Pseudomonas aeruginosa ventricular assist device infections: findings from ineffective phage therapies in five cases

Abstract

Left ventricular assist devices (LVAD) are increasingly used for management of heart failure; infection remains a frequent complication. Phage therapy has been successful in a variety of antibiotic refractory infections and is of interest in treating LVAD infections. We performed a retrospective review of four patients that underwent five separate courses of intravenous (IV) phage therapy with concomitant antibiotic for treatment of endovascular Pseudomonas aeruginosa LVAD infection. We assessed phage susceptibility, bacterial strain sequencing, serum neutralization, biofilm activity, and shelf-life of phage preparations. Five treatments of one to four wild-type virulent phage(s) were administered for 14-51 days after informed consent and regulatory approval. There was no successful outcome. Breakthrough bacteremia occurred in four of five treatments. Two patients died from the underlying infection. We noted a variable decline in phage susceptibility following three of five treatments, four of four tested developed serum neutralization, and prophage presence was confirmed in isolates of two tested patients. Two phage preparations showed an initial titer drop. Phage biofilm activity was confirmed in two. Phage susceptibility alone was not predictive of clinical efficacy in P. aeruginosa endovascular LVAD infection. IV phage was associated with serum neutralization in most cases though lack of clinical effect may be multifactorial including presence of multiple bacterial isolates with varying phage susceptibility, presence of prophages, decline in phage titers, and possible lack of biofilm activity. Breakthrough bacteremia occurred frequently (while the organism remained susceptible to administered phage) and is an important safety consideration.

Keywords: LVAD; MDRO; Pseudomonas aeruginosa; device-related infection; phage therapy.

Fig 1

Timeline depicting phage duration, onset of positive blood cultures, and antibiotic therapy in five cases of multidrug-resistant Pseudomonas aeruginosa left ventricular assist device infections.

Fig 2

Susceptibility of Pseudomonas aeruginosa isolates from Case 1 to phages used in the treatment of the patient using the Biolog method. This consisted of inoculation of standardized bacterial suspensions with bacteriophages individually and in combination in 96-well microtiter plates incubated at 37°C in a Biolog machine for 24 hours. Bacterial respiration led to a reduction of the tetrazolium dye leading to a color change which is depicted as relative units of bacterial growth.

Fig 3

Antimicrobial susceptibility and relatedness of Pseudomonas aeruginosa isolates for Cases 2.1 and 2.2. (A) Phage and antibiotic susceptibility of isolates PaD1–13. Susceptibility was determined by spotting 4 µL of 10⁹ PFU from a library of Pseudomonas phages to determine the three candidate virulent myoviruses with the highest activity, PAK_P1, PAK_P5, and E217. The isolate ID indicates the order of isolation. The source, day, and timeline of sample collection are listed, with day 0 being the start of first course phage administration (Case 2.1). Phage susceptibility was tested using efficiency of plating for both Cases 2.1 and 2.2 phages. Phage susceptibility is indicated as sensitive (blue), partial clearing as intermediate (yellow), no plaquing as resistant (red), or not determined (white). Antibiotic susceptibility was determined using VITEK 2 in the clinical microbiologic laboratory. (B) Phylogenetic tree assembled with the complete genome sequence of case originating isolate PaD1 and short read sequences of PaD2 to 13. Branch colors indicate clades, and lengths indicate relative evolutionary distance.

Fig 4

Phage susceptibility testing for Case 3. (A) Plaque morphologies of baseline Pseudomonas aeruginosa isolates, Sh1 and Sh2. (B) Growth curves of baseline Sh1, Sh2, and combined culture of the two strains in 1:1 ratio as affected by the phage. Graphs are average of three replicates and standard deviations (SD) are shown. (C) Growth curves of Sh1 and Sh2 combined culture in the presence of various sub-inhibitory levels of antibiotics: ciprofloxacin (cipro), gentamicin (genta), meropenem (mero). Graphs are average of three replicates and SD are shown. (D) Plaque morphologies of post-phage Pseudomonas aeruginosa isolates, Sh3 and Sh4. (E and F) Effect of PASA16 on the strains SH1 and SH2 which were isolated before the treatment, in biofilm setting in two assays; bacterial colony forming unit (CFU) count in a 96-well plate static model (E) and biomass detection by crystal violet staining in a flow model (F). * denotes differences with P-value <0.05.

Fig 5

Phage susceptibility testing for Case 4. (A) Plaque forming units (PFU) of phage PASA16 on the Pseudomonas aeruginosa isolates, C393 (left panel) and C442 (right panel). As a positive control for PASA16, efficacy served the strain PA14. (B) Growth curves of C393 alone and with PASA16 phage. Graphs are average of three replicates and standard deviations (SD) are shown. (C) Growth curves of C393 in the presence of combinations of PASA16 and various antibiotics in concentration of their minimum inhibitory concentration (MIC) and 0.1 MIC. Graphs are average of three replicates and SD are shown. (D) Bacterial colony forming unit (CFU) count of the cultures presented in C, at the endpoint of experiment. (E) Effect of patient’s serum on the PFU of PASA16 pre- and post-treatment.