CRISPR-Armed Phage Therapy Shows Promise in a Rare Multidrug-Resistant E. coli Infection

A peer-reviewed clinical case published in Clinical Infectious Diseases has provided one of the most unusual demonstrations yet of how engineered bacteriophages might eventually be integrated into precision antimicrobial medicine. Investigators from the University of California San Diego, working with the Danish biotechnology company SNIPR Biome, administered SNIPR001, a cocktail of four genetically engineered CRISPR-Cas-armed bacteriophages, to a kidney-transplant recipient suffering from progressive malakoplakia caused by multidrug-resistant Escherichia coli. The therapy was provided under a single-patient emergency Investigational New Drug authorization from the US Food and Drug Administration and was used alongside an extensive conventional treatment regimen rather than as a stand-alone therapy.

The case is remarkable not simply because an engineered phage product was administered to a patient with a resistant bacterial infection. It brings together several of the most difficult problems confronting modern phage therapy at once: antimicrobial resistance, immunosuppression, persistent bacterial reservoirs, an infection involving organisms capable of surviving within host immune cells, anatomically inaccessible disease, limited surgical options and the clinical use of bacteriophages that have themselves been deliberately modified to carry a programmable antibacterial CRISPR-Cas system.

Malakoplakia is an exceptionally unusual setting in which to test such a strategy. The disease is a chronic granulomatous inflammatory disorder associated with defective bacterial killing by macrophages. Instead of efficiently eliminating engulfed microorganisms, macrophages can fail to complete intracellular destruction, allowing bacterial material and sometimes viable organisms to persist. Histologically, the disease is characterized by the formation of Michaelis-Gutmann bodies, distinctive mineralized inclusions found within macrophages and giant cells. Escherichia coli is among the organisms most frequently associated with the condition, particularly in urinary and genitourinary disease. In immunocompromised patients, including transplant recipients, this failure of intracellular clearance can become extraordinarily difficult to manage.

The patient described by Saima Aslam and colleagues was a 65-year-old man who had undergone a deceased-donor kidney transplant in February 2023 for focal segmental glomerulosclerosis. His post-transplant course was subsequently complicated by recurrent E. coli bacteremia, complicated urinary tract infections, necrotizing pyelonephritis of the transplanted kidney, chronic prostatitis and a persistent urinary leak. Seven months after transplantation, an intra-abdominal mass originating around the bladder wall and prostate progressed into a large exophytic lesion extending through the abdominal wall. Biopsies of both the prostate and abdominal lesion demonstrated Michaelis-Gutmann bodies, confirming malakoplakia, while repeated cultures continued to recover E. coli.

The microbiological trajectory was particularly concerning. The bacterial isolate was already resistant to quinolones, cephalosporins and trimethoprim-sulfamethoxazole, antibiotic classes that would normally include several of the agents favoured in malakoplakia because of their capacity to reach intracellular bacterial compartments. Over the course of the illness, the E. coli subsequently became carbapenem resistant. Despite multiple prolonged courses of susceptibility-guided antibiotics and local wound management, the mass continued to enlarge. Surgical correction was considered extremely high risk because the inflammatory mass involved the distal ureter of the transplanted kidney and lay close to critical graft vasculature.

Genomic characterization identified the patient's organism as E. coli phylogroup B2, multilocus sequence type ST1193 and serotype O18. ST1193 belongs to a globally disseminated extraintestinal pathogenic E. coli lineage that has attracted increasing attention because of its association with urinary tract and bloodstream infections and the accumulation of clinically important antimicrobial resistance determinants. In this patient, the combination of progressive disease, extensive anatomical involvement, immunosuppression and increasing resistance created precisely the sort of therapeutic dead end in which experimental antimicrobial strategies are most likely to be considered.

SNIPR001 differs fundamentally from a conventional wild-type bacteriophage cocktail. It contains four complementary engineered bacteriophages developed not only to infect E. coli through the normal phage replication cycle but also to deliver an antibacterial CRISPR-Cas system into the bacterial cell. In the clinical paper, the authors describe these phages as carrying CRISPR-Cas systems programmed to induce DNA breaks in essential bacterial genes, providing an additional sequence-directed mechanism of bacterial killing.

The engineering history behind SNIPR001 is unusually extensive. In the original preclinical development programme published in Nature Biotechnology, researchers began with a library of 162 lytic bacteriophages and tested them against a phylogenetically diverse collection of E. coli isolates. Candidate viruses were selected according to host-range coverage, complementary receptor usage and their capacity to tolerate genetic modification. Selected phages were then subjected to tail-fiber engineering and equipped with CRISPR-Cas machinery, eventually producing four complementary CRISPR-armed phages that became SNIPR001.

The CRISPR component is based on a type I-E system and includes the Cas3 nuclease-helicase together with the Cascade recognition machinery and programmable CRISPR arrays. Once expressed inside a susceptible bacterium, the guide sequences direct the complex toward selected E. coli genomic targets. Cas3 then contributes to extensive target-DNA degradation rather than generating only a localized double-strand break. In preclinical development, CRISPR arrays were programmed against multiple bacterial virulence or essential genes, providing a genetically directed killing mechanism superimposed upon ordinary phage infection.

This architecture is conceptually important. A conventional lytic bacteriophage relies upon successful adsorption, intracellular replication, assembly and lysis. A CRISPR-armed phage potentially delivers two antibacterial pressures at once: the biological consequences of phage infection itself and targeted destruction of bacterial DNA. In theory, this could increase bacterial killing, influence resistance trajectories or retain antibacterial activity under conditions in which the normal phage life cycle is partially compromised. The present case, however, does not establish that CRISPR engineering itself was responsible for the clinical outcome, and the authors explicitly caution against drawing that conclusion.

Before treatment, the patient's E. coli isolate was tested against SNIPR001. The complete cocktail reached a measured titer of approximately 7.4 × 10^10 PFU/mL on the clinical isolate, while individual testing indicated strong antibacterial activity for two of the four component phages. These data provided the microbiological rationale required before compassionate administration. Importantly, this illustrates why even a commercially developed or standardized phage cocktail cannot simply be assumed to work against every isolate of the target species. Phage susceptibility remains strain dependent, and patient-specific microbiological testing continues to be an essential component of precision phage therapy.

The treatment strategy was notable for using three distinct routes of administration. Intravenous SNIPR001 was given at 2 mL per dose at a concentration of 1 × 10^9 PFU/mL twice daily for two weeks. The same concentration was applied topically to visible abdominal lesions twice daily, with the preparation allowed to remain in contact with the lesions before dressings were placed. In addition, an interventional radiologist injected 6 mL of SNIPR001 directly into different regions of the intra-abdominal mass under ultrasound guidance once weekly for eight weeks.

This multimodal delivery strategy is scientifically interesting in its own right. Systemic administration provides circulating phage particles with an opportunity to reach disseminated or vascularized bacterial reservoirs. Topical administration bypasses many pharmacokinetic barriers and exposes surface lesions directly to very high local concentrations. Intralesional injection takes the same logic further by delivering phages directly into an anatomically difficult mass where vascular penetration of systemic antimicrobial agents may be incomplete. In a disease characterized by granulomatous inflammation and poorly accessible bacterial reservoirs, local administration may be especially relevant, although this single case cannot establish the contribution made by each route.

SNIPR001 was not administered in isolation. Eleven days before phage therapy began, the patient had already been placed on a combination including intravenous tigecycline and meropenem, oral fosfomycin, high-dose ascorbic acid and bethanechol. Phage therapy was added to that regimen. Tigecycline and fosfomycin subsequently continued for an extended period, ultimately reaching approximately one year of treatment. This is one of the most important facts to retain when interpreting the case: the outcome cannot be causally attributed to SNIPR001 alone.

Nevertheless, the clinical trajectory after phage initiation was striking. Within approximately one week, the abdominal cutaneous lesions began improving markedly, and several had completely healed within four weeks. Serial radiological measurements showed that the intra-abdominal malakoplakia mass decreased from approximately 744.6 cm³ before phage treatment to 373.4 cm³ after eight weeks, representing roughly a 50% reduction during the period in which phage treatment was being actively administered. At one year, the mass measured approximately 82 cm³, an overall reduction of about 89% from baseline.

https://doi.org/10.1093/cid/ciag446

The timeline reproduced as Figure 1 of the Clinical Infectious Diseases paper makes this temporal relationship particularly clear. It overlays years of recurrent E. coli-positive cultures and successive antibiotic regimens with the introduction of intravenous, intralesional and topical SNIPR001. Disease volume reached approximately 745 cm³ shortly before phage treatment and then declined sharply during the following months, eventually reaching 82 cm³. The figure does not establish causality, but it illustrates why the clinicians considered the response clinically meaningful after a prolonged period of progression.

The urinary leak that had contributed to the patient's complicated clinical course also resolved without surgical correction. A urine culture obtained 88 days after phage initiation was negative, as was a tissue culture from the residual intra-abdominal mass collected 161 days after treatment began. The authors report no subsequent growth of E. coli in post-phage cultures.

Safety observations were equally important because this patient was immunosuppressed following kidney transplantation and received engineered phages through repeated intravenous, intralesional and topical administration. The investigators reported no adverse events attributed to SNIPR001. The patient did develop pancytopenia while receiving high-dose meropenem and tigecycline; this had begun before phage treatment, worsened after its initiation, and resolved after meropenem was discontinued, leading the authors not to attribute the event to SNIPR001.

The investigators also monitored one of the most important immunological variables in repeated phage therapy: neutralizing antibodies. Serum obtained before treatment and periodically during the following eight weeks was incubated with SNIPR001 to determine whether circulating antibodies could progressively block phage infectivity. No serum-mediated phage neutralization was detected through week eight.

That observation is encouraging but should be interpreted cautiously. Intravenous exposure lasted only two weeks, while the longer treatment period consisted predominantly of local administration. The absence of measurable neutralization in a single immunosuppressed transplant recipient cannot establish that repeated systemic administration of SNIPR001 will avoid neutralizing immunity more generally. Antibody responses to bacteriophages vary considerably between individuals, phages, doses and routes of administration, and the patient's immunosuppressive therapy may itself have influenced the response. The authors explicitly acknowledge these possibilities.

Perhaps the most scientifically provocative aspect of the case is the intracellular nature of malakoplakia. Phage therapy is often conceptualized around extracellular bacteria accessible in blood, urine, wounds or biofilms. Here, the pathology is defined partly by bacteria surviving within macrophages whose bactericidal mechanisms are impaired. Whether therapeutically administered phages entered macrophages, remained infectious intracellularly and directly killed intracellular E. coli was not demonstrated.

Several mechanisms could theoretically account for the observed improvement without requiring efficient intracellular phage replication. Reducing extracellular bacterial burden could prevent continual reseeding of macrophage reservoirs. Phages might disrupt bacteria associated with biofilm-like or necrotic regions of the lesion. Direct intralesional administration could produce high local concentrations around infected tissue. Antibiotic-phage interactions could increase bacterial clearance, while tigecycline and fosfomycin may have independently acted within intracellular compartments. The Clinical Infectious Diseases authors explicitly identify intracellular killing, extracellular bacterial reduction, biofilm disruption and antibiotic synergy as possibilities but emphasize that none was experimentally distinguished in this patient.

This mechanistic uncertainty is not a weakness peculiar to this report. It reflects one of the central limitations of compassionate-use phage therapy. When patients are critically ill or have exhausted conventional treatment options, clinicians understandably combine every rational intervention available. That can produce clinically meaningful outcomes while making it exceptionally difficult to determine which component was responsible.

In this case, antibiotics, SNIPR001, high-dose ascorbic acid, bethanechol, changes in immunosuppression and local wound management all formed part of the therapeutic environment. The temporal association between phage initiation and rapid reduction in disease volume is intriguing, particularly given the preceding progression, but the case cannot demonstrate that phages caused the response or quantify their independent contribution.

The authors are appropriately explicit on this point. They state that it remains unclear whether the CRISPR-Cas engineering itself influenced the outcome, whether ordinary bacteriophage activity was primarily responsible or whether the response resulted from the combined effect of phages, antibiotics and the other adjunctive therapies. This restraint is scientifically important because the phrase “CRISPR phage therapy” can easily create the impression that the programmable CRISPR component has already been clinically validated. It has not.

The clinical case is nevertheless particularly valuable because SNIPR001 does not arrive in the clinic as an isolated compassionate-use phage preparation. It has already followed a comparatively advanced translational development pathway. The original Nature Biotechnology programme systematically screened wild-type phages, characterized bacterial receptor usage, engineered selected candidates with both modified tail fibers and CRISPR-Cas systems, examined activity against large panels of clinical E. coli isolates, tested biofilm activity, evaluated resistance and performed in vivo studies in mice and minipigs before selecting the final four-phage combination.

The four SNIPR001 phages were also selected partly for complementary receptor usage, an important strategy for reducing the probability that a single bacterial surface mutation produces resistance to every cocktail component. Preclinical genomic analysis found no known integrases or transposases suggestive of temperate behaviour, no detected antimicrobial-resistance or virulence genes, and no evidence of generalized transduction above the experimental detection limit used in those studies. These features were part of the product's safety-oriented development programme rather than observations made specifically in the malakoplakia patient.

SNIPR001 had also entered humans before this compassionate-use case. A randomized, double-blind phase 1 study in 36 healthy participants evaluated repeated oral administration of the product. Published in The Lancet Microbe in 2026, that trial supported short-term safety and tolerability, found that functional SNIPR001 was predominantly recovered from the gastrointestinal tract rather than systemically, and did not detect significant differences in overall microbiome composition between treatment and placebo groups. The highest-dose cohort showed a numerically larger reduction in E. coli abundance, although the reduction was not statistically significant.

The malakoplakia case therefore extends the clinical story in an important new direction. The phase 1 trial examined oral delivery in healthy volunteers primarily to characterize safety, recovery and pharmacodynamics. This patient instead received the product therapeutically against an established, severe, resistant infection using intravenous, topical and intralesional routes. The indication, anatomy, route of delivery and clinical objective were entirely different.

SNIPR001's principal formal development programme remains focused on preventing E. coli bloodstream infections in high-risk patients with hematological malignancies. A phase 1b/2a randomized, double-blind study is evaluating oral SNIPR001 in patients undergoing allogeneic hematopoietic stem-cell transplantation who carry fluoroquinolone-resistant E. coli in the intestine. The trial is designed to investigate safety, pharmacokinetics, pharmacodynamics and preliminary efficacy in approximately 24 participants. The sponsor's August 2026 press release states that recruitment has been completed; ClinicalTrials.gov still displayed the study as “Recruiting” in its public record, but that record was last updated in October 2025, making the company's newer statement the more recent information available.
The distinction between prevention and treatment deserves attention. SNIPR001 was initially designed to selectively reduce intestinal E. coli carriage before bacteria can translocate across a damaged intestinal barrier and cause bloodstream infection in vulnerable oncology patients. The present case suggests that the same engineered platform may potentially have applications against active infections, including anatomically complex disease caused by resistant E. coli. SNIPR states that it is exploring this broader therapeutic direction, but SNIPR001 remains investigational and has not been approved by the FDA, EMA or any other regulatory authority.

This case also highlights an increasingly important distinction between personalized phage therapy and engineered standardized products. Classical compassionate phage therapy often begins with a patient's bacterial isolate, followed by screening of a phage bank and preparation of an individualized cocktail. SNIPR001 follows a different philosophy. It is a predefined four-phage product engineered and selected to cover a broad spectrum of E. coli strains. The patient isolate is still tested for susceptibility, but the product itself is not redesigned for each individual patient.

If products of this type prove effective in larger trials, they could potentially address one of the logistical challenges that has historically constrained phage therapy: the need to identify, manufacture and release a personalized preparation for each bacterial isolate. A sufficiently broad engineered cocktail could theoretically be manufactured in advance and stored as a standardized pharmaceutical product while retaining the strain specificity that distinguishes bacteriophages from conventional broad-spectrum antibiotics.

At the same time, this approach creates its own regulatory and evolutionary questions. A fixed cocktail must maintain clinically useful coverage across a genetically heterogeneous bacterial species. Resistance can still emerge through receptor modification, intracellular defence mechanisms or CRISPR target mutation. Engineered genetic cargo must remain stable during manufacturing and replication. Each phage component must be characterized independently and as part of the final product. Pharmacokinetics may change according to the site of infection and route of administration.

The use of CRISPR also adds another layer of biological specificity. Conventional bacteriophages discriminate primarily through host recognition and intracellular compatibility. CRISPR-armed phages can theoretically introduce an additional sequence-specific checkpoint because bacterial killing depends upon recognition of defined genomic targets. This could allow future engineered phages to distinguish bacterial populations according not only to species or receptor phenotype but also to particular genetic determinants. That possibility remains more advanced conceptually than clinically, but it is one reason CRISPR-phage platforms have attracted considerable attention.

The therapeutic significance of such specificity could extend beyond direct killing. Broad-spectrum antibiotics frequently perturb commensal bacterial communities while selecting resistance across many species simultaneously. A phage platform designed to attack E. coli while largely preserving unrelated intestinal organisms could potentially reduce some of this ecological disruption. Preclinical and phase 1 SNIPR001 studies support biological selectivity toward E. coli under the conditions tested, although whether this advantage translates into superior patient outcomes remains to be demonstrated.

The malakoplakia case also illustrates why route of administration deserves far greater attention in phage medicine. Intravenous therapy is frequently discussed as though systemic circulation were sufficient for every deep infection. In reality, granulomatous tissue, necrotic cavities, abscesses, implants and poorly vascularized lesions may expose phages to the same penetration problems encountered by conventional antimicrobials. Direct intralesional administration, as performed here under image guidance, represents one way of bypassing those anatomical barriers.

Such strategies may become particularly relevant as phage therapy moves toward highly individualized infectious-disease interventions. Rather than viewing the phage itself as the entire treatment, clinicians may increasingly need to design a delivery architecture around each infection: selecting systemic, inhaled, topical, local, intraventricular, intra-articular or direct lesion administration according to where the bacteria reside.

The case therefore offers a useful illustration of how future phage therapy may look less like prescribing a conventional antibiotic and more like designing a multimodal intervention around the biological geography of infection.

There are equally important reasons not to overinterpret the result. This is one patient. There is no control group. Multiple active antimicrobial and adjunctive therapies were administered simultaneously. Malakoplakia can have a variable clinical course. The relative contribution of antibiotics, phages, CRISPR engineering, local administration and changes in bacterial ecology cannot be separated. Long-term recurrence remains possible. Absence of phage-neutralizing antibodies in one immunosuppressed individual cannot be generalized.

Yet case reports can still matter profoundly when they test something that has rarely or never been attempted clinically. Their role is not to establish efficacy but to reveal feasibility, identify unexpected toxicities, generate mechanistic questions and define what should be examined systematically next.

Here, the clinically important observations are concrete. An engineered four-phage CRISPR-Cas product could be administered intravenously, topically and intralesionally to an immunosuppressed kidney-transplant recipient without identified phage-related adverse events. The patient's resistant E. coli isolate was susceptible to the product before treatment. Cutaneous lesions healed rapidly after therapy was introduced. The intra-abdominal disease burden decreased by approximately 50% during the first eight weeks and by 89% over one year. Follow-up cultures from urine and tissue were negative. No serum neutralization of the phage cocktail was detected during the monitored period.
What remains unknown is arguably even more scientifically interesting. Did intact bacteriophages reach intracellular E. coli inside macrophages? Did local phage replication occur inside the malakoplakia lesion? Did the CRISPR cargo materially increase bacterial killing? Did phages disrupt biofilm or extracellular bacterial reservoirs and thereby allow macrophages and antibiotics to regain control? Was there meaningful phage-antibiotic synergy? Could the response have occurred with the corresponding non-engineered phages? How much did intralesional delivery contribute relative to intravenous treatment?

The authors themselves argue that future studies should directly examine phage uptake, intracellular persistence and bactericidal activity within macrophages, both alone and in combination with antibiotics. These experiments could be particularly important because intracellular infections represent one of the least understood frontiers in bacteriophage pharmacology.

If phages can reliably access bacteria residing within professional phagocytes or other host cells, the implications could extend beyond malakoplakia. Numerous clinically important pathogens establish intracellular or partially intracellular reservoirs, and these niches can substantially limit conventional antimicrobial activity. If phages cannot efficiently reach such compartments, understanding how extracellular bacterial depletion indirectly affects intracellular disease will be equally important.

The report also reflects a broader maturation of engineered bacteriophage therapeutics. The field has progressed from identifying natural viruses, to modifying receptor-binding proteins, to inserting programmable antibacterial systems and now to administering CRISPR-armed phages in carefully controlled human settings. SNIPR001 is among the most advanced examples of this progression, combining classical bacteriophage biology with synthetic biology, bacterial genomics and pharmaceutical development.

Its development has received support through CARB-X and multiple public and philanthropic funders. The Clinical Infectious Diseases paper notes that CARB-X supported development of SNIPR001 with funding originating from BARDA, Wellcome and Germany's federal research ministry, among others, while the company describes additional support across its broader programme. SNIPR001 itself was supplied without charge for this compassionate-use treatment.

This matters because engineered phage therapy will require a fundamentally different infrastructure from isolated compassionate-use preparations. Genetically modified phages must be manufactured reproducibly, their engineered sequences must remain stable, their bacterial production hosts must be controlled and their biological activity must be measured using assays suitable for pharmaceutical release. Regulatory authorities must evaluate both the conventional risks associated with live bacteriophage products and additional questions generated by their engineered genetic payloads.

For now, the most defensible interpretation of the malakoplakia case is neither that CRISPR phage therapy has been proven effective nor that the clinical response can be dismissed because antibiotics were administered simultaneously. The case occupies the scientifically important territory between those extremes.

It demonstrates that a pharmaceutical-grade, CRISPR-enhanced bacteriophage cocktail could be deployed as adjunctive therapy against a severe multidrug-resistant E. coli infection in a profoundly complex clinical setting, using several administration routes, without identified phage-related toxicity. It also documents a substantial and durable regression of disease that began rapidly after phage therapy was introduced following prolonged preceding progression. Whether SNIPR001 caused that improvement, amplified the effect of antibiotics or contributed through some more complex interaction remains unresolved.

That uncertainty is exactly why the case deserves attention rather than sensationalism.

Phage therapy is often described as a century-old treatment returning because antibiotics are failing. SNIPR001 represents something different. It combines that century-old biological principle with programmable CRISPR-Cas antibacterial activity and a pharmaceutical development strategy intended to produce a standardized precision medicine.

The patient described in Clinical Infectious Diseases does not prove that this strategy works broadly. But the case moves engineered CRISPR phages from the conceptual and preclinical domain into the reality of treating an individual with a progressive, anatomically difficult, multidrug-resistant infection.

For a field attempting to transform bacteriophages from biological curiosities and compassionate-use interventions into reproducible antimicrobial medicines, that is a meaningful step.

The next milestone must be considerably harder: demonstrating, in controlled clinical studies and mechanistic experiments, exactly how much engineered phages contribute, which patients benefit, which routes of administration are optimal, and whether CRISPR provides a clinically measurable advantage beyond the remarkable antibacterial biology that bacteriophages already possess.

Primary scientific source:

Aslam S, Brubaker A, Shah M, Martin TCS, Mandt T, Windham-Herman AM, Zhang H, Bryde T, Gençay YE, Gram A, Grøndahl C, Haaber JK, Hallström B, van der Helm E, Hink J, Maccario L, Østergaard A, Troy A, Yavari N, Schooley RT, Mehta S. First-in-Human Adjunctive Bacteriophage Therapy for Malakoplakia. Clinical Infectious Diseases. Published online 11 August 2026.

https://doi.org/10.1093/cid/ciag446

Additional scientific context: Gencay YE et al. Engineered phage with antibacterial CRISPR–Cas selectively reduce E. coli burden in mice. Nature Biotechnology. 2024;42:265–274. https://doi.org/10.1038/s41587-023-01759-y

Petersen AØ et al. Safety, recovery, and pharmacodynamics of CRISPR-Cas therapeutic SNIPR001: a phase 1, randomised, double-blind, first-in-human, dose-escalation study. The Lancet Microbe. 2026;7:101257. https://doi.org/10.1016/j.lanmic.2025.101257