Ewingella americana to Treat Cancer

Evidence Review created on 10/17/2026 using AI4L / Opus 5.5

Also known as: E. americana, Enteric Group 40

Motivation

Ewingella americana is a rod-shaped bacterium first described in 1983 from human clinical samples. For four decades it was known mainly as an uncommon opportunistic pathogen in hospital patients and as a spoilage organism on cultivated mushrooms. Interest shifted when researchers in Japan isolated a strain of it from the gut of the Japanese tree frog and tested it as a living cancer treatment given intravenously, on the premise that certain bacteria settle inside solid tumors, where oxygen is scarce and immune surveillance is weak.

Using live bacteria against tumors is not new. Physicians injected bacterial preparations into inoperable tumors in the late nineteenth century, and weakened Salmonella and Clostridium strains have since been carried into early human trials. What is new here is the use of an unmodified organism taken from a wild amphibian rather than a genetically engineered laboratory strain.

This review examines what has actually been measured for Ewingella americana as a cancer treatment, what the same organism is known to do as a human pathogen, and how far that evidence reaches toward use in people who monitor cancer risk closely.

Benefits - Risks - Protocol - Conclusion

Grokipedia

Ewingella americana

Covers the organism end to end — taxonomy, mushroom pathogenicity, human case series, antibiotic resistance patterns — and carries a dedicated antitumor-applications section, making it the fastest orientation to both sides of this organism.

Examine

No Examine article exists for Ewingella americana; the site covers supplements, nutrients and dietary compounds rather than live bacterial species under investigation as cancer therapies.

ConsumerLab

No ConsumerLab article exists for Ewingella americana; ConsumerLab tests consumer supplement and food products for identity, potency and contamination, and this organism is not sold as one.

Systematic Reviews

No systematic reviews or meta-analyses for Ewingella americana were found on PubMed as of October 11, 2026.

Neither side of the trade-off is represented: no systematic review or meta-analysis covers the claimed antitumor effect, and none covers the principal risk, invasive infection.

Mechanism of Action

Ewingella americana is a facultative anaerobe (it grows with or without oxygen), and that property drives the proposed mechanism. Injected intravenously, it clears from blood within a day but expands inside tumors, whose cores are hypoxic (oxygen-starved) and locally immunosuppressed. Intratumoral counts rose roughly three-thousand-fold between three and twenty-four hours, with no colonies recoverable from lung, liver, spleen, kidney or heart (Iwata et al., 2025).

Two effector arms follow. The bacterium secretes pore-forming toxins — hemolysin and an exotoxin — that lyse tumor cells directly, which is what destroyed tumor-cell clusters in culture. In animals, its cell-wall molecules (pathogen-associated patterns, the signatures innate immune receptors such as Toll-like and NOD-like receptors read as “microbe”) recruited neutrophils, T cells and B cells within six hours, with rising interferon-gamma and tumor necrosis factor-alpha (inflammatory signaling proteins) and widespread apoptosis (programmed cell death) (Iwata et al., 2025).

Which arm matters more is unsettled: the authors emphasize immune recruitment, since neutrophils rose most, while the culture work shows direct lysis suffices without immunity. Tumor selectivity is also contested — low oxygen alone cannot explain it, since intestinal crypts and kidney medulla are hypoxic yet never colonized, so the authors invoke additional tumor features including “don’t-eat-me” signaling, necrosis (tissue death) and leaky vessels.

There is no drug-metabolizing pathway to describe: distribution is tumor-restricted and clearance is immune-mediated rather than hepatic or renal.

Historical Context & Evolution

The organism’s own history is unremarkable. Described in 1983 from ten American clinical isolates, previously catalogued only as “enteric group 40”, named for the bacteriologist William H. Ewing (Ioannou et al., 2024). For four decades it appeared as two separate things: a cause of browning and stem necrosis on cultivated mushrooms (Reyes et al., 2004), and an occasional hospital isolate whose status as a true pathogen was debated.

Treating cancer with live bacteria has a longer history. Nineteenth-century surgeons injected bacterial preparations into inoperable tumors after noticing regressions following skin infections; the survivor in oncology is BCG (bacille Calmette-Guérin, a weakened cattle tuberculosis bacterium), still licensed for bladder carcinoma in situ (cancer confined to the surface layer) (U.S. Food and Drug Administration, TICE BCG). Modern attempts engineered known pathogens, lately pairing bacteria with immune-checkpoint inhibitors (drugs that release the brakes tumors place on immune cells, such as pembrolizumab and anti-PD-L1 antibodies). Intravenous attenuated Salmonella typhimurium reached patients in 2002: tumor colonization occurred in three of twenty-five, dose-limiting toxicity (side effects severe enough to cap the dose) appeared at the top dose, and no tumor shrank (Toso et al., 2002). Clostridium novyi-NT spores injected into tumors lysed ten of twenty-four masses but caused three life-threatening complications in large lesions (Janku et al., 2021).

What changed in 2025 was the search strategy, not the biology: instead of engineering a pathogen, one laboratory screened wild-animal gut bacteria for strains already tumor-homing and tolerated. Whether that reframing survives human testing is open.

Expected Benefits

High 🟩 🟩 🟩

No benefit reaches High: there is no human outcome evidence of any kind — the efficacy record consists of mouse tumor experiments and cell-culture assays from a single laboratory.

Medium 🟩 🟩

No benefit reaches Medium either: no human trial, controlled or not, and no comparison-group observational human data exist, so even single-trial human evidence is absent.

Low 🟩

Speculative 🟨

Complete Tumor Regression in a Mouse Colorectal Model

One intravenous dose cleared established tumors in every treated mouse and extended survival. It outperformed an immune-checkpoint antibody and liposomal chemotherapy in the same experiment. No human data exist (Iwata et al., 2025).

Durable Resistance to Tumor Rechallenge in Mice

Cured mice rejected re-implanted tumor cells while naive controls all grew tumors, indicating immune memory past sixty days. The basis is one rechallenge experiment in a single mouse strain (Iwata et al., 2025).

Direct Killing of Tumor Cell Clusters in Culture

Three-dimensional tumor-cell clusters were largely destroyed within a day at the highest bacterial concentration, with killing attributed to secreted pore-forming toxins. Cell-culture cytotoxicity cannot predict clinical benefit (Iwata et al., 2025).

Benefit-Modifying Factors

Genetic variation in innate immune sensing: Response depends on reading bacterial cell-wall signals, so loss-of-function variants in TLR4 (a receptor that detects bacterial surface molecules) or NOD2 (a receptor that detects bacterial cell-wall fragments) would plausibly blunt the immune arm. Untested for this organism.
Iron availability: The only live-bacterium trial dosing intravenously supplies desferrioxamine to provide usable iron, because the bacteria need it to grow. Low iron stores would plausibly limit intratumoral expansion; iron overload is excluded in that protocol (NCT05120596).
Tumor hypoxia and size: Selective colonization depends on an oxygen-poor tumor core, so small, well-oxygenated or diffusely infiltrating tumors would plausibly accumulate fewer bacteria. The mouse experiments began only once tumors reached about 200 cubic millimetres (Iwata et al., 2025).
Baseline neutrophil count: Neutrophils were the immune population that rose most after dosing, so chemotherapy-induced neutropenia (a low neutrophil count) would plausibly reduce benefit while raising infection risk — the two move in opposite directions (Iwata et al., 2025).
Immunosuppressive medication and conditions: The efficacy experiments used mice with intact immune systems. Transplant immunosuppression, systemic corticosteroids and acquired immunodeficiency are excluded from live-bacterium trials for safety, and would also remove the immune half of the effect (NCT05120596).
Sex: No sex-stratified efficacy data exist, in animals or people. The human infection series ran 47 percent male with no sex difference in outcome reported, which says nothing about therapeutic response (Ioannou et al., 2024).
Age: Immune ageing would be expected to weaken the recruitment arm at the older end of the target range, and the one intravenous live-bacterium trial caps enrolment at 74 years (NCT05120596). No age-stratified data exist.

Potential Risks & Side Effects

High 🟥 🟥 🟥

No risk reaches High: no trial of Ewingella americana as a therapy exists, so there are no adverse-event data from independent research groups — only unconnected case reports of naturally acquired infection.

Medium 🟥 🟥

No risk reaches Medium: no single trial and no comparison-group observational study has measured harm from this organism; the case literature has no control group.

Low 🟥

Invasive Infection, Including Bloodstream Infection ⚠️ Conflicted

Reported human infections involve the bloodstream, peritoneum and lung, usually in immunocompromised or catheterized patients; one was fatal. All were naturally acquired, not therapeutically dosed — a route mismatch. Mouse dosing produced no toxicity, so the evidence conflicts. Net: the organism infects vulnerable patients; therapeutic dosing is untested.

Magnitude: Nineteen human infections across sixteen published reports, with one death from sepsis (a life-threatening whole-body reaction to infection) following adrenal haemorrhage in a previously healthy 74-year-old; no control group (Ioannou et al., 2024; Tsokos, 2003). Against this, mouse safety testing at the therapeutic dose reported no significant difference from control animals in any blood count, liver, kidney or metabolic parameter, and no organ damage on histology (Iwata et al., 2025).

Multidrug-Resistant Isolates That Narrow Antibiotic Rescue

The species carries a chromosomal AmpC beta-lactamase (an enzyme that destroys many penicillins and cephalosporins) and is naturally resistant to several antibiotic classes. One reported isolate resisted everything tested except three agents. Resistance data come from clinical and environmental isolates, not a therapeutic strain — a strain mismatch.

Magnitude: All twenty strains tested produced a beta-lactamase, class C in the representative strains examined, and were naturally resistant or only intermediately susceptible to many agents, including cefaclor, cefazolin, cefuroxime, cefoxitin, benzylpenicillin, oxacillin, fosfomycin, erythromycin, clarithromycin, linezolid and rifampicin; no control group (Stock et al., 2003). In one case report the isolate was susceptible only to trimethoprim–sulfamethoxazole, ticarcillin–clavulanate and cefotetan (Pound et al., 2007).

Speculative 🟨

Dose-Dependent Lethality Above the Mouse Tolerated Dose

Dose-finding identified the therapeutic dose as also the maximum tolerated dose; doses above it were not tolerated, leaving a narrow window. Basis is murine dose-ranging only, with no human equivalent (Iwata et al., 2025).

Transient Systemic Inflammatory Response After Dosing

Treated mice lost weight briefly and mounted a mild inflammatory response that resolved within ten days. Whether this escalates to sepsis in people is unknown; basis is animal data (Iwata et al., 2025).

Seeding of Prosthetic Implants and Indwelling Devices

Live bacteria in the bloodstream can colonize valves, joints and catheters, which is why the engineered-bacterium trial excludes such patients. No case of this with therapeutic dosing exists; basis is mechanism and trial design (NCT05120596).

Secreted Toxins of Unmeasured Relevance to Mammals

Beyond pore-forming toxins, it releases volatile compounds that damage mushroom tissue, and its genome encodes secretion systems, virulence effectors and toxins. Neither has been assayed in mammals (Wei et al., 2025; Liu et al., 2020).

Risk-Modifying Factors

Genetic variation in immune defence: Variants reducing complement (blood proteins that puncture bacteria), mannose-binding lectin (a protein that tags sugars on microbes) or NOD2 signalling impair clearance of gram-negative bacteria. None is linked to Ewingella americana specifically.
Baseline biomarker levels: No biomarker predicts invasive infection. Procalcitonin (a blood marker that rises in bacterial infection) tracked severity: below 0.12 nanograms per millilitre in a mild case, 113 in the fatal one (Esposito et al., 2019; Tsokos, 2003).
Pre-existing conditions: Most severe reported cases involved malignancy, chemotherapy, kidney failure on dialysis, an indwelling catheter or the neonatal period, though the one fatal case struck a previously healthy 74-year-old (Tsokos, 2003).
Sex: No sex difference in risk has been reported; the pooled case series was 47 percent male, too small to detect one (Ioannou et al., 2024).
Age: The two most severe documented outcomes sit at the extremes — meningitis (infection of the brain’s lining) in a term newborn and the one fatal case at 74 (Meisler et al., 2020; Tsokos, 2003).

Key Interactions & Contraindications

Systemic antibiotics (ceftriaxone, ceftazidime, meropenem, ciprofloxacin): Avoid during dosing (theoretical) — these are the organism’s effective treatments, so concurrent use would eradicate the therapeutic bacteria and abolish the effect. The engineered-bacterium protocol requires five antibiotic half-lives to elapse first (NCT05120596).
Systemic corticosteroids and other immunosuppressants (drugs that damp down the immune system: prednisone, tacrolimus, mycophenolate, rituximab): Avoid (theoretical) — removes the immune arm while raising the chance a bloodstream bacterium disseminates. Any systemic corticosteroid dose is an exclusion (NCT05120596).
Immune-checkpoint inhibitors (pembrolizumab, nivolumab, atezolizumab): Potentially additive (theoretical) — the mechanisms differ, one releasing a brake on existing T cells and the other starting a new inflammatory response. Both live-bacterium programmes pair their organism with pembrolizumab (NCT03435952).
Iron chelators and iron salts (drugs that bind iron: desferrioxamine, ferrous sulfate): Monitor (theoretical) — iron availability governs bacterial growth, so supplying or withholding it would shift both efficacy and infection risk. The intravenous trial supplies desferrioxamine and excludes iron overload (NCT05120596).
Anticoagulants and antiplatelet agents (blood thinners: warfarin, apixaban, aspirin, clopidogrel): Caution (theoretical) — bleeding at an injected tumor site. Therapeutic anticoagulation is excluded unless the injected lesion is superficial and low-risk; lower-dose aspirin or clopidogrel may be permitted at investigator discretion (NCT05120596).
Over-the-counter antipyretics and anti-inflammatories (fever and pain reducers: acetaminophen, ibuprofen, naproxen): Caution (theoretical) — suppressing fever masks the earliest sign that bacteria are circulating rather than confined to the tumor, and the anti-inflammatory effect may blunt the immune arm. Neither has been measured.
Over-the-counter gut and oral antibacterials (bismuth subsalicylate, chlorhexidine mouthwash): No expected interaction (theoretical) — these act in the gut lumen or mouth rather than the bloodstream, and the intravenous route bypasses both compartments entirely.
Supplements with antibacterial activity (berberine, oregano oil, allicin-standardized garlic, high-dose zinc): Caution (theoretical) — systemic antibacterial exposure could reduce the viable bacterial load reaching the tumor. None has been tested against this organism.
Immune-stimulating supplements (beta-glucans, mushroom extracts, Echinacea): Potentially additive (theoretical) — these engage the same innate receptors the bacterium triggers, so inflammatory response could be amplified rather than merely duplicated. This organism is itself a pathogen of cultivated button mushrooms (Agaricus bisporus) (Reyes et al., 2004).
High-dose antioxidant supplements (vitamin C, N-acetylcysteine, alpha-lipoic acid): Caution (theoretical) — part of the killing is oxidative and neutrophil-driven, so large antioxidant doses could blunt it. No study has measured this for any bacterial cancer therapy.
Other interventions, radiotherapy and hyperbaric oxygen: Opposite directions (theoretical) — radiation increases tumor necrosis and hypoxia, favouring colonization, while hyperbaric oxygen removes the hypoxia the organism depends on. Neither combination has been tested.

Populations who should avoid Ewingella americana:

Implanted prostheses that cannot easily be removed — artificial heart valves, joint replacements, arterial grafts — or any history of bacterial endocarditis (infection of the heart lining), regardless of organism (NCT05120596 exclusion criteria)
Prior allogeneic bone-marrow or stem-cell transplant, congenital or acquired immunodeficiency, or any dose of systemic corticosteroid (NCT05120596)
Active uncontrolled infection, or a major infection such as pneumonia within the previous 4 weeks (NCT05120596)
Documented recent infection with, or detectable baseline carriage of, the therapeutic organism (NCT05120596)
Directly injected tumors larger than 8 cm — all three dose-limiting toxicities in the intratumoral spore trial, two life-threatening sepsis events and one gas gangrene (gas-forming tissue death), occurred above that size (Janku et al., 2021)
Age above 74 years, the upper enrolment limit of the one intravenous live-bacterium trial (NCT05120596)
New York Heart Association (NYHA) class II–IV heart failure (symptoms on ordinary activity or worse), major cardiac arrhythmia or ischaemia requiring medication, or coronary angioplasty within 6 months (NCT05120596)
Allergy to the antibiotics that would serve as rescue, or to chloramphenicol (NCT05120596)
Known central nervous system metastases, a bleeding disorder, or high risk of bowel perforation including recent diverticulitis (inflamed pouches in the colon wall) or intra-abdominal abscess (NCT05120596)
End-stage kidney disease on dialysis with an indwelling peritoneal catheter — among the most frequently reported sites of human infection with this organism (Ioannou et al., 2024)

Risk Mitigation Strategies

Parameters below without a citation reflect common practice in early-phase live-bacterial oncology trials, not an established protocol for this organism.

Antibiotic rescue in place before the first dose: The organism remains susceptible to ceftriaxone, ceftazidime and meropenem, the stated safety advantage over engineered strains; an effective agent on hand caps an uncontrolled bloodstream infection (Stock et al., 2003).
Susceptibility testing of the therapeutic lot: One reported human isolate resisted everything tested except three agents, so confirming the specific lot’s susceptibility before dosing prevents a situation with no rescue option (Pound et al., 2007).
Dose fractionation below the single-dose ceiling: Because the effective and maximum tolerated doses coincided in mice, splitting the total load across days is the authors’ own proposal for widening the margin and limiting peak bacterial load (Iwata et al., 2025).
An 8-centimetre ceiling on directly injected tumors: All three dose-limiting toxicities in the intratumoral spore trial — sepsis and gas gangrene — occurred above that size, so smaller lesions are selected first (Janku et al., 2021).
Blood-culture clearance as the stop signal: Bacteria persisting in the blood were a dose-limiting toxicity in the attenuated-Salmonella trial; antibiotics are started if cultures remain positive beyond 48 hours, well past the murine clearance window (Toso et al., 2002).
Exclusion of prosthetic-material carriers: Circulating bacteria colonize artificial valves, joints and grafts, where eradication requires removing the device; screening them out prevents a device infection that antibiotics alone cannot clear (NCT05120596).
Planned antibiotic eradication at the end of therapy: The intravenous live-bacterium protocol gives antibiotics to all patients on completion, removing any residual organism and the risk of late dissemination (NCT05120596).
Holding immunosuppressive medication where feasible: Corticosteroids and transplant drugs both remove the immune contribution and raise dissemination risk, which is why trials exclude them outright rather than dose-adjusting (NCT05120596).

Therapeutic Protocol

No human protocol exists, because no person has received Ewingella americana as a therapy. Parameters other than doses that carry no citation — timing, cycling, titration steps — reflect common practice in early-phase live-bacterial oncology trials rather than an established regimen for this organism.

Only dosed regimen (preclinical): One intravenous dose of roughly 1 × 10⁹ colony-forming units (CFU, live bacteria able to grow) in mice — simultaneously the optimal and the maximum tolerated dose (Iwata et al., 2025).
Comparator doses in the same experiment: Anti-PD-L1 antibody and liposomal doxorubicin were each given intravenously at 2.5 milligrams per kilogram every other day for four doses, against the bacterium’s single dose (Iwata et al., 2025).
Competing approach, direct tumor injection: Both the monotherapy and the pembrolizumab-combination ceilings landed at 1 × 10⁶ spores injected once, with pembrolizumab 200 milligrams intravenously every 3 weeks (Nelson et al., 2025).
Competing approach, intravenous engineered bacterium: The one such trial gives low-dose desferrioxamine throughout to supply usable iron, and antibiotics to every patient on completion; dose is still being escalated (NCT05120596).
Who developed it: The approach belongs to Miyako’s group at the Japan Advanced Institute of Science and Technology, which has published the same intravenous single-dose design for three other wild-isolate bacteria (Miyahara et al., 2025; Iwata et al., 2026).
Best time of day: No data. Mouse dosing was not time-controlled and nothing in the mechanism implies a circadian window, though neutrophil counts and tumor blood flow both vary across the day (Iwata et al., 2025).
Clearance rather than half-life: A live organism has no pharmacological half-life. Blood counts fell to undetectable by 24 hours while intratumoral load rose roughly three-thousand-fold between 3 and 24 hours, setting the window for combination timing (Iwata et al., 2025).
Single versus split dosing: A single dose sufficed in mice; the authors propose fractionation for human use to avoid a peak systemic load, which has not been tested in any species (Iwata et al., 2025).
Genetic variants bearing on dose: None identified. The relevant candidates would be innate-sensing variants such as TLR4 and NOD2 and iron-handling variants such as HFE (the gene behind hereditary iron overload), none studied here.
Sex-based dose differences: None reported. The mouse experiments used one inbred strain and did not report sex-stratified dosing or response (Iwata et al., 2025).
Age-related dose considerations: None established. The one intravenous live-bacterium trial restricts enrolment to 18 to 74 years, the only age boundary anywhere in this field (NCT05120596).
Baseline biomarkers governing response: Iron stores, tumor hypoxia and neutrophil count are the mechanistically relevant inputs; the intravenous trial screens iron stores and organ function before dosing (NCT05120596). None has a response threshold.
Pre-existing conditions altering response: Tumor type and burden matter most; the mouse work used one colorectal line under the skin. Trials select advanced solid tumors with good performance status (NCT05120596); the authors propose requiring documented hypoxia (Iwata et al., 2025).

Discontinuation & Cycling

Single course by design, not lifelong: The regimen is one dose. Nothing in the mechanism implies maintenance dosing, and the durable rechallenge protection in mice suggests the effect, if real, outlasts the organism (Iwata et al., 2025).
Deliberate eradication rather than tapering: Therapy ends by killing the organism with antibiotics, as the intravenous live-bacterium protocol does for every patient on completion. There is nothing to taper (NCT05120596).
No withdrawal effects described: None reported in mice, and none plausible — the organism is cleared rather than withdrawn. What persists is immune memory, not dependence (Iwata et al., 2025).
Repeat dosing and cycling untested: No study has given a second dose. Antibody responses to the first exposure would plausibly accelerate clearance and blunt a repeat, which cuts against cycling as a strategy.
Stopping a checkpoint-inhibitor pairing: If paired with pembrolizumab, as both live-bacterium programmes do, the bacterium is a single dose while the antibody continues for up to 24 months, so the two stop at different times (Nelson et al., 2025).

Sourcing and Quality

No clinical-grade product exists: The therapeutic strain is a frog-gut isolate held by the originating laboratory and not publicly deposited. There is no supplier, no compounding route and no pharmacy that can furnish it (Iwata et al., 2025).
Reference strains are research-only: The type strain, catalogued as ATCC 33852 and NCTC 12157 in the public culture collections, is sold for laboratory use. It is a human clinical isolate, not the frog strain, and meets no standard for human administration.
What a usable lot would need: Manufacture under good manufacturing practice (GMP, the regulated quality system for medicines), a master cell bank with whole-genome strain identity, endotoxin and virulence-gene screening, viability counts, and a documented antibiotic-susceptibility certificate.
Strain identity matters more than purity here: Isolates of this species differ markedly in aggressiveness, and farm strains cluster in a clade closest to nonpathogenic environmental strains, so a lot verified only to species level is not verified (Hamidizade et al., 2023).
Food-borne isolates are not a source: The species is common on retail mushrooms, where it was the dominant gut-type bacterium identified across three cultivated species. That is contamination, not a therapeutic preparation (Reyes et al., 2004).

Practical Considerations

Time to effect: In mice, tumors became undetectable within days of a single dose, with intratumoral bacterial expansion already measurable by 24 hours. No human timeline exists (Iwata et al., 2025).
Common pitfalls: Reading a complete-response rate in five mice as a human expectation; conflating the frog isolate with clinical or mushroom isolates of the same species; treating a declaration of no competing interests as the absence of commercial interest.
Regulatory status: No regulatory approval exists for this organism anywhere. The only live bacterium licensed in oncology is BCG instilled into the bladder, for bladder carcinoma in situ and for preventing recurrence after resection (U.S. Food and Drug Administration, TICE BCG).
Cost and accessibility: Unobtainable — no product, no trial, no supplier. A single-dose live bacterium would be far cheaper to make than checkpoint antibodies, so payers and manufacturers face opposite incentives over what gets funded and studied, a structural bias in this field.

Interaction with Foundational Habits

Sleep: Indirect, no direct interaction established. The transient fever and inflammatory response after dosing in mice would be expected to fragment sleep for a few days, through the same inflammatory signalling that drives sickness behaviour. Nothing has been measured in people; no timing adjustment has evidence behind it (Iwata et al., 2025).
Nutrition: Indirect and potentiating through one channel only — iron. Bacterial growth inside the tumor depends on usable iron, which is why the intravenous live-bacterium protocol supplies desferrioxamine and excludes iron overload. Beyond iron status, no dietary pattern, fasting protocol or nutrient has been linked to this organism (NCT05120596).
Exercise: No direct interaction; indirect and potentially blunting. Heavy training raises circulating cortisol and transiently suppresses some immune functions, which overlaps the window when the immune arm of the effect is forming. The febrile phase after dosing also overlaps the period when exertion raises cardiovascular strain. Untested.
Stress management: Indirect and blunting. Sustained cortisol elevation damps the innate and T-cell recruitment the effect depends on, and trials exclude patients on any systemic corticosteroid for exactly that reason. Whether behavioural stress reduction changes the outcome has never been studied (NCT05120596).

Monitoring Protocol & Defining Success

No monitoring protocol exists for Ewingella americana, because no one has been dosed with it. The schedule below is assembled from what the mouse work tracked and from what the two live-bacterial oncology programmes monitor, and is presented as what would have to be watched rather than as an established protocol.

Before any dose, the relevant baseline is a safety baseline: blood cultures, a full blood count, liver and kidney chemistry, inflammatory markers, and iron studies, since the engineered-bacterium trial required adequate iron stores without overload.

During the first week after dosing, the live-bacterial trials monitor intensively — in the mouse work, blood counts and chemistry were sampled at 24, 48, 72 and 240 hours (Iwata et al., 2025). A human schedule would check blood cultures and inflammatory markers daily for the first 72 hours, then at 1 week, 4 weeks, and every 3 months while tumor response is followed.

Biomarker	Optimal Functional Range	Why Measure It?	Context/Notes
Blood culture, aerobic and anaerobic	No established target — track clearance to no growth	Safety check: bacteria persisting in the blood was a dose-limiting toxicity in the attenuated-Salmonella trial	Draw before dosing, then at 24, 48 and 72 hours; two sets from separate sites; draw before any antibiotic is given (Toso et al., 2002)
Absolute neutrophil count	1.5–8.0 × 10⁹/L (standard reference range)	Both: a safety check, since low counts mark the patients in whom this organism has caused invasive infection, and the cell population that rose most after dosing	Reported as part of a full blood count; a low count also defines who is excluded from live-bacterium trials (Iwata et al., 2025; NCT03435952 requires a count above 1.5 × 10⁹/L)
Platelet count	150–400 × 10⁹/L (standard reference range)	Safety check: thrombocytopenia (a low platelet count) was among the dose-limiting toxicities of intravenous attenuated Salmonella	Falls with sepsis and with consumptive clotting; interpret alongside blood cultures rather than alone (Toso et al., 2002)
C-reactive protein	Below 5 mg/L (standard reference range)	Marker the intervention is expected to change: the inflammatory response is the intended immune arm, so a rise is expected rather than alarming	Non-specific; interpret with procalcitonin and blood cultures, never alone
Procalcitonin	Below 0.1 ng/mL (standard reference range)	Safety check separating an intended inflammatory response from bacterial sepsis	A mild childhood case of this infection ran below 0.12 ng/mL; the one fatal case ran 113 ng/mL (Esposito et al., 2019; Tsokos, 2003)
Alanine aminotransferase	7–55 U/L (standard reference range)	Safety check: the mouse study screened liver chemistry for organ toxicity and found none, so a rise would be a new signal	A liver enzyme; draw with the baseline panel and again at 1 week (Iwata et al., 2025)
Serum creatinine with estimated filtration rate	Creatinine 0.6–1.3 mg/dL; filtration rate above 60 mL/min/1.73 m² (standard reference range)	Safety check: peritoneal infection during dialysis for kidney failure is among this organism’s most frequently reported human infections	Estimated glomerular filtration rate (eGFR) is a calculated measure of kidney filtering capacity (Ioannou et al., 2024)
Serum ferritin with transferrin saturation	Ferritin 30–300 ng/mL in men, 15–200 ng/mL in women; transferrin saturation 20–50% (standard reference range)	Safety check: iron overload is an exclusion in the one intravenous trial, and bacterial growth inside the tumor depends on usable iron	Fasting morning draw; ferritin rises with inflammation, so interpret with C-reactive protein (NCT05120596)
Tumor burden on imaging	No established target — track change from the individual’s own baseline measurements	Marker the intervention is expected to change: this is the efficacy endpoint	Baseline scan before dosing, then at 6–8 weeks; the same lesions measured each time

Qualitative markers worth tracking alongside the laboratory values:

Fever, chills and shaking in the first 72 hours after dosing
Pain, swelling or redness at a site where a tumor was injected directly, which rose transiently in the intratumoral spore trial
Fatigue and appetite, as the clearest signal of whether an inflammatory response is resolving
Sleep continuity through the febrile window

Emerging Research

No registered trial of this organism: A ClinicalTrials.gov search returns no study of Ewingella americana in any condition, and none of bacterial cancer therapy using a non-engineered wild isolate. Until one opens, every statement about human efficacy or human safety at a therapeutic dose remains an extrapolation from mice.
First-in-human intravenous live bacterium: NCT05120596, sponsored by T3 Pharmaceuticals, is dosing a genetically modified Yersinia enterocolitica intravenously and intratumorally in about 100 patients with advanced solid tumors, with estimated completion October 2027. A tolerable intravenous dose would make a trial of this organism conceivable; a toxicity failure would close the route.
Checkpoint pairing past its registered end date: NCT03435952 pairs pembrolizumab with intratumoral Clostridium novyi-NT spores; it is listed active, not recruiting, its 1 October 2026 completion date passed. Phase Ib results are published — 4 of 16 patients responded (Nelson et al., 2025) — enough to keep bacterium-plus-checkpoint combinations alive, not to establish them.
Completed intratumoral spore trial: NCT01924689 completed in 2017; its results showed tumor lysis in 10 of 24 patients alongside three life-threatening toxicities, and remain the most informative human safety dataset for a tumor-colonizing bacterium (Janku et al., 2021).
The same laboratory’s other wild isolates: Photobacterium angustum prolonged survival with direct tumor lysis (Miyahara et al., 2025) and a Proteus mirabilis–Rhodopseudomonas palustris consortium produced complete remissions (Iwata et al., 2026), both by the same route. Replication strengthens the platform; one group behind every result weakens confidence in any single strain.
Published critique of the platform: Li and Gong questioned the group’s earlier oncolytic-bacteria work (Li & Gong, 2025); Miyako replied defending the rationale, methods and outcomes (Miyako, 2025). The exchange concerns study design and translational claims, not data integrity, and neither side has been adjudicated by independent replication.
Antibiotic co-exposure as a confounder: In a pooled analysis of 38 studies, antibiotics given shortly before checkpoint-inhibitor therapy tracked with worse survival (hazard ratio 1.80, a measure of relative event rate; 95 percent confidence interval 1.44 to 2.26) (Yu et al., 2021). Rescue antibiotics could carry a similar penalty.
Native-organ and multi-model testing: The authors name models in which the tumor grows in its native organ, other tumor types and combination schedules as next steps (Iwata et al., 2025). A regression in a native-organ tumor would be the most informative next result; failure there would confine the finding to flank tumors.

Conclusion

Ewingella americana is an ordinary environmental bacterium carrying an extraordinary recent claim. One Japanese laboratory, screening gut bacteria from frogs, newts and lizards, reported that a frog-derived strain settled inside implanted tumors in mice, killed tumor cells directly, recruited immune cells, cleared the tumors, and left the animals able to reject a second tumor. In the same experiments it did more than a widely used immune-releasing antibody and a widely used chemotherapy drug.

All of that is animal and laboratory work from one group, in one tumor type placed under the skin. Nothing has been measured in a person. The organism’s own clinical history runs the other way: scattered reports of infection in the blood, abdomen, lungs and nervous system, almost always in already-weakened patients, and in one instance fatal. Some isolates have resisted nearly every antibiotic tried.

The authors declare no competing financial interests, though part of the funding came from a government programme for turning research into companies, and the firms running the adjacent human trials of other bacteria hold the same stake in the approach succeeding. A cheap single-dose living treatment would also sit awkwardly beside the far more expensive drugs it was compared with, and that shapes whose money flows toward testing it.

For anyone weighing early signals rather than settled practice, the honest position is that the strain cannot be obtained, no dose has ever been given to a person, and the distance between a cured mouse and a treated person is entirely unmeasured.

Top - Benefits - Risks - Protocol

Ewingella americana to Treat Cancer

Motivation

Recommended Reading

Grokipedia

Examine

ConsumerLab

Systematic Reviews

Mechanism of Action

Historical Context & Evolution

Expected Benefits

High 🟩 🟩 🟩

Medium 🟩 🟩

Low 🟩

Speculative 🟨

Complete Tumor Regression in a Mouse Colorectal Model

Durable Resistance to Tumor Rechallenge in Mice

Direct Killing of Tumor Cell Clusters in Culture

Benefit-Modifying Factors

Potential Risks & Side Effects

High 🟥 🟥 🟥

Medium 🟥 🟥

Low 🟥

Invasive Infection, Including Bloodstream Infection ⚠️ Conflicted

Multidrug-Resistant Isolates That Narrow Antibiotic Rescue

Speculative 🟨

Dose-Dependent Lethality Above the Mouse Tolerated Dose

Transient Systemic Inflammatory Response After Dosing

Seeding of Prosthetic Implants and Indwelling Devices

Secreted Toxins of Unmeasured Relevance to Mammals

Risk-Modifying Factors

Key Interactions & Contraindications

Risk Mitigation Strategies

Therapeutic Protocol

Discontinuation & Cycling

Sourcing and Quality

Practical Considerations

Interaction with Foundational Habits

Monitoring Protocol & Defining Success

Emerging Research

Conclusion