The end of antibiotics, again: phage therapy, bacteriocins, and precision antimicrobials

The “end of antibiotics” is a phrase that keeps returning because it captures a real sensation: the ground moving under medicine’s feet. Infections that were once predictable are becoming less so, and in some settings, clinicians are again forced into older, harsher drugs or into combinations that feel like educated guesswork.

Antimicrobial resistance (AMR) sits behind the anxiety. The World Health Organization (WHO) describes it as a major global threat, noting estimates that bacterial AMR was directly responsible for 1.27 million deaths in 2019 and contributed to 4.95 million. A large Lancet analysis published in 2024 projected that, without stronger action, deaths linked to bacterial AMR will rise sharply between 2025 and 2050.

Yet the “replacement antibiotic” pipeline has not arrived in force. That is not a failure of imagination. It is, in large part, an economic design flaw.

AMR in plain English: evolution, but with receipts

Bacteria reproduce fast and share genetic tricks. When an antibiotic kills susceptible bacteria, any microbes with resistance mutations or resistance genes are more likely to survive and multiply. Repeat this across hospitals, farms, water systems and communities, and resistance becomes less a rare event and more a statistical certainty.

Misuse and overuse accelerate the process. WHO highlights inappropriate use in humans, animals and plants as a key driver. Add weak infection prevention and control, poor sanitation, and patchy access to diagnostics, and the evolutionary pressure intensifies.

A crucial subtlety is that resistance is not only “superbugs” in intensive care. It is also everyday infections that become slower to treat, require more expensive drugs, and carry higher risks.

Why incentives are broken: the market punishes good antibiotics

If you invent a life-saving cancer drug, you are rewarded when it is used widely. If you invent a truly valuable antibiotic, public health asks you to hold it back. Stewardship demands that new drugs are reserved for the sickest patients and the hardest infections, precisely to slow resistance.

This creates the classic antibiotic market failure: companies face high research costs but low and unpredictable revenues, especially when responsible use is a success. Health economists and policy analysts have argued for “delinked” pull incentives, where payment is based on value rather than volume.

The UK has been a global test case. The National Health Service (NHS) has trialled a subscription-style payment model that pays a fixed annual fee based largely on assessed value, not on how many doses are sold. This is an attempt to make “use less” compatible with “invest more”.

Precision antimicrobials: narrow spectrum as feature and bug

Traditional antibiotics are often broad spectrum. They hit the pathogen, but also collateral damage. The gut microbiome is disrupted; opportunistic infections can follow; resistance genes get a wider training ground.

Next-generation antimicrobials aim for something closer to a guided missile: narrow-spectrum agents that target a specific species, strain, or virulence factor. The pitch is compelling: treat the infection without flattening the ecosystem.

The bug is also obvious. If a therapy is narrow, you need to know what you are treating. That means rapid diagnostics, tighter clinical pathways, and trials designed around a moving target.

Bacteriophages: viruses that hunt bacteria

A bacteriophage (phage) is a virus that infects bacteria. Phages attach to a bacterial surface, inject genetic material, and replicate. Many phages burst the host cell, releasing more phage particles.

Phage therapy is often described as old science becoming new again. It has a long history in parts of Eastern Europe, and renewed interest elsewhere as resistance rises. Reviews in 2024 and 2025 describe an expanding landscape of compassionate-use cases and structured clinical trials, alongside persistent challenges in standardisation and regulation.

Why phages are attractive
Phages can be exquisitely specific. That specificity can spare bystander microbes and reduce some side effects linked to broad-spectrum antibiotics.

Why phages are difficult
Specificity means you often need the right phage for the right strain. Bacteria also evolve resistance to phages, so many approaches use phage cocktails, mixtures designed to broaden coverage and reduce escape. Resistance evolution does not disappear, it changes shape.

Clinical evidence is still patchy. The PhagoBurn trial, a randomised controlled phase 1/2 study in burn wound infections caused by Pseudomonas aeruginosa, is often cited as a cautionary tale. It ran into practical issues including phage preparation stability and dosing, underlining how manufacturing quality can determine clinical feasibility.

More recently, cystic fibrosis has emerged as a testing ground, including trials of inhaled phage products for chronic Pseudomonas aeruginosa infection.

Manufacturing and regulation: living products that can change

Phages are biological entities, and they can evolve. Regulators therefore care about identity, potency, purity, stability, and consistency.

In October 2025, the European Medicines Agency (EMA) released a draft guideline focused on quality documentation for phage therapy medicinal products, addressing manufacture, characterisation, specifications and stability. The US Food and Drug Administration (FDA) has also discussed the difficulty of standardising phage stocks and ensuring quality control, given viability and potential mutation over time.

This is the unglamorous core. If you cannot manufacture reliably, you cannot run decisive trials. If you cannot run decisive trials, reimbursement becomes guesswork.

Bacteriocins: bacteria’s own targeted weapons

Bacteriocins are antimicrobial compounds produced by bacteria, often to compete with closely related strains. Many have narrow activity spectra, which makes them interesting as microbiome-sparing agents.

In late 2025, researchers described a genomics-guided atlas of class II bacteriocins, framing unmodified bacteriocins as potential precision antimicrobials that could spare bystander microbes. This kind of mapping work matters because it turns bacteriocins from a scattered literature into a searchable repertoire.

The challenges resemble phages in miniature: delivery to the right place, stability, and the possibility that bacteria evolve around the attack. Bacteriocins also face a translation problem. Potent activity in a dish does not guarantee useful behaviour in a human body, where proteins can be degraded, diluted or neutralised.

Antimicrobial peptides: nature’s templates, clinical headaches

Antimicrobial peptides (AMPs) are short protein-like molecules found across life, part of innate immune defences. They can disrupt membranes or interfere with microbial functions, and they are being explored as templates for new drugs.

Reviews continue to describe AMPs as promising against multidrug-resistant pathogens, while stressing well-known barriers: toxicity, stability, bioavailability, and cost of manufacture. Their potency can come with a price, since anything that disrupts membranes can end up disturbing human cells too, depending on dose and formulation.

Some of the more realistic near-term uses may be topical, local or device-associated, where high local concentrations are possible and systemic exposure is limited.

Microbiome-sparing approaches: treat the pathogen, keep the ecosystem

Precision antimicrobials also include strategies designed to avoid broad collateral damage: agents that target virulence factors rather than bacterial survival, narrow-spectrum antibiotics paired with diagnostics, or approaches that maintain microbiome resilience during treatment.

The common thread is the same: the therapy is only as good as your ability to identify the right patient at the right time.

Trials: why the evidence is hard to generate

Precision changes trial design.

With broad-spectrum antibiotics, you can often enrol “suspected infection” and adjust later. With phages or narrow bacteriocins, you may need confirmed pathogen identification, strain typing, and rapid turnaround.

Manufacturing adds another complication. If phage products are personalised or frequently updated, what exactly is the “same” product for a trial? Regulators are beginning to clarify quality expectations, but the field is still building shared frameworks.

Endpoints are also tricky. For stubborn infections, patients are often on multiple antibiotics, have complex comorbidities, and have fluctuating bacterial loads. Demonstrating additive benefit is possible, but not trivial.

Reimbursement: paying for readiness, not volume

Even if a precision antimicrobial works, the economic problem remains. The best tool is the one you rarely use.

The UK subscription model is an attempt to pay for availability and value, not sales. A similar logic may be required for some next-generation antimicrobials, especially those designed for resistant infections that, with good stewardship, remain relatively uncommon.

The political challenge is that you are paying for a fire extinguisher before the smoke appears.

Stewardship and prevention: the practical section

Precision antimicrobials do not replace prevention. They increase the value of using existing drugs carefully.

Practical priorities remain stubbornly familiar:

• Vaccination: fewer infections means fewer antibiotic courses, which reduces selection pressure. WHO emphasises prevention as part of AMR control.
• Infection prevention and control: hand hygiene, clean water, isolation protocols, and safe surgical practice reduce spread.
• Diagnostics: rapid tests that distinguish viral from bacterial infections, and identify pathogens quickly, make narrow-spectrum treatment feasible.
• Antimicrobial stewardship: prescribing antibiotics only when needed, choosing the right drug, dose and duration, and avoiding “just in case” courses. The UK Health Security Agency (UKHSA) tracks usage and resistance through ESPAUR, reinforcing how surveillance supports stewardship.
• One Health action: antibiotic use in agriculture and environmental contamination contribute to resistance pressure, so solutions must span sectors.

What success would look like

What success would look like

• Rapid diagnostics in routine care, so narrow therapies are practical rather than aspirational.
• A handful of validated phage or bacteriocin products with consistent manufacturing and clear indications.
• Trials that demonstrate patient-centred outcomes, not only bacterial counts.
• Resistance management plans baked into product development, including cocktail updates and stewardship protocols.
• Reimbursement models that reward readiness and value, not volume.
• Surveillance systems that detect resistance shifts early, so treatment strategies adapt before crises.

Glossary (8 terms)

• Antimicrobial resistance (AMR): when microbes evolve to survive drugs designed to kill them.
• Bacteriophage (phage): a virus that infects bacteria.
• Phage cocktail: a mixture of phages intended to broaden coverage and reduce bacterial escape.
• Bacteriocin: an antimicrobial compound produced by bacteria, often active against closely related strains.
• Antimicrobial peptide (AMP): a short peptide with antimicrobial activity, often part of innate immunity.
• Stewardship: policies and practices that reduce unnecessary antimicrobial use.
• Narrow spectrum: activity against a limited range of microbes.
• One Health: a framework linking human, animal and environmental health in addressing AMR.

Realistic timelines: steady progress, uneven adoption

Phage therapy is moving from case reports towards more formal programmes, and regulators are actively clarifying quality expectations. Bacteriocins and AMPs are expanding in discovery and engineering, but many remain in earlier stages, where translation problems tend to appear.

The most realistic near-term change is not a sudden end to antibiotics. It is a gradual expansion of the antimicrobial toolkit, combined with better diagnostics and payment models that stop treating antibiotics as ordinary commodities. The end, if it comes, will be to the idea that one broad-spectrum pill can carry modern medicine indefinitely.

The weird world of senescence: ‘zombie’ cells, anti-ageing hype, and what the science actually says

Standfirst (40 words)
Senescent cells are damaged cells that stop dividing but refuse to go quietly, releasing inflammatory signals that can disrupt tissues. Drugs that clear them, senolytics, have transformed ageing research in mice. In humans, evidence is emerging, mixed, and not yet a shortcut to youth.

A senescent cell is sometimes called a zombie cell, which is unfair to zombies and flattering to marketing. Senescence is a stress response. A cell that has suffered DNA damage, telomere shortening, oxidative stress, or other insults can enter a stable growth arrest, essentially stepping out of the division cycle. It does this to prevent cancer-like proliferation. That part is protective.

The strange bit is what happens next. Many senescent cells develop a loud chemical voice, secreting inflammatory and tissue-remodelling molecules known as the senescence-associated secretory phenotype (SASP). In the right context, SASP can help with wound healing and tissue repair. In the wrong context, chronic SASP can drive inflammation, fibrosis, and dysfunction.

This is why senescence sits at the heart of ageing debates. It is not simply “bad cells”. It is a trade-off that, over time, can become maladaptive.

What senescent cells are, and why they accumulate

Senescent cells accumulate with age partly because damage accumulates, and partly because immune clearance may decline. They can appear in many tissues, including fat, blood vessels, lung, and joints, particularly under chronic stress and disease conditions. Reviews describe senescence as a hallmark-like process linked to diverse age-associated diseases.

Senescent cells also differ by tissue and trigger. Some are transient and useful. Others persist.

The SASP adds complexity. A major 2024 review in Nature Reviews Molecular Cell Biology describes the SASP as heterogeneous and dynamic, with both beneficial and detrimental roles depending on context, and discusses its use as a biomarker and therapeutic target.

Senolytics: the idea, and the unease

Senolytics are drugs designed to selectively eliminate senescent cells by targeting survival pathways that these cells rely on. The appeal is direct: reduce SASP-driven inflammation and tissue disruption by clearing the source.

The unease is equally direct: senescence also has physiological roles, including wound healing and tumour suppression. Some reviews emphasise this duality, warning that indiscriminate clearance could disrupt beneficial processes.

A separate class, senomorphics, aims to dampen the SASP without killing cells, shifting the risk profile but raising its own questions about long-term effects.

Animals versus humans: where the evidence stands

In mice, senolytics have repeatedly improved measures of physical function and extended lifespan in certain settings. That does not guarantee human benefits, partly because mouse lifespans are short, and partly because human ageing is not one disease.

In humans, the evidence is early and fragmented, but it exists.

A 2019 pilot clinical trial in individuals with diabetic kidney disease reported that a combination of dasatinib and quercetin reduced markers of senescent cell burden in adipose tissue and skin. Another 2019 first-in-human pilot in idiopathic pulmonary fibrosis suggested feasibility and provided preliminary signals that senolytics might alleviate physical dysfunction, while strongly implying that larger randomised trials are needed.

These studies are not definitive. They are proof that the biology can be probed in people, and that measuring senescence-related outcomes is possible, though still technically challenging.

The measurement problem: you cannot treat what you cannot track

A recurring barrier is that senescence is not a single marker. It is a state. Biomarkers include cell-cycle inhibitors, DNA damage signatures, and SASP components, but these can vary by tissue and context.

The SASP review literature increasingly treats measurement as central, not secondary, including the need for accessible biomarkers and careful interpretation of inflammatory signals.

This matters for hype. If a therapy claims to “reduce biological age” without credible, validated measures tied to health outcomes, it is usually selling a story, not evidence.

Risks: cancer, wound healing, and immune effects

Senolytics are not inherently gentle. Many candidates target anti-apoptotic pathways, and some originate from oncology contexts.

Risks include:

• Cancer dynamics: senescence helps suppress tumour formation by halting cell division. Clearing senescent cells might reduce chronic inflammation that can promote cancer, but could also interfere with tumour-suppressive processes, depending on context and timing. Reviews stress that roles differ across tissues and disease states.
• Wound healing: senescent cells can be involved in repair processes, and senescence has been described as having roles in wound repair. The question is not whether senescence is “good” or “bad”, it is when and where.
• Immune effects: senescence and immune signalling are intertwined, and interventions that shift inflammatory states can have unpredictable effects, particularly in older populations.

This is why serious clinical programmes focus first on clear disease indications, not on selling senolytics as lifestyle upgrades.

The regulatory and commercial landscape: ageing is not an approved indication

Regulators approve drugs for diseases, not for the abstract concept of ageing. That forces developers to pick specific indications where senescence is plausibly causal and measurable, such as fibrotic lung disease, metabolic disease complications, eye disease, or frailty-related syndromes.

A 2024 review in Endocrine Reviews notes that more than 30 clinical trials of senolytic and senomorphic agents have been completed, are underway, or planned across indications, highlighting both momentum and the early-stage nature of much of the evidence.

Commercially, the field has seen both excitement and retrenchment. UNITY Biotechnology, an early high-profile company in senolytics, published a sham-controlled trial of its senolytic UBX1325 in diabetic macular oedema, reporting acceptable short-term safety and trends suggestive of efficacy, while emphasising the need for larger trials. It also later cut staff heavily while exploring strategic alternatives, illustrating how difficult translation can be even with strong narratives.

The pattern is familiar in biotech. A plausible mechanism is not a product.

Hype audit: what senolytics can claim, and what they cannot

Hype audit

• Claim you can take seriously: senescent cells and SASP contribute to multiple diseases and can be targeted in principle.
• Claim that needs careful framing: clearing senescent cells may improve specific functional outcomes in specific diseases; early human pilots support feasibility, not certainty.
• Claim to treat with suspicion: “reverse ageing” in healthy people, especially if based on unvalidated biomarkers or short-term studies.
• Claim that is usually marketing: supplement-driven senolysis as a proven route to longer life.

How to read a longevity claim (12 questions)

1. Is the claim about lifespan or healthspan, and is that defined in measurable terms?
2. Is the evidence from mice, other animals, or humans?
3. If human, is it a randomised controlled trial (RCT) or a small pilot?
4. What is the endpoint: a clinical outcome, functional test, or a biomarker proxy?
5. If biomarkers, are they validated and linked to meaningful health outcomes? The SASP literature stresses heterogeneity and context.
6. Which tissue is affected, and does that map to the claimed benefit?
7. What is the proposed mechanism: senolytic clearance or SASP suppression?
8. What is known about off-target effects, especially for drugs with oncology origins?
9. How long is follow-up, and are delayed effects plausible?
10. Is there evidence of harm to wound repair, immune function, or tissue regeneration pathways?
11. Who funded and ran the study, and is the protocol and analysis transparent?
12. Does the claim generalise from a disease population to healthy ageing without evidence?

What the science actually says, in one sentence

Senescence is a real biological process with both protective and harmful roles, senolytics are a plausible therapeutic strategy, and human evidence is emerging but not yet decisive.

Realistic timelines: what to expect over the next decade

Over the next five years, the most credible progress is likely in targeted indications where senescence-related biology is strong and endpoints are measurable, including fibrotic diseases, metabolic complications, and some eye conditions.

Over the next decade, two things will determine whether senolytics become mainstream therapeutics:

• Measurement: robust, accessible biomarkers and functional endpoints that show benefit beyond placebo and beyond short-term inflammation shifts.
• Selectivity and safety: drugs that clear the “harmful” senescent cells without destabilising repair, immunity, or tumour suppression dynamics.

If those pieces fall into place, senolytics may become part of ordinary medicine for specific diseases, not a universal anti-ageing reset. If they do not, the field will still have delivered something valuable: a sharper understanding of how ageing biology becomes disease, and how hard it is to turn that knowledge into safe, scalable treatment.