Bacteriophage therapy for infections in CF
Benjamin K. Chan1, Gail Stanley2,
Mrinalini Modak3, Jon L. Koff2, Paul
E. Turner1,4,#
1 Department of Ecology and Evolutionary Biology, Yale
University, New Haven, CT 06511
2 Department of Internal Medicine, Yale School of
Medicine, New Haven, CT 06520
3 Yale New Haven Hospital, New Haven, CT 06520
4 Program in Microbiology, Yale School of Medicine,
New Haven, CT 06520# Correspondence, paul.turner@yale.edu
Abstract. Pseudomonas aeruginosa and Staphylococcus aureusare bacterial pathogens frequently associated with pulmonary
complications and disease progression in cystic fibrosis (CF) patients.
However, these bacteria increasingly show multiple resistance to
antibiotics, necessitating novel management strategies. One possibility
is phage therapy, where lytic bacteriophages (phages; bacteria-specific
viruses) are administered to kill target bacterial pathogens. Recent
publication of case reports of phage-therapy treatment of
antibiotic-resistant lung infections in CF has garnered significant
attention. These cases exemplify the renewed interest in phage therapy,
as an older concept that is newly updated to include rigorous collection
and analysis of patient data to assess clinical benefit, while informing
the development of clinical trials. As outcomes of these trials become
public, the results will valuably gauge the potential usefulness of
phage therapy to address the rise in antibiotic-resistant bacterial
infections. In addition, we highlight the further need for basic
research on accurately predicting the different responses of target
bacterial pathogens when phages are administered alone, sequentially or
as mixtures (cocktails), and whether within-cocktail interactions among
phages hold consequences for the efficacy of phage therapy in patient
treatment.
Extensive antibiotic use during the previous century has resulted in an
alarming rise in antibiotic resistant infections. These infections,
particularly for bacterial pathogens which are resistant to multiple
classes of antibiotics [also known as multi-drug resistant (MDR)],
significantly contribute to increased morbidity and mortality
[1]–[4]. As a result, we have been forced to increase our
reliance on drugs of last resort [5], [6]. Predictably, bacteria
have now emerged that are resistant to even these drugs [5],
[6]. While in some cases these isolates may not have spread widely,
concern is warranted, and the development of new antibiotics and
management strategies is urgently needed.
While antibiotic resistance increases, the antibiotic development
pipeline is lagging far behind with limited evidence that novel
antibiotics are being discovered and developed [7]. Therefore,
efforts by the World Health Organization have resulted in the creation
of a list of ‘priority pathogens’ for antibiotic development, due to
their public health relevance. This list contains six bacterial
pathogens notable for their high levels of antibiotic resistance, as
well as their ability to escape conventional therapies. These bacteria
are Enterococcus faecium , Staphylococcus aureus ,Klebsiella pneumoniae , Acinetobacter baumanii ,Pseudomonas aeruginosa , and Enterobacter spp. are named by
the ESKAPE acronym because they are becoming increasingly prevalent as
MDR and pan-drug resistant (PDR) organisms that frequently escape
approved treatments. For cystic fibrosis (CF) patients and their care
teams, P. aeruginosa and S. aureus are common pathogens
associated with pulmonary complications and disease progression. Because
of the need to frequently treat pulmonary exacerbations with antibiotics
the CF community has been wrestling with the challenges of increased
antibiotic resistance for some time.
S. aureus is a Gram-positive bacterium that is highly relevant in
the CF lung, where it is frequently the first bacterial pathogen
cultured from those with CF [8]. Strains with resistance to
methicillin (i.e., methicillin resistant S. aureus [MRSA]),
have been shown to correlate with more rapid pulmonary decline, relative
to S. aureus strains which are sensitive to methicillin [9].
Attempts to actively eradicate MRSA upon detection have been successful,
resulting in improved FEV1 and BMI; however, spontaneous
eradication in the absence of intervention was also observed in the same
study [10]. Ultimately, presence of S. aureus can increase
the risk of subsequent P. aeruginosa infection by damaging lung
tissue and creating an environment favorable for growth [11]. Thus,
because S. aureus is one of the first identified pathogens in the
lungs of those with CF that paves the way for eventual P.
aeruginosa infection, control strategies are needed.
P. aeruginosa is a Gram-negative bacterium that thrives in myriad
natural and artificial environments, which vary from household sink
drains to the natural environment (soil and water sources) to clinics
and hospitals [12]. Thus, individuals with CF readily encounterP. aeruginosa in their home environment and when they receive
medical care. While P. aeruginosa is less common in pediatric CF
sputum, by adulthood, the majority of CF patients have P.
aeruginosa in sputum, which negatively correlates with CF lung function
[13], [14]. P. aeruginosa employs multiple strategies to
colonize the CF lung, and several of these are virulence factors that
induce lung inflammation and tissue damage, which may contribute to the
severity of pulmonary exacerbations. While the CF community also suffers
from a lack of new antibiotics, the recent development of CFTR
modulators may affect sputum colonization of P. aeruginosa and
the frequency of pulmonary exacerbations [15]. Although the
long-term durability of these effects remains may be limited [16].
While CF clinical outcomes and survival in the past 30 years has
improved, the persistence of S. aureus , P. aeruginosa , and
other Gram-negative bacteria (e.g., Burkholderia andAchromobacter ) that are increasingly resistant to antibiotics,
and the emergence of non-tuberculous mycobacteria, highlight the need to
develop novel antimicrobial therapeutics that, ideally, can also
decrease the inflammation and tissue-damaging virulence factors.
Ideally, such therapeutics will also have limited, if any, off-target
effects, which would provide an intervention that greatly benefits the
CF patients. One approach which has recently received an abundance of
attention, particularly in the CF community, is phage therapy.
Phage therapy is the use of lytic bacteriophages (viruses of bacteria)
to kill infectious bacteria. Such an approach harnesses phages, which
are abundant in nature, to efficiently and effectively kill specific
bacteria. Discovered in the early 1900s, phages have been used for human
therapeutics since then, but after the discovery of penicillin, phage
therapy continued in the former Soviet Union and Eastern Europe more
than Western countries. While much of the available data from this
experience have been translated from Russian, these reports estimate
that over 300,000 individuals have received phage therapy over the last
100 years. While there have been few reports of adverse events, access
to the data is limited. Furthermore, general consensus for phage therapy
is that it is safe in humans without evidence for toxicity
[17]–[20].
Because of the emergence of MDR bacteria, phage therapy is again being
seriously considered in Europe, the United States and elsewhere as a
potential approach to manage antibiotic resistant infections. In
addition to more clinical experience with phage therapy, which is
summarized below, the CF Foundation has provided support to the Center
for Innovative Phage Applications and Therapeutics at the University of
California, San Diego, and funding for phage therapy clinical trials in
CF from Armata Pharmaceuticals and Yale University (personal
communication, J.P. Clancy M.D., Vice President of Clinical Research, CF
Foundation).
The existing phage therapy clinical trials have studied infectious
diarrhea, wound infections, and chronic otitis media with different
metrics of efficacy, but no significant side effects. While no clinical
trials have been completed in CF at this time, there are published
reports of individuals with CF being treated with phage therapy before
or after lung transplantation. In 2008, Kutateladze and Adamia described
the treatment of pulmonary infections at the Eliava Institute of
Bacteriophages, Microbiology, and Virology in Tbilisi, Georgia [21].
This included adding phages to CF care. CF patients included both
infants and adults, who received phage therapy to treat S. aureusand P. aeruginosa by nebulizer over approximately 1 week. The
authors reported that phage therapy caused decreased densities of sputum
bacteria, improvements in patient health, and extension of time until
subsequent bacterial infection. Kvachadze et al., reported a case of
phage therapy in a pediatric CF patient who received phage therapy to
treat S. aureus and P. aeruginosa via nebulization a total
of 9 times once every 4 to 6 weeks [22]. Phage therapy was reported
to result in reduction in bacterial titers and a 50% reduction in
administered antibiotics over 9 months of treatment. While there was no
change in chest imaging, both S. aureus and P. aeruginosawere undetectable after phage therapy. Subsequently, Aslam and
colleagues reported their experience with phage therapy in three
post-lung transplant patients [23]. One of these patients had CF
whose post-transplant course was complicated with Burkholderia
dolosa . Phage therapy was administered with antibiotics. Initial phage
therapy resulted in decreased fever and leukocytosis with evidence for
improved consolidations on chest imaging. Continued therapy over
approximately six weeks led to improved functional status. However,B. dolosa after ten weeks resulted in pneumonia, sepsis, and
ultimately multi-organ failure, and this patient was transitioned to
hospice. A 26-year-old woman with CF listed for lung transplant was
treated with intravenous phage cocktail of four lytic phages every six
hours for eight weeks in addition to antibiotics for MDR P.
aeruginosa . Phage therapy resulted in decreased supplemental oxygen use
and reduced sputum production without evidence for recurrent
exacerbation for 100 days, and this patient subsequently received lung
transplant nine months later. Most recently, Dedrick et. al. [24],
reported the use of phage cocktail, which included an engineered phage,
to treat disseminated Mycobacterium abscessus in a fifteen
year-old individual with CF who was post-lung transplant. In this case,
phage therapy resulted in decreased size of skin lesions, and
improvement in lung function, liver function, chest imaging and weight
gain. Here at Yale, we’ve successfully treated multiple CF associated
infections, observing a rapid decrease in sputum bacteria density and
correlating with significant improvement of FEV1%pred. This experience,
and those of others have been invaluable as we and others proceed with
clinical trials.
In summary, while the existing published literature in CF is limited,
there are encouraging signs that phage therapy may provide clinical
benefit, which needs to be confirmed in rigorous, controlled clinical
trials. In addition, these cases highlight potential differences in
route of phage administration (e.g., intravenous vs. nebulized),
concomitant use of antibiotics, short-term or longer-term duration of
therapy, choice of phage, and single phage vs. multi-phage
“cocktails”. These differences could have a major impact on outcomes,
and through studies propelled by recent interest, the specific
contribution of each approach can be identified.
Above, we discuss how the resurged interest in phage therapy and
promising outcomes from administered treatment in human patients
highlight the need for abundant basic and applied research. Here, we
highlight two examples that pertain to (i) effects of differing
strategies for administering phages and whether these would cause
problems in accurately predicting how target bacteria will
evolutionarily respond, and (ii) cocktail approaches to phage therapy
and how the supposed benefits of cocktails for covering ‘genotype space’
of target bacterial strains might be offset by the costs of phage-phage
interactions that diminish cocktail efficacy.
Recently, Wright et al. [25] examined evolution of bacterial
resistance to phages, by studying interactions between P.
aeruginosa strain PA01 and phages that bind to host cells by utilizing
either lipopolysaccharide (LPS) or the type-IV pilus. In particular,
they assessed whether sequential exposure to the two different phages
caused mutational evolution of phage resistance in the target bacterial
population that was distinct from resistance evolution when the two
phages were placed together in a cocktail (mixture). This possible
differing outcome relates to the general idea that biological evolution
in response to selection by a single environmental challenge (each phage
alone), may differ from molecular evolution in response to the same
environmental challenges when experienced simultaneously (phage mixture)
(e.g., see Kassen 2002 on evolution in simple versus complex
environments [26]) Results showed that the genetic basis of
bacterial resistance in response to sequential phage exposure generally
led to mutations in genes associated with the binding target (LPS or
type-IV pilus), indicating evolution of additive resistance in the
target bacterial populations. Whereas, this accumulation of multiple
receptor-specific resistance mutations was never observed when phage
pairs targeting different binding sites were applied simultaneously.
Rather, half of the observed phage-cocktail-resistant mutants showed
only LPS changes and no detected mutations in genes for the type-IV
pilus; the remaining mutants either surprisingly presented no mutational
changes or showed duplications in genes for bacterial recombination that
could not easily explain their supposed relationship to evolved phage
resistance. In this study, the data indicated that bacterial evolution
in response to sequential phage exposure was both more predictable and
interpretable, relative to bacterial evolution against the same phages
when mixed together as a cocktail [25]. In turn, because
phage-steering approaches are designed to leverage phage killing of
target bacteria alongside predictable clinically-useful evolution of
resistance trade-offs in bacterial pathogens, the study by Wright et al.
[25] suggests that administering phage cocktails can cause the
accuracies of these predictions to break down.
If bacteria evolve resistance to phage attack, this phenotypic change
may be costly, such as incurring a reduced growth ability in phage
resistant mutants. Such pleiotropic costs of phage resistance should be
commonplace, owing to the tendency for phages to evolve specific binding
to highly-conserved structures on host cells [27], [28];
altering (or deleting) these binding targets should therefore debilitate
bacterial growth. By this logic, simultaneous evolution of
cross-resistance to phages that bind to different cellular targets may
be more costly for host bacteria, due to evolutionary constraints (i.e.,
cross-resistance should be unlikely when phages bind differently)
combined with the expectation that distinct resistance mutations for
each phage will cause fitness to decrease additively or negatively
synergistically. Clearly, the evolution of high-cost resistance in
target bacteria would be advantageous for phage therapy applications,
because this outcome should improve the goal of reducing bacterial load.
However, there is mixed evidence for the prediction that simultaneous
evolution of resistance to multi-phage cocktails should be more costly
for bacteria, relative to fitness costs associated with sequential phage
application. Wright et al. [25] observed that sequential resistance
was more costly than simultaneous resistance in half of their
replicates. Similarly, in a separate study involving additional phage
pairs and P. aeruginosa strain PA01, in vitro data showed
no differences between sequential and simultaneous phage exposure, in
terms of bacterial clearance or cost of resistance [29] . Again, the
equivocal results for phenotypic costs of resistance against sequential
versus simultaneous phages in these studies suggest that cocktails are
not necessarily the best therapeutic option; the resulting resistant
bacteria may be equally debilitated in growth, whereas only
sequentially-administered phages tend to cause predictable underlying
genetic changes in resistant mutants [25].
Although phage cocktails are popularly touted in the literature, it is
known that the temporal dynamics of multi-phage therapy can be
complicated [30]. As an example, phages may be mixed together in
equimolar amounts and applied simultaneously, but (as with any
biological entities) there is no guarantee that the phages will be
equivalent in their rates of cell entry, intracellular rates of genome
replication and packaging, lysis timings, and burst sizes. These
different characteristics would easily cause the phage cocktail to
deviate from the intended equal representation of virus types in the
mixture as phage particles interact with host cells during therapy,
resulting in different relative killing abilities and selection
pressures exerted on the bacteria. In theory, any concerns over this
possible imbalance could be eliminated if the phages in the cocktail
were chosen specifically because they were equal in the various phage
traits. However, this notion seems unrealistic, given that even single
mutations can radically alter traits of phages that are otherwise
isogenic [31], [32]. Moreover, this strategy would likely
undermine the generally popular goal of discovering and combining
functionally distinct phages into a single cocktail based solely on
their different host range (e.g., cell-binding) properties. Aside from
differing growth capabilities of phages that can alter the intended
equal representation of phages in an administered cocktail, effects of
phage-phage intracellular competition can complicate predictions of
phage-therapy efficacy. Multiplicity of infection (MOI) is defined as
the ratio of phage particles to susceptible bacterial cells, when these
microbes are combined. MOI is easily controlled in laboratory
experiments with phage and bacteria, because it is trivial to accurately
measure particle titers and cell densities to achieve the intended
initial ratios [33]. However, MOI cannot be controlled in
vivo ; phage titer is known upon delivery, but the exact size of the
infecting bacterial population is very difficult to ascertain. This
becomes important because elevated MOIs (i.e., ratios exceeding 1.0)
allow higher probabilities that multiple phage particles co-infect the
same host cell, producing phage-phage competition that can select for
virus traits which foster antagonistic interactions between viruses
[34]–[37]. An extreme example is the evolution of “cheating”
where phages are selected to selfishly utilize proteins coded by other
co-infecting viruses, causing both wildtype and non-cheater variants of
viruses to be competitively disadvantaged for intracellular growth
[38], [39]. The net result is for cheater variants to take over
the phage population, which surprisingly can occur even if there is a
high phenotypic cost of cheating in the phage. For the latter reason,
phage-phage competition can both reduce variability of the overall phage
community, as well as decrease the efficacy of phage particle production
if phage competitiveness does not correlate with an advantage in burst
size [36]. Obviously, the concern over adverse phage-phage
interactions become minimized if phages are used alone or sequentially,
relative to the complexities that might occur when cocktails are
employed. Nevertheless, we note that these and other possible negative
consequences of phage-phage competition remain poorly studied in phage
therapy, making it currently difficult to predict how differing
treatment strategies would be affected by the phenomenon.
As planned trials are initiated and completed, the outcomes should allow
us to determine the impact of phage therapy on pulmonary infections
associated with CF. Individual cases treated by our group and others
have suggested a clinical benefit from the use of phage therapy in
complicated cases, but until we have data from a controlled clinical
trial, we can only speculate.