Large scale phage-antibiotic combination studies reveal key combinations for urinary tract infection and urosepsis treatments

Technical Summary: Large-Scale Phage-Antibiotic Combination Studies for UTI and Urosepsis

Problem Statement
The escalating crisis of antimicrobial resistance (AMR) has severely limited treatment options for common bacterial infections, particularly urinary tract infections (UTIs) and urosepsis caused by Escherichia coli and Klebsiella pneumoniae. While bacteriophage therapy has emerged as a potential alternative, the clinical application of phage-antibiotic combinations remains hindered by a lack of comprehensive data. Previous studies have been limited in scale, often examining small numbers of phages and antibiotics under heterogeneous conditions, leading to inconsistent reports of synergy, additivity, or antagonism. Consequently, the principles governing phage-antibiotic interactions are incompletely understood, and the benefits of such combinations are highly context-dependent. There is a critical need for large-scale, systematic screening to identify reproducible interaction patterns that can inform rational combination therapy development.

Methodology
This study employed a high-throughput screening approach to assess thousands of phage-antibiotic interactions using a clinically validated platform.

• Strain and Phage Selection: The researchers screened 61 E. coli and 132 Klebsiella spp. clinical isolates against 21 coliphages and 57 Klebsiella phages, respectively. From these, 17 highly resistant bacterial strains (8 E. coli, 9 K. pneumoniae) and 43 phages (20 E. coli, 23 Klebsiella) were selected based on their resistance profiles, phage susceptibility, and ability to be propagated to high titers. The selected strains represented diverse sequence types (STs) and clades, including the globally disseminated MDR ST131.
• Experimental Platform: Interactions were quantified using Sensititre GN7F plates, a commercial, quality-controlled system containing 24 different antibiotics across multiple concentrations. This allowed for the simultaneous assessment of Minimum Inhibitory Concentration (MIC) shifts and susceptibility changes.
• Assay Conditions: Phages were added to bacterial cultures at a multiplicity of infection (MOI) of 1 (~1×10⁵ PFU/mL and ~1×10⁵ CFU/mL). Bacterial growth was measured via optical density and image analysis of pellet opacity to determine fold changes relative to antibiotic-only controls.
• Data Analysis: A total of 13,485 interaction data points were generated. Statistical significance was determined using 2-way ANOVA with false discovery rate (FDR) correction. Hierarchical clustering and heatmaps were used to identify patterns based on phage taxonomy, antibiotic class, and bacterial strain. Genomic analysis (ANI, SNP calling, and phylogenetic trees) was performed to correlate genetic relatedness with phenotypic outcomes.

Key Results

• Prevalence of Additivity: Additive interactions were the predominant outcome, particularly in E. coli (11% of interactions) and even more so in K. pneumoniae (24%). Antagonistic interactions were rare (<1% for E. coli and 0.3% for K. pneumoniae).
• Species-Specific Profiles: K. pneumoniae interactions showed a greater overall impact and frequency of significant changes compared to E. coli. E. coli interactions were characterized by strong additivity with β-lactams, while K. pneumoniae showed more varied patterns, including specific antagonism with carbapenems in certain phage lineages.
• Phage Genus and Lineage Effects:
  • Tequatroviruses (E. coli): A strong, consistent additive signal was observed when combining Tequatroviruses with β-lactams. However, this was not universal; closely related phages (e.g., 148 and 162) lacked this effect, while genetically divergent members retained it.
  • Genomic Divergence vs. Phenotype: Closely related phages (>99.9% Average Nucleotide Identity) often displayed divergent phenotypes. For instance, phages 112 and 130 (differing by only two genes) showed opposite effects with nitrofurantoin and minocycline (additive vs. antagonistic). Similarly, phages JK19 and JK21 differed in host range and interaction profiles despite high genomic similarity.
  • Sugarlandviruses (K. pneumoniae): Significant intragenus variation was observed. BMCPR_006049 showed strong additivity across nearly all antibiotic classes, whereas RedSea1_var2 and Ponyo_var3 (sharing >99.5% ANI with others in the group) displayed antagonism with carbapenems and tetracyclines.
• MIC and Susceptibility Shifts: Approximately 13% of combinations resulted in a change in MIC, and 8.4% resulted in a change in clinical susceptibility classification. In E. coli, phage addition generally reduced MICs for β-lactams (e.g., ampicillin, ceftriaxone), whereas in K. pneumoniae, reductions were also observed but accompanied by specific increases for tetracyclines.
• Functional Convergence: Unrelated phages (e.g., RedSea1_var2 and SoFaint_var3) displayed strikingly similar antagonistic profiles with carbapenems, suggesting functional convergence independent of taxonomy. Conversely, some phages (e.g., EJ77E, KAH1) showed no measurable interaction, indicating they could be safely included in cocktails without compromising antibiotic efficacy.

Significance and Claims
The authors claim this study represents the first large-scale, systematic analysis of phage-antibiotic interactions spanning multiple bacterial species, phage genera, and antibiotic classes. By integrating phages into a standard clinical MIC determination platform, the study establishes a reproducible framework for identifying optimized therapeutic pairings.

Key claims regarding the study's contribution include:

1. Evidence-Based Foundation: The data provides a rational basis for prioritizing phage-antibiotic combinations, moving beyond anecdotal evidence to identify species-specific patterns (e.g., the strong additivity of Tequatroviruses with β-lactams in E. coli).
2. Genetic vs. Phenotypic Complexity: The study demonstrates that interaction outcomes cannot be inferred solely from taxonomic relatedness or genomic similarity; minor genomic differences can lead to divergent phenotypes, while unrelated phages can exhibit functional convergence.
3. Clinical Utility: The findings suggest that phages can act as functional enhancers of existing antibiotics, potentially restoring the efficacy of compromised drugs (particularly β-lactams) rather than serving as standalone replacements.
4. Scalability: The use of a commercially available, regulatory-approved platform (Sensititre) suggests a practical pathway for future clinical susceptibility testing and the development of evidence-based combination therapies for multidrug-resistant infections.

The authors maintain a modest tone regarding clinical translation, noting that while in vitro MIC shifts are promising, they do not yet reflect in vivo pharmacokinetics or complex host-pathogen interactions. The study positions itself as a foundational step toward defining reproducible patterns of interaction to guide future mechanistic and in vivo validation studies.