Combination of pre-adapted bacteriophage therapy and antibiotics for treatment of fracture-related infection due to pandrug-resistant *Klebsiella pneumoniae*

Fracture-related infection (FRI) caused by multidrug-resistant bacteria is increasingly difficult to treat due to biofilm formation, immune evasion, and acquisition of resistance determinants such as ESBLs and carbapenemases in pathogens like Klebsiella pneumoniae. Biofilms harbor persister cells that tolerate antibiotics, often necessitating prolonged, high-dose treatments with significant toxicity and suboptimal outcomes. Phage therapy—using bacteriophages targeted to patient isolates—has re-emerged as a potential adjunct against MDR infections, including orthopedic-device-related infections. The study addresses whether a personalized, pre-adapted phage combined with antibiotics can salvage a longstanding, otherwise untreatable FRI due to pandrug-resistant K. pneumoniae and improve clinical outcomes, supported by in vitro efficacy including biofilm models.

Background highlights include: increasing MDR FRI burden; K. pneumoniae’s capacity for biofilm formation and resistance via ESBLs and carbapenemases; persister biology contributing to antibiotic tolerance. Personalized phage therapy selects phages active against patient isolates and can be pre-adapted to reduce resistance emergence (Appelmans method). Prior work demonstrates growing interest in phage therapy for orthopedic infections and synergistic effects of phage-antibiotic combinations in models (e.g., linezolid with phages in MRSA diabetic foot ulcers) and cases of biofilm-associated K. pneumoniae prosthetic infection. Combination therapy is advocated for carbapenem-resistant Enterobacteriaceae to mitigate resistance emergence. However, standardized phage cocktails have had mixed results in RCTs, suggesting individualized approaches may be preferable.

Design: Single-patient case report with microbiological, genomic, and in vitro experimental evaluations of phage-antibiotic combinations. Patient and clinical course: A 30-year-old woman sustained polytrauma in March 2016. Initial management included debridements and external fixation. Early cultures were polymicrobial. After months of multiple high-dose antibiotics (e.g., piperacillin/tazobactam, meropenem, colistin, aminoglycosides), complications occurred (e.g., febrile neutropenia with meropenem, ototoxicity with amikacin, colistin nephrotoxicity). Invasive mucormycosis required gastrectomy and splenectomy plus antifungals. In September 2016, FRI was suspected; femur biopsies grew two K. pneumoniae strains (one XDR), plus S. aureus and Mycobacterium xenopi. Despite prolonged therapy, infection persisted with nonhealing wounds and nonunion. Phage selection and pre-adaptation: Two day-170 K. pneumoniae isolates (Kp040741, Kp040762) were sent to Eliava Institute. From 12 Klebsiella phage clones, vB_KpnM_M1 (M1), a myovirus of genus Slopekvirus, was selected and pre-adapted via 15 rounds of Appelmans co-evolution on both isolates to enhance lytic activity and reduce resistance. Phage characterization: TEM morphology; host range (~65% across Klebsiella spp. isolates); adsorption (91% in 2 min, 99% in 6 min); latent period ~35 min; burst size ~43 PFU/cell; pH stability 5–10; thermostable to 50 °C. Genome analysis confirmed absence of genes associated with toxicity, lysogeny, or antibiotic resistance; a missense mutation (Thr281Arg) in the distal tail fiber hinge connector was identified in the pre-adapted variant. Ethics and administration: Ethical approval obtained; informed consent signed. On day 702, during surgery (radical debridement and rifampicin-impregnated autologous bone grafts), phage therapy was applied locally. Regimen: 100 ml of phage preparation at 1×10^8 PFU/ml diluted in 1000 ml saline instilled via a catheter at surgery end; catheter left in place for local instillations of 20 ml undiluted phage three times daily for 5 days (total 6-day course), alongside ongoing antibiotics. Due to limited pharmacological data and lack of consensus, no IV phage was given. Antibiotic regimens: Prior prolonged combinations included meropenem, colistin, fosfomycin, etc. After phage initiation, therapy was switched to ceftazidime/avibactam (2 g/0.5 g q8h; predicted Cmax 90.4 µg/ml and 14.6 µg/ml respectively), plus high-dose tigecycline and ciprofloxacin (later moxifloxacin). Tigecycline was stopped due to pancreatitis; moxifloxacin used due to nausea with ciprofloxacin. Antibiotics continued for 3 months post-surgery, then stopped. Microbiology and genomics: Day-702 isolate (Kp36336) was pandrug-resistant by EUCAST/VITEK 2. All K. pneumoniae isolates (ST893; capsular type K20) were sequenced using Illumina and Oxford Nanopore; hybrid assemblies with Unicycler; annotations with eggNOG-mapper, ABRicate, PHASTER, IslandViewer. Plasmid analyses revealed OXA-48 plasmid in all isolates; variable plasmid content (Kp040741: 3 plasmids; Kp040762: 5 plasmids). Kp36336 carried multiple ARGs (e.g., blaSHV-145, blaOXA-1, blaTEM-1, blaCTX-M-15; AME genes aac(6’)-Ib, aph(3")-Ib, aadA1, aac(3)-IIe, aph(6)-Id; qnrB1; oqxA5/oqxB19; dfrA14, sul2, tetA, fosA6, catA1). SNP analysis indicated limited within-host adaptation. Phage genome sequenced and compared to KP15; mutation identified by snippy; protein modeled by AlphaFold. In vitro assays: Phage activity tested against patient isolates at various MOIs (10.0, 1.0, 0.1). Growth kinetics measured via Omnilog respiration in 96-well plates with phage and antibiotics (ceftazidime/avibactam at 0.125–0.5 mg/L; meropenem 4 mg/L; colistin 1 mg/L), alone and in combination. Mature biofilm assay: 7-day biofilms of Kp36336 in 96-well plates; daily exposures for 3 days to phage (7.7×10^7–7.7×10^10 PFU/ml), ceftazidime/avibactam (0–90/22.5 µg/ml), and combinations. CFUs quantified after sonication; residual phage neutralized with 10 µM ammonium iron(II) sulfate. Statistics: Mann–Whitney U comparing combinations vs antibiotic alone. Immunogenicity: Serial sera collected at days 0, 1, 2, 5, 8, 18, 161, and 3 years post-phage to assess phage neutralization (Adams method).

- Clinical outcome: Following local application of pre-adapted phage M1 with ongoing antibiotics and subsequent switch to ceftazidime/avibactam (plus transient tigecycline and fluoroquinolone), the patient showed marked improvement within 2 days and sustained resolution by 3 months: vascularized viable graft, closure of sinus tract, no purulence at pin sites, normalized labs, and CT evidence of partial fracture consolidation. Antibiotics stopped at 3 months post-surgery; external fixator removed a week later. Multiple cultures from detached bone fragments and subsequent wound samples showed no growth. At 3-year follow-up, the patient had no recurrence and regained mobility. - Safety and immunogenicity: No adverse events attributable to phages; neutralizing antibodies emerged between days 8–18 post-application but were undetectable at 3 years. - Microbiology: Day-702 K. pneumoniae isolate (Kp36336) was pandrug-resistant yet susceptible to ceftazidime/avibactam (MIC 2.0/0.5 µg/ml). All isolates were ST893, K20 capsule type, with OXA-48 plasmid and multiple ARGs; limited within-host SNP changes. The pre-adapted phage retained activity against the PDR day-702 isolate. - Phage M1 characteristics: Broad Klebsiella host range (~65% of tested clinical isolates); rapid adsorption (91% at 2 min, 99% at 6 min), latent period ~35 min, burst size ~43 PFU/infected cell; stable across pH 5–10 and up to 50 °C; no lysogeny/toxin/ARG genes; pre-adapted variant carried a Thr281Arg mutation in distal tail fiber hinge connector. - In vitro synergy in biofilms: Ceftazidime/avibactam reduced biofilm bacterial counts dose- and time-dependently but did not eradicate at systemic Cmax (90/22.5 µg/ml). High phage doses alone also failed to eradicate. Combinations of high-dose phage and moderate ceftazidime/avibactam achieved significantly greater reductions than antibiotic alone (Mann–Whitney U p=0.0238 and p=0.0079 at tested conditions). - In vitro synergy in suspension: Combinations of pre-adapted phage M1 with ceftazidime/avibactam (0.125–0.5 mg/L) and with meropenem (4 mg/L) markedly suppressed bacterial proliferation more than either agent alone across MOIs 0.01–1.0. No synergy observed with colistin (1 mg/L).

This case demonstrates that individualized, pre-adapted phage therapy combined with appropriate antibiotics can control a recalcitrant FRI due to pandrug-resistant K. pneumoniae after prolonged prior treatment failure. The clinical resolution correlates with in vitro evidence of phage–antibiotic synergy in both planktonic and biofilm states, a critical feature for implant-associated infections where biofilms protect persisters. The pre-adaptation of phage M1 likely enhanced lytic efficacy and reduced resistance emergence, consistent with literature advocating tailored phage approaches. The observed synergy with beta-lactam/beta-lactamase inhibitor and carbapenem but not with colistin suggests antibiotic class-dependent interactions, aligning with combination therapy strategies for carbapenem-resistant Enterobacteriaceae. Although surgical debridement and rifampicin-impregnated bone grafting may have contributed, rifampicin lacked in vitro activity against the K. pneumoniae strains, underscoring the phage-antibiotic role. The development of neutralizing antibodies did not impede short-course local therapy; long-term absence of neutralization supports limited sustained immunogenicity. These findings support integrating personalized phages with selected antibiotics as an adjunctive modality for MDR FRI, particularly where biofilms dominate.

Pre-adapted phage M1 used locally in combination with antibiotics (initially meropenem/colistin and subsequently ceftazidime/avibactam) salvaged a longstanding, pandrug-resistant K. pneumoniae fracture-related infection, yielding durable clinical, microbiological, and radiological improvement with no phage-related adverse events. In vitro data corroborated strong phage–antibiotic synergy against the patient’s strain in mature biofilms and suspensions. The study underscores the promise of individualized phage–antibiotic combinations for MDR, biofilm-associated infections and supports further development of patient-tailored phage therapy pathways. Future research should include controlled studies to define optimal dosing, timing (sequential vs simultaneous), delivery routes (local vs systemic), pharmacokinetics/pharmacodynamics in bone and biofilms, and genetic determinants of phage efficacy (e.g., tail fiber mutations).

- Single-patient case report limits generalizability and causal inference. - Concomitant surgical interventions (radical debridement, rifampicin-impregnated autologous bone grafts) are potential confounders; contribution of surgery vs phage–antibiotic therapy cannot be fully disentangled, though rifampicin had no in vitro activity against the isolates. - Local, short-course phage administration without systemic pharmacokinetic data; optimal duration and dosing remain undefined. - No intraoperative biopsies at fixator removal to avoid disrupting consolidation; reliance on cultures from detached fragments and subsequent wound cultures. - Genotype–phenotype link for the phage tail fiber mutation remains inferential; engineered reintroduction to prove causality was not achieved. - In vitro models (Omnilog respiration, static 7-day biofilms) may not fully recapitulate in vivo conditions; synergy with antibiotics was class- and concentration-dependent.