Personalized aerosolised bacteriophage treatment of a chronic lung infection due to multidrug-resistant Pseudomonas aeruginosa

Bacteriophage therapy has been suggested as an alternative or complementary strategy for the treatment of multidrug resistant (MDR) bacterial infections. Here, we report the favourable clinical evolution of a 41-year-old male patient with a Kartagener syndrome complicated by a life-threatening chronic MDR Pseudomonas aeruginosa infection, who is treated successfully with iterative aerosolized phage treatments specifically directed against the patient’s isolate. We follow the longitudinal evolution of both phage and bacterial loads during and after phage administration in respiratory samples. Phage titres in consecutive sputum samples indicate in patient phage replication. Phenotypic analysis and whole genome sequencing of sequential bacterial isolates reveals a clonal, but phenotypically diverse population of hypermutator strains. The MDR phenotype in the collected isolates is multifactorial and mainly due to spontaneous chromosomal mutations. All isolates recovered after phage treatment remain phage susceptible. These results demonstrate that clinically significant improvement is achievable by personalised phage therapy even in the absence of complete eradication of P. aeruginosa lung colonization.


The electron micrographs presented in
do not add much to the manuscript and could be removed. The TEM of the phage shows a contracted myophage that is of low quality, and the disrupted cell does not tell us anything about the phage or host other than that the phage can lyse it, which is already known.
6. For Fig. 3C, state the criteria for determining sensitive/intermediate/resistant for the phages.
7. In Fig. 3F, there are two lines for breakpoints, it is not clear what this means.
8. Fig. 4 should have a scale bar and node bootstrap values. 9. In Fig. 5, panels A and B could be merged into one figure with side-by-side bars for 24 and 72 h; color coding for each isolate is not needed as the axes are labeled.
10. In the Methods, primers are named but I could not identify a table containing a list of primer sequences. This needs to be included with the manuscript as a main or supplementary table.
11. Throughout the manuscript the mu (µ) symbol is not displayed correctly.
Reviewer #2 (Remarks to the Author): The paper gives a nice exposition of a simple clinical case, in which a colonised/infected gentleman with background CF suffered for months with Pseudomonas superinfection including classic hypermutator phenotypes, as expected in this setting and well demonstrated here.
He was offered phage therapy in desperation, with good clinical response, and then again later, again with good response.
The clinical story is strongly suggestive that phage was successful but provides no definitive evidence. I think the clinical anecdote is too long and could be greatly simplified.
Of note is the evidence of intra-pulmonary/ intra-airways phage amplification after inhalation only, and the fact that recovering Pseudomonal populations are not significantly affected in terms of phage or antibiotic adaptation. These data address an important assumption in phage therapy -that is, that the nature of the interaction is so mature that efficacy can be expected without eradication and that relapse can be expected without resistance, so that retreatment is entirely feasible and practicable.
The emphasis in the abstract is about right I think. The case report element, currently around a thousand words, could be abbreviated in the text, perhaps by leaning more on Fig 1 and its legend, as n=1 phage therapy anecdotes have little value other than to illustrate the phage-bacterial interplay, which is the more important element of the story for this readership.
The evidence presented of phage amplification in the airways is important but does not appear to be accompanied by evidence of phage in blood. That does not appear to have been tested and is probably worth a mention because the question of whether Pseud DNA spills into blood or phages are detectable in light of the hint at an inflammatory response early on.
The early clinical decline suggestive of an inflammatory response to phage therapy is not discussed but a Jarisch-Herxheimer type reaction vs simply too much endotoxin needs to be considered in the text. His calculated endotoxin per dose is 5mL over 20 mins @ (less than) 1250EU/mL means that he may have received ~6000EU over 20 mins, which is for a 70kg man means ~85 EU/kg over 20 mins. This is well in excess of the pyrogenic threshold of 5EU/kg/hr which we are advised (e.g. by FDA) not to exceed for intravenous preparations and (logically) for nebulised preparations. It is noteworthy that this was so well tolerated thereafter. This should be briefly discussed.
In The virulence studies are well presented but probably could be more summary as well. The most important data are the wax moth and cell culture data. The assertion of pyoverdine synthesis as an explanation for slow growth could be reference by alluding to any one of many refs that show the relationship between pyoverdine synthesis and bacterial killing. TEM images add colour rather than data and could also be trimmed away to save space, at the editors discretion.
Reviewer #3 (Remarks to the Author): [1] Was there any possibility to try to identify the poly-microbial species present?
[2] Pseudomonas aeruginosa is cleared by 300-400 days, but then comes back by around 466 onwards. Where is coming from? Is it an internal source, a niche that harbours a smaller number of undetected bacteria or is the source external? Previous studies have shown that the nasopharynx may act as a potential reservoir for re-infecting the lower airways. Did the authors look at upper respiratory tract colonisation?
[3] I don't see the point of comparing the clinical isolates to PA01 in figure-5. PA01 will be very different to these host adapted strains, so the value of this comparision is minimal. Why not compare the clinical isolates from this patient to other isolates from other patients at different stages of host adaptation?
[4] The Galleria models does not really tell us anything new or useful about these strains. The model itself is an acute infection model so it is not surprising that PA01 kills the larvae while biofilm forming host adapted clinical strains do not. I think the ideal model to investigate these clinical strains would have been a mouse model of chronic lung infection, one which allows persistence of infection, biofilm formation and interactions with both humoral and cellualr immunity.
Reviewer #4 (Remarks to the Author): The authors report in details on a single case of phage use in a patient with an MDR pseudomonas infection. Although similar reports have been published before, it is refreshing to see such details in a case report.
My overall suggestions/comments: 1. The manuscript is extremely long for a case report, esp as a type of 'communication'. Some of the details may be more suited as a story, but may lack scientific value. I would recommend: (a) Reduction in number of, or details of previous studies on phage therapy: no need to review all the previous experience in a cummunication as the readers would be lost long before reading the actual report.
(b) Many details are not important: e.g. p127-132: specific dates are not important. If authors want to make a point on duration, they can change dates to number of days, which may be more useful in giving context to the 'story'. Or line 108-113: Can simply indicated phage therapy was initiated after consent was obtained from patient using phage vFB297... Words like 'last resort' makes the reading more entertaining but also distracting.
(c) Authors should summarize the clinical course more succinctly, focusing not on providing as much detail as possible, but thinking about what is most important in their goal of the report. Responses to the reviewer's comments are shown in blue and newly added or modified text is highlighted by quotation marks.

Reviewer #1 (Remarks to the Author):
The manuscript by Köhler et al. describes a single case report of a phage therapy intervention for treatment of a persistent MDR P. aeruginosa infection. The persistent nature of the infection and multiple rounds of antibiotic therapy prior to phage treatment supports the idea that the phage treatment played some role in clinical improvement. Strengths of the work include collection and sequencing of multiple bacterial samples, which shows the infection was a single acquisition of a pathogen followed by diversification by mutation. Interestingly, phage resistance did not appear during treatment and bacterial virulence and antibiotic sensitivity did not change during or after treatment. COMMENTS 1. The phage vFB297 was used for treatment but its characteristics are not described in the manuscript. The phage is described as a member of the Pakpunavirus genus, so I suspect that the phage genome has already been sequenced. At a minimum, the phage genome sequence should be deposited in an INSDC repository (NCBI or EMBL) and some basic features, including relationships to other known phages, described in the manuscript.
Response: We thank the reviewer for this comment. Indeed, the phage had been sequenced (phage genome size of 90,516 bp) We have submitted the genome sequence to DFAST (https://dfast.ddbj.nig.ac.jp/) for annotation. As with many phage genomes, most of the ORFs are annotated as hypothetical proteins. A complete map with generated annotations has been added to supplementary data as a novel Figure (Figure S1). The relationship to homologous phages has been added to the legend of Figure S1 to avoid interrupting the flow in the main manuscript. The phage genome, including annotations, has been deposited to NCBI GenBank and is available as accession # OQ921398.
We have added the following figure legend, which contains all information requested by the reviewer: " Fig. S1. Genomic map of phage vFB297. BLAST search in the NCBI database retrieved Pseudomonas phages 908-1, SRT6, PaYy-2 and vB_PaeM_SCUT-S2 as showing more than 94% nucleotide identity with vFB297. These phages are annotated as belonging to the genus Pakpunavirus. Their complete genome sizes (90 -95 kbp) are comparable to those of phage vFB297 (90,516 bp). The genome sequence of vFB297 was annotated using DFAST (https://dfast.ddbj.nig.ac.jp/help_annotation) and the genome map generated with SnapGene. Arrows in green identify genes with functional predictions, while all other arrows indicate genes encoding hypothetical proteins. The phage genome was submitted to NCBI GenBank (accession # OQ921398). " 2. Somewhere in the manuscript (e.g., the heading for Table 1) should describe the rationale for the strain naming scheme so they can be confidently interpreted by the reader. It looks like the scheme is [day#].

Response:
The footnote to Table 1 in the submitted version already described the designation and meaning of isolate names, but this might not have been visible enough: "Isolate nomenclature designates day of sampling (D), starting with first day of phage administration (D0), and is followed by the number of the individual isolate, when more than one colony was recovered from the sputum sample (e.g. D0.1, D0.2, etc). D-100 designates isolate for phage screening recovered 100 days before the first phage administration." To make this naming more visible to the reader we have added the following sentence to the legend of figure 1, where isolate names appear for the first time: "Bacterial isolates as well as sampling days and CT scan dates are numbered in reference to the day of first phage treatment onset (D0)." We have also added to the Methods section (first paragraph) the sentence: "Isolates are named according to their day of isolation starting with the first day of the initial phage administration (day 0, D0)." Fig. 1 has a number of issues. A magenta star in the legend denotes "Rehabilitation" but this symbol does not appear on the figure. The "no antibiotic" arrow feature does not also need the word "STOP" above it, as its meaning is already conveyed in the legend. The CT images should be enlarged significantly; they could be a separate panel or made large enough so that they are as wide as the timeline image (i.e., each CT image is ~ ¼ the width of the whole image). If possible, the green antibiotic treatment line should be disaggregated into the different drugs used for treatment, rather than all lumped together into a single "Antibiotherapy" line.

Response:
We thank the reviewer for this comment. To improve the reading of figure 1, we have modified it as follows: • Stop was removed • "rehabilitation period" indicated by a star was replaced by a neutral white rectangle and indicated as such in the figure legend • CT scans were enlarged and fitted below the timeline as suggested by this reviewer • However, we decided to keep the green "generic" antibiotic treatment line. Given that the patient received several antibiotics and with different treatment periods, we felt it would be too confusing for the reader to detail all treatment regimens; if requested the different antibiotics could be added to the supplementary data 4. In Fig. 2, it is not clear if points at the X-axis detection limit are counted or if this means that no CFU or PFU were observed (i.e., counts were below the detection limit). There should be an X-axis title for both figures.
Response: We agree with the reviewer that this was confusing. For the qPCR assay in panel 2A, we used the term limit of quantification (LoQ), meaning that our qPCR assay detected a signal but the values were not in a range for allowing precise quantification. For the CFU and PFU determinations in panel 2B, our limit of detection (LoD) is 200 CFU or PFU/ml. We indicated this by a stippled line at 2x10 2 , meaning that in samples below this line, we did not count any CFU or PFU. We added definitions for abbreviations "LoQ" and "LoD" to the revised figure legend and also added the x-axis title (time (days)) to both panels. Fig. 3a do not add much to the manuscript and could be removed. The TEM of the phage shows a contracted myophage that is of low quality, and the disrupted cell does not tell us anything about the phage or host other than that the phage can lyse it, which is already known.

Response:
We partially agree with the reviewer; indeed, such phage images have been shown previously. However, we would prefer to keep the images of the phage as an illustration and leave the decision to remove them at the editor's discretion.
6. For Fig. 3C, state the criteria for determining sensitive/intermediate/resistant for the phages.
Response: In the submitted version we have given the following definition for phage susceptibility in the legend to Fig. 3: Susceptibility of patient isolates to phage vFB297 was determined by plaque assays (full data set see We feel that these "arbitrary" criteria are sufficiently clearly defined and further supported visually by the supplementary Figure S2. We have only added the word "visible" behind the word "plaques", hoping this helps further clarifying the definition. 8. Fig. 4 should have a scale bar and node bootstrap values. Response: To improve figure 4 as suggested we have prepared a new version of the phylogeny tree of the isolates (Fig. 4) figure 4 also shows the scale bar as requested; the figure legend was modified accordingly.
9. In Fig. 5, panels A and B could be merged into one figure with side-by-side bars for 24 and 72 h; color coding for each isolate is not needed as the axes are labeled.
Response: We agree with the reviewer's comment that for space and simplicity reasons, the initial two panels A and B should be merged into a single graph (now panel A). This also allows removing the isolate name labelling on the x-axis. We also now show the individual data points for each replicate. Panel B showing data from the Galleria mellonella remains unchanged.  I could not identify a table containing a list of  primer sequences. This needs to be included with the manuscript as a main or supplementary  table. Response: We apologize for this missing information. We had prepared such a Table, but apparently omitted to join it during submission. The new Table (Table S4) showing the primer sequences used is now part of the supplementary information.
11. Throughout the manuscript the mu (µ) symbol is not displayed correctly.

Response:
We have checked the µ symbol throughout the manuscript and made replacements where necessary.

Reviewer #2 (Remarks to the Author):
The paper gives a nice exposition of a simple clinical case, in which a colonised/infected gentleman with background CF suffered for months with Pseudomonas superinfection including classic hypermutator phenotypes, as expected in this setting and well demonstrated here.
He was offered phage therapy in desperation, with good clinical response, and then again later, again with good response.
The clinical story is strongly suggestive that phage was successful but provides no definitive evidence. I think the clinical anecdote is too long and could be greatly simplified. Of note is the evidence of intra-pulmonary/ intra-airways phage amplification after inhalation only, and the fact that recovering Pseudomonal populations are not significantly affected in terms of phage or antibiotic adaptation. These data address an important assumption in phage therapythat is, that the nature of the interaction is so mature that efficacy can be expected without eradication and that relapse can be expected without resistance, so that retreatment is entirely feasible and practicable.
Response: we thank the reviewer for his positive global comment. Comments: 1. The emphasis in the abstract is about right I think. The case report element, currently around a thousand words, could be abbreviated in the text, perhaps by leaning more on Fig 1  and its legend, as n=1 phage therapy anecdote have little value other than to illustrate the phage-bacterial interplay, which is the more important element of the story for this readership.
Response: We agree with this comment. The initial case description was meant to provide the reviewers with a complete clinical picture, but this made the description indeed too long. As suggested, we have significantly shortened the patient's history and treatment description and removed "anecdotal" descriptions, thereby reducing the length of this section from 991 words to 597 words in the revised version.
2. The evidence presented of phage amplification in the airways is important but does not appear to be accompanied by evidence of phage in blood. That does not appear to have been tested and is probably worth a mention because the question of whether Pseud DNA spills into blood or phages are detectable in light of the hint at an inflammatory response early on.