The problem begins with a paradox inherent to phages themselves. These viruses are essentially nucleic acids wrapped in protein coats, decorated with tail fibers and baseplate structures that must remain precisely folded to recognize and inject their genetic payload into bacterial hosts. Unlike small-molecule drugs, which can often tolerate considerable environmental variation, phage infectivity depends on conformational integrity at the nanometer scale. A tail fiber that unfolds, a capsid that cracks, or a protein subunit that detaches can render a particle noninfectious even though the genetic material remains fully intact. Liquid formulations, which are preferable for many clinical settings because they avoid the stresses and costs of lyophilization, expose particles continuously to the surfaces of their containers, to each other, and to solution conditions that shift subtly over time.

Surface adsorption is the first and often the fastest loss mechanism. Phage particles, like many proteins, carry a mosaic of charged and hydrophobic patches that attract them to interfaces. Glass vials, plastic syringes, rubber stoppers and tubing all present surfaces to which virions can adhere, sometimes irreversibly. For low-concentration preparations, intended for pediatric dosing or for phages that are difficult to propagate to high titers, even a small fraction of adsorbed particles can translate into a clinically significant loss of dose. The study underscores that adsorption is not uniform across container materials: hydrophobic polymers tend to sequester particles more aggressively than borosilicate glass, but glass introduces its own risks through leached metal ions and alkaline surface chemistry. The authors emphasize that adsorption must be evaluated for every combination of phage, container and buffer, because a formulation that works beautifully for one phage type may fail entirely for another.

The second threat, aggregation, arises when particles collide and stick together, forming dimers, larger clusters and eventually visible precipitates. Aggregation is driven by electrostatic attraction between oppositely charged regions of adjacent particles, by hydrophobic interactions between exposed protein patches, and by the removal of stabilizing hydration shells during freezing and thawing or during shifts in ionic strength. Aggregated phages are problematic on multiple levels. Aggregates sediment during storage, creating heterogeneous dosing; they are cleared more rapidly from the bloodstream after injection; and they can trigger immune responses or exceed particulate limits set by pharmacopoeial standards for injectable products. Crucially, aggregation and infectivity loss are not always correlated in simple ways, which means that turbidity measurements or particle counts alone cannot substitute for plaque assays when assessing formulation quality.

The third mechanism, structural degradation, is the slowest but perhaps most insidious. Over weeks and months in liquid storage, capsid proteins can undergo deamidation, oxidation and backbone cleavage, while tail structures can gradually lose the precise geometry required for host recognition. Temperature accelerates these processes dramatically, which is why cold-chain dependence has been such a burden on phage therapy programs. The research highlights that structural degradation is often detected only through functional assays: a formulation may look perfectly clear and show normal particle counts by electron microscopy, yet exhibit a steady decline in plaque-forming units as more and more particles lose their ability to infect. This decoupling of appearance from activity is a central argument for rigorous, infectivity-based stability testing.

Against this backdrop, the study evaluates the protective roles of specific classes of excipients, the pharmaceutically accepted, nonactive ingredients that surround the phage in the final product. Sugars and sugar alcohols, including sucrose, trehalose and mannitol, emerge as foundational stabilizers. These molecules act through preferential exclusion, a thermodynamic effect in which the sugar is excluded from the surface of the protein, effectively making the folded state energetically favorable and discouraging both unfolding and aggregation. Trehalose in particular has a long history in the stabilization of biological drugs and vaccines, and the findings reinforce its value for phages, where it buffers particles against osmotic shocks and reduces interfacial stress. Polysorbates and other nonionic surfactants occupy a complementary niche: by adsorbing competitively to container surfaces and air-liquid interfaces, they deny phage particles access to the very interfaces where adsorption and surface-induced denaturation occur. Even low concentrations of surfactant can dramatically reduce the fraction of virions lost to a plastic syringe barrel.

Amino acids and buffering agents form the third pillar of the protection strategy. Arginine, glycine and histidine appear repeatedly in biologics formulations because they disrupt electrostatic attractions that drive aggregation, while histidine buffer offers mild pH control near neutrality, where most phages are comfortable. Plasma proteins and serum albumin, historically used as catch-all stabilizers in early phage work, are discussed in the context of modern pharmaceutical expectations: although effective, animal-derived proteins raise regulatory and supply concerns, pushing formulators toward chemically defined alternatives. Salts deserve careful handling, the authors note, because ionic strength simultaneously affects electrostatic repulsion between particles, adsorption to charged surfaces and the osmotic environment of the capsid; there is no universally optimal salt concentration, and each phage-buffer-container system must be characterized empirically.

Perhaps the most practically important message of the work is that excipient effects are context-dependent and phage-specific. A tailed, contractile Myoviridae-like phage with a large, complex capsid responds differently to stress than a small, robust Podoviridae-like particle. The study therefore advocates a structured development workflow: characterize the phage’s isoelectric point and hydrodynamic behavior, screen a panel of excipient candidates across relevant pH ranges, stress-test the leading candidates with accelerated temperature challenges, and validate the winners under real storage conditions in the intended container closure system. Analytical tools discussed in the article include plaque assays for infectivity, dynamic light scattering and nanoparticle tracking for aggregation, electron microscopy for morphological integrity, and differential scanning fluorimetry or similar methods for probing thermal stability. Combining these orthogonal readouts, rather than relying on any single metric, gives formulators the confidence that a liquid product will remain potent throughout its shelf life.

The implications for the phage therapy field are considerable. Liquid formulations stored at standard refrigeration temperatures, or even at room temperature in optimized compositions, would remove one of the most burdensome logistics constraints on clinical deployment, particularly in regions where deep-freeze distribution is unreliable. Standardized, excipient-protected formulations would also simplify regulatory pathways, since regulators can evaluate excipient safety profiles that are already well established for other biologics. As phage products progress from bespoke compassionate-use preparations toward licensed medicines, the kind of systematic formulation science presented in this study will be indispensable. It transforms the stabilization of phages from a craft of trial and error into a rational engineering discipline, giving the next generation of antiviral antibacterials a fighting chance to arrive intact, potent and ready at the patient’s bedside.

Subject of Research: Use of pharmaceutical excipients to stabilize bacteriophages in liquid formulations by preventing surface adsorption, aggregation and structural degradation.

Article Title: Investigating the role of excipients in mitigating surface adsorption, aggregation and structural degradation in liquid phage formulations

Article References: Siafakas, E., Duong, H. T. T., & Iredell, J. R. (2026). Investigating the role of excipients in mitigating surface adsorption, aggregation and structural degradation in liquid phage formulations. npj Viruses. https://doi.org/10.1038/s44298-026-00233-1

Image Credits: AI Generated

DOI: 10.1038/s44298-026-00233-1

Keywords: bacteriophage, phage therapy, liquid formulation, excipients, trehalose, polysorbate, surface adsorption, aggregation, capsid stability, biologics formulation, antibiotic resistance, npj Viruses

The research, which appears under the title Performance of large language models as a source of clinical information on bacteriophage therapy, was motivated by a practical reality. Physicians considering phage therapy for a patient with a drug-resistant infection often cannot consult a colleague with phage expertise, and formal clinical guidance remains limited. Large language models promise instant, fluent answers to complex medical questions, and surveys suggest that both clinicians and patients increasingly turn to such tools for health information. Whether those answers are accurate, complete, and safe in a niche therapeutic domain like phage therapy had not been systematically assessed, leaving a gap between the enthusiasm for AI-assisted medicine and the evidence needed to support it.

The logic of the evaluation reflects how these models actually work. Large language models are trained on vast corpora of text and generate responses by predicting likely continuations of a prompt rather than by retrieving verified facts from a database. This architecture produces fluent, confident prose regardless of whether the underlying information is correct, a phenomenon often described as hallucination. In a specialized field such as phage therapy, where the training data may be thinner and more heterogeneous than in mainstream medicine, the risk of confident but inaccurate statements is a central concern. The study therefore set out to measure not just whether the models could talk about phage therapy, but whether what they said could be trusted at the bedside.

Phage therapy presents particular challenges for such an assessment. Unlike antibiotics, which are standardized pharmaceutical products, therapeutic phage preparations are typically tailored to the bacterial strain infecting an individual patient. The process involves phage selection, susceptibility testing, formulation, dosing, and monitoring for outcomes that range from bacterial clearance to immune reactions. Clinical evidence includes case reports, small cohort studies, compassionate-use programs, and a limited number of randomized controlled trials, each with different methodological rigor. An information source that conflates experimental findings with established practice, or that presents anecdotal successes as generalizable results, could mislead clinicians in consequential ways.

The evaluation framework used in the study mirrors the standards applied to other emerging medical information tools. Responses generated by the models were assessed for factual accuracy against the primary literature, for completeness in covering the essential elements of a clinical question, for internal consistency, and for the presence of appropriate caveats and safety information. Questions posed to the models spanned the practical spectrum of phage therapy: indications for use, the process of matching phages to bacterial pathogens, dosing and route of administration, known adverse effects, interactions with antibiotics, regulatory status, and the strength of the clinical evidence base. This breadth matters because a model might perform well on general background questions while failing on the specific, operational details that determine whether a therapy is used correctly.

The results highlight a pattern that has emerged across evaluations of AI in medicine. Large language models generally perform well on questions with abundant, well-established answers in the training data. Basic descriptions of what bacteriophages are, how they kill bacteria, and why they are being reconsidered in the era of antimicrobial resistance tend to be accurate and clearly expressed. The models are also effective at summarizing the general rationale for phage therapy and at explaining concepts such as phage specificity and the importance of susceptibility testing. For a clinician seeking orientation in an unfamiliar field, this level of performance can be genuinely useful, providing a readable entry point that would once have required hours of literature searching.

Performance degrades, however, as questions move from general principles to specific clinical detail. The study found that models can produce answers that are partially correct but incomplete, omitting critical caveats such as the experimental status of many phage therapy protocols or the limited availability of approved phage products in most jurisdictions. Some responses blended established facts with outdated or unsupported claims, presenting them with equal confidence. In a domain where treatment decisions depend on precise, current information about phage-bacterium matching and evolving regulatory frameworks, such subtle inaccuracies are not trivial. A response that is ninety percent correct can still be clinically dangerous if the incorrect ten percent concerns dosing, safety, or the evidence supporting a therapeutic claim.

Another dimension of the evaluation concerns how the models communicate uncertainty. Trustworthy medical information sources distinguish clearly between what is proven, what is plausible, and what is speculative. The study indicates that large language models vary considerably in this respect, sometimes providing appropriate disclaimers about the experimental nature of phage therapy and sometimes presenting contested or preliminary findings as settled. This variability is itself informative, because it suggests that clinicians cannot assume a consistent standard of epistemic caution across different questions or different models. The fluency of AI-generated text can mask this inconsistency, making careful verification more important, not less.

The implications extend beyond phage therapy to the broader integration of artificial intelligence into clinical practice. The study’s authors frame their work as a caution against treating chatbots as authoritative references, particularly in specialized and rapidly evolving fields. At the same time, the findings do not support dismissing these tools outright. Used as a starting point for literature exploration, a drafting aid, or a way to formulate better questions for specialists, large language models can add real value. The critical requirement is human oversight: clinicians with domain knowledge must remain in the loop, verifying AI-generated claims against primary sources before any of that information influences patient care. This is the same standard applied to other secondary sources of medical information, and the study argues it should apply with equal force to AI.

The research also points toward what would be needed for large language models to become genuinely reliable clinical resources. Improvements are likely to come from several directions: grounding model responses in curated, up-to-date medical databases rather than relying solely on static training data; developing domain-specific evaluations that test models against expert-validated question sets; and building transparency features that allow users to trace claims back to their sources. For phage therapy specifically, a field whose evidence base is growing quickly as new trials are completed, the ability to incorporate current literature is essential. Until such systems mature, the study’s central message stands: large language models can be informative conversational partners on phage therapy, but their outputs should be regarded as provisional drafts of knowledge, subject to expert review, rather than as substitutes for the primary literature and clinical judgment on which safe patient care ultimately depends.

Subject of Research: Evaluation of large language models as sources of clinical information on bacteriophage therapy

Article Title: Performance of large language models as a source of clinical information on bacteriophage therapy

Article References: Walter, N., Amanatullah, D. F., Debarbieux, L., Doub, J. B., Ferry, T., Groß, J., Międzybrodzki, R., Mirzaei, M. K., Deng, L., Rācenis, K., Suh, G. A., Que, Y.-A., Górski, A., & Rupp, M. (2026). Performance of large language models as a source of clinical information on bacteriophage therapy. npj Viruses, 4(1), Article 41. https://doi.org/10.1038/s44298-026-00224-2

Image Credits: AI Generated

DOI: 10.1038/s44298-026-00224-2

Keywords: bacteriophage therapy, large language models, artificial intelligence, antimicrobial resistance, clinical information, medical AI, npj Viruses, hallucination, infectious diseases, phage selection, clinical decision support, evidence quality