and two capsule-specific monophages against a clinically representative collection of 448 Klebsiella pneumoniae isolates comprising 56 capsule types, collected in 2018-2025 from 12 medical centers. Using a modified Gratia titration assay, we defined host range (HR) as specific lysis at any dilution and putative therapeutic applicability (PTA) as lysis at dilutions corresponding to ≥10⁶ plaque-forming units per mL.

We identified a systematic discrepancy between these two measures of cocktail efficacy. HR coverage reached 64%, whereas PTA was significantly lower, with a median HR-PTA difference of 36%. This discrepancy persisted for capsule-specific monophages tested against KL2 isolates (n = 69), indicating that neither nominal capsule matching nor low component titres fully explained the loss of activity. This pattern provides indirect functional evidence that post-adsorption barriers contribute to the HR-PTA discrepancy, although adsorption and intracellular antiphage mechanisms were not assessed directly. Capsule type was the principal and most robust predictor of efficacy, whereas spatiotemporal factors, including year and medical centre, had only a weak effect on PTA.

Capsule matching can therefore help estimate population-level coverage but does not fully predict high-titre activity. Phage selection should incorporate serial-dilution testing rather than rely on spot-test lysis alone. Cocktail design should account for both receptor coverage and functional activity against representative isolates within individual capsule types.