Bacteriophage Clinical Use as Antibacterial “Drugs”: Utility and Precedent

ABSTRACT

For phage therapy—the treatment of bacterial infections using bacterial viruses—a key issue is the conflict between apparent ease of clinical application, on the one hand, and on the other hand, numerous difficulties that can be associated with undertaking preclinical development. These conflicts between achieving efficacy in the real world versus rigorously understanding that efficacy should not be surprising because equivalent conflicts have been observed in applied biology for millennia: exploiting the inherent, holistic tendencies of useful systems, e.g., of dairy cows, inevitably is easier than modeling those systems or maintaining effectiveness while reducing such systems to isolated parts. Trial and error alone, in other words, can be a powerful means toward technological development. Undertaking trial and error-based programs, especially in the clinic, nonetheless is highly dependent on those technologies possessing both inherent safety and intrinsic tendencies toward effectiveness, but in this modern era we tend to forget that ideally there would exist antibacterials which could be thus developed, that is, with tendencies toward both safety and effectiveness, and which are even relatively inexpensive. Consequently, we tend to demand rigor as well as expense of development even to the point of potentially squandering such utility, were it to exist. In this review I lay out evidence that in phage therapy such potential, in fact, does exist. Advancement of phage therapy unquestionably requires effective regulation as well as rigorous demonstration of efficacy, but after nearly 100 years of clinical practice, perhaps not as much emphasis on strictly laboratory-based proof of principle.

INTRODUCTION

Phage therapy is the use of bacterial viruses to reduce or eliminate bacterial infections. As such, phage therapy has been employed clinically for approximately 100 years. It is possible, nevertheless, that more English-language reviews and commentaries on bacteriophage use as antibacterial agents are published yearly (e.g., for 2015 [1–19] and for approximately the first third of 2016 [20–30]), than there are people within the borders of the United States who are subject to officially sanctioned phage treatment. This contrasts with well over 100 million courses of antibiotic that are prescribed per year (31).

Phages—or bacteriophages, as phages are more formally known—more broadly can be employed as a form of biocontrol within a number of nonclinical contexts. For example, there is phage treatment of foods such as for reducing pathogen numbers, most notably of the foodborne pathogen Listeria monocytogenes (13, 28, 29, 32, 33). Also, for example, there is the treatment of plants for reducing the quantities of bacterial plant pathogens found in association with crops, particularly as marketed by the Utah-based company OmniLytics, Inc. (34–38). Phages especially represent alternatives as well as adjuvants to the use of small-molecule antibiotics. Notably, they are nonxenobiotic, that is, they do not consist of chemically unnatural materials, and typically they are effective against antibiotic-resistant bacteria. These viruses can be useful for combating unwanted bacteria in part because they have evolved for billions of years as specifically acting antibacterial agents. Phages are highly abundant and diverse, generally easy to work with, and represent two of seven “tier 1” alternatives to antibiotics for systemic use as considered by Czaplewski et al. (39). Particularly given informed phage choice, along with appropriate purification, phages do not tend to display substantial toxicities toward, for example, human tissues, or even against nontarget portions of the human microbiome.

Notwithstanding these positive factors, phage therapy is not widely employed in clinical practice. Part of the reason for its relative absence is historical and accompanying tradition, part is due to regulatory constraints, and another part (40) is that only relatively recently has there been widespread recognition of a need for alternatives to antibiotics as antibacterial drugs. This review provides an overview of the arguments for why further development of phages as antibacterial “drugs” to supplement or in some cases even replace small-molecule antibiotics may be warranted. A related aim is to suggest that substantial clinical development already has occurred.

Glossary of Terms

Toward introducing phage therapy and related concepts, in this section I define a number of relevant terms.

Biocontrol Here this term refers to the use of phages as antibacterial agents; phage therapy is a type of phage-mediated biological control of bacteria, i.e., “biocontrol” as used here.

Crude lysate Phage lysate of host bacteria from which phages have not been substantially purified; e.g., at most a crude lysate has been subject to filter sterilization to remove intact bacteria along with relatively large pieces of bacterial debris but not removal of smaller debris or sufficient reduction of contamination with endotoxin.

Endotoxin Lipid-based component of the outer membranes of Gram-negative bacteria.

Lysate Typically fluid product of phage-induced lysis of bacterial cultures.

Lysogen Bacterial cell hosting one or more lysogenic cycles/prophages.

Lysogeny Description of a phage infection that is ongoing, usually over large numbers of bacterial generations, and which does not produce phage virions unless lysogeny is exited; lysogeny is the product of a lysogenic infection/lysogenic cycle.

Phage Virus that can productively infect bacterial cells; “phage” can refer to the virions or the virus more generally, e.g., “phage infection”; also known as bacteriophage or bacterial virus.

Phage therapy Application of phages as medicinals, particularly as used in the clinic or by veterinarians; a specific type of phage-mediated biocontrol of bacteria.

Professionally lytic phage Phage that is not temperate, nor closely related to temperate phages, and which releases phage progeny via lytic cycles; “virulent,” “strictly lytic,” “obligately lytic,” etc. are also descriptions of phages that are both lytic and not temperate, but also that are not necessarily distantly related to temperate phages; professionally lytic phages are preferred for phage-mediated biocontrol.

Prophage Phage genome as observed during lysogenic cycles; prophages traditionally are seen as integrated into the chromosomes of their bacterial host, though plasmid prophages exist as well.

Temperate phage Phage that is capable of displaying lysogenic cycles; most temperate phages are also lytic phages, that is, releasing virions from productive infections via host-cell lysis; contrast temperate with virulent, strictly lytic, obligately lytic, professionally lytic, or obligately productive phages.

Staphage lysate Formulation consisting of phage-lysed Staphylococcus culture in which associated phages may or may not represent an active ingredient.

Transduction Virus-mediated movement, especially of nonviral DNA from one potential host cell to another; specialized transduction is the movement of relatively small amounts of bacterial DNA as mediated by excising prophages, whereas generalized transduction is the movement of relatively large amounts of bacterial DNA as mediated by phages which otherwise are less destructive to bacterial genomes and less selective when packaging DNA into virions.

PHAGE THERAPY AND PHAGE-MEDIATED BIOCONTROL

The use of phages as antibacterial agents can be differentiated into uses that, in their utility, are more antibiotic-like versus more disinfectant-like. Because successful introduction of phages into clinics may be informed by efforts to employ phages more widely as antibacterials, in this section I differentiate the therapeutic use of phages from a broader utility as biological control agents of undesirable bacteria. This discussion is provided in part to briefly overview this nontherapeutic antibacterial use of phages.

Phage Therapy

Phage therapy literally is the use of bacterial viruses as therapeutic agents (see Table 1 for a summary of published modern human treatment). Usually these therapeutic aspects are assumed to stem directly from phage antibacterial activity. There exist, however, other therapeutic utilities of phages, for example, as delivery vehicles of toxins or DNA (e.g., 27, 41, 42) or as immune system modulators and characterizers (43–46). Among bactericidal phage therapies, it is also possible to distinguish prophylaxes from the targeting of already contaminating, colonizing, or infecting bacteria.

TABLE 1.

Modern clinical or human experimental phage therapy (English-language literature)

Targets	Comment	Number treateda	Substantial efficacy	Authors	Year	Reference
Various	Various treatments	62	47%b	Łusiak-Szelachowska et al.	2017	142
S. aureus	Diabetic foot ulcers	9	100%	Fish et al.	2016	183
E. coli	Diarrhea	79	0%c	Sarker et al.	2016	114
S. aureus	Eye treatment	1	100%	Fadlallah et al.	2015	185
Various	Various treatments	>100	95%d	Kutateladze	2015	186
P. aeruginosa, S. aureus	Safety trial (topical dosing)	9	NAe	Rose et al.	2014	116
E. coli, Proteus	Safety trial (oral dosing)	15	NA	McCallin et al.	2013	91
Various	Various treatments	153	40%	Międzybrodzki et al.	2012	102
E. coli	Safety trial (oral dosing)	15	NA	Sarker et al.	2012	113
P. aeruginosa	Urinary tract infection	1	100%	Khawaldeh et al.	2011	188
P. aeruginosa, S. aureus	Cystic fibrosis patient	1	NAf	Kvachadze et al.	2011	189
NA	Cystic fibrosis patients	NA	NA	Kutateladze and Adamia	2010	144
E. faecalis	Chronic bacterial prostatitis	3	100%	Letkiewicz et al.	2009	190
E. coli, P. aeruginosa, S. aureus	Safety trial (topical dosing)	39	NAg	Rhoads et al.	2009	117
P. aeruginosa	Chronic otitis	12h	25%i	Wright et al.	2009	118
NA	Cystic fibrosis patient	1	NA	Kutateladze and Adamia	2008	137
S. aureus	Gastrointestinal colonization	1	100%	Leszczyński et al.	2008	191
S. aureus	Economics of methicillin-resistant S. aureus treatment	6	NA	Międzybrodzki et al.	2007	206
P. aeruginosa	Burn wound infection	1	100%	Marza et al.	2006	192
Staphylococcus spp.	Otitis media	1	0%j	Weber-Dąbrowska et al.	2006	207
E. coli	Safety trial (oral dosing)	15	NA	Bruttin and Brüssow	2005	112
S. aureus	Radiation burn and PhagoBioDerm	2	100%	Jikia et al.	2005	193
S. aureus	Hand-washing experiment	NA	NA	O’Flaherty et al.	2005	68
Various	Septicemia	94	85%	Weber-Dąbrowska et al.	2003	208
Various	Wounds, ulcerations, and PhagoBioDerm	96	70%	Markoishvili et al.	2002	194
S. aureus	Peripheral neutrophil functioning	37	73%	Weber-Dąbrowska et al.	2002	209
Various	Infections of cancer patients	20	100%	Weber-Dąbrowska et al.	2001	210
Various	Various treatments	1,307	96%	Weber-Dąbrowska et al.	2000	187
Various	Chronic SBIk of skin	31	74%	Cisło et al.	1987	211
Various	Determination of neutralizing antibody	57	77%	Kucharewicz-Krukowska and Ślopek	1987	139
Various	SBI	550	85%l	Ślopek et al.	1987	181
Various	SBI	56	88%	Weber-Dąbrowska et al.	1987	182
Various	SBI	370	85%	Ślopek et al.	1985	178
Various	SBI in children	114	89%	Ślopek et al.	1985	179
Staphylococcus	SBI	254	85%	Ślopek et al.	1985	180
Various	SBI	150	81%	Ślopek et al.	1984	177
Various	SBI	138	88%	Ślopek et al.	1983	103
Various	SBI	184	88%	Ślopek et al.	1983	176
S. aureus (or other)	Treatment of hidradenitis suppurativa with Staphage Lysatem	8	75–100%n	Kress et al.	1981	212

^a Throughout this review, numbers treated refers to non-placebo-treated individuals for whom studies or treatments were completed.

^b Clinical improvement or better.

^c Relatively little effort in this trial appears to have been devoted to ensuring that sufficient numbers of phages of desired specificity would be present in the vicinity of target bacteria.

^d It is not obvious what types of cases this percentage refers to, e.g., both acute and chronic or just acute.

^e NA, either not available or not applicable.

^f Endpoints of treatment of cystic fibrosis patients are sufficiently ambiguous that “NA,” not applicable, has been assigned despite arguably positive results.

^g Phage choice in this trial was biased toward better phage in vitro characterization rather than toward ability to treat those strains of wound-infecting bacteria encountered.

^h Plus 12 placebo controls.

ⁱ Based on only single phage dosing in the course of a phase I/II clinical trial.

^j Phage therapy had some positive impact, but clearance did not occur until subsequent lactoferrin treatment was employed.

^k SBI, suppurative bacterial infections.

^l This publication is an overview of previously published Ślopek et al. articles; here and in these others, results indicated as “++++” or “+++” are counted as “substantial efficacy” versus merely positive results; typically, these are treatments of chronic bacterial infections that had already been subject to antibiotic therapy, and in some cases (about one-quarter) antibiotic treatment also coincided with phage treatment.

^m Staphylococcal phage lysate.

ⁿ This article is not necessarily looking at phage therapy in the sense of phages penetrating to and then directly killing target bacteria; i.e., efficacy instead may be a consequence of immune system stimulation. From the abstract: “Six of the 8 patients reported noticeable improvement in odor, consistency, and amount of drainage and considerable decreases in pain. Seven of the 8 patients reported improvement in the ability of lesions to drain spontaneously, and a decrease in the frequency of inflammatory nodules. All 8 patients reported that the inflammatory periods were definitely shorter.”

Considering just phage antibacterial aspects, we can view phage therapy as a subset of phage-mediated biological control of bacteria (47), or simply “biocontrol” for short. Elsewhere, and somewhat consistent with usage by others, I have suggested that the phrase “phage therapy,” as a subset of biocontrol, should be used particularly in those circumstances where organisms, e.g., ourselves, are individually treated with phages to reduce the numbers of bacteria that in some manner are affecting the treated individual. All other uses of phages as antibacterial agents, by contrast, may be referred to as different forms of biocontrol (next section) rather than explicitly as phage therapy (48). Thus, phage-mediated removal of colonizing pathogenic bacteria from individual affected humans or from affected individual animals is unambiguously phage therapy. Given such treatment of individuals directly with phages, as though phage formulations were complex drugs, we can consider the process of phage therapy from a pharmacological perspective. Drugs, in particular, have specific chemical properties, are delivered to bodies in some manner, are able to reach and maintain different concentrations in different regions of the body, and can impact the body in both intended and unintended ways.

A number of reviews have considered various aspects of phage therapy pharmacology (e.g., 49–53). Furthermore, a diversity of pharmacologically familiar application approaches have been employed clinically or experimentally in phage therapy. These include local application, systemic application, and the attachment of phages to surfaces to prevent subsequent colonization. Local application traditionally includes application to body surfaces and wounds, inhalation into the lungs, and delivery to various mucus membranes including the oral cavity and vagina. Systemic delivery, at least in principle, may include per os, via suppositories, or via injection. Surgically, it is also possible to infuse phages relatively deeply into tissues. Phages, in addition, may be packaged in various ways prior to delivery, including into liposomes (54, 55) or encapsidated by various additional means (6). To minimize complications, typically, phages that are lytic, that are not capable of displaying lysogeny (and thus are not temperate), and that are not closely related to temperate phages are preferred for phage therapy purposes. These are phages that can be described as “professionally lytic” (56).

Phage-Mediated Biocontrol

Biocontrol, other than that which can be described unambiguously as “phage therapy” (previous section), includes the disinfection of nonliving surfaces or environments (e.g., 57–61). Representative forms of biocontrol thus are the treatment of aquatic environments to remove unwanted cyanobacteria (for references, see reference 62) or phage-mediated removal of certain bacteria during water treatment (23). So too, one can view as forms of biocontrol phage application to foods, to food animals preharvest, or to food-handling equipment, in each of these cases to remove foodborne pathogens which otherwise could affect humans (6, 13, 14, 22, 28, 29, 32, 33, 63) or, instead, to remove organisms which can affect food quality (64, 65). The treatment of biofilms is similar, particularly other than in the course of treating diseases (e.g., 1, 66). Thus, for example, the removal of biofouling biofilms from inanimate surfaces (25), e.g., membrane filters (67), would constitute of a form of biocontrol but should not be described as medicinal therapy.

More ambiguous is the disinfection of surfaces of our own bodies such as in the course of hand-washing (68). This is biocontrol, but it also, at least arguably, could be viewed as therapeutic as well, and particularly so if removal is of more than simply transient microbiota, e.g., clearance of methicillin-resistant Staphylococcus aureus from body surfaces prior to surgery (69). Also ambiguous is the application of phages to plants (70–73) or to fungi such as mushrooms (74) to remove bacterial pathogens affecting these organisms. These are examples simply of biocontrol to the extent that phages are applied to multiple individuals simultaneously, e.g., spraying fields to treat bacterial diseases affecting plant species versus, for example, soaking individual plants or seeds prior to planting, with the latter arguably being more therapeutic in its nature. So too is the newly explored application of phages to control bacterial disease in honey bees, which also can include phage application in feed (75) versus phage application less specifically to entire hives. This distinction can be seen with poultry as well, i.e., spraying pens to inoculate animals against, for example, colibacillosis (76) versus feeding phages directly to animals (77).

The use of phages in the context of aquaculture (e.g., 78) may be viewed similarly. Ambiguity between biocontrol and phage therapy is seen particularly with the release of phages directly into water (79) rather than more individually directed application such as feed-borne phage delivery (80), with the latter approach, toward treating or preventing disease in husbanded animals, e.g., farmed fish, more reasonably described as phage therapy rather than biocontrol. Phage treatment of invertebrate feed animals in aquaculture to remove pathogens capable of impacting farmed fish (81), however, would clearly constitute a form of biocontrol rather than therapy, just as is the case with antipathogen treatment of foods prior to human consumption (above). In each of these cases, broader phage application (biocontrol) has greater potential to impact bacteria whether or not those bacteria are directly associated with potentially negatively affected host organisms. This contrasts with phages predominantly reaching only those bacteria currently colonizing treated individuals, i.e., as seen with phage therapy, which is the primary emphasis of this review.

UTILITY

The utility of phages as antibacterial agents is derived from two basic phage properties: (i) phage virions primarily consist of proteins and nucleic acid, that is, they are not xenobiotic (59); and (ii) the mechanisms of phage antibacterial activity, by and large, are somewhat robust, consisting of multiple mechanisms per phage, and are not thought to overlap substantially in terms of molecular targets with those of existing antibiotics (e.g., 82). This section considers the safety of phages themselves, i.e., virions as well as phage infections, the safety of the formulations within which therapeutic phages are carried, the impact of phages on nontarget microbiota, and the safety track record of phage administration—preclinical, clinical, and in terms of clinical trials. Also relevant to phage utility, as nonxenobiotic pharmaceuticals, is the idea of emergent property pharmacology. For balance, this section also considers various countering issues, including discussion of the potential for phages to interact with immune systems. The latter, it should be noted upfront, and rightly or wrongly, is typically seen as much less of an issue within the phage therapy community (e.g., 83) than as typically viewed from outside of that community. Formal listings of “pros and cons of phage therapy” can be found in a number of publications (e.g., 1, 4, 59, 84–86). Issues of efficacy are addressed in the subsequent section.

Low Toxicity of Well-Chosen Phages

Despite the great genetic diversity that is seen among phages (e.g., 87), the chemical diversity of phage virions—as basically protein- and nucleic acid-based entities—pales in comparison to the chemical diversity seen among drugs generally, e.g., antibacterial drugs (88). The latter, especially at higher concentrations, can be toxic to bodies by mechanisms that can be unrelated to the drugs’ mechanisms of antibacterial activity. Phages, by contrast, tend to be relatively inert within bodies other than while displaying bioactivity during their interactions with target bacteria. As a consequence, minimum toxic concentrations for phages within bodies tend, at least so far as is understood, to be somewhat higher than minimum effective concentrations. Due to the resultantly large therapeutic index, i.e., toxic dose relative to effective dose, the application of phages to bodies tends not to result in phage-virion-associated side effects, and this is true even given the potential for phages to amplify their numbers in situ in the course of eradicating target bacteria. Crucially, the relative lack of chemical diversity of phage virions, i.e., their nonxenobiotic nature, can greatly simplify the process of developing specific phage products.

Despite the relative lack of direct, chemical toxicities associated with phage virions, and their comparative inertness in the absence of interaction with target bacteria, a variety of concerns typically are raised regarding the potential for phages to harm bodies. These include the prospect of phages interacting with immune systems, their potential to encode harmful gene products such as exotoxins, and their ability to modify target bacteria in ways that are potentially harmful to us, e.g., by converting bacteria to lytic products or by transducing DNA between bacteria. It is important to note, though, that especially given proper phage choice (34, 89, 90), particularly of appropriately characterized and purified professionally lytic phages, these concerns in practice have not been known to result in substantial harm to treated organisms or environments. Furthermore, Pirnay et al. (53) describe concern about transduction as “optional.” Thus, an important utility as well as a safety characteristic of well-chosen phages is their generally low toxicity.

Formulated Products

Phage formulated products consist not just of phage virions, which can be viewed as active ingredients, but also of various carriage materials (34, 89). The latter most notably can include lysis products produced during phage preparation—which can be toxic and/or contribute to formulation immunogenicity—though carriage materials can also include intrinsic components of growth media as well as materials released from bacteria even in the absence of lysis. These latter issues to a degree can be controlled in the course of choosing the bacterial host and growth media used to generate phage stocks. It is especially useful to avoid, if possible, hosts for phage-stock generation that are capable of producing toxic products since those hosts not only may produce those products but will also carry potentially transducible genes that underlie that production. Included in the list of “genes” to avoid are also prophages, because these can potentially contaminate phage stocks as virions and also transduce host genes. In principle, though, it should be possible to analyze phage stocks or formulated products (91, 92) for even relatively low-level phage-virion carriage of bacterial genes (e.g. 93, 94).

In terms of intrinsic media components, a reasonable argument can be made that it can be best to avoid ingredients that are of animal origin, e.g., to avoid the introduction of prions or animal viruses into formulations (34, 85, 95). Generally, however, the most-considered carriage material-associated hazard is endotoxin. Endotoxin can be released from Gram-negative bacteria both with and without phage-induced lysis, and a variety of approaches have been developed for separating endotoxin from phage virions (34, 89, 96, 97). Such purification nevertheless is not essential under all circumstances, and indeed is generally less of a concern the further dosing is from parenteral. Thus, for example, topical dosing of wounds or oral dosing does not require nearly as much phage purification prior to use (34, 43), in contrast to intravenous (i.v.) dosing (26, 45). In general, therefore, while it is relevant to consider the toxicity of the formulations within which therapeutic phages are suspended, such toxicities tend to be mostly avoidable issues or seldom tend to be highly problematic in practice. See also Vandenheuvel et al. (17) for a recent review of issues of phage formulated products with an emphasis on enhancing stability.

Low Phage Impact on Nontarget Microbiota

The potential for phage impact on nontarget organisms is an issue predominantly because of how phages can be contrasted in this regard with nonphage antibacterial agents, that is, rather than this being an issue emanating from phage properties themselves. Further, this issue of impact on other than target bacteria can be broadened to that of low phage-virion influence on nontarget organisms more generally, that is, phage interaction with our own cells and tissues as well as our microbiome. Phages, as cytotoxic entities in particular, tend to be highly specialized in terms of what they are cytotoxic toward.

The range of bacteria that a typical phage can infect tends to be relatively narrow (98); indeed, Górski et al. (5) report, despite extensive experience, an ability to access what they call “an active specific phage” against S. aureus, Enterococcus faecalis, and Pseudomonas aeruginosa strains only 90%, 77%, and 72% of the time, respectively. The result is that phages tend to have relatively little impact not just on our own tissues but also on our microbiomes, beyond their impact, that is, on target bacteria. The caveat, however, is that at least in principle a phage’s transductive host range might be wider than its virion-productive host range or bactericidal host range. Transduction issues can be minimized by employing professionally lytic phages, and otherwise in the course of phage characterization (92), though as previously noted there are differences of opinion over whether transduction should be a concern beyond avoiding phages which have obviously high transduction likelihoods (53, 83). The potential for additional bactericidal activity also can be minimized in the course of phage characterization using spot testing, which ideally (99) is able to detect an ability of phages to kill host bacteria even if those bacteria cannot support phage replication. Even without such effort, it at least has not been obvious that phage therapy has substantial unintended impacts on normal microbiota. Antibiotic use, by contrast, can be somewhat disruptive of microbiomes, leading to negative health consequences (100).

Low Toxicity in Use

The points discussed in the three previous sections are somewhat theoretical. That is, there are reasons to expect that therapeutic phage products can be fairly safe, thus not resulting in any more than minor side effects. An important remaining question nevertheless is whether, empirically, phages are nontoxic or otherwise not damaging to bodies or environments during actual use. This question can be framed as one of emergent property pharmacology (subsequent section). That is, despite arguments for the potential for phages to be safe upon use, are they? There are four general areas of evidence for actual phage safety: animal testing, a substantial amount of clinical use, actual (though limited in number) clinical trials (17), and the nature of our interaction with phages as normally found in our environments.

In short, with animal testing (e.g., feeding phages to rats [101]), modern clinical use (102), or in the course of clinical safety trials (next section), there have been only relatively minor side effects associated with phage application, particularly relative to the symptoms associated with the infection being treated. According to Międzybrodzki et al. (102, p. 111)—and keeping in mind that crude lysates apparently were employed for these procedures (61, 103) so the basis of noted adverse reactions cannot be readily assigned—“The most frequent adverse reactions to phage preparations found in patients subjected to PT were symptoms from the digestive tract (nausea, abdominal pain, loss of appetite), local reactions at the site of administration of a phage preparation (redness, various uncomfortable feelings, but also individual cases of atopic dermatitis, urticarial blistering, as well as purulent blistering), superinfections (as seen in 4.6% of patients), and a rise in body temperature (3.3% subfebrile and 3.3% febrile).” Arguably, even fewer instances of side effects are seen given sufficient phage purification prior to use, particularly when avoiding more invasive application of crude phage lysates (e.g., parenteral), though use of crude lysates appears to have been the case during less modern phage use as therapeutics (e.g., references 85, 104, and 105).

Further, we are surrounded by phages. Except for those which are known to contribute to disease (e.g., exotoxin-encoding temperate phages [106, 107], which ideally will be avoided for phage therapy), so-called endogenous phages (108) with which our bodies regularly interact are not known to directly give rise to disease. For example, the human oral cavity may contain on the order of 100 billion (10¹¹) viruses, many of which are bacteriophages (109). Arguments can be made that this point simply has not been sufficiently well studied or that endogenous phages might give rise to dysbioses (e.g., 109, 110). Nonetheless, such issues can be mitigated in terms of phage therapy through proper phage choice: use of phages which display appropriately narrow host ranges and which are not temperate and do not otherwise encode potentially dangerous gene products. In general, the application of well-chosen and perhaps especially well-purified phages to bodies has not been known to result in a substantial worsening of patient health.

Safety Trials

A handful of phage therapy safety trials have been performed over the previous 20 or so years. These I review as follows. More complete discussion of phage clinical trials can be found, e.g., in references 7, 17, 61, and 111 and in a subsequent section.

Kutter et al. (61) provide a personal communication-based overview of an otherwise unpublished study from 2000 undertaken by Exponential Biotherapies, Inc. A dose of 5 × 10⁶ anti-Enterococcus phages was supplied intravenously to 12 healthy volunteers, which worked out to roughly 10³ phages per ml of blood. This is a dosage which I would argue could be many orders of magnitude too low for therapeutic purposes (51) unless in situ phage replication were able to raise these numbers substantially (50). Reported side effects consisted of a transient rash that was observed in one subject. No mention is made of the degree to which phages were purified prior to use, but a conservative assumption would be that they were substantially purified.

In 2005, Bruttin and Brüssow (112) published the results of a phase I phage-therapy safety trial. Phage T4 was supplied orally to 15 healthy adults in 150 ml of mineral water at concentrations of 10³ or 10⁵ per ml, thrice daily for 2 days. No side effects that the authors attributed to phage dosing were reported. In otherwise equivalent trials, similar results were seen with higher phage densities (113) as well as when treating children, the latter employing a commercially available Russian phage product (91). See too the more recent study of Sarker et al. (114), which is considered below. Brüssow (115, p. 138) noted that “as phages are considered relatively low risk when the [U.S. Food and Drug Administration] selection criteria are met, not very high numbers of treated humans are needed to demonstrate their safety.”

Rose et al. (116) describe the treatment of nine volunteers in a 2007 burn-treatment phage-therapy safety trial. A phage cocktail active against both P. aeruginosa (two phages) and S. aureus (one phage) was employed. Phages were present at a titer of 10⁹/ml, were separated from endotoxin prior to use, were applied only once at a density of 10⁷ per cm², and were used in conjunction with standard protocols, i.e., other than phage treatment. The authors concluded (p. 70), “No adverse events were reported and no clinical or laboratory test abnormalities related to the application of phages were observed.”

In 2009, Rhoads et al. (117) studied the impact of topical application of a cocktail of eight phages (10⁹/ml diluted 12.5-fold) active individually against Escherichia coli, P. aeruginosa, and S. aureus. Application was topical in combination with ultrasonic debridement to chronic venous leg ulcers of 18 volunteers (plus 21 placebo controls) over a 12-week period, with one application per week. No negative impacts in terms of either side effects or interference with wound healing were observed. See also the discussion of this trial in reference 61.

Also in 2009, Wright et al. (118) published a study in which chronic otitis was treated topically in 24 volunteers, none currently undergoing antibiotic therapy, using anti-Pseudomonas phages (cocktail of six). Half were treated once in one ear with 10⁵ phages and half with placebo. This was a phase I/II trial, that is, safety as well as preliminary efficacy were being studied. The author’s conclusion regarding safety (p. 353) was that “there were no reportable side effects, and no evidence of local or systemic toxicity.”

Emergent Property Pharmacology

An emergent property is a higher-level characteristic of a system that is difficult to predict from lower-level aspects. Thus, the collective properties of groups of organisms can be difficult to predict given knowledge gleaned solely from the study of individual organisms in isolation. In the case of emergent property pharmacology, the difficulty in question is prediction of the pathophysiological consequences of drug application solely from a drug’s chemical or other characteristics as determined in isolation from actual use. This issue arises in pharmacology particularly because the number of possible targets of drug action within a body can be vast, and at least some of the resulting interactions can be disruptive of homeostasis. In short, it can be difficult to predict all of a drug’s potential side effects prior to undertaking animal testing or clinical trials.

Emergent properties can substantially increase both the cost and duration of drug development because “most ‘hits’ [antibiotics] fail to proceed to clinical stages because of the subsequent discovery of mammalian toxicity” (40, p. 326). An additional consequence is the commercialization of drugs despite known toxicities, i.e., the standard package insert disclaimers associated with most pharmaceuticals. In principle, a phage which comes to display unexpected toxicities—or insufficient host range, e.g., pyophage development as discussed in reference 61—should be easily replaced given the genetic diversity of phages available for isolation. Considered instead in this section, however, is the relatively low potential for unexpected phage toxicities to begin with.

Two basic strategies can be employed to minimize a drug’s emergent properties. The first is to improve the predictive power such that drug physiological properties may be more readily projected based on chemical properties. In this way, drugs which are anticipated to ultimately display toxicities may be sidelined earlier during the development process. Alternatively, one can seek out drug candidates which, independent of knowledge of chemical properties, tend to be biased toward high specificity in terms of their interaction with body tissues and toward reduced intrinsic toxicity, as indeed can be the case generally for therapeutic protein products (119). In many cases, natural products also have been honed by evolution toward substantial specificity in their physiological interactions (120).

This idea of natural products displaying greater specificity for the sake of greater if narrower effectiveness is a variation on a broader concept of what can be described as antagonistic pleiotropy, that is, the tendency for enhancement of one aspect of an organism—here, the natural product—to give rise to changes, often in some negative manner, in other phenotypic aspects of the same organism (121, 122). Thus, the evolution of increased effectiveness against a specific target, e.g., what an agent binds to, can result in decreased effectiveness against other targets. In terms of emergent property pharmacology, the important general point is that it is a drug interacting with other body locations, or inappropriate interactions with primary targets, that often can lead to secondary pharmacodynamic consequences, i.e., side effects.

In terms of phage therapy emergent property pharmacology (59), the result of antagonistic pleiotropy is that phages, in many cases, have been honed by evolution toward limiting their physiological interactions with entities other than those bacteria with which the phage has some potential to subsequently produce phage progeny. In practice, this means phages tend to display relatively narrow host ranges, i.e., narrow spectra of activity as considered in pharmacological terms. As a consequence, phages tend not to be highly effective at interacting, especially in negative terms, with eukaryotic cells and tissues; see, though, the issue of lysogeny and lysogenic conversion such as the ability of certain temperate phages to encode bacterial exotoxins (106, 123).

Another way of making these points is that while phages, as viruses, can serve as highly cytotoxic agents, that cytotoxicity tends to be limited to those organisms into which phage virions can deliver their nucleic acid genomes (e.g., 124). In other words, phage virions represent vehicles of cytotoxic agent delivery, but both the delivery of those agents and, to a large extent, the cytotoxicity itself tend to be specific to an only minor subset of possible targets. Indeed, given this specificity, it is typical to not even consider phages to be cytotoxic agents.

The relative lack of emergent property pharmacology associated with phages as potential medicinals makes it possible to characterize phages in vitro toward use in vivo without substantial concern about unanticipated side effects. In this modern age of phage therapy, this characterization can include the sequencing and annotation of phage genomes, which addresses the other means of avoiding emergent properties, that of achieving substantial predictive in vitro characterization. Even without such sequencing, however, phages have not tended to display emergent toxicities during use (previous sections). Indeed, an argument can be made that regulatory approval of phage formulations for therapy could apply more in terms of specific approaches to characterization rather than in terms of specific phage isolates. The latter is just how annual influenza vaccines are developed, which a number of authors have suggested, discussed, and/or advocated as a model for the regulation of phage therapeutics (16, 85, 117, 125–134). Even yearly approval of new formulations, however, might unreasonably curtail phage therapy’s potential to quickly respond, through development of new therapeutic phages, to the need to target newly arising pathogen strains (135). As biocontrol agents of foods, the designation of phages following proper characterization as GRAS, that is, generally recognized as safe, can result in regulatory implementation of this idea of phage interchangeability based on criteria for characterization versus completely independent trials (32). GRAS designation functions, however, within the context of biocontrol rather than of phage therapy, sensu stricto.

Addressing Potential Caveats

Most or all concerns with phage use as antibacterial agents, particularly within a clinical context, are either hypothetical or readily addressable. The potential for phages to encode bacterial virulence factors, for example, can be addressed with increasing bioinformatic power given phage genome sequencing. This potential, however, (i) generally has not been reported as an issue during actual phage therapy, whether clinical or preclinical, (ii) is avoidable to a large extent by avoiding the use of temperate phages for phage therapy, i.e., particularly by employing professionally lytic phages, and (iii) otherwise could become apparent in the course of toxicity testing. The potential for phages to transduce genes can be avoided by not using phage types which are known to be prone to transduction (including temperate phages), by avoiding hosts for phage-stock generation which carry undesirable genes or prophages, by testing phage stocks for the presence of such genes, and even by testing the potential for phages to transduce from or to multiple additional strains. Toxins and other dangerous contaminants that potentially may be found in phage lysates are, as noted, not always an issue but nonetheless can be addressed through various purification approaches as well as by avoiding growth media ingredients that can potentially carry these factors.

Also relevant is the issue of presumptive treatment. Broad-spectrum antibiotics are convenient to prescribe because substantial prior knowledge of target organism susceptibility often is not required. Indeed, this convenience likely contributed in earlier decades to the rise of antibiotics to treat bacterial infections instead of phages as selectively toxic antibacterials (an abbreviated history of phage therapy is provided below). It is difficult to develop phage formulations that possess as wide a spectrum of activity as those of typical broad-spectrum antibiotics, however. Two not mutually exclusive technical solutions nonetheless exist which potentially can address phage therapy’s presumptive treatment issue. The first is to employ phage cocktails as therapeutic agents since, by mixing different phage types into a single product, its spectrum of activity can be substantially broadened (126, 128, 136). The result can be an increase in the likelihood that any given formulation will cover most of the strains of a given bacterial pathogen that may be encountered. In addition, it is possible to include multiple phages to multiple species to allow the treatment of infections without first identifying target species (137), though this approach is more complicated because it involves more phages as well as more targets. Notwithstanding such complication in development, Kutter et al. (61, p. 260) report that one cocktail produced in the former Soviet Republic of Georgia contains “hundreds of different phages that together target over 28 bacterial species.”

The second approach to addressing the issue of presumptive treatment is to improve diagnostics, particularly so that the phage susceptibility of target pathogens can be ascertained rapidly, ideally in minutes, as well as reasonably inexpensively (29, 138). A goal would be that precise diagnosis in terms of the phage vulnerability of presumptive pathogens could be determined relatively soon after examining patients. The need for rapidity in diagnostics as well as presumptive treatment more generally, however, is less important given the treatment of chronic versus more acute bacterial infections. This is because with chronic bacterial infections more time and resources—a hospital’s versus those of stand-alone medical practices—can be available to match phages to bacterial targets prior to use.

Phages and Immune Systems

Lastly in terms of potential detriments to phage utility is the question of phage interaction with animal immune systems (139–142). An often-leveled critique of phage therapy is that phage virions predominantly consist of proteins, and proteins are immunogenic. The immunogenicity of phage virions has long been appreciated, and it raises at least two concerns vis-à-vis phage therapy: phage-immune system interactions will negatively impact patients, and phage-immune system interactions will interfere with the effectiveness of phage therapy. Both of these concerns can be relevant to protein-based therapeutics generally, of which currently over 200 are being marketed (119), and to at least some extent both of these concerns are valid for phages. Nevertheless, these issues are not necessarily “game changers” so far as the practice of phage therapy is concerned.

First, substantial negative consequences of phage-immune system interactions during therapy are relatively rare, are typically not life threatening (i.e., are easily controlled or reversed), and especially as reported in the early phage literature, are difficult to distinguish from the interaction of phage carriage material with immune systems, that is, nonphage materials found in phage lysates (above). A product consisting essentially of phage-induced S. aureus lysate at one time was deliberately employed clinically to at least hundreds of patients (607 reported), including both intradermally and subcutaneously, for immune-system stimulation, but no anaphylactic reactions or serum sickness were reported despite weeks and months of application (104). Note, though, that side effects were reported in at least one patient, including abdominal pain following intranasal application; in addition, known responses to staphylococcus exposure were also observed, particularly temporary liver enlargement and tenderness, which the authors categorized as “temporary discomfort,” with treatment nevertheless generally continued in these cases. See too the discussion of Górski et al. (43). Second, immune system-mediated inactivation of phages does not necessarily translate into phage therapy failure (140, 142).

One circumstance in which problematic phage-immune system interactions would be expected to be greatest is with direct phage administration to the blood, i.e., intravenous (i.v.) treatment. The i.v. administration of phages is an issue because of fears of anaphylaxis and concerns about the development of humoral immune responses to phages, though it should be noted that i.v. exposure in and of itself appears to result in lower protein-based drug immunogenicity than injection into solid tissues such as intramuscular or subcutaneous injection or inhalation into the lungs (119). In addition is the potential nonspecific impact of the mononuclear phagocyte system, also known as the reticuloendothelial system. The latter has been shown to substantially reduce the duration of circulation of certain phage virions (143). Notwithstanding these issues, i.v. application of phage virions during phage therapy has substantial historical clinical precedence, which is remarkable given the crude methods of virion purification which were employed during the early decades of phage therapy. Indeed, Speck and Smithyman (26) recently made a case for the i.v. application of phages for therapy, which they suggest may be particularly effective against bacteremias along with endocarditis by S. aureus and as treatments for typhoid fever. See also the discussion by Kutateladze and Adamia (144) of i.v. phage application by the Eliava Institute.

PRECEDENT

The practice of phage therapy dates back to the very dawn of phage research (145, 146), and indeed, early research on the nature of phages might be reasonably viewed as a consequence of interest in phage therapy rather than the other way around. Phage therapy was explored clinically in many locations around the globe, though it persisted to differing degrees from region to region. Overall, a fairly substantial body of published research on clinical phage therapy has been built up. Not all has been published in the English-language primary literature, however, though a fair amount is English-language accessible within reviews. In this section I provide a primer on, in particular, the English-language literature on clinical phage therapy. For numerous reviews of this subject, see references 43, 61, 85, 92, 125, 147, and 148.

100 Years of Phage Therapy

Phages were “codiscovered” by Frederick Twort and Félix d’Hérelle in 1915 (149, 150) and 1917 (151, 152), respectively. Earlier studies have been cited as possibly involving phages (e.g., see references 153 and 154), particularly that of Hankin (155) from 1896, though the evidence for phage involvement in that study is not strong, given that heating bacteria-disinfecting water in unsealed tubes destroyed antibacterial activity while heating in “hermetically sealed tubes” did not (156). The first published hint of phage therapy can be found in d’Hérelle (151), including the title of that 1917 paper (emphasis mine): “On an Invisible Microbe Antagonistic to Dysentery Bacilli” (152). D’Hérelle’s emphasis on phages serving as bacterial antagonists, versus phages simply existing, may be supported by his later claims (157, p. 45) that the simple evidence of phage existence, in the form of plaques, “was ordinary enough, so banal indeed that many bacteriologists had certainly made it before on a variety of cultures.” As also reported by d’Hérelle in that same article, the notion that phages are a “filterable virus” was not even his original idea but instead was a suggestion of Charles Nicolle, at the time (probably around 1915) director of the Pasteur Institute in Tunis. Even d’Hérelle’s term, “bacteriophage” (“un bactériophage obligatoire”), is suggestive of his emphasis on the antibacterial nature of phages (phage, from Greek, meaning “to eat” or “devour”) or simply the “dissolution of bacteria” (158, p. 6), rather than on their potential viral nature, of which, as noted, he nevertheless appeared to be well aware (152, 158).

In 1918, d’Hérelle (159, p. 972) provided what may be the first suggestion of the therapeutic use of phages as antibacterial agents: “…il semble donc logique de proposer comme traitement de la dysenterie bacillaire l’administration, dès l’apparition des premiers symptômes, de cultures actives du microbe bactériophage.” [Translation by Google: “…it seems logical to propose as a treatment for shigellosis administration, from the onset of symptoms, active cultures of the microbe bacteriophage.”] By 1919, d’Hérelle (145, p. 934) had described the treatment of birds with phages (translation initially by Google): “…abrupt cessation of the enzootic the day of administration of the microbe bacteriophage confirms indisputably the role of the microbe as an immunity agent.” By at least 1921, we have a publication of phage therapy in humans (160), in this case anti-Staphylococcus treatment of skin via phage injection. Indeed, the first monograph considering phage therapy, Le Bactériophage: Son Rôle dans l’Immunité, was published by d’Hérelle in 1921. In the 1922 English translation of that book (161, p. 266) we find mention of “ingestion of cultures of the anti-Shiga bacteriophage, this treatment was applied for therapeutic purposes to patients affected with bacillary dysentery.” Thus, we are nearing the 100th anniversary of the clinical use of phages as antibacterial agents. Though it is difficult to say exactly how many individuals have been treated with phages over those years, Debarbieux et al. (162, p. 1) speculate that the number is in the “thousands if not millions.”

Grab Bag of Pre-“Modern” Human Use

Summers (163) encapsulates the history of phage therapy as a sequence of four periods consisting of enthusiasm, skepticism, abandonment, and then a recent revival. Break points are roughly the early 1930s (164), mid 1940s, and mid to late 1990s. Abandonment, corresponding to the rise of antibiotics, was not complete and, indeed, varied from region to region, with substantial retention of phage use as antibacterial agents seen in the former Soviet Union as well as Europe (125, 137, 147). A recent English-language summary of work published in the Soviet Union throughout much of the abandonment period is provided in book form by Chanishvili (148). Substantial phage use as a human therapeutic continues to take place to this day, especially in association with the George Eliava Institute of Bacteriophages, Microbiology, and Virology in Tbilisi, Georgia, which was founded, with the help of d’Hérelle, in 1923 (e.g., 137, 148), as well as the Phage Therapy Unit established in 2005 in association with the Ludwik Hirszfeld Institute of Immunology and Experimental Therapy in Wrocław, Poland (43, 102). In this section I summarize various English-language studies dating from the prerevival period of phage therapy. For a more detailed exploration of the early phage-therapy literature, particularly with regard to phage treatment of lung-associated infections, see reference 105.

In 1929, Larkum (165) supplied a list of justifications for the utility as well as disadvantages of phage use as antibacterial therapeutic agents, with the disadvantage being particularly that of excessive specificity. In addition, a “problem of controls” is explicitly indicated regarding the interpretation of outcomes. A summary of 13 studies and over 450 cases in total of phage treatment of S. aureus infections in humans is presented. Larkum states (p. 37), “Although these results cannot be adequately summarized, it is evident that from 75 to 80% of the patients had marked and immediate improvement. The details of many of these cases are such as to convince the reader that something other than coincidence was responsible for the results reported. Nor can the enthusiasm of the individual worker be entirely responsible, for it is a noteworthy fact that whereas in bacteriophage treatment of other infections, typhoid fever and dysentery for example, many workers have given unfavorable reports, only one which is not exceedingly favorable has as yet appeared concerning staphylococcus [sic] treatments.” Larkum presents an additional ∼240 cases of phage treatment of furunculosis involving subcutaneous phage injection. Larkum describes the immediate results (p. 39) by stating, “The lesions present regress, the exudate changes from purulent to serous, the pain is promptly relieved, and the lesions start healing within 24 hours from the time of inoculation. There are exceptions to be sure, but they are exceedingly rare.” Fully successful treatment was reported in 78% of cases. In 19% there was mostly mild recurrence. In 3% of cases there was no improvement.

As is the case in many of these older studies, the following statement tends to hold (165, p. 39): “…the inoculation of bacteriophage gives reactions in the majority of instances. These are as a rule extremely mild…” Note, though, that in 1% of patients the response to phages is described as severe. Both in terms of side effects and efficacy, it is important to keep in mind, however, that the formulations employed likely represented at least an approximation of crude lysates, which for Staphylococcus treatments may or may not function independently of the phages themselves in treating infections (104). Larkum notes, though, that some of the patients had previously received vaccines, presumably to the etiology, with no benefit, and he suggests that these patients might be viewed as controls.

In 1935, Dunlap (166) reported the successful treatment of S. aureus meningitis using phages, citing two previous publications by others in which three additional cases were also successfully treated using phages. The etiology was identified from a spinal tap 5 days following the beginning of symptoms, at which time antimeningococcus serum was supplied intraspinally (days 5 and 6). Treatment also consisted (p. 1595) of “copious spinal drainage by lumbar taps, approximately twice a day” along with intraspinal acriflavine administration on days 6 and 7. Phages were tested against the bacteria in vitro and were supplied, also intraspinally, on days 7, 8, 9, and 10, plus subcutaneously on day 16. Dunlap states (p. 1595) that “the first negative spinal fluid culture resulted after the administration of the second ampule of bacteriophage and remained negative throughout the duration of the illness.” Recovery was described as complete, though it is not obvious that phage action alone was responsible for that recovery.

In 1941, MacNeal et al. (167) described a (presumably) New York-based “service” of supplying staphylococcal phages to physicians. They present detailed instructions for how to dose intravenously since (p. 550) “strange as it may seem, our records show several instances in which intravenous therapy was abandoned because of difficulty in venipuncture.” They note further that “in severe sepsis, vein puncture may need to be done several times daily over long periods of time and the importance of careful preservation of the veins cannot be overemphasized.” In phage treatment of septicemia they report that a “shock reaction” is expected and apparently tolerated by the patient as well as desired by the physician, though the authors note (pp. 552–553) that it “…is rather alarming to those who have not previously observed it and is not entirely free from danger.” They recommend treating with sulfamethylthiazole or sulfathiazole simultaneously with phages. Three case studies are presented in this article, two of which were successful. Of the third, in which the patient apparently succumbed to infection, the authors note retrospectively that it may have been desirable to have employed higher doses of intravenous bacteriophage prior to identification of the target etiology.

In 1946, also from MacNeal et al. (168), came the combined use of phages and penicillin against various etiologies as published near the dawn of widespread antibiotic use. Notably, concern is expressed over resistance to penicillin. They note as well, in introducing the topic, that it had been shown previously, and then confirmed by these authors, that combined use of phages and penicillin in vitro was more effective against “moderately resistant” staphylococci than either alone and, further, that this effect could be seen as well with the normally penicillin-resistant Gram-negative “coli.” The article otherwise recounts a number of case studies—chronic osteomyelitis, facial carbuncle, bacterial endocarditis, intestinal perforation, bacteremia—in which phages and penicillin were administered both intramuscularly and intravenously, with various Staphylococcus and “coli” etiologies as well as Streptococcus, all with favorable outcomes. Unfortunately, it is not obvious that phage-only versus penicillin-only controls were sufficiently employed to conclusively rule in a synergistic effect between penicillin and phages in vivo. Nevertheless, as the authors suggest (pp. 980–981), “This combined action in the test tube may be designated as a potentiating, conjoined, or synergistic effect… For practical purposes it is significant that this combined action is also manifest in the clinical use of these agents in treating otherwise malignant infections, particularly those due to staphylococci, colon bacilli, and selected members of the streptococcus group.”

In 1945, Morton and Engely (169) presented a review of phage use to treat dysentery. They caution that it is necessary to confirm that phage preparations are active against target organisms in order to evaluate the effectiveness of treatments and also that phage presence, whether endogenous or therapeutically supplied, can interfere with the laboratory enumeration of target bacteria. The review first considers nine clinical studies which had indicated poor phage therapeutic performance along with, in many cases, suggestions of where those specific studies may have been flawed (e.g., possible phage gastric inactivation, lack of demonstration of phage effectiveness in vitro, low phage titer applied, too few cases reported, and a lack of controls). The review then considers 19 clinical studies which had been cited as evidence for the effectiveness of antidysentery phage treatment, again with suggestions of how these studies in their opinion likely were flawed. While concluding that these treatment reports together are inconclusive with regard to addressing the question of phage treatment efficacy against dysentery in humans, their conclusions regarding prophylaxis are positive. From their review of experimental treatment in animals they concluded that (p. 591) “dysentery phage has demonstrated an unmistakably therapeutic action.” It is important to note, however, that the animal disease models employed were not necessarily of dysentery per se.

In 1956, Mills (170) reported on the use of filter-sterilized S. aureus lysates containing between 10⁸ and 10¹⁰ phages/ml to treat sinusitis via nasal inhalation of aerosolized phages. Initially, 0.25 ml was applied, with a gradual increase to 1.0 ml per dose and with treatments taking place over 2 to 6 weeks, involving 10 to 20 treatments. Results are reported as “excellent” in 45% of cases and “good” in 33%. Fatigue, mild aches and pain, chills and fever, etc. are described as side effects observed early during treatment, and these are attributed, in part, to successful lysis of targeted bacteria. Treatments of 60 patients are reported, with two cases described in detail.

An important overall summary is that results from clinical phage therapy have been both reported for thousands of patients as well as critically reviewed in the older literature. This literature is well worth the consideration of modern-day phage therapists because it can supply invaluable clues as to why phage treatments may or may not have worked, i.e., as toward development of phage therapy best practices (171). It should be noted as well that in many cases this earlier literature involves the initiation of phage treatment closer to the point of diagnosis of infection than is the case for more modern treatment, especially of chronic bacterial infections, a bias that we can speculate would be due to a relative lack of viable alternative treatment approaches prior to the advent of widespread antibiotic use. This would be in contrast to the more modern treatment of chronic bacterial infections (next section), that is, infections which typically have been proven over long periods prior to the initiation of phage treatment to not be susceptible to treatment using more conventional therapies, particularly antibiotics (e.g., 102).

I have been collecting over the past decade and more a list of phage therapy, biocontrol, and related publications (>1,000; see publications.phage-therapy.org), and there are few that involve human treatment between 1945 and the 1980s, that is, during Summers’ period of abandonment, at least as published in English. Peitzman (164) published in 1969 an engaging history of phage therapy during the enthusiasm and skepticism periods that provides a number of clues to what went wrong to result in abandonment of the technique, especially in North America. Mostly, I believe, problems can be summed up as consisting, typically, of enthusiasm outpacing both understanding and rigor. Presumably, however, a death blow was dealt by the widespread introduction of antibiotics, which it should be noted—and to be fair—also have displayed a lag in terms of enthusiasm outpacing both understanding and rigor, i.e., as we are now coming to appreciate problems associated with routine antibiotic use in terms of both the evolution of antibiotic resistance and the disruption of microbiomes. The story ultimately was somewhat different regarding the fate of phage therapy behind the so-called Iron Curtain, as Chanishvili ably recounts (148). In Fig. 1 I provide my justification for labeling the mid to late 1990s to early 2000s as the point of transition to Summers’ revival period of phage therapy study.

FIGURE 1.

Prevalence of use of the phrase “phage therapy” in the literature. Google Scholar searches were performed without “include patents” or “include citations” checked. Shown are searches on “phage therapy” (in quotation marks, ●), “antibiotic resistant” OR “antibiotic resistance” (as written, ○), “bacteriophage” (▼), and “microbiology” (△). To limit presentation of spurious results, the y axis begins at four hits. Results and discussion: ●: “phage therapy” as a phrase is present to a small degree during the skepticism period (early 1930s through mid 1940s) and just prior to the recent revival (mid 1980s through mid to late 1990s) but then displays what appears to be renewed enthusiasm leading up to 2001. Starting from 2003 there then is a steady if less steep climb with a doubling approximately every 4 years. Relative lack of use especially prior to 1950 could reflect the popularity of alternative phrasing for “phage therapy,” including in non-English publications, though this possibility was not explored. ○: Reference to antibiotic resistance goes through an initial spurt beginning around 1950 and peaking in about 1960. This is followed by a nadir in 1963 and then a steady if less steep rise with a doubling approximately every 7 years. The presented curve first comes to exceed that of “bacteriophage” in 2005. ▼: Reference to the word “bacteriophage” likely is less representative early on due to non-English publications. Nevertheless, again there is early enthusiasm that is discernible starting around 1920 and peaking in the early 1930s. This is followed by a slow decline that levels out during 1943 through 1945, i.e., toward the end of World War II. A steady climb follows that parallels the rise in the use of the term “microbiology,” as indeed, so does the rise in reference to antibiotic resistance, but which to a degree plateaus for “bacteriophage” around 1990. Though not explored here, it can be speculated that the slower increase in use of “bacteriophage” starting in 1990 is a consequence of a greater prominence of the use of “phage” instead in publications, though alternatively, this switch may reflect a real decline in the rate of growth of the field; for additional data on the prevalence of bacteriophage publications, see reference 205. △: Reference to “microbiology” is presented as a general growth-in-the-literature control. The peak in 2007 and subsequent steady, ultimately multifold decline in its use is inexplicable, but similar drops are also seen with searches on “biology,” “chemistry,” “physics,” and “physiology” (not shown), so this likely reflects indexing lags by Google.

Modern Use

A substantial rekindling of interest in phage therapy began during the mid to late 1990s as reflected by a rapid surge in publications using that phrase, increasing, according to Google Scholar, from four such publications each in 1995, 1996, and 1997 to a peak of about 140 publications in 2001 (Fig. 1). This span seems to represent a second period of enthusiasm. A decline in usage followed, with about 78 publications in 2003, a drop which likely captured a second if brief period of skepticism, or at least a popping of the “bubble” of this second period of enthusiasm. I would like to suggest that Summers’ (163) “recent revival,” as proposed in 2011, be dubbed simply “revival” and be considered to span from the mid to late 1990s through approximately 2002. Since 2003, as noted, there has been a steady if slower increase in the use of the phrase “phage therapy.” I would like to suggest, somewhat optimistically, that this increase represents a “premainstream” period of phage therapy development. That is, there seems to be an ongoing increase in enthusiasm for phage therapy but as yet only minimal expansion of its clinical use worldwide. For recent calls for such expansion, see for example, references 8 and 162.

Notwithstanding the above discussion, in hindsight and with some consensus, a modern period of phage therapy can be described in terms of English-language publication that begins during the 1980s with the animal work of H. Williams Smith and collaborators in England (172–175) and the clinical work reported by S. Ślopek and collaborators in Poland (103, 139, 176–182). As above, I point especially to Chanishvili (148) for discussion of the Russian and Georgian literature, though see also reference 61 for a broader discussion. In this section I provide overviews of recent phage therapy use in the clinic as has been published over approximately the past 10 years. These specifically are other than strictly safety trials, i.e., as were discussed above. These are published primarily in English, mostly in the primary literature, and are presented in descending-date order. Earlier as well as non-English-published clinical work has been reviewed elsewhere (e.g., 43, 85, 148). See Table 1 for a more complete listing of English-language human phage therapy articles (mostly primary literature) published since 1980.

Fish et al. (183) used an S. aureus phage to treat nine S. aureus-associated diabetic foot ulcers, specifically as found on toes, with six cases presented, one of them methicillin-resistant S. aureus associated. The ulcers were poorly vascularized, were not responding to antibiotic therapy including as topically applied, and toe amputation was indicated prior to phage treatment but not undertaken. Tissue was removed (debridement) in three cases, and in one case phage treatment was used to prevent infection. Topical phage application was once per week using volumes of 0.1 to 0.5 ml (of 10⁷ to 10⁸ phages/ml), varying with the size of the area treated, applied to gauze packed over the wound; patients were instructed to leave the dressing in place for 48 h. All ulcers healed in an average of 7 weeks. Because these were compassionate-use clinical treatments, no controls were employed other than a failure of (“poor response to”) conventional treatments employed prior to initiation of phage treatment.

Sarker et al. (114) followed up previous anti-E. coli trials by the same group (91, 112, 113, 115, 184) involving antidiarrhea treatment of children, using oral delivery. Treatment success, however, was not achieved. It is probable that insufficient phage numbers were employed since E. coli densities found in the gastrointestinal tract likely were insufficient to substantially boost phage numbers in situ to densities required to result in adequate bacterial eradication. In addition, the host range of the phages employed may not have included the etiologies involved.

Fadlallah et al. (185) report on the use of a commercially available staphylococcal phage against chronic vancomycin-intermediate S. aureus. This involved treatment of interstitial keratitis and corneal abscess of a 65-year-old woman’s left eye at the Phage Therapy Center in Tbilisi, Georgia. The patient had experienced various infections, vancomycin treatments, and staphylococcal carriage over an 11-year, initially postoperative period. The phage was delivered topically using eye drops as well as via a nasal spray, and also systemically intravenously. The duration of treatment was 4 weeks. As confirmed during testing 3 and 6 months later, the result was an absence of ocular and nasal carriage of the target organism.

Kutateladze (186) describes the activity of the Eliava Phage Therapy Center in Tbilisi, Georgia, from 2012 through 2014. She describes over 3,000 patient visits for phage therapy, though there is no indication that these are unique patients versus repeat visits by some of the same patients. Of these patients, 39 were described as “foreign.” She also reports that phage preparations were supplied to 130 patients from abroad. Efficacy is indicated with the statement (p. 81), “more than 95% exhibiting significant improvement and recovery.” In addition, a complete lack of “complications or side effects after phage application” was reported (p. 81). In the case of treatment of chronic bacterial infections, there is an indication that phage therapy was useful primarily for reducing ongoing antibiotic requirements as well as increasing the length of periods of remission rather than necessarily resulting in outright cure.

Międzybrodzki et al. (102) provide a summary of 153 patients treated from 2008 to 2010 at the Phage Therapy Unit in Wrocław, Poland. Their emphasis is on what they describe as the treatment of “otherwise untreatable chronic bacterial infections,” that is, where for various reasons treatment success using antibiotics, including due to resistance, simply was not possible, and with ongoing antibiotic treatment not necessarily discontinued upon the initiation of phage treatment. Treatments did not involve surgery or hospitalization (61). The infections treated are described as orthopedic, respiratory, soft tissue, and urogenital and of 43-month median duration prior to initiation of phage treatment (thus all are chronic). Per their description (102, p. 86), “Phage preparations against Staphylococcus, Enterococcus, E. coli, Pseudomonas, Klebsiella, Enterobacter, Proteus, Citrobacter, Salmonella, and/or Stenotrophomonas… were administered to patients topically, orally, intrarectally, intravaginally, or as inhalations of aerosol.” These phage preparations consistently were composed of only a single phage type, though in the case of mixed infections more than one phage type would be applied in alternation. Phage sensitivity of target organisms was determined as necessary, i.e., before treatment, in response to unsatisfactory results, and after completion of treatments, with treatments lasting a median of 55 days.

Pathogen eradication was observed in 18.3% of cases, what they describe as a “good clinical result” in another 8.5% of patients, and “clinical improvement” for a further 13.1% of patients, for a total of 39.9% “good response” to phage therapy. The largest success (65.7% good results) was seen against enterococcal infections as well as oral or intrarectal application (72.2% “good responses” and 44% eradication or recovery, respectively). There was no significant difference between rates of success with versus without use of nonphage antibacterials in conjunction with phage treatment. A pertinent observation is that statistically significant differences in outcomes were seen depending upon the site of phage administration employed, suggesting that treatments, though not blinded, nonetheless were to a degree controlled between patients. It is noted, however, that the treatment success rate is somewhat less than has been reported previously in Poland (181, 187), though one has to wonder whether the lack of the use of surgery in the treatments reported by Międzybrodzki et al. may have contributed to this difference. The authors consider the “lack of significant side effects” (p. 115) to be consistent with historical results.

Khawaldeh et al. (188) describe treatment, in Australia, of a P. aeruginosa urinary tract infection of a 67-year-old woman following uretic stent placement. Antibiotic treatment took place over a 2-year period. Subsequent treatment involved the use of a six-phage cocktail active against the target bacterium, obtained from the Eliava Institute (phage preparation involved filter sterilization, presumably to remove contaminating cells and larger cellular debris). Bladder instillations took place twice a day for 20 days with 20 ml of approximately 2 × 10⁷ phages/ml. The patient was also treated with the antibiotics colistin and meropenem, starting on day 6, with the latter as had been employed previously. Bacterial viable counts were reduced from 2 × 10⁶/ml on day 0 to roughly 10-fold lower on days 1 through 5, 50-fold lower on day 6 (prior to antibiotic treatment), 500-fold lower by day 7, and reduced to 0 on day 8. Phages alone thus appeared to contribute to at least an approximately 50-fold decline in bacterial numbers and may have contributed to the further observed reductions.

Kvachadze et al. (189) reported on the treatment of a 7-year-old cystic fibrosis patient with both anti-Staphylococcus and anti-Pseudomonas phages. Phages were delivered via nebulizer every 4 to 6 weeks. Improvement in the patient’s general condition was reported as well as an ability to cut antibiotic dosing in half.

Kutateladze and Adamia (144) summarize the use of phages as supplements to standard cystic fibrosis treatments performed at the National Center of Cystic Fibrosis in Tbilisi, Georgia. Patients included both infants and adults, with phages supplied by nebulizer over spans of approximately 1 week. The authors suggest that phage use had the effect of causing “a substantial decrease in the concentration of bacterial cells” found in sputum samples (p. 4). Improvements in patient health as well as extensions of time until subsequent bacterial colonization were also attributed to phage use. No use of mock-treatment controls or details of individual cases were reported.

Letkiewicz et al. (190) describe phage treatment, at the Phage Therapy Unit in Wrocław, Poland, of three patients suffering from E. faecalis prostate infections which had previously resisted antibiotics and other treatment strategies. Between 10⁷ and 10⁹ phages/ml were present in formulations, which were matched to target bacteria (102). Phage preparations were applied rectally twice daily for 28 to 33 days, resulting in eradication of the target pathogen in all three cases.

Wright et al. (118) reported the results of a successful phase I/II clinical trial treating P. aeruginosa chronic otitis infections of duration prior to phage treatment ranging from up to 2 to more than 50 years. Half of the 24 volunteers were treated with phages and half with placebo. Additional treatment details—involving in all cases only a single phage dose—are described under the heading of “Safety Trials,” above. Phage replication in the course of treatment was observed. No placebo-treated volunteers experienced improvement, while the blinded clinical investigator informally noted that “after 20 patients it was clear… that there was a marked effect in approximately half the patients” (p. 356). Three (of 12) phage-treated volunteers experienced substantial symptomatic improvement (“almost complete recovery”) as well as elimination of the pathogen to below detectable numbers.

Leszczyński et al. (191) treated one patient orally to eliminate gastrointestinal colonization by methicillin-resistant S. aureus that appeared to have given rise to urinary tract infection. A sterile lysate (10 ml) of three different phages with titer of 7 × 10⁸ phages/ml was applied three times per day. This phage application followed gastric juice neutralization (187). Treatment spanned 4 weeks but appeared to have eliminated the target organism, as determined via rectal swabs, after 1 week of phage treatment.

Marza et al. (192), in a single case study, demonstrated phage multiplication in association with a P. aeruginosa burn wound infection. Phage treatment coincided with elimination of the target pathogen but also a febrile episode. The latter, though, was also observed in the same patient in the absence of phage application. Intravenous ceftazidime dosing was employed as well, obscuring the phage role in the process.

Jikia et al. (193) recount the treatment, in Tbilisi, Georgia, of multiply antibiotic-resistant S. aureus-infected radiation burns using a phage- and antibiotic-impregnated biopolymer, PhagoBioDerm (ciprofloxacin was the antibiotic, to which the target bacteria were already resistant). Antibiotics were applied beginning 1 month prior to the start of phage treatment, but a lack of sufficient impact of those treatments resulted in the decision to employ phages. In 2 days after the start of phage treatment, wounds decreased in size along with the extent of purulent drainage, and by day 7 wounds were S. aureus-negative.

Markoishvili et al. (194) describe the treatment of infected wounds and/or ulcerations associated with 96 patients with PhagoBioDerm after a failure of “standard clinical therapy” (as described for 22 cases). In addition to phages, PhagoBioDerm contains ciprofloxacin and benzocaine, the latter to reduce pain. Complete recovery was indicated for 70% of individuals.

FURTHER DEVELOPMENT

Preclinical development of phages for therapeutic purposes is relevant only to the extent that useful data are obtained. Whether data are useful depends on such obvious issues as the question of whether animal or in vitro infection models are appropriate and whether associated phage dosing protocols are realistic, such as in terms of timing. This section briefly addresses these issues. Also considered, though not in depth, is that successful development of phage therapeutics is highly dependent on the phage therapy regulatory environment, the economics of antibacterial drug development in general, and the economics of phage therapeutic development more specifically.

Preclinical Efforts

Phage characterization prior to introduction into the clinic can involve determination of whether therapeutic phages are able to display lysogenic cycles (they should not), whether they carry potentially dangerous genes (ditto), possibly whether they are capable of effecting generalized transduction (ditto as well), and more recently whether complete genome sequencing and annotation has been achieved. In addition, monitoring for toxicity is crucial as a standard aspect of pharmaceutical development. Phage host range determination is also important, as too can be consideration of incompatibilities between different phage types or excessive overlap in antibacterial abilities during cocktail development. What these various determinations all have in common is that they are relatively independent of assay conditions. Specifically, they do not necessarily involve assessment of phage therapy efficacy as an antibacterial agent given accurate approximation of real-world conditions.

Unfortunately, in many cases preclinical efforts appear to have been expended more toward identifying experimental protocols which may give rise to what can be interpreted as efficacious results rather than toward the development of more realistic though potentially also more challenging experimental conditions. It is likely that these tendencies result from a combination of the relative ease with which laboratories can take up phage research, the duration of graduate student careers, and the duration of grants. In short, and for good reason, much preclinical phage therapy research in the modern era has tended to prioritize demonstration of phage therapy success rather than critical product development. In many instances there also has been a tendency toward only limited in vitro phage characterization before moving on to animal testing, tendencies which one can speculate might stem from greater experience among various researchers with handling animals versus characterizing phages.

These tendencies have several important consequences. The first is that numerous proof-of-principle results likely are of limited utility for either specific or more general phage therapy development. The second is that it really pays, as in any field of science, to understand the system being considered, here especially phage phenotypic characteristics as well as the basics of phage therapy pharmacology. The third issue, and potentially the most difficult to address, is that there exists a crucial need both to understand the real-world conditions under which phage therapy is to be undertaken and to develop realistic model systems for evaluating efficacy. For example, if chronic bacterial infections are a target, e.g., such as Staphylococcus infections that have resisted treatment over long time frames, then some realistic approximation of chronic bacterial infections ought to serve as the basis for experimentation both in vitro and in vivo (195). Quoting Larkum, from 1929 (165, p. 35), “It is possible to produce staphylococcus infection in many animals, and it is possible to terminate such processes through use of bacteriophage, but these infections are not to be compared with naturally acquired diseases, nor can the results of experimentation be translated into terms of human disease and treatment.” Furthering this sentiment, nearly 90 years later, Czaplewski et al. (39, p. 240) noted, “Studies should define and test clear go or no-go decision points for product progression. Programmes of work that are mainly in vitro or those focused entirely on surrogate endpoints (eg, characterising cytokines rather than pathology, microbiology, or clinical response) might not be competitive for funding.” Or, as noted by Debarbieux et al. (162, p. 2), “…instead of expecting basic science to run almost unlimited investigations since many molecular aspects of phage therapy are still not understood. If we had waited for immunology to be fully understood at the molecular level before the use of vaccines, most of us would not be reading this paper!”

Regulation and Economics

The other important aspect of phage therapy development is issues of regulation and economics, as has been considered by numerous authors. Regulatory issues in particular have been addressed, for example, in references 32, 115, 134, and 196–199. The economics of development of phage therapeutics have been discussed in references 39 and 115, with intellectual property rights aspects considered in reference 197. See Ventola (200) for a discussion of the economics and regulation of antibiotic development more generally, including their observation vis-à-vis the latter (p. 280) that “studies comparing antibiotics with placebo are considered to be unethical,” which appears to have been an ongoing complication of phage therapy development, as too, perhaps, are expectations based on past experience that antibacterial drugs should be relatively inexpensive. As a consequence of these and other impediments to antibacterial development, 155 nonphage antibacterial agents have been approved for clinical use since 1938, of which only 62% are still in use, and only three have been approved recently, i.e., during “the current decade” (201).

Harnessing the full potential of phages and antibacterial agents more generally will require regulatory mechanisms that explicitly do not place the full economic burden of full-blown clinical trials on every possible modification of a product. The justification for this need is that, unlike many drug targets, the drug targets of phages are not static but instead are capable of evolving both individually (that is, over the course of individual infections) and across communities (with new strains entering into as well as evolving within communities). Therefore, there is reason to argue that the regulation of all antimicrobial agents should be such that it is at least recognized that there can be an ongoing need for new agents not just for the sake of generating new profits as patents expire but instead because new strains of target organisms literally represent new diseases, or at least new variants on established diseases, as viewed from a pharmacological perspective. Or, as stated by Xu et al. (18, p. 11), the problem which needs to be addressed is that “the rate of development of new antibiotics is slower than the rate of the appearance of antibiotic resistance.” What phages offer that is relatively unique, given this latter perspective, is large numbers of potential candidate drugs that are relatively lacking in pharmaceutically emergent properties. Or, as De Vos and Pirnay (3) noted, the “development of new phage preparations [is] quick and cheap” (see also reference 4). Thus, given repeated emergence of pharmacologically new bacterial disease types in combination with an abundance of new potential drugs to combat those new diseases, logic at least would be consistent with seeking effective ways of addressing the former with the latter.

In light of this context, would it be reasonable for phage therapy to demand favored treatment from regulators? That is, might phages be viewed as “special” from a regulatory standpoint? To a degree the answer to that question may effectively be “Yes” because regulators do appear to “get it,” that is, to understand that there can be worth in attempting to work with those phage qualities that are relatively unique rather than against those same qualities. Nevertheless, what would be the justification for carving out a relatively unique track for regulatory approval of phages? I offer four answers to that question: phages are abundant, they are simple in structure to the point that many are uncomfortable with the suggestion that they are even organisms, they can be relatively safe, and they possess a long history of relatively rigorous clinical use. Indeed, as I have stated elsewhere (50, p. 39), “If an effectively inexhaustible supply of antibiotic types were available to which cross-resistance did not excessively occur and which displayed very high therapeutic indices (ratio of a drug’s toxic to therapeutic dose), then there would not be an antibiotic crisis.”

CONCLUSION

Pratt (202, p. 3) has stated, “The ideal antibiotic would have no deleterious effect on the patient but would be lethal to the organism. There is no ideal antibiotic.” Penicillin G in nonallergic individuals, however, is cited as coming closest to this ideal. Perhaps phages come close as well. Indeed, to a degree, the apparent safety of phage interaction with the human body underlies the basis of phage designation as Generally Recognized As Safe (GRAS) additives to food. Within this context, the simplicity of phage development as drugs provides an advantage over many other antibacterial drug candidates, with a possible exception of certain vaccines (i.e., influenza) or monoclonal antibodies. An argument also can be made that phages exist, at least absent genetic engineering (10, 203), as naturally occurring entities and as such could, in principle, be marketed as natural products (196) or instead as probiotics (147, 197) rather than as medicinals.

A distinction between phage use as antibacterial agents versus newer, especially synthetic small-molecule antibacterials is that phages have been subject to many decades of clinical use that at least arguably has been effective, appears to have not raised significant safety concerns, and has involved large numbers of people, i.e., thousands if not millions. Without that history, phage therapy would exist simply as another “new” approach to treating bacterial infections. With that history, however, we rightly should view phage therapy as not just another new approach to treating bacterial infections but instead as a relatively mature antibacterial technology, one that is ripe for further clinical implementation and development. As has been suggested elsewhere (8, p. 685), such clinical implementation by interested physicians can be achieved in the United States via “compassionate use to lay the groundwork for physician and public acceptance as well as full-blown clinical trials.” Preclinical research will still be important to furthering phage therapy development, however. Especially there is a need to put greater effort toward improving in vitro and in vivo models of chronic bacterial infections, e.g., as can require weeks of phage treatment to achieve reasonable levels of efficacy (102, 171, 204). More generally, rather than focusing on relatively facile proof-of-principle studies, instead specific impediments to antibacterial success may need to be systematically determined and then overcome in the course of rigorous experimental phage therapy exploration.