Phage Therapy in the Resistance Era: Where Do We Stand and Where Are We Going?

The ability to control infectious diseases marked a revolution in human health, shifting the global burden of disease from infectious diseases to noncommunicable diseases, which are now considered a leading cause of death. The 20th century brought an appreciation of bacteriophages (often known simply as phages) and their relationship to infectious diseases, which resulted in the implementation of phage therapy to decrease the incidence of several infectious diseases. From the first successful administration of phages in 1921 at Hôpital des Enfants-Malades, Paris, France, treating children with Shiga dysentery to its widespread use in humans around the world, the effectiveness of phage therapy was highly controversial. In 1945, a new era began with the golden age of antibiotics, and phage products were taken off the market. However, shortly thereafter, penicillin resistance became a substantial clinical problem. In response, new classes of antibiotics and modifications to old classes were developed and deployed.

However, the antibiotic pipeline has been dry for decades. Today, multidrug-resistant (MDR) bacteria are becoming commonplace, making available antibiotics that are much less effective. In 2017, the World Health Organization released a global priority pathogens list of antibiotic-resistant bacteria (Table I). On the basis of diverse criteria, such as mortality, prevalence of resistance, treatability and current pipeline, the situation is critical for most difficult-to-treat, community-acquired, health care–associated, and nosocomial infections caused by ESKAPEE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter species, and Escherichia coli) pathogens. We included Mycobacterium tuberculosis to the critical category as an additional priority global health threat. The list is an excellent guide for the prioritization of research and development aimed at discovering new antibacterial therapies, which are critically needed.^,

The inevitable demise of all available antibiotics has reignited the development of phage therapy. Phages are naturally occurring viruses that infect both gram-positive and gram-negative bacteria. As such, they are generally unaffected by antibiotic resistance and (unlike most antibiotics) are able to target bacteria encased in biofilms.^, A substantial volume of preclinical data and an increasing body of clinical evidence indicate the immense therapeutic potential across a wide range of infectious diseases. For decades, the Eliava Institute in Georgia and the Ludwik Hirszfeld Institute in Poland have provided phage therapies to hundreds of nationals and internationals. The Eliava Institute has routinely provided off-the-shelf fixed-phage products and personalized single-patient phage treatments as extempore medications. The Ludwik Hirszfeld Institute provides phage therapy when the standard of care has failed, under the Declaration of Helsinki (compassionate use). Recently, Belgium implemented single-patient phage therapy regulation as magistral preparations (ie, pharmacy compounding), with phage products prepared and deployed by the Queen Astrid Military Hospital. France has adopted a similar exceptional single-patient authorization dispensed by hospital pharmacies to patients with a prescription. In the United States, the Food and Drug Administration recently approved multiple phage therapy clinical trials and several phage therapies as expanded access Investigational New Drug (eIND) for patients with an immediately life-threatening condition or serious disease where standard of care has failed.^{,  ,} 

In this review, we summarize the increasing body of clinical data on phage therapy and discuss how phages transition from bench to bedside and beyond. Our aim is to highlight the importance of approaching translational and clinical research from various angles and develop innovative ways to overcome the many challenges with implementing phage therapy.

Phages are bacteria-restricted viruses that exploit all natural environments, including the human body.^, With >10³¹ phages estimated to be in the biosphere, they are readily isolated from soil, bodies of water, feces, and sewage. Although phages come in a variety of morphologic structures and genome types, most discovered phages are nonenveloped icosahedral head and tail in structure and have double-stranded DNA genomes. Viral tails are unique to phages and can be contractile, short noncontractile, or long noncontractile, which allow for viral classification as myoviruses, podoviruses, and siphoviruses, respectively. Phage genomes have enormous genetic diversity, with as much as 80% of the predicted genes having no known function. In addition, phages exhibit substantial gene diversity even with phage strains that infect a common bacterial host. Together with their enormous abundance, great diversity, and relative ease of isolation, phages provide an unlimited source of antibacterial agents against all human bacterial pathogens (Fig. 1).

When virulent double-stranded DNA phages infect susceptible host bacteria in vitro, the number of phages in the culture supernatant does not increase until after, generally, 30–40 min at 37 °C. This period is required for a phage particle to inject its genome into the host cell, hijack host metabolism, express its viral genes, assemble phage particles, and release progeny into the environment (Fig. 1). Cell lysis is the result of viral endolysins cleaving bonds within the cell wall peptidoglycan, which causes the cell wall to destabilize and osmotic rupture to occur under the high internal pressure. A successful infection by a virulent phage always leads to cell lysis.

By contrast, infection by a temperate phage can lead to lysogeny. That is, temperate phages infect bacteria as a symbiotic prophage and replicate along with the bacterial chromosome during bacterial cell division. When exposed to environmental cues or stressors, the prophage resumes a lytic cycle to create phage particles. An inherent capacity of prophages to mediate the transfer of genes between bacteria by specialized transduction—an event that may increase bacterial virulence or promote antibiotic resistance—generally disqualifies temperate phages for therapeutic purposes. However, advances in synthetic biology have provided new opportunities for the use of temperate phages as a therapy against bacterial infections by removing so-called lysogeny genes from their genome.^, This is beneficial for infections caused by bacterial species where little to no virulent phages have been discovered, such as Clostridium difficile and Mycobacterium abscessus.

The parasitic and enzymatic bacterial killing by phages are different from the antibacterial mode of action by all other drugs, which allows them to be effective against most MDR bacteria (Fig. 2). In addition, self-replication in the presence of susceptible bacteria provides density-dependent dosing at the site of infection. Reducing the number of daily doses through extended-release drugs can reduce the relapse of disease, symptoms associated with drug therapy (ie, drug-related problems), patient nonadherence, and costs. However, the association between the number of phage doses per day and relapse in infections remains uninvestigated .

A unique aspect of phage therapy is the potential for low-dose phage treatments, relying on self-replication to bring up the phage concentration at the site of infection. For self-dosing to occur, target host cell densities need to be over a replication threshold of ≥10⁴/CFU/mL.^,

Phage infection can only take place after the viral tail fibers bind irreversibly to specific receptors on the bacterial cell wall, including lipopolysaccharide (endotoxin), teichoic acid, pili, outer membrane proteins, efflux pumps, and polysaccharide. This stringent compatibility causes phages to be specific against bacterial targets and prevent tropism for mammalian cells. Nevertheless, phages can rapidly translocate throughout the body, even penetrating compartments previously considered sterile.^, However, despite the increasing number of clinical applications, there is a limited understanding of the interactions between phages and human cells.

It is well recognized that all recipients of phage therapy have so far been treated empirically, with incomplete information on, for example, the phages, adequate route of administration, dosing, duration, and antibiotic compatibility. Figure 1, Figure 3 and the subsections below briefly summarize select phage therapy clinical trials and single-patient reports between 2005 and 2020 grouped by route of administration. Further case data are summarized in Supplemental Table 1. With limited clinical trial data, lessons learned from personalized case studies under eIND have advanced medicinal phage technologies and highlighted potential risks and challenges, which we review in subsequent sections.

Not all phages are candidates for therapeutic purposes. As mentioned, virulent phages with obliquely lytic replication cycles are preferred (see Fig. 1) because cell lysis is indeterminate and there is a reduced risk of viral mediated horizontal gene transfer via transduction. Genome sequencing and bioinformatic workflows^{,  ,  ,  ,}  permit identification of undesirable gene features to rapidly rule out unsuitable phage strains based on homologies to known genes on existing databases. For example, carrying lysogeny-associated repressors and integrases, bacterial virulence, toxin, and/or antibiotic-resistance genes would disqualify a phage for human therapy.^, However, it may be possible to knock out genes with undesirable traits when, for example, virulent phages are difficult to isolate.^, Continued efforts to improve the quality and accuracy of archived phage genomic information will contribute to increasing the tolerability of phage therapy.

In general, a good phage candidate for therapy is one that infects a wide range of bacterial strains. There is evidence that myoviruses exhibit a broader host range. For example, a single phage lysed 17 of 28 S aureus clinical isolates. In contrast, narrow host range phages may present new therapeutic opportunities. For instance, targeting a prevalent and specific strain of pathogenic bacteria could avoid adverse effects associated with host microbiome dysbiosis.

Another feature important in phage selection is their natural ability to disrupt structured communities of bacterial cells that attach to surfaces and enmesh in biofilm. Many infectious diseases in humans are the result of, or exacerbated by, biofilms. Self-made exopolysaccharide matrix provides an essential scaffold for biofilm development, promoting bacterial adhesion to surfaces and cohesion as well as hindering diffusion. Although it is often assumed that biofilms confer resistance to phages, most phages readily infect bacteria within biofilms. Indeed, phages have coevolved with bacterial biofilms; thus, their infection of encased bacteria is expected. For instance, some phage tail fibers and tail-spikes even carry depolymerases that degrade exopolysaccharides to unmask cell receptors on planktonic cells and disrupt the exopolysaccharide matrix of biofilms.^{,  ,  ,}  In vitro, phages have reduced both single-species^,, and dual-species^{,  ,}  biofilms. In an artificial CF sputum, phages could reduce P aeruginosa biofilm by 3-logs, indicating that phages readily penetrate and lyse bacteria encapsulated in biofilms. Because phages lyse distal bacteria, interior cells awaken and become more metabolically active because of increased oxygenation and nutrient exposure, thus becoming more sensitive to phage attack. Certain phages, however, poorly penetrate the biofilm matrix, and others can promote a mucoid phenotype of host bacteria that enhances biofilm formation. Although biofilm infections are common and cause clinically significant and potentially fatal infections, much work remains in the quest for effective antibiofilm phages in clinical settings.

The extended storage of phages is another important consideration. Phages keep in short-term cold storage at 4 °C when purified in buffered solution, concentrated, and shielded from light.^{,  ,}  Even then, phage counts generally trickle down for months. Cryopreservation can minimize the loss of phages, but phage particles are inactivated with freezing and thawing cycles. Long-term storage is also feasible with phage lyophilization (freeze-drying) using stabilizing additives. Although lyophilization is not compatible for most clinical applications, it may be suitable for inhaled phage therapies.^, Although many phages can be stored in these conditions, specific storage conditions for each phage strain require evaluation.

In clinical practice, phages are typically administered as a mix (cocktail) of viral strains.^,,, Well-planned phage combinations have several benefits over use of a single-phage strain. These benefits include, (i) broadening target bacterial strain spectrums, (ii) targeting multiple bacterial species at once, (iii) increasing dose potency by multiple phage strains attacking the same bacterial cell, and (iv) limiting resistance by forcing the target bacterium to evolve resistance to multiple phages simultaneously to survive.^{,  ,  ,  ,}  The drawback is that individual phages generally require a reduction in concentration when mixed into a single dose. There is also the potential that phages will interfere with one another, for example, compete for the same bacterial cell receptor and drive cross-resistance.

Conversely, a single-phage strain can precisely target a specific bacterial strain.^, The drawback of using a single-phage strain is that it requires careful definition of the etiologic bacterium before therapy. Thus, this approach is primarily for proof of concept, efficacy, and tolerability in vitro and animal studies.

Generally, 3 pipelines exist for phage drug products: off the shelf, pharmacy compounding, and on-demand de novo. Off-the-shelf, fixed-phage products are designed to target not-yet-identified pathogens (eg, multiple-phage strains targeting multiple bacterial species) or identified pathogens (eg, multiple phages targeting a single bacterial species). They are manufactured under current Good Manufacturing Practice production, and final products undergo phased clinical safety and efficacy testing. Fixed products enhance tolerability, quality, and uniformity. However, the clinical usefulness of fixed-phage products is unclear because of, for example, high strain variability of target bacterial pathogens, rapidity of phage resistance, encountering multispecies infections, and varied phage storage stability.

Compounding by a licensed pharmacist with a physician's prescription can create phage medications that are tailored to the individual needs of a single patient. This approach may be required when the patient's infection isolates are not sensitive to phages in the fixed product, dosage is unavailable, or the fixed-phage product is not well tolerated (see the Drug-Related Problems section below). Compounding also allows the insertion of phages into topical creams, transdermal gels, or other dosage forms.

To implement phage compounding, large and diverse collections of purified phages (phage banks) are a prerequisite.^, Indeed, bacterial susceptibility and phage strain diversity have been the driving forces in library construction. For instance, the Queen Astrid Military Hospital has an extensive phage collection. However, only 15 of 260 patient isolates were sensitive to phages in the collection. Likewise, in a bank with >10,000 mycobacteria phages, only 6 phages lysed M abscessus isolated from a patient with CF. Adaptive Phage Therapeutics diverse and well-characterized PhageBank collection also did not contain a phage strain suitable against a patient's K pneumoniae isolate. Thus, proper phage selection criteria and library size have yet to be identified. A potential improvement may be to ensure a diverse collection of phages that use different bacterial receptors.

On demand, de novo personalized phage products have so far provided better overall efficacy.^,,, This pipeline allows phages to be isolated specifically against bacterial isolates directly from the patient.^, Thus, phage products can be rapidly and continually modified to respond to, for example, phage resistance.^,, The problem remains that on-demand phage products can take weeks to months to manufacture.

The pharmacokinetic and pharmacodynamic properties of phages in vivo are complicated and largely unknown.^, Phage PK properties are unique because of their relatively large size (approximately 100 times larger than antibiotics and 10 times larger than antibodies). In addition, phages consist of several different proteins, which can increase their rate of clearance by the mononuclear phagocyte system and neutralization by antiphage antibodies.^, Phage efficacy in entering animal or human bodies strongly depends on the route of administration. Oral administration effectively delivers phages to the gastrointestinal tract, and efficiency of gut transit is dose dependent. Fecal phage counts peaked at 3.0 × 10⁴ PFU/g after oral consumption of 1.5 × 10⁷ PFU in mineral water 3 times daily. Patients orally treated with phages for acute cholera at a dose of 10¹³ PFU, together with phage propagation at sites of infection, had fecal concentrations of >10¹¹ PFU/g. This finding suggests that phage passage through the stomach can be efficient. Absorption of phages into the blood, however, was inefficient, reaching only 10² PFU/mL. Phages can easily penetrate most organs, although they readily accumulate in the spleen and liver.^{,  ,} 

For phages to be effective, they must reach the site of infection in sufficient numbers and infect as many target bacteria as possible, without causing patient harm. Conversely, removal of phages from the site of infection is an important factor in terminating action. Phages delivered systemically can penetrate the lumen of murine lungs. Phages in the plasma and tissues (spleen, kidney, liver, and lung) of rats after a single 1-mL intravenous bolus of 10¹⁰ PFU/mL exhibited a half-life of 2.3 and 9 h, respectively. Continuous infusion of 0.1 mL/h for 24 h of the same phage mix and titer resulted, however, in plasma progressively increased to reach a plateau of 10⁷ PFU/mL after 6 h, which is comparable with the concentration achieved after bolus injection. In humans, after a dose of 8.5 × 10⁷ PFU in 4 mL of lactated Ringer solution, administered through a peripherally inserted central catheter, phage levels were detectable after 5 min in blood samples but were undetectable after 50 min.

Studies examining the pharmacodynamic properties of phages, defining synergy, additivity, indifference, antagonism, and host immune system interaction, are scarce. The monotherapy versus mix versus phage-antibiotic combination therapy debates have not yet considered the potential additional value of the patient immune system. It is still unclear whether the initial phage dose or phage replication most influences outcomes as well as what is a sufficient level of target bacteria to allow phage replication to overtake bacterial replication. Thus, the effectiveness of phage therapy likely increases as the infection worsens, inflating the likelihood that phages will encounter a bacterial host. Inefficient outcomes, however, may arise from the tendency for phages to bind to bacterial debris.

With the uniqueness of phage therapy, studying pharmacokinetic and pharmacodynamic properties is challenging. Most proof-of-principle studies of phage therapeutic applications use very high titers of phage to achieve efficacy, which is unlikely to be achieved in humans. However, given the low toxicity of phages, phage therapy should be successful provided that phages reach the site of infection in sufficient numbers and are able to replicate within target bacteria.

Drug-related problems actually or potentially interfere with desired health outcomes, including adverse reactions and emerged resistance. They are relatively common in hospitalized patients and can result in morbidity and mortality as well as increased costs. The range of adverse reactions during phage therapy varies widely (summarized in Table II). A 2-year-old experienced anaphylaxis after intravenously administered phages, and a 15-year-old patient with CF became flushed after initial intravenous phage treatments. Topical application of S aureus phages has caused mild adverse reactions, such as rhinalgia, oropharyngeal pain, and metabolic acidosis. Multiple patients experienced hypoxemia and hypothermia during topical Pseudomonas phage treatment of burn wounds. Oral administration of E coli phages has caused transient abdominal pain, dyspepsia and toothache in adults.

We speculate that adverse reactions can arise because of bacterial lysis by the phages and/or direct phage-mammalian host immune system interactions and/or residual bacterial toxins in phage products.^, Phage-induced rapid release of bacterial cell wall components may have immediate adverse effects for the patient. Animal studies have found that phage-induced endotoxin release is significantly lower in lysis-deficient phages compared with lytic phages, which leads to decreased proinflammatory cytokines (eg, tumor necrosis factor α and interleukin [IL] 6).^, However, the amount of endotoxin released during antibiotic action is also clinically important. For instance, some classes of β-lactam antibiotics lead to markedly increased levels of free endotoxins, whereas treatment with carbapenems and aminoglycosides produces relatively low amounts of endotoxins. Dufour et al found that phage-mediated bacterial lysis–released endotoxins were similar to β-lactam antibiotics in vitro.

Direct interactions between phages and the host immune system can induce signaling cascades that can result in elevated cytokines, increased immune cell infiltration, increased phagocytosis, and adaptive immunity activation (T-cell response and antibody production by B cells). Phages stimulate antiviral interferons through activation of Toll-like receptor 9 and play crucial roles in the innate immune system by recognizing microbial DNA as a pathogen-associated molecular pattern. Treating mice with phages led to immune cell expansion in the gut and increased interferon-γ production, which worsened murine colitis. Human lung epithelial cells increased IL-6 and tumor necrosis factor α production in response to exposure to certain Pseudomonas phages. Similarly, certain staphylococcal and pseudomonal phages induce proinflammatory cytokines, such as IL-6 and IL-1β, in murine dendritic cells, whereas other staphylococcal phages did not elicit human dendritic cell cytokine expression. Together, data suggest that immune responses are phage strain dependent.

Moreover, some phages can bind to cell surface polysialic acids as a receptor for cell entry, whereas other phages will internalize through epithelial cell pinocytosis. However, the physiologic and immunologic consequences of harboring internalized phage particles are unclear.

Certain structural proteins of phages can elicit weak humoral responses after prolonged exposure, which can lead to the antibody neutralization of phage particles.^, Antiphage antibodies have been recovered from patients who received phage therapy as well as those who have not, presumably because of natural contact with phages during infections with associated bacteria. A phage safety trial in humans found no antiphage IgG, IgM, or IgA antibodies detected in the blood of recipients after oral ingestion of phages for 2 consecutive days. In animal models, however, continuous oral consumption of phages can induce weak antiphage antibodies (IgG after 36 days and IgA after 79 days). A different phage type induced IgG and IgM antibodies after prolonged murine oral ingestion, and Staphylococcal phages induced IgM, IgG, and IgA antibodies in blood. Notably, phage numbers in murine models are significantly higher compared with human therapeutic doses. Although phages appear to induce only weak antibody responses, antiphage antibody responses may lead to the activation of the complement system. Human blood from recipients of phage therapy was found to contain antiphage IgG, IgM, and IgA antibodies in vitro. However, no correlation was found between the induction of phage-neutralizing antibodies and the outcome of phage therapy. Antiphage antibodies, although likely not harmful to patients, could limit the effectiveness of phages given therapeutically. To date, this problem has not been reported. Identifying immunologic interactions with phages that contribute to the development of adverse reactions may aid in the prevention of problems related to phage therapy.

Another complication of phage therapy is that bacteria can thwart phage attack through an arsenal of antiviral mechanisms, targeting almost every step of the phage replication cycle. These mechanisms include spontaneous chromosomal mutations, the ability to block the entry of phage genetic material (eg, superinfection exclusion), DNA restriction-modification enzymes, abortive infection, and CRISPR-Cas adaptive immunity. A major mechanism that drives phage resistance is spontaneous chromosomal mutations governed by Darwinian dynamics. During treatment of an infection, a small number of mutations will spontaneously arise (rate of approximately 10⁻⁸), followed by phage-resistant subpopulation regrowth.^{,  ,}  Animal studies have found bacteria can rapidly shrug off phage attack, leading to uncontrolled proliferation of phage-resistant mutants and treatment failure.^{,  ,}  However, target bacterial populations are likely to experience large, cyclic swings (predator-prey dynamics) in population size before resistance dominates.^,

Among 12 phage therapy clinical studies, 7 cases confirmed the emergence of phage resistance during treatment. For example, A baumannii isolates from blood cultures exhibited resistance to all 8 phages after only 1 week of therapy. Similarly, resistance developed during topical treatment but after nearly continuous dosing for 3 months. Most studies have not adequately tested for the emergence of phage resistance, likely because of successful outcomes. However, cases continue to indicate that resistance emerges during phage therapy, which might be attributable to unreported treatment failures.

Despite the near certainty that phage resistance will arise, multiple scenarios occur with which phage resistance does not impede a positive therapeutic outcome.^, The patient immune system, given an adequate response, can eradicate emerged resistant bacterial mutants. If resistance to phage treatment arises, switching to new phages with different binding sites or order of exposure can be therapeutic strategies to improve efficacy. The maintenance of bacterial antiviral defense systems comes with a cost, such as DNA restriction-modification and CRISPR-Cas enzymes.^, Evolution of resistance can lead to attenuation of bacterial virulence^,, or promote collateral sensitivity to antibiotics.^, Lastly, phages also harbor their own defenses to counteract bacterial immunity, such as antirestriction and anti-CRISPR enzymes,^, as well as counteradapt to reinfect resistant bacteria.^,

One approach that has received little exploration is the use of phage adjuvants, also referred to as phage potentiators and resistance breakers. These adjuvants are active compounds with little or no effect on bacterial growth in isolation and when c-administered with phages act to enhance phage activity or block phage resistance. The combination of synergistic antimicrobials are also adjuvants. Chemical antibiotics can act as phage adjuvants by boosting phage particle production. For instance, a subinhibitory concentration of cephalosporin increased an E coli cell's production of the phage particles by 7-fold. Phage-infected Burkholderia cenocepacia cells produced higher phage counts in the presence of subinhibitory concentrations of meropenem, ciprofloxacin, and tetracycline. Antibiotics that act by inhibiting cell division (β-lactam, quinolones, and nalidixic acid) or cause an elongated morphologic structure (ciprofloxacin and meropenem) can permit higher production of phage particles.^, Likewise, subinhibitory concentration of tetracycline causes cell clustering, which may promote increased phage infections by minimizing lateral travel between adjoined cells, thereby enhancing contact with phage receptors on uninfected cells. Importantly, elevated phage production is unaltered when bacterial cells become antibiotic resistant. Combining phages with subinhibitory concentrations of ciprofloxacin or meropenem can also inhibit regrowth of phage-resistant mutants in a murine endocarditis model and thus improve the phage therapy outcome.

Bacterial biofilms represent a major obstacle in the fight against bacterial infections because they are inherently refractory to various antibiotics. A potential adjuvant to phage therapy include DNAs that can degrade extracellular DNA, which plays important roles in both the aggregation of bacteria and the interaction of the resulting biofilm with polymorphonuclear leukocytes during an inflammatory response. Sugar alcohols, such as xylitol, sorbitol, and erythritol, can also be phage adjuvants. They inhibit bacterial growth by diffusing into the biofilm and accumulate as a toxic, nonmetabolizable sugar alcohol phosphate. Together, these findings suggest that phage adjuvants may be a way to improve the efficacy of bacterial killing during treatment.

Conversely, phages in combination with antibiotics offer an orthogonal approach to newer antibiotic discovery.^, For example, some phages infect gram-negative bacteria by binding to TolC or its homologues, a common component of a wide variety of multidrug efflux pumps.^,, Phage infection of MDR bacteria can lead to phage resistance (eg, tolC mutants), which can induce a collaterally sensitive response to antibiotics. Similarly, the phage OMKO1 infects P aeruginosa by binding to the efflux pump MexAB/MexXY. Mutations in the gene encoding of the outer membrane porin M, which is part of the efflux pump complex, impaired phage infection but restored the sensitivity to ciprofloxacin. E faecalis is a human intestinal pathobiont with intrinsic and acquired resistance to many antibiotics, including vancomycin. Despite gaining phage resistance, mutant strains exhibited a loss of resistance to cell wall–targeting antibiotics in vitro. Identifying evolutionary trade-offs is an emerging strategy for combating antibiotic resistance, such as prescribing sequences of synergistic antimicrobials wherein the evolution of resistance to the first induces susceptibility to the second.^,