Prosthetic vascular graft infections (PVGI) are complicated events associated with high morbidity and mortality rates. PVGI incidence is between 1 and 6%, showing a death rate range between 15 to 75% with a rate of major amputation that may reach 70%, which is especially caused by aortic graft (1). PVGI are usually caused by the more virulent microorganisms, such as Staphylococcus aureus, Escherichia coli, Pseudomonas aeruginosa, Klebsiella spp, Proteus spp., and Enterobacter spp. (2). Carbapenem-resistant Pseudomonas aeruginosa is one the most critical pathogens according to the World Health Organization (WHO) and poses a particular threat in hospitals, nursing homes, and among patients whose care requires devices such as ventilators and blood catheters (3). Several mechanisms of resistance to carbapenems have been described in clinical isolates of Pseudomonas aeruginosa, among them, we can highlight -lactamases, mutations in porins, and overexpression of efflux pumps (4).

Phage therapy is a promising new treatment against infections produced by multi-drug resistant pathogens (5). To improve phage therapy application, it would be necessary to know more about the clinical response and bacterial host-phage interactions. The monitoring of these interactions can be done by massive sequencing, thus identifying the genes affected by mutations that occur during therapy, and therefore directing the way phage therapy should be applied.

In this work, we analyzed the clinical, microbiological, and molecular features of P. aeruginosa isolates in a case of prosthetic vascular graft infection (PVGI) after phage administration was deemed unsuccessful. This knowledge could allow the development of strategies to improve the use of clinical use of phage therapy.

This work presents the clinical case of a man in his fifties who developed recurrent bloodstream infections (BSI) caused by ceftazidime and piperacillin-tazobactam resistant Pseudomonas aeruginosa. In August 2020, an axillo-bifemoral bypass was performed in this patient because of a severe infra-renal aorta atherosclerotic occlusion (Leriche syndrome). Shortly after, the patient first developed a BSI, which was interpreted as secondary to an infected sacral ulcer, and was treated with meropenem at dose of 1 g every 8 h for 17 days. From November to June 2021, the patient suffered several recurrences of BSI due to P. aeruginosa. After the second relapse, ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET/CT was performed, showing radioactive-labeled glucose uptake along the pre-clavicular graft region (SUVmax value, 5.29), suggesting prosthetic vascular graft infection (PVGI). Graft replacement was rejected due to high risk of contralateral leg ischemia and bypass thrombosis if partial bypass replacement was done. The patient subsequently received different targeted antibiotherapies for 2 to 6 weeks (Figure 1). Two new P. aeruginosa BSI relapses occurred between March and May 2021. Ceftolozane-tazobactam was not used (not available), and off-label tebipenem use was denied; treatment was therefore as shown in Figure 1. Simultaneously, therapeutic phages were obtained from the Queen Astrid Military Hospital (Brussels, Belgium) in view of a possible compassionate use of phage therapy.

Phage susceptibility of the clinical isolate was determined in the A Coruña University Hospital, and three P. aeruginosa phages were selected. Two (14/01 and PT07) were myoviruses, and one (PNM) was a podovirus. Two of the phages had known receptors, being lipopolysaccharide (LPS) for phage 14/01, and the type IV pili for PNM (6, 7). In July 2021, 70 ml of the bacteriophage cocktail, consisting of the three phages, each at a concentration of 10⁷ plaque forming units (PFU)/mL for a total dose of 2.1 × 10⁹ PFU/day, was administered intravenously, once a day in a 6-h infusion for 3 days in an inpatient regimen. Thereafter, and for four additional days, this cocktail of phages alongside ceftazidime-avibactam were administered by outpatient parenteral antimicrobial therapy (OPAT) at the patient's request. In summary, phage therapy was used for 1 week, with ceftazidime-avibactam being used for 8 weeks (from 6 weeks before, to 2 weeks after phage therapy). Importantly, no adverse events were observed.

Ceftazidime-avibactam was applied for 8 weeks (from 6 weeks before to 2 weeks after the phage therapy). In August 2021, in the absence of antimicrobial therapy, a new P. aeruginosa BSI with a septic residual limb metastasis occurred, now involving a wild-type strain being susceptible to ß-lactams and quinolones. Finally, upon a multidisciplinary discussion, a proximal vascular prosthesis replacement combined with antibiotherapy in OPAT was performed. As of this writing, 2023, after 10 months without treatment, the patient remains asymptomatic.

Eight blood culture isolates obtained between November 2020 and August 2021 were characterized by microbiological analysis. In August 2021, these isolates were identified by matrix assisted laser desorption/ionization—time of flight mass spectrometry (MALDI-TOF MS, Bruker Daltonics) as P. aeruginosa. Antimicrobial susceptibility testing was performed by broth microdilution (Microscan, Beckman Coulter) and interpreted using European Committee on Antimicrobial Susceptibility Testing (EUCAST) criteria. Pulsed-field gel electrophoresis (PFGE) revealed that four representative isolates, from 2020 and 2021, showed the same pattern and therefore were clonal (Supplementary Figure S1). Genomic sequence of four representative isolates showed that these isolates belonged to the high-risk clone ST308 with core genome MLST (MLST)cgST2675. High-risk clones are those with a wide dissemination and a global spread associated with a multidrug resistant (MDR) or extensively drug resistant (XDR) profiles including extended-spectrum β-lactamases (ESBLs) and carbapenemases (8). A search for antimicrobial resistance genes in the ResFinder database did not identify any acquired resistance genes such as ß-lactamases (9). Analysis of 40 chromosomal resistance genes revealed natural polymorphisms (SNPs) in numerous genes, including some changes with unknown effect (10). The profile of all isolates was identical for all genes studied, except for the nfxB gene, a transcriptional repressor that regulates the efflux pump MexCD-OprJ (11), in which multiple amino acid changes were observed in the last 3 isolates (Supplementary Table S1). Moreover, comparison of the genomes of P. aeruginosa isolates HE2011471 (previous to phage administration) and HE2105886 (1 month after phage treatment) revealed important mutations (Figure 2). Interestingly, several detected genomic changes could be involved in phage resistance (Supplementary Table S2, Figure 3). P. aeruginosa isolate HE2105886, recovered 1 month after phage treatment, exhibited a decrease in the Minimum Inhibitory Concentration (MICs) of ß-lactam and quinolone antibiotics (Figure 1) probably due to the mutations of the membrane receptors, regulators and efflux pumps. However, we did not test any of these isolates against the individual phages in the cocktail to track the sensitivity or resistance to each of them.

Finally, the post-phage therapy isolate exhibited increased production of biofilm and phage-resistant mutants. These two tests were deemed relevant, as bacteria can form biofilms as a response to environmental stress, such as the presence of antibiotics, and this, alongside the emergence of phage-resistance could explain the observed failure to eradicate the infection with phage therapy (Figures 4A, B) (16).

This case highlights three key issues. The first relates to the use of combined therapy (antibiotics and phages) vs. monotherapy (antibiotic or phages) as a definitive treatment for P. aeruginosa BSI. So far, combination therapy has not been associated with reduced mortality or any advantages in terms of clinical outcome or successful treatment of recurrent/persistent bacteraemia (17). The second issue concerns prosthetic vascular graft infection (PVGI) diagnosis and its best management. At the early stages of PVGI, a high degree of suspicion is essential. Indium-111-labeled white blood cell scintigraphy plus single-photon emission computed tomography (SPECT/CT) could reduce the false positive rates observed with PET/CT (18). In the present case, PET/CT was not performed at the early postoperative stage, because of the possibility of a false positive result. Regarding the treatment, graft excision is the preferred surgical approach. However, some patients are considered unacceptable surgical candidates due to underlying comorbidities or technically unfeasible surgery. When this occurs, lifelong suppressive antimicrobial therapy is an option, but is not free of side-effects or the risk of development of antimicrobial resistance. As the patient was initially denied surgical treatment, long-term suppressive treatment with quinolones was administered. However, ciprofloxacin-resistance developed, and surgery was finally deemed key to a favorable outcome. Furthermore, there is no evidence-based recommendation for either PVGI antimicrobial treatment or its optimal duration; minimum intravenous therapy for 6 weeks followed by oral antibiotherapy for up to 6 months has been proposed (19, 20). ¹⁸F-FDG PET/CT-guided treatment duration seems feasible and would allow treatment to be tailored to individual patients (21). The third highlighted issue is related to phage therapy. Phage therapy involves the targeted application of strictly virulent phages that can specifically infect and lyse the targeted pathogenic bacteria they encounter, hereby releasing virion progeny that continues the lytic cycle. A major advantage of phages is the minimal impact on non-target bacteria or body tissues (22). A recent systematic review suggested that phage therapy is safe and may be effective in different difficult-to-treat infections (23). Interestingly, a previously-reported case of ß-lactam resistance in an MDR P. aeruginosa isolate causing an aortic graft infection, showed the reversion of the infection after treatment with a phage whose receptor was the outer membrane protein M (OprM) of the MexAB- and MexXY-multidrug efflux pumps, associated with antibiotic resistance and which was no longer expressed in the selected phage-resistant bacterial strain (24). In the present case, even though phage therapy did not cure the infection, the posterior infection recurrence was caused by an antibiotic susceptible isolate that belonged to the same lineage as the one that was causing the pre-phage treatment episodes of infection. However, the recurrent isolate was recovered 1 month after the phage therapy, and it is possible that the resensitization (to -lactams and quinolones) could have been due to phage action which produced genomic changes in the membrane receptors, regulators and efflux pumps (Supplementary Table S2) or to a spontaneous evolution. In order to confirm it, further studies could be carried out as to demonstrate that these mutations located in the membrane receptors, regulators and efflux pumps proteins indeed render a strain phage-resistant and susceptible to antimicrobials. These studies would involve constructing several plasmids encoding the wild-type genes with lack of the mutations and overexpressing them in the P. aeruginosa isolate HE2105886 to restore the phage susceptibility and the resistance to antimicrobials.

One strength of this study is that phage therapy was able to be safely administered in OPAT for greater patient convenience. To our knowledge, this has only been reported once before, in a series of six cases, without phage-related adverse events (25).

The main limitations of our study were: first, we did not perform in vitro studies of the effect of the phages in combination with antibiotics, especially ß-lactams; second, we could not study the P. aeruginosa clinical isolates during the seven days of application of phage therapy. Finally, it should be noticed that there are no national or local phage banks with characterized phages against P. aeruginosa in which specific phages could be selected to provide personalized approach in patients with complex infections caused by these bacteria.

In conclusion, in complex P. aeruginosa infections the choice of antibiotic therapy and its duration are crucial for minimizing antibiotic pressure and development of resistance. Although we have not demonstrated that phage treatment was effective in this case, studying the molecular mechanisms of resistance to phages and bacteria-phage interactions are key to improving phage therapy in the near future.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary material.

Written informed consent was obtained from the patient for the publication of any potentially identifiable images or data included in this article.

LB, IL-H, MR-F, JP-F, and CC-S conducted the experiments, analyzed the results, and wrote the manuscript. SD, MM, and J-PP revised the results and phage therapy administration. JR-B, MT, and LL obtained the research funding, directed the clinical settings, and supervised the writing of the manuscript. All authors contributed to the article and approved the submitted version.