Bacteriophage-antibiotic combination therapy against extensively drug-resistant Pseudomonas aeruginosa infection to allow liver transplantation in a toddler

Abstract

Post-operative bacterial infections are a leading cause of mortality and morbidity after ongoing liver transplantation. Bacteria causing these infections in the hospital setting can exhibit high degrees of resistance to multiple types of antibiotics, which leads to major therapeutic hurdles. Alternate ways of treating these antibiotic-resistant infections are thus urgently needed. Phage therapy is one of them and consists in using selected bacteriophage viruses - viruses who specifically prey on bacteria, naturally found in various environmental samples - as bactericidal agents in replacement or in combination with antibiotics. The use of phage therapy raises various research questions to further characterize what determines therapeutic success or failure. In this work, we report the story of a toddler who suffered from extensively drug-resistant Pseudomonas aeruginosa sepsis after liver transplantation. He was treated by a bacteriophage-antibiotic intravenous combination therapy for 86 days. This salvage therapy was well tolerated, without antibody-mediated phage neutralization. It was associated with objective clinical and microbiological improvement, eventually allowing for liver retransplantation and complete resolution of all infections. Clear in vitro phage-antibiotic synergies were observed. The occurrence of bacterial phage resistance did not result in therapeutic failure, possibly due to phage-induced virulence tradeoffs, which we investigated in different experimental models.

Fig. 1. Clinical timeline.

Timeline of the most relevant surgical procedures (light green), microbiological results (light blue), antibiotic therapies (dark gray), pre- and post-liver transplantation protocols (blue-green), and phage therapy (magenta). The child developed several early-onset post-operative complications, including liver rejection, anaphylactic shock on rituximab, biliary digestive anastomosis perforation followed by an Escherichia coli and Enterococcus faecium sepsis, and cytomegalovirus (CMV) infection. At day 53 post-LDLT, he entered a severe septicemia due to XDR Pseudomonas aeruginosa. ABOc ABO compatible, ABOi ABO incompatible, BFC1 “bacteriofaagcocktail 1”, CMV cytomegalovirus, DDLT deceased-donor liver transplantation, EC ethics committee, IA intra-abdominal, IL intralesional, IV intravenous, LDLT living-donor liver transplantation, pfu plaque forming unit, PICU pediatric intensive care unit, VIM Verona integron-encoded metallo-β-lactamase, XDR extensively drug-resistant.

Fig. 2. Circular chromosomic view (CCV) of the bacterial genomes of seven Pseudomonas aeruginosa isolates retrieved just before (Pa1_BS) and during (Pa2_BR–Pa7_BS) phage therapy.

The CCV algorithm chose strain Pa3_LS as reference (inner circle). From the inner to outer rings of the CCV, two green rings display the genomic variations in phage PNM-susceptible isolates Pa1_BS and Pa7_BS, and four outer red rings display the genomic variations in phage PNM-resistant isolates Pa2_BR, Pa4_BR, Pa5_LR, and Pa6_LR. The two multi-colored outer rings display the protein annotations (categories) as present in the database of Clusters of Orthologous Groups of proteins (COGs). aa amino acid, bp basepairs, CDS coding sequence, IS insertion sequence, Mb megabases, nt nucleotide, PAGI Pseudomonas aeruginosa genomic island, PTM post-translational modification; SNP single-nucleotide polymorphism.

Fig. 3. Virulence assays.

a Chronological evolution of the mean activity score of Gm larvae injected with phosphate-buffered saline (control) or any of the seven Pseudomonas aeruginosa isolates retrieved just before (Pa1_BS) and during (Pa2_BR−Pa7_BS) phage therapy. Full lines represent the mean activity scores of Gm larvae injected with any of four PNM-resistant P. aeruginosa isolates (Pa2_BR and Pa4_BR−Pa6_LR) or phosphate-buffered saline (PBS). Dashed lines were used for the three PNM-susceptible isolates (Pa1_BS, Pa3_LS, and Pa7_BS). Note that first shown measurement after injection starts at +5 h. Real scale standard deviations were not visually implementable, but the dot width of each final mean measurement correlates with its standard deviation. n = 10 biologically independent animals. b Log-rank test finds the three liver abscess-borne strains (Pa3_LS, Pa5_LR, and Pa6_LR) significantly more virulent than the bloodborne strains (Pa1_BS, Pa2_BR, Pa4_BR, and Pa7_BS): p value = 0.0011. c Conversely, Log-rank test shows no significant difference in virulence when comparing the three phage-susceptible strains (Pa1_BS, Pa3_LS, Pa7_BS) with the four phage-resistant strains (Pa2_BR, Pa4_BR, Pa5_LR, Pa6_LR): p value = 0.7552. d The absorbance at 570 nm was measured to evaluate the viability of HeLa cells. The highly virulent P. aeruginosa strain PA14 was used as a positive control whereas HeLa cells without the addition of a P. aeruginosa culture served as a negative control. A connected letter report was created to pairwise compare all isolates and controls (two-sided Student’s t test, p value = 0.05, n = 6 biologically independent samples, no adjustment made for multiple comparisons). Pa1_BS was found to not significantly reduce the viability of the cells, compared to the negative control. Box-plots elements: center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, individual values. Source data are provided as a Source Data file.

Fig. 4. Phage-antibiotic synergy assay results.

Activities of phage PNM (at multiplicity of infection [MOI] 1.0) and a/d colistin (0.5 mg/L), b/e aztreonam (8 mg/L), and c/f gentamycin (2 mg/L), as well as combinations of phage PNM and these antibiotics against Pseudomonas aeruginosa isolate Pa1_BS (a–c) and Pa7_BS (d–f) were determined using an OmniLog® system. Bacterial proliferation is presented through relative units of cellular respiration. Efficacious phages, antibiotics, and combinations thereof, suppress bacterial proliferation. Results are presented as mean values of three experiments (biological replicates) with error bars representing the standard deviations of the means. Source data are provided as a Source Data file.

Fig. 5. Phage immune neutralization (PIN) assay.

a Schematic overview of the assay’s methodology (T^C: control titer; T^5’: titer after 5 min of serum-phage co-incubation; T^30’: titer after 30 min of serum-phage co-incubation). b Chronological phage immune neutralization (PIN) activity against phage ISP of ten patient sera collected before, during and after phage therapy. The evolution over time of the PIN activity against phage ISP is shown. Serum PIN activity is shown as % phage titer loss (compared to a pre-PT control serum) after incubation of phage ISP with sequential serum samples for 30 min. PIN activity appeared during the fifth week of PT initiation and had disappeared in a one-year post re-transplantation serum sample. Data are presented as mean values with error bars representing the standard deviation of the means. Each serum sample was tested on at least three independent occasions: six times for the 24-12-18 sample, four times for the 07-01-19 and 18-01-19 samples, and three times for all the other samples. c, d Persistent biological immunosuppression over the course of PT is illustrated through repeated quantification of the patient’s blood lymphocytes (normal values 4.0–10.5 × 10³ cells/microliter) and gamma-globulins (normal values 7.0–15.0 grams/liter). Source data are provided as a Source Data file.