Phage Therapy in the Postantibiotic Era

Antibiotic resistance is a naturally occurring phenomenon that predates the development and use of antibiotics by humans. Across diverse environments, both microorganisms and higher eukaryotes produce a plethora of biologically active molecules that have antibacterial properties, some of which have been repurposed as modern-day antibiotics (8). However, the concentrations of these chemicals within these natural environments are often below clinically relevant thresholds, suggesting that resistance does not arise exclusively to escape their toxic activity. It has been proposed that in nature, antibiotics and their interplay with antibiotic resistance mechanisms serve as a communication channel between the members of a microbial community rather than as strict antimicrobial agents (9). These compounds have been demonstrated to prompt adaptive phenotypic and genotypic responses and shape the composition of the community (9). When considering antibiotics from this biological perspective, it is unsurprising that genes conferring resistance to modern antibiotics have been identified in ancient microbial populations, such as unpolluted arctic permafrost, suggesting that antibiotic resistance exists even in the absence of anthropogenic influence (10–12).

Human activity, particularly clinical and industrial overuse of antibiotics, greatly aggravates the problem of antimicrobial resistance. Antibiotics are used to enhance the growth of livestock, treat crop and fish diseases in agriculture and aquaculture, respectively, and, of course, treat infectious diseases in humans (13). In fact, agricultural use of antibiotics has been estimated to be as high as 180 mg of active antibiotic agent per kg of meat produced in the United States and is reported to be even higher in other countries (14). An unintended side effect of their agricultural use is the release of millions of tons of antibiotics into water effluents and environmental reservoirs (15). This is further exacerbated by a lack of appropriate water treatment, the contribution of pharmaceutical waste, and the reduction in distance between farmlands and cities, all of which have resulted in increasing levels of antibiotic release and persistence in the environment. This continued exposure of environmental microbial communities to diverse antibiotics has led to the accelerated evolution and expanded repertoire of antibiotic resistance genes persisting in natural reservoirs (15).

In addition to spontaneous mutations in chromosomal genes selected by the pressure of the drug, resistance can arise from the process of horizontal gene transfer—the movement of genetic information between organisms—via transformation, conjugation, or transduction (16). Antibiotic-resistant bacteria selected for in the guts of humans and animals treated with antibiotics are a leading source of resistance determinants for horizontal transfer (17). Furthermore, bacterial cells can achieve transient, non-genetically encoded resistance through processes such as growth in biofilms, swarming adaptation, metabolic dormancy, and persistence (18). The myriad of resistance mechanisms can impede antibiotic action at every step of their passage through the bacterial cell: bacteria can alter the structure of their cell envelope to deny entrance of the drug or synthesize efflux pumps to expel it; they can modify (and even destroy) the compounds through the production of enzymes such as beta-lactamases or stop the production of enzymes required for antibiotic activation; they can modify, hide, or quantitatively adjust the intended target of the drug; or, finally, they can activate alternative metabolic pathways to circumvent the toxic action of the antibiotic (16, 19). For an illustrative graphical summary of antibiotic resistance mechanisms and the genotypic basis of antibiotic resistance, refer to the work of Yelin and Kishony (19).

The “golden age” of antibiotics began in the 1940s and continued for over four decades, with more than 40 antibiotics being discovered and introduced for clinical use (Fig. 1). During this period, emergence of resistance against a specific antibiotic was met with minimal concern, as newer compounds, more often than not exhibiting better pharmacokinetic and pharmacodynamic characteristics, were quickly developed, fueling a cycle of antibiotic discovery, use/overuse, and concomitant appearance of resistance (13, 20). From the decade of the 1990s, however, the consequences of this already-counterproductive cycle became more apparent, as the number of novel antibiotics introduced steadily decreased. The phenomenon has been described as a “dry pipeline” in antibiotic research and development, with the majority of recently introduced antibiotics being either modified or combined versions of previously known compounds.

Novel drugs must go through the challenging process of proving not only their efficacy but also their safety, favorable pharmacokinetic profile, and cost-effectiveness. It is estimated that only five out of 5,000 to 10,000 candidate molecules reach phase I studies and that only one out of those five receives regulatory approval for human use (21). The drug development process is as expensive as it is long, and pharmaceutical companies hesitate to invest in antibiotics when the odds are against them. Furthermore, antibiotics are consumed in short regimes, should be overseen by strict stewardship programs, and are vulnerable to the emergence of resistance, all of which can severely diminish the revenues a company receives (22). It is even believed that newly marketed antibiotics will be outpaced by the high rate at which antibiotic-resistant bacteria are emerging, highlighting the need for novel therapeutics (23). The dry pipeline aggravates the problem of antibiotic resistance, as it has stalled our arsenal of therapeutic options.

The result of consecutive acquisition of antibiotic resistance traits is the generation of multidrug-resistant (MDR), extensively drug-resistant (XDR), and even pan-drug-resistant (PDR) pathogens (24). A group of bacterial species have caught the attention of researchers, clinicians, and public health officials, as they cause the most frequent and severe health care-associated MDR infections. This group includes the pathogens Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp. and is identified with the acronym ESKAPE, referencing the capacity of its members to escape the biocidal activity of antibiotics through multiple resistance mechanisms (25). Consequently, the last four of these pathogens, specifically the carbapenem- and cephalosporin-resistant strains, were recently listed by the World Health Organization (WHO) as being a critical priority for the research and development of new antibiotics (26). Additional species, listed under high priority, included the remaining ESKAPE members, the digestive tract pathogens Helicobacter pylori, Campylobacter spp., and Salmonella spp., and the sexually transmissible agent Neisseria gonorrhoeae (26). Patients with underlying medical conditions, those with any degree of immunocompromise, and those hospitalized (especially in intensive care units, surgical wards or burns units) are at a higher risk of contracting MDR infections. However, an ever-growing number of reports warn about “commonplace” community-acquired infections, in otherwise-healthy patients, becoming unresponsive to antibiotic treatment (27, 28). In the postantibiotic era, even common infections and minor injuries can kill (29).

The burden of antimicrobial resistance can be measured by morbidity and mortality rates and financial costs, but even a combination of these indicators fails to encompass its magnitude. From an oversimplified description, patients with MDR infections have poorer prognoses, higher mortality rates, lengthier hospital stays, higher risks of complications or permanent sequelae, and increased treatment failure compared with their counterparts with antibiotic-sensitive infections (7, 13, 29, 30). A recent review proposes that by the year 2050 ten million lives will be lost annually due to antimicrobial resistance, with a cost to the world economies equivalent to US$100 trillion (31). It is a major threat to global health that is capable of affecting all individuals, regardless of their age, socioeconomic status, or country of residence (32).

Any strategy to combat the problem of antimicrobial resistance has to be multidimensional, multidisciplinary, and global. A key component to be included is the implementation of regulatory measures for the human and animal use of antibiotics (33). In many countries, particularly in the developing world, self-medication and easy purchase of antibiotics without a prescription are worryingly common and need to be addressed. Even more troubling is the everyday use of antibiotics in animal feeds and agriculture, which should be restricted. Furthermore, inappropriate prescription of antibiotics by health care professionals, such as prescribing antibiotics to treat viral or fungal infections, unnecessarily prolonging antibiotic courses, or using antimicrobials with a broader spectrum than reasonably needed for a specific infection, also occurs. The issue of the dry pipeline could be mitigated by providing administrative and financial stimuli to the field (22) or through the “revival” of old antibiotics. The latter is the case for polymyxins and chloramphenicol, which were abandoned due to toxicity concerns (nephrotoxicity/neurotoxicity and rare but potentially fatal hematological side effects, respectively) and are now making a comeback (34, 35). Most importantly, it is essential to encourage research on new therapeutic alternatives and to rekindle the interest in neglected ones, such as phage therapy.

Bacteriophage therapy, which was first used almost a century ago, is now going through a revival driven mostly by the antibiotic resistance crisis. This renewed interest in phage therapy has been facilitated by our improved understanding of phage biology, genetics, immunology, and pharmacology. Key aspects of phage therapy have now been standardized to improve treatment success. Minimum suggested regulatory requirements for the therapeutic use of phages call for strictly lytic phages, confirmed antimicrobial activity against the target pathogen, and removal of contaminating bacterial debris and endotoxins (36). In addition, the identity of the bacterial host receptor for any therapeutic phage should be established, which will provide important information on emergence of phage resistance, evolutionary trade-offs, and use of combination therapies that are less likely to generate phage-resistant hosts.

Lytic phage infection begins with adsorption to specific receptors on the bacterial host’s surface. These receptors may be located on either Gram-positive or -negative cell walls, as well as polysaccharide capsules or even appendages such as pili and flagella (37). The lock-and-key relationship between a phage and the bacterial receptors will typically determine the host range that the virus is able to infect, and the list of characterized phage receptors is constantly growing. After adsorption, the virus will eject its genetic material into the host. The majority of described lytic phages associated with human pathogens belong to the orders Caudovirales and Microviridae and have double- or single-stranded DNA genomes (38). Next, the virus takes over the bacterial replication machinery, creating the next generation of phage progeny in the process. Replication will continue until phage-encoded proteins (see “Phage-Derived Proteins” below) are activated in order to lyse the cell, effectively killing the host and allowing the newly synthesized viruses to escape and reinitiate the cycle. The lysis time, or latent period, is the amount of time taken by a phage to complete this intracellular life cycle.

As mentioned above, the use of temperate or lysogenic phages for phage therapy is normally inadvisable, not only because their killing capacity is hampered by the quick arise of homoimmunity but also because of the possible harmful consequences of lysogenic conversion. Through lysogenic conversion, bacteria can acquire new, often pathogenic, genetic traits, such as phage-encoded toxins capable of greatly enhancing their virulence (39) or potentially even antibiotic resistance determinants (40). However, for relevant pathogens such as Clostridium difficile no strictly lytic phages have been isolated, and the use of temperate phages may be necessary (41). Similarly, in cases of emergency, time constraints could justify the therapeutic use of temperate phages when lytic phages are unavailable. Unfortunately, this does not mean that the use of lytic phages is exempt from concerns. Lytic phage genomes can contain greater than 50% hypothetical genes with no known function (42) or encode auxiliary proteins that alter bacterial physiology in ways that are not fully understood. During abortive infection, where phage DNA is ejected into the cell and some genes expressed, without production of viral progeny, the bacterial host could potentially act as a reservoir for foreign DNA of unknown function. This reasoning should warrant continued research into phage genetics as a way to ensure the safety of phage therapy.

Although technically not living organisms, phages are certainly dynamic entities, and the lytic cycle is the cornerstone of phage-based therapeutics. In contrast, antibiotics are chemicals capable only of selective disruption of certain bacterial physiological processes, such as protein or cell wall synthesis. A quick comparison between phages and antibiotics demonstrates how strikingly different their mechanisms of action are. A summary of these differences, and some similarities, between the two can be found in Table 1 (43, 44). However, the following paragraphs elaborate on some of the most therapeutically relevant comparative points.

Phages have been classically described as highly specific for their hosts. However, it was recently demonstrated that phages are able to “jump” hosts, and in the gut that process is facilitated by the microbiota (45). Thus, phage-host specificity may evolve and adapt over time. This specificity is simultaneously the greatest advantage and greatest disadvantage of phage therapy. Phage therapy aims directly at the pathogenic bacteria, whereas antibiotic treatment carries collateral damage as it disrupts the microbiome. Due to its lack of off-target effects, phage therapy is exempt from side effects related to microbiome disturbances, such as mucosal candidiasis, antibiotic-associated diarrhea, pseudomembranous colitis caused by Clostridium difficile, and even long-term metabolic and immunological disorders (46). Conversely, that specificity also demands accurate diagnosis of the infection and identification of the etiological agent, sometimes to the strain level, a process that can be difficult and time- and resource-consuming (47). Moreover, early initiation of phage therapy has been shown to be critical to its success, and delays as short as 6 h can result in a significant decline in treatment effectiveness (48). Together, these facts justify the practice of establishing and expanding phage collections or automated pipelines to quickly isolate and identify candidate phages. The collections, or libraries, are made up of readily available, well-characterized phages, isolated from natural sources, that exhibit biological traits theoretically desirable for phage therapy, such as short latency time, large burst size, and broad host range (49).

Both antibiotics and phages interact with the immune system. First, the most severe, and potentially lethal, side effect of antibiotic therapy is hypersensitivity, a group of immune-mediated phenomena that include IgE-mediated reactions, anaphylaxis, Stevens-Johnson syndrome, and toxic epidermal necrolysis (50). A 24-year retrospective study in a tertiary care hospital found that antibiotics, particularly beta-lactams, were the most common triggers of anaphylaxis (51). Comparatively, there have been no documented adverse anaphylaxis cases associated with phage therapy in humans (52, 53), although the collective body of evidence is decidedly smaller. Second, the action of the immune system complements the antibacterial activity of both antibiotics and phages. Bactericidal agents are able to kill bacteria, whereas bacteriostatic agents only prevent their growth. These effects usually depend on the mechanism of action and dose of a given antibiotic, but at therapeutic doses, protein synthesis inhibitors such as macrolides and tetracyclines are mostly bacteriostatic (54). Bacteriostatic antibiotics heavily rely on the immune system of a patient to clear the infection (55). Although lytic phages are by definition “natural killers,” phage therapy on its own will not completely clear an infection either, at least theoretically. This is because eradication of the host would also result in termination of viral replication. Instead, phages engage in “kill-the-winner” dynamics with their bacterial hosts, rapidly reducing host abundance before reaching a dynamic equilibrium (56, 57). The immune system is still needed to eliminate the lingering bacterial population for phage therapy to be successful. This collaboration between phages and the immune system has been termed “immunophage synergy” (58). It could be argued that every treatment of an infectious disease in an immunocompetent patient (and to a lesser extent in an immunocompromised patient) is, by definition, combinational therapy due to the natural action of the immune system. Finally, it has been demonstrated that phages elicit both innate and acquired immune responses against them (59). More interestingly, the immune activation and inflammatory environment created by the bacterial infection could heighten the inhibition of phage therapy, although the clinical relevance of these phenomena remains to be determined (59).

As with antibiotics, phage therapy is affected by bacterial resistance. Lytic phages impart strong antimicrobial selective pressures on their hosts that rapidly select for phage-resistant bacterial mutants (2). Bacterial antiphage systems can be encountered along every step of the phage replication cycle. The best-known processes include modification of the receptors used during phage adsorption, superinfection exclusion (Sie) systems to prevent viral DNA entry, restriction-modification systems that protect host DNA while leaving foreign DNA vulnerable to the action of restriction enzymes, and clustered regularly interspaced short palindromic repeat (CRISPR)-Cas systems that recognize and degrade previously encountered foreign DNA (60). The sum of phage resistance mechanisms constitutes a true prokaryotic “immune system.” However, the characterization of these mechanisms has come from experiments using a small number of model phage-host pairs and may not necessarily be applicable to less-investigated pathogens and their phages. Regarding this issue, Doron et al. (61) recently searched more than 45,000 bacterial and archaeal genomes and discovered nine new families of antiphage defense systems, whose molecular mechanisms are yet to be fully understood. Closing the knowledge gap about phage-resistance mechanisms may allow for therapeutic and biotechnological exploitation (see “Exploitation of Phage Resistance” below).

A final benefit of phage therapy is its versatility. Due to their genetic diversity, abundance, and ubiquity, there is a virtually limitless source of phages. Furthermore, phage therapy can be delivered through different approaches, each one adaptable to the available resources, type of infections, and characteristics of the patients. Phages can be administered as a tailor-made, personalized therapy or, conversely, subjected to automated high-throughput production of phage cocktails. During the last decades of work with phages for therapeutic purposes, the in vitro and preclinical findings have begun to be translated into carefully designed clinical trials and case studies, highlighting important lessons to be considered moving forward.

In the early years of human phage therapy (Fig. 1), phages were used to treat conditions including typhoid fever, dysentery, skin and surgical wound infections, peritonitis, septicemia, urinary tract infections, and external otitis (62). During the 1930s, however, a series of analytical reviews posed important questions about the validity of the presented results (Fig. 1). Criticism was aimed at the lack of proper methodological design, controls, and standardized production and characterization of the phage preparations, as well as contradictory results (63–65). After the introduction of antibiotics, when the efforts in phage therapy studies were primarily relegated to a few countries in Eastern Europe, language also became a barrier for the widespread dissemination of the results. Sulakvelidze et al. (66) reviewed studies from the Georgian, Russian, and Polish literature, finding successful use of phages for the treatment of the previously mentioned conditions, as well as pneumonia, meningitis, osteomyelitis, and postsurgical infections in cancer patients. Although some early and encouraging work at the end of the 20th century presented the use of phage therapy in animal models (67–69), it was not until the start of the new millennium that the English literature rediscovered human phage therapy trials (62). Here, we discuss some of the results of these contemporary studies.

The results of the first phase I randomized, placebo-controlled phage therapy trial conducted in the United States were published by Rhoads et al. (70) in 2009. The study investigated the safety of a 12-week topical treatment, with a total follow-up period of 24 weeks, of a phage preparation targeting S. aureus, P. aeruginosa, and Escherichia coli in chronic venous leg ulcers. Their results showed no significant difference in the incidence of adverse events between groups. With a sample size of only 42 patients and no previous in vitro demonstration of susceptibility of the patients’ infectious agents to the phage preparation, the study was not designed to assess effectiveness of the intervention. Expectedly, the rate and frequency of healing were the same between treatment groups. Previously, in Georgia, Markoishvili et al. (71) had demonstrated successful healing of poorly vascularized ulcers with the application of a biodegradable polymer impregnated with antibiotic and lytic phages. However, the independent effect of the phages could not be discriminated from that of the antibiotic, and the authors recommended further studies.

The effectiveness of topical administration of phage therapy has been assessed by at least two phase I/II trials. In 2009, Wright et al. (72) established a randomized, double-blind, placebo-controlled study where they tested the action of a phage cocktail in the treatment of chronic otitis due to antibiotic-resistant P. aeruginosa in 24 patients. The measured outcomes included physician-assessed signs of inflammation, patient-reported symptoms, and quantification of the bacterial and viral loads. The phage preparation improved all of the measured outcomes compared to placebo after the 42-day follow-up period. In the second case, the randomized, multicenter, single-blind, and open study Phagoburn ran from 2015 to 2017 and evaluated the treatment of burn wound infections by P. aeruginosa in 25 patients, using a cocktail of 12 phages. The published report highlights this as being the first clinical trial of phage therapy ever performed according to both good manufacturing practices (GMP) and good clinical practices (GCP) (73). Additionally, no unwanted side effects attributable to the phage cocktail were reported. The authors also showed that the intervention significantly decreased the pathogen load in the wounds, but unfortunately, this happened at a lower rate than in the control arm, which consisted of standard care with 1% sulfadiazine silver emulsion cream. The authors explained these results with three claims. First, after delays associated with manufacturing and administrative challenges, the length of the recruitment period was nearly halved, leading to a small patient sample size. Second, the titer of the phage cocktail was found to have significantly decreased after manufacturing, leading to patients receiving a lower dose of phages than originally intended. Third, bacteria from patients in whom the phage treatment failed were shown to be resistant to low phage doses (73). According to the authors, further studies that address these issues are warranted. Finally, an ongoing phase II trial (ClinicalTrials registration no. NCT02664740) is looking at the topical treatment of diabetic foot ulcers infected by S. aureus with a phage cocktail.

Studies focusing on the oral administration of phages have also been carried out. A T4-like phage preparation, targeted against E. coli and designed for the treatment of diarrheal disease, has been assessed in phase I placebo-controlled trials in healthy adults from Switzerland (74) and Bangladesh (52), in 2005 and 2012, respectively, and in healthy and diseased children from Bangladesh in 2017 (75). No adverse effects from oral administration of phages were found by self-report, physical examination, and laboratory testing of hepatic, renal, or hematological function. The studies also provided insights into the bioavailability and activity of oral phage preparations. Phages did not amplify in the gut of healthy individuals, and only small, dose-dependent fractions of the initial phage dose were recovered from their feces. There was no evidence of phages or phage-specific antibodies in the bloodstream, and the phage preparation did not disturb the composition of the gut microbiota. However, the same lack of viral replication was seen in the children with diarrheal disease, and the treatment did not have significant favorable effects compared to standard rehydration therapy. While the studies had been designed primarily around safety, the latter observations prompted the authors to question the value of phage therapy to treat these low-abundance infections within the gut (75). In a different approach, Gindin et al. (76) recruited 32 adults with mild to moderate gastrointestinal complaints, to carry out a phase I randomized, double blind, placebo-controlled, crossover trial in 2018. The study was based on the premise that the gut microbiota can regulate human health, that dysbiosis can lead to disease states, and that modulation of its components, namely, eradication of specific detrimental organisms, can have beneficial effects in patients suffering from gastrointestinal distress. The intervention, a 28-day oral treatment with capsules containing 4 strains of bacteriophages targeting recognized gastrointestinal pathogens, was proven to be safe and tolerable. After controlling for sequence effects given the crossover design of the study, the treatment was more effective than placebo at reducing symptoms of colon pain and abnormal gastric function, comparable to placebo in reducing small-intestine pain, but ineffective in decreasing perceived gastrointestinal inflammation. Further studies could open the door for phages to be used as novel prebiotics.

Regarding intravenous phage therapy in humans, Speck and Smithyman (53) recently reviewed the evidence supporting its use. Their primary focus was on severe infections such as typhoid fever and S. aureus bacteremia, with data obtained from reports spanning the last 80 years. They reference studies that collectively account for almost 1,000 patients successfully treated with intravenous phage therapy, with a negligible number of side effects. In addition, all of the severe side effects (reactions resembling shock or serum sickness) could be attributed to contaminants from early phage preparations. Even though historical evidence tends to be dismissed because of its age or lack of compliance with modern clinical standards, many of the described studies came from groups in France, Canada, and the United States, countries with well-regulated medical systems. The authors concluded that there is a strong possibility that intravenous administration of phages will be safe and efficacious (53). Moreover, an ongoing phase II/III randomized, double-blind clinical trial (ClinicalTrials registration no. NCT03140085) is addressing an alternative route of parenteral phage administration. The trial will look at the efficacy of intravesical administration of the commercial phage cocktail Pyophage (PYO) versus oral antibiotics or intravesical placebo for the treatment of urinary tract infections in patients undergoing transurethral prostatectomy.

Parallel to full-scale clinical trials, the pathway toward widespread use of phage therapy and its translation from the lab bench to the patient’s bedside could be shortened by cases of compassionate use. In the absence of alternative treatments, or in terminally ill patients, phage therapy can obtain “off-license” approval for use. The advantages of the approach include immediate clinical usage, obtaining data that could be used to inform future work, and that it can be used for all forms of phage therapy (77). Nevertheless, the results from this approach are usually hard to replicate as they are limited to a single patient and do not directly lead to the approval of the therapeutic intervention.

Further use of intravenous and intracavitary phage therapy in humans has been assessed primarily through individual case studies of compassionate use. In 2017, Schooley et al. (78) reported the first case of successful intravenous and intracavitary phage therapy targeting a systemic MDR infection in the United States. The patient was a 68-year-old diabetic man with necrotizing pancreatitis complicated by an MDR A. baumannii infection. After the infection completely stopped responding to antibiotic therapy and the patient’s condition severely deteriorated, phage therapy was initiated. Several aspects of this case report are worth highlighting. First, treatment was personalized and quickly available, as the bacterial strain causing the infection was tested against preestablished A. baumannii-specific phage libraries. Second, treatment consisted of sequential administration of phage cocktails, which is necessary for countering the emergence of phage resistance (see “Conventional Phage Therapy” below). Third, phage therapy was well tolerated by the patient. In 2018, Chan et al. (79) reported a comparable case. A 76-year-old man underwent aortic arch replacement surgery with a Dacron graft and was later diagnosed with an MDR P. aeruginosa graft infection. As the patient was considered at too high risk for surgical replacement of the graft and treatment with intravenous ceftazidime and superficial chest wall debridement proved to be unsatisfactory, phage therapy was used. The intervention consisted of the simultaneous, single intracavitary application of phage OMKO1 and ceftazidime. The treatment was well tolerated, and even though the patient exhibited further complications expected in aortic graft carriers, the P. aeruginosa infection receded without any recurrence despite the discontinuation of antibiotics. Additional mechanistic aspects that contributed to the success of these two cases is discussed in “Exploitation of Phage Resistance” below. Lastly, Jennes et al. (80) reported the case of a 61-year-old patient who developed septicemia caused by a P. aeruginosa strain that was sensitive exclusively to colistin, which likely originated from local colonization of pressure sores. When the antibiotic treatment, along with the patient’s underlying conditions, led to acute kidney injury, therapy with a phage cocktail was initiated. The two-phage cocktail was administered intravenously every 6 h, and in topical irrigation of the wounds every 8 h, for a total of 10 days. Phage therapy immediately turned blood cultures negative and reduced the levels of C-reactive protein and the patient’s temperature, with kidney function being restored after a few days. Despite this favorable clinical course, the patient’s multiple comorbidities remained, leading to his death four months after phage therapy due to sepsis by a different pathogen.

In the coming years, we certainly expect to see an increase in phase I phage therapy clinical trials and their progress toward phases II and III. Notably, most of the presented studies have primarily assessed the so-called “conventional” approach to phage therapy. In the next section, we explore the notions behind this approach, but more importantly, we include innovative approaches that will progressively reach clinical use, including phage-antibiotic combinations, phage-derived enzymes, exploitation of phage resistance, and phage bioengineering.

The use of phages as the only therapeutic agents administered to a patient during the course of a bacterial infection is referred to as conventional phage therapy (Fig. 2A). Additionally, the terms monophage and polyphage therapy clarify whether a single phage or a combination of phages is used, respectively (81).

Monophage therapy has been used primarily in laboratory settings with animal models, as proof of concept during the design and testing of phage preparations. As expected, focus is placed on troublesome infections by MDR pathogens, such as carbapenem-resistant A. baumannii pneumonia (82, 83) and vancomycin-resistant E. faecium septicemia (84). Nevertheless, the usefulness of monophage therapy in clinical practice may be hampered by the emergence of bacterial phage resistance, as it can appear relatively quickly, even in small bacterial populations (85, 86). Within a natural environment, phages and their bacterial host will continually evolve and adapt in order to preserve infective capacity or withstand it, respectively, entering an evolutionary arms race (87). Even though this evolutionary capacity of phages to overcome bacterial resistance is a further advantage over antibiotic therapy, the process may not be efficient enough to warrant the clinical use of monophage therapy. A further disadvantage of monophage therapy is that it requires precise matching between the etiological agent and the phage. This match is typically performed in a laboratory setting, yet the in vitro and in vivo lytic properties of a phage may not always coincide (88).

To address the issues of monophage therapy, polyphage therapy, also known as the use of phage cocktails, is a commonly used strategy (Fig. 2A). Phage cocktails can be designed to target a single bacterial strain, multiple strains of a single bacterial species, or multiple species, typically grouped by the clinical syndromes they cause. Unfortunately, these cocktails require longer and more complex preparation and purification processes, increasing the likelihood of eliciting immune responses, and have reduced predictability of phage pharmacokinetic and pharmacodynamic properties (81). To limit these shortcomings, the comparative efficacy of the individual phages within a cocktail should be assessed (89) and unnecessary, poorly active phages removed. A final approach combines the notions of monophage and polyphage therapies by using sequential, instead of simultaneous, administration of phage cocktails. With sequential administration, even if resistance arises, bacteria are continually exposed to new phages to which they were not initially resistant. This strategy would maintain the bacterial population density at a lower level for longer, facilitating the action of the immune system. An in vivo study using a wax moth larva model suggests that sequential administration of phage cocktails may provide better results over longer time scales regarding reduction of bacterial populations and emergence of phage resistance (90).

Phage cocktails are available as over-the-counter medication in Russia and Georgia for the treatment of bacterial infections of broad etiology. The Eliava Institute is associated with the production and biannual update of two of the most widely used generic phage cocktails: Pyophage (PYO) and Intestiphage (91). PYO contains phages targeting S. aureus, E. coli, P. aeruginosa, Proteus vulgaris, Proteus mirabilis, and Streptococcus pyogenes, whereas Intestiphage targets approximately 23 different enteric bacteria; they are marketed for the treatment of pyoinflammatory and enteric diseases, respectively (91). Western regulatory agencies demand, among other requirements, a thorough characterization of these products before approving them for use in humans. Recently, metagenomic analyses of PYO and an additional phage cocktail have been carried out (92, 93). The first of these studies, by Villarroel et al. (92), compared the compositions of three batches of PYO from a period between 1997 and 2014, finding sequences from up to 30 different phages and a surprisingly stable composition throughout the years, despite the aforementioned biannual “upgrade” of the product. However, most commercially available phage products in Eastern Europe, at least historically, have not been produced under the full manufacture controls required by these regulatory agencies.

In addition to the previously discussed clinical trials (see “Experience in Clinical Trials” above), results from preclinical studies of polyphage therapy have greatly increased in recent years. The targeted conditions or microorganisms include methicillin-resistant S. aureus (MRSA) osteomyelitis (94), P. aeruginosa colonization of the sputa of patients with cystic fibrosis (95), catheter-associated urinary tract infections by Proteus mirabilis (96), and C. difficile infection (97). The consensus demonstrates favorable safety profiles and encouraging evidence of efficacy.

In clinical practice many conditions are treated, sometimes exclusively, using combination therapy. Highly active antiretroviral therapy (HAART) against HIV and multiple-drug therapy for tuberculosis or malaria are examples where multiple treatment approaches are used to combat a disease. Yet the concept transcends the field of infectious diseases, with similar tactics being used in chemotherapy protocols against specific types of cancer or in the control of refractory hypertension. Amid the looming threat of bacterial resistance, a commonly used strategy is the combination of two or more antibiotics (98) with the hope of achieving a synergistic effect. Pharmacological synergy occurs when the combined effect of two therapeutic agents is greater than the sum of their individual effects, a phenomenon that brings about significantly higher rates of treatment success (99). In some cases, even the addition of compounds that lack an inherent antimicrobial activity can be favorable; this is the case for adjuvants that can block certain resistance mechanisms or pharmacokinetically improve the action of the drug (100). All these premises can be summarized by the evolutionary principle that dictates that “two sufficiently different selective pressures are likely to be more effective than either alone” (99). This logic supports the practice of combining phages and antibiotics as a therapeutic strategy against MDR infections.

The term phage-antibiotic synergy (PAS) was first used by Comeau et al. (101) to describe the serendipitous discovery that sublethal concentrations of antibiotics could greatly increase the bacterial production of lytic phages. The phenomenon was attributed to the increased biomass and biosynthetic potential of bacteria in the presence of antibiotic concentrations high enough to inhibit cell division but not to cause cell death. For the virus, this resulted in a shorter latent period and larger burst size, allowing quicker spread and reduction of the bacterial population. This initial description, made with beta-lactams and quinolones on E. coli, has since been replicated with species including P. aeruginosa, Burkholderia cepacia, and methicillin-resistant S. aureus (MRSA) and with additional antibiotic groups, such as tetracyclines and oxazolidinones (102–104). We believe that this traditional concept of PAS can be greatly expanded, as additional mechanisms through which synergy between phages and antibiotics occurs have since been discovered (see “Phage-Derived Proteins” and “Exploitation of Phage Resistance” below). Finally, it has been shown that an “order” effect can arise when combining phages and antibiotics, whereby treatment with phages before antibiotics may achieve maximum killing. The effect has been described using protein synthesis inhibitors with bacteriostatic effects, reflecting the need of actively replicating hosts for phages to propagate and kill. The results show how optimizing the timing of administration of combinational therapy can potentiate its efficacy (105, 106).

The clinical exploitation of PAS would have two additional benefits. First, by limiting the amount of antibiotics used, it would assist with antibiotic stewardship and managing emergence of resistance. Second, it would provide antibiotics a second wind against MDR pathogens through combination treatments with phages (99). Previous studies have demonstrated the effectiveness of phage-antibiotic combinational therapy, both in vitro and in vivo, for ESKAPE bacteria. Chhibber et al. (107) demonstrated that the combination of a lytic phage with linezolid was more effective than either treatment alone in managing MRSA infections of diabetic foot ulcers in a murine model. The results are particularly remarkable as they were observed in a model of diabetes, a disease characterized by systemic immunosuppression, where reduced therapeutic effects would have been expected. Oechslin et al. (108) showed synergism between a single-dose phage cocktail and ciprofloxacin treatment against experimental endocarditis caused by P. aeruginosa in rats. One of the hallmarks of endocarditis is the formation of vegetations that protect bacteria from antibiotics and the host’s immune response. The combination treatment killed >6 log CFU per gram of vegetations, as opposed to 2.5 log CFU/g of vegetations for each of the single-agent treatments (108). Additionally, the study showed that emergence of phage-resistant mutants occurred exclusively in vitro, suggesting that those mutations had detrimental effects on the bacterial population in vivo. Finally, Valério et al. (109) demonstrated increased synergy between a monophage therapy and bactericidal agents against E. coli grown in human urine samples, with a significantly lower emergence of both singly and doubly resistant bacterial mutants when combination therapy was used. Together, the findings underpin the idea that phage therapy should not aim to replace antibiotics. The combination therapy with antibiotics and phages can work exceedingly well against complicated infections.

Phage genomes encode a number of proteins and enzymes required to breach the bacterial cell during infection (Fig. 2B). During phage adsorption to the host and ejection of its genome, two groups of proteins are typically required: virion-associated peptidoglycan hydrolases (VAPGH) and polysaccharide depolymerases (58). VAPGH are structural components of the virus that are typically located on the phage base plate and act to locally degrade the peptidoglycan layer, thereby allowing the phage tail tube structure to eject its genomic material into the host. The other major proteins are phage-encoded depolymerases, which target the polysaccharide components of the bacterial cell envelope, such as the bacterial capsule, the Gram-negative lipopolysaccharide (LPS), or the extracellular matrix of biofilms. Phage depolymerase activity facilitates access to secondary host receptors located at the cell wall by degrading the structural polysaccharide components and may have broader effects for the degradation of biofilms (110).

Bacterial capsule and biofilm structures are important virulence determinants that can block the action of antibiotics, disinfectants, and the immune response. As such, the use of phage-derived depolymerases to target and remove these structures is a viable treatment option. Bacterial capsules are essential virulence determinants for many important pathogenic bacterial species, including Streptococcus pneumoniae, Haemophilus influenzae, and E. coli. Capsules help the pathogen escape the action of phagocytes and the complement system and facilitate epithelial colonization, cell invasion, and intravascular survival (111). Similarly, biofilm structures protect bacterial communities from the immune response and, in addition, can increase their metabolic efficiency and genetic exchange, among multiple other benefits. Bedi et al. (112) demonstrated synergy between monophage therapy and amoxicillin for eradication of K. pneumoniae biofilm, which they attributed to the action of phage-encoded depolymerases breaking down the extracellular matrix of the biofilm, allowing better penetration of the antibiotic and whole phage particles. Relatedly, Mushtaq et al. (113) observed how stripping off the capsule of E. coli K1, a common causative agent of neonatal meningitis, using a phage-derived depolymerase increased the phagocytic action of macrophages and protected against bacteremia in rat models. Born et al. (114) demonstrated that a tail-associated phage depolymerase could degrade the capsule of the plant pathogen Erwinia amylovora and, by doing so, increase its susceptibility to phages it was previously resistant against. These studies suggest that phage-derived depolymerases are the protagonists of numerous mechanisms for synergy between phages, antibiotics, and the immune system (Fig. 2C).

Towards the end of the lytic cycle, phages use a different set of enzymes to facilitate lysis of the host and release of the viral progeny (Fig. 2B). Even though single-gene lysis systems have been characterized for phages with small, single-strand genomes (115), the systems used by tailed phages rely on the coordinated action of at least two types of proteins: holins and endolysins. Holins are hydrophobic proteins that passively accumulate in the inner bacterial cell membrane until reaching a specific concentration that triggers their arrangement into holes. The timing of holin triggering is allele specific, in that it can be advanced or retarded by missense mutations that can result in fine-tuning of phage lysis times (116). Holins thus are responsible for a generalized permeabilization of the inner cell membrane but are not independently capable of cell lysis (58). Next, endolysins use those holes to translocate from the cytoplasm to the periplasmic space, gaining access to their substrate: the bacterial cell wall polymer peptidoglycan. For Gram-positive bacteria, degradation of the bacterial cell wall is sufficient to cause cell lysis and release of the viral progeny. For Gram-negative bacteria, however, the outer membrane must be disrupted to allow for efficient cell lysis and release of viruses. This is achieved by the action of spanins, which catalyze the fusion of the inner and outer membranes, leading to cell lysis (117). A variation of this pathway is represented by the pinholin/signal anchor release (SAR) endolysin system. Here, inactive SAR endolysins are secreted by the host sec system (118). Along with pinholins, they accumulate on the inner cell membrane until the pinholins trigger its depolarization, which in turn prompts the activation of the SAR endolysins to begin the enzymatic destruction of the cell wall (117).

Endolysins are classified according to the peptidoglycan structure they target, with three major classes being glycosidases, amidases, and endopeptidases (58). Furthermore, their host range has been shown to be variable, with serovar-specific (119), multispecies (120), and even multigenus (121) endolysins having been isolated. The natural cell wall-lysing properties of endolysins have given them great potential as antimicrobial agents. Nevertheless, the majority of reports of successful endolysin use for therapeutic purposes have been for use of the enzymes against Gram-positive pathogens only. This is unsurprising, as the outer membranes of Gram-negative organisms act as a barrier to these enzymes. Few natural endolysins have the ability to bypass the outer membrane and thus possess relevant anti-Gram-negative activity (122). For this reason, additional tactics to bypass the outer membrane obstacle have been devised, including the coadministration of chemicals that permeabilize the outer membrane, such as polymyxins, aminoglycosides, or chelating agents such as EDTA (123), and bioengineered endolysins. One example of such bioengineering is seen in artilysins, where the addition of polycationic oligopeptides to the endolysin grants outer membrane-breaching ability and the capacity to kill Gram-negative MDR pathogens such as A. baumannii and P. aeruginosa with a 4- to 5-log reduction within 30 min in vitro (124).

One prospective target for endolysin therapy is the eradication of S. aureus, including MRSA, from the nose and skin (125, 126). Nasal and skin carriage of S. aureus in patients and health care workers is a risk factor for infections by this pathogen, and the topical antibiotic mupirocin has been recommended to eradicate it (127). A study by Pastagia et al. (128) demonstrated how an engineered lysin achieved an additional 1-log reduction in CFU of S. aureus compared to mupirocin treatment in a murine model. Similar results have been reported for the eradication of Streptococcus pneumoniae from the upper respiratory tract (129) and of Streptococcus agalactiae, which is associated with preterm labor and neonatal infection, from the urogenital tract of mice (130). Experiments have also assessed the protective effect and safety of intraperitoneal endolysin administration against S. aureus septicemia in mice (126, 131). Remarkably, one of the studies demonstrated synergy between an endolysin and the antibiotic oxacillin, as doses that were not protective individually efficiently protected from septic death when combined. One current antistaphylococcal intravenous endolysin therapy (SAL200) is currently in phase I clinical trials to assess its pharmacokinetic profile, establish its ideal dosing schedule, and evaluate its safety, with encouraging preliminary results (132, 133). Another example of current endolysin therapies is the topical use of the recombinant endolysin Staphefekt to treat skin flora dysbiosis caused by overgrowth of S. aureus. Staphefekt was able to decrease the load of S. aureus, restore the balance of the local flora, and clinically improve the condition in three patients (134) and has now progressed through to phase II clinical trials (135).

The use of phages combined with antibiotics and the natural immune response can be expanded beyond the classic concept of PAS. Here, we focus on the “trade-offs” of bacterial evolution toward phage resistance, following the evolutionary rationale proposed by Torres-Barceló and Hochberg (99). In this rationale, it is suggested that bacterial evolution has genetic constraints whereby acquisition of resistance against phages may come with a cost. The diverse range of evolutionary trade-offs has now been demonstrated in numerous studies. Using a fitness competition experiment with Pseudomonas fluorescens in soil, Gómez and Buckling demonstrated that acquisition of phage resistance reduced bacterial fitness, with a decrease in growth of 36% (136). Through coincubation of lytic phages with their corresponding bacterial hosts, Capparelli et al. were able to obtain phage-resistant S. aureus mutants with reduced growth rates and impaired production of capsular polysaccharide (137) and Salmonella enterica serovar Paratyphi B mutants that had completely lost virulence and had a shorter life span (138). Evans et al. (139) observed how resistance against a flagellatropic phage in the plant pathogen Erwinia carotovora subsp. atroseptica impaired bacterial motility, consequently reducing its virulence and resulting in easier-to-control infections. Finally, and perhaps more interestingly, Chan et al. (140), demonstrated that applying the selective pressure of phage OMKO1 to a population of MDR P. aeruginosa resulted in the selection of mutants that had a significant increase in antibiotic sensitivity. A review by León and Bastías (141) discusses how, in many cases, the trade-offs of phage resistance can be traced back to mutations in the phage receptors. For example, if a phage uses bacterial pili as its receptor, then a mutation in those pili could result in loss of adsorption, leading to phage resistance but also loss of function of the pilus structure, the consequences of which may include impaired bacterial adherence to epithelial surfaces or reduced virulence. For the case of antibiotic resensitization presented by Chan et al., it was shown that the receptor-binding site for phage OMKO1 was part of two multidrug efflux systems and that phage resistance caused loss of function of the efflux pumps, restoring antibiotic sensitivity. In summary, the natural evolution of phage and bacteria presents us with a number of opportunities to take advantage of bacterial phage resistance.

The clinical translation of the aforementioned in vitro studies has taken its first steps. The cases presented by Schooley et al. (78) and Chan et al. (79), described in “Experience in Clinical Trials” above, contain interesting insights related to this therapeutic approach. In the former, when the A. baumannii strain causing the infection acquired phage resistance, two fitness costs were observed: loss of the bacterial capsule and resensitization to minocycline. The trade-offs were effectively exploited in order to eradicate the infection, as minocycline was added to the antibiotic regime and the patient’s immune system could have theoretically improved its phagocytic capacity against the mutant pathogen. In the latter, the phage that was used, OMKO1, had been previously shown to direct the evolution of P. aeruginosa toward antibiotic resensitization in vitro. The phage was administered to the patient along with ceftazidime, an antibiotic that the pathogen was resistant against. Even though the phage-antibiotic combination was shown to reduce bacterial densities in the graft material in vitro, and ultimately a favorable outcome in the patient was achieved, the authors did not recapitulate the resensitization event in vivo.

This evolutionary trade-off approach could usher in the next generation of phage therapy: the use of selected phages that force an evolutionary cost on their bacterial host. In order to consistently and reliably take advantage of the evolutionary trade-offs associated with phage resistance in the clinical setting, we must use well-characterized phages with defined host receptors and molecular mechanisms of infection and resistance. Current research suggests that phage resistance is a nonphylogenetically conserved, highly species-specific, evolutionary trait (142). Indeed, a number of studies have highlighted that phage resistance can repeatably and predictably occur, as seen in Enterococcus faecalis mutations within the integral outer membrane protein granting phage resistance (143). However, further research is required to assess the broad applicability of these evolutionary mechanisms across critical MDR bacterial pathogens within a therapeutic setting.

We have entered the age of “synthetic biology,” where the principles of biology and engineering are conjugated to design, create, or modify existing biological entities to perform tasks they would not naturally do (144). Bioengineering of phages could dramatically increase their therapeutic potential via a range of mechanisms, including expanded host range, switching host tropism, delivery of exogenous genes, or modification of phage capsids (Fig. 2D). Mahichi et al. (145) were able to expand the host range of E. coli phage T2 by incorporating the long tail fiber genes of phage IP008 through homologous recombination. The resulting chimeric phage exhibited the broader host range of phage IP008, while still maintaining the strong lytic activity of phage T2 (145). Furthermore, Lu and Collins (146) modified E. coli phage T7 to express the enzyme dispersin B, which is capable of degrading one of the key components of bacterial biofilms, within the bacterial host cell upon infection. The engineered virus was capable of reducing biofilm cell counts by >100-fold compared to the wild-type phage. Finally, Kim et al. (147) found that by conjugating polyethylene glycol (PEG) to phages, their blood circulation time was increased, possibly through evasion of the T-cell-mediated immune response. Similar modifications could be used to improve other pharmacokinetic parameters of phage therapy (148, 149).

Bioengineered phages have also been used to increase antibiotic specificity and to combat MDR bacterial pathogens. Yacoby et al. (150) devised a way to attach chloramphenicol molecules to lytic phages and used the viral specificity to deliver the antibiotic directly to bacterial cells, improving the in vitro potency of the drug by a factor of ∼20,000 and potentially eliminating the side effects caused by its interaction with human cells and the microbiota. Similarly, phages have been used to deliver photosensitizing agents to target bacteria, making them susceptible to photodynamic inactivation without disturbing the rest of the microbial community, an approach that has been demonstrated against MRSA (151) and the fungal pathogen Candida albicans (152). Finally, genetic engineering can utilize phage-mediated gene delivery for therapeutic purposes (Fig. 2D). The genetic material that can be delivered or inserted into the bacterial cell includes dominant sensitive genes to reverse antibiotic resistance (153), CRISPR/Cas9 sequences to inactivate virulence genes (154), modified lethal transcription regulators (155), small regulatory RNAs to silence antibiotic resistance determinants (156), and even genes that code for proteins capable of increasing the susceptibility to specific antibiotics (157).

Bioengineered phages have also found use in the field of vaccinology. Vaccines have arguably had a larger positive effect on global health than antibiotics, and vaccine development has a leading role in the fight against antibiotic resistance (158). Phage display is the most significant strategy for phage-based vaccine design and antigen expression. Here, the vaccine sequence, or antigen, is cloned into one of the phage capsid proteins, typically either the major capsid protein or an accessory protein. As a result, the antigen is expressed on the phage surface and readily presented to the immune system (159). Conversely, in phage DNA vaccines a eukaryotic gene expression cassette contains the antigen sequence and is cloned into the phage genome. After administration, the phages are recognized and taken up by antigen-presenting cells (APCs) such as macrophages and dendritic cells, which will in turn express the antigen and initiate an adaptive immune response (159). An unrelated mechanism is the use of bacteria that have lost their virulence, as a result of acquisition of phage resistance, as attenuated vaccines, since they retain their immunogenicity (137, 138). With innovative approaches being constantly tested, we foresee a remarkably vast potential in therapeutic phage bioengineering.

We can use phage therapy as a proficient vehicle for education and scientific outreach.

Although efficacy is the keystone for adequate translation, education plays an important role in the acceptance, uptake, and dissemination of new therapeutic strategies by a population. Conversely, poor health literacy can lead to frank misuse of health care resources (as exemplified by antibiotics) or to dissemination of nonfactual information and active dissent (as seen with the antivaccine movement). Phage therapy is at a critical point where we are seeing increased therapeutic applications, media attention, and public awareness. There needs to be a concerted effort to promote education on phage therapy and avoid the pitfalls seen with other health care resources. With the rebirth of phage therapy, we propose four guiding educational principles to maximize its benefits. First, we must encourage, maintain, and capitalize on the extensive media attention phage therapy is already receiving. Second, we should direct outreach to all stakeholders: scientists, clinicians, pharmaceutical and biotechnology companies, public health officials, policy makers and lawmakers, politicians, and of course, the broader community. Education must be pervasive through all social layers. Third, we must send messages that are clear, approachable, and honest. Improving the public perception of phage therapy is certainly desirable, but transparency will ultimately gain the trust of the public. Fourth, we should incorporate all the fields of human knowledge in the initiative. Artistic and cultural approaches—museum exhibitions, narrative literature, filmmaking, music—have great potential for receiving inspiration from and disseminating information on phage therapy. The Pulitzer Award-winning novel Arrowsmith, for example, tells the story of a medical doctor using phages to control an outbreak of bubonic plague. However, we must be careful not to overhype the findings and translation possibilities. Complete information must be shared, contrasting success stories with the obstacles that are encountered.

The pathway of phage therapy from the lab bench to the bedside is still hindered. Current accessibility to phage therapy for a given patient is extremely low, with only three options presently available. (i) Medical tourism to established phage therapy clinics is possible, with destinations including Georgia, Poland, Russia, or the United States. This treatment approach incurs a high financial cost (typically at the patient’s expense), sometimes requires intricate paperwork regarding the transport of clinical samples and phages, and may require multiple visits to be effective. (ii) There are extended-access (compassionate-use) programs in some countries, including the United States, Canada, Australia, Japan, China, the United Kingdom, France, Belgium, and Italy (164). Unfortunately, the legal process can be long and convoluted, the programs are currently limited to life-threatening cases, they require a physician willing to consider and apply phage therapy, and there may be limited access to clinically relevant phages. (iii) Clinical trials are currently operating. There are databases of privately and publicly funded studies conducted around the world (such as clinicaltrials.gov). Nevertheless, the recruitment processes are typically difficult and highly selective, studies are limited to specific infections, and the patients face the possibility of being randomized to the placebo control groups.

Equity issues associated with phage therapy arise when considering, for example, the high personal expense associated with medical tourism. There is an existing public perception that phage therapy is available only to highly connected, wealthy individuals with the ability to mobilize large medical teams. Consequently, we recognize that significant efforts will be required to make phage therapy scalable to a larger population and not exclusively a personalized medicine intervention. In any case, increased accessibility must be accompanied by patient education, creating realistic expectations on the possibilities and limitations of therapeutic success.

Interestingly, social media, online platforms, and educational programs can facilitate the communication between patients and phage researchers, helping improve accessibility to phage therapy. Phage Directory (@phagedirectory), for example, is an online “catalogue” of researchers working on open phage therapy for streamlined collaborations between researchers, patients, doctors, and regulatory bodies. Likewise, Phages for Global Health (phagesforglobalhealth.org) develops programs aimed at teaching phage biology to scientists in developing countries and at creating international, multidisciplinary teams that codevelop phage products for specific applications, also in developing countries. Alternatively, the program SEA-PHAGES (Science Education Alliance-Phage Hunters Advancing Genomics and Evolutionary Science; seaphages.org) has engaged undergraduate students from over 73 institutions in phage research by allowing them to isolate, characterize, and name new phages (165). These resources can act as matchmaking applications between phage experts and people needing phages or interested in them, and they have the potential to empower the patients, generate new leaders in the field, and expand the availability of phages, in both the short and long terms.

Efficacy, education, accessibility, and public awareness will all effectively increase the pressure on regulatory agencies to accelerate the establishment of preliminary legal frameworks on the extensive production, testing, use, and vigilance of phage therapy. An approach that has been devised in Belgium is considering tailor-made phage medicines as magistral preparations, thus enabling a preexisting legal framework to be applied (166). Each country should take insights from its particular constitution and medical and industrial systems to develop the regulatory frameworks needed for the use of phage therapy. One of the expected challenges will be the management of intellectual property (IP) rights. Naturally occurring biological organisms, including phages, cannot be patented. As such, IP will likely be focused around methodologies, with first-mover advantage and clinical data on the efficacy of the phage product establishing the market value.

On a brighter side, we believe that the widespread use of phage therapy has the potential to catalyze economic growth. The established pharmaceutical industry has been called a dormant stakeholder for phage therapy whose involvement is limited by the lower degree of urgency it perceives (167). This is the reason why the primary investment into phage therapy translation will likely come from government bodies (168), private foundations, investment firms, commercial linkages, and universities. The pipeline of therapeutic phage manufacturing can be divided into smaller enterprises, including phage isolation, phage library expansion, purification and safety assessment, screening and personalized therapy, automation of methods, large-scale production, storage, and transport. This means that the bulk of this new economic activity could be undertaken by emerging start-ups and biotechnology companies, calling for experts in multiple fields and creating new job opportunities. We acknowledge that the field of phage therapy manufacture can be described as high-risk/high-reward, but successful trailblazers stand to gain considerable revenues and market share.