If you’ve read very much of this blog, you know that my opinion of phage science in general is not very high. I haven’t been especially nice about expressing this opinion either, tossing out epithets like “weak”, “porcine” and “innumerate”.
The latter, I suspect, riled up Stephanie Strathdee, one of the co-authors of the paper describing an apparently successful PT treatment of a patient with a systemic Acinetobacter infection who had failed antibiotic therapy. You can read the thread here.
She suggested that my knowledge of the field was several years out of date, and that I was ignoring the great work done recently. That might be a fair and true criticism – before I wrote that blog post I searched clinicaltrials.gov for “phage therapy” and PubMed for phage therapy with a clinical trial filter. No search strategy is perfect and perhaps I missed something good. Dr. Strathdee didn’t respond to my request for examples, so I went back to PubMed to do a broader search, reasoning that perhaps there were some compelling animal model studies that moved the field forward more than I expected.
That’s not what I found. I read through a few papers that were drearily incomplete with small data sets and very limited explorations of the unfortunately vast experimental space that phage therapy occupies. But I did find a nice review of the state of the science that was just recently published.
Martina Rossito and colleagues have critically (“argumentatively” is their word) appraised the state of PT as applied to cystic fibrosis. CF patients almost inevitably contract Pseudomonas lung infections. Pseudomonas, more than most bacteria, develops antibiotic resistance over the course of treatment, and patients frequently suffer severe infections that become untreatable.
This is a high-value/high-feasibility application for PT: a fairly common disease (about 30K CF patients in the US), a very typical infection etiology, high rates of resistance, high rates of biofilm formation. The lung is a good compartment for applying PT, as nebulizer and powder delivery technology is well-advanced. Clearance rates are slow, allowing the phages to persist and attack the infecting bacteria.
Rossito et al base their appraisal on Steve Abedon’s rules/suggestions for the types of data that clinical and preclinical studies should publish in order to move the field from anecdotal one-offs (where it has been since the 1920s), to a pyramid of data that the PT community can use to systematically build a safe and effective mode of therapy. Steve is one of the few scientists in the field whose writings and presentations at meetings impress me as being rigorous and grounded in the realities of drug development. No one gets to make the rules for everyone else of course, but his rules, if adopted, would do much to make PT a reality.
That’s not what’s happening, though. I’ll spare you the details, but quote a few sentences from Rossito et al, describing the 24 studies included in their review:
Only three studies in vitro reported phage functional characterization estimated by latent phase and burst size (the mean number of phages released per bacterial cell; Garbe et al., 2010; Danis-Wlodarczyk et al., 2015, 2016). Only two studies in vitro selected and tested phage activity and host range by efficiency of plating (EOP) (Henry et al., 2013; Uchiyama et al., 2016)…
Only three studies reported phage stability assessment on newly-isolated phages alone (Danis-Wlodarczyk et al., 2015, 2016) and in a phage cocktail (Supplementary Tables 3, 4; Lehman et al., 2016)….
Only three of the nine in vivo studies reported the formulated products used for administering the phage suspension to host animal models (Supplementary Table 4; Debarbieux et al., 2010; Morello et al., 2011; Alemayehu et al., 2012). Few papers, therefore, gave essential information on phage functional characterization, selection, activity, genomic sequence, stability, and the formulated products used in phage suspensions useful for further studies on phage therapy in CF PA.
And so on.
It’s true that these sorts of data are not commonly reported in preclinical studies of drug candidates. But that’s because these candidates are usually developed by for-profit entities who want to maintain trade secrets so as not to enable competition. But Merck or Pfizer or J&J are not going to develop PT. If it is going to happen, it’s going to be the work of clinicians and not-for-profit entities. They have to share everything.
These groups are used to doing R&D; they will have to learn to do product development, a very different discipline. Doing just enough to get a publication is not going to cut it. You have to thoroughly characterize every step of your formulation and therapeutic protocol. Not just doing things that have been working, but deliberately and systematically altering every parameter so that you know what makes the product fail or become unreliable. The use of statistical design of experiment techniques – something most clinicians and biologists are utterly ignorant or suspicious of – is essential.
I suspect Dr. Strathdee has been feeling pretty good about that patient that got better (and perhaps others yet unpublished) and was offended to get called out for doing insufficient work. There have been a number of kumbaya articles claiming that PT has finally arrived.
But here’s the thing – we don’t really know that PT worked. Yes, the patient got better, but it took months of daily phage doses. If my analysis of phage kinetics is correct, it might have worked much faster if they had administered 100X more phage, thereby sparing the patient weeks of suffering. They couldn’t because endotoxin levels were limiting. Did they have a truly optimized protocol for maximizing phage concentration and limiting endotoxin levels? I don’t know, but I’ll bet they don’t either – that would require a systematic exploration of the design space for purification.
This is not nitpicking. Doing it right could lead to an important therapy that reliably saves lives and reduces suffering. Putzing along for another twenty years, reporting anecdotal cures is not acceptable.
Yeah, I was kind of nasty. But reality – particularly drug development reality – is a bitch. Kumbaya doesn’t cut it. We need rigorous quantitative analysis and evaluation every step of the way. Accepting anything less means accepting avoidable suffering, and I won’t do it.
1 thought on “Phage therapy science – still weak”